Mark-recapture results and changes in bat abundance at the cave of Na Turoldu, Czech Republic

Transcription

1

2 Folia Zool. 51(1): 1 10 (2002) Mark-recapture results and changes in bat abundance at the cave of Na Turoldu, Czech Republic Jifií GAISLER 1 and Josef CHYTIL 2 1 Department of Zoology and Ecology, Faculty of Science, Masaryk University, Kotláfiská 2, CZ Brno, Czech Republic; gaisler@sci.muni.cz 2 Administration of the Biosphere Reserve Pálava, Námûstí 32, CZ Mikulov, Czech Republic; jchytil@palava.cz Received 25 October 2000; Accepted 9 November 2001 A b s t r a c t. Between , 3,148 records of hibernating bats were made in a natural limestone cave, 92 % of them concerning Rhinolophus hipposideros. Other species included Myotis myotis, M. mystacinus, M. brandtii, M. emarginatus, M. nattereri, M. daubentonii, Plecotus auritus, P. austriacus and Myotis blythii (which was not recorded after 1973). The abundance of R. hipposideros (and of all bats combined) decreased between , probably due to the impact of mark-recapture work, and between due to the devastation of the cave. After the cave was put under protection and the marking of bats was stopped, the hibernating assemblage recovered so that, between , numbers of bats showed significant increases resulting in numbers more than twice those at the beginning of the monitoring period. Between , 517 bats were netted and banded from spring until autumn in front of the cave. Nettings of species not found in winter (Myotis bechsteinii, Eptesicus serotinus, Vespertilio murinus, Nyctalus noctula and Barbastella barbastellus) increased the total number of species to 15. In N. noctula and V. murinus echolocation signals were also detected and, in the latter a display flight with mating calls was recorded. Samples of netted bats were compared with samples of bats observed in the same winter, and significant differences were found in species composition, diversity and in the dominance values of individual species. It is suggested that considerable numbers of bats other than R. hipposideros occur at the locality in winter, but have remained undetected due to their hidden hibernation. We analysed 1,038 individuals marked at Turold and 237 recaptured, plus 10 of bats marked elsewhere and recaptured at Turold. The longest movement recorded (66 km) was in a female M. blythii, the highest age (19.5 years) was in a male R. hipposideros. Key words: hibernating bats, fluctuation in numbers, summer bat community, results of banding Introduction The cave of Na Turoldu is the largest known bat hibernaculum in the territory of the Pálava UNESCO Biosphere Reserve, and a previous study (Gaisler et al. 1996) has shown bats to represent the most diverse community among the reserve s small mammals. While this study reported data collected between , bats in the Na Turoldu Cave have been monitored from 1958 up to the present day. This 42 year period represents the third longest continuous bat monitoring record from any site in either the Czech Republic or the former Czechoslovakia, and is exceeded only by certain localities south of Prague (44 years) and in the Moravian Karst (43 years) ( ehák 1997, ehák & Gaisler 1999). Although the cave in question has not been visited every year, the sample is large enough to be analysed with respect to shifts in numbers of hibernating bats - one of the goals of this paper. However, the previous study also showed that certain species, although common in the territory of the 1

3 reserve, were found in the cave only rarely or not at all. Therefore, the second goal of this paper was to compare species composition and diversity, as well as the dominance values of individual species, between samples obtained both inside and outside the cave. By this we intend to demonstrate just how many bat species can occur within a very small territory (i.e. <1 ha - the cave interior plus its surroundings). Material and Methods Bats hibernating inside the cave have been monitored from The cave was visited once a year in January or February except in 1963 (when it was checked in March) and in , and 1983 (no surveys made). Until 1980 bats were banded and those found with a band were recaptured. To avoid any confusion when comparing papers about batbanding and bird-ringing we make it clear that, for bats, the terms banding and ringing are synonymous. The bands used to mark bats were very similar to the rings applied to passerine birds (for methodological details see Hanák et al. (1962) and Roer (1995)). From 1981 until the present day, bats found in the cave were censused without marking or disturbance, and only those banded previously, when found, were examined closely to check the band number (Bauerová et al. 1989). Ten species were recorded, the numbers of individuals in each sample are specified in the paragraph Winter census, and total numbers are given in Table 1. Table 1. Total number of records (N) and mean number of bats (Mean) per one winter check in the Na Turoldu Cave, ; n = 35 checks. Species N Mean Rhinolophus hipposideros Myotis myotis Myotis blythii Myotis mystacinus 1 <0.1 Myotis brandtii 1 <0.1 Myotis emarginatus Myotis nattereri Myotis daubentonii Plecotus auritus Plecotus austriacus indetermined Total From , netting was performed outside the cave, up to 20 m from its entrance. This entrance, however, is closed by a solid gate which does not allow the bats to fly in and out (see the locality description). Bats were netted irregularly from April until November each year, a period termed summer for the sake of simplicity. Usually three m long nets were set one hour before sunset and lifted shortly after midnight. Netted individuals were banded and released. Thirteen species were netted (the numbers of individuals in each sample will be presented elsewhere) and the total number of banded bats can be seen in Table 2 and those of recaptures in Table 3. The cave was not visited at each netting but, when visited, no bats were seen inside. When the two samples (winter and summer) were combined, the total number of species recorded was 15, and the total number of records 3,741. In addition to the above data, a few ultrasound records of Nyctalus noctula and Vespertilio murinus were made with a Skye bat detector. 2

4 The following abbreviations of bat species names are used in the list of records from the winter census: Rhip = Rhinolophus hipposideros, Mmyo = Myotis myotis, Mbly = M. blythii, Mmys = M. mystacinus, Mbra = M. brandtii, Mema = M. emarginatus, Mnat = M. nattereri, Mdau = M. daubentonii, Paur = Plecotus auritus, Paus = P. austriacus. Calendar dates are given with date first, followed by month and year, e.g = 10 February Further explanation of symbols used in the paper: m = male, f = female, T = Turold, both inside and outside the cave, O = other localities, B = banding (ringing), R = recapture, T/T = banded and recaptured at Turold, T/O = banded at Turold, recaptured elsewhere (analogously O/T), W = winter, inside the cave, S = summer, outside the cave (netting), W/1W = banded in winter, recaptured the next winter (analogously /2W, /3W = recaptured the second, the third winter after banding; S/ = banded in summer). Study Area The cave of Na Turoldu (= at Turold) is a natural limestone cave which consists of a complicated system of corridors and small domes forming three storeys situated at elevations between m a.s.l. The total length of known underground caverns is 1,100 m. There are numerous crevices in the ceiling, the walls and in the bases of underground corridors and domes. Some of the crevices open to the surface and obviously allow bats to move in and out, but the exact locations of these bat passages are unknown. Winter air temperatures in the cave vary between C (n = 16). The cave is named after the 385 m high hill Turold, situated on the periphery of the small town of Mikulov (population = 8,000). Between , Turold was affected by the output of Ernstbrunn dolomitic limestone. During the last 66 years, however, this limestone quarry has not operated and it has become overgrown - mostly by natural vegetation. The locality has been protected since 1946 and became a nature reserve in The quarry forms part of the Pálava Biosphere Reserve, which is situated in southern Moravia, close to the Czech- Austrian border. The Pálava Reserve (area = 83 km 2 ) consists of a core Upper Jurassic limestone ridge 12 km long and 3-4 km wide, with maximum elevation of 550 m (for further details see Gaisler et al. (1996) and âtyrok et al. (1998)). Results Winter census List of records: Rhip 51 m, 33 f; Mbly 2 m, 3 f; Mema 11 m, 12 f; Paus 1 m Rhip 31 m, 13 f; Mmyo 2 f; Mbly 5 m, 9 f; Mema 1 m; Paus 1 f Rhip 40; Mema Rhip 40, Mbly 2 m, 5 f; Mema 3 m, 1 f Rhip 55 m, 27 f; Mbly 2 m, 4 f; Mema 6 m, 4 f Rhip 51 m, 16 f; Mmyo 1 m, 1 f; Mbly 3 f; Mema 3 m, 7 f Rhip 31 m, 19 f; Mmyo 1 f; Mbly 2 m, 2 f; Mema 1 f Rhip 4 m, 2 f; Mema 5 f; Paus 1 m Rhip 8 m, 7 f; Mbly 1 f; Mema 2 m; Mnat 1 m; Mdau 1 f; Paus 1m Rhip 1 f; Mema 2 f; Paus 1 f Rhip 3 m, 1 f; Paur 1 m, 1 f; Paus 1 f Rhip 21 m, 11 f; Mnat 1 m; Paus 1 m Rhip 17 m, 12 f; Mema 2 m, 3 f; Mbra 1 m Rhip 11 m, 3 f; Mmyo 1 m; Mema 1 f Rhip 16 m, 4 f; Paus 2 m R hip 16 m, 16 f Rhip 23; Mema 1 f Rhip 21; Mema 1 m Rhip 53; Mema 5; Mnat 1; Paus Rhip Rhip 36; Mmyo 1; Mnat 9; Mdau 4; Paur Rhip 48; Mmyo 1; Mema 3; Mnat Rhip 62; Mema 3

5 Rhip 116; Mema Rhip 97; Mema 4; Mnat 3; Mdau Rhip 123; Mdau Rhip Rhip 142; Mema Rhip 172; Mmyo 1; Mema 6; Mdau Rhip 208; Mmyo 2; Mema 11; Mnat 1; Mmys 1; Myotis sp Rhip 243; Mmyo 1; Mema 4; Mnat 7; Mdau Rhip 202; Mema 3; Mnat 3; Mdau 1; Paus Rhip 201; Mema 8; Mnat 5; Mdau 2; Paur Rhip 200; Mema 13, Mnat 1; Paur Rhip 214; Mema 6. Rhinolophus hipposideros was the most common bat recorded hibernating in the cave, with a mean of 83 individuals per survey for the whole monitoring period (Table 1). Myotis emarginatus was the second commonest, with an average of only four individuals per survey. With regard to other species, the biggest difference between early and subsequent years was the decline and disappearance of M. blythii. Up to 14 M. blythii per survey were recorded between , and up to four between After 1973 the species could no longer be found in the cave. The remaining species were recorded in low numbers (average one individual per survey, see Table 1). Changes in abundance The fluctuations in numbers of R. hipposideros and of all other bats combined (Fig. 1) are characterised by the following trends: (1) a decrease from , (2) a decrease from , and (3) an increase after 1983 until the end of monitoring. The first two decreasing trends are statistically insignificant due to small sample sizes (n = 3 and 6 years, respectively). In contrast, the increasing trend during the final years of monitoring is highly significant (p < 0.001). However, in species other than R. hipposideros numbers are too small to show any trends. Rhinolophus hipposideros hibernates either singly or in groups termed colonies. Fluctuations in total numbers are reflected in fluctuations within the largest winter colony. If present, this assemblage has invariably been situated at the same place in the so-called Bat Fig. 1. Fluctuations in the numbers of hibernating bats recorded in the Na Turoldu Cave, (y axis = number of bats, x axis = year). The arrow pointing downwards denotes the year the gate closing the cave was broken down, the arrow pointing upwards represents the year a new gate was installed, thus closing the cave once more. 4

6 Dome. At the beginning of research (in February 1958) the colony consisted of 40 individuals (see G aisler et al. 1990: plate 1). The colony disappeared in the early 1970s and was relocated in In 1989 it consisted of 61 individuals and continued to increase to 200 individuals by During the last survey (in January 2000) we counted 170 individuals. Possible reasons for the shifts in numbers observed, such as the breaking down of the gate closing the entrance to the cave in 1970 (Fig. 1) will be discussed. Summer netting and comparison with winter census During 31 netting sessions we banded 517 individuals from 13 species. In addition, netting resulted in 69 (13.4 %) recaptures of eight species. Three species were recorded at the locality only by netting, i.e. M. bechsteinii, E. serotinus and B. barbastellus, and two more, V. murinus and N. noctula, were detected by bat detector. During the netting on , both a display flight and the mating calls of V. murinus were recorded. The bat, obviously a male, flew in circles from till h (Central European Time) and uttered loud calls at about 15 khz in a regular sequence. Since bats were only netted between , the sample of comparable winter records was limited (Table 2). The differences between the winter and the summer samples included both data on species composition and data on the quantitative representation of each species recorded. The differences between the numbers of individuals of the eight species common to both were significant (χ 2 Goodness-of-fit test, p < 0.001). While about ten times more bats were recorded per one winter survey than per one summer netting, the overall species diversity was eight times higher in the summer than in winter. The winter sample was dominated by one superabundant species, R. hipposideros, which accounted for 95.5% of all records. In contrast, the quantitative representation of species was more balanced in summer, with M. nattereri, M. daubentonii, R. hipposideros and P. auritus each accounting for >10% of all records. The ranks of species according to their dominance values are completely different when the two samples are compared (Table 2). Table 2. Comparison of results of the winter census and summer netting, N = number of bats seen or netted; D = species dominance (%); Mean = mean number of bats per one census or netting (n = 10 winter checks, 31 summer nettings); H' = species diversity. Species Winter Summer N D Mean N D Mean R. hipposideros M. myotis M. mystacinus M. emarginatus M. nattereri M. bechsteinii M. daubentonii E. serotinus V. murinus <0.1 N. noctula <0.1 P. auritus P. austriacus B. barbastellus <0.1 Total H'

7 M a r k - r e c a p t u r e a n a l y s i s During the entire time span , 1,038 bats of 15 species were banded at Turold (Table 3). Males were more commonly banded in all species with the exceptions of M. myotis and M. blythii. In the former, the two sexes were equally represented, but females dominated in the latter. Altogether, 237 recaptures were obtained, representing 22.8% of the total bats banded. Of these 161 (68%) were made in winter and 76 (32%) in summer. At least one recapture was obtained for nine species and 151 marked individuals. The recovery rate is higher for winter bats (30.9 %) than summer (14.7 %). There are 36 records of bat movements (15.2 % of all recaptures), including 26 concerning bats banded at Turold and recaptured outside this locality (15 R. hipposideros, 10 M. emarginatus, 1 M. blythii) and 10 of bats banded elsewhere and recaptured at Turold (6 R. hipposideros, 2 M. myotis, 1 M. blythii, 1 M. nattereri). Recapture data with respect to individual species are given below. Table 3. Survey of the number of all banded bats and of all recaptures, In the last column, the number of summer recaptures is in brackets. Species / Sex Winter Summer Total Recaptures males females males females banded R. hipposideros (23) M. myotis (1) M. blythii M. mystacinus M. brandtii M. emarginatus (15) M. nattereri (15) M. bechsteinii M. daubentonii (4) E. serotinus (1) V. murinus N. noctula P. auritus (13) P. austriacus (4) B. barbastellus Total (76) Rhinolophus hipposideros T/T recaptures: m W/1W 43 x, W/2W 14 x, W/4W 4 x, W/5W 2 x, W/6W 1 x, W/11W 1 x, W/12W 1 x, S/1W 12 x, S/2W 6 x, S/3W 5 x, S/4W 2 x, S/2S 1 x, S/3S 1 x; f W/1W 11 x, W/2W 6 x, W/3W 2 x, W/4W 1 x, W/7W 1 x, W/8W 1 x, W/12W 1 x, S/1W 5 x, S/2W 2 x, S/3W 1 x, S/4W 2 x, S/8W 1 x. In both males and females, the longest period between banding and recapture was 12 years and the individuals concerned were found in the cave (type W/W). While in type S/W the longest period was eight years in a female, and three years for an S/S type male, no recaptures of type W/S were made. There were 15 recaptures of type T/O - 10 of them W/S. The bats were recaptured in the summer of the same or subsequent year, mostly at Lednice or Bfieclav, up to 25 km east of the banding site. One record is exceptional: f, B , was repeatedly found hibernating in the cave on , and ; on it was pregnant and recaptured in a breeding colony in the loft of a building at Bulhary, Panensk ml n, (7 km NE). Another record represents the only recapture abroad: f, B

8 76, R Rabensburg, Austria, (27 km SE). The following recaptures of bats banded at Turold in summer are also worth mentioning: m, B , R artificial cave, Lednice, (13 km E); f, B, , R in a cellar at Janohrad, Lednice, (14 km E); f, B , R in a cellar at Insel, Mikulov, (4.5 km SES); m, B , R game reserve feeding trough, Klentnice, (3 km NE). Four recaptures of type O/T represent records showing the connection between the caves of the Moravian Karst and Turold: 2 m, B Jedovnické propadání Cave, , R Na Turoldu Cave, (58 km S); m, B Jedovnické propadání Cave, , R Na Turoldu Cave, (same distance, but after 19 years); f, B Ochozská jeskynû Cave, , R Na Turoldu Cave, (49 km S). Two O/T recaptures are of females who moved from breeding colonies to hibernacula: f, B Lednice, castle loft, , R Na Turoldu Cave, (13 km W); f, B Lednice, castle loft, , R Na Turoldu Cave, (same). The highest numbers of R. hipposideros were marked during winter checks in 1969 and On the former, 54 m and 27 f were banded, and 21 m and 12 f on the latter date. There were 30 recaptures (56%) of males banded in 1969 and recaptured in 1970, but only four (15%) of the females treated at the same time. This species has the second highest overall recovery rate (36.9%) i.e. 149 recaptures of 404 bats banded (Table 3). Myotis myotis Two recaptures of the O/T type: m, B B ãí skála Cave, Moravian Karst, , R Na Turoldu Cave, (53 km S); m, B Bulhary, netted in Milovice Forest, , R Turold, netted, (6 km W). As the second animal was recaptured only five weeks later (i.e. within the same season) the two localities probably belonged within its foraging range. The recovery rate was low (5.7%) for this species. Myotis blythii T/T recaptures: m W/3W 1 x, W/10W 2 x, W/11W 1 x, W/12W 1 x; f W/1W 3 x, W/2W 2 x, W/3W 1 x, W/9W 1 x, W/11W 1 x. The last three records of males and the last two of females concern only one animal each. The male was banded in 1959 and recaptured in 1969, 1970 and 1971 in the cave Na Turoldu, but it was also recaptured in Katefiinská Cave, Moravian Karst, (63 km N). The female was banded in 1959 and recaptured in 1968 and 1970 at the same place. There is one record of type O/T: f, B Erichova Cave, Moravian Karst, , R Na Turoldu Cave, (66 km S). Myotis blythii had the highest recovery rate (46.4%). Myotis emarginatus T/T recaptures: m W/1W 1 x, S/1S 1 x, S/2S 1 x, S/3S 1 x, S/7S 1 x, S/2W 1 x; f W/1W 1 x, S/2S 1 x. In addition to recaptures at the same locality, there are 10 cases of movements (T/O) - all of them concerning females banded in winter inside the cave and recaptured at summer breeding colonies at Lednice. The breeding colonies were situated in three different lofts of the central Lednice Castle, as well as in lofts of two smaller hunting castles up to 13 km away (Rybniãní and Tfii Grácie). One animal was checked six times: f, B , R Lednice, loft, , , , , It was immature in 1970 and either pregnant or lactating on subsequent recaptures. The overall recovery rate was 18 % in this species. Myotis nattereri T/T recaptures: m S/S (same season) 1 x, S/1S 8 x, S/2S 4 x, S/3S 1 x, S/4S 2 x, S/1W 2 x, S/2W 2 x, S/3W 1 x; f S/5W 1 x. 7

9 One movement (O/T) was documented between a forest habitat and a hibernaculum: m, B Pavlov, netted at Soutûska, , R Na Turoldu Cave, (5.5 km S). Myotis nattereri had a low recovery rate (11.6%), but there were large differences between the sexes: in males the recovery rate was 15.5%, but only one female was recaptured (1.8%). Myotis daubentonii T/T recaptures: m S/S (same season) 1 x, S/1W 1 x, S/1S 2 x; f S/1W 1 x. The lack of data is related to the very low recovery rate (5.1%). Eptesicus serotinus T/T recapture: f - S/2S 1 x. With only one recapture, the recovery rate for this species (3.4%) was the lowest recorded. Plecotus spp. T/T recaptures: P. auritus: m W/1S 1 x, S/1S 5 x, S/2S 3 x, S/4S 2 x, S/5S 1 x, S/2W 2 x; f S/3S 1 x. P. austriacus: m S/S (same season) 1 x, S/5S 1 x, S/6S 1 x, S/1W 4 x, S/5W 1 x; f W/2W 1 x, W/3W 1 x, S/1S 1 x. The recovery rates were high: 23.8% in P. auritus and 25.6% in P. austriacus. Discussion The results of our checks show that Na Turoldu Cave hosts large numbers of R. hipposideros but negligible numbers of bats of other species. This resembles, for example, Hermannshöhle in Austria (Baar et al. 1986) and certain caves in northern Moravia ( ehák & Gaisler 1999). Due to its exposed hibernation, R. hipposideros can be found relatively easily so that numbers ascertained at a hibernaculum are close to reality (Bezem et al. 1964). As the species is sedentary (Roer 1995, S c hober & Grimmberger 1998) changes observed in its abundance at large hibernacula may reflect changes in the population density of the local population. This, however, does not seem to be the case for the decrease recorded at the beginning of our research. It seems most likely that the decreases from were the result of disturbance to the bats associated with banding (Reiter 1998). On the other hand, the decrease between was the result of disturbance by unauthorised human activity. In summer 1971, the gate closing the entrance to the cave was broken down and, during subsequent visits by intruders, the cave was devastated. The finding of only one R. hipposideros and a total of four bats in 1974 undoubtedly reflect this impact. The cave was closed and again put under protection in autumn The period (and probably until 1983, although there was no survey that year) can be viewed as representing the recovery of the hibernating assemblage, when the number of bats started to grow (Fig. 1). The absence of banding and other negative impacts upon hibernating bats seems to be only one reason for this positive trend. Even more obvious than the increase in the total numbers of R. hipposideros recorded in the cave, is the increase in the size of the winter colony from 40 in 1959 to 200 in 1996, i.e. five fold. This corresponds with data from certain other hibernacula situated close to the species northern border in Europe. After a general decrease in R. hipposideros numbers in the 1960s and 1970s (Daan et al. 1980, Ohlendorf 1995) numbers stabilised and started to increase (McAney 1994, ehák 1997, ehák & Gaisler 1999). This increase may reflect the successful breeding of local lesser horseshoe bat populations over the last fifteen years or so. Is the superabundance of R. hipposideros observed in Na Turoldu and some other caves real or an artefact of the sampling method? We do not believe that the large numbers of bats (e.g. M. nattereri, M. daubentonii or P. auritus) caught in summer in front of the cave (Table 2) do not 8

10 hibernate at the locality. These, and some other species netted at Turold, tend to hide in protected places when hibernating, and so are difficult to find (Bezem et al. 1964, S c hober & Grimmberger 1998). Baagoe et al. (1988) found that the number of M. daubentonii recorded departing from a Danish hibernaculum in spring was much greater than the number of hibernating individuals of the same species recorded by a visual census inside the same quarters in winter. Similar results were obtained in a study of M. nattereri and M. daubentonii wintering in a bunker in Poland (Jurczyszyn 1998). In both species the number of departing bats netted in late winter and spring was higher (nearly two times so in some years) than the number of bats seen inside. We therefore suspect that an unknown, yet considerable, number of bats netted outside the cave hibernate inside it. If this is the case, hibernation must occur either in protected places inaccessible to an observer, and/or they hibernate in deep slots in the rocks near to the outside the cave, which also makes them difficult to locate. If this is true, then the actual species diversity of the hibernating bat community must be greater (and the dominance of R. hipposideros lower) than that revealed from the winter samples. Irrespective of this rather speculative conclusion, the number of 15 species (nearly the half of European chiropterofauna) recorded at the locality is high and warrants Turold s the status as a nature reserve. The results of marking and recapturing bats are generally in agreement with those from previous bat banding studies in the former Czechoslovakia (H anák et al. 1962, Gaisler & Hanák 1969) - for example, the finding that R. hipposideros moves up to 27 km between summer and winter quarters. Similar movements between summer nursing quarters and a hibernaculum up to 13 km away have also been recorded in M. emarginatus (cf. Gaisler et al. 1989). New findings, however, are the movements between Turold and the Moravian Karst, i.e. between winter quarters up to 66 km, as recorded in R. hipposideros, M. myotis and M. blythii. A male R. hipposideros, banded in Moravian Karst area, was recaptured at Na Turoldu Cave after 19 years. As this individual must have been born in the summer prior to marking, its age at recapture was at least 19.5 years. Several recaptures of R. hipposideros and one in each M. myotis and M. nattereri document movements between a shelter and a foraging area or between two foraging areas at distances of up to 6 km. The overall recovery rate for banded bats of 22.8% is high (Hanák et al = 8.1%, Gaisler & Hanák 1969 = 9.1%, Hanák et al = 9.7%). Recovery was higher in winter (30.9%) than in summer (14.7%) because of the much higher representation of R. hipposideros in the winter samples. If we omit M. blythii, the actual status of which is unclear (cf. Gaisler et al. 1988), R. hipposideros has the highest recovery rate of 36.9%. This results not only from its exposed hibernation, but also from its sedentary way of life. Sedentary Plecotus spp. also have high recovery rates. R eiter (1998) analysed recapture data at a locality where R. hipposideros was absent and found recovery rates of 20 25% in Plecotus, Barbastella and Eptesicus, while the recovery rates of Myotis spp. were lower. Masing et al. (1999) showed that second year recapture rates usually varied between 20 50% in sedentary bat species in Estonia, but that values were significantly lower in migratory species. We have only few records of bats banded in winter and recaptured in summer (or vice versa) at Turold. This may (in part?) be due to the different times at which we banded bats: hibernating bats were only banded in , and bats netted outside the cave only in Another source of error may be the potentially lower trapability of rhinolophids compared to vespertilionids in summer due to their echolocation differences. The bias makes it difficult to assess differences in interseasonal recovery rates for individual species, which might contribute to elucidating the problem of their real representation at the locality in different times of year. 9

11 A c k n o w l e d g e m e n t s We thank the following colleagues and friends who took part in the fieldwork, the results of which are subject of this paper. In alphabetic order (titles and first names omitted): BaroÀ, Bartáková, Bauerová, Beljanin, Gajdo ík, Hanák, Hoat, Chytilová, KaÀuch, Klima, Koutn, Macháãek, Málek, Málková, Matu ka, Nepra, Piro, Rödl, ehák, evãík, Tau, Vla ín, Zejda, Zima and Zukal. Beljanin is Russian, Hoat is Vietnamese, KaÀuch is Slovak and Klima, though a native Czech, has German citizenship. The other people are Czech citizens and mostly live in Brno or at Mikulov. We thank also V. Jarsk for providing mark-and-recapture data on bats from the database of the Czech Agency for Nature Conservation (Prague), and two anonymous reviewers for their comments. The research was funded by GA CR Grant No 206/99/1519. LITE R ATUR E BAAGOE, H. J., DEGN, H. J. & NIELSEN, P., 1988: Departure dynamics of Myotis daubentoni (Chiroptera) leaving a large hibernaculum. Vidensk. Meddr. dansk naturh. Foren., 147: BAAR, A., MAYER, A. & WIRTH, J., 1986: 150 Jahre Fledermausforschung in der Hermannshöhle. Ann. Nathist. Mus. Wien B, 88-89: BAUEROVÁ, Z., GAISLER, J., KOVA ÍK, M. & ZIMA, J., 1989: Variation in numbers of hibernating bats in the Moravian Karst: results of visual censuses in In: Hanák, V., Horáãek, I. & Gaisler, J. (eds.), European bat research Charles Univ. Press, Praha: BEZEM, J. J., SLUITER, J. W. & van HEERDT, P. F., 1964: Some characteristics of the hibernating locations of various species of bats in South Limburg. Proc. Koninkl. Nederl. Akad. Wet., Amsterdam, 67: âtyrok, P., DANIHELKA, J., CHYTIL, J., STUCHLÍK, J. & VIDLÁK, S., 1998: Pfiírodní rezervace Turold [Nature reserve Turold]. ARC Mikulov, 24 pp. (in Czech). DAAN, S., 1980: Long term changes in bat populations in the Netherlands: a summary. Lutra, 22: GAISLER, J., BAUEROVÁ, Z., VLA ÍN, M. & CHYTIL, J., 1988: The bats of S-Moravian lowlands over thirty years: Rhinolophus and large Myotis. Folia Zool., 37: GAISLER, J. & HANÁK, V., 1969: Ergebnisse der zwanzigjährigen Beringung von Fledermäusen (Chiroptera) in der Tschechoslowakei: Acta Sc. Nat. Brno, 3 (5): GAISLER, J., CHYTIL, J. & VLA ÍN, M., 1990: The bats of S-Moravian lowlands (Czechoslovakia) over thirty years. Acta Sc. Nat. Brno, 24 (9): GAISLER, J., VLA ÍN, M. & BAUEROVÁ, Z., 1989: The bats of S-Moravian lowlands over thirty years: small Myotis. Folia Zool., 38: GAISLER, J., ZUKAL, J., NESVADBOVÁ, J., CHYTIL, J. & OBUCH, J., 1996: Species diversity and relative abundance of small mammals (Insectivora, Chiroptera, Rodentia) in the Pálava Biosphere Reserve of UNESCO. Acta Soc. Zool. Bohem., 60: HANÁK, V., GAISLER, J. & FIGALA, J., 1962: Results of bat-banding in Czechoslovakia, Acta Univ. Carolinae Biol., 1962 (1): HANÁK, V., REITER, A. & BENDA, P., 1996: Pfiehled terestrick ch obratlovcû Ledov ch slují (The review of terrestric vertebrates of the Ledové sluje area). Pfiíroda sbor. prací z oboru ochr. pfiír., Praha, 3: (in Czech, with a summary in English). JURCZYSZYN, M., 1998: The dynamics of Myotis nattereri and M. daubentoni (Chiroptera) observed during hibernation season as an artefact in some type of hibernacula. Myotis, 36: MASING, M., POOTS, L., RANDLA, T. & LUTSAR, L., 1999: 50 years of bat-ringing in Estonia: methods and the main results. Plecotus et al., Moskva, 2: McANEY, C. M., 1994: The lesser horseshoe bat in Ireland past, present and future. Folia Zool., 43: OHLENDORF, B. (ed.), 1995: Zur Situation der Hufeisennasen in Europa. Tagungsband, Nebra, Deutschland, 182 pp. REITER, A., 1998: Is banding a real threat to bats? Vespertilio, Revúca, Praha, 3: ROER, H., 1995: 60 years of bat-banding in Europe results and tasks for future research. Myotis, 32-33: EHÁK, Z., 1997: Long-term changes in the size of bat populations in Central Europe. Vespertilio, Revúca, Praha, 2: EHÁK, Z. & GAISLER, J., 1999: Long-term changes in the number of bats in the largest man-made hibernaculum of the Czech Republic. Acta Chiropterologica, 1: SCHOBER, W. & GRIMMBERGER, E., 1998: Die Fledermäuse Europas. Kosmos, Stuttgart, 265 pp. 10

12 Folia Zool. 51(1): (2002) Diet estimation by faeces analysis: sampling optimisation for the European hare Krisztián KATONA 1 and Vilmos ALTBÄCKER 2 1 St.Stephen University, Department of Wildlife Biology and Management, Páter K. u. 1, H-2103 GödöllŒ, Hungary; katonak@ns.vvt.gau.hu 2 Eötvös Loránd University, Department of Ethology, Jávorka u. 14, H-2131 Göd, Hungary; altbac@ludens.elte.hu Received 20 June 2001; Accepted 30 October 2001 Abstract. We investigated how the sampling process of microhistological faeces analysis could be optimised for an accurate estimation. Spring diet composition of European hare (Lepus europaeus Pallas, 1778) was determined in a juniper shrubland at Bugac, Hungary. Both inter and intraobserver reliability was high permitting us to separate the components of variance due to the methodological steps in the faeces analysis. Estimates varied depending on the number of independent droppings, pellets/individual, subsamples/pellet and epidermis/subsample. The variance was much higher among than within the independent pellet groups. The cumulative frequency estimate stabilised at around 100 epidermis fragments per pellet. We conclude that the most critical steps of the sampling procedure are the collection of independent droppings and the identification of a sufficient number of epidermis fragments. We propose to collect at least 10 independent droppings, one pellet/individual, and analyse 100 epidermis fragments as an optimum for estimating the relative frequency of forage classes reliably. The importance of the individual variability in the diet should be emphasised. Key words: microhistology, faecal pellet, accuracy, minimum sample size, Lepus europaeus Introduction Feeding habits of mammals are in the centre of interest of population biology (Lodé 1996) and ecology (Mátrai et al. 1998). Several methods have already been involved in the investigation of food composition (Holechek et al. 1982). The most widely used indirect technique for determining diet composition of herbivores is the microhistological identification of epidermis fragments in the stomach content or faecal pellet (B a u m g a r t n e r 1939, Dusi 1949). The effect of sample size on the estimated diet composition has already been proved (Hanson & Graybill 1956). The greater the diet diversity of a species and the smaller the similarity of individuals, the larger the required minimum sample size is (K o v á c s & T ö r ö k 1997). Data on individual variability in the diet of herbivorous mammals, however, are very scarce (M átrai & Kabai 1989, H omolka & Heroldová 1992). Microhistological faeces analysis method includes multiple successive sampling from the individuals, pellets and epidermis fragments. Sampling size, therefore, could affect the estimate in all consecutive sampling steps. The lack of the information on the precision of the faecal sampling has already been emphasised (Holechek et al. 1982). The applied minimum sample size could vary according to study species, as for deer (Anthony & Smith 1974), hare (H omolka 1987a) and rabbit (C hapuis 1980). Sample size, 11

13 however, could also differ from study to study as the characteristics and size of study area change (Hanson & Graybill 1956, Homolka 1987b). Problems to design a valid diet composition analysis derive from the difficulties of the determination of the sample sizes required. To optimise the sampling process, a general knowledge on the variances of different sampling levels should be established. For this comparison a single basic sample of faecal pellets is needed, which has never been used by earlier investigators studying the minimum sample sizes on consecutive sampling levels. The aim of our investigation was to assess a valid sampling strategy to obtain precise results. In order to optimise faecal sampling, therefore, we compared the variances observed at the consecutive steps of faeces analysis in European hare (Lepus europaeus Pallas, 1778). Materials and Methods The study area (1200 by 1400 m, 168 ha) was situated in the Bugac Juniper Forest core area in Kiskunság National Park, central Hungary (46 38 N; E). This protected area has shrubland vegetation (for a detailed description of it see Kertész et al. (1993) and Kalapos (1989)). Fresh faecal pellets of European hare were collected on a single day, 22 April, Altogether 10 independent droppings, each containing five pellets, were collected. (Faeces of European hare is called droppings, which could include only one or several faecal pellets). Sampling points were separated by at least 200 m intervals (x ± SD = 578 ± 272 m). Spring diet composition of hares was determined by microhistological faeces analysis. Proportion of diet components was estimated in each pellet by the number of fragments for a particular forage class relative to the total number of fragments. The distinguished forage classes were monocotyledons (grasses), dicotyledons (forbs and browses), gymnosperms (evergreens), barks and seeds. For microhistological analysis the following preparation procedures were conducted (Mátrai et al. 1989). A subsample mixed from five parts of g was taken out from each pellet into test tubes. From five pellets of different droppings a subsample of one part of 0.01 g was taken parallely out into five other test tubes. After boiling in 20 % nitric acid solution for 1.5 minutes epidermis fragments were dispersed on microscopical slides into a mixture of 0.1 ml of 87 % glycerine and 0.05 ml of 0.1 % Toluidine-Blue. All fragments found on the slides were identified under 160X magnification. We had the following investigations. A) One pellet from each droppings was chosen and 200 epidermis fragments of it were identified. Just the minimum required number of fragments (100) were then categorised from the remaining pellets. Minimum sample sizes for different sampling units were determined by rarefaction analyses (Kovács & Török 1997) and by saturation curves of the proportion of forage categories. B) Intra and interobserver reliability were determined by analysing one pellet from each droppings two times. Statistical analysis of observer reliabilities was performed by Spearman-correlation (Caro et al. 1980). C) Diet composition estimated from the same pellets by subsamples including one bigger or five smaller parts were compared by Wilcoxon matched-pairs signed rank test. D) Determining diet composition of all five pellets of five droppings variances within and among droppings were tested by Kruskal-Wallis one-way analyses of variance by ranks. 12

14 To quantify diet overlaps between different samples Renkonen s proportional similarity index was always calculated (H urlbert 1978): S is = Σmin(P 1,i ; P 2,i ) where P 1,i is the proportion of prey category i in one individual, P 2,i in the other individual. Similarity and diversity indices and rarefaction curves were calculated using the software NICHE (Schluter, D., University of British Columbia, Vancouver). Results There was no significant negative correlation between the distance of sampling points of droppings and their diet similarity (Spearman-correlation: N=45, r s =0.24, p=0.12) supporting the independence of droppings. A) Rarefaction analyses and saturation curves indicated that a sample size of 100 epidermis fragments (Fig. 1), 10 independent droppings and one pellet of a single droppings were surely sufficient for obtaining unbiased estimates of diet composition. B) Intra and interobserver reliability were measured to be high (Spearman-correlation: N=10, r s >0.8 in all cases for monocots, dicots and gymnosperms; r s >0.6 for seeds and barks). C) Diet composition estimated from either a single 0.01 g subsample or from a mixture of five g subsamples seemed to be identical (Wilcoxon matched-pairs test: N=5, p>0.38 for all forage classes). D) Based on the analysis of all five pellets of five droppings, diet composition of pellets significantly differed among droppings (Kruskal-Wallis test: df=4, p<0.05 for all forage classes) (Fig. 2), but within droppings they were identical (Kruskal-Wallis test: df=4, p>0.7 for all forage classes). In the course of sampling procedure similarity was the lowest, therefore variance was the highest among independent droppings as shown in Table 1. Fig. 1. Stabilisation curves of epidermis fragments identifying to broad forage classes. Figure shows typical curves to determine an optimum sample size of the epidermis fragments based on the analysis of 200 fragments to broad forage classes in a single pellet. Relative occurrence was given by the number of fragments for a particular forage category relative to the total number of fragments. Confidence interval of five percent is indicated. 13

15 Fig. 2. Variability in the diet composition among independent hare droppings. Diet composition of a single pellet from each droppings is provided after identifying 200 epidermis fragments per pellet to broad forage classes. Relative occurrence was given by the number of fragments for a particular forage category relative to the total number of fragments. Table 1. Variability in the sampling procedure of microhistological faeces analysis. Sample size (N), value of Renkonen s proportional similarity index (Sis) and statistical significance of differences for the given sampling factors are shown. Sampling variability Sample size (N) Proportional similarity (Sis) Significance Among droppings p<0.05 Within droppings NS Between subsamples NS Interobserver reliability NS Observer1 reliability NS Observer2 reliability NS Discussion Our results suggest that sampling procedure could significantly affect the result of microhistological analysis. In our study, the largest contribution to overall variance was revealed among droppings, indicating considerable individual differences among the hares. Mátrai & Kabai (1989) and Homolka & Heroldová (1992) have already proved considerable individual differences in deer species. We have also found a low similarity among independent droppings of hares, but cumulative composition analysis indicated that 10 droppings gave a reliable estimate. This is very similar to the minimum sample size suggested by Homolka (1987a) for hares and by Chapuis (1980) for rabbit colonies inhabiting a well-defined smaller habitat. This individuality of diet composition casts doubt on the validity of many diet analyses done from a faecal pellet mixture. These investigations might reveal the pressure on the vegetation by the given herbivore population rather than the actual food choice of the individuals. Similarity of the pellet compositions was much higher within than among droppings, similarly to Chapuis (1980), suggesting the importance to collect more droppings and only one pellet/individual. 14

16 Number of identified fragments should be the other main factor of sampling design as Homolka (1987a) and C hapuis (1980) have already pointed out. There is no consensus in the literature on the minimum sample size, but it often seems to be larger than necessary. In rabbits in our study area Mátrai et al. (1998) identified fragments in different seasons. C hapuis (1980) applied a sample size of 400 fragments, but estimated 200 to be sufficient. In summary, the different steps in the sampling procedure influence the accuracy in different ways. The two major factors in faecal sampling optimisation are the sample size of droppings and epidermis fragments. Estimation of diet composition is less sensitive to the observer reliabilities and to within-droppings and within-pellet variance. We are confident that our directives will help to design future faecal analyses to gain accurate and biologically interpretable results with optimal effort and without underestimating the variety of dietary habits of animals. Acknowledgements We are indebted to K. Mátrai for her help in applying microhistological techniques and to K. Nagy for her participation in the interobserver reliability study. Two anonymous referees made constructive comments on the manuscript, while L. Sári revised the final English version. Kiskunság National Park permitted us to work in the protected study area. The work was funded by the Hungarian Science Foundation (T to V. Altbäcker). LITERATURE ANTHONY, R. G. & SMITH, N. S., 1974: Comparison of rumen and faecal analysis to describe deer diets. J. Wildl. Manage., 38(3): BAUMGARTNER, L L. & MARTIN, A. C., 1939: Plant histology as an aid in squirrel food-habits studies. J. Wildlife Manage., 3: CARO, T. M., ROPER, R., YOUNG, M. & DANK, G. R., 1980: Inter-observer reliability. Behaviour, 69 (3 4): CHAPUIS, J. L., 1980: Méthode d étude du régime alimentaire du lapin de garenne, Oryctolagus cuniculus (L.) par l analyse micrographique des fèces. Rev. Ecol.- Terre Vie, 34: (in French, with a summary in English). DUSI, J. L., 1949: Methods for the determination of food habits by plant microtechniques and histology and their application to cottontail rabbit food habits. J. Wildlife Manage., 13(3): HANSON, W. R. & GRAYBILL, F., 1956: Sample size in food-habits analyses. J. Wildlife Manage., 20(1): HOLECHEK, J. L., VAVRA, M. & PIEPER, R. D., 1982: Botanical composition determination of range herbivore diets: a review. J. Range Manage., 35(3): HOMOLKA, M., 1987a: Problems associated with investigations to the diet of the European hare. Folia Zool., 36(3): HOMOLKA, M., 1987b: The diet of brown hare (Lepus europaeus) in central Bohemia. Folia Zool., 36(2): HOMOLKA, M. & HEROLDOVÁ, M., 1992: Similarity of the results of stomach and faecal contents analyses in studies of the ungulate diet. Folia Zool., 41(3): HURLBERT, S. H., 1978: The measurement of niche overlap and some relatives. Ecology, 59(1): KALAPOS, T., 1989: Drought adaptive plant strategies in a semiarid sandy grassland. Abstracta Botanica, 13: KERTÉSZ, M., SZABÓ, J. & ALTBÄCKER, V., 1993: The Bugac Rabbit Project. Part I: Description of the study site and vegetation map. Abstracta Botanica, 17: KOVÁCS, T. & TÖRÖK, J., 1997: Determination of minimum sample size to estimate diet diversity in anuran species. Herpetol. J., 7: LODÉ, T., 1996: Polecat predation on frogs and toads at breeding sites in western France. Ethol. Ecol. Evol., 8: MÁTRAI, K. & KABAI, P., 1989: Winter plant selection by red and roe deer in a forest habitat in Hungary. Acta Theriol., 34(15): MÁTRAI, K., ALTBÄCKER, V. & HAHN, I., 1998: Seasonal diet of rabbits and their browsing effect on juniper in Bugac Juniper Forest (Hungary). Acta Theriol., 43(1):

17 Folia Zool. 51(1): (2002) Aggression in reciprocal crosses of two subspecies of wild house mouse Radka VOLFOVÁ, Pavel MUNCLINGER and Daniel FRYNTA Department of Zoology, Faculty of Science, Charles University, Viniãná 7, Praha 2, Czech Republic; Received 25 September 2000; Accepted 9 November 2001 Abstract. We studied reciprocal hybrids of Mus musculus musculus and M. m. domesticus. These two subspecies of house mouse were found to differ in their social behaviour, the former being less aggressive than the latter. The paternal effect on aggression (observed repeatedly in laboratory mice) was not found. However, F 2 generation mice were less aggressive than those from the F 1 generation, and the maternal effect was also significant in a homogenous test set. Key words: wild mice, hybrid zone, social behaviour, hybrid advantage, Y chromosome Introduction Aggression is an important component of mouse social behaviour. The level of aggression is species specific and may vary, even between populations or strains (B rain& Parmigiani 1990, G a ne m & Searle 1996, Sluyter et al. 1996b). At least in part, these differences are determined by genotype, as has been made clear from model experiments on laboratory mice (Cairns et al. 1983, Sluyter et al. 1996b). It has been repeatedly demonstrated that artificial selection has a strong and immediate effect on measures of aggression recorded in standardised tests (Hood 1988a, Hood 1988b, Sluyter et al. 1996b). Special attention has been paid to paternal effects, which have been repeatedly reported in studies quantifying mouse aggression. In their study dealing with reciprocal crosses of strains DBA/1/Bg and C57BL/10/Bg, Selmanoff et al. (1975) hypothesised that the Y chromosome was responsible for the variation observed in mouse aggression. Y chromosome correlates on aggression were subsequently shown in several inbred strains and in lines bidirectionally selected for attack latency (see review by Carlier et al. 1990). Moreover, some comparisons of congenic strains confirmed the Y chromosome effect on aggression - usually the more aggressive hybrid has the Y of the more aggressive strain (Guillot et al. 1995, Carlier et al. 1990, Mohanan & Maxson 1998). Clear, fixed Y chromosome differences have been found between the subspecies Mus musculus musculus L., 1758 and Mus musculus domesticus Schwarz & Schwarz, 1943 (Boissinot & Boursot 1997) and it is of interest that these subspecies also differ in their social behaviour. The phenomenon whereby M. m. domesticus males are more aggressive than those of M. m. musculus has been shown by authors studying mouse populations from the north-western part of Europe (Thuesen 1977, Zegeren& O o r t m e r s e n 1981). We have also demonstrated the same phenomenon by comparing the social behaviour of house mice from two distant populations (i.e. Bohemia and eastern Turkey), each of which belong to different subspecies (M u n clinger & Frynta 2000). Male mice from Turkey (M. m. domesticus) were undoubtedly more aggressive than Bohemian mice (M. m. musculus). 17

18 In this paper we have attempted to demonstrate the differences between reciprocal F 1 males from crosses of two subspecies of wild house mouse to prove the paternal effect on aggression; this is the obligatory first step for testing the Y chromosome effect. In addition, we also compared their descendants (F 2 hybrids) to randomise the effects of autosomal genes. Material and Methods Parental generation mice were born in the laboratory (about 4 th - 6 th out-bred generation in captivity). Mus m. musculus stock were derived from 20 males and 20 females captured in the Bohemian villages of âerno ice, Soutice and Satalice (Czech Republic) during winter Mus m. domesticus were descendants of eight pairs of mice captured in the vicinity of the town of Kilkis, Kilkis District (Greece) in May Reciprocal hybrids of the above populations were used as experimental subjects. Twenty males of each combination of generation (F 1 and F 2 ) and cross direction, i.e. female M. m. domesticus x male M. m. musculus (hereafter MdMmF 1 males) and female M. m. musculus x male M. m. domesticus (hereafter MmMdF 1 males) were used. Experimental mice were adult (i.e. at least two months old), socially experienced males housed in heterosexual pairs (i.e. together with an adult female of the identical filial generation and cross direction). All animals were kept under an artificial 12 L: 12 D light cycle and housed in pairs in 30 x 15 x 15 cm plastic cages. Water and food (ST1 mouse and rat breeder diet, wheat etc.) were provided ad libitum. Each cage contained sawdust bedding, nesting material (paper) and shelters. The experiments were performed in January-March A standard neutral cage procedure was used. Encounters between mice were carried out in a 50 x 30 x 35 cm glass cage, divided into two equal parts by a thick card partition. During testing, the cage was illuminated by a single 40 W red light bulb. Mice were tested during the dark phase of their light-dark cycle. At the beginning of each experimental session, two mice were placed in the pen on the opposite sides of the partition, and left for five minutes. The central partition was then removed and video recording by a single VHS camera commenced. The video camera was stopped at the end of the session (ten minutes after the moment at which one or both animals paid attention to the other for the first time). The cage was thoroughly cleaned using 96% ethanol after each session. Repeated tests on the same individual were undertaken at a minimum interval of 24 hours. In the first set we tested mice in homogenous dyads, i.e. with opponents of the same generation and cross direction. In this set 120 dyadic encounters were staged (i.e. 30 encounters for every combination of generation and cross direction). Each animal was tested with different opponents three times. In the second set, the same individuals were tested in heterogeneous dyads consisting of individuals belonging to the same generation, but different cross direction. 20 and 17 heterogeneous dyads were performed on the F 1 and F 2 generations, respectively. The video records of the encounters were subsequently observed, and the duration of behavioural elements was quantified using the computer program package ACTIVITIES (Vrba & Donát 1993). We used the standard catalogue of 34 behavioural elements for the purpose of data collection (see â iháková & Frynta 1996 for details). Selmanoff et al. (1976) and M a x s o n et al. (1979) suggested the any aggression score as being a more robust and reliable index of fighting behaviour than any single 18

19 agonistic act. Therefore, only the following four variables were derived from the primary data and used in further analyses: 1. Aggressive behaviour the sum of the time intervals spent in the following elements of aggressive behaviour: Threat, Aggressive upright, Attack, Chase, Roll-over fight. 2. Defensive behaviour the sum of the time intervals spent by defensive elements: Defensive upright, Defensive threat, Avoid, Retreat, Flee, Jump avoid, Freeze, Submissive posture. 3. Agonistic behaviour the total time spent in aggressive, defensive, and neutral (Neutral upright, Box, To-fro - repeated approach and avoidance, Tail rattling) elements of behaviour. We believe this variable is the best indicator of hostile motivation in interacting mice. This well-defined and continuous variable shows nearly log-normal distribution, so we used it as first choice in our testing. 4. Attack latency latency of the first attack expressed in seconds. This variable was highly correlated with time spent in agonistic behaviour (Spearman s r = -0.78, n = 157). The behaviour of both interacting animals in a particular dyad is obviously intercorrelated. Thus, in the first set of experiments dealing with comparisons of homogenous dyads we summed the scores of agonistic and aggressive behaviour recorded in both members of a single dyad. These were log-transformed and further treated by analyses of variance/covariance (Statgraphics v. 5.0). The scores for defensive behaviour as well as attack latency were treated by non-parametric statistics. Defensive behaviour was used solely for comparisons within heterogeneous dyads and analysed with Wilcoxon matchedpair tests. Results The mean time spent in agonistic behaviour in the first set of experiments is shown in Fig. 1. The proportion of time spent in agonistic behaviour by MdMmF 1 males (F 1 generation = 22%, F 2 = 12%) was higher than in MmMdF 1 males (F 1 generation = 11%, F 2 = 4%). Fig.1. Average total time spend in agonistic behaviour by homogenous dyads. The results of parental population tests are shown for comparison. Key: Md M. m. domesticus males, Mm - M. m. musculus males, F1: MdMm, F1: MmMd, F2: MdMm, F2: MmMd MdMm different generations and types of hybrids, see Material and Methods for more details. 19

20 Analysis of variance, as controlled for body weight (covariate F = 8.59, p = ), revealed significant differences in the time spent in agonistic behaviour between reciprocal hybrids (F = 18.27, p < ) and between the F 1 and F 2 generations (F = 6.51, p = ). The effect of interaction between the above factors was also important (F = 5.63, p = ). The differences between reciprocal hybrids were also supported when aggressive behaviour was treated separately (F = 3.99, p = ). Attack latency was not affected by the cross direction in the F 1 generation (Mann- Whitney test: z = -1.29, p = ) and only a marginal effect was found in the F 2 (z = -2.05, p = ). Attack latency was considerably longer in both MmMdF 1 and MdMmF 1 hybrids in the F 2 than in the F 1 generation (cross direction pooled: z = -3.84, p = ; average rank = 72.0 and 49.0, respectively). Unlike homogenous set tests, further experiments involving both reciprocal hybrids in a single dyad enabled matched pair comparisons. No effect of reciprocal hybridisation was found, either on the time spent in defensive behaviour (Wilcoxon matched-pair tests: z = , p = , n = 37 dyads, F 1 and F 2 generations pooled) or on attack latency (z = 0.28, p = ). The F 1 generation also remained more agonistic than the F 2 generation in this experiment (ANOVA: F = 5.79, p = ). Discussion We did not find any paternal effect on agonistic behaviour in male mice; moreover, we found a significant maternal effect. Surprisingly, the more aggressive hybrid did not carry the Y chromosome of the more aggressive subspecies. Studies carried out on laboratory mice have not led to consensus on how the Y chromosome determines the level of aggression in males. Roubertoux et al. (1994) failed to demonstrate the effect of the non-pseudoautosomal region of the Y on mouse aggression. Other studies have suggested influences from the pseudoautosomal region, genetic background and the maternal environment (Carlier et al. 1991, Roubertoux et al. 1994). However, there is some evidence that artificial selection for male aggression leads to a correlated response in female aggression (H ood & Cairns 1988). This finding provides indirect evidence for the effect of genes not located on the Y chromosome, because the Y chromosome is absent in females. To simulate natural conditions we kept mice in pairs, and the offspring had equal chances to be influenced by their fathers and their mothers. In spite of this, we found a significant maternal effect in the first set of experiments. We have no cross-fostering data to demonstrate that the differences between reciprocal hybrids are of genetic nature, and therefore prenatal and/or postnatal environmental effect cannot be excluded. Such environmental effects were previously revealed by studies of laboratory mice (Carlier et al. 1991, but see Sluyter et al. 1995, 1996a). It should be noted that aggression is a context-specific phenomenon (e.g. Suchomelová & Frynta 2000) and therefore details of experimental procedure can considerably influence results. In order to reduce stress, we studied mice in a standard neutral cage environment, i.e. outside the context of territoriality in which mice fight more seriously (G r a y & Hurst 1995), and we avoided the use of inbred laboratory mice as standard opponents. We suppose that this design is simply not applicable in wild mouse studies. Moreover, Guillot et al. (1995) showed that the standard opponent test is less 20

21 sensitive to parental (Y chromosome) effects than homogenous set tests. Nevertheless, social behaviour is determined by both interacting animals, so it is not surprising that the results of our first and second set were not the same. Unlike the total time spent in agonistic behaviour, attack latency was found to be only slightly effected by cross direction even though these two variables were highly correlated. This may suggest that the total time spent in agonistic behaviour is a more sensitive indicator of hostility than attack latency is. Despite differences between reciprocal hybrids, all the F 1 hybrids were highly aggressive and their mean scores for agonistic behaviour (201 s) even exceeded those previously recorded in the more aggressive (M. m. domesticus) parental population (191 s). We suggest that the fact that the F 1 hybrids consistently exhibited more agonistic behaviour than the F 2 in both experimental sets may be explained by recombination of the loci affecting agonistic behaviour. We conclude that paternal effects are not easily demonstrable in our wild mice crosses. The differences observed between reciprocal hybrids, as well as between F 1 and F 2 generations, in their social behaviour cannot be attributed to the Y chromosome as a single factor, and the possible effects of genes on other chromosomes, hybrid advantage, and even non-genetic maternal effects cannot be rejected. Acknowledgements We thank Dr. Radka D a n dová and Mr. Milan Kaftan for technical assistance in keeping the experimental animals. The project was supported by Ministry of Education, Youth and Sport of the Czech Republic (grant VZ ). The participation of P.M. was also supported by Ministry of Education, Youth and Sport of the Czech Republic (grant no. VS 97102). LITERATURE BOISSINOT, S. & BOURSOT, P., 1997: Discordant phylogeographic patterns between the Y chromosome and mitochondrial DNA in the house mouse: selection on the Y chromosome? Genetics, 146: BRAIN, P. F. & PARMIGIANI, S., 1990: Variation in aggressiveness in house mouse populations. Biol. J. Linn. Soc., 41: CAIRNS, R. B., MACCOMBIE, D. J. & HOOD, K.E., 1983: A developmental-genetic analysis of aggressive behaviour in mice: I. behavioural outcomes. J. Comp. Psychol., 97: CARLIER, M., ROUBERTOUX, P. L., KOTTLER, M. L. & DEGRELLE, H., 1990: Y chromosome and aggression in strains of laboratory mice. Behav. Genet., 20: CARLIER, M., ROUBERTOUX, P. L. & PASTORET, C., 1991: The Y chromosome effect on intermale aggression in mice depends on the maternal environment. Genetics, 129: âiháková, J. & FRYNTA, D., 1996: Intraspecific and interspecific behavioural interactions in the wood mouse (Apodemus sylvaticus) and the yellow-necked mouse (Apodemus flavicollis) in a neutral cage. Folia Zool., 45: GANEM, G. & SEARLE, J. B., 1996: Behavioural discrimination among chromosomal races of the house mouse (Mus musculus domesticus). J. Evol. Biol., 9: GRAY, S. & HURST, J.L., 1995: The effects of cage cleaning on aggression within groups of male laboratory mice. Anim. Behav., 49: GUILLOT, P.-V., CARLIER, M., MAXSON, S.C. & ROUBERTOUX, P.L., 1995: Intermale aggression tested in two procedures, using four inbred strains of mice and their reciprocal congenics: Y chromosomal implications. Behav. Genet., 25: HOOD, K. E., 1988a: Selective breeding - selective rearing interactions and the ontogeny of aggressive behavior. Behav. Brain Sci., 11:

