Predator refuges for conservation biological control in an intermediately disturbed system: the rise and fall of a simple solution

1. Managed systems harvested at intermediate time-scales have advantages over annual short-cycled systems in maintaining top-down control of insect herbivores, and the flexible harvest regimes in these systems provide opportunities for habitat management that can stabilize predator–prey population dynamics across harvests – resulting in reduced risk of pest outbreaks. 2. In a large-scale field experiment, we explored whether retaining refuges, that is preserving parts of the stand to reduce predator mortality, could reduce the risk of pest insect outbreaks in willow short-rotation coppice. Population densities of three omnivorous predator species and three outbreaking herbivorous leaf beetle species were monitored over four years after coppice (stem harvest) in eight stands with refuges (treatment) and eight stands without refuges (control). Predation pressure was estimated in years three and four. 3. Contrary to our predictions, leaf beetle densities were higher in stands with refuges and predator densities were higher in stands without refuges. Leaf beetle egg mortality increased with total predator density, but did not differ between stands with and without refuges. 4. These unexpected results can be attributed to interactions between dispersal and patch age. The altered phenology of coppiced stems may have triggered leaf beetle aggregation in refuges and migration from stands without refuges. A behavioural response to resource concentration in retained old patches likely transformed the predator refuge from a ‘source’ to a ‘sink’. 5. Synthesis and applications. This study shows that retaining refuges in willow short-rotation coppice to facilitate predator population recovery after harvest can come at the cost of more attractive herbivore habitats – and thus increased pest problems. We conclude that crop refuges in systems with intermediate disturbance regimes pose new challenges for conservation biological control, in particular the need to consider how patch age affects dispersal and recolonization of both pest and predators.