SEVERAL fundamental studies in community ecology

2008, No. 2 COPEIA June 4 Copeia 2008, No. 2, 261 272 Niche Relationships and Interspecific Interactions in Antiguan Lizard Communities Jason J. Kolbe 1, Paul L. Colbert 2, and Brian E. Smith 2 Anolis lizards are the focus of most Caribbean lizard community ecology studies with few studies including other common species that might influence community structure. To study niche relationships and interspecific interactions in Antiguan lizard communities, we used five offshore islands with varying combinations of three diurnal lizards: Ameiva griswoldi, Anolis leachii, and Anolis wattsi. We collected data on perch height, substrate, thermal microhabitat, body size, head length, daily activity, and abundance to characterize the ecological niche of each species. Ameiva griswoldi was more similar to A. leachii in size and daily activity, but more similar to A. wattsi in perch height, and A. leachii and A. wattsi were more similar in thermal microhabitat. This pattern of niche differentiation was consistent with niche complementarity, where species are similar on some niche axes but differ on others. Using the same niche characteristics as in species comparisons, we tested for a niche shift among islands for A. wattsi. In the absence of A. griswoldi, A. wattsi used lower perches, sunnier microhabitats, and was found more often on the ground. In contrast, with A. leachii absent, A. wattsi perched higher, more often in the shade, and on trunks. Furthermore, A. wattsi was most abundant when with A. leachii only, but least abundant when alone with A. griswoldi. These results suggest interspecific interactions, most likely due to competition and intraguild predation, are important for structuring Antiguan lizard communities. SEVERAL fundamental studies in community ecology involve Caribbean Anolis lizards, including pioneering work on resource partitioning (Schoener, 1968, 1974), interspecific competition (Roughgarden et al., 1983; Rummel and Roughgarden, 1985), thermal ecology (Huey and Webster, 1976), and behavioral ecology (Moermond, 1979a, 1979b). Anoles are typically small, arboreal insectivores, and most Caribbean islands contain at least one or two species (Losos, 1994). Few studies of Caribbean lizard communities, however, include other lizard species likely to interact with anoles (but see Powell et al., 1996; Simmons et al., 2005). In particular, terrestrial lizards in the genera Ameiva and Leiocephalus are widespread and common in the Caribbean (Schwartz and Henderson, 1991). Furthermore, recent studies detailing interactions between Anolis sagrei and Leiocephalus carinatus in the Bahamas show strong interspecific interactions including predation (Schoener et al., 2002; Losos et al., 2004). Because most lizard communities in the Caribbean with Anolis also have Ameiva, Leiocephalus, or both species, the potential for interspecific interactions between anoles and more distantly related species exists. Therefore, a broader taxonomic scope is warranted when studying Caribbean lizard communities. Recent studies evaluating classic niche dimensions food, space, and time in tropical lizard communities show local interactions are strongest between species within a family (e.g., Polychrotidae or Teiidae) and weakest between species from different families (Vitt and Zani, 1996; Vitt et al., 1999). If niche conservatism is found in Caribbean lizard communities as well, then interactions between Anolis species are expected to be stronger than interactions between Anolis and either Ameiva or Leiocephalus. Although evidence exists for interspecific interactions between Anolis species in present-day Caribbean lizard communities (Losos, 1994; Roughgarden, 1995; Leal et al., 1998), interactions with more distantly related species are little known. Comparisons of local communities that vary in composition over a small spatial scale can provide evidence for how interspecific interactions might influence niche relationships (Schoener, 1974). We tested these ideas for niche relationships and interspecific interactions in Antiguan lizard communities, which are composed of varying combinations of three diurnal lizards. Anolis leachii is a medium-sized, highly arboreal lizard endemic to Antigua and Barbuda and A. wattsi is a small lizard inhabiting the ground and low vegetation on Antigua. Both species exhibit the typical sit-and-wait foraging style of anoles (Roughgarden, 1995). Aggressive interactions between these two species usually result in A. leachii displacing A. wattsi (Schwartz and Henderson, 1991; J. Kolbe, pers. obs.). Anolis leachii also preys on A. wattsi (J. Kolbe, pers. obs.; O. Davis, pers. comm.). The Ground Lizard, Ameiva griswoldi, is a medium to large, actively foraging lizard almost completely restricted to the ground 1 Department of Biology, Campus Box 1137, Washington University, St. Louis, Missouri 63130. Present address: Museum of Vertebrate Zoology, 3101 Valley Life Sciences Building, University of California, Berkeley, California 94720; E-mail: kolbe@berkeley.edu. Send reprint requests to this address. 2 Department of Biology, Black Hills State University, 1200 University Street Unit 9044, Spearfish, South Dakota 57799. Submitted: 18 January 2007. Accepted: 30 October 2007. Associate Editor: J. W. Snodgrass. F 2008 by the American Society of Ichthyologists and Herpetologists DOI: 10.1643/CE-07-011

262 Copeia 2008, No. 2 Island is 2.1 ha and also contains both Anolis species, but A. griswoldi has never been reported from this island. At 0.9 ha, Red Head Island is the smallest island in this study, and it has A. wattsi and A. griswoldi, but no A. leachii. Fig. 1. Map of the Northern Lesser Antilles with an inset of the North Sound area of Antigua showing the five offshore islands used in this study. and is endemic to Antigua and Barbuda. Ameiva griswoldi chases A. wattsi in aggressive interactions (J. Kolbe, pers. obs.) and Ameiva are well-known predators of other lizards (Vitt and Colli, 1994; Vitt and Zani, 1996). We are not aware of any direct interactions between A. griswoldi and A. leachii. We collected data from five offshore islands in Antigua that varied in their species composition: one island had A. griswoldi and A. wattsi, two islands had A. leachii and A. wattsi, and two islands had all three species. First, we pooled data across islands for the three species to delimit the niche space occupied by each species using the niche axes of daily activity, microhabitat use, and morphology. Second, we tested for the presence of a niche shift in A. wattsi. We focused on A. wattsi because it