nent (first PC axis) would describe most of the variation in
the data set, as in our data on life-history traits. Thus the
first principal component summarizes traits that change to-
gether, and such traits will have high and even factor load-
ings on the first principal component. Factor loadings tell us
which variables change the most along the principal com-
ponent. The analysis also shows us whether the magnitudes
of traits vary together in a positive or inverse manner, re-
vealed by the sign of the factor loadings (viz., either the same
or opposite signs). Gaillard et al. (1989) also suggested that
a secondary life-history tactic to the continuum was evident
in their analyses, which reflected an altricial-precocial di-
chotomy among species. This hypothesis can be tested by
looking at PC axes after the first one, if the first axis reflects
the fast-slow continuum.
Like previous studies (e.g., Western 1979; Western and
Ssemakula 1982; Peters 1983; Schmidt-Nielson 1983; Cal-
der 1984), we found that adult body mass was significantly
associated with life-history traits (table 8.1). When we sta-
tistically adjusted the life-history data for family-level phy-
logenetic effects, however, body mass had little additional
influence. Body mass varies among families of rodents, and
thus influences of size and phylogeny cannot be easily sep-
arated. Thus removal of influences of body mass and phy-
logeny from the life-history variables was done together.
Body mass and phylogeny accounted for large amounts (up
to 90%) of variation in life-history variables. In particular,
removal of these sources of variation left little residual vari-
ation in age at maturity, adult survival, and fecundity. This
is important to keep in mind when considering residual pat-
terns of life histories, and we expected that PCA might re-
flect the fast-slow continuum more poorly when influences
of body size and family-level phylogeny were removed.
Loadings of life-history variables (unadjusted for body
size and phylogeny) on the first principal component clearly
reflected the fast-slow continuum: timing variables (aand
v) and survival were all strongly associated on the first prin-
cipal component (table 8.2). In addition, reproduction var-
ied in a negative way along this axis. These results sup-
port the idea that a fast-slow continuum exists in rodents,
from species with short lives and greater reproductive in-
vestments during a relatively short period of time to species
with long lives and limited but repeated investments in re-
production. A tradeoff between reproductive and somatic
investments may be reflected along this axis. Our results
also supported the suggestion that the fast-slow contin-
uum may be evolutionarily influenced by mortality patterns
(Read and Harvey 1989). High mortality was associated
with the fast end of the continuum.
We adjusted our data statistically to control for influ-
ences of body mass and family-level phylogenetic effects. In
doing this, most of the variation among populations in life-
history traits was removed (see the preceding). Despite this,
loadings on the first PC still indicated a fast-slow contin-
uum, with high positive loadings for timing variables and
survival, though with a weaker pattern of changes in adult
survival (table 8.2). The greater strength of the loading of
juvenile compared to adult survival on the first PC was in
agreement with the scenario of life-history evolution pre-
sented by Promislow and Harvey (1990, 1991), that life his-
tories evolve in response to mortality patterns, particularly
among juveniles. Also in this analysis, reproduction exhib-
ited variation that was statistically independent of the con-
tinuum, as reflected by the high positive loading of fecun-
dity on the second PC axis. Thus reproduction did not vary
in quite the same way as when the life-history traits were
not adjusted for body size and phylogeny.
The second PC axis of the unadjusted life-history data
accounted for such a low amount of variation in life-history
traits that it may not be useful to evaluate this axis further.
When adult body mass was included in the analysis, mass
exhibited variations that were different from its association
with the fast-slow continuum (table 8.2). In this analysis,
the third PC axis exhibited a virtually identical pattern to
the second PC when body mass was not included, but the
third axis accounted for even less of the variation in life-
history traits. When adjustments were made to the data for
body mass and family-level phylogeny, substantial residual
variations in life-history traits was explained by the second
and third PC axes. Each axis primarily reflected changes
in single life-history variables, fecundity and adult survival
respectively, that were statistically independent of the fast-
slow continuum. In the former case, reproduction appeared
to vary independently of the fast-slow continuum. Little
variation in life-history traits remained after statistical ad-
justment of the data for body size and phylogeny (table 8.1),
however, so interpretation of these axes might be risky.
Oli and Dobson (2003, 2005) suggested that the ratio of
fertility to age at maturity (F/a) might be a good index of
the fast-slow continuum. This expectation was borne out
by strong associations of F/aand m/awith the first princi-
pal component of the unadjusted data set (table 8.3). Sur-
prisingly, PA/awas very closely associated with the fast-
slow continuum, as the latter was reflected by scores on the
first PC. This ratio reflects mortality patterns relative to
the length of prereproductive development. When the life-
history variables were statistically adjusted for body mass
and family-level phylogeny, however, all of these signifi-
cant associations became nonsignificant. This result may
have been due to removal of so much of the variation in life-
history traits when statistical adjustments were made.
The mass that offspring attain at birth and at the end104 Chapter Eight