Temporal variation in site fidelity: scale-dependent effects of forage abundance and predation risk in a non-migratory large herbivore.

Large herbivores are typically confronted by considerable spatial and temporal variation in forage abundance and predation risk. Although animals can employ a range of behaviours to balance these limiting factors, scale-dependent movement patterns are expected to be an effective strategy to reduce predation risk and optimise foraging opportunities. We tested this prediction by quantifying site fidelity of global positioning system-collared, non-migratory female elk (Cervus canadensis manitobensis) across multiple nested temporal scales using a long-established elk-wolf (Canis lupus) system in Manitoba, Canada. Using a hierarchical analytical approach, we determined the combined effect of forage abundance and predation risk on variation in site fidelity within four seasons across four nested temporal scales: monthly, biweekly, weekly, daily. Site fidelity of female elk was positively related to forage-rich habitat across all seasons and most temporal scales. At the biweekly, weekly and daily scales, elk became increasingly attached to low forage habitat when risk was high (e.g. when wolves were close or pack sizes were large), which supports the notion that predator-avoidance movements lead to a trade-off between energetic requirements and safety. Unexpectedly, predation risk at the monthly scale increased fidelity, which may indicate that elk use multiple behavioural responses (e.g. movement, vigilance, and aggregation) simultaneously to dilute predation risk, especially at longer temporal scales. Our study clearly shows that forage abundance and predation risk are important scale-dependent determinants of variation in site fidelity of non-migratory female elk and that their combined effect is most apparent at short temporal scales. Insight into the scale-dependent behavioural responses of ungulate populations to limiting factors such as predation risk and forage variability is essential to infer the fitness costs incurred.