Discussion
Divergence in reproduction-related traits has been documented in congeneric species of salamanders and includes shifts in reproductive phenology (Burkhart 2017), in habitat selection (Caspers et al. 2015, Kieren et al. 2018, Weitere et al. 2004), or in clutch and egg sizes (Petranka 1987). Previous research suggests that these traits can diverge rapidly and can be coupled with genetic differentiation indicative of local adaptation (Evans et al. 2020, Weitere et al. 2004). This is consistent with our experiment as all measured traits remained well differentiated between A. barbouri and A. texanumwhen bred and raised under a common treatment and supports previous comparative studies (Sih and Maurer 1996). This suggests genetically determined differentiation between lineages rather than plasticity in response to different environmental conditions these species occupy.
The two species exhibited a different strategy in egg deposition, generally consistent with literature descriptions (Petranka and Straus 1989). Ambystoma barbouri females prioritized attaching their eggs to the undersides of rocks; however, once the available rocks were expended, they attached their eggs to vegetation (Figure 4). A. barbouri produced larger and fewer eggs per clutch, and the attachment of those eggs to rocks and plants was very firm and difficult to manually remove from the surface. Ambystoma texanum females used artificial plants as an oviposition site, and once all available vegetation was expended, eggs were scattered loosely throughout the aquarium (Figure 4). Ambystoma texanum produced smaller eggs but larger clutches which were laid in small clumps, lacked firm attachment, and would rapidly fall off the vegetation if disturbed. Upon oviposition, the A. barbouri eggs had a longer embryonic stage and larvae were bigger upon hatching but took a shorter time to metamorphosis and the metamorphs were smaller in comparison withA. texanum metamorphs.
These aforementioned traits are likely under selective pressure in the respective breeding habitats of the two species. Small streams, the preferred breeding habitat of A. barbouri , are generally more prone to strong flooding events in the early spring when eggs are deposited, and desiccation in the late spring and early summer when larvae are developing (Eng et al. 2016, Holomuzki 1991, Johnson et al. 2009, Petranka and Sih 1986, Petranka and Sih 1987). Therefore, we would expect A. barbouri to have adaptations to resist flowing streams and flooding events, such as the laying of eggs under rocks with strong adhesion and a longer egg development time to avoid larvae washing downstream in fish inhabited sections of streams (Petranka and Sih 1986). This strategy has been seen in other stream breeding salamanders and may adhere to the safe harbor hypothesis (Nussbaum 1987, Shine 1978). Larger initial larval size and shorter larval stage may then serve to cope with early stream desiccation -in contrast to more slowly desiccating vernal pools-, via faster metamorphosis as seen in other amphibians with rapidly desiccating ephemeral habitats (Nussbaum 1987). Additionally, access to available prey is critical for fast larval development and metamorphosis, and a larger body size has been shown to increase capture rates of large prey (Takagi and Miyashita 2019). Although additional exploration of specific A. barbouri stream invertebrate assemblages is necessary, streams have been generally documented as having high invertebrate populations and low zooplankton concentrations (Holomuzki 1991, Wan Maznah et al. 2018). Hatching at a larger size may therefore give newly hatched A. barbouri access to the large invertebrate prey and facilitate quick metamorphosis (Mcwilliams 1989, Mcwilliams and Bachmann 1989, Smith and Petranka 1987, Takagi and Miyashita 2019, Nussbaum 1987).
The larvae of A. barbouri also maintained darker pigmentation than those of A. texanum (Figure 3.). Larval pigmentation is also likely under selective pressure, with A. barbouri likely coming under higher ultraviolet (UV) stress than A. texanum due to shallowness of first-order streams inhabited by A. barbouri and higher levels of UV penetration (Storfer et al. 1999). Previous studies have noted the adaptive advantage of the darker coloration of A. barbouri larvae living in shallow streams with high levels of UV radiation as opposed to deeper, darker vernal pools inhabited byA. texanum larvae (Garcia et al. 2003, Garcia et al. 2004, Garcia and Sih 2003, Storfer et al. 1999).
The selective pressures faced by stream breeding A. barbouri are in contrast with the vernal pools occupied by A. texanum , which do not tend to experience flooding events resulting in strong currents, often maintain water well into summer, and typically have plentiful zooplankton populations, as reflected in the diet of young A. texanum (Gamble and Mitsch 2009, McWilliams and Bachmann 1989). Selection against A. texanum traits in stream habitats might therefore be quite strong. If a female A. texanumoviposits in a stream on either vegetation or on loose substrate, the eggs may then be easily carried away by flooding events, and the newly hatched larvae may not have small enough prey available to them. Our results suggest that populations of A. barbouri and A. texanum are under a number of strong selection pressures to oviposit in the correct breeding habitats.
