arXiv subjects

Jason Matthiopoulos

Publications and source records attributed to Jason Matthiopoulos.

3 recordsLinked to original sources

arXiv2018

Inference in MCMC step selection models

Habitat selection models are used in ecology to link the distribution of animals to environmental covariates, and identify habitats that are important for conservation. The most widely used models of this type, resource selection functions, assume independence between the observed locations of an animal. This is unrealistic when location data display spatio-temporal autocorrelation. Alternatively, step selection functions embed habitat selection in a model of animal movement, to account for the autocorrelation. However, inferences from step selection functions depend on the movement model, and they cannot readily be used to predict long-term space use. We recently suggested that a Markov chain Monte Carlo (MCMC) algorithm could define a step selection model with an explicit stationary distribution: the target distribution. Here, we explain how the likelihood of a MCMC step selection model is derived, and how maximum likelihood estimation can be used for inference about parameters of movement and habitat selection. We describe the local Gibbs sampler, a rejection-free MCMC scheme designed to capture important features of real animal movement. The sampler can be used as the basis for a flexible class of movement models, and we derive the likelihood function for several important special cases. In a simulation study, we verify that maximum likelihood estimation can be used to recover all model parameters. We illustrate the application of the method with data from a plains zebra.

stat.AP↗

arXiv2017

Linking resource selection and step selection models for habitat preferences in animals

The two dominant approaches for the analysis of species-habitat associations in animals have been shown to reach divergent conclusions. Models fitted from the viewpoint of an individual (step selection functions), once scaled up, do not agree with models fitted from a population viewpoint (resource selection functions). We explain this fundamental incompatibility, and propose a solution by introducing to the animal movement field a novel use for the well-known family of Markov chain Monte Carlo (MCMC) algorithms. By design, the step selection rules of MCMC lead to a steady-state distribution that coincides with a given underlying function: the target distribution. We therefore propose an analogy between the movements of an animal and the movements of a MCMC sampler, to guarantee convergence of the step selection rules to the parameters underlying the population's utilisation distribution. We introduce a rejection-free MCMC algorithm, the local Gibbs sampler, that better resembles real animal movement, and discuss the wide range of biological assumptions that it can accommodate. We illustrate our method with simulations on a known utilisation distribution, and show theoretically and empirically that locations simulated from the local Gibbs sampler give rise to the correct resource selection function. Using simulated data, we demonstrate how this framework can be used to estimate resource selection and movement parameters.

q-bio.QM↗

arXiv2014

Indirect effects of primary prey population dynamics on alternative prey

We develop a theory of generalist predation showing how alternative prey species are affected by changes in both mean abundance and variability (coefficient of variation) of their predator's primary prey. The theory is motivated by the indirect effects of cyclic rodent populations on ground-breeding birds, and developed through progressive analytic simplifications of an empirically-based model. It applies nonetheless to many other systems where primary prey have fast life-histories and can become locally superabundant, which facilitates impact on alternative prey species. In contrast to classic apparent competition theory based on symmetric interactions, our results suggest that predator effects on alternative prey should generally decrease with mean primary prey abundance, and increase with primary prey variability (low to high CV) - unless predators have strong aggregative responses, in which case these results can be reversed. Approximations of models including predator dynamics (general numerical response with possible delays) confirm these results but further suggest that negative temporal correlation between predator and primary prey is harmful to alternative prey. We find in general that predator numerical responses are crucial to predict the response of ecosystems to changes in key prey species exhibiting outbreaks, and extend the apparent competition/mutualism theory to asymmetric interactions.

q-bio.PE↗