BEHAVIORAL MECHANISMS OF COEXISTENCE IN SYMPATRIC SPECIES OF DESERT RODENTS, DIPODOMYS ORDII AND D. MERRIAMI

Transcription

1 BEHAVIORAL MECHANISMS OF COEXISTENCE IN SYMPATRIC SPECIES OF DESERT RODENTS, DIPODOMYS ORDII AND D. MERRIAMI LEONORA M. PERRI AND JAN A. RANDALL Department of Biology, San Francisco State University, San Francisco, CA We investigated intraspecific and interspecific behaviors that function as mechanisms of social spacing and tolerance to allow coexistence of sympatric species of desert rodents. Radiotracking data showed overlap in home ranges of two species of kangaroo rats, Dipodomys ordii and D. merriami. Home ranges of males of both species overlapped females; females exhibited exclusive home ranges. Although D. ordii dominated D. merriami in paired encounters in the laboratory, D. merriami avoided D. ordii, and aggression was infrequent Tests for recognition by olfactory cues revealed that kangaroo rats did not respond to heterospecific scent and behaviors of intraspecific recognition of D. ordii were similar to those described previously for D. merriami. Females responded to the scent of con specific females and preferred the scent of familiar, con specific males; males showed no preferences. These two species of kangaroo rats probably coexist through interspecific avoidance and intraspecific spacing mediated by familiarity and neighbor recognition in females. Key words: Dipodomys ordii, Dipodomys merriami, home range, sympatry, behavior, coexistence, olfaction, dominance, avoidance Our goal was to investigate intraspecific and interspecific behaviors that might function as mechanisms of social spacing and tolerance to allow biologically similar, sympatric species to coexist. Coexistence of ecologically similar rodents in desert communities has long been of interest to community ecologists (see reviews by Brown and Harney, 1993; Randall, 1993; Reichman and Price, 1993). Small rodents, especially members of the Heteromyidae (kangaroo rats, Dipodomys; kangaroo mice, Microdipodops; pocket mice, Perognathus and Chaetodipus), have been of special interest because of the use of seeds as the same ephemeral food resources by coexisting species (Heske et ai., 1994; Reichman and Price, 1993; Rosenzweig and Winakur, 1969). Up to six species of heteromyids can coexist in desert communities (Brown, 1973). Because similar species often overlap in use of resources, the potential for compe- tition may be high. Roles of resource partitioning and competition in structuring heteromyid communities, therefore, have received much attention in the quest to answer how closely related species in North American deserts coexist (Brown and Harney, 1993; Heske et al., 1994). It is well accepted that the diversity of sympatric granivores is associated with complexity of habitat (Reichman and Price, 1993). Rodents might partition resources by habitat selection (M'Closkey, 1976, 1978; Schroder and Rosenzweig, 1975), exploitative competition for seeds in different microhabitats (Bowers, 1982; Price, 1978), optimizing foraging choices (Price and Heinz, 1984; Price and Reichman, 1987; Reichman and Oberstein, 1977), and differences in hoarding behavior (Jenkins and Breck, 1998) and seed husking (Dayan and Simberloff, 1994; Randall, 1993). Risk of predation might promote coexistence based on occupation of micohabitats where each spe- Journal of Mammalogy. 80(4): ,

2 1298 JOURNAL OF MAMMALOGY Vol. 80, No.4 cies is best adapted to compete for resources while avoiding predation (Kotler et ai., 1994). Use of habitat could vary temporally (Haim and Rozenfeld, 1993), and some sympatric heteromyids are active during distinctly different times of the night or season (Kenagy, 1973; O'Farrell, 1974) and use similar habitats at different times of the year (M'Closkey, 1976; O'Farrell, 1980). Coexisting heteromyids also vary in body size, which could influence habitat use and help to space species. Typical body sizes of granivores in desert communities differ by ratios of > 1.5 (Bowers and Brown, 1982; Brown, 1973). Brown and Lieberman (1973) predicted that heteromyids reduce competition by selecting different-sized seeds based upon these differences in body size. However, the seed-sizeselection hypothesis has not been supported (Lemen, 1978; Price, 1983), and resource partitioning of seeds by size is not an important mechanism of coexistence in heteromyids (Randall, 1993). Social interactions are neglected in most of the above investigations. Even when behavior is considered, aggressive interactions are assumed to be instrumental in interference competition as a means of shaping community structure (Brown and Munger, 1985; Frye, 1983; Reichman, 1983). Larger species of heteromyids are believed to dominate smaller species by aggressive interference (Basset, 1995; Blaustein and Risser, 1976; Frye, 1983). Social behaviors that minimize aggression, such as recognition of neighbors, also could function in structuring of heteromyid communities (Randall, 1993). Closely related species often exhibit distinctly different behavior patterns (Banks et al., 1979; Cranford and Derting, 1983; Dempster et ai., 1992; Gouat et al., 1996; Livoreil et ai., 1993; Wolff et ai., 1983), and some species are naturally tolerant, rather than aggressive (Gouat et ai., 1996; Livoreil et ai., 1993). We chose to study behavioral mechanisms of coexistence of Dipodomys ordii and D. merriami for several reasons. First, these solitary, nocturnal rodents are sympatric over a large part of their geographic ranges in North America (Hall, 1981). They overlap in about one-third of the geographic range of D. ordii and about one-half of the geographic range of D. merriami in a variety of habitats (Brown, 1973; M'Closkey, 1978; Rosenzweig and Winakur, 1969). Second, population fluctuations of D. ordii and D. merriami are correlated positively (Brown and Heske, 1990a, 1990b; Brown and Zeng, 1989; Heske et ai., 1994). Because both species may experience high population densities at the same time, competition should increase in regions of sympatry. Both species also are similar in body size with D. ordii slightly larger than D. merriami. This ratio of body sizes is < 1.5 (ca. 1.15), which is the ratio of body sizes of desert granivores that is considered necessary for coexistence (Bowers and Brown, 1982; Brown and Harney, 1993). Finally, in addition to similar morphology and use of habitat, D. ordii and D. merriami exhibit extensive intraspecific and interspecific overlap in home ranges (Schroder, 1987). Schroder (1987) proposed that the ecological similarity between D. ordii and D. merriami may result in neither species having a clear competitive advantage. Intraspecific densities would be kept below saturation to permit the two species to coexist in habitats different from the habitat occupied by either species alone. Because aggressive interference would be ineffective and costly, other behavioral mechanisms of coexistence should evolve. To understand behaviors that mediate coexistence of D. merriami and D. ordii, we conducted studies in the laboratory and field. We measured home-range overlap with radiotelemetry, made behavioral observations in the field, and conducted olfactory preference tests and behavioral encounters in the laboratory. We investigated interspecific and intraspecific olfactory communication because chemical cues have been found to maintain separation between sympatric species of desert rodents (Haim and Rozenfeld, 1993;

