Ecologists are reluctant to make use of existing data and information when conducting analyses because of concerns regarding inconsistencies in data collection, analysis and sample sizes. This means that ecologists often feel that they must collect all the data required for a study or, conversely, that there is insufficient data to conduct the study. But there is SO much data out there! Bayesian statistical methods enable us to utilise existing knowledge in the form of informative priors, but this requires that the available information be synthesised in such a way that it is useful.
By collating information on bird dispersal distances from around the world, we were able to develop a model that predicts median dispersal distance from species’ traits, which can be used as informative priors in a Bayesian analysis. Using this prior information, we were able to show – for the first time – that bird species with poor predicted dispersal ability show a greater response to habitat fragmentation than those species who are predicted to be able to disperse much larger distances (Figure 1: Garrard et al. 2011).
Figure 1*. Relationship between response to habitat fragmentation and predicted dispersal ability for 57 woodland bird species in northern Victoria. Response to habitat fragmentation (or its opposite, aggregation) is measured by the size of the influence that aggregation of habitat has on the occupancy rate of each species in the landscape. In other words, high values on the y-axis indicate those species for whom habitat aggregation has a large influence on occupancy in a landscape. There is a negative relationship between the size of response to aggregation and the predicted dispersal ability of the species (slope = -1.33), indicating that species with longer predicted dispersal distances appear less affected by habitat fragmentation than species with limited dispersal ability.
The belief that a species’ response to habitat loss and fragmentation is influenced by its ability to move between habitat patches has long been held, but demonstration of this relationship has been hampered by limited species-specific dispersal information. Only by synthesising dispersal information from multiple studies were we able to demonstrate this relationship for Victorian woodland birds.