METHODS

Adult female red-eared sliders (Trachemys scripta elegans) were collected on the Joe Hogan Fish Hatchery in Lonoke, Arkansas (34.78°N; 91.90°W), on 20 May 1997 (4), 3 May 1998 (8), and 1 May 1999 (12). Their removal was encouraged by fisheries personnel because the turtles compete with the fish stocks for the food pellets that they provided. Those I removed were destined for elimination by hatchery staff. This population received supplemental feeding, so its reproductive output could be elevated (but see Results). Oviposition in 6 females from 1999 was induced using oxytocin injection (1 unit/100 g body mass) following the method of Ewert and Legler (1978) and the turtles were rereleased. All other females (n = 18) were humanely euthanized with an intracoelomic injection of sodium pentobarbital (ca. 400 mg/ml) and their reproductive tracts removed for examination (e.g., Kuchling 1999). Each female was measured (maximum carapace length, CL, in mm; maximum plastron length, PL, in mm; Iverson and Lewis 2018), and weighed (body mass, BM, in grams) within 48 hrs of capture.

In order to compare the reproductive data from Arkansas with those across the species’ range, I compiled data on reproductive output for populations of T. scripta (Table 1). Mean parameters included population latitude, carapace length (CL), body mass (BM), clutch size (CS), egg length (EL), egg width (EW), egg mass (EM), clutch mass (CM = EM × CS), relative clutch mass (RCM = %CM/BM), relative egg mass (REM = %EM/BM), and clutch frequency (CF, in number of clutches per year). When possible, missing data were estimated from other data published in the original source. If not reported, CL or PL were estimated by the PL/CL ratio of 0.942 as reported in Lewis et al. (2018). When not reported, BM for T. s. elegans was estimated from data from 209 females and juveniles from Indiana (BM = 0.000520 × CL^2.7474; r² = 0.998; p < 0.0001; Lewis et al. 2018 and unpublished), and that for T. s. scripta was estimated from the formula in Iverson (1984) based on turtles from Georgia. Egg mass (EM) was estimated from EL and/or EW based on data from 340 eggs from across 4 states (EM = 0.329EL + 0.773EW − 18.563; r² = 0.90; p < 0.0001; J.B.I., M.A. Ewert, and E.O. Moll, unpubl. data). I excluded introduced populations from this analysis, as well as reports based on single individuals (e.g., Taylor 1935; Lardie 1976; Tucker 1996). No data were available for T. s. troostii. Means are reported ± 1 standard deviation (SD). I performed all calculations (simple statistics, t-tests, and least-squares linear regression analyses) on MacIntosh hardware with Statview™ software (Abacus Concepts, Berkeley, CA, USA).

RESULTS

Reproductive Output in Arkansas. — Arkansas females in this study (n = 24) averaged 216 ± 13 mm CL, 201 ± 13 mm PL, and 1404 ± 278 g BM (Table 2). Only 1 of the 18 females captured and dissected had oviposited prior to its capture on 1 May (based on counts of her corpora lutea). Three others lacked oviducal eggs or corpora lutea, indicating that they had not yet produced their first clutch of the year. All others were gravid, including the 6 that were administered oxytocin (20 total). All dissected females had 2 or 3 sets of enlarged ovarian follicles (Table 2). The largest sets were 18–21 mm in diameter (ovulatory sized), with secondary sets in discrete size classes down to 10-mm diameter (following Moll 1979; Kuchling 1999; Legler and Vogt 2013), strongly suggesting the potential for subsequent clutches. Mean estimated clutch frequency for 18 dissected females based on these sets of follicles was 3.3, ranged only from 3 to 4, and increased with CL (F_1,16 =14.63; r² = 0.48; p = 0.0015). Counts of corpora lutea matched counts of oviducal eggs in all cases. There was no evidence of any female skipping reproduction in the year of capture (i.e., ovaries lacking enlarged follicles or corpora lutea).

