Chersina angulata is endemic to southern Africa, ranging from southwestern Namibia south along the west coast of South Africa and along the south coast to Eastern Cape Province. The species occurs from the coastal plains, all along the escarpment, to altitudes of approximately 1200 m but does not penetrate far into the interior of southern Africa. Chersina angulata generally lives in regions with low annual rainfall (< 100–600 mm), but in some areas along the south coast rainfall can exceed 1000 mm. The west coast receives winter rains (Mediterranean climate) whereas the south has year-round rain (temperate climate). Depending on the region, summer temperatures are mild to very hot (Hofmeyr 2009).

The female reproductive cycle of C. angulata is unusual for turtles and other reptiles in Mediterranean to temperate climatic zones: females are reproductively active over a prolonged season during which they ovulate sequentially and oviposit multiple clutches usually containing a single egg, occasionally two. Ovulations can occur from late January to late November and eggs are deposited from March to December. Hatching in C. angulata occurs from March to April at the start of the rainy season. Consequently, early clutches of this species incubate for 12 mo whereas late clutches incubate for 3–4 mo (Hofmeyr 2004, 2009).

It is not known which mechanisms enable all C. angulata eggs to hatch in autumn no matter when they were laid. As a general rule for the order Testudines, egg development inside the female is arrested at the gastrula stage (preovipositional developmental arrest) until oviposition, when development recommences once eggs are deposited into the nest. Following oviposition, development can again be arrested at different stages: embryonic diapause arrests development prior to the main morphogenesis (somite formation); it arrests development during environmental conditions (e.g., temperatures) when embryonic development could normally proceed. For embryonic development to recommence, diapause has to be broken, usually triggered by specific environmental conditions which can be warming temperatures following cooler periods. After completion of embryonic morphogenesis, another mechanism that can delay hatching and prolong incubation is embryonic aestivation during which the metabolism of the embryo is depressed; it is a form of quiescence (Ewert 1985; Rafferty and Reina 2012). For C. angulata eggs laid in autumn, winter, and spring, the mechanism extending their incubation period could either be embryonic diapause, embryonic aestivation, or a combination of both. In this article we test the hypothesis that diapause is involved in the variable incubation durations of eggs from early to late C. angulata clutches (April–December) by evaluating development stages of eggs incubated under ambient temperatures in Cape Town, South Africa.

Recently it has also been demonstrated in the wild as well as in captivity that C. angulata females can retain eggs until embryonic development has progressed to the hatching stage (Hofmeyr and Kuchling 2017), conforming to the accepted definition of viviparity in reptiles (Shine 1985; Van Dyke et al. 2014); although the last clutches of the season are often oviposited in early summer (December), under extreme summer heat females can retain the eggs until the time hatching normally occurs in early autumn, then lay them without nesting on the ground, where they hatch in between a few days. This phenotypically plastic response to environmental extremes has the potential to buffer the developing embryos of the season's last clutch from overheating during summer heat waves: females can thermoregulate by moving among different microhabitats whereas, once the nest site is chosen, eggs of non-nest-guarding reptiles are passively exposed to hot ambient temperatures (Kuchling and Hofmeyr 2022). However, the mechanisms behind this unique reproductive lability of C. angulata are currently unknown.

The first step for a transition from oviparity to viviparity in reptiles is considered to be prolonged egg retention (Shine 1985; Blackburn 1999). Hofmeyr (2004) showed that the oviducal phase of C. angulata eggs can vary from 23 to 212 d without females becoming egg-bound, but the embryonic stages at oviposition following prolonged retention in the oviducts have not been investigated. If the constraints on intrauterine embryonic development in C. angulata are relaxed for the last clutch of the season, embryonic stages laid late in the season, in early summer, should be more advanced than the gastrula stage. We tested the hypothesis that eggs oviposited in late December are developmentally more advanced than the gastrula stage. This would constitute an intermediate stage postulated by life history theory for the evolution of viviparity in an oviparous group like the order Testudines (Williams 1992).