was found on all five islands, but in varying combinations with the other species. If A. wattsi shifts its microhabitat use, daily activity, or morphology depending on the species with which it coexists on an island, then evidence may exist for interspecific interactions. Explicit predictions for microhabitat use and abundance of A. wattsi are: A. wattsi will use higher perches and will perch more often on trunks in the absence of A. leachii and will use lower perches and will be found more often on the ground in the absence of A. griswoldi; A. wattsi will use shadier thermal microhabitats in the absence of A. leachii and sunnier thermal microhabitats in the absence of A. griswoldi; and A. wattsi abundance will be higher on twospecies islands than on three-species islands. If some or all of these predictions hold, then evidence exists for interspecific interactions among species in Antiguan lizard communities. MATERIALS AND METHODS Study site. We conducted this study on five islands varying in size from 0.9 to 43 ha in the North Sound area of Antigua, West Indies, during July August 2001 (Fig. 1). Islands were primarily composed of Caribbean littoral forest and dense scrub with grassy sandbars and rocky ridges on some islands (B. Smith, unpubl.). To facilitate comparisons among islands, we restricted data collection to the forest habitat on each island. Species composition varied among the islands: Green (43 ha) and Great Bird (9.9 ha) are the largest islands and have all three species. At the time of our study, the Antiguan Racer Snake, Alsophis antiguae, was present only on Great Bird Island (Daltry et al., 2001). Maiden Island is 8.3 ha and has both Anolis species, but A. griswoldi was last seen on this island in 1997 ( J. Daltry, pers. comm.). Rabbit Habitat availability. To determine if islands differed in vegetation structure, which could potentially influence lizard microhabitat use, we collected data at established survey sites along transects within the forest habitat on each island to quantify available habitat. We measured percent canopy cover with a spherical densiometer held at a height of 1.5 m, percent vegetation density by visual estimate, modal vegetation height (m), and maximum vegetation height (m) within each 6.28 m 2 survey site. Measurements were taken from an average of 22.4 sites per island (range 5 12 35; B. Smith, unpubl.). We used multivariate analysis of variance (MANOVA) to test for an overall difference among islands in these four habitat variables. If significant differences existed using the MANOVA, we conducted analysis of variance (ANOVA) for each habitat variable to characterize the differences and Tukey s honestly significant difference (HSD) post hoc tests to determine which island comparisons were significantly different. Percentages were arcsin square root transformed and vegetation heights were ln-transformed prior to analyses, and statistical tests were conducted in JMP (vers. 5.1, SAS Institute, Inc., Cary, North Carolina, 2003). Niche data collection. To test for niche differences among species and for a niche shift among islands in A. wattsi, we collected data on microhabitat use, morphology, and daily activity. We recorded microhabitat use data for each lizard encountered when its initial location was observed. We recorded species, sex, time, perch height (cm), thermal microhabitat (full sun, filtered sun, or shade), and substrate (agave, branch, ground, rock, trunk, or twig). We caught a subset of these individuals for morphological measurements. For captured individuals, we measured snout vent length (SVL) and head length (HL), both in mm. We obtained at least 50 microhabitat use observations and captured 20 individuals per sex for both A. leachii and A. wattsi on each of the five islands, with at least five microhabitat use observations each hour of the day from 0500 1900 h on each island. Microhabitat data and morphological measurements for A. griswoldi were taken from Davis (2001). We conducted visual encounter surveys (VES) to assess changes in microhabitat use and activity of each species throughout the day (Crump and Scott, 1994). One 15- minute VES was conducted in each half-hour time block from sunrise to sunset (i.e., 0500 or 0530 to 1900 h) for a total of 27 or 28 surveys on each island. Visual encounter surveys were conducted by walking at a constant pace and recording the species, time, sex, and estimated perch height to the nearest 5 mm for all individuals observed. The same general area was covered on each island during VES to ensure that variation in microhabitat use and abundance throughout the day could be attributed to daily patterns of activity and not spatial variation within an island. Microhabitat use. We evaluated three components of microhabitat use: substrate type, thermal microhabitat, and perch height. We tested for differences in substrate and thermal microhabitat use among species using contin-

Kolbe et al. Antiguan lizard community ecology 263 gency tests. Perch height data collected during VES were used in repeated-measures ANOVA to test for differences among species, times of day, and their interaction. Repeated-measures ANOVA was used because the VES repeatedly surveyed the same area within an island, varying only the time of day for each survey. In contrast, perch height data from habitat observations were collected over a period of several days; thus, we used a two-way factorial ANOVA to test for differences among species and times of day for these data. To determine if A. wattsi shifts its microhabitat use depending on the species with which it co-exists, we tested for differences among islands and times of day in substrate and thermal microhabitat using contingency tests. A twoway factorial ANOVA with island and times of day as treatment factors was used to determine if A. wattsi shifts its perch height use depending on the absence of A. griswoldi or A. leachii. Additionally, we used Kolmogorov Smirnov tests to determine if the perch height distribution of A. wattsi shifts between islands. Perch heights were ln-transformed prior to analyses. Morphology. Body size influences many aspects of a lizard s life including energetics, locomotion, and biotic interactions such as competition and predation (Peters, 1983; Naganuma and Roughgarden, 1990; Gerber and Echternacht, 2000). Furthermore, body size and head size are correlated with gape limits and prey size in Anolis (Schoener, 1968; Herrel et al., 2006) and lizards in general (Pianka and Vitt, 2003). Thus, larger head size allows individuals to consume larger prey, although not necessarily to the exclusion of smaller