Table 2. Reproductive data from female pond sliders (Trachemys scripta) from central Arkansas. Abbreviations are CL (maximum carapace length in mm), PL (maximum plastron length in mm), BM (gravid body mass in g), CS (clutch size), EL (egg length in mm), EW (egg width in mm), EM (egg mass in g), CF (clutch frequency), Next CS (largest set of preovulatory follicles presumed to be the next clutch), 3rd CS (second set of enlarged follicles assumed to be another clutch), 4th CS (third set of enlarged follicles assumed to be the fourth clutch of the season), CM (clutch mass in g), RCM (relative clutch mass, CM/[BM − CM] in %), and REM (relative egg mass, RCM/CS in %). ND = no clutch frequency data for females with oxytocin-induced oviposition; NC = not counted but present.

View Fullscreen

Clutch size (CS) averaged 10.7 ± 2.65 eggs (n = 20 clutches), but was not related to female CL (F_1,18 =1.75; r² = 0.09; p = 0.20; n = 20); however, if one outlier was removed (CS = 18), they were related (CS = 0.095CL − 10.454; F_1,17 = 8.44; r² = 0.33; p = 0.01; log CS = 2.124log CL − 3.97; F_1,17 = 8.72; r² = 0.58; p = 0.009). Mean clutch sizes for estimated subsequent clutches (from follicular counts; i.e., not necessarily the eventual numbers oviposited) were 11.7, 10.5, and 10.6 (Table 2), suggesting possible average annual production of 44 eggs in average-sized clutches.

Mean clutch egg mass (EM) averaged 10.73 ± 1.28 g (n = 20 clutches) and was related to EL and EW by the equation EM = 0.296EL + 0.789EW − 17.675 (F_2,180 = 1872; r² = 0.96; p < 0.0001; n = 183). In addition, EM = 1.170EW − 15.365 (F_1,182 = 979; r² = 0.84; p < 0.0001). There was no relationship to CL of mean clutch EL (r² = 0.03; p = 0.46) nor EW (r² = 0.10; p = 0.17) nor EM (r² = 0.003; p = 0.81; n = 20 for each), but mean clutch EM and CS were inversely related (F_1,18 = 16.28; r² = 0.48; p = 0.0008; n = 20).

Clutch mass (CM) was positively related to CL (F_1,18 = 7.94; r² = 0.31; p = 0.011; n = 20). Relative clutch mass (RCM) averaged 7.86 ± 1.36% (n = 20), but was highly variable (range, 5.4–10.1%), and inversely related to CL (F_1,18 = 6.06; r² = 0.25; p = 0.02). Relative egg mass (REM; mean = 0.77 ± 0.17%; range 0.54–0.11%; n = 20) was negatively correlated with CL (F_1,18 = 29.2; r² = 0.62; p < 0.0001). Together, these data suggest that clutch size tends to increase slightly with body size in T. scripta in Arkansas, but that egg size remains relatively constant.

Population Comparisons. — Comparisons with other populations revealed that all measures of reproductive output in Arkansas were similar to those from across the range of T. s. elegans (Table 1). However, those comparisons revealed significant differences in reproductive output between T. s. scripta and T. s. elegans. Although the two subspecies had similar average carapace lengths (Table 3), they differed significantly in body mass (presumably due to the more domed shell of the nominate subspecies; Carr 1952, p. 242). Despite their larger body volume, T. s. scripta had smaller clutches, lower clutch masses (both actual and relative to body mass), and lower relative egg mass, but eggs of nearly identical size (Table 3). Further analyses were done separately by subspecies because of these differences.

Table 3. Comparisons (t-tests) of reproductive data between red-eared sliders (Trachemys scripta elegans) and yellow-bellied sliders (T. s. scripta). Abbreviations are CL (maximum carapace length in mm), BM (body mass in g), CS (clutch size), EL (egg length in mm), EW (egg width in mm), EM (egg mass in g), CM (clutch mass in g), RCM (relative clutch mass, CM/[BM − CM] in %), REM (relative clutch mass, RCM/CS in %), and CF (clutch frequency). Sample sizes in parentheses.

