You are here:
COMPARING ECOLOGICALLY SCALED LANDSCAPE INDICES WITH A SPATIALLY EXPLICIT POPULATION MODEL
Citation:
White, R D. COMPARING ECOLOGICALLY SCALED LANDSCAPE INDICES WITH A SPATIALLY EXPLICIT POPULATION MODEL. Presented at Seminar at University of Maine, Orono, ME, April 19, 2004.
Description:
Vos et al. (2001) proposed a class of landscape indices they called ecologically scaled. By this they meant that the indices incorporate species-specific characteristics that are assumed to be important for population viability. I used their two ideas of species carrying capacity and landscape connectivity, as further developed by Polasky et al. (submitted), to create nine versions of ecologically scaled landscape indices (ESLI). I then tested these indices against data from Schumaker et al. (2004) on the population persistence of 16 species of birds and mammals in alternative scenario landscapes for the Willamette River Basin, a 30,000 sq km watershed in western Oregon. The comparisons used as common currency the percent change in the various measures between the baseline scenario and four alternatives that were set both in the future and in the past. I then computed the difference between the percent change in the nine ESLI measures and the percent change in population size as simulated by the PATCH spatially explicit population model in Schumaker et al. I also computed the difference between the percent change in a simple measure of suitability-weighted habitat quantity and the percent change in population size, to give a total of ten different estimates of how well landscape pattern measures compared with results from a spatially explicit population model. The versions I used of the ESLI included those varying the weight of the contribution of neighboring patches relative to the weight of the patch itself, those varying the steepness of the exponential weighting function for neighboring patch contributions, and those varying the scaling of habitat suitability values. To evaluate the ten different measures I calculated the mean absolute difference, the median absolute difference, and the number of sign changes for each measure across 64 combinations of 16 species and four alternative landscape scenarios. Sign changes were species-landscape combinations where the percent change in the PATCH population size was positive, or negative, and the percent change in the landscape pattern measure was negative, or positive, respectively. Because one of the alternative landscape scenarios consisted of an estimate of a pre-EuroAmerican settlement landscape based on nineteenth century survey records, and therefore was quite different in method and structure from the other landscape scenarios, I also examined the performance of the ten measures for just the three other scenarios. The ESLI measure that added neighboring patches' contributions to the contribution from the patch itself scored consistently well in all tests. Several other versions of the ESLI scored well in some of the tests. Somewhat surprisingly, however, the simple measure of suitability-weighted habitat quantity also scored well in most of the tests. Although the behavioral basis of the ESLI suggests these indices as promising surrogates for more demanding estimates of population persistence, such as those from PATCH, continued research on these measures is necessary.