EVOLUTION OF BODY SIZE IN THE MAP TURTLES AND SAWBACKS (EMYDIDAE: DEIROCHELYINAE: GRAPTEMYS)

Herpetologica, 64(1), 2008, 32 46 E 2008 by The Herpetologists League, Inc. EVOLUTION OF BODY SIZE IN THE MAP TURTLES AND SAWBACKS (EMYDIDAE: DEIROCHELYINAE: GRAPTEMYS) PETER V. LINDEMAN 1 Department of Biology and Health Services, 150 Cooper Hall, Edinboro University of Pennsylvania, Edinboro, PA 16444, USA ABSTRACT: Map turtles and sawbacks (Emydidae: Deirochelyinae: Graptemys) are a diverse group of turtles that are of ecological interest due to their diversity in trophic morphology, particularly in females, and their extreme sexual size dimorphism (with females larger). I used comparative analyses (independent contrasts in correlation analyses and GLM analyses) to examine hypotheses regarding the evolution of body size in the genus. Evolutionary changes in body size of both males and females were positively related to inferred shifts in latitude. Trophic morphology (relative head width, expressed either as a continuous or a discrete variable) was not an additive source of variation in male body size, but was for female body size, primarily due to the large body sizes exhibited by the megacephalic southern clade (G. pulchra, G. ernsti, G. gibbonsi, and G. barbouri). Status as allopatric or sympatric to other species of Graptemys was not an additive source of variation in female body size. These results argue against the hypothesis that character displacement of body size in females was important in the radiation of map turtle and sawback species and instead suggest a functional relationship whereby degree of molluscivory in females covaries with body size. Sexual size dimorphism was found to increase with body size of females, also due primarily to the large females of the megacephalic clade; this result means Graptemys is an exception to Rensch s Rule. In a latitude-corrected analysis of body size evolution in deirochelyine turtles, the exceptional degree of sexual size dimorphism in Graptemys appears to result more from reduction in male size than increase in female size. Decreased male size may indicate relaxed selection for large body size in Graptemys males as a consequence of the fact that they rarely leave the water for terrestrial excursions, because selection for larger male body size in other deirochelyines may be mediated by (a) predation pressure imposed by terrestrial predators, (b) enhancement of overland mobility for mate searching, and (c) the need for resistance to desiccation during overland excursions. Alternatively, male body size reduction in Graptemys may be explained by energetic requirements of searching for mates in the fluvial environment or unknown differences in social structure among deirochelyine turtles. Key words: Bergmann s rule; Body size; Character displacement; Comparative methods; Deirochelyinae; Emydidae; Evolution; Graptemys; Map turtles; Rensch s rule; Sawbacks; Sexual size dimorphism BODY SIZE covaries in important ways with several life-history parameters (Blanckenhorn, 2000; Blueweiss et al., 1978; Stearns, 1992). Selective pressures that may influence body size include ecological interactions with competitors and predators (Abrams and Rowe, 1996; Butler and Losos, 1997; Endler, 1986; Kozłowski, 1996; Radtkey et al., 1997). Sexual size dimorphism (SSD) is indicative of differential results of natural and sexual selection between the sexes. Large size in females may confer advantages via increased fecundity or larger offspring size and hence increased offspring fitness (Shine, 1988). Large size in males may confer advantages via male-male competition for mates (Fairbairn, 1997). Phylogenetically, shifts in SSD occur as selective pressures on the sexes change and males and females decrease, increase, or 1 CORRESPONDENCE: e-mail, plindeman@edinboro.edu remain stable in size (Hormiga et al., 2000). Rensch s Rule holds that sexual size dimorphism increases in magnitude with increased body size when males are the larger sex but decreases in magnitude with increasing body size when females are larger (Rensch, 1960). The leading hypothesis proposed to explain the rule concerns intrasexual selection among males competing for mates (Abouheif and Fairbairn, 1997; Fairbairn, 1997). Body size is an intriguing parameter in the map turtles and sawbacks (Emydidae: Deirochelyinae: Graptemys) for two reasons. First, map turtles and sawbacks have diverse trophic morphologies, with adult females falling into three groups comprised of four species each: microcephalic species with narrow heads, which feed mainly on nonmolluscan prey; mesocephalic species with moderately broad heads, which feed heavily on mollusks but generally supplement their diet with other 32

March 2008] HERPETOLOGICA 33 prey; and megacephalic species with exceptionally broad heads, which feed almost exclusively on mollusks (Lindeman, 2000). All microcephalic species live in sympatry with either meso- or megacephalic species, a pattern suggestive of character displacement and character assortment (Lindeman, 2000). Body size covaries with trophic morphology in a way that increases the difference in trophic morphology for co-occurring species, further suggesting character displacement for cooccurring species (Lindeman, 2000). Second, species of Graptemys have the most extreme sexual size dimorphism of any well-studied turtle clade, with mature females being up to 2.58 times longer in carapace length than conspecific adult males (Gibbons and Lovich, 1990). Sexual size dimorphism in ectothermic vertebrates results largely from bimaturism (Shine, 1990), a finding consistent with data on the genus Graptemys, in which males typically mature in their second or third years while females do not mature until their seventh to fifteenth years (Lindeman, 1999, 2005). In the present study, I used comparative analyses focused on four major objectives. The first two are directed at testing the hypotheses that changes in body size are adaptations to divergence in prey that have arisen via character displacement: (1) determination of whether changes in trophic morphology and diet during the radiation of Graptemys have been associated with changes in body size and (2) determination of whether status as allopatric or sympatric with congeners has influenced changes in body size. The third objective was to test for Graptemys and the remaining Deirochelyinae the prediction of Rensch s Rule that the magnitude of sexual size dimorphism should decrease with increased body size. The fourth objective was to determine whether change in body size in either males or females has played a disproportionate role in the exceptional SSD exhibited by Graptemys species. I also included an analysis of the confounding effect of latitude on body-size evolution, because many deirochelyine turtle species in the temperate zone have body sizes that are positively correlated with latitude (Cagle, 1954; Iverson and Smith, 