Background

One of the most fundamental aspects of any sexually reproducing organism is its phenotypic sex, as this profoundly influences many aspects of its life history and eventual reproductive success [1]. In squamates, sexual development is controlled by a variety of mechanisms resulting from a dynamic evolutionary history [2]. These can be broadly categorised into temperature-dependent sex determination, genetic sex determination [1, 3,4,5,6,7] and systems where genotype and environment interact to determine sex [8, 9]. Temperature-dependent sex determination (TSD)—where sex is determined by incubation temperature during the “thermosensitive period”—occurs in all crocodiles, many turtles, the tuatara, and seems to be the predominant mechanism of sex determination for lizards [10,11,12,13,14,15]. By contrast, genetically controlled sex determination (where genes on sex chromosomes determine sexual phenotypes; GSD) occurs in snakes and some lizards and turtles [3, 16,17,18,19,20]. The evolutionary history of sex-determining mechanisms (SDMs) is remarkably diverse in squamates when compared with mammals, whose sex chromosomes have a single origin [21]. Squamate sex chromosomes have independently evolved in many lineages, and transitions from TSD to GSD systems can occur within short evolutionary time frames [12, 18, 22,23,24,25].

The conserved development of external genitalia (hereafter referred to as genitalia) in squamates is thought to be controlled by hormones secreted after sex determination, a process generally regarded as being unperturbed by squamates’ various SDMs [26,27,28,29,30,31]. However, as any comparative study of genital development would be phylogenetically confounded, this assumption has not been properly tested. Given that cell-autonomous sex has been demonstrated in birds, and there are instances of intersexuality and gynandromorphism in squamates, it is possible that genital development is influenced by mechanisms other than gonadal hormones, which could be perturbed by different SDMs [32,33,34,35,36,37].

To investigate the developmental effects of different SDMs, particularly on genital morphology, we used an experimental approach in a unique model system, Pogona vitticeps. This species exhibits genotypic sex determination (ZZ/ZW female heterogametic system [38]), but incubation temperatures at or above 32 °C can cause the complete phenotypic feminisation of genetically male (ZZ) individuals [8, 14, 39]. P. vitticeps is one of only two reptile species known to exhibit thermally triggered sex reversal in wild populations (the other being the Eastern Three-Lined Skink, Bassiana duperreyi [39]). P. vitticeps is also the only reptile in which a rapid transition from GSD to TSD has been experimentally triggered through the mating of male and female homogametic individuals [14]. This provides a unique opportunity to examine embryonic development under both chromosomal and temperature influence within the same species.

Our study is the first to characterise and compare the developmental effects of different incubation temperatures on offspring from concordant (ZWf) and sex-reversed (ZZf) mothers in P. vitticeps, including the first assessment of developmental patterns associated with temperature-induced sex reversal. For this purpose, we provide a comprehensive embryonic staging table for P. vitticeps, with a particular focus on describing the effects of temperature and genetic sex determination on the development of male and female genitalia.

Using a new molecular approach to identify embryonic genotypes, we assess for the first time whether development, particularly of the genitalia, is perturbed by differing SDMs (GSD vs. TSD) or sex reversal in the same species. We also ask whether staging accurately describes gross embryonic development in different incubation regimes. This allows us to provide the first macroevolutionary perspective on how sex determination mechanisms may impact on the phenotype of the body and particularly genitalia of squamates.

Methods

Breeding and incubation treatments

To assess developmental differences between GSD and TSD breeding lines of P. vitticeps, we crossed ZZ males with ZWf (concordant) and ZZf (sex-reversed) females. Eggs were collected upon laying and allocated into four experimental treatments to produce all offspring phenotypes resulting from combinations of high and low temperatures (28, 36 °C) and maternal genotypes (ZZ, ZW; Fig. 1). The 28ZW treatment provided a baseline for normal development, as sex under these conditions is genetically determined (ZZ males, ZW females). The 36ZW treatment is expected to yield approximately 50:50 concordant (ZWf) and sex-reversed (ZZf) females, making it possible to compare concordant and sex-reversed development at the same temperature. The 36ZZ treatment documented the development of sex-reversed females from sex-reversed mothers, while the 28ZZ treatment yielded concordant males from sex-reversed mothers.