METHODS

Maintenance of the Captive Colony. — In 1999, 16 adult wild C. angulata females and 5 males were transferred from the West Coast National Park (WCNP; 33°13′48″S, 18°08′07″E) into a 20 × 10-m outdoor enclosure at Kuilsrivier (33°55′39″S, 18°41′08″E), 90 km from the WCNP. The enclosure contained Kikuyu grass (Pennisetum clandestinum) that provided food and shelter, bushes for shelter, artificial shelters, and sandy areas for nesting (Fig. 1). The animals received supplemental food (fresh vegetables, occasional fruits, and chicken eggshell) and drinking water. Each female was marked with a unique number on her posterior shell to allow identification when nesting. Over the years, offspring of the original group were placed in a similar but separate enclosure. In late 2016, the females were temporarily moved into another, structurally similar compound in front of the kitchen window to allow continuous closer observation of their behavior.

View Figure

• Download as Powerpoint
• Download Figure

Egg Incubation and Egg Candling. — M.D.H. collected 84 freshly laid eggs in the captive C. angulata colony from April to December 2018 and penciled laying date and ID of the mother on the upper side of their shells. Eggs were incubated half-covered in dry soil in open plastic containers in the family kitchen at ambient room temperature in a dark corner away from windows (Fig. 2). Temperatures were not recorded, but the kitchen was neither heated nor air-conditioned, so temperatures followed the seasonal cycles at Cape Town. All eggs were candled with a torch light on 27 November and 23 December 2018 and on 1 and 17 January 2019. The appearance of egg content, embryos, and blood vessels of embryonic membranes was noted and respective photos of developing eggs were taken on 1 and 17 January 2019. One egg laid on 28 December 2018 was, in addition, candled and photographed in-between 24 hrs of oviposition on 29 December 2018.

View Figure

• Download as Powerpoint
• Download Figure

G.K. estimated embryonic stages from the descriptions as well as the photos of candled eggs. Embryo shapes and sizes and the appearance of blood vessels of extraembryonic membranes of candled eggs were compared with 1) the 26 developmental stages described in detail for embryos of Testudo hermanni boettgeri (Guyot et al. 1994), which conform to the 26 embryonic stages described by Yntema (1968) for Chelydra serpentina; and 2) Ewert's (1985) description of the developmental chronology of extraembryonic membranes and their blood circulation for turtles generally as it relates to the embryonic stages of Yntema (1968). Developmental stages were classified (Fig. 3) based on the following criteria. 1) Prior to Yntema stage 5: yolk sedimented and/or banding or air bubble visible. 2) Yntema stages 5–10: red spots or red rim indicates blood islands in yolk sac membrane and start of vitelline circulation; embryo, if visible, thin and straight to slightly bent. 3) Yntema stages 11–13: terminal sinus of the yolk sack circulation and the anterior vitelline vein discernible; allantoic veins raise under the shell; the embryo starts to have a curled appearance, the eyes commence to become pigmented. 4) Yntema stages 14–16: allantois forms well-vascularized lobes between shell and embryo, blood vessels under shell become prominent; head with dark eyespot; carapace usually not discernible or dorsally not clearly round. 5) Yntema stages 17+: embryo becomes large and prominent, carapace dorsally round and discernible; in later stages only large shadow of embryo and some thick blood vessels visible.

View Figure

• Download as Powerpoint
• Download Figure

RESULTS

Of the 85 artificially incubated eggs of 2018, 34 (40%) were classified through candling on or before 17 January 2019 as nondeveloping or with embryos having died early during development prior to Yntema stage 5, 6 (7.1%) were assessed as containing dead but slightly advanced embryos (approximately at Yntema stages 6–12), and 45 eggs (52.9%) were classified as containing developing embryos (Fig. 3).

There was no significant correlation between the percentage of live developing embryos by mid-January and the number of months since oviposition, suggesting length of artificial incubation had not affected the viability of embryos (Fig. 4). However, the trend of the regression suggests the nonsignificance could be due to our limited data set.

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

Embryonic Development of Eggs Laid in Autumn, Winter, and Spring. — By 27 November 2018, 31.6% (n = 6) of the developing eggs laid from April to August had not yet reached Yntema stage 5, 57.9% (n = 11) were at Yntema stages 5–10, and 10.5% (n = 2) had reached Yntema stages 11–13. Of the eggs laid in September and October, 94% (n = 17) were at Yntema stages 5–10 and 1 (6%) was at stages 11–13; 100% (n = 7) of the eggs laid in November had not yet reached Yntema stage 5. By 23 December, all developing eggs laid from April to November were more advanced than Yntema stage 5, and all eggs kept advancing further in development by the candling dates of 1 January and 17 January 2019. By 17 January, all 19 developing eggs (100%) laid from April to August had large embryos beyond Yntema stage 16, as did 72% (n = 13) of eggs laid in October and November, with the other 28% (n = 5) were at Yntema stages 14–16. By 17 January, 100% (n = 7) of the eggs laid in November were also at Yntema stages 14–16 (Table 1).