1993; Lindeman, 1997; Moll, 1973; Seigel, 1980; Tucker et al., 1998). METHODS Data Sets and General Analytical Methods To represent body size, I used measurements of midline plastron length (PL; measured to the nearest mm with a plastic ruler) for 15 populations of 12 species of Graptemys, taken as part of earlier studies (Appendix; Lindeman, 2000; Lindeman and Sharkey, 2001). I included two populations of the wide-ranging species G. pseudogeographica and three of the wide-ranging species G. ouachitensis (including both putative subspecies for each) in order to include latitudinal variation in body size. Data used for a third wide-ranging species, G. geographica, were from a Wisconsin population while data for each of the other nine species, all with much more restricted geographic ranges (Lamb et al., 1994), came from various localities within their ranges. I expressed body size as logarithmic transformations of minimum, mean, and maximum PL of mature males (as recognized by their enlarged tail bases; Ernst et al., 1994) and mean and maximum PL of mature females. Maturity in female Graptemys is not evidenced by secondary sexual characteristics; hence, minimum size at maturity is not known for females of some species. To obtain means, I averaged PL measurements of all female specimens with a PL of at least 70% of the maximum PL recorded for their species or population. This figure is based on the relatively constant ratio of minimum female size to maximum female size for turtles in general (Shine and Iverson, 1995) and ratios reported for some species of Graptemys (Cagle, 1952; Horne et al., 2003; Jones and Hartfield, 1995; Killebrew and Porter, 1989; Lahanas, 1982; Lindeman, 1999, 2005; Porter, 1990; Timken, 1968). I recorded latitude to the nearest 0.5uN for sample populations of G. geographica (Wisconsin), G. o. ouachitensis (Wisconsin and Kentucky), and G. pseudogeographica (Wisconsin and Kentucky). For the other nine species and G. o. sabinensis, I used a midpoint latitude based on the range of their sites of specimen collection. Data on trophic morphology were coded discretely by assignment to three groups based on female head width (micro-, meso-, and megacephaly) in some analyses and continuously in others by using

34 HERPETOLOGICA [Vol. 64, No. 1 the ratio of predicted female head width at maximum recorded female PL (HW/PL), based on allometric regressions (Lindeman, 2000). I also recorded whether each species or population of Graptemys was sympatric or allopatric with respect to congeners. For analysis of Rensch s Rule (see below), I combined the above data on Graptemys with published data from populations of other species of deirochelyine emydids. I used only temperate-zone species to be able to include interspecific analysis of the positive correlation of body size with latitude that has been reported for some temperate-zone deirochelyines (Cagle, 1954; Iverson and Smith, 1993; Lindeman, 1997; Moll, 1973; Seigel, 1980; Tucker et al., 1998) and other North American freshwater turtles (Edmonds and Brooks, 1996; Iverson et al., 1997; Litzgus and Brooks, 1998; Tinkle, 1961). To the 15 data sets for Graptemys, I added 11 other deirochelyine taxa for which studies reported mean and maximum sizes for each sex plus minimum sizes for males (Appendix). I analyzed data by generating independent contrasts (Felsenstein, 1985) in COMPARE 4.4 (Martins, 2001). Regression analyses employing the general linear model (GLM) were conducted after importing contrasts into S-PLUS 6.1 (Insightful Corporation, 2001) and were constrained to have a zero intercept (Garland et al., 1992). The phylogenies and branch lengths I used were taken from a combined parsimony analysis of all molecular and morphological data for emydid turtles (Fig. 1; adapted from Stephens and Wiens, 2003, their Fig. 7). To separate two populations of G. o. ouachitensis, I added short branch lengths (1, compared to an average branch length of 23 for the rest of the tree). Tropical and subtropical species and species without data were pruned from the tree in the analysis of Rensch s Rule in the Deirochelyinae. Body Size, Trophic Morphology, and Sympatry/Allopatry Relationships I used a general linear models procedure to relate contrasts in body size to contrasts in latitude and trophic morphology (Garland et al., 1993). Contrasts in latitude were entered to account for the potentially confounding influence of latitude on body size. Contrasts in body size were represented alternately by male minimum, mean, or maximum PL or by female mean or maximum PL. In GLM models, contrasts in trophic morphology were represented as a categorical variable (microcephalic 5 0, mesocephalic 5 1, or megacephalic 5 2), with state changes based on parsimony, or as independent contrasts in a continuous variable (HW/PL). For analyses in which trophic morphology was entered as a categorical variable, there are relatively few transitions among the three states (Lindeman, 2000). Because the independent contrasts method transforms raw data to comparisons of adjacent nodes (Felsenstein, 1985), the result of having a low number of state changes in a phylogenetic ANCOVA is that numerous contrast values are zero and few have nonzero values (Garland et al., 1993), the latter representing speciation events at which one sister species is estimated to have changed states. To follow up analyses using categorical representations of trophic morphology, I explored the relationship of trophic morphology to female body size by combining categories in pairs: (a) microcephalic species vs. meso- + megacephalic species and (b) micro- + mesocephalic species vs. megacephalic species. I again used contrasts in latitude as a covariate. These analyses allowed me to compare the influence of transitions to and from the two derived states to determine the stronger influence upon body size evolution in females of the genus. I also expanded models based on latitude and the trophic morphology of females (coded categorically as micro-, meso-, or megacephaly) to include a third independent variable that denoted allopatry (0) or sympatry (1) with regard to congeners. Ancestral states for sympatry and allopatry were determined via parsimony for these analyses, with the ancestral state constrained to be allopatry. Models with a sympatry/allopatry variable allowed me to test the hypothesis that female body size has exhibited character displacement in response to the presence or absence of congeneric competitors (Lindeman, 2000). The above analyses were interpreted at an a level of 0.05 with a sequential Bonferroni

March 2008] HERPETOLOGICA 35 FIG. 1. Phylogeny used for analyses, modified from Stephens and Wiens (2003, their Fig. 7). Letters at nodes denote contrasts, with branches arranged such that in models relating body size to latitude, subtraction of each lower value from its corresponding upper value yields a positive contrast for latitude (see Fig. 2).