22 HOOD, K. E., 1988b: Female aggression in mice: development, social experience, and the effects of selective breeding. J. Comp. Psychol., 2: HOOD, K. E. & CAIRNS, R. B., 1988: A developmental-genetic analysis of aggressive behavior in mice: II. Cross-sex inheritance. Behav. Genet., 18: MAXSON, S. C., GINSBURG, B. E. & TRATTNER, A., 1979: Interaction of Y-chromosomal and autosomal gene(s) in the development of intermale aggression in mice. Behav. Genet., 9: MOHANAN, E. J. & MAXSON, S. C., 1998: Y chromosome, urinary chemosignals, and an agonistic behaviour (offense) in mice. Physiol. Behav., 64: MUNCLINGER, P. & FRYNTA, D., 2000: Social interactions within and between two distant populations of house mouse. Folia Zool., 49: 1-6. ROUBERTOUX, P. L., CARLIER, M., DEGRELLE, H., HAAS-DUPERTUIS, M.-C., PHILLIPS, J. & MOUTIER, R., 1994: Co-segregation of intermale aggression with the pseudoautosomal region of the Y chromosome in mice. Genetics, 135: SELMANOFF, M. K., JUMONVILLE, J. E., MAXSON, S. C. & GINSBURG, B. E., 1975: Evidence for a Y chromosomal contribution to an aggressive phenotype in inbred mice. Nature, 253: SELMANOFF, M. K., MAXSON, S. C. & GINSBURG, B. E., 1976: Chromosomal determinants of intermale aggressive behavior in inbred mice. Behav. Genet., 6: SLUYTER, F., MEIJERINGH, B. J., VAN OORTMERSSEN, G. A. & KOOLHAAS, J. M., 1995: Studies on wild house mice (VIII): Postnatal maternal influences on intermale aggression in reciprocal F 1 s. Behav. Genet., 25: SLUYTER, F., VAN DER VLUGT, J. J., VAN OORTMERSSEN, G. A., KOOLHAAS, J. M., VAN DER HOEVEN, F. & DE BOER, P., 1996a: Studies on wild house mice. Prenatal maternal environment and aggression. Behav. Genet., 26: SLUYTER, F., VAN OORTMERSSEN, G. A., DE RUITER, A. J. H. & KOOLHAAS, J. M., 1996b: Aggression in wild house mice: Current state of affairs. Behav. Genet., 26: SUCHOMELOVÁ, E. & FRYNTA, D., 2000: Intraspecific behavioural interactions in Apodemus microps: a peaceful mouse? Acta Theriol., 45: THUESEN, P., 1977: A comparison of the agonistic behaviour of Mus musculus musculus L. and Mus musculus domesticus Rutty (Mammalia, Rodentia). Vidensk. Meddr. dansk naturh. Foren., 140: VAN ZEGEREN, K. & VAN OORTMERSSEN, G.A., 1981: Frontier disputes between the West- and East- European house mouse in Schleswig-Holstein, West Germany. Z. Saugetierk., 46: VRBA, I. & DONÁT, P., 1993: Activities version 2.1. Computer programme for behavioural studies. 22

23 Folia Zool. 51(1): (2002) Food intake, feeding behaviour and stock losses of cormorants, Phalacrocorax carbo, and grey herons, Ardea cinerea, at a fish farm in Arcachon Bay (Southwest France) during breeding and non-breeding season Jesús M. LEKUONA c/ Virgen del Puy n 5, 7 D, E Pamplona, Navarra, Spain; jmlekuona@wanadoo.es Received 15 May 2000; Accepted 6 August 2001 Abstract. The feeding ecology of cormorants and grey herons were investigated at a fish farm in Arcachon Bay (southwest France) during both breeding and non-breeding season. Cormorants were mainly recorded during winter and grey herons during both breeding and wintering seasons. Adult cormorants and herons were the most abundant age clas at the fish farm. Adult cormorants and herons were more successful at feeding than first-years and although younger birds spent more time feeding and their biomass intake rates remained lower than those of adults birds. Cormorants and herons took the same biomass intake per feeding session at the fish farm during the non-breeding season, about 200 g. The impact of the two ichthyophagous birds (cormorant and grey heron) was estimated as 53.0% (average predation of cormorant per year) and 10.8% (mean predation of heron per year) of the annual yield of the fish farm. This imposed a significant economic loss due to low productivity of the farm. Key words: aquaculture, diet, foraging parameters, impact, piscivorous birds Introduction In French coastal and inland waters the cormorant (Phalacrocorax carbo) and the grey heron (Ardea cinerea) are commonly perceived as two great predator of fish stocks at fish farms (Hafner & Moser 1980, I m & Hafner 1984, Marion 1983, 1990, Marion & Marion 1983, 1987, Perennou 1987, T rolliet 1993a, 1993b). In other countries of Europe most studies with these two fish-eating birds have focused on extensive aquacultures (D raulans et al. 1985, C arss & Marquiss 1991, Janda 1993, C a r s s 1994, D irksen et al. 1995, K eller 1995, S uter 1995, Van Eerden & Gregersen 1995, Dekker 1997) rather than intensive systems (C a r s s 1990, C a r s s 1993a,1993b, L e k u o n a 1998) where fish stocks area reared at high densities. Their feeding activities had been often considered to result in important economic losses (Meyer 1981, M arion 1983,1990, U tschick 1983, D raulans & Van Vessem 1985, Carss 1991, Osieck 1991, Dirksen et al. 1995, Keller 1995, Warke & Day 1995, Pérez-Hurtado et al. 1997, Lekuona 1998). Recently, C a r s s (1993a) found that these losses are small compared with other forms of fish mortality. A wide variety of deterrent techniques (acoustic, visual and biological) has been tried in attempts to reduce cormorant and heron impact (Barlow & Bock 1984, Draulans 23

24 1987, M oerbeek et al. 1987, F eare 1988, O sieck 1991, T rolliet 1993a, 1993b, Kirby et al. 1996, V a n D a m & Asbirk 1997, Lekuona 1998). The present study gives some data on heron and cormorant predation at a fish farm in terms of piscivorous bird numbers during breeding and non-breeding seasons, food, feeding parameters and stock losses. The study had several aims: 1) to know heron and cormorant abundance and age structure of their population in Arcachon bay and at a fish farm, 2) to determine the food and some feeding parameters of two species of piscivorous birds and 3) to quantify their impact on fish stocks. Study Area, Material and Methods The study was carried out in Arcachon Bay (Fig. 1), southwest France (44 40 N, 1 10 W), a coastal lagoon with variations in water level due to tides, triangular in shape, approximately 20 km wide and km 2 in area (106.6 km 2 of mud and 54.7 km 2 of deep channels, Lekuona 1997). A colony of Ardeids breed there (Lekuona 1993, Lekuona & Campos 1995, Lekuona 1997). Fig. 1. Arcachon bay study area showing location of grey heron colony ( ), winter roosts of cormorants ( ) and the study fish farm (Certes). The farm was located in the oriental extreme of Arcachon bay. The fisf farm of Certes had an area of 142 ha, with two kind of fish ponds: deep ponds (1.5-2 m of depth and 3-4 m of width) and shallow ponds ( m of depth and m 2 of area, Lekuona 1993). Fish stocks (eel Anguilla anguilla, mullets Quelon labrosus and Liza ramada, gilthead Sparus aurata and bass Dicentrarcus labrax) were mainly reared at low densities in shallow ponds. The total yield per year of the fish farm was 9.1 Tm (35 kg.ha -1 for eel, for mullets 100 kg.ha -1, for gilthead 5 kg.ha -1 and for bass 30 kg.ha -1 ). 24

25 The observations on the both cormorant and grey heron s activities were carried out between February 1992 and December 1993, using a x telescope. One day a week, between h and h the bay and the fish farm were visited to obtain an index of abundance of two piscivorous birds by transect counts on a standard route of land. The cormorant s and heron s indeces of abundance were each the maximum of four counts made each month throughout the study period. During non-breeding season cormorant and heron roosts were censused fortnightly at dusk. Two age-classes were taken into account both for cormorants and herons (adult and first-year birds). Birds were categorized on their plumage characteristics (C ramp & Simmons 1977). Most of birds could not be recognized individually (only few cormorants had coloured rings) and some of observations on different study days could have been of the same individuals. However, the data were recorded over several months (February 1992 to December 1993) and the foraging sessions were assumed to be independent. Cormorants and herons were always observed at a distance <100 m. Each day the following data were noted for each bird in both natural feeding areas (Arcachon bay) and at fish farm (Certes): 1) Total time spent (Tt) in a feeding site, as the time elapsed from a bird s arrival to its departure (defined as a foraging session). 2) Foraging time (Ff), as the time used only for foraging within Tt by the two piscivorous bird species. 3) The number of feeding attempts during Ft and their results (successful and unsuccessful). 4) Dive time, only for cormorants, as the average of all dives made during each foraging bout in seconds. 5) Number, type and size of the captured prey. For herons four broad taxonomic groups of prey were used: amphibians, fish, insects, and crustaceans. The size of the prey was calculated only for fish (which were the most abundant prey, see below) in relation to the size of the both cormorant and grey heron s beak (8 cm and 12 cm, respectively, Cramp & Simmons 1977, Voisin 1991). In this way, for grey heron five size classes of fish were established: very small (<12 cm), small (12-18 cm), medium (19-25 cm), large (26-32 cm) and very large (>32 cm), and for cormorant five size classes were also established: very small (<8 cm), small (8-14 cm), medium (15-21 cm), large (22-28 cm) and very large (>28 cm). The biomass (g wet weight) of fish was calculated according to the formulae of R i c h n e r (1986) for eel and flounder Pleuronectes flesus. For mullets, Thick-lipped grey mullet and thin-lipped grey mullet the equation W = L (W = wet weight in g, L = total length in mm) was calculated according to the data of A rne (1938). The biomasses (g wet weight) of gilthead and bass taken were calculated according to the average weight of similar-sized specimens captured in the study area. Cormorant and heron predation (the estimated biomass of fish removed, in kg) was calculated on monthly and per species bases as: Σ Y ij = N j x C x P ij (Suter 1995) where N j is the number of cormorant days during the month j, C is the daily food consumption (g/day) and P ij is the proportion of the species i in the diet (% in biomass) recorded in the month j. Two averages were compared using the Student s t-test (parametric data) and Mann- Whitney Z-test (non-parametric data) and frequencies were compared using the χ 2 test, with Yates correction as necessary (S okal & Rohlf 1979, Fowler & Cohen 1992). The Bonferroni sequential correction test was made in multiple comparisons to avoid 25

26 making the Type I error (reject the null hypothesis when, in fact, it is true) and for keeping the overall error of the test set below 5%. Chandler (1995) showed that sometimes this kind of correction was extremely conservative. Thus, our statistical results have been explained basically without correction, when their biological meaning was very evident. Results Heron and cormorant counts and age ratio Grey heron numbers increased in spring (Fig. 2) due to start of the breeding season in the near ardeid colony located in a riparian wood near the fish farm (<10 km) and rose during the summer (post-fledgling period of juvenile grey herons). During the breeding season adult herons were the most important age class in the study area. During summer juvenile numbers were more abundant in Arcachon bay, falling during autumn when adults showed a high level. Cormorant roost numbers increased in autumn, remained relatively high during the winter, before falling during the spring to a low level in summer (Fig. 3). Adult birds were more abundant during autumn and winter than during spring and summer, when first-year birds had a high level. Seasonal changes were found in the abundance index of two species of fish-eating birds at the fish farm during the present study (Fig. 4): for herons due to management desiccation of extensive ponds (second week of March 1922 and first week of may 1993) and for cormorants due to a change in the feeding behaviour at the farm: very often cormorants foraged solitarily but in November 1993 social fishing (>300 birds) was observed in several shallow ponds, increasing the predation on fish stock (see below). Diet, feeding behaviour and fish stock losses The eel was the main fish species captured by both herons and cormorants in Arcachon bay and at the fish farm (Table 1). Flounders were also taken by two species of piscivorous birds but only in natural habitat. Gilthead and bass were taken by cormorants at low numbers at the fish farm. Other prey species taken by herons were not important (crustaceans, amphibians and insects). The main amphibian and invertebrate taxa identified as a heron food were: frogs Rana perezi, insects (mainly beetles, Coleoptera) and shrimp, Crangon spp. and shore crab, Carcinus maenas. During the breeding season, grey heron spent less total time at the fish farm (t=13.6, p<0.001), more foraging time (t=18.4, p<0.001) and had a higher biomass intake (t=3.2, 0<0.01) than in Arcachon bay (Table 2). During winter some feeding parameters showed the same pattern as the breeding period: grey heron spent less total time at fish farm (t=4.44, p<0.001) and took more biomass (t=1.94, p<0.05) with less feeding time (t=2.79, p<0.01). The Bonferroni sequential test gave the same statistical results. The agression rate observed at the fish farm was higher than in the bay during the breeding season (Z=9.2, p<0,01). In the area under study agressions were not found during winter season. During winter cormorant (Table 3) spent more time fishing in the bay than at the fish farm (t=7.2, p<0.001) and the dive time was also greater in bay than at fish farm (t=16.2, p<0.001). No differences were found between feeding habitats in the biomass intake of 26

27 F-92 A J A O D F A J A O D Fig. 2. Number of grey herons (upper) and age-structure population (lower) during the study period in Arcachon Bay (southwest France). Data are percentage numbers, black: adult birds, white: of first-year birds. Table 1. Food of cormorants and grey herons in natural feeding habitats and at the Certes fish farm in Arcachon Bay (southwest France) during the breeding (B) and non-breeding season (NB). Piscivorous bird Grey heron Cormorant Feeding habitat Bay Fish farm Bay Fish farm Prey B NB B NB NB NB Eel Mullet Flounder Gilthead Bass Unidentified fish Other prey species Total

28 F-92 A J A O D F A J A O D F-92 A J A O D F A J A O D Fig. 3. Number of cormorants (upper) and age-structure population (lower) during the study period in Arcachon Bay (southwest France). Data are percentage numbers, black: adult birds, white: first-year birds. wintering cormorants (t=0.35, NS). Cormorants and herons had the same biomass intake per feeding session at the fish farm during non-breeding period (t=0.47, NS). Both adult grey heron and cormorants (Table 4) were more successful at feeding than first-year birds and younger birds (χ 2 =41.7, 1df, p<0.001 and χ 2 =26.4, 1df, p<0.001, respectively), spent less time feeding (t=16.3, p<0.001 and t=11.7, p<0.001, respectively) and their biomass intake remained higher (t=9.6, p<0.001 and t=10.8, p<0.001, respectively) than those of juveniles at fish farm. Great differences were found in the length of fishes captured by both cormorants and herons in the bay and at the fish farm (χ 2 =32.1, 4df, p<0.001 and χ 2 =27.9, 4df, p<0.001, respectively, Fig. 5). At the fish farm cormorants caught very large fishes whereas grey heron captured large and very large fishes. In Arcachon bay grey heron took significantly more small and medium size fishes than at the fish farm. During breeding season fish losses by grey herons were 12.7% in 1992 and 8.9% in 1993 of the total fish farm yield (the average impact per year was 10.8%). During the study 28

29 F M A My J F-92 A J A O D F A J A O D Fig. 4. Census of grey herons at the fish farm during the breeding season in 1992 and 1993 (upper) and number of cormorants during the study period (lower). Table 2. Foraging studied parameters in feeding bouts made by grey herons in Arcachon bay and at Certes fish farm during breeding and non-breeding season. N: number of foraging bouts, Tt: total time spent in the habitat (min), %: percentage of success, Ft: foraging time (min), B: biomass intake (g wet weight) and A (aggression rate, Ag min -1 ). Tt, Ft, B and A as average±standard deviation. Habitat Bay Fish farm Bay Fish farm Season Breeding Non-breeding N Tt 136.4± ± ± ±23.0 % Ft 27.7± ± ± ±10.3 B 197.7± ± ± ±61.7 A 0.086± ± ± ±0.00 period cormorant losses were 38.4% in 1992 and 67.6% in 1993 (the average predation per year was 53.0%). The habit of mass fishing by cormorants observed in winter 1993 significantly increased the annual impact. 29

30 Table 3. Foraging studied parameters in feeding bouts made by cormorants in Arcachon bay and at fish farm during non-breeding season. N: number of foraging bouts, %: percentage of success, Ft: foraging time (min), D: dive time (s) and B: biomass intake (g wet weight). Ft, D and B as average±standard deviation. Feeding habitat Bay Fish farm N % Ft 27.9± ±2.6 D 24.3± ±3.3 B 195.8± ±58.9 Table 4. Foraging parameters of cormorants and grey herons at a fish farm in southwest France taking into account the age classes. N: number of foraging bouts, %: percentage of success, Tt: total time spent in the feeding habitat, Ft: feeding time (min), D: dive time (s), B: biomass taken (g wet weight). Tt, Ft, D and B are given as average±standard deviation. Species N % Tt Ft D B Grey Heron Adult ± ± ±46.2 First-year ± ± ±34.7 Cormorant Adult ± ± ± ±41.4 First-year ± ± ± ±19.9 Fig. 5. Length of prey (cm) captured by cormorants (upper) and grey herons (lower) at Certes fish farm (black) and in Arcachon bay (white). Data are percentages. 30

31 Discussion As a consequence of the rapid increase of both cormorant (V a n Eerden &Gregersen 1995) and heron populations (V o i s i n 1991) in Europe, interference with fisheries has risen (Marion & Marion 1983, 1987, Moerbeek et al. 1987, Marion 1990, Carss 1994, V a n D a m & Asbirk 1997, Callaghan et al. 1998, Leopold et al. 1998). Arcachon bay area was located within a region with high heron and cormorant density (Campos & Lekuona 1994, L e kuona 1999), where adult birds were the most abundant age class in coastal and inland waters (L e k u o n a 1997). In the present study cormorant numbers increased at the fish farm during winter and the heron number increased during breeding season due to some management techniques. Both adults and juveniles birds visited the fish farm during all the study period, but they showed significant agerelated differences in feeding performance (C arss 1993b). Young birds were able to compensate their lower success by attempting more feeding strikes or dives (Desgranges 1981, L ekuona 1997). In this study both age classes took fish of a similar size as did those studied by D raulans (1987) and C arss (1993b), but juvenile birds fed for more time at the fish farm than did adults. Draulans (1987) also found that first-year herons were less successful at foraging thatn adults at low prey densities; similar data were found in the present study. Cormorants and herons could not be recognized individually and some observations could have been made on the same bird, and this fact may affect the comparisons between two age classes. However, the data were recorded over 24 months with great differences always between age classes so that the foraging sessions were assumed to be independent and representative of the impact made by cormorants and herons at the fish farm. An adult (heron and cormorant) needs a biomass intake per day of g of fish (Cramp & Simmons 1977, M arion 1990, C arss 1993b, S uter 1995, Lekuona & Campos 1997, V a n D a m & A sbirk 1997). The biomass intake rates at the farm in relation to the feeding time of the two species were considerably higher than other published rates (C o o k 1978, R i c h n e r 1986), but similar data were found for grey heron at cage fish farms in Argyll (Carss 1993b). Thus, the daily requirements of food could be met at the fish farm making a second foraging session and many of the cormorants and herons probably caught all their prey there. At the fish farm, food requirements of the two ichthyophagous species were satisfied within a shorter foraging time than in Arcachon bay. Foraging activity of breeding grey herons and wintering cormorants in Arcachon bay was strongly influenced by variations in sea water level due to tides (L ekuona 1997, 1999), but no significant differences were found in their feeding activity at the fish farm during tides (L e k u o n a 1997) because the water level was constant during most of the study period. At the study fish farm the cormorant predation on fish stocks was higher than heron impact, but these rates represented a considerable economic loss. Similar date were found in other regions of Europe (I m & H afner 1984, P erennou 1987, M arion 1990, Osieck 1991, Lekuona 1998). Most studies on fish-eating birds have been focused in estimate direct loss of fish stocks (Marion 1990, O sieck 1991), but some other factors must be taken into account (parasites, physiological poor condition, blindness, wounded fish...) (C a r s s 1990, 1993b). Also in some regions of Europe available data at fish farms are anecdotal or based on 31

32 second-hand information (C a r s s 1993b), also this author found that not all of the fishes taken by grey heron were healthy. Parasited trouts moved slowly just below the surface and they formed a high proportion of the prey caught by herons. Thus, such predation did not represent a direct loss. It is obvious that very often it is extremely complicated to quantify the interaction between piscivorous birds and fisheries. Acknowledgements I would like to thank to the Asociación de Amigos de la Universidad de Navarra (1992) and to the Caja de Ahorros Municipal de Pamplona (1993) for their grants awarded to me for the completion of this study. My thanks go to all those who helped with the fieldwork, especially Alain Fleury, Claude Feigné and Veronique Hidalgo (Parc Ornithologique du Teich), Alberto Artazcoz and Itziar Gastón. Iam very grateful to the Certes s fish managers and their staff who kindly allowed me free access to the fish farm during the study period. Dr. P. Blahák, Professor R.H.K M a n n and an anonymous reviewer gave useful comments on a earlier draft of this paper. LITERATURE ARNE, P., 1938: Contribution à l étude de la biologie des muges du Golfe de Gascogne. P.V. Reun. Com. Explor. Scient. Mer Mediterranée, 11: BARLOW, C. G. & BOCK, K., 1994: Predatioin of fish in farm dams by cormorants, Phalacrocorax spp. Australian Wildl. Res., 11: CAMPOS, F. & LEKUONA, J. M., 1994: La población invernante de Cormorán Grande (Phalacrocorax carbo) en el Norte de España y Suroeste de Francia. Ardeola, 41: CALLAGHAN, D. A., KIRBY, J. S., BELL, M. C. & SPRAY, C. J., Cormorant Phalacrocorax carbo occupancy and impact at stillwater game fisheries in England and Wales. Bird Study, 45: CARSS, D. N., 1990: Beak-prints help in war againts aerial invaders. Fish Farmer, 13: CARSS, D. N., 1991: Avian predation at farmed and natural fisheries. Proceedings of the Institute of Fisheries Management. September 1991, University of Aberdeen. CARSS, D. N., 1993a: Cormorants Phalacrocorax carbo at cage fish farms in Argyll, western Scotland. Seabirds, 15: CARSS, D. N., 1993b: Grey Heron, Ardea cinerea L., predation at cage fish farms in Argyll, western Scotland. Aquaculture and Fisheries Management, 24: CARSS, D. N., 1994: Killing of piscivorous birds at Scottish fin fish farms, Biol. Conserv., 68: CARSS, D. N. & MARQUISS, M., 1991: Avian predation at farmed and natural fisheries. In: Lucas, M. C., Diak, I. & Laird, L. (eds.), Interaction between fisheries and environment. Proc. Institute of Fisheries Management, 22nd Annual Study Course, University of Aberdeen. CHANDLER, C. R., 1995: Practical considerations in the use of simultaneous inference for multiple test. Anim. Behav., 49: COOK, D. C., 1978: Foraging behaviour and food of grey herons on the Ythan Estuary. Bird Study, 54: CRAMP, S. & SIMMONS, K. E. L., 1977: The birds of the Western Palearctic. Vol. I. Oxford University Press, Oxford, U.K. DEKKER, W., 1997: The impact of cormorants and fykenet discards on the fish yield from Lake Ijsselmeer, The Netherlands. In: Van Dam, Ch. & Asbirk, S. (eds.), Proceedings of the Workshop an International Conservation and Management Plan for the Great Cormorant (Phalacarcrocorax carbo). October 1996, Lelystad, The Netherlands: DESGRANGES, J. L., 1981: Observations sur l alimentation du grand heron (Ardea herodias) en Quebec (Canada). Alauda, 49: DIRKSEN, S., BOUDEWIJN, T.J., NOORDHUIS, R. & MAERTEIJN, E. C. L., 1995: Cormorants Phalacrocorax carbo sinensis in shallow eutrophic freshwater lakes: prey choice and fish consumption in the non-breeding period and effects of large-scale fish removal. Ardea, 83: DRAULANS, D., 1987: The effectiveness of attempts to reduce predation by fish-eating birds: a review. Biological Conservation, 41:

33 DRAULANS, D. & VAN VESSEM, J., 1985: The effect of disturbance on nocturnal abundance and behaviour of Grey Herons (Ardea cinerea) at a fish farm in winter. Journal of Applied Ecology, 22: DRAULANS, D. & VAN VESSEM, J. & DE BONT, A. F., 1985: The effects of killing and removal on the abundance of Grey Heron at fish farms. XVIIth Congress of the International Union of Game Biologists, pp. FEARE, C. J., 1988: Cormorants as predators at freshwater fisheries. Inst. Fisheries Management Ann. Stud. Course, 18: FOWLER, J. & COHEN, L., 1992: Statistics for Ornithologists. BTO Guide 22. HAFNER, H. & MOSER, M., 1980: Les hérons et la pisciculture en Camargue. Colloque sur les zones humides et littorales, l aquaculture et la faune sauvage. France. Bull. de l office National de la Chasse, p IM, B. H. & HAFNER, H., 1984: Impact des oisesaux piscivores et plus particulièrement du grand cormoran Phalacrocorax carbo sinensis sur les exploitations piscicoles en Camargue, France. Le Sambuc, Station Biologique de la Tour du Valat. JANDA, J., 1993: La situation en Tchécoslovaquie. Bulletin Mensuel Office National de la Chasse, 178: KELLER, T., 1995: Food of Cormorants Phalacrocorax carbo sinensis wintering in Bavaria, southern Germany. Ardea, 83: KIRBY, J. S., HOLMES, J. S. & SELLERS, R. M., 1996: Cormorants Phalacrocorax carbo as fish predators: a appraisal of their conservation and management in Great Britain. Biol. Conserv., 75: LEKUONA, J. M., 1993: Uso de hábitat, alimentación y reproducción de la Garza Real (Ardea cinerea) en la Bahía de Arcachon (Francia). Tesis de Licenciatura, Universidad de Navarra. LEKUONA, J. M., 1997: Importancia de las aves ictiófagas: Cormorán Grande (Phalacrocorax carbo) y Garza Real (Ardea cinerea) en el norte de España y suroeste de Francia. Tesis Doctoral, Universidad de Navarra. LEKUONA, J. M., 1998: Impacto del Cormorán Grande Phalacrocorax carbo sinensis, la Gaviota Reidora Larus ridibundus y la Garza Real Ardea cinerea en un piscifactoría de Navarra durante la época invernal. Ardeola, 45: LEKUONA, J. M., 1999: Food and foraging activity of grey herons, Ardea cinerea, in a coastal area during the breeding season. Folia Zool., 48: LEKUONA, J. M. & CAMPOS, F., 1995: Le succès de reproduction du Héron cendré (Ardea cinerea) dans le Bassin d Arcachon. Alauda, 63: LEKUONA, J. M. & CAMPOS, F., 1997: Foraging ecology of cormorants (Phalacrocorax carbo) wintering in northern Spain. Folia Zool., 46: LEOPOLD, M. F., VAN DAMME, C. J. G. & VAN DER VEER, H. W., 1998: Diet of cormorants and the impact of cormorant predation on juvenile flatfish in the Ducht Wadden Sea. Journal of Sea Research, 40: MARION, L., 1983: Etat de nos connaissances sur les relations oiseaux piscivores et populations piscicole naturelles et artificielles et propositions de recherches prioritaires pour la période Rapport du Groupe de Travail sur les oiseaux piscivores, Ministère de l Environnement, Paris. MARION, L., 1990: Les oiseaux piscivores et les activités piscicoles: impact et protection. Secrétariat d Etat chargé de l environnement & Ministère de l agriculture et de la forêt, Paris. MARION, P. & MARION, L., 1983: Etude de la predation par le Héron cendré sur les piscicultures intensives et sur le milieu naturel. Rapport Laboratoire d Evolution des Systèmes Naturels et Modifiés. Ministère de l Environnement, Paris. MARION, P. & MARION, L., 1987: Impact de la predation exercée par le Héron cendré sur une pisciculture. Revue d Ecologie (Terre et Vie), 4: MEYER, J., 1981: Room for birds and fish. Fish Farmer, 4: MOERBEEK, D. J., VAN DOBBEN, W. H., OSIECK, E. R., BOERE, G. C. & BUNGENBERG DE JONG, C. M., 1987: Cormorant damage prevention at a fish farm in the Netherlands. Biol. Conserv., 39: OSIECK, E. R., 1991: Prevention of Cormorant damage at the Lelystad fish farm. In: Van Eerden, M. R. & Ziljstra, M. (eds.), Proceedings Workshop 1989 on Cormorants (Phalacrocorax carbo). Rijkswaterstaat Directorate Flevoland, Lelystad: PERENNOU, C., 1987: L impact de Héron cendré Ardea cinerea en bassin d alevinage. L Oiseau et R. F. O., 57: PÉREZ-HURTADO, A., MASERO, J. A., ARROYO, G., PIZARRRO, S R., HORTAS, F. & CASTRO, M., 1997: Estudio de la incidencia de la Avifauna Ictiófaga en los cultivos marinos y otras especies orníticas en el Parque Natural de la Bahía de Cádiz. Consejería de Medio Ambiente de la Junta de Andalucía, Cádiz. RICHNER, H., 1986: Winter feeding strategies of individually marked herons. Anim. Behav., 34: SOKAL, R. R. & ROHLF, F. J., 1979: Biometría. Blume, Madrid, Spain. 33

34 SUTER, W., 1995: The effect of predation by wintering cormorants Phalacrocorax carbo on grayling Thymallus thymallus and trout (Salmonidae) populations: two case studies from Swiss rivers. J. of Applied Ecology, 32: TROLLIET, B., 1993a: Moyens préventifs de limitation dl impact du grand cormoran sur la pisciculture extensive. Bulletin mensuel Office National de la Chasse, 178: TROLLIET, B., 1993b: Un nouveau moyen d effarouchement: le fusil laser. Bulletin mensuel Office National de la Chasse, 178: UTSCHICK, H., 1983: Abwehrstrategie und Abwehrmassnahmen gegen den Graureiher (Ardea cinerea) und Fischgewässern. Garm. Vogelk. Ber., 12: VAN EERDEN, M. R. & GREGERSEN, J., 1995: Long-term changes in the northwest European population of Cormorants Phalacrocorax carbo sinensis. Ardea, 83: VAN DAM, C. & ASBIRK, S., 1997: Cormorants and human interests. Proceedings of the Workshop towards an International Conservation and Management Plan for the Great Cormorant (Phalacrocorax carbo), 3 4 October 1996, Lelystad, The Netherlands. VOISIN, C., 1991: The Herons of Europe. T & DA Poyser, London. WARKE, G.M.A. & DAY, K. R., 1995: Changes in abundance of cyprinid and percid prey affect rate of predation by Cormorants Phalacrocorax carbo carbo on Salmon Salmo salar smolt in Nothern Ireland. Ardea, 83:

35 Folia Zool. 51(1): (2002) Importance of colony size and breeding synchrony on behaviour, reproductive success and paternity in house sparrows Passer domesticus Radovan VÁCLAV 1* and Herbert HOI 2 1 Institute of Zoology, Slovak Academy of Sciences, Dúbravská cesta 9, SK Bratislava, Slovakia; uzaevacl@savba.sk 2 Konrad Lorenz Institut für Vergleichende Verhaltensforschung, Österreichische Akademie der Wissenschaften, Savoyenstraße 1a, A-1160 Vienna, Austria; h.hoi@klivv.oeaw.ac.at Received 20 March 2001; Accepted 12 November 2001 A b s t r a c t. Several comparative studies have previously identified breeding density and synchrony as potential determinants of reproductive success and extra-pair mating. However, the mechanisms and interaction of these two factors are poorly known. Here, we examined the effects of breeding density and synchrony on the behaviour, reproductive success and paternity losses in house sparrows. In order to test the effects of colony size, we created nest sites with varying numbers of nest-boxes. Our results show that there is an interaction between breeding synchrony and density, namely that breeding synchrony decreased with colony size. Neither colony size nor breeding synchrony seemed to influence brood size at fledging, although birds in larger colonies laid larger clutches. Moreover mate guarding behaviour was not influenced substantially by these two factors. Only nest guarding was significantly related to colony size and breeding synchrony. Paternity losses were not significantly related to colony size but they appeared to decrease with increasing synchrony. This finding supports the idea that extra-pair fertilisations are under male rather than female control. Key words: breeding density, synchrony, house sparrows, reproductive success, paternity Introduction Ecological factors like food availability, weather or predator pressure are well known to influence life history and reproductive success of a species (L a c k 1968). In addition, they were recognised to effect parental effort (H o i et al. 1995), mating system (Emlen & Oring 1977, Hoi et al. 1995), sexual strategies (see Birkhead & M ø ller 1992) and paternity (Hoi & Hoi-Leitner 1997). In respect of the latter two, colony size and breeding synchrony have been recently suggested as important socio-ecological determinants (Birkhead & Biggins 1987, Birkhead & Mø ller 1992). Both factors may determine intraspecific competition for resources such as food, nest sites or mates (W ittenberger & Hunt 1985, Brown 1987). In this way, breeding density and synchrony may directly influence breeding success, but may also effect sexual strategies like extra-pair behaviour or paternity guards and consequently the realised reproductive success. M ø ller & Birkhead (1993) reported that cuckoldry is higher in colonial than solitary breeding bird species. They suggested that this might be due to insufficient mate guarding or increased possibility to encounter sexual partners. W agner (1993) hypothesised that females may even promote breeding in higher densities in order to facilitate extra-pair matings. Nevertheless, only some studies support this prediction (e.g. M ø l l e r * Corresponding author 35

36 1985, Birkhead et al. 1987, H ill et al. 1994, M ø ller & Ninni 1998). A comparative study by W estneat & Sherman (1997) did not provide general support for this idea between species (but see M ø ller & Ninni 1998) although the authors showed that cuckoldry increases with density within species. W estneat & Sherman (1997) further suggested that density might perhaps be linked with additional factors such as habitat structure, female behaviour affecting encounter rate with extra-pair males or breeding synchrony. Indeed, breeding synchrony was found to favour extra-pair mating in some species (Stutchbury et al. 1994). However, the role of synchrony for extra-pair behaviour is still controversial (see W eatherhead 1997). Namely, it is not clear whether extra-pair paternity (EPP) should increase (Stutchbury & Morton 1995) or decrease with synchrony of breeding (Birkhead& Mø ller 1992). A recent comparative study by M ø ller & Ninni (1998), however, suggests that among species paternity decreases with breeding asynchrony. Most of the non-comparative studies on extra-pair paternity tend to examine only one factor at the time, whether it is breeding density or synchrony (T husius et al. 2001). In this study, therefore, we try to investigate simultaneously the importance of both colony size and breeding synchrony on reproductive success, paternity assurance, nest guarding behaviours and paternity losses in house sparrows Passer domesticus. The house sparrow is an ideal species for such a study because it nests solitarily as well as in colonies of variable sizes (S ummer-smith 1954, Cramp 1994, T ost 1994). Moreover, sperm competition is intense in this species as indicated by high within-pair copulation frequency (M ø ller 1987) and occurrence of EPP (e.g. W etton & Parkin 1991, Cordero et al. 1999, V á cl a v et al., in press). In order to study the effect of colony size and breeding synchrony, we offered house sparrows breeding situations in the plots with varying number of nest-boxes and experimentally altered start of egg-laying. Material and Methods Our study was conducted in the Schönbrunn Zoo in Vienna, Austria during the breeding seasons 1999 and In these two years, house sparrows nested in 80 and 69 nest-boxes, respectively. Five to ten nest-boxes were hung on buildings in 1999 and three nest-boxes in In this way, we created four plots with five nest-boxes and ten plots with ten nest-boxes in 1999 and 15 plots with three nest-boxes in All plots were occupied in 1999 but eight plots were unsettled or settled only during one breeding attempt in The nest-boxes were placed m apart in all plots. Birds were usually trapped after completing their clutches. We weighed all birds and measured the length (mm) of tarsus, wing and tail. We found no difference in the size of the male ornament (mean badge size: mm 2 in 1999 and mm 2 in 2000; t-test: t = 0.39, p = 0.70, n = 16, 14). Since badge size is considered to be a condition-dependent trait (Griffith 2000), we think that the phenotype of birds did not markedly influence breeding patterns between years. In both breeding seasons, we started to observe activity around nest-boxes several months before the first eggs were laid (25 March in both years). Clutch and brood size were regularly monitored in all nest-boxes at least every second day during the whole breeding season. Brood size at fledging refers to an estimate of fledging numbers when chicks were days old. Accordingly, fledging success was calculated as the number of chicks in the 36

37 nest reaching the age of days divided by the number of laid eggs. We analysed the behaviour of 28, 29 and 25 pairs during three consecutive breeding attempts during the egg laying in Every nest site was observed daily during 15 min when we recorded the presence of birds inside and around nest-boxes, the time mates stayed together and withinpair copulation frequency. In 2000, we manipulated breeding synchrony experimentally. To achieve asynchrony, we swapped up to the half of nest contents from the nest of a focus pair to the nest of a neighbouring pair that was at a similar stage of breeding. On the other hand, we attained synchrony by equalising nest contents between pairs at apparently different breeding stages. The focus pair was, in this case, one that nested formerly at a later breeding stage. This was not the most suitable technique in the second and mainly third breeding attempt when the volume of nest material was more or less stable and was hence an unreliable estimate of breeding stage. For this reason, we used behavioural observations to estimate breeding stage and manipulated nest contents in the same logic as described above. Moreover, in the third breeding attempt, we delayed more advanced breeders by reducing their clutch size by taking up to two eggs and hence extending their laying period by laying additional eggs. In this way, we could increase synchrony between pairs with reduced clutches and focus pairs by more than 40% (mean clutch size in our study was 4.6 eggs in 1999 and 4.3 in 2000). Consequently, synchrony refers in this part of the results to the cases when we achieved either full or at least 80% breeding synchrony, whereas asynchrony means that the focus and neighbouring pairs did not overlap their laying periods. Synchrony from the correlational part of results refers to the proportion of the laying period that overlapped between pairs within a nesting site. For analysing paternity losses, we used multi-locus DNA fingerprinting. We examined genetic parentage of 34 families. Laboratory procedures followed those previously described by Krokene et al. (1996). We used band-sharing and number of novel bands (i.e. bands present in chicks but not in the putative parents) when excluding parents (e.g. Krokene et al. 1996). The method and assumptions for paternity exclusion are presented in more detail in W etton et al. (1987) and W estneat (1990). Paternity losses refer for each pair to the proportion of extra-pair young in the brood. Parametric tests were used in our analyses only when the requirements for these tests were met. Normality was tested with the Liliefors and Shapiro-Wilk tests. In order to achieve normality, data on paternity losses were log transformed. In parametric tests, post hoc comparisons were calculated with the Newman-Keuls test. If not stated otherwise, presented values are means and standard errors. The majority of studied pairs (94 % in 1999 and 93 % in 2000, respectively) contributed more than one clutch within breeding season. Hence, in order to avoid pseudo-replication, we examined seasonal and annual variation of reproductive success and behaviour with repeated ANOVAs. In addition, since the number of pairs at nest sites fluctuated between breeding cycles, variation of different behaviours was analysed using mean values for each nest site. Synchrony decreased with increasing size of colony (multiple regression: r 2 = 0.22, F 2,27 = 3.83, p = 0.034, effect of colony size ß = -0.47, p = 0.011, effect of set-up ß = -0.02, p = 0.90). To control for this interrelationship and the effect of nest-box set-up, we included in multiple regression analyses all of the three factors. Hence, colony size, breeding synchrony and the type of setup (5 and 10 nest-box sites) were entered as independent variables and reproductive success, behaviour and paternity losses as dependent variables. 37

38 Results Seasonal and annual variation in reproductive success and behaviour during egg laying Although the first eggs were recorded on the same day, 25th March, pairs nesting in 2000 tended to initiate their clutches later than pairs in 1999 and this difference increased and was significant in the two subsequent breeding attempts (Fig. 1). Fig. 1. Start of egg laying (means ± s.e.) for the three breeding attempts in 1999 and Median-test of the annual difference between first breeding attempts: χ 2 = 3.43, p = 0.068, n = 78; second breeding attempts: χ 2 = 12.36, p < 0.01, n = 81; and third breeding attempts: χ 2 = 20.13, p < 0.01, n = 73. There was seasonal variation in reproductive success (Fig. 2). Clutch size, hatching size and brood size at fledging varied significantly among the three breeding attempts (Fig. 2a, b and c). Reproductive success in terms of brood size at fledging peaked in the second breeding attempt in 1999, whereas in 2000 it appeared to decline with the season. Importantly, when taking clutch size into account, efficiency of fledging chicks did not change significantly between the three breeding cycles (Fig. 2d). Examining annual variation in reproductive success, we found that larger clutches were laid in 1999 but hatching size and brood size at fledging were similar between years (Fig. 2a, b and c). In addition, pairs achieved similar fledging success in both years (Fig. 2d). Interactions between the effect of year and the time of season for clutch and brood size at fledging, and mainly fledging success (Fig. 2a, c and d) suggest that pairs in 1999 allocated most reproductive effort to the second breeding attempt rather than the first as in Seasonal variation of colony size, synchrony and behaviour Depending on the number of available nest-boxes in the plots (plots with 3, 5 and 10 nestboxes), 87%, 77% and 46% of the nest-boxes were occupied (Fig. 3). While the number of breeding pairs did not differ between the plots with 5 and 10 nest-boxes, fewer pairs nested 38

39 Fig. 2. Seasonal and annual variation in four measures (means ± s.e.) of reproductive success. (a) Clutch size (2-way repeated measures ANOVA, seasonal effect: F 2,96 = 13.54, p < 0.001, year effect: F 1,48 = 15.29, p < 0.001, interaction: F 2,96 = 5.90, p < 0.01), (b) Hatching size (seasonal effect: F 2,96 = 5.51, p < 0.01, year effect: F 1,48 = 1.99, p = 0.17, interaction: F 2,96 = 2.67, p = 0.07), (c) Brood size at fledging (seasonal effect: F 2,96 = 5.49, p < 0.01, year effect: F 1,48 = 0.04, p = 0.84, interaction: F 2,96 = 6.24, p < 0.01), (d) Fledging success (seasonal effect: F 2,96 = 0.85, p = 0.43, year effect: F 1,48 = 0.35, p = 0.55, interaction: F 2,96 = 4.50, p = 0.013). in the plots with 3 nest-boxes (the mean number (± se) of pairs in plots: with 10 nest-boxes: 4.56 ± 0.42; 5 nest-boxes: 3.83 ± 0.34; 3 nest-boxes: 2.5 ± 0.1; F 2,69 = 23.25, p < 0.01; post hoc 10-3** and 5-3**; ** p < 0.01). In contrast to the set-ups with 5 and 10 nest-boxes, there was a seasonal decline in the colony occupancy with breeding season for the set-up with 3 nest-boxes (Fig. 3). On average, 25% of the laying period (days) overlapped between pairs within a nesting site. We found no change in breeding synchrony between the three successive breeding attempts (first attempt: 26.45% ± 5.31; second attempt: 21.09% ± 3.27; third attempt: 26.43% ± 7.51, Kruskal-Wallis: H 2,30 = 0.51, p = 0.75). We did not detect significant differences between three breeding attempts in the time that mates stayed at the nest together during the time of egg laying, although mate guarding 39

40 Fig. 3. Seasonal variation of nest site occupancy (means ± s.e.) in relation to the number of installed nest-boxes. (10 nest-boxes: Kruskal-Wallis H 2,18 = 1.83, p = 0.40; 5 nest-boxes: H 2,12 = 1.18, p = 0.55; 3 nest-boxes: H 2,42 = 6.02, p = 0.049). appeared to decline towards the end of season (Fig. 4a). Copulation frequency, however, significantly declined with the breeding season (Fig. 4b). In contrast to males, the time females stayed during egg laying at the nest alone decreased with the time of season (Fig. 4c). Moreover, males and females seemed to reduce the time they stayed inside the nestboxes during egg laying after the first breeding attempt (Fig. 4d). Paternity losses (proportion of extra-pair chicks/brood) did not significantly differ between breeding attempts (first attempt: 6.7%; second attempt: 28.8%; third attempt: 24%; Kruskal-Wallis H 2,25 = 3.78, p = 0.15). The role of colony size for reproductive success, behaviour during egg laying and paternity losses After controlling for the effect of season, multiple regression analyses revealed a relationship only between clutch size and the number of breeding pairs. Specifically, clutch size increased with colony size (Fig. 5) (including the 5 and 10 nest-boxes set-up only). In addition, the partial regression coefficient suggests that clutch size was larger in five than in ten nest-box plots (see Table 1). We found a relationship between nest guarding behaviour and colony size (Table 2). Nevertheless, birds of both sexes breeding in the plots with five nest-boxes stayed in their nests during egg laying significantly longer than birds in the plots with ten nest-boxes (see Fig. 6, Table 2). The time males and females stayed at the nest alone was not related to colony size or the type of set-up (see Table 2). Similarly, paternity assurance behaviours do not seem to be largely related to colony size (Table 2). There is, however, indication of a link between copulation frequency and colony size (see Table 2). Finally, the relationship between colony size and paternity losses seemed to be positive but not significantly (r = 0.18, p > 0.4, n = 24). 40

41 Fig. 4. Seasonal variation of mate guarding behaviour (means ± s.e.) in terms of (a) time mates stayed at nests together (1-way repeated measures ANOVA: F 2,14 = 2.34, p = 0.13) and, (b) copulation frequency (F 2,14 = 5.02, p = 0.02; 1-3* and 2-3*), and nest guarding behaviour (means ± s.e.) in terms of (c) time males stayed at nests alone (F 2,14 = 0.90, p = 0.43; time females alone (F 2,14 = 4.09, p = 0.04; 1-3*), (d) time males inside the nest (F 2,14 = 6.32, p = 0.01; 1-2* and 1-3*) and time females inside the nest (F 2,14 = 2.81, p = 0.09). After tests are showed post hoc comparisons (* p < 0.05). The role of breeding synchrony for reproductive success, behaviour during laying and paternity losses After controlling for the effect of season, a multiple regression analysis revealed a relationship only between hatching success and breeding synchrony (Fig. 6a). Specifically, the number of hatched chicks increased when egg laying was more synchronous at a colony. However, the influence of breeding synchrony on brood size at fledging (Fig. 6b) and fledging success was not significant (see Table 1). Experimental manipulation of the laying synchrony in 2000 supported our result from the natural condition about the existence of a relationship between hatching size and laying synchrony. Although we found no significant difference in clutch size between synchronous 41

42 Fig. 5. The relationship between clutch size and the number of pairs at nest site. Values of clutch size were standardised for the time of season. Lines are linear regressions for each types of nest site. Table 1. Partial regression coefficients (p) from stepwise multiple regression models between socio-ecological factors and parameters of reproductive success. Whole model values are r 2, (n) and corresponding p. Variables of reproductive success were controlled for the time of season (we used residual values). Socio-ecological variables Parameters of reproductive breeding type of set-up whole model success colony size synchrony clutch size 0.50 (0.013) 0.20 (0.28) 0.45 (0.01) 0.32 (30)* hatching size 0.38 (0.07) 0.44 (0.03) 0.27 (0.13) 0.24 (30) (0.07) hatching failure 0.36 (0.05) (30)* brood size at fledging 0.22 (0.25) (30) (n.s.) fledging success 0.21 (0.27) (30) (n.s.) * p < 0.05 Table 2. Partial regression coefficients (p) from stepwise multiple regression models between socio-ecological factors and behaviours related to mate and nest guarding. Whole model values are r 2, (n) and corresponding p. All variables referring to behaviour were controlled for the time of season (we used residual values). Socio-ecological variables Mate and nest guarding breeding type of set-up whole model behaviours colony size synchrony copulation frequency 0.21 (0.32) (26) (n.s.) time mates together a 0.11 (0.63) 0.07 (0.77) 0.04 (0.87) 0.01 (26) (n.s.) time males in the nest 0.28 (0.12) 0.45 (0.02) 0.46 (0.01) 0.41 (26)** time females in the nest 0.33 (0.08) 0.38 (0.04) 0.30 (26)* time males at the nest alone 0.35 (0.10) 0.27 (0.20) 0.30 (0.14) 0.20 (26) (n.s.) time females at the nest alone 0.29 (0.13) 0.33 (0.09) 0.23 (26)* a none of the factors was significant and therefore we performed standard multiple regression p < 0.05, ** p <

43 Fig. 6. Hatching size (a) and brood size at fledging (b) in relation to breeding synchrony. Values of hatching size and brood size at fledging were standardised for the time of season. Lines are linear regressions for each types of nest site. and asynchronous pairs (t = 1.09, p = 0.28, n = 28, 20), pairs nesting synchronously hatched significantly more chicks (t = 2.37, p = 0.022, n = 28, 20). Moreover, differences in brood size at fledging and fledging success between synchronously and asynchronously laying pairs were marginally significant (brood size at fledging: t = 1.97, p = 0.054, n = 28, 20; fledging success: t = 1.86, p = 0.069, n = 28, 20). With respect to behaviour, synchrony seems to have a large influence only on nest guarding behaviour. Both males and females spent a longer time in the nest with decreasing synchrony (Fig. 7a and b, Table 2). We did not find any stronger relationship between synchrony and paternity assurance behaviours (see Table 2). Similarly, there was no significant relationship between breeding synchrony and paternity losses (Fig. 8). However, 43

44 Fig. 7. Nest guarding behaviours in relation to breeding synchrony. (a) Time males stayed in the nest. (b) Time females stayed in the nest. Duration of nest guarding was standardised for the time of season. the shape of the relationship between synchrony and paternity losses indicates that males suffered a loss of their paternity less often with increasing synchrony. Discussion Our results suggest that colony size does not have a strong effect on reproductive success, behaviour and paternity losses. Clutch size seemed to increase with colony size but finally there was no difference in the reproductive success in relation to colony size. Surprisingly, we did not find a substantial effect of colony size on investment in paternity guards or nest guarding behaviours and consequently only a weak effect on paternity losses. This is in contrast to some other studies, which found a density dependent effect on paternity losses (Gowaty & Bridges 1991, Hill et al. 1994, H o i & H o i - L e i t n e r 1997) 44

45 Fig. 8. The relationship between paternity losses and breeding synchrony (r = -0.21, p = 0.31, n = 24). Paternity losses were log transformed to achieve normality. as well as on investment in paternity assurance tactics (B irkhead et al. 1987, Hatchwell 1988, for house sparrows see T o s t 1994). Hence, although sperm competition occurred in our study population, it did not seem to be directly linked to colony size. An explanation could be that extra-pair copulations occurred at the foraging sites and colony size would hence not reflect the level of sperm competition. This is, however, unlikely since W etton et al. (1995) showed specifically for house sparrows and Birkhead & M ø ller (1992) for other colonially breeding species that extra-pair males are often neighbours. There is still a possibility that density may be important for extra-pair fertilisations since we manipulated only colony size but not density and the distance between nest-boxes remained similar in all colonies. In contrast to T o s t (1994), who found that house sparrows adjust their paternity guards in relation to colony size, we may not have found such a relationship for another reason. T ost s (1994) population nested under natural conditions, whereas our study provides an example of an unnatural nest-box population. There are several important differences between natural and nest-box populations that need to be stressed. Firstly, in nest-box populations, quality of nesting sites does not necessarily reflect territory quality, such as food abundance. Secondly, if females seek extra-pair fertilisations (EPF) according to the male territorial quality, equal quality of nest-boxes disables females to judge the quality of males because male dominance is no longer reflected in nest site quality. Thirdly, the cost of searching extra-pair males may be too high in uniform and high quality nesting habitats (Slagsvold & Dale 1991, W estneat 1992). Our results are, in respect of the latter two points, similar to the study by V eiga & Boto (2000), who did not find in their nest-box colony higher levels of EPP than those reported from lower density nest-box colonies of house sparrows. Nonetheless, other factors like breeding synchrony (see later, Stutchbury & Morton 1995), territory quality (H o i et al. 1995) or intense intraspecific competition (V eiga 1992) might have outweighed the role of breeding density in our study. Also, it is important to mention that the phenotype of birds was not considered in our study due to a small sample size and hence we might lose an important component of variation in our analyses. 45

46 The missing relationship between colony size, behaviour, reproductive success and achieved paternity may also be due to a nonlinear relationship between them. T o s t (1994) found a considerable variation of breeding sociality, ranging from rare cases of solitarily breeding pairs up to colonies of 200 breeding pairs. Yet, even within these huge colonies, he could usually distinguish clusters with two to seven pairs. This suggests that birds avoided to nest solitarily as well as in large groups and the observed aggregations reflected the preferred colony size. In line with this, we found that, regardless of the nest-box availability, there was no significant difference in the colony size between plots with 5 and 10 nestboxes. Pairs nested most frequently in groups of four to five pairs in both plots with 5 or 10 nest-boxes, whereas in the plots with 3 nest-boxes, naturally, in smaller groups of two to three pairs. Although our results about the role of nesting site type need to be treated with caution because plot size was confounded to the year of study, it seems that breeding in groups of four to five pairs at the nest site may have been beneficial. We have two indications for this assumption. First, pairs in the plots with 3 nest-boxes started breeding at approximately same time as the pairs in the plots with 5 and 10 nest-boxes, but it took them longer to fledge their chicks (Fig.1). This may be disadvantageous since it is known that offspring produced late in the season have a shorter time to build up their energetic reserves for the winter, reach a lower hierarchical status and suffer higher mortality (L a c k 1954, Klomp 1970, see D aan & Tinbergen 1997). Second, despite the positive relationship between clutch size and colony size, pairs in 5 nest-box plots had significantly larger clutches than pairs in 10 nest-box plots (Table 1). Therefore, it seems that availability of nest-boxes could serve as a habitat cue reflecting future colony size. Consequently, birds may try to avoid nesting sites that deviate from the preferred colony size. In contrast to colony size, we found that breeding synchrony influenced hatching success. Namely, more chicks hatched under both natural and experimental conditions in situations when pairs laid eggs at the nest site more synchronously. Again, breeding synchrony did not seem to largely influence paternity assurance behaviour but paternity losses appeared to decrease with higher breeding synchrony (Fig. 8). Hence, this finding supports Birkhead & Biggins (1987) hypothesis that EPP decreases with breeding synchrony due to the incompatibility of males to guard their mates and seek extra-pair copulations simultaneously. This is in contrast to the hypothesis of S tutchbury & Morton (1995), which implies an increase in extra-pair paternity with breeding synchrony because females should have better possibilities to compare the quality of possible copulation partners that all breed at the same time. In contrast to paternity assurance behaviour, we found that breeding synchrony significantly influenced investment in nest guarding behaviour. Synchrony is usually thought to be advantageous because it decreases the risk of nest depredation (e.g. W ittenberger & Hunt 1985, I m m s 1990) but may also reflect the time when the environmental conditions for breeding are at best (Brown & Brown 1996). However, in our population house sparrows had three breeding attempts with fluctuating reproductive success, but without obvious fluctuations of breeding synchrony. Synchrony may, however, mediate different levels of intraspecific competition. Indeed we found that nest guarding behaviour apparently increased with decreasing synchrony. V e i g a (1990, 1992) showed that competition between females over the superior males, eventually leading to egg destruction, decreases with breeding synchrony. Since we have shown that egg losses are in general very high in house sparrows and substantially influence reproductive success also in our population, nest guarding seems to be very important in this species. Moreover, our previous works (V á c l a v & H o i in 46

