Page 15 - Dedola2011

algorithms implemented in the software package BAPS 5.2 (Corander and Tang 2007)
and STRUCTURE 2.2.3 (Pritchard et al. 2000).

BAPS analysis is used to infer the number of genetic clusters (K) in the data set,
where K is treated as an unknown parameter to be estimated. To compare the BAPS
results with the STRUCTURE results, the module for the nonspatial clustering of
individuals was used. We performed ten independent analyses to assess the reliability of
results. In each analysis the number of clusters (K) was estimated by a range of possible
values (K = 2 up to 35). The algorithm was reiterated on each predefined value of K five
times. The clustering solution with the highest mean posterior probability computed
over the ten independent runs was chosen as the correct partitioning on which the
admixture analysis was subsequently performed (Vaha et al. 2007).

STRUCTURE analysis, which assumes that K is known in advance, was used to
assess the occurrence of hierarchical levels of genetic structure. The method is hereafter
described: in a given dataset STRUCTURE identifies the uppermost hierarchical
structure that corresponds to the minimum number of clusters that captures the major
structure in the data (Pritchard and Wen 2004); after this first round data are partitioned
into smaller datasets according to the best clustering solution, and subsequent rounds
are performed on each subset of data. The procedure is reiterated until it is not possible
to partition the data any further. To detect the uppermost hierarchical structure present
in a given partition of data we used the method ofEvanno et al. (2005). Next, we took
into account the posterior probability of data for a given K. If both methods failed, the
individuai assignment patterns were examined using q value thresholds of 0.2/0.8 to
denote membership in the cluster (Vaha et al. 2007). For each data partition and for
each value ofK ten independent runs of STRUCTURE were performed by applying the
admixture model with correlated allelic frequencies (Falush et al. 2003, 2007). Each run

15

Gìan Luca Dedola: ANALISI DELLA VARIABILITÀ GENETICA DI PATELLA FéRRUGINEA, PATEL.LA UL l'SSIPONENSIS (MOLLUSCA: GASTROPODA) E PINNA
NOBILIS (MOLLUSCA: BIVALVIA): lLCONTRJBUTO DEl DATI MOLECOLARI ALLA CONSERVAZIONE DI SPECIE MINACCIATE - Tesi di douorato in Scienze
della natura e delle sue risorse • indiri7.zo BK>logi.a Ambientale, Università degli Studi di Sassarl