prey from their diet (Sexton et al., 1972), thereby assessing potential dietary differences among species. We used ANOVAs to test for differences in HL and SVL among species and among islands for A. wattsi. Head length and SVL were ln-transformed prior to analyses. density (F 4,107 5 4.43, P 5 0.0024, R 2 5 0.14), modal vegetation height (F 4,107 5 5.08, P 5 0.0009, R 2 5 0.16), and maximum vegetation height (F 4,107 5 3.20, P 5 0.0158, R 2 5 0.11), but only a small portion of the variation is explained for each habitat variable. Based on Tukey s HSD tests, Red Head Island has a significantly shorter, more open canopy with denser vegetation compared to the other islands, whereas the other four islands do not differ significantly (Table 1). Niche differences among species. Microhabitat differences exist among all three species. Anolis leachii primarily uses the upper portion of tree trunks and high branches, and A. wattsi uses the lower portion of tree trunks and the ground, whereas A. griswoldi uses the ground almost exclusively (substrate: x 2 5 1092.11, df 5 2; P, 0.0001, n 5 1383; Tables 2, 3; Figs. 2, 3). The two anoles are each found in shady thermal microhabitats 73% of the time, whereas A. griswoldi is in sunny areas 84% of the time (x 2 5 313.28, df 5 2; P, 0.0001, n 5 1226). Head length differs substantially between A. wattsi and the other two species (F 2,452 5 399.80, P, 0.0001, R 2 5 0.64) as does SVL (F 2,452 5 480.02, P, 0.0001, R 2 5 0.68; Fig. 4), whereas A. griswoldi and A. leachii are similar in both HL and SVL. Both Anolis species are active from 0500 1900 h, but A. griswoldi is not active until 0600 h (Fig. 5). The significant interaction between species and times of day suggests that the pattern of daily activity differs among species (Table 4; Fig. 5). Daily activity for A. griswoldi peaks at 0900 h with lower abundance earlier in the morning and later in the Daily activity and abundance. Differences in daily activity were evaluated using abundance data from the VES. We grouped the abundance data into two-hour time blocks from 0500 1900 h and used repeated-measures ANOVA to test for differences among species, times of day, and their interaction. Island could not be included in the repeated-measures ANOVA due to lack of replication. Therefore, we used a three-way ANOVA with species, island, times of day, and the interaction between species and times of day to test for an island effect on abundance. A species 3 island interaction could not be included because not all species island combinations existed. Instead, we conducted two-way factorial ANOVAs with island and times of day for each species separately. This allowed us to evaluate abundance differences among islands for each species (focusing primarily on A. wattsi) and determine if daily patterns of activity differed among islands depending on which species were present. Abundance data were analyzed as the mean number of lizards per 15-minute VES for all surveys within a given time block. RESULTS Habitat differences among islands. Islands differ in habitat availability (MANOVA: Wilks l 5 0.75, df 5 12, 278, P 5 0.0025). Follow-up ANOVAs show islands differ in canopy cover (F 4,107 5 2.34, P 5 0.0600, R 2 5 0.08), vegetation Fig. 2. Frequency of substrate use for the three Antiguan lizard species. Sample sizes are shown in parentheses.

264 Copeia 2008, No. 2 Fig. 3. Perch heights (mean 6 1 SE) for the three Antiguan lizard species in one-hour time increments from 0500 to 1900 h. afternoon. In contrast, A. wattsi is most active in the morning (,0600 h) and evening (,1800 h) and decreases in abundance from 0800 1400 h, which corresponds to the daily increase in abundance of A. griswoldi (Fig. 5). Abundance of A. leachii is low throughout the day. Additionally, both A. leachii and A. wattsi use higher perches during midday (Fig. 3), which corresponds to the increase in A. griswoldi activity on the ground (Fig. 5). Overall, A. wattsi is three times as abundant as A. griswoldi and five times as abundant as A. leachii (Table 4; Fig. 5). The islands with all three species have the lowest combined abundances (Green: 4.57 6 0.65 and Great Bird: 5.67 6 0.77), whereas the two islands with A. leachii and A. wattsi have the highest abundances (Maiden: 13.61 6 1.08 and Rabbit: 13.95 6 1.98) and abundance on the island with A. griswoldi and A. wattsi was intermediate (Red Head: 8.18 6 0.87). Perch height and substrate differences among islands for A. wattsi. Perch heights of A. wattsi differ significantly among islands and times of day (ANOVA: Islands, F 4,532 5 19.55, P, 0.0001; Times of day, F 13,532 5 8.86, P, 0.0001, R 2 5 0.40). Mean perch height is lowest on Rabbit Island (no A. griswoldi) at 6 cm, highest on Red Head Island (no A. leachii) at 40 cm, and intermediate on the other three islands, ranging from 22 31 cm (Fig. 6). Mean perch heights are lower during the early morning (0500 0800 h) and late evening (1600 1900 h), times with low A. griswoldi activity (Figs. 3, 5, 6). Mean perch heights reach between 70 110 cm on islands with A. griswoldi at midday, whereas on islands without A. griswoldi, perch height use did not increase markedly during midday and mean perch height never exceeded 40 cm. The significant island 3 times of day interaction confirms that the daily pattern of perch height use for A. wattsi differs among islands (ANOVA: island 3 times of day, F 52,532 5 2.42, P, 0.0001). Kolmogorov Smirnov tests reveal A. wattsi perch height distributions also shifted depending on the other species present; perch heights for A. wattsi shifted lower on islands without A. Table 1. Summary of Habitat Measurements on the Five Offshore Islands in Antigua. Mean (61 SD) and range in parentheses for each variable. Like letters indicate no significant difference between islands using Tukey s HSD post hoc test. Island Canopy Cover (%) Vegetation Density (%) Modal Veg. Height (m) Maximum Veg. Height (m) Great Bird 37.2 6 29.6 (0 86) A B 16.3 6 17.0 (1 72) B 1.9 6 0.6 (1.0 3.5) A 3.4 6 1.7 (1.2 7.8) A Green 35.5 6 31.3 (0 86) A B 24.2 6 21.3 (4 89) B 1.6 6 0.6 (0.7 3.5) A 3.1 6 1.2 (1.8 6.7) A Maiden 37.3 6 30.2 (0 96) A 27.0 6 22.0 (0 90) A B 2.2 6 1.9 (0.2 8.8) A 3.2 6 1.8 (1.1 8.8) A Rabbit 26.9 6 27.6 (0 79) A B 24.3 6 13.1 (8 56) A B 1.7 6 1.1 (0.5 5.2) A 2.7 6 1.1 (0.9 5.2) A B Red Head 14.9 6 27.3 (0 75) B 42.3 6 19.7 (20 96) A 0.9 6 0.4 (0.3 1.6) B 2.0 6 0.7 (0.8 2.8) B

Kolbe et al. Antiguan lizard community ecology 265 Fig. 4. Distributions of female and male head lengths and snout vent lengths (SVL) for the three Antiguan lizard species. Sample sizes are shown in parentheses.