Table 1. Number of eggs per month of oviposition at particular Yntema developmental-stage classes during candling from late November 2018 to mid-January 2019.

View Fullscreen

The development pattern from late November to mid-January demonstrates that all eggs laid from April to about mid-October underwent a period of embryonic diapause (developmental arrest) prior to resuming development from late October to late November when diapause is terminated. The variability of developmental stages of eggs laid in spring from late October to late November indicate that some resume development after a very short period of diapause and some appear to proceed with development without undergoing any diapause (Table 1).

Developmental Stages of the Egg Laid in Summer. — Only 1 egg was laid in early summer, on 28 December 2018. When candled on 29 December 2018, the embryo was at Yntema stage 12: the terminal sinus of the yolk sack circulation and the anterior vitelline vein were visible; the allantoic veins had raised and reached the shell; and the embryo was lying on its side and had a curled appearance. On 1 January 2019, it had reached stages 13–14: the vitelline circulation spread out further around the yolk; the chorioallantois extended below the embryo, with the allantoic blood vessels spreading out under the shell visibly clearer than the vitelline vessels which were deeper under the shell; and the embryo was curled up so that the head was opposite the tail and the eye pigmentation became discernible. On 17 January 2019, it had reached Yntema stages 17–18: head and eye were prominent and the carapace was discernible and round dorsally (Fig. 6).

View Figure

• Download as Powerpoint
• Download Figure

DISCUSSION

The developmental success of the artificially incubated C. angulata eggs laid during 2018, or the percent proceeding with development by mid-January 2019, was just over one half of the eggs (52.9%) in this study. This could be related to the relatively simple incubation setup under ambient environmental temperatures, which fluctuated roughly parallel with Cape Town's seasonal temperature variations, and to the lack of humidity control. Temperature conditions, humidity, and developmental or hatching success in wild nests of C. angulata are unknown. Artificial incubation of C. angulata eggs under more sophisticated and better-controlled conditions by breeders in Europe and the United States also appears to be challenging:

A great number of eggs will not develop at all, a few will start developing and then stop, and even fewer will actually hatch. Different incubator types, various substrates, different temperatures, and higher or lower humidity levels, incubation with or without cooling (diapause), or incubation with and without night temperature drop—nothing seems to help to bring the hatching rate up. Very dry conditions in the incubator (humidity only about 50%) or at the place where the egg had been laid in the enclosure, and a sudden increase in humidity at the end of the usual incubation period seems to cause higher hatch rates. (Pfau 2022, p. 36)

The fact that some eggs in our study did not develop or died during development appears not to be unusual for artificially incubated eggs of this species, and we assume it did not skew our results.

Diapause. — Under natural and seminatural conditions, egg incubation periods of turtles are not necessarily restricted to the time needed to complete development—they can also accommodate time to wait for conditions to become optimal for the survival of the hatchling. Embryonic diapause, cold torpor, and embryonic aestivation are mechanisms to prolong the free-living egg stage. Embryonic diapause arrests development prior to main morphogenesis (somite formation); it maintains developmental arrest during environmental conditions when embryonic development could normally proceed and synchronizes the timing for advanced embryonic development to progress to completion during the best possible environmental conditions (Ewert 1985).

Egg candling in this study revealed that, in all C. angulata eggs laid in autumn, winter, and early to mid-spring (April–October), no development beyond Yntema stage 5 could be ascertained by candling until late October. However, by late November the majority of eggs were beyond Yntema stage 5 (Table 1). Eggs laid during November reached this stage only about 4 wks later (23 December) and, by that time, embryonic development in all the eggs had progressed beyond Yntema stage 5. This suggests that eggs laid in autumn, winter, and early spring underwent diapause and resumed development by late October or early to mid-November once ambient temperature increased due to the seasonal warming trend. Embryonic development in most eggs laid in November progressed without diapause and in some probably after a very short diapause. A comparable pattern has been described for Kinosternon baurii in Florida: eggs go into diapause when laid in autumn, but eggs laid in spring initiate development immediately; both autumn and spring eggs hatch in the following autumn (Ewert and Wilson 1996). As in K. baurii, diapause in C. angulata is facultative—its occurrence depends on the season of oviposition: for all eggs laid from autumn to spring, this mechanism synchronizes advanced embryonic development to proceed over the summer months and hatching to occur in autumn.

