Evidence of Pelvic and Nonpelvic Constraint on Egg Size in Two Species of Kinosternon from Mexico
Abstract
Optimal egg size theory predicts that natural selection optimizes egg size within populations and most of the variation in reproductive output is attributable to clutch size variation driven by body size, available resources, and age. For small-bodied turtles, morphological (pelvic) constraint on egg size has been considered the main explanation when populations exhibit considerable variation in egg size, because the pelvis could be under selection for other functions besides reproduction. Kinosternids, a small-bodied and semiterrestrial lineage of turtles, show evidence for both pelvic and nonpelvic constraint on egg size. In order to test if small species show a tendency toward pelvic constraint on egg size, we examined possible pelvic constraints in 1 population of the small-bodied Kinosternon chimalhuaca from western Mexico and in 3 populations of medium- to large-bodied Kinosternon integrum from central Mexico. Gravid females were X-rayed to measure both pelvic aperture and egg width. To test for pelvic constraint on each population we compared the slopes of pelvic aperture and egg width (mean and maximum) to body size (plastron length) with analysis of covariance (ANCOVA); we also compared egg elongation with fresh egg measurements among populations using ANCOVA (with body size as covariate), and we conducted an allometric analysis in order to test for egg size optimization. We found evidence of pelvic constraint in 1 population of K. integrum, and evidence of nonpelvic constraint in the K. chimalhuaca population and in the other 2 populations of K. integrum. Our data did not support the supposed adaptive compromise and pelvic constraints on egg size in small-bodied turtles reported in other studies. Environmental factors such local pressure on egg size (stability of the environment) could explain this pelvic constraint discrepancy in kinosternids.
Energy allocated to reproductive output is driven by maternal effects (body size) and environmental factors that impact fitness (Stearns 1977, 1992; Roff 2002). Variation in life-history traits such as offspring size and number are determined by many different abiotic and biotic factors and the trade-off between producing more or bigger offspring (Iverson 1992; Iverson and Smith 1993; Hofmeyr et al. 2005; Wilkinson and Gibbons 2005). Under this scenario the fitness of the mother and her offspring would increase by producing large (and more viable) offspring when environmental conditions are harsh and unpredictable, and by producing more, but not larger, offspring when environmental conditions are stable and predictable (if mother body size allows it) (Smith and Fretwell 1974; Iverson et al. 1993; Ryan and Lindeman 2007; Rollinson and Brooks 2008).
The morphological constraint hypothesis suggests that anatomical factors, such as the mother's pelvic aperture (Tucker et al. 1978; Congdon and Gibbons 1987; Wilkinson and Gibbons 2005) or the caudal gap (Clark et al. 2001) may constrain egg size and influence the balance between offspring size and offspring number (Smith and Fretwell 1974), so that small mothers produce smaller eggs than would be optimal. Furthermore, as proposed by Ryan and Lindeman (2007) and confirmed by Macip-Ríos et al. (2012), to increase their fitness those small females should produce the greatest number of the largest eggs that they are physically capable of producing.
Egg constraint by the pelvic aperture in reptiles is taxon-specific. In lizards and turtles it has been considered a by-product of other uses of the pelvis (i.e., support and locomotion; Congdon and Gibbons 1987; Sinervo and Licht 1991; Lovich et al. 2012), although this is not a problem for snakes (Ji et al. 2009). If the pelvis is under selection for other functions besides reproduction (such as limb retraction or suspension of body mass during terrestrial locomotion), a supposed “adaptive compromise” between reproduction and locomotion should be found (Congdon and Gibbons 1987). This compromise should be more stringent in terrestrial and semiterrestrial turtles than in fully aquatic species (Sinervo and Licht 1991). Congdon and Gibbons (1987) suggested that pelvic constraint on egg size should be more common in small-sized species or populations with small body size, since in small and semiterrestrial turtle species the forces of selection may favor locomotion, walking stability, and limb retraction at the expense of optimal egg size; however, this trend should be sex-biased, with females having a wider pelvic girdle than males in order to achieve an optimum egg size (Long and Rose 1989).