266 Copeia 2008, No. 2 Table 2. Summary of Niche Dimensions for the Three Antiguan Lizards. Mean (61 SE) for perch height, head length, percentage shady thermal microhabitat use, and peak daily activity time. Data are from habitat observations except peak daily activity, which is from visual encounter surveys. Niche dimension n Ameiva griswoldi Anolis leachii Anolis wattsi Perch height (cm) 1437 0 6 2.39 122 6 2.43 24 6 2.01 Head length (mm) 455 22 6 0.45 22 6 0.33 13 6 0.29 Shady microhabitat 1226 16% 73% 73% Peak daily activity (h) 2817 0900 0900 0600 and 1800 griswoldi and higher on islands where A. leachii was absent (Table 5). These results are consistent with the prediction that perch height use depends on species composition except for Maiden Island, where A. wattsi perched at intermediate heights despite the absence of A. griswoldi. Substrate use differs both among islands (x 2 5 132.5, df 5 4, P, 0.0001, n 5 548) and times of day (x 2 5 147.9, df 5 13; P, 0.0001, n 5 548) for A. wattsi. Lizards on Rabbit Island are most often on the ground, whereas lizards on Red Head Island are most often on tree trunks. Anolis wattsi uses the ground most between 0500 0700 h and 1600 1900 h, whereas trunks are used most during the rest of the day. Thermal microhabitat and morphological differences among islands for A. wattsi. Anolis wattsi thermal microhabitat use differs both among islands (x 2 5 81.8, df 5 4; P, 0.0001, n 5 548) and times of day (x 2 5 66.8, df 5 13; P, 0.0001, n 5 548; Fig. 7). Lizards on Rabbit and Maiden Islands (no A. griswoldi) use sunny microhabitats 38% and 57% of the time, respectively, which is substantially more than the overall average of 27% (Table 2). Conversely, on the three islands where A. griswoldi is present, A. wattsi uses sunny microhabitats only 10% to 21% of the time, with lizards on Red Head Island (no A. leachii) using sunny microhabitats the least (Fig. 7). These results confirm predictions that A. wattsi uses sunny microhabitats more in the absence of A. griswoldi and shady microhabitats more in the absence of A. leachii. Temporal differences in thermal microhabitat use are characterized by use of sunny areas from 0900 to 1600 h on Rabbit and Maiden Islands (no A. griswoldi), whereas sunny microhabitat use never exceeds 40% within a time Table 3. Perch Height Use among Species and Times of Day for the Three Antiguan Lizards. Repeated-measures ANOVA used perch heights from visual encounter surveys, and the two-way factorial ANOVA used perch heights from habitat observation data. Repeated-measure ANOVA: Species 2 52.73,0.0001 Times of day 6 10.68 0.0191 Species 3 Times of day 12 2.58 0.0932 Two-way factorial ANOVA: Species 2, 1416 689.25,0.0001 Times of day 6, 1416 4.98,0.0001 Species 3 Times of day 12, 1416 7.31,0.0001 R 2 5 0.68 block on the three islands with A. griswoldi. Within-island analyses show significant differences in thermal microhabitat use among times of day on all five islands as well (all P, 0.05). Neither SVL (F 4,201 5 0.84, P 5 0.5006) nor HL (F 4,201 5 1.93, P 5 0.1073) differs among islands for A. wattsi. Daily activity and abundance differences among islands for A. wattsi. Anolis wattsi abundance differs significantly among islands and times of day (Table 6; Fig. 8). Anolis wattsi is present in highest abundance on Rabbit Island, with a mean of 26.7 lizards per 15-minute VES, followed by Maiden Island with a mean of 19.6 lizards. Both of these islands lack A. griswoldi. Mean abundance on the other three islands ranges from 7.5 to 13.7 lizards with Red Head Island having the lowest abundance of A. wattsi (Fig. 8). The low abundance on Red Head Island is contrary to the prediction for two-species islands, although the other two-species islands have the two highest abundances. All islands show a similar pattern of daily activity with a few exceptions. The general pattern is for higher abundance of A. wattsi in the early morning and late evening with the lowest abundance around midday (1100 1200 h; Fig. 8). Exceptions include the lack of an increase in abundance late in the day on Great Bird Island and delay in early morning peak abundance on Rabbit and Maiden Islands until 0600 0700 h. The general pattern