47 press, V á cl a v et al. in press) show that egg losses were highest in the nests of males that were cuckolded most often, possibly reflecting male harassment upon females. Hence, since males are cuckolded less often and females suffer lower clutch losses when breeding synchronously, intra-sexual competition in both sexes may favour breeding synchrony and consequently genetic monogamy. The reason why house sparrows do not prefer to breed in large colonies and prefer to stay in loose groups might therefore be rooted in avoidance of breeding asynchronously. To summarise, our study showed that colony size does not seem to be a factor substantially explaining reproductive success, extra-pair behaviour and paternity losses in house sparrows. Although clutch size increased with colony size, birds preferred nesting in rather small groups. Higher hatching success and possibly also higher fledging success and lower risk of paternity losses increased when birds nested more synchronously. Since synchrony decreased with colony size in our study population, this might be a reason why house sparrows of both sexes prefer to breed in small groups. A c k n o w l e d g e m e n t s We thank the staff of Vienna Zoo in Schönbrunn for their permission to carry out this study and their field assistance during two years. We thank Donald B l o m q v i s t and an anonymous referee for their comments on this manuscript. R.V. was funded from the exchange programme between Austrian Academy of Sciences and Slovak Academy of Sciences and by grants from the Jubiläumsfonds der Stadt Wien STI LITE R ATUR E BIRKHEAD, T. R., ATKIN, L. & MØLLER, A. P., 1987: Copulation behaviour of birds. Behaviour, 101: BIRKHEAD, T. R. & BIGGINS, J. D., 1987: Reproductive synchrony and extra-pair copulations in birds. Ethology, 74: BIRKHEAD, T. R. & MØLLER, A. P., 1992: Sperm competition in birds: evolutionary causes and consequences. Academic Press, London. BROWN, J. L., 1987: Helping and Communal Breeding in Birds: Ecology and Evolution. Princeton University Press, Princeton. BROWN, C. R. & BROWN, M. B., 1996: Coloniality in the cliff swallow: the effect of group size on social behaviour. University of Chicago Press, Chicago. CORDERO P. J., WETTON, J. H. & PARKIN, D. T., 1999: Extra-pair paternity and male badge size in the House Sparrow. J. Avian Biol., 30: CRAMP, S., 1994: The birds of the Western Palearctic, Vol. IX.: Buntings and New World Warblers. Oxford University Press, Oxford. DAAN, S. & TINBERGEN, J. M., 1997: Adaptations of Life Histories. In: Krebs, J. R. & Davies, N. B. (eds.), Behavioural ecology: an evolutionary approach. Blackwell Science, Oxford: EMLEN, S. T. & ORING, L. W., 1977: Ecology, sexual selection and the evolution of mating systems. Science, 197: GOWATY, P. A. & BRIDGES, W. C., 1991: Nestbox availability affects extra-pair fertilisations and conspecific nest parasitism in eastern bluebirds, Sialia sialis. Anim. Behav., 41: GRIFFITH, S. C., 2000: A trade-off between reproduction and a condition-dependent sexually selected ornament in the house sparrow Passer domesticus. Proc. R. Soc. Lond., B 267: HATCHWELL, B. J., 1988: Intraspecific variation in extra-pair copulation and mate defence in common guillemots Uria aalge. Behaviour, 107: HILL, G. E., MONTGOMERIE, R., ROEDER, C. & BOAG, P. T., 1994: Sexual selection and cuckoldry in a monogamous songbird: implications for sexual selection theory. Behav. Ecol. Sociobiol., 35: HOI, H. & HOI-LEITNER, M., 1997: An alternative route to coloniality in the bearded tit: females pursue extrapair fertilizations. Behav. Ecol., 8:

48 HOI, H., KLEINDORFER, S., ILLE, R. & DITTAMI, J., 1995: Prey abundance and male parental behaviour in Acrocephalus warblers. Ibis, 137: IMMS, A. D., 1990: On the adaptive value of reproductive synchrony as a predator-swamping strategy. Am. Nat., 136: KLOMP, H., 1970: Regulations of the size of bird populations by means of territorial behaviour. Netherlands Journal of Zoology, 22: KROKENE, C., ANTHONISEN, K., LIFJELD, J. T. & AMUNDSEN, T., 1996: Paternity and paternity assurance behaviour in the bluethroat, Luscinia s. svecica. Anim. Behav., 52: LACK, D., 1954: The Natural Regulations of Animal Numbers. Clarendon Press, Oxford. LACK, D., 1968: Ecological Adaptations for Breeding in Birds. Chapman & Hall, London. MØLLER, A. P., 1985: Mixed reproductive strategy and mate guarding in semi-colonial passerine, the swallow Hirundo rustica. Behav. Ecol. Sociobiol., 17: MØLLER, A. P., 1987: House sparrow (Passer domesticus) communal displays. Anim. Behav., 35: MØLLER, A. P. & BIRKHEAD, T. R., 1993: Cuckoldry and sociality a comparative-study of birds. Am. Nat. 142: MØLLER, A. P. & NINNI, P., 1998: Sperm competition and sexual selection: a meta-analysis of paternity studies of birds. Behav. Ecol. Sociobiol., 43: SLAGSVOLD, T. & DALE, S., 1991: Mate choice models: can cost of searching and cost of courtship explain mating patterns of female pied flycatchers? Ornis Scand., 22: STUTCHBURY, B. J. & MORTON, E. S., 1995: The effect of breeding synchrony on extra-pair mating systems in songbirds. Behaviour, 132: STUTCHBURY, B., RHYMER, J. M. & MORTON, E. S., 1994: Extra-pair paternity in the hooded warbler. Behav. Ecol., 5: SUMMER-SMITH, D., 1954: The communal display of the house sparrow Passer domesticus. Ibis, 96: THUSIUS, K.J., DUNN, P. O., PETERSON, K. A. & WHITTINGHAM, L. A., 2001: Extrapair paternity is influenced by breeding synchrony and density in the common yellowthroat. Behav. Ecol., 12: TOST, J., 1994: The effects of colony size and male quality on paternity assurance strategies in house sparrows (Passer domesticus L.). Unpublished M.Sc. thesis. University of Vienna, Vienna. VÁCLAV, R. & HOI, H., 2002: Different reproductive tactics in house sparrows signalled by badge size: is there a benefit of being average? Ethology (in press). VÁCLAV, R., HOI, H. & BLOMQVIST, D., 2002: Badge size, paternity assurance behaviours and paternity losses in male house sparrows. J. Avian. Biol. (in press). VEIGA, J. P., 1990: Sexual conflict in the house sparrow: interference between polygynously mated females vs. asymmetric male investment. Behav. Ecol. Sociobiol., 27: VEIGA, J. P., 1992: Why are house sparrows predominantly monogamous? A test of hypotheses. Anim. Behav., 43: VEIGA, J. P. & BOTO, L., 2000: Low frequency of extra-pair fertilisations in House Sparrows breeding at high density. J. Avian Biol., 31: WAGNER, R., 1993: The pursuit of extra-pair copulations by female birds: a new hypothesis of colony formation. J. Theor. Biol., 163: WEATHERHEAD, P. J., 1997: Breeding synchrony and extra-pair mating in red-winged blackbirds. Behav. Ecol. Sociobiol., 40: WESTNEAT, D. F., 1992: Do female red-winged blackbirds pursue a mixed-mating strategy? Ethology, 92: WESTNEAT, D. F. & SHERMAN, P. W., 1997: Density and extra-pair fertilizations in birds: a comparative analysis. Behav. Ecol. Sociobiol., 41: WESTNEAT, D. F., SHERMAN, P. W. & MORTON, M. L., 1990: The ecology and evolution of extra-pair copulations in birds. In: Power, D. M. (ed.), Current ornithology. Vol. 7. Plenum Press, New York: WETTON, J. H., BURKE, T., PARKIN, D. T. & CAIRNS, E., 1995: Single-locus DNA fingerprinting reveals that male reproductive success increases with age through extra-pair paternity in the house sparrow (Passer domesticus). Proc. R. Soc. Lond B, 260: WETTON, J. H, CARTER, R. E., PARKIN, D. T. & WALTERS, D., 1987: Demographic study of a wild House Sparrow population by DNA fingerprinting. Nature, 327: WETTON, J. H. & PARKIN, D. T., 1991: An association between fertility and cuckoldry in the house sparrow, Passer domesticus. Proc. R. Soc. Lond B, 245: WITTENBERGER, J. L. & HUNT, G. L. Jr., 1985: The adaptive significance of coloniality in birds. In: Farner, D. S., King, J. R. & Parkes, K. C. (eds.), Avian Biology. Vol. 8. Academic Press, Orlando:

49 Folia Zool. 51(1): (2002) Bagrichthys majusculus, a new catfish from Indochina (Teleostei, Bagridae) Heok Hee NG Fish Division, Museum of Zoology, University of Michigan, 1109 Geddes Avenue, Ann Arbor, Michigan , USA; heokheen@umich.edu Received 20 November 2000; Accepted 29 August 2001 A b st r a c t. Bagrichthys majusculus, a new species of bagrid catfish from Indochina, is very similar to B. macracanthus and B. vaillantii, and has been previously identified as the former species. It differs from congeners in having a unique combination of the following characters: relatively large and broad mouth, well-developed oral dentition with homodont teeth, gill rakers, moderately-long dorsal spine with serrations, 9 pectoral-fin rays, inner and outer mandibular barbels with straight margins, pectoral-spine length % SL (standard length), dorsal-spine length % SL, length of adipose-fin base % SL, adipose maximum height % SL, depth of caudal peduncle % SL, and head depth % SL. Key words: new species, B. macracanthus, B. vaillantii, Southeast Asia Introduction The genus Bagrichthys consists of highly-specialised bagrid catfishes found in large rivers throughout Southeast Asia. They are characterised by their elongate and laterally compressed caudal peduncle, dorsally-directed serrations on the posterior edge of the dorsal-fin spine, gill membranes united at the isthmus, and a long adipose fin without a free posterior margin. The taxonomy of Bagrichthys has been the subject of recent work (R o b e r t s 1989, N g 1999, 2000), and six species are currently recognised, viz. B. hypselopterus (Bleeker, 1852), B. macropterus (Bleeker, 1853), B. macracanthus (Bleeker, 1854), B. vaillantii (Popta, 1906), B. micranodus Roberts, 1989, and B. obscurus Ng, Whilst comparing specimens identified as B. macracanthus from Indochina with those from Sumatra and Borneo, several significant differences in morphology were observed between the two populations. The population from Indochina is thus described in this study as belonging to a new species, B. majusculus. Material and Methods Measurements were made point to point with dial callipers and data recorded to 0.1 mm. Counts and measurements were made on the left side of specimens whenever possible. Subunits of the head are presented as proportions of head length (HL). Head length itself and measurements of body parts are given as proportions of standard length (SL). Measurements and counts were made following N g (2000). Fin rays were counted under a binocular dissecting microscope using transmitted light. Vertebral counts were taken from radiographs following the method of R o b e r t s (1994). Numbers in parentheses following a particular fin-ray, branchiostegal-ray, gill-raker or vertebral count indicate the number of specimens with that count. Institutional codes follow Eschmeyer (1998). 49

50 Bagrichthys majusculus, new species (Fig. 1) Pseudobagrichthys macracanthus (non Bleeker, 1854) Bleeker 1865a: 34, 1865b: 175. Bagroides macracanthus (non Bleeker, 1854) Sauvage 1883: 161, Durand 1940: 30, S m i t h 1945: 378, Suvatti 1936: 77, 1950: 292, Kuronuma 1961: 6, T a k i 1968: 20, 1974: 50, Fig. 51, O r si 1974: 161, D e soutter 1975: 458, K o ttelat 1985: 270, M a i & Nguyen 1988: 49, Maiet al. 1992: 185. Bagrichthys hypselopterus (non Bleeker, 1852) Chevey 1932: 17, Chaux & Fang 1949: 342, Kuronuma 1961: 6. Bagroides hypselopterus (non Bleeker, 1852) K o ttelat 1989: 13. Bagrichthys macracanthus (non Bleeker, 1854) Jayaram 1968: 378 (in part), Roberts 1993: 33, Rainboth 1996: 139. Holotype: UMMZ , male, mm SL; Thailand: Ubon Ratchathani province, Khong Chiam district, Mun River at Ban Dan, 3 km upstream from Mekong River confluence; W. J. Rainboth, R. E. Arden & Y. Dhammigbavorn, 12 June Paratypes: CAS 92487, 1 male, mm SL; 1 female, mm L; Laos: Mekong at Ban Hang Khone, just below Khone Falls; T. R. Roberts, June July MCZ 51318, 1 male, mm SL, 1 female, mm SL; Thailand: Nong Khai market; T. R. Roberts, 13 May UMMZ , 1 female, 96.8 mm SL; Vietnam: Phung Ding province, S side of Can Tho island, 3 km SE of Can Tho; E. D. Buskirk, UMMZ , 1 female, mm SL; Thailand: Khon Kaen province, Nam Pong Reservoir, 50 m from E shore, 3 km S of fish landing; W. J. Rainboth et al., September UMMZ , 1 male, mm SL; 1 female, mm SL; Cambodia: Stung Treng morning market; W. J. Rainboth, N. van Zalinga & C. Rotha; 26 January UMMZ , 1 female, mm SL; Cambodia: Kandal province, Tonle Sap; W. J. Rainboth & C. Rotha, 13 February UMMZ , 1 male, mm SL; Laos: Mekong River at Ban Hang Khone, just below Khone Falls; I. Baird, date unknown. USNM , 1 female, mm SL; Thailand: Roi Et province, Lam Chi, 1.5 km below highway 23 bridge, 4 km W of Selaphum; WBD- Mekong expedition, 31 January USNM , 1 female, mm SL; Thailand: Ubon Ratchathani province, market in Ubon at edge of Mun River; WBD-Mekong expedition, 14 September Diagnosis: Bagrichthys majusculus differs from other congeners in having a unique combination of the following characters: relatively large and broad mouth, well-developed oral dentition with homodont teeth, gill rakers, moderately long dorsal spine with Fig. 1. Bagrichthys majusculus, paratype male, MCZ 51318, mm SL; Thailand: Nong Khai. 50

51 15 27 serrations, 8 9 pectoral-fin rays, inner and outer mandibular barbels with straight margins, pectoral-spine length % SL, dorsal-spine length % SL, length of adipose-fin base % SL, adipose maximum height % SL, caudal peduncle depth % SL, and head depth % SL. Description: Body long and compressed. Dorsal profile rising evenly and steeply from tip of snout to origin of dorsal fin, then sloping ventrally from there to end of caudal peduncle. Ventral profile flat to anal-fin base, then sloping dorsally from there to end of caudal peduncle. Anus and urogenital openings located at vertical through middle of adpressed pelvic fin. Skin smooth. Lateral line complete and midlateral, with long tubular epidermal extensions of the sensory pores. Head somewhat compressed and narrow, anterior part produced into a prominent protruding snout with somewhat truncate margin when viewed laterally. Mouth inferior and relatively small, with papillate lips. Gill openings wide, extending from exposed surface of posttemporal to beyond isthmus. Gill membranes free from, and not attached across, isthmus. Eye ovoid, horizontal axis longest; located entirely in dorsal half of head. Orbit with free margin. Barbels in four pairs. Maxillary barbel moderately long, extending to opercular flap. Nasal barbel slender and short, extending to posterior margin of orbit (females) or to midway between posterior margin of orbit and base of supraoccipital (males). Inner mandibular-barbel origin close to midline; barbel thicker and longer than nasal barbel and extending to level of anterior margin of orbit. Outer mandibular barbel originates posterolateral of inner mandibular barbel, extending to level of centre of orbit. Oral teeth small and in irregular rows on all tooth-bearing surfaces. Premaxillary tooth band rounded, of equal width throughout. Dentary tooth band broader than premaxillary tooth band, widening laterally and then tapering to a sharp point poterolaterally. Vomerine tooth band unpaired, continuous across midline; strongly arched along anterior margin; band width broader than premaxillary band. Dorsal fin located at middle of body; origin nearer tip of snout than caudal flexure. Dorsal-fin margin convex, usually with anterior branch of fin rays longer than other branches. Last dorsal-fin ray without posterior membranous connection to body. Dorsal-fin spine short, straight and robust, with serrations on posterior margin. Adipose fin with margin convex for entire length; posterior end deeply incised. Caudal fin deeply forked; upper and lower lobes pointed, with outermost principal fin-rays produced into filaments. Procurrent rays symmetrical and extending only slightly anterior to fin base. Anal-fin base ventral to posterior half of adipose fin. Fin margin curved or straight. Last anal-fin ray without posterior membranous connection to body. Pelvic-fin origin at vertical through posterior end of dorsal-fin base. Pelvic-fin margin slightly convex, tip of adpressed fin not reaching anal-fin origin. Pectoral fin with stout spine, sharply pointed at tip. Anterior spine margin smooth; posterior spine margin with serrations along entire length. Pectoralfin margin straight anteriorly, convex posteriorly. In % SL: head length , head width , head depth , predorsal distance , preanal length , prepelvic length , prepectoral length , body depth at anus , length of caudal peduncle , caudal peduncle depth , pectoralspine length , pectoral-fin length , dorsal-spine length , length of dorsal-fin base , pelvic-fin length , length of anal-fin base , caudal-fin length , length of adipose-fin base , adipose maximum height , post-adipose distance ; in % HL: snout length , interorbital distance , eye diameter , nasal barbel length (females) or 51

52 (males), maxillary barbel length (females) or (male), inner mandibular barbel length (females) or (males), outer mandibular barbel length (females) or (males). Branchiostegal rays 5 (7) or 6 (1). Gill rakers 3+7 (2), 3+9 (2), 4+8 (1) or 3+10 (1). Vertebrae 17+25=42 (1), 18+24=42 (1), 18+25=43 (3), 18+26=44 (2), 19+25=44 (4) or 19+26=45 (1). Fin ray counts: dorsal II,7 (8); pectoral I,8 (1) or I,9 (7); pelvic i,5 (8); anal iii,10 (1), iv,9 (1), iv,10 (2), iii,11 (1), iv,11 (2) or v,10 (1); caudal 8/9 (8). Colour in alcohol: dorsal surface of head and body uniform brown or dark brown, with a cream-coloured midlateral streak running along entire length of body. Large regularly spaced cream-coloured patches occasionally present on nape and sides of body below anterior third of adipose fin and immediately above posteriormost anal-fin ray; patches sometimes coalescing to form bands. Ventral surfaces of head and body cream; adipose fin and fin rays of all fins brown; inter-radial membranes of all fins with scattered melanophores. Distribution: Known from the Mekong and Chao Phraya River drainages in Indochina. Ecology: Bagrichthys majusculus feeds on crustaceans and other small benthic animals (T a k i 1978). It spawns at the beginning of the rainy season in flooded riparian forests, with juveniles beginning to appear in August (R a i n b o t h 1996). Etymology: From the Latin majusculus, meaning somewhat greater, in reference to the relatively larger adipose fin base, pectoral and dorsal spines of this species when compared to B. macracanthus, its closest congener. Remarks: Bagrichthys majusculus differs from B. hypselopterus in having fewer gill rakers on the first gill arch (10 13 vs ), serrations on the posterior margin of the pectoral spine (15 27 vs. 60 or more) and pectoral-fin rays (8 9 vs ), the nape and dorsal-fin base slightly (vs. very strongly) elevated, inner and outer mandibular barbels with straight margins (vs. convoluted anterior margins), and jaw teeth homodont, i.e. of one kind (vs. markedly heterodont, i.e. of different kinds). It differs from B. macropterus, B. micranodus, and B. obscurus in having a relatively large and broad (vs. small and narrow) mouth with well-developed, i.e. with exposed teeth forming well defined tooth bands (vs. extremely reduced, i.e. with few scattered teeth deeply buried in soft tissue) oral dentition, and a longer pectoral spine ( % SL vs ). Bagrichthys majusculus most closely resembles B. macracanthus and B. vaillantii, but can be differentiated from both species in having a longer pectoral spine (pectoral-spine length % SL vs ), the former in having a larger adipose fin (length of adipose-fin base % SL vs ), deeper caudal peduncle (depth of caudal peduncle % SL vs ), and the latter in having a longer dorsal spine (dorsal-spine length % SL vs ), more slender head (head depth % SL vs ) and a deeper adipose fin (adipose maximum height % SL vs ). From the results of this study and a previous one by N g (1999), it is clear that there are two species of Bagrichthys in Indochina that are closely related to morphologically similar Sundaic species. Although R a i n b o t h (1996) postulated that there may be more than one species of Bagrichthys with long dorsal spines in the Mekong River, the results here indicate that B. majusculus is the only such species present in the Mekong. The historical biogeography of Indochinese Bagrichthys is probably similar to that for Hemisilurus proposed by B ornbusch & Lundberg (1989), i.e. an ancestral Bagrichthys in the North Sunda River system splitting into the B. hypselopterus + B. macracanthus + B. majusculus (diagnosed as having long dorsal spines with 18 or more serrations in adults, and exposed jaw and palatal teeth forming well defined tooth bands) 52

53 and B. macropterus + B. micranodus + B. obscurus lineages (diagnosed as having short dorsal spines with 15 or less serrations in adults, and scattered jaw and palatal teeth deeply buried in soft tissue) not later than the Pleistocene, with subsequent divergence by the post- Pleistocene isolation of the Indochinese and Sundaic populations. Comparative material: Bagrichthys hypselopterus: CAS 49366, 2 ex., mm SL; Borneo: Kalimantan Barat, fish market at Sintang. CAS 49367, 1 ex., mm SL; Borneo: Kalimantan Barat, Sungai Tawang near Pengembung. ZRC 40472, 1 ex., mm SL; ZRC 41532, 15 ex., mm SL; ZRC 41898, 1 ex., mm SL; Sumatra: Jambi, Pasar Angso Duo (fish market). Bagrichthys macracanthus: CMK 11278, 2 ex., Sumatra: Jambi, Danau Arang Arang. ZMA , 3 ex., mm SL; Sumatra: Jambi, Batang Hari. ZRC 14251, 1 ex., mm SL; Peninsular Malaysia: Selangor, North Selangor peat swamp forest, pools at Sungai Bernam headworks. ZRC 38547, 2 ex., mm SL; Sumatra: Jambi, Sungai Kumpeh in Arang Arang. ZRC 38633, 1 ex., mm SL; Sumatra: Jambi, Danau Arang Arang. ZRC 39034, 1 ex., mm SL; Sumatra: Riau, Sungai Bengkwan, tributary of Indragiri River, 4 hours downstream from Rengat. ZRC 41533, 1 ex., mm SL; ZRC 41897, 1 ex., mm SL; ZRC 44118, 12 ex., mm SL; Sumatra: Jambi, Pasar Angso Duo (fish market). ZRC 42601, 1 ex., 52.8 mm SL; Sumatra: Jambi. ZRC 44161, 4 ex., mm SL; Sumatra: Jambi, Sungai Alai. ZRC 46030, 2 ex., mm SL; Borneo: Sarawak, Kapit market. Bagrichthys macropterus: USNM , 4 ex., mm SL; Borneo: Kalimantan Barat, Kapuas mainstream 58 km NE of Sintang and 1 km downstream from Sebruang. ZRC 38997, 1 ex., mm SL; ZRC 41534, 5 ex., mm SL; Sumatra: Jambi: Pasar Angso Duo (fish market). Bagrichthys micranodus: CAS 49369, 2 paratypes, mm SL; USNM , 3 paratypes, mm SL; Borneo: Kalimantan Barat, Kapuas River 58 km NE of Sintang and 1 km downstream from Sebruang. CMK 7015, 1 ex., mm SL; Borneo: Kalimantan Barat, Kapuas River at Nanga Embaloh. Bagrichthys obscurus: USNM , holotype, mm SL; Thailand: Roi Et province, Lam Chi, 1.5 km below highway 23 bridge, 4 km W of Selaphum. CAS 92567, 1 paratype, mm SL; Laos: Mekong below Khone Falls. CAS 92580, 2 paratypes, mm SL; Laos: Mekong at Ban Hang Khone, just below Khone Falls. UMMZ , 1 paratype, mm SL; UMMZ , 1 paratype, mm SL; Thailand: Ubon Ratchathani province, Khong Chiam district, Mun River at Ban Dan, 3 km upstream from confluence with Mekong River. UMMZ , 2 paratypes, mm SL; Cambodia: Stung Treng morning market. USNM , CAS 61918, 22 ex., mm SL; Thailand: fish market at Ubon Ratchathani. Bagrichthys vaillantii: RMNH 7839, 1 ex., 90.6 mm SL; Borneo: Tepoe. CAS 93398, 1 ex., mm SL; eastern Borneo. CAS 93403, 1 ex., mm SL; Borneo: Kalimantan Timur, Sungai Belayan from mouth of Sungai Sentekan southwards for about 6 km. CMK 7635, 5 ex., mm SL; Borneo: Kalimantan Timur, stream connecting Mahakam River to Danau Semajang. Acknowledgements I am grateful to the following for permission to examine material under their care: William E s c h m e y e r (CAS), Maurice K o t t e l a t (CMK), Karsten H a r t e l (MCZ), Martien van O i j e n (RMNH), Douglas Nelson (UMMZ), Lynne Parenti (USNM), Isaäc I s brücker (ZMA), Peter N g (ZRC). I also thank H. H. T a n for assistance in obtaining data. This work was funded by support from the Rackham School of Graduate Studies, University of Michigan. 53

54 LITERATURE BLEEKER, P., 1865a: Nouvelle notice sur la faune ichtyologique de Siam. Ned. Tijdschr. Dierk., 2: BLEEKER, P., 1865b: Sixième notice sur la faune ichtyologique de Siam. Ned. Tijdschr. Dierk., 2: BORNBUSCH, A. H. & LUNDBERG, J. G., 1989: A new species of Hemisilurus (Siluriformes, Siluridae) from the Mekong River, with comments on its relationships and historical biogeography. Copeia, 1989: CHAUX, J. & FANG, P. W., 1949: Catalogue des Siluroidei d Indochine de la collection du laboratoire des pêches coloniales au muséum, avec la description de six espèces nouvelles. (Suite et fin). Bull. Mus. Natn. Hist. Nat., 2e sér., 21: CHEVEY, P., 1932: Inventaire de la faune ichtyologique de l Indochine. Deuxième liste. Notes Inst. Océanogr. Indochine, 19: DESOUTTER, M., 1975: Étude de quelques Bagridae (Siluriformes, Pisces) du Cambodge. Description d une espèce nouvelle : Mystus aubentoni. Bull. Mus. Natn. Hist. Nat., 3e sér., 206: DURAND, J., 1940: Notes sur quelques poissons d espèces nouvelles ou peu connues des eaux douces cambodgiennes. Notes Inst. Océanogr. Indochine, 36: ESCHMEYER, W. N., 1998: Catalog of Fishes. California Academy of Sciences, San Francisco. JAYARAM, K. C., 1968: Contributions to the study of bagrid fishes (Siluroidea: Bagridae). 3. A systematic account of the Japanese, Chinese, Malayan, and Indonesian genera. Treubia, 27: KOTTELAT, M., 1985: Fresh-water fishes of Kampuchea A provisory annotated check-list. Hydrobiologia, 121: KOTTELAT, M., 1989: Zoogeography of the fishes from Indochinese inland waters with an annotated check-list. Bull. Zoöl. Mus., 12: KURONUMA, K., 1961: A Check List of the Fishes of Vietnam. Division of Agriculture and Natural Resources United States Operations Mission to Vietnam. MAI, D. Y. & NGUYEN, V. T., 1988: Species composition and distribution of the freshwater fish fauna of southern Vietnam. Hydrobiologia, 160: MAI, D. Y., NGUY ~ EN, V. TRỌNG, NGUY ~ EN, V. THIỆN, LÊ, H. Y. & HÚ A, B. L., 1992: Dịnh loại các loàicá nu ó c ngọt Nam Bộ. [Identification of the Fresh-water Fishes of South Vietnam]. Nhà Xuâ t Bán Khoa Học Và Ky ~ Thuật, Hanoi. NG, H. H., 1999: Bagrichthys obscurus, a new species of bagrid catfish from Indochina (Teleostei: Bagridae). Rev. Biol. Trop., 47: NG, H. H., 2000: Bagrichthys vaillantii (Popta, 1906), a valid species of bagrid catfish from eastern Borneo (Teleostei: Siluriformes). Zool. Med., 73: ORSI, J. J., 1974: A check list of the marine and freshwater fishes of Vietnam. Pub. Seto Mar. Biol. Lab., 21: RAINBOTH, W. J., 1996: Fishes of the Cambodian Mekong. FAO Species Identification Field Guide for Fishery Purposes, FAO, Rome. ROBERTS, T. R., 1989: The freshwater fishes of Western Borneo (Kalimantan Barat, Indonesia). Mem. California Acad. Sci., 14: ROBERTS, T. R., 1993: Artisanal fisheries and fish ecology below the great waterfalls of the Mekong river in southern Laos. Nat. Hist. Bull. Siam Soc., 41: ROBERTS, T. R., 1994: Systematic revision of the Asian bagrid catfishes of the genus Mystus sensu stricto, with a new species from Thailand and Cambodia. Ichthyol Explor. Freshwaters, 5: SAUVAGE, H. E., 1883: Sur une collection de poissons recueillis dans le Mé-Nam (Siam) par M. Harmand. Bull. Soc. Philom. Paris, Ser. 7, 7: SMITH, H. M., 1945: The freshwater fishes of Siam or Thailand. Bull. U. S. Natn. Mus., 188: SUVATTI, N. C., 1936: Index to Fishes of Siam. Bureau of Fisheries, Bangkok. SUVATTI, N. C., 1950: The Fauna of Thailand. Department of Fisheries, Bangkok. TAKI, Y., 1968: Notes on a collection of fishes from lowland Laos. United States Agency for International Development Mission to Laos Agriculture Division. TAKI, Y., 1974: Fishes of the Lao Mekong basin. United States Agency for International Development Mission to Laos Agriculture Division. TAKI, Y., 1978: An analytical study of the fish fauna of the Mekong basin as a biological production system in nature. Research Institute of Evolutionary Biology Special Publications, 1, 77 pp. 54

55 Folia Zool. 51(1): (2002) Movements of barbel, Barbus barbus (Pisces: Cyprinidae) Dedicated to the late Professor Antonín Lelek - our friend and colleague Milan PE ÁZ, Vlastimil BARU, Miroslav PROKE and Miloslav HOMOLKA Institute of Vertebrate Biology, Academy of Sciences of the Czech Republic Kvûtná 8, Brno, Czech Republic; penaz@brno.cas.cz Received 10 May 2001; Accepted 7 August 2001 A b s t r a c t. Altogether 701 adult barbel, Barbus barbus were captured by electrofishing and individually tagged to study their local displacement and movements in a stretch of the River Jihlava (Czech Republic). A total of 149 fish were recaptured and 105 of them (70.47 %) were considered as resident because they were always recaptured in the same, relatively restricted ( m) stream section, which always contained a pool and was demarcated naturally by riffles on both edges. The remaining 44 recaptured specimens (29.53 %) belonged to the mobile part of population, their movements encompassing two (or exceptionally more) adjacent stream sections and at maximum a distance of 1680 m downstream or 2020 m upstream. The proportion of mobile barbel, relatively low in smaller and middle size classes, increased in the largest size classes ( mm of SL). A rather limited extent of movements also suggests a relatively small area of home range in the studied stretch, which nevertheless provides satisfactory resources and favourable conditions required by barbel over their entire life cycle. The extent of movements and corresponding proportion of mobile fish appear to be increasing with diminishing habitat patchiness. In the stretch of River Jihlava studied, with a rich patchy heterogenous habitat and well developed riffle-pool-raceway structure, each section (pool) can be considered as a more or less isolated spatial unit containing its own, and in a certain degree, isolated component of a metapopulation. Key words: tagging, resident and mobile fish, territoriality, metapopulation, riffle-pool development Introduction The barbel Barbus barbus (Linnaeus, 1758) is a popular and much studied species whose biology is still insufficiently understood. Unfortunately it is also a species currently vanishing from many European rivers. Already four International Round Tables (Montpellier France in 1989, Liége - Belgium in 1993, Liblice Czech Republic in 1995 and Thessaloniki Greece in 1997) have been devoted to this and other Barbus species, providing a good information tool of actual knowledge and a solid base stimulating their further research. Within this context, the barbel is also the subject of an extensive research project on the River Jihlava, Czech Republic. The construction and operation of hydroelectric complex on this stream and its close vicinity created conditions that interfered with natural environmental stimuli (temperature, photoperiod) triggering gonad maturation and spawning of the barbel (PeÀáz et al. 1999). The population appeared to be suppressed for a period after the sudden change in habitat caused by the start of hydroelectric operation operating energy complex. Subsequently, the adaptation processes seems to have been completed and the barbel population was found again viable, prosperous and dominating the fish community (P eàáz et al. 1999). The aim of present partial study was to analyse and 55

56 quantify the movements of barbel as well as to assess the results relative to situation existing in some other rivers and barbel populations studied in a similar way (Steinmann et al. 1937, Hunt & Jones 1974, Pelz & Kästle 1989, Baras 1995, 1997, Lucas & Batley 1996). Study Area Our survey took place in a 3.0 km stretch of the River Jihlava (Czech Republic), a fifth-order stream (tributary to the River Svratka, Danube basin) near the village of Hrub ice between river kilometers (RK) and (49 07 N, E) (Fig. 1). The study stretch is characteristic of submountain streams, running through an incised valley, morphologically heterogenous, corresponding to a barbel zone (H u e t 1949) and actually influenced by an upstream located energy operating hydro- and nuclear complex. The main environmental impacts exerted on the river, described in particular by PeÀáz et al. (1999), were a regulated discharge and reduced temperature variations. These, rather favourable habitat changes were followed by considerable changes in composition of fish assemblage. The former fish community dominated by barbel and other stream cyprinids was replaced by a salmonid community dominated by brown trout Salmo trutta m. fario, however, the barbel maintained a strong and in fact increased presence in present fish community, which consists of at least 25 species. Except of barbel, species recorded during experimental electrofishing were rainbow trout Oncorhynchus mykiss, brook trout Salvelinus fontinalis, grayling Thymallus thymallus, roach Rutilus rutilus, chub Leuciscus cephalus, dace L. leuciscus, rudd Scardinius erythrophthalmus, tench Tinca tinca, nase Chondrostoma nasus, gudgeon Gobio Fig. 1. Map of the study area. 56

57 gobio, bleak Alburnus alburnus, spirlin Alburnoides bipunctatus, crucian carp Carassius carassius, common carp Cyprinus carpio, eel Anguilla anguilla and, in addition, by sport fishing were recorded also common bream Abramis brama, Eurasian perch Perca fluviatilis, pike-perch Stizostedion lucioperca, asp Aspius aspius, ide Leuciscus idus, goldfish Carassius auratus and grass carp Ctenopharyngodon idella (PeÀáz et al. 1999). The habitat of river under study was very heterogenous and had a typical patchy riffle pool raceway structure, with fully natural and stabilised river bed and banks, with abundant spawning grounds but with no tributaries habitable for barbel (Table 1). The study stretch was divided into seven sections (A G), each containing a complete set of various microhabitats, mainly a deeper pool, long raceway and separated between adjacent sections by riffles as natural boundaries. These patches differ mainly by depth and flow velocity: riffles were characterised by a water velocity of m.s -1 and a depth of cm, raceway by m.s -1 and cm while the pools by m.s -1 and a depth of cm. The shallow parts of raceways are often overgrown by dense macrophytes (Batrachium fluitans, Callitriche sp.) during spring and summer. Table 1. Sections of the stretch of River Jihlava under study and their environmental characters (RK = river kilometers). Section RK Area Length Proportion of patches, % hectares m riffle pool raceway A B C D E F G Total Methods and Material A gasoline powered electroshocker (DC, 250 V, A, 50 Hz) was used to collect the fish. Electrofishing excursions were carried out on 15 dates: , , , , , , , , , , , , , and All fish except of barbel 120 mm of SL were after capture immediately released. After being measured, weighed and examined for sex according to external features (mainly running sexual products) recognisable only during reproduction season, 701 barbel were individually marked by means of the anchor full plastic tags (Floy Tag - type FD-94). Tags were fixed on the left body side into the dorsal musculature, just below the anterior edge of dorsal fin. Different colours of tags were applied in consecutive years (yellow 1999, white 2000, red 2001). The effect of tags on fish behaviour, survival and growth was minimised by the type of tags used and their careful application. The tags proved to be durable and easy to recognise in the field even after long periods (2 years). The tagged and recaptured fish were always released in centres of the same particular sections (see Table 1). The distance between centres of sections where the fish were released 57

58 and centres of sections where they were recaptured, was considered as distance that a recaptured fish had moved. All recaptures were obtained by the experimental fishing exclusively. Barbel do not represent a favourite coarse fish in sport fishery conducted in the stretch of River Jihlava under study, where presently the brown trout and other salmonid fishes have been preferably appraised, nevertheless, 30 to 87 individuals were caught by anglers annually during in the fishing ground Jihlava 5B located between RK 45.9 and 55.6 (19 hectares) and encompassing also the river stretch under study. Despite the fact that the fishermen were informed about the tagging project, none have so far reported capturing a tagged barbel. Occasionally, the fishing ground concerned is also stocked by artificially reared one-yearold barbel in a quantity amounting individuals annually. The fish were classed according to their recapture rate and location as: 1) specimens with restricted home range (i.e. inhabiting permanently only the same single stream section where they were captured, after tagging released and eventually recaptured); they are designated as resident (also home or stationary) fish; and 2) specimens with a larger home range (extending over the two or more neighbour stream sections, i.e. those recaptured at least once beyond the boundaries of the home section where they were originally released); these specimens are designated as mobile or stray fish. The differences in frequencies of recaptures of individual size classes and stream sections were analysed by means of chi-square test (S ok a l & R oh l f 1981). The unequal numbers of fish tagged in individual size categories and subsequent stream sections, as well as different length and area of these sections, were respected when computing the expected frequencies. The stream section G, where only one fish was tagged, was excluded from the statistical analyses. Results Recapture rate In total 149 of 701 tagged barbel (21.26 %) were recaptured (Table 2). Of them, 117 were recaptured once, 24 twice, seven specimens three times and one fish was recaptured four times. The distribution of recaptures among subsequent stream sections fluctuated % (Table 2). Recapture frequencies plotted against consecutive recapture categories showed a distinct exponential relation (Fig. 2), which may have resulted from a higher natural mortality, eventual loss of tags. However, most likely it is caused by the fact that always only a minor part of population has been caught and the probability of recapture was thus relatively low. The expected values of successive multiple recaptures, calculated by means of the recapture coefficient C r were not significantly different from the observed values in total (χ 2 = 1.465; P<0.05) nor in all subsequent recapture categories (Table 3). The C r coefficient can be thus considered, under these environmental conditions and using the fishing methods described, valid for calculating the quantities of the single or multiple recaptures of marked fish. Resident and mobile fish and their home range Of the 149 barbel recaptured and analysed for their movements, 105 specimens (70.47 %) could be classed as resident, i.e. steadily living in a single stream section - a fairly confined 58

59 Table 2. Tagging and recapture results with barbel in the River Jihlava; proportion of resident and mobile fish. Section Number of Recaptured fish, N/% tagged fish resident 1 mobile 2 all A B C D E F G Total/Mean fish recaptured in the same of successive stream section as they were released after tagging 2 fish recaptured at least once outside of the home section where they were captured and released when tagged 3 S.D. = ±7.26 Fig. 2. Recapture rate of the barbel from the River Jihlava, Czech Republic, in successive multiple recaptures. Recapture category 1-4 explained in the legenda to Table 3. ( m) area demarcated by the two riffles (Table 2). This group of fish did not move beyond stream riffles and respected them as boundaries of their home range. The remaining 44 (29.53 %) of recaptured barbels moved in various, however rather restricted, extent from their home section where they were first caught and after tagging again released. According to our arbitrary and perhaps rather rigorous criteria, these fish were categorised as mobile part of barbel population under study. The home range of mobile barbel is thus formed by the two or even more of stream sections. Mobile barbel crossed the riffles, however, 93.2 % of them have been recaptured within a distance (1000 m from the place of their release after tagging. No difference was found between the numbers of mobile fish wandering upstream and those wandering downstream (Table 4). 59

60 Table 3. Recapture frequencies of the barbel in the River Jihlava and the difference between actual and expected values in consecutive recapture categories (1 number of fishes recaptured 1 4 times; 2 = fish recaptured 2 4x; 3 fish recaptured 3 4x; 4 fish recaptured 4 times; 5 fish recaptured 5 times); number of tagged fish = 701. Recapture Number of recaptured fish Recapture χ 2 P df category actual theoretical expected 1 coefficient 2 N a N t N e C r > > > Sum > Expected frequency N e = N t /C rn, where n = recapture category (1 4). 2 Recapture coefficient was computed as weighted mean of shares N t /N o of successive recapture categories (1 4) Table 4. The distribution of mobile barbel according to the distance of movement realised in the River Jihlava, Czech Republic. Distance from Fish recaptured downstream Fish recaptured upstream Home section (m) number % number % All An above-average number of barbel were recaptured in that stream section exhibiting a lower proportion of pools (D), whereas the lower recapture rate was found in a section with a higher pool proportion and low area of riffles (A). Resident fish were more frequently recaptured, compared to the mobile ones, in the section with low proportion of pools (D) whereas the mobile fish dominated the resident ones in the section with the above-average proportion of pools and riffles and low proportion of raceways (Table 5). Table 5. Chi-squared test for significant difference (P < 0.05; denoted by a < or > symbol) between prevailing habitat patches and the numbers of tagged and recaptured fish in subsequent stream sections (< = observed values less than expected; > = observed value greater than expected; R = resident > mobile; M = mobile > resident; NS = nonsignificant difference, P > 0.05). Stream Riffle Pool Raceway No of fish Resid/mobile section tagged recaptured fish A < > NS > < NS B NS NS NS > NS NS C < < < > NS NS D NS < NS < > R E > > < < NS M F NS NS NS < NS NS 60

61 Influence of fish size All tagged fish as well as all recaptured fish were divided into 50 mm size (SL) classes and the recaptured fish additionally subdivided into the resident and mobile parts (Table 6). Only one tagged fish was recaptured within the smallest size class ( mm), however, the proportion of recaptured fish increased in consecutive size classes within the size range mm of SL (Fig. 3) and this relation showed to be significant (Pearson correlation coefficient, r = 0.915; P<0.001; N = 10). The probability of recapture thus increases with the size of fish, which may also result from higher susceptibility of larger fish towards the D.C. electricity. With an apparent fluctuation in individual size classes, the proportion of resident barbels predominated in the size range until SL of 450 mm, whereas the proportion of mobile fish exceeded that of resident ones in the two largest, less frequented size classes ( mm SL) (Table 6). A further analysis, which followed after the size classes were collected by a 100 mm interval into four size categories only, resulted in following conclusions: 1) The number of recaptured individuals belonging to subsequent size categories was not equal in the total sample of fish (χ 2 = 10.49; P < 0.025; d.f. = 3) whereby an above-average number of mobile individuals was captured in the smallest size category (p < 0.005). 2) Neither it was equal in the subsample of resident individuals (χ 2 = ; P < 0.025; d.f. = 3) in which the above-average number was found in the size category mm (p < 0.025). 3) The unequal number of recaptures existed also in mobile fish (χ 2 = ; P < 0.005; d.f. = 3) and they were higher than the expected one in the highest size category mm (P < 0.005). The comparison of frequencies of recaptured mobile and resident fish (contingency table 2x4) showed that they were not equal (χ 2 = 15.51; P < 0.05; df = 3). Against expectations, more resident fish were caught than mobile ones in the size category mm (P < 0.025), whereas more mobile than resident individuals were recaptured in the largest size category mm (p < 0.005). Table 6. Recapture data of Barbus barbus analysed according t the size (SL) and residentiality; (size related with the date of tagging). Size No of No of recaptured fish Proportion* of class tagged all fish repeatedly recaptured resident fishmobile fish mm, SL fish n % 1x 2x 3x 4x n n Total * Proportion expressed in per cent of total number of all recaptured fish 61

62 Fig. 3. Dependence of recapture rate and mobility of barbel upon their size in the River Jihlava, Czech Republic. (Expressed as percentages of the fish recaptured against fish tagged in consecutive size classes). Discussion General nature of barbel s movements According to G e r k i n g (1953), many stream fishes live in very restricted areas during most, if not all, of their life. The problem of their local movements has to be recognised from the two viewpoints. Firstly, home range (sensu G e r k i n g 1953) is understood as the area over which the fish normally travels, searching for the optimum element of habitat (patch, microhabitat) required at given time and phase of its life cycle (foraging, spawning, nursery, resting, hiding, wintering etc.). In the gregarious barbel, home range selection and occupation is conditioned not only by the availability of suitable physical habitat, mainly feeding areas, but also by the actual status and abundance of the population and by the presence of sufficiently numerous shoals (B a r a s 1997). In our study, home range and its area were considered only indirectly, based on the location where tagged fish have been released and repeatedly recaptured and on the extent of movements exerted. Secondly, territory is any area that is defended, mostly with elements of aggressive behaviour. Generally, territorial behaviour concerns mainly reproductive territories and is markedly pronounced in monogamic species spawning in couples, in nest-building and in the egg and offspring protecting or guarding species and as such the territoriality could rather be a seasonal phenomenon. Nothing of this seems to be the case in barbel, which spawn collectively in numerous aggregates, and are neither hiding nor guarding their eggs and not exhibiting any kind of parental care for young. However, some features of territorial behaviour could be recognised in barbels spawning behaviour. The chase-away interactions between males, mainly between the courting ones, have a character of spatial competition and defence (Hancock et al. 1976, Baras 1994, P o ncin et al. 1996). Furthermore, the territoriality may be a true behaviour pattern of the barbel population under study because of its sound structure, characterised by a high reproduction potential and recruitment whilst simultaneously subjected to a low predation and fishing pressure. This 62

63 situation is expected to increase competition for territories and food resources, eventually creating aggressive behaviours and the subsequent emigration into underpopulated areas of submissive (smaller, younger) individuals, which may appear as strays (mobile fish) in our observations. A similar pattern has been reported for by-passed sections of the upper River Rhône, where reductions in deep-water habitat due to flow to a hydroelectric plant resulted in smaller and younger nase to emigrate because the available deep-water habitat was occupied by older and larger nase (P ersat & Chessel 1989). Such aspects of territorial behaviour, however, still were not satisfactorily understood in the barbel. Strong recruitment of young fish, together with behavioural dominance of old fish would substantiate the expectation that rather young fish would become mobile and search for new territories. However, rather the opposite situation seems to be true in the River Jihlava, where the mobility rate increased with increasing size (and age) of fish. Similar fact was generally stated also by Gerking (1953) and specially for the barbel found by Hancock et al. (1976). This pattern may be associated with a higher reproduction activity in larger barbel as well as their greater susceptibility to being caught by electrofishing. However, B aras (1995) found no relation between fish size and patterns of locomotory activity. The barbel s locomotory activity and extent of movements exhibit clear diel and seasonal patterns and are mainly related to the temperature and length of day. Foraging was found the main stimulus determining the localised activity in the diel scale, whereas spawning is the main motivation for movements of the seasonal scale (Lucas & Batley 1996). B a r a s (1995) found the diel and seasonal locomotory activity patterns in the barbel thermal-dependent and a temperature of 4.0 C, when barbel entered a dormancy period, as a thermal lower limit for its activity. However, water temperatures below 4 C are less common in our study area during winter months due to the influence of energy operating complex (see PeÀáz et al. 1999). Spatial dimension of movements in different rivers Bibliographic sources report much variation in the barbel s movements in various riverine habitats and often this species is considered as a wandering fish. Some of barbels tagged by Steinmann et al. (1937) were re-captured in a distance more than 300 km from their release site. Regardless of ability of some individuals to move over long distances, most barbel populations have been characterised as resident, of course often with different criteria used for the distance fixed as the boundary between the residential and mobile movement patterns. For example Steinmann et al. (1937) considered as resident all fish recaptured within a ± 5 km distance from the site where they were released. These authors found that resident were 48 % of 216 recaptured individuals in the German stretch of the Danube, 54 % of 130 recaptures in the Austrian Danube, whereas 73.7 % of 38 recaptured fish in the River Main. In the River Severn, the barbel population is also divisible into the mobile and resident components (Hunt & Jones 1974b), however, in a much larger scale than found in the River Jihlava: 86 % of 531 recaptured barbels, considered as resident, moved within a 5 km distance from the point of release in the Severn (of them 54 % did not move at all and permanently occupied the same place). The remaining 14 % wandered more widely, both upstream and downstream, up to a maximum of 34 km from the site of tagging (the mobile fish). Total range over which the barbel were recaptured was 54 km. Similarly, in the River Nidd (a rather deep stream, sectioned by weirs, NE England, tributary to Ouse) some individuals moved nearly 20 km upstream. The radiotracking carried out in this river 63

64 brought also the evidence that some tracked barbels moved regularly between the Rivers Nidd and Ouse and it was demonstrated that each of these rivers is used by them at different times of the year (Lucas & Batley 1996, Lucas & Frear 1997). A short-term (12 days) radiotracking survey conducted on eight barbel in the River Nidda (tributary to the Main, Germany) showed that the fish behaved as significantly resident moving within a distance of m only (Pelz & Kästle 1989). In large rivers, the habitats required by stream fishes are often separated by very large distances relative to a species dispersal abilities and considerable movements must be thus exhibited by the fish. Migration behaviour of the barbel could be also inferred from their seasonal occurence in fish ladders. For instance in a fish ladder built adjacent to the Stfiekov weir on the River Labe, the barbel was the most frequently (19 49 %) entering fish species during a survey conducted in and a successful passage has been confirmed (Novotn & Punãocháfi 1996). When compared to most of published data, the extent of movements and consequently the home range of barbel in the River Jihlava seem to be very limited and the resident character of behaviour is probably the most profoundly expressed of all studied populations despite the fact that we have fixed the maximum limit distance for movements of resident specimens to a relatively very low value of m. The very heterogenous and patchy character of the study stretch (and in each of the stream sections as defined in our survey) may contribute to the observed patterns by ensuring very favourable conditions for all phases of the life cycle (gonad maturation, spawning, larval and juvenile nursery, foraging, hiding, wintering etc.) within a relatively very small area compared to other locations. Especially, the richness of available spawning sites reduces the need for extended spawning migrations and this is reflected in home range fidelity as well as their behaviour patterns. Patchy habitat structure is followed by patchy population dispersion According to Gerking (1953), in streams with a well developed riffle pool structure, such as the River Jihlava, each pool (or better each stream section separated by riffles) can be considered as a more or less isolated unit containing a natural population of its own. The fish population of a small stream may thus be thought as a series of discrete, natural units rather than as a single, homogenous and freely missing group. This approach is, of course, near to the theory of metapopulation dynamics (H anski 1996, Hanski & Simberloff 1997). Consequently, the set of local, more or less isolated, barbel population of the River Jihlava, exhibiting a dynamic turnover (extinction and colonisation; Harrison 1998) could thus be considered a case of metapopulation. However, all this does not mean a fragmentation either of the physical habitat, or of the gene pool of the given population. Despite restricted extent of home range and well expressed residential behaviour, a sufficient gene exchange undoubtedly exists there as only a small genetic differentiation exists among the barbel populations of the Czech Republic ( lechtová et al. 1998). Importance of aquatic system connectivity Riffles, despite not being crossed by majority of barbel population, i.e. by resident specimens, do not fragment the stream physically, in fact they easily could be crossed by the fish. This 64

65 connectivity plays an important role in the formation and functioning of metapopulation (Schmutz & Jungwirth 1999). Barbel movements are notably restricted or impeded by the construction of dams, weirs and other water retention structures. This problem seems to be especially serious in those streams where natural habitat physical heterogeneity and riffle-pool development are in decline and the spawning migrations to remote sites are necessary. Longitudinal and/or lateral connectivity is thus crucial, as is instream and riparian cover (C o p p & Bennets 1996), to the continued existence of barbel populations (Baras et al. 1994, Lucas & Batley 1996, Lusk 1996, Lucas & Frear 1997, Schmutz & Jungwirth 1999). Acknowledgements We much appreciate valuable comments to the earlier version of manuscript by Dr. Juraj H o lãík and Gordon H. C op p as well the field assistance by Mr. Jan K ocián and Mr. Josef Melkus. This research was funded through grants No 206/01/0586 by the Grant Agency of the Czech Republic and Nos AV and KSK by the Grant Agency of the Academy of Sciences of the Czech Republic. LITERATURE BARAS, E., 1994: Constraints imposed by high-densities on behavioural spawning strategies in the barbel, Barbus barbus. Folia Zool., 43 (3): BARAS, E., 1995: Seasonal activities of Barbus barbus effect of temperature on time-budgeting. J.Fish.Biol., 46 (5): BARAS, E., 1997: Environmental determinants of residence area selection by Barbus barbus in the River Ourthe. Aquatic Living Resources, 10 (4): BARAS, E., LAMBERT, H. & PHILIPPART, I.C., 1994: A comprehensive assessment of the failure of Barbus barbus spawning migrations through a fish pass in the canalized river Meuse (Belgium). Aquatic Living Resources, 7 (3): COPP, G.H. & BENNETTS, T.A., 1996: Short-term effects of removing riparian and instream cover on barbel (Barbus barbus) and other fish populations in a stretch of English chalk river. Folia Zool., 45 (3): GERKING, S.D., 1953: Evidence for the concepts of home range and territory in stream fishes. Ecology, 34: GUNNING, G.E., 1963: The concepts of home range and homing in stream fishes. Ergebnisse der Biologie, 26: HANCOCK, R.S., JONES, J.W. & SHAW, R., 1976: A preliminary report on the spawning behaviour and the nature of sexual selection in the barbel, Barbus barbus (L.). J.Fish.Biol., 9: HANSKI, I., 1996: Metapopulation ecology. In: Rhodes, O.E. Jr., Chesser, R.K. & Smith, M.H. (eds): Population dynamics in ecological space and time. University of Chicago Press, Chicago: HANSKI, I. & SIMBERLOFF, D., 1997: The metapopulation approach, its history, conceptual domain, and application to conservation. In: Hanski, I. & Gilpin, M. (eds): Metapopulation biology: ecology, genetics and evolution. Academic Press, New York: HARRISON, S., 1998: Do taxa persist as metapopulations in evolutionary time? In: McKinney, M.L. & Drake, J. A. (eds), Biodiversity dynamics: Turnover of populations, taxa, and communities. Columbia University Press, New York: HUET, M., 1949: Aperçu de la relation entre la pente et les populations piscicoles des eaux courantes. Schweitz. Z. Hydrol., 11: HUNT, P.C. & JONES, J.W., 1974: A population study of Barbus barbus (L.) in the River Severn, England: II. Movements. J.Fish.Biol., 6: LUCAS, M.C. & BATLEY, E., 1996: Seasonal movements and behaviour of adult barbel Barbus barbus, a riverine cyprinid fish: Implications for river management. Journal of Applied Ecology, 33 (6):

