Predation, Competition, and Zooplankton Community Structure: An Experimental Study

Transcription

1 Predation, Competition, and Zooplankton Community Structure: An Experimental Study Michael Lynch Limnology and Oceanography, Vol. 24, No. 2. (Mar., 1979), pp Stable URL: Limnology and Oceanography is currently published by American Society of Limnology and Oceanography. Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. The JSTOR Archive is a trusted digital repository providing for long-term preservation and access to leading academic journals and scholarly literature from around the world. The Archive is supported by libraries, scholarly societies, publishers, and foundations. It is an initiative of JSTOR, a not-for-profit organization with a mission to help the scholarly community take advantage of advances in technology. For more information regarding JSTOR, please contact support@jstor.org. Thu Aug 30 11:56:

2 Limnol. Oceanogr., 24(2), 1979, by the American Society of Limnology and Oceanography, Inc. Predation, competition, and zooplankton community structure: An experimental study1y2 Michael Lynch Department of Ecology, Ethology, and Evolution, University of Illinois, Urbana Abstract An experimental investigation of the zooplankton community of a small Minnesota pond was conducted for 2 years to determine the mechanisms maintaining its structure and to show that it is predictably organized. A mechanistic interpretation of the strncture of this community cannot be made solely on the basis of predation, but also requires evaluation of the relative competitive abilities of the herbivores. The presence of Chaoborus and fish place predictable constraints on the abundance of zooplankton species in this pond. The competitive dominant (Ceriodaphnia reticulata) is of intermediate size and is removed when either predator is abundant. In the presence of intense Chaoborus predation, Ceriodaphnia is replaced by a larger subordinate competitor, Daphnia pulex; when fish predation is intense, smaller species (Bosmina longirostris and rotifers) increase. These small species are also able to maintain large populations in vertebrate-free environments when Chaoborus is rare. When these small herbivores are abundant, two additional invertebrate predators (Cyclops vernalis and Asplanchna priodonta) arrive, neither of which seems able to reduce its prey to extinction. Recent studies provide ample evi- nature. The information they are based dence that predator-prey interactions are on does not allow a mechanistic interpreof major significance in structuring fresh- tation of zooplankton community strucwater zooplankton communities (see Hall ture. For instance, while it is well docuet al. 1976). Vertebrate planktivores (fish mented that large herbivores disappear and salamanders) feed visually and re- when vertebrate planktivores are intromove the largest and most conspicuous duced to a community, no data exist to zooplankton (Brooks 1968; Werner and determine whether their extinction is en- Hall 1974; Zaret and Kerfoot 1975; tirely a result of direct removal by ver- O'Brien et al. 1976); invertebrate preda- tebrates. It is possible that vertebrates, tors (predaceous copepods and rotifers, through their effects on the rest of the midge larvae, and Leptodora) cannot community, may impose other deleterihandle the large herbivores but prey ex- ous conditions on the large herbivores tensively on small species (Dodson (e.g. increased competition with newly 1974a; Fedorenko 1975~; Kerfoot 1977); established smaller herbivores). Furtherwhen vertebrates are present the larger, more, there is no direct evidence that inmore conspicuous invertebrate predators vertebrate predation can be sufficient to are replaced by smaller, less conspicuous cause extinction of small herbivores. Fiones (Dodson 1970, 1974~). nally, the role of herbivore competition These generalizations form the bases in structuring zooplankton communities for recent conceptual models of zoo- is poorly understood (Lynch 1977a), and plankton community structure (Dodson nothing is known about competition be- 1974a; Zaret in prep.). However, despite tween invertebrate predators. their success at predicting the distribu- To eliminate these ambiguities in intion of different zooplankton species, terpreting zooplankton community structhese models are phenomenological in ture, I examined simultaneously the predatory and competitive interactions in the community of a small Minnesota Contribution 172 from the Limnological Repond. Here I report on the mechanisms search Center, University of Minnesota. Work supported by National Science Founda- determining the relative abundances of tion grant EMS to J. Shapiro. species in this pond and show that the

3 254 Lynch Table 1. Bluegill sunfish added and recovered from enclosures. Enclosures 1and 7 were controls. Since its one fish only lived for 5 days, enclosure 2 is treated as a control. Enclosure Added-18 Jun Recovered-2 Aug Length (mm) Wt (g) Length (mm) Wt (9) zooplankton community is organized predictably. Although this study has shortcomings (as most community-level investigations do), it permits rejection and support of existing hypotheses and the development of a general theory for zooplankton community structure. I thank J. Lynch, W. Combs, B. Forsberg, G. Jacobson, G. Lie, G. Lindmark, B. Monson, J. Shapiro, V. Smith, and E. Swain for help with fieldwork. S. Anderson, S. Dodson, D. Hall, J. Shapiro, R. Taylor, and T. Zaret provided criticisms throughout this study. Methods Pleasant Pond (about 0.25 ha; max depth, 2.5 m) is about 10 km north of St. Paul, Minnesota. The basin, which was dug about 1955, has no inlet or outlet and lies in a sandy outwash plain of glacial origin. One side is bordered by a gravel road but otherwise the pond is surrounded by oak forest and some secondary growth. The pond was completely devoid of fish at the beginning of this study (June 1975) and remained so until summer It has natural populations of salamanders (Necturus maculosus) and painted turtles (Chrysemys picta), but I found no evidence that they were significant predators on zooplankton. From 15 June to 25 July 1975 several enclosures were suspended from wooden floats in the pond. Twelve 1-m-diameter polyethylene bags (1.8 m deep and closed at the bottom) were filled with surface water and stocked with five different densities of bluegill sunfish (Lepomis macrochirus) in duplicate (leaving two controls). Two of the fish enclosures were destroyed during the experiment, and fish escaped from two others. Therefore, the results for only eight of these enclosures are considered here (Table 1). In addition to bags 1and 7, bag 2 is treated as a control in the following discussion since its one fish lived for only 5 days. The control bags provided an effective test of the effects of Chaoborus predation on the zooplankton community in the pond. Since most Chaoborus were near the bottom of the pond while the bags were being filled, they were effectively excluded from the enclosures. Only after midge eggs began to hatch in the bags in the middle of July did Chaoborus begin to increase, and even then it was much more numerous in the pond itself. On 27 April 1976, the pond was divided in half with a double curtain of nylonreinforced polyethylene film weighted firmly in the sediments with steel chain (0.95cm) and held about 25 cm above the surface by cable and styrofoam floats. The north half (hereafter Pleasant Pond North or PP N) was stocked with 10,000 walleye (Stixostedion uitreum) fry (aged 1day) on 5 May. Although we seined the pond extensively throughout summer, we neither saw nor captured a walleye. In early October, the Minnesota Department of Natural Resources intensively trap-netted Pleasant Pond North for 2 days and captured 306 fish (total weight, 7.7 kg). On 5 June 1976 several mature fathead minnows (Pimephales promelas) were noticed in the center of the southern half of the pond (Pleasant Pond South or PP S). These minnows were extremely pro-

