Egg morphometric data are important for the study of questions relating to freshwater turtle life history, such as temperature-dependent sex determination (Roosenburg 1996), optimal egg size theory (Congdon and Gibbons 1985), and the relationship between hatchling size and survival (Brooks et al. 1991; Bobyn and Brooks 1994; Congdon et al. 1999; Packard et al. 1999; Janzen et al. 2000; Packard and Packard 2001). However, freshwater turtles have a cryptic reproductive cycle in which mating occurs in water (Ernst et al. 1994) and eggs are oviposited in terrestrial nests during a short period in early summer. Methods that do not require nest excavation, such as x-radiation or ultrasound, have been used to estimate clutch size and egg size (Gibbons and Greene 1979; Hinton et al. 1997; Kuchling 1998), but these methods cannot be used to measure egg mass and are prone to inaccurate measurements of egg size because they provide only 2-dimensional views of 3-dimensional eggs. Kuchling (1998) further noted that the number of oviductal eggs, measured with x-radiation or ultrasound, may not be an accurate measure of fecundity, as some eggs may be retained after laying and subsequently discarded in water. Excavating turtle nests immediately after oviposition is thus a preferred method to obtain fecundity data and accurate egg morphometrics.

Handling of turtle eggs has been criticized because it may have adverse effects on the survival of the embryos: Chrysemys picta, Trachemys scripta, and Terrapene carolina (Drajeske 1974; Simon 1975; Ewert 1979); Chelonia mydas (Parmenter 1980); C. picta, Malaclemys terrapin, and Chelydra serpentina (Feldman 1983; Ewert 1984); Dermochelys coriacea (Chan 1989); and Pelodiscus sinensis (Chou and Choo 1995). Evidence of the acceptance of the hypothesis that handling eggs has deleterious effects can be found in recent literature (e.g., Bjurlin and Bissonette 2004): “Because jarring and reorientation of eggs may affect embryo survival (Ewert 1979), we did not manipulate or uniquely mark eggs prior to incubation.” Previous work on effects of handling and manipulating eggs has been conducted in laboratories under artificial incubation conditions and cannot be extrapolated with confidence to wild nests. Additionally, eggs in previous studies were exposed to levels and types of handling (e.g., single 90° rotations at varying times throughout incubation [Chou and Choo 1995]) that were not representative of the manipulations performed by researchers collecting morphometric data. And, even though Drajeske (1974), Feldman (1983), and Ewert (1984) found no effect of rotation on painted turtle eggs, Ewert (1984) stated that in painted turtle eggs, adhesion of the vitelline membrane to the egg (the basis for movement sensitivity) sometimes occurs as soon as the first hour following laying.

We assessed the effect of egg handling on the survival of the embryos in a population of midland painted turtle (Chrysemys picta marginata) where, from 1983 until the present study, all observed nests were excavated and the eggs measured and reburied immediately following oviposition.

Methods

A long-term mark–recapture study of midland painted turtles has been conducted at Wolf Howl Pond (WHP) and West Rose Lake (WRL) which are less than 1000 m apart in Algonquin Park, Ontario (45°34′N, 78°41′W), Canada, for 21 summers between 1978 and 2003, and a study of reproductive ecology was initiated in 1983. The probability of observing most nesting females is very high because this population is located in a coniferous habitat that has very few suitable nesting sites. More precisely, all nests are found on a sandy railway embankment (now a popular hiking trail) a few meters wide and approximately 250 m long at WHP and 750 m long at WRL. A detailed description of the sites can be found elsewhere (Schwarzkopf and Brooks 1985).

