Sex ratio is a fundamental characteristic of a population capable of affecting its demography and fate (Cotton and Wedekind, 2009; Earl, 2019). In species where gonadal sex is determined by sex chromosomes, Fisher (1930) predicted that sex ratios should be equal at birth, hatching, or metamorphosis. However, observations of sex ratios at early life stages are frequently biased from parity in a variety of species (Hardy, 2002). In mammals and birds, most research on the drivers of early-life sex ratio variation has stressed parental sex allocation – where parents manipulate the sex of their offspring prior to or soon after fertilization by disproportionately allocating resources to a given sex – as a leading mechanism modulating sex ratio in early life stages (Charnov, 1982; West, 2009). However, research in other vertebrates, like many fishes and non-avian reptiles, emphasizes a role for sex determination in shaping offspring sex ratios (Sarre et al., 2004; Ospina-Alvarez and Piferrer, 2008). Sex determination in vertebrates occurs along a spectrum (Sarre et al., 2004). At one end, an organism’s phenotypic sex is controlled entirely by genes (genotypic sex determination) and at the opposite end by environmental conditions like temperature (environmental sex determination). In between these extremes there can be roles for both genetic and environmental factors to contribute to sex determination. Such species can sex reverse, where environmental conditions cause individuals to develop their phenotypic gonadal sex opposite from their genotypic sex (Sarre et al., 2004; Quinn et al., 2007; Ospina-Alvarez and Piferrer, 2008; Alho et al., 2010; Holleley et al., 2015, 2016; Lambert et al., 2019; Miko et al., 2021). In mammals and birds, phenotypic sex is genetically determined whereas fishes and reptiles show a diversity of sex-determining modes, including environmental sex reversal (Bachtrog et al., 2014).

Amphibians have been used as model organisms for over a century to study how environmental conditions influence sex ratios in developing animals (King, 1909, 1910; Witschi, 1929; Wallace and Wallace, 2000; Pettersson and Berg, 2007; Lambert, 2015; Lambert et al., 2016a, 2018; Miko et al., 2021). Sex ratio variation in metamorphosing amphibians is largely assumed to be the result of sex reversal and phenotypic sex ratios at metamorphosis are a proxy of sex reversal in laboratory and field studies (Wallace and Wallace, 2000; Pettersson and Berg, 2007; Papoulias et al., 2013; Lambert, 2015; Lambert et al., 2015, 2018). In such studies, female-biased phenotypic sex ratios are assumed to result from male-to-female sex reversal and male-biased phenotypic sex ratios from female-to-male sex reversal. However, this assumption is rarely tested. Evaluating whether sex ratio variation is due to sex reversal has remained elusive, in large part because genotypically sexing amphibians is notoriously challenging (Alho et al., 2010; Lambert et al., 2016b, 2019; Nemeshazi et al., 2020). Recent genomic advances permit the development of sex-linked molecular markers that allow researchers to genotypically sex large numbers of individual amphibians (Lambert et al., 2016b, 2019; Nemeshazi et al., 2020). Sex-linked markers have recently been used to demonstrate that sex reversal is occurring in wild amphibian populations (Lambert et al., 2016b, 2019; Nemeshazi et al., 2020). Even so, no study has yet to directly evaluate sex reversal as a mechanism underlying phenotypic sex ratio variation.

An alternate, and non-exclusive, hypothesis to sex reversal is that amphibian sex ratio variation is driven by sex-biased mortality (Figure 1). Sex-biased mortality prior to birth or hatching has been increasingly observed in wild animal populations, particularly in mammals and birds, but also occasionally in fish (Kruuk et al., 1999; Cichon et al., 2005; Svensson et al., 2007; Orzack et al., 2015; Moran et al., 2016; Kato et al., 2017; Firman, 2019). Multiple factors, including temperature stress and parental condition, have been identified as factors underlying sex-biased mortality, suggesting that sex-biased mortality may not be an adaptive process, as has been suggested for sex allocation, but a maladaptive response to stressful rearing conditions (Kruuk et al., 1999; Goth and Booth, 2005; Eiby et al., 2008; DuRant et al., 2016; Firman, 2019). Thus, sex-biased mortality is important to consider as it can directly shape sex ratios (Szekely et al., 2014).