36 HERPETOLOGICA [Vol. 64, No. 1 adjustment made within any general linear model in which multiple independent variables yielded P, 0.05 (Rice, 1989). I did not make sequential Bonferroni adjustments to a across tests, in spite of the large number of tests, because I view these tests as alternative choices five measures of body size, three expressions of trophic morphology for examining the hypotheses that body-size evolution within Graptemys has been influenced by (a) trophic morphology (and its associated dietary ecology) and (b) status as sympatric or allopatric, above and beyond the confounding influence of latitude on body size. Among tests that differ by choice of body-size variable or coding of trophic morphology, consistency with regard to whether or not null hypotheses are rejected is thus of greater interest than the results of any one test. Rensch s Rule Rensch s Rule holds that in cases such as Graptemys in which females are larger, the magnitude of sexual size dimorphism tends to decrease with increasing body size (Fairbairn, 1997). If this rule holds for the radiation that produced extant species of Graptemys, I would expect positive allometry in a regression of log(male size) on log(female size), as indicated by a slope.1, with significance assessed by examining 95% confidence limits (CL) on slope. I tested this hypothesis using regression analysis of independent contrasts for (a) the 15 populations of Graptemys and (b) the expanded data set that also included the other 11 deirochelyine populations. Identifying Major Evolutionary Shifts in Deirochelyine Body Size To determine whether male or female evolutionary changes have been more responsible for the large magnitude of SSD in Graptemys, I examined changes in body size at all 25 nodes in the evolution of the temperate-zone deirochelyines for which I had data (15 data sets for Graptemys and 11 data sets for other species). I used a GLM model in which contrasts in minimum, mean, or maximum PL for one sex were related to two independent variables, contrasts in latitude and a categorical variable termed Node. I coded Node as 1 for the contrast representing a single node of interest and as 0 for all other nodes, with each of the 25 nodes in the phylogeny taking its turn serving as the node of interest. These analyses test the null hypothesis that the origin of a particular clade in the phylogeny is associated with a nonsignificant difference in magnitude of evolutionary change in body size. Because 25 tests were conducted on each set of body-size data, however, sequential Bonferroni adjustments to a would result in few if any significant results in such analyses (i.e., at least one P- value for the partial correlation of Node within a set of analyses would have to be,0.05/25 5 0.002). I therefore simply ranked P-values for nodes within each of the five sets of analyses (based on the five choices of bodysize variables) to determine the relative magnitude of evolutionary change at each node. Special emphasis is given below to the rankings of nodes relating to the origin and diversification of Graptemys and how these rankings compare among analyses using male and female body-size variables. Graphically, I show the results of these analyses in regression through the origin of contrasts in body size on contrasts in latitude, with the latter all calculated as positive values (Garland et al., 1992). Departure from the resulting regression lines represents the degree to which a contrast departs from the expected change in body size (Garland et al., 1992). RESULTS Body Size and Trophic Morphology Latitude was significantly positively related to body size in four of the six models relating contrasts in male body size to contrasts in relative head width (failing to be significant only for models employing minimum male PL) and in all four of the models relating contrasts in female body size to contrasts in relative head width (Table 1). Various expressions of relative head width failed to explain significant additional variation in the contrasts in male body sizes but were significant in all but one of the analyses involving contrasts in female body sizes (Table 1). I repeated the analyses that employed discrete character states for relative head width, combining pairs of categories to better

March 2008] HERPETOLOGICA 37 TABLE 1. Results (P-values relating to partial correlation) of general linear models relating contrasts in female body size in Graptemys to contrasts in latitude and relative head width. Relative head width was represented in models as a discrete variable (micro-, meso-, or megacephaly) with ancestral nodes determined by parsimony or as a continuous variable (predicted head width [HW] at maximum female plastron length [PL]). Asterisks denote significance following a sequential Bonferroni correction to a within each statistical model (see text). Relative head width Latitude Head width Males: Minimum PL at maturity Categorical 0.16 0.44 Continuous (HW/PL) 0.22 0.97 Males: Mean PL Categorical,0.0001* 0.19 Continuous (HW/PL),0.0001* 0.34 Males: Maximum PL Categorical 0.0001* 0.11 Continuous (HW/PL) 0.0009* 0.78 Females: Mean PL Categorical,0.0001* 0.045 Continuous (HW/PL),0.0001* 0.008* Females: Maximum PL Categorical,0.0001* 0.025* Continuous (HW/PL),0.0001* 0.013* identify the changes in trophic morphology associated with significant changes in female body size. When megacephaly and mesocephaly were combined as broad-headed (1) to compare against microcephaly, or narrowheaded (0), the most parsimonious scenario for character-state change had two contrasts scored as +1, indicating two transitions from the former state to the latter. The covariate latitude was a significant model term for both mean and maximum female PL (P, 0.0001) while the relative head width variable was nonsignificant in both models (P 5 0.45 for mean PL and P 5 0.30 for maximum PL). When mesocephaly and microcephaly were combined as narrow-headed (0) to compare against megacephaly, or broad-headed (1), the most parsimonious scenario had a single contrast scored as 21, indicating a single transition from the former state to the latter. The covariate latitude was once again a significant model term in both analyses (P, 0.0001), but in this case the relative head width variable was also significant in both models (P 5 0.017 for mean female PL and P 5 0.022 for maximum female PL). Hence the TABLE 