Page 4 - vol64-2011-215-222deidun

a) b)
Figure 2. a) The landmarks for pronotum (N = 4: 1-3-9-10) and the semi-landmarks for pronotum (N = 6) and
b) adopted for the P. bimaculata individuals studied. The landmarks for elytra (N = 4: 1-2-3-16) and the semi-
landmarks (N = 12) were digitized on half of each structure to remove the variability introduced by an eventual
asymmetry.
collected and digitized by means of the tpsDIG 2 soft-
ware (Rohlf, 2004). In order to better perform statistical possible reason for such a discrepancy might be the
analysis, the landmarks and semi-landmarks were suc-
cessively recognized with tpsUtil 1.45 (Rohlf, 2008).
The resulting coordinates were subjected to a General-
ized Procrustes Analysis (GPA), which removes all the
information unrelated to shape (Rohlf and Slice, 1990).
For the comparison of the configurations of landmarks
and semi-landmarks, the Relative Warps Method was
used (Bookstein, 1991; Rohlf, 1993). The relative warps
are principal component vectors of the partial warps,
variables generated for thin-plate spline transformations
(Bookstein, 1989), and these were used to describe the
major trends in shape variation among specimens within
the sample (Rohlf, 1993; 1996). Thin-plate spline de-
formation grids were generated to facilitate description
of shape variation. The analyses were performed by
means of Relative Warps 1.39 software (Rohlf, 2004).
In order to single out a putative relationship between
site and shape, a clustering of the studied specimens was
performed on the basis of their geographical position,
through the Canonical Variate Analysis (CVA). The
CVA of the morphometric data was performed using
MorphoJ 1.01 software (Klingenberg, 2011) and used to
simplify the identification of differences between prede-
fined group means (Zelditch et al., 2004).
Results and discussion
P. bimaculata individuals were recorded at 20 different
locations, of which 2 were located in Sicily, 9 in
circum-Sicilian islands and 9 in the Maltese Islands. In
addition, three P. acuminata populations were also re-
corded along southern Sicilian shores (table 1). The two
Phaleria species were never recorded in sympatry
within the current study. Such an observation was unex-
pected for the Maltese beaches of Golden Bay and
Mgiebah where a previous sympatry between the two
218
Phaleria species was reported (Deidun et al., 2010). A
snapshot nature of the sampling conducted in the pre-
sent study which might have resulted in P. acuminata
not being represented in collections, although there are
no records of significant phenological differences be-
tween the two species.
Major body shape differences seem to be probably re-
lated to geometric differences in elytra, as evident from
deformation grids (figure 3): the elytra of the individu-
als from Vulcano (points 31-45 in the plot) were more
elongated and elliptical in comparison to the Maltese
individuals (1-15) and the Sicilian individuals (16-30),
with the latter being positioned in the central area of the
plot.
The CVA showed a cumulative variance in the first
two axis of 100%, for both pronotum and elytra analy-
sis. In the CVA scatterplot, the three groups, Vulcano,
Malta, Sicily, can be clearly distinguished (figure 4).
The Mahalanobis distances (table 2) showed, in all
pairwise comparisons, strong significant differences (p
< 0.001) in body shape, highlighting the differentiation
of elytra-shape between the site Vulcano and the rest of
samples.
The observed inter-population shape differentiation
clustered the populations into three groups, according to
a spatial structure pattern, which could be described as
“site-specific”. A model of isolation by distance cannot
Table 2. Mahalanobis distances among groups. Below
elytra, above pronotum analysis.
Malta Malta Sicily Vulcano
Sicily 3.1369*** 3.0108***
Vulcano 3.9024*** 3.1567***
6.8089*** 5.8927***
P-values from permutation tests (10000 permutation
rounds); ***P < 0.001.