Locating bat roosts through the coupling of diffusion-type models and static acoustic detectors

Lead Research Organisation: CARDIFF UNIVERSITY
Department Name: Sch of Mathematics

Abstract

To improve knowledge of bat ecology and achieve more effective conservation, it is frequently necessary to locate roosts where the animals congregate, sometimes in large numbers. Unfortunately, finding roosts, or even determining whether a roost is likely to be present in a given area, is extremely difficult for both ecological consultants and research scientists. Specifically, field data collection is expensive and time consuming. Creating a theoretical framework for predictions, which can be validated in particular locations using smaller amounts of field data, is a novel and welcomed approach.
To understand better the patchy distribution of bats in the landscape and to maximise the success of locating roosts, we propose to combine the ecological expertise of Prof. Fiona Mathews (FM) and the modelling skills of Dr Thomas Woolley (TW) and Prof. Owen Jones (OJ). FM and TW have already worked together to produce a preliminary analysis showing that the probability of recording bats using acoustic surveys is strongly dependent on roost proximity. In order to consolidate this knowledge and extend the results this studentship will:
* create a stochastic random motion model of the bat population;
* parameterise the model using field data;
* simulate and provide analytical predictions of bat spatial spread and roost location;
* use information theory to estimate the optimal number of microphones (based on optimising field worker effort, error tolerances and financial constraints) and their spatial spread for bat detection;
* adapt and refine the model to include biased motion depending on real environmental data;
* code software for the biologists to provide immediate solutions to the problems above based on variable parameters that they choose;
The program of work has a well-structured learning curve in that many insights can be gained from simple stochastic models. As the student's confidence increases, we will seek to include further complicating factors, such as environmental heterogeneity and directed motion, to provide more realism in our results.

Publications

10 25 50