Page 7 - MICHAUX2002

P. 7

Body size increase in insular rodent populations 433

Body size (g in log) 1.60 (a) variables. These variables were the contrasts in area, the con-
trasts in total number of predators and the contrasts in total
1.55 number of competitors. Using the whole dataset, we found
1.50 that only the total number of predators was correlated with
1.45 body size (P = 0.04). Removing the outlier, we found that both
1.40 area (partial P = 0.027) and total number of predators (partial
1.35 P = 0.010) were correlated with body size (P = 0.033).
1.30
1.25 DISCUSSION
1.20
Our comparative analyses using the independent contrasts
0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 method show clearly that the increase in body size of
A. sylvaticus occurred on small islands with a low number of
Island area (km2 in log) predator species. In general, the explanation for the relation-
ship between increase in body size of small species and island
Residuals of contrasts in body size 0.02 (b) n = 11 size is not clear. Melton (1982) and Adler & Levins (1994)
(controlled for island area) r = –0.76 proposed that on small islands, small-bodied species should
0.01 P = 0.0047 have higher population densities than on continents or larger
0.00 islands. Adler & Levins (1994) suggested that this favours
a K-strategy, leading to an increase in life span and to a
–0.01 decrease in the number of offspring. This reduces the overall
reproductive effort and results in re-allocation of energy to
–0.02 growth and maintenance and then large body size. However,
in the case of woodmouse populations on Mediterranean
–0.03 islands, their densities are often lower than those of continental
populations (Sarà & Casamento, 1993; J.R. Michaux,
–0.04 0.08 0.12 0.17 unpublished data). This results probably from competition
–0.08 –0.03 0.02 with the black rat (Rattus rattus) and the house mouse (Mus
musculus), wild populations of both of which are present on
Residuals of contrasts in sum of predator species many Mediterranean islands at high densities (Granjon &
(controlled for island area) Cheylan, 1988, 1990).

Residuals in body size 0.10 (c) n = 12 A role for predators
(controlled for island area) r = –0.54
0.08 P = 0.070 Our analyses indicate that the number of predator species
plays a more important role than the number of competitor
0.06 species. Indeed, Lawlor (1982) and Lomolino (1985), who
associated the appearance of insular gigantism with the
0.04 number of competitor species, based their hypothesis on the
fact that, generally, the number of competitor species is lower
0.02 on small islands than on continents (see also Alcover &
0.00 Gosalbez, 1988). This leads to ‘empty’ niche space, which
will be more easily ﬁlled by individuals having a larger size
–0.02 0 0.1 0.2 0.3 0.4 (Lister, 1976). However, in our case the number of competitor
–0.04 species observed on small islands with large-bodied Apodemus
–0.06 is of the same order as those observed on large islands
(Corsica, Sicily, Sardinia) where no gigantism is observed
–0.4 –0.3 –0.2 –0.1 (Contoli et al., 1988a,b; Libois & Fons, 1990). Moreover, the
densities of the competitor species (principally R. rattus and
Residuals in sum of competitor species M. musculus) currently living on the small Mediterranean
(controlled for island area) islands are drastically high (Granjon & Cheylan, 1988,
1990), which would seem to rule out easy access to the ‘free’
Fig. 2 (a) Relationship between island area and average body size of ecological niche space for the woodmouse.
Apodemus sylvaticus. Two regressions were obtained. The signiﬁcant
one (dashed line) was obtained when the outlier Port-Cros (left
bottom point) was removed from the analysis. (b) Relationship
between the residuals of body size of A. sylvaticus (controlled for
island area) and the residuals of numbers of predator species
(controlled for island area) using independent contrasts. (c)
Relationship between the residuals of body size of A. sylvaticus
(controlled for island area) and the residuals of numbers of
competitor species (controlled for island area) using raw data.

2 3 4 5 6 7 8 9 10