of daily activity holds even on the islands without A. griswoldi. Significant differences among islands also exist in A. leachii and A. griswoldi abundances (Table 6). Ameiva griswoldi is most abundant on Red Head Island with a mean of 8.9 lizards and least abundant on Great Bird Island with a mean of 2.0 lizards. Maiden Island had the most A. leachii at 7.7 lizards per 15-minute VES, whereas the other three islands with A. leachii have between 1.2 1.4 lizards. DISCUSSION Niche relationships among Antiguan lizards. Anolis wattsi is similar to A. griswoldi in perch height and to A. leachii in thermal microhabitat, but differs from both species in size and daily activity. In turn, A. griswoldi and A. leachii differ in their perch height and thermal microhabitat, but are similar in size and daily activity (Table 2; Figs. 3 5). The two anoles are similar in thermal microhabitat only, thus the overall pattern of niche differentiation among Antiguan lizards is more consistent with niche complementarity, where species similar on one (or more) niche axis differ along other axes (Schoener, 1974), than with niche conservatism. Two predictions follow from this finding. First, competition between Anolis species is expected to be no stronger than competition between each anole and A. griswoldi, except for thermal microhabitats. Second, interactions between A. griswoldi and either anole are as likely to influence niche

Kolbe et al. Antiguan lizard community ecology 267 Fig. 5. Abundance (mean 6 1 SE) for the three Antiguan lizard species in one-hour time increments from 0500 1900 h. Data are from visual encounter surveys. structure as interactions between the two anoles (Losos et al., 2003). Given the niche relationships identified for Antiguan lizards (Table 2), how might these species interact? Evidence exists that time is an unimportant niche dimension in Central and South American lizard communities due to the broad overlap in activity among species (Vitt and Carvalho, Table 4. Abundance among Species, Times of Day, and Islands for the Three Antiguan Lizards. Both repeated-measures and three-way ANOVAs used abundances from visual encounter surveys. Note that not all species occur on each island. Repeated-measures ANOVA: Species 2 5.20 0.0315 Times of day 6 15.57 0.0096 Species 3 Times of day 12 13.56 0.0005 Three-way ANOVA: Species 2, 285 154.50,0.0001 Times of day 13, 285 2.40 0.0044 Islands 4, 285 37.34,0.0001 Species 3 Times of day 26, 285 4.72,0.0001 R 2 5 0.69 1995; Vitt and Zani, 1996, 1998a, 1998b). In contrast, marked differences in daily activity exist among Antiguan lizard species. In particular, A. wattsi is most active in the morning and evening when A. griswoldi is least active, and A. griswoldi is most active at midday when A. wattsi is least active, whereas A. leachii activity is constant and relatively low throughout the day (Fig. 5). Thermoregulatory constraints on active foraging may restrict Ameiva activity in the cooler morning and evening hours (Vitt and Colli, 1994; Vitt and Zani, 1996; Vitt et al., 2000), providing an opportunity for A. wattsi to forage on the ground without A. griswoldi present. Some temporal overlap between A. griswoldi and A. wattsi occurs during midday from 0800 1100 h (Figs. 5, 8); however, A. wattsi increases its perch height during this time, potentially limiting the opportunity for interactions with the ground-dwelling A. griswoldi (Figs. 3, 6). This pattern of limited ground activity by Anolis in the presence of Ameiva is observed in other Caribbean communities (Fobes et al., 1992; Eaton et al., 2002), and may be due to the threat of predation by Ameiva (Hodge et al., 2003; Simmons et al., 2005). One of three trials that placed a tethered A. wattsi within 1.5 m of a foraging A. griswoldi resulted in a lethal attack, whereas the A. wattsi was ignored in the other two instances. When A. wattsi was placed within 0.5 m of A. leachii the results were more dramatic, four of seven trials resulted in lethal attacks, most of which ended with A. leachii consuming A. wattsi (P. Colbert, unpubl.). Thus, A. wattsi habitat use and daily activity is consistent with the threat of aggression and predation by the two larger lizard species.

