Mule deer may be adversely affected by high densities of elk through several pathways such as direct displacement, forage competition, and altered predator-prey dynamics. However, the utility of reducing elk as a measure to promote mule deer populations is uncertain. During 2014-2024, the Starkey Experimental Forest and Range facility in eastern Oregon, USA, experimentally reduced elk density by approximately 80% while intensively monitoring sympatric mule deer demography before, during, and after elk reduction. The experiment was conducted within a large (7,762 ha) fenced exclosure which was fully permeable to predators, semi-permeable to mule deer, and impermeable to elk. Mule deer abundance declined consistently throughout the study from an estimated 88 adult female deer in 2014 to 18 in 2024. The sharpest mule deer population declines coincided with the 3 years of intensive elk reduction, with slower declines occurring during the 3-4 year periods of consistently high (before reduction) or low (after reduction) elk density. Mule deer pregnancy, litter size, and birth weights were apparently healthy, but annual survival of fawns and adult females were well below the expected rates for a stable population. Predation accounted for the large majority of adult and fawn deer mortality. It is uncertain whether the initially high elk density was a cause of decline for this mule deer population, but it is clear that aggressively reducing elk did not stabilize the mule deer population within a period of 4-6 years. More information is needed on the topics of residual damage to mule deer habitat following high elk densities, and how a sudden reduction of elk as a prey resource transfers to mule deer predation risk. Ungulate dynamics within a large-area exclosure may not apply directly to free-ranging populations; however, these results indicate that while high elk density may contribute to mule deer declines, elk reduction may not provide a quick-fix for struggling mule deer populations.
Mule deer may be adversely affected by high densities of elk through several pathways such as direct displacement, forage competition, and altered predator-prey dynamics. However, the utility of reducing elk as a measure to promote mule deer populations is uncertain. The Starkey Experimental Forest and Range facility in eastern Oregon, USA, has manipulated elk populations for >30 years within large semi-captive ungulate enclosures (fully permeable to predators, semi-permeable to mule deer, and impermeable to elk) to measure ecosystem effects. Mule deer demographic outcomes are reported here. Starting from a robust mule deer population of 250-400 deer within the experimental forest in the 1990s, an experiment was conducted during 1999-2001 by producing extremely high elk density (~20 elk/sq.km.) within a 1,453 ha. ungulate enclosure. Within 3 years, the high elk density resulted in the extirpation of a localized herd of ~30 mule deer from the enclosure. The high elk density enclosure was subsequently reduced to 6-10 elk/sq.km., but no mule deer returned over 20 years of subsequent monitoring. During 2014-2024, a larger (7,762 ha) study enclosure reduced elk density by 80% from moderate (4.6 elk/sq.km.) to low (0.8 elk/sq.km.) levels, while monitoring sympatric mule deer demography before, during, and after elk reduction. Mule deer abundance declined consistently throughout the study from an estimated 88 adult female deer in 2014 to 18 in 2024. Mule deer pregnancy, litter size, and birth weights were apparently healthy, but annual survival of fawns and adult females were well below the expected rates for a stable population. Predation accounted for the large majority of adult and fawn deer mortality. These results indicate a possible asymmetry in mule deer responses to elk density, where high elk density may induce mule deer declines, but elk reduction may not induce mule deer recovery. As the mule deer population of Starkey Experimental forest has declined by >90% within 30 years, elk densities are likely part of a larger pattern of degraded mule deer conditions. More information is needed on the topics of residual damage to mule deer habitat following high elk densities, and how a sudden reduction of elk as a prey resource transfers to mule deer predation risk.
Annual estimates of adult female elk abundance were calculated using an Integrated Population Model (IPM) by Loonam et al. (in review).
Annual estimates of adult female mule deer abundance were calculated using a mark-resight model for radio-collared mule deer on a summer camera survey grid during 2017–2023.
A Leslie Matrix projection of adult female mule deer population growth was calculated using radio-collared mule deer parameters during 2014–2024. For the sake of comparison, the mule deer matrix projection was fit to the 2018 mark-resight abundance estimate (51 adult females), and then projected to future and past years.
Ultimately, an IPM that combines the collared deer sample and mark-resight population parameters should offer the best mule deer abundance estimates. Given the high agreement of the collared sample and mark-resight estimates, a reliable IPM fit for 2014-2024 mule deer population seems likely.
The map below summarizes 1,658 mule deer detection events during 2017-2023 on the summer camera survey grid.
Camera locations were consistent among years. Collared adult female mule deer had unique belting markers and high-intensity GPS relocation intervals, which allowed nearly all June-August camera detections of collared females to be identified to the individual. The marked and unmarked adult female mule deer detections were then used to fit annual mark-resight population estimates.
Survival Rates were calculated for adult female and fawns (both sexes) using cumulative incidence functions (CIFs). A Kaplan-Meier function was used for overall survival and an Aalen-Johansen function was used for cause-specific rates. A 1-day time step was used with staggered entry for animals captured during a biological year.
Annual survival was estimated from a biological year starting on June 1st - the subsequent May 31st (e.g., 2023 represents survival from June 1 2023 – May 31 2024). Adult female survival was calculated by actual date, with live females rolling-over into the next biological year on 1 June. Fawn survival was calculated in days from birth to 365 days-old, which was then approximated to June 1 for population summaries. In other words, the actual dates of fawn birth, capture, and mortality/censor were used to calculate survival, then the resulting estimates were re-assigned to a universal 1 June birth date for population modeling.
In addition to survival, other demographic parameters measured for the collared mule deer that can be applied to a population model included mean litter size (1.56 fawns from VIT birth site captures) and sex ratio at birth (46% females). The presence of a few un-collared siblings on the camera survey demonstrates that we occasionally failed to find the full litter at VIT searches, so I slightly boosted litter size to 1.7 in the Leslie Matrix to account for imperfect detection.
The daily survival curves and cause-specific rates were calculated on a biological year starting on June 1st, but with all years pooled into a single 365-day sample.