66 LUCAS, M.C. & FREAR, P.A., 1997: Effects of a flow-gauging weir on the migratory behaviour of adult barbel, a riverine cyprinid. J.Fish.Biol., 50: LUSK, S., 1996: Development and status of populations of Barbus barbus in the waters of the Czech Republic. Folia Zool., 45 (Suppl. 1): NOVOTN, V. & PUNâOCHÁ, P. (eds), 1996: Die Fischfauna der Elbe. International Commission zum Schutz der Elbe, Magdeburg, pp Tables. PELZ, G.R. & KÄSTLE, A., 1989: Ortsbewegungen der Barbe Barbus barbus (L.) radiotelemetrische Standortbestimmungen in der Nidda (Frankfurt/Main). Fischökologie, 1/2: PE ÁZ, M., BARU, V. & PROKE, M., 1999: Changes in the structure of fish assemblages in a river used for energy production. Regul. Rivers: Res. Mgmt., 15: PERSAT, H. & CHESSEL, D., 1989: Typologie de distributions en classes de taille: intéret dans l étude des populations de poissons et d invertébrés. Acta Oecologia, 10 (2): PONCIN, P., PHILIPPART, J.C. & RUWET, J.C., 1996: Territorial and non-territorial behaviour in the bream. J.Fish.Biol., 49 (4): SCHMUTZ, S. & JUNGWIRTH, M., 1999: Fish as indicators of large river connectivity: the Danube and its tributaries. Archiv f. Hydrobiologie, 3 (Suppl. 115): SOKAL, R.R. & ROHLF, F.J., 1981: Biometry. W.H. Freeman and Company, 859 pp. STEINMANN, P., KOCH, W. & SCHEURING, L., 1937: Die Wanderungen unserer Süsswasserfische dargestellt auf Grund von Markierungsversuchen. Z. f. Fischerei, 35: (Viz Heuschman: Handbuch d. Binnenfischerei Mitteleuropas). LECHTOVÁ, V., LECHTA, V., LUSKOVÁ, V. & LUSK, S., 1998: Genetic variability of common barbel, Barbus barbus populations in the Czech Republic. Folia Zool., 47 (Suppl. 1):

67 Folia Zool. 51(1): (2002) Morphometric study of the Iberian Aphanius (Actinopterygii, Cyprinodontiformes), with description of a new species Ignacio DOADRIO 1, José A. CARMONA 1 and Carlos FERNÁNDEZ-DELGADO 2 1 National Museum of Natural Sciences, Department of Biodiversity, c/josé Gutiérrez Abascal 2, Madrid, Spain; mcnd147@mncn.csic.es 2 Faculty of Sciences, Department of Animal Biology, University of Córdoba, Av. San Alberto Magno s/n, Córdoba, Spain Received 30 June 2000; Accepted 13 November 2001 A b s t r a c t. Morphometric variation in Aphanius iberus was analysed to demostrate the remarkable genetic divergence between Mediterranean and Atlantic population of Iberia. Four discrete morphotypes in males and three in females were distinguished. Morphometric data discriminated the Atlantic and Mediterranean populations, but revealed the Villena population as the morphologically most differentiated. Atlantic populations were described as a new species, Aphanius baeticus sp. nov., which differs from A. iberus in the overall shape, coloration pattern and number of branched rays on the dorsal and anal fins. The Villena population was retained in A. iberus because, despite of its morphological differentiation, show high genetic introgression at the nuclear genome with neighbouring populations. The range of A. baeticus sp. nov. is restricted to the eight localities of the Atlantic slope of the Iberian Peninsula. This new species should be considered Critically Endangered (CR) according to the IUCN Red List Categories. Key words: killifishes, taxonomy, Lebias Introduction The Aphanius Nardo, 1827 cyprinodontid killifishes are likely descendants of the Tethys fauna inhabiting coastal lagoons, marshes and salty rivers around the circum-mediterranean area as well as the Arabian Peninsula as far as Iran and Pakistan (V illwock 1999). As a consequence of the paleogeographic history of the region, which supported dramatical changes during the Cenozoic period, species of Aphanius show an irregular distribution pattern with isolated populations. Hence, the genus Aphanius in the eastern Mediterranean basin and adjacent regions (the Middle and the Near East) is highly diverse, with at least ten different species (H u b e r 1996). In contrast, in the western Mediterranean area, only three species are known Aphanius apodus (Gervais, 1853), A. fasciatus (Valenciennes in Humboldt & Valenciennes, 1821) and A. iberus (Valenciennes in Cuvier & Valenciennes, 1846) (Villwock 1999, Doadrio 1994). A. iberus has been reported in a wide area along the Mediterranean coast of southern France (Arnoult 1957), and Spain (Paracuellos & Nevado 1994, Fernández-Pedrosa et al. 1995), some localities of the Atlantic Iberian coast (Hernando 1975) and in North Africa in Algeria and Morocco (P ellegrin 1921, Villwock & Scholl 1982). A. iberus now appears to be extinct in France (Allardi & K e i t h 1991), and the African populations are both morphologically and genetically different from those of Iberia (V illwock & Scholl 1982) and might deserve distinct taxonomic recognition (Doadrio et al. 1996). 67

68 In the course of recent genetic studies of different populations of A. iberus using mitochondrial DNA and allozyme markers, the Atlantic populations were found to be genetically differentiated from the Mediterranean ones (D oadrio et al. 1996, Perdices et al. 2001). Similarly, restriction fragment length polymorphism analysis of mitochondrial DNA markers showed that the population of the endorrheic Villena Lagoon is highly differenciated within the Mediterranean populations (Fernández-Pedrosa et al. 1995). Since diagnostic loci and high genetic distances discriminate the Atlantic and Mediterranean populations, morphometric variation analysis is needed to characterise these two discrete gene pools. In this paper, we conduct a morphometric study of several Atlantic and Mediterranean populations to characterise the genetically differentiated populations of A. iberus. Both genetic and morphological data are used in order to revise the taxonomic status of the populations of A. iberus. Materials and Methods The specimens originated from varios locations in Spain and one location in Greece (Table 1). The samples housed in the Museo Nacional de Ciencias Naturales of Madrid (listed in the Appendix) were also revised as a comparative material. Twenty-two morphometric variables were measured. All measurements are in millimetres and were log-transformed for morphometric analysis. Only adult specimens of the 0+ and 1+ age groups were used for comparative morphometric analysis. The following abreviations were used for morphometric and meristic characters (Tables 2, 3 and 6): SL, standard length; HL, head length; HD, head depth; PrOL, preorbital length; ED, eye diameter; PrDD, predorsal distance; PrPD, prepectoral distance; PrVD, preventral distance; PrAD, preanal distance; CPL, caudal peduncle length; APL, anal peduncle length; PVL, pectoral-ventral length; VAL, ventral-anal length; DFL, dorsal fin length; DFH, dorsal fin height; PFL, pectoral fin length; VFL, ventral fin length; AFL, anal fin length; AFH, anal fin height; CFL, caudal fin length; BD, body depth; BLD, body least depth; D, dorsal fin rays; A, anal fin rays; P, branched pectoral fin rays; V, branched ventral fin rays; C, branched caudal fin rays; LLS, lateral line scale rows; Abd. vert., abdominal vertebrae; Cau. vert., caudal vertebrae; GR, gill rakers. Only branched caudal, pectoral and ventral fins rays were counted. The osteological characters were studied Table 1. Localities and sample size of Aphanius populations used for morphological, osteological and meristic analysis. Aphanius iberus morphometry osteology & meristics Locality males females males females Salinas de San Pedro del Pinatar, Murcia, Spain Albufera de Valencia, Valencia, Spain River Adra, Adra, Almeria, Spain Villena Lagoon, Villena, Alicante, Spain River Salado, Lebrija, Sevilla, Spain Iro River, Chiclana de la Frontera, Cádiz, Spain River San Pedro, Paterna de la Ribera, Cádiz, Spain Aphanius fasciatus males females males females Messolongi Marshes, Greece

69 Appendix Aphanius iberus. France. MNHN Lectotype, MNHN Syntypes. Spain. MNCN , , , , , , Albufera de Valencia, Valencia. MNCN , , 28787, Albufera de Valencia, Silla, Valencia. MNCN , , 15343, Prat de Cabanes, Cabanes, Castellón. MNCN , Albufera de Torreblanca, Torreblanca, Castellón. MNCN , , , , , , , , , Salinas de San Pedro del Pinatar, Murcia. MNCN Salinas de Marchamalo, Murcia. MNCN , Segura River, Guardamar de Segura, Alicante. MNCN , Salinas de Santa Pola, Alicante. MNCN Albatera, Alicante. MNCN , La Cava, Tortosa, Tarragona. MNCN Pantano del Hondo, Elche, Alicante. MNCN Delta del Ebro, La Tancada, Tarragona. MNCN Aiguamolls del Empordá, Ampurias, Gerona. MNCN , Albuixec, Valencia. MNCN Balsas de riego, Adra, Almería. MNCN Adra River, Adra, Almeria. Aphanius baeticus sp. nov. MNCN Sevilla. from cleared and stained specimens (W a s s e r s u g 1976). Caudal vertebres nomenclature follows Coelho et al. (1998). To test for sexual dimorphism and population variation, we used analysis of variance (ANOVA) for unbalanced design to test for differences in means (for groups or variables) because it is more statistically powerful than the simple t test, permitting the test of each variable whilst controlling for all others as well as detection of interaction effects between variables. A multivariate analysis of variance (MANOVA) was performed to test for two dependent variables at a time (i.e. between species and sexes) Differences in body shape among populations were analysed using Principal Component Analysis (PCA). Burnaby s method was used to correct for size effects (B u r n a b y 1966) using an iml procedure written in SAS-pc by Dr. L.F. Marcus (corrected from version in Appendix in Rholf & Bookstein 1987) (see Bekele et al. 1993). Canonical variate analyses (CVA) and Mahalanobis distances between group centroids were used to summarise differences among populations, which were clustered in a phenogram using the Unweighted Pair Group Method (UPGMA). All the computations were performed using SAS 11, (SAS Institute Inc. 1996) and NTSYS-pc version 1.8 (Rohlf 1993). We maintain the use of Aphanius against Lebias until the International Commission on Zoological Nomenclature publishes the decision on the submitted application referring to Lazara s Lebias type species (see K ottelat 1997). Institutional acronyms: MNHN Museum National d Histoire Naturelle, Paris. MNCN Museo Nacional de Ciencias Naturales, Madrid. UCOBA Departamento de Biología Animal, Universidad de Córdoba, Córdoba. Results Significant differences (ANOVA) among populations for the morphometric variables were mainly due to differences in SL (Table 2), with sexual dimorphism statistically significant in practically all characters. These results suggest that females are larger than males and show overall smaller ventral fins. There was interaction between sexes and populations for most of the variables, which should be due to size differences between sexes for each population. Significant differences (MANOVA) were found in all characters among populations and between sexes and between populations by sexes, justifying the separate treatment of males and females. 69

70 Table 2. Two-way analysis of variance testing for sexual dimorphism, population variation and their interaction (Pop*Sex). Mean squares from SAS type III glm procedure for unbalanced designs. Except ventral fin length for sexual dimorphism and anal fin height for interaction between sexes and species, all variables were significant (P< 0.01). Variables are described in Methods. (n.s.= not significant). Variable Population Sex Pop*Sex SL HL HH PrOL ED PrDD PrPD PrVD PrAD CPL APL PVL VAL DFL DFH PFL VFL n.s AFL AFH n.s. CFL BD BLD Table 3. Eigenvectors and eigenvalues for the first four principal components for 22 variables for males and females. Variable codes are given in the Method section. Males Females Eigenvectors I II III IV I II III IV SL CFL HL HH PROL ED PRDD PRPD PRVD PRAD CPL APL PVL VAL DFL DFH PFL VFL AFL AFH BD BLD Eigenvalue (53.35) (68.24) (77.30) (82.75) (71.09) (77.74) (83.69) (87.29) 70

71 Fig. 1. A) Plots of first two principal components for 22 Burnaby corrected variables (males-left and femalesright). B) Plots from canonical variates analysed (males-left and females-right). River Salado, Lebrija, Sevilla. River Iro, Chiclana de la Frontera, Cádiz. River San Pedro, Paterna de la Ribera, Cádiz. River Adra, Adra, Almeria. Albufera de Valencia, Valencia. Salinas de San Pedro del Pinatar, Mar Menor, Murcia. Villena, Alicante, and A. fasciatus Messolongi Marshes, Greece. The first four principal components accounted for 88.5% of variance in males and 92.6% in females. The areas of the scores of each component diferred significantly (MANOVA) among populations for these first four components. Normalized eigenvectors on the first principal components showed close values for both sexes and with the same sign, suggesting an influence of body size. Variation in body size could be due to local adaptation to different habitats like marshes, irrigation chanels, lagoons or salty springs. A PCA with 22 Burnaby s corrected variables was carried out to remove size effect. The first two principal components accounted for 68.2% of variation in males and 77.7% in females (Table 3). For both sexes, the highest eigenvectors were dorsal fin length and caudal peduncle length with five different groups of males (Fig. 1). This five groups represent: Adra population, Villena population, the remaining Mediterranean populations, Atlantic populations and A. fasciatus. In females, this structure is less evident and the Adra population overlaps with Atlantic populations. Significant differences in morphology were found between the populations (Fig. 1), but differentiation was more evident in males. Mahalanobis distances D2 and its posterior cluster (Fig. 2) were congruent with previous genetic studies (F e r n á n d e z- P e d r o s a et al. 1995, Doadrio et al. 1996), although Adra population was not studied in there. 71

72 Fig. 2. UPGMA clusters on Mahalanobis distances between groups of centroids. Top Tree for males: Discussion Taxonomic considerations A. baeticus sp. nov. A. iberus. Adra population. A. fasciatus. Villena population. Bottom Tree for females: A. baeticus sp. nov. and Adra population. A. iberus. Villena population. A. fasciatus. Morphological analyses revealed the existence of four morphotypes in the Iberian populations. Eventhough morphotypes have previously been used as the base of species description, many species show a high degree of phenotypic plasticity and therefore it is necesary to corroborate the presence of a well-differentiated gene pool for taxonomical purposes. Hence, we consider the presence of diagnostic loci as a strong criterion for species recognition (B eerli et al. 1995, Mendoza-Qijano et al. 1998) since, in a concordant geographic framework, we can distinguish populations that may be isolated ephemerally only from those that have been isolated long enough to differentiate unlinked characters (B aum & Donoghue 1995). Therefore, the two morphotypes corresponding to the Villena and Adra populations can not be considered as new species due to the lack of diagnostic loci. The Adra population showed a different morphotype in males and currently is isolated from other Aphanius populations by at least for 350 kilometers of coastal line, nevertheless, its low genetic divergence seems to indicate a recent speciation of this populations. On the contrary, the drainage channel in Villena Lagoon connected this population with the neighbouring Mediterranean populations favouring a nuclear DNA introgression in this century. The introgression process and the persistent small population size could have propitiated the high morphological divergence of this population. 72

73 The Atlantic populations showed such genetic and morphological divergence with respect to other populations that merit to be distinguished at the species level. No available name can be applied to the Atlantic populations, and therefore, we choose to designate these populations A. baeticus sp. nov. Hence, A. iberus is maintained for the Mediterranean populations. Aphanius baeticus sp. nov. is characterized by morphological characters that refer to mensural and meristic characters (Tables 4 and 5). Moreover, using molecular markers A. baeticus sp. nov. showed diagnostic loci in different enzymatic systems (Doadrio et al. 1996), and highly differentiated haplotypes of mitochondrial DNA (P e r d i c e s et al. 2001). An interesting finding was that males showed a higher number of differentiated morphotypes than females, indicating that sexual dimorphism varies from one species to another. Table 4. Frequency distribution of Peduncle indeces (CPL/BLD) in Aphanius from Iberia. Caudal peduncle index Species male A. iberus (Villena) A. iberus A. baeticus sp. nov females A. iberus (Villena) A. iberus A. baeticus sp. nov Table 5. Summary of diagnostic characters of Iberian Aphanius. Characters A. iberus A. baeticus sp. nov. Number of branched rays on dorsal fin 8-9(10) 8(9) Number of branched rays on anal fin 8-9 (9)10(11) Preorbital distance long short Coloration pattern in males narrow silver wide silver transversal bars transversal bars Coloration pattern in females numerous and few large black small black spotts spotts in flanks Unique alleles in allozymes IDHP-1*100 IDHP-1*73 IDHP-2*100 IDHP-2*110 smdh-2*100 smdh-2*110 The high degree of differentiation among populations of A. iberus is similar to that detected in the closely related species A. fasciatus. In the latter species, population differentiation corresponds to its naturally fragmented distribution and to a restricted gene flow among populations, which favours a genetic structure through isolation by distance (Maltagliati 1998,1999, Maltagliati & C amilli 2000). The main divergence between the Atlantic and Mediterranean populations of the Iberian Aphanius probably occurred after the Mediterranean dessication and subsequent interruption of water communication during the Lower Pliocene (D oadrio et al. 1996). The same pattern of speciation, based on a paleogeographical event, was proposed for the Anatolian species Aphanius asquamatus (Sözer, 1942) which probably diverged from the central 73

74 Anatolian cyprinodonts during the Upper Miocene-Lower Pliocene (see V illwock 1999). During this period, Anatolia was raised several meters above sea level whilst the entire region was separated in two large systems. Afterwards, the central Anatolian cyprinodonts gradually split during the Pliocene and Pleistocene periods, rendering distinguishable species, such as A. chantrei Gaillard, 1895 and A. anatoliae Leidenfrost, 1912, besides many subspecies and micro populations (V illwock 1963). Although Aphanius are euryhaline and a coastal route of colonization should be used, the strait of Gibraltar appears to act as an important barrier. Thus, for instance, the exotic species Fundulus heteroclitus, which quickly colonized the southern Atlantic area of the Iberian Peninsula, has not been able to colonize the Mediterranean basin through the strait of Gibraltar. Natural geographic barriers and life-history traits, which are similar to those observed in A. fasciatus e. g. benthic eggs and absence of larval stages (Maltagliati & C a m i l l i 2000), should prevent gene flow between Atlantic and Mediterranean populations. Aphanius iberus is critically endangered (Blanco & González 1992, Maitland& Crivelli 1996, Council Directive 92/43/ECC). The distribution range of A. baeticus sp. nov. is relatively narrow and pollution, exotic species and hydraulic alterations render the species critically endangered too. Description of new species Aphanius baeticus sp. nov. Diagnosis. Differs from all other known species of Aphanius in the combination of: ten branched rays in the anal fin, eight branched rays in the dorsal fin, long and low caudal peduncle, short preorbital length, males show light wide transversal bars and females large black spots. Presence of unique alleles at IDHP-1*, IDHP-2* and smdh-2* (Doadrio et al. 1996). Holotype. (Fig. 1). MNCN Male, 31.3 mm SL. Salado River, Lebrija, Sevilla. Guadalquivir basin, Spain. 20 m altitude. Leg. C. Fernández-Delgado. 6. VI Paratypes. MNCN , 29 individuals. Salado River, Lebrija, Sevilla. Guadalquivir basin, Spain. Leg. C. Fernández-Delgado. 6.VI MNCN , 45 individuals. Salado River, Lebrija, Sevilla. Guadalquivir basin, Spain. Leg. C. Fernández- Delgado. 17.VII MNCN , 34 individuals. Salado River, Lebrija, Sevilla. Guadalquivir basin, Spain. Leg. C. García Utrilla. 25.I UCOBA , 55 individuals. Salado River, Lebrija, Lebrija, Sevilla, Spain. Guadalquivir basin. Leg. Fernández-Delgado. MNCN , 65 individuals. Hondón Lagoon, Parque Nacional de Doñana, Huelva, Spain. Leg. L. Dominguez. 5.XI MNCN , 12 individuals. Salinas de San Carlos, Sanlúcar de Barrameda, Cádiz, Spain. Leg. J.A. Hernando. 15.VI MNCN , 28 individuals. Salado de San Pedro River, Paterna de la Ribera, Cádiz, Spain. Leg. C. Fernández-Delgado. UCOBA 1-96, 96 individuals. Salado de San Pedro River, Paterna de la Ribera, Cádiz, Spain. Leg. C. Fernández-Delgado. UCOBA , 30 individuals. Roche River, Roche, Cádiz, Spain. Leg. C. Fernández-Delgado. MNCN , 35 individuals. Iro River, Chiclana de la Frontera, Cádiz, Spain. Leg. Fernández-Delgado. 21.IX UCOBA , 50 individuals. Iro River, Chiclana de la Frontera, Cádiz, Spain. Leg. C. Fernández-Delgado. Description. D I-II 8(9), A I-(9)10(11), P (8)9-10(11), V (3)4-5, C (13) LLS (24)25(26), Abd. vert. (10)11, Cau. vert. (14,15)16(17), G. R. 8-10(11). The body form of 74

75 Fig. 3. Aphanius baeticus sp. nov. Top: Holotype, male, MNCN Bottom. Paratype, female, MNCN Salado River, Lebrija, Sevilla. Spain. Scale 1 cm. Drawing by M. Merino. Aphanius baeticus sp. nov. (Fig. 3, Table 6) is characterised by a deep but more elongated body than in A. iberus. Minimal body depth is (1.7) times in the caudal peduncle in females and (1.8) in males. Minimum body depth (1.2) times in the anal peduncle in females and (1.2) in males. Short head comprising (3.5) times the standard length in females and (3.4) in males. Preorbital length is small, reaching (1.7) times the eye diameter in females and (1.8) in males. Insertion of the ventral fin is anterior to dorsal fin origin. Preventral length is (1.1) times the predorsal length in females and (1.2) in males. Fin size moderately large. Robust premaxille (maximun length / maximum high = (2.3)) with (7)8-10(11) teeth. Dentary with (7)8-11(12,13) teeth. Pigmentation pattern. Males. Light wide vertical bars along the body side, being more evident in juveniles (Fig. 4). The number of bars on the flanks is typically and 4-5 on the caudal fin. Four rows of dark spots extend onto the dorsal and anal fins. Pectoral and ventral fins are ligth orange. Females. Light brown with few and large dark spots on the body flanks. Two rows of spots are shown, one of them along the lateral line and another in the ventral side. Frequently, the spots are arranged as bars on the middle of the body. An additional row of spots is shown on the caudal peduncle. Caudal fin has 4-6 dark vertical bars. Sexual dimorphism. Males are smaller overall than females with proportionally larger ventral fins. Distribution. A. baeticus sp. nov. occurs in the lower reaches of the River Guadalquivir and streams located on the southern Atlantic slope of Spain (Fig. 5). The species is currently found in the following areas: River Santiago, El Palmar de Troya, Sevilla. River Salado, Lebrija, Sevilla. Parque Nacional de Doñana, Huelva. Salinas de Sanlúcar de Barrameda, Sanlúcar de Barrameda, Cádiz. River Salado de San Pedro, Paterna de la Ribera, Cádiz. 75

76 Table 6. Statistical parameters for the morphometric characters of A. baeticus sp. nov. Second row (SL/) represents standard length / each variable. Variables are described in Methods. (SD = Standard deviation). Aphanius baeticus sp. nov. males (n = 44) females (n = 36) Variable range mean SD range mean SD SL CFL SL/ HL SL/ HH SL/ PROL SL/ ED SL/ PRDD SL/ PRPD SL/ PRVD SL/ PRAD SL/ CPL SL/ APL SL/ PVL SL/ VAL SL/ DFL SL/ DFH SL/ PFL SL/ VFL SL/ AFL SL/ AFH SL/ BD SL/ BLD SL/ River Iro, Chiclana de la Frontera, Cádiz. River Roche, Roche, Cádiz. River Salado de Conil, Conil, Cádiz (Moreno-Amich et al. 1999). Etymology. The species name baeticus derives from the roman name of the Guadalquivir River valley. The range of the species overlaps almost exactly with this area. Remarks. A. baeticus sp. nov. has been found in three different types of habitat: lagoons (one population), tidal channels (one population) and small streams (six populations). The 76

77 A B Fig. 4. Vertical bar and shape patterns of juvenile males in Iberian Aphanius. (A) A. baeticus sp. nov. (B) A. iberus. Fig. 5. Geographical distribution of Iberian species of the genus Aphanius. lagoon has permanent, very shallow (30 cm mean deep), freshwater with a silty clay bottom and scarce aquatic vegetation (Juncus sp., Phragmithes sp.). The most common habitats of A. baeticus sp. nov. are first order streams (2-3 m width, 0.5 m deep), close to the sea with a silty clay bottom and scarce bank vegetation dominated by Salicornia sp., Tamarix sp. and/or Juncus sp. (F ernández-delgado et al. 1999). Water salinity is highly variable, two populations are located in freshwaters and the rest are living in hyperhaline 77

78 water (reaching up to 110%). Aquatic vegetation is dominated by Cladophora sp. In the hyperhaline small streams, A. baeticus sp. nov. does not coexist with any other fish species and occurs in high densities. At the remaining locations, the species coexists with one or another of the introduced fish species, the mosquito-fish (Gambusia holbrooki Baird et Girard, 1853) and the salt-marsh killifish (Fundulus heteroclitus L.), when this occurs, the density of Aphanius is very low (Fernández-Delgado et al. 1999). Aphanius baeticus sp. nov. shows a very short life-span, with a winter age-structure of 95.9% 0+, 3.9% 1+ and 0.2% 2+ fish. The growth period is from April to September and the annulus on the scales is formed in April. The species spawns intermittently during the reproductive period. From April to July, 1+ specimens (7 to 12 months old) reproduce. Their offspring (0+ group; 3-4 months old) reproduce from July to September. Very few of the 1+ group specimens survive to spawn the following year. Males mature before females, but females are larger than males. The overall sex ratio does not differ from 1:1, but males outnumber females considerably during the spawning period (F e r n á n d e z- D e l g a d o et al. 1988). Conservation. Four of the eight populations reported are in a critical situation. This is due to the fact that the species occurs in habitats with G. holbrooki and/or F. heteroclitus. Apparently, the other four populations are well protected, occurring in high densities. The most serious threat to this new species is that all the populations are isolated, without any connection, and hence genetic flow between them. In our opinion, given the restricted number of known populations and their actual status, this species should be catalogued as critically endangered (CR) according to the IUCN Red List Categoories. Acknowledgements Part of this study was financially sponsored by the Consejería de Medio Ambiente (Autonomous Government of Andalusia) under the Plan de recuperación de las poblaciones del género Lebias en Andalucía and DGES project PB to I. Doadrio. For the assistance with the field collections, we thank L. Dominguez, P. Garzón, F. Gomez-Caruana, A. Machordom, R. Moreno-Amich, J. C. Nevado, S. Peiró, M. Planelles, P. Risueñ o and A. Sostoa. We are grateful to J. Castroviejo, T. Chapa, A. Garcia-Valdecasas, M. Kottelat, A. Perdices, A. Machordom, M. Merino, S. Reig, R. Márquez and D. Buckley for suggestions and improve the English. We are also grateful to Prof. Guy Duhamel (MNHN-Paris) for facilities in the designation of a lectotype of A. iberus from the syntype series. Special thanks is to Palmira G u a r n i z o for her invaluable field and laboratory assistance. LITERATURE ALLARDI, J. & KEITH, P., 1991: Atlas preliminaire des poissons d eau douce de France. Muséum National d Histoire Naturelle, Paris, 232 pp. ARNOULT, J., 1957: Sur quelques poissons rares et peu connus dans les eaux douces de France. Bull. Museum Natl. d Histoire Nat., 6: BAUM, D. A. & DONOGHUE, M. J., 1995: Chosing among alternative phylogenetic species concepts. System. Biol., 20: BEERLI, P., HOTZ, H. & UZZELL, T., 1995: Geologically dated sea barriers calibrate a protein clock for Aegean water frogs. Evolution, 50: BEKELE, A., CAPANNA, E., CORTI, M., MARCUS, L. F. & SCHLITTER, D. A., 1993: Systematics and geographic variation of Ethiopian Arvicanthis (Rodentia, Muridae). J. Zool. London, 230: BLANCO, J.C. & GONZÁLEZ, J.L., 1992: Libro rojo de los vertebrados de España. ICONA, colección técnica, Madrid. BURNABY, T. P., 1966: Growth-invariant discriminant functions and generalized distances. Biometrics, 22: COELHO, M. M., BOGUTSKAYA, N. G., RODRIGUES, J. A. & COLLARES-PEREIRA, M. J., 1998: Leuciscus torgalensis and Leuciscus aradensis, two new cyprinids for Portuguese fresh waters. J. Fish Biol., 5:

79 DOADRIO, I., 1994: Freshwater fish fauna of North Africa and its biogeography. Konink. Mus. Midd. Afri. Terv. Belg. Annal. Zool. Wetenschap., 275: DOADRIO, I., PERDICES, A. & MACHORDOM, A., 1996: Allozyme variation of the endangered killifish Aphanius iberus and its application to conservation. Environ. Biol. Fish., 45: FERNÁNDEZ-DELGADO, C., HERNANDO, J. A., HERRERA, M. & BELLIDO, M., 1988: Age, growth and reproduction of Aphanius iberus (Cuv. & Val., 1846) in the lower reaches of the Guadalquivir River (southwest Spain). Freshwat. Biol., 20: FERNÁNDEZ-DELGADO, C., TORRALVA, M. M., OLIVA, F. & PINTOS, R., 1999: Caracterización ecológica del hábitat del fartet (Lebias ibera Valenciennes, 1846) en una pequeña cuenca hidrográfica del bajo Guadalquivir. In: Generalitat Valenciana (ed.), Peces cyprinodontidos ibéricos: Fartet y Samaruc. Consellería de Medio Ambiente, Valencia: FERNANDEZ-PEDROSA, V., GONZÁLEZ, A., PLANELLES, M., MOYA, A. & LATORRE, A., 1995: Mitochondrial DNA variability in three Mediterranean populations of Aphanius iberus. Biol. Conserv., 72: HERNANDO, J. A., 1975: Notas sobre la distribución de los peces fluviales en el Suroeste de España. Doñ. Acta Verteb., 2: HUBER, J. H., 1996: Killi-Data Liste actualisée des noms taxonomiques, des localites de pêche et des références bibliographiques des poissons cyprinodontes ovipares (Atherinomorpha, Pisces). Société Francais d Ichtyologie, Paris, 399 pp. KOTTELAT, M., 1997: European freshwater fishes. An heuristic checklist of the freshwater fishes of Europe (exclusive of former USSR), with an introduction for non-systematics and comments on nomenclature and conservation. Biologia (Bratislava), 52 (suppl. 5): MAITLAND, P. S. & CRIVELLI, A. J., 1996: Conservation of freshwater fish. Conserv. Mediter. Wetl., 7: MALTAGLIATI, F., 1998: A preliminary investigation of allozyme genetic variation and population geographical structure in Aphanius fasciatus from Italian brackish-water habitats. J. Fish Biol., 52: MALTAGLIATI, F., 1999: Genetic divergence in natural populations of the Mediterranean brackish-water killifish Aphanius fasciatus. Mar. Ecol. Prog. Ser., 179: MALTAGLIATI, F. & CAMILLI, L., 2000: Temporal genetic variation in a population of Aphanius fasciatus (Cyprinodontidae) from a brackish-water habitat at Elba Island (Italy). Environ. Biol. Fish., 57: MENDOZA-QIJANO, F., FLORES-VILLELA, O. & SITES, J. W., 1998: Genetic variation, species status and phylogenetic relationships in rose-belied lizards (Variabilis group) of the genus Sceloporus (Squamata: Phrynosomatidae). Copeia, 2: MORENO-AMICH, R., PLANELLES-GOMIS, M., FERNÁNDEZ-DELGADO, C. & GARCÍA-BERTHOU, E., 1999: Distribución geográfica de los ciprinodontiformes en la Península Ibérica. In: Generalitat Valenciana (ed.), Peces cyprinodontidos ibéricos: Fartet y Samaruc. Consellería de Medio Ambiente, Valencia: PARACUELLOS, M. & NEVADO, J. C., 1994: Localización del Fartet, Aphanius iberus, en la cuenca del río Adra (Almeria, sudeste ibérico). Doñ. Acta Verteb., 21 (2): PELLEGRIN, J., 1921: Les poissons des eaux douces de l Afrique du Nord francaise, Moroc, Algérie, Tunisie, Sahara. Mem. Soc. Sci. Nat. Maroc, 1: PERDICES, A., CARMONA. J. A. & DOADRIO, I., 2001: Nuclear and mitochondrial data reveal high genetic divergence among Atlantic and Mediterranean populations of the Iberian killifish Aphanius iberus (Teleostei: Cyprinodontidae) Heredity, 87: ROHLF, F. J., 1993: NTSYS-pc. Numerical taxonomy and multivariate analysis system, Version Setauket, New York. ROHLF, F. J. & BOOKSTEIN, F. L., 1987: A comment on shearing as a method for size correction. Sys. Zool., 36: SAS Institute Inc., 1996: SAS, User s guide statistics, Version Ed. Cary, North Carolina, SAS Institute, USA. VILLWOCK, W., 1963: Genetische Analyse des Merkmals Beschuppung bei anatolischen Zahnkarpfen (Pisces, Cyprinodontidae) im Auflöserversuch. Zool. Anz., 170: VILLWOCK, W., 1999: Biogeography of the cyprinodontiform fishes (Teleostei: Cyprinodontidae) of the mediterranean region. In: Generalitat Valenciana (ed.), Peces cyprinodontidos ibéricos: Fartet y Samaruc. Consellería de Medio Ambiente, Valencia: VILLWOCK, W. & SCHOLL, A., 1982: Ergänzende Mittteilungen über Aphanius aus der Oase Azraq/Jordanien sowie Betrachtungen zum taxonomischen Status eines neuen A. iberus (Cyprinodontidae: Pisces) aus dem Oued Zousfana, Igli/Nordwest-Algerien. Mitteil. Hamburg. Zool. Mus. Zool. Inst., 79: WASSERSUG, R. J., 1976: A procedure for differential staining of cartilage and bone in whole formalin-fixed vertebrates. Stain Technol, 51:

80 Folia Zool. 51(2): (2002) Genetic variation in house mice (Mus, Muridae, Rodentia) from the Czech and Slovak Republics Pavel MUNCLINGER 1, Eva BOÎÍKOVÁ 1, Monika UGERKOVÁ 2,3, Jaroslav PIÁLEK 4 and Milo MACHOLÁN 2 1 Biodiversity Research Group, Department of Zoology, Faculty of Science, Charles University, Viniãná 7, Praha 2, Czech Republic; muncling@natur.cuni.cz 2 Institute of Animal Physiology and Genetics, AS CR, Vevefií 97, Brno, Czech Republic; macholan@iach.cz 3 Department of Zoology and Ecology, Faculty of Sciences, Masaryk University, Kotláfiská 2, Brno, Czech Republic; e- mail: sugerkova@centrum.cz 4 Institute of Vertebrate Biology, AS CR, Studenec 122, Konû ín, Czech Republic; jpialek@brno.cas.cz Abstract. Genetic variation among populations of commensal house mice was studied across the territories of the Czech and Slovak Republics and in some adjacent areas of Germany. We used six diagnostic allozyme loci (Es-2, Gpd-1, Idh-1, Mpi, Np, Sod-1) and the following molecular markers: B1 insertion in the Btk gene (X chromosome), Zfy2 18-bp deletion (Y chromosome), BamH I restriction site in the mt-nd1 gene (mtdna) and Hba-4ps 16-bp insertion (diagnosing the presence of t haplotypes). In total, 544 individuals taken from 49 localities were examined. Almost the entire territories of the Czech Republic and Slovakia were found to be occupied by Mus musculus, the only exception being the westernmost parts of the Czech Republic, where M. musculus meets the range of M. domesticus and forms a narrow belt of hybrid populations. Despite this, domesticus-type alleles of some allozyme markers (notably Es-2) were also found at sites well within the range of M. musculus, either tens or hundreds of kilometres behind the hybrid zone. This provides evidence of either: (1) introgression of some markers into the species genome due to free gene flow through the zone, or (2) human-mediated long-distance migrations, or (3) incomplete lineage sorting. Conversely, variants of molecular markers typical for M. domesticus in Btk, Zfy2 and mt-nd1 were only found in the westernmost populations studied. t haplotypes were quite frequent in some populations, irrespective of whether M. domesticus, M. musculus or their hybrids, yet no t/t homozygotes were found. The mean frequency of t/+ heterozygotes found within the study populations was 13%. Key words: house mouse, Mus musculus, M. domesticus, genetic variation, allozymes, sex chromosomes, mtdna, t haplotype Introduction Notwithstanding taxonomic issues (e.g. A u f fray et al. (1990), Musser & Carleton (1993), Sage et al. (1993), D i n et al. (1996), Mitchell-Jones et al. (1999)) there is general agreement that two commensal house mice taxa occur in Europe (here considered as distinct species): Mus domesticus Schwarz et Schwarz, 1943 (Corbet 1988) in the west and south, and M. musculus L., 1758 in the north and east. Where their ranges abut, a narrow belt of hybrid populations is formed. This hybrid zone stretches across the Jutland Peninsula and from the Baltic coast of East Holstein, through central Europe and the Balkan Peninsula to the Bulgarian Black Sea coast (see Boursot et al. 1993, Sage et al. 1993). Some sections of the Mus hybrid zone have been studied thoroughly, e.g. in Denmark (Hunt & Selander 1973, V anlerberghe et al. 1986, 1988b, D o d et al. 1993), East Holstein (Prager et al. 1993, 1997), Bulgaria (Boursot et al. 1984, Vanlerberghe et al. 1986, 1988a) and southern Germany (S age et al. 1986b, 81

81 Tucker et al. 1992) (for mtdna see also S age et al. 1990, P rager et al. 1996). Despite this, information from a major portion of the zone is scarce, so the precise ranges of the two Mus taxa are not well understood. This is particularly apparent in Central Europe. Here, apart from data from a transect across the hybrid zone in southern Germany and Austria (Sage et al. 1986a,b, T ucker et al. 1992, Prager et al. 1996), mtdna data from Germany (Sage et al. 1990) and morphological data from Bavaria (Kraft 1984), little (at best) is known from the vast area of the former East Germany (DDR). In the Czech Republic (Czechia), the results of a pilot study on the identity of house mice from the western part of Czechia were presented by Macholán & Zima (1994), who provided evidence for the presence of M. domesticus in the westernmost parts of the country; this was later corroborated by morphological studies (L a z a r o v á 1999). So, despite growing data on the position and dynamics of the M. domesticus/m. musculus hybrid zone in Europe, its precise position in Czechia remains undetermined. This paper focuses on the following: (i) evaluation of the suitability of various molecular markers for diagnosis of the two taxa [We selected markers on sex chromosomes and a marker for mtdna as well, since this part of mouse genome has been shown to behave differently when compared with allozymes in some parts of the hybrid zone (see references in discussion)] and demonstration of their utility in surveys of the Czech part of the hybrid zone; (ii) to extend the study of Macholán & Zima (1994) to a larger area of former Czechoslovakia (i.e. the current Czech and Slovak republics) and adjacent areas of Germany, thus increasing both the number of animals investigated, and the markers used, to obtain more data on the distribution of the two taxa in this part of Europe; (iii) although t haplotypes were studied extensively in M. domesticus, only limited information is available on their frequency in M. musculus populations (see R uvinsky et al. 1991, and references therein) so we decided to type the mice studied for the presence of t haplotypes. Material and Methods Mice Depending on the marker used, between 215 and 544 mice were studied. These were collected from 49 sites scattered across Czechia, Slovakia and adjacent areas of NE Bavaria and S Thuringia. The list of localities and numbers of animals examined are given in Table 1 (see also Figs 1, 2 for site locations). Allozymes Altogether, 544 individuals were studied. Kidneys were dissected from freshly killed animals (in some cases, animals dead for a few hours were also successfully scored) and stored at -80 C until processed. Portions of tissue were crushed in homogenisation buffer (P a s t e u r et al. 1988) and centrifuged at 10,000 RPM for five minutes. Standard horizontal starch gel electrophoresis (Harris & Hopkinson 1976, Pasteur et al. 1988) was then carried out. Six presumptive enzymatic loci, considered diagnostic (or nearly so) between M. domesticus and M. musculus (Bonhomme et al. 1984, Milishnikov &Rafiev 1990, T ucker et al. 1992, Dod et al. 1993) were scored: Isocitrate dehydrogenase-1 (E.C ; Idh1, Chromosome 1), Glucose dehydrogenase-1 (E. C ; Gpd-1, Chr. 4), Superoxide dismutase- 1 (E.C ; Sod-1, Chr. 16), Nucleoside phosphorylase (E.C ; Np, Chr. 14), Esterase- 2 (E.C ; Es-2, Chr. 8) and Mannose phosphate isomerase (E.C ; Mpi, Chr. 9). 82

82 Fig. 1. Map of the study localities in the Czech and Slovak Republics. For each site, a pie diagram is shown depicting the values of the hybrid index HI6, i.e. frequencies of domesticus alleles averaged across six allozyme loci (HI6) per site (black sections). The rectangle indicates the area where hybridisation occurs between the two taxa (shown in detail in Fig. 2). Numbers of localities are the same as in Table 1. Fig. 2. Detail of the westernmost part of Czech Republic and adjacent areas of NE Bavaria. Pie diagrams depict frequencies of domesticus alleles averaged across six allozyme loci (HI6) at each site (black sections). An abrupt transition between the two taxa is apparent from the picture. Numbers of sites are as in Table 1. Laboratory mice of the inbred C57BL/6J strain were used as standards and run alongside unknown samples on the same gel. This strain is autosomally M. domesticus, though its Y chromosome is of M. musculus origin (Nagamine et al. 1992). Since, in all but a single case, only two alleles were shown in the study material, the alleles were designated m (musculus-like) or d (domesticus-like). In Nucleoside phosphorylase, one fast allele was present in M. domesticus (Np 100 ) whereas two alleles were found in M. musculus, slow (Np 70 =m 1 ) and medium (Np 90 =m 2 ); for simplicity, the last two alleles were considered as a single musculus allele throughout this paper. A simple hybrid index was calculated from the allozyme data, given as a frequency of domesticus alleles averaged across all alleles within individual populations. Two hybrid indices were used, one based on all six loci scored (HI6), and another one excluding Es-2 (HI5). In addition, frequencies of domesticus alleles were computed for each locus. DNA isolation DNA was isolated from frozen or ethanol-preserved tissues using proteinase K digestion and subsequent extraction with phenol-chloroform and ethanol precipitation (H o e l z e l & G r e e n 1992). 83

83 X chromosome Primers 5 -AATGGGCTAGCGTAGTGCAG-3 and 5 -AGGGGACGTACACTCAGCTTT- 3 were designed to flank the B1 insertion in the Bruton agammaglobulinemia tyrosine kinase gene (Btk; GeneBank U , B1 is located at positions ) on the X chromosome. The insertion appears to be fixed in M. domesticus populations but is absent in M. musculus (Munclinger et al. unpubl.). PCR conditions were 2 mm MgCl 2, 200 µm concentration of dntps, 0.5 u Taq polymerase (Promega) and 0.5 µm concentration of each primer. The 10 Buffer diluted 1:10 has a final composition of 10mM Tris-HCl, 50 mm KCl and 0,1% Triton X-100. Reactions were performed using an MJ thermal cycler under the following conditions: after an initial three minute incubation at 95 C, 35 cycles were performed at 95 C, 62 C, and 72 C, each step for 45 s, with a final extension for ten min at 72 C. PCR products were run on 2% agarose gels. A 342-bp fragment can be observed when the insertion was present, whereas a 206-bp fragment indicated the absence of the insertion. Both fragments are visible in heterozygotes, although the 342-bp band is slightly weaker. Y chromosome Male mice were typed for the presence or absence of an 18-bp deletion, located within the last exon of the Zinc finger protein 2, Y-linked gene (Zfy2) using the method given in Orth et al. (1996). The deletion was found to be fixed in M. musculus and absent in M. domesticus (Nagamine et al. 1992, Boissinot & Boursot 1997). Electrophoresis was run on a mixture of 3% NuSieve and 0.8% Serva agarose gels. Mitochondrial DNA (mtdna) A 430-bp mtdna fragment was amplified using the primers 5 -TTACTTCTGCCAGCCTGACC- 3 (positions in the complete mtdna; Bibb et al. 1981) and 5 - ATGGTGGTACTCCCGCTGTA-3 (positions in the complete mtdna; B i b b Fig. 3. Mitochondrial DNA of house mice from Czech Republic. An amplified 430-bp mtdna fragment was digested with BamH I. The fragment contains the BamH I restriction site in M. domesticus whereas it is absent in M. musculus. Thus a single 430-bp fragment indicates absence of the restriction site (localities Hlavno, lanes 1, 2, 5; and Bouãí, lane 7) while two fragments, 295-bp and 135-bp, indicate its presence (locality Poustka, lanes 3 and 4). Lane C contains PCR mixture without template DNA. Lane 6 shows an unsuccessful amplification attempt of a template DNA. Lane L contains a 100-bp DNA Ladder 323-1S (New England BioLabs Inc.). The five smallest fragments are (in base pairs): 100, 200, 300, 400 and 500. Electrophoresis was run on a 2% agarose gel. 84

84 et al. 1981). The PCR conditions included 1.5 mm MgCl 2, 200 µm concentration of dntps, 0.5 u Taq polymerase (Promega) and 0.5 µm concentration of each primer. The 10 Buffer diluted 1:10 had a final composition of 10mM Tris-HCl, 50 mm KCl and 0,1% Triton X-100. Reactions were performed using an MJ thermal cycler under the following conditions: 95 C for 3 min and 35 cycles at 95 C, 55 C, and 72 C, each step for 30 s. PCR products were then digested with the restriction endonuclease BamH I and subsequently analysed on 2% agarose gels. The amplified fragment contained a BamH I restriction site (position 3565 in the complete mtdna; Bibb et al. 1981) in M. domesticus but is absent in M. musculus (Boursot et al. 1996). Thus, a single 430-bp fragment indicates absence of the restriction site, while two fragments (295-bp and 135-bp) indicate its presence (Fig. 3). t haplotypes Primers Hb.1 and Hb.2 (Schimenti & Hammer 1990) were used to genotype mice for the presence of t haplotypes. The primers flank a 16-bp insertion in the Hba-4ps locus, which is supposed to be present in most of the t haplotypes. PCR and electrophoretic conditions followed protocols given in Schimenti &Hammer (1990). Results As shown in Table 1, most of Czechia and Slovakia are occupied by M. musculus. This is apparent especially in the sex chromosomes markers (Btk, Zfy2) and mtdna (mt-nd1). Conversely, M. domesticus was found only in Germany (S Thuringia, NE Bavaria) and at sites located in the westernmost part of Czechia (Table 1). A slightly different pattern was seen in allozymes (first two columns in Table 1), where domesticus-like alleles were found at low frequencies at sites even hundreds of kilometres east of the zone (Fig. 1). However, comparison of hybrid indices based on the full set of six loci (HI6) and those without Es-2 (HI5) clearly shows the extensive introgression to be due to the esterase locus (this is shown more explicitly in Table 2, where hybrid indices are listed for each allozyme locus separately). If we only consider areas east of the hybrid zone (site nos ), domesticus alleles either do not appear at all (Sod-1, Idh-1) or they are found at two (Np, Mpi) or three (Gpd-1) sites, respectively. In contrast, Es-2 was found to segregate in as many as nine of 27 M. musculus localities. Regardless of this large-scale introgression by some alleles, the transition of allozyme markers from the musculus-side to the domesticus-side is rather abrupt (Fig. 2), and even more so for the sex chromosomes and mtdna markers (cf. Table 1). Except for the westernmost localities (sites 1 16) all the mice studied were light-bellied, with the relative tail length (tail: body ratio) seldom exceeding 100%. Conversely, mice from within the M. domesticus range (except locality No. 1, Waldau) were dark-bellied, with tails longer than the body. Hybrids formed a mixture of dark-bellied, light-bellied and intermediate animals with various tail lengths. Values for tail: body ratios for each locality are listed in the Appendix. The presence of t haplotypes was found in almost half the localities studied (23 of 49). The mean frequency of t/+ heterozygotes was 0.13 (range = ). Even if samples comprising only a single specimen are excluded, the maximum frequency reaches (Hohenberg). Owing to the low number of M. domesticus samples, we could not test if frequencies of t haplotypes were significantly different from those in M. musculus and/or hybrids. We did not find any t/t homozygotes. 85

85 Table 1. List of collecting sites with frequencies of domesticus-type markers for allozymes, Btk, Zfy2, and mt-nd1. The last column shows frequencies of t/+ heterozygotes, which were mapped with a 16-bp insertion at the Hba- 4ps locus. Allozyme data are summarised as hybrid indices computed as frequencies of domesticus alleles averaged across all loci within individual populations (HI6 = hybrid index base on the whole set of six loci; HI5 = hybrid index based on five loci, excluding Es-2). In parentheses, numbers of animals scored; in markers other than HI6, numbers are only indicated when different from allozymes. No. Locality HI5 HI6 Btk Zfy2 mt-nd1 Hba-4ps 1 Waldau 1 (1) Röslau 1 (2) (1) Thierstein (14) (13) (8) 1 (12) Neuenreuth (11) (7) Hohenberg (6) Trojmezí (3) Hazlov (9) (1) Poustka (38) (36) (16) (36) (36) 9 LuÏná (12) (4) Dolní Pelhfiimov (6) (4) Plesná (123) (46) (26) (44) (46) 12 KfiiÏovatka (9) (8) (6) (8) (8) 13 StfiíÏov (7) (5) 0 (2) (5) (6) 14 Horní Ves (20) (18) (12) (18) (18) 15 Dolnice (30) (19) (28) 0 16 Kopanina (13) (6) Nebanice (8) (2) Kacefiov (14) (8) Dolina (21) (10) Bouãí (10) (7) Hlavno (6) (3) Kostelní Bfiíza (24) (8) Osvinov (3) StráÏ nad Ohfií 0 (4) (2) Horní Jifietín (7) (1) Filipov 0 (2) Nov DÛm 0 (1) Na Kokrdech 0 (7) (4) 0 (1) 0 (4) 0 (4) 29 Doupno 0 (2) 0 0 (2) 0 (1) 0 (2) (2) 30 BoÏtû ice 0 (5) 0 0 (4) 0 (3) 0 (4) 0 (4) 31 Mlad Smolivec (3) (1) 0 (2) Jaro kov 0 (1) Bene ova Hora 0 (23) (2) 0 (2) 0 (1) 0 (2) 34 Masákova Lhota 0 (4) (3) Zdíkovec 0 (33) (20) 0 (18) 0 (20) 0 (20) 36 Vimperk 0 (9) (4) 0 (8) 0 37 Dvory nad LuÏnicí (2) (1) Telã 0 (1) Havraníky 0 (1) atov 0 (3) Brno (3) Tûchov 0 (3) (1) Rudické propadání 0 (1) Vilémovice 0 (2) (1) 0 (1) 0 (1) 0 (1) 45 Lipovec 0 (1) Kulífiov (13) (11) 0 (4) 0 (11) (10) 47 Zemianska Oºãa 0 (8) (4) Necpaly 0 (8) (1) HraÀ (7) (3)

86 Table 2. Frequencies of domesticus alleles for six diagnostic allozyme loci scored at each locality under study. No Locality Es-2 Gpd-1 Idh-1 Mpi Np Sod-1 1 Waldau (1) 1 (1) 1 (1) 1 (1) 1 (1) 1 (1) 2 Röslau 1 (2) 1 (2) 1 (2) 1 (2) 1 (2) 1 (2) 3 Thierstein (11) (14) (14) 1 (14) 1 (14) (13) 4 Neuenreuth (10) 1 (11) (11) (11) (11) (11) 5 Hohenberg (6) 1 (5) (6) 1 (6) (6) 1 (6) 6 Trojmezí (3) 1 (3) 1 (3) (3) (3) 1 (3) 7 Hazlov (9) 1 (9) (9) (9) (9) 1 (9) 8 Poustka (38) (31) (32) (34) (38) (37) 9 LuÏná (12) 1 (12) (12) (12) (12) 1 (12) 10 Dolní Pelhfiimov (6) (3) (6) (6) (6) (6) 11 Plesná (77) (70) (122) (122) (123) (122) 12 KfiiÏovatka (9) (8) (9) 1 (8) (9) (8) 13 StfiíÏov (7) (7) (6) (7) (7) (7) 14 Horní Ves (20) (19) (20) (17) (20) (20) 15 Dolnice (30) (25) (29) (30) (30) (28) 16 Kopanina (13) (9) 1 (13) (9) (13) 1 (13) 17 Nebanice (8) (8) (8) (8) (8) (8) 18 Kacefiov (14) (13) 0 (14) (14) (14) (14) 19 Dolina (20) (20) (20) 0 (21) (20) (20) 20 Bouãí (9) 0 (10) 0 (10) (10) (10) (10) 21 Hlavno (6) (6) 0 (6) (6) 0 (6) (6) 22 Kostelní Bfiíza 0 (19) 0 (15) 0 (24) (24) 0 (24) 0 (20) 23 Osvinov 0 (3) 0 (3) 0 (3) (3) 0 (3) 0 (3) 24 StráÏ nad Ohfií 0 (4) 0 (3) 0 (4) 0 (4) 0 (4) 0 (4) 25 Horní Jifietín (7) 0 (6) 0 (7) (7) 0 (6) 0 (7) 26 Filipov 0 (2) 0 (2) 0 (2) 0 (2) 0 (2) 0 (2) 27 Nov DÛm 0 (1) 0 (1) 0 (1) 0 (1) 0 (1) 0 (1) 28 Na Kokrdech (7) 0 (7) 0 (7) 0 (7) 0 (7) 0 (7) 29 Doupno 0 (2) 0 (2) 0 (2) 0 (2) 0 (2) 0 (2) 30 BoÏtû ice 0 (5) 0 (5) 0 (5) 0 (5) 0 (5) 0 (5) 31 Mlad Smolivec 0 (3) (3) 0 (1) 0 (3) 0 (3) 0 (3) 32 Jaro kov 0 (1) 0 (1) 0 (1) 0 (1) 0 (1) 0 (1) 33 Bene ova Hora (14) 0 (23) 0 (23) 0 (23) 0 (23) 0 (23) 34 Masákova Lhota 0 (4) 0 (4) 0 (4) 0 (4) 0 (4) 0 (4) 35 Zdíkovec (21) 0 (33) 0 (21) 0 (33) 0 (33) 0 (33) 36 Vimperk (9) 0 (9) 0 (4) 0 (9) 0 (9) 0 (9) 37 Dvory nad LuÏnicí 0 (2) 0 (2) 0 (2) 0 (2) (2) 0 (2) 38 Telã 0 (1) 0 (1) 0 (1) 0 (1) 0 (1) 0 (1) 39 Havraníky -- 0 (1) -- 0 (1) 0 (1) 0 (1) 40 atov (3) 0 (3) 0 (3) 0 (3) 0 (3) 0 (3) 41 Brno 0 (2) (3) 0 (3) 0 (3) 0 (3) 0 (3) 42 Tûchov 0 (3) 0 (3) 0 (3) 0 (3) 0 (3) 0 (3) 43 Rudické propadání 0 (1) 0 (1) -- 0 (1) 0 (1) 0 (1) 44 Vilémovice (2) 0 (2) 0 (2) 0 (2) 0 (2) 0 (2) 45 Lipovec 0 (1) 0 (1) 0 (1) 0 (1) 0 (1) 0 (1) 46 Kulífiov (13) 0 (13) 0 (9) 0 (13) (13) 0 (13) 47 Zemianska Oºãa 0 (8) 0 (8) 0 (8) 0 (8) 0 (8) 0 (8) 48 Necpaly (8) 0 (8) 0 (8) 0 (8) 0 (8) 0 (8) 49 HraÀ 0 (6) (7) 0 (4) 0 (6) 0 (6) 0 (6) Discussion Our genetic investigations have revealed that almost the entire territory of the Czech and Slovak Republics is populated by M. musculus. However, the westernmost part of Czechia harbours populations of M. musculus/m. domesticus hybrids and/or pure M. domesticus as shown by Macholán & Zima (1994) and Lazarová (1999). Although a detailed 87

