The role of bet-hedging in the life-history strategy of the red-crowned toadlet, Pseudophryne australis (Gray, 1835) (Anura: Myobatrachidae)

<p dir="ltr">The red-crowned toadlet, Pseudophryne australis, an Australian myobatrachid frog, was used to test predictions and assumptions about the evolution of life-history strategies. Field studies showed that this frog uses a reproductive strategy not common among amphibians. It breeds throughout the year producing small clutches of relatively large eggs, and thus provides an opportunity to investigate the evolution of iteroparity (repeated breeding). Iteroparity enables the spreading of reproductive effort over many events and it is generally assumed that it is an adaptation to prevent total reproductive loss. Two hypotheses have been proposed for the evolution of iteroparity. The first predicts that it evolves in uniform environments and that the spreading of reproductive effort enables the organism to optimise its reproductive success in response to directional selection. The second hypothesis is almost the opposite. It proposes that iteroparity is selected for in stochastic environments, is a form of 'bet-hedging', and is a response to disruptive selection. It is the variance in reproductive success in these stochastic environments that is hypothesised to drive selection for this bet-hedging strategy. To test these hypotheses it is possible to compare the predicted reproductive output (offspring phenotypes) and life history features. If selected in a uniform environment offspring should have low phenotypic variation and traits should reach an optimum. If selected in a stochastic environment offspring should be variable, without an optimum phenotype, as selection is disruptive. I tested these alternative hypotheses on the red-crowned toadlet by examining the assumptions they were based on and the predictions they make. I first investigated the nature and extent of iteroparity in the red-crowned toadlet. This frog is 'continuously iteroparous' in contrast with the majority of frogs in the mesic climatic zones of Australia. Field studies show that this frog breeds in all seasons of the year, and over many years. Males call for mates all through the year, and individual females return repeatedly to terrestrial 'nest' sites to oviposit. An average of five clutches a year over a sequence of 34 clutches was documented for one female. I tested the prediction that this repeated breeding is a selective response to environmental stochasticity. Field studies show that high mortality in the embryonic stages is caused by desiccation of eggs and embryos (recruitment rate 0.8% over three years at one study site). I conclude that iteroparity in this species has evolved in response to disruptive selection related to the stochastic nature of a specific habitat component of the environment in which they live, and represents a bet-hedging strategy. Unpredictable rainfall, when combined with the fact that nests are terrestrial, and the ponds on which their tadpoles rely are small and ephemeral, leads to high variance in reproductive success. Other frogs in the region which lay their eggs in permanent ponds or creeks are less affected by the stochasticity of rainfall, as it does not result in desiccation of their breeding sites and high reproductive losses. A further prediction which stems from the hypothesis that iteroparity evolves response to stochastic environmental conditions, is that clutch size should be variable across reproductive events. In nature and in the laboratory clutch size varied considerably. Clutches ranged from 9 to 51 eggs with a mean of 25 eggs, and varied considerably within and among individuals. The large residual reproductive potential (the difference between 25 and 51) provides strong supporting evidence for the hypothesis that producing fewer eggs may be a way of reducing the cost of total losses, and thus reducing the mean variance in reproductive success. In the field and in laboratory experiments embryos of red-crowned toadlets hatch over a wide range of stages (Gosner 24 - 36) and ages (15 - 119 days). This staggering of hatching stages and age (length of embryonic period) was independent of the environmental conditions. I present the case that these traits represent a form of 'diversified bet-hedging', a concept known from the investigation of insect diapause and plant germination. It has been demonstrated in chestnut weevils, for example, that diapause for one sibship is staggered over a period of 2 - 3 years as a way of spreading risk and avoiding total reproductive failure. In desert annuals, seeds that do not germinate under good conditions in the first year germinate under the same conditions in following years. This is the first time 'diversified bet-hedging' has been related to the hatching dynamics of an amphibian. Investigation of whether maternal provisioning, measured as the size of the yolk, was the proximate cause of variable hatching dynamics, indicated that this diversification of offspring phenotypes was not related to size per se. It is hypothesised that it may be related to genetic factors or possibly the nutritive value of the yolk independent of size. Red-crowned toadlets lay eggs when weather conditions are suitable, rather than at regular intervals i.e. they are opportunistic breeders. This may explain why egg sizes are variable within clutches and between clutches. The coefficient of variation (CV) of egg sizes within a clutch ranges between 2.5 and 8.5%. It may also be an explanation for why the predicted trade-off between egg size and clutch size is not observed in this species (r = -.25; P = 0.43). Clutch size was positively correlated to the length of the period between oviposition (r = 0.69; P < 0.00 I). Concomitantly the coefficient of egg sizes declines within a clutch (r = 0.6; P = 0.02). I found no evidence that variation in egg sizes within a clutch was adaptive in relation to the hatching dynamics of this species. Furthermore, the high variability of egg sizes and clutch sizes contrasted with predictions from optimality theory and modelling that species that breed more frequently should have uniform egg and clutch sizes, as they have more opportunities to optimise their reproductive investment. The variability observed in clutch and egg size extend to features of larval growth and development. Larval duration was variable, and although there was some indication of a plastic response to desiccation of their ponds (the prime factor affecting fitness), there was strong evidence of diversified bet-hedging, with a wide spread of ages at metamorphosis, irrespective of treatment. As metamorphs with longer larval periods were greater in mass (r2 = 0.55; P < 0.001), which is considered to confer selective advantages in some circumstances, it is possible that the selection for rapid metamorphosis was countered by selection for larger size at metamorphosis. Thus iteroparity and its associated consequences as a life history strategy can evolve under very differing circumstances. On the one hand, if iteroparity evolves within a uniform environment, the prediction is that clutch and egg size as well as embryonic growth and development will be uniform and optimised. On the other hand, when iteroparity evolves in response to stochastic environmental conditions, the outcome is variation in all these traits.</p>