due to the presence of migratory and wintering raptors. The
winter brings a reprieve for the kangaroo rats as the rattle-
snakes go torpid, and the presence of owls scares the ro-
dents into preferring the bush microhabitat (Brown 1989).
In addition to climate-driven shifts in predation risk, des-
erts see pulses of productivity following winter and /or sum-
mer rains. During this period desert rodents almost uni-
versally employ one of two strategies: food caching and /or
seasonal dormancy. For pocket mice (Perognathus and
Chaedopidus) and ground squirrels dormancy means that
the rodents only experience risk during their activity sea-
son. Owls may time their breeding to correspond to peri-
ods of peak activity and densities of pocket mice, and
indeed pocket mice are more vulnerable to owl predation
than are kangaroo rats (Kotler 1985). Caching behavior
provides desert rodents with a powerful antipredator tool:
namely, the opportunity to focus foraging behavior toward
periods high in food and /or low in predation risk. Cach-
ing behavior probably evolved in response to seasonal
food production and the low perishability of seeds. Once
evolved, however, it becomes a “bank account” that al-
lows the rodents to forgo any costly foraging periods. The
latitude of antipredator behaviors afforded by caching
may be what renders desert rodent systems noncyclic and
m-driven.
Conclusions and Future Prospects
If we draw a line from the Negev desert in Israel to Finnish
Lappland by Kilpisjärvi and Pallasjärvi, both well studied
in terms of empirical and theoretical rodent ecology, we
have a productivity and cover gradient over a range of fa-
miliar vegetation types. Productivity is very low in the des-
ert, increases strongly in temperate grasslands and forests,
remains high in boreal habitats and decreases again toward
northern subarctic habitats (T. Oksanen and Henttonen
1996). Cover and shelter from predation follow the same
curve but change in northern boreal and subarctic habitats,
where strong seasonality and permanent snow during win-
ter provide shelter from cold and predation (Hansson and
Henttonen 1985). Rodent abundance is low in the desert,
high and fluctuating in temperate regions (Tkadlec and
Stenseth 2001), and high but strongly fluctuating (and of-
ten cyclic) in boreal and subarctic habitats (e.g., Hanski
et al. 2001).
Predator assemblages change less, however, having the
greatest diversity and numbers in temperate regions and
southern parts of boreal habitats. Stronger seasonality and
long-lasting permanent snow cover cause a strong seasonal
change in the predator guild in the north. Most raptors de-
part the areas during winter, leaving a small number of res-
ident mammalian predators and owls (T. Oksanen et al.
2000). Let us now try to combine the impact of productiv-
ity, shelter, and prey and predator numbers to assess the
joint per capita predation riskfor gerbils, temperate-region
mice, and voles across the gradient (table 28.1).
Desert, temperate, and boreal regions probably offer
rodents quite different risk regimes. In deserts, a very high
per-unit-foraging-time predation risk likely promotes im-
mediate and density-dependent responses in gerbils to risk.
Gerbils respond by spending little time active per night.
Temperate regions see high numbers of prey together with
high numbers of diverse predators, but due to a dilution ef-
fect, the per-unit-foraging-time predation risk is lower than
in the desert. Effects of predators on prey populations —
and behavior — are directly density-dependent and probably
more N-driven than in desert systems and more m-driven
than in boreal systems (T. Oksanen et al. 2000). While risk
per unit activity is lower than in deserts, each individual
probably spends more time actively foraging. In the north-
ern boreal areas and the subarctic, per capita predation risk
even on a per-foraging-time basis is higher than for tem-
perate mice, probably due to the impact of vole-specialist
predators such as weasels and stoats. The dynamics of the
predators and voles exhibit time lags and delayed density-
dependent effects. Furthermore, boreal voles spend a rela-
tively large amount of time foraging and exposed to preda-
tion risk, and predation risk may remain relatively high
even when the voles are in their burrows.
The gradient of fear, foraging, breeding, and population
dynamics proposed here is not yet proven but is rather a se-
ries of interconnected hypotheses (fig. 28.4). As a more syn-
thetic conceptual framework develops that is applicable to
all rodents, new research avenues will develop that apply
similar research tools over similar spatial and time scales.
Research tools that can be used more systematically across
desert, temperate, and boreal systems include censusing
protocols, measurements of populations, social and breed-
ing structures, GUDs, the duration and scale of enclosure
and aviary experiments, and the protocols for food enrich-
ments, habitat modifications, and manipulations of actual
or perceived predation risk.
The synthesis of conceptual tools will require an under-
standing of how fear influences foraging, breeding, and so-
cial behaviors and of how predation risk interacts with
specific environmental circumstances to produce m-driven
versus N-driven predator-prey systems. Fear, the nonlethal
effect of predators on their prey, has far-reaching conse-
quences for the life histories of rodents. Rodent feeding,
breeding, and social behaviors are all impacted by fear ofFear and the Foraging, Breeding, and Sociality of Rodents 339