87 Appendix. Values of the mean relative tail length for each sample for which relevant data were available. TL/HBL*100, the tail length to head-and-body length ratio (expressed as a percentage); SE, standard error of the mean; in parentheses, numbers of animals measured. Site numbers correspond to those in Table 1. No. Locality TL/HBL*100±SE No. Locality TL/HBL*100±SE 1 Waldau (1) 25 Horní Jifietín ± 2.33 (7) 2 Röslau ± 3.81 (2) 26 Filipov ± (2) 3 Thierstein ± 1.70 (14) 27 Nov DÛm (1) 4 Neuenreuth ± 2.60 (8) 28 Na Kokrdech ± 4.23 (4) 5 Hohenberg ± 2.52 (6) 29 Doupno ± 1.95 (2) 6 Trojmezí ± 3.34 (3) 30 BoÏtû ice ± 0.92 (4) 7 Hazlov ± 1.76 (9) 31 Mlad Smolivec ± 0.35 (3) 8 Poustka ± 1.19 (35) 32 Jaro kov (1) 9 LuÏná ± 1.66 (15) 33 Bene ova Hora ± 2.30 (8) 10 Dolní Pelhfiimov ± 3.21 (6) 34 Masákova Lhota ± 5.59 (4) 11 Plesná ± 0.74 (109) 35 Zdíkovec ± 1.75 (19) 12 KfiiÏovatka ± 4.18 (8) 36 Vimperk ± 2.41 (8) 13 StfiíÏov ± 3.96 (14) 37 Dvory nad LuÏnicí ± 8.56 (2) 14 Horní Ves ± 1.92 (18) 39 Havraníky (1) 15 Dolnice ± 1.61 (29) 40 atov ± 3.41 (3) 16 Kopanina ± 1.98 (13) 41 Brno ± 2.44 (2) 17 Nebanice ± 1.53 (15) 42 Tûchov ± 4.77 (4) 18 Kacefiov ± 1.79 (13) 43 Rudické propadání (1) 19 Dolina ± 3.11 (21) 45 Lipovec (1) 20 Bouãí ± 3.11 (11) 46 Kulífiov ± 2.34 (8) 21 Hlavno ± 0.74 (6) 47 Zemianska Oºãa ± 1.86 (8) 22 Kostelní Bfiíza ± 1.37 (23) 48 Necpaly ± 2.61 (8) 23 Osvinov ± 0.79 (3) 49 HraÀ ± 2.41 (7) 24 StráÏ nad Ohfií ± 1.79 (4) study of the hybrid zone was not the focus of the present paper, it appears that the zone is quite narrow, and that the transition of allozyme markers between the two species takes place within less than 20 km. The position of the transition zone fits with the results of Kraft s (1984) morphological studies from the entirety of Bavaria, as well as with the results of genetic and parasitological studies along a transect across the hybrid zone in southern Bavaria and north-western Austria (Sage et al. 1986a,b, T ucker et al. 1992). In contrast, we contend that the precise position of the M. domesticus/m. musculus hybrid zone north of the area studied by us should still be treated as undetermined. As far as external features are concerned, the animals under study corresponded with the data in published reports (e.g. M arshall & Sage 1981, Macholán 1996 and references therein). However, as already pointed out by Macholán (1996), morphological criteria such as coloration and relative tail length should only be considered as a rough guide. Mice from as many as five sites (nos. 20, 28, 37, 40, 43) from well within the M. musculus range included mice with average tail: body ratios exceeding 100% but, on the other hand, a mouse from Waldau (Thuringia) had a relatively short tail (93.07%, see Appendix). Obviously, the level of commensalism of a population, and perhaps other factors as well, play a role so that strictly commensal mice are generally darker and longer-tailed, whereas mice living outdoors tend to be lighter and shorter-tailed (Macholán 1996). Extensive introgression of some allozyme alleles has been observed in Denmark (Hunt & Selander 1973, V anlerberghe et al. 1986, 1988b, D o d et al. 1993), Bulgaria (Boursot et al. 1984, V anlerberghe et al. 1986, 1988a) and southern Germany (Tucker et al. 1992). In the present study, we found domesticus alleles as far east of the zone as southern Moravia (SE Czechia) and eastern Slovakia, i.e. hundreds of kilometres 88

88 apart. However, occurrence of M. domesticus allozyme markers is rather incidental and their frequencies are usually extremely low (except in very small samples, cf. Table 1). Hence, it is not clear if this phenomenon has resulted from human-mediated long-distance dispersal, or from gene flow across the zone due to a weak barrier against supposedly neutral markers, or from incomplete fixation of diagnostic alleles due to incomplete lineage sorting for autosomal markers. [Under an assumption of neutrality, the expected time for lineage sorting is four times longer for nuclear markers than for mitochondrial ones due to a four times larger effective population size; see A vise (2000).] The last possibility seems to correspond well with our finding no such introgression in the mtdna marker (Table 1). Finally, we cannot rule out the possibility that seemingly introgressed markers are, in fact, newly arisen alleles, unidentified by our methods. On the other hand, the introgression in western Czechia appears to be systematic at the locus. It is apparent from Table 1 that Es-2 tends to decrease values of the hybrid index in the westernmost sites and, conversely, increases it at sites east of the supposed centre of the hybrid zone; this results in a broader cline. The highest introgression of domesticus alleles at the Es-2 locus was also found in Denmark (H unt & Selander 1973) and Bulgaria (B oursot et al. 1984, Vanlerberghe et al. 1988a). However, it should be noted that large penetration of Mpi was also reported in Bulgaria (B oursot et al. 1984) and this was expressed more explicitly by V anlerberghe et al. (1986), who found the Es-2 introgression rather intermediate (being lower than that of Mpi, Np and Pgm-1). Similar results were reported from Denmark, where the Es-2 introgression was lower than in Mpi and Pgm-1 (Vanlerberghe et al. 1986). Mus domesticus alleles at Mpi were also found at least 50 km eastwards of the East Holstein Peninsula (Prager et al. 1993). In addition, T ucker et al. (1992) found domesticus-like alleles at Np-1 up to 100 km east of the centre of the hybrid zone along a transect in southern Germany and western Austria. In all these cases, the introgression was asymmetrical i.e. higher for M. domesticus alleles penetrating into the range of M. musculus. Unfortunately, these results can not be assessed with our data owing to the lack of appropriate sampling sites on the M. domesticus side. Limited intogression of the X and Y chromosome markers is in accord with strongly impeded movement of sex chromosome loci found in other sections of the hybrid zone (Vanlerberghe et al. 1986, T ucker et al. 1992, Dod et al. 1993, Prager et al. 1997). A general lack of populations with both variants of the X chromosome marker (Btk) in the study area supports the view that X-linked markers are strongly selected against in the hybrid genome (D o d et al. 1993). Conversely, our analysis of the Y chromosome marker revealed an unexpectedly gradual transition between domesticus-type and musculus-type Zfy2 alleles. Moreover, none but one of the localities studied was found to carry the fixed M. domesticus Y chromosome (the only exception was Röslau, where only a single male was studied). This contradicts the results of studies carried out in Denmark (D o d et al. 1993), East Holstein (P rager et al. 1997), southern Germany (T ucker et al. 1992) and Bulgaria (V anlerberghe et al. 1986) which report steep clines for the Y chromosome. Our results should therefore be verified by comparison with material from southern Germany. Since mtdna is almost exclusively matrilineal and inherited as a single unit due to a lack of recombination (but see Gyllensten et al. 1991, W allis 1999, Ladoukakis & Zouros 2001) one simple marker can be used to differentiate between the M. domesticus or M. musculus origin of the mtdna in hybrid animals. In accord with previous studies on mtdna variation in Europe (Sage et al. 1990, Prager et al. 1996) we have demonstrated the prevalence of musculus-type mtdna throughout the 89

89 study area, domesticus-type mtdna being confined to the westernmost parts. Similarly to the Btk gene, the transition from the musculus side to the domesticus side was rather steep; this is in agreement with Bulgarian data (Boursot et al. 1984, V anlerberghe et al. 1988a). Limited evidence presented by Prager et al. (1996) suggests a rather different pattern, including the presence of both domesticus and musculus mtdna about 100 km eastwards of the centre of the hybrid zone (locality 1 in the Bavarian transect, Branau). A remarkably different situation occurs in Scandinavia, where populations of M. musculus have been found to carry exclusively M. domesticus mtdna (F erris et al. 1983a,b, Gyllensten & Wilson 1987, V anlerberghe et al. 1988b, Prager et al. 1993). This pattern was suggested as having resulted from a supposed colonisation event by M. domesticus females from East Holstein, where a high proportion of the mice appeared to carry the types of mtdna found in Scandinavian M. musculus, as do most mice from the musculus side of the Holstein hybrid zone (G yllensten & Wilson 1987, Prager et al. 1993). Thus the presence of M. domesticus mtdna in northern Denmark, Sweden and Finland is not a result of permanent gene flow across the zone, but rather a consequence of a founder effect before the establishment of the Danish hybrid zone. The frequency of t/+ heterozygotes ascertained in this study falls into the range based on previously studied populations of the house mouse (reviewed in A r d l i e 1998). However, it is worth noting that populations (both of M. musculus and those with a high proportion of M. domesticus alleles) with more than 50% of t/+ animals were not exceptional. Since all known M. musculus t haplotypes belong to the same t w73 group (Klein et al. 1984, Forejt et al. 1988, R uvinsky et al. 1991, M azin 1994) all t/t carrying individuals in this species probably die early in gestation. Therefore, the absence of t/t homozygotes in M. musculus populations is not surprising. t haplotypes belonging to other complementation groups can only be expected in M. domesticus and in hybrid populations. Unfortunately, the method used did not allow us to ascertain any details about the haplotypes or complementation groups of the studied t complexes. Laboratory crosses with mice carrying known t haplotypes or microsatellite typing of t/+ animals (as suggested in Ardlie & Silver 1996) are needed to obtain more information about the t haplotypes found in the present study. We conclude that the molecular markers and allozymes used in this study are capable of mapping the position of the hybrid zone in central Europe, and of revealing some details of the genetic features of wild mice populations, e.g. the frequency of t haplotypes. Since PCRbased markers are practicable for even low quality templates such as DNA from museum specimens (cf. P rager et al. 1998) they probably represent the best option for future studies. These will extend our knowledge of the exact position of the hybrid zone, which is still not satisfactorily resolved in some areas. Acknowledgement We are grateful to Barbara D o d and Pierre Boursot for helpful suggestions and providing some of primers used. We are also obliged to anonymous referees for notably useful comments (although not all of them have been accepted). Our special thanks are to Jan Zima (Charles University, Praha and Academy of Science of the Czech Republic, Brno), Andrea Mack (Ökologische Bildugsstädtte, Hohenberg/Eger) and Karel BroÏ (Soos Reserve, Skalná) whose aid was substantial for the entirety of the project. The help of Milo Andûra, Barbora Bímová, Josef Bryja, Miroslav Kovafiík, Martin Lípa, David Luk, Pavel MackÛ, Ivan MaÀásek, Ján Obuch, Zdenûk Vo lajer and lots of local people in collecting material is also much acknowledged. This research was supported by the Grant Agency of the Academy of Science of the Czech Republic (grants no. A and A ). The participation of PM and EB was additionally supported by the Grant Agency of the Czech Republic (grant no. 206/01/0989). We are also grateful to the Ministry of Education, Youth and Sport of the Czech Republic, whose grant VZ formed a framework for a part of this study. 90

90 LITERATURE ARDLIE, K. G., 1998: Putting the brake on drive: meiotic drive of t haplotypes in natural populations of mice. Trends Genet., 14: ARDLIE, K. & SILVER, L. M., 1996: Recent evolution of mouse t haplotypes at polymorphic microsatellites associated with the t complex responder (Tcr) locus. Genet. Res., 67: AUFFRAY, J.-C., MARSHALL, J. T., THALER, L. & BONHOMME, F., 1990: Focus on the nomenclature of European species of Mus. Mouse Genome, 88: 7 8. AVISE, J. C., 2000: Phylogeography. The history and formation of species. Harvard University Press, Cambridge, Massachusetts. BIBB, M. J., VAN ETTEN, R. A., WRIGHT, C. T., WALBERG, M. W. & CLAYTON, D. A., 1981: Sequence and gene organization of mouse mitochondrial DNA. Cell: 26, BOISSINOT, S. & BOURSOT, P., 1997: Discordant phylogeographic patterns between the Y chromosome and mitochondrial DNA in the house mouse: selection on the Y chromosome? Genetics, 146: BONHOMME, F., CATALAN, J., BRITTON-DAVIDIAN, J., CHAPMAN, V. M., MORIWAKI, K., NEVO, E. & THALER, L., 1984: Biochemical diversity and evolution in the genus Mus. Biochem. Genet., 22: BOURSOT, P., AUFFRAY, J.-C., BRITTON-DAVIDIAN, J. & BONHOMME, F., 1993: The evolution of house mice. Ann. Rev. Ecol. Syst., 24: BOURSOT, P., BONHOMME, F., BRITTON-DAVIDIAN, J., CATALAN, J., YONEKAWA, H., ORSINI, P., GUERASIMOV, S. & THALER, L., 1984: Introgression différentielle des génomes nucléaires et mitochondriaux chez deux semi-espèces européennes de Souris. C. R. Acad. Sc. Paris, Ser. III, 299: BOURSOT, P., DIN, W., ANAND, R., DARVICHE, D., DOD, B., VAN DEIMLING, F., TALWAR, G. P. & BONHOMME, F., 1996: Origin and radiation of the house mouse: mitochondrial DNA phylogeny. J. Evol. Biol., 9: CORBET, G. B., 1988: Mus musculus domesticus Schwarz & Schwarz, 1943 (Mammalia, Rodentia): proposed conservation. Bull. Zool. Nomen., 45: DOD, B., JERMIIN, L. S., BOURSOT, P., CHAPMAN, V. M., NIELSEN, J. T. & BONHOMME, F., 1993: Counterselection on sex chromosomes in the Mus musculus European hybrid zone. J. Evol. Biol., 6: DIN, W., ANAND, R., BOURSOT, P., DARVICHE, D., DOD, B., JOUVIN-MARCHE, E., ORTH, A., TALWAR, G. P., CAZENAVE, P.-A. & BONHOMME, F. 1996: Origin and radiation of the house mouse: clues from nuclear genes. J. Evol. Biol., 9: FERRIS, S. D., SAGE, R. D., HUANG, C.-M., NIELSEN, J. T., RITTE, U. & WILSON, A. C., 1983a: Flow of mitochondrial DNA across a species boundary. Proc. Natl. Acad. Sci. USA, 80: FERRIS, S. D., SAGE, R. D., PRAGER, E. M., RITTE, U. & WILSON, A. C., 1983b: Mitochondrial DNA evolution in mice. Genetics, 105: FOREJT, J., GREGOROVÁ, S. & JANSA, P., 1988: Three new t-haplotypes of Mus musculus reveal structural similarities to t-haplotypes of Mus domesticus. Genet. Res., 51: GYLLENSTEN, U., WHARTON, D., JOSEFSSON, A. & WILSON, A. C., 1991: Paternal inheritance of mitochondrial DNA in mice. Nature, 352: GYLLENSTEN, U. & WILSON, A. C., 1987: Interspecific mitochondrial DNA transfer and the colonization of Scandinavia by mice. Genet. Res., 49: HARRIS, H. & HOPKINSON, D. A., 1976: Handbook of enzyme electrophoresis in human genetics. North- Holland Publ. Comp., Amsterdam, Oxford. HOELZEL, A. R. & GREEN, A., 1992: Analysis of population-level variation by sequencing PCR-amplified DNA. In: Hoelzel, A. R. (ed.), Molecular genetic analysis of populations. A practical approach. Oxford University Press, New York: HUNT, W. G. & SELANDER, R. K., 1973: Biochemical genetics of hybridisation in European house mice. Heredity, 31: KLEIN, J., SIPOS, P. & FIGUEROA, F., 1984: Polymorphism of t-complex genes in European wild mice. Genet. Res., 44: KRAFT, R. V., 1984: Merkmale und Verbreitung der Hausmäuse Mus musculus musculus L., 1758, und Mus musculus domesticus Rutty, 1772 (Rodentia, Muridae) in Bayern. Säugetierk. Mitt., 32: LADOUKAKIS, M. & ZOUROS, E., 2001: Recombination in animal mitochondrial DNA. 8 th Congr. ESEB, Aarhus: 256. LAZAROVÁ, J., 1999: Epigenetic variation and fluctuating asymmetry of the house mouse (Mus) in the Czech Republic. Folia Zool., 48 (Suppl. 1):

91 MACHOLÁN, M. & ZIMA, J., 1994: Mus domesticus in Western Bohemia: a new mammal for the Czech Republic. Folia Zool., 43: MACHOLÁN, M., 1996: Key to European house mice (Mus). Folia Zool., 45: MARSHALL, J. T. Jr. & SAGE, R. D., 1981: Taxonomy of the house mouse. Symp. zool. Soc. Lond., 47: MAZIN, S. M., 1994: Distribution of t-haplotypes. In: Kotenkova, E. V. & Bulatova, N. Sh. (eds), The house mouse. Origin, distribution, systematics, behaviour. Nauka, Moscow: MILISHNIKOV, A. N. & RAFIEV, A. N., 1990: High level of introgression of nuclear protein genes of Mus domesticus in a Mus musculus s. str. population from Azerbaijan. Proc. 5 th Congr. of the All-Union Teriol. Soc. AS USSR, Moscow, Vol. 1: MITCHELL-JONES, A. J., AMORI, G., BOGDANOWICZ, W., KRY TUFEK, B., REIJNDERS, P. J. H., SPITZENBERGER, F., STUBBE, M., THISSEN, J. B. M., VOHRALÍK, V. & ZIMA, J., 1999: The atlas of European mammals. Academic Press, London. MUSSER, G. G. & CARLETON, M. D., 1993: Family Muridae. In: Wilson, D. E. & Reeder, D. A. M. (eds), Mammal species of the world. A taxonomic and geographic reference. (2 nd ed). Smithsonian Institution Press, Washington and London: NAGAMINE, C. M., NISHIOKA, Y., MORIWAKI, K., BOURSOT, P., BONHOMME, F. & LAU, Y. F. C., 1992: The musculus-type Y chromosome of the laboratory mouse is of Asian origin. Mamm. Genome, 3: ORTH, A., LYAPUNOVA, E., KANDAUROV, A., BOISSINOT, S., BOURSOT, P., V ORONTSOV, N., & BONHOMME, F., 1996: L espèce polytypique Mus musculus in Transcaucasie. C. R. Acad. Sci. Paris, Sci. Vie., 319: PASTEUR, N., PASTEUR, G., BONHOMME, F., CATALAN, J. & BRITTON-DAVIDIAN, J., 1988: Practical isozyme genetics. Ellis Horwood Ltd., Chichester. PRAGER, E. M., BOURSOT, P., & SAGE, R. D., 1997: New assays for Y chromosome and p53 pseudogene clines among East Holstein house mice. Mamm. Genome, 8: PRAGER, E. M., ORREGO, C., & SAGE, R. D., 1998: Genetic variation and phylogeography of central Asian and other house mice, including a major new mitochondrial lineage in Yemen. Genetics, 150: PRAGER, E. M., SAGE, R. D., GYLLENSTEN, U., THOMAS, W. K., HÜBNER, R., JONES, C. S., NOBLE, L., SEARLE, J. B. & WILSON, A. C., 1993: Mitochondrial DNA sequence diversity and the colonization of Scandinavia by house mice from East Holstein. Biol. J. Linn. Soc., 50: PRAGER, E. M., TICHY, H., & SAGE, R. D., 1996: Mitochondrial DNA sequence variation in the eastern house mouse, Mus musculus: Comparison with other house mouse mice and report of 75-bp tandem repeat. Genetics, 143: RUVINSKY, A., POLYAKOV, A., AGULNIK, A., TICHY, H., FIGUEROA, F. & KLEIN, J., 1991: Low diversity of t haplotypes in the eastern form of the house mouse, Mus musculus L. Genetics, 127: SAGE, R. D., ATCHLEY, W. R. & CAPANNA, E., 1993: House mice as models in systematic biology. Syst. Biol., 42: SAGE, R. D., HEYNEMAN, D., LIM, K.-C. & WILSON, A. C., 1986a: Wormy mice in a hybrid zone. Nature, 324: SAGE, R. D., PRAGER, E. M., TICHY, H., & WILSON, A. C., 1990: Mitochondrial DNA variation in house mice, Mus domesticus (Rutty). Biol. J. Linn. Soc., 41: SAGE, R. D., WHITNEY III, J. B. & WILSON, A. C., 1986b: Genetic analysis of a hybrid zone between domesticus and musculus mice (Mus musculus complex): Hemoglobin polymorphisms. Curr. Top. Microbiol. Immunol., 127: SCHIMENTI, J. & HAMMER, M., 1990: Rapid indentification of mouse t haplotypes by PCR polymorphism (PCRP). Mouse Genome, 87: 108. TUCKER, P. K., SAGE, R. D., WARNER, J., WILSON, A. C. & EICHER, E. M., 1992: Abrupt cline for sex chromosomes in a hybrid zone between two species of mice. Evolution, 46: VANLERBERGHE, F., BOURSOT, P., CATALAN, J., GERASIMOV, S., BONHOMME, F., BOTEV, B. A. & THALER, L., 1988a: Analyse génétique de la zone d hybridation entre les deux sous-espèces de souris Mus musculus domesticus et Mus musculus musculus en Bulgarie. Genome, 30: VANLERBERGHE, F., BOURSOT, P., NIELSEN, J. T. & BONHOMME, F., 1988b: A steep cline for mitochondrial DNA in Danish mice. Genet. Res., 52: VANLERBERGHE, F., DOD, B., BOURSOT, P., BELLIS, M. & BONHOMME, F., 1986: Absence of Y- chromosome introgression across the hybrid zone between Mus musculus domesticus and Mus musculus musculus. Genet. Res., 48: WALLIS, G. P., 1999: Do animal mitochondrial genomes recombine? Trends Ecol. Evol., 14:

92 Folia Zool. 51(2): (2002) Food storage, prey remains and notes on occasional vertebrates in the diet of the Eurasian water shrew, Neomys fodiens Werner HABERL Hamburgerstrasse 11, A-1050 Vienna, Austria; Received 25 June 2001; Accepted 7 January 2002 Abstract. The food remains of Neomys fodiens (particularly trichopteran larvae, Gastropoda and Amphibia) found on the banks of ponds and small creeks in Lower Austria are described. Characteristic bite marks, the manner of opening the cases and shells, as well as data on feeding patterns are presented. Food caches mainly consisted of caddis fly larvae and snails, but also contained non-palatable items which shrews apparently had confused with real prey and retrieved. The composition of the caches varied seasonally, showing a marked mid-summer decline and a shift in the proportion of Trichoptera and Mollusca in late summer and autumn. Shrews employed particular methods when breaking snail shells and opening caddis fly cases, and in the consumption of vertebrate carcasses. Key words: shrews, ecology, foraging behaviour, food cache, frogs, carrion Introduction The Eurasian water shrew Neomys fodiens (Pennant, 1771) inhabits various riparian and littoral habitats of the Palaearctic including the shores of marine habitats, lakes and rivers, but also lives in swamps, humid woodland, wet meadows and, in the northern part of its range, even fields and drainage ditches (Churchfield 1990, Spitzenberger 1999). In the south it is confined to mountainous areas where it lives in the reedbeds of lakes, mountain brooks, small rivers and riverine forests. Water shrews take both terrestrial and aquatic prey, the latter particularly including caddis fly larvae, crustaceans and snails. Whereas most small prey items can easily be detected by gut or faecal analysis from undigested chitinous exoskeletal fragments (C hurchfield 1985), other prey are difficult to detect from such analyses, even though their mass intake may be high. This is especially true for snails and occasional vertebrate prey such as amphibians, fish and occasionally utilised small mammal carcasses (H a berl 1993). Water shrews usually have favourite landing and feeding places as well as caches in secluded places amongst vegetation or under stones and logs, where it is possible to find immobilised prey and prey remains. The literature (see H a b e r l 1995) includes only a few reports on water shrew food storage places, and all refer to N. fodiens rather than N. anomalus. Formozov (1948) found a few snails in a shrew burrow, but the first reports of larger prey accumulations are from the Bialowieza National Park in Poland (B uchalczyk & P u c e k 1963, W o l k 1976). These storage places included caddis fly larvae, snails, partly eaten frogs and fish. Amongst caddis fly and fish remains, K r a f t & Pleyer (1978) found over 4,000 snail shells at ca. 60 feeding places on the banks of German fishponds. Descriptions of the behaviour of water shrews when opening snail shells are sparse, controversial and limited to Lymnaeidae (B uchalczyk & Pucek 1963, W o l k 1976, K r af t & Pleyer 1978, K r af t 1980, Köhler 1984). However, fixed patterns 93

93 of prey handling have been described for other shrew species (N. anomalus, Sorex araneus, S. minutus, Crocidura suaveolens) including predation on snails and other prey items (Hutterer 1976, H a berl 1998). This paper deals with the food remains of Neomys fodiens found in a study area in Austria. The main concerns of the study were prey composition at storage places during different parts of the year, the positioning of prey storage places, and the description of patterns of prey utilisation. Material and Methods The data were collected in the course of a study of small mammals in a wetland habitat in the Waldviertel (Schönbach, Lower Austria; H:48 27,33, V:15 2,44 ) between This wetland was characterized by several underground springs and brooks leading into two small ponds that drained into the plain of the Kleiner Kamp river. Dominant plants on the study plot were Polygonum bistorta, Filipendula ulmaria, Deschampsia caespitosa and several Carex species (see H ab e r l 1993 for details). Live trapping was conducted in addition to collecting tracks on smoked paper. The banks of streams and ponds in and around the study area were regularly searched for storage places and the food remains found were collected. Prey items were classified as being those of N. fodiens on the following criteria: (1) The storage places conformed to those described in the literature, (2) they were close to water, (3) live traps placed near the storage places yielded only N. fodiens, (4) tracks on smoked paper were found to be those of N. fodiens when compared with reference tracks from captive N. fodiens and N. anomalus, (5) faeces found nearby were those of N. fodiens (containing aquatic prey, mainly Gammarus sp.), and (6) snail shells showed the same bite marks as found on snails fed to captive Neomys, and as described by Buchalczyk & Pucek (1963) and K r a f t & Pleyer (1978). In order to obtain reference material on patterns of prey handling and carcass consumption, behavioural observations and experiments were conducted upon captive shrews. Animals were caged separately in glass vivaria containing a small water bowl, a ground layer of 5 10 cm of peat, soil and adequate cover (pieces of wood, stones, foliage and grass-tufts with rootstocks). Captive shrews were fed a basic diet of Tenebrio larvae ad libitum. Additional food items, particularly snails, and also fish and small mammal carrion were offered to captive shrews, and their utilisation was documented. Single experiments were conducted using prey dummies presented to the shrews in dry bowls: pieces of plastic drinking straws and gelatine capsules (either empty or filled with mealworms), clean cigarette filters and balls of aluminium foil. Small mammal carrion was also laid out in the field. Rodent carcasses (Murinae, Microtinae) (n = 64) and tracking tunnels (smoked paper placed in PVC pipes) were randomly deposited in the study area. Carrion was fixed to the ground by a stake and sheltered from rain by a plastic dish placed ca. 5 cm above ground. Carrion was inspected regularly, weighed and photographed, and the presence of bite marks and faeces were documented. Results Food caches and prey remains A few empty trichopteran larval cases were found in a clump close to a fishpond in Larger accumulations were discovered along the banks of a small pond (ca. 30 m 2 ) in the Lower Austrian study area from May The majority of the feeding places were on the very steep 94

94 banks of this pond (slope >60º), located 5 30 cm away from the water (d max = 90 cm). Prey remains were concentrated in a small area of 5 10 cm in diameter, and these often were very difficult to find when hidden amongst vegetation. Live-trapping showed that the banks of this small pond were inhabited by 1 3 water shrews (captures from July to November 1989). In total, 568 items from 66 food caches in the study area were found between July 1988 and June For the most part, these included trichopteran larvae (79.2%, mostly Limnephilidae: Limnephilus sp., Glyphotelius sp.) and aquatic Mollusca (9.7%, mostly Radix peregra). Terrestrial Gastropoda (Succinea putris, Perforatella incarnata, Cepea hortensis, Eucobresia diaphana, Vitrina pellucida) constituted only 4%. Other aquatic and terrestrial invertebrates were Hirudinea and the larvae and pupae of Ephemeroptera, Odonata and Coleoptera (mainly Curculionidae, Carabidae, Dytiscidae) (3,2%). Non-palatable items such as twigs, stones and artificial objects such as a broken ping-pong ball and a rubber ring were also identified (3,9%). The caches also contained two colour-marked prey dummies (empty snail shells) that I had positioned near the pond during an earlier experiment. In winter (November to April), even in snowless periods, there were no caches evident and the only finds were two partly eaten frogs (Rana temporaria) collected on and (not included in calculations). All years combined, the largest numbers of food remains were found between May and June and again in October. During late summer (August to September) such finds were rare. Considering Trichoptera and aquatic Mollusca only, a seasonal separation of the data reveals that from August to October Trichoptera predominated (54.1%) but there was a significantly higher proportion of limnic gastropods than in late spring and early summer. In contrast, from May to July limnic gastropods represented 2.3% of the prey found, whilst Trichoptera comprised 87.1% of the food remains (χ 2 = 118; df = 1; P = 0.05) (Fig. 1). Feeding patterns and experimental studies Trichoptera The caddis fly cases found (most were those of pupae) consisted of either crude sand or small stones or small twigs (Limnephilidae use a combination of these materials). Trichopteran Fig. 1. Seasonal changes in the composition of food remains of N. fodiens at the feeding places (n1=433, n2=135) (all years combined). 95

95 cases were sorted and evaluated according to eight categories: (A) intact (5%) (B) opened to a small extent from one end (22,6%) (C) opened from one end to a maximum extent of half the case (26,1%) (D) opened from one end and leaving less than half the case (20,3%) (E) opened from both ends (1,7%) (F) small middle parts (7,9%) (G) opened from the side (12,6%) (H) opened from the side and from one end (3,8%). The distribution deviated significantly from random (χ 2 = 257; df = 7; P = 0.01; n = 522). Shrews mainly opened trichopteran cases from either on one end (patterns B, C, and D) or from the side (patterns G and H) (69% and 16.4%, respectively) (Fig. 2). Mollusca The 55 Radix peregra found at the storage places included 21 that were alive and obviously uninjured and one shell-less body, which had been pulled out of its shell and cached by a water shrew. 29% of the empty shells had been opened from the apex by breaking at least one turn of the shell. Another 15% were opened from the side, but leaving the lip intact. The remainder were not broken or only the apex was missing. This pattern, employed when breaking conical shells, and also observed in captivity (e.g. Radix, Limnaea, Succinea) was not used for larger, spherical terrestrial snails (e.g. Helix, Arianta, Cepea). The latter were either eaten without the shell being broken or, occasionally, by the opening of the shell from the lip, thus breaking away up to one half of the final whorl or more (a method similar to that employed by Sorex araneus) (Fig. 2). There seems to be great intra-individual variation in the pattern in which their shells are opened (also see K r a f t 1980). There were also marked individual differences in the time of the shrew s first acceptance of snails as prey. Snails were immediately smelled and investigated when new objects in vivaria and sometimes marked with faeces, however, they were usually attacked and eaten on only after several days or even sometimes weeks. Fig. 2. The most characteristic patterns for opening trichopteran larvae cases and shelled Mollusca. Experiments using prey dummies In captivity, prey dummies (see Material and Methods) were removed by the shrews from feeding bowls and accumulated ( hoarded ) in small groups of 3 17 objects (occasionally mixed with meal worms or snails) in different places in the vivarium. The position of these caches varied: most were in runways or under pieces of wood, some were placed in the open or near or even in water bowls. In the course of several days these caches were 96

96 reorganised, i.e. individual pieces were shifted from one place to another usually over night. It is remarkable that in some cases a few dummies of similar size and make were found cached next to each other, although they were initially mixed when dispensed in the bowl. The plastic straw dummies remained unopened, but gelatine capsules and cigarette filters showed distinct bite marks. 55% of the gelatine capsules (n=40) were opened from the side a pattern similar to that seen in trichopteran cases. Vertebrate prey and carcass consumption Studies both in captivity and in the field showed that distinct eating patterns are not only employed in the consumption of invertebrates, but also of vertebrate prey. The carcasses of small mammals (e.g. Microtus, Apodemus, Mus) usually were cut open, starting with the area behind the head and subsequently opening the thorax by nibbling through the rib-cage (Fig. 3). The lungs, the heart and the liver were consumed first. The limbs were sometimes skinned from inside the thoracic cavity by the shrews pulling them inwards and eating the muscles, so leaving the skin turned inside out. The carrion continued to be eaten for several days. Fig. 3. A field mouse carcass (Microtus arvalis) utilised by shrews. (Photo by W. Haberl) In the field, 59% (n = 64) of the murid carcasses laid out in the study area were utilised by shrews (Sorex and Neomys), at least in part within the first 4 7 days (March November). That this was the work of shrews was obvious from comparing the eating patterns with those found in captive shrews, from faeces deposited on or near the carcasses, and from tracks on smoked paper. After 2 3 weeks most pieces were almost completely consumed so that mainly the skeleton and the skin remained. Some carrion was eaten on the first day or even within a few hours, whereas other carcasses, placed only a few metres away, remained untouched for long periods, even if defecated on by shrews. However, according to the shrews home ranges within the study grid (see H a berl 1993), most carrion utilisation, except for corpses placed on the banks of the pond, could be attributed to Sorex spp., presumably S. araneus. Dead fish fed to captive water shrews were eaten starting from behind the head and continuing in a distinct pattern along the dorsal muscles (Fig. 4). This was also observed in a large trout (Salmo gairdneri) found caught in the drainage net of a small fishpond after drainage; here consumption continued for several days. Of particular interest were the remains of two partially eaten adult frogs (Rana temporaria) found in April Although still breathing, the frogs were immobile and stiff, showed numerous bite marks, and had their hind legs stretched open; spawn protruded from wounds. About one week later I found the remains of one of the frogs about one metre away; it had been almost completely consumed apart from the skin and skeleton. These frog remains closely resembled those reported by Schmidt (1986). 97

97 Discussion Although live-trapping and the collection of tracks on smoked paper revealed the presence of water shrews along the banks of most creeks and ponds in the study area, it is notable that feeding places were found only in certain areas. The majority were located on a small pond in the main capture grid. The reasons for this may be twofold: (1) the particularly steep bank of this pond provided dry nesting places, and (2) the steepness of the slope provided ideal diving conditions, shown to be of great importance for successful shrew foraging (see R u t h a r d t & Schröpfer 1985, Schröpfer 1985), although L a rdet (1988) has shown that N. fodiens prefers shallow water. Underwater burrow exits (S p a nnhof 1952) could not be found in this study (also see Illing et al. 1981). I carefully searched the banks of all other water bodies in and around the study area (small creeks, lakes and fishponds) for over two years but could not find any other similar accumulation of caches. On one occasion I did find a few empty caddis fly cases on a large rock in a creek close to the main study area (Kleiner Kamp). This suggests that a shrew had used the rock as a place for landing prey (and had probably started foraging dives from there) as previously observed by S c h l o e t h (1980), D. Cantoni (pers. comm. by P. Vogel) and P. Vogel (pers. comm.). The scarcity of feeding places in summer (July to September) might be connected to the abundance and availability of terrestrial invertebrate prey. The observed seasonal shift in the ratio of Trichoptera / Mollusca (a higher proportion of snails in late summer and autumn) was also reported by W o l k (1976). The limited numbers of Trichoptera in small ponds, as Fig. 4. Fish carrion (Cyprinidae) utilised by Neomys fodiens. (Photo by W. Haberl) 98

98 well as the timing of metamorphosis, may well account for the higher percentage of Mollusca caught in late summer and autumn (see also K r a f t & Pleyer 1978). Although I found a steady decline in trichopteran case numbers over the course of the year, the comparative rarity of findings in mid summer represents a notable decrease that cannot be explained by a progressively diminishing population (sensu W o l k 1976) alone. One possible explanation would be a higher predation by young shrews before they disperse from their mothers home ranges. This effect may also account for the larger numbers of food remains found in caches in late spring and early summer. In addition, the abundance of terrestrial prey in mid summer and late summer and a corresponding dietary shift could be responsible. Churchfield (1984) also reports of two peaks (May and October) and a mid-summer decline in the trichopteran diet of N. fodiens. Shrews are generalists in terms of the spectrum of food types taken, and yet some degree of specialisation does occur (C hurchfield 1990). In summer Trichoptera pupae may represent an easy, defenceless and valuable prey resource. Thus, when Trichoptera are pupating, foraging by N. fodiens could be regarded as harvesting a foraging mode that has already been reported for Sorex spp. (Hutterer 1982, Kuvikova 1986, Pierce 1987). The patterns of removal of the bodies of larger snails have been reported by Cranbrook (1959), K r af t (1980) and K öhler (1984). The maximum size of broken snail shells was within the range of the largest shells offered to captive animals and found in nature. The mm strong lip of adult Helicidae can easily be broken by N. fodiens and by S. araneus. This confirms that shrews can break even thicker shells (see Broadbrooks 1939, C rowcroft 1957, C r a nbrook 1959). The critical thickness of 0.3 mm for N. fodiens as reported by Köhler (1984) is even outperformed by S. araneus and the size limits of snails (3 5 cm in Lymnaeidae) may well represent realistic values when compared to those given by C r a nbrook (1959). The opening patterns found for Radix peregra in this study show a higher degree of variability than those described for Lymnea stagnalis (Kraft & Pleyer 1978, K r af t 1980). The overall picture presented by Buchalczyk & Pucek (1963) shows a much greater similarity to my findings. While the main bulk of the water shrew s diet comprise small invertebrates, there are many examples of vertebrate prey being caught and eaten. Although there is no evidence from food remains that water shrews kill and eat small rodents in the wild, they may well feed on mice and microtines that have died in their nests or have been left by cats or other predators (S chlüter 1980, K osoi 1982, H a berl 1993). However, my studies on both captive and free-ranging shrews, in which I offered rodent carcasses and observed their utilisation, reveals a distinct eating pattern. Interestingly, feeding on carcasses is not due to lack of other food, since both in the wild and in captivity additional food was abundant. The carcasses are fed on continuously over several days even if, especially in summer, the state of decay is obvious and the carrion smells strongly. This contradicts Crowcroft (1957) who stated that...the meat supplied to shrews must be fresh for they will starve to death rather than feed on tainted flesh (also see S p a nnhof 1952, B a xter & Meester 1980, 1982). However, in foraging theory such carrion represents a food patch that ensures a continuous food supply. S c h l ü t e r (1980) suggests that carrion eating by Sorex araneus and N. fodiens may be most important in winter. Carcasses of reptiles were never found in the study area but small lizards should be considered as easy prey for water shrews (see Mezhzerin 1958) and even Suncus etruscus can kill small lizards with bites to the throat (G e r a e t s 1972). Larger lizards are 99

99 unlikely to be killed by water shrews: an adult male Lacerta agilis and a toad, Bufo bufo, survived three weeks in the vivarium of a water shrew (H a berl 1993). The finding of the remains of the two partially eaten adult R. temporaria are of interest. Frogs seem to be distasteful to shrews because of their slimy and toxic skin secretions (Formanowicz & Brodie 1979, Brodie & Formanowicz 1981, Köhler 1984). In this study predated frogs were found in April March had been unusally warm and the frogs had started to breed, however, there was a sudden cold snap in April. Both frogs were discoverd early in the morning, when ambient temperatures were around freezing, causing immobility and a reduction in skin secretion production (also see Haberl & Wilkinson 1997). This suggests that random climatic fluctuations can make adult frogs vulnerable to shrew predation. Rozmus (1961) states that it is likely that young frogs dispersing and occurring in masses after metamorphosis temporarily represent the main source of food for shrews (also see R örig 1905, W ilcke 1938, S e r geev 1973). O gnev (1959) counts amphibians to the preferred prey of N. fodiens. Buchalczyk & Pucek (1963) and W o l k (1976) report of approximately 20 42% of frogs in the water shrew s diet, whereas Kraft & Pleyer (1978) found no frog remains. L orenz (1952) and Pucek (1959) report that also adult, medium-sized frogs can be killed by N. fodiens and Pernetta (1976) reports the predation on newts (Triturus vulgaris). It is significant that other, non-palatable, objects found in the feeding places closely resembled real prey. The size of found objects was slightly above the mean size of Trichoptera larvae. This suggests a preference for supra-optimal prey size and, hence the strong effect of supra-optimal dummies. According to C h u r c h f i e l d (1990), a surplus of prey seems to stimulate caching behaviour. Hitherto only a few experiments using prey dummies have been conducted with shrews: Shull (1907) used sand-filled snail shells with Blarina brevicauda; Dummies made of yarn and tissue resembling Calliphora larvae could not be discriminated by N. fodiens under water in the experiments by C hurchfield & White (unpublished data, c.f. Churchfield 1985). Pierce (1987) used meal-worm segments in pieces of straws in order to examine the foraging strategy of S. araneus. Crowcroft (1955) observed that S. araneus carried small stones into their nest and interpreted this behaviour as food hoarding that was triggered by an object resembling real prey, once it has been seized and taken into the mouth. As in former studies on N. fodiens, the terms feeding place and storage place / cache have been used synonymously in this report. However, it is recommended that, in future, these terms should be used more specifically whenever possible. It is very difficult to draw a line between the terms as, when finding food remains in the wild, it is almost impossible to differentiate between a storage place (i.e. caught prey that is immobilised and kept for later usage) and a feeding place (i.e. an area where prey is taken to shore and is eaten, leaving only the unpalatable parts). Finds of prey remains can only be evaluated with the aid of direct observations and frequent monitoring of known landing places, preferably by colour marking the remains found and keeping a record of their position. The results also suggest that monitoring should continue over longer periods of time, since the occurrence of feeding places may be sporadic and their composition may undergo seasonal changes. In this respect it is interesting that other studies on the diet and habitat of water shrews do not mention food caches at all (e.g. N iethammer 1977, 1978, S chloeth 1980, C hurchfield 1984, Kuvikova 1985, L a rdet 1988, DuPasquier & Cantoni 1992). 100

100 In order to evaluate the complete dietary spectrum of N. fodiens and knowledge of its foraging behaviour and habitat preferences, it would be necessary to search its habitats for storage places and prey remains regularly, as well as to conduct experiments using captive shrews. Acknowledgements I would like to thank Prof. Dr P. Vogel (University of Lausanne), Dr B. Kry tufek (Slovenian Museum of Natural History) and Dr H. Griffiths (University of Hull) for critically commenting on earlier versions of the manuscript. Two anonymous reviewers are thanked for their valuable comments and corrections. LITERATURE BAXTER, R. M. & MEESTER, J., 1980: Notes on the captive behaviour of five species of southern African shrews. Säugetierkundl. Mitt., 28 (1): BAXTER, R. M. & MEESTER, J., 1982: The captive behavior of the red musk shrew, Crocidura f. flavescens (I. Geoffroy, 1827) (Soricidae: Crocidurinae). Mammalia, 46 (1): BROADBROOKS, H. E., 1939: Food habits of the vagrant shrew. The Murrelet, 20: BRODIE, E. D. JR. & FORMANOWICZ, D. R. JR., 1981: Palatability and antipredator behavior of the treefrog Hyla versicolor to the shrew Blarina brevicauda. J. Herpetol., 15 (2): BUCHALCZYK, T. & PUCEK, Z., 1963: Food storage of the European water shrew, Neomys fodiens (Pennant, 1771). Acta Theriol., 7 (19): CHURCHFIELD, S., 1984: Dietary seperation in three species of shrew inhabiting water-cress beds. J. Zoology, 204: CHURCHFIELD, S., 1985: The feeding ecology of the European water shrew. Mamm. Rev., 15 (1): CHURCHFIELD, S., 1990: The natural history of shrews. Christopher Helm, London. CRANBROOK, THE EARL OF, 1959: The feeding habits of the water shrew, N. fodiens bicolor Shaw, in captivity and the effect of its attack upon its prey. Proc. Zool. Soc. Lond., 133 (2): CROWCROFT, P., 1955: Notes on the behaviour of shrews. Behaviour, 8: CROWCROFT, P., 1957: The life of the shrew. Max Reinhardt, London. DUPASQUIER, A. & CANTONI, D., 1992: Shifts in benthic macroinvertebrate community and food habits of the water shrew, Neomys fodiens (Soricidae, Insectivora). Acta Oecol., 13 (1): FORMANOWICZ, D. R. JR. & BRODIE, E. D. JR., 1979: Palatability and antipredator behaviour of selected Rana to the shrew Blarina. Am. Midl. Nat., 101: FORMOZOV, A. N., 1948: The small rodents and insectivores of Sharinskiy District of Kostroma Province in Fauna i Ekologija Gryzunov (MOIP, Moscow), 3: (in Russian). GERAETS, A., 1972: Aktivitätsmuster und Nahrungsbedarf bei Suncus etruscus. Bonn. Zool. Beitr., 23 (3): HABERL W., 1993: Zur Ökologie einheimischer Spitzmäuse (Soricidae, Insectivora) und ihres Lebensraumes am Beispiel eines Waldviertler Feuchtbiotops und experimentelle Bearbeitung ausgewählter ethologischer Fragestellungen. Dissertation, University of Vienna, Vienna. HABERL W., 1995: The Shrew Bibliography. A collection of more than 6000 references to research on the biology of the Soricidae (Insectivora, Mammalia) and small mammal ecology. I. CD-ROM Version, W. Haberl, Vienna. HABERL, W., 1998: A note on prey handling times and partial prey consumption in five species of European shrews (Soricidae, Insectivora). Pak. J. Biol. Sci., 1 (1): HABERL, W. & WILKINSON, J., 1997: A note on the Unkenreflex and similar defensive postures in Rana temporaria (Anura, Amphibia). Bull. Brit. Herpetol. Soc., 61: HUTTERER, R., 1976: Deskriptive und vergleichende Verhaltensstudien an der Zwergspitzmaus, Sorex minutus L., und der Waldspitzmaus, Sorex araneus L. (Soricidae Insectivora Mammalia). Dissertation, University of Vienna, Vienna. HUTTERER, R., 1982: Biologische und morphologische Beobachtungen an Alpenspitzmäusen (Sorex alpinus). Bonn. Zool. Beitr., 33 (1): ILLING, K., ILLING, R. & KRAFT, R., 1981: Freilandbeobachtungen zur Lebensweise und zum Revierverhalten der europäischen Wasserspitzmaus, Neomys fodiens (Pennant, 1771). Zool. Beitr., 27 (1):

101 KOSOI, M. E., 1982: An experimental study of eating carcasses of the same species in small mammals. Zool. Zhur., 61 (5): (in Russian, with a summary in English). KÖHLER, D., 1984: Zum Verhaltensinventar der Wasserspitzmaus (Neomys fodiens). Säugetierkundl. Informat., 2 (8): KRAFT, R., 1980: Fraßspuren der Europäischen Wasserspitzmaus, Neomys fodiens (Pennant, 1771), und der Waldspitzmaus, Sorex araneus Linné, 1758, an Gehäusen der Spitzschlammschnecke, Lymnaea stagnalis Linné, Säugetierkundl. Mitt., 28 (4): KRAFT, R. & PLEYER, G., 1978: Zur Ernährungsbiologie der Europäischen Wasserspitzmaus, Neomys fodiens (Pennant, 1771), an Fischteichen. Z. Säugetierkd., 43 (6): KUVIKOVA, A., 1985: Zur Nahrung der Wasserspitzmaus, Neomys fodiens (Pennant, 1771) in der Slowakei. Biológia (Bratislava), 40 (6): KUVIKOVA, A., 1986: Nahrung und Nahrungsansprüche der Alpenspitzmaus (Sorex alpinus, Mammalia, Soricidae) unter den Bedingungen der tschechoslowakischen Karpaten. Folia Zool., 35 (5): LARDET, J.-P., 1988: Spatial behaviour and activity patterns of the water shrew Neomys fodiens in the field. Acta Theriol., 33 (21): LORENZ, K., 1952: King Solomon s Ring. Methuen, London. MEZHZERIN, V. A., 1958: On the feeding habits of Sorex araneus and Sorex minutus. Zool. Zhur., 37: (in Russian). NIETHAMMER, J., 1977: Ein syntopes Vorkommen der Wasserspitzmäuse Neomys fodiens und N. anomalus. Z. Säugetierkd., 42: 1 6. NIETHAMMER, J., 1978: Weitere Beobachtungen über syntope Wasserspitzmäuse der Arten Neomys fodiens und N. anomalus. Z. Säugetierkd., 43: OGNEV, S. I., 1959: Säugetiere und ihre Welt. Akademie Verlag, Berlin. PERNETTA, J. C., 1976: A note on predation of the smooth newt, Triturus vulgaris by European water-shrew, Neomys fodiens bicolor. J. Zool., 179: PIERCE, G. J., 1987: Search paths of foraging common shrews Sorex araneus. Anim. Behav., 35: PUCEK, M., 1959: The effect of the venom of the European water shrew (Neomys fodiens fodiens Pennant) on certain experimental animals. Acta Theriol., 3 (6): RÖRIG, G., 1905: Über den Nahrungsverbrauch einer Spitzmaus. Arbeiten der Biologischen Abteilung für Landund Forstwirtschaft am Kaiserlichen Gesundheitsamte, 4: ROZMUS, T., 1961: Les observations sur la conquête de la proie par le Sorex araneus Linnaeus, Acta Theriol., 4 (14): RUTHARDT, M. & SCHRÖPFER, R., 1985: Zum Verhalten der Wasserspitzmaus Neomys fodiens (Pennant, 1771) unter Wasser. Z. Angewandte Zool., 72: SCHLOETH, R., 1980: Freilandbeobachtungen an der Wasserspitzmaus, Neomys fodiens (Pennant, 1771) im Schweizerischen Nationalpark. Rev. Suisse Zool., 87 (4): SCHLÜTER, A., 1980: Waldspitzmaus (Sorex araneus) und Wasserspitzmaus (Neomys fodiens) als Aasfresser im Winter. Säugetierkundl. Mitt., 40: SCHMIDT, E., 1986: Tödlicher Laichvorgang beim Moorfrosch (Rana arvalis). Beiträge zur Naturkunde Wetterau, 6 (2): SCHRÖPFER, R., 1985: Ufergebundenes Verhalten und Habitatselektion bei der Wasserspitzmaus Neomys fodiens (Pennant, 1771). Z. Angewandte Zool., 72: SERGEEV, V. E., 1973: Nourishing material of shrews (Soricidae) in the flood-plain of the river Ob in the forest zone of West Siberia. Izv. Sib. Otd. Akad. Nauk SSSR, Ser. Biol. Nauk, 5: (in Russian). SHULL, A. F., 1907: Habits of the short-tailed shrew, Blarina brevicauda (Say). Am. Nat., 41: SPANNHOF, L., 1952: Spitzmäuse. Die Neue Brehm Bücherei, Akademische Verlagsgesellschaft Geest & Portig K.-G., Leipzig. SPITZENBERGER, F., 1999: Neomys fodiens. In: Mitchell-Jones, A. J., Amori, G., Bogdanowicz, W., Kry tufek, B., Reijnders, P. H. J., Spitzenberger, F., Stubbe, M., Thissen, J. B. M., Vohralík, V. & Zima, J. (eds), The Atlas of European Mammals. T. & A. D. Poyser for the Societas Europaea Mammalogica, Academic Press, London: WILCKE, G., 1938: Freiland- und Gefangenschaftsbeobachtungen an Sorex araneus L. Z. Säugetierkd., 12: WOLK, K., 1976: The winter food of the European water-shrew. Acta Theriol., 21 (6):

102 Folia Zool. 51(2): (2002) Habitat segregation among the woodchat shrike, Lanius senator, the red-backed shrike, Lanius collurio, and the masked shrike, Lanius nubicus, in NE Greece Csaba MOSKÁT and Tibor István FUISZ Animal Ecology Research Group of the Hungarian Academy of Sciences (HAS), c/o Hungarian Natural History Museum, Baross u. 13., H-1088 Budapest, Hungary; Received 5 February 2001; Accepted 12 October 2001 A b s t r a c t We studied habitat selection of the woodchat Lanius senator, red-backed Lanius collurio, and masked Lanius nubicus shrikes in NE Greece, where they occur sympatrically. During the breeding season the masked shrike is most distinct, but woodchat and red-backed shrikes highly overlap in their habitat use. Multivariate discriminant analysis revealed the best separating variables from a set of 13 vegetation variables measured around perch sites. For axis one the best separating variables were identified as the woodland character, and the shrub character for axis two. By the help of the multivariate discriminant analysis the habitat selection in these avian species can be separated with high probability (64%, 64%, and 86% for the woodchat, red-backed and masked shrikes, respectively). Key words: shrikes, sympatry, habitat selection, multivariate analysis Introduction Although shrikes (Laniidae) are medium-sized songbirds, they hunt for large insects and small vertebrates, combining the insectivorous and carnivorous mode of feeding (C a d e 1995). In Europe most of the species show a severe decline and regional extinction in several countries (T ucker & Heath 1994, Yosef 1994, Y osef & Lohrer 1995, H a g e m e i j e r & B l a i r 1997). Extreme examples are those of the red-backed shrike Lanius collurio that disappeared from Britain in 1989 (Peakall 1995), and the lesser grey shrike L. minor, which is almost extinct in France (Lefranc 1997). In Europe there are five breeding shrike species, namely the red-backed shrike, the lesser grey shrike, the great grey shrike L. excubitor, the woodchat shrike L. senator, and the masked shrike L. nubicus. Additionally, the southern subspecies of the great grey shrike L. e. meridionalis, inhabits southern France and the Iberian peninsula, and is treated by certain authors as a separate species L. meridionalis (Lefranc 1995, Lefranc & Worfolk 1997). The coexistence of these small predators in the same geographical area is affected by habitat preference and interspecific competition. Generally, there is a diversity of factors that affects habitat use of avian species, and even ecologically similar species segregate by fine tuned differences in habitat use (C o d y 1985a). Earlier in the last century, in Europe there were more places where three or four shrike species occurred sympatrically, like the woodchat, red-backed, great grey and lesser grey shrikes in SE Gemany (U l l r i c h 1971). The latter three species still breed sympatrically in Slovakia (Kri tín, in litt.). In recent years the multispecies shrike assemblages tend to disappear because of the strong decline of some shrike species (Y osef 1994). At present NE Greece is almost the only area in Europe where 103

103 the breeding range of four shrike species, namely that of the woodchat, the red-backed, the lesser grey and the masked shrikes, overlap. This hot spot area for shrikes partly extends to the southern part of Bulgaria and northern-western Turkey. The masked shrike has only a limited range in Europe, and is concentrated in the north-eastern part of Greece, including a few islands in the Aegean Sea. It also occurs along the western cost of Turkey, and the island of Cyprus (Lefranc & Worfolk 1997), and extends to Israel (Inbar 1995). Although the Greek population has declined considerably, its population is stable in Bulgaria. However, the Bulgarian population is small, and contains no more than 100 pairs, but appears to be increasing (T ucker & Heath 1994). Unfortunately, no data are available whether these species coexist in the same breeding habitats, and the type of specialization on habitat structure that is to be found at the regional or the local spatial scales. The aim of the present study was to examine how the three sympatric shrike species breeding in NE Greece are separated by habitat requirements. We analyse how vegetation structure affects habitat occupancy pattern of the three shrike species. We wanted to determine which habitat variables separates them. Study Area and Methods The study was conducted in the surroundings of the town Alexandroupolis, NE Greece ( N, E), in the vicinity of the villages Avas, Lefkimi, Dadia, and Loutros during 6 and 13 June, This period coincides with the breeding season for all of the studied species. No fledglings were observed at that time, but several nests had welldeveloped chicks. Although some of the habitats in the study area seemed to be similar to the breeding habitat of the lesser grey shrike in central Europe (Kri tín 1995, K ri tín et al. 2000, Lovászi et al. 2000), we observed only one lesser grey shrike, so this species was excluded from the analysis. Therefore we sampled habitat parameters of the woodchat, red-backed and masked shrikes. The study area is characterised by hills (up to 300 m), covered mainly by bushy pastures, scrubs, and small woodlands. Riverine forests and agricultural areas occur in the valleys, but closed forests were found only at higher elevations. In the field we characterised each shrike site by the main habitat type. This is not easy because shrikes occupy habitats that exhibit relatively fine-grain patchiness at the scale of individual territories (Y osef & Grubb 1992), and they are often a mixture of two or more habitat types (e.g. cereal field and olive grove). All of the sites were recorded on video tape for later identification. We characterised six habitat types: (1) Forest edge: edge of broad-leaved forest (Quercus spp.), sometimes with scattered pines (Pinus spp.), with thorny shrubs, or with small patches of grassland; (2) Olive, almond, and walnut groves, or the mixture of them (generally old, but rarely young stands); (3) Cereals: cereal fields with hedges or sparse trees; (4) Bushy pasture and scrub (xerophytic): dense scrub on hillside, or small patches of thorny shrubs, occasionally Junipers, or scattered trees; (5) Bushy pasture and scrub (mesophytic): patches of shrubs with scattered trees, mainly in the valleys; (6) Riverine woodland: dense wooded vegetation with well-developed bushes along the streams in the valleys, occasionally with old Platanus trees. Following the guidelines for habitat description (James & Shugart 1970, Noon 1981), we measured 13 habitat variables in the field within a 25 m radius circle around the perch sites of the shrikes: percent foliage cover (%), mean height of trees (m), mean diameter of tree canopies (m), mean distance between tree canopies (m), mean diameter of 104