Evidence for nonpelvic constraint on egg size has been reported for large freshwater turtles such as Trachemys scripta (carapace length [CL] = approximately 210–220 mm; Congdon and Gibbons 1987), and Pseudemys floridana (CL = 235 mm; Wilkinson and Gibbons 2005), and also for smaller turtles such as Sternotherus odoratus (CL = 78 mm; Wilkinson and Gibbons 2005), Sternotherus carinatus (CL = 117 mm; Iverson 2002), Kinosternon flavescens (CL = 102 mm; Iverson 1991), Kinosternon sonoriense (CL = 130 mm; van Loben Sels et al. 1997; CL = 122 mm; Lovich et al. 2012), and Mauremys leprosa (CL = 180 mm; Lovich et al. 2010; Naimi et al. 2012). On the other hand, evidence for pelvic constraint has been reported in small species such as Deirochelys reticularia (CL = 160 mm; Congdon et al. 1983; Congdon and Gibbons 1987; but see Iverson and Smith 1993 for Chrysemys picta bellii), Chrysemys picta marginata (CL = 125–135 mm; Congdon and Gibbons 1987), Kinosternon subrubrum (CL = 87 mm; Wilkinson and Gibbons 2005), and the extreme and so far unique case of Homopus signatus (CL = 92.6 mm; Hofmeyr et al. 2005).
The available evidence for pelvic and nonpelvic constraint on egg size does not show a clear pattern related exclusively to body size, terrestriality, or even phylogenetic inertia, a possible result of the diversity of architectures exhibited by the pelvic girdle and associated skeleton of turtles (Lovich et al. 2012). Our aims in this study were to test the main predictions made by Congdon and Gibbons (1987) on 1) pelvic constraint on egg size in small and semiterrestrial species, 2) a reduction in egg size variation among populations inhabiting relatively stable environments (optimal egg size theory), and 3) variation in clutch size; under optimal offspring size theory, more variation in clutch size rather than egg size (width) should be expected across a range of female body size (Lovich et al. 2012). We contextualized our analyses on the 5 constraint/optimization outcome tests of Lovich et al. (2012). Based on previous research done on kinosternids (van Loben Sels et al. 1997; Wilkinson and Gibbons 2005; Lovich et al. 2012; Macip-Ríos et al. 2012), we expect a range of responses on egg size constraint from unconstrained/optimized to constrained/optimized outcomes (see fig. 1 of Lovich et al. 2012).
Citation: Chelonian Conservation and Biology 12, 2; 10.2744/CCB-1038.1
METHODS
Study Species
The Mexican mud turtle (Kinosternon integrum) is the freshwater turtle with the broadest distribution in Mexico (Iverson et al. 1998). The species typically inhabits small and shallow aquatic habitats such as ponds, wells, or perennial and seasonal streams, but they can also be found also in large river systems (Iverson 1999). Mexican mud turtle populations reach densities as high as 1020 individuals/ha (Macip-Ríos et al. 2009). A preliminary study of reproductive characteristics across several populations of K. integrum showed little geographic variation in life history traits, with positive correlations of clutch size and egg mass to maternal body size (Iverson 1999). Mexican mud turtles nest from May to October, with a mean clutch size of 4 (1–12) eggs and a clutch frequency ranging from 1 to 4 clutches per breeding season (Iverson 1999; Macip-Ríos et al. 2009).
Kinosternon chimalhuaca is a small-bodied endemic turtle from the Chamela-Barra de Navidad region in Jalisco and Colima in western Mexico (Berry et al. 1997). This turtle inhabits shallow seasonal ponds and streams along the coast, occurring from the Ciuhuatlan to the Tomatlan river basins (Casas-Andreu 2002). Very few data are available regarding its natural history, but according with Berry et al. (1997) this species probably nests from May to August, with an average clutch size of 3.7 (range 2–5); males mature at about 100 mm carapace length whereas females mature between 97 and 107 mm carapace length.