2. Results (P-values relating to partial correlation) of general linear models relating contrasts in female body size in Graptemys to contrasts in latitude, relative head width (micro-, meso-, or megacephaly), and status of species as allopatric or sympatric with other species of Graptemys. Asterisks denote significance following sequential Bonferroni correction to a within each statistical model (see text). Reconstruction 1 Latitude Head width Allopatry/Sympatry Mean female PL A,0.0001* 0.073 0.85 B,0.0001* 0.073 0.84 C,0.0001* 0.094 0.30 D,0.0001* 0.094 0.29 Maximum female PL A,0.0001* 0.054 0.38 B,0.0001* 0.053 0.41 C,0.0001* 0.056 0.14 D,0.0001* 0.055 0.16 1 Reconstruction A has five transitions from allopatry to sympatry; B and C have four such transitions and one reversal to allopatry each; and D has three transitions to sympatry and two reversals. single instance of the evolution of megacephaly in Graptemys, in the common ancestor of a clade of four species (Clade T in Fig. 1: barbouri, ernsti, gibbonsi, and pulchra), appears to account for much of the influence relative head width has had upon female body size, as described by the above analyses. Sympatry and Allopatry Four equally most parsimonious reconstructions were analyzed, each with five changes in state (transitions to sympatry and reversals to allopatry). Latitude was a highly significant covariate (P, 0.0001) and female head width was a marginally nonsignificant covariate (0.05, P, 0.10) in all eight of the models assessed (Table 2). The allopatry/ sympatry variable was not significant in any of the models (all P $ 0.14). Rensch s Rule Contrasts in male PL were significantly related to contrasts in female PL within the radiation of Graptemys, for both means (MPL 5 0.710FPL, r 2 5 0.85, P, 0.001) and maximum values (MPL 5 0.560FPL, r 2 5 0.72, P 5 0.001). Instead of the positive allometry expected under Rensch s Rule, both slopes indicated significant negative allometry (95% CL on slope 0.534 0.885 for means and 0.349 0.772 for maximum values). In the

38 HERPETOLOGICA [Vol. 64, No. 1 TABLE 3. Rankings of five lowest P-values and P-values for three nodes of special interest, for the partial correlation of the term Node for 25 permutations of the model PL 5 Latitude*X1 + Node*X2. Each of 25 ancestral nodes in the deirochelyine phylogeny (Fig. 1) acted as the node of interest. Values underneath each entry in the table are P-values. Variable 1 st 2 nd 3 rd 4 th 5 th clade (Node G) Malaclemys + Graptemys Graptemys clade (Node H) Megacephalic Graptemys clade (Node S) Male PL min X O D M L 17 th 10 th 13 th 0.0016 0.0038 0.017 0.26 0.32 0.76 0.50 0.68 Male PL mean X O E F D 6 th 9 th 16 th 0.0032 0.087 0.093 0.11 0.20 0.21 0.36 0.64 Male PL max X G E C W 2 nd 13 th 20 th 0.012 0.059 0.062 0.17 0.22 0.059 0.49 0.88 Female PL mean S B F Y L 11 th 18 th 1 st 0.034 0.034 0.045 0.11 0.12 0.50 0.76 0.034 Female PL max B S W F Y 9 th 25 th 2 nd 0.040 0.055 0.12 0.13 0.15 0.44 0.97 0.055 Deirochelyinae overall, the relationship of contrasts in male PL with contrasts in female PL was significant, with slopes in both analyses indicating nonsignificant negative allometry (Mean values: MPL 5 0.847FPL, r 2 5 0.62, P, 0.001, 95% CL 0.567 1.127; Maximum values: MPL 5 0.849FPL, r 2 5 0.58, P, 0.001, 95% CL 0.543 1.154). Evolutionary Shifts in Deirochelyine Body Size In 125 phylogenetic ANCOVAs assessing the magnitude of change in body size at individual nodes in the deirochelyine phylogeny (i.e., 25 contrasts 3 5 choices of PL variable), the model term Latitude was positively associated with body size change in all 125 and was a significant model variable (P, 0.05 for partial correlation) in every analysis except one involving minimum mature PL of males. In Table 3, I list the contrasts producing the five smallest P-values associated with the model term Node for each body-size variable, as well as the rankings of P- values for three contrasts of special interest: the origin of the Malaclemys/Graptemys clade, the origin of Graptemys, and the origin of the megacephalic clade of Graptemys (G. barbouri, G. ernsti, G. gibbonsi, and G. pulchra). For males, the small body size of Pseudemys texana relative to two congeners caused its contrast (X) to be most divergent from the regression line for contrasts in minimum, mean, and maximum PL regressed on contrasts in latitude (Fig. 2A C). Other highly divergent contrasts were not consistent across the three analyses. Relative to their sister taxa, the ancestors of the Malaclemys + Graptemys clade (contrast G) and of Graptemys (contrast H) were shown to exhibit moderate to strong size reduction in males, after correction for latitudinal change. The placement of contrasts associated with the origin of the megacephalic clade (contrast S) suggested that the ancestor of the clade exhibited a slight increase in male body size, after correction for latitudinal change. In both analyses for females, four data points were most divergent from the regression lines: (B) the contrast in the small body size in the genus Chrysemys versus the larger body sizes typical of its sister group, a clade of four genera; (S) a contrast within the Graptemys radiation, associated with the large body sizes of the megacephalic clade; (F) the contrast representing the divergence between the larger body sizes of Pseudemys relative to their sister clade, the Trachemys + Malaclemys + Graptemys clade; and (Y) the contrast representing the divergence between the smaller, more northern P. rubriventris and the larger, more southern P. concinna. Origin of the Malaclemys + Graptemys clade (G) was associated with a moderately strong decrease in body size, while origin of the Graptemys clade (H) was associated with a comparatively slighter increase in body size (both after correction for latitudinal change). One other contrast within Graptemys, between a smaller-bodied clade of five microcephalic species and two populations of the

March 2008] HERPETOLOGICA 39 FIG. 2. Relationships of the contrast in body size with the contrast in latitude (un) for deirochelyine turtles. Letters correspond to labels at nodes in Fig. 1. Regressions are constrained to have zero intercepts. larger-bodied mesocephalic species, G. pseudogeographica (L), also diverged substantially from both regression lines. It is apparent that evolutionary changes in male and female body sizes have been highly correlated among the temperate-zone deirochelyines. Residuals from the two regressions of contrasts in mean PL on contrasts in latitude (Fig. 2B and 2D) were significantly positively correlated (r 5 0.433, P 5 0.0237),