A potential caveat to our findings is transgenerational plasticity in breeding adult traits (i.e., oviposition location, clutch size, offspring size), where traits of offspring are determined by environmental conditions experienced by their parents (Richter-Boix et al. 2014). In our system, this would mean that environmental differences during early development of adults (i.e., development in stream or vernal pool) would affect their breeding behavior and traits of their offspring (Bell and Hellmann 2019, Herman and Sultan 2011, Richter-Boix et al. 2014). Although transgenerational plasticity can have effects on traits such as the size and morphology of metamorphs, it typically arises due to differences in environmental conditions (e.g., high versus low food availability), which were identical for A. barbouri andA. texanum in this study. Additionally, transgenerational plasticity does not influence complex reproductive and developmental traits, as would be necessary in this system (Orizaola et al. 2016, Richter-Boix et al. 2014). Therefore, we believe that the observed differences in reproductive and developmental traits is the result of differing selection pressures that these species face in their respective environments.
Despite the trait differentiation between A. barbouri andA. texanum, the two species readily hybridized. The hybridization between these two species has been suspected (Denton et al. 2014), but had never been documented, bar one induced hybridization (Petranka 1987). In our experiment, hybridization readily occurred at similar rates between conspecifics and heterospecifics. This is despite reported differences in courtship between A. barbouri and A. texanum , such as more cryptic courtship and generally less nudging behavior by females towards males in A. barbouri in comparison with A. texanum (Arnold 1972, Garton 1972, McWilliams 1992, Licht and Bogart 1990, Petranka 1984, Wyman 1971).
We did not observe any immediate negative fitness consequences of hybridization. The hybrid crosses did not exhibit reduced viability of eggs or larvae, or reduced survival to metamorphosis. However, we recognize that hybrid vigor is common (Birchler et al. 2010, Cogălniceanu et al. 2020, Fitzpatrick and Shaffer 2007, Hotz et al. 1999, Nzau Matondo et al. 2007), and that fitness consequences, if any, are more likely to manifest in natural habitats. Indeed, in species and populations that exhibit divergent traits linked to their local environments, hybrids are typically expected to have inferior fitness in either parental habitat (Moser et al. 2016). Interestingly, offspring ofA. texanum females had very poor survival in mesocosms, despite mesocosms closely mimicking non-flowing, high leaf litter, high zooplankton environments (Kneitel and Lessin 2010). This is in comparison to A. barbouri female offspring, which had very high survival in mesocosms. We are uncertain of what may have caused this result, but it may have been due to a difference between experimental conditions and typical A. texanum habitat. These potential differences include increased water hardness, higher larvae density resulting in cannibalism or bullying, and higher UV radiation, as mesocosms were placed in fields with no canopy cover. This result warrants further investigation as its cause may be inhibiting A. texanum female migration into the geographic range of A. barbouri .
One mechanism that may lessen the negative impact of hybridization between phenotypically divergent species is the presence of maternal effect, where phenotypes of the offspring are more greatly influenced by the mother than would be expected from its genotype (Pfennig and Martin 2009). Maternal effects have been noted in amphibians previously (Crespi and Lessig 2004, Kaplan 1987, Pfennig and Martin 2009), especially when it comes to initial egg/larval size and further larval development influenced by the egg size (Bernardo 1996, Dziminski and Roberts 2006, Kaplan 1985, Kaplan 1992, Komoroski et al. 1998). Our hybrids exhibited a maternal effect in some traits associated with oviposition and fitness, such as clutch size, initial larval size, pigmentation, and survival to metamorphosis, especially in mesocosms (Figure 5). Conversely, a paternal effect was only observed in size at metamorphosis. We believe maternal effects are the result of selection pressure as intermediate traits would exhibit inferior fitness in either habitat as argued above. As the preferred oviposition substrate is dependent on breeding location, females are likely to be under selective pressure to have offspring which exhibit life history traits suited to their oviposition location, rather than paternal genotype. This maternal effect may further serve as an explanation to why hybridization between these species appears to be present, but intermediary phenotypes are not reported in the wild. Even if hybridization in the wild is readily occurring, egg masses and larvae will still appear to be the “correct” species in each breeding habitat. Selection strength on these traits, however, might differ among populations and depend on local environmental conditions. Indeed, our populations of both species came from their respective “typical” habitats, where we would expect selection pressures to be maximized. Other habitats where both species have been observed, such as roadside ditches, may impose weaker or different selection pressures on hybrid populations.
In summary, our results suggest that genetic adaptation, rather than phenotypic plasticity, plays a larger role in determining traits defining A. barbouri and A. texanum . Despite the propensity for A. barbouri and A. texanum to hybridize and the lack of fitness consequences in experimental conditions, the divergent selective pressures and resulting divergence in life history traits may be due to the two species being on distinct evolutionary trajectories (Abbott et al. 2013). Maternal effects on hybrid offspring may be impeding the divergence of these salamanders but may also be reducing the negative impact of hybridization. Further investigation into the possible presence of intermediary traits in non-typical habitats, coupled with genomic assessments should be conducted to fully understand the extent and consequences of hybridization between these two species of salamanders.