Study Sites and Data Collection
From October 2003 to October 2008 we surveyed and collected adult Kinosternon integrum females from 3 localities in central Mexico: Tonatico (18°45′N, 99°41′W), Tejupilco (18°45′N, 100°36′W) in Estado de Mexico, and Nuevo Urecho (19°09′40″N, 101°55′00″W) in Michoacan, and from 13 to 18 August 2007 we looked for adult K. chimalhuaca in the Chamela-Barra de Navidad region in the state of Jalisco, western Mexico (19°31′39″N, 105°04′25″W; Fig. 1). The 4 study sites are located along the same general latitude, but elevation was highly variable between localities (Tonatico is at 1600 m above sea level, Tejupilco is at 600 m, Nuevo Urecho at 450 m, and the Chamela region is near sea level). They also experience different environmental conditions such as temperature, rainfall, and rainfall predictability. In order to estimate the stability of each environment and to test our second prediction, we gathered annual rainfall and annual mean temperature data from 1976 to 1988 (the only data available) from the 4 study sites. Climatic data were provided by the Instituto Meteorologico Nacional (National Institute of Meteorology). We followed Hofmeyr et al. (2005) and calculated rainfall predictability (rainfall predictability = 100 – the coefficient of variation of annual rainfall in %).
Turtles were captured by hand, seines, hoop nets, and fyke nets. We measured plastron length (PL) with a plastic ruler (to 1 mm) and body mass with a triple-beam balance (to 1 g). We consider those K. integrum female specimens larger than 96 mm in PL to be adults, following Iverson (1999) and our previous work (Macip-Ríos et al. 2011); for K. chimalhuaca, those females larger than 99 mm in carapace length (the smallest body size reported for a K. chimalhuaca female; Berry et al. 1997) were considered to be adults. Turtles were individually marked by shell-notching and radiographed in the laboratory to determine gravidity and clutch size, and to have a measureable image of eggs and the pelvic girdle. While in the laboratory, females were individually maintained in 20-l buckets of approximately 6 cm of water. Females were fed every 4 days with commercial turtle food pellets and raw chicken. The X-ray procedure was done with a 300-mA Philips 1.50 veterinary X-ray machine, using nonscreened cardboard cassettes (tabletop) at 200 mA and a 70-kV peak for 0.7 sec at a distance of 1 m (Gibbons and Greene 1979; Hinton et al. 1997). After the X-radiographs were taken most of the females were returned to their original populations after a maximum of 4 wk.
Egg width (EW) and the width of the pelvic aperture (the closest distance between the ilia) (PA) were measured in the X-ray photographs in a (homemade) viewer box using a Truper® dial caliper (0.02 mm). PA and EW measurements were performed by the same person in order to avoid bias and reduce measurement error. From parallel studies we obtained a series of eggs from gravid females by injecting a 1.5-ml/kg dose of oxytocin (Ewert and Legler 1978). We used the extracted eggs' data to compare EW from fresh-laid eggs with EW from the X-radiographs (XREW) in order to validate the egg size accuracy from the X-ray technique. We also used fresh-laid eggs to obtain egg length measurements.
Statistical Analysis
To test for environmental stability at our study localities, we analyzed the available climatic data comparing annual rainfall and mean annual temperature from 1976 to 1988 between localities with a one-way analysis of variance (ANOVA). An honestly significant difference (HSD) Tukey test was used as an a posteriori test to compare climate means among localities. Since the annual rainfall predictability was based on the coefficient of variation, we compared this variable directly between localities without using any other statistical test (Hofmeyr et al. 2005).
To validate our X-ray technique we calculated the ratio and standard deviation between the actual EW in fresh eggs and EW for the same eggs measured in the X-ray photographs. We did not follow Graham and Petokas (1989) for correcting X-ray magnification because we did not use any known-size object in the X-ray photograph to correct magnifications; however, we consider our estimation of egg size (from X-ray photographs) accurate because the ratio fell between 0.9 and 1.1 with a standard deviation ≤ 0.05 (95% of confidence interval [CI]). We calculated mean and maximum egg elongation with fresh-egg data by using the simple egg length/EW ratio (Maritz and Douglas 1994); then in order to search for pelvic constraint we used both mean X-ray EW (meanXREW) and maximum X-ray EW (maxXREW; per clutch) in the analyses.