The protocol for nest and egg handling was the same in all years. Upon observation of a nesting female, the identity of the female was recorded along with environmental variables, nest location, and time. We only used the first nests of the year—approximately 10% of the females will lay a second nest in a season (Samson 2003)—to minimize biases in survival rates arising from the date of oviposition. In 2002, we monitored 13 handled nests and 33 nonhandled nests at WRL. In 2003, we monitored 38 handled nests and 40 nonhandled nests at WHP. In 2004, we monitored 18 handled nests and 25 nonhandled nests at WHP. The differences in sample sizes and sites reflect the availability of clutches with respect to other ongoing research projects in this population. Handled nests were treated as follows. The eggs were carefully removed from the nest without recording their original orientations and pencil-marked according to female ID and egg order (e.g., A02 1, meaning the first egg to be collected from the nest of female A02). Care was taken to retain the original nest shape as much as possible. The eggs were then brought back to the Algonquin Park Wildlife Research Station in aerated plastic containers filled with moist vermiculite. Egg width, length, and mass were measured and, within 24 hours, the eggs were reburied in their original nests in the same orientation given when first placed in the vermiculite. To protect the eggs from mammalian predators, we covered each nest with a hardware cloth box (approx. 12 × 12 × 6 cm) secured with 2 metal spikes (Fig. 1) (Hughes 2003). Nonhandled nests were simply located and covered with a hardware cloth box. It was essential to protect the nests from mammalian predators because more than 90% of nests were depredated throughout the long-term study period (Samson 2003). All nests were monitored daily for signs of depredation or soil erosion from oviposition until mid-July, and then monitored weekly until early September. Additional sand was added to hide the top of the nest protectors as required.

View Figure

• Download as Powerpoint
• Download Figure

The handling of the eggs in this study can be described as intensive compared to previously published studies where simple rotations were performed; however, the duration of the handling was restricted to the first 24 hours postoviposition as compared to handling throughout incubation (Marcellini and Davis 1982; Feldman 1983; Chou and Choo 1995). All eggs sampled in the population since 1983 have undergone the following steps: 1) excavation from the nest and random-orientation placement in a moist vermiculite-filled box, 2) 10–20 minutes' walk to a car, 3) 30 minutes of driving on gravel roads and highway, 4) cleaning of the eggs to remove sand particles, 5) measurements of egg length, width, and mass, 6) repetition of steps 3 and 2, and 7) reburial in nest chamber in the same orientation as when placed in vermiculite.

Hatchling painted turtles do not normally emerge from the nest in the fall following hatching, but remain in the nest over winter (Ewert 1979). Hence, we excavated the nests in the springs of 2003, 2004, and 2005 as soon as the ground thawed (late April) and brought all hatchlings, eggs, and eggshells back to the Wildlife Research Station. We categorized the fate of each egg as 1) unfertilized (no evidence of an embryo), 2) empty (depredated by insects or roots or representing a hatchling that escaped from the nest protector), 3) not fully developed, 4) dead hatchling (hatched successfully, but failed to overwinter successfully), or 5) live hatchling (survived until excavated in April of the following year). These categories were created for a concurrent project (Hughes 2003). To test our hypothesis, we compared the “not fully developed” category (3) with the “dead hatchling” plus “live hatchling” categories (4 + 5) and also compared the “not fully developed” category with the “live hatchling” category. Hence, in this paper “embryo survival” will comprise “survival to hatch” and “survival to emergence.” Survival rates were estimated by dividing the number of individuals that hatched or emerged by the clutch size. Comparisons of embryo survival rates between treatments were made by t-tests adjusted for unequal sample size after performing an arcsine transformation on the data to normalize them.

Results

We could not estimate the survival rates of a few nests because the eggs were removed from the nests at an early stage, usually by predators. Seven nonhandled nests were depredated in 2002, 7 handled nests and 3 nonhandled nests were depredated in 2003, and 8 handled nests and 3 nonhandled nests were depredated in 2004. A moose trampled one nest protector, and curious hikers removed 21 nest protectors (over all 3 years) that had been exposed by red foxes (Vulpes vulpes), raccoons (Procyon lotor), or soil erosion. The majority of these unprotected nests were subsequently depredated. These depredations resulted in sample sizes in 2002 of 13 handled nests and 26 nonhandled nests; in 2003, 31 handled nests and 37 nonhandled nests; and in 2004, 15 handled nests and 17 nonhandled nests.