3 November 1999 PERRI AND RANDALL-BEHAVIOR OF KANGAROO RATS 1299 Johnston and Robinson, 1993). We tested species recognition through a comparison of responses to heterospecific odors from sebaceous glands of D. ordii and D. merriami. We conducted intraspecific test of gender recognition and familiarity in D. ordii to compare with a previous study of D. merriami (Randall, 1981). Finally, we compared behavioral interactions of D. ordii and D. merriami in paired encounters in an arena sufficiently large to allow avoidance behavior (Ostfeld,1985). MATERIALS AND METHODS Radiotracking and observations.-we monitored movements and observed behavior of D. ordii and D. merriami on a 200-by-240-m study site 7 kin NE of Portal, Cochise Co., Arizona, 1 July-12 August 1988 and 2-26 July 1989 (Randall, 1984). D. merriami and D. ordii occupied part of the site (ca. 200 by 140 m), while a more open area was occupied by D. spectabilis. Numbered stakes positioned at 15-m intervals served as trapping stations and for radiotracking coordinates. Resident kangaroo rats were trapped with Sherman live traps and marked with numbered eartags covered with color-coded reflectant tape for individual recognition at night. We trapped animals regularly to census and weigh them, attach ear tags, and check reproductive condition. In 1988, we trapped and marked 24 D. merriami (15 males, 9 females) and 38 D. ordii (13 males, 25 females). D. merriami had an average weight of 44.3 ± 1.10 g (males 48.9 ± 0.98, females 41.9± 1.29 g) and D. ordii 50.1 ± 1.4 g (males 51.4 ± 1.03, females 49.4 ± 1.68 g) for a body-size ratio of We implanted transmitters in 17 D. merriami (10 males, 7 females) and 11 D. ordii (5 males, 6 females). A sharp decline occurred in the population in 1989, and we trapped 7 D. merriami (5 males, 2 females) and 10 D. ordii (2 males, 8 females). Transmitters were implanted into three males and one female D. merriami and two males and four females of D. ordii. We implanted mouse-style transmitters with an internal antenna and individually tuned frequencies (AVM Instrument Inc., Livermore, CA) in animals captured on the study site and transported to the laboratory at the Southwestern Research Station. We anesthetized kangaroo rats with a light dosage of sodium pentobarbital (0.02 cm3 /40 g body weight) and implanted the 3-g transmitter-battery unit, encapsulated in beeswax, subcutaneously and lateral to the dorsal midline. Animals were allowed to recover and released at the point of capture the following night. We located kangaroo rats by approaching them on foot with a 12-channel receiver and a hand-held Yagi antenna. We located individuals at I-h intervals for 6-8 h/night, with day burrows located before sunset, and recorded coordinates of locations from the trapping grid by speaking into a hand-held tape recorder. In 1988, we tracked 12 D. merriami a minimum of 10 and a maximum of 40 nights and 11 D. ordii from 7 to 28 nights. Insufficient data were collected from other rats with transmitters because of failure of transmitters or disappearance or death of the animal. In 1989, we tracked four D. merriami (3 males, 1 female) and six D. ordii (2 males, 4 females) an average of 14.2 nights. We also made 62 h of focal-animal observations in We located a kangaroo rat via the radio signal and then followed it until it went into a burrow or disappeared. We recorded interactions with other kangaroo rats into a handheld tape recorder, including chases, distance foraging apart, and nonaggressive contact between any part of the animals' bodies. We estimated home ranges with the minimum-convex-polygon (MCP) and adaptive-kernel (AK) methods in the CALHOME program (Kie et al., 1996; see Gehrt and Fritzell, 1997, for a discussion of the advantages and disadvantages of MCP and AK methods). We considered the MCP a better representation of home ranges of kangaroo rats than AK and used it in our calculations and illustrations of overlap of home ranges. We calculated home ranges with a 95% distribution of data. We analyzed the following number of radiotracking locations for individuals of each species and sex in 1988: ex = 217) for male D. merriami (n = 6), (X = 119.5) for female D. merriami (n = 6), (X = 95) for male D. ordii (n = 5), (X = 180.4) for female D. ordii (n = 6). In 1989, we analyzed locations (X = 87) for four D. merriami and (X = 100.8) for six D. ordii. We measured overlap between home ranges by graphing the CALHOME output using Lotus, Freelance Graphics 96. From data points in Lo-

4 1300 JOURNAL OF MAMMALOGY Vol. 80, No.4 tus, we generated overlapping polygons for each individual we wanted to measure and used Scion ImagePC (NIH Image for Macintosh by Scion Corporation available at to create additional polygons where home ranges overlapped. We calculated from the polygons of each individual kangaroo rat the percentage of non-overlapping area (the area occupied exclusively by that individual) and the percentage of area that overlapped with other individuals. Olfactory tests.-we collected 12 D. ordii and 21 D. merriami for our laboratory studies during 4 weeks of trapping in 1991 and added five and seven more animals, respectively, during 3 weeks of trapping in July We limited our trapping to the area southeast of Portal, Cochise Co., Arizona, where both D. ordii and D. merriami lived sympatrically and which was near the area where our radiotracking data were obtained. We transported animals by automobile to the laboratory at San Francisco State University where they were housed on 4-5 cm of sand in individual plastic cages (26 by 16 by 13 cm) with wire lids and a tin can burrow. They were kept in a windowless room under a 14L: lod cycle and provided with wild-birdseed mix and leaves of lettuce on alternate days. All animals were reproductively mature; males had descended testes and females exhibited recurrent swelling of the vaginal area as an indication of estrus. In the interspecific olfactory tests, we tested seven males and seven females of each species in October 1992 for their