268 Copeia 2008, No. 2 Table 5. Perch Height Distribution Shifts between Three-Species and Two-Species Islands for A. wattsi. Kolmogorov Smirnov test results and the direction of perch height distribution shift. The Bonferroni corrected P- value is 0.0083. Three-species island Two-species island D P Shift direction Great Bird Maiden 0.2077 0.008 Lower Great Bird Rabbit 0.5049,0.0001 Lower Great Bird Red Head 0.1229 0.261 Higher Green Maiden 0.1708 0.058 Lower Green Rabbit 0.3090,0.0001 Lower Green Red Head 0.2540 0.001 Higher evaluate prey items directly, some aspects of morphology and habitat use suggest the potential for species differences. Perch height differences may provide access to different prey types (Schoener, 1968; Pianka, 1973), with A. griswoldi and A. wattsi both using the ground extensively and A. leachii perching more than 1 m higher on average than the other species (Fig. 3). Head length differences suggest A. griswoldi and A. leachii can eat substantially larger prey than A. wattsi (Schoener, 1968; Herrel et al., 2006; Fig. 4). The prediction that body size and habitat use leads to differences in diet among species needs to be tested. Intraguild predation among Anolis occurs in Caribbean and Central American communities (Gerber, 1999), and experimental evidence shows intraguild predation in Anolis with juveniles being eaten by adults twice their size (Gerber and Echternacht, 2000). In Antigua, both A. griswoldi and A. leachii are much larger than A. wattsi (Fig. 4), providing the opportunity for intraguild predation of not only juvenile A. wattsi, but also adults. In fact, both of the larger lizards eat A. wattsi; for example, a male A. leachii (SVL 5 65 mm) was caught in the process of consuming a female A. wattsi (SVL 5,40 mm) approximately two-thirds its size (J. Kolbe, pers. obs.). Future experiments investigating the relative importance of intraguild predation and interspecific competition for explaining the pattern of niche segregation observed in this study for Antiguan lizard communities would be valuable. Fig. 6. Mean perch height of Anolis wattsi in one-hour time increments from 0500 1900 h on each of the five offshore islands. Perch height data from both visual encounter surveys and habitat observations are shown. SE bars are omitted for clarity. Previous studies of New World tropical lizard communities that include both Anolis and Ameiva show evidence for niche conservatism (Vitt and Zani, 1996; Vitt et al., 1999). In particular, closely related species tend to eat similar prey even though these lizards do not necessarily live in the most similar microhabitats (Vitt et al., 1999). Although we did not Niche shift among islands in A. wattsi. Anolis wattsi shifts its perch height, thermal microhabitat, and daily activity in the absence of either A. griswoldi or A. leachii, suggesting interspecific interactions are important for niche partitioning (Schoener, 1975). Given the perch height differences among species (Fig. 3), A. wattsi shifts both its mean and distribution in the predicted directions, higher on Red Head Island (no A. leachii) and lower on Maiden and Rabbit Islands (no A. griswoldi; Table 5; Fig. 6). Additionally, the daily pattern of perch height use for A. wattsi varies among islands: A. wattsi increases its perch height during midday most on Red Head Island (no A. leachii), whereas no midday increase in perch height occurs on Maiden and Rabbit Islands (no A. griswoldi; Fig. 6). Several studies show Anolis limits ground activity when Ameiva are active (Eaton et al., 2002; Simmons et al., 2005), and shifts perch heights in the absence of congeners (Jenssen, 1973; Schoener, 1975; Losos et al., 1993). For example, Anolis schwartzi on St. Eustatius, which is ecologically similar to A. wattsi, decreases its perch height under experimental conditions with the higher

Kolbe et al. Antiguan lizard community ecology 269 Fig. 7. Percentage of sunny or shady thermal microhabitat observations for Anolis wattsi in two-hour time increments from 0500 1900 h on each of the five offshore islands. Fig. 8. Mean abundance of Ameiva griswoldi, Anolis leachii, and Anolis wattsi in one-hour time increments from 0500 1900 h on each of the five offshore islands. SE bars are omitted for clarity. perching Anolis bimaculatus present (Rummel and Roughgarden, 1985). Habitat availability may affect perch height use as suggested by other studies (Harris et al., 2004; Simmons et al., 2005; Johnson et al., 2006). Despite Red Head Island having a significantly shorter canopy than the other four islands, A. wattsi on this island perched the highest (Table 1; Fig. 6), suggesting interspecific interactions with A. griswoldi influenced its perch height use more than the habitat availability. On the other hand, the short canopy on Red Head Island may explain the absence of A. leachii on this island.

270 Copeia 2008, No. 2 Table 6. Abundance among Islands and Times of Day for the Three Antiguan Lizards. Two-way ANOVA results using data from visual encounter surveys. Ameiva griswoldi: Island 2, 62 29.11,0.0001 Times of day 6, 62 16.91,0.0001 Island 3 Times of day 12, 62 2.93 0.0028 R 2 5 0.76 Anolis leachii: Island 3, 82 83.94,0.0001 Times of day 6, 82 2.81 0.0154 Island 3 Times of day 18, 82 1.19 0.2879 R 2 5 0.78 Anolis wattsi: Island 4, 103 83.23,0.0001 Times of day 6, 103 26.87,0.0001 Island 3 Times of day 24, 103 2.94,0.0001 R 2 5 0.85 Thermal microhabitat use among islands for A. wattsi shows the predicted shift: shadier microhabitats are used on Red Head Island (no A. leachii) and sunnier microhabitats are used on Maiden and Rabbit Islands (no A. griswoldi; Fig. 7). This is consistent with experimental studies on St. Eustatius where A. schwartzi used hotter microclimates when with the larger A. bimaculatus (Rummel and Roughgarden, 1985). However, in that experiment, A. schwartzi shifted its activity from midday to late afternoon when in the presence of A. bimaculatus. In our study, daily activity is similar on all islands: A. wattsi is generally more active in the morning and evening on all islands regardless