A predictive model of bird dispersal

The use of Bayesian statistical methods is becoming increasingly popular in ecology. A key feature of these methods is that they can use prior information when modelling systems and making predictions. Prior information is usually incorporated into the modelling process in the form of a probability density function, which may be estimated from existing data or through expert elicitation. While Bayesian analyses are becoming more popular, the use of strongly informative
priors remains rare.

Dispersal distance is an important ecological parameter, but is difficult to measure and estimates are scarce. General models that provide informative prior estimates of dispersal distances will therefore be valuable. Using a world-wide data set on birds, the researchers developed a predictive model of median natal dispersal distance that includes body mass, wingspan, sex and feeding guild. The model predicts median dispersal distance well when using the fitted data and
an independent test data set.

Using this model, the researchers were able to predict a priori estimates of median dispersal distance for 57 woodland-dependent bird species in northern Victoria. These estimates are then used to investigate the relationship between dispersal ability and vulnerability to landscapescale changes in habitat cover and fragmentation. They found evidence that woodland bird species with poor predicted dispersal ability are more vulnerable to habitat fragmentation than those
species with longer predicted dispersal distances, thus improving the understanding of this important phenomenon.

The value of constructing informative priors from existing information was also demonstrated. When used as informative priors for four example species, predicted dispersal distances reduced the 95% credible intervals of posterior estimates of dispersal distance by 8–19%. Further, should we have wished to collect information on avian dispersal distances and relate it to species’ responses to habitat loss and fragmentation, data from 221 individuals across 57 species
would have been required to obtain estimates with the same precision as those provided by the general model.

More info: Georgia Garrard ggarrard@unimelb.edu.au

Garrard GE, MA McCarthy, PA Vesk, JQ Radford & AF Bennett (2011). A predictive model of avian natal dispersal distance provides prior information for investigating response to landscape change. Journal of Animal Ecology, DOI:10.1111/j.1365-2656.2011.01891.x


Leave a Reply