A 1-way ANOVA was used to compare body size among populations. An HSD Tukey test was used as an a posteriori test to compare means between populations. To test for pelvic constraint on egg size we searched for the following evidence: 1) greater egg elongation in small females and 2) equality in the slopes of regression lines for EW and PA on PL (Congdon et al. 1983). We also tested for egg size optimization using Ryan and Lindeman's (2007) criterion of no correlation between egg size and body size under an allometric approach; this criterion was used as complementary evidence of morphological constraint due to pelvic morphology.
We used simple linear regression to test if egg elongation was correlated with body size. We followed King's (2000) methodology to generate the slope values within populations and to determine allometric correlations between EW and body size, and PA and body size. We also follow Ryan and Lindeman (2007) and Naimi et al. (2012) in expecting a slope of 1 when regressions involve the analysis of 2 linear measures. A slope with a 95% CI positioned below the expected value was considered as a significantly hypoallometric relationship and slopes with a 95% CI including the expected value were considered to be isometric. A full factorial analysis of covariance (ANCOVA) was used to compare slopes and origins in the PA and EW regressions to body size in order to test for pelvic constraint. Under optimal offspring theory more variation in clutch size rather egg size should be found within populations. To test that we compared coefficients of variation for clutch size and egg size in each of the populations studied (Lovich et al. 2012). All statistical analyses were performed using JMP ver. 5.0.1 (SAS 2002) with α = 0.05.
RESULTS
We found significant variation in both mean annual temperature (F3,52 = 242.47, p < 0.0001) and annual rainfall (F3,52 = 9.34, p = < 0.0001) among localities (Table 1). Mean annual temperature was statistically different among all localities, but annual rainfall fell into 2 statistical groups, one including Tejupilco and Tonatico with a range of 838–1230 mm annual rainfall, and the other including Chamela and Nuevo Urecho with a range of 625–943 mm annual rainfall (Table 1). Rain predictability did not vary significantly across locality, from 74.73% in Tonatico to 82.12% in Tejupilco (Table 1). Considering the 3 climatic variables analyzed we consider environmental stability at the 4 localities unequal (Table 1).
Our measurements of egg size on the X-ray photographs were accurate estimates of actual egg size (using the X-ray/fresh eggs ratio) only for the K. chimalhuaca population (x¯ ± SD) (0.96 ± 0.02, n = 36 eggs) and for the K. integrum population at Tonatico (0.96 ± 0.02, n = 32 eggs); however, for the K. integrum populations at Nuevo Urecho (0.91 ± 0.05, n = 56 eggs) and at Tejupilco (0.92 ± 0.04, n = 38 eggs), our estimates of egg size from X-ray were below the alpha level of 0.05. We recognize that our PA and X-ray mean EW measures broke the imposed confidence limit of 95%, but it is our best measure and stills fit in the confidence limit of 90%. We found significant variation in body size (PL) (F3,66 = 20.27, p < 0.0001) among the 4 turtle populations. The larger turtles were from Nuevo Urecho, followed by those in Tonatico and Tejupilco (not significantly), and finally the significantly smaller-bodied population was from the K. chimalhuaca population of Chamela, Jalisco (Table 2). Egg elongation (mean and max) did not show any correlation with body size in K. chimalhuaca (Chamela) and the K. integrum population from Nuevo Urecho. The Tonatico and Tejupilco (partially, only for mean egg elongation) populations both showed a negative correlation between mean and maximum egg elongation with body size (see Table 3 for all results). The coefficients of variation for clutch size in the 4 populations surveyed were 4.5 to 9.5 times higher than coefficients of variation for maxEW and meanEW (Table 2).
For the single K. chimalhuaca population the slopes of the regressions of meanXREW to PL, and PA to PL were marginally statistically different (F3,19 = 4.44, p = 0.051), with the same result for the regressions for maxXREW to PL, and PA to PL (F3,19 = 4.34, p = 0.053; Fig. 2). Intercepts were also significantly different in both sets of regressions (F2,19 = 90.38, p < 0.0001 for meanXREW, PA to PL, and F2,19 = 66.51, p < 0.0001 for maxXREW, PA to PL). No significant correlations were observed between PA or mean EW or maxXREW and body size (PL) (Table 4).
