Page 9 - THE_ISLAND_RULE_2012

THE ISLAND RULE IN LARGE MAMMALS 1739

later cohabited with new species evolving in situ (Vos 1996, direction of this variation agrees with the influence of bio-
2000). Depending on whether one recognizes one or two logical interaction—mainly competition in the case of un-
colonization events, either four or three deer species origi- gulates and resource base in the case of carnivores—and con-
nated on this island (Capasso Barbato 1988, 1990; Vos 1996, tradicts the notion of an optimal size. Instead, smaller sizes
2000). The three largest, and stratigraphically younger (Cer- may be selected for in insular ungulates because resources
vus rethymnensis, C. dorothoensis, and C. major), were as that are no longer needed for antipredator behavior and in-
large as or even larger (much larger in the case of C. major; terspecific competition can be safely reallocated to repro-
online Appendix 1) than mainland red deer. The size of youn- duction (Brown 1995; Raia et al. 2003; Lomolino 2005),
ger Crete deer contradicts the common rule for large mam- resulting in lower adult body mass (Roff 2000; Stearns et al.
mals (i.e., dwarfism, Lomolino 2005) but is expected ac- 2000; Charnov 2001). However, once the smallest size clas-
cording to our hypothesis. The overdispersion of Crete deer ses were occupied, new species evolved larger body size.
sizes (which is robust irrespective of the taxonomy one ap- This should be not viewed as maladaptive. We argue that
plies) further indicates the importance of competition among species arriving later simply faced different conditions, for
them. The regression of M 3 lengths in island-mainland pairs which the size they attained was probably adaptive (Case
had a slope not statistically different from one, indicating 1978). Indeed, larger size allows the inclusion of new (lower
that the degree of dwarfism is independent of ancestor size. quality but often superabundant) resources, at least in her-
This is striking given that our dataset is disproportionately bivores (Demment and Van Soest 1985; Illius and Gordon
rich in islands where elephants were the only large herbivores 1992; Clauss and Hummel 2005). Competition with smaller
(and therefore reducing the most in size relative to their main- species is thus reduced. The only clear disadvantage of larger-
land ancestors, according to our expectations). sized ungulates seems to be an extended growth period. In
Our results suggest that carnivore sizes are influenced by carnivores, body size seems mainly related to the ability to
resources and little else. However, whether it is resource acquire resources. Superabundant, energetically rich foods
biomass available or prey size or both that are important is (e.g., salmon runs), will promote gigantism (Case and Schwa-
unclear. Carnivore size is affected neither by ancestral body ner 1993; Goltsman et al. 2005), but prey size is also a major
size per se (Meiri et al. 2004, 2006; this study); nor by island factor: in the absence of large prey, carnivores can dwarf
area, isolation, diet, and phylogenetic affinities (Goltsman et (e.g., island foxes, Urocyon littoralis dwarf in the absence of
al. 2005; Meiri et al. 2004, 2005a, 2006; this study); nor by rabbits, see also Jessop et al. 2006), but when large prey is
most within-guild interspecific interactions (this study). Al- available carnivores remain large.
though modal prey size may differ between males and females In summation, we believe size evolution on islands is not
of the same species (e.g., Loy et al. 2004; Radloff and Du directly dependent on abiotic attributes such as area and iso-
Toit 2004), the effect of the nature of the resource base is lation. Such factors can have an indirect effect on animal
shared among the sexes, and SSD does not increase on is- body sizes because they influence the numbers and identities
lands. Unfortunately, we cannot test for SSD in fossil un- of species that occur on islands. However, we emphasize that
gulates (except for a few deer populations) because of the species are real interacting entities, not merely numbers to
difficulties in sexing fossils. Thus, the claims of large SSD
be added. Size of insular mammals probably depends on the
(and paedomorphosis) in island fossil ungulates remain an-
peculiar biotic characteristics of a given island in a given
ecdotal (e.g., Ambrosetti 1968; Malatesta 1980; Capasso Bar-
time, and on the biological attributes of resident species.
bato 1988; but see Palombo 2001; Raia et al. 2003). Nev-
These attributes may be weakly correlated with absolute body
ertheless, we predict that increased SSD is likely to occur in
size, but a better understanding of the autecological char-
ungulates because they are not constrained by the need to
acteristics of a given insular population is the key to pre-
remain large in order to subdue prey. We could not quantify
dicting the direction of magnitude of size evolution on islands
resource availability for fossil ungulates; yet, the highly sig-
and elsewhere.
nificant effects of competition and predation indicate that
resource limitations could not have been very important. Pri-
ACKNOWLEDGMENTS
mary productivity seems to have little influence on energy
flow in island ecosystems (Schoener 1989). We thank R. Asher (Museum fur Naturkunde, Humboldt
Another class of explanation of size evolution rests on Universita¨t, Berlin), D. Balkwill and M. Gosselin, (Canadian
optimal body size hypothesis (Brown et al. 1993; Damuth Museum of Nature), Y. Chang Man (Raffles Museum), J.
1993; Marquet and Taper 1998; cf. Blackburn and Gaston Chupasko (MCZ, Harvard), Elisabetta Cioppi (Museo di Pa-
1996; Kozlowski and Teriokhin 1999; Roy et al. 2000; Meiri leontologia, Universita` di Firenze), Andy Currant and J.J.
et al. 2004, 2005b). Lomolino (2005) distinguished different Hooker (Natural History Museum, London), J. Dawson
optima for different taxa (or underlying bauplans). Our data (New-Walk Museum), M. del Re (Museo di Paleontologia,
suggest that body size on islands evolves according to pre- Universita` di Napoli), C. Di Patti (Museo Rocco Gemellaro,
vailing ecological conditions. In Pleistocene Mediterranean Universita` di Palermo), J. Eger and S. Woodward (Royal
islands, there were several cases of a mainland species re- Ontario Museum), H. Endo (National Science Museum, To-
peatedly colonizing islands. SR varied from 0.23 to 0.82 in kyo), B. Engesser (Naturhistorischen Museum, Basel), G.
E. antiquus, from 0.6 to 1.35 for the C. elaphus, from 0.47 Fornasiero (Museo di Paleontologia, Universita` di Padova),
to 0.91 for Praemegaceros verticornis, and so on. Similarly, D. Harrison and M. Perch (Harrison Zoological Institute), D.
SR often varies in carnivores on different islands (e.g., in Hills (Natural History Museum, London), T. Holmes (Uni-
Paradoxurus hermaphroditus it ranges from 1.1 to 0.88). The versity of Kansas Museum of Natural History), G. Jarrell