40 HERPETOLOGICA [Vol. 64, No. 1 as were the residuals from the two regressions of contrasts in maximum PL on contrasts in latitude (Fig. 2C and 2E; r 5 0.527, P 5 0.0045). DISCUSSION For a variety of phylogenetically-corrected linear models, latitude was a strongly consistent, significant influence on body size evolution, both within Graptemys and for the temperate-zone Deirochelyinae as a whole. These results demonstrate that increased body size at northern latitudes, previously regarded as an intraspecific phenomenon in the deirochelyine genera Chrysemys, Trachemys, and Malaclemys (Cagle, 1954; Iverson and Smith, 1993; Lindeman, 1997; Moll, 1973; Seigel, 1980; Tucker et al., 1998) as well as in widely distributed freshwater species of other turtle clades in North America (Edmonds and Brooks, 1996; Iverson et al., 1997; Litzgus and Brooks, 1998; Tinkle, 1961) is also an interspecific phenomenon within the deirochelyines. The phenomenon of increased body size at more northerly latitudes within a species is known as Bergmann s Rule and was originally proposed for endotherms, but also holds for many ectotherms (including most turtles; Ashton and Feldman, 2003). Interestingly, the rule was originally formulated for genera but subsequently was investigated primarily on an intraspecific basis (reviewed in de Queiroz and Ashton, 2004). The present study supports the conclusion of of de Queiroz and Ashton (2004) that the tendency to conform to Bergmann s Rule can show phylogenetic signal. Iverson and Smith (1993) showed that latitude is a convenient surrogate for climatic variables and reviewed five hypotheses that might explain the adaptive nature of a latitudinal increase in body size for the painted turtle, Chrysemys picta. Larger body size in the north may be thermally adaptive, by (1) promoting retention of body heat in cooler climates, (2) allowing for greater storage of energy for longer winters, or (3) allowing deeper nests to be dug to enhance survival of overwintering hatchlings. Alternatively, larger body size in the north may result from (4) competitive release, as turtle diversity in North America declines with latitude, or it may be necessary for (5) production of larger clutches that compensate for increased mortality of eggs or juveniles at northern latitudes. These hypotheses may apply to other deirochelyines, including species of Graptemys (although only the basal species in the genus, G. geographica, is known to have hatchlings that overwinter in the nest; Baker et al., 2003; Nagle et al., 2004). The widespread conformity to Bergmann s Rule among endothermic and ectothermic tetrapods is thought to be an argument in favor of the hypothesis that larger body size is adaptive in more strongly seasonal environments by allowing for greater storage of energy and nutrients, but other advantages may also occur additively (de Queiroz and Ashton, 2004). Within each sex, results of analyses were relatively consistent regardless of whether minimum, mean, or maximum adult body size was used. Minimum and maximum sizes may be biased by inadequate sampling. In addition, bias in minimum body sizes may occur due to intraspecific variation in body size at maturity, while bias in mean and maximum body sizes may occur due to differences among species in adult survivorship rates or in capture techniques used to collect the specimens used in the present study. Consistency in results suggests at best a weak influence of the biases that may occur with each choice of body size variable. The hypothesis that body-size evolution within Graptemys has been driven by character displacement, as a means of increasing the divergence in trophic morphology that typifies coexisting species (Lindeman, 2000), is not supported by the present analyses. Models incorporating a sympatry/allopatry variable did not significantly improve the description of evolutionary changes in body size. Instead, differences in the magnitude of body-size shifts were detected in association with origins of megacephaly, mesocephaly, and microcephaly in females, with the difference associated with the origin of megacephaly being the most dramatic; shifts between sympatry or allopatry did not help to explain residual variation. Similar differences in the magnitude of body-size changes were not found to be associated with origins of the three states among males, whose separation

March 2008] HERPETOLOGICA 41 based on head and alveolar width is not nearly as evident as it is for females (Lindeman, 2000). Taken together, these results suggest a rather simple primary result: while latitude influences body size in Graptemys, the four species with megacephalic females have substantially larger adult female body sizes than their southern distributions would suggest. It is likely that trophic adaptation to nearly exclusive feeding on large, hard-shelled molluscan prey has driven this increase in body size. Females of these species may require not only the hypertrophied head width and wide alveolar surfaces that adapt them to crushing their prey (Lindeman, 2000; Lindeman and Sharkey, 2001), but large body size as well, as it is likely that such large heads would scarcely be biomechanically functional if they were attached to smaller bodies. The lack of support for character displacement of body size in Graptemys is not surprising, given the low level of support for character displacement for trophic morphology in a previous analysis. Mapping trophic morphology characters on the phylogeny of the genus showed few changes in character states (Lindeman, 2000), a situation not dramatically changed under a more recent phylogeny (Stephens and Wiens, 2003). To the extent that interspecific competition may have played a role in the radiation of these turtles it has probably acted primarily by character assortment (sensu Losos, 1990), following character-state changes, with at most two incidents of character displacement having produced character-state changes early in the radiation of Graptemys (Lindeman, 2000). Neither the genus Graptemys nor temperate-zone deirochelyine turtles conformed to Rensch s Rule. The predicted pattern would be positive allometry in regression of contrasts in male size on