104 tree trunks at breast height (m), mean distance between tree trunks (m), percent bush cover (%), mean height of bushes (m), mean bush diameter (m), mean distance between bushes (m), percent grass cover (%), mean height of grass (m), percent bar ground cover (%). When tree or bush cover was 0% at a sampling circle, variables mean distance between tree canopies and mean distance between tree trunks, were set up as the diameter of the sampling circle (50m). The same value was given to variable mean distance between bushes if bush cover was 0%. By this recalculation we avoided giving the same value (zero) to two separate extreme cases: (i) when distances between canopies, tree trunks or bushes were zero because of the high density of trees or bushes, and (ii) when these variables were not measurable because of the lack of trees or bushes within the sampling circles. Variables deviating from the normal distribution were transformed (D unn 1981). All percentage variables were arcsine transformed (Za r 1984). Although we did not colour-ring the populations, we measured only one set of habitat variables within a territory in order to avoid pseudoreplication in the data set. Then we moved on until we found a new territory. Overlap in habitat selection was calculated by the Petraitis formula (P etraitis 1979). The main advantage of this likelihood measure is that a significance test is available for it (P etraitis 1985). The computer program SPOVRLAP was used for calculation (Ludwig & Reynolds 1988). The multivariate discriminant analysis (MDA) is a suitable technique for evaluating the best discriminator variables between the species and to test the separation among the populations within the canonical space (M orrison et al. 1997). Different variants of MDA have already been applied to compare niche partitioning of closely-related avian species (C ody 1985b), to compare habitat selection during migration and breeding (Moskát et al. 1993), or the habitat use of taxonomically different species, like the loggerhead shrike Lanius ludovicianus and the American kestrel Falco sparverius (Gawlik & Bildstein 1995). MDA was performed with SPSS/PC+ (N orusis 1990). Both discriminant function coefficients and discriminant loadings are used to explain the relationships between the discriminant functions and the original variables, but discriminant loadings offer a more appropriate approach for interpretation because of the potential multicollinearity between variables (Morrison et al. 1997, Sharma 1996). Results Altogether 117 samples on the vegetation structure around perch sites were collected. Woodchat shrikes and red-backed shrikes were common in pastures, but the masked shrike was found only in special habitats, like olive groves and orchards, in valleys with patchy but well-developed vegetation, and generally along streams. Vegetation samples were collected at 56 sites for woodchat shrikes, 33 for red-backed shrikes, and 28 for masked shrikes. None of the species were found in all of the six habitat categories, but the woodchat and the red-backed shrikes were observed in five habitats. The masked shrike showed the narrowest range and occurred in only four habitats (Table 1). More than half of the woodchat shrikes (53.6%) were found in open xerotherm bushy pasture or woodlands. This habitat was also preferred by the red-backed shrike (42.4%) in addition to the forest edge habitat (36.4%). No masked shrike occurred in the xerotherm open bushy pasture or woodland habitat, but preferred the olive grove and orchard habitat (53.6%), the mesophil 105

105 Table 1. Habitat distribution of the woodchat shrike, red-backed shrike and masked shrike in NE Greece (percentages and total number of observations). Habitat Woodchat Red-backed Masked Shrike Shrike Shrike Forest edge Olive grove or orchard Cereals Bushy pasture and woodland (xerotherm) Bushy pasture and woodland (mesofil) Riverine woodland N Table 2. Specific pairwise overlap (SO) of habitat distribution of the woodchat, red-backed, and masked shrikes in NE Greece, calculated by the Petraitis formula. Pairs of shrike species SO U df woodchat - red-backed ** 5 red-backed - woodchat * 5 woodchat - masked *** 5 masked - woodchat *** 5 red-backed - masked *** 5 masked - red-backed *** 5 * = P < 0.05, ** = P < 0.01, *** = P < Table 3. Wilks Lambda (U-statistic), univariate F-ratio with 2 and 114 degrees of freedom, standardized discriminant coefficients, and correlations between dicriminating variables and canonical discriminant functions (discriminant loadings) (PFC: percent foliage cover (%), MHT: mean height of trees (m), MDTC: mean diameter of tree canopies (m), MDBTC: mean distance between tree canopies (m), MDT: mean diameter of tree trunks at breast height (m), MDBTT: mean distance between tree trunks (m), PBC: percentage bush cover (%), MHB: mean height of bushes (m), MBD: mean bush diameter (m), MDBB: mean distance between bushes (m), PGC: percentage grass cover (%), MHG: mean height of grass (m), PBGC: percentage bare ground cover (%)). Habitat Wilks F Standardized discriminant Discriminant loadings variable Lambda coefficients Function 1 Function 2 Function 1 Function 2 PFC ** MHT *** MDTC *** MDBTC ** MDT *** MDBTT PBC *** MHB *** MBD ** MDBB * PGC *** MHG * PBGC * * P < 0.05, ** P < 0.01, *** P <

106 bushy pasture or woodland (21.4%), and the riverine woodland habitats (21.4%). Pairwise comparison of the three shrike species in habitat overlap revealed a higher similarity between the woodchat and red-backed shrikes (0.839 < SP < 0.848), and a much lower overlap with the masked shrike (0.000 < SP < 0.007; Table 2.), when the value of the general overlap (GO) was Although all of the differences in habitat occupancy of any pair of shrikes were statistically significant (P < 0.05 for all), the differences between the masked shrike and the woodchat or red-backed shrikes proved to be the widest (P < 0.001). Vegetation variables were tested by the univariate Wilks Lambda statistic. It tests the differences in the means of the three species for each habitat variable (Table 3.). Except for the variable mean distance between tree trunks, all of the variables measured showed significant differences among the three groups. Multiple discriminant analysis of the habitat variables grouped for the three shrike species revealed highly significant differences between the groups (Table 4). In the case of the first discriminant function the highest positive loadings were for the variables mean height of trees, mean diameter of tree canopies, mean diameter of trunks, percent foliage cover and the highest negative loadings were for percent grass cover and mean distance between tree canopies. In the case of the second discriminant function the highest positive loadings were found for the variables percentage bush cover, mean bush diameter, mean height of bushes, and the highest negative loading were get for mean distance between bushes. So we can interpret axis one as the woodland character, and axis two as the bush character. The classification probability (i.e. the probability that a habitat sample that belongs to a given shrike species can be classified correctly) was 86% for the masked shrike and 64% for both the woodchat and the redbacked shrikes (Table 5). Table 4. Statistical properties of the discriminant functions. Function 1 Function 2 Eigenvalue Percent of variance Cumulative pct of variance Canonical correlation Wilks Lambda Chi-square df Significance P < P < Table 5. Results of classification derived from discriminant analysis showing actual and predicted species (group) membership for woodchat shrike, red-backed shrike and masked shrike based on vegetation structural data around perch site within territories. Predicted group membership woodchat shrike red-backed shrike masked shrike woodchat shrike 64% 18% 18% red-backed shrike 21% 64% 15% masked shrike 7% 7% 86% 107

107 Discussion Segregation mechanisms on the local scale play an important role in the coexistence of related species. Including the much rarer lesser grey shrike, four Lanius species coexist in NE Greece, and this overlap is restricted to a limited area around Alexandroupolis. At present the masked shrike sparsely inhabits Greece and occurs only east of Asprovalta (H a n d r i n o s & Akriotis 1997). It shows a higher frequency in the Alexandroupolis area, but earlier it was a regular breeder along the east coast between Thessaloniki and Athens. On the Turkish side of the border these species also overlap locally. In the Alexandroupolis area we found that masked shrikes separated well from the woodchat and red-backed shrikes by habitat, using riverine and mesophil bushy woodlands and bushy edges of forests. We found a high overlap between the woodchat and the red-backed shrikes in the Alexandraoupolis area, but the woodchat shrike showed a tendency to use xerotherm bushy pastures, and did not avoid cereal fields when some bushes were present. The red-backed shrike showed a greater tendency to occupy forest edges, mesophil bushy pastures and woodlands. This is similar to the habitats in Central Europe where this species is found. In central Italy G uerrieri et al. (1995) studied the habitat selection of three coexisting shrike species, namely the woodchat shrike, the red-backed shrike, and the lesser grey shrike. The latter species was rare and was found in only 2.6% of the 10x10 km quadrates. Shrikes were segregated mainly by vegetation characteristics. The lesser grey shrike preferred habitats with scattered low shrubs and reduced arboreal components. The red-backed shrike occurred in each habitat type, but preferred edges of dense arboreal and shrubby formations. The woodchat shrike overlapped with the other shrike species, especially the red-backed shrike. Some altitudinal differences can be observed in the distribution of shrike species in Greece (Handrinos & Akriotis 1997). The woodchat shrike occurs mainly on the coastal plains, and in the lower foothills, up to 500 m a.s.l., but occasionally goes up to 1000 m. The red-backed shrike breeds mainly between m, but sometimes up to 1700 m. Locally, as in our study area, it is also found at sea level and in the lower hills. The lesser grey shrike is restricted to altitudes from sea level to 1000 m. We found most of the woodchat and redbacked shrikes in NE Greece at lower elevations, from the coastal plains up to 300 m, because xerotypic pastures mainly occur in the lower elevations. The masked shrike was found in the lower parts of the valleys, and the coastal lowlands in olive groves and orchards. We do not consider elevation as a separating factor among the shrike species, because the availability of suitable habitats changes with elevation. At higher elevations the closed forest is generally not suitable for shrikes. The red-backed shrike shows the greatest affinity to small open patches with bushes and young trees in the mountains. F uisz & Yosef (1997) showed the preference of post-breeding red-backed shrikes for forest edges and pastures in Hungary, but they did not find red-backed shrikes in closed broad-leaved forests. Our results explain this by higher preference of this species for denser vegetation than the woodchat shrike. On Mount Hermon in Israel the woodchat, red-backed, and masked shrikes, and also the southern form of the great grey shrike ( meridionalis group) occur sympatrically in the m elevational zone (I nbar 1995). The highest occurrence of shrikes is consistent with the distribution of bushes and tall trees. An obvious tendency for separation by habitat characteristics was described for the area. The great grey shrike preferred habitats with scattered thorny trees and bushes; the woodchat shrike inhabited areas with thorny bushes and shrubs, and also in sparse orchards; the red-backed shrike occurred in the steppe 108

108 forest with thorny bushes and trees in the elevational zone m, and also at lower elevations; and the masked shrike was restricted to lower elevations, mainly to 1000m. This species preferred cherry and apple orchards. On the local geographical scale, taxonomically similar species are separated by habitat. If they coexist in the same habitat, these species generally show divergent evolution, if not, they have become reproductively isolated (C ody 1973). In our study only the masked shrike showed clear separation by habitat. Territorial behaviour, differences in morphology, diet and hunting techniques may also be important. Availability of high quality perch sites also has an importance for site occupancy (Moskát et al. 2000). Although shrikes are interspecifically territorial (Cramp & Perrins 1993), several studies have reported no territorial interactions among different shrike species ( Inbar 1995). Sometimes shrikes spread sparsely over the area, having a population density below the potential carrying capacity of the habitat (T emple 1995). Although, in our study area at least, the woodchat and red-backed shrikes had generally neighbouring territories, we did not observe any interspecific conflict among any of the shrike pairs. However, one of the authors (C. M.) observed in northern Corfu a male woodchat shrike, which attacked and chased out of his territory an intruding lesser grey shrike. In the Kalahari desert, in Botswana, wintering lesser grey shrikes dominate overwintering red-backed shrikes in aggressive encounters (Herremans 1998), and also in the breeding ground in Slovakia (H. H oi and A. Kri tín pers. com.). We suppose that interspecific conflicts are more severe at the beginning of the breeding season when boundaries of the territories are being established. In northern Japan, the bull-headed shrike Lanius bucephalus occupies the optimal habitats before the arrival of the brown shrike Lanius cristatus (Takagi & Ogawa 1995). There are other types of segregation mechanisms working on the time scale; e.g. the great grey shrike is a winter resident in Hungary, but the lesser grey shrike is a breeding species. They can never be seen simultaneously (Keve 1984). R e c o m m e n d a t i o n s f o r c o n s e r v a t i o n Throughout Europe changes in land use and management seem to be the most important factors responsible for the decline of shrike species (T ucker & Heath 1994, Lefranc 1997). Multivariate analysis of the habitat structure proved to be helpful in finding the differences in the habitat selection on a fine-scale habitat level. Multiple discriminant analysis revealed the most important structural variables affecting habitat separation among the three sympatric shrike species. Most of the woodchat shrikes inhabited habitats containing bushy pastures, therefore sheep and goat grazing help to preserve these habitats from forest succession. The government in Greece subsidises this traditional farming method in order to maintain this habitat. Although this subvention is due to the black vulture Aegipius monachus population and for other raptor species in the surrounding of the WWF station in Dadia, this conservation management is also favourable for the woodchat shrike. The red-backed shrike population is not threatened in this area. It mainly breeds at forest edges and also in shrubby pastures, overlapping with the woodchat shrike. The masked shrike seems to be the most vulnerable shrike species in the area. As a considerable decline was recently shown in the species (T ucker & Heath 1994), nature conservation has to focus on this species in NE Greece. We do not expect that the masked shrike population can survive in the future without preserving old olive, almond and walnut groves, and additionally old Platanus trees along the streams. 109

109 A c k n o w l e d g e m e n t s. The authors are indebted to Revuen Y osef and two anonymous reviewers for their comments on the ms. Kostas P o i r a z i d e s (WWF, Dadia), and Skevis T h e o d o r o s (Theo Tours, Budapest) kindly helped to organise our field-work. We are grateful to Prof. R. Mann for improving the English. The study was supported by the grant No. T16510 of the Hungarian Scientific Research Fund (OTKA) to C. M. LITE R ATUR E CADE, T., J., 1995: Shrikes as predators. (Proc. of the First Intern. Shrike Symposium) Proc. West. Found. Vert. Zool., 6: 1 5. CODY, M. L., 1973: Parallel evolution in bird niches. In: di Castri, F. & Mooney, H. A. (eds), Mediterranean type ecosystems. Springer, Heidelberg: CODY, M. L., 1985a: An introduction to habitat selection in birds. In: Cody, M. L. (ed.), Habitat selection in birds. Academic Press, Orlando: CODY, M. L., 1985b: Habitat selection in the sylviine warblers of Western Europe and North Africa. In: Cody, M. L. (ed.), Habitat selection in birds. Academic Press, Orlando: CRAMP, S. & PERRINS, C. M., 1993: The birds of the western Palearctic. Vol. VII. Oxford University Press, Oxford. DUNN, J. E., 1981: Data-based transformations in multivariate analysis. In: Capen, D. E. (ed.), The use of multivariate techniques in studies of wildlife habitat. USDA Gen. Tech. Rep. RM-87, Rocky Mountain Forest and Range Exp. St., Forest Service, Fort Collins, CO: FUISZ, T. I. & YOSEF, R., 1997: Habitat choice of post-breeding Red-backed Shrikes (Lanius collurio) in northeastern Hungary. (Proc. of the Second Intern. Shrike Symposium) IBCE Technical Publications, 6: GAWLIK, D. & BILDSTEIN, K. L., 1995: Differential habitat use by sympatric Loggerhead Shrikes and American Kestrels in South Carolina. (Proc. of the First Intern. Shrike Symposium) Proc. West. Found. Vert. Zool., 6: GUERRIERI, G., PIETRELLI, L. & BIONDI, M., 1995: Status and reproductive habitat selection of three species of Shrikes, Lanius collurio, L. senator, and L. minor, in a mediterranean area. (Proc. of the First Intern. Shrike Symposium) Found. Vert. Zool., 6: HAGEMEIJER, W. J. M. & BLAIR, M. J. (eds), 1997: The EBCC atlas of European breeding birds: their distribution and abundance. T & A D Poyser, London. HANDRINOS, G. & AKRIOTIS, T., 1997: The birds of Greece. Christopher Helm, London. HERREMANS, M., 1998: Habitat segregation of male and female Red-backed Shrikes Lanius collurio and Lesser Grey Shrikes Lanius minor in the Kalahari basin, Botswana. J. Avian Biol., 28: INBAR, R., 1995: Shrikes nesting on Mount Hermon, Israel. (Proc. of the First Intern. Shrike Symposium) Proc. West. Found. Vert. Zool., 6: JAMES, F. C. & SHUGART, H. H., 1970: A quantitative description of avian habitat. Audubon Field-Notes, 24: KEVE, A., 1984: Nomenclator Avium Hungariae. Akadémiai Kiadó, Budapest (in Hungarian and German). KRI TÍN, A., 1995: Why the Lesser Grey Shrike (Lanius minor) survives in Slovakia: Food and habitat preferences, breeding biology. Folia Zool., 44: KRI TÍN, A., HOI, H., VALERA, F. & HOI, C., 2000: Breeding biology and breeding success of the Lesser Grey Shrike Lanius minor in a stable and dense population. Ibis, 142: LEFRANC, N., Identification des pies-griéches grises de France et du Paléarctique occidental. Ornithos, 2: LEFRANC, N., 1997: Shrikes and the farmed landscape in France. In: Pain, D. J. & Pienkowski, M. W. (eds), Farming and birds in Europe: the Common agricultural policy and its implementations for bird conservation. Academic Press, London: LEFRANC, N. & WORFOLK, T., 1997: Shrikes. A guide to the shrikes of the world. Pica Press, Sussex, U.K. LOVÁSZI, P., BÁRTOL, I. & MOSKÁT, C., Nest site selection and breeding success of the lesser grey shrike (Lanius minor) in Hungary. (Proc. of the Third Intern. Shrike Symposium) Ring, 22:

110 LUDWIG, J. A. & REYNOLDS, J. F., 1988: Statistical ecology. Wiley, New York. MORRISON, M. L., MARCOT, B. G. & MANNAN, R. W., 1997: Wildlife-habitat relationships. Concepts and applications. The Univ. of Wisconsin Press, Madison. MOSKÁT, C., BÁLDI, A. & WALICZKY, Z., 1993: Habitat selection of breeding and migrating icterine warblers Hippolais icterina: a multivariate study. Ecography, 16: MOSKÁT, C., FUJIMAKI, Y. & YAMAGISHI, S., 2000: Perch site preference of the Bull-headed Shrike (Lanius bucephalus) during the breeding season in Japan. (Proc. of the Third Intern. Shrike Symposium) Ring, 22: NOON, B. H., 1981: Techniques for sampling avian habitats. In: Capen, D. E. (ed.), The use of multivariate techniques in studies of wildlife habitat. USDA Gen. Tech. Rep. RM-87, Rocky Mountain Forest and Range Exp. St., Forest Service, Fort Collins, CO: NORUSIS, M. J., 1990: SPSS/PC+. Advanced Statistics 4.0 for the IBM PC/XT/AT. SPSS Inc., Chicago. PEAKALL, D. B., 1995: The decline and fall of the Red-backed Shrike in Britain. (Proc. of the First Intern. Shrike Symposium) Proc. West. Found. Vert. Zool., 6: PETRAITIS, P. S., 1979: Likelihood measures of niche breadth and overlap. Ecology, 60: PETRAITIS, P. S., 1985: The relationship between likelihood niche measures and replicated tests for goodness of fit. Ecology, 66: SHARMA, S., 1996: Applied multivariate techniques. Wiley, New York. TAKAGI, M. & OGAWA, I., 1995: Comparative studies on nest sites and diet of Lanius bucephalus and L. cristatus in Northern Japan. (Proc. of the First Intern. Shrike Symposium) Proc. West. Found. Vert. Zool., 6: TEMPLE, S. A., 1995: When and where are shrike populations limited? (Proc. of the First Intern. Shrike Symposium) Proc. West. Found. Vert. Zool., 6: TUCKER, G. M. & HEATH, M. F., 1994: Birds in Europe: their conservation status. BirdLife International (Birdlife Conservation Series No. 3.) Cambridge, U.K. ULLRICH, B., 1971: Untersuchungen zur Ethologie und Ökologie des Rotkopfwürgers (Lanius senator) in Südwestdeutchland im Vergleich zu Raubwürger (Lanius excubitor), Schwarzstirnwürger (Lanius minor) un Neuntöter (Lanius collurio). Vogelwarte, 26: YOSEF, R., 1994: Evaluation of the global decline in the true shrikes (family Laniidae). Auk, 111: YOSEF, R. & GRUBB, T. C., Jr., 1992: Territory size influences nutritional condition in nonbreeding Loggerhead Shrikes (Lanius ludovicianus): A ptilochronology approach. Conserv. Biol., 6: YOSEF, R. & LOHRER, F. (eds.), 1995: Shrikes (Laniidae) of the world: biology and conservation. (Proc. of the First Intern. Shrike Symposium) Proc. Wetern Foundation of Vertebrate Zoology, 6: ZAR, J. H., 1984: Biostatistical analysis. Prentice Hall, Englewood Cliffs, New Jersey. 111

111 Folia Zool. 51(2): (2002) Effects of age composition of pairs and weather condition on the breeding performance of tawny owls, Strix aluco Lajos SASVÁRI 1 and Zoltán HEGYI 2 1 Department of General Zoology, Eötvös University, Pázmány Péter sétány 1/C VI. em, H-1117 Budapest, Hungary; sas5890@mail.iif.hu 2 Management of Duna-Ipoly National Park, Hıvösvölgyi út 52, Hungary Received 28 May 2001; Accepted 12 October 2001 Abstract. Nest boxes for breeding tawny owls Strix aluco were sited in a mixed oak/hornbeam/beech forest located in the Pilis Biosphere Reserve, 30 km northwest of Budapest, Hungary, during the period We hypothesized that both within-pair differences in age composition and weather conditions affect the breeding success of the parents. To test our hypothesis, we marked the owl parents in their first known breeding year and related their reproductive performance was to within-pair age composition through three subsequent breeding seasons. Sex-related age differences affected the breeding performance of the parents in their first and second known breeding year: number of eggs and hatching success were influenced by the age of females, while fledging success was influenced by the age of males. Within-pair age differences did not affect the breeding performance at their third known breeding year. Reproductive performance was lower in snowy years than in years without snow cover in all of their three subsequent breeding seasons. On basis of the lower mean body mass and shorter mean tarsi of the nestlings before fledging recorded in adverse weather conditions, we suggest that parents favour the sex that demands less investment in care. As a consequence, the sex-ratio becomes skewed towards the male in a poor food environment. Key words: fledging success, hatching success, owl, parental age, reproductive performance, Strix, weather condition Introduction Studies of lifetime reproductive success on long-lived bird species have shown that young parents have a lower breeding performance than older birds (Ollason & Dunnet 1988, Scott 1988, Bradley et al. 1989, N e wton & Rothery 1997). Young parents invest less in reproduction than do more mature birds, which may be due to the young birds poorer condition or their inexperience in foraging and knowledge of the habitat (Orians 1969, R ec h er & R ec h er 1969, Pugesek & Diem 1983, Thomas & Coulson 1988, W e i m e r s k i r c h 1992). Long-term studies on owls have also revealed age dependent effects on breeding performance (K orpimäki 1988, Gehlbach 1989, Saurola 1989). The aim of our recent study was to examine the age dependent reproductive success of tawny owls Strix aluco which maintained their pair bond in subsequent breeding periods. Tawny owl is a resident strongly monogamous owl species that can be attracted to artificial nest boxes. We marked individual parents and determined the age composition of the pairs in their first known breeding years. We focused our study on the question: do within-pair age differences influence the breeding performance of the parents? Nevertheless we also 113

112 hypothesized that weather conditions affect the reproductive success of all pairs whether young or old. S outhern & Lowe (1968) suggested that food availability for owl species preying on small mammals is determined by ground cover and papers have reported that owls changed to secondary foods, when snow prevents them from hunting rodents (Goszczynski 1981, M ikkola 1983, K irk 1992). We thus examined the breeding performance achieved by the pairs in snowy years and in years when snow did not cover the ground during the incubation and early nestling periods. In addition, we recorded the body mass and length of tarsus of nestlings before fledging and related these measures to the number of breeding seasons of the parents and weather effects. It has been suggested that, if the different sex of offspring require different care in a brood, the young which demand more parental effort should be limited in a poor food environment in order to reduce the total reproductive cost to the parents (S lagsvold 1990, Clutton Brock 1991, W e atherhead & Teather 1991). On the basis of the sexual dimorphism in tawny owls we presumed that the lighter offspring, which demand less parental investment, survive better during brood reduction. In order to examine the reduction of reproductive cost by elimination of heavier offspring, we measured the body mass of nestlings before fledging and related this to hatching order. Methods One hundred and eighty nest-boxes for breeding tawny owls were sited in a mixed oak/hornbeam/beech forest, with year old trees, in the Pilis Biosphere Reserve, 30 km north-west of Budapest (47 35 N; E) during the period Six to eight nestboxes were grouped together with m between them, the groups being separated by 2 5 km. Nest-boxes were checked at 4 8 day intervals from the first week in February. Ninety-eight females and 85 males were captured during the nestling period by placing a net over the entrance of the boxes while the birds were inside, and the birds were marked with different combinations of coloured rings to ensure individual identification. From these, 39 pairs were identified which bred through three subsequent breeding seasons, which did not divorce in the following breeding season, and did not lose their nests through predation. These pairs were chosen for our study of the relationship between the reproductive performance, parental age composition and weather conditions. Age determination was based on the pattern of the primaries and secondaries (P e t t y 1992), and four categories were distinguished for the age composition of the pairs in their first known breeding season: (1) parents were two years old or less; (2) male was more than two years old, female was two years old or less; (3) female was more than two years old, male was two years old or less; (4) parents were more than two years old. As a consequence of the 500 m altitude range of the study area (from m), snow covered the ground longer in the higher nest areas than the lower ones. A year when snow completely covered the ground for a radius of 1km from the owl nest during the incubation and early nestling period (until the 10 days after the hatching of the last young) was considered as a snowy breeding year for any given pair. Statistical analyses were carried out using the SPSS statistical package. Multiple comparisons calculated by GT2-method were used for testing the relationships between breeding performance and within-pair age composition of the parents. Hatching success and fledging success were calculated by number of eggs hatched per number of eggs laid and number of nestlings fledged per number of nestlings hatched, respectively. 114

113 Results Breeding performance of the pairs of various age composition Within-pair age differences influenced both the number of eggs and hatching success in the first and second breeding seasons of the parents (first: F 3,35 = 4.22, P = and F 3,35 = 3.36, P = 0.027; second: F 3,35 = 3.94, P = and F 3,35 = 3.69, P = 0.029; Table 1.) but effects were not found in their third breeding season (F 3,35 = 2.34, P = and F 3,35 = 2.09, P = 0.215). Multiple comparisons showed that pairs where females were more than two years old produced more eggs with higher hatching success than pairs where females were two years old or less in their first and second known breeding year (Table 2). Within-pair age differences also influenced fledging success in the first and second breeding season (first: F 3,35 = 6.24, P < 0.001; second: F 3,35 = 3.36, P < 0.001), but effects were not recorded in the third breeding season (F 3,35 = 2.07, P = 0.229). Multiple comparisons supported the idea that pairs where males were more than two years old reached higher fledging success than pairs where males were two years old or younger. More fledglings were raised by the pairs where males were more than two years old in the first and second breeding year (F 3,35 = 4.89, P < and F 3,35 = 5.23, P = 0.002), but no differences were recorded between pairs according to their within-pair age composition in their third breeding season (F 3,35 = 2.66, P = 0.072). Parents produced more fledglings in their second known breeding year than in their first (t = 2.89, P < 0.01) and raised more young in their third breeding year than in their second (t = 3.16, P < 0.01). Relationships between the age composition and weather condition influencing the breeding performance Parents laid fewer eggs in years when snow covered the ground than in years when snow did not cover the ground in their first (t = 4.53, P < 0.001), second (t = 3.28, P < 0.01) and third breeding season (t = 2.18, P < 0.05), but differences were not recorded for hatching success Table 1. Number of eggs, hatching success, fledging success and number of fledglings produced by owl parents of various age composition in their first second and third breeding years. In parentheses, number of pairs. Age composition Breeding Number of Hatching Fledging Number of Years eggs success success fledglings Parents two years First 2.15± ± ± ±0.46 old or less than Second 2.77± ± ± ±0.45 two years old (13) Third 3.31± ± ± ±0.66 Male more than First 2.34± ± ± ±0.83 two years old, Second 3.22± ± ± ±0.92 female two years Third 3.56± ± ± ±0.83 old or less (9) Female more than First 2.89± ± ± ±0.46 two years old, Second 3.33± ± ± ±0.76 male two years Third 3.44± ± ± ±0.83 old or less (9) Parents older First 3.13± ± ± ±0.49 than two Second 3.75± ± ± ±0.48 years old (8) Third 3.88± ± ± ±

114 Table 2. Multiple comparisons (GT2-method) among the number of eggs, hatching success and fledging success produced by pairs of various age composition. Numbers in parentheses indicate the four classes of age composition. Upper numbers denote the number of eggs, hatching success and fledging success in the first, lower numbers denote the number of eggs, hatching success and fledging success in the third breeding year. Asterisks indicate significant differences at the 0.05 level. Values calculated for second breeding year are not presented because the significant differences showed the same relationships in first and second breeding seasons. Minimal significant differences (MSD) Number of eggs Hatching success Fledging success (1) (2) (3) (4) (1) (2) (3) (4) (1) (2) (3) (4) Parents two years old or less than (1) two years old Male more than (2) * Differences two years old, in absolute female two years valueold or less between the groups Female more than two years old, 0.75 * 0.39 * * 0.07 * * male two years (3) old or less Parents older than 0.94 * 0.58 * * 0.09 * * * -- two years old (4) Table 3. Number of eggs, hatching success, fledging success and number of fledglings produced in snowy years and in years without snow cover. In parentheses, number of breeding pairs. Breeding Snow cover Number of Hatching Fledging Number of years on the ground eggs success success fledglings First Yes (16) 1.94± ± ± ±0.75 No (23) 3.00± ± ± ±0.88 Second Yes (12) 2.67± ± ± ±0.75 No (27) 3.41± ± ± ±0.88 Third Yes (14) 3.36± ± ± ±0.39 No (25) 3.60± ± ± ±0.46 (t = 0.16, t = 0.54, t = 0.10, n.s. in all occasions; Table 3). Pairs achieved lower fledging success and fledged fewer nestlings in snowy years than in years without snow cover in all breeding seasons (first: t = 2.87, P < 0.01 and t = 3.24, P < 0.01; second: t = 3.70, P < and t = 2.45, P < 0.01; third: t = 2.70, P < 0.02 and t = 2.59, P < 0.02). Three-way ANOVA shows significant differences in the breeding performance for the relationships between within-pair age composition and breeding years (Table 4). Within-pair age composition affected the number of eggs, hatching and fledging success in the first and second breeding seasons, but did not influence the performance of the parents in later breeding years. Significant differences in breeding performance were found for the relationships between within-pair age composition and weather effects, except during the incubation. i.e. pairs laid fewer eggs and achieved lower fledging success in all age categories when snow covered the ground but hatching success was not influenced by the 116

115 Table 4. Three-way ANOVA for the relationships between the breeding performance and within-pair age composition, number of breeding years and weather effects. Within-pair age composition was distinguished into four categories: (1) parents two years old or less; (2) male more than two years old, female two years old or less; (3) female more than two years old, male two years old or less; (4) parents more than two years old. Breeding years were categorized for three subsequent breeding seasons. Weather effects were distinguished into two categories: snow covered the ground and snow did not cover the ground during the incubation period and early nestling period (until the age of 10 days old of last-hatched nestlings). Sample size: 117 broods. Source of variation d.f. F P Number of eggs <0.001 Within-pair age composition x Breeding years Hatching success <0.001 Fledging success Number of eggs Within-pair age composition x Weather effects Hatching success Fledging success Number of eggs Breeding years x Weather effects Hatching success Fledging success Within-pair age composition x Breeding years Number of eggs x Weather effects Hatching success Fledging success Table 5. Mean body mass (g±sd) and mean length of tarsus (mm±sd) of nestlings before fledging per brood. n = number of broods. Number of breeding years Snow cover n Body mass Length of tarsus First Yes ± ±2.5 No ± ±3.1 Second Yes ± ±2.1 No ± ±2.3 Third Yes ± ±2.8 No ± ±2.7 relationship between weather and within-pair age differences. Also, differences in numbers of eggs and fledging success were recorded for the relationships between breeding years and weather effects. Pairs achieved higher performance in their third than in their first and second breeding years, but lower performance in snowy years than in years without snow cover in all breeding seasons. These relationships did not affect hatching success. Body mass and length of tarsus of nestlings before fledging We recorded 137 nestlings before fledging and 42 nestlings which died in 61 broods. Mean body mass of nestlings per brood was heavier and mean tarsal length of nestling per brood was longer in the second broods of the parents than in the first (t = 2.45, P < 0.02 and t = 2.64, P < 0.02) and nestlings were heavier with longer tarsal lengths in the third than in second breeding years (t = 2.20, P < 0.05 and t = 2.76, P < 0.02; Table 5). Parents fledged nestlings with higher mean body mass and longer mean tarsal length per brood in years without snow cover than in snowy 117

116 years both in their first (t = 2.90, P < 0.01 and t = 2.60, P < 0.02), second (t = 3.97, P<0.001 and t = 4.02, P < 0.001) and third breeding seasons (t =5.37, P<0.001 and t = 3.39, P < 0.01). Discussion Our findings on the relationships between parental age and reproductive performance may be summarized as follows. (1) Reproductive performance of pairs increased through successive breeding years. (2) Sex-related age-differences influenced the breeding performance of pairs in the first and second known breeding years, the within-pair age differences did not affect in the third breeding season. (3) Age of females affected the number of eggs and hatching success (older females laid more eggs with higher hatching success) while age of the males affected fledging success (older males were associated with higher fledging success). Low fecundity of young females compared to older females has been documented both in passerines (P e rrins 1979, D hondt 1989, N e wton 1989) and non-passerines (Newton 1989, S a et h er 1990, Sydeman et al. 1991), and the small clutch size with low hatching success of tawny owl females may be due also to a low fecundity in their first and second breeding years. Male owls feed the females during the incubation period, but surprisingly egg survival was not influenced by male age. The high rate of nestling mortality recorded in the ten days after hatching suggests that nestlings are most at risk during the brooding season. Both females and nestlings are supplied with food by the males during the brooding period and the higher fledging success achieved by older males in their first and second breeding year is due to their more efficient food provision. The differences in number of offspring fledged reflected also the effects of age of the males: pairs raised more offspring where the males were old than pairs where the males were young. Longitudinal studies on bird species have shown that clutch size, hatching success and fledging success increased with age to a plateau and decreased among the oldest breeders (Newton et al. 1983, T homas 1983, N isbet et al. 1984, N e wton 1989, Pugesek & Wood 1992). We have not yet conducted such long term studies of the tawny owl that we can examine the possible effects of senescence, and we followed only the period of their life span where success rates are increasing. It has been found both in passerines (H a r v e y et al. 1988) and non-passerines (S c o t t 1988) that experience of breeding, irrespective of age, increased breeding success, and we suggest that maintainance of the pair-bond between females and males and the breeding experience acquired together lead the parents of various ages to the highest level of the reproductive performance after the same number of breeding years. When the common breeding experience resulted in high reproductive success in their third breeding season, the effects of within-pair age differences ceased for breeding. Predators may respond to fluctuations in the abundance of prey numerically or functionally (S o l o m o n 1949). The numerical responses are due to reproductive success, survival and immigration/emigration rate, whereas the functional responses are reflected in a change in prey composition (A n d e r s s o n & E r l i n g e 1977). Studies have shown numerical responses in nocturnal raptors: the productivity of owls increased in a rich food environment but decreased when the prey declined both in boreal and temperate regions in northern latitudes (A d a m c i k et al. 1978, 1979, K o r p i m ä k i & N o r r d a h l 1991, J e d r z e j e w s k i et al. 1994). In this study we also found that pairs even with common breeding experience of many years achieved lower reproductive success in adverse weather 118

117 conditions. The analyses of diet composition of tawny owls showed that males delivered more birds for the females and their offspring when the availability of small mammals was low because of snow cover (S a s v á r i et al. 2000). The older males were better than young males at providing alternative prey. However, both the young and old males delivered fewer food items with lower food mass during the incubation and early nestling period in snowy years than in years when snow did not cover the ground. Neither the inexperienced nor experienced males were able to fully compensate by secondary prey types for the loss of primary food. The lower mean body mass and mean shorter tarsi of the nestlings in a brood raised in snowy years may be due to the food shortage and the preference of the parents which favours the sex requiring less investment. Adult females are heavier and larger than adult males in tawny owls, and on the basis of the findings that, by the time of fledging, dimorphism in size and body mass is usually as pronounced as adulthood (N e w t o n & M a r q u i s 1979, F i a l a 1981, R i c h t e r 1983, B o r t o l o t t i 1986). Hence we may suppose that the differences in body mass and tarsal length measured on nestlings before fledging reflected the dimorphism of offspring. If one sex requires proportionally more food than the other then, when food resources are scarce, this sex should experience increased mortality and the sex-ratio becomes skewed towards the less costly sex (S l a g s vol d 1990, W e a t h e r h e a d & T e a t h e r 1991, C l u t t o n - B r o c k 1991). We did not identify the sex of the owlets but, on basis of body mass and length of tarsi, we may assume the sex ratio to be biased in favour of the cheaper male. The reduction in effort maintains the parents in better condition and promotes their survival and future reproductive performance. Acknowledgements We are indebted to Ian N e w t o n for his valuable comments and suggestions. We are grateful to Susan T o t t e r d e l l, Department of Pharmacology, University of Oxford, for her linguistic corrections. This work was supported by the Department of General Zoology of Eötvös University, and Hungarian National Foundation for Scientific Research (project number: T ). We are thankful to István H a h n, Department of Plant Taxonomy and Ecology, Eötvös University, Budapest, for his help in the statistical analyses. We also grateful to the Frank M. C h a p m a n Memorial Fund, The American Museum of Natural History, New York, for its financial support. LITE R ATUR E ADAMCIK, R. S., TODD, A. W. & KEITH, L. B., 1978: Demographic and dietary responses of Great Horned Owls during a snowshoe hare fluctuation. Can. Field Nat., 92: ADAMCIK, R. S., TODD, A. W. & KEITH, L. B., 1979: Demographic and dietary responses of Red-tailed Hawks during a snowshoe hare fluctuation. Can. Field Nat., 93: ANDERSSON, M. & ERLINGE, S., 1977: Influence of predation on rodent populations. Oikos, 29: BORTOLOTTI, G. R., 1986: Influence of sibling competition on nestling sex ratios of sexually dimorphic birds. Am. Nat., 127: BRADLEY, J. S., WOOLLER, R. D., SHIRA, I. J. & SERVENTY, D. L., 1989: Age-dependent survival of breeding Short-tailed Shearwaters Puffinus tenuirostris. J. Anim. Ecol., 58: CLUTTON-BROCK, T. H., 1991: The evolution of parental care. Princeton University Press, Princeton. DHONDT, A. A., 1989: The effect of old age on the reproduction of Great Tits Parus major and Blue Tits P. caeruleus. Ibis, 131: FIALA, K. L., 1981: Reproductive costs and the sex ratio in red-winged blackbirds. In: Alexander, R.D. & Tinkle, D.W. (eds), Natural Selection and Social Behavior. Chiron Press, New York: GEHLBACH, F. R., 1989: Screech-Owl. In: Newton, I. (ed.), Lifetime reproduction in birds. Academic Press, London: GOSZCZYNSKI, J., 1981: Comparative analysis of food of owls in agrocenoses. Ekol. Pol., 29: HARVEY, P. H., STENNING, M. J. & CAMPBELL, B., 1988: Factors influencing reproductive success in the Pied Flycatcher. In: Clutton-Brock, T. H. (ed.), Reproductive success. Chicago University Press, Illinois, USA:

118 JEDRZEJEWSKI, W., JEDRZEJEWSKA, B., ZUB, K., RUPRECHT A. I. & BYSTROWSKI, C., 1994: Resource use by Tawny Owls Strix aluco in relation to rodent fluctuation in Bialowieza National Park, Poland. J. Avian Biol., 25: KIRK, D. A., 1992: Diet change in breeding tawny owls (Strix aluco). J. Raptor. Res., 26: KORPIMÄKI, E., 1988: Effects of age on breeding performance of Tengmalm s Owl Aegolius funereus in western Finland. Ornis Scand., 19: KORPIMÄKI, E. & NORRDAHL, K., 1991: Numerical and functional responses of kestrels, short-eared owls, and long-eared owls to vole densities. Ecology, 72: MIKKOLA, H., 1983: Owls of Europe. T. & A. D. Poyser, Calton. NEWTON, I. 1989: Lifetime reproduction in birds. Academic Press, London. NEWTON, I. & MARQUISS, M., 1979: Sex ratio among nestlings of the European sparrow-hawk. Am. Nat., 113: NEWTON, I., MARQUISS, M. & ROTHERY P., 1983: Age structure and survival in a Sparrowhawk population. J. Anim. Ecol., 52: NEWTON, I. & ROTHERY, P., 1997: Senescence and reproductive value in sparrowhawks. Ecology, 78: NISBET, I. C., WINCHELL, J. M. & HEISE, A. E., 1984: Influence of age on the breeding biology of Common Terns. Colon. Waterb., 7: OLLASON, J. C. & DUNNETT, G. M., 1988: Variations in breeding success in Fulmars. In: Clutton-Brock, T. H. (ed.), Reproductive success. Chicago University Press, Chicago, Illinois, USA: ORIANS, G. H., 1969: Age and hunting success in the Brown Pelican (Pelecanus occidentalis). Anim. Behav., 17: PERRINS, C. M., 1979: British tits. Collins, London, England. PETTY, S. J., 1992: A guide to age determination of Tawny Owl Strix aluco. In Galbraith, C., Taylor, A. I. R. & Percival, S. (eds), The ecological and conservation of European owls. Joint Nature Conservation Committee, Peterborough: PUGESEK, B. H. & DIEM, K. L., 1983: A multivariate study of the relationship of parental age to reproductive success in California gulls. Ecology, 64: PUGESEK, B. H. & WOOD, P., 1992: Alternative reproductive strategies in the California Gull. Evolut. Ecol., 6: RECHER, H.F. & RECHER, J. A., 1969: Comparative foraging efficiency of adult and immature Little Blue Herons (Florida caerulea). Anim. Behav., 17: RICHTER, W., 1983: Balanced sex ratios in dimorphic altricial birds: The contribution of sex-specific growth dynamics. Am. Nat., 121: SAETHER, B. E., 1990: Age-specific variation in reproductive performance of birds. Current Ornithology, 7: SASVÁRI, L., HEGYI, Z., CSÖRGÃ, T. & HAHN, I., 2000: Age-related diet change, parental care and reproductive cost in tawny owls Strix aluco L. Acta Oecol., 21: SAUROLA, P., 1989: Ural Owl. In: Newton, I. (ed.), Lifetime reproduction in birds. Academic Press, London: SCOTT, D. H., 1988: Reproductive success in Bewick s Swans. In: Clutton-Brock, T. H. (ed.), Reproductive success. Chicago University Press, Chicago, Illinois, USA: SLAGSVOLD, T., 1990: Fisher s sex-ratio theory may explain hatching patterns in birds. Evolution, 44: SOLOMON, M. E., 1949: The natural control of animal population. J. Anim. Ecol., 18: SOUTHERN, H. N. & LOWE, V. P. W., 1968: The pattern of distribution of prey and predation in tawny owl territories. J. Anim. Ecol., 37: SYDEMAN, W. J., PENNIMAN, J. F., PENNIMAN, T. M., PYLE, P. & AINLEY, D. G., 1991: Breeding performance in the western gull: effects of parental age, timing of breeding and year in relation to food availability. J. Anim. Ecol., 60: THOMAS, C. S., 1983: The relationship between breeding experience, egg volume, and reproductive success of the Kittiwake Rissa tridactyla. Ibis, 125: THOMAS, C. S. & COULSON, J. C., 1988: Reproductive success of Kittiwake Rissa tridactyla. In: Clutton- Brock, T. H. (ed.), Reproductive success. Chicago University Press, Chicago, Illinois, USA: WEATHERHEAD, P. J. & TEATHER, K. L., 1991: Are skewed fledgling sex-ratios in sexually dimorphic birds adaptive? Am. Nat., 138: WEIMERSKIRCH, H., 1992: Reproductive effort in long-lived birds: age. specific patterns of condition, reproduction and survival in the Wandering Albatross. Oikos, 64:

119 Folia Zool. 51(2): (2002) Effects of habitat structure, floral composition and diversity on a forest bird community in north-western Italy Paola LAIOLO Università di Torino, Dipartimento di Biologia Animale e dell Uomo, Via Accademia Albertina 17, Turin, Italy; laiolo@dba.unito.it Received 2 February 2001; Accepted 26 November 2001 A b s t r a c t During the breeding period, I analysed bird community-habitat relationships in a managed broad-leaved forest of north-western Italy (a newly established nature reserve). Species richness, diversity, biomass and abundance of some ecological groups of birds were analysed with respect to habitat variables, summarising habitat structure, canopy and understorey floral composition. Two major patterns of relationships among bird and habitat were traced: the first involved changes in tree structure during their growth (canopy height and diameter of the dominant tree), the second was related to characteristics associated with floral richness. Bird diversity, species richness, the amount of hole nesters, of trunk and ground feeders were positively associated with stands age. The abundance of some groups of birds was positively related to plant species richness: understorey species richness influenced shrub feeders, shrub nesters and edge species; canopy species richness affected trunk feeders. Canopy species richness also affected bird diversity, richness, biomass, abundance of hole nesters, trunk feeders and forest interior species. Key words: diversity, species richness, broad-leaved woodland Introduction The effect of spatial configuration, productivity and diversity of vegetation on bird communities has been widely documented (C o d y 1985, W i e n s 1989). Much of the current information concerning bird assemblages derives from studies on temperate-forest birds (J a m e s 1971, F u l l e r 1995). These forests offer a wide variety of habitats that differ in extent, physical structure, and food and nest availability to birds over both space and time (C o d y 1985). Aspects of the environment that influence distributional patterns in forest birds include the presence or availability of foraging resources, which in turn are affected by the diversity and composition of the habitat. Preferred nesting, roosting or perching sites are also affected by forest configuration. Elevation, sensitivity to hunting or other human activities and interaction among bird species may also influence the distribution of birds among habitats. The availability of resources is by far the better predictor of the structure of forest birds assemblages, but it is hard to quantify (B ellamy et al. 2000). Hence, some general features of habitat that are more easily characterised have been used to analyse bird-habitat relationships. Traditionally, the emphasis in studies of bird habitats has been on features relating to the structural configuration or floral composition of the habitat (W i e n s 1989). The structure of a woodland can be broken down into a large number of components that may be used as explanatory variables of birds community structure. The stage of forest growth exerts a strong affect on bird assemblages. Bird species richness tends to increase with stand age ( Lack 1939, Lack & Lack 1951, M oss 1978, H elle & Mönkkönen 1990), although many species prefer a particular stage of growth. Most hole-nesters select those stages with mature trees, while many migrant passerines are confined to the earliest stages when the vegetation is open (F u l l e r 1995). A prominent 121

120 influence on bird communities is played by the vertical foliage profile (S o l o n e n 1996, F r e i f e l d 1999). Woodland birds differ considerably in the heights at which they forage and nest; the canopy, the understorey and the field layers all have their own species. It is widely accepted that increased structural complexity generally provides a wider diversity of foraging sites upon which birds can specialise, allowing the addition of more species in existing assemblages (K arr 1990). The availability of tree holes is fundamental for woodland birds, and can result in severe competition among hole-nesters. Another important micro-structural feature for forest birds is dead wood: the huge invertebrate community it supports is a rich food resource for many birds (N ilsson 1979). M acarthur & MacArthur (1961) have suggested that the structural variables mentioned above are important in explaining variations in woodland bird communities, conversely, tree species composition or diversity are often irrelevant. However, several studies have demonstrated that floral composition and plant diversity may be important determinants of habitat use in birds (H olmes & Robinson 1981, Robinson & Holmes 1984). In Europe, species exhibit differences in habitat use, foraging behaviour and even morphology as a function of leaf type (K arr 1990). Since floral diversity strongly influences physical structure (Horn 1971), it is difficult to separate the effects of these two factors. In the present study I analysed bird-habitat relationships in a managed broad-leaved forest of north-western Italy during the breeding season. In this area, a protected reserve was recently established. Bird species diversity, richness and biomass were analysed in relation to measured features of habitat structure, floral composition and diversity. Since the analysis of habitat relationships is an important part of wildlife management and conservation biology, I also analysed how the variations in the abundance of ecological groups were associated with certain habitat characteristics. Material and Methods The study was carried out in the Val Sarmassa Nature Reserve (44 48 N, 8 20 E; north-western Italy). The reserve, ranging from 140 to 270 m a.s.l., comprises mostly broad-leaved woods, but scattered farmland is present. Human impact has always been strong in the study area, as woods have been harvested and managed since the XVII century. The reserve mainly consists of a mosaic of small coppice patches at different growth stages. Some coppice is grown with scattered single-stem trees (standards), which are fallen on a longer rotation time than underwood. The predominant tree species were sweet chestnuts (Castanea sativa), oaks (Quercus petraea, Q. cerris, Q. pubescens and Q. robur) and locust trees (Robinia pseudoacacia). In the study area young locust tree woods (established in many abandoned vineyards), small clearances, poplar plantations (in the valleys) and vineyards (on the top of the hills) also occur. I surveyed species presence using a single visit point-count method (Blondel et al. 1970, Blondel 1975, Järvinen & Lokki 1978). This technique involved a count of all birds seen or heard inside or outside a 50 m radius circular plot; overflying birds that did not land in trees or on the ground were recorded but not included in the analysis. Each point was sampled for 10 minutes between sunrise and 3hr later, since this morning period yields the best results ( L ynch 1995). Field work was carried out from 5 March to 28 June 2000, as census performances for counts from the beginning to the end of the breeding season proved to be significantly better than counts that were concentrated in either one of these periods (D rapeau et al. 1999). Overall, 95 survey plots were distributed through a wide range of environmental conditions within the forest. 122

121 In each census station, habitat data were collected. Measurements were made after each session of point counts was completed, in plots centred in the avian census plots (50 m radius). I measured the following abiotic variables: (1) ground slope, (2) elevation (m above see level) and (3) distance from the forest edge (m). The structure of vegetation was quantified by measuring: (1) mean diameter at breast height (DBH) of the dominant tree species (the closest four canopy trees of the dominant species were located, one within each quartile: north, south, east and west; DBH was measured directly with a tape measure); (2) mean canopy height (measured on the same trees chosen for DBH, a hypsometer was used), (3) mean understorey height (the closest four shrubs were located, one within each quartile: north, south, east and west), (4) mean canopy density (the distance among the four closest trees was paced in metres; the inverse of distance gave the measure of density), (5) understorey density (the distance among the four closest shrubs was paced; gave the measure of density), (6) percentage of trees with climbers (such as ivy Hedera helix). The abundance of canopy species and understorey species was expressed as percentage of cover. Canopy species richness and understorey species richness were also calculated. In the interest of clarity, I used the term canopy to denote trees higher than 8 m, and understorey to point to shrub layer (as synonymous of midstorey in North American literature). I calculated bird species diversity (Shannon s index), richness and total biomass (i.e. the sum of birds weights) at each individual points. Bird weights were those reported in Cramp (1988, 1992) and C ramp & Perrins (1993, 1994a, 1994b). To avoid including birds that were outside the habitat sampling plots, I used detections within the 50 m radius plot only, in keeping with Pagen et al. (2000), Sallabanks et al. (2000). Bird abundance was expressed as percentage of occurrence, i.e. no. of times a species occurred in the point-counts divided by the total number of point-counts. Principal component analysis (PCA, G a u n c h 1984) was chosen to reveal patterns in the data for structure, using standardised data (zero mean and unit standard deviation). The paucity of data did not allow a detailed analysis at the species-level, so species were grouped according to their ecological group: edge species, forest interior species, hole nesters, tree nesters, understorey nesters, ground feeders, trunk and branches surface feeders, shrub feeders, tree feeders and generalist feeders ( Fuller 1995, J okimäki & Huhta 1996). Bird species diversity, species richness, total biomass and the abundance of ecological groups were used as dependent variables in stepwise multiple regression analyses. The variables used as predictors were the three abiotic variables, canopy species richness, understorey species richness, as well as PCA scores for components one and two for habitat structure. Significance levels were Bonferroni corrected. Associations among bird and plant or understorey species were examined using canonical correspondence analysis (CCA, T e r Braak 1986). Despite CCA is unable itself to test hypotheses, the resulting plot is a compromise solution expressing the association between species, variables and samples (T e r Braak & Vendonschot 1995). I included in the analysis only bird species that were spotted at least four times. The relative abundance of the three commonest tree species (chestnut, locust tree and oaks) and of the four commonest understorey species (elder Sambucus nigra, hazel Corylus avellana, hawthorn Crataegus monogyna, blackthorn Prunus spinosa) were used as environmental variables. The results of CCA can be presented in a diagram containing the environmental variables (here, plant and understorey species composition) plotted as arrows emanating from the center of the graph, along with points for the bird taxa (Fig. 1). 123

122 Fig. 1. Relationship between bird species and the commonest plant species, as resulted from canonical correspondence analysis. The arrows representing plant species indicate the direction of maximum change of those species across the diagram. For bird species acronyms see Appendix. Table 1. Results of principal component analysis carried out on habitat structure data in the Val Sarmassa Nature Reserve. The highest loadings are given in bold type. Factor loadings PC1 PC2 DBH canopy height understorey height canopy density understorey density percentage of trees with climbers eigenvalue % total variance Results The first two principal component (PC1, PC2) accounted for almost 56% of the total variation in the habitat structure matrix (eigenvalues > 1) (Table 1). Canopy height and diameter at breast height of the dominant trees (DBH) showed the highest correlation with PC1 scores, which could therefore be considered as accounting for forest age. The percentage of climbers and understorey density provided the major loading on PC2, which thus accounted for structural complexity. Thirthy species of landbirds were recorded in 95 individual point-counts in the forest (Appendix). In percentage of occurrence, robin was the most frequently encountered bird, followed by great tit, blackcap and blackbird. In total numbers, robin was again the most common bird, followed by blackcap and great tit. In the forest, the mean diversity per point count was 0.5; a mean of 3.9 species was recorded. Both bird diversity and species richness 124