Citation: Chelonian Conservation and Biology 12, 2; 10.2744/CCB-1038.1
For the 3 populations of K. integrum analyzed we found the following results: in the Nuevo Urecho population the slopes of the regressions of meanXREW to PL, and PA to PL were not statistically different (F3,17 = 0.55, p = 0.46), with the same result for the regressions for maxXREW to PL, and PA to PL (F3,17 = 0.62, p = 0.44; Fig. 3). Intercepts were significantly different in meanXREW and PA to PL regressions (F2,17 = 163.18, p < 0.0001) and in maxXREW and PA to PL regressions (F2,17 = 161.79, p = 0.0001). The correlation between PA and body size (PL) was hypoallometric; meanwhile neither meanXREW nor maxXREW showed any correlation to body size (Table 4).
Citation: Chelonian Conservation and Biology 12, 2; 10.2744/CCB-1038.1
For the Tejupilco population the slopes of the regressions of meanXREW to PL, and PA to PL were statistically different (F3,25 = 21.12, p = 0.0001). The same result was found for the regressions between maxXREW to PL, and PA to PL (F3,25 = 17.91, p = 0.0003; Fig. 4). Intercepts were also significantly different in both sets of regressions (F2,25 = 63.81, p < 0.0001 for meanXREW and PA to PL and F2,25 = 47.03, p < 0.0001 for maxXREW and PA to PL) (Table 4). The correlation between PA and body size (PL) was isometric, but we found no correlation between meanXREW or maxXREW and PL (Table 4).
Citation: Chelonian Conservation and Biology 12, 2; 10.2744/CCB-1038.1
Finally, for the Tonatico population the slopes of the regressions of meanXREW to PL, and PA to PL were not statistically different (F3,69 = 3.36, p = 0.071), with the same result for the regressions of maxXREW to PL, and PA to PL (F3,69 = 2.96, p = 0.09; Fig. 5). Intercepts were significantly different in both sets of regressions (F2,69 = 309.44, p < 0.0001 for meanXREW and PA to PL and F2,69 = 260.96, p < 0.0001 for maxXREW and PA to PL; Table 4). The correlation between PA and body size (PL) was isometric, whereas the correlations between meanXREW and maxXREW to PL were hypoallometric (Table 4).
Citation: Chelonian Conservation and Biology 12, 2; 10.2744/CCB-1038.1
Figure 6 simultaneously presents all the slopes of the regression on PA to body size and EW to body size for comparative purposes. Mean EW was not constrained by PA width in K. chimalhuaca (the smaller species), whereas the K. integrum populations showed variants of the same general pattern, with only the Tonatico population suggesting a possible pelvic constraint. This figure also shows how in larger animals such those of Nuevo Urecho, PA is more variable than in small animals.
Citation: Chelonian Conservation and Biology 12, 2; 10.2744/CCB-1038.1
DISCUSSION
According to the criteria used to determine pelvic constraint (equal slopes in egg size/PA to body size and greater egg elongation in small females) we only found strong support for pelvic constraint in the Tonatico population and some indication of morphological constraint in the Nuevo Urecho population (equal slopes in egg size/PA to body size) and Tejupilco (an inverse correlation between mean egg elongation with body size). We did not found any evidence of pelvic constraint for the population of K. chimalhuaca. Our overall data analyses differ from Congdon and Gibbons' (1987) prediction of pelvic constraint in small-bodied populations of turtles, since most of our data on the populations K. chimalhuca from Chamela and K. integrum from Tejupilco showed no pelvic constraint despite small bodies (mean PL = 116.4 and 129.61, respectively). Our data agree with others who found no pelvic constraint in small kinosternids (Iverson 1991, 2002; van Loben Sels et al. 1997; Wilkinson and Gibbons 2005; Lovich et al. 2012).