4 Zooplankton community structure 255 lific. By 13 June, schools of newly hatched minnows were visible near the shore, and by 3 July many schools were swimming through the center of the pond. On 7 June, a sampling of a 19-m2 area with a 3-mm-mesh seine yielded 20 minnows (avg length, 2 cm). A second sampling on 3 August yielded about 500 minnows (avg length, 4 cm). The origin of these fish is unknown. However, since neither these nor any other species invaded PP N and since very few walleye survived, we had a useful whole pond experiment, even though it was not as originally designed. Zooplankton were sampled weekly throughout summer by vertical tows (duplicates in the pond, single in the enclosures) with a Wisconsin net (15-cm-diameter opening, 64-pm netting). The average coefficients of variation for four samplings on three occasions were 0.36 for cladocerans, 0.45 for nauplii, 0.59 for copepodites and adult copepods, and 0.59 for rotifers. All samples were immediately fixed with Formalin and sucrose (Haney and Hall 1973) to prevent loss of eggs; pond samples were anesthetized with carbonated water before fixing to prevent evacuation of the guts. At least two 1-ml subsamples were counted in entirety under loox to determine zooplankton abundance. Adult copepods and copepodites were identified and grouped together. All other individuals, except for a few rotifers, were identified to species. All Chaoborus within a sample were counted but not differentiated into instars. In addition, for the cladocerans, sizefrequency distributions were determined by measuring (to the nearest 0.01 mm) 50 random individuals of each species from the anterior margin of the head to the base of the tail spine. Mean clutch sizes were determined for size classes for which at least three separate measurements were available; usually these were means of from 15 to 30 individuals. Egg volumes were estimated by measuring the long and short axes of newly devel- oped eggs under 200x and using the formula for an oblate spheroid. Instantaneous birth and death rates were estimated from egg:female ratios from a random sample of 50 individuals (Paloheimo 1974). Egg development times were taken from Knutson (1970) for Daphnia pulex, Kwik and Carter (1975) and Hall et al. (1970) for Ceriodaphnia reticulata, and Kwik and Carter (1975) and Kerfoot (1974) for Bosmina longirostris. Larger invertebrates (insects, hydracarinids) were sampled biweekly in both halves of the pond in Twenty to forty vertical hauls were taken with a 30- cm-diameter cylinder with a bottom of 1.5-mm screening. Since these predaceous invertebrates are rapid swimmers, this sampling technique underestimates actual densities. Seine hauls on 7 June and 3 August 1976 provided samples of vertebrates for stomach analyses. Pimephales and Necturus were netted at midday, fixed immediately in Formalin, and stored at 4 C. Pimephales was measured from the anterior margin of the snout to the fork of the caudal fin, Necturus from the snout to the posterior edge of the vent. Complete stomach contents of all animals examined were identified and counted un- der 4 0 magnification. ~ In August 1976 several laboratory experiments were done to estimate the relative vulnerability of the herbivore species to Chaoborus predation. Chaoborus americanus was collected from Williams Pond (a nearby pond with similar community structure), since the species was rare at that time in Pleasant Pond, and B. longirostris from Loch Loso (a similar pond, but having dense populations of planktivorous fish). All other prey species were taken from PP N. Only those D. pulex <1.5 mm were used; all of the Diaptomus clavipes were adults (about 2.5 mm long); C. reticulata and B. longirostris were chosen randomly with respect to size. No oviparous individuals were used. Predation experiments were done in amber bottles at the prevailing

5 JUN I JUL I AUG I SEP I APR, MAY, JUN, JUL, BUG 600r 8 Ceriodaphnia reticulata.~o.''o'o 0 80,o, Bosmina (\1 Qs P-dh, lonpirostris F 0 %=-I= I Daphnia galeafa mendo fae A I Ceriodaphnia reficulata 4 Daphnia parvulo (\1 - r o,,j=yo \.0'41'.~ u Cyclops vernalis q~ La-, 0 4,O0Or Cyclops vernalis Nauplii JUN JUL AUG SEP 1975 Fig. 1. The 1975 abundances of main zooplankton species in Pleasant Pond. Dotted lines represent mean abundances of species in enclosures in which Chaoborus was very rare. temperature and 1ight:dark cycle in Pleasant Pond water filtered through 64- pm netting. Two fourth instar Chaoborus were enclosed per chamber with variable numbers and types of prey; after 24 h, the samples were fixed with Formalin, and all remaining prey items counted. Those prey missing or reduced to chitinous balls were considered eaten. In triplicate controls without Chaoborus, only one of thirty Daphnia, Ceriodaphnia, and Diaptomus died. Similar experiments were done with Notonecta sp. and hydracarinids as predators. APR MAY JUN JUL AUG 1976 Fig. 2. The 1976 abundances of main zooplankton species in Pleasant Pond North having no fish (solid lines) and Pleasant Pond South in which fathead minnows began to appear in June (dashed lines). The competitive interactions between D. pulex, C, reticulata, and B, longirostris were examined in PP N. Two experiments were done in 3.5-liter Plexiglas chambers having two small windows of 102-pm Nitex netting which allowed some flow but excluded all zooplankton (except rotifers and some nauplii). Ten individuals of each species were placed in their respective chambers (starting densities varied in the first experiment) and incubated at a depth of 0.5 m for 4 weeks. All species pairs were done in duplicate or triplicate, and triplicate controls were run for each species (see Lynch 1978). Results Pleasant Pond zooplankton community-in 1975, there was a period of in-

