et al. 2001 for a review). Little is known about such pro-
cesses in more visually oriented, diurnal rodents, but ample
opportunities exist for extension of such learning to the
visual domain once interactions outside the nest begin.
Development of a preference for conspecifics provides a
schema against which other organisms in a rodent’s envi-
ronment might be compared. Those predators whose form
contrasts most with this schema, i.e., snakes and rap-
tors, should be least likely to be misidentified as a con-
specific. In fact, laboratory-born, snake-inexperienced Cal-
ifornia ground squirrels respond differently to snakes and
heterospecific mammals (Owings and Coss 1977; Coss
1991a; Coss 1993; Coss and Biardi 1997). This difference
in response to novel mammalian and reptilian species may
be based in part on differences in their similarity to the
learned ground-squirrel schema.
Conversely, associative learning typically involves gen-
eralization to similar configurations — for example, to the
similar forms of other mammals, such as other rodents and
cats. Consistent with this prediction, laboratory-reared Cal-
ifornia ground squirrel pups and juveniles responded with
much greater caution to a gopher snake than to either a
guinea pig (Cavia porcellus) or a domestic cat (Felis catus;
Coss 1991a; Coss and Biardi 1997). In contrast, an experi-
mentally presented dog evoked more cautious responses by
juveniles than by adults a few weeks postemergence, sug-
gesting that rapid learning of the danger posed by mam-
malian predators counters this generalization effect (Han-
son and Coss 1997). The juveniles in this study were more
aroused than the adults (Hanson and Coss 2001a), a con-
dition known to facilitate associative learning (McGaugh
1989). Such learning may have been mediated by the evoca-
tiveness of this species’ alarm calls, which serve as uncondi-
tioned stimuli, evoking antipredator behavior in lab-born,
predator-inexperienced members of this species (Tromborg
1999). So encounters with mammalian predators may in-
volve pairing conspecific vocal unconditioned stimuli for
antipredator behavior with conditioned stimuli, such as the
sight of novel mammalian predators (see also Shriner 1999;
Mateo, chap. 17 this volume).
Ontogenetic adjustment of the stress-response system
The hypothalamic-pituitary-adrenal (HPA) stress-response
system that supports social and antipredator behavior var-
ies adaptively in ways that are shaped through both proxi-
mate and ultimate processes. This is an ancient system that
has undergone evolutionary diversification in its structure
and function (Stoddart and Bradley 1991; Eilam et al. 1999;
DeVries 2002; Romero 2002). And this system varies intra-
specifically in the details of its functional organization, de-
pending on the environmental conditions prevailing during
each individual’s development, as influenced by such fac-
tors as the levels of food availability, predation intensity, or
social conflict (Francis and Meaney 1999). However, many
such developmentally important environmental conditions
exist outside the nest, and so are unavailable for direct as-
sessment by the developing neonate.
Work with laboratory rats indicates that pups adjust to
such environmental variation through differences in the ma-
ternal care they receive (Francis and Meaney 1999). Con-
ditions that curtail contact between mother and infant limit
the amount of licking, grooming, and arched-back nurs-
ing that the mother provides to her infants. Pups respond
to such low levels of maternal behavior by developing more
active stress-response systems. Conversely, conditions that
foster licking, grooming, and arched-back nursing between
mother and infant also facilitate development of a less ex-
citable stress-response system in the offspring. These effects
have been discovered through experimenter manipulation
of pups, and confirmed through laboratory studies of natu-
ral variation in maternal behavior (Liu et al. 1997). Such ef-
fects on HPA system development have not been demon-
strated in the field, and therefore raise important questions
for future field research (see Mateo, chap. 17 this volume).Two-way connections between proximate
and ultimate processes
The work on maternal effects on HPA function illustrates
how developmental processes are not only shaped through
evolution but also influence evolutionary processes. Mother
rats with more active HPA systems engage in less licking,
grooming, and arched-back nursing with their pups, and
this induces more active HPA systems in their pups (Francis
and Meaney 1999). Consequently, such variation in HPA
activity is heritable via maternal effects, and so is potential
raw material for the action of natural selection.
Patterns of development may also influence the evolu-
tionary persistence of a phenotypic system. Evidence for per-
sistence of systems under relaxed selection was revealed by
systematic comparisons of the antisnake behavior of eleven
populations of California ground squirrels, six of which
had been relatively free of selection from rattlesnakes or go-
pher snakes for estimated time frames of 70,000 to 300,000
years (Coss 1991a; Coss 1993; Coss 1999). Estimates of
these time frames were derived by analyzing genetic varia-
tion among thirty-one populations and calibrating this vari-
ation to time using geological evidence for the formation
of a barrier to gene flow. These results were combined with
data on interpopulation variation in rattlesnake venom re-
sistance and current densities of rattlesnakes and gopher
snakes. It was assumed that snakes not present during the
current warm interglacial period would have been even less314 Chapter Twenty-Six