Fig. 1
figure 1

Experimental design encompassing all procedures used in this study. Circle denotes approximate day of hatching for eggs incubated at 36 °C (46.7 ± 1.6 SD) and diamond for eggs incubated at 28 °C (73 ± 3.5 SD) based on estimates from Holleley et al. [14]

Full size image

During the 2015–2016 breeding seasons, a total of 254 eggs were incubated and sampled. Of these, 221 eggs were obtained from the University of Canberra’s (UC) captive breeding colony (1–3 generations from animals sourced from a wild population in northern New South Wales/South West Queensland). An additional 33 eggs were sourced from the commercial pet trade and incubated at the University of Queensland (10 sampled in the 36ZW treatment and 23 in the 28ZW treatment). All specimens were staged and photographed using a Dino-Lite Edge digital microscope after formalin preservation.

Due to issues with formalin preservation, early developmental stages (prior to stage 4), including stage at oviposition, were not captured during this initial sampling effort. To obtain these stages, 8 eggs were sampled on the day of oviposition (four different mothers; two ZZ and two ZW) and three stage 2–4 embryos (single ZZ mother, incubated at 36 °C) were obtained from UC’s colony during the 2017 breeding season. All specimens were staged and photographed using a Leica Wild MZ8 dissection microscope prior to formalin preservation.

All eggs were incubated in damp vermiculite (four parts vermiculite to five parts water by weight) in constant temperature incubators with high humidity and minimal temperature fluctuations outside of the set range (± 1 °C range, excluding fluctuations arising from examination of the eggs). Eggs from all mothers (n = 17 breeding females) were allocated to the four treatments and sampled across development (Fig. 1). Initially, the entirety of development was surveyed by randomly assigning eggs to sampling days to establish baseline data on developmental timing (every 3 days at 36 °C and every 5 days at 28 °C). This survey identified critical periods of development, which required higher-resolution sampling to determine the timing and order of developmental events (Fig. 1). Sampling intensity was greater for the 36ZZ treatment (sex-reversed offspring of sex-reversed mothers) to ensure that any morphological diversity associated with temperature-induced sex reversal was adequately described.

Embryos and intact yolks were dissected from the egg, and all embryos sampled after the first third of the incubation period were humanely euthanised by intracranial injection of 100 µl of sodium pentobarbitone (60 mg/ml; [40]). Embryos were kept in 10% neutral-buffered formalin fixative for a minimum of 24 h (no more than 72 h), then rinsed in water and stored in 70% ethanol. After ethanol preservation, which stabilises the yolk and embryo for handling, all embryos and yolks were weighed separately for analysis of growth and yolk absorption rates. Ethanol dehydrates tissues; thus, the embryo and yolk weights in this study may slightly underestimate the weight prior to preservation. However, because all specimens were subjected to the same preservation conditions, this approach is unlikely to have introduced systematic bias in our data and is suitable for a general assessment of growth patterns.

Molecular sex identification in embryonic samples

We developed a novel approach that is non-destructive to the embryo for molecular sex identification of embryonic specimens. Embryonic blood from the inside of the eggshell was swabbed onto a FTA® Elute Micro Card (Whatman) immediately after egg dissection. DNA was extracted following the manufacturer’s instructions with a protocol adapted for automated high-throughput analysis on the Eppendorf EPmotion 5075 liquid-handling platform. Specifically, in a 96-well plate, 3mmdiameter FTA card punches were individually washed in water, boiled for 30 min (100 °C) in a total volume of 100 µl water and then vortexed for 2 min to release DNA from the card. To concentrate the DNA, samples were incubated in a heat block at 40–50 °C for approximately 2 h, until the elution volume was reduced by 50% via evaporation. To confirm that sufficient DNA was extracted from embryonic blood collected on FTA cards, DNA concentration was quantified for a subset of samples (n = 92) both prior to and after controlled sample evaporation using a NanoDrop 1000 spectrophotometer (Thermo Scientific). The DNA concentration of embryonic samples was compared to the DNA concentration from FTA® card extractions of adult P. vitticeps blood samples (n = 30).

We then conducted a PCR-based test, which is diagnostic for the presence of the W chromosome. PCR conditions followed Holleley et al. [14]; however, due to the likelihood of low DNA concentrations from embryonic material, we increased the volume of DNA added to PCRs (3 µl per reaction; approximately 65 ng DNA per PCR). Using primers H2 and F [41], two bands amplify in ZW individuals, whereas a single control band amplifies in ZZ individuals. Animals showing genotype–phenotype discordance were classified as sex-reversed. To test the level of confidence with which we could assign a genotype to embryos sampled with the non-destructive approach, we compared genotype results from blood taken directly from the embryonic heart to the non-destructively sampled eggshell swabs in a subset of specimens (n = 23). All PCR tests were run in triplicate, and maternal contamination was investigated by checking phenotypically male embryos (obligate ZZ genotype) for the presence of W chromosome contamination.

Developmental staging

Staging was based on Sanger et al. [40] staging system for Anolis spp, but also included characters from Wise et al. [13] staging system for the leopard gecko (Eublepharis macularis). Stages based on traits not present in P. vitticeps (digital pad, toe lamellae), or that were not diagnostic for a given stage in P. vitticeps (scale anlagen, first full scales, pigmentation), were renamed. In addition, we developed novel staging criteria that described genital development. Specimens obtained from the commercially bred line (n = 33) were not used to establish pigmentation development, as pigmentation patterning obviously differed to that of the wild-derived breeding colony (likely due to selective breeding for colour variation in the pet trade).

To quantify how well age as a function of stage explained embryo growth (defined as embryo weight over age), and whether there were differences between treatments, models were fit to a linear equation (Stage = a + b * Age) with treatment as fixed effect, using the nls function in R version 3.2.2. Subsequently, we investigated whether the relationship between age and stage was different between temperature and maternal type (sex-reversed ZZ mother or concordant ZW mother) treatments using the nlme function of the nlme package. A random maternal effect was incorporated into the model to account for maternal effects as clutches from 17 different mothers were distributed across the study. Our data set was too small to incorporate maternal types (ZZ/ZW mothers) across both temperature treatments while including the effects of having 17 mothers in total as well, so we first compared the growth of ZZ versus ZW treatments within temperatures. If these regressions were not significantly different in slope and intercept, we pooled them and compared these pooled data between temperatures.

Embryo growth and yolk consumption

Embryo growth was estimated using the relationship between weight (g) over time (age, days post-oviposition), with an exponential curve fitted for each treatment using the nls function in the nlme package for R version 3.2.2, with treatment as fixed effect and mother as random effect. As with the age versus stage comparison, we first compared ZZ/ZW treatments within temperatures, and if no significant differences were found, we pooled treatments and compared between temperatures.

For a visual assessment of the relationship between embryo growth, embryo stages, and yolk consumption in the four treatments, we also plotted log embryo weight and log yolk weight against age (days post-oviposition, dpo).

Results

Staging and age prediction by stage

For the staging table and relevant morphology, see Figs. 2, 3; Table 1 and videos of live early-stage embryos (see Additional files 1: Video S1, 2: Video S2 and 3: Video S3). For specimen staging, ageing, genotyping and weights, see Additional file 4: Table S1. Embryos at day of oviposition were less developed than the earliest stages at lay described for Anolis spp. and E. macularius [13, 40]. While embryos can develop in the oviducts before oviposition, introducing variation in stage of development at lay, we found that eggs were consistently laid at stage 1 (late pre-limb bud; Fig. 1).