6 Zooplankton community structure 257 Table 2. Approximate size range (A), smallest size at first reproduction (B), and egg volume (C) for planktonic cladocerans of Pleasant Pond. All measurements are from anterior margin of head to base of tail spine, except those for Daphnia galeata mendotae which extend from eyespot to base of tail spine. Daphnia pulex Daphnia galeata mendotae Daphnia parvula Daphnia ambigua Ceriodaphnia reticulata Bosmina lontzirostris tense Chaoborus predation. When sampling began in mid-june, species of herbivorous cladocerans present, in order of decreasing abundance, were D. pulex, C. reticulata, B. longirostris, and Daphnia galeata mendotae (Fig. 1).Daphnia ambigua and Daphnia parvula were also present in June in very low numbers (<2,000.m-2). Size characteristics are given in Table 2. The predatory copepod Cyclops vernalis (max length, 1.3 mm) was abundant only in June. Accompanying its decline was an increase in numbers of the herbivorous copepod Diaptomus siciloides (max length, 1.5 mm). In early June larvae of the predaceous midge C. americanus were very abundant. By August they had reached extremely high densities consisting largely of newly hatched larvae, but fourth instar larvae were also present. Although September 1975 levels of Chaoborus seem low relative to the population in August, the midge larvae were still quite abundant at that time (2-5.liter-l), and almost all were third and fourth instars. There were significant changes in the composition of the zooplankton community along with the increase of Chaoborus. Bosmina declined first, followed by declines of Ceriodaphnia and nauplii, and then of Diaptomus. By August, these species had nearly disappeared, and even the Daphnia population had declined to a very low density. While all of these crustaceans were rare, the rotifer Polyarthra vulgaris increased, followed by an increase in the predatory rotifer Asplanchna priodonta. Only after the Chaoborus population declined in September 1975 did the Daphnia population begin to recover, and even then all other species remained rare. Invertebrate predation was mild in Pleasant Pond North in Despite the continued absence of vertebrate planktivores from PP N, the sequence of events was quite different from that of 1975 (Fig. 2). Chaoborus was only a fifth as dense as in The large (max length, 2.5 mm), colorful D. clavipes, an herbivorous calanoid copepod absent in 1975, appeared in late May and had become a dominant member of the community by July. Several of the herbivores which disappeared in late 1975 did not reappear in the PP N community: B, longirostris, D. ambigua, D. galeata mendotae, D. parvula, and D, siciloides. Rotifers remained rare throughout the summer. In the absence of these other herbivores, D. pulex became twice as abundant as in However, at the end of July, while Chaoborus was particularly rare, Ceriodaphnia suddenly replaced Daphnia, reaching densities four times greater than any recorded in Pleasant Pond South experienced intense fish predation in At the end of June, the fathead minnow population had become well established, and the zooplankton community of PP S began to diverge from that of PP N (Fig. 2). Daphnia pulex was completely eradicated by the end of July, and Ceriodaphnia and D. clavipes remained very rare. At the end of summer, two small daphnids, D. ambigua and D. parvula, began to appear, and rotifers became abundant (prin-

7 258 Lynch Table 3. Results of laboratory experiments showing mean predation rate (prey ingested per predator per day) for Chaoborus, Notonecta, and Hydracarina predation Predator (No.) Vol Prey Prey (ml) Ex~ts Predahon Chaoborus americanus (2): Ceriodaphnia Ceriodaphnia Ceriodaphnia Daphnia Bosmina Diaptomus clauipes Daphnia 10 ', Ceriodaphnia Daphnia Ceriodaphnia Daphnia D. clauipes Ceriodaphnia D. clauipes Daphnia Ceriodaphnia 10 - * 3.2 D, clauipes 10, Notonecta sp. (1): Daphnia 1: Ceriodaphnia 0.0 Daphnia Ceriodaphnia D. clauipes Hydracarina (2): Daphnia Ceriodaphnia : } * Three Chaoborus used cipal species were Brachionus angularis, Conochiloides sp., Filinia longiseta, Keratella quadrata, and Trichotria sp.). Despite all these changes in the zooplankton community, numbers of Cyclops and nauplii remained similar in both halves of the pond. As Fig. 2 shows, Chaoborus densities were actually greater in PP S than in PP N. However, while all of the midge larvae in PP S were in their first or second instar, most of those in PP N were in their third or fourth. Thus, although less abundant, the larvae in PP N were probably more intense predators. Predators-The laboratory experiments provided direct evidence that Chaoborus predation is of potential significance to the Pleasant Pond herbivores. Over 24 h, individual Chaoborus consumed several of any of the cladocer- ans (B. longirostris, C. reticulata, D. pulex); D, clavipes was less vulnerable (Table 3). The measured rates of predation on the three cladocerans almost certainly underestimate what the fourth instar larvae can do, since invariably nearly all prey were eaten. Predation by individual Chaoborus larvae is probably less intense in the natural community than in the laboratory. Lower prey densities and spatial separation of larvae and prey will reduce the rate at which Chaoborus encounters prey items. Chaoborus predation was not measured on D. siciloides, C. vernalis, nauplii, and rotifers. Fortunately, other studies are consistent enough to make some general statements about the likelihood of such predation. Swiiste et al. (1973) estimated predation rates on Diaptomus

