Community ecology of southwest arroyo toads: Joint dynamics of toads, predators, and habitat

Background/Question/Methods

We demonstrate an approach to both modeling and inference that deals simultaneously with the dynamics of a focal species, habitat state, and predators of the focal species.  We used site occupancy models to construct a Markov-chain model that projects annual changes in occupancy of the southwest arroyo toad, availability of stream breeding habitat, and the predators of the toad.  At the site level, we estimated the degree to which dynamics of our state variables (toad, habitat, and predators) were conditional on the state of the other variables.  We used the approach to examine how disturbance affects co-occurrence of toads and predators.  Model parameters were estimated using a multi-season multi-state occupancy estimator for data collected over 7 years at 360 sites which occurred in low-disturbance (perennial) and high-disturbance (ephemeral) streams. Prospective modeling tools were then used to make inferences from the Markov model, including estimating stationary states and conducting a sensitivity analysis.

Results/Conclusions

Estimated occupancy dynamics for toads and predators were similar between stream types while, as expected, habitat dynamics differed substantially between ephemeral and perennial streams.  Ephemeral streams were characterized by frequent annual drying, leaving sites unavailable to breeding toads and their predators.  The mean occupancy probability across time of toads was lower in the ephemeral streams but the probability predators were present at sites occupied by toads was also substantially lower in ephemeral habitats.  The relationship between low co-occurrence and higher disturbance was corroborated by sensitivity analysis.  In both stream types, increasing the annual probability that suitable habitat shifts to unsuitable (disturbance) reduced the probability toads co-occurred with predators.  Habitat dynamics have an important indirect effect on predator-prey interaction frequency in this system, where increased disturbance led to reduced interactions between toads and predators. The ability to make this conclusion was only possible because we were able to jointly estimate and model the dynamics of our focal species, habitat, and predators.