Rodent Societies: An Ecological & Evolutionary Perspective

and 60% of this correlation is phylogenetically constrained;
(3) Litter size and increases in social complexity are nega-
tively correlated, and there is no correlation between ges-
tation time and level of social complexity. Given appar-
ent problems with existing taxonomic treatments of some
groups of sciurids, recent molecular studies on ground-
dwelling sciurids provide the necessary components for the
derivation of a well-resolved phylogenetic framework that
can be used in future studies of social complexity and the
evolution of life history traits. As an example of how this
phylogeny can be used, we recoded the numerical values
shown in the appendix of Blumstein and Armitage (1998a),
and used MacClade 4.05 (Maddison and Maddison 1992)
to map character states on the consensus phylogeny com-
piled from existing molecular data (fig. 2.8; Steppan et al.
1999; Harrison et al. 2003; Herron et al. 2004). The con-
version from continuous to discrete characters required
partitioning values into non-overlapping subsets (e.g., high
complexity 0.8 to1, medium complexity0.8 to 0.4, low
complexity 0.2 to 0.4). Despite the problems with such
character coding, some of the results can be compared to
patterns observed by Blumstein and Armitage (1998a). For
instance, the ancestral litter sizes are large (gray), whereas
lineages with high social complexity show a decrease in lit-
ter size (black), which seems to be an ancestral state in all

species of Marmota,including the less socially complex

Marmota monax(fig. 2.8a). In addition, the ancestral state

for ground-dwelling squirrels appears to be low social com-

plexity, with socially complex groups arising independently

at least twice (fig. 2.8b). The clade containing the highly

socially complex Cynomysreveals an early trend toward

increasing complexity in the ancestor.The ancestral state

for percent females breeding is low and a decrease ap-

pears ancestral to the clade containing the socially complex

Marmota flaviventris, M. olympus,and M. caligata,but

is possibly independently derived in the clade containing

M. caudataand M. marmota. A similar pattern is seen for

the Cynomysclade, characterized by an intermediate per-

centage of females breeding, except for Cynomys leucurus,

which has secondarily derived a high percentage of females

breeding.

Summary

As can be seen in the preceding sections, the order Roden-

tia represents a major branch of the mammalian tree of life,

and unlike many branches, they have experienced little ex-

tinction, with many of the extant families dating almost to

the beginning of the radiation. Rodents are excellent colo-

Rodent Evolution, Phylogenetics, and Biogeography 21

Figure 2.7 Phylogeny of the caviomorph superfamily Cavioidea,derived from a combined maximum parsimony analysis

of one mitochondrial (12S rRNA) and two nuclear sequences (exon 10 of growth hormone receptor and intron 1 of pre-

albumin transthyretin). Bootstrap values (100 replicates), followed by Bremer decay indices, are given above the nodes for

the MP analysis. A positive sign () indicates Bremer support of greater than 20. Bootstrap values (100 replicates) below

the nodes correspond to the ML analysis. Based on the results of Rowe and Honeycutt (2002).