Embryonic development of turtles highly depends on environmental factors such as temperature and moisture (Deeming and Ferguson 1991; Packard 1999). Numerous studies have found that incubation temperature profoundly affects the developmental rate and energy mobilization of embryos (Choo and Chou 1987; Gutzke and Packard 1987; Packard and Packard 1988; Rhen and Lang 1999a; Du and Ji 2001), as well as the phenotypic traits of hatchlings, including morphology, behavioral performances, growth, and gender (Brooks et al. 1991; Janzen and Paukstis 1991; Janzen 1993; Rhen and Lang 1995; Shine 1999; Janzen and Morjan 2002; Du and Ji 2003; Ji et al. 2003; Freedberg et al. 2004). Constant temperature incubations have indicated that extremely high or low temperatures may be harmful or even lethal for reptilian embryos, whereas moderate temperature facilitates eggs to produce relatively larger hatchlings that performed better than hatchlings produced at high or low temperatures (Packard and Packard 1988; Deeming and Ferguson 1991). More recently, both constant temperatures and fluctuating incubation temperatures have been incorporated into experiments that explored the effects of temperatures on turtle eggs and hatchlings; such studies indicated that the fluctuating temperatures could also affect hatchling traits (Doody 1999; Kolbe and Janzen 2001; Ashmore and Janzen 2003).

Although the effects of temperatures on hatchling traits have been well published for turtles, there is significant geographic bias in the scientific literature on this topic due to historical reasons. For example, compared with the abundant outcome from North American and European countries, studies on Asian faunas are extremely scarce. Currently, many species of Asian turtles are highly threatened by habitat loss and human exploitation (Altherr and Freyer 2000). For the conservation of these species, enhancing the survival of embryos and hatchlings is an important measure. Unfortunately, information on egg incubation is available for only a few Asian species (Ewert 1985; Choo and Chou 1987; Du and Ji 2003), and the influences of temperatures on eggs and hatchlings remain unknown in most of these turtles. Thus, research on the effects of temperatures during egg incubation on hatchling traits will contribute significantly to our knowledge of these species and shed light on management of their wild populations. The Chinese three-keeled pond turtle (Chinemys reevesii) is a species exposed to overexploitation.

In this study, we incubated C. reevesii eggs at 4 constant temperatures ranging from 24° to 33°C to assess the influence of incubation temperatures on hatching success, morphology, locomotor performance, and critical thermal minima of hatchlings. Our goals were to determine 1) what incubation temperatures yield the best quality hatchlings, and 2) whether incubation temperatures affect cold tolerance of the hatchlings. Thus, our research has important implications for conservation of this species.

Methods

The Chinese three-keeled pond turtle is a widely distributed aquatic emydid found in the central and southern provinces of China and southeastern Asia (Zhao and Adler 1993). From late May until early August, female C. reevesii produces multiple clutches, with a mean clutch size of 5.1 eggs (Liu et al. 1988). Incubation temperatures can significantly affect developmental rate, oxygen consumption (Wang et al. 1990), and sex (Hou 1985) of C. reevesii embryos. In contrast, substrate water potential has little effect on hatching success and hatchling traits for this species (Du and Zheng 2004).

In mid-July 2001, 136 freshly laid eggs from an unknown number of females were collected from a private hatchery at Shanghai, eastern China. All eggs were transported to our laboratory at Hangzhou Normal College, where they were measured (maximum and minimum diameter to the nearest 0.1 mm) and weighed (to the nearest 1 mg). A white patch on the shell surface indicated that eggs were fertilized and viable. Viable eggs (n = 111) were randomly assigned to 4 incubation treatments at temperatures of 24°, 27°, 30°, and 33°C. Eggs were placed in plastic boxes containing moist vermiculite (Du and Zheng 2004). The white patch of the shell was always kept upward. Boxes were covered with a perforated plastic membrane to retard water loss, and then placed into LRH-250G incubators (Guangdong Medical Instruments, China). We moved the boxes among shelves daily according to a predetermined schedule to minimize any effects of thermal gradients inside the incubators. We maintained the water potential of the substrate constant by adding water every other day.

Due to logistic reasons, we did not replicate our experiments at each temperature with different incubators. This shortcoming could have exposed our results to putative incubator effects (Hurlbert 1984). We reduced potential problems in our experimental design by calibrating all incubators with a standard thermometer (Du and Ji 2003) prior to the experiment. The 136 eggs came from about 27 clutches because the average clutch size is 5 eggs in this species. The relative large number of clutch origins and random assignment of eggs among the temperatures would largely reduce any clutch effects in this study.