123 were positively correlated with PC1 (forest age) and with canopy species richness; all the other variables were excluded from the model (Table 2). Bird biomass was positively affected only by canopy species richness (Table 2). The abundance of hole nesters, trunk and branches surface feeders and interior species was positively correlated with PC1 (forest age) and with canopy species richness Trunk feeders were also negatively correlated with PC2 (structural complexity, also related to climber presence). Shrub feeders abundance was positively related with shrub species richness and negatively related with the distance from the forest edge. Only canopy species richness affected tree feeders, whereas no structural variable significantly affected edge species, tree nesters and generalists. Understorey nesters were related to understorey species richness. The first two CCA axes accounted for 53.4 of the variability in the data (CA1: 31.3 %, CA2: 22.1%). Chaffinch, woodpigeon, blue tit, and great spotted woodpecker appear associated with oak- or chestnut-rich plots of increasing proportion. Blackcap, long-tailed tit and robin abundance increased when blackthorn was present; wren and chiffchaff seemed to prefer hazel rich stations, whilst turtle dove appeared to use hawthorn understorey. Table 2. Results of stepwise multiple regression analyses. The response variables were bird species diversity, bird species richness, the abundance of ecological groups of birds and bird biomass (g in weight); the explanatory variables (independent variables) were abiotic and structural parameters. Only the parameters that significantly entered in the model are listed. PC1 of structural variables could be considered a forest age factor, while PC2 of structural variables was related to structural complexity. Data were Bonferroni corrected. Variable Coefficient SE t-value P R 2 F P Species diversity Intercept < F 2,92 = 14.5 <0.05 PC1 of structural variables <0.05 Canopy species richness <0.05 Species richness Intercept < F 2,92 = 13.3 <0.05 PC1 of structural variables <0.05 Canopy species richness <0.05 Interior species Intercept < F 2,92 = 16.4 <0.05 Canopy species richness <0.01 PC1 of structural variables <0.01 Hole nesters Intercept F 2,92 = 10.5 <0.05 PC1 of structural variables <0.05 Canopy species richness <0.05 Understorey nesters Intercept < F 1,93 = 5.3 <0.05 Understorey species richness <0.05 Trunk and branches surface feeders Intercept F 3,91 = 7.6 <0.05 PC1 of structural variables <0.05 PC2 of structural variables <0.05 Canopy species richness <0.05 Ground feeders Intercept < F 1,93 = 6.9 <0.05 PC1 structural variables <0.05 Shrub feeders Intercept F 2,92 = 9.01 <0.05 Understorey species richness <0.01 Distance from the forest edge <0.05 Tree feeders Intercept F 1,93 = 8.26 <0.05 Canopy species richness <0.05 Total biomass Intercept F 1,93 = 13 <0.01 Canopy species richness <

124 Appendix. Bird species recorded during point-counts. Birds were allocated to different ecological groups, defining location (E = edge species, W = interior species), nest habit (H = hole nester, TR= tree nester, CP = understorey nester) and foraging guild (C = mostly tree feeder; G = mostly ground feeder, T =mostly trunk and branches surface feeder; S = mostly shrub feeder; F = generalist feeder). Categorisation is based on H a r r i s o n (1988), F u l l e r (1995) and Jokimäki & Huhta (1996). Biomass in keeping with Cramp (1988, 1992) and Cramp & Perrins (1993, 1994a, 1994b). Acronyms are given for species included in canonical correspondence analysis. Species Category Biomass % % total (g) occurrence numbers woodpigeon (Columba palumbus) WP W, TR, C turtle dove (Streptopelia turtur) TD E, CP, G hoopoe (Upupa epops) E, H, G wryneck (Jynx torquilla) W, H, G green woodpecker (Picus viridis) W, H, G great spotted woodpecker (Dendrocopos major) GW W, H, T wren (Troglodytes troglodytes) WR W, CP, S robin (Erithacus rubecola) RO W, CP, G redstart (Phoenicurus phoenicurus) E, H, G blackbird (Turdus merula) BB W, CP, G song thrush (Turdus philomelos) ST W, CP, G mistle thrush (Turdus viscivorus) MT E, TR, G blackcap (Sylvia atricapilla) BC E, CP, S wood warbler (Phylloscopus sibilatrix) W, CP, C chiffchaff (Phylloscopus collybita) CC E, CP, C spotted flycatcher (Muscicapa striata) E, TR, C long-tailed tit (Aeghitalos caudatus) LT E, CP, G marsh tit (Parus palustris) MTI W, H, C blue tit (Parus coeruleus) BT W, H, C great tit (Parus major) GT W, H, C nuthatch (Sitta europaea) W, H, T short-toed treecreeper (Certhia brachydactyla) TC W, H, T jay (Garrulus glandarius) JY W, TR, F hooded crow (Corvus corone cornix) E, TR, F starling (Sturnus vulgaris) E, TR, F chaffinch (Fringilla coelebs) CF E, TR, G greenfinch (Carduelis chloris) E, TR, G goldfinch (Carduelis carduelis) E, TR, G hawfinch (Coccothraustes coccothraustes) HF W, TR, C cirl bunting (Emberiza cirlus) E, CP, G Discussion Both woodland structure and floral diversity influenced the broad attributes of the bird community studied, although the relative importance of structural and vegetation composition differed among ecological groups of birds. Tree height and diameter (i.e. stand age) had a weak but constant and significant effect in most multiple regression analyses. Bird diversity, species richness, and the amount of hole nesters, interior species, trunk and ground feeders were all positively related to with stand age. In many European forests, the number of species and overall density of birds increase with forest maturation ( Lack 1939, L ack & Lack 1951, M oss 1978, Helle & Mönkkönen 1990). Stand age is positively correlated with tree volume and, in turn, with productivity of forest areas, thus its effect on bird communities (especially on interior species) can be very strong ( Jokimäki & Huhta 1996). Thus, hole nester abundance increased significantly with stand age, as these species are often confined to the oldest stands of a wood because of the greater availability of holes and crevices in mature 126

125 trees (S mith et al.1985). Trunk and branch feeders density also peaked in old stands, where invertebrate food in decaying trunks or branches is greater than young-growth. Older trees also have a greater surface area of bark to forage on, and the bark has a greater structural complexity ( Nilsson 1979, S chieck et al. 1995). Trunk feeders were negatively influenced by the density of climbers, which can hide the potential food resources on trunks and therefore have an adverse effect on feeding activities. Climbers may also make it more difficult for trunk feeders to move up and down the trunk. Trunk feeders appeared to be negatively affected by understorey density, but the reason for this remains unknown. Besides the covariation between stand age and bird abundance, another major factor affecting birds was floral diversity. Bird species diversity, richness and biomass decreased in single species woods. Most pure woods in the study site consisted of coppices of sweet chestnut or young woods of locust tree. Chestnut coppice may be deserted by species that require a vigorous shrub layer, as chestnut is typically grown in dense monoculture, which rapidly shades out the low foliage (Fuller 1995). Young locust tree wood has few holes, thus it could be avoided by hole nesters. Furthermore, there is possibly less insect food available in locust-tree woods than in mixed or pure oakwoods ( Fuller 1995). The number of shrub feeders was related significantly to understorey species richness, whereas that of trunk feeders was mainly influenced by canopy richness. Again, this may be explained by the greater diversity of food, or stability of food supply (in the form of insects or berries), found in woodlands of varied vegetation composition. Different understorey nesters appeared to be associated with specific shrub species: wren and chiffchaff, for instance, preferred hazel-tree, turtle dove favoured hawthorn, whereas blackcap, robin and long-tailed tit used blackthorn understorey. Habitat structure and plant heterogeneity should thus be considered in future forest management and nature reserve creation plans. In my study area, coppice management created a large variety of habitat types, as different growth stages are present. The bird community in the study site is clearly related to this habitat. Considering that edge effects also exist between forest compartments of different ages (far from the actual woodland edges), it is not surprising that most species recorded can be classified as edge species ( F u l l e r 1995, J o k i m ä k i & H u h t a 1996). The high proportion of young-aged forest with a rich understorey present in the study area would certainly benefit many habitat generalists, understorey nesters and edge birds. However, we must consider that these species are already very abundant and need no special assistance. Conversely, species that require large areas of integral forest habitat, such as interior species, hole nesters and virgin forest birds (e.g. woodpeckers), are of particular concern. Since this study emphasised the importance of stand age for many ecological groups of birds, it would be prudent to retain adequate amounts of mature trees in the managed landscape. Acnowledgements This research is part of a three year project on the bird communities in the Nature Reserves of Val Sarmassa and Rocchetta Tanaro, funded by the Ente Parchi Astigiani. I am grateful to G. Baldizzone, G. Miroglio and to all the staff of Val Sarmassa Nature Reserve for encouraging this research. Special thanks are due to two anonymous referees and to G. C o p p for improving the manuscript. LITERATURE BELLAMY, P.E., ROTHERY, P., HINSLEY, S. A. & NEWTON, I., 2000: Variation in the relationship between numbers of breeding pairs and woodland area for passerines in a fragmented habitat. Ecography, 23: BLONDEL, J., 1975: L analyse des peuplements d oiseaux, éléments d un diagnostic écologique. I. La méthode des échantillonages fréquentiels progressifs (E.F.P.). Terre et Vie, 29 :

126 BLONDEL, J., FERRY, C. & FROCHOT, B., 1970: La méthode des indices ponctuels (I.P.A.) ou des relevés d avifaune par «station d écoute». Alauda, 38 : CODY, M. L., 1995: Habitat selection in Birds. Academic Press, London. CRAMP, S., 1988: The birds of the western Palearctic. Vol. 5. Oxford University Press, Oxford. CRAMP, S., 1992: The birds of the western Palearctic. Vol. 6. Oxford University Press, Oxford. CRAMP, S. & PERRINS, C. M., 1993: The birds of the western Palearctic. Vol. 7. Oxford University Press, Oxford. CRAMP, S. & PERRINS, C. M., 1994a: The birds of the western Palearctic. Vol. 8. Oxford University Press, Oxford. CRAMP, S. & PERRINS, C. M., 1994b: The birds of the western Palearctic. Vol. 9. Oxford University Press, Oxford. DRAPEAU, P., LEDUC, A. & MCNEIL, R., 1999 : Refining the use of point counts at the scale of individual points in studies of bird-habitat relationships. J. Avian Biol., 30: FREIFELD, H. B., 1999: Habitat relationships of forest birds on Tutuila Island, American Samoa. J. Biogeogr., 26: FULLER, R. J., 1995: Bird life of woodland and forest. Cambridge University Press, Cambridge. GAUNCH, H.G. jr., 1984: Multivariate analysis in community ecology. Cambridge University Press, Cambridge. HARRISON, C., 1988: Nests, eggs and nestlings of British and European birds. William Collins Sons & Co., London. HELLE, P. & MÖNKKÖNEN, M., 1990: Forest succession and bird communities: theoretical aspects and practical implications. In: Keast, A. (ed.), Biogeography and Ecology of Forest Bird Communities. SPB Academic Publishing, The Hague: HOLMES, R. T. & ROBINSON, S. K., 1981: Tree species preferences of foraging insectivorous birds in a northern hardwoods forest. Oecologia, 48: HORN, H. S., 1971: The adaptive geometry of trees. Monographs in Population Biology 3, Princeton University Press, Princeton. JAMES, F. C., 1971 : Ordinations of habitat relationships among breeding birds. Wilson Bull., 83: JÄRVINEN, O. & LOKKI, J., 1978: Indices of community structure in bird censuses based on single visit: effects of variation in species efficiency. Ornis Scand., 9: JOKIMÄKI, J. & HUHTA, E., 1996: Effects of landscape matrix and habitat structure on a bird community in Northern Finland: a multi-scale approach. Ornis Fenn. 73: KARR, J. R., 1990: Interactions between forest birds and their habitats: a comparative synthesis. In: Keast, A. (ed.), Biogeography and Ecology of Forest Bird Communities. SPB Academic Publishing, The Hague: LACK, D. & LACK, E., 1951: Further changes in birdlife caused by afforestation. J. Anim. Ecol., 20: LACK, D., 1939: Further changes in the Breckland avifauna caused by afforestation. J. Anim. Ecol., 8: LYNCH, J. F., 1995: Effects of point count duration, time-of-day, and aural stimuli on detectability of migratory and resident bird species in Quintana Roo, Mexico. In: Ralph, C. J., Sauer, S. R. & Droege, S. (eds), Monitoring bird populations by point counts. Pacific SW For. Res. Stn., Gentile, Tech. Rep, PSW-GTR-149. MACARTHUR, R. H. & MACARTHUR, J. W., 1961: On bird species diversity. Ecology, 42: MOSS, D., 1978: Diversity of woodland song-bird populations. J. Anim. Ecol., 47: NILSSON, S. G., 1979: Density and species richness of some forest bird community in south Sweden. Oikos, 33: PAGEN, R. W., THOMPSON III, F. R. & BURHANS, D. E., 2000 : Breeding and post-breeding habitat use by forest migrant songbirds in the Missoury Ozarks. Condor, 102: ROBINSON, S. K. & HOLMES, R. T., 1984: Effects of plant species and foliage structure on the foraging behavior of forest birds. Auk, 101: SALLABANKS, R., WALTERS, J. R. & COLLAZO, J. A., 2000: Breeding bird abundance in bottomland hardwood forests: habitat, edge, and patch size effect. Condor, 102: SCHIECK, J., NIETFELD, M. & STELFOX, J. B., 1995: Differences in bird species richness and abundance among three succesional stages of aspen-dominates boreal forests. Can. J. Zool., 73: SMITH, K.W., AVERIS, B. & MARTIN, B., 1985: The breeding bird community of oak plantations in the forest of Dean, southern England. Acta Oecol./Oecol. Gener., 8: SOLONEN, T., 1996: Patterns and variations in the structure of forest bird communities in southern Finland. Ornis Fenn., 73: TER BRAAK, C.J.F., 1986: Canononical Correspondence Analysis: a new eigenvector technique for multivariate direct gradient analysis. Ecology, 67: TER BRAAK, C.J.F. & VERDONSCHOT, P.F.M., 1995: Canonical correspondence analysis and related multivariate methods in aquatic ecology. Aquat. Sci., 57: WIENS, J. A., 1989: The Ecology of Bird Communities. Vol. I. Cambridge University Press, Cambridge. 128

127 Folia Zool. 51(2): (2002) Sexual dimorphism in fire-bellied toads Bombina spp. from the central Balkans Jelena M. RADOJIâIå 1, Dragana D. CVETKOVIå 1*, Ljiljana M. TOMOVIå 1, Georg V. DÎUKIå 2 and Milo L. KALEZIå 1 1 Institute of Zoology, Faculty of Biology, University of Belgrade, Studentski trg 16, Belgrade, Yugoslavia; dragana@bf.bio.bg.ac.yu 2 Institute for Biological Research Sini a Stankoviç, 29. Novembra 142, Belgrade, Yugoslavia Received 29 May 2001; Accepted 3 January 2002 A b s t r a c t The direction and level of sexual size dimorphism (SSD) was examined in firebellied toads from the central Balkans. Samples were taken from 12 populations: three Bombina bombina, three B. variegata variegata and six B. variegata scabra populations. Intersexual variation of 20 morphometric characters was analysed. In addition, correspondence analysis of eight qualitative characters was performed. The results showed that, though body length had an inconsistent pattern of intersexual variation across taxa, other traits contributed to statistically significant level of SSD. The analysis on multivariate level revealed a highly significant effect of population in all three taxa, and significant effect of sex on character variation. Discriminant analysis confirmed a higher level of intersexual differences in B. variegata compared to B. bombina. Correspondence analysis showed that females and males were similar with respect to qualitative traits in all three taxa. At the univariate level, the most prominent features were: significant differences in tibia length in all three taxa, and highly significant differences in head width in B. bombina and in humerus length in B. v. scabra. These results are discussed with respect to specific reproductive behaviour and possible ecological differences between sexes. Key words: morphometric characters, sexual size dimorphism, Bombina spp., Balkans Introduction Sexual dimorphism (SD) is a widespread phenomenon. In numerous species, pronounced differences between the sexes are found in various characteristics of morphology (body size and shape), coloration, ornaments, etc. (e.g. Darwin 1871, Anderson 1994, Halliday & Tejedo 1995). A special case of sexual dimorphism, sexual size dimorphism (SSD), is defined as any statistically significant difference in the mean length or weight of sexually mature organisms from the same population during a given time interval ( L o v i c h & Gibbons 1992). Since Darwin (1871) developed the concept of sexual selection, it has most often been used to explain the observed differences between sexes. Darwin emphasized the distinction between sexual and natural selection, pointing out that they can have opposite directions. Sexual selection can operate through active choice of mates, through differential success in achieving dominance in direct combat or in contests where strength or agility are most important, in competition to attract females or in sperm competition (Halliday 1983, Anderson 1994, Reynolds & Harvey 1994). Apart from sexual selection, other mechanisms can produce sexual size dimorphism, such as ecological divergence between sexes, i.e. differential resource utilization (S h i n e * Corresponding author 129

128 1989). For instance, as a result of difference in dietary preferences (different types of food or feeding rates) sexes may diverge in head dimensions or total body size. Fecundity selection is also one of the possible explanations. In many animal species, females are larger than males. The general assumption is that selection favors larger females because they can produce more offspring. Positive correlation between fecundity and body size was found in a number of species, but other life-history traits, such as different age at maturation or sexspecific mortality, may have more influence on SSD (Madsen & Shine 1994). Sexual size dimorphism has been demonstrated in a great variety of species, invertebrates as well as vertebrates (e.g. Clutton-Brock et al. 1977, Price 1984, Cheverud et al. 1985, F airbairn 1990). In amphibians in general, sexual dimorphism includes a diverse array of characteristics size and shape differences, dorsal crests, mating calls, etc. (Duellman & Trueb 1986). Many amphibian species exhibit SSD. In a review of published data, S h i n e (1979) estimated that females are larger than males in approximately 61% of urodeles (of 79 species reviewed) and 90% of 589 anuran species reviewed. The opposite trend was found mainly among species reported to have aggressive behaviour and male combat (19% of urodele and 5% of anuran species). Though these figures were criticized (H alliday & Verrell 1986) on the grounds that heterogeneity of sources and types of data hindered precise estimates, it seems that in amphibians, like in many other poikilothermic vertebrates, the general pattern is that females are larger than males (e.g. H a l l i d a y & V e r r e l l 1986, H a l l iday & Tejedo 1995). In amphibians and reptiles, the usual explanation for intersexual differences in body size was sexual selection. Thus, S hine (1979) proposed that, in species with male combat, selection is supposed to favor larger males because they are more successful in intrasexual struggles. However, numerous factors and their interactions may influence the total SSD, such as: differences in growth rates (e.g. Cvetkoviç et al. 1997), delayed maturation in one sex (M a d s e n & S hine 1994) or sex-specific mortality, correlated response to selection (F a i r b a i r n 1997), characteristics of mating system including the operational sex-ratio that may change with time and duration of reproduction period (H a l l i d a y & Tejedo 1995). It has been emphasized (Halliday & Verrell 1986, Madsen & Shine 1994) that the direction of sexual size dimorphism (i.e. which sex is larger) depends on the relative advantage of larger body size in which sex is larger body size selectively more advantageous. For instance, if large body size gives more advantage to females (through greater fecundity) than to males (through greater reproductive success), females will be the larger sex. In addition, small body size in males may be favored in intrasexual contests where agility is more important than strength; in both sexes, small body size is advantageous in terms of earlier maturation and shorter generation time (A nderson 1994). Thus, the effects of selection in both sexes should be considered (Reynolds & Harvey 1994), as well as correlated size changes in both sexes (e.g. F airbairn 1997). European fire-bellied toads (Bombina spp.) were subjects of numerous studies (e.g. Arntzen 1978, G ollmann 1984, S zymura 1993, S zymura & Barton 1986, 1991, M accallum et al. 1998, S zymura et al. 2000), especially those concerning interspecific hybridization and hybrid zones. Diverse aspects of variation at intra- and interspecific level were intensively studied, but sexual dimorphism in Bombina populations has attracted little attention so far. Though SSD was examined in anurans in 130

129 general, the majority of data concerns species of Rana, Bufo and Hyla (e.g. H oward 1981, W oolbright 1983, S ullivan 1984, Halliday & Verrell 1986, Emerson 1994). The aim of this study was to examine the direction and magnitude of intersexual differences in two species of European fire-bellied toads, Bombina bombina and B. variegata. We analysed populations from the central Balkans, the region where both species occur, and where, additionally, two subspecies of B. variegata (B. v. variegata and B. v. scabra) are found. Fig. 1. Map of sampling sites in FR Yugoslavia and Former Yugoslav Republic of Macedonia. B. bombina: 1. Melenci, 2. Panãevo, 3. Banatska Palanka; B. v. variegata: 4. Andrevlje, 5. Kamenolom, 6. Cvetanovac; B. v. scabra: 7. Vasiljev Vrh, 8. Bjelo i, 9. Prekornica, 10. Livari, 11. Prohor Pãinjski, 12. Podgorci. 131

130 Material and Methods Analysis were conducted on 12 Bombina spp. populations: three Bombina bombina, three B. v. variegata and six B. v. scabra populations from the central Balkans (Fig. 1). Three Bombina bombina samples were collected from ponds near Melenci (n = 65), Panãevo (n = 30) and from Deliblato Sand, locality Banatska Palanka (n = 18). Three samples of B. v. variegata were from Fru ka Gora Mt. (loc. Andrevlje, n = 28, and Kamenolom n = 21) and from Miroã Mt. (loc. Cvetanovac, n = 28). B. v. scabra populations were from: Javor Mt. (Vasiljev vrh, n = 48), Prekornica Mt. (pond Ponikvica, n = 32), Lovçen Mt. (Bjelo i village, n = 31), the Skadar Lake (Livari village, n = 20), Prohor Pãinjski (n = 46) and Starac Mt. (loc. Podgorci, n = 21). All individuals included in analyses were adults. In order to examine morphological variation, we measured the following 20 morphological characters to the nearest 0.1 mm: L body length (from the tip of snout to the edge of cloaca), F femur length, T tibia length, N distance from tibiotarsal ankle to the tip of the longest toe, P foot length (from the metatarsal ankle to the tip of the longest toe), H humerus length, M forearm length (to the tip of the longest finger), DPPA length of the first finger of fore leg, DSPA length of the second finger of fore leg, DPPP length of the first toe of hind leg, CINT metatarsal tubercle length, LC head length (from the tip of snout to the edge of jaw), LTC maximum head width (at the point of jaw articulation), SPP minimal interorbital distance, SPI distance between nasal pores, SPCR snout width (at the front edge of orbits), LO eye length, LTP maximum eyelid width, DRO snout-eye distance, and DNO distance between nasal pore and eye. For bilateral characters, measurements on the right side of body were taken. Eight qualitative characters, mostly defined on the basis of previous references (Radovanoviç 1951, Micha owski & Madej 1969, L ang 1988) were examined as well: I pigmentation of dorsal gland structures, II appearance of dorsal warts, III colour of ventral side of body, IV colour of patches on ventral side, V presence of white spots on ventral side, VI presence of neck (i.e. slight constriction between head and body), VII colour of dorsal side of body and VIII pigmentation of tips of fingers. The results of gel electrophoresis were used to confirm the subspecies status of different B. variegata populations; B. v. variegata and B. v. scabra showed substantial differences in allelic frequencies on Idh2 and Ldh2 loci (unpublished data). Statistical analyses of morphometric characters included: descriptive statistics, One-way and Two-way analysis of variance (ANOVA) as well as multivariate analysis of variance (MANOVA) and discriminant analysis. Mahalanobis multivariate distances (D 2 ) between sexes were also calculated. Correspondence analysis was performed to examine differences in qualitative characters between sexes. Results Preliminary analyses (Appendix) of 20 morphological traits in female and male Bombina bombina, B. v. variegata and B. v. scabra, revealed a somewhat inconsistent pattern of intersexual size differences. Total length (L), the trait most commonly used as indicator of body size, showed higher mean values in B. bombina males than in females, and the opposite trend in both B. variegata groups, though none of these differences was statistically 132

131 significant. When other traits are considered, males were larger in all cases where significant intersexual differences were found. This raises the question of differences between sexes at the multivariate level. The results of One-way MANOVA showed significant SSD in all three taxa (Rao s R: Bombina bombina = 1.69, p < 0.05; B. v. variegata = 2.75, p < 0.01; B. v. scabra = 7.86, p<0.001). Since the sample was heterogeneous in a geographic sense, Two-way ANOVA with sex as fixed and locality as random factor was performed, in order to separate the effects of sex from the effects of locality (i.e. population) on morphological variation. The results are given in Table 1. The effect of sex was significant for three traits in B. bombina (T, LTC, SPI), two traits in B. v. variegata (L,T) and seven traits in B. v. scabra (T, N, P, H, SPCR, DRO, DNO). Factor population showed a strong effect in all three taxa, being highly significant in B. bombina and B. v. scabra. The effect of interaction was, with almost no exception, insignificant. The most prominent features were: significant differences in Tin all three taxa, and highly significant difference in traits LTC in B. bombina and H in B. v. scabra. Multivariate analysis (Two-way MANOVA, Table 2) showed highly significant effects of population and insignificant effects of interaction in all three taxa, and the growing significance of the effect of sex on character variation in B. bombina, B. v. variegata and B. v. scabra, respectively. Discriminant analysis confirmed higher level of intersexual differences in B. variegata compared to Bombina bombina. Discriminant scores centroids (Fig. 2) show that, while species and subspecies are delimited along the first and the second discriminant axis, sexes are clearly separated along the third discriminant axis. The same general pattern is observed when the population centroids (Fig. 3) are analysed, with the exception that in Bombina bombina sexes are not clearly separated along the third discriminant axis. Mahalanobis multivariate distances between females and males were calculated within each group. The significance of Mahalanobis distance between sexes was lowest in B. bombina and highest in B. v. scabra (Z values: B. bombina = 1.67, p = 0.05; B. v. variegata = 2.40, p < 0.01; B. v. scabra = 7.97, p < 0.001). Correspondence analysis, employed to examine differences in qualitative characters, showed that females and males were similar with respect to all analysed qualitative traits in all three taxa. The proximity of sexes in the space of correspondent axes indicates that distributions of character states between females and males are very similar. All dissimilarities in qualitative traits were related to distinction of the taxa (appearance of dorsal warts, colour of ventral side of body and the patches, presence of white spots on the ventral side of body and the presence of slight constriction between head and body). Thus, none of the analysed qualitative characters showed sexual dimorphism. Discussion Our results revealed the existence of statistically significant dimorphism in size (SSD) in the Bombina species and subspecies analysed. The pattern of intersexual differences was complex. Body length (L), the trait most often used in this kind of study, showed inconsistent differences across taxa. In B. bombina males were larger than females, in B. v. scabra females were larger, but in both cases the differences were not statistically significant. Only in B. v. variegata were females significantly larger than males. However, intersexual differences on 133

132 Table 1. Two way ANOVA for the effects of sex, population and their interaction for 20 morphometric characters in Bombina spp. ( * p < 0.05, ** p < 0.01, *** p < 0.001). See text for explanation of character codes. B. bombina B. v. variegata B. v. scabra Trait Sex Population Interaction Sex Population Interaction Sex Population Interaction L *** ** 3.57 * *** 1.45 F *** ** *** 2.49 * T 3.91 * *** ** *** *** 2.57 * N *** * * *** 1.63 P *** ** ** *** 1.08 H *** ** *** *** 0.83 M *** ** *** 1.11 DPPA *** *** 1.30 DSPA *** *** 2.08 DPPP *** * *** 2.14 CINT *** *** 0.27 LC * *** *** 1.95 LTC *** *** *** 1.16 SPP *** *** 1.94 SPI 8.20 ** *** *** 1.88 SPCR *** * 4.78 *** 1.54 LO *** *** 1.81 LTP ** ** 2.23 DRO *** ** 6.35 *** 1.04 DNO * 4.99 * 3.68 **

133 Table 2. Two-way MANOVA for the effects of sex, population, and their interactions in Bombina spp. ( * p < 0.05, **p < 0.01, *** p < 0.001). B. bombina B. v. variegata B. v. scabra Effect Wilks λ Rao s R Wilks λ Rao s R Wilks λ Rao s R Population *** *** *** Sex * ** *** Interaction a) b) Fig. 2. a) Discrimination of B. bombina (b), B. v. variegata (v) and B. v. scabra (s) along the first and the second discriminant axis; b) discrimination of sexes along the second and the third discriminant axis. 135

134 a) b) Fig. 3. a) Ordination of population centroids along the first and the second discriminant axis; b) ordination of population centroids along the second and the third discriminant axis. multivariate level were statistically significant in all three taxa, with growing level of significance of the effect of sex on character variation in B. bombina, B. v. variegata and B. v. scabra, respectively. It is interesting to note that S z y m u r a (1993), in the analysis of 136

135 Appendix. Means and standars errors (x ± S.E., in mm) for 20 morphometric characters in Bombina spp. (f females, m males, n sample size) and the significance of intersexual differences; (ANOVA; * p < 0.05, ** p < 0.01, *** p < 0.001). See text for explanation of character codes. B. bombina B. v. variegata B. v. scabra f n = 52 m n = 61 f n = 28 m n = 49 f n = 87 m n=111 Trait x SE x SE ANOVA x SE x SE ANOVA x SE x SE ANOVA L F T * ** ** N P ** H *** M * DPPA DSPA * DPPP * CINT * LC * LTC *** SPP SPI ** SPCR LO LTP DRO * DNO

136 patterns of genetic variation, found that B. variegata populations were more differentiated than B. bombina populations and that, within B. variegata, northern groups of populations were more similar than the southern ones. Body length, which is commonly used as the indicator of body size (many comparative studies of SSD are based on this trait), is not always appropriate for this purpose. In some species, as is clearly the case here, differences in other characters (related to specific reproductive behaviour, feeding, locomotion) could be more important. This is in line with models based on ecological divergence between sexes and differences in food utilization (e.g. S hine 1989). It has been already suggested (e.g. H oward & Kluge 1985, Halliday & Verrell 1986) that body length is not always the most important factor for male mating success in amphibians. In addition, one of the patterns revealed in our study is the strong effect of populations (i.e. localities) on character variation, indicating the substantial level of geographic differentiation among populations. This raises another issue to reach the accurate estimate of SSD for a certain species, samples from different populations are needed. When the effect of population was separated from the effect of sex, the most prominent results were: significant intersexual differences in tibia length (T) in all three taxa, and highly significant intersexual difference in head width (LTC) in B. bombina and humerus length (H) in B. v. scabra. Longer legs in males could be explained with respect to reproductive behaviour; agility is generally considered important in competition to attract females. Longer legs may also be connected with sex-specific territorial behaviour. In this context, it is interesting to mention that, though B. variegata males were not considered to be territorial (e.g. S zymura 1993), S eidel (1999) reported that a number of males performed sex-specific territorial behaviour, producing water-waves that demarcated territories with their hind legs. Humerus was significantly longer in male than in female B. v. scabra; contrary to expectations, differences in this trait in the other two taxa were insignificant. Arm length is considered important for male mating success in anurans, and the general assumption is that longer arms allow a more secure grip of females. Howard & Kluge (1985) reported that arm length was more important for reproductive success than body length in male Rana sylvatica. Highly significant difference in head width in B. bombina is related to the presence of vocal sacs in males and their role in reproduction, but the possibilty of a certain degree of trophic divergence may be also worth examining. Thus, we conclude that the pattern of SSD in analysed taxa was complex, body size was insignificantly different, but all significantly different traits were larger in males, the only exception being L in female B. variegata variegata. The general level of SSD was higher in B. variegata than in B. bombina. In addition, our results emphasize that great caution is needed when analysing SSD on basis of body length only. This problem is pronounced in comparative studies based on previously published data, with heterogeneous indicators of body size, in studies where body length (L) is the only indicator of body size, and studies on geographically differentiated species where samples are not taken from different populations. A c k n o w l e d g e m e n t s The authors thank P. Simonoviç, J.W. Arntzen and two anonymous reviewers for constructive comments. 138

137 L I T E R A T U R E ANDERSON, S., 1994: Sexual Selection. Princeton University Press, Princeton, New Jersey. ARNTZEN, J. W., 1978: Some hypotheses on postglacial migrations of the fire-bellied toad, Bombina bombina (Linnaeus) and the yellow-bellied toad, Bombina variegata (Linnaeus). Journal of Biogeography, 5: CLUTTON-BROCK, T. H., HARVEY, P. H. & RUDDER, B., 1977: Sexual dimorphism, socionomic sex ratio and body weight in primates. Nature, 269: CHEVERUD, J. M., DOW, M. M. & LEUTENEGGER, W., 1985: The quantitative assesment of phylogenetic constraints in comparative analyses: sexual dimorphism in body weight amnog primates. Evolution, 39: CVETKOVIå, D., KALEZIå, M. L. & DÎUKIå, G., 1997: Sexual size and shape difference in the crested newt (Triturus carnifex): ontogenetic growth aspects. Alytes, 15: DARWIN, C., 1871: The descent of man and selection in relation to sex. Murray, London. DUELLMAN, W. E. & TRUEB, L., Biology of amphibians. McGraw-Hill, New York. EMERSON, S. B., 1994: Testing pattern predictions of sexual selection: A frog example. Am. Nat., 143: FAIRBAIRN, D. J., 1990: Factors influencing sexual size dimorphism in temperate waterstriders. Am. Nat., 136: FAIRBAIRN, D. J., 1997: Allometry for sexual size dimorphism: pattern and process in the coevolution of body size in males and females. Ann. Rev. Ecol. Syst., 29: GOLLMANN, G., 1984: Allozymic and morphological variation in the hybrid zone between Bombina bombina and B. variegata (Anura, Discoglossidae) in northeastern Austria. Z. Zool. Syst. Evol., 22: HALLIDAY, T. R., 1983: The study of mate choice. In: Bateson, P. (ed.), Mate Choice. Cambridge University Press, Cambridge: HALLIDAY, T. R. & VERRELL, P. A., 1986: Sexual selection and body size in amphibians. Herpetological Journal, 1: HALLIDAY, T. R. & TEJEDO, M., 1995: Intrasexual selection and alternative mating behavior. In: Heatwole, H. & Sullivan, B. K. (eds), Amphibian Biology, Vol. 2, Social Behaviour. Surrey Beatty and Sons, Chipping Norton, NSW: HOWARD, R. D., 1981: Sexual dimorphism in bullfrogs. Ecology, 62: HOWARD, R. D. & KLUGE, A. G., 1985: Proximate mechanisms of sexual selection in wood frogs. Evolution, 39: LANG, M., 1988: Notes on the genus Bombina OKEN (Anura, Bombinatoridae). The British Herpetological Society Bulletin, 26: LOVICH, J. E. & GIBBONS, J. W., 1992: A review of techniques for quantifying sexual size dimorphism. Growth, Development and Aging, 56: MACCALLUM, C. J., NÜRNBERGER, B., BARTON, N. H. & SZYMURA, J. M., 1998: Habitat preference in the Bombina hybrid zone in Croatia. Evolution, 52: MADSEN, T. & SHINE, R., 1994: Costs of reproduction influence the evolution of sexual size dimorphism in snakes. Evolution, 48: MICHA OWSKI, J. & MADEJ, Z., 1969: Studies on the relationship of Bombina bombina L. and Bombina variegata L. III Taxonomic characters of both species from laboratory and in interspecific hybrids. Acta Zool Cracov, 14: PRICE, T. D., 1984: The evolution of sexual size dimorphism in Darwin s finches. Am. Nat., 123: RADOVANOVIå, M., 1951: Vodozemci i gmizavci nase zemlje. [The Amphibians and Reptiles of Our Country]. Naucna knjiga, Beograd (in Serbian). REYNOLDS, J. D. & HARVEY, P. H., 1994: Sexual selection and the evolution of sex differences. In: Short, R.V. & Balaban, E., (eds), The Differences between the Sexes. Cambridge University Press, Cambridge: SEIDEL, B., 1999: Water-wave communication between territorial male Bombina variegata. Journal of Herpetology, 33: SHINE, R., 1979: Sexual selection and sexual dimorphism in the Amphibia. Copeia, 2: SHINE, R., 1989: Ecological causes for the evolution of sexual dimorphism: A review of the evidence. Quarterly Review of Biology, 64: SULLIVAN, B. K., 1984: Size dimorphism in anurans: A comment. Am. Nat., 123: SZYMURA, J. M., 1993: Analysis of hybrid zones with Bombina. In: Harrison, R.G. (ed.), Hybrid Zones and the Evolutionary Process. Oxford University Press, New York and Oxford:

138 SZYMURA, J. M. & BARTON, N. H., 1986: Genetic analysis of a hybrid zone between the fire-bellied toads, Bombina bombina and B. variegata near Cracow in southern Poland. Evolution, 40: SZYMURA, J. M. & BARTON, N. H., 1991: The genetic structure of the hybrid zone between the fire-bellied toads, Bombina bombina and B. variegata: comparisons between transects and between loci. Evolution, 45: SZYMURA, J. M., UZZELL, T. & SPOLSKY, C., 2000: Mitochondrial DNA variation in the hybridizing firebellied toads, Bombina bombina and B. variegata. Molecular Ecology, 9: WOOLBRIGHT, L. L., 1983: Sexual selection and size dimorphism in anuran amphibia. Am. Nat., 121:

139 Folia Zool. 51(2): (2002) Reproductive biology of burbot, Lota lota lota, in Lake Haƒcza, Poland Maria BRYLI SKA 1, ucjan CHYBOWSKI 2 and Adam BOGUSZEWSKI 1 1 Department of Zoology, University of Warmia and Mazury in Olsztyn, M. Oczapowski St. 5, Olsztyn, Poland; kzool@.uwm.edu.pl 2 Inland Fisheries Institute, M. Oczapowski St. 10, Olsztyn, Poland Received 2 September 1998; Accepted 2 April 2002 Abstract. Absolute fecundity (Fa) of burbot Lota lota lota females from Lake Haƒcza was , the relative fecundity based on eviscerated weight (Fw o ) was , and that based of total weight (Fw) was Regression equations were calculated, approximating the most significant relationships between fecundity and total body length of burbot (Fa = Sl , Fa = 0,0099 Tl ) and weight (Fa = W ). Mean gonadosomatic indexes (GSI) increased slightly with body length from 3.9 to 8.8% for females and from 3.7 to 20.6% for males. Key words: fish, burbot, fecundity, condition, gonadosomatic index Introduction Burbot (Lota lota lota L.) is a popular fish in clean rivers, streams and lakes. It prefers well oxygenated, clean and cold water. In deep lake waters, if the conditions are suitable, burbot will feed throughout the year and attain considerable size (H e w s o n 1955, M ı l l e r 1960, 1970, Kainz & Gollmann 1996). River regulation and water pollution have decreased considerably the available habitats for this fish (C o p p 1990). In many waters burbot has been included into the red list of protected species (H e r z i g- S t r a s c h i l 1991, H arsányi & Aschenbrenner 1992, Maitland & Lyle 1996, Kainz & Gollmann 1996). The objective of our study was to broaden our knowledge on the reproduction of this still little known cold-water species in particulare that of the lake population in Lake Haƒcza, the deepest lake of Poland. Study Area, Material and Methods Lake Haƒcza is located in a nature reserve in Suwa ki Landscape Park. Its maximal depth is m, average depth 38.7 m, area ha. It receives waters from the Czarna Haƒcza River, three small streams, and from underground sources. Water visibility was 8 9 m. Hillbricht-Ilkowska (1995) classified this lake as mesotrophic based on the content of total phosphorus. Lake waters were well oxygenated to the bottom, with a characteristic peak of oxygen concentration in the metalimnion, at the depth of m. Burbot were collected with trammel nets (mesh size mm) set at the depth of 30 to 60 m in winter 1996 and summer Table 1 presents number and characteristic of the collected burbot. Catches carried out in winter 1996 had to yield mature gonads for fecundity studies and gonad development. Material collected in summer 1997 was used only to check gonad development. Gonads used to determine fish fecundity were wrapped in 141

140 gauze and preserved in 4 % formaldehyde solution. Ovaries were then taken out, eggs separated from the ovary tissue, and diameter (E) of 50 eggs was measured up to 0.01 mm. Absolute fecundity (Fa) was calculated using gravimetric methods (B ryliƒska & Bryliƒski 1972). Relative fecundity (Fw O ) was determined by dividing the number of trophoplasmatic oocytes (viz. absolute fecundity) by the weight of eviscerated fish (Wo). Fish intestines were removed to avoid errors due to different filling of the digestive tracts with food. In some cases, however, total body weigh (W) was also used to compare the data of relative fecundity (named Fw) with these from the literature. For histological analysis, the material was preserved in Bonin-Holland solution, and then passed through different concentrations of alcohol, xylene, and immersed in paraffin. Sections were stained with Delafield s haematoxylin. Table 1. Characteristics of burbot collected in Lake Haƒcza (Tl total length; W body weight; Wo eviscerated body weight; T testis; O ovary). Date (n) Sex Minima - maxima Tl (cm) W (g) Wo (g) T or O (g) males females males , females , The least squares method was used to calculate the regression equations for the dependence between fecundity absolute and relative, and total (Tl in cm) and standard body length (Sl in cm) as well as total body (W) and eviscerated weight of fish (Wo in g), ovary weight (O in g), egg diameter (E in mm) and egg weight (E in g). The results were treated statistically. Significance of Pearson correlation coefficients was determined using the respective tables (r tab , Platt 1974) for n - 1 = 30 degrees of freedom, the accepted level of significance being P = Analyses were performed to determine the best equation for the dependence between absolute fecundity (Fa) and the other parameters (Tl, Sl, W, Wo, O, E). In order to compare the condition of burbot males and females, data on body length and weight were analysed. Gonadosomatic indexes (GSI) calculated for males and females from the equation GSI = O(T)/W * 100 (in %). Results Absolute fecundity of burbot from Lake Haƒcza ranged from 47.3 to thousand eggs. The mean values minima and maxima are presented in Table 2. Absolute fecundity increased more rapidly than body length and weight increments; the respective regression line is polynomial (curvilinear). Absolute fecundity of burbot increased with increasing gonad weight (Table 3) in a similar way as with body weight. There were no significant correlations between absolute fecundity and egg diameter and egg weight (E). Absolute fecundity of burbot appeare to depend most of all on the fish size (Table 3), gonad weight (O) being more important than body length (Sl, Tl) and body weight (W). 142

141 Table 2. Mean (x ), minima and maxima (x 1 - x n ) of absolute (Fa) and relative fecundity (Fw, Fw o ), total body length (Tl in cm), body weight (W in g), body weight of gutted fish (Wo in g), ovary weight (O in g), egg diameter (E n in mm) and weight (E g in g) of burbot from Lake Haƒcza (n = 30, in winter). Tl Fa*10 3 O x 30.8 x x x 1 -x n x 1 -x n x 1 -x n W Fw o *10 3 E n x x x x 1 -x n x 1 -x n x 1 -x n Wo Fw*10 3 E g x x x x 1 -x n x 1 -x n x 1 -x n Table 3. Pearson coefficients of correlation (r) for curvilinear and linear regression functions between absolute fecundity (Fa 10 3 ) and relative fecundity (Fw o 10 3 / Wo) and body length (Tl, Sl in cm) body weight (W, Wo in g) ovary weight (O in g) (n - 1 = 29). y - x r y = ax b r y = ax+b Fa - Tl Fa= Tl Fa= Tl Fa - Sl Fa= Sl Fa= Sl Fa - W Fa= W Fa= W Fa - Wo Fa=1.186 Wo Fa= Wo Fa - O Fa= O Fa= O Relative fecundity (Fw o ) ranged from to 1572 thousand eggs, its mean value being thousand eggs (Table 2). Relative fecundity (Fw o ) did not increase with body length, but remained at an almost unchanged level. Coefficients (r) for the relationship between relative fecundity and the parameters under study (Tl, Sl, W, Wo, O, E) were all insignificant, showing that no relationship can be found between these variables (r = ). Body weight was related significantly to total length with females (W = Tl 2.642, r = 0.953) showing slightly better condition than males (W = Tl , r = 0.937), as body weight of females was slightly higher than of males of the same length. Mean GSI increased slightly with body length (Table 4), from 3.9 to 8.8 % for females, and from 3.7 to 20.6 % for males. Mean male GSI was 12.5 %, being noticeably higher than the respective value calculated for females (5.6 %). The relationship between gonad and body weight (Fig. 1) was significant in females (O = W ; r = 0.936) and in males (T = W ; r = 0.803) revealing that males had proportionally greater gonads weight than females of the same size. Mean GSI for females and males after spawning (on 29 July 1997) was the same 1 %. Many protoplasmatic oocytes can be seen (O.p.), in the female gonads (Figs 2, 3) with a characteristically large number of nucleoli in the oocyte nuclei (Cj) and a dark cytoplasm (haematoxylin staining). In addition to this, single resorbed oocytes can also be seen (o.t.); these are the eggs, that had not been spawned during the last reproduction. The single resorbed oocytes in burbot ovary are good be seen in the end of July, this is probably 4 to 7 month after spawned time. Gonads of a few females (Fig. 3a) had a single row of vacuole in the oocyte cytoplasm, suggesting that these oocytes have already commenced the 143

142 Table 4. Gonadosomatic index (GSI) of females and males of burbot in total body length classes (Tl in cm). Date Sex Parameter Tl Σ GSI Females X 1 -X n n GSI Males X 1 -X n n GSI Females X 1 -X n n GSI Males X 1 -X n n trophoplasmatic growth stage (o.t.). Male testes (Fig. 3b) contain sprematocytes, which can be seen close to the canals (with empty light). All female and male gonads collected in 1996 and 1997 were fertile as they contained reproductive cells in different stages of development, corresponding to the annual cycle of sexual development. Discussion Burbot commence spawning in winter, from December to March (Siergieyev 1959, Sorokin 1968,1971, Volodin 1964), a little later in lakes from January onwards O/T (g/g) W (g) Fig. 1. Relationship between ovary weight (O in g), testis weight (T in g) and body weight (W in g) of burbot in winter. Equations given in the text. 144

143 Fig. 2. Ovary of a burbot (July 1997). A female no 9, Tl = 35 cm; B female no 12, Tl = 36 cm; o.p. oocyte of protoplasmatic stage; r.o.t. resorbed oocyte of trophoplasmatic growth stage non spawned during the last reproduction. (Kainz & Gollmann 1996). Water temperature in this period is from 0 to 4 C (H ewson 1955, Struganov 1962, Sorokin 1971, Gerster & Guthruf 1987 after Kainz & Gollmann 1994), or even up to 5.7 C (F a r k a s 1993 after Kainz & Gollmann 1994). Spawning period is characterised by low temperatures and is extended in time. Elsewhere, the diameter of trophoplasmatic oocytes immediately prior to spawning is from 0.80 to mm (Munth & Smith 1974, Volodin & Raspopova 1994). G erster & Guthruf (1987) after P atzner & Reihl (1992) stated that burbot eggs diameter ranged from 0.5 to 1.70 mm (mean 1.00 mm). Patzner & Reihl (1992) described burbot eggs are transparent, yellowish, with micropyle close to animal pole, with a canal diameter 2.5 to 3.0 µm, having one lipid droplet some 0.4 µm in diameter. Oocyte diameter in burbot ovaries from Lake Haƒcza were collected the end of November ranged from to mm, being a little smaller than that observed in a comparable period by Munth & Smith (1974). Data presented in the literature suggest that absolute fecundity of burbot females weighing from 44.5 to 3164 g was from 32.2 to eggs, this range (Table 5, Fig. 4) being broader than in the population from Lake Haƒcza, which had high absolute fecundity. Also upper values of relative fecundity (Fw) in Lake Haƒcza burbot ( ) suggests that were higher then in other populations (Table 5). Higher relative fecundity (Fw) was characteristic of 145

144 Fig. 3. Ovary (A) and testes (B) of burbot (July 1997). A female no 9, Tl = 35 cm, w single row of vacuoli, j nucleoli, cj nuclei of oocytes; B male no 10, Tl = 32.5 cm, kn canals with empty leight, s spermatocytes. the smallest and often the largest females in relation to those of medium body weight, although variability within particular groups was quite high. The same has been observed elsewhere (Volodin & Raspopova 1994, K ainz & Gollmann 1996). Gonadosomatic index (GSI) increased from July to November respectively from to % and from to % of body weight (W). Vostradovská (1963) found that females GSI in burbot from Lipno Reservoir in November ranged from 4.37 to 16.90, whereas it in males ranged from 8.06 to 25.8 % of body weight. These data suggest also that GSI in burbot is higher for the testes than the ovaries. According to Volodin (1994), in Rybin ski Reservoir female GSI for burbot weighing from 311 to 530 g reached in January from 16.2 (± 0.7) to 20.0 % of body weight. Studies on gonad maturation in different burbot populations (P ulliainen et al. 1992, P ulliainen & Korhonen 1993, 1994) from the northern coast of the Bothnian Bay (23 93 %), Lake Kemijaervi in northern Finland (44 %), Kemi River (24 %), Kitanien River (86 %), and Tonio River (98 %) revealed high percentage of non spawning ( sterile ) individuals. The proportion of non spawning burbot was a little lower in big fish (over 39 cm in length, Tl), ranging from 21 to 51 %. The high proportion of non spawning burbot was said to be related to environmentally toxic chemicals (PCB, PCDD, Al, Cr, Pb 146

145 Fa Tl (cm) Fig. 4. Relationship between absolute fecundity (Fa) and total body length (Tl) of several populations of burbot. a, b, c absolute fecundity of burbot in Lake Haƒcza (a, Fa = Tl 2,812, b, Fa = Tl , c one female). 1. Variation in fecundity (brooks near Tachov); 2. River Vltava (K oufiil et al. 1985); 3. Absolute fecundity (Upper Tuloma Reservoir, N e l i c h i k 1975); 4. Absolute fecundity (Rybinskoe Reservoir, Volodin 1994); 5. Lake Traum and Stream Stimnitz (Kainz & Gollmann 1996) Table 5. Minima and maxima for absolute (Fa) and relative (Fw) fecundity of burbot from different populations. W (g) Fa 10 3 Fw 10 3 Tl (cm) min max min max min max Authors Podubsk & tûdronsk (1967) Koufiil et al. (1985) Kainz & Gollmann (1996) Nelichik (1975) (±22) Volodin (1994) (±21) 515 (±71) present study and Hg), which accumulated in the liver and coincided with irregular growth of otoliths and bone respiation (Pulliainen et al. 1992). Biological effects such as morphological and reproductive deformation incited by organic pollutants such as PCBs has be reported for the Baltic Sea (Baltic Sea Environment Proceedings 1990). Studies on gonad maturation in burbot from Lake Haƒcza did not reveal any sterile fish amongst the sexually mature individuals, but water of this lake is exceptionally clean. 147

146 LITERATURE Baltic Sea Environment Proceedings, 1992: Second periodic assessment of the state of the marine environment of the Baltic Sea, Back round document No 35B, Baltic Mar. Environ. Prot. Commis., Helsinki Commission. BRYLI SKA, M. & BRYLI SKI, E., 1972: Metody okreêlania p odnoêci ryb na przyk adzie leszcza (Abramis brama L.) (Methods for estimation of fish fecundity on the example of bream (Abramis brama L.)). Rocz. nauk roln., 94 H 2: 7 40 (in Polish, with summaries in English and Russian). COPP, G. H., 1990: Effect of regulation on 0+ fish recruitment in the Great Ouse, a lowland river. Regul. Rivers Res. Manage., 5 (3): HARSÁNYI, A. & ASCHENBRENNER, P., 1992: Die Rutte (Lota lota Linnaeus, 1758). Biol. und Aufzucht, Fischer und Teichwirtschaft, 10: HEWSON, L. C., 1955: Age, maturity, spawning and food of burbot, Lota lota in Lake Winnipeg. J. Fish. Res. Board Can., 12: HILLBRICHT-ILKOWSKA, A., 1995: Charakterystyka limnologiczna ekosystemu wodnego rezerwatu przyrody Jeziora Ha cza [Limnological characteristic of water landscape nature reserve Haƒcza Lake]. Inst. Ekol. PAN, Miko ajki (unpubl., in Polish). KAINZ, E. & GOLLMANN, P., 1996: Laichgewinnung, Erbrütung und erste Aufzuchtversuche bei Aalrutten (Lota lota). Österr. Fisch., 49 (17): KOU IL, J., LINHART, O., DUBSK, K. & KVASNIâKA, P., 1985: The fertility of female and male burbot (Lota lota) reproduced by stripping. Pr. VÚRH VodÀany, 14: MAITLAND, P. S. & LYLE, A. A., 1996: Threatened freshwater fishes of Great Britain. In: Kirchhofer, A. & Hefti, D. (eds), Conservation of Endangered Freshwater Fishes in Europe. Birkhäuser Verlag: MUNTH, K. & SMITH, L., 1974: The burbot fishery in Lake of the Woods. Agricult. Exp. Station, Univ. Minnesota: MÜLLER, W., 1960: Beiträge zur Biologie der Quappe (Lota lota L.) nach Untersuchungen in den Gewässern zwischen Elbe und Oder. Z. Binnen Fischerei, 9: MÜLLER, K., 1970: Beobachtungen über das Laichen der Quappe Lota lota. Oikos Suppl., 13: NELICHIK, V. A., 1975: [Fecundity of the burbot in the upper Tuloma Reservoir]. Izv. GosNIORKH, Leningrad, 93: (in Russian). PATZNER, R. A. & RIEHL, R., 1992: The eggs of native fishes. 1. Burbot, Lota lota L. (1758), (Gadidae). Österr. Fisch., 45(10): PLATT, Cz., 1974: [Problems of the calculus probability and mathematical statistics]. PWN, Warszawa (in Polish). PODUBSK, V. & TùDRONSK, E., 1953: Beitrag zur Biologie der Quappe (Lota lota). Sb. âsazv, Îiv. V roba., 26: PULLIAINEN, E. P. & KORHONEN, K., 1993: Does the burbot, Lota lota, have rest years between normal spawning seasons? J. Fish. Biol., 43(3): PULLIAINEN, E. P. & KORHONEN, K., 1994: Sagittal otolith growth patterns in regularly and irregularly spawning burbot, Lota lota, in northern Finland. Environ. Biol. Fish., 40: PULLIAINEN, E. P., KORHONEN, K., KANKAANRANTA, L. & MÄKI, K., 1992: No spawning burbot on the northern coast of the Bothnian Bay. AMBIO, 21(2): SIERGIEYEV, R. S., 1959: [Data of burbot biology from Rybinskoe Reservoir]. Tr. Inst. Biol. Vodkh., 1(4): (in Russian). SOROKIN, V. N., 1968: [Biology of young burbot Lota lota (L.)]. Vopr. Ikhtiol., 8(3): (in Russian). SOROKIN, V. N., 1971: [The spawning and spawning grounds of the burbot]. Vopr. Ikhtiol., 11: (in Russian). STROGANOV, N. S., 1962: [Ecological physiology of fishes]. Izd. MGU: 420 pp. (in Russian). VOLODIN, V. M. & RASPOPOVA, Z. A., 1994: [The state of stocks and reproductive potential of some fish species in the Sheksna reach Rybinskoe Reservoir]. Vopr. Ikhtiol., 34(4): (in Russian). VOLODIN, V. M., 1994: [Status of breading stocks of burbot Lota lota in Sheksna reach of Rybinskoe Reservoir and after the emergency sewage disposal in year]. Vopr. Ikhtiol., 34(3): (in Russian). VOSTRADOVSKÁ, M., 1963: The biology of the burbot (Lota lota) in the Lipno Reservoir. I. Variations of the liver and gonad weight as well as the condition of the burbot during the pre-spawning period. Pr. VÚRH VodÀany, 3:

