Page 3 - Espinosa_alii_2013

268 F. Espinosa et al.

TABLE 1 Summary statistics for the size distribution of Patella ferruginea (only for sites where the species was detected). Site numbers
(Supplementary Table S1; Fig. 1) are in parentheses.

Location (site) n % ,3 cm Max. size (cm) Min. size (cm) Mean size±SD (cm) Skew±SE
Galeria (1) 40 47.5 6 0.6 3.14±1.15 0.046±0.374
Cargese (2) 10 20 6.2 1.9 4.16±1.32 −0.516±0.687
Tizzano (3–4) 4 75 5.6 1.6 2.97±1.78 1.735±1.014
Bonifacio (5) 1 100 2.4 2.4 2.4
Asinara (10) 40 12.5 9.3 1.9 6.14±2.03 −0.724±0.374
Capo Falcone (11) 8 75 3.4 1.4 2.43±0.66 −0.012±0.752
Olbia (18) 11 36.33 3.8 1.4 2.97±0.71 −0.993±0.661
Cala Moresca (20) 1 0 4.9 4.9 4.9
Porto Rafael (24) 19 47.36 6.6 0.4 3.05±1.63 0.295±0.524
Capo Testa (25) 36 8.33 6.6 1.9 4.36±1.05 −0.118±0.393
Punta Mugnona (33) 4 0 6.5 4 4.92±1.19 0.921±1.014
Grotta Preseppe (34) 3 0 6.5 5.5 5.66±0.76 0.935±1.225
Punta Sottile (35) 3 33 4.6 2.1 3.23±1.26 0.801±1.225
Punta Rosso di Nica (39) 1 100 1.9 1.9 1.9
El Haouaria (44) 16 100 2.6 1.2 2.08±0.40 −0.726±0.564
Kelibia (45) 16 68.75 4.6 0.9 2.57±0.89 0.239±0.564
Zembra (51–55) 850 27.17 8.8 0.3 4.29±2.26 −0.353±0.084*

*Skew is considered significant when its value is .2 x SE.

equal variances is not met. Differences between means were (Algeria; Espinosa, 2009) and Melilla (North Africa,
examined a posteriori using the Student–Newman–Keuls Spain; González-García et al., 2006). Univariate and
test. Multivariate MDS (non-metric multidimensional multivariate analyses were carried out using SPSS v. 15.0
scaling) statistics were also used, based on the UPGMA (SPSS, Chicago, USA) and PRIMER v. 6.0 (Clarke & Gorley,
method (unweighted pair-group method using arithmetic 2006).
averages), along with the Bray–Curtis similarity index.
MDS was used to test differences in population structure
among sites (i.e. number of specimens in each 1-cm size
Results
class from 0 to 10 cm). Clusters of sites identified as
statistically significant using the profile test SIMPROF In total 1,063 individuals of P. ferruginea, distributed non-
(P , 0.05) were considered to have a similar population homogeneously (Supplementary Table S1; Table 1; Fig. 2),
structure. Kruskal’s stress coefficient was used to test were recorded and measured. The species is extremely
the ordination (Kruskal & Wish, 1978). PERMANOVA scarce on mainland Tunisia (Fig. 2e,f; in the Cap Bon area,
(permutational multivariate analysis of variance) was used although there are other small populations from the
to test hypotheses regarding differences in population Gulf of Tunis to Monastir, see Discussion). A large and
structure between types of substrate (natural vs artificial) well-structured population was recorded in the Zembra
−1
and accessibility, assuming that at less accessible sites the archipelago (mean density 2.65 m ), with high levels of
limpets are more protected from collection (protected vs recruitment and large, reproductively mature specimens
unprotected). Analyses were performed using a log(x+1) (Fig. 2b). On the Egadi islands we found four individuals
transformation for the MDS analysis and fourth-root trans- (4.0, 4.0, 5.2, 6.5 cm in length) at Punta Mugnona, three at
formation for the PERMANOVA test, to lessen the Grotta Preseppe (5.0, 5.5, 6.5 cm; both sites on Marettimo
influence of the more abundant size classes. The abundance Island), and three at Punta Sottile (2.1, 3.0, 4.6 cm; Favignana
of each size class was standardized to density (individuals Island). Only one individual (1.9 cm) was recorded on
per m of transect) to avoid artefactual differences among Pantelleria, at Punta Rosso di Nica. On Corsica and Sardinia
populations resulting from the differing sampling effort at the populations were highly fragmented and densities
each site. Data for the MDS and PERMANOVA analyses were low, with only one individual (2.4 cm) found at
were compiled from the present study, augmented by Bonifacio and four at Tizzano (1.6, 2.2, 2.5, 5.6 cm), although
information from Granada and Alboran island (Spain; some sites, such as Asinara (Sardinia), had more individuals
Barba et al., 2006), Crinavis (Algeciras Bay, Spain; Navarro- (Fig. 2a). The species was absent from mainland Italy and
Barranco, 2010), San Felipe breakwater (Algeciras Bay, Sicily. There were significant differences in mean shell
Spain; Espinosa et al., 2005), Ceuta (North Africa, Spain; length among populations, with the largest individuals at
Espinosa et al., 2009a), Habibas and Plane islands Asinara (Table 2).

© 2013 Fauna & Flora International, Oryx, 48(2), 266–275

http://journals.cambridge.org Downloaded: 14 Mar 2014 IP address: 150.214.246.48