Page 5 - capula_eal2009

MORPHOLOGICAL VARIATION IN INSULAR LACERTID LIZARDS
allometry (sensu Klingenberg, 1996). The following
biometric measurements were registered: SVL = snout-
vent length; HL = head length, HH = head height, AL =
hind leg length, PL = forefoot length, FL = hind foot
length, TL = tail length (original tails only). Ranges of
variation and covariation of biometric characters were
explored through ANOVA and MANOVA. Static
allometry was investigated through regression analysis
and Partial Least Square analysis using tpsPLS (Rohlf,
2006) comparing centroid size, shape variables, and
biometric characters.
All analyses were performed at the Laboratorio di
Morfometria Geometrica of the Museo Civico di
Zoologia di Roma.
RESULTS
Interspecific variation
To depict the morphometric relationships among
Podarcis raffonei, P. sicula, and P. wagleriana, the
partial warp scores and the uniform components of all
samples were used as input data for relative warps and
canonical variate analysis. The results are shown in fig.
3. Ordination of individuals along the first and second
relative warps (accounting for 28.43 % of cumulative
variance) shows a clear separation of P. sicula from P.
wagleriana, while P. raffonei is characterized by a wide
range of variation and overlap with the other two
species. On the other hand, multivariate discrimination
between the three taxa is highly significant (Wilk’s
lambda = 0.1813; F = 81.59; P < 0.0001), and the species
are clearly distinct along both the first and the second
canonical variate axes (Fig. 4). Transformation grids
related to extremes of variation along the first canonical
axis (accounting for 96 % of total variance) indicate that
major differences regard the shape of the Parietal, the
Sopraocular, the Posterior supraciliar granule, and the
Anterior supraciliar granule scales (respectively P, SO,
PSG, ASG in fig. 2). The skull of Podarcis sicula, lying
at one extreme of variation, is characterized by a
narrower rostral and nasal region with respect to the
central portion, while P. wagleriana shows the opposite
pattern, i.e. the rostral and nasal regions are wider than
the central part. In this context, the skull of P. raffonei is provided further information concerning the extent and
intermediate between P. sicula and P. wagleriana. The
shape of the Sopraocular and Parietal scales appears to
be useful for the taxonomic analysis. More specifically,
Sopraocular scales are much larger and wider in
Podarcis sicula than in P. wagleriana, while in P.
raffonei the shape of these scales is intermediate between
P. sicula and P. wagleriana. This is clearly recognizable
in the pictures shown in fig. 5. The observed
modifications in the shape of the Sopraocular scale are
221
related to the modifications of the shape of the Parietal
scale, which appears to be diagnostic too between P.
sicula and P. wagleriana.
MANOVA of centroid size for the three species
indicates a significant size difference (F = 5.45; P =
0.0053). The skull of Podarcis sicula (mean CSPs =
5886) is larger than that of P. wagleriana (mean CSPw =
5528), and smaller than that of P. raffonei (mean CSPr =
6094).
In tab. 2 the means and standard errors of the seven
biometric characters analysed are reported. Podarcis
wagleriana shows the smallest body length (Mean SVL =
61.81 ± 0.90), while P. raffonei is the largest species
(Mean SVL = 71.71 ± 2.00). The snout-vent length of
Podarcis sicula is intermediate between those of P.
raffonei and P. wagleriana (Mean SVL = 66.57 ± 0.88).
No relationship was found between the shape and the
centroid size of the skull, as indicated by the regression
coefficient of the PLS analysis (r = 0.44087; percentage
of correlation = 12 %; percent of squared singular value =
93 %). Differences in the size of the skull (HL) are not
related to differences in snout-vent length (SVL) (r =
0.066; P = 0.23, fig. 6), suggesting a different allometric
static pattern among the three species (sensu Klingenberg,
1996). Podarcis wagleriana is characterized both by a
small head and body length, while P. sicula has a head
larger than P. wagleriana and a body length which is
intermediate between P. wagleriana and P. raffonei.
Podarcis raffonei shows the largest body length, but its
head size is intermediate between P. sicula and P.
wagleriana. The allometric patterns within the three
species were investigated through the analysis of
variation of snout-vent length (SVL) vs head length (HL).
As it is evident from fig. 7, P. raffonei shows the most
diverging pattern, differing in both intercept and slope
coefficients (intercept = 6.18; slope = 0.14; r = 0.91; F =
126.53, P < 0.0001), while P. wagleriana and P. sicula
show similar patterns (intercept = 1.30 and -0.23,
respectively; slope = 0.22 and 0.26, respectively).
Intraspecific variation
The analysis of relative warp scores for each sample
distribution of morphometric variation among
populations of each species. Box plots of the first (RW1)
and the second (RW2) relative warps scores (accounting
for 15.99 % and 12,43 % of total variance respectively)
for individual populations are shown in fig. 8. Podarcis
sicula is characterized by reduced morphometric
variation of the skull, and no morphometric gap is
evident between the samples from the small islands
(Favignana Island, Lipari Island) and those from Sicily.
The situation is different in P. wagleriana, where the