8 Zooplankton community structure 259 gracilis (a species similar in size to D. siciloides) to be as high as 8.2 prey ingested per Chaoborus per day. For a slightly larger species, Diaptomus tyrrelli, Fedorenko (1975b) measured maximum daily predation rates of 30 prey ingested per Chaoborus. Independent estimates are not available for midge predation on cyclopoids, but they appear to prefer both cladocerans and calanoids, and daily rates seldom exceed 1 cyclopoid ingested per Chaoborus when alternate prey are available (Allan 1973; Anderson and Raasveldt 1974). Nauplii may be intensely preyed upon; Fedorenko (1975b) reported rates as high as 19 prey ingested per day by second instar Chaoborus. Rotifers may be eaten, although they are not normally an important component of the diet (Fedorenko 1975~). My results are not sufficient to order a preference between prey species. However, other studies have consistently shown Chaoborus to prey selectively on calanoid copepods over Daphnia and cyclopoids of similar size (Sprules 1972; Swiiste et al. 1973; Anderson and Raasveldt 1974). Although none of these studies considered predation on calanoids relative to nondaphnid cladocerans, it is well known that small cladocerans are preferred to larger ones (Dodson 1970; Sprules 1972; Allan 1973; Anderson and Raasveldt 1974). Very large species, such as D. pulex and D. clavipes in Pleasant Pond, have a distinct advantage since they can grow larger than Chaoborus can handle. I found no evidence that either of the diaptomids in Pleasant Pond is predaceous. The guts of D. siciloides and D. clavipes contained only phytoplankton. Triplicate laboratory experiments similar to those used for measuring Chaoborus predation were done with 10 D, clavipes enclosed with 10 D. pulex (1.5 mm) and 10 C. reticulata (unsized). All individuals survived. Cyclops vernalis is r:arnivorous. In laboratory experiments it consumed small cladocerans (such as Bosmina and Cerio- daphnia) at rates of about 1.d-' and preferred small ones to larger ones (Brandl and Fernando 1974; Kerfoot 1977). It also preys on nauplii and young copepodites but cannot handle large Daphnia (Anderson 1970). Although large invertebrates (other than Chaoborus) seemed uncommon in 1975, no samples were taken, and no accurate statement can be made about their abundance. In 1976 no macroinvertebrates were found in PP S after 11 July (Table 4). It is unclear if any macroinvertebrates are significant predators on zooplankton. Laboratory feeding experiments with Notonecta sp. and hydracarinids as predators show that they can prey on Daphnia, Ceriodaphnia, and Diaptomus at the rate of several individuals per day (Table 3). The hydracarinids seemed particularly voracious consuming nearly all available prey. There is little information regarding predation on zooplankton by the other species in Table 4. Anderson and Raasveldt (1974) found no evidence of such predatory activity by Hyallela. The other species listed probably can consume zooplankton, but other macroinvertebrates seem to be preferred (Bay 1974). Hall et al. (1970) presented indirect evidence that when these macroinvertebrates do prey on zooplankton, they selectively remove the largest species. Chaoborus, because of its size, is probably preferred over any of the other zooplankton. The salamander Necturus spends most of its time foraging on the bottom, as evidenced by the frequent predominance of sediment and filamentous and vascular plants in the gut and by the dominance of the benthic amphipod Hyallela axteca in the diet (Table 5). However, in early June, before most macroinvertebrates had become abundant, Daphnia comprised a large proportion of the diet. By August the Necturus in PP N consumed Hyallela almost exclusively, while those in PP S concentrated on Chaoborus larvae. This dietary shift in PP S was probably because of the rarity of macroinvertebrates at that time (see Table 4).

9 Table 4. Estimated densities of macroinvertebrates in 1976 (individua1s.m-=). No macroinvertebrates found in Pleasant Pond South after 11 July. Pleasant Pond North Pleasant Pond South Jun Jun Jul Jul Aug Sep Jun Jun Jul Amphipoda Hyalleh azteca Coleoptera Haliplus sp. (adults) (larvae) Laccophilus sp. Ephemeroptera Callibaetis sp Hemiptera Hydrometra sp. 0.7 Notonecta sp Sigara sp Hydracarina Odonata Enallagma sp. Plecoptera Isogenus sp Table 5. Stomach contents of vertebrate predators. 7 Jun 3 Aug 7 Jun 3 Aug 7 Jun 3 Aug Length (mm) Sample size Mean proportion of total animal items Cladocera Ceriodaphnia reticulata Daphnia pulex Pleuroxus procuruus Cyclopoda Cyclops uernalis Diaptomus clauipes Nauplii Rotifera Chaoborus americanus (larvae) Hyallela azteca Haliplus sp. adults 0.01 Callibaetis sp. < Notonecta sp Sigara sp. < Himdinea Hydracarina c Anax sp Isogenus sp Mean proportion of total biomass by groups Zooplankton Macroinvertebrates Filamentous algae 0.25 Vascular plants 0.25 Sediment 0.60

10 Zooplankton community structure 1 lncreasing / Increasing No Fish 1 Fish + No Fish 1 Fish 4 1 Predation [ Predation u Fig. 3, Mean and range for abundance of crustaceans, Chaoborus, and Asplanchna for last three sampling dates of enclosure experiment. Early in its colonizing phase, the minnow Pimephales consumed zooplankton (particularly D. pulex) almost exclusively (Table 5). However, by August most of the zooplankters had disappeared or were substantially reduced, and Pimephales switched to feeding primarily on the sediments. Thus. both of these vertebrates have plastic foraging strategies. Pimephales prefers zooplankton when available, but can switch to a detritus-based diet when necessary; Necturus concentrates on larger invertebrates, but can also consume Daphnia, Chaoborus, and vascular plant material. Herbivore competition-since the results of the herbivore competition experiments have been discussed elsewhere (Lynch 1978), I only consider the major points here. The largest herbivore in Pleasant Pond, D. pulex, is not generally the dominant competitor. Although I used a variety of starting densities and size distributions in the first experiment, Daphnia went extinct in the presence of Ceriodaphnia in nine out of nine cases; both species survived and reproduced in their control enclosures (Table 6). Since this experiment was run concuirently with the replacement of Daphnia by Ceriodaphnia in PP N, the results strongly suggest that the midsummer decline of Daphnia was a direct outcome of a coexploitative interaction with Ceriodaphnia. All other zooplankton species were rare in the pond during that period (Fig. 2), so that it is unlikely that they were involved in this succession. Enclosure experiments confirmed that the demise of Daphnia was not a result of predation by

11 262 Lynch cl densities in the presence of Ceriodaphnia and Bosmina than in its controls (Taangularis I ble 6). However, while Ceriodaphnia 0 I and Bosmina declined in these chambers relative to their own controls, they did not die out during a period of 4 weeks. quadridenfalo;' I Thus, although the sensitivity of Daph- F O nia to coexploitative interactions varied I significantly throughout the year, in no case did it exclude Ceriodaphnia or Bosmina. Although Ceriodaphnia cannot depress the food supply to the extent that Daphnia can, its survival in the presence of Daphnia results from its lower sensitivity to a reduced food supply (Lynch 1978). The response of all three species to coexploitative interactions was a reduction in juvenile survivorship. Adult reproduction was less sensitive to competitive interactions. a The ability of Bosmina to thrive in its i1 Increasing control chambers indicates that its ab- No Fish1Fish ----t sence from the pond in 1976 was not a 1 Predation result of unsuitable physical conditions. Its greatly reduced density in chambers Fig. 4. Mean and range for abundance of pre- with Daphnia, Ceriodaphnia, or both dominant rotifers for last three sampling dates of (Table 6) suggests that its exclusion from enclosure experiment. the pond may have been associated with competitive interactions with these vertebrates or macroinvertebrates (Lynch species. Its recovery in 1976 may have 1978). also been inhibited by Chaoborus pre- Four weeks later, in the second exper- dation since few alternate prey were iment, Daphnia actually attained higher available at that time. (Even Ceriodaph- Table 6. Final results of competition experiments, mean densities i0.95 confidence limits (individuals.liter-'). Daphnia + Daphnia + Daphnia + Ceriodaphnia Ceriodaphnia Daphnia Ceriodaphnia Bosmina Ceriodaphnia Bosmina + Bosmina + Bosmina Experiment 1 (19 July-14 August) Daphnia i19.0 Ceriodaphnia i53.4 i42.3 Experiment 2 (16 August-10 September) Daphnia i14.0 i15.4 i6.7 Ceriodaphnia i40.2 i16.0 ~29.1 k2.6 Bosmina k ~6.2 i23.5 k5.6