Upon hatching, each hatchling was weighed and measured (carapace length, carapace width, limb length). After measuring, we assessed crawling capacity of the hatchlings by chasing them along a cycle racetrack with a 0.6 m diameter. The surface of the racetrack was covered with 5-mm–thick sand to mimic the natural substrate on which the turtles run after emerging from nests. On the second day after hatching, swimming capacity of the hatchlings was assessed by chasing them along a 1-m–long direct racetrack filled with 50-mm–deep water. Because locomotor performance is highly sensitive to body temperature in reptiles, we conducted these tests at a constant temperature of 30°C. Locomotor performance of each turtle was tested twice with a half hour resting period between trials and recorded with a Panasonic NV-MX3 digital video camera. Videotapes were then examined. For crawling performance, we scored crawling speed in the fastest 100-mm intervals and maximal crawling distance in 30 seconds. For swimming performance, we scored swimming speed in the fastest 100-mm intervals and swimming time for 800-mm distance. After the locomotor performance test, we determined critical thermal minima of the hatchlings in an LRH-250G incubator. Hatchling turtles were cooled from 28°C at a rate of 0.25°C per minute. The temperatures associated with a transient loss of righting response (test animals not responding to intense stimulation and losing their righting reflex when turned over) were defined as critical thermal minima.

We used the G-test to analyze the effect of temperature on hatching success, and one-way analysis of variance (ANOVA) and analysis of covariance (ANCOVA) to analyze the effect of temperature on hatchling traits. Homogeneity of slopes was checked prior to testing for differences in adjusted means. Tukey's test was used for multiple comparisons.

Results

The mean mass of eggs used in this experiment was 6.97 ± 0.099 g (range: 4.652–10.998), the maximum egg diameter was 32.8 ± 0.28 mm (range: 26.82–46.20), and the minimum egg diameter was 18.9 ± 0.10 mm (range: 16.11–21.23). Of the 111 incubated eggs, 94 hatched (84.7%). Hatching success was independent of incubation temperatures (G = 6.51, df = 3, p > 0.05). Incubation length was not affected by initial egg mass (all p > 0.05), but varied considerably among different temperature treatments (ANOVA- F_{3, 90} = 676.59, p < 0.0001). Eggs incubated at warm temperatures had a shorter incubation period than eggs at cool temperatures. For every 3° increase in temperature from 24° to 33°C, the incubation length decreased on average of 26, 5.3, and 3.2 days, respectively (Table 1).

Table 1. Incubation length and hatching success of Chinemys reevesii eggs incubated in different thermal environments. Values for incubation length are expressed as mean ± SE.

View Fullscreen

Incubation temperature affected hatchling body mass and carapace size (length and width). Hatchlings from eggs incubated at 27° were heavier than those from 33°C, while hatchlings from eggs incubated at 24° and 27°C had larger carapaces than did their counterparts from 30° and 33°C (Table 2). Incubation temperatures also affected limb length of the hatchlings, with hatchlings from eggs incubated at 24°C having longer forelimbs than did those from the other 3 temperatures (Table 2).

Table 2. Body mass, carapace size, and limb length of hatchlings from eggs incubated at 4 different temperatures for Chinemys reevesii. Analysis of covariance was used to analyze the effects of incubation temperatures on hatchling traits. Initial egg mass was used as the covariate in body mass and carapace size, whereas hatchling carapace length was used as the covariate in limb length. Data are expressed as adjusted mean ± SE. Adjusted means with different superscripts on each line are statistically different (Tukey's test). Asterisks immediately after F values represent significant level, * p < 0.05, ** p < 0.01, *** p < 0.0001.

View Fullscreen

During the locomotor experiment, 27 animals (10/24 from 24°, 3/23 from 27°, 3/22 from 30°, 11/22 from 33°C) did not move during crawling and swimming tests. These hatchlings were then excluded from the subsequent statistical analyses of locomotor capacity. Because locomotor capacity was independent of hatchling sizes (both carapace length and body mass) for all treatments (all p > 0.05), we used ANOVA to analyze the effect of temperature on hatchling locomotor capacity. The analysis showed that incubation thermal environments significantly affected maximal crawling distance (F_3,35 = 4.41, p < 0.01) (Fig. 1A), swimming time (F_3,60 = 14.69, p = 0) (Fig. 1C), and swimming speed (F_3,60 = 6.58, p < 0.001) (Fig. 1D), but not crawling speed (F_3,35 = 1.76, p = 0.17) (Fig. 1B). Overall, hatchlings from 27° and 30°C had better locomotor performance than those from 24° and 33°C.

