Effects of biological control on long-term population dynamics: identifying unexpected outcomes.
Attempts to control natural systems through management have often met with success but have also led to unexpected and often undesirable outcomes. Unfortunately, the ultimate result of such management programmes may not be apparent until long after the control efforts have begun. This is particularly true for forest-defoliating species that exhibit long-period cycles such as the invasive gypