MODELING FAUNAL RESPONSES TO CLIMATIC GRADIENTS WITH GIS - LAND SNAILS AS A CASE-STUDY

This study describes and exemplifies a multivariate approach based on geographical information system (GIS) technology for the analysis of f aunal responses to climatic gradients. The approach is particularly su itable for the analysis of museum collections, and it combines the dat a reduction capabilities of multivariate techniques (ordination and cl assification) with GIS tools for the manipulation of geographically re ferenced data. The specific steps of the analysis are (1) construction of a grid covering the study area at an appropriate resolution, (2) o verlaying the grid with a data base of spatially referenced species re cords, (3) determining the number of times each species was recorded i n each cell, (4) constructing a matrix of grid cells by species where each entry indicates the number of records of a particular species in a particular cell, (5) removing cells whose total number of records is lower than some threshold (to be determined) from the matrix, (6) ana lysis of the reduced matrix using ordination and classification techni ques, (7) construction of maps representing the results of the multiva riate analyses, and (8) analysis of these maps with respect to digital maps of relevant climatological factors. The applicability of this ap proach was evaluated by analysing the response of the land snail fauna of Israel to regional variation in mean annual rainfall. As was expec ted, patterns of faunal variation were significantly correlated with u nderlying variation in rainfall. However, the per-unit effect of rainf all on the composition of the studied fauna was much greater in dry re gions than in more rainy areas. Above 450 mm, no relationships could b e detected between the observed patterns of faunal variation and rainf all. These patterns were consistent over a wide range of grid cell siz es (5-20 km), and were robust to outliers. The overall results indicat e that the integration of GIS tools with standard multivariate techniq ues may serve as a valuable methodology for the identification and int erpretation of regional patterns of faunal variation.