Simple individual-based models effectively represent Afrotropical forest bird movement in complex landscapes.
Abstract
Reliable estimates of dispersal rates between habitat patches (i.e. functional connectivity) are critical for predicting long-term effects of habitat fragmentation on population persistence. Connectivity measures are frequently derived from least cost path or graph-based approaches, despite the fact that these methods make biologically unrealistic assumptions. Individual-based models (IBMs) have been proposed as an alternative as they allow modelling movement behaviour in response to landscape resistance. However, IBMs typically require excessive data to be useful for management. Here, we test the extent to which an IBM requiring only an uncomplicated set of movement rules [the 'stochastic movement simulator' (SMS)] can predict animal movement behaviour in real-world landscapes. Movement behaviour of two forest birds, the Cabanis's greenbul Phyllastrephus cabanisi (a forest specialist) and the white-starred robin Pogonocichla stellata (a habitat generalist), across an Afrotropical matrix was simulated using SMS. Predictions from SMS were evaluated against a set of detailed movement paths collected by radiotracking homing individuals. SMS was capable of generating credible predictions of bird movement, although simulations were sensitive to the cost values and the movement rules specified. Model performance was generally highest when movement was simulated across low-contrasting cost surfaces and when virtual individuals were assigned low directional persistence and limited perceptual range. SMS better predicted movements of the habitat specialist than the habitat generalist, which highlights its potential to model functional connectivity when species movements are affected by the matrix. Synthesis and applications. Modelling the dispersal process with greater biological realism is likely to be critical for improving our predictive capability regarding functional connectivity and population persistence. For more realistic models to be widely applied, it is vital that their application is not overly complicated or data demanding. Here, we show that given relatively basic understanding of a species' dispersal ecology, the stochastic movement simulator represents a promising tool for estimating connectivity, which can help improve the design of functional ecological networks aimed at successful species conservation. Modelling the dispersal process with greater biological realism is likely to be critical for improving our predictive capability regarding functional connectivity and population persistence. For more realistic models to be widely applied, it is vital that their application is not overly complicated or data demanding. Here, we show that given relatively basic understanding of a species' dispersal ecology, the stochastic movement simulator represents a promising tool for estimating connectivity, which can help improve the design of functional ecological networks aimed at successful species conservation.