To our knowledge, sex-biased mortality in pre-metamorphic amphibians remains largely unexplored. As with sex reversal, sex-biased mortality has likely remained understudied due to the technical limitations in genotypically sexing amphibians. Only two studies that we are aware of have considered sex-biased mortality in developing amphibians. One study attempted to infer sex-biased mortality by regressing phenotypic sex ratio variation against mortality rates at metamorphosis (Lambert et al., 2016a). This study concluded that sex ratio variation was due to sex reversal and not sex-biased mortality, however, molecular sex data are necessary to better clarify the processes underlying sex ratio variation. Another laboratory experiment found evidence for sex-biased mortality in tadpoles in response to high temperature even though that was not the intention of the study (Miko et al., 2021). Because sex reversal and sex-biased mortality could co-occur, distinguishing the relative contribution of these processes is crucial as both may produce similar phenotypic sex ratios but different genotypic sex ratios which can have implications for the evolutionary ecology and persistence of populations (Cotton and Wedekind, 2009; Earl, 2019).

In this study, we explore sex-biased mortality from embryogenesis to metamorphosis in two natural populations of the Green Frog (Rana clamitans) by sampling the genotypic sex of multiple developmental stages directly from a forested pond and a suburban pond. One of the few studies to assess phenotypic sex ratios in wild metamorphosing amphibians found that metamorphosing R. clamitans tadpole phenotypic sex ratios were male-biased in uncontaminated forest ponds and equal or female-based in suburban ponds that were contaminated by chemicals known to impact sexual differentiation (Lambert et al., 2015). Thus, R. clamitans sex ratio variation could suggest female-to-male sex reversal in natural populations and perhaps male-to-female reversal in environments impacted by suburban pollutants. Subsequent work used a novel molecular sexing approach and found that sex reversal was prevalent across 16 populations of free-living R. clamitans in both forest and suburban ponds and that both female-to-male and male-to-female reversal occur (Lambert et al., 2019). Although this study demonstrated that sex reversal is common in wild R. clamitans, it assessed sex reversal in adult frogs. Because adult R. clamitans populations represent multiple overlapping offspring cohorts, this study could not determine whether sex reversal contributes to sex ratio variation in metamorphosing R. clamitans larvae. Here, our two goals were to (1) document whether sex-biased mortality occurs and alters phenotypic sex ratios and (2) test whether sex reversal is correlated with phenotypic sex ratio variation. We genotype the sex of over 900 R. clamitans across ontogeny from embryo to free-swimming tadpole and metamorphosis, illustrating the important contribution of sex-biased mortality, but secondary and inconsistent effects of sex reversal, on phenotypic sex ratios.

All methods were approved by Yale IACUC protocols 2013-10361 and 2015-10681 and CT DEEP Permit 0116019b.

The pattern of genotypic sex ratio in multiple populations from the embryo stage to metamorphosis is consistent with an important role for sex-biased mortality in amphibians. Although there is evidence of sex-biased mortality in juvenile mammals, birds, and even some fish (Kruuk et al., 1999; Cichon et al., 2005; Svensson et al., 2007; Orzack et al., 2015; Moran et al., 2016; Kato et al., 2017; Firman, 2019), to our knowledge, this is the first record of sex-biased mortality in larval amphibians in the wild. Importantly, we find the same pattern of male-biased mortality across two independent frog populations which inhabit contrasting environments, suggesting sex-biased mortality may be a relatively common phenomenon. Although we identified both directions of sex reversal, we found little evidence that sex reversal frequencies are consistently associated with phenotypic sex ratios. Even though sex reversal may not be the dominant mechanism undergirding phenotypic sex ratio variation, sex reversal still contributes to phenotypic sex ratios. In the populations we studied, male-biased mortality produced proportionally more genotypic females over larval development and additional sex reversal led to proportionally more phenotypic females than expected from genotypic sex ratios alone. The degree of sex-biased mortality we detected was modest (3–5% percentage change) but the degree of sex ratio change due to sex-biased mortality is sufficient to have important demographic consequences (Wedekind, 2017). To be clear, we identified sex-biased mortality as a mechanism that shapes sex ratios of individual wild amphibian populations, but our study was not designed to document whether sex-biased mortality shapes sex ratio variation across populations. Additionally, we note that our limited sample size of six populations may have had limited power to detect a relationship between sex reversal and phenotypic sex ratios, although the lack of a detectable pattern with these six population is notable. Sex-biased mortality, to our knowledge, is a previously undocumented process in larval amphibians and acts in concert with other processes like sex reversal to shape the phenotypic sex ratios of entire cohorts.