View Fullscreen

Across populations, mean CS increased with mean CL for both T. s. elegans and for T. s. scripta, although the slopes differed significantly (for elegans, b =0.16 ± 0.02 standard error [SE]; for scripta, b = 0.06 ± 0.01 SE; Fig. 1). However, mean clutch egg mass increased with mean CL across populations only for T. s. scripta (F_1,9 = 27.49; r² = 0.75; p = 0.0005; n = 11), but not for T. s. elegans (F_1,19 = 0.22; r² = 0.01; p > 0.64; n = 21). Reflecting the low variation in egg size across populations, mean REM was negatively correlated with mean CL for populations of both T. s. elegans (F_1,18 = 137.2; r² = 0.88; p < 0.0001; n = 20) and T. s. scripta (F_1,8 = 77.36; r² = 0.91; p < 0.0001; n = 10). Mean RCM was unrelated to mean CL (Fig. 2) for both T. s. elegans (F_1,18 = 0.11; r² = 0.006; p = 0.75; n = 20) and T. s. scripta (F_1,8 = 3.03; r² = 0.27; p = 0.12; n = 10). Mean CF did not differ between subspecies (Table 3) and was not related to mean CL for either T. s. elegans (F_1,6 = 0.05; r² = 0.008; p = 0.83; n = 8) or T. s. scripta (F_1,3 = 0.31; r² = 0.09; p = 0.62; n = 5) or for the subspecies combined (F_1,11 = 0.02; r² = 0.002; p = 0.90; n = 13).

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

Latitudinal Effects. — Mean CL among populations increased significantly with latitude (Fig. 3) for T. s. elegans (F_1,20 = 33.64; r² = 0.63; p < 0.0001; n = 22), but not for T. s. scripta (F_1,12 = 1.00; r² = 0.08; p = 0.34; n = 14). Similarly, mean clutch size increased significantly with latitude (Fig. 4) for T. s. elegans (F_1,24 = 23.44; r² = 0.49; p < 0.0001; n = 26), but not for T. s. scripta (F_1,12 = 0.99; r² = 0.048; p = 0.45; n = 14).

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

Mean clutch egg mass was not related to latitude in either T. s. elegans (F_1,22 = 0.02; r² = 0.001; p = 0.90; n = 24) or T. s. scripta (F_1,10 = 1.26; r² = 0.11; p = 0.29; n = 12), nor was mean RCM (Fig. 5; T. s. elegans, F_1,18 = 0.12, r² = 0.006, p = 0.74; n = 20; T. s. scripta, F_1,8 = 2.12, r² = 0.21, p = 0.18; n = 10). Mean REM decreased with latitude in T. s. elegans (F_1,18 = 47.08; r² = 0.72; p < 0.0001; n = 20), but not in T. s. scripta (F_1,8 = 0.68; r² = 0.08; p = 0.43; n = 10). Finally, mean CF was not related to latitude for either T. s. elegans (F_1,8 = 0.19; r² = 0.023; p = 0.68; n = 10) or T. s. scripta (F_1,3 = 0.36; r² = 0.11; p = 0.59; n = 5).

View Figure

• Download as Powerpoint
• Download Figure

DISCUSSION

The ponds at the Joe Hogan Fish Hatchery were extensive, shallow, and unshaded, which likely resulted in higher water temperatures than nearby natural habitats. In addition, although aquatic vegetation was limited in these hatchery ponds as a result of rotational draining for fish removal, commercial food pellets were provided to ponds daily and readily consumed by turtles (to the chagrin of fisheries staff). However, despite this optimal turtle habitat, reproductive parameters and patterns from my sample were generally consistent with previously reported data for T. s. elegans (compare Tables 2 and 3).