of species composition (Fig. 8). This suggests other factors, such as thermoregulation, may also play a role in governing A. wattsi activity in addition to interspecific interactions (Hertz, 1992; Nicholson et al., 2005). Although a niche shift in thermal microhabitat for A. wattsi may be due to interspecific interactions (Schoener, 1975), vegetation structure may also affect thermal microhabitats. On Red Head Island, the canopy is more open, but the understory vegetation is denser, likely creating a patchwork of warmer, open areas and cooler, densely vegetated areas. Unfortunately, it is difficult to determine how this habitat structure affects A. wattsi thermal microhabitat use and whether vegetation structure had a greater or lesser affect than co-existence with A. griswoldi. If the abundance of A.wattsi when co-existing with different species provides some insight into the strength of interspecific interactions (Roughgarden et al., 1983), then our results suggest stronger negative interactions with A. griswoldi than with A. leachii. Anolis wattsi abundance is highest on Maiden and Rabbit Islands (no A. griswoldi) and lowest on Red Head Island (no A. leachii; Fig. 8). Furthermore, on islands where A. wattsi and A. griswoldi co-occur, A. wattsi abundance increases as A. griswoldi abundance decreases, but no such relationship exists between A. wattsi and A. leachii. That is not to say A. leachii does not influence A. wattsi abundance. Extremely low A. leachii abundance on Rabbit is accompanied by the highest abundance of A. wattsi, whereas on Maiden the abundance of A. leachii is higher and abundance of A. wattsi is correspondingly lower. Experiments testing whether the strength and type of interactions among species affect abundances should follow-up these preliminary observations. Differences among islands in prey or predator abundances may also affect lizard abundances. Although we do not have data on prey abundances, the composition of avian predators does not appear to vary substantially among islands and their close proximity suggests birds could easily travel among islands (V. Joseph, J. Prosper, and A. Otto, unpubl.). Lastly, the lizard-eating snake, Alsophis antiguae, was found only on Great Bird Island during this study (Daltry et al., 2001), potentially confounding comparisons using this island. Microhabitat use, daily activity, and abundance of A. wattsi on Great Bird Island compared to other islands do not suggest a confounding effect of A. antiguae. In fact, both three-species islands, Great Bird and Green, are similar with regard to A. wattsi niche characteristics and abundance (Figs. 6 8). The pattern of niche complementarity among Antiguan lizard species and niche shift among islands for A. wattsi suggest interspecific interactions with the more distantly related A. griswoldi have a strong influence on A. wattsi microhabitat use, daily activity, and abundance. Future studies should focus on experimental evaluation of competition and intraguild predation as the mechanisms causing the patterns of niche partitioning found in this study. Furthermore, similar Anolis and Ameiva communities occur throughout the Lesser Antilles and provide replication for testing the generality of niche relationships found in Antiguan lizard communities. ACKNOWLEDGMENTS We thank the Columbus Zoo (Columbus, OH, USA), Cleveland Zoo (Cleveland, OH, USA), John Ball Zoo Society (Grand Rapids, MI, USA), Black Hills State University (BHSU), the Nelson Scholarship Fund (BHSU), Fauna and Flora International (FFI), and a National Science Foundation Grant (DEB-9982736) to J. Losos for funding that allowed us to complete this project. Support by the Environmental Awareness Group (EAG) of Antigua and Barbuda and the Forestry Unit (Ministry of Agriculture, Government of Antigua) was critical to the success of this project. The people of the village of Seatons, especially A. and R. Nicholas, helped us in innumerable ways. This is a scientific contribution of the Antiguan Racer Conservation Project, which is a partnership of the EAG of Antigua and Barbuda, the Forestry Unit, FFI, BHSU, the Durrell Wildlife Conservation Trust, and the Island Resources Foundation. Comments from L. Harmon, J. Losos, and Losos Lab members improved this manuscript and we thank O. Davis, N. Douglas, and B. Simmons for help in the field. LITERATURE CITED Crump, M. L., and N. J. Scott, Jr. 1994. Visual encounter surveys, p. 84 92. In: Measuring and Monitoring Biological Diversity: Standard Methods for Amphibians. W. R. Heyer, M. A. Donnelly, R. W. McDiarmid, L. A. C. Hayek,

Kolbe et al. Antiguan lizard community ecology 271 and M. S. Foster (eds.). Smithsonian Institution Press, Washington, D.C. Daltry, J. C., Q. Bloxam, G. Cooper, M. L. Day, J. Hartley, M. Henry, K. Lindsay, and B. E. Smith. 2001. Five years of conserving the world s rarest snake, the Antiguan racer Alsophis antiguae. Oryx 35:119 127. Davis, O. 2001. Selection of Alsophis antiguae reintroduction islands based on prey demography. Unpubl. Master s thesis, University of Derby, U.K. Eaton, J. M., S. C. Larimer, K. G. Howard, R. Powell, and J. S. Parmerlee, Jr. 2002. Population densities and ecological release of a solitary species: Anolis gingivinus on Anguilla, West Indies. Caribbean Journal of Science 38:27 36. Fobes, T. M., R. Powell, J. S. Parmerlee, Jr., A. Lathrop, and D. D. Smith. 1992. Natural history of Anolis cybotes (Sauria: Polychridae) from an altered habitat in Barahona, Dominican Republic. Caribbean Journal of Science 28:200 207. Gerber, G. L. 1999. A review of intraguild predation and cannibalism in Anolis, p. 28 39. In: Anolis Newletter. V. J. B. Losos and M. Leal (eds.). Washington University, St. Louis, Missouri. Gerber, G. L., and A. C. Echternacht. 2000. Evidence for asymmetrical intraguild predation between native and introduced Anolis lizards. Oecologia 124:599 607. Harris, B. R., D. T. Yorks, C. A. Bohnert, J. S. Parmerlee, Jr., and R. Powell. 2004. Population densities and structural habitats in lowland populations of Anolis lizards on Grenada. Caribbean Journal of Science 40:31 40. Herrel, A., R. Joachim, B. Vanhooydonck, and D. J. Irschick. 