The abundant centre model (ACM) predicts that the suitability of environmental conditions for a species decreases from the centre of its distribution towards its range periphery and consequently its populations will become scarcer, smaller and more isolated, resulting in lower genetic diversity and increased differentiation. However, little is known about whether genetic diversity shows similar patterns along elevational and latitudinal gradients with similar changes in important environmental conditions. Using microsatellite markers we studied the genetic diversity and structure of 20 populations each of Anthyllis vulneraria along elevational gradients in the Alps from the valleys to the elevational limit (2500 m), and along a latitudinal gradient (2500 km) from Central Europe to the range margin in northern Scandinavia. Both types of gradients corresponded to a 11.5 °C difference in mean annual temperature. Genetic diversity strongly declined and differentiation increased with latitude in line with the predictions of the ACM. However, as population size did not decline with latitude and genetic diversity was not related to population size in A. vulneraria, this pattern is not likely to be due to less favourable conditions in the North, but due to serial founder effects during the post-glacial recolonization process. Genetic diversity was not related to elevation, but we found significant isolation by distance along both gradients, although the elevational gradient was shorter by orders of magnitude. Subarctic populations differed genetically from alpine populations indicating that the northern populations did not originate from high elevational Alpine ones. Our results support the notion that postglacial latitudinal colonization over large distances resulted in a larger loss of genetic diversity than elevational range shifts. The lack of genetic diversity in subarctic populations may threaten their long-term persistence in the face of climate change, whereas alpine populations could benefit from gene flow from low-elevation populations.

Genotyping using microsatellite markers

We extracted genomic DNA using a DNeasy Plant Mini Kit (QIAGEN, Hilden, Germany) starting from approximately 10 mg of dried material. Samples were genotyped at 17 microsatellite loci (AV2, AV3, AV7, AV8, AV10, AV12, AV14, AV23, AV-000290, AV-002128, AV-004868, AV-005692, AV-015354, AV-020270, AV-021012, AV-021224, AV-021803, for references see Van Glabeke et al. 2007 and Kesselring et al. 2013) in four multiplex reactions using the QIAGEN Multiplex PCR Kit (QIAGEN, Hilden, Germany).

Multiplex 1 contained loci AV23 (Van Glabeke et al., 2007), AV-000290 and AV-015354 (Kesselring et al., 2013). Multiplex 2 contained loci AV2, AV3, AV12 (Van Glabeke et al., 2007), AV-021012 and AV-021224 (Kesselring et al., 2013). Multiplex 3 contained loci AV7, AV8, AV10 (Van Glabeke et al. 2007) and AV-004868 (Kesselring et al., 2013). Multiplex 4 contained loci AV14 (Van Glabeke et al., 2007), AV-002128, AV-005692, AV-020270 and AV-021803 (Kesselring et al., 2013). We amplified each multiplex using the QIAGEN multiplex Kit (QIAGEN, Hilden, Germany).  Each multiplex reaction contained 1 x QIAGEN multiplex master mix and 0.2 μM of each primer in a total volume of 6 μl.

The PCR conditions were: 5’ at 95 °C, 30 cycles of 30’’ at 95 °C, 90’’ at 53 °C (55 °C for Multiplex 1) and 30’’ at 72 °C and a last step of 30’ at 68 °C. Reactions were performed using a Mastercycler nexus (Eppendorf, Hamburg, Germany). PCR products were separated using an automated sequencer (3730xl DNA Analyzer, Applied Biosystems). The data were analysed using Geneious 11.1.5 (https ://www.geneious.com, Kearse et al., 2012).

To estimate the error rate, we extracted and genotyped 5% of the samples twice. The mean error rate per sample was calculated as the number of errors divided by the total number of analysed loci within replicated samples. We randomly chose one of the repeated samples to continue with the analyses.

Spreadsheet programm (e.g. Excel)