A forager should abandon a depletable food patch when the benefits derived from its present harvest rate no longer exceed the costs of foraging in the patch (Brown 1988
; Brown et al. 1997
). Consequently, foragers should demand higher harvest rates from resource patches with higher foraging costs and quit them at higher quitting harvest rates and giving up densities. We presented foxes with identical food patches in terms of their metabolic, predation, and missed opportunity costs, but with different risks of injury. Thus, the differences in the GUDs between the patches must be attributed to the differences in the risk of injury between the patches.
Our results show that foxes leave higher GUDs in resource patches with higher risk of injury than they do in safer patches. This result strongly suggests that foragers treat risk of injury as a foraging cost. Foxes also leave higher GUDs when they are less hungry. These state-dependent changes in the foxes' foraging strategies may have come about via two main pathways: (1) As the forager grows hungrier, its missed opportunities cost of foraging decreases because acquiring food becomes a more valuable activity for it, and it spends more time foraging. (2) The cost of injury for a forager is a product of the injury risk and the marginal rate of substitution of energy for injury (MRSei
is the quotient of two components: survivor's fitness and the marginal value of energy (F
; Brown 1992
; Kotler 1997
). Thus, hungrier foragers (higher marginal value of energy) are expected to take higher risks and be willing to spend more time foraging in risky patches. Our results demonstrate that foxes treat the foraging cost arising from the risk of injury in a similar manner to which other foragers treat the foraging cost arising from the risk of predation. Hence, an optimally foraging animal should forage in a patch as long as its harvest rate exceeds the sum of its metabolic, predation, missed opportunities, and injury costs associated with foraging in that patch.
The riskiness of the patch affected both the slopes of the harvest rate curves and their curvatures. The different shapes of the curves can reveal information about the foraging tactics of the fox. The slopes of the gain curves in the safe patches were steeper, i.e., the rate of food gain was higher in these patches. As both food patches contained an identical initial amount of food mixed thoroughly throughout the patch, the change in the rate of food gain between the patches has to be a result of differences in the fox's foraging behavior while exploiting them. A slower intake rate could reflect a more cautious approach to the patch that would increase the time spent searching for and extracting food from the patch. Our video footage supports the notion that the fox approached the risky patches with much more care than the safe patches. Thus, the increase in foraging costs in the risky patch is likely due to elevated searching and handling time in that patch, i.e., the fox reduced its intake rate in order to decrease the chances of injuring itself. As an injury could reduce the future intake rate of the fox, such behavior might reflect a trade-off between long-term and short-term intake rates.
The state of the fox also affected its harvest rate. When the fox was very hungry, the slopes of the curves were overall gentler, at least compared to the fox exploiting the safer patch when it was less hungry. This at first may appear to be somewhat counterintuitive, as a hungrier animal can be expected to forage more rapidly and less carefully and thus have a higher rate of food consumption. We believe that the slower overall intake rate of the very hungry fox is due to more thorough patch exploitation by the fox. Because of its higher marginal value of energy when very hungry, the fox chose a lower GUD and therefore spent substantially more time foraging in each patch (Fig. ). That is, we recorded many more data points for the very hungry fox at especially long foraging times. These times were associated with low harvest rates due to extensive patch depletion by that time. When averaging overall rate, these points lower the average. Furthermore, the approximately equal numbers of points for the hungry and very hungry treatments are spread differently. Those for the hungry state tend to be bunched at times less than 20 min; those for the very hungry state are much more evenly spread across the entire range of observations. Thus, we have a better rendering of the shape of the harvest rate curve for the very hungry treatment, and it is weighted more heavily for the lower harvest rates incurred at long foraging times. In fact, for the first 20 min of patch exploitation while the fox removed the easy-to-get food items on top, there is little difference in harvest rate for any patch and condition. Any differences there are more statistical than biologically meaningful.
There were no significant interactions between the effects of the patch type and the state of the fox on its harvest rate curves when only three-way interactions are included. Thus, the riskiness of the patch and the fox's state have an additive effect on the fox behavior. Interestingly, when we include the four-way interaction of risk × state × time × time2, it is significant, but then neither the lower order interactions nor the main effects are. Four-way interactions can be extremely difficult to interpret, so we say with great caution that this may offer support for the interaction of state and risk in determining the curvature of the harvest rate curves in a manner consistent with the use of daring, i.e., stronger separation between the curves when the fox is very hungry.
Most foragers are likely to face both risks—predation and injury—at the same time. However, they may differ in the relative weights of the different foraging costs. Top predators may be especially susceptible to risk of injury. Predators must pursue, capture, and kill others in order to eat, and in so doing, they may hurt themselves. In extreme cases, predators may actually be killed by their intended victims, as may happen when African lions attack buffalo (Mangani 1962
; Beyers 1964
) or wolves attack moose (Mech 1966
). Thus, for top predators, which are often assumed to be free of predation costs, the heightened chances of risk of injury may be their main foraging cost, equivalent to the predation cost of their prey.
Just as prey individuals manage risk of predation, using behavioral tools of time allocation and vigilance (Brown 1999
; Bouskila 2001
), predators too should manage their own risks while foraging. To this end, predators may have several behavioral tools at their disposal such as: (1) selecting for prey size and type (e.g., Rutten et al. 2006
); (2) using time allocation to decide how long to hunt in a certain area and how long to pursue a prey item; and (3) varying their degree of daring. Daring can be defined as the willingness of the animal to risk injury. A more daring predator is likely to be more lethal (higher capture rate), but also may suffer a higher risk of injury while foraging. The willingness of the predator to take these risks may strongly depend on its energetic state. Predators may use any or all of these behavioral tools to maximize their expected fitness while facing a trade-off between energy reward and the risk of injury they face while foraging (Gilchrist et al. 1998
; Brown and Kotler 2004
; Rutten et al. 2006
In our experiment, the fox used the behavioral tools of both time allocation and daring for managing its risk of injury. It spent more time foraging in the safe patches even though it was exposed to both types of patches for exactly the same amount of time. Also, despite its apparent discomfort (personal observation) and the risk of injuring itself, the fox increased the amount of time foraging in the risky patch when it was hungrier. This willingness to risk injury could be regarded as daring behavior. The fox was willing to stay longer in the risky patch in order to increase his energetic gain. Daring behavior, while risky, may increase the lethality of a predator (e.g., an owl increasing its diving speed, a bird eating a butterfly despite its warning colors, etc). This, in turn, may cause a change in the prey behavior and in the long run may even have an effect on the evolution of certain prey species (Sherratt 2003