Page 2 - Renaud_Michaux_2007

S. Renaud and J. R. Michaux

sequences and variations in restriction fragment length
INTRODUCTION
polymorphism (Michaux et al., 1996b, 1998a, 2002a, 2003;
The interaction between contingency and determinism drives Libois et al., 2001). Therefore, patterns of morphological
morphological evolution, but the relative importance of both divergence can be interpreted within a well-defined phylo-
processes remains the subject of considerable discussion (e.g. genetic framework. Two main clades emerge, corresponding to
Losos et al., 1998). Insular evolution presents a unique a western and central European group and to one of Italo-
opportunity to address this question as chance effects gener- Balkanic affinity. Overall, Mediterranean insular populations
ating divergence are likely to be enhanced due to the process of of wood mice are closely related to mainland relatives, in
colonization and subsequent drift in a reduced population. On agreement with Holocene dates for the colonization by the
the other hand, directional evolutionary trends can be wood mouse of large Mediterranean islands (Vigne, 1999).
favoured on islands where isolation enhances the ability of a Sicilian populations are an exception in that they emerge as a
population to track local environmental optima (Whittaker, differentiated clade.
1998; Lenormand, 2002). Both contingency and determinism Morphological differentiation between populations was
have been invoked to explain the insular evolution of rodent estimated based on the size and shape of mandibles and first
populations (Berry, 1973, 1996). The wood mouse, Apodemus, upper molars using Fourier analysis of the outline. Results
is of interest in this framework because of its notable ability to were compared between characters and sorted by genetic
colonize islands. Insular populations often display a larger grouping, insularity and latitudinal location. Pattern coher-
body size than their closest mainland relatives, suggesting the ency and deviation identified key factors determining the
occurrence of an ‘insular syndrome’ (Angerbjo ¨rn, 1986; morphological differentiation. Size vs. shape and mandible
Michaux et al., 1996b, 2002b; Lomolino, 2005). On the other vs. molar emerged as characters varying according to
hand, some insular populations show apparently random different factors, and prone to either ecological or genetic
morphological differentiation (Renaud & Millien, 2001; influences.
Renaud & Michaux, 2003). In contrast, deterministic factors
seem to play an important role in structuring a latitudinal
MATERIALS AND METHODS
gradient in response to environmental conditions in mainland
populations (Renaud & Millien, 2001; Renaud & Michaux,
Materials
2003).
As latitudinal gradients may cause morphological differ- The study is based on a sample set of 239 wood mice (A.
entiation, with a response amplitude that is itself variable with sylvaticus): 216 mandibles and 239 first upper molars (M1)
latitude (Yom-Tov et al., 1999), focusing on a limited latitu- were measured (Table 1). Only mature specimens with the
dinal range may help decipher the key factors shaping insular third molar erupted were considered. The mice were either
divergence within a given climatic realm. The European wood trapped by J. R. Michaux and R. Libois, or were obtained from
mouse, Apodemus sylvaticus (Linnaeus, 1758), has been present the Museum National d’Histoire Naturelle (Paris, France).
in the Western Palearctic region for the past 3 Myr (Michaux All animals from the former collection were diagnosed
& Pasquier, 1974). From this wide geographical distribution, genetically as A. sylvaticus. These specimens were trapped in
we focus on only south-western European populations. Within 33 localities (Table 1) on the mainland and islands (Fig. 1).
this geographical range, size and shape should display a simple The mainland localities were selected as reference points from
gradient with latitude (Renaud & Michaux, 2003). The range each island. The islands were selected to sample variable
2
restriction might impede the detection of gross latitudinal geographical conditions and included small (< 20 km )to
2
variation but will enhance the detection of insular divergence. large islands (> 20,000 km ) and islands close to (distance
A previous study (Renaud & Michaux, 2003) was based on the < 10 km) or remote from (distance > 200 km) the mainland.
analysis of the mandible. This character varies with body size Corsica, Sardinia and Ibiza were isolated during the Pleisto-
(Renaud, 2005) and is involved in the feeding function. Hence, cene, whereas other islands were connected to the mainland
its variation with several ecological factors invoked in the insular during sea-level low-stands of the last glacial–interglacial cycles
syndrome, such as increased body size (Foster, 1964; Lomolino, (Van Andel, 1989, 1990).
1985; Michaux et al., 2002b) and diversification of the diet The majority of wood mice belong to the western European
(Orsini & Cheylan, 1988), is unsurprising. Other characters, genetic clade, with the exception of Corsica, Sardinia and Elba,
such as the teeth, are less likely to vary with body size at an which were occupied by wood mouse populations of the Italo-
intraspecific level but are also involved in the feeding function. Balkanic genetic clade. Sicily is characterized by wood mice
Such features are important for evolutionary studies of small belonging to a distinct genetic group related to the Italo-
mammals since analysis of the fossil record of small mammals Balkanic clade. Sicily is a large, elevated island, separated from
relies heavily on teeth. Generalization of conclusions drawn the mainland only by the Strait of Messina. The strong currents
from studies based on body or mandible size to fossil data in the Strait of Messina at present form an efficient barrier to
requires direct evidence that teeth co-vary with other characters. gene flow from the mainland, but the depth of the strait
Finally, the genetic structure of the European wood mouse is (c. 80 m) allowed connections between Sicily and the mainland
now well understood based on mitochondrial cytochrome b during the Late Pleistocene.

340 Journal of Biogeography 34, 339–355
ª 2006 The Authors. Journal compilation ª 2006 Blackwell Publishing Ltd