Fig. 2
figure 2

Early developmental stages for Pogona vitticeps; stages 1 (day of lay) to 4 (see Table 1). All specimens were photographed prior to formalin preservation. Mes. Mesencephalon, Met. Metencephalon, C.F. choroid fissure, O.P. olfactory placodes

Full size image
Fig. 3
figure 3

Developmental staging series for Pogona vitticeps, depicting stages 5–18 observed across all experimental treatments (see Table 1). All specimens were photographed after formalin preservation

Full size image
Table 1 Developmental staging table for P. vitticeps based on Sanger et al. [40] staging system for Anolis lizards
Full size table

Staging is easiest and most accurate early in development when organogenesis and limb development events are more discrete and becomes more difficult and less accurate as the embryos approach hatching because the morphological changes become less distinct. Often P. vitticeps embryos showed a combination of traits across two stages, and so were denoted as 0.5 of a stage. Generally, development progressed similarly to Anolis (the species described in the original system upon which we based ours), with exception of the far earlier onset and development of pigmentation, and more rapid development of the eyelid in P. vitticeps. Early in development (stages 1–3), the somites extend beyond the developing hindlimb towards the tail bud, increasing their number, whereas in Anolis the somites do not extend past the hindlimb.

For each of the treatments, stage as a function of age explained embryo growth very well (Fig. 4; Table 2). Incubation temperature and sex reversal did not influence the order of development of any phenotype. Slopes and intercepts of ZZ and ZW age versus stage fits were not significantly different, although a relatively low p value (0.09) of the slope comparison suggests a tendency of 36ZZ specimens to go through later stages slightly more quickly (Fig. 4). A comparison of the slopes between the 36 °C treatments and 28 °C treatments found they were significantly different (Fig. 4; Table 2).

Fig. 4
figure 4

Plot of specimen stages against specimen post-oviposition ages. Growth and stage development are accelerated at high temperatures (36 vs. 28 °C), but are unaffected by the sex chromosome complement of the mother (ZZ vs. ZW)

Full size image
Table 2 Summary of age versus stage regressions within temperatures and between temperatures, including variance explained by maternal effects
Full size table

Sex chromosome genotyping

For details of the embryo genotyping results, refer to Additional file 5: Table S2. As expected, DNA extracted from embryonic material yielded less DNA than an equivalent extraction from adult blood (50.58 ng/µl ± 8.05 SE), both before (13.09 ng/µl ± 1.99 SE) and after evaporative DNA concentration (21.63 ng/µl ± 3.36 SE). However, embryonic DNA yield was sufficient to generate highly reproducible molecular sex identification results. Ninety-nine percentage of samples run in triplicate (n = 184 embryonic blood or heart samples) returned concordant results (i.e. all tests ZZ or all tests ZW). Genotype concordance was 100% when comparing embryonic blood with invasively sampled heart blood from the same embryo (n = 23). We did not detect any evidence of maternal contamination (presence of W chromosome) in obligate ZZ male embryos from wild-type crosses (n = 9). Thus, we are confident that the embryonic blood supply sampled from the interior of an eggshell can be used to assign the genotypic sex of the developing embryo.

Genital development

In all sexes (concordant males and females, and sex-reversed females), genital development begins as small paired phallic swellings form on either side of the developing cloaca (between stages 5 and 8 in all treatments, Fig. 5a; score 1 Additional file 4: Table S1). The swellings increase in size until they achieve a club-shaped appearance and are enclosed by distinct anterior and posterior cloacal lips (approximately between stages 9 and 13 in all treatments, Fig. 5b; score 2 Additional file 4: Table S1). This club shape becomes more pronounced as development progresses until the distal tip of each hemipenis is bifurcated, creating the characteristic bilobed appearance of mature hemipenes in all sexes (from approximately stage 11 in all treatments, Fig. 5c; score 3 Additional file 4: Table S1).

Fig. 5
figure 5

Genital development in female Pogona vitticeps embryos. Development progresses from rounded paired swellings between the hindlimbs (a, white arrowheads; score 1), club shape (b, blue arrows; score 2), bilobed hemipenes (c, deepening invaginations create distal bifurcation: black arrow; score 3). The lobes become increasingly accentuated as this invagination deepens (black and white arrows). In females, the hemipenes begin to regress, but maintain their bilobed appearance (d, black arrowhead: cloacal opening; score 4). Females eventually possess hemiclitores (e, blue arrowheads; score 5) before they too regress completely to the pedicel (f; score 6). The anterior and posterior cloacal lips in specimen e were removed to expose the hemiclitores. Scale bar = 1 mm