responses to body oils, presented on cotton swabs, of same-sex conspecifics and heterospecifics in a counterbalanced order. Body oils were a mixture of secretions from the non-specialized sebaceous glands associated with hair and a specialized sebaceous gland located on the dorsum. Sebum from this gland forms a gradient down the sides of the animal (Westerhaus, 1983). We determined a dorsal gland to be functional by presence of waxy sebaceous secretions (Quay, 1953). Both male and female D. ordii had functional dorsal glands, but because the dorsal gland in D. merriami is sexually dimorphic (Quay, 1953), only males had functional glands (Lepri and Randall, 1983). We collected body oils from donors by slowly and gently wiping a cotton swab, three times, along each side of the animal. Because sandbathing removes excess body oils, we deprived animals of sand for h in empty plastic cages (15 by 8 by 7 cm) with wire lids to ensure sufficient amounts of oil for all olfactory tests. We tested animals in a circular aluminum arena (90 cm in diameter and 60 cm high) on ca. 3 cm of sand. The arena had removable sides to facilitate easy removal of sand and cleaning of the bottom of the arena with 95% alcohol between tests to remove odors. We imprinted two 17 -cm diameter circles in the test arena by pressing a circular plastic container into the sand on opposite sides of the arena 7 cm from the arena's edge. Into the center of one circle we placed a 2.5-cm long cotton swab with the body oil stimulus upright in the sand and in the other circle we placed an identical swab with no scent as a control. We transported animals between their cages and the test arena within their tin-can burrows and placed the can within the arena. Immediately after the animal vacated the can, we removed it and began to record the amount of time sniffing and approaching the swabs (tabulated as time investigating) within the imprinted circle for a 5-min test (Randall, 1991). We also tabulated number of times kangaroo rats entered each circle (approach) and sandbathing frequencies. Testing began ca. 2 h into the animal's dark cycle under a red light placed directly above the arena. All tests were conducted blind, so we were unaware of the gender and species of the test animal, stimulus presented and position of the stimulus. We controlled for position effects by changing position of stimuli within the arena (north, south, east, or west) and position of the observer relative to the stimuli. Because of asynchronous estrous cycles between the two species, we were unable to control totally for estrous stages. One female of each species was in an estrous condition during 50% of testing, while all remaining females were in a non-estrous condition throughout the testing period. Of the 28 animals tested, 9 D. ordii and 12 D. merriami had been tested previously in the arena in a sandbathing study. We habituated seven kangaroo rats not previously used in a test to the arena for 5 min. Number of animals tested differed each day because we used some test animals as scent donors. We tested 6-8 animals (at least one female and one male of each species) each day. Because of the testing schedule, ca. 75% of animals were tested prior to donating scent and 25% were tested after donating. Over a 4-day period, 20 animals (five of each species and of each sex) were

5 November 1999 PERRI AND RANDALL-BEHAVIOR OF KANGAROO RATS 1301 tested first and became donors the following day (24-h after testing), while eight animals (two of each species and of each sex) donated first and were tested h after donating. The above sequence was repeated over the following 4-day period to complete the second test for each animal. Each animal was tested twice and used twice as a scent donor. We conducted intraspecific olfactory tests 1-15 January We tested gender recognition in D. ordii to compare with D. merriami (Randall, 1981) by testing responses of D. ordii to body oils of opposite-sex conspecifics. Testing procedures and measured responses were similar to the interspecific test above, except we chose to control for estrous condition, instead of controlling the schedule of testing, because stimuli were opposite-sex individuals. Also, each individual was tested only once with one stimulus. Testing schedule depended upon the time at which females entered a non-estrous condition. Forty-eight hours after a female entered a nonestrous condition (indicated by a closed non-perforate vagina), we either tested her or deprived her of sand to become a scent donor. To test for familiarity, we tested seven nonestrous females and seven sexually mature males for their choice of scent of familiar and unfamiliar opposite-sex conspecifics. Test procedures and measured responses were similar to those in the previous swab investigations, except that both cotton swabs contained body oils. To establish familiarity, male and female pairs cohabited in cages (26 by 16 by 13 cm with wire lids) with a Plexiglas barrier that had 15 5-mm diameter holes drilled to allow exchange of odor and soil containing body oils. The seven pairs lived together for 6 weeks before testing during which time each female completed one to two complete estrous cycles. No females, however, exhibited estrus at the time of testing 5-11 March All animals had prior experience in the test arena. During the test, animals had a choice between cotton swabs with the body oils of their familiar cage-mate and body oils of an unfamiliar conspecific. Each animal donated scent once as a familiar cage-mate and once as an unfamiliar cage-mate. We separated scent donors from their cage-mates and deprived them of sand for 48 h before donating scent. After the 1st testing day, we housed scent donors and test animals together for 24 h before separating cage-mates again to deprive the second set of scent donors of sand for 48 h. Paired encounters.