contrasts in female size (Fairbairn, 1997), but significantly negative allometry was the result of the analysis within Graptemys, with the deirochelyine data also trending toward negative allometry, albeit not in a manner significantly different from isometry. Rensch s Rule has previously been found to hold, after consideration of phylogeny, for all turtles and separately for the kinosternid turtles (Fairbairn, 1997). Exceptions to the rule are found most commonly, however, among taxa in which most species have larger females (Abouheif and Fairbairn, 1997; Fairbairn, 1997), as is the universal pattern among Graptemys and other deirochelyines. As a group, turtles contain a minority of species in which males are larger, with most kinosternids having only slight size dimorphism (Gibbons and Lovich, 1990). The most promising explanation for Rensch s Rule concerns male-male combat (Abouheif and Fairbairn, 1997; Fairbairn, 1997), a phenomenon that occurs in some turtles but which is unknown (and unlikely, given smaller male body sizes) in Graptemys and other deirochelyines (Berry and Shine, 1980). Its absence among the deirochelyines may therefore explain the status of these turtles as exceptions to the rule. The exceptional degree of sexual size dimorphism that distinguishes Graptemys from other turtles (Gibbons and Lovich, 1990) appears to be more a result of changes in male body size than changes in female body size. The Malaclemys + Graptemys clade shows size reduction in both sexes relative to what would be expected based solely on latitudinal effects within the Deirochelyinae; however, the divergence from expectation was considerably more dramatic for mean and maximum sizes of males than for females, thus partially explaining the high degree of sexual size dimorphism shared by these two sister genera. The origin of the Graptemys clade is typified by a small increase in female size and a more substantial decrease in male size, explaining the greater degree of size dimorphism in Graptemys than in Malaclemys (Gibbons and Lovich, 1990). Why male sizes have shown so much size reduction is unclear, but hypotheses include the influence of habitat on predation pressures and mate-searching behavior as well as potential differences in social systems relative to other deirochelyine turtles. In other deirochelyine species in which male-male combat for mating opportunities is unknown, it has been suggested that males mature upon growing to a size that is relatively resistant to predators (e.g., in Trachemys scripta; Gibbons et al., 1981, and Chrysemys picta, Frazer et

42 HERPETOLOGICA [Vol. 64, No. 1 al., 1993). Overland movements among lentic habitats are relatively common in males of Trachemys, Chrysemys, and some Pseudemys and expose them to a variety of primarily mammalian predators (Aresco, 2005; Gibbons et al., 1983, 1990; Lindeman and Rabe, 1990; Sexton, 1959; Wilbur, 1975). Conversely, male Graptemys are rarely found on land away from their aquatic habitats (Bodie and Semlitsch, 2000; Craig, 1992; Flaherty, 1982; Jones, 1996; Sanderson, 1974; Steen et al., 2006), which are predominantly large to medium rivers (Ernst et al., 1994). While many freshwater fish would seem to be potential predators of small Graptemys in the aquatic environment, experiments suggest that even live turtles as small as hatchlings (including Graptemys) are resistant to fish predation due to noxious behaviors (Britson, 1998; Britson and Gutzke, 1993; Semlitsch and Gibbons, 1989). While American alligators (Alligator mississippiensis) and alligator snapping turtles (Macrochelys temminkii) are significant aquatic predators of freshwater turtles in the southeastern United States (Delany and Abercrombie, 1986; Elsey, 2006; Iverson and Hudson, 2005; Sloan et al., 1996; Taylor, 1986; Valentine et al., 1972), only the latter has been recorded to eat Graptemys (and only one confirmed specimen; Elsey, 2006), perhaps due to the two predators preferences for more lentic waters. Therefore, strongly aquatic male Graptemys may be relatively released from predation pressures compared to the males of other deirochelyine species, allowing them to reap the fitness advantage of rapid maturity at smaller body sizes. Gibbons and Lovich (1990) suggested that increased handling time of turtles is possible for predators in a terrestrial setting and that this would explain reduced sexual size dimorphism in more fully terrestrial turtle species; perhaps the same applies to aquatic deirochelyines that are more prone to terrestrial movements. They also suggested that the terrestrial setting may select for larger body size to travel overland more effectively and to avoid desiccation, selective agents that might also contribute to size differences among male deirochelyines based on their terrestrial tendencies. Although they occupy the same habitats, female Graptemys may be constrained from maturing at smaller sizes because of the effect of body size on both egg and clutch size (Lindeman, 2005; Ryan and Lindeman, 2007) and possibly also because of potential exposure to mammalian predators during terrestrial nesting excursions. Alternatively, male Graptemys primarily inhabit flowing rivers and thus swim in strong currents when searching for mates. Smaller body size may provide energetic advantages related to male mobility (although it should be noted that males are unable to swim as fast as females in current; Pluto and Bellis, 1986). This scenario would be an example of sexual selection against large body size in males (sensu Blanckenhorn, 2000). Finally, there may be as-yet undiscovered differences among deirochelyine genera in mating systems which would confer more advantage to large size on males of other genera and less on males of species of Graptemys. The correlation between residual contrasts in female and male body sizes after correction for latitudinal changes is an interesting phenomenon. Intraspecific genetic correlation between male and female body sizes is a key component of hypotheses that seek to explain interspecific allometry in sexual dimorphism (Abouheif and Fairbairn, 1997; Fairbairn, 1997). Genetic models suggest that such correlation will occur while selection is producing changes in body size, but that the correlation will disappear at equilibrium (Lande, 1980). While Graptemys and other deirochelyines do not conform to Rensch s Rule, they do show evidence of genetic correlation of body size between males and females, possibly as a result of continuing selective pressures on body size. Acknowledgments. I thank J. Iverson and P. Stephens for their many helpful comments on the manuscript. LITERATURE CITED ABOUHEIF, E., AND D. J. FAIRBAIRN. 1997. A comparative analysis of allometry for sexual size dimorphism: assessing Rensch s Rule. American Naturalist 149:540 562. ABRAMS, P. A., AND L. ROWE. 1996. The effects of predation on the age and size of maturity of prey. Evolution 50:1052 1061. ARESCO, M. J. 2005. Mitigation measures to reduce highway mortality of turtles and other herpetofauna at

