When effects of body mass and phylogeny were statis-
tically controlled, scores on the first PC axis were not sig-
nificantly associated with other measures of the fast-slow
continuum. Scores on the second PC (primarily reflecting fe-
cundity, table 8.2) were significantly associated with longer
subadult periods relative to the life span (viz., the a/vra-
tio). Fecundity, however, was only weakly associated with
the a/vratio (r0.275, n43, P0.07). Significant as-
sociations with the third PC (primarily reflecting variation
in PA, table 8.2) with PAand PA/awere likely due to the
confounded influence of PA.
Discussion
Explaining the evolution of life histories is an important
goal for ecologists and evolutionary biologists. Interspecific
studies attempt to evaluate the way in which life histories
evolve by comparing traits of different species to aspects of
the environments in which species live. Two major prob-
lems have faced such research. The first is that critical as-
pects of environmental influences are often difficult to iden-
tify and quantify. A possible solution to this problem is
to use demographic traits as indicators of environmental
changes, and then look for associated changes in life his-
tories. This tack was taken by Murphy (1968) and later
Stearns (1976) to introduce what came to be known as the
bet-hedging model of life-history evolution. Harvey and his
colleagues (reviewed by Harvey and Purvis 1999) also used
this line of thought to argue that life-history patterns should
be evolutionarily molded by mortality patterns. The pri-
mary problem with this research tactic is that demographic
traits themselves are also life-history traits, so it is difficult
to discern the interplay of cause and effect.
This first major problem leads us to a second problem:
difficulties with drawing inferences from interspecific stud-
ies. The problem is whether inferences about adaptation can
be made for species that have historical affiliations. Intra-
specific studies should yield strong inferences about evolved
responses of life histories to environmental conditions, but
the range of variation in life histories may be narrow. Inter-
specific studies examine broader ranges of variation in both
life histories and environments. Differences among spe-
cies may be historical, however, and thus reflect evolved re-
sponses to past environments (Dobson 1985). Reconstruct-
ing past environments and discerning how ancestors
responded to them is exceedingly difficult. Alternatively,
differences among species may reflect current adaptation to
modern environments. It is these evolutionarily maintained
variations in life histories that reveal the clearest answers to
our question about how life histories evolve. A possible so-
lution to analyzing interspecific patterns of life histories is
to attempt to remove influences of historical patterns by sta-
tistically adjusting for phylogeny (e.g., Miles and Dunham
1992, Martins and Hansen 1997). Such procedures also
provide at least partial solutions to sampling problems as-
sociated with interspecific studies (Harvey and Pagel 1991).
Considerable interspecific research on life histories has
focused on mammals. Studies of the relationship of life his-
tories to body size used comparisons of mammalian spe-
cies, and concluded that much of the variation in life histo-
ries could be explained by differences among species in size
(reviewed by Harvey and Purvis 1999). Read and Harvey
(1989) countered this argument by pointing out that if the
influence of body mass was statistically removed from mam-
malian life-history traits, very similar associations among
the traits remained. These associations seemed to describe a
continuum of life histories, from fast traits such as high rate
of reproduction, early maturity, and low survival, to slow
traits such as late maturity, high survival, and low rates of
repeated reproduction. This fast-slow continuum was evi-
dent not only among mammals, but within mammalian or-
ders as well. Promislow and Harvey (1991) suggested that
fast life histories evolved due to environmentally caused
mortality rates that were high, and slow life histories evolved
when mortality rates were low.
The purpose of our present interspecific study was to re-
examine the fast-slow continuum in the order Rodentia. We
improved on past analyses in two ways. First, we looked at
a restricted set of life-history traits that summarize the life
cycles of species. For this, we borrowed from the concept
of partial life-cycle models (Caswell 2001; Oli and Zinner
2001). A partial life-cycle model can be parameterized from
demographic traits that are sufficient to describe changes in
population size or growth rate: age at maturity, age at last
reproduction, juvenile and adult survival, and fecundity.
Since these traits are sufficient to describe population dy-
namics, they should be key life-history characteristics. In ad-
dition, with the exception of age at last reproduction, these
life-history variables have generally been used to measure
the evolutionary fitness of other characteristics of organ-
isms (Endler 1986).
Our second improvement was to borrow the idea of us-
ing principal component analysis to evaluate life-history
patterns, after Stearns (1983) and Gaillard et al. (1989).
Gaillard et al. (1989) used fewer traits to describe life his-
tories, but identified the fast-slow continuum as statisti-
cally independent of body size. Principal component anal-
ysis identifies the major axis of variation in variables in a
hyperspace of several axes. We can think of variables in
hyperspace as similar to a rugby ball in three-dimensional
space. The first principal component would run along a line
drawn through the long axis of the rugby ball. If the rugby
ball were “skinny” around the middle, the first compo-Fast and Slow Life Histories of Rodents 103