-we conducted paired encounters between same-sex pairs in the laboratory to detect differences in aggression and ability of one species to displace the other. We paired 12 D. ordii (six males, six females) with each other and with 10 D. merriami that were used as stimulus animals. Two D. merriami were paired twice. Because data for conspecifics paired together were not independent, we met assumptions for independence by averaging (total score of each behavior in the encounter divided by two) scores of behavior tabulated in the six conspecific encounters and the two scores of the same two animals in interactions with heterospecifics. We combined data for males and females and analyzed the averages of the six pairs with a Wilcoxin-sign rank test. We video-taped behavior (Table 1) during 15- min encounters with a camera (8 mm Sony EVe-X7 video camcorder) mounted directly above the same arena used in olfactory investigations. We placed one individual of a designated pair on each side of a removable cardboard barrier in the arena, removed the barrier, and began the test. At the end of the 15-min test, we separated individuals by placing the barrier between them. We distinguished individuals in paired encounters from a mark with a water-soluble marker on the mid-dorsal surface just posterior to the head. The same body area of the other individual also was marked with a moist cotton swab, ca. 4 h prior to testing. We paired animals that differed by <5 g and tested each animal on two separate days, 48 h apart. We tested one to six female D. ordii over 5 testing days and one to five male D. ordii males over 7 days. The schedule for males differed from the schedule for females because we used two male D. merriami twice as heterospecific stimuli, and we spaced encounters of these males 48 h apart. For reasons similar to those given for the previous investigations, we were unable to control completely estrous condition in females. One female of each species was in estrus, while all remaining females were in a non-estrous condition throughout testing March Because all but one D. merriami had no experience in the arena and 1 year had lapsed since all D. ordii had experience in the arena, we gave each animal 5 min alone in the arena on sand,

6 1302 JOURNAL OF MAMMALOGY Vol. 80, No.4 TABLE i.-descriptions of behaviors tabulated for Dipodomys ordii in paired encounters with D. ordii and D. merriami. Behavior Approach other Approached by other Approached each other Displaced other Displaced by other Not displaced Initiates nasal contact Allows nasal contact Initiates physical contact Sandbathing behavior 24 h prior to testing, to habituate them to the arena and video camera. We analyzed the video tapes frame by frame. We copied the 8 mm video tapes to VHS with running time burned into a window dub (accurate to 1130 s) and viewed tapes using a Sony i9-inch television monitor and a ProScan PSVR8i VHS player with jog-and-shuttle control. For time measurements, we recorded frame times at the beginning and end of measured behaviors (Table i). We then converted these frames into real time. We used SYSTAT (Wil- 250, , 250 Description The test animal decreases distance between itself and the other animal in any noncontinuous decrease in distance (pause of 2/30 s counts as a separate approach). The other animal decreases distance between itself and the test animal. Distance between both animals is decreased simultaneously. The test animal decreases distance between itself and the other animal as the other animal increases distance between itself and the test animal. The other animal decreases distance between itself and the test animal as the test animal increases distance between itself and the other animal. The other animal decreases distance between itself and the test animal, but the test animal does not increase distance. The test animal moves its rostrum to within I cm of the other animal. The test animal does not increase distance while the other animal initiates nasal contact. Same as nasal contact except uses other parts of body. As described by Randall (1981). 200 ' B 50- i;..;.., '.:':> '....' U FIG. i.-distribution of home ranges of A) Dipodomys merriami and B) D. ordii and individual symbols of day burrows on the same isoby-2s0-m area. Males are represented by dashed lines and closed symbols; females are solid lines with open symbols. kinson, i990) for all statistical analyses as specified. RESULTS Spacing.-We found considerable overlap in home ranges between the two species in 1988 (Fig. 1). Home ranges of D. ordii overlapped those of D. merriami by 36.4%, and D. merriami overlapped D. ordii by 49.5%. The true overlap probably was more extensive than our data show because only a portion of kangaroo rats on the study site had radiotransmitters. Both species typically used individual burrows during the day, and many individuals had more than one day burrow in the home area. Entrances of day burrows of the two species could be in proximity, and we found kangaroo rats using day burrows ca. 1 m apart on opposite ends of the same clump of bushes. Kangaroo rats, however, seemed to place their day burrows in areas of their home ranges that overlapped least with other kangaroo rats (Fig. 1). Both species showed gender-specific patterns of home-range overlap in intraspecific spacing. Conspecific females tended to have exclusive home ranges, while males overlapped females and other males (Fig. 1; Table 2). Home ranges of females over-

7 November /999 PERRI AND RANDALL-BEHAVIOR OF KANGAROO RATS 1303 TABLE 2.-Percentage of overlap in home ranges of Dipodomys ordii and D. merriami with conspecifics in high (1988) and low (1989) population densities. Interaction Year Species Female-female 1988 D.ordii 3.14 D. merriami D.ordii 0 D. merriami... c: (1) () en C> c: cts C> en (1) > c: (1) E I Male Female Male Female Conspecific Heterospecific FIG. 2.-Time investigating ex ::!:: SEs) cotton swabs with conspecific and heterospecific scent from same-sex animals and blank swabs with no scent as a control by male and female A) Dipodomys ordii and B) D. merriami during 5-min tests (n = 7 for each sex and species). Male-male Male on female Female on male lapped <10%; males overlapped 10-20%. Conspecifics of the opposite sex overlapped most with 50% overlap of male D. merriami on home ranges of females of D. mer- A riami (Table 2). We found no significant difference in size of home ranges. Male D. merriami, however, tended to have larger home ranges than either female D. merriami or D. ordii in 1988 (two-way ANOVA, species and sex interaction: F = 3.77; d.f. = 1,19; P = 0.067). Home ranges of male D. merriami averaged 1,671.7 ± m 2 compared to ± m 2 for female D. merriami, ± m 2 for male D. ordii, and ± m 2 for female D. ordii. Population densities in 1989 were lower than in 1988, and there was less overlap among kangaroo rats (Table 2). Despite the lower population density, neither species exhibited expanded home ranges. Sizes of home ranges of male and female D. ordii were similar, and the home ranges of six animals averaged ± 58.6 m 2 Home ranges of D. merriami averaged 1,302.0 ± m 2 The female D. merriami had a home range of m 2, and the three males averaged 1,638.7 ± 1,030.5 m 2 Olfactory tests.-d. ordii and D. merriami spent about the same amount of time investigating the cotton swab with body oils of heterospecifics as they spent investigating the control swab with no scent (Fig. 2A and 2B). There was no significant difference in time spent investigating or frequency of approaches toward scent of heterospecifics and the control by males and females of either species.

