Genetic Differentiation and Localization of Glacial Refuges

The analysis of the allele frequencies of 61 samples of dataset A by both cluster analysis and PCA reveals concordant patterns. Whereas the samples from the Pyrenees - a comparably small geographic range - exhibit a striking genetic structure, partly with large genetic distances, the remaining populations sampled in a large part of the distribution area of C. auronitens show comparably small genetic differentiation (Fig. 3).

Fig. 3 (a) Neighbor-joining dendrogram based on the genetic distances (Nei 1978) of 61 populations of C. auronitens (dataset A). For population numbers, see Fig. 2. (b) Principal component analysis of C. auronitens using dataset A. Populations from the Pyrenees are separated from the other populations on the first two factors, which explain 40.5% of the total variance (factor 1: 27.5%, eigenvalue 7.98; factor 2: 13.0%, eigenvalue 3.77)

Fig. 3 (a) Neighbor-joining dendrogram based on the genetic distances (Nei 1978) of 61 populations of C. auronitens (dataset A). For population numbers, see Fig. 2. (b) Principal component analysis of C. auronitens using dataset A. Populations from the Pyrenees are separated from the other populations on the first two factors, which explain 40.5% of the total variance (factor 1: 27.5%, eigenvalue 7.98; factor 2: 13.0%, eigenvalue 3.77)

Moreover, the latter samples are clearly separated from those from the Pyrenees. A special case provides sample no. 29 from the easternmost part of the Pyrenees (Mt. Canigou). Whereas Assmann and Weber (1997) could not decide to which of the two groups (Pyrenees or non-Pyrenees) this sample clustered, we find evidence for a similarity to the Pyrenean C. auronitens populations but with a relatively strong genetic distance.

The analysis of dataset B (excluding the populations from the Pyrenees) offers more detailed information. Despite a generally lower genetic differentiation, three groups can be distinguished by means of PCA that can also be found in the tree: the samples originating from the disjunct areas of the Montagne Noire and the region around Rodez form their own groups, whereas the other populations cluster together (Fig. 4). Populations originating from sample sites that are localized South of the Holdhaus line are highlighted in Fig. 4a to give an impression of possible glacial refuge areas. In contrast to samples from Rodez and Montagne Noire, populations from the Cevennes (nos. 37-40) and populations from the Southern part of the Auvergne (nos. 70-71) do not cluster together but can be found - from a genetic point of view - in the midst of the samples from all over the Western part of the large distribution range of the study species.

The results support the hypothesis of multiple glacial refuge populations of C. auronitens (already given by Assmann et al. 1994), most of which behaved

Fig. 4 (a) Neighbor-joining dendrogram based on the genetic distances (Nei 1978) of 48 populations of C. auronitens (dataset B). For population numbers, see Figure 2. (b) Principal component analysis of C. auronitens using dataset B. Populations from Montagne Noire and Rodez are separated from the other populations on the first two factors, which explain 42.0% of the total variance (factor 1: 23.0%, eigenvalue 2.98; factor 2: 19.0 %, eigenvalue 2.47)

Fig. 4 (a) Neighbor-joining dendrogram based on the genetic distances (Nei 1978) of 48 populations of C. auronitens (dataset B). For population numbers, see Figure 2. (b) Principal component analysis of C. auronitens using dataset B. Populations from Montagne Noire and Rodez are separated from the other populations on the first two factors, which explain 42.0% of the total variance (factor 1: 23.0%, eigenvalue 2.98; factor 2: 19.0 %, eigenvalue 2.47)

Fig. 5 Glacial refuge areas of C. auronitens. Arrows show postglacial recolonization from two of these refuges, "?" indicates a putative additional refuge area or unknown migration routes (see text). The black line indicates the Holdhaus line

endemically or showed only restricted dispersal (i.e., in the Pyrenees or the Montagne Noire), whereas the populations from the Cevennes and the Southern Auvergne recolonized large parts of Western and Central Europe (Fig. 5). The strong power of dispersal of the populations at the edge of the distribution range is demonstrated by a recent recolonization event of C. auronitens populations in Northwestern Germany after anthropogenic landscape changes (Drees et al. 2008).

Was this article helpful?

0 0

Post a comment