Vigilance behavior can aid in the detection of predators and may also play a role in observation of conspecifics, in food acquisition, and in the prevention of kleptoparasitism. However, in most occasions, vigilance is most important as an antipredator function. Generally, factors that increase the risk of predation also increase the amount of vigilance. We examined whether the reintroduction of the large predator, the wolf, in Yellowstone National Park (YNP) would influence coyote vigilance and foraging ecology. From December 1997 to July 2000, we collected 1743 h of coyote activity budgets. Coyote home ranges occurred within wolf territories (termed high-use or nonbuffer zone areas) and also between them in buffer zones. In high wolf use areas as well as when wolves were present, coyotes fed on carcasses much more; however, they increased the amount of vigilance and decreased rest to prevent predation. Wolf kills may provide a quick source of food and be energetically advantageous to coyotes; however, costs include increased vigilance, decreased rest, and a higher predation risk. Vigilance and avoidance behavioral responses to the reintroduction of large predators may ultimately be more common outcomes than actual killing by competing carnivores of prey. Keystone carnivore reintroductions have a variety of cascading effects throughout the ecosystem and can be driven by both numeric responses (trophic cascades) and behavioral responses ("behavioral cascades"). Behavioral cascades resulting from increased vigilance or spatial changes may lead ultimately to numeric changes and trophic cascades.

Long-distance vocalizations by canids play an important role in communication among individuals. I evaluated efficacy of broadcasted coyote (Canis latrans) group-yip calls and gray wolf (C. lupus) lone howls to elicit vocal responses from 18 GPS-collared coyotes on 144 occasions. I concluded that eliciting coyote vocalizations where wolves are present will not bias responses, and recommend eliciting coyote vocalizations using recorded coyote group-yip howls during July–September to estimate species’ presence or density. From foraging theory, generalist predators should increase consumption of prey if prey availability increases. I estimated densities for coyotes, adult deer, and fawns, and collected coyote scat to estimate occurrence and biomass of adult and fawn deer consumed by coyotes during 2 periods. I suggest that consumption rates of coyotes was associated positively with increases in fawn density, and fawn consumption by coyotes follows predictions of foraging theory during this pulsed resource event.

Food webs consist of a combination of bottom-up (resource-driven) and top-down (predator-driven) effects. The strength of these effects depends on the context in which they occur. I investigated food web (trophic) relationships between wolf (Canis lupus) predation, elk (Cervus elaphus) herbivory, aspen (Populus tremuloides Michaux) recruitment, and fire. The study setting, in the central portion of the Crown of the Continent Ecosystem, spans the US/Canada border and encompasses Glacier National Park (GNP), Montana and Waterton Lakes National Park (WLNP), Alberta. I stratified my observations across three spatially distinct areas, the North Fork Valley, in the western portion of GNP; the Waterton Valley, in the eastern portion of WLNP; and the Saint Mary Valley, in the eastern portion of GNP. All valleys are elk winter range (low-lying grasslands with patches of aspen). The valleys have three different observed wolf population levels (Saint Mary: low; Waterton: moderate; North Fork: high), which represent three levels of long-term predation risk (the probability of an elk encountering a wolf). Ecological characteristics (e.g., climate, soils, elevation, plant associations) are comparable among valleys. Fire has occurred in 90% of the North Fork. My objective was to examine the relative influence of bottom-up (fire) and top-down (predation risk) factors and the context-dependence of these relationships via data gathered during a three-year time span. I found complex elk responses to bottom-up and top-down factors that could influence habitat use by elk. Pellet transect data demonstrated that elk exhibited the same risk reduction behavior at all wolf population levels, even at very low levels. Predation risk variables that provided impediments to detecting or escaping wolves had a similar and negative influence on occurrence of elk (pellet piles), regardless of wolf population density. Fire had a negative effect on elk density and a positive effect on wolf density (per scat piles) in aspen communities where a high wolf population existed. Aspen cover, which may be riskier than open grassland, also had a negative effect on elk density, except at very high wolf levels without fire. The risk of wolf predation alone did not drive elk behavior. Conversely, focal animal (elk vigilance behavior) data suggested a positive relationship between wolf population and elk vigilance. However, when I deconstructed vigilance, elk demonstrated complex, context-dependent adaptive behavior in response to the long-term risk of predation by wolves. Commonly identified drivers of elk vigilance (group size, impediments to wolf detection and escape) appeared to be important drivers at an intermediate level of long-term predation risk (e.g., Waterton). These drivers ceased to function in this manner when the long-term predation risk level increased (The North Fork). At high levels of long-term predation risk, vigilance was high, but not driven by these common factors. In some cases, the relationship between vigilance and risk factors was reversed (e.g., group size). And at a low level of long-term predation risk (Saint Mary), elk did not respond to these drivers of vigilance. When I measured aspen demography (browse, recruitment), browse was lower in the North Fork, where there was a high wolf population, suggesting a top-down effect. However, I found low aspen recruitment in the absence of fire in all valleys, which indicates a bottom-up effect in that aspen is highly fire-dependent. Top-down predictors of aspen recruitment (e.g., plot position and stand size, which are related to predation risk) had no effect on browse levels regardless of wolf population level. In sum, the risk of wolf predation alone did not drive the food web relationships I observed. Bottom-up and top-down forces worked together in valleys that contained well-established wolf populations, and to a lesser degree in a valley with a low wolf population. Commonly used measures of predation risk responses (e.g., vigilance) reversed their relationship as the wolf population increased. Low aspen recruitment in the absence of fire demonstrates the importance of bottom-up effects. Bottom-up and top-down effects may be important joint engineers of aspen communities. My findings invite deeper inquiry into the interaction between bottom-up and top-down effects in large mammal systems. / Graduation date: 2012