12 Zooplankton community structure Food Size Classes (,urn3) Fig. 5. Mean abundance of phytoplankton in size classes available to herbivores for last three sampling dates for enclosure experiment. Pond, 1,7, and 2 have no fish; 8,9, 3, 11, and 6 are in order of increasing fish abundance. nia did not begin to reach significant numbers in PP N until the end of July when Chaoborus declined.) Community level effects of Chaoborus andfish predation-of all the planktonic crustaceans in the pond, only D, clavipes was absent at the start of the enclosure experiment. Since most potential colonists were present, the results of this experiment indicate the direction in which the Pleasant Pond community tends to move when exposed to different intensities of Chaoborus and fish predation. The average zooplankton communities under different treatments for the last 3 weeks of the experiment are shown in Figs. 3 and 4. Chaoborus predation of the intensity that occurred in the pond in 1975 has a severe impact on the abundance of Cerio-

13 264 Lynch Table 7. Mean and range (in parentheses) for instantaneous death rates for Ceriodaphnia and Bosmina in enclosure experiment for last three sampling dates. Instantaneous death rates (day-') Enclo- sure Ceriodaphnia Bosmina Pond 0.31( ) 0.37 Controls ( ) 0.20 ( ) ( ) 0.20 ( ) ( ) 0.14 ( ) Mean Fish ( ) 0.02 ( ) ( ) 0.24 ( ) ( ) 0.08 ( ) ( ) 0.02 ( ) ( ) 0.04 ( ) Mean Control F~sh * Enclosure Enclosure I 0 8 * 1 7 A 9 * A 2 ** Bosmino longlrostr~s A daphnia and Bosmina (and perhaps Keratella). The dotted lines in Fig. 1give the mean densities of species in the controls for this experiment. In the absence of Chaoborus, Ceriodaphnia and Bosmina maintained higher numbers in the bags than in the pond. Rotifers (particularly Keratella cochlearis) also increased in the control bags. The disappearance of these species from the pond cannot be attributed to competition with Daphnia. The abundance of all sizes of food particles was greater in the pond than in any of the control bags (Fig. 5). If anything, the presence of Ceriodaphnia and Bosmina depressed D, pulex. Not only did Daphnia appear to decline in the presence of Ceriodaphnia and Bosmina, (Fig. 3), but Daphnia clutch sizes in the control bags were lower than those in the pond (Fig. 6). Since Cyclops declined at identical rates in the controls and the pond, its density does not seem to have been regulated by Chaoborus predation. Adding fish to enclosures also had a dramatic effect on the zooplankton community. Three species completely disappeared from all fish enclosures: D, pule~,d. galeata mendotae, and Chaoborus. Three others appeared only in fish enclo- 31 A Carapace Length (mm) Fig. 6. Mean clutch sizes for Daphnia pulex (0.1-mm size classes) and Ceriodaphnia and Bosmina (0.03-mm size classes) for last three sampling dates for enclosure experiment. sures at the end of the experiment: D. ambigua, D. parvula, and A. priodonta. In the absence of Daphnia, food levels increased over the controls in all of the fish enclosures (Fig. S), and the number of eggs carried by Ceriodaphnia increased severalfold (Fig. 6). However, the abundance of Ceriodaphnia in these enclosures was dictated by predation intensity, and Ceriodaphnia was rare at the highest levels of fish predation (Fig. 3). Mean instantaneous death rates were very low in the control enclosures ( d-l), while those in enclosures with fish (and in the pond where Chaoborus predation was intense) were many times higher ( d-') (Table 7). Since Cyclops was much more abundant in the

14 Zooplankton community structure fish enclosures, predation by Cyclops as well as by fish may have been responsible for the increase in Ceriodaphnia mortality. (The three fish enclosures in which Ceriodaphnia was rarest also had the greatest abundance of Cyclops: Fig. 3.) Bosmina was least vulnerable of the cladocerans to fish predation and indeed increased with the intensity of fish predation (Fig. 3). The enhanced food supply in the fish enclosures was also reflected in the clutches of Bosmina, which were much larger in the presence of fish than in their absence (Fig. 6). Not only were conditions better for reproduction in the presence of fish, but instantaneous death rates also dropped severalfold (Table 7). Average death rates for the last three sampling dates were much higher in the controls ( d-l) than in fish enclosures ( d-i), except for bag 9 (0.24. d-l) which also had a very dense Cyclops population. The general increase in abundance and decrease in the death rate of Bosmina in the fish enclosures suggest that neither Lepomis nor Cyclops is directly capable of restricting its distribution. o.5 Since Cyclops was not restricted by Chaoborus predation in the control enclosures, its increase in fish enclosures (Fig. 3) must have resulted from an enhanced food supply. Several possible food items (Bosmina, rotifers, and algae) were more abundant in the fish enclosures. 0 : Pond I The results of this experiment provide Fig. 7. Linear regressions for estimated rates of some indirect evidence for ~~~l~~~pre- Cyclops predation in pond, 1975, and control enclosures (containing no fish). Method of estimation dation on the small cladocerans. I esti- given in text, Ceriodaphnia: = i5, )x - mated rates of Cyclops predation on cla- 0,286,,= 0.93; ~o~rnina: Y = (3.96 x io-6)xdocerans in the pond and in the controls 0.017, r = 0.52; Daphnia: Y = (1.30 x 1W6)xfor davs on which Chaoborus was absent 0.034> = by multiplying the density of the prey species by its instantaneous death rate and dividing by the density of Cyclops (adults ~ lus copepodites). Predation rates estimated in this manner will overestimate the actual rates, since the method assumes that cladoceran mortality is entirely a result of Cyclops predation. However, when the estimated predation rates are plotted against prey density, the fit with a linear model is reasonably good, especially for Ceriodaphnia (Fig. 7). The regression lines suggest that Cyclops preys more intensely on Ceriodaphnia than on Bosmina, and very little at all on Daphnia.

