Portal de Periódicos da CAPES

HABITAT DISTRIBUTION MODELS: ARE MUTUALIST DISTRIBUTIONS GOOD PREDICTORS OF THEIR ASSOCIATES?

2005; Wiley; Volume: 15; Issue: 1 Linguagem: Inglês

10.1890/03-5344

1939-5582

David Gutiérrez, Pilar Fernández, Adrian S. Seymour, Diego Jordano,

Ecology and Vegetation Dynamics Studies

Because of time and resource limitations, extensive studies of distributions at fine resolutions over entire landscapes are not viable, and distribution maps must be delineated using static habitat models. In cases where short life cycles and time limitations prevent the collection of detailed abundance and distributional data to generate habitat models for target species, one potential alternative is to use models of the abundance or occurrence of closely associated species (e.g., prey, host plant, mutualist) to predict the abundance or occurrence of the target species. We present a predictive model for the habitat of the butterfly Plebejus argus in scrubland in a protected area in southern Spain (Doñana National Park), based on the distribution of its mutualist ant Lasius niger, field habitat data, and topographical variables from a digital elevation model. The frequency of nests of L. niger was by far the major predictor of P. argus abundance and presence–absence. In turn, high L. niger frequency was associated with low elevations, presence of heathland vegetation, and intermediate latitudes within the site, reflecting the depth of the water table. L. niger frequency predicted by models from 50 calibration 100-m grid squares was significantly correlated with observed frequencies in 30 independent evaluation squares. Observed P. argus abundance in the same evaluation sites was more closely correlated with predicted L. niger frequency alone than with predicted P. argus abundances from a model including L. niger and additional topographical variables. In contrast, P. argus presence– absence was better predicted by models containing topographic variables as well as L. niger frequency, although it was also closely related with predicted L. niger frequency alone. Thus, our study shows how to use modeled species distributions to predict those of their associates. We further suggest that the models could be used to assess the potential effects of the declining water table depth and the resulting habitat changes on P. argus distribution and abundance in Doñana.