Environmental (and Random?) Sex Determination in Endangered and Invasive Phelsuma Geckos

Most vertebrate species are gonochoristic; i.e., their populations consist of male and female individuals. Sex determination is a crucial moment that is important for the individual, but also for the whole population, as it contributes significantly to the sex ratio [1, 2]. In amniotes, there are two basic modes of sex determination: genotypic sex determination (GSD) and environmental sex determination (ESD), which can be distinguished by the presence or absence of consistent sex-specific differences in genotypes between the sexes [3]. In reptiles, ESD takes the form of temperature-dependent sex determination (TSD), where incubation temperature during a sensitive period of incubation determines the sex of the individual for the rest of their life. The adaptiveness of ESD is still in question, with the most supported model being that proposed by Charnov and Bull [4]. According to this model, sex should be environmentally determined when the environment affects the relative fitness of females and males differently. Nicolas Perrin [5] emphasised that in addition to genotype versus environment, there is a third axis in sex determination represented by randomness. He predicted that random components of sex determination can be expected under conditions where the environment has no sex-specific effect on fitness or in species possessing life history traits that buffer sex ratio biases (e.g., high longevity, strong dispersal). Nevertheless, this view remains somewhat isolated, perhaps because randomness is difficult to study and there are only a few systems in which it is supported.

Current knowledge suggests that ESD is much rarer than GSD in amniotes, occurring in maybe around 5% of species [6]. ESD is restricted to crocodiles, many turtles, the tuatara, and two lineages of squamate reptiles – geckos [7] and possibly dragon lizards ([8, 9]; but see [10]). ESD has been suggested by some authors to be the ancestral sex determination system in amniotes [3], but this question is far from being resolved. One of the major obstacles is the lack of even basic knowledge about sex determination in most species. With advances in genomic techniques, the situation is improving dramatically in GSD lineages, where sex chromosomes are now routinely detected in many lineages (reviewed by [11]). On the other hand, the number of species with proven ESD is not increasing, because the necessary incubation experiments are time-consuming, labour- and material-intensive, and require access to freshly laid eggs, and thus mostly the establishment of breeding colonies of non-model species.

The logistical difficulties in documenting ESD can be illustrated by geckos, the amniote group that probably exhibits the greatest variability in sex determination mechanisms [12]. To date, we know the mode of sex determination in less than 0.5% out of about 2,200 described species [13]. While GSD and sex chromosomes have recently been discovered in several species from various families [13, 14], we have reliable data on ESD in only around 12 gekkotan species with limited and dubious data in a few others [15].

We also know very little about sex determination in day geckos of the genus Phelsuma. This genus with about 50 species [16] radiated in Madagascar and adjacent islands (Mascarenes, Comores, Seychelles). Both GSD and TSD have been proposed in them, based on unpublished data coming from a single incubation temperature or a very small number of hatchlings [17]. Reliable incubation experiments at controlled, constant incubation temperatures are virtually non-existent for this genus, and data are minimal, reviewed in [12], although incomplete data suggest TSD in several species [18‒21]. The only exception with an adequate sample size at multiple incubation temperatures is Phelsuma grandis, which clearly exhibits a TSD of the female-male-female type, where females hatched at the lowest and highest temperatures examined and males peaked at the intermediate temperature [17].

We established breeding pairs and performed constant temperature incubation experiments on two species: the vulnerable (according to IUCN Red List) Phelsuma nigristriata, endemic to the Mayotte Islands [16], and the invasive Phelsuma laticauda originating from northern Madagascar and nearby islands and successfully colonising, for example, Hawaii, Florida, and French Polynesia [22‒24]. The Charnov-Bull model predicts a sex-specific effect of incubation temperature on fitness. The model is difficult to test, because the environment can have a sex-specific effect on many phenotypic traits. However, a comparative study showed a relationship between the shape of the reaction norm for sex ratio and sexual differences in body size, supporting that sexual size dimorphism may be the phenotypic trait that behaves in a manner consistent with the Charnov-Bull model [25]. Therefore, we used body size as a proxy for fitness and tested whether incubation temperature has a long-term sex-specific effect on body size in these male-larger geckos.

