Changes in rodent community structure in the Chihuahuan Desert Mexico: comparisons between two habitats

From 1996 to 2001, we conducted a study of two adjacent rodent communities in the Mapimi Basin of the Chihuahuan desert in Mexico. The objective was to test the resource level hypothesis in explaining differences in two contrasting but spatially close habitats: a tobosa grassland area (Hilaria mutica) and a mesquite (Prosopis glandulosa)-creosote (Larrea tridentata)-prickly pear (Opuntia rastrera) shrubland. Based on the inherent differences in the two habitats, we made the following predictions relative to the resource level hypothesis: (1) shrublands should have higher inherent density, diversity, and richness of small mammals than grasslands; (2) annual and seasonal changes in density, diversity, richness, and community composition should be more pronounced in grassland than in shrubland. Concerning the first set of predictions, rodent density was higher in the shrubland (17.7 +/- 1.1 ind/ha) than the grassland (5.8 +/- 0.7 ind/ha; F-1,F-48 = 292.2, p less than or equal to 0.001). Overall species richness was equal in both habitats but the richness at any given time was significantly higher in shrubland (7.5 +/- 0.2 species vs. 5.6 +/- 0.3 species, F-1,F-48 = 35.0, p less than or equal to 0.001). Species diversity, however, was equal in both habitats. Concerning the second set of predictions, rodent densities differed significantly in both areas annually (grassland: F-5,F-24 = 10.9, p < 0.001; shrubland F-5,F-24 = 11.0, p < 0.001) and seasonally (grassland: F-1,F-24 = 27.2, p < 0.001; shrubland: F-1,F-24 = 15.3, p < 0.001). However, the variability in the relative number of individuals captured per year was higher in the grassland compared to the shrubland. Also, rodent densities in grasslands responded directly to the amount of precipitation of the previous calendar year while those in shrubland did not respond until November. The composition of the rodent community, as expressed by the first principal component of a principal components analysis varied yearly and seasonally for both shrubland (annual: F-5,F-24 = 14.6, p<0.001, seasonal: F-1,F-34 = 12.1, p > 0.001) and the grassland (annual: F-5,F-24 = 3.01, p = 0.03, seasonal: F-1,F-24 = 13.5, p = 0.001). However, we found changes in community structure in grasslands but not in shrubland were related to previous year precipitation levels. We suggest that differences between the two areas reflected the higher productivity and biomass stability in the shrubland compared to the grassland. Consequently, we concluded that the results in general supported the resource level hypothesis in explaining the structure and dynamics of rodent communities in the Chihuahuan desert. (C) 2004 Elsevier Ltd. All rights reserved.