The causes for sex-biased mortality in this system remain to be studied, but both chromosomal and ecological processes may contribute. Heterogametic sex chromosomes (Y-chromosomes in species with XX-XY sex chromosome and W-chromosome in species with ZZ-ZW sex chromosomes) are often degenerate and accumulate deleterious alleles (Charlesworth and Charlesworth, 1997; Gordo and Charlesworth, 2001; Bachtrog, 2008; Miura et al., 2012; but see Kratochvil et al., 2021). As such, these degraded sex chromosomes can reduce survival and longevity in a diversity of animals (Xirocostas et al., 2020). Given R. clamitans has an XX-XY sex chromosome system (Lambert et al., 2016b), R. clamitans Y-chromosomes may be degraded (although there is no evidence for sex chromosome heteromorphy) which could explain higher rates of genotypic male mortality early in life. The absence of YY-individuals in our study may provide additional support for this hypothesis if YY-embryos are frequently unfit, as is often expected (Bull, 1983; Charlesworth, 1996). However, embryos with two Y-chromosomes may also never exist in wild populations if ecological factors limit reproduction by male-to-female reversed frogs.

Ecological differences between genotypic female and male larvae may also contribute to sex-biased mortality. For instance, pilot laboratory data suggest that male wood frog (Rana sylvatica) tadpoles exhibit higher activity patterns than female tadpoles (Skelly et al., unpubl.). Higher activity in tadpoles is associated with higher mortality due to predation (Skelly, 1994). If R. clamitans larvae similarly display sex-specific behaviors, then there is the possibility for sex-biased predation that could contribute to the sex-biased mortality we observed. Further, sex-specific behaviors may also influence sex-specific sampling efficacy. For instance, if sexes differ in microhabitat preferences, different types of sampling approaches may be biased toward one sex or another. Although identifying this issue is beyond the scope of our study, we sampled through the entire water body of each pond, including the benthos and surface and so presumably this bias should not influence our results. In mammals and birds, evidence also suggests that stress or poor health in parents can lead to sex-biased mortality in embryos and offspring (Kruuk et al., 1999; Muller et al., 2005; Perez et al., 2006; Firman, 2019). It is possible that reduced body condition in breeding R. clamitans similarly contributes to sex-biased mortality of embryos and tadpoles. Additionally, sex-biased mortality may be a result of female and male differences in tolerance of various environmental conditions like temperature, as has been seen in birds (Goth and Booth, 2005; DuRant et al., 2016; Eiby et al., 2008). Support for these and other hypotheses require further research.

Our analyses provide some evidence for environmental conditions associated with sex reversal. Specifically, our data suggest that natural water chemistry (dissolved oxygen, pH, and temperature) are related to genotypic female R. clamitans developing as phenotypic males. Experimental evidence in fishes shows that sex reversal can result from variation in dissolved oxygen and pH, and so these conditions may be relevant to amphibian sexual development (Romer and Beisenherz, 1996; Oldfield, 2005; Shang et al., 2006). We also have correlative evidence of temperature-mediated sex reversal where cooler temperatures contribute to higher rates of female-to-male sex reversal. Intriguingly, the opposite trend occurred for male-to-female sex reversal where warmer temperatures contribute to more male-to-female reversal, although the strength of evidence was weak. Temperature-mediated sex reversal in amphibians has been explored in laboratory experiments (Witschi, 1929; Wallace and Wallace, 2000; Lambert et al., 2018) but, to our knowledge, this is the first evidence from wild populations. Sex reversal in amphibians has, in recent decades, been assumed to be strictly caused by environmental pollutants or to unnatural conditions imposed by laboratory experiments (Hayes, 1998; Eggert, 2004; Nakamura, 2013; Orton and Tyler, 2015). This contrasts with work over a century ago that assumed sex reversal was largely a natural process (King, 1909, 1910; Witschi, 1929). Our findings here and previously in both metamorphosing and adult R. clamitans – in addition to recent work on European Agile Frogs (Rana dalmatina) – suggest that sex reversal can be natural and not necessarily a response to abnormal conditions because sex reversal has been observed under natural conditions in the wild for both species (Lambert et al., 2019; Nemeshazi et al., 2020). The conditions (dissolved oxygen, pH, and temperature) correlated with sex reversal vary naturally but can also be influenced by human land use (Holgerson et al., 2018; Lambert et al., 2019). However, in our study, these variables were uncorrelated with suburban land use (all p ≥ 0.14), suggesting that female-to-male reversal is largely natural. Although, our analysis did provide limited support that suburban land used increased the rate of male-to-female sex reversal. These patterns should encourage further work on sex reversal and the relative effect of natural and anthropogenic causes in wild amphibians.

Temporal variation may complicate identifying relationships between the environment, sex ratios, and sex reversal. For instance, in our two focal populations (Forest5 and Septic7), the presence, direction, and frequency of sex reversal varied not only between the two populations but also between years (Figure 5). This resulted in among-year differences in phenotypic and genotypic sex ratios (Figure 6). Among-year variation may explain why our results seemingly contrast with previous research. In prior work in this system, metamorphs sampled across 13 populations in 2012 consistently had male-biased phenotypic sex ratios across multiple forest ponds but equal or female-biased sex ratios in suburban ponds (Lambert et al., 2015). We found no such association between the environment and sex ratios in metamorphs sampled in 2016. Additionally, due to drought, we had a smaller number of populations (n = 6) available for study. Although this sample size is like what is often used in similar studies (Hayes et al., 2003; Orton and Routledge, 2011; Papoulias et al., 2013), it is half of the sample size of our prior work (Lambert et al., 2015). How yearly differences in environmental conditions influence sex ratios and sex reversal is unclear but documenting interannual variation is likely important for understanding the natural and anthropogenic contributions to wild amphibian sexual demography. Importantly, work in fishes and reptiles suggests that climate change is having and will continue to shape the sexual demography of wild animal populations (Ospina-Alvarez and Piferrer, 2008; Montalvo et al., 2012; Holleley et al., 2015). Additionally, experimental research in amphibians provides unambiguous evidence that temperature can modulate amphibian sex reversal (Lambert et al., 2018; Miko et al., 2021). Given the associations between drought, temperature, sex reversal, and sex ratios observed in R. clamitans here, there is a need to understand how climate change will shape sex reversal and population demography in wild amphibians.

We also found that genotypic sex ratios of individual clutches are variable and can deviate from parity. Breeding by sex-reversed individuals could produce a diversity of potential genotypic sex ratios (Figure 7) which, in aggregate, could hypothetically shape cohort sex ratios. Four clutches (two from each pond) had male-biased genotypic sex ratios. The most likely explanation for this pattern is that male-to-female reversed adults successfully breed, although further work is needed to document breeding by sex-reversed frogs. Interestingly, we also found two clutches in our focal forest pond which had female-biased genotypic sex ratios. Mating between only two frogs – a sex-reversed genotypic female and a non-sex reversed female – would produce a clutch with only genotypically female embryos (i.e., all XX sex chromosome genotypes). However, the female-biased clutch genotypic sex ratios we observed were ∼ 2/3 female, not entirely genotypically female. The only other study, to our knowledge, to assess clutch sex ratios was on European Common Frogs (Rana temporaria) and also found several clutches that were ∼ 2/3 genotypic female (Alho et al., 2010). Alho et al. (2010) interpreted this as a result of multiple paternity where one father was a sex-reversed XX-male. While the drivers of female-biased (but not entirely female) clutches are unclear, multiple paternity and “clutch piracy” is known to occur in frogs (Laurila and Seppa, 1998; Vietes et al., 2004) and so these sex ratios could potentially be the result of mating between a sex-reversed genotypic female, a sex-concordant male, and a sex-concordant female. Our individual clutch genotypic sex ratios provide suggestive evidence that sex-reversed frogs breed in the wild which may skew individual clutch and entire cohort genotypic sex ratios at fertilization (Figure 1). Direct observations are still needed to confirm whether sex-reversed adults successfully reproduce and whether this breeding can sufficiently skew entire cohort genotypic sex ratios. We note that because our clutch samples represent field-collected embryos that were hatched in the lab, it is possible that sex-biased mortality occurred during lab hatching, skewing our inferences of clutch sex ratios. Even so, mortality in our lab-hatched embryos was negligible and likely did not impact our inferences here.

Our work has implications for how we study the role sex variation plays in population, community, and ecosystem biology. The growing recognition that sex reversal in amphibians is common including in environments assumed to be natural and minimally impacted by people – including our observations here – highlights how a binary model of sex (i.e., female vs. male) limits how we study the biology of sex (Alho et al., 2010; Lambert et al., 2019; Nemeshazi et al., 2020). In the case of amphibian sex reversal, we can conceive of at least four sex categories depending on the combination of an individual’s genotypic and phenotypic sex. However, these sex categories may still be incomplete as sexual phenotypes and functions may occur along a gradient rather than discrete categories and sex-reversal may create new emergent behavioral and morphological phenotypes that differ from expectations based on constituent phenotypic or genotypic sexes (Li et al., 2016). Sex ratios can influence intraspecific interactions in a population and sex differences in behavior or morphology can shape predation rates and impacts on community and ecosystem processes (Lode et al., 2004, 2005; Weir et al., 2011; Liker et al., 2014; Fryxell et al., 2015). Given this, sex reversal combined with sex-biased mortality has the potential to influence the role of individuals within populations as well as how populations of different genotypic and phenotypic sex compositions influence the broader biological community and ecosystem properties. There has been an increasing call for biologists to rethink our understanding of sex roles and sexual variation in amphibians may open opportunities to expand this area of inquiry (Ah-King and Ahnesjo, 2013; Monk et al., 2019; Warkentin, 2021).

We document multiple processes that work together to shape phenotypic sex ratios of metamorphosing tadpoles. Particularly noteworthy is the consistent pattern of male-biased mortality between two disparate populations. Sex-biased mortality among pre-metamorphic life stages has, to our knowledge, never been documented before in amphibians and represents a previously-unknown contribution to amphibian population demography. Sex-biased mortality, particularly in the context of sex reversal, has the potential to shape amphibian ecological and evolutionary dynamics due to the underappreciated influence on both the genotypic and phenotypic sexual makeup of populations. We further highlight how phenotypic sex ratios may not be a reliable proxy for important biological processes. For example, sex-biased mortality could skew genotypic sex ratios in one direction, but sex reversal could bring phenotypic sex ratios back to parity; superficially this equal sex ratio would appear unremarkable whereas it is in fact shaped by the joint effects of sex-biased mortality and sex reversal. Continued exploration of the processes underlying the sexual demography of wild populations will be essential not only to our general understanding of population biology but to conserving populations and species in the face of global environmental change.

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.

The animal study was reviewed and approved by Yale IACUC Prototocols 2013-10361 and 2015-10681.

ML, TE, and DS conceived the project idea, contributed to the writing and editing of the manuscript. ML performed field and laboratory procedures and conducted statistical analyses. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.