Figure 6.

Results of STRUCTURE and LAPEA analyses for all 230 AFLP loci. A STRUCTURE and Laplacean Eigenvector analysis (LAPEA) was performed on 369 gulls, representing seven distinct species (taxon names indicated in the top row boxes in panel A). (A). STRUCTURE was used to explore values of K = 1 to K = 10. For each K 25 independent runs were performed. Consistency among runs was explored by the combined use of Structure Harvester, CLUMPP, and DISTRUCT. For each value of K, the most frequent mode is displayed, with the number of independent runs with this mode indicated to the right of each multicoloured panel. Individuals are depicted as vertical bars segmented in their calculated membership of each of the various clusters (reflecting ancestral populations). Each colour reflects the estimated relative contribution of each of the populations to that individual's AFLP-based genome. Individuals are posteriori sorted according to their taxon definition. K = 6 is the most optimal overall number of clusters in the total dataset, with 23 out of 25 runs with exactly the same mode of differentiation. At K = 7 two clear modes were visible, with the one shown here being most frequent (13 out of 25), but failed to differentiate fuscus from cachinnans. The other (12 out of 25, not shown) did represent a mode that could differentiate all seven taxa. (B). A graph showing for each of the 25 replicate runs for K = 1 to k = 10, the Ln probability of the data estimated by STRUCTURE. (C). LAPEA of 230 AFLP loci of all 369 gulls from seven different taxa. Shown are the first two Eigenvectors and the percentage of the total variance they explain. Each dot represents a single individual and its colour corresponds with those of the seven taxa in panel (A). None of the seven taxa could be clearly differentiated.

Sternkopf et al. BMC Evolutionary Biology 2010 10:348   doi:10.1186/1471-2148-10-348
Download authors' original image