Eggs were collected from twelve pairs of P. laticauda and eight pairs of P. nigristriata. The parents came from several different sources. They were always captive bred geckos. The P. laticauda animals were obtained from a private breeder, while most of the P. nigristriata animals came from the breeding stock of the Plzeň Zoo. The experiments took place between 2020 and 2023 at the Zoological and Botanical Gardens of the City of Pilsen. Glass cages (15 cm × 35 cm × 40 cm) with fluorescent lighting for one breeding pair were placed in a room with a daytime temperature of 30–31°C for 12 h and a night-time temperature of 26–27°C. The walls between the terrariums of each pair were made of white opaque glass to prevent the animals from seeing other individuals, as the day geckos are very territorial. A layer of sand at the bottom and a layer of peat covered the floor of the cage equipped with live plants that provided sufficient moisture and two bamboo sticks to provide shelter for the animals and a convenient place to lay eggs. The breeding animals were fed twice a week with six domestic crickets (Acheta domestica) per individual. The insects were supplemented with the vitamin-mineral preparation Repashy Calcium Plus HyD. The geckos also had access to crushed cuttlefish bones as an additional source of calcium. Water was provided by misting the interior of the terrarium daily with a spray bottle. Vitamins A and D3 were added to the water once a month in the form of liquid Kombisol AD₃.

The breeding pairs were inspected every 2 days to collect eggs. Females of both species laid eggs in hollow bamboo stalks. Clutches consisted of two eggs that were glued together so that they could not be separated. Exceptionally, females laid only a single egg. The incubation took place at three constant temperatures (24, 28, and 31°C) in incubators FB-50 (JAEGER & PFROMMER) with the declared precision 0.1°C equipped with a digital temperature controller. The controller automatically compensates for changes in room temperature; however, the room with the incubators had controlled ambient temperature as well. The choice of the incubation temperatures was based on our previous experience with gecko breeding. The temperature in each incubator was controlled using a calibrated Comet UO141M data logger recording the temperature to the nearest tenth of a degree Celsius every 5 s. The system allows alarm SMS and JSON messages to be sent via a GPRS data connection when set limits are exceeded. We incubated the eggs in small plastic boxes. The bottom of the incubation box was covered with a layer of moist vermiculite, a dish of dry vermiculite was placed on top, and the eggs were placed on top of it. This prevented the shell from coming into contact with the moist substrate, which can lead to increased embryo mortality. The lid of the incubation box contained several holes for air exchange.

The juveniles of both species were kept individually from hatching in plastic boxes (14 cm × 18 cm × 7 cm) with a plastic mesh front, again with fluorescent lights installed as in the breeding pairs. The boxes had only a paper towel on the bottom and two thin bamboo sticks for climbing. We measured the body mass of the hatchlings immediately after hatching and the snout-vent length (SVL) when the animals reached 1 year of age. Feeding was carried out in the same way as for the adults, just with crickets of the adequate size. Sex was determined by visual assessment of the presence or absence of femoral pores. In both species, sex can be in this way determined very easily, as the glands open on the underside of the thighs of males at a relatively early age (2 months in P. nigristriata and 4 months in P. laticauda). Statistical analysis and visualisation were performed in the R software packages “ggplot2,” “ggpubr,” “gplots,” “car,” “MASS,” “lme4,” “gam,” and “dplyr.” To assess the effect of incubation temperature on sex, we included kinship as a random factor in the general linear mixed model, the reaction norm of sex ratios across different temperatures was estimated by the generalised additive model (GAM) approach, and however due to limitations of GAM modelling we used GLM for P. laticauda. To compare SVL, we used a linear mixed model with temperature as a factor, and body mass at hatching as a covariate, controlling for the random effect of family. Model selection was based on the Akaike information criterion (AIC [26]). The difference in AIC <2 indicated that the models were equivalent and the simpler model was then selected, while the more complex model was selected when ΔAIC >2.

A total of 118 and 137 eggs were collected from P. laticauda and P. nigristriata, respectively. In both species, the reaction norms modelled with GAM and GLM indicated the female-male-female pattern of TSD known from other geckos such as Eublepharis macularius [27] and P. grandis [17]. However, sex is affected by incubation temperature in a different way in each species (Table 1; Fig. 1). In P. laticauda, the temperature 24°C produced only females, whereas at 28°C the majority of sexed individuals were males. At 31°C, the overwhelming majority of individuals were again females. In P. nigristriata, no males were observed at 24°C, while males predominated at both 28 and 31°C. However, mortality in this species differed significantly among temperatures (χ² test, p < 0.001; Table 1). In addition, ten offspring in P. nigristriata coming from all incubation temperatures and from different pairs suffered from morphological abnormalities (curvature of the lumbar spine or caudal region). However, even taking into account the mortality, it seems that at the highest temperature tested, more males are produced in P. nigristriata than in P. laticauda.

