59248

Towards mechanistically predicting animal population abundance and distribution within a real landscape

Favoritar este trabalho

We present preliminary results concerning a central problem in Ecology with applied implications: to mechanistically predict animal population abundance and distribution within a real landscape. We approach it by combining Resource-Area-Dependence Analysis (RADA) with individual-based modeling (IBM). Common buzzards Buteo buteo in lowland UK were used to exemplify it. RADA determined that a buzzard requires, on average, a tree for roosting, 0.56ha of rough-ground and 15ha of grassland (good habitats for small mammals). This information was used to define, in the IBM, virtual animals with realistic resource-related parameters and their values. Rules concerning maximum foraging distances and territorial behavior were then included. The model was run on the 1990 Land Cover Map of Great Britain. Its outputs were: at the individual-level, home-range area, perimeter and proportion of overlap (proxy for territoriality), and, at the population-level, overall range, local density and carrying capacity of the landscape for buzzards. Virtual buzzards’ home-ranges and their pattern of overlap were indistinguishable from those of wild buzzards. When compared with two independent field-based estimates, predictions for carrying capacity of 211-226 individuals (100 model runs) suggested buzzards had recovered from previous low levels and reached equilibrium density in the area. As our approach relies on remote sensing for data acquisition, it allows for modeling animals roaming over areas that are huge, dangerous or difficult to access, and for using historical and contemporary datasets and techniques. Future advancements should allow for modeling of social or non-territorial species, and for considering landscape management scenarios in a climate change context.