15 266 Lynch Table 8. Mean egg volume (t0.95 confidence limits) for Ceriodaphnia reticulata and Daphnia pulex for periods under influence of different predators. Predators Clutch size Sam- ple size Mean egg vol Ceriodaphnia reticulata Jun 75 Few Chaoborus, many Cyclops 3 Jul-19 Aug 75 Many Chaoborus, few Cyclops Pleasant Pond North Jul 76 Few Chaoborus, very few Cyclops 5-27 Aug 76 Few Chaoborus, moderate Cyclops Daphnia pulex 20 Jun 75 Few Chaoborus, many Cyclops 11 Jul-17 Sep 75 Many Chaoborus, few Cyclops Pleasant Pond North 28 Apr-30 Jul 76 Few Chaoborus, very few Cyclops Aug 76 Few Chaoborus, moderate Cyclops Pleasant Pond South 28 Apr4 Jun 76 Few Chaoborus, very few Cyclops, few fish 11 Jun-15 Jul 76 Few Chaoborus, few Cyclops, many fish Cladoceran size strategies-as a result dation. Instead, in the face of intense of its intermediate size, C. reticulata is Chaoborus predation, smaller morphs of vulnerable to most aquatic predators. Ceriodaphnia are favored. As Chaoborus Next to the much larger D, pulex it is the became abundant in July 1975, the maxspecies most vulnerable to vertebrate imum and minimum sizes and the size at predators in Pleasant Pond. Furthermore, first reproduction of Ceriodaphnia all deunlike many other cladocerans of its size, clined (Fig. 8).By the end of August, the Ceriodaphnia has no spiny appurte- largest individuals observed were only nances which might thwart invertebrate half the size of those noted before the inpredators (Dodson 1974b). crease in Chaoborus. As Chaoborus pre- Ceriodaphnia seems to have different dation relaxed in PP N in 1976, the minsize strategies to cope with different imum size and the size at first types of predators. It does not grow near- reproduction increased to previous levels. ly large enough to avoid Chaoborus pre- Measurements of egg volumes for Cerio-

16 Zooplankton community structure daphnia are also consistent with the selection for smaller size under the influence of Chaoborus predation. Although the sample size is small, the eggs of Ceriodaphnia were smaller in late summer 1975 than at any other time during this study (Table 8). The sharp decline in the maximum size of Ceriodaphnia in summer 1975 may have been a result of selective predation by Chaoborus on the larger individuals; Dodson (1974~) noted that Chaoborus increases its selectivity for prey as body size increases in the size range of Ceriodaphnia. An increase in predation on all sizes of individuals would also lower the probability of any attaining a large size. Ceriodaphnia could probably have grown larger in late 1975 if not for the presence of Chaoborus, since much larger animals were apparent in early However, it is unlikely that the reduction in the size at birth and the size at first reproduction resulted from an insufficient food supply in 1975, since food was much more abundant than in 1976 (when a very low food supply had little effect on Ceriodaphnia: Lynch 1978). Unlike the much larger Chaoborus, C. vernalis prefers small Ceriodaphnia to larger ones. Thus, when Cyclops predation is intense there is an advantage to Ceriodaphnia in producing large offspring and growing to a large size at the expense of early reproduction. In Pleasant Pond, Cyclops was most abundant when Chaoborus was rare-early summer 1975 and late summer 1976; during those periods Ceriodaphnia produced large offspring and began to reproduce at a large body size (Fig. 8). Cyclops also often became very abundant when vertebrate predators were present. Since the maximum body size declines in the face of fish predation (Fig. 9), Ceriodaphnia cannot thwart Cyclops by growing large without increasing its vulnerability to fish. Under such circumstances, Ceriodaphnia produces large offspring (Fig. 10) which are less vulnerable to Cyclops and also begins reproduction at a small size (Fig. 9). Pleasant Pond, r ---, > O3 JUN JUL AUG SEP Fig. 8. Mean size of largest and smallest 5% of population (solid lines) and size at first reproduc- tion (dashed lines) for Ceriodaphnia. Invertebrate predators are less of a problem for D. pulex. It is too large to be of much interest to Cyclops, and it can grow large enough to escape Chaoborus predation. When Chaoborus predation was very intense in late summer 1975, Daphnia began reproduction at a large size and produced large offspring (Fig. 11, Table 8). As Chaoborus predation relaxed in 1976, the size at first reproduction declined considerably and smaller offspring were produced. Further work is needed to determine whether these size changes are a response to an alteration in the physical or nutritional status of the environment or are true shifts in the genetic structure of the populations. Discussion These results suggest a general framework for the Pleasant Pond zooplankton community. When vertebrate predators are rare, the composition of the community is most closely related to abundance of Chaoborus (Fig. 12). However, since several of the factors that may be responsible for regulating the midge populations are unpredictable (abundance of prey in previous generation, abundance

17 268 Lynch s 4 P, 0.6 Control Enclosures Fish Enclosures Pond 8 Mean Size of I I Upper 5% '\ ' m '\.J' Size at First Reproduction Days Fig. 9. Changes in length of largest 5% of Ceriodaphnia measured and in size at first reproduction in control and fish enclosures. of macroinvertebrate predators, physical constraints such as overwintering conditions, or timing of emergence and reproductive success of adults), the composition of the zooplankton community may vary significantly from year to year when planktivorous fish are absent. The most abundant herbivore is always D. pulex or C, reticulata in the absence of vertebrate predators. When Chaoborus is rare or absent, Ceriodaphnia is able to express its competitive superiority over D. pulex and is the dominant herbivore. " 501 I I I I Carapace Length (rnrn) Fig. 10. Egg volume measurements for Ceriodaphnia carrying one egg in enclosure experiment. Solid symbols-enclosures without fish; open symbols-fish enclosures. I I I I I I APR MAY JUN JUL AUG SEP Fig. 11. Mean size of largest and smallest 5%of population (solid lines) and size at first reproduction (dashed lines) for Daphnia pulex.

