Accounting for food web dynamics when assessing the impact of mesopredator control on declining prey populations

• Published in: The Journal of applied ecology Vol. 58; no. 1; pp. 104 - 113
• Main Authors: Henden, John‐André, Ehrich, Dorothee, Soininen, Eeva M., Ims, Rolf A., Sales, Lilian
• Format: Journal Article
• Language: English
• Published: Oxford Blackwell Publishing Ltd 01-01-2021; Wiley
• Subjects: Animal behavior; Animal populations; Bird populations; Birds; carrion; Culling; Food chains; Food webs; Foxes; generalist predators; Impact analysis; management evaluation; Monitoring; Polar environments; Population decline; Population density; Predation; Prey; red fox; Replication; rodents; Tundra; Wildlife; Wildlife management; willow ptarmigan
• ISSN: 0021-8901; 1365-2664; 1365-2664
• DOI: 10.1111/1365-2664.13793

Increasing populations of mesopredators are suspected to cause declines in vulnerable wildlife to the extent that mesopredator decimation actions (culling) have become commonplace. Design constraints, especially a lack of spatial replication, often hamper the assessment of the impact of such actions. However, extensive temporal replication (i.e. time series) and accounting for potentially confounding variables may alleviate this problem.
In alpine‐arctic tundra, the red fox Vulpes vulpes is increasing, while many bird species are declining, likely due to increased predation. Here, we assessed the impact of a long‐term (12‐year) and spatially extensive (~3,500 km2) red fox culling action on the red‐listed willow ptarmigan Lagopus lagopus in the Norwegian Arctic. Ptarmigan populations were monitored annually in the impact area and in an adjacent no‐action area, including a 5‐year period before the action commenced. While logistical constraints prohibited monitoring of red fox population densities, the number of culled foxes and three influential food web covariates were monitored after the onset of the culling action.
A Before‐After‐Control‐Impact‐Paired‐Series (BACIPS) analysis without food web covariates indicated that red fox culling curbed the decline of the population in the impact area, and that ptarmigan population density became ~25% higher than in the reference area.
Spatially and temporally variable drivers within the food web confounded the simple BACIPS analysis. Accounting for three food web drivers as covariates in a linear mixed model after the onset of action, yielded a more unbiased impact estimate that amounted to ~40% higher ptarmigan population density (4.3 more ptarmigan/km2) in the red fox impact area.
Synthesis and applications. We provide the first evidence of the role of the recent expansion of red fox in the decline of bird populations in tundra. We also show that red fox culling may be able to curb such declines, given that management actions are large scale and long term. As mesopredator culling campaigns are often expensive and controversial, it is important that their impacts are accurately assessed. We demonstrate that the accuracy of impact assessments can be profoundly increased by monitoring drivers of food web dynamics that impinge on the target species so that such drivers can be included as covariates in the analysis. This applies in particular to declining bird populations in boreal and arctic food webs ruled by strong multi‐annual interaction cycles.
We provide the first evidence of the role of the recent expansion of red fox in the decline of bird populations in tundra. We also show that red fox culling may be able to curb such declines, given that management actions are large scale and long term. As mesopredator culling campaigns are often expensive and controversial, it is important that their impacts are accurately assessed. We demonstrate that the accuracy of impact assessments can be profoundly increased by monitoring drivers of food web dynamics that impinge on the target species so that such drivers can be included as covariates in the analysis. This applies in particular to declining bird populations in boreal and arctic food webs ruled by strong multi‐annual interaction cycles.