Although it is known that C. angulata eggs hatch and hatchlings emerge in autumn (Hofmeyr 2009), the time-span over which hatching occurs has not been investigated in detail. We do not know if, in addition to diapause, embryonic aestivation may also play a role in the timing of hatching so that it coincides with favorable environmental conditions for hatchling survival. This remains an area for future research.

Intermediate Stages Between Oviparity and Viviparity. — Prolonged egg retention can be seen as the first step toward a transition from oviparity to viviparity (Blackburn 1999). Chersina angulata clearly satisfies this first requirement and can retain eggs for variable periods often, however, for longer in the wet/cold than in the dry/hot season (Hofmeyr 2004). It appears that females can complete the oviducal phase of eggs to full calcification in 23–25 d (the minimum periods to oviposition in 2 consecutive years, respectively), but retention in utero without further calcification could persist up to 212 d. We observed only 1 female that retained an over-calcified egg for 316 d (Hofmeyr and Kuchling 2023). However, prolonged egg retention on its own without progression of embryonic development will not lead to viviparity.

In all chelonians the oviducal period of eggs accommodates early embryonic development up to the gastrula stage. The generally accepted wisdom is that, no matter how long turtle females may retain eggs, intrauterine development of embryos is then arrested until oviposition. The description of pre ovipositional developmental arrest is based on egg incubation and embryological data of aquatic turtles of various families (Ewert 1985; Rafferty and Reina 2012; Williamson et al. 2017) for which this mechanism appears to be adaptive: the vast majority of aquatic turtle species nests on land and environmental vagaries, such as flash flooding, may temporarily prevent the use of prime nesting habitat. From the gastrula stage onward, usually following oviposition, development of eggs includes chalking: the vitelline sac expands and its upper part, the embryonic disc and adjacent vitelline membrane rise to the uppermost portion of the egg and adhere to the inner shell membrane. In turtle eggs with pliable shells, this adhesion determines subsequent embryonic orientation. Movement of eggs following adhesion could increase embryonic mortality if eggs were deposited in the nest with their embryos attached to the lower side, which is detrimental to some species (Ewert 1985). Preovipositional developmental arrest allows the embryo to wait at the gastrula stage until the female succeeds to nest; it could be a mechanism to avoid ovipositing eggs at a movement-sensitive embryonic stage, with the vitelline membrane already attached to the shell membrane. Preovipositional developmental arrest provides flexibility for aquatic turtle females to choose appropriate conditions for selecting nesting sites and avoiding predators (Ewert 1985). In susceptible species, movement-induced mortality may constrain the evolution of viviparity (Williamson et al. 2017).

However, through gradual slippage of the embryo between the adhering layers in eggs with brittle and hard shells, embryonic orientation can still change following chalking (Ewert 1985). Movement-induced mortality, thus, may be less of a problem for the brittle- or hard-shelled eggs of tortoises. To our knowledge, Testudo hermanni boettgeri is the only tortoise for which the embryonic development has been described from oviposition to hatching: in a captive colony kept under seminatural conditions at Gonfaron, France, eggs were at the gastrula stage at oviposition (Guyot et al. 1994), comparable to aquatic turtles. This suggests that preovipositional developmental arrest at the gastrula stage may also be the norm for the genus Testudo and could generally be the norm in the family Testudinidae.

Our observation of an embryo advanced to Yntema stage 12 in an egg oviposited in late December demonstrates that, in C. angulata, preovipositional developmental arrest is facultative—its occurrence appears to depend on the season. Although the lack of advanced development until October seen in candled eggs oviposited from April to October suggests preovipositional developmental arrest as well as diapause occurred in those eggs, at least for the last clutch of their long nesting season embryonic development can proceed in utero beyond the gastrula stage. Due to the lability of embryonic orientation following chalking in brittle-or hard-shelled eggs, laying of advanced embryos may be less constrained by movement-induced mortality for tortoises than it is for many aquatic turtles.

According to the egg incubation periods needed to complete development under natural or seminatural conditions, turtles can be roughly divided into rapid developers and slow developers. Although the total range of incubation periods for turtle eggs can extend from 28 d (Pelodiscus sinensis) to 420 d (Stigmochelys pardalis), eggs of rapid developers normally hatch after about 40–55 d of incubation at naturally relatively high temperatures (e.g., 30°C and above), whereas eggs of slow developers have highly variable incubation periods of, generally, over 75 d and at ambient temperatures of 24°C–30°C, often of about 150–250 d (Ewert 1985).