March 2008] HERPETOLOGICA 43 a north Florida lake. Journal of Wildlife Management 69:549 560. ASHTON, K. G., AND C. R. FELDMAN. 2003. Bergmann s rule in nonavian reptiles: Turtles follow it, lizards and snakes reverse it. Evolution 57:1151 1163. BAKER, P. J., J. P. COSTANZO, J.B.IVERSON, AND R. E. LEE, JR. 2003. Adaptations to terrestrial overwintering of hatchling northern map turtles, Graptemys geographica. Journal of Comparative Physiology B 173:643 651. BERRY, J. F., AND R. SHINE. 1980. Sexual size dimorphism and sexual selection in turtles (order Testudines). Oecologia 44:185 191. BLANCKENHORN, W. U. 2000. The evolution of body size: what keeps organisms small? Quarterly Review of Biology 75:385 407. BLUEWEISS, L., H. FOX, V. KUDZMA, D. NAKASHIMA, R. PETERS, AND S. SAMS. 1978. Relationships between body size and some life history parameters. Oecologia 37:257 272. BODIE, J. R., AND R. D. SEMLITSCH. 2000. Spatial and temporal use of floodplain habitats by lentic and lotic species of aquatic turtles. Oecologia 122:138 146. BRITSON, C. A. 1998. Predatory responses of largemouth bass (Micropterus salmoides) to conspicuous and cryptic hatchling turtles: a comparative experiment. Copeia 1998:383 390. BRITSON, C. A., AND W. H. N. GUTZKE. 1993. Antipredator mechanisms of hatchling freshwater turtles. Copeia 1993:435 440. BUTLER, M. A., AND J. B. LOSOS. 1997. Testing for unequal amounts of evolution in a continuous character on different branches of a phylogenetic tree using linear and squared-change parsimony: an example using Lesser Antilles Anolis lizards. Evolution 51:1623 1635. CAGLE, F. R. 1952. The status of the turtles Graptemys pulchra Baur and Graptemys barbouri Carr and Marchand, with notes on their natural history. Copeia 1952:223 234. CAGLE, F. R. 1954. Observations on the life cycles of painted turtles (genus Chrysemys). American Midland Naturalist 52:225 235. CRAIG, M. J. 1992. Radio-telemetry and tagging study of movement patterns, activity cycles, and habitat utilization in Cagle s map turtle, Graptemys caglei. M.S. Thesis, West Texas State University, Canyon, Texas, U.S.A. DELANY, M. F., AND C. L. ABERCROMBIE. 1986. American alligator food habits in northcentral Florida. Journal of Wildlife Management 5:348 353. DE QUEIROZ, A., AND K. G. ASHTON. 2004. The phylogeny of a species-level tendency: Species heritability and possible deep origins of Bergmann s rule in tetrapods. Evolution 58:1674 1684. EDMONDS, J. H., AND R. J. BROOKS. 1996. Demography, sex ratio, and sexual size dimorphism in a northern population of common musk turtles (Sternotherus odoratus). Canadian Journal of Zoology 74:918 925. ELSEY, R. M. 2006. Food habits of Macrochelys temminckii (alligator snapping turtle) from Arkansas and Louisiana. Southeastern Naturalist 5:443 452. ENDLER, J. A. 1986. Natural Selection in the Wild. Princeton University Press, Princeton, New Jersey, U.S.A. ERNST, C. H., J. E. LOVICH, AND R. W. BARBOUR. 1994. Turtles of the United States and Canada. Smithsonian Institution Press, Washington, D.C., U.S.A. FAIRBAIRN, D. J. 1997. Allometry for sexual size dimorphism: pattern and process in the coevolution of body size in males and females. Annual Review of Ecology and Systematics 28:659 687. FELSENSTEIN, J. 1985. Phylogenies and the comparative method. American Naturalist 125:1 15. FLAHERTY, N. C. 1982. Home range, movement, and habitat selection in a population of map turtle, Graptemys geographica (Le Sueur), in southwestern Quebec. M.S. Thesis, McGill University, Montreal, Quebec, Canada. FRAZER, N. B., J. L. GREENE, AND J. W. GIBBONS. 1993. Temporal variation in growth rate and age at maturity of male painted turtles, Chrysemys picta. American Midland Naturalist 130:314 324. GARLAND, T., JR.., P. H. HARVEY, AND A. R. IVES. 1992. Procedures for the analysis of comparative data using phylogenetically independent contrasts. Systematic Biology 41:18 32. GARLAND, T., JR.., A. W. DICKERMAN, C.M.JANIS, AND J. A. JONES. 1993. Phylogenetic analysis of covariance by computer simulation. Systematic Biology 42:265 292. GIBBONS, J. W. 1969. Ecology and population dynamics of the chicken turtle, Deirochelys reticularia. Copeia 1969:669 676. GIBBONS, J. W., AND J. L. GREENE. 1978. Selected aspects of the ecology of the chicken turtle, Deirochelys reticularia (Latreille) (Reptilia, Testudines, Emydidae). Journal of Herpetology 12:237 241. GIBBONS, J. W., AND J. E. LOVICH. 1990. Sexual dimorphism in turtles with emphasis on the slider turtle (Trachemys scripta). Herpetological Monographs 4:1 29. GIBBONS, J. W., R. D. SEMLITSCH, J.L.GREENE, AND J. P. SCHUBAUER. 1981. Variation in age and size at maturity of the slider turtle (Pseudemys scripta). American Naturalist 117:841 845. GIBBONS, J. W., J. L. GREENE, AND J. D. CONGDON. 1983. Drought-related responses of aquatic turtle populations. Journal of Herpetology 17:242 246. GIBBONS, J. W., J. L. GREENE, AND J. D. CONGDON. 1990. Temporal and spatial movement patterns of sliders and other turtles. Pp. 210 215. In J. W. Gibbons (Ed.), Life History and Ecology of the Slider Turtle. Smithsonian Institution Press, Washington, D.C., U.S.A. HORMIGA, G., N. SCHARFF, AND J. A. CODDINGTON. 2000. The phylogenetic basis of sexual size dimorphism in orb-weaving spiders (Araneae, Orbiculariae). Systematic Biology 49:435 462. HORNE, B. D., R. J. BRAUMAN, M.J.C.MOORE, AND R. A. SEIGEL. 2003. Reproductive and nesting ecology of the yellow-blotched map turtle, Graptemys flavimaculata: implications for conservation and management. Copeia 2003:729 738. INSIGHTFUL CORPORATION. 2001, S-PLUS 6 for Windows User s Guide. Insightful Corporation, Seattle, Washington, U.S.A. IVERSON, J. B., AND R. HUDSON. 2005. Macrochelys temminckii (alligator snapping turtle). Diet. Herpetological Review 36:312 313.