Population Comparisons. — Comparisons of the Arkansas data with those from other populations of T. scripta revealed several clear patterns, some (but not all) previously reported. Perhaps the most surprising result was the distinct difference in reproductive strategies between populations of T. s. elegans and those of T. s. scripta (e.g., Table 3). Despite the more highly domed shell of the latter (providing greater abdominal volume), populations of T. s. scripta actually produce fewer eggs per clutch than those of T. s. elegans, but their eggs are nearly identical in size across both subspecies, suggesting selection for optimal egg size (Smith and Fretwell 1974; Brockelman 1975; Hendry et al. 2001). Thus, the reproductive output of T. s. elegans (as measured by RCM per clutch) is more than twice that of T. s. scripta (see also Moll and Moll 1990), despite apparently similar nesting season lengths (Ernst and Lovich 2009:459) and annual clutch frequencies (Table 3), and much greater known longevity (47 yrs in T. s. elegans, Ernst and Lovich 2009; 25 yrs in T. s. scriptaGibbons 1987). The reason that T. s. elegans evolved such high reproductive output over a longer life compared with T. s. scripta is not yet clear, but no doubt contributes to its success as an invasive taxon around the world (Lowe et al. 2000; Reshetnikov et al. 2023).

The significant divergence in reproductive strategies between the two subspecies demonstrated here is surprising, given that intergradation has been demonstrated genetically (Parham et al. 2020; Vamberger et al. 2020) and the uncorrected p distance between the subspecies for the cyt b gene was only 0.59% (Vargas-Ramírez et al. 2017).

The increase in clutch size with body size found in my Arkansas sample has also been reported for nearly all populations of T. scripta cited in Table 1. Only Webb (1961; Oklahoma, but n = 6), Mitchell and Pague (1990; Virginia), and Palmer and Braswell (1995; North Carolina: table 19) found no such correlation.

Across populations of both T. s. scripta and T. s. elegans, mean clutch size also tended to increase with body size (Fig. 1). However, egg size tended to increase with body size only across populations of T. s. scripta and not T. s. elegans, demonstrating another fundamental difference in the reproductive strategies of the two subspecies. The adaptive significance of this difference is unclear.

Although egg size did not vary with body size in my sample or those reported by Cagle (1950; Illinois), Webb (1961; Oklahoma, but n = 6), and Palmer and Braswell (1995; North Carolina, data from table 19), egg size was found to increase with body size by Congdon and Gibbons (1983, 1990; South Carolina), Glidewell (1984; Texas), and Tucker and Moll (1997; Illinois; see also Tucker 2000b, 2001a). Clearly, egg size is much less likely than clutch size to increase with body size in this species. I had no measurements of pelvic aperture diameter, but Congdon and Gibbons (1987) demonstrated that egg size in T. scripta is not constrained by pelvic anatomy.

Egg size and clutch size were inversely related in Arkansas as they were in Illinois (Tucker et al. 1998b); however, Congdon and Gibbons (1983) found no correlation between the two variables in South Carolina. This relationship suggests that a trade-off exists between egg size and clutch size in T. s. elegans, but not in T. s. scripta. Relative egg mass decreased with body size in Arkansas as it did in Missouri (Thomas 1993) and South Carolina (Congdon and Gibbons (1985). Hence, larger females tend to produce relatively smaller eggs across populations.

In Arkansas, RCM decreased with body size, whereas Thomas (1993) found no relationship between RCM and body size in Missouri. No other study has reported on the relationship of these two variables, and hence no universal pattern is yet evident.

Clutch frequency increased with body size in the Arkansas population, as it did in 2 populations of T. scripta in Missouri (Thomas 1993) and in 1 of 2 populations in Illinois (Thornhill 1982). Although this pattern likely applies across the range of T. s. elegans, it remains unstudied in T. s. scripta.

The typical reproductive pattern for T. s. elegans (both within and across populations) is increasing clutch size with female size, with egg size increasing only weakly if at all with body size. However, within populations of T. s. scripta both clutch and egg size were weakly related to body size if at all. In at least T. s. elegans, increases in reproductive output are accomplished primarily by increases in clutch size rather than egg size. More data are needed to test this pattern in T. s. scripta.

Latitudinal Effects. — Lewis et al. (2018) reported that populations of female T. s. elegans exhibited increasing mean body size with latitude (Bergmann’s Rule; Angielczyk et al. 2015), whereas those of T. s. scripta did not (Fig. 3). This pattern was confirmed with the additional data compiled here. Clutch size is generally correlated with body size in T. s. elegans (see above), so CS also increased with latitude in that subspecies. However, neither body size nor clutch size increased with latitude in T. s. scripta (Figs. 3 and 4). Egg size did not vary with latitude in either subspecies, further suggesting that selection is operating to optimize egg size in the species.