8 1304 JOURNAL OF MAMMALOGY Vol. 80, No.4 -c: Q).. 0 (/) 0).. c: CtI 0) (/) Q) > c: Q) E i= Male Female Scent D Blank Fig. 3.-Time investigating ex ::!:: SEs) opposite-sex body oils and a no scent control (blank) by female (n = 7) and male (n = 7) Dipodomys ordii during 5-min tests. Females of both species, however, responded to scent of same-sex conspecifics (Fig. 2). Female D. ordii spent more time investigating the swab with body oils of conspecific females than the control with no scent (paired t = 2.44, d.! = 6, P = 0.051; Fig. 2A), and they approached the scent of female D. ordii at higher frequencies (8.86 ± 2.1) than they approached the control (5.14 ± 1.03; t = 2.52, dj = 6, P = 0.045). Similarly, female D. merriami spent more time investigating the scent of female D. merriami compared with the control (t = 2.67, dj = 6, P = 0.037; Fig. 2B), but they did not approach the scent of female D. merriami at higher frequencies than they approached the control. In contrast to females, males exhibited no preference for scent of conspecific males. Male D. ordii and D. merriami spent equal time investigating and approaching scented and unscented cotton swabs (Fig. 2A and 2B). Female D. ordii exhibited a preference for scent of male D. ordii and spent significantly more time investigating body oils from males than the control (t = 3.37, dj = 5, P = 0.02; Fig. 3). Males, however, spent equal time investigating scent of fe- 1: 10 VJ C) 8 c:: ; 6.5 Q) E i= VJ o 10 0 co e c.. 8 c.. co -6 0 c:: Q) 4 :l Q" 2 u.. Familiar scent DUnfamiliar scent 12r o Male Female FIG. 4.-A) Time investigating ex ::!:: SEs) and B) number of approaches to scent of familiar and unfamiliar opposite-sex donors by male (n = 7) and female (n = 7) Dipodomys ordii during 5-min choice tests. males compared to the control (Fig. 3). Neither females nor males approached opposite-sex scent (8.86 ± 1.72 times and 6.71 ± 1.97 times, respectively) significantly more than the control (6.47 ± 2.01 times and 5.71 ± 1.13 times, respectively). Female D. ordii preferred scent of familiar males to unfamiliar males and spent more time investigating scent of their familiar, male cage-mates compared with unfamiliar males (t = 4.12, dj = 6, P = 0.006; Fig. 4A). Females approached scent of familiar males at higher frequencies than the scent of unfamiliar males (4.43 ± 1.13, t = 4.56, d.! = 6, P = 0.004; Fig. 4B). A B

9 November 1999 PERRI AND RANDALL-BEHAVIOR OF KANGAROO RATS 1305 TABLE 3.-Frequency of behavior of D. ordii in 5-min paired encounter with Dipodomys ordii and with D. merriami (n = 6 pairs; see Table 1 for definitions of behavior). Encounters Behavior Frequency of approach To other 62.4 By other To each other 8.01 Frequency to displace Other By other Frequency not displaced "P < X D.ordii Male D. ordii, however, showed no preference for scent of familiar females. There was no significant difference in time spent investigating or in frequency of approaches toward familiar and unfamiliar scent of females by males (Fig. 4A and 4B). Behavioral interactions.-in general, D. ordii interacted more with conspecifics than with D. merriami and was approached and displaced by conspecifics more than by D. merriami. (Table 3). Conspecifics approached D. ordii significantly more than they were approached by D. merriami (Wilcoxin sign-rank test, Z = 0.046), and D. ordii was displaced by conspecifics significantly more times than displaced by D. merriami (z = 0.031). D. ordii displaced D. merriami twice as often as D. merriami displaced D. ordii, but that comparison was not significant (z = 0.37, Table 3). No other behavioral comparisons differed significantly. Quick and non-injurious aggressive contact between individuals occurred in 25% of encounters. Two male D. ordii initiated physical contact six times with D. merriami. Two female D. ordii initiated physical contact with conspecifics 9 times and heterospecifics 31 times. Nasal contact and sandbathing occurred infrequently. We observed 11 interspecific interactions D. merriami SE X SE " " during focal-animal observation of D. ordii and D. merriami in the field D. ordii chased D. merriami in two interactions. However, D. merriami immediately returned after the brief chase to about the same location and began to forage. D. merriami ran away and avoided D. ordii without a chase in a third interaction. There was mutual tolerance in the remaining eight observations in which the two species foraged < 10m apart. In three cases, kangaroo rats were only ca. 1 m apart. Two interactions were observed between D. ordii. In one case, there was a chase, and in the other rats tolerated each other while foraging 1- m apart. In interactions between two D. merriami, two resulted in a chase, one in mutual tolerance and two in amicable contact. DISCUSSION Interference competition.-our radiotracking data and field observations revealed that D. ordii and D. merriami coexisted at high population densities in home ranges with considerable overlap. Our behavioral data suggest that avoidance, rather than interference, probably mediates coexistence of D. ordii and D. merriami. Although D. ordii dominated D. merriami in paired encounters, D. merriami displayed