2006. Ecological consequences of ontogenetic changes in head shape and bite performance in the Jamaican lizard Anolis lineatopus. Biological Journal of the Linnean Society 89:443 454. Hertz, P. E. 1992. Temperature regulation in Puerto Rican Anolis lizards: a field test using null hypotheses. Ecology 73:1405 1417. Hodge, K. V. D., E. J. Censky, and R. Powell. 2003. The Reptiles and Amphibians of Anguilla, Bristish West Indies. Anguilla National Trust, Anguilla. Huey, R. B., and T. P. Webster. 1976. Thermal biology of Anolis lizards in a complex fauna: the cristatellus group on Puerto Rico. Ecology 57:985 994. Jenssen, T. A. 1973. Shift in the structural habitat of Anolis opalinus due to congeneric competition. Ecology 54: 863 869. Johnson, M. A., R. Kirby, S. Wang, and J. B. Losos. 2006. What drives variation in habitat use by Anolis lizards: habitat availability or selectivity? Canadian Journal of Zoology 84:877 886. Leal, M., J. A. Rodriguez-Robles, and J. B. Losos. 1998. An experimental study of interspecific interactions between two Puerto Rican Anolis lizards. Oecologia 117:273 278. Losos, J. B. 1994. Integrative approaches to evolutionary ecology: Anolis lizards as model systems. Annual Review of Ecology and Systematics 25:467 493. Losos, J. B., M. Leal, R. E. Glor, K. de Queiroz, P. E. Hertz, L. Rodríguez-Schettino, A. Chamizo Lara, T. R. Jackman, and A. Larson. 2003. Niche lability in the evolution of a Caribbean lizard community. Nature 424:542 545. Losos, J. B., J. C. Marks, and T. W. Schoener. 1993. Habitat use and ecological interactions of an introduced and a native species of Anolis lizard on Grand Cayman, with a review of the outcomes of anole introductions. Oecologia 95:525 532. Losos, J. B., T. W. Schoener, and D. A. Spiller. 2004. Predator-induced behaviour shifts and natural selection in field-experimental lizard populations. Nature 432: 505 508. Moermond, T. C. 1979a. Habitat constraints on the behavior, morphology, and community structure of Anolis lizards. Ecology 60:152 164. Moermond, T. C. 1979b. The influence of habitat structure on Anolis foraging behavior. Behaviour 70:147 167. Naganuma, K. H., and J. D. Roughgarden. 1990. Optimal body size in Lesser Antillean Anolis lizards a mechanistic approach. Ecological Monographs 60:239 256. Nicholson, K. L., S. M. Torrence, D. M. Ghioca, J. Bhattacharjee, A. E. Andrei, J. Owen, N. A. Radke, and G. Perry. 2005. The influence of temperature and humidity on activity patterns of the lizards Anolis stratulus and Ameiva exsul in the British Virgin Islands. Caribbean Journal of Science 41:870 873. Peters, R. H. 1983. The Ecological Implications of Body Size. Cambridge University Press, Cambridge, U.K. Pianka, E. R. 1973. The structure of lizard communities. Annual Review of Ecology and Systematics 4:53 74. Pianka, E. R., and L. J. Vitt. 2003. Lizards: Windows to the Evolution of Diversity. University of California Press, Berkeley, California. Powell, R., J. S. Parmerlee, Jr., and D. D. Smith. 1996. Evidence of spatial niche partitioning by a Hispaniolan community in a xeric habitat, p. 317 326. In: Contributions to West Indian Herpetology: A Tribute to Albert Schwartz. R. Powell and R. W. Henderson (eds.). Society for the Study of Amphibians and Reptiles, Ithaca, New York. Roughgarden, J. 1995. Anolis Lizards of the Caribbean: Ecology, Evolution, and Plate Tectonics. Oxford University Press, Oxford, U.K. Roughgarden, J., J. D. Rummel, and S. W. Pacala. 1983. Experimental evidence of strong present day competition between the Anolis populations of the Anguilla Bank a preliminary report, p. 499 506. In: Advances in Herpetology and Evolutionary Biology. A. Rhodin and K. Miyata (eds.). Museum of Comparative Zoology, Cambridge, Massachusetts. Rummel, J. D., and J. Roughgarden. 1985. Effects of reduced perch-height separation on competition between two Anolis lizards. Ecology 66:430 444. Schoener, T. W. 1968. The Anolis lizards of Bimini: resource partitioning in a complex fauna. Ecology 49:704 726. Schoener, T. W. 1974. Resource partitioning in ecological communities. Science 185:27 39. Schoener, T. W. 1975. Presence and absence of habitat shift in some widespread lizard species. Ecological Monographs 45:233 258. Schoener, T. W., D. A. Spiller, and J. B. Losos. 2002. Predation on a common Anolis lizard: Can the food-web effects of a devastating predator be reversed? Ecological Monographs 72:383 407. Schwartz, A., and R. W. Henderson. 1991. Amphibians and Reptiles of the West Indies: Descriptions, Distributions, and Natural History. University of Florida Press, Gainesville, Florida. Sexton, O. J., J. Bauman, and E. Ortleb. 1972. Seasonal food habits on Anolis limifrons. Ecology 53:182 186.

272 Copeia 2008, No. 2 Simmons, P. M., B. T. Greene, K. E. Williamson, R. Powell, and J. S. Parmerlee, Jr. 2005. Ecological interactions within a lizard community on Grenada. Herpetologica 61:124 134. Vitt, L. J., and C. M. Carvalho. 1995. Niche partitioning in a tropical wet season: lizards in the lavrado area of northern Brazil. Copeia 1995:305 329. Vitt, L. J., and G. R. Colli. 1994. Geographical ecology of a Neotropical lizard: Ameiva ameiva (Teiidae) in Brazil. Canadian Journal of Zoology 72:1986 2008. Vitt, L. J., S. S. Sartorius, T. C. S. Avila-Pires, M. C. Espósito, and D. B. Miles. 2000. Niche segregation among sympatric Amazonian teiid lizards. Oecologia 122:410 420. Vitt, L. J., and P. A. Zani. 1996. Organization of a taxonomically diverse lizard assemblage in Amazonian Ecuador. Candian Journal of Zoology 74:1313 1335. Vitt, L. J., and P. A. Zani. 1998a. Prey use among sympatric lizard species in lowland rain forest of Nicaragua. Journal of Tropical Ecology 14:537 559. Vitt, L. J., and P. A. Zani. 1998b. Ecological relationships among sympatric lizards in a transitional forest in the Northern Amazon of Brazil. Journal of Tropical Ecology 14:63 86. Vitt, L. J., P. A. Zani, and M. C. Esposito. 1999. Historical ecology of Amazonian lizards: implications for community ecology. Oikos 87:286 294.