Based on incubation time, an over half-a-century-old report already suggests eggs of another tortoise species could also be laid at a more advanced embryonic stage than the gastrula stage: in an outdoor captive Geochelone elegans colony (1 female, 3 males) at Bhubaneswar, Odisha, India, 7 eggs of 1 clutch were incubated individually in glass jars, buried 6–8 cm deep in soil, and kept indoors in a cupboard at ambient temperatures. Two eggs hatched after 47 d of incubation, 1 after 54 d, and 4 after 122–147 d of incubation. The authors interpreted this phenomenon as bimodality, with quick-developing and slow-developing eggs in a single clutch (Jayakar and Spurway 1964, 1966). The incubation times of the 4 eggs that took longer to hatch were similar to those reported for G. elegans by breeders in Europe and the United States (86–147 d; review in https://startortoises.net/incubation.html, accessed 10 February 2022), but the first 3 eggs hatched after about one-half of the shortest incubation times reported since by other breeders of this species. Because the 45–54-d incubation period for part of a clutch of G. elegans reported by Jayakar and Spurway (1964, 1966) is shorter than any other incubation period known for the family Testudinidae, it “. . . seems to require an alternative explanation in place of single clutch bimodality” (Ewert 1985, p. 192). We agree with Ewert's (1985) assessment of Jayakar and Spurway's interpretation of their results, but Ewert (1985) did not propose any alternative explanation. We suggest that the most parsimonious interpretation is that a clutch of 3 eggs was retained by the female until a second clutch of 4 eggs was ovulated and shelled, and both clutches were then simultaneously oviposited into one nest, with the first 3 retained eggs at oviposition already more advanced in development stages than gastrulas and, thus, requiring shorter incubation periods to complete development and hatch than did the 4 other eggs. In addition to C. angulata, G. elegans could be another tortoise species with labile preovipositional developmental arrest.

Facultative abolishment of preovipositional developmental arrest and advanced embryonic development in utero, as demonstrated by the advanced embryo in the C. angulata egg oviposited on 28 December 2018 (Fig. 6), represents a transitional, intermediate stage postulated by life history theory for the evolution of viviparity (Williams 1992). According to Andrews and Mathies (2000, p. 234): “. . . once the ability to maintain gravidity is in place [prolonged egg retention in C. angulata—see above], the further shift to embryonic development in utero, and ultimately to viviparity, might be easier than if selection had to act on both traits simultaneously.” Thus the reproductive patterns of C. angulata make this species a prime candidate among turtles to study the evolution of extended intrauterine embryonic development and viviparity. There is no general agreement on how viviparity should be defined—it is open for interpretation if deposition of eggs with fully developed hatchlings (Hofmeyr and Kuchling 2017; Kuchling and Hofmeyr 2022) should be called viviparity. For example, Lodé (2012) proposed “oviparity” for lecithotrophic reproduction with most of embryonic development taking place outside the body of the female; “ovo-viviparity” for internal incubation, or a prolonged retention of eggs during embryonic development, but with no direct nutritive exchange between the parent and the embryo; “histotrophic viviparity” for the development of embryos inside the female's body with a steady supply of nutrients; and “hemotrophic viviparity” when the embryo draws continuous nourishment from the mother through a placenta or similar structure. Under this classification, the observed egg retention by C. angulata until completion of embryonic development would fall under “ovo-viviparity.” However, under definitions preferred by reptile researchers, viviparity occurs even if there is only physiological exchange of respiratory gases (not necessarily nutrients), as long as structures/mechanisms to facilitate respiratory gas exchange exist in the mother (Blackburn 1999; Van Dijk et al. 2014). Although we did not investigate if this is the case in C. angulata, we assume it is. However, if there are no such structures or mechanisms and the eggs literally develop with no help for gas exchange from the C. angulata mothers (something we consider unlikely), then under any definition, ovo-viviparity would be the most appropriate term for this phenomenon.

