Reproductive Trends in Captive Heosemys grandis (Geoemydidae)
Abstract
A 20-year record of captive breeding of a female Heosemys grandis revealed a tradeoff between egg size and clutch size across the years when she produced 2 clutches per breeding season. First clutches had few large eggs and second clutches had a large number of smaller eggs. Four F1 progeny from this founder female began their reproductive years with much smaller eggs; however, their eggs increased in size over successive years until they were the same size as those of the long-term breeder.
Variation in egg size can be viewed as an adaptive form of bet-hedging (Kaplan and Cooper 1984). In this context, variation in egg size from a given female may be predictable within a short term, such as a year, but is at random in the long term. For instance, the growth of ovarian follicles to become eggs may occur during weather that is good or bad for ovarian growth, but the resultant hatchlings may face unrelated bad or good conditions for their type (e.g., large or small) because the weather has changed (Kaplan and Cooper 1984). Greater attention regarding turtles has focused on an energetic or space-limited reproductive tradeoff between egg size and clutch size (Elgar and Heaphy 1989). In natural populations, demonstration of a significant inverse correlation nearly always has required a statistical adjustment for female body size (Rowe 1994; Tucker et al. 1998), and even then, that correlation has not always been supported (Nieuwolt-Decanay 1997; Clark et al. 2001). Further, the statistical adjustment may confound interactions between female body size, age-related changes in the female reproductive system (Congdon et al. 2003), and a reversible, perhaps random tradeoff within fully mature females. Only one study has documented a significant egg size-clutch size tradeoff without adjustment for female body size and this tradeoff represented just one seasonal sample among three seasons (Roosenburg and Dunham 1997).
The world population of the Asian turtle Heosemys grandis (greater orange-headed earth turtle) is being depleted through international trade. It is ranked by the IUCN as Vulnerable (www.iucn.org) and by CITES as Appendix II listing (www.cites.org). Rather little information is available on the natural history of the species in the wild. A female collected and held briefly captive in western Thailand (from 19 November to January) evidently nested because the holding pen yielded juveniles and eggshells by July (van Dijk 1998). Some adults presumably taken from the wild for international trade and “rescued” subsequently oviposited in captivity during the winter (Hudson and Buhlmann 2002). The species has temperature-dependent sex determination (Ewert et al. 1994).
The main focus of the present account concerns the 20-year record of reproduction of 1 captive founder female Heosemys grandis (“Founder or Mother”), 4 of her F1 progeny (daughters), and 1 sire for all (see Goode 1997, for a partial summary account). These data allowed us to examine the tradeoff between egg size and clutch size, age-related changes in reproduction, and some temporal aspects of reproduction particular to H. grandis, all in an artificially stable environment in captivity at the Columbus Zoo (CZ). Other accounts of reproduction in captivity include Foust (1989) and Rudolphi and Weser (2000).
Methods
The parental pair and female young, upon reaching adulthood, were housed at CZ in a display area with 3.7 × 2.2 m of floor space. This space was divided between two-thirds land and one-third water (to 30 cm deep) parts. The land part included a sandpit (ca. 1 m2, > 25 cm deep) for nesting. Air temperature remained close to 27°C and water temperature was usually 23°–25°C. The photo-period was set at 12 light:12 dark from mid-fall to mid spring and 15 L:9 D from late spring to early fall. Food was regularly available in excess of that consumed.
During the nesting season, female turtles were palpated to detect eggs, and in most cases left to nest on the display. On a few occasions, oviposition was induced with oxytocin. Although rare in occurrence, eggs laid after a main clutch could be associated with a given clutch because they were fully shelled, apparently overmature, and laid within 3–4 weeks of the main clutch, too soon for egg maturation from subsequent ovulation. Some of the eggs were incubated in containers of damp vermiculite (ca. −100 to −300 kPa) in warm or cooler parts of the room space (26°–30°C). Other eggs were incubated at closely controlled temperatures (25°, 28°, 30°C) in upright incubators at Indiana University.
Statistical applications included simple ANOVAs, linear correlation and regression (with residuals analyses where stated to adjust for female body mass), and Spearman's rho or Wilcoxon's signed rank tests in most cases involving binomial data. Most measurements include ± standard deviation (SD).