Full size image

Male and female development diverges from stage 11 (73% through development). In males, ongoing development of the hemipenes is characterised by deepening invaginations on the bilobes, which considerably increases their surface area. In all male specimens, the hemipenes were consistently everted; however, in both treatments (28ZW and 28ZZ) a total of fourstage 18 specimens exhibited no everted hemipenes. It was unclear as to whether they were simply folded within the vent as the specimens approached hatching, or were truly absent. In the 28ZW treatment, there were two unexpected phenotypes: one stage 17 (55 dpo) male exhibited reduced hemipenes, while one stage 18 (70 dpo) male exhibited hemiclitores.

In females, hemipenis regression commences at approximately stage 11 with an overall shortening of the hemipenes, but is characterised by retention of the lobes throughout the hemipenis regression process (approximately between stages 11 and 17, Fig. 5d; score 4 Additional file 4: Table S1). The lobes ultimately regress to hemiclitores, which are characterised as two small swellings without bilobes slightly protruding from the margins of the pedicel (from approximately stage 16.5, Fig. 5e; score 5 Additional file 4: Table S1). As embryos approach hatching (stage 18), the hemiclitores reduce completely so that only the pedicel is present at hatching (see Fig. 4f; score 6 in Additional file 4: Table S1). Hence, it is possible to sex hatchlings using traditional methods such as hemipenal eversion [42] and transillumination [43].

Comparison of embryo growth and yolk consumption

Embryo growth follows an exponential curve (Fig. 6), with no differences in slopes between ZZ and ZW offspring within temperature treatments, so we pooled all specimens for each temperature treatment. As predicted, significant differences exist between slopes of the two temperature groups. In all cases, maternal effects were far smaller than the residual variation (Table 3). Early in development, yolk weight is highly variable and not clearly associated with embryo weight (Fig. 7). Later in development, the embryo becomes heavier than the yolk (see shaded areas in Fig. 7). There are no significant differences between offspring from breeder versus wild-obtained mothers in the 36ZW and 28ZW treatments (Additional file 6: Figure S1).

Fig. 6
figure 6

Embryo weight over age follows an exponential curve in all treatments. Growth is faster in the 36 °C treatments than the 28 °C treatments, and growth is unaffected by maternal type (ZZ vs. ZW)

Full size image
Table 3 Summary of growth curve (embryo weight over time) comparisons within temperatures and between temperatures, including variance explained by maternal effects (Var. Mat. Effect)
Full size table
Fig. 7
figure 7

Log embryo (blue) vs. log yolk (orange) weight over time in days post-oviposition (dpo) for each treatment. Shaded rectangle highlights the time at which embryo weight rapidly increases at the expense of yolk weight

Full size image

Discussion

In this study, we provide the first morphological characterisation of external development in P. vitticeps under normal and sex-reversing temperatures. Regardless of the sex-determining cue (temperature or sex chromosomes), genital development is a highly conserved process that does not differ between males and females for much of embryonic development. Female development is characterised by the growth, retention, and eventual regression of hemipenes, which are normally characteristic of the male genital phenotype. A review of the literature (Additional file 7: Table S3) reveals that the development of male genitalia in P. vitticeps is consistent with the gross morphological processes described for other squamate species. Across temperatures and maternal type, the genital development remains synchronised with the development of other parts of the body, which are also not perturbed in their sequence by either temperature or sex determination mechanism. This observation differs from results in turtles where low temperatures extended the retention of some earlier phenotypes [44]. However, it is possible that similar effects might occur in P. vitticeps in particularly cold incubations, which were not included in this study. Regardless, the robustness of genital and phenotypic development to these influences is interesting because in adult sex-reversed females there are differences in fecundity [14], behaviour [45], gene expression [46], and some morphological traits [45]. In contrast, we did not observe any sex-reversal-specific differences in the timing, sequence, or structure of morphological development.