View Figure

• Download as Powerpoint
• Download Figure

Incubation temperature significantly influenced cold tolerance of the hatchlings. The critical thermal minima were lower for hatchlings from eggs incubated at 24° and 27°C than that for hatchlings from 30° and 33°C (ANOVA- F_3,85 =  4.36, p < 0.01) (Fig. 2).

View Figure

• Download as Powerpoint
• Download Figure

Discussion

Overall, the current and previous studies on C. reevesii indicate that incubation temperatures significantly affect incubation length, hatching success, size, locomotor performance, cold tolerance, and gender of hatchlings. Hatchlings from relatively low temperatures (24° and 27°C) had larger body size than those from high temperatures (30° and 33°C) (Table 2), whereas hatchlings from moderate temperatures (27° and 30°C) had better locomotor performance than their counterparts from extremely low and high temperatures (24° and 33°C) (Fig. 1). These results are largely consistent with those from previous studies in turtles with genotypic sex determination (GSD) or temperature-dependent sex determination (TSD) (Janzen 1993; Hewavisenthi and Parmenter 2001; Du and Ji 2003). The smaller hatchlings at high temperatures might be related to the higher metabolic expenditure by these embryos and less energy transference from yolk to body (Booth and Astill 2001; Du and Ji 2001). The decreased locomotor performance of hatchlings incubated at extreme temperatures may reflect an adverse effect on the physiological and behavioral performance of the turtles.

Incubation temperature can influence cold tolerance of hatchling C. reevesii. Hatchlings derived from relatively low temperatures tolerated lower temperatures than did individuals from relatively high temperatures. Wild hatchlings of C. reevesii overwinter in the nest and emerge in March or April (Fukada 1965). The ability to tolerate cold temperature directly affects the first winter survival of this species and other reptilian hatchlings (St. Clair and Gregory 1990; Costanzo et al. 2004). Our results suggest that incubation temperature might affect winter survival of hatchlings, although understanding the ecological importance of temperature effects on cold tolerance of hatchlings depends on data on winter temperatures in field nests as well as the acclimatization of C. reevesii. Nonetheless, our study provides evidence that incubation temperature can affect thermal physiology of reptilian hatchlings. Several recent studies have also indicated that incubation temperature can modulate posthatching thermoregulation behavior in young reptiles (Chelydra serpentina, O'Steen 1998; Rhen and Lang 1999b; Paroedura pictus, Blumberg et al. 2002).

Because C. reevesii is a TSD species, with hatchlings from 27° and downwards being males and 32°C being females (Hou 1985), the temperature alone cannot entirely explain the above-mentioned influences on hatchling traits. The sex of hatchlings may also partly account for the phenotypic variation in hatchling traits. Unfortunately, this study could not tease apart the effects of temperature and sex on hatchling traits. To identify the temperature or sex source of variation in hatchling phenotype, we need to run incubation experiments that use aromatase inhibitor and/or estrogen to manipulate the sex of embryos (Rhen and Lang 1995; Crews 1996; Rhen and Crews 1999). Findings from such studies would provide insight into the adaptive significance of TSD in reptiles. Nonetheless, the outcome of this study and Hou (1985) would urge us to consider both temperature and sex effects when making practical management decisions for the conservation of this species.

Given that human exploitation of turtles is a tradition in China, turtle husbandry is believed to benefit both wild population conservation and the local economy. Currently, some species of Chinese turtles (i.e., Pelodiscus sinensis and C. reevesii) have been widely farmed by local people for food. Nevertheless, more work is needed to improve husbandry practices. For example, Chinese farmers usually incubate turtle eggs in relative high temperatures (around 30°C) to shorten the incubation period, and thereby reduce the cost of temperature control during artificial incubation. However, our studies on Chinese turtles indicate that medium temperatures produce large hatchlings with improved performance and high growth rate. Hatchling P. sinensis from temperatures around 27°C are better in locomotor performance and posthatching growth than individuals from higher temperatures (Du and Ji 2003; Ji et al. 2003). This study demonstrates that, for constant temperature incubation, 27°C should also be the optimal temperature that produces the best quality hatchlings of C. reevesii. However, unlike P. sinensis, in addition to morphological and behavioral traits, the sex of hatchlings in C. reevesii is determined by incubation temperature. Eggs incubated at 27°C only produced males (Hou 1985), and thus is not the only temperature that should be recommended for egg incubation in this species. Practically, we conclude that temperatures between 27° and 30°C are optimal for egg incubation in C. reevesii. Within this temperature range, hatchlings are relatively large and demonstrate good locomotor performance; and the sex of hatchlings can be chosen artificially as needed by changing incubation temperature. For instance, if a male-biased population is needed, the eggs should be incubated at temperatures close to 27°, whereas the eggs should be incubated at temperatures close to 30°C if a farmer wants to produce more female turtles.