147 Folia Zool. 51(2): (2002) Meristic and mensural morphological characters of juvenile sterlet reared in the Czech Republic and Slovak Republic Miroslav PROKE 1, Vlastimil BARU 1, Milo MACHOLÁN 2, Ivan KRUPKA 3 and Juraj MASÁR 4 1 Institute of Vertebrate Biology, Academy of Sciences of the Czech Republic, Kvûtná 8, Brno, Czech Republic; prokes@brno.cas.cz 2 Laboratory of Genetics and Embryology, Institute of Animal Physiology and Genetics, Academy of Sciences of the Czech Republic, Vevefií 97, Brno, Czech Republic; macholan@iach.cz 3 Majerníkova 10, Bratislava, Slovak Republic 4 Research Institute of Animal Production, Institute of Fisheries and Aquaculture, âastá, Slovak Republic Received 6 November 2000; Accepted 13 February 2002 Abstract. Documented were the values of 7 morphological meristic and 37 mensural characters in samples of 0+ juvenile specimens of the sterlet reared in special aquacultural facilities in the Czech Republic (and originating from Russia), and in the Slovak Republic (originating from the Danube Slovak section). Individual character values were compared with literature data using the method size-pooled samples. Specimens reared in the Czech Republic (CR) and in the Slovak Republic (SR) did not differ in the maximum-minimum range of their meristic characters from the literature systematic description presented for sterlet (Acipenser ruthenus Linnaeus, 1758). Regards differences between mean values, the sample from the Czech Republic differed significantly in 4 (of 6) meristic characters, and in 14 (of 27) mensural characters compared. On the basis of multivariate morphometrical analysis (PCA, UPGMA) it was found that the sample of sterlet from the CR clusters well with other samples of the sterlet populations, being morphologically closest to a pair of samples from the Danube River. We suggest that, because of the numerous transfers into the Danube of sterlets reared by aquaculture in the CR, it will be impossible to discriminate them morphologically from those reared naturally in the Danube Slovak section. Key words: morphometrical characters, juvenile Acipenser ruthenus, aquaculture form, comparative morphometrical analysis, repeated introduction Introduction In the first half of the 20 th century, sterlet Acipenser ruthenus (Linnaeus, 1758) occurred naturally in the territory of the present Czech Republic in the lower reaches of the Morava and Dyje rivers. In the second half of this century, gradual and significant reductions of its occurrence took place (Baru & Oliva 1995), and during the last decade, its natural occurrence could not be proved at all (Lusk & Hanel 1996a,b, 2000, Lusk et al. 1996, 2000), even though reports on its rare catches by angling do exist (Hanel 1992, Prá il 2000, Lusk et al. 2000, Zelinka 2000). In the Slovak Republic, the sterlet occurs naturally, first of all, in the Danube and Tisa rivers (Holãík 1998, Krupka 2000). Attempts to artificial rear this species in ponds of the Czech Republic occurred as early as at the end of the 19 th century in the environs of TfieboÀ, and during 1935 and in the environs of Velké Mezifiíãí and KfiiÏanov (Kostomarov 1947a,b, Hubáãek 1950, Baru & Oliva 1995). Its artificial culture, of course, was not successful because the artificial reproduction failed in this species. Other attempts focused on the sterlet culture in special aquaculture facilities in the CR were launched during early 1990s in the Hluboká n. Vlt. Pond Fishery (Jirásek 1999a,b; Proke et al. 1995, 1997a,b,c, 1999, 2000a,b, and 149

148 others). From fertilized eggs imported from Russia, juvenile and adult specimens were reared in the Mydlovary hatchery, which became then economical subjects of that pond fishery company. Since there was a trend of sterlet artificial culture in the territory of the CR, and incidental or deliberate transfers of juvenile specimens into free waters (especially into rivers and flow reservoirs) might not be excluded, the documentation of meristic and morphometrical characteristics in imported and reared specimens was needet. That was done in the Institute of Vertebrate Biology AS CR in Brno. In parallel, the analysis of morphometrical and meristic characters was performed in a sample of sterlet juvenile specimens reared experimentally in the Institute of Fisheries and Aquaculture at âastá (Slovak Republic). Reproductive specimens for stripping were sampled in the Danube, belonging to the autochthonous population. Artificial reproduction in sterlet has been carried out currently in Slovakia since 1987 (K r u p k a 2000). The biometrical results and their analysis are subjects of the present study, and they complete the research by Flaj hans & Vajcová (2000) who, measuring the DNA content in nuclei by flow cytometry and by other methods, analysed the ploidy level in sturgeon juvenile specimens (including sterlet) reared in the Mydlovary Fish Farm during (Proke et al. 2000a). According to the literature data, sterlet is included in the species group with 120 chromosomes, distributed in the Adriatic-Ponto-Caspian zoogeographical zone (F o n t a n a 1994, Fontana et al. 1977, Holãík 1989, Serebryakova 1979, Sokolov & Va s i l y e v 1989, R á b 1986, and others). It belongs to the genus Acipenser Linnaeus, 1758, and it was included recently in the subgenus Sterleta Gueldenstaedt, 1772 (B e r g 1948, A rtyukhin & Romanov 1997). Fontana (1994) and Kuzmin (1996) consider the sturgeon species with 120 chromosomes to be oligochromosomic, diploid, and the species with 240 chromosomes to be polychromosomic, tetraploid. In contrast, B i r s t e i n & Vasilyev (1987), B irstein et al. (1993, 1997) consider the sturgeon species with 120 chromosomes to be tetraploid and those with 240 chromosomes to be octoploid. The research on meristic and mensural characters in sterlet autochthonous populations, according to the literature data known to us, was dealt with by A bdurakhmanov (1962), Banarescu (1964), H olãík (1989), J ankoviç (1958), K rylova (1980), Lukin (1979), Lukin et al. (1981), M e n s h i k o v & Bukirev (1934), Nikolyukin (1972), Oliva & Chitravadivelu (1972) and Pavlov (1967, 1968). Electrophoretic studies on proteins of great sturgeon, sterlet, bester and Russian sturgeon were realized by Dobrovolov & Dobrovolova (1983). Genome structure in interspecific fish hybrids was studied by Vladycheskaya & Kedrova (1982). Material and Methods For biometrical analysis, 40 sterlet juvenile specimens reared in the Czech Republic (CR) and 30 sterlet juvenile specimens reared in the Slovak Republic (SR) were used. Fertilized eggs were imported from the Rybnoye Warm-Water Production Farm (Dmitrovskiy Region, Moscow Province, Russia) on Hatching started on Eggs were hatched out and free embryos, larvae and 0+ juvenile specimens were reared under heated water (t = 21 C, O 2 = 8.0 mg) in the Mydlovary Aquaculture Farm (Hluboká n. Vlt. Pond Fishery Co.) during Further culture of this material was realized within experimental research in the Institute of Landscape Ecology (Institute of Vertebrate Biology at present) AS CR in Brno during Average water temperature during whole breeding period was about 7.1 C higher than exists in natural conditions in the lower parts 150

149 of Dyje and Morava rivers (in Czech Republic). On , some juvenile specimens were fixed in 4% formaldehyde solution. The fixed material was treated by identifying and reprocessing the values of selected meristic and mensural characters during The material obtained from sterlet reproductive specimens reared in the SR had its origin from the Slovak Danube section. They were wild specimens before spawning sampled by seine nets, and were considered to be members of the original autochthonous population. Their stripping and culture (at a water temperature, that of the natural resembling Danube river near Bratislava) of juvenile specimens were carried out in the âastá experimental aquaculture facility, which was in an accessory field station of the Institute of Fisheries and Aquaculture in Bratislava. For morphological analysis, 7 meristic and 37 mensural characters were used. Calculations with characters were performed using the standard methods according to Pravdin (1966), H olãík (1989) and B a r u & O liva (1995). Analysed were absolute and relative values of characters. The relative values of body characters were related to total length (TL), and those of head characters to TL and head length. Analysing the meristic characters, we compared in case of individual characters the values of ranges and means using Student s t-test. The presupposition of insignificant dependence between the respective character and TL values was verified in specimens under study. In order to reveal multivariate morphomeristic affinity of the sample under study, principal component analysis (PCA) was carried out, based on the variance-covariance matrix computed from sample means. Since variances are heavily influenced by the magnitude of raw measurements, the variables were log-transformed prior to analysis. Subsequently, a morphomeristic similarity among the samples, based on matrix of Euclidean distances, was computed from logtransformed variables. In the total, 8 various samples of sterlet, 4 samples of bester (H. huso x A. ruthenus) and 6 samples of great sturgeon were compared (Tables 4 6, Figs 3,4). Fig. 1. Juvenile 0+ aquaculture form of sterlet, TL = 346 mm, w = 207,1 g, age = 186 days after hatching. The artificial rearing from eggs was realised in the Mydlovary Hatchery, Hluboká nad Vlt. Pond Fishery, A.S. (CR) and later (from age D80) in Institute of Vertebrate Biology AS CR in Brno. Original by Petr Pelikán, Brno Analysing the morphometric characters, we first pooled samples to obtain very close or the same mean TL values. In the case of reciprocal comparison of samples from the CR and SR with available individual values for plotting up the mean samples, the data of specimens were used, whose TL occurred within the same size range. In the other case, heterogeneous samples of the TL mean value (literature data were compared), and samples were pooled mathematically from the calculation of regression coefficient values, characterizing the relationship between the respective character mean value and TL mean value in available samples. For this purpose exclusively, character absolute values (in mm) and two regression types, linear or non-linear (polynomial), were used (Tables 3,5,6). For calculations of the condition coefficient and length-weight relationship in the CR material, the TL value used. The basic length and weight characteristics in the own material are presented in Table 1. For making it more simple and unambiguous, we designate below the sterlet aquaculture form reared in the CR as sterlet from the CR (Fig. 1). 151

150 Table 1. Basic length and weight values for 0+ juvenile specimens of aquaculture form of sterlet from the CR and SR explanation for morphological meristic and mensural analysis. Explanations: TL = total length, FL = fork length, SL = standard length, w = weight, R 2 = determination coefficient, CR = Czech Republic, SR = Slovak Republic, S.D. = standard deviation. Character Sample Range Mean S.D. TL (mm) CR SR FL (mm) CR SR SL (mm) CR SR w (g) CR SR FL(CR) = TL, R 2 = ; FL(SR) = TL, R 2 = SL(CR) = TL, R 2 = ; SL(SR) = TL, R 2 = w(cr) = 3E-07. TL , R 2 = ; w(sr) = 3E-06. TL , R 2 = Results Meristic characters Values for the maximum-minimum range of ray number in D and A, number of scutae in dorsal, lateral and ventral lines and number of gill rakers did not differ from those reported in the literature (Table 2). This fact was found in sterlet samples from the CR and also from the SR. However, higher mean values of ray number were found in D and A. In D, the differences in ray mean number from the mean value calculated, presented below as so-called standard values (Table 2), were statistically insignificant (CR diff. 4.95%, SR diff. 1.4%). In A, the statistically significant difference was found in the material from the CR (CR diff %, SR diff. 2.88%). Variation of the ray number mean values in all ten samples examined was significantly higher in Athan in D (A max. diff %, D max. diff %). Comparing the maximum-minimum number of scutae in dorsal and ventral lines, as found in specimens from the CR and SR, with the data reported in the literature, we found no significant differences (Table 2). The maximum number of scutae in lateral line (73) found in the sample from the SR was by 2 scutae higher than reported hitherto in the literature (Table 2). The mean values of scutae number in dorsal line differed statistically significantly within all samples; in lateral and ventral lines, differences were insignificant (max. diff. 2.58%). Analysing the gill raker number range (our material versus literature data), we found no statistically significant differences. However, between the maximum and minimum mean values of gill rakers number within all ten samples analysed, the statistically significant difference (30.38%) was found. The number of fulcrae was in the material from the CR (mean 35.9) and in that from the SR (mean 35.5). The value range of fulcrae number, found by us, was in both samples (CR and SR) statistically significantly higher than the values reported in the literature. The mean values, of course, did not differ statistically significantly. Comparing in total the material from the CR versus SR regards meristic values, we found in the 7 characters analysed statistically significant differences in mean values of 3 characters: rays in A(diff. 8.40%), scutae in ventral line (diff. 9.23%) and scutae in dorsal line (diff %). 152

151 Table 2. Meristic characters of 0+ juvenile specimens of sterlet aquaculture form from the Czech Republic and Slovak Republic (our results), and of free-living populations from other localities (literature data). Explanations: 1 Oliva & Chitravadivelu (1972); 2 Holãík (1983 cit. Holãík 1989); 3 our results, SR; 4 Pavlov (1968); 5 Jankoviç (1958); 6 Men shikov & Bukirev (1934); 7 Pavlov (1968); 8 Kuchina (1967 cit S okolov & Vasilyev 1989); 9 Lukin et al. (1981); 10 our results, CR. Explanations: Du = rays in D, Au = rays in A, SD = scutae dorsales, SLa = scutae laterales, SV = scutae ventrales, Sp. br. = spinae branchiales. Char. Du Au SD Author range mean s x S.D. n range mean s x S.D. n range mean s x S.D. n Danube Others Total Char. SLa SV Sp. br. Author range mean s x S.D. n range mean s x S.D. n range mean s x S.D. n Danube Others Total In the case of multivariate morphomeristic analysis applications, the first principal component explains by far the largest amount of total variation (87.42%), while the other two PCs explain 7.34% and 2.98%, respectively. The first three components account for 97.74% of total morphometrical variation. As shown in Fig. 3, PC1 discriminates fairly well the two sturgeon species, with the bester being in between. The analysed sample falls well within the group of the sterlet (AR). 153

152 Fig. 2. The head (ventral view) of 0+ juvenile aquaculture form of sterlet, TL = 346 mm, w = g. Original by Petr Pelikán, Brno Fig. 3. Principal component analysis of morphometrical afinity into the sturgeon and hybrid samples after 6 meristic markers presented in Table 2. Explanations: AR = A. ruthenus; BES = bester (intergeneric hybrid of A. ruthenus x H. huso); HH = H. huso; samples 1-10 see Table 2; sample (sa.) 12 Krylova (1980); samples 13 and 14 = F1 and F2 generation of bester after Krylova (1980); sa. 15 Krylova (1980); sa. 16 Proke et al. (1995); sa. 17 Pavlov (1967); sa. 18 B e r g (1948); samples after A bdurakhmanov (1962). The first eigenvector is explained mainly by Du and Sp. br. against SLa and SV whereas the second one is explained mostly by Du and Au against Sp.br. Results of cluster analysis are shown in Fig. 4. There are two main clusters on the dendrogram, great sturgeon (HH) on the one hand, and sterlet (AR), bester and AR10 (CR) on the other hand. Interestingly, both the bester samples are very close to sterlet according to 154

153 the 6 meristic markers used and the variation within the whole group is even lower than within the group of great sturgeon populations. As in the case of PCA, the AR10 sample clusters well with other sterlet (AR) populations, being morphologically closest to a pair of samples from the Danube River, analysed by O liva & Chitravadivelu (1972), and Pavlov (1968), respectively. Mensural characters Compared size-pooled samples of sterlet from the CR and SR in 18 body characters studied (Table 3), statistically significant differences (P 0.05) in one character (P length, diff. = -7.16%) and statistically highly significant differences (P 0.01, P 0.001) in four characters (Alength, D height, caudal peduncle length and Aheight) with diff % were found. In total, a significant difference was found in 27.78% of body characters compared. In contrast, the least differences (diff. less than 1%) were found in 4 characters (preventral distance, preanal distance, body depth and Smitt length). From the above data it is evident that specimens from the CR compared with those from the SR were noted for shorter pectoral fins, larger anal fin (longer and higher), shorter caudal peduncle and higher dorsal fin. The body shape, location of fins on the body and head length did not differ significantly. When 14 selected mensural characters measured on the head were similarly compared, sterlets from the CR (versus SR) were noted for greater interocular distance, greater distance between snout tip and barbel bases, smaller distance between barbel bases and mouth margin and for wider head (Table 3). In specimens from the CR (versus SR), the significantly higher mean value of condition coefficient (CR: , SR: ) and a steeper course of the length-weight relationship curve were found (Table 1). Within morphometric comparisons of sterlet reared in the CR with sterlet mainly from natural conditions of the environment using the comparison of two size-pooled samples (one from the CR and the other as a mean from available data, see Tables 4 and 5), significant differences were found in 6 of 14 body characters analysed, i.e. in 42.9% of characters (Table 5). Within the range from minimum to maximum differences, the following characters are concerned: Alength (9.21%), V length (12.77%), Aheight (14.05%), D length (-14.50%), body depth (14.99%) and D height (-20.66%). From these data it is evident that sterlet reared in the CR differed from the other sterlets (standards) compared in particular for longer head, greater anal fin (longer and higher), longer ventral fins, shorter pectoral fins, greater body depth and lower dorsal fin. In most cases (except for head length), the dependence between TL and analysed individual characters was of linear character. The determination coefficient (R 2 ) values were Trends of differences between size-pooled samples of bester reared in aquaculture (F1 and F2) and the sterlet mean sample (mainly from the natural environment) were analogous to differences found between the sample from CR and mean sample values from the literature data (Tables 4 and 5). For parameter dependence determination between head length and TL, the most suitable was the polynomial function with convex curve (Tables 4 and 5). Analysing 13 mensural characters measurable on head (in sample from the CR versus literary data Table 6), we found significant differences in 6 characters (i.e., in 46.2%). Specimens reared in the CR differed from the mean sample by longer snout (diff. 8.72%), greater interocular distance (24.22%), higher head (16.23%), greater distance between snout tip and barbel bases (12.59%), smaller distance between barbel bases and mouth margins (-15.25%) and by wider mouth (18.27%). In 7 characters, more suitable for expression of 155

154 Table 3. Mensural characters of 0+ juvenile specimens of sterlet aquaculture form from the CR and SR, represented and compared in absolute values (in mm). Explanation: a, b, R 2 = regression and determination coefficients of linear regression between TL and other mensural characters; comparisons = comparison of calculated values in individuals of the same size (TL = mm, mean = 275 mm) from the both examples by mean differences (in % and by of Student s t- test, n 1 +n 2-2 = 60; * P 0.05, ** P 0.01, *** P Character Sa. Coefficient Comparisons a b R 2 mean S.D. dif.(%) t P Preventral distance SR CR Preanal distance SR CR Body depth SR CR FL SR CR Predorsal distance SR CR SL SR CR Body width SR CR Body depth min. SR CR Range P-V SR CR Head length SR CR Length D SR CR Length V SR CR Range V-ASR CR Length P SR * CR Length ASR ** CR Height D SR E-07*** CR Length of C peduncle SR *** CR Height ASR *** CR

155 Table 4. Value of characters (in mm) used for comparison. Explanations: abbreviations of characters see Methods; sample 1 (A. ruthenus) Nikolyukin (1972); sa. 2 (A. ruthenus) Krylova (1980); sa. 3 (A. ruthenus) Pavlov (1968); sa. 4 (bester F1) Krylova (1980); sa. 5 (bester F2) Krylova (1980); sa. 6 (bester F1) Nikolyukin (1972); sa. 7 (A. ruthenus) our data from the SR; sa. 8 (bester F1) Nikolyukin (1972); sa. 9 sterlet aquaculture form from the CR (our results); sa. 10 (A. ruthenus) Men shikov & Bukirev (1934); sa. 11 (A. ruthenus) Lukin et al. (1981); sa. 12 (A. ruthenus) Kuchina (1967 cit. S o k o l o v & V asilyev 1989); bester F1 and F2 = hybrids of H. huso x A. ruthenus. Character A.r. A.r. A.r. Hyb. F1 Hyb. F2 H F1 A.r. (SR) H F1 A.r. (CR) A.r. A.r. A.r. Sample TL Predorsal distance Preventral dist Preanal dist Range P-V Range V-A Length of C peduncle Body depth Body depth min Length D Height D Length A Height A17.I Length P Length V Head length Preocular distance Eye diamater Supraocular distance Interocular width Head depth Head depth at eye Snout - mouth distance Snout - barbel dist Barbel - mouth dist Barbel length Snout width at barbels Head width Head width at mouth Mouth width

156 Table 5. Comparison of selected body mensural characters in sterlet (literature published data) and sterlet aquaculture form (CR sa.). Explanations: a,b,r 2 (1) = regression and determination coefficients of the linear regression for relation, dependence between TL and other mensural characters; literature published data = calculated mean values (in mm) of the concrete characters for individuals of A. ruthenus (see Table 3) with theoretical TL 289 mm (by presented regression); CR sa. = sample from the Czech Republic, absolute mean values (in mm), relative mean values (in % of TL); differ. (%) = differ. of mean value of literature data (summarized sa.) and CR sa.; differ. R 2 (1)/R 2 (2) = difference between determination coeff. of linear regression R 2 (1) and deter. coeff. of polynomial regression R 2 (2); differ. H (in %) = difference between A. ruthenus (literature published data) and hybrid F1 (bester) and between A. ruthenus (literature published data) and hybrid F2 (bester). Character Regr. coeff. Deter. coeff. Liter. data CR sa. Differ. Differ. Diff.H (%) a b R 2 (1) R 2 (2) (mm) (%TL) (mm) (%TL) (%) R 2 F1 F2 TL Predors. dist Preanal dist Length D P-V Length C peduncle V-A Body depth min Head length Length A Length V Height A Length P , , Body depth Height D

157 Table 6. Comparison of selected body mensural characters in sterlet (literature published data) and sterlet aquaculture form (CR sa.). Explanations: a, b, c, R 2 = regression and determination coefficients of the linear and polynomial regressions for relation between TL and other head mensural characters; literature published data, CR sa., differ. (%) and diff. H (in %) = see explanation in Table 5; % lca = relative value of characters in % of head length. Character Coefficients Liter. data CR sa. Differ. Diff. H (in %) a b c R 2 (mm) (% lca) (mm) (% lca) (%) F1 F2 Head length E Preocular dist E Eye diameter E E Supraocular dist Interocular width Head depth max E Head depth at eye E E Dist. snout-mouth E Dist. snout-barb E Dist. barb.-mouth E Barbel length Snout width at barb Head width at mouth E Mouth width

158 Fig. 4. UPGMAphenogram of morphometric affinity into the 11 sterlet samples, 2 bester samples and 6 great sturgeon samples. Explanations see Fig. 3. Fig. 5. Relationship between the mouth width and TL in specimens of sterlet, giant sturgeon and intergeneric hybrids, H. huso x A. ruthenus (bester) and in specimens reared in the CR. dependence between the respective character value and TL was the polynomial function with convex curve (Table 6), which corresponds to the fact that the head relative size decreases with increasing TL and age in juvenile and adult specimens. The character of 160

159 differences between size-pooled samples of bester (F1, F2) and the sterlet mean sample (standard value) was not analogical, as it was in the case of differences between the sample from the CR and standard of sterlet mainly from autochthonous populations (Table 5). The dependence course in important determinative mensural (head) character, mouth width, is expressed conclusively in Fig. 5. It is evident that the mouth width in sterlet from the CR is significantly greater than that calculated in so-called sterlet standard (diff %), but simultaneously smaller than that in bester (diff. from -17.3% to 43.0%). The mouth shape in the material (juvenile specimens) from the CR is identical with that typical for sterlet. As well, the presence of barbel lashes (papillae) was proved in specimens from the CR and SR (Fig. 6). Fig. 6. Detail of barbels with touch papillae (lashes) situated in the lower third of their total length (A,D = outside barbels; B,C = inside barbels; line = 2 mm). Aquaculture form of sterlet reared in the CR. Photo by B. Koubková Discussion The used method of morphometrical data comparisons between samples from the CR and SR and the data reported in the literature, using calculations of parameters of so-called standard regressions for each character analysed individually and in absolute values (in mm), appears in our case really applicable and needed regards elimination of errors, which could arise in comparisons of specially non-adjusted data, or in comparisons of relative values. According to the conclusions by J a n k o v i ç (1958), P a v l o v (1967, 1968) and other authors, sterlet is not noted for sexual dimorphism and significant biometrical differences within natural populations inhabiting various catchments. Therefore, it is possible to pool parameters of different samples into one set representing the so-called pure species (zoological taxon). Since the amount of the literature available data on meristics and mensural morphological characters of sterlet were largely limited as to sample numbers, we had to use all data available to us for calculations of needed regressions. The conclusions found by analyses of meristic characters enable to the opinion that the samples from the CR and SR did not differ by character value ranges from the actually 161

160 accepted morphometrical description for the sterlet species (Du 32 49, Au 16 34(39), SD 11 18, SL 56 71, SV 10 20, Sp. br , Fu 25 45; H olãík 1989). However, as regards the mean values, 4 of 6 assessed meristic characters (66.6%) in the sample from the CR differed from the mean so-called standard data and this topic is discussed below. The trend of changes was in all cases towards the values found for bester F1 and F2. Significantly higher number of rays in Ain specimens from the CR (27.1) as compared with the standard (but within in individual samples) may be assessed as the result of environmental effects in special culture facilities, that of natural variability (cf. Pavlov 1968), but also as the result of artificial reproduction and breeding. Krylova (1980) found the mean number of rays in Ain great sturgeon 30.79, in sterlet of 25.87, and in bester F1 and F and respectively. The increased number of scutes (25.4) in the dorsal line in the CR sample, as compared with the standard, can be similarly of ecological and breeding consequences. K rylova (1980) found in great sturgeon scutae, in sterlet 13.7 scutae and in bester F1 and F and scutae respectively. Significantly reduced number of scutae (13.0) in the ventral line in the SR sample, as compared with the standard (14.1) occurs outside the range of all other samples ( ) and is close to the value found for bester F1 (13.06; K rylova 1980). Statistically insignificantly increased value in ray number in D (in sample from the CR = 44.5), but outside the value range of all other samples (sterlet range = ), occurs towards the value in bester F2 (47.96; K rylova 1980). Statistically insignificantly decreased value in the mean number of scutae in lateral line in sample from the CR (44.5), again outside the species range ( ), occured towards the value found in great sturgeon (45.38; Krylova 1980). Significantly greater mouth width in sterlet from the CR is, according to our opinion, a phenomenon manifesting the adaptation to more productive aquaculture rearing, as we also found in Siberian sturgeon (Proke et al. 1997a,b). As regards the fact that, according to the results by F laj hans & Vajcová (2000), potential theoretical influences by sturgeon with 240 chromosomes can be excluded in the case of sterlet sample from the CR. On the basis of assessed and compared characters, we consider specimens reared in the Mydlovary Water Production Farm in the CR to be sterlet aquaculture population, with their meristic, mensural and other typical morphological characters, are close to the sterlet autochthonous population. Specimens reared under aquaculture in the SR approach very close to the Danube autochthonous sterlet population. Acknowledgements Financial support was given by the Grant Agency of the Academy of Sciences of the Czech Republic, grants no. A and no. A LITERATURE ABDURAKHMANOV, Yu. A., 1962: [The freshwater fishes of Azerbaidzhan]. Izdat. Akad. Nauk Azerb. SSR, Baku, 407 pp. (in Russian). ARTYUKHIN, E.N. & ROMANOV, A.G., 1997: [About inside genus classification of sturgeon taxonomy]. In: First congress of ichthyologists of Russia. Theses of reports (Astrakhan, September 1997). Moscow, VNIRO Publisher House, p. 8 (in Russian). BANARESCU, P., 1964: Fauna Republici Populare Romine. Pisces Osteichthyes. Pesti ganoizi si ososi [Fauna of the Romanian People s Republic. Fishes Osteichthyes]. Vol. XIII. Ed. Acad. R.P.R., Bucuresti, 962 pp. (in Rumanian). 162

161 BARU, V. & OLIVA, O. (eds), 1995: Fauna âr a SR, sv. 28/1. Mihulovci Petromyzontes a ryby Osteichthyes (Fauna of the Czech Republic and Slovak Republic, vol. 28/1. Lampreys Petromyzontes and Fishes Osteichthyes). Academia, Praha, 623 pp. (in Czech, with a summary in English). BERG, L. S., 1948: [The freshwater fishes of the USSR and adjacent countries I]. Izd. AN SSSR, Moskva Leningrad, 466 pp. (in Russian). BIRSTEIN, V.J. & VASILYEV, V.P., 1987: Tetraploid octoploid relationships and karyological evolution in the order Acipenseriformes (Pisces). Karyotypes, nucleoli, and nucleolus organizer regions in four acipenserid species. Genetica, 72(1): BIRSTEIN, V. J., HANNER, R. & DE SALLE, R., 1997: Phylogeny of the Acipenseriformes: cytogenetic and molecular approaches. Environ. Biol. Fish., 48: BIRSTEIN, V. J., POLETAEV, A. I. & GONCHAROV, B. F., 1993: The DNA content in Eurasian sturgeon species determined by flow cytometry. Cytometry, 14: DOBROVOLOV, I. & DOBROVOLOVA, S., 1983: Electrophoretic studies on proteins of great sturgeon, Huso huso (L.), the sterlet, Acipenser ruthenus L., bester (H. huso (L.) x A. ruthenus (L.)) and the Russian sturgeon, Acipenser gueldenstaedti Brandt. Varna Proc. Inst. Fish., Varna, 20: FLAJ HANS, M. & VAJCOVÁ, V., 2000: Odd ploidy levels in sturgeons suggest a back cross of interspecific hexaploid sturgeon hybrids to evolutionarily tetraploid and/or octaploid parental species. Folia Zool., 49(2): FONTANA, F., JANKOVIâ, D. & ÎIVKOVIâ, S., 1977: Somatic chromosomes of Acipenser ruthenus L. Arch. Biol. Nauka, Beograd, 27(1 2): FONTANA, F., 1994: Chromosomal nucleolar organizer regions in four sturgeon species as markers of karyotype evolution in Acipenseriformes (Pisces). Genome, 37(5): HANEL, L., 1992: Poznáváme na e ryby [We recognize our fishes]. Brázda, Praha, 288 pp. (in Czech). HOLâÍK, J. (ed.), 1989: The freshwater fishes of Europe. Vol. 1, Part II. General introduction to fishes Acipenseriformes. AULA Verlag Wiesbaden, 469 pp. HOLâÍK, J., 1998: Ichtyológia [Ichthyology]. Príroda, Bratislava, 310 pp. (in Slovak). HUBÁâEK, J., 1950: Jeseter mal v na ich rybnících [The sterlet in our ponds]. âeskoslovensk rybáfi, 5(9): (in Czech). JANKOVIå, D., 1958: Ekologija dunavske keãige (Acipenser ruthenus L.) [Ecology of the Danube sterlet (Acipenser ruthenus L.)]. Biolo ki institut N.R. Srbije, Posebna izdanja, knjiga 2, Beograd, 145 pp. (in Serbian). JIRÁSEK, J., 1999a: Jesetefii v akvakulturách [The sturgeons in aquacultures]. Rybáfiství, 10/1999: (in Czech). JIRÁSEK, J., 1999b: PfieÏijí jesetefii 20. století? [Survival of sturgeons behind the 20 th century?]. Rybáfiství, 11/1999: (in Czech). KOSTOMAROV, B., 1947a: O jeseteru malém (1. ãást) [About the sterlet (l st part)]. âeskoslovensk rybáfi, 2(7): (in Czech). KOSTOMAROV, B., 1947b: O jeseteru malém (2. ãást). [About the sterlet (2 nd part)]. âeskoslovensk rybáfi, 2(9): (in Czech). KRUPKA, I., 2000: Boli doma aj u nás [At home this were also in our country]. Poºovníctvo a rybárstvo, 5(3): (in Slovak). KRYLOVA, V. D., 1980: [Variability and inheritance of characters in F sub(1) and F sub(2) hybrids of great sturgeon (beluga) and sterlet Huso huso (L.) x Acipenser ruthenus L., in connection with selection work]. Vopr. Ikhtiol., 20(2): (in Russian). KUZMIN, E.V., 1996: The albumins system of blood serum of Acipenseriformes in the fluviatile period of their life. J. Ichthyol., 36(1): LUKIN, A.V., 1979: [The sterlet of the Kuibyshev water reservoir. In: Biologicheskie osnovy razvitiya osetrovogo khozyaistva v vodoemakh SSSR]. Izd. Nauka, Moskva: (in Russian). LUKIN, A.V., KUZNETSOV, V.A., KHALITOV, K.P., DANILOV, N.N., TIKHONOV, K.P. & MELENT EVA, R.R., 1981: The sterlet of the Kuibyshev water reservoir and ways of his adaptation in new existence. Izd. Kazanskogo univ., Kazan, cited after Holãík LUSK, S. & HANEL, L., 1996a: âerven seznam mihulí a ryb âeské republiky verze 1995 (The red list of lampreys and fishes in the Czech Republic version 1995). In: Lusk, S. & Halaãka, K. (eds), Biodiversity of fishes in the Czech Republic (I). Ústav ekologie krajiny AV âr Brno: (in Czech, with a summary in English). LUSK, S. & HANEL, L., 1996b: Druhová diverzita ichtyofauny âeské republiky (Biodiversity of the ichthyofauna in the Czech Republic). In: Lusk, S. & Halaãka, K. (eds), Biodiversity of fishes in the Czech Republic (I). Ústav ekologie krajiny AV âr Brno: 5 15 (in Czech, with a summary in English). 163

162 LUSK, S. & HANEL, L., 2000: âerven seznam mihulí a ryb âeské republiky verze 2000 (The red list of lampreys and fishes in the Czech Republic version 2000). In: Lusk, S. & Halaãka, K. (eds), Biodiversity of fishes in the Czech Republic (III). Ústav biologie obratlovcû AV âr Brno: 5 13 (in Czech, with a summary in English). LUSK, S., HALAâKA, K., LUSKOVÁ, V. & PRAÎÁK, O., 1996: Fish assemblages of the Soutok area in Southern Moravia, Czech Republic. Acta Univ. Carolinae, Biologica, 40(1 2): LUSK, S., LUSKOVÁ, V., HALAâKA, K. & LOJKÁSEK, B., 2000: Zmûny v druhové skladbû ichtyofauny na území âeské republiky po roce 1990 (Changes in the species composition of the ichthyofauna in the territory of the Czech Republic after 1990). In: Lusk, S. & Halaãka, K. (eds), Biodiversity of fishes in the Czech Republic (III). Ústav biologie obratlovcû AV âr Brno: (in Czech, with a summary in English). MEN SHIKOV, M.I. & BUKIREV, A.I., 1934: [Fishes and fishery in the upper part of the River Kama]. Trudy Biol. Nauchno-issl. Inst. Permsk. univ., 6: (in Russian). NIKOLYUKIN, N. I., 1972: [Distant hybridization of acipenserid and bony fishes. Theory and practice]. Pishchevaya promyshlennost, Moskva 1972, 335 pp. (in Russian). OLIVA, O. & CHITRAVADIVELU, K., 1972: On the systematics of the sterlet, Acipenser ruthenus Linnaeus, 1758 (Osteichthyes, Acipenseridae). Vûstník ãs. Spol. zool., 36: PAVLOV, P.I., 1967: Morphological peculiarities of Danube beluga Huso huso ponticus Sal nikov et Maliatskiy. Gidrobiologicheskiy zhurnal, 3: (in Russian, with a summary in English). PAVLOV, P.I., 1968: On the variability degree of Acipenser ruthenus L. of the Danube and Dnieper. Gidrobiologicheskiy zhurnal, 4: (in Russian, with a summary in English). PRÁ IL, O., 2000: Zpráva o úlovcích nejvût ích ryb v roce 1999 [Report about catches of the largest fishes in 1999]. Rybáfiství, 3/2000: (in Czech). PRAVDIN, I. F., 1966: [Handbook of ichthyology]. Izd. Pishchevaya promyshlennost, Moskva, 374 pp. (in Russian). PROKE, M., BARU, V. & PE ÁZ, M., 1995: Morphological analysis of 0+ juvenile giant sturgeon (Huso huso) reared in the Czech Republic for first time. Folia Zool., 44(3): PROKE, M., BARU, V. & PE ÁZ, M., 1997a: Meristic and plastic characters of juvenile Siberian sturgeon (Acipenser baerii) imported to the Czech Republic in Folia Zool., 46(1): PROKE, M., BARU, V. & PE ÁZ, M., 1997b: Parameters of Siberian sturgeon domestication in aquaculture. In: First congress of ichthyologists of Russia. Theses of Reports (Astrakhan, September 1997). Moscow, VNIRO Publisher House, p. 295 (in Russian). PROKE, M., BARU, V. & PE ÁZ, M., 1997c: Comparative growth of juvenile sterlet (Acipenser ruthenus) and Siberian sturgeon (A. baerii) under identical experimental condition. Folia Zool., 46(2): PROKE, M., BARU, V. & PE ÁZ, M., 1999: Morphometry, systematics and growth parameters of sturgeons (Acipenseridae) introduced in the Czech Republic. In: Biennial report Institute of Vertebrate Biology, Academy of Sciences of the Czech Republic, Brno 1999, pp PROKE, M., BARU, V. & PE ÁZ, M., 2000a: Morfometrická a rûstová rozmanitost u druhû jeseterû chovan ch v âeské republice v letech (Morphometry and growth variety of sturgeon species reared in the Czech Republic in ). In: Lusk, S. & Halaãka, K. (eds), Biodiversity of fishes in the Czech Republic (III). Ústav biologie obratlovcû AV âr Brno: (in Czech, with a summary in English). PROKE, M., BARU, V. & PE ÁZ, M., 2000b: Akvakulturní chov jeseterû v âeské republice (Aquaculture rearing of sturgeons in the Czech Republic). In: Mike ová, J. (ed.), Sb. referátû ze IV. âeské ichtyologické konference. Jihoãeská univerzita v âesk ch Budûjovicích, V zkumn ústav rybáfisk a hydrobiologick ve VodÀanech, VodÀany, pp (in Czech, with a summary in English). RÁB, P., 1986: Anote on the karyotype of the sterlet, Acipenser ruthenus (Pisces, Acipenseridae). Folia Zool., 35(1): SEREBRYAKOVA, J.V., 1979: Peculiarity of karyotypes in Acipenseriformes. 3. Europ. Ichthyol. Congr., Warszawa Sept., Abstracts, pp SOKOLOV, L.I. & VASILYEV, V.P., 1989: Acipenser ruthenus Linnaeus, In: Holãík, J. (ed.), The freshwater fishes of Europe. Vol. 1, Part II. General introduction to fishes Acipenseriformes. AULA Verlag Wiesbaden, pp VLADYCHESKAYA, N.S. & KEDROVA, O.S., 1982: Genome structure in interspecific fish hybrids. Genetica, 18(10): ZELINKA, M., 2000: Chránûné druhy ryb [The protected fish species]. Sportovní rybáfi, 4(1): (in Czech). 164

163 Folia Zool. 51(2): (2002) Aberrant meioses in diploid and triploid progeny of gynogenetic diploids produced from eggs of natural tetraploid loach, Misgurnus anguillicaudatus Quanqi ZHANG 1, Kumiko HANADA 2 and Katsutoshi ARAI 3 * 1 Marine Life Science College, Ocean University of Qingdao, #5 Yushan road, Qingdao , China 2 Faculty of Applied Biological Science, Hiroshima University, Kagamiyama, Higashi-hiroshima , Japan 3 Graduate School of Fisheries Sciences, Hokkaido University, Hakodate , Japan; araikt@fish.hokudai.ac.jp Received 30 April 2001; Accepted 20 March 2002 A b s t r a c t. Gynogenetic diploids were produced from the eggs of natural tetraploid loach Misgurnus anguillicaudatus (Pisces: Cobitidae) without any manipulation for chromosome duplication. When eggs of a four-year-old diploid gynogenetic individual were fertilized with spermatozoa of specimens from normal diploid and natural tetraploid lines, viable diploid and triploid progeny were produced, respectively. Thus, egg nucleus of the diploid gynogen is haploid. In the gonads of diploid progeny, diploid (2n = 50) and tetraploid (4n = 100) mitotic metaphases were observed. The majority of oocytes (76%) showed regular 25 bivalents as in normal diploids, but the other 16% showed a few univalents. The remaining 8% exhibited about 50 bivalents, suggesting chromosome duplication by premeiotic endomitosis. In the testes, a few spermatocytes (6%) showed normal 25 bivalents, but 86% contained various number of univalents and the remaining 8% contained about 50 bivalents. No peaks of spermatozoa were identified in the testis by flow cytometry. In the triploid progeny, triploid (3n = 75) and hexaploid (6n = 150) mitotic metaphases were observed in both ovaries and testes. Most meiotic figures (about 90%) contained approximately 25 bivalents and 25 univalents in both sexes; the rest contained approximately 75 or more bivalents. Spermatozoa were not identified in the testis by flow cytometry. Thus, the diploid males between the diploid gynogens and common diploid, and both sexes of triploids between the diploid gynogens and tetraploid, show aberrant meioses such as frequent formation of univalents, but the diploid females seem to be less affected. Key words: polyploid, gynogenesis, univalent Introduction In the loach Misgurnus anguillicaudatus (Pisces: Cobitidae), most individuals in Japanese populations have 50 chromosomes (2n = 50), but tetraploid individuals with 100 chromosomes have been found in the specimens obtained from fish dealers (O j i m a & Takai 1979, Arai et al. 1991a). The individuals with 100 chromosomes are considered genetically true tetraploids with four sets of homologous chromosomes, because viable gynogenetic and androgenetic diploids were produced using their gametes without treatment to double their chromosomes (Arai et al. 1993, 1995). Although the actual origin of these tetraploids has not been identified and no tetraploid individuals were found in a total 1815 wild individuals collected from 33 locations in Japan (Z ha n g & Arai 1999a), the same species from Hubei Province, China was reported to be tetraploid with 100 chromosomes (L i et * Corresponding author 165

164 al. 1983). Thus, the tetraploid specimens are likely to originate from the Asian continent and are genetically different from common diploid loach in Japan. Triploid males produced by fertilizing eggs of common diploids with spermatozoa of tetraploids formed no spermatozoa and were absolutely sterile (Matsubara et al. 1995, Zhang & Arai 1996), but triploid females simultaneously laid large and small eggs, with triploid and haploid female pronucleus after fertilization, respectively (M a t s u b a r a et al. 1995). Multilocus DNA fingerprinting revealed that most triploid gynogens artificially induced from the large eggs of such triploids belonged to a clone genetically identical to their mother (Arai & Mukaino 1997). Cytogenetic studies demonstrated that the unreduced triploid eggs were developed from hexaploid oocytes generated by premeiotic endomitosis, whereas haploid eggs were produced from triploid oocytes with both bivalents and univalents in the first meiotic prophase, but eliminated unpaired univalents during later processes of meiosis (Z ha n g et al. 1998). These atypical mechanisms of reproduction may be caused by the imbalance of genomes between diploid and tetraploid parents. To obtain further insight into the genetic incompatibility between normal diploid and natural tetraploid loach, diploid gynogens were artificially produced by inseminating eggs of tetraploids with UV-irradiated heterospecific spermatozoa. Thereby, the resultant diploid gynogens should have two genomes (sets of chromosomes) derived from the tetraploid loach. In the present study, we fertilized the eggs of a mature diploid gynogen with spermatozoa of normal diploid and natural tetraploid lines, respectively, and then produced diploid progeny comprising one genome from diploid and one genome from tetraploid as well as triploid progeny comprising three genomes from tetraploid loach. We observed the survival potential of these two types of progeny as well as metaphase spreads of premeiotic cells and counted chromosomes to examine the occurrence of endomitosis during gametogenesis. We also observed chromosome configurations in the early meiotic stages of oocytes and spermatocytes to examine the abnormalities of meiotic chromosome behaviors in the diploid and triploid progeny. Materials and Methods Tetraploid individuals of loach were found among the specimens obtained from a fish dealer in Mie Prefecture, Japan (Arai et al. 1991a) and tetraploid line was made by mating between them (Arai et al. 1991b, 1993). Normal diploid individuals were caught in paddy field in Sera town, Hiroshima Prefecture, Japan. Diploid gynogens (G2n) were produced by inseminating eggs of a tetraploid female with UV-irradiated carp (Cyprinus carpio) spermatozoa in 1993 (Z ha n g & Arai 1996). One surviving diploid gynogen reached maturation in Using the methods previously reported (Arai et al. 1993), the eggs of the mature gynogen were fertilized with spermatozoa of two normal diploid males (A and B; G2n 2nA and G2n 2nB, respectively). The eggs were also fertilized with spermatozoa of a laboratory-reared tetraploid male (G2n 4n) that was selected from the progeny of a natural tetraploid pair. The ploidy status of all parental fish used in the crosses, as well as six- and eight-monthold progeny, was determined by DNA-content flow cytometry according to the method described by Z ha n g & A rai (1996). Nine-day-old progeny were minced finely with forceps in two drops of Ca, Mg-free phosphate buffered saline solution (CMF-PBS, 166

165 Zhang & Arai 1996) in 1.5 ml microcentrifuge tubes, and then all suspension pipetted with a 1-ml syringe equipped with a 22 gauge needle after adding 0.5 ml of CMF-PBS. The cell suspensions were centrifuged at 130g (1200 rpm) for 5 min, decanted, stained and DNA content was measured by flow cytometry. Chromosome preparation of eyed embryos and hatched progeny was performed according to Arai et al. (1991b). At eight months of age, two females and four males ( cm in total length) were taken from each group of G2n 2nA and G2n 4n crosses and sacrificed for chromosomal examination after confirmation of their ploidy by flow cytometry. The testes of two diploids and two triploids were assayed by flow cytometry (Z ha n g & Arai 1996) to detect the ploidy status of gonadal cells. The remaining individuals were used for chromosome observation in the gill, spleen and gonad from eight-month-old fish according to the previously described method (Z h a n g et al. 1998). Results Survival of the progeny The gynogenetic individuals induced from eggs of tetraploids were confirmed to be diploid by flow cytometry. When a mature gynogenetic individual was crossed with two normal diploids (G2n 2n) and one tetraploid (G2n 4n), the mean egg diameter of this gynogenetic progeny (1.11±0.03 SD mm, n = 35) was not different from that of normal diploid one (1.11±0.06 SD mm, n = 49). One diploid male (A) and a tetraploid male gave fair fertilization rates (Table 1), whereas the other diploid male (B) showed lower fertilization rate. A higher hatching rate was observed in G2n 2nA, but a low rate of normal progeny (4.1%) was recorded because 82.7% (321/ total 388) of progeny were abnormal at hatching. At three weeks after fertilization, a small proportion of individuals survived (Table 1). G2n 2nB cross showed poor rates of hatching and normal progeny(0.7%). Abnormal progeny comprised 74.2% (72/total 97) at hatching and few individuals survived at three weeks after fertilization. Comparatively, G2n 4n cross showed lower hatching rate than G2n 2nA cross did, but higher rate of normal progeny (7.4%). A small proportion of progeny were abnormal at hatching (Table 1), a small proportion survived to three weeks after fertilization. Ploidy status of the progeny Of the 31 specimens from the G2n x 2nA cross, 29 were diploid and two were triploid (Table 2). On the contrary, all 18 specimens from the G2n x 4n cross were triploid. Table 1. Developmental potential of the progeny of the diploid gynogen (G2n) produced from eggs of natural tetraploid loach when crossed with common diploid (G2n 2n) and natural tetraploid (G2n 4n). 1: Relative to no. of eggs used, 2: Relative to no. of hatching, 3: Relative to no. of eggs used; Recorded at three weeks after fertilization. Cross No. of Fertilization Hatching Normal progeny Survival eggs No. (%) 1 No. (%) 1 No. (%) 2 No. (%) 3 G2n 2n A G2n 2nB G2n 4n

166 Table 2. Ploidy status of the progeny of the diploid gynogenetic individual (G2n) produced from eggs of natural tetraploid loach when crossed with common diploid (G2n 2n A) and natural tetraploid (G2n 4n) loaches. 1: Ploidy was determined by counting chromosome numbers (CHM), 2: Ploidy was determined by DNA -content flow cytometry (FCM). Cross Age Ploidy No. of Ploidy determination sample Diploid (2n) Triploid (3n) G2n 2n A Eyed embryo CHM Hatched progeny CHM day-old progeny FCM month-old fish FCM month-old fish FCM Total G2n 4n Eyed embryo CHM Hatched progeny CHM day-old progeny FCM month-old fish FCM month-old fish FCM Total Chromosomes of gonadal cells In two diploid females derived from the G2n 2nA cross (Table 3), all 56 examined cells from the gills and spleen showed the diploid chromosome number (2n = 50) (Fig. 1a), whereas in their ovaries 6.5% of the metaphase spreads were tetraploid (4n = 100) (Fig. 1b). All 92 somatic cells of the two diploid males were diploid (Table 3). In the testes, however, 13% were polyploid (including tetraploid, hexaploid and octaploid) metaphase spreads, although the other cells were diploid (Table 3). Somatic cells in the gills and spleen of two 8-month-old triploid (G2n 4n) females showed only triploid metaphase figures (Table 3, Fig. 1c). In contrast to the triploidy of somatic tissues, the ovaries of one female contained about 17% hexaploid cells (Fig. 1d). In Table 3. Ploidy status of somatic (gill / spleen) and gonadal (ovary / testis) cells in the diploid (G2n 2n A) and triploid (G2n 4n) progeny of the diploid gynogenetic individual produced from eggs of natural tetraploid loach. Cross Sex FishGill/spleen Ovary/testis # 2n 3n Sum 2n 3n 4n 6n 8n 9n 12n Sum G2n 2n A Female Sum Male Sum G2n 4n Female Sum Male Sum

167 the ovaries of the other female, only 8 triploid cells were observed because of few readable metaphase spreads, and no hexaploid cells were found. Testes of the triploid males contained 14% hexaploid cells, and 1.7% 9n and 3.3% 12n cells (Table 3). In the testes, 19% were polyploid. DNA content of testicular cells Flow cytometrical analysis of DNA content of the testes of normal diploids revealed one prominent peak of 1C spermatozoa, one low peak of 2C somatic cells and secondary Fig. 1. Mitotic metaphases prepared from diploid (G2n 2n) and triploid (G2n 4n) progeny. (a) diploid cell (2n = 50) from gill of the diploid progeny, (b) tetraploid cell (4n = 100) from ovary of the diploid progeny, (c) triploid cell (3n = 75) from gill of the triploid progeny, (d) hexaploid cell (6n = 150) from ovary of the triploid progeny. 169

168 Fig. 2. Flow cytometric histograms in testes of the progeny of diploid gynogen produced from eggs of tetraploid. (a) normal diploid showing 1C spermatozoa, interphase 2C and DNA replicated 4C cells, (b) diploid progeny from G2n x 2nA cross showing 2C, 4C and a small 8C ranges of cells but no peak representing spermatozoa and spermatids, (c) triploid progeny from G2n x 4n cross showing 3C, 6C and a small 12C ranges of cells but no peak corresponding to spermatozoa and spermatids. The 2C and 4C peaks in diploid progeny and the 3C and 6C peaks in triploid progeny are markedly higher than the corresponding 2C and 4C peaks of normal diploid. spermatocytes, and one low 4C peak of DNA replicated somatic cells and primary spermatocytes (Fig. 2a). Nevertheless, the testes of diploid progeny derived from the G2n 2nA cross exhibited two distinct peaks of 2C and 4C cells (Fig. 2b). Both peaks were markedly higher than the corresponding peaks of normal diploids. A low peak in the 8C range, representing the G2 phase of tetraploid cells, was also detected. However, no peak representing spermatozoa was recognized (Fig. 2b). Similarly, triploid males derived from the G2n 4n cross contained unusually high peaks in the 3C and 6C ranges and one low peak in the 12C range, respectively (Fig. 2c). No peak of lower ploidy level representing spermatozoa was detected. Meioses In the ovaries of eight-month-old diploids from the G2n x 2nA cross, 37 pachytene oocytes were observed and they were categorized into two groups according to their sizes (Table 4): 170

169 small oocytes represented 91.9% of which 75.7% contained only 25 bivalents (Fig. 3a), whereas the remaining 16.2% of small oocytes showed bivalents fewer than 25 and four, six, or infrequently 20 univalents (Fig. 3b,c); large oocytes represented 8.1% and had 50 or close to 50 bivalents (Fig. 3d). In the two diploid male progeny from G2n 2nA cross (Table 4), spermatocytes were observed and were categorized into two types. 91.7% were small spermatocytes of which 5.6% showed regular 25 bivalents (Fig. 3e); whereas the other 86.1% spermatocytes contained 21 to 24 bivalents and 2 to 8 unpaired univalents (Fig. 3f-h). These univalents generally congregated in one side of the cell and some of them showed delayed condensation (Fig. 3g,h). Large spermatocytes represented 8.3% and the numbers of chromosome elements were difficult to count exactly. Two showed approximately 50 bivalents and the third one showed about 50 bivalents and a few univalents (Fig. 3i). Ovaries of triploid progeny from the G2n x 4n cross also had small and large pachytene oocytes (Table 4). The small oocytes represented 91.4% and had about 25 thick elements, probably bivalents, and about 25 thin elements, probably univalents (Fig. 4a). The five large oocytes (8.6%) had approximately 75 to no fewer than 60 countable bivalents and a few thin chromosome elements (Fig. 4b). Neither large nor small oocytes beyond the pachytene stage were observed. In triploid males, 92.3% of the spermatocytes showed about 25 bivalents and about 25 univalents at meiosis I (Table 4, Fig. 4c). In the remaining spermatocytes (7.7%), 6.7% showed approximately 75 bivalents (Fig. 4d), but the eighth cell gave more than 100 bivalents. Table 4. Variation in numbers of bivalent and univalent chromosomes observed in the diploid (G2n 2n A) and triploid (G2n 4n) progeny of the diploid gynogenetic individual produced from eggs of natural tetraploid loach. 1: The numbers of bivalents in these large oocytes / spermatocytes were approximate numbers because of some overlaps. Progeny Sex No. of No. of Small oocytes / spermatocytes 25 II+ 24 II+ 23 II+ 22 II+ 21 II+ 15 II+ 25 II+ Sum (Cross) fish meiosis 0 I 2 I 4 I 6 I 8 I 20 I 25 I Diploid F (G2n 2n A) M Triploid F (G2n 4n) M Progeny Sex No. of No. of Large oocytes / spermatocytes 50 II Sum (Cross) fish meiosis II 1 II 1 Diploid F (G2n 2n A) M Triploid F (G2n 4n) M

170 Fig. 3. Meiotic chromosome configurations in diploid progeny (G2n 2nA). Typical bivalents and univalents indicated by arrows and arrowheads, respectively, (a) 25 bivalents in a pachytene stage oocytes, (b) 23 bivalents and 4 univalents in a pachytene stage oocyte, (c) 15 bivalents and 20 univalents in a pachytene stage oocyte, (d) 50 bivalents in a large pachytene stage oocyte, (e) 25 bivalents in a spermatocyte, (f) a spermatocyte showing 24 bivalents and 2 univalents, (g) a spermatocyte showing 23 bivalents and 4 univalents, (h) a spermatocyte showing 22 bivalents and 6 univalents, (i) a spermatocyte of diploid showing approximately 50 bivalents and a few univalents. Some of the univalents in (g) and (h) showed delayed condensation compared with the bivalents. Discussion Viable gynogenetic diploid (G2n) individuals produced from eggs of a tetraploid individual female in 1993 reached maturation in A large number of mature eggs were obtained from one of the diploid gynogens. Viable diploid and triploid progeny were produced from 172

171 Fig. 4. Meiotic chromosomes configurations in triploid progeny (G2n 4n). Arrows and arrowheads show typical bivalents and univalents, respectively. (a) approximately 25 bivalents and 25 univalents in pachytene stage oocyte of triploid, (b) a large oocyte showing at least 60 bivalents and a few univalents, (c) approximately 25 bivalent and 25 univalent chromosomes in a meiosis I spermatocyte of triploid, (d) a large spermatocyte showing approximately 75 bivalents. these eggs by fertilization with spermatozoa of diploid (2n) and tetraploid (4n) males, respectively. Although the hatching and normal rates obtained in these crosses (Table 1) were lower than those normally observed in the crosses by common diploid pairs (42 to 68 % in hatching and 42 to 68% in normal rates; Suwa et al. 1994), our results indicated that the gynogenetic diploid (G2n) laid functional haploid eggs. However, low viability of the progeny can be explained by aneuploid eggs spawned by the G2n and subsequent death due to aneuploid embryos. On the other hand, the haploidy of the eggs provided additional evidence for the hypothesis of genetically true tetraploidy of the original grandparent (Arai et al. 1993, 1995), because the gynogens derived from eggs of tetraploids were not only viable, but also able to form haploid eggs by means of normal meiosis. The small number of 173