Toxin cues affect cannibalism responses of cane toad tadpoles

In many species cannibalism is uncommon and involves non-selective consumption of conspecifics as well as heterospecifics. However, within their invasive Australian range cane toad larvae (Rhinella marina) specifically target and voraciously consume the eggs and hatchlings of conspecifics, often extirpating entire clutches. In contrast, toad larvae rarely consume the eggs and hatchlings of native frogs. Here, we use laboratory studies to demonstrate that this selective consumption is triggered by species-specific chemical cues: maternally-invested bufadienolide toxins that otherwise defend cane toad eggs and hatchlings against predators. We find that these cues stimulate feeding behaviors in toad tadpoles, such that the addition of bufadienolide toxins to the water column increases predation on eggs, not only of conspecifics, but also of native anuran species that are otherwise usually ignored. In contrast, we find that cannibalism rates on conspecific hatchlings are high and unaffected by the addition of bufadienolide cues. The maternally-invested toxins present in conspecific eggs may therefore be more easily detected post-hatching, at which point tadpole feeding behaviors are induced whether or not additional toxin cues are present. As bufadienolide cues have previously been found to attract toad tadpoles to vulnerable hatchlings, our present findings demonstrate that the same toxin cues that attract cannibalistic tadpoles also induce them to feed, thereby facilitating cannibalism through multiple behavioral effects. Because native fauna do not produce bufadienolide toxins, the species-specificity of these chemical cues in the Australian landscape may have facilitated the evolution of targeted (species-specific) cannibalism in invasive cane toad populations. Thus, producing bufadienalide toxins confers a cost (increased vulnerability to cannibalism in early life-stages) as well as a benefit (reduced vulnerability to predation by other taxa).

Methods

The data are from a series of laboratory experiments conducted in Australia, where eggs/hatchlings of invasive cane toads (Rhinella marina) and a native frog (Litoria tornieri) were exposed to a predatory cane toad tadpole in the absence versus presence of toxin (bufadienolide) cues. The response variable was the number of eggs/hatchlings consumed. Data were analysed as a binomial response to treatment.

Usage Notes

The data provided are the raw data as collected during the experiments. When we analysed the data, there were several instances where the models failed to reach convergence due to 100% survival in the Control treament. To account for this and obtain a conservative estimate of effect, we allocated one tadpole to have died in the Control treatment. Specifically, we did this for: Experiment 1 C. albogutata (TadPresent versus TadAbsent), P. ornatum (TadPresent versus TadAbsent), R.marina (TadPresent versus TadAbsent) Experiment 2 Egg stage 14 analysis (toxin absent versus present). Details of these adjustments are given in our manuscript and in the attached README file. To follow our analyses in these specific instances, others need to make these adjustments. Also, note that the data sheet "Expt 2 Toad Egg-Hatchling Data" has data for both number of eggs eaten/not eaten and number of hatchlings eaten/not eaten. Thus, while eggs were developing and not yet hatched in this data set, the columns for TotalNSurvivingAtHatching, HatchlingsEaten and HatchlingsNotEaten are not relevant because eggs have yet to reach the hatchling stage (i.e., there are no hatchling data). In these instances, the cells for TotalNSurvivingAtHatching, HatchlingsEaten and HatchlingsNotEaten are filled with "n/a" (not applicable).