The conserved developmental sequence across temperature treatments and sex determination mechanisms allows an accurate prediction of specimen age from stage for a given temperature in all treatments. Staging is often criticised because there is no standard practice, it usually does not account for the effects of incubation temperature, or differences between field and laboratory raised animals, and often uses small sample sizes [44, 47]. However, these factors had little influence on the accuracy of P. vitticeps staging, suggesting that staging remains an ideal method for categorising development. In particular, staging is a powerful method to visually calibrate sampling points in future studies of P. vitticeps development, avoiding the need for heavy replication to capture a specific sexual phenotype in this emerging model organism [8, 12, 14, 24, 45, 48, 49].

Our results provide intriguing evidence that sex determination mechanisms (SDMs) do not impact on the formation of P. vitticeps genitalia. This suggests that the molecular underpinnings of genital formation through hormonal signalling and dosage from the gonads after sex determination follow the same pattern regardless of whether sex is genetically or temperature-determined [26, 50,51,52]. This lack of connection between SDMs and genital formation also suggests that the evolution of genital development and SDMs are not closely linked based on current evidence (Additional file 6: Figure S1). However, this needs further investigation across squamates with different SDMs as well as other dual-SDM systems [53, 54].

A robust developmental programme of genital development is not unexpected, as mating success depends on the proper formation of genitalia [26]. However, genitalia are highly diverse within squamates and evolve faster than other phenotypic traits [26, 27, 29]. Based on our results, intraspecific variability or switches in SDM are unlikely to be a source for this diversity; future comparative study of squamate genital phenotypes may provide further insights into the mechanisms driving the evolution of squamate genital morphology.

The extended retention of male traits in female P. vitticeps is interesting in an evolutionary context because female genitalia exhibit a far wider range of genital phenotypes than males, but these phenotypes are generally based on the default of a hemipenis form. Female genitalia in squamates vary from structures resembling rudimentary hemipenes to species where females have longer hemipenes and associated musculature than males [33, 35, 50, 55,56,57,58,59]. In P. vitticeps, extended developmental hemipenis retention in females and male sex chromosome homogamety suggest that the ancestral programme of genital development may be biased towards hemipenis formation. The acquisition of a developmental pathway for hemipenis regression, which seems to be a secondary occurrence in P. vitticeps, may also occur in other species, possibly driven by sexual selection. Although this is speculative, it is consistent with suggestions that the developmental programme governing hemipenis formation is extremely conserved in amniotes [26]. However, limited data exist on female genital development in squamates, and the mechanistic underpinnings of their growth remain poorly understood [51]. This is in contrast to work on males, which is considerably more detailed and addresses the evolutionary and genetic processes governing hemipenis development (Additional file 6: Figure S1). Future studies should consider female development, in particular the developmental processes governing the growth of the genitalia, to improve our understanding of sexual development, particularly in sexually labile species such as P. vitticeps.

We observed that P. vitticeps eggs were consistently laid at stage 1, which is earlier than described for most other squamates (Fig. 2; Additional file 7: Table S3). Anolis were laid at stage 4 (early limb bud), while E. macularius were laid at stage 2. A final interesting observation was the variability of yolk weights compared to embryo weight, particularly early in development, across all treatments (Fig. 7). After this phase of large variability, a rapid decrease in yolk beginning from stages 13–18 coincides with the completion of organogenesis (Table 1). This suggests that the majority of yolk consumption occurs when the embryo has a complete body plan and begins to gain weight in preparation for hatching.

Conclusions

Our investigation clarifies the evolutionary relationship between sex determination and embryological development in P. vitticeps, providing an encouraging perspective on the stability of squamate development, even under environmental extremes that might be expected under climate change. Our data suggest that even substantial perturbations at the very beginning of development (+ 8 °C) may not de-stabilise the formation of functional genital phenotypes. However, exposure to feminising temperatures could negatively impact on population viability if sex ratios become skewed [60]. We highlight the need for larger-scale study of female squamate genitalia, particularly with a view to the intriguing possibility that male and female squamate genitalia might represent variations of a hemipenis-like structure. This also raises questions about the main determinants of squamate genital formation, which appears to be unaffected by the sex-determining mechanism. Our staging system and ability to confidently sex embryos using small quantities of embryonic blood will facilitate further molecular based work on this question.