Page 6 - Patella_ferruginea_Casu_Rivera_ali2011

1298 Genetica (2011) 139:1293–1308

bootstrap confidence interval did not include r = 0 for a tests are expected under recent population expansion fol-
given distance class. lowing a bottleneck, selective sweeps or purifying selection
(Fu 1997; Ramos-Onsins and Rozas 2002). The signifi-
Mitochondrial DNA analysis cance was tested by performing coalescent simulations
(10,000 iterations) in DNASP 5.0.
Sequence data from 12S and 16S mitochondrial coding
genome have been used to infer the occurrence of phy-
logeographic patterns of limpets (Bird et al. 2007, 2011; Results
Nakano et al. 2010; Gonza ´lez-Wevar et al. 2011a, b).
Furthermore, studies on the COI gene of other Patella ISSR
species, most of which were designed to reconstruct the
phylogeny of the genus, also showed its potential use in The number and size of bands resolved per primer ranged
detecting major geographical discontinuities (Sa ´-Pinto from ten to 11 and from 500 to 1,800 bp, respectively
et al. 2008, 2010). Thus, the 12S and 16S genes were (Table 2). The Bayesian model-based clustering analyses
sequenced for 174 individuals, whereas the COI gene was showed a greater degree of genetic structuring in the SCR
sequenced for 110 individuals, being 64 COI sequences than SAS group (Fig. 3a, b). Indeed, the uppermost hier-
available on GenBank (accession no. AY996038- archical level of genetic structuring, as estimated from the
AY996043; GQ469872-GQ469874; GQ469876) (Table 1). first round of STRUCTURE analyses (Table 3), resulted in
Universal primers were used for COI (Folmer et al. 1994), a clustering solution of K = 3: cluster A roughly corre-
12S and 16S (Nakano and Ozawa 2004). PCR amplifica- sponded to samples from North-Western Sardinia (SCR),
tion was performed as described in Espinosa and Ozawa and cluster B roughly corresponded to samples from Cor-
(2006) with the following annealing temperatures: COI, sica, North-Eastern Sardinia, and Argentiera (SCR); cluster
54°C; 12S and 16S, 49°C. PCR products were purified with C corresponded to the all samples from the SAS group
ExoSAP-IT (USB Corporation) and were sequenced using (Table 3).
an external sequencing core service (Macrogen Inc., Seoul, The second and third round of STRUCTURE analyses
Korea). The sequences obtained were aligned using the (overall K = 7 and K = 11, respectively, Table 3) showed
software CLUSTAL W Multiple alignment tool in BIO- a further substructuring that was highly consistent with the
EDIT 7.0.5.2 (Hall 1999). clustering solution found by BAPS analysis which identi-
The number of haplotypes (h), number of polymorphic fied eight genetic clusters (Fig. 3b) seven of which corre-
sites (S), nucleotide diversity (p), and haplotype diversity sponded to groups of individuals found after the third
(H) were calculated to obtain estimates of the genetic round of STRUCTURE analyses (Fig. 3a). Both analyses
divergence among populations using the software package are consistent in showing a clear spatial genetic structuring
DNASP 5.0 (Rozas and Rozas 1999). The genetic rela- in North-Western Sardinia (SCR), where all the samples
tionships among the haplotypes were inferred using the analysed correspond to a different cluster (Fig. 3a, b). As
Median Joining Network analysis (Bandelt et al. 1999) well, samples from North-Eastern Sardinia were parti-
with the software package NETWORK 4.5.0.1 (www. tioned in two subclusters which however showed an
fluxus-engineering.com). overlapping range (Fig. 3a, b). In cluster C (Fig. 3a, b), two
The best probabilistic model of sequence evolution was main clusters were defined by both STRUCTURE (from
determined using the program JModeltest 0.1.1 (Posada the second round) (Table 3) and BAPS analyses, roughly
2008) with a maximum likelihood optimised search, using corresponding to the westernmost (Strait of Gibraltar and
the Akaike Information Criterion (AIC). JModelTest cal- Alboran Sea) and the easternmost (Siculo-Tunisian Strait)
culated as the appropriate models of nucleotide substitution samples from the SAS group. Nonetheless, Cape Bon
the following models: HKY for 12S; F81 for 16S; and sample grouped with the westernmost samples; similarly,
TPM2uf for COI. some individuals from the Alboran Island grouped with the
Maximum Likelihood (ML) analysis was performed easternmost samples (Table 3).
with the program PhyML 3.0 (Guindon and Gascuel 2003), The first two principal coordinates that resulted from
allowing the estimation of proportion of invariable sites PCA carried out on the entire dataset accounted for 60.24%
and the gamma distribution. Best of SPR and NNI branch of the total variation, and identified three groups of indi-
rearrangements were applied on ten BioNJ random starting viduals (Fig. 4a). These groups are consistent with the
trees, and bootstrapping with 1,000 replicates was set. highest hierarchical structure found in the first round of
The departures from mutation-drift equilibrium were STRUCTURE analyses (Fig. 3a). Furthermore, PCA ordi-
assessed using Tajima’s D (Tajima 1989) and Fu’s Fs (Fu nation based on subsets of individuals corresponding to the
1997) neutrality tests. Significant negative values for both aforementioned groups produced plots which tended to

123