likely to be present during past colder glacial periods. The
systems subserving the antisnake behavior of California
ground squirrels proved to be remarkably persistent, span-
ning gaps of up to 300,000 years in which selection from
snakes had been relaxed. This persistence was not simply
the result of occasional encounters with rattlesnakes in
snake-rare areas. Not a single population in a snake-rare
area exhibited high levels of venom resistance, even though
they continued to recognize snakes as dangerous (Coss and
Goldthwaite 1995).
Persistence of behavioral antisnake defenses may be a
byproduct of selection for reliable expression of this im-
portant antipredator system in individuals large enough to
withstand envenomation. Such selection may have had the
effect of locating the neural substrates for antisnake behav-
ior on the early-developing proximal processes of develop-
ing neurons (Coss 1991b; Coss 1999). This pattern of early
development protects the neural system from alterations via
the less predictable experiences mediating neural growth
as the older animal engages with a broader, less-predictable
world outside the nest (Coss 1991b). But early “installa-
tion” of the neural substrate subserving antisnake behav-
ior produces developmental entrenchment of this system by
making it part of the foundation upon which development
of the rest of the phenotype depends (as in Schank and
Wimsatt 2000). Entrenchment in turn increases the nega-
tive developmental consequences of modifying the system
and thereby adds to its evolutionary persistence (Coss and
Moore 2002).
Installation of the neural bases of antisnake behavior
early in development has had a cascading development con-
sequence that may well have modified the social compo-
nent of antisnake behavior. As we described in our model
episode, California ground squirrel pups express adult-
like antisnake behavior early, before they have the venom-
neutralizing capacity or behavioral skills to handle the risk
posed by confronting a rattlesnake (Poran and Coss 1990;
Coss 1991a; Owings 2002). Such precocious, risky con-
frontation of snakes not only places pups in greater danger
but also complicates the protective mother’s social task of
managing her pups while dealing with snakes (see fig. 26.1).
Precocious snake confrontation does not appear to involve
nepotistic self-sacrifice for siblings; the antisnake behavior
of these pups is more self-interested than nepotistic (Hersek
and Owings 1994).
Finally, a major theme of this chapter, the sharing of
components by different systems, also appears to have
had evolutionary consequences. Another important part of
California ground squirrel defense against rattlesnakes is
the venom-neutralizing system, a system that shows much
greater evidence of adaptive intraspecies variation than the
behavioral defense system (Poran et al. 1987; Coss 1999).
This indicates that the venom-neutralizing system is more
evolutionarily labile than the behavioral system, a difference
that may be due in part to differential sharing of compo-
nents with other systems. The venom-neutralizing system is
dedicated to that function (Biardi et al. 2000), whereas the
behavior system shares many parts with other functional
systems, especially social systems, as discussed in this chap-
ter. Continuing selection associated with those social func-
tions has the potential to maintain a multiplexed system like
the behavioral antisnake system even when selection asso-
ciated with the antisnake functions has been relaxed for
thousands of generations (Hodgkin 1998; Coss 1999). This
idea — that dedicated systems disintegrate more rapidly than
multiplex systems — seems more applicable to these results
than the idea that this differential persistence is generated
by differential maintenance costs. From the perspective of
differential cost, the behavioral system should disintegrate
more rapidly because it is far more complex and thus pre-
sumably more costly.Questions for Future ResearchThe observations and ideas discussed in this chapter raise
some important questions for future research. First, we have
seen that shared components can cause one system to limit
useful change in another. What other kinds of connections
between organismic systems can have that effect? Means-
end relations are another obvious candidate, but are there
additional ones? Next, the kinds of factors that we have dis-
cussed also indicate that adaptive modification in organis-
mic systems may be limited by the need for the organism
to remain functionally coherent. Given this constraint, how
is it that complex organisms manage to adjust individual
systems to new conditions? For evolution, learning, or any
other form of change to be feasible, organismic systems need
to be at least quasi-independent (Lewontin and Levins 1978;
Hodgkin 1998). Such independence may be fostered by such
processes as gene copying and the establishment of modu-
lar systems (Schank and Wimsatt 2000). These considera-
tions help us to identify one of the most challenging current
questions in the study of the adaptive adjustment of behav-
ioral systems: how does adaptive adjustment of individual
organismic systems coexist with functional coherence of
whole organisms (Lewontin and Levins 1978)?SummaryThis chapter has explored connections and similarities be-
tween rodent social and antipredator behavior as well as
the processes that generate them. We have relied heavily onSocial and Antipredator Systems: Intertwining Links in Multiple Time Frames 315