Contribution of Captive Colonies to Viviparity Research. — Many aspects of the transition from ovipary to vivipary are challenging and difficult to study in wild populations. Although our first observation of the hatching of a C. angulata egg in-between 24 hrs of oviposition took place in the wild population at Dassen Island in March 1999 (Kuchling and Hofmeyr 2022), in that case the hatchling had a microphthalmic condition (Fig. 7), suggesting the possibility of teratogenic influences during its development. This problem could have been a trade off for intrauterine development and was a reason we originally abstained from publishing our observation. The breakthrough for our understanding of facultative viviparity in C. angulata came when 2 eggs were oviposited in the captive colony on 31 March or 1 April 2016 and pipped in-between 5 d or less. Both hatchlings had emerged from their eggs by 9 April 2016. Although it could be argued that emerging from an egg 9 d after oviposition does not constitute viviparity (live birth) but rather prolonged egg retention, it is normal for C. angulata hatchlings to take 6–7 d from pipping to leaving the egg (Branch 2008). This period is, supposedly, the time needed to internalize the residual yolk. It can be argued that pipping is the relevant point in time to consider for the completion of embryonic development, as Ewert (1985) did by defining the incubation period as the time elapsed from oviposition to pipping of the eggshell as opposed to leaving it behind by the hatchling.

View Figure

• Download as Powerpoint
• Download Figure

It is notable that the only previous observation that suggests labile or facultative preovipositional developmental arrest in another tortoise species was also made in a captive colony (Jayakar and Spurway 1964, 1966), in a range country outdoor setup and egg incubation regime (ambient temperature) comparable to our C. angulata study. Because tortoises (family Testudinidae) are now bred in captivity around the world, it would be interesting for breeders to report any unusually short egg incubation times and to actually assess through candling embryonic stages at oviposition under various environmental conditions and stimuli rather than to assume oviposition always occurs at the gastrula stage. However, when females retain eggs longer than considered to be normal, breeders often fear the female may become egg bound and take her to a veterinarian (Pfau 2022). Oviposition may then be induced, e.g., by oxytocin injection. This could be one of the reasons why steps toward viviparity, such as reduced incubation periods, have so far not been reported by breeders of C. angulata. Other reasons, particularly when tortoises are kept indoors in cold climates, may include the avoidance of environmental extremes that appear to be the natural triggers for females to retain developing eggs, such as droughts and heat waves (Kuchling and Hofmeyr 2022). Reported egg incubation times to hatching of C. angulata in Europe and the United States are in the range of 90–135 d (Pfau 2022) and do not reflect the wide variation of egg incubation times in wild populations or in our seminatural outdoor captive colony inside the natural range of the species—from approximately 1 yr (Hofmeyr 2009) to 1 d (Kuchling and Hofmeyr 2022), with the latter corresponding to the widely accepted definition of viviparity in reptiles (Shine 1985; Van Dyke et al. 2014).

CONCLUSIONS AND OUTLOOK

The results of this study and a critical review of the literature challenge several long-held assumptions regarding reproductive traits of turtles, including that all turtles oviposit embryos at the gastrula stage and that assume a total lack of traits toward viviparity in chelonians. In addition, the unusual suite of reproductive strategies of C. angulata, including facultative preovipositional embryonic arrest (this study) and facultative viviparity (Hofmeyr and Kuchling 2017; Kuchling and Hofmeyr 2022), does not conform with several basic assumptions regarding the evolution of oviparity and viviparity in reptiles, including that the traits of oviparity or viviparity in reptiles are controlled by genetics rather than by phenotypic plasticity, i.e., changing a female's environment does not change her mode of reproduction (Shine 2014); and the reproductive mode of reptiles (oviparity or viviparity) is consistent among all female reptiles in a given population (Shine 2014).

Whittington et al. (2022) recently speculated that, as a bet-hedging strategy in variable environments, transitional phenotypes might persist in populations for long periods if, under certain environmental conditions, they are more advantageous than oviparity or viviparity. Because fitness costs of retaining egg clutches appear to be low for C. angulata compared with many other turtles and squamates (Kuchling and Hofmeyr 2022), further study of its transitional stages may offer an excellent prospect for understanding the major evolutionary shift between oviparity and viviparity. Our incubation results (this article) and previously reported observations (Hofmeyr and Kuchling 2017; Kuchling and Hofmeyr 2022) demonstrate what is possible, but we don't know yet how common these phenomena are within populations and among populations in different climatic zones or among the genetic subgroups of the species (Daniels et al. 2007; Spitzweg et al. 2020). The predictions and hypotheses based on our observations are testable both empirically and theoretically. More research is needed into the proximate causes and the mechanism underlying the unusual reproductive mode of C. angulata as well as on the ecological and evolutionary implications.