Document: AND-3-83-4

Can static habitat models recover process information?

TYRE, A.J.* ¹, H.P.POSSINGHAM ¹ and D.B.LINDENMAYER ²

University of Adelaide, Glen Osmond, SA, Australia ¹
Australian National University, Canberra, ACT, Australia ²

Abstract:
Habitat models are widely used in wildlife ecology to identify good habitat for conservation actions like reservation. These models are usually based on correlations between species presence or abundance, and habitat characteristics like vegetation, geomorphology, or climate. Habitat models contain no temporal component. However, habitat quality for conservation is best measured as the population growth rate the difference between birth and death rates. Good quality habitat is habitat that allows a positive population growth rate, at least at low population densities. We test the ability of static habitat modelling to recover information about habitat quality with a spatially explicit, individual based simulation with a fractal landscape. The advantage of a fractal landscape is that the amount of spatial autocorrelation in habitat characteristics is completely controllable. Only a single habitat characteristic influences either survival or fecundity, and hence habitat quality, on the fractal landscape. Our "virtual ecologist" randomly samples territories on the landscape and records whether or not the territory is occupied and what the local habitat characteristic is. The virtual ecologist makes no sampling errors, and measures the only habitat characteristic that influences habitat quality. We find that even under these ideal circumstances, a static habitat model can only explain about 50% of the variation in territory occupancy; the unexplained variation is a consequence of demographic stochasticity alone. We find that the probability of dispersing to a site has a greater influence on whether or not it is occupied by a species than the quality of the site. These results have significant implications for the interpretation of habitat models based on static correlations, and the value of traditional habitat models for wildlife conservation.

Keywords: habitat model, habitat quality, neutral landscape model

This abstract is being presented at: 2:15 PM in session:
Oral Session #46: Modeling Populations and Statistical Ecology.