We found evidence of egg size optimization (no pelvic constraint and no correlation between egg sizes to body size) in the Tejupilco population (Fig. 4) and the K. chimalhuaca population (Fig. 2). In addition we also found support for an optimized size in the Nuevo Urecho population, despite the equal slopes in PA and egg size (possible pelvic constraint evidence). For example, we found no overlap between the smallest PA and the largest EW in this population (Fig. 3). We believe that this population did not show a pelvic constraint on egg size, even though PA showed a hypoallometric correlation with body size. This result implies that parallel slopes should not be used as absolute evidence for pelvic constraint (see also Iverson and Smith, 1993) and confirms our approach of using multiple lines of evidence in to diagnose pelvic constraint. Naimi et al. (2012) found similar results and they argued for nonpelvic constraint in Mauremys leprosa in Morocco. Additional evidence of lack of a PA constraint in large females of K. integrum was provided by Macip-Ríos et al. (2012), in which a threshold size of 140 mm in PL was found, below which there were no morphological constraints. As we show in this paper, the Nuevo Urecho population had the largest body size of the four populations surveyed (Table 2), with maturity starting at 140 mm of PL. These results place this population at the upper size limit previously described for this species (Macip-Ríos et al. 2012).
The Tonatico population did not show an optimized egg size (Table 4; Fig. 5), but we found a hypoallometric correlation of egg size to body size. Both results agree with the evidence of equal slopes of egg size and PA found in the ANCOVA test, and the greater elongation of eggs found in small females confirms a pattern of pelvic constraint on egg size in this population.
The observed pattern of no pelvic constraint and lack of correlation between egg size and body size reported in the present study suggest that egg size (width) could be optimized in small-bodied populations. This trend has been reported in other small turtle species such as Chrysemys picta (Iverson and Smith 1993; Rollinson and Brooks 2008), K. sonoriense (Lovich et al. 2012), and K. subrubrum (Wilkinson and Gibbons 2005), and could be driven by wider pelvic girdles of females than males (Long and Rose 1989). To increase their fitness, turtles with optimized eggs simply add more eggs to their clutch rather than increase egg size (Macip-Ríos et al. 2012). On the other hand, the pelvic constraint pattern found in the Tonatico population indicates a nonoptimized egg size. These results corroborate with the model of maternal investment proposed by Kaplan and Cooper (1984), which implies that reproduction in unpredictable environments leads to high variation in egg size among clutches and females. Hofmeyr et al. (2005) and Lovich et al. (2012) suggested that environmental stability could drive reproductive output in turtles; in more stable environments egg size tends to stabilize compared with less stable environments. The possible explanation for the optimized egg size but significant correlation between PA and body size in the K. integrum population from Nuevo Urecho could be that this population inhabits a qualitatively more eutrophic environment compared with the other populations studied. This could drive attainment of large body size (Lovich et al. 2010; Naimi et al. 2012) and then lead to more variation in other morphological characteristics such PA, but not necessarily to an increase in egg size (Germano 2010). Despite that this result can definitely be population-specific, it is not completely unknown for turtles, since 2 populations of Malaclemys terrapin (one population occurring in a habitat stressed by extensive human use and human-caused mortality, and the other occurring in a much less disturbed habitat) showed differences in egg size to body size relationships (Lovich, pers. comm., March 2013).
The environmental stability gradient observed coincides with the pelvic and nonpelvic constraint evidence found in the 4 populations analyzed. The populations inhabiting more stable environments (Tejupilco and Chamela) showed clear evidence of no pelvic constraint and egg size seems to be optimized (but see the egg elongation to body size inverse correlation in Tejupilco), whereas at less stable sites the evidence points to pelvic constraint and nonoptimal egg size at Tonatico, but optimal egg size, and some contradictory evidence of pelvic constraint at Nuevo Urecho. Lovich et al. (2012) proposed that EW could be less variable in hydrologically stable environments in K. sonoriense. Our data are in partial agreement, since the Tonatico population inhabits a seasonal pond (less-stable habitat) and the Nuevo Urecho population inhabits a perennial creek. Our contradictory evidence at Nuevo Urecho also could be driven by the small sample size compared with the other populations; more work is needed on this population in order to clarify the trends we found.