18 Zooplankton community structure 269 Chaoporus &./' t '1 c Ceriodaphnia Bosmina Diaptomus.O _k, + $ --'Cyclpps Daphnia pu/ex FISH, Dophnlo pulex/ x,-, D~optomus Dophnio goleato chwborus--2~er1odo~hnlo,' +,,.x Bosmma%'-:--Cyclops'--+ Rotifers '\ 4. ' '0 b 8 b ;s Chaoborus Ji O.= \ Ceriodaphnia ; Diaptomus '0 Daphnia pu/ex c, :: Daphnia pulex \ 'I 1 Rotifers -Asplunchno/- Fig. 12. Generalized framework for composition of Pleasant Pond zooplankton community when vertebrate predators are absent. Species underlined are abundant; others are rare. Solid arrows point from predator to prey and indicate intense predation; dashed arrows indicate less intense predation. The smaller forms (Bosmina and rotifers), their predator, Cyclops, and Diaptomus may also be present under these conditions, since they are not often eaten by Chaoborus. As the abundance of Chaoborus increases, the smaller herbivore species are removed. Increased predation on Ceriodaphnia offsets its competitive ability, and D. pulex becomes the most abundant herbivore. Diaptomus is less preferred by Chaoborus and remains abundant. In the absence of sufficient prey and in the face of increased losses to Chaoborus, Cyclops also disappears. When Chaoborus predation is intense, all crustacean herbivores except D. pulex are removed, and even Daphnia becomes quite rare. A consequent increase Fig. 13. Generalized framework for composition of Pleasant Pond zooplankton community under different intensities of fish predation. Details as in Fig. 12. in the food supply allows the small rotifers and their predator, Asplanchna, to increase rapidly enough to offset potential losses to Chaoborus. The most immediate effect of fish predation is the reduction of the two largest species, Chaoborus and D. pulex (Fig. 13).In the absence of Chaoborus, Ceriodaphnia becomes the dominant herbivore at low levels of fish predation. However, since Ceriodaphnia is unable to reduce the food supply as much as D. pulex, several other small herbivores (D. galeata, Bosmina, and rotifers) can become quite abundant. This expanding food supply also allows Cyclops to become more numerous. As fish become more abundant D. pule~,d. galeata, and Chaoborus completely disappear, and Ceriodaphnia is significantly reduced. In response to an increased food supply and the absence of Chaoborus, the smallest herbivores (Bosmina and rotifers) become abundant and, in turn, support a large population of Cyclops. Two small da~hnids, D. ambigua and D. parvula, also appear in low numbers.

19 As the intensity of fish (and Cyclops) predation becomes very great, Ceriodaphnia is removed and Diaptomus is significantly reduced. Bosmina continues to increase. However, intense fish predation prevents the larger D. ambigua and D. parvula from increasing any further. Rotifers continue to reproduce at a high rate, but their abundance is less than at lower levels of fish predation because of the arrival of the predatory rotifer Asplanchna. While the presence of Chczoborus and fish imposes predictable constraints on the Pleasant Pond community, an unambiguous interpretation of the structure of this community requires us to consider competitive interactions between herbivores. Thus, the general utility of this description depends on the nature of the competitively dominant herbivore in other zooplankton communities. Unfortunately, the competitive ability of different species may vary from lake to lake and even between years or seasons in a given lake (Lynch 1977b, 1978). Such circumstances limit the construction of a theory for zooplankton community structure that could yield precise predictions over a broad geographic area. However, a more genera1 approach can be taken to interpret the composition of zooplankton communities. 1. In communities where the competitive dominant is a large herbivore, community composition will follow the models of Brooks and Dodson (1965)and Dodson (1974a).Then, in the absence of vertebrate predators, the smaller herbivores will be competitively constrained from increasing at a high enough rate to offset losses to invertebrate predators. As fish predation increases, these small herbivores will be released from competition with the dominant herbivore and from predation by large invertebrates. 2. If the dominant competitor is of intermediate size, such as Ceriodaphnia in Pleasant Pond, the numerical dominance of that species will be restricted to particular circumstances. It will be rare whenever large invertebrate predators or vertebrate predators are common. The larger subordinate competitors will be most abundant when large invertebrate predators are common, and the smaller subordinate competitors will dominate in the presence of vertebrate predators. 3. If the competitive dominant is small enough to be of little interest to vertebrate predators, its distribution will be primarily dictated by the abundance of invertebrate predators. In the absence of vertebrate predators, it will only be dominant when large invertebrate predators are rare; otherwise larger herbivores will be most abundant. However, when vertebrate predators are present, such a small competitive dominant will always increase at the expense of larger herbivores. These arguments admit that in most situations large numbers of large and small herbivores will tend not to coexist, just as most previous theory predicts (Brooks and Dodson 1965; Dodson 1974~).However, a significant departure from previous theory is the recognition that small herbivores may dominate in vertebratefree environments under appropriate conditions. The numerical dominance of small herbivores in a vertebrate-free environment is by no means unique to Pleasant Pond. In a series of north Scandinavian lakes having no fish, Nilsson and Pejler (1973) found the small Bosmina coregoni to be as abundant as or more abundant than the much larger Waphnia longispina. Significantly, large invertebrate predators were quite scarce in these lakes (no Chaoborus or Leptodora were reported, and the predaceous calanoid Heterocope saliens was noted only rarely). Lakes in eastern Quebec that lack fish are often dominated by the small B. longirostris and have few of the much larger D. pulex despite their abundance of Chaoborus (Pope and Carter 1975). At low intensities of fish predation, Chaoborus is reduced and several other invertebrate predators (Leptodora, Epischura, and Mesocyclops) increase; then Bosmina declines, and Waphnia becomes more numerous.