RESULTS AND DISCUSSION
Founder. — The Founder Female (Mother) was an adult when obtained on 27 July 1967 and died on 2 March 1998. Her mass averaged 5160 g (range 5075–5225 g, independent of age). Her carapace length (CL) remained at 295 mm throughout the study. The male (sire) measured 345 mm CL soon after arriving at CZ on 2 June 1977 and gradually grew to 368 mm (7060 g) by 1992.
During 20 years of observation, Founder produced 2 clutches seasonally during 15 years, 1 clutch seasonally during 2 years and did not reproduce during at least two of three other seasons (Fig. 1). Across all seasons, Founder's clutch production occurred between 5 November (1986) and 11 March (1992). During two-clutch seasons, the first clutch was laid between 5 November and 11 January (range 67 days for 15 clutches, mean = 10 December). Second clutches were laid 21 December–11 March (range 81 days for 15 clutches, mean = 1 February). The average interval between clutches was 55 ± 9 days (n = 15, range 46–79 days). The first and second oviposition dates were significantly correlated across years (r = +0.91, p < 0.0001). During seasons with a single clutch, Founder laid these on 2 and 3 December, i.e., ca. 1 week ahead of the average first clutch.
Citation: Chelonian Conservation and Biology 5, 1; 10.2744/1071-8443(2006)5[165:RTICHG]2.0.CO;2
During the two-clutch seasons, the first clutch averaged significantly smaller than the second clutch by one egg (Fig. 1; first = 4.87 ± 1.06 eggs vs. second = 5.87 ± 1.457 eggs; F1,14 = 9.54, p = 0.008; z = −2.51, p = 0.012). Founder produced the largest clutches (9 and 10 eggs) during each of the two seasons when she produced just one clutch.
During the two-clutch seasons, average egg mass was significantly larger in the first clutch than in the second clutch (Fig. 2; first = 52.22 ± 2.75 g vs. second = 48.23 ± 3.42 g; F1,14 = 21.1, p = 0.0004). Mean egg size in the one-clutch seasons was not consistent relative to egg size in two-clutch seasons (Fig. 2).
Citation: Chelonian Conservation and Biology 5, 1; 10.2744/1071-8443(2006)5[165:RTICHG]2.0.CO;2
The trend in egg size required estimating of egg mass for some clutches because data on eggs from early in the study (11 clutches) included only linear measurements. We used the apparent density of eggs with measured masses and linear measurements of later eggs (19 clutches) to estimate the masses of the early eggs. The mean egg mass of 87 weighed eggs was 48.8 ± 5.4 g (range 36.1–59.0). As estimated from the formula for a prolate ellipsoid, the densities of these eggs averaged 1.174 (range 0.994–1.395). The mean estimated mass of 75 unweighed eggs was 50.69 ± 3.5 g. The densities of 3 eggs directly measured through water displacement was 1.169–1.199 (mean = 1.180). Given the close correspondence between measured and estimated egg densities, the average egg shape of H. grandis (i.e., Founder) seems to conform closely to a prolate ellipsoid (see Iverson and Ewert 1991).
During Founder's two-clutch seasons, the fresh clutch mass of the first clutch (with 1 less egg on average) averaged slightly lighter than for the second clutch (260.1 g vs. 270.8 g, respectively; F1,12 = 5.31, p < 0.04), which relative to Founder's average mass (5160 g) gave relative clutch masses (RCM) of 0.050 and 0.052, respectively, and a seasonal RCM of 0.102. In the 2 years with single clutches, RCM was 0.095 and 0.085.
Mean clutch egg mass was inversely associated with clutch size across all of Founder's clutches (r = −0.36, p = 0.043; Spearman's rho = −0.375, p = 0.04, n = 32). If one clutch (with just 1 unbroken egg for weighing) is removed the association is stronger (r = −0.49, p = 0.0054, Spearman's rho = −0.45, p = 0.014, n = 31).
F1 Progeny
Six of Founder's progeny were retained at CZ until they had matured and produced at least one clutch. Four of these were retained for several additional years. The age at producing the first (primiparous) clutch in 5 of the 6 was 7.6–7.7 years; 1 was younger, 5.7 years. The size at the primiparous clutch was quite uniform (4496 ± 110 g, range = 4325–4650 g; 283 ± 6 mm CL, range 278–291 mm).