Our third prediction was based on optimal offspring size theory that predicts more variation in clutch size than egg size (width) across a range of female body size (Smith and Fretwell 1974). Our results indicate that clutch size varied 4.5 to 9.5 on egg size (Table 2). The intriguing finding was that even in the population with ample evidence for pelvic constraint (Tonatico), we detected support for our prediction, further suggesting a tendency to optimize egg size. This result coincides with that of Ryan and Lindeman (2007) for Graptemys geographica, in which turtles produce the widest eggs that they are physically capable of laying, but with egg size always showing less variation than clutch size. Iverson and Smith (1993; see their table 12), Rollinson and Brooks (2008), Rollinson et al. (2012) and Naimi et al. (2012) also found more variation on clutch size compared with egg size.
The prediction of pelvic constraint in small and semiterrestrial turtles was not supported by our main findings. Our results did not agree with a supposed expected adaptive compromise between the use of the pelvis for reproduction and for locomotion proposed by Congdon and Gibbons (1987). Three of the 4 populations (not Tonatico) of 2 species surveyed tended toward optimality in egg size, showing less variance in egg size than in clutch size. We also found that a simple slope comparison of egg size to body size vs. PA to body size was not enough to test for pelvic constraint. Rather, the use of at least 3 different criteria (slope comparison, allometric correlations between body size and egg size, and the study of overlap between wider eggs and smallest pelvic aperture) is needed to test for pelvic constraint (Congdon and Gibbons 1987; Iverson and Smith 1993; Ryan and Lindeman 2007; Lovich et al. 2012).
Finally, local environmental factors affecting water availability seem to play an important role in reproductive output in these turtle populations (Macip-Ríos 2010). Apparently, besides the important role of body size in reproductive output in turtles (Iverson 1992; Iverson and Smith 1993, Rollinson et al. 2012), environmental predictability (Murphy 1968; Cunnington and Brooks 1996; Rollinson and Brooks 2007) and stability (Lovich et al. 2012; Naimi et al. 2012) play important roles in local reproductive output strategies. Future work on geographic variation in the reproductive ecology of the wide-ranging K. integrum, which spans circa 11° of latitude, 14° of longitude, and 2200 m of elevation will be valuable in better understanding the basis for the variation we have documented.
Study area and the localities where Kinosternon integrum and K. chimalhuaca (Chamela) were captured.
Comparison of slopes for the relationships of plastron length to size of pelvic aperture (closed circles), or mean (open circles) and maximum (triangles) egg width for the studied population of Kinosternon chimalhuaca in Chamela, Jalisco (n = 10).
Comparison of slopes for the relationships of plastron length to size of pelvic aperture (closed circles), or mean (open circles) and maximum (triangles) egg width for the studied population of Kinosternon integrum from Nuevo Urecho, Michoacan (n = 9).
Comparison of slopes for the relationships of plastron length to size of pelvic aperture (closed circles), and mean (open circles) and maximum (triangles) egg width for the studied population of Kinosternon integrum from Tejupilco, Estado de Mexico (n = 13).
Comparison of slopes for the relationships of plastron length to size of pelvic aperture (closed circles), and mean (open circles) and maximum (triangles) egg width for the studied population of Kinosternon integrum from Tonatico, Estado de Mexico (n = 35).
Comparison of slopes for the relationships of plastron length to size of pelvic aperture (open symbols) and mean egg width (closed symbols) for the 4 populations of Kinosternon analyzed. Circles represent Kinosternon chimalhuaca; triangles represent K. integrum from Nuevo Urecho, Michoacan; squares represent K. integrum from Tejupilco, Estado de Mexico; and diamonds represent K. integrum from Tonatico, Estado de Mexico. For K. integrum, dashed lines represent pelvic aperture to body size slope and solid lines represent mean egg width to body size slopes. The long dashed lines punctuated with points represent the slopes of the regression of pelvic aperture and mean egg width to body size of K. chimalhuaca.
Contributor Notes