20 Zooplankton community structure 271 A small pond studied by Kwik and Carter (1975) which seemed to have no predators had a zooplankton community dominated by small species (B. longirostris, Ceriodaphnia quadrangula, and D. ambigua).sunfish Lake, a small meromictic lake in southern Ontario, has neither Chaoborus nor Leptodora and no pelagic fish (Clark and Carter 1974). Its most abundant zooplankters are of intermedi- ate size-daphnia rosea, Daphnia retrocurva, and Diaphanosoma leuchtenbergianum. Finally Hall et al. (1970) found C, reticulata to be the most abundant species in a series of experimental ponds lacking fish; D. pulex was only present early in summer. Chaoborus, Notonecta, and Buenoa were present in these ponds. It is not clear that the small species dominating these lakes are superior competitors, but their abundance in the absence of vertebrate predators indicates that a theory more complex than that proposed by previous workers is necessary to explain adequately the distribution of zooplankton species. The future development of a theory for zooplankton community structure requires a finer understanding of the factors influencing the competitive ability of different species. It is particularly important that the mechanism of competition (the relation of reproduction and survival to resource abundance) be understood, since the demographic consequences of food limitation may increase the sensitivity of a species to other adverse community interactions. For instance. the eradication of large herbivores by "ertebrate predators may be accelerated by the appearance of smaller competing species which cause reduced juvenile survival, smaller clutch sizes, or both. Furthermore, the significance of invertebrate predators in structuring zooplankton communities cannot be assessed until the factors regulating their populations are known (especially in vertebrate-free environments). In Pleasant Pond, Chaoborus could reach high enough densi- ties to cause extinction of herbivores: Cu-, " clops could not. Chaoborus may become abundant enough to eradicate any one prey species because it can accept alternate prey. The ability of predaceous copepods to exclude prey species may be limited for two reasons. First, because of their small size (compared to Chaoborus),they are restricted to a narrow range of prey items and may be unable to sustain themselves once a preferred prey species nears extinction. Second, an abundant population of any herbivore, necessary to sustain a large population of predaceous copepods, will also be in direct competition with herbivorous nauvlii. Finally, adding precision to any theory for zooplankton community structure will require knowledge of the relative sensitivity of different species to abiotic factors and to dispersal barriers. For instance, several species common in nearby lakes were never found in Pleasant Pond-D. retrocurua, Diaphanosoma brachyurum, Leptodora kindtii. We need to elucidate the mechanisms excluding them from the pond. References ALLAN, J. D Competition and the relative abundances of two cladocerans. Ecology 54: ANDERSON,R. S Predator-prey relationships and predation rates for crustacean zooplankters from some lakes in western Canada. Can. J. Zool. 48: , AND L. G. RAASVELDT Gammarus predation and the possible effects of Gammarus and Chaoborus feeding on the zooplankton composition of some small lakes and ponds in western Canada. Can. Wildl. Serv. Occas. Pap p. BAY, E. C Predator-prey relationships among aquatic insects. Annu. Rev. Entomol. 19: BRANDL,Z., AND C. H. FERNANDO Feeding of the copepod Acanthocyclops vernalis on the cladoceran Ceriodaphnia reticulata under laboratory conditions. Can. J. Zool. 52: BROOKS,J. L The effects of prey size selection by lake planktivores. Syst. Zool. 17: , AND S. I. DODSON Predation, body size, and composition of plankton. Science 150: CLARK,A. S., AND J. C. CARTER Population dynamics of cladocerans in Sunfish Lake, Ontario. Can. J. Zool. 52:

21 Lynch DODSON, S. I Complementary feeding plankton community structure. Limnol. Oceanniches sustained by size-selective predation. ogr. 22: Limnol. Oceanogr. 15: b. Fitness and optimal body size in a. Zooplankton competition and pre- zooplankton populations. Ecology 58: 763- dation: An experimental test of the size effi ciency hypothesis. Ecology 55: Complex interactions between nat-, 1974b. Adaptive change in plankton mor- ural coexploiters-daphnia and Ceriodaphnia. phology in response to size-selective preda- Ecology 59: tion: A new hypothesis of cyclomorphosis. Limnol. Oceanogr. 19: NILSSON, N., AND B. PEJLER On the relation between fish fauna and zooplankton composi- FEDORENKO,A. Y. 1975a. Instar and species-specific diets in two species of Chaoborus. Limnol. tion in north Swedish lakes. Rep. Inst. Freshwater Res. Drottningholm 53: Oceanogr. 20: O'BRIEN, W. J., N. A. SLADE, AND G. L. VINYARD b. Feeding characteristics and preda Apparent size as the determinant of prey tion impact of Chaoborus (Diptera, Chaoboridae) larvae in a small lake. Limnol. Oceanogr. 20: HALL, D. J., W. E. COOPER, AND E. E. WERNER An experimental approach to the production dynamics and structure of freshwater animal communities. Limnol. Oceanogr. 15: , S. T. THRELKELD, C. W. BURNS, AND P. H. CROWLEY The size-efficiency hypothesis and the size structure of zooplankton communities. Annu. Rev. Ecol. Syst. 7: HANEY,J. F.,AND D. J. HALL Sugar-coated Daphnia: A preservation technique for Cladocera. Limnol. Oceanogr. 18: KERFOOT, W. C Egg-size cycle of a cladocselection by bluegill sunfish (Lepomis macrochirus). Ecology 57: PALOHEIMO,J. E Calculation of instantaneous birth rate. Limnol. Oceanogr. 19: POPE, G. F., AND J. C. CARTER Crustacean plankton communities of the Matamek River system and their variation with predation. J. Fish. Res. Bd. Can. 32: SPRULES, W. G Effects of size-selective predation and food consumption on high altitude zooplankton communities. Ecology 53: SWWSTE,H. F., R. CREMER, AND S. PARMA Selective predation by larvae of Chaoborus flavicans (Diptera, Chaoboridae). Int. Ver. -. eran. Ecology 55: Theor. Angew. Limnol. Verh. 18: Implications of copepod predation. WERNER, E. E., AND D. J. HALL Optimal Limnol. Oceanogr. 22: KNUTSON,K. M Plankton ecology of Lake Ashtabula Reservoir, Valley City, North Dakota. Ph.D. thesis, North Dakota State Univ., Fargo. 99 p. KWIK. I. K.. AND I. C. CARTER Povulation > ", dynamics of limnetic Cladocera in a beaver pond. J. Fish. Res. Bd. Can. 32: LYNCH, M Zooplankton competition and foraging and the size selection of prey by the bluegill sunfish (Lepomis macrochirus). Ecology 55: ZARET, T. M., AND W. C. KERFOOT Fish predation on Bosmina longirostris: Body-size selection versus visibility selection. Ecology 56: Submitted: 25 October 1977 Accepted: 16 August 1978