A1207
Title: Tensions in ecological data collection: Integrating data sources in Bayesian design for species distribution modelling
Authors: Max Savery - Ghent University (Belgium) [presenting]
Stijn Luca - Ghent University (Belgium)
Abstract: An analysis of Bayesian design of surveys for ecological data collection when integrating Presence-Only (PO) and Presence-Absence (PA) data in a species distribution model is presented. Working in a simulated virtual environment, the effect of PA and PO data balance and contribution in varying data availability environments-both scarce and abundant-is analysed. The optimization of structured survey sites as well as unstructured locations for citizen scientists, conditional on previously collected data, is considered. Additionally, a number of important issues regarding the behaviour of the integrated model in the designs are addressed, including correlation between detection probability and sampling bias, the use of single-visit PA surveys, the prior availability of citizen science data, and the tension between likelihood components. Furthermore, design-based and model-based Bayesian design strategies are compared, highlighting how the balance of PA and PO data affects the outcome of different strategies in unexpected ways, particularly when considering prior data availability. Finally, the ability of the designs to detect changes in species distribution over time, an essential requirement of designs for biodiversity monitoring in a rapidly changing environment, is evaluated.
