Page 2 - Fiorentino_alii_2008
P. 2

810   V. FIORENTINO ET AL.

             (Marmorana)],  the   central-southern  Apennines  sample areas was planned considering the patchy
             [assigned to Marmorana (Ambigua)] and Sicily      distribution of Marmorana in the field. This patchi-
             [assigned to Marmorana (Murella)]. The Sicilian   ness is determined by the rock-dwelling life style of
             Marmorana have impressive conchological diversity,  this group which is confined to limestone rock faces.
             with globular to flat-shelled populations, which has  For shell measurement, ten specimens were col-
             led to the description of approximately sixty nominal  lected randomly in each site, making 20 specimens
             taxa, now reduced to five species (Manganelli et al.,  per area and a total of 570 specimens. Only shells
             1995) or five species with 59 subspecies (Bank, 2007):  with a reflected lip were used because this indicates
             Marmorana globularis (Philippi, 1836), Marmorana  cessation of growth and maturity of the snail. Several
             nebrodensis (Pirajno, 1840), Marmorana scabriuscula  shell features (Fig. 1A) were measured to the nearest
             (Deshayes, 1832), Marmorana platychela (Menke,    0.01 mm using digital images (Adobe Photoshop,
             1830) and Marmorana muralis (Müller, 1774). Unfor-  version 7.0.1) of standard views of shell with a Nikon
             tunately, due to punctiform distribution of shell  Coolpix 4500 digital camera (Fiorentino et al., 2008).
             morphs, loss of typical material, vaguely indicated  For each site, five sexually developed specimens
             type locality (frequently only ‘Sicily’) and no consis-  (genitalia at similar stages of development), randomly
             tent or controversial application of names to different  selected among those analysed for shell features, were
             morphs, it is impossible to unequivocally assign popu-  dissected under a light microscope (Wild M5A) using
             lations to a particular taxon and, with the present  fine-pointed watchmaker’s tweezers. Twelve linear
             state of the art, nomenclatural and taxonomic revi-  variables (Fig. 1B) of isolated genitalia were measured
             sion is difficult.                                using a millimetric lens under a light microscope
               As Arbogast & Kenagy (2001) note, ‘the goal of  (0.01 mm) (Fiorentino et al., 2008).
             phylogeography (Avise et al., 1987) is to characterize
             the phylogenetic deployment of genealogical lineages
             across the geographical landscape’. Phylogeographic      ANALYSIS OF MORPHOLOGICAL DATA
             analyses have often shown cryptic divergent evolu-  Variables were log-transformed to obtain linear rela-
             tionary lineages that are not reproduced in the   tionships. Discriminant function analysis (DFA) was
             current taxonomy, as well as the existence of nominal  performed considering all the shell variables mea-
             species found to be poly- or paraphyletic.        sured (Jennrich, 1977; Chiu et al., 2002; Alonso et al.,
               The present study aimed: (1) to determine the   2006b) to discover which measurements contributed
             phylogenetic relationships of Sicilian Marmorana spe-  to discriminate groups defined a priori, in this case
             cies; (2) to test for shared patterns among putative  the populations sampled. A sequential chi-square test
             Marmorana species by morphological (shell and     was used to quantify the significance with which the
             anatomy) and molecular analysis [mitochondrial    discriminant functions significantly separated the
             (mt)DNA genes]; (3) to investigate the evolution of  groups. Structure and canonical coefficient tables
             shell shape (globular versus keeled flat shell); and (4)  were used to establish the contribution of each mea-
             to investigate the taxonomic implications of the three  surement to the first two discriminant functions.
             character sets analysed (conchological, anatomical  Two-way analysis of variance (ANOVA) was used to
             and molecular).                                   distinguish populations on the basis of the variables
               Accordingly, we obtained a molecular phylogeny of  contributing most to the DFA functions. The a poste-
             Sicilian populations of the genus Marmorana by    riori Tukey test (P < 0.05) was used to check group
             means of sequences of two commonly used mtDNA     significance.
             genes, cytochrome oxidase subunit I (CO I) and the  DFA was then performed considering all genital
             large ribosomal subunit (16S rRNA).               variables measured. The analysis was run with the
                                                               groups identified by ANOVA, in an endeavour to
                                                               check for differences in genitalia among recognized
                                                               shell groups. With this analysis we assessed which
                      MATERIAL AND METHODS
                                                               measurements   contributed  to  discrimination  of
                  SAMPLING FOR MORPHOLOGICAL ANALYSIS          groups defined a priori, in this case the populations
             A two-factor sampling design was used for morpho-  sampled. A sequential chi-square test was used to
                                            2
             logical analysis: area factor (6 km ) and site factor  quantify the extent to which each discriminant func-
             (100 m ) (for sampling design, see Fiorentino, Manga-  tion significantly separated groups and structure and
                   2
             nelli & Giusti, 2008). The site was a two-level random  canonical coefficient tables were used to establish the
             nested factor: two sites were randomly selected in  contribution of each measurement to the first two
             each of the 30 sampling areas (in five areas, two-site  discriminant functions.
             sampling was not possible for logistic reasons) (see  Size is generally the first determinant of biometric
             Appendix). The sampling design and number of      variation (Cadima & Jolliffe, 1996). Techniques that

                                © 2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 94, 809–823


  Downloaded from https://academic.oup.com/biolinnean/article-abstract/94/4/809/2701256
  by guest
  on 18 November 2017
   1   2   3   4   5   6   7