It was very interesting that siblings from the same clutch incubated at temperatures that produced both sexes were of the opposite sex. Specifically, a female and a male hatched from two clutches, each from a different pair, in P. laticauda eggs incubated at 28°C. In P. nigristriata, this occurred in a total of five clutches, each from a different pair, at 28 and 31°C. This observation is quite surprising because the egg pair was always glued together and incubated side by side in a small plastic container in the incubators. The thermal gradient must have been minimal. Earlier, it was reported in another TSD gecko (E. macularius) that eggs from the same clutch incubated side by side produced offspring of the same sex [28], but probably a small temperature difference is sufficient to break this link, as siblings from the same clutch incubated separately even in the same incubator can be of the opposite sex [29]. In the case of the day geckos, we cannot exclude the possibility that each offspring inherited an allele controlling a different response to the same environment. However, this possibility does not seem very likely, especially in P. nigristriata, which must be rather inbred, as can be deduced from the occurrence of malformed juveniles. It seems to us that the situation can be best attributed to a certain degree of stochasticity in sex determination, as suggested by Perrin (see [5] for other putative cases). More recently, the stochastic factor remained the most likely explanation after a careful investigation of several possibilities in sex determination in the nematode Bursaphelenchus okinawaensis [30]. The authors believe that the sex determination in this species is influenced by random expression of an unknown trigger gene and/or developmental stochasticity independent of environmental conditions, and discuss practical difficulties of studying the random effect [30]. We stress that further experimental work would be required in the day geckos in order to clarify the factors that drive the differences in sex determination in the full siblings from eggs glued together.

Perrin [5] predicted that random components of sex determination would be expected under conditions where the environment has no sex-specific effect on fitness, or in species with life history traits that buffer sex ratio biases (e.g., high longevity, high dispersal). It is difficult to test these predictions in the day geckos. In captivity, P. laticauda can live for at least 8 years [31], which seems long enough to buffer random fluctuations. We compared adult SVL among females from different temperature treatments and among males and found no significant effect of temperature on adult size (online suppl. material Appendix 1; for all online suppl. material, see https://doi.org/10.1159/000538906, online suppl. Table A2). The variation in SVL appeared to be due to differences in post-hatching body mass rather than the incubation temperature. The lack of differences in adult body size induced by the incubation temperature could explain why there is no strong selection against producing females at “male” temperatures: they may not get any significant disadvantage, which would allow some stochasticity in sex determination. However, we cannot exclude the possibility that adult body size is not directly related to fitness and that other phenotypic traits such as behaviour or locomotor performance may be a better proxy for fitness [32, 33]. Future studies should further investigate whether the environment has a sex-specific effect on phenotype in the day geckos.

The documentation of ESD in two species of the genus Phelsuma extends our knowledge of sex determination in squamates. All three from the comprehensively studied species of the genus exhibit TSD. They are phylogenetically quite distant [34] and represent a putative basal split of the genus. We can therefore conclude that ESD may be widespread in the day geckos and may be ancestral to the genus. Research in outgroups is needed to determine whether TSD could be a plesiomorphic character for the day geckos.

The results presented here have implications for environmental management or ex situ conservation practice. There is a current concern about whether TSD species would be able to cope with the global warming changing incubation environment [35]. Less discussed is whether TSD species are limited in their invasiveness due to their specific requirements for incubation conditions that ensure appropriate population sex ratios. Nevertheless, Dubos and colleagues [36] already suggested that ESD may limit the spread of the day geckos during the invasions. Among the most invasive reptile species, we can find the red-eared slider turtle (Trachemys scripta elegans) with ESD, but its widespread alien distribution across Europe and Asia is thought to be dependent on the recruitment of new individuals to rising pseudo-populations due to additional releases, rather than an establishment of reproducing populations [37]. The case of highly invasive species such as P. laticauda raises the intriguing question of how ESD species are able to spread and establish populations in environments likely thermally different from their native ranges.

We thank Tomáš Peš, sr., Lucie Pešová, and other employees of Plzeň Zoo for their help with experiments.

The animal facility of the Faculty of Science, Charles University, is accredited by the Ministry of Education, Youth and Sports of the Czech Republic (accreditation number CZ11760285).

The authors declare no conflict of interest.

The study was supported by the Czech Science Foundation (project No. 23-07658S).

Conception of the work, interpretation of data, and drafting and final approval of the manuscript: T.P., B.S., and L.K.; acquisition of data: T.P.; data analysis: B.S.

The datasets supporting this article have been uploaded as part of the supplementary material.