10 1306 JOURNAL OF MAMMALOGY Vol. 80. No.4 an overall avoidance of D. ordii, and D. ordii was displaced by conspecifics significantly more than it was displaced by D. merriami. Agonistic behaviors always were initiated by D. ordii, but these occurred infrequently in both laboratory and field. This tolerance is in striking contrast to the agonistic behavior of D. ordii toward D. merriami observed by Blaustein and Risser (1976). We believe that our study, conducted in a larger arena than that used by Blaustein and Risser and coupled with observations in the field, reflects more accurately interspecific behavior between these species. If provided a large enough space, D. merriami usually retreats before any physical contract can occur between it and a dominant species (Congdon, 1974). Amount of agonistic behavior directed toward a competitor should reflect degree of true competition (MacArthur, 1972). Because individuals of a species have a greater impact on reproductive success of members of their own species than on members of another species, exploitation of resources should be reflected more in intraspecific interactions than in interspecific interactions. Aggression can be costly in time and energy, and it only should be employed if a resource is defensible and the benefit gained is greater than the cost. D. ordii, therefore, should not direct aggression toward D. merriami, which avoids them, but they should direct aggression toward conspecifics that do not avoid them and with which they are probably in more direct competition. Also, the slight dominance of D. ordii toward D. merriami in the laboratory may not result in competitive dominance under natural conditions in which individuals within their established home ranges exhibit site-specific dominance. For example, Peromyscus maniculatus dominates P. leucopus in the laboratory; but, in an area of sympatry where home ranges of species overlap, individuals dominated in core areas of their individual home ranges (Wolff et al., 1983). Distribution of day burrows on our study site suggests that kangaroo rats also may attempt to avoid the core area of other species. Even if interference interactions between the two species occurs, costly aggression and fighting are unnecessary to maintain separation. If individuals of both species coincide in time and space to use a similar resource, behavioral dominance of D. ordii may cause D. merriami to alter its pattern of resource use to avoid D. ordii, such as adopting a different pattern of seed hoarding (Jenkins and Breck, 1998). Encounters in the field often consisted of a brief chase and the immediate return of both animals to their respective foraging sites. In times of extreme limitation of resources, however, an increase in interspecific interactions as resources become scarce might occur (Kaufmann, 1983). Avoidance behavior could be a common component of species coexistence in deserts where animals increase home ranges and foraging efficiency rather than to increase aggressive interference (Dempster et al., 1992; Gouat et al., 1996). Olfaction and spacing.-we found no evidence that olfaction mediates interactions between D. ordii and D. merriami. Neither species responded to scent of the other species in our tests. Females of both species, however, responded to scent of female conspecifics, which they typically exclude from their home ranges, and they did not respond to scent of either heterospecific females or conspecific males, with which their home ranges overlap. Males also did not respond to scent of either conspecific or heterospecific males. In D. merriami (Behrends et al., 1986a, 1986b; 0' Farrell, 1980; Randall, 1989) and D. ordii (this study), spatial arrangement of conspecific neighbors exhibits a distinct gender-related pattern in which females occupy exclusive home ranges. Females may space themselves through olfactory information about the location of conspeific female neighbors. Males that do not respond to scent of other kangaroo rats share home ranges with them. Familiarity also may be an important el-

11 November 1999 PERRI AND RANDALL-BEHAVIOR OF KANGAROO RATS 1307 ement in social structure of D. ordii. The preference of female D. ordii for scent of familiar males suggests that (as in D. merriami-randall, 1991) olfactory cues may function in neighbor recognition. In both species, familiarity may play an important role, not only to space neighbors, but in mating interactions. In D. ordii (this investigation) and D. merriami (Randall, 1991), females preferred scent of familiar to unfamiliar males, whereas males did not discriminate the scent of either familiar or unfamiliar females. Female D. merriami allowed familiar, but not unfamiliar, males to contact them, while males attempted to contact both familiar and unfamiliar females equally (Randall, 1989). This suggests that familiarity is important to females, especially in allowing contact by males. Because mating occurs primarily between familiar individuals in D. merriami (Randall, 1989) and familiarity is established by repeated exposure between close neighbors, males may compete for preferred locations within proximity to conspecific females (Daly, 1977 ; DeVries et al., 1997). Community structure.-the community composition of D. ordii and D. merriami seems shaped by intraspecific spacing behavior and interspecific interactions in which the species avoid each other. Intraspecific interactions are probably more important in shaping heteromyid communities of D. ordii and D. merriami where they coexist than interspecific interactions. Intraspecific behaviors of recognition in D. ordii (this study) and D. merriami (Randall, 1991) may act to space conspecifics in similar ways. Spacing behavior has an important effect on ecological processes, such as regulation of population densities and use of resources (Brown and Orians, 1970; Gliwicz, 1988; Ims, 1988). Perhaps, intraspecific spacing behavior regulates populations of D. ordii and D. merriami and allows them to coexist ecologically. Perturbation experiments of sympatric D. ordii and D. merriami revealed a strong intraspecific response from contiguous populations of the two species of heteromyids (Schroder and Rosenzweig, 1975), which suggests that individuals of a species are in greater competition with conspecifics