Prior to the present study, very few authors had examined RCM or its correlates in T. scripta, and those cases were complicated because some authors (e.g., Congdon and Gibbons 1985; Jackson 1988) used gravid body mass as the divisor, and others (e.g., Moll and Moll 1990; Thomas 1993; Tucker et al. 1998a, 1998b) used spent body mass (yielding spent relative clutch mass [SRCM]). The latter were all recalculated in Table 1 herein using gravid body mass as the divisor, and RCM shows considerable consistency within subspecies (Fig. 5) and no evidence of latitudinal variation. Thus, despite the general pattern of increasing body size and clutch size (but not egg size) with latitude demonstrated for T. s. elegans, RCM does not vary with body size or latitude. This suggests that RCM may be under significant genetic constraint range wide.

Our understanding of latitudinal variation in reproductive output in T. scripta has been hampered by small sample size (e.g., Cagle 1950) and by the inclusion of tropical populations now considered to be separate species (Moll and Moll 1990; Tucker et al. 1998b). For example, Cagle (1950) found no difference in mean clutch size between populations of T. s. elegans from Louisiana and Illinois, but smaller eggs in Louisiana, contrary to the pattern demonstrated here. Similarly, Tucker et al. (1998b) reported no correlation between clutch size or egg size and latitude in T. scripta, but their data included 3 tropical populations and they also lumped data across the 2 southern subspecies of T. scripta. The more thorough analysis presented here showed a clear latitudinal increase in clutch size in T. s. elegans, but not in T. s. scripta, and no effect of latitude on egg mass for either subspecies.

My estimate of CF (3.3/yr) was based on the assumption that all enlarged ovarian follicles > 10 mm in diameter would be ovulated during the current nesting season (e.g., Moll 1979, but see Cagle 1950, p. 38, who found some enlarged follicles present in a few females in September and October). That estimate slightly exceeded all others for populations of T. scripta, except Jackson’s (1988) estimate of 3–5 clutches/yr for T. s. scripta from northern Florida, but may reflect the effects of supplemental feeding of the Arkansas population. I expected CF to decrease with latitude as a result of presumed constraints on the length of the nesting season in the north; however, the data available to date do not support that prediction. Of all the reproductive traits quantified here, CF is known to be the most elusive and difficult to quantify (Gibbons and Greene 1990), and will require much more intensive field study.

The latitudinal pattern of increasing CS and CL, but not EM (found in T. s. elegans) is the most common pattern found among wide-ranging North American turtles (reviews in Iverson et al. 1993 and Lewis et al. 2018; see also Litzgus and Mousseau 2006, Lovich et al. 2018, Willey et al. 2021). The advantages to a turtle of larger body size at high latitude (Bergmann’s Rule) were reviewed by Lewis et al. (2018), and those advantages include the ability of large turtles to produce more eggs per clutch (the fecundity hypothesis), possibly as a response to the effects of climatic uncertainty on survival at all life stages and the shorter reproductive season at high latitude (Iverson et al. 1993). However, for at least North American turtles, egg size is consistently less variable both within and among populations, even across latitude, suggesting that egg size is under stronger selection than clutch size (Elgar and Heaphy 1989; Iverson et al. 2019; among others), and that egg size may have evolved as expected from optimal egg size theory (i.e., increases in reproductive output should manifest in increased clutch size rather than egg size; Smith and Fretwell 1974; Brockelman 1975).

Although the comparisons made here included studies with small sample sizes and estimated data, the patterns identified here seem consistent within subspecies. These patterns can be tested with new data from additional populations. Future studies of reproductive output in the southern range of T. s. scripta are particularly needed (especially in Georgia and northern Florida) for a better understanding of latitudinal variation in that subspecies. In addition, studies in the area of presumed intergradation of the subspecies in Alabama are needed to determine where and if the differing subspecific strategies merge.