44 HERPETOLOGICA [Vol. 64, No. 1 IVERSON, J. B., AND G. R. SMITH. 1993. Reproductive ecology of the painted turtle (Chrysemys picta) in the Nebraska sandhills and across its range. Copeia 1993:1 21. IVERSON, J. B., H. HIGGINS, A. SIRULNIK, AND C. GRIFFITHS. 1997. Local and geographic variation in the reproductive biology of the snapping turtle (Chelydra serpentina). Herpetologica 53:96 117. JONES, R. L. 1996. Home range and seasonal movements of the turtle Graptemys flavimaculata. Journal of Herpetology 30:376 385. JONES, R. L., AND P. D. HARTFIELD. 1995. Population size and growth in the turtle Graptemys oculifera. Journal of Herpetology 29:426 436. KILLEBREW, F. C., AND D. PORTER. 1989. Graptemys caglei (Cagle s map turtle). Size maxima. Herpetological Review 20:70. KOZłOWSKI, J. 1996. Optimal initial size and adult size of animals: Consequences for macroevolution and community structure. American Naturalist 147:101 114. LAHANAS, P. N. 1982. Aspects of the life history of the southern black-knobbed sawback, Graptemys nigrinoda delticola Folkerts and Mount. M.S. Thesis, Auburn University, Auburn, Alabama, U.S.A. LAMB, T., C. LYDEARD, R. B. WALKER, AND J. W. GIBBONS. 1994. Molecular systematics of map turtles (Graptemys): A comparison of mitochondrial restriction site versus sequence data. Systematic Biology 43:543 559. LANDE, R. 1980. Sexual dimorphism, sexual selection and adaptation in polygenic characters. Evolution 34:292 307. LINDEMAN, P. V. 1996. Comparative life history of painted turtles (Chrysemys picta) in two habitats in the inland Pacific Northwest. Copeia 1996:114 130. LINDEMAN, P. V. 1997. Does life-history variation in the turtle Chrysemys picta have a subspecific component? Journal of Herpetology 31:55 61. LINDEMAN, P. V. 1999. Growth curves for Graptemys, with a comparison to other emydid turtles. American Midland Naturalist 142:141 151. LINDEMAN, P. V. 2000. Evolution of the relative width of the head and alveolar surfaces in map turtles (Testudines: Emydidae: Graptemys). Biological Journal of the Linnean Society 69:549 576. LINDEMAN, P. V. 2005. Aspects of the life history of the Texas map turtle (Graptemys versa). American Midland Naturalist 153:378 388. LINDEMAN, P. V. 2007. Diet, growth, body size, and reproductive potential of the Texas river cooter (Pseudemys texana) in the South Llano River, Texas. Southwestern Naturalist 52:586 594. LINDEMAN, P. V., AND F. W. RABE. 1990. Effect of drought on the western painted turtle, Chrysemys picta belli, in a small wetland ecosystem. Journal of Freshwater Ecology 5:359 364. LINDEMAN, P. V., AND M. J. SHARKEY. 2001. Comparative analyses of functional relationships in the evolution of trophic morphology in the map turtles (Emydidae: Graptemys). Herpetologica 57:313 318. LITZGUS, J. D., AND R. J. BROOKS. 1998. Reproduction in a northern population of Clemmys guttata. Journal of Herpetology 32:252 259. LOSOS, J. B. 1990. A phylogenetic analysis of character displacement in Caribbean Anolis lizards. Evolution 44:558 569. LOVICH, J. E., AND J. W. GIBBONS. 1990. Age at maturity influences adult sex ratio in the turtle Malaclemys terrapin. Oikos 59:126 134. MARTINS, E. P. 2001. COMPARE. Version 4.4. Computer programs for the statistical analysis of comparative data. Available at http://compare.bio.indiana.edu/. Department of Biology, Indiana University, Bloomington, Indiana, U.S.A. MITCHELL, J. C. 1985a. Variation in the male reproductive cycle in a population of painted turtles, Chrysemys picta, from Virginia. Herpetologica 41:45 51. MITCHELL, J. C. 1985b. Female reproductive cycle and life history attributes in a Virginia population of painted turtles, Chrysemys picta. Journal of Herpetology 19: 218 226. MOLL, E. O. 1973. Latitudinal and intersubspecific variation in reproduction of the painted turtle, Chrysemys picta. Herpetologica 29:307 318. NAGLE, R. D., C. L. LUTZ, AND A. L. PYLE. 2004. Overwintering in the nest by hatchling map turtles (Graptemys geographica). Canadian Journal of Zoology 82:1211 1218. PLUTO, T. G., AND E. D. BELLIS. 1986. Habitat utilization by the turtle, Graptemys geographica, along a river. Journal of Herpetology 20:22 31. PORTER, D. A. 1990. Feeding ecology of Graptemys caglei Haynes and McKown in the Guadalupe River, Dewitt County, Texas. M.S. Thesis, West Texas State University, Canyon, Texas, U.S.A. RADTKEY, R. R., S. M. FALLON, AND T. J. CASE. 1997. Character displacement in some Cnemidophorus lizards revisited: a phylogenetic analysis. Proceedings of the National Academy of Sciences U.S.A. 94:9740 9745. RENSCH, B. 1960. Evolution Above the Species Level. Columbia University Press, New York, New York, U.S.A. RICE, W. R. 1989. Analyzing tables of statistical tests. Evolution 43:223 225. RYAN, K. M., AND P. V. LINDEMAN. 2007. Reproductive allometry in the common map turtle, Graptemys geographica. American Midland Naturalist 158:49 59. SANDERSON, R. A. 1974. Sexual dimorphism in the Barbour s map turtle, Malaclemys barbouri (Carr and Marchand). M.A. Thesis, University of South Florida, Tampa, Florida, U.S.A.. SEIGEL, R. A. 1980. Nesting habits of diamondback terrapin (Malaclemys terrapin) on the Atlantic Coast of Florida. Transactions of the Kansas Academy of Science 83:239 246. SEMLITSCH, R. D., AND J. W. GIBBONS. 1989. Lack of largemouth bass predation on hatchling turtles (Trachemys scripta). Copeia 1989:1030 1031. SEXTON, O. J. 1959. Spatial and temporal movements of a population of the painted turtle, Chrysemys picta marginata (Agassiz). Ecological Monographs 29:113 140. SHINE, R. 1988. The evolution of large body size in females: a critique of Darwin s fecundity advantage model. American Naturalist 131:124 131. SHINE, R. 1990. Proximate determinants of sexual difference in adult body size. American Naturalist 135:278 283.