We did not know the clutch size of nonhandled nests because they were not dug up at the time of oviposition; hence we estimated clutch size following excavation in the spring. We also estimated the clutch size of handled nests following excavation in the spring and compared the true clutch size with the spring estimate to assess if the spring clutch sizes were underestimated. In 2002, a total of 7 eggs were not accounted for in the spring excavation of handled nests, representing an underestimation of 7.7% in clutch size, 24 eggs were not accounted for in 2003 (10.8%) and 25 eggs were not accounted for in 2004 (17.7%). We calculated embryo survival (Table 1) based on the clutch size estimated during the spring sample so that the comparison between treatments was unbiased, assuming that the clutch size underestimation was the same for both treatments. We excluded “empty” eggs from the analysis because we could not determine if the eggs were empty because the embryo had failed to develop, because the egg was depredated, or because the hatchling had successfully escaped from the nest protector.

Table 1. Mean and standard deviations of the survival of embryos to hatch (Φ_h) and to emergence (Φ_e) in handled and nonhandled nests of midland painted turtles.^a

View Fullscreen

The average nest dates in 2002, 2003, and 2004 were 22 June (18 June–3 July; SD = 2.6 days), 16 June (7–26 June; SD = 4.2 days), and 18 June (11 June–4 July; SD = 6.7 days), respectively. The coefficient of determination between nest Julian date and embryo survival was approximately 3%, which suggests that the date of oviposition of the monitored nests had a negligible effect on survival.

Embryo survival to hatch did not differ between nonhandled nests and handled nests in either year (Table 2). In 2002, the difference was very large numerically; however, it became statistically insignificant when uneven sample sizes were taken into account. Embryo survival to emergence did not differ between nonhandled nests and handled nests in any year (Table 2). In addition, the standard deviations of the survival rates were much larger than the difference in average survival rates, suggesting extremely high variability in survival rates regardless of treatments.

Table 2. t-tests comparing survival to hatch (Φ_h) and survival to emergence (Φ_e) in handled and nonhandled nests.^a

View Fullscreen

Discussion

Handling freshly laid midland painted turtle eggs does not have a deleterious impact on embryo survival. Moreover, the large standard deviations depict the high stochasticity observed in this population regarding embryo survival (Samson 2003). These results indicate that the effect of egg handling on embryo survival is very small and difficult to detect across the noisy environment in which such a field experiment must be done. The data presented in this study provide field support for laboratory-based studies showing that handling has no effect on the survival of painted turtle eggs (Drajeske 1974; Feldman 1983; Ewert 1984), despite the fact that the handling occurred after the vitelline membrane had attached to the egg, which sometimes occurs within one hour following oviposition (Ewert 1984). Nonetheless, our goal was not to estimate quantitatively the impact of handling eggs on embryo survival per se, but rather to assess qualitatively the costs of excavating turtle nests in the wild. Our results suggest that the benefits, in terms of obtaining accurate data on fecundity and egg morphometrics, far outweigh the very small potential increase in mortality that could arise from handling the eggs.

Many populations of freshwater turtles have precarious demographies because their life-history strategies are particularly sensitive to increased adult mortality (Brooks et al. 1991; Congdon et al. 1993, 1994). Management plans often rely on protecting eggs and/or headstarting hatchlings in artificial enclosures (Heppell et al. 1996; Seigel and Dodd 2000; Spinks et al. 2003) because the anthropogenic sources of adult mortality (e.g., road kill, poaching, alteration of habitat) are difficult to halt. We suggest that excavating freshly laid nests of turtles that lay small eggs does not significantly alter the recruitment of these freshwater turtles. Given special precautions taken regarding predation, variations in egg size and shell type and turtle species, measuring eggs shortly after oviposition should be beneficial for the management of turtle species because it provides crucial information on the fecundity of typically poorly known populations and because the eggs can subsequently be relocated, protected, or incubated for headstarting programs when needed.