Aim: We investigated the beta-diversity patterns of fleas parasitic on rodents along environmental and host turnover gradients using linear and nonlinear approaches. We asked (1) which factors explain a larger proportion of the variation in flea beta diversity and (2) whether the results of the linear versus the nonlinear approach are similar in indicating the relative roles of environment versus host turnover. Location: Mongolia. Methods: The linear approach was represented by a partial linear regression of flea turnover against rodent turnover, environmental variables and a spatial term. The nonlinear approach was represented by generalized dissimilarity modelling (GDM) with the response variable being the dissimilarity of flea composition and the predictors being dissimilarity in rodent composition, dissimilarity in environmental variables and geographic distances between localities. To test for the response of rodent beta diversity to environmental gradients, this factor was used as a response variable/matrix. Results: Partial regression analyses explained only 24% of observed variance. Environmental variables (mainly air temperature) and rodent turnover independently explained similarly small portions of flea turnover. Gradients of air temperature and rodent turnover were the most important factors in explaining flea turnover across space, whereas the altitudinal gradient and the gradient of annual variation in precipitation were the most important in explaining rodent turnover. The GDM resulted in 68.4% of explained deviance. The best predictor of flea species turnover was the air temperature gradient followed by rodent host beta diversity and, to a lesser degree, the precipitation gradient. The responses of rodent and flea turnover to environmental gradients differed. Main conclusions: Air temperature and rodent host turnover were the main factors affecting flea beta diversity with the latter responding to climate directly and not being mediated by host responses. Accommodation of the nonlinearity of species turnover responses to gradients allows patterns obscured by linear approaches to be revealed.
ASJC Scopus subject areas