The oviposition dates of the progeny clutches (23 November–8 March) occurred within the range of Founder's dates. Three of the 4 retained progeny produced 1 clutch per season during their first 2–4 years of breeding; the fourth progeny started with 2 clutches. Each of these 4 progeny eventually had 2 two-clutch seasons. The mean interval between these clutches averaged 48 ± 10 days (n = 8, range 36–67 days), briefer than Founder's interval between clutches (55 days). Collectively, progeny first and second oviposition dates were significantly associated across years (r = +0.85, p < 0.008).
At CZ, the 7 females produced a total of 61 clutches during a 127-day period between early November and mid-March. At the Turtle Back Zoo (New Jersey), one female produced 7 clutches within the same period (Foust 1989). Thus, nesting in captivity was strongly seasonal, despite the artificial conditions. The range in Julian days includes that of one natural nesting in Thailand (van Dijk 1998) and represents the dry season in western central Thailand (Wernstedt 1959; van Dijk 1998), but includes the last 1–2 months of the wet season in the more southerly Malaysian range of H. grandis. Cues for reproduction at CZ may have been artificial reduction of the photoperiod, naturally drier winter air and accelerated evaporative cooling of the water in the enclosure.
During one-clutch seasons, the progeny produced the 4 largest clutches (1 with 11 eggs and 3 with 10 eggs). The largest clutches produced during two-clutch seasons were 9 eggs (1 season) and 8 eggs (3 seasons). In progeny two-clutch seasons, the first clutch did not differ significantly in size from the second. In two-clutch seasons, progeny egg mass in 7 of 8 first clutches was larger than in the second clutches from the same season. However, the trend was not statistically significant.
Upon achieving maturity, progeny egg size was small in their primiparous clutches (mean egg sizes of 29.4–39.4 g) and tended to remain small in the next few clutches (Fig. 3). In 3 of the 4 retained progeny, production of eggs more nearly as heavy as Founder's eggs (typically > 41 g) occurred only after three to four seasons. Thus, during this prolonged early period of reproduction, each progeny produced 3 to 5 clutches of small eggs. Progeny eggs weighing 50 g (common for Founder) first appeared after f4 seasons and in just 2 of the 4 retained progeny. Overall, the mean mass of progeny eggs was 38.6 ± 6.5 g (range 26.0–58.3 g, n = 158). The range of variation in progeny egg masses was much larger than for Founder (variances of 42.88 vs. 28.16, respectively). Estimated densities of progeny eggs averaged 1.187 ± 0.052 (range 1.014–1.288, n = 156), which lies within the density range determined through water displacement.
Citation: Chelonian Conservation and Biology 5, 1; 10.2744/1071-8443(2006)5[165:RTICHG]2.0.CO;2
Excluding the primiparous clutches from each progeny (clutches were atypically small), an inverse correlation was present in a contrast of mean clutch size with mean clutch egg mass (without adjustment for body size: Speaman's rho = −0.62, p = 0.0038, n = 23). During this period, the variation across female masses was slight, deviating only up to ±295 g from a mean of 4664 g. For the entire CZ series, the egg size-clutch size tradeoff with adjustment for female mass was strong (r = −0.498, p < 0.0001), and remained compelling without adjusting for female mass (Spearman's rho = −0.438, p = 0.0004, n = 60 clutches). The association prevailed because the progeny produced relatively large clutches of small eggs (6.1 eggs; 41.6 g, respectively, n = 29 clutches) and Founder averaged smaller clutches of larger eggs (5.6 eggs; 50.5 g, respectively, n = 32 clutches).
Progeny RCMs averaged quite small for each of their primiparous clutches (mean 0.039, range 0.020–0.055, n = 6). Over the next two seasons, progeny seasonal RCMs increased to equal those of Founder (Fig. 4). Excluding primiparous clutches, progeny RCMs across one-clutch seasons averaged larger than RCMs of either clutch taken singly across two-clutch seasons (0.068 ± 0.011, range 0.052–0.083, n = 7; vs. 0.054 ± 0.014, range 0.032–0.071, n = 14), but below statistical significance (Mann-Whitney U, p = 0.08). Progeny seasonal RCMs in two-clutch seasons averaged 0.113 ± 0.019, range 0.083–0.129). The RCM's of H. grandis are well within the range given for small samples of geoemydid turtles (Moll and Moll 1990) and for large southeastern USA emydids (Jackson 1988).
Citation: Chelonian Conservation and Biology 5, 1; 10.2744/1071-8443(2006)5[165:RTICHG]2.0.CO;2
An Additional Female
Another female (mass, 3599 g; 282 mm CL) maintained at IU, independently from the CZ females, produced only one clutch per season (2 eggs, 25 March 1976; 3 eggs, 7 August 1990; 3 eggs, 2 October 1991; 3 eggs, 30 January 1993). The first 2 clutches, while seemingly laid out-of-season, were also problematic (infertile or over-mature eggs). Egg size increased from 33.2–33.9 g in the first clutch to 52.9–61.0 g in the last 3 clutches. With this female and the CZ females included, egg mass across all females ranged from 26.0 g (47 × 30 mm, from a progeny) to 61.0 g (69 × 37 mm, from this non-CZ female).
Hatchlings
Hatchlings were produced following incubation at 3 constant temperatures. The embryos developed directly (no post-ovipositional developmental arrest) and took the following times to pip: 25°C, 142.0 ± 11.3 days (range 133–158 days, n = 4); 28°C, 102.3 ± 4.5 days (range 94–106 days, n = 6); 30°C, 89.4 ± 4.6 days (range 85–98 days, n = 6). The live masses of 33 hatchlings (with internalized yolk sacs) from eggs of known weight averaged 31.5 ± 5.4 g (range 20.4–40.9 g) and were strongly isometric (r = 0.92) as a proportion (65.6%) of the fresh masses of their eggs. These hatchlings measured 52.7 ± 3.9 mm CL (range 45.2–60.2 mm) and 48.1 ± 3.6 mm PL (range 41–55 mm).
Across all females, the difference in egg size was 2.34-fold (2.27-fold for just Founder and her progeny; range 26–59 g). The range in masses of eggs that realized hatchlings was 28.2–61.0 g (2.2-fold). These ratios support the general observation that viable eggs within turtle populations tend to range 2- to 3-fold in mass (Ewert 1979).
Conclusions
In considering variation in size as an adaptive feature of eggs, such as in bet-hedging (Kaplan and Cooper 1984), the egg mass variance for Founder and her progeny was 63.1 (cv = 18.0, n = 339 eggs). This value included a variance of 21.8 for Founder alone (cv = 9.4, n = 169 eggs), and 41.1 for the progeny alone (cv = 16.6, n = 170 eggs). Thus, although there was ample variation in egg size overall, the F1 progeny were responsible for most of it (Fig. 3). As shown above, much of the variation was nonrandom. The egg size-clutch size tradeoff was significant for just Founder, just her progeny, and overall. This tradeoff may be compatible with bet-hedging. However, the strong within-season trend for producing a few large eggs initially and several smaller eggs later (Fig. 2) contributed to this tradeoff, and may have little to do with bet-hedging. It is likely that this within-season decline in egg mass was a physiological consequence of the annual yolking of ovarian follicles for both clutches. The large change in progeny egg mass was age-related (Fig. 3) and thus non-random. This age-related increase may result from decreasing investment in somatic growth relative to investment in reproduction (Congdon et al. 2003), as expressed in seasonal RCM (Fig. 4). Residual variation in egg mass that remained unexplained and potentially attributable to bet-hedging could include the large decline in mean egg mass that occurred early in the second half of Founder's reproductive record (Fig. 2).
Heosemys grandis, seasonal records of clutch size during the life span of one individual, Founder Female (Mother). Open circles labeled 1 and 2 and connected with a solid line represent first and second clutches, respectively, within a given season. Solid circles labeled 1 represent seasons with a single clutch. The record follows continuously from 1 November 1977. Reproduction occurred every year into 1993, but did not occur during the subsequent three seasons.
Heosemys grandis, seasonal records of clutch mean egg size during the life span of one individual, Founder Female (Mother). Symbols and other formats are as in Fig. 1.
Heosemys grandis, continuous record of clutch mean egg size of Founder Female (Mother, solid circles) and F1 Progeny (open circles). The record follows continuously from 1 November 1977. The progeny hatched from eggs produced prior to day 1000 and began producing their own eggs when their own ages were 5.7 to 7.7 years.
Heosemys grandis, continuous record of clutch mean egg size in Founder Female (Mother, solid circles) and F1 Progeny (open circles). The record commences with the first season in which an individual turtle produced a clutch and is continuous except for one season noted with a question mark. Note: Founder was >15 years old at her first recorded oviposition; the F1 progeny were <8 years old at their first recorded oviposition.