than closely related heterospecifics. D. ordii may be unable to exclude D. merriami from a sympatric area, even when densities appear high, as evidenced by coexistence of both species at high densities in the Portal area (Heske et al., 1994; Valone et al., 1995). The Portal area may represent a transitional habitat (Brown and Heske, 1990) that differs from habitats occupied by either species alone to cause the species to be more strongly influenced by intraspecific rather than interspecific interactions (Ayala, 1970). We suggest that D. ordii and D. merriami coexist in the Portal area because populations of each species are kept below saturation for shared resources. The successful increase of D. ordii on the study site without extermination of existing species implies that the original community was unsaturated (Pianka, 1988). D. ordii began to appear in low numbers on the Portal site in 1982, which had been occupied by only D. merriami and D. spectabilis since Densities on this site and a nearby area fluctuated in synchrony, peaked in 1988, and declined substantially in the following years (Brown and Heske, 1990a; Heske et ai., 1994). The two species continued to coexist, however, for 10 years without excluding each other on our study site and on another study site nearby (Heske et al., 1994). Because spacing behavior in female rodents may limit number of possible breeding territories and density of mature females in the population (Bondrup-Nielsen, 1986; Saitoh, 1981), space available for home ranges of females also may limit Dipodomys to cause the populations of each species to exist below the density that the habitat is capable of supporting and below the density where competitive interactions between species normally operate. Schroder (1987) suggested that the ecological similarity between D. ordii and D. merriami may result in neither species having a clear competitive advan-

12 1308 JOURNAL OF MAMMALOGY Vol. 80. No.4 tage. Fluctuations of immigration from and emigration to allopatric populations, predation, and disease may strongly influence interactions to keep intraspecific densities below saturation and permit coexistence in a habitat that differs from the habitat occupied by either species alone, even if species use similar resources. ACKNOWLEDGMENTS We are grateful to all the students who helped with radiotracking and trapping, especially J. Shore and M. B. Stone, and for support of the staff at the Southwestern Research Station. We especially thank B. Robbins and R. Larson for their support and assistance and to them, S. Jenkins, M. Price and an anonymous reviewer for a critical review of the manuscript. J. A. Randall appreciates the continued support of her research on desert rodents by the National Science Foundation and the National Geographic Society. LITERATURE CITED AYALA, F Competition, coexistence and evolution. Pp , in Essays in evolution and genetics (M. K. Hecht and W. D. Steere, eds.). Appleton-Century-Crofts, New York. BANKS E. M., U. W. HUCK, AND N. J. MANKOVICH Interspecific aggression in captive male lemmings. Animal Behaviour, 27: BASSET, A Body size-related coexistence: an approach through allometric constraints on homerange use. Ecology, 76: BEHRENDS, P., M. DALY, AND M. I. WILSON. 1986a. Range use patterns and spatial relationships of Merriam's kangaroo rats (Dipodomys merriami). Behaviour, 96: b. Above-ground activity of Merriam's kangaroo rats (Dipodomys merriami) in relation to sex and reproduction. Behaviour, 96: BLAUSTEIN, A R., AND A C. RISSER Interspecific interactions between three sympatric species of kangaroo rats (Dipodomys). Animal Behaviour, 24: BONDRUP-NIELSEN, S Investigation of spacing behaviour of Clethrionomys gapperi by experimentation. The Journal of Animal Ecology, 55: BOWERS, M.A Foraging behavior of heteromyid rodents: field evidence of resource partitioning. Journal of Mammalogy, 63: BOWERS, M. A, AND J. H. BROWN Body size and coexistence in desert rodents: chance or community structure? Ecology, 63: BROWN, J. H Species diversity of seed-eating desert rodents in sand dune habitats. Ecology, 54: BROWN, J. H., AND B. A. HARNEY Population and community ecology of heteromyid rodents in temperate habitats. pp , in Biology of the Heteromydae (H. H. Genoways and J. H. Brown, eds.). Special Publication, The American Society of Marnmalogists, 10: BROWN, J. H., AND E. J. HESKE. 1990a. Temporal changes in a Chihuahuan Desert rodent community. Oikos, 59: b. Control of a desert-grassland transition by a keystone rodent guild. Science, 250: BROWN, J. H., AND G. A. LIEBERMAN Resource utilization and coexistence of seed-eating desert rodents in sand dune habitats. Ecology, 54: BROWN, J. H., AND J. C. MUNGER Experimental manipulation of a desert rodent community: food addition and species removal. Ecology, 66: BROWN, J. H., AND Z. ZENG Comparative population ecology of eleven species of rodents in the Chihuahuan Desert. Ecology, 70: BROWN, J. L., AND G. H. ORIANS Spacing patterns in mobile animals. Annual Review of Ecology and Systematics, 1 : CONGDON, J Effect of habitat quality on distributions of three sympatric species of desert rodents. Journal of Marnmalogy, 55: CRANFORD, J. A, AND T. L. DERTING Intra- and interspecific behavior of Microtus pennsylvanicus and Microtus pinetorum. Behavioral Ecology and Sociobiology, 13:7-11. DALY, M Some experimental tests of the functional significance of scent-marking by gerbils (Meriones unguiculatus). Journal of Comparative Physiological Psychology, 91: DAYAN, T., AND D. SIMBERLOFF Morphological relationships among coexisting heteromyids: an incisive dental character. The American Naturalist, 143: DEMPSTER, E. R., R. DEMPSTER, AND M. R. PERRIN A comparative study of the behaviour of six taxa of male and female gerbils (Rodentia) in intra- and interspecific encounters. Ethology, 91: DEVRIES, A C., C. L. JOHNSON, AND S. C. CARTER Familiarity and gender influence social preferences in prairie voles (Microtus ochrogaster). Canadian Journal of Zoology, 79: FRYE, R. J Experimental field evidence of interspecific aggression between two species of kangaroo rats (Dipodomys). Oecologia, 59: GEHRT, S. D., AND E. K. FRITZELL Sexual differences in home ranges of raccoons. Journal of Mammalogy, 78: GLIWICZ, J The role of dispersal in models of small rodent population dynamics. Oikos, 52: GOUAT, P., I. E. YAHYAOUI, V. MANDIER, B. LIVOREIL, AND C. BAUDOIN Social behaviour and the use of space in two sympatric ground squirrels, Spermophilus spp., from the Chihuahua desert. Pp. 9-20, in Fifth international conference rodens & spatium, biodiversity and adaptation (A Zaime, ed.). Acetes Editions, Rabat, Morroco. HAIM, A., AND F. M. ROZENFELD Temporal

13 November 1999 PERRI AND RANDALL-BEHAVIOR OF KANGAROO RATS 1309 segregation in coexisting Acomys species: the role of odour. Physiology and Behavior, 54: HALL, E. R The mammals of North America. Second ed. John Wiley & Sons, New York, 1: HESKE, E. J., J. H. BROWN, AND S. MISTRY Longterm experimental study of a Chihuahuan desert rodent community: 13 years of competition. Ecology, 75: IMS, R. A Spatial clumping of sexually receptive females induces space sharing among male voles. Nature, 335: JENKINS, S. H., AND S. W. BECK Differences in food hoarding among six species of heteromyid rodents. Journal of Mammalogy, 79: JOHNSTON, R. E., AND T. A. ROBINSON Crossspecies discrimination of individual odors by hamsters (Muridae: Mesocricetus auratus, Phodopus campbelli). Ethology, 94: KAUFMANN, J. H On the definitions and functions of dominance and territoriality. Biological Review, 58:1-20. KENAGY, G. J Daily and seasonal patterns of activity and energetics in a heteromyid rodent community. Ecology, 54: KIE, J. G., J. A. BALDWIN, AND C. J. EVANS CALHOME: a program for estimating animal home ranges. Wildlife Society Bulletin, 24: KOTLER, B. P., J. S. BROWN, AND W. A. MITCHELL The role of predation in shaping the behaviour, morphology and community organisation of desert rodents. Australian Journal of Zoology, 42: LEMEN, C. A Seed size selection in heteromyids: a second look. Oecologia, 35: LEPRI, J. J., AND J. A. RANDALL Hormonal regulation of sandbathing in male kangaroo rats (Dipodomys merriami). Behavioral Neural Biology, 37: LIVOREIL, B., P. GOUAT, AND C. BAUDOIN A comparative study of social behaviour of two sympatric ground squirrels (Spermophilus spilosoma, S. mexicanus). Ethology, 93: MACARTHUR, R. H Geographical ecology: patterns in the distribution of species. Harper & Row, New York. M'CLOSKEY, R. T Community structure in sympatric rodents. Ecology, 57: Niche separation and assembly in four species of Sonoran desert rodents. The American Naturalist, 112: O'FARRELL, M. J Seasonal activity patterns of rodents in a sagebrush community. Journal of Mammalogy, 55: Spatial relationships of rodents in a sagebrush community. Journal of Mammalogy, 61: OSTFELD, R. S Limiting resources and territoriality in microtine rodents. The American Naturalist. 126:1-15. PIANKA, E. R Evolutionary ecology. Harper and Row, New York. PRICE, M. V The role of microhabitat in structuring desert rodent communities. Ecology, 59: Laboratory studies of seed size and seed species selection by heteromyid rodents. Oecologia, 60: PRICE, M. V., AND K. M. HEINZ Effects of body size, seed density, and soil characteristics on rates of seed harvest by heteromyid rodents. Oecologia, 61 : PRICE, M. v., AND O. J. REICHMAN Distribution of seeds in Sonoran desert soils: implications for heteromyid rodent foraging. Ecology, 68: QUAY, W. B Seasonal and sexual differences in the dorsal skin gland of the kangaroo rat (Dipodomys). Journal of Mammalogy, 34:1-14. RANDALL, J. A Olfactory communication at sandbathing loci by sympatric species of kangaroo rats. Journal of Mammalogy, 62: Territorial defense and advertisement by footdrumming in bannertail kangaroo rats (Dipodomys spectabilis) at high and low population densities. Behavioral Ecology and Sociobiology, 16: Neighbor recognition in a solitary desert rodent Dipodomys merriami. Ethology, 81: Sandbathing to establish familiarity in the Merriam's kangaroo rat, Dipodomys merriami. Animal Behaviour, 41: Behavioural adaptations of desert rodents (Heteromyidae). Animal Behaviour, 45: REICHMAN, O. J Behavior of desert heteromyids. Great Basin Naturalist Memoirs, 7: REICHMAN, O. J., AND D. OBERSTEIN Selection of seed distribution types by Dipodomys merriami and Perognathus amplus. Ecology, 58: REICHMAN, O. J., AND M. V. PRICE Ecological aspects of heteromyid foraging, pp , in Biology of the Heteromydae (H. H. Genoways and J. H. Brown, eds.). Special Publication, The American Society of Mammalogists, 10: ROSENZWEIG, M. L., AND J. WINAKUR Population ecology of desert rodent communities: habitats and environmental complexity. Ecology, 50: SAITOH, T Control of female maturation in high density populations of the red-backed vole Clethrionomys rufocanus bedfordiae. The Journal of Animal Ecology, 50: SCHRODER, G. D Mechanisms for coexistence among three species of Dipodomys: habitat selection and an alternative. Ecology, 68: SCHRODER, G. D., AND M. L. ROSENZWEIG Perturbation analysis of competition and overlap in habitat utilization between Dipodomys ordii and Dipodomys merriami. Oecologia, 19:9-28. VALONE, T. J., J. H. BROWN, AND C. L. JACOBI Catastrophic decline of a desert rodent, Dipodomys spectabilis: insights from a long-term study. Journal of Mammalogy, 76: WESTERHAUS, M. D A histological comparison of the dorsal and generalized holocrine skin glands