How do CEMs Work

Before CEMs can be computed it is necessary to compile a set of species occurrence records and a set of suitable predictor variables (e.g. GIS layers containing information on climatic parameters). Some examples of freely available climate data sets are given in Kozak et al. 2008; Rödder et al. 2008, Chap. 22. The layers for each climate scenario commonly include the minimum and maximum temperatures and the mean precipitation per month (= 36 climatic parameters). Based on these monthly layers, bioclimatic parameters can be generated, e.g. with DIVA-GIS (Hijmans et al. 2001). Bioclimatic parameters are more useful than monthly values, since they are independent of latitudinal variations. This becomes obvious considering that the 'maximum temperature of the warmest month' is more useful than the maximum temperature of a specific month, since not everywhere the same month might also be the warmest.

Multi-collinearity among predictor variables may hamper the analysis of species-environment relationships because ecologically more causal variables may be excluded from models if other correlated variables explain the variation in response variable better in statistical terms (Heikkinen et al. 2006). Therefore, variable selection should be guided by a throughout assessment of the target species' ecology, and a rather minimalist set of predictors should be preferred. Specific adjustment of variable sets according to specific ecological needs of the target species may improve the model output.

Once species records and predictor variables have been compiled, a subsequent step involves the development of a multidimensional view of the climatic niche of a species, which is a considerable challenge given the complex nature of species' niches (Peterson and Vargas 1993). In plain text: in CEM, climatic information for species presence localities are summarized to an 'ideal' climatic niche for the target species (note that this can also be done with reliable absence data) that is afterwards compared to climatic conditions at the query localities, i.e. where the presence/ absence of the species is unknown. The results are geographic maps showing the similarity of an area with the 'ideal' climatic niche (Figs. 2a, and 3a). The selection of a suitable algorithm for the computation of the CEM depends on the amount of distribution records available, their quality, and the specific goal of the study (for a brief overview of available algorithms see Jeschke and Strayer 2008 or Rödder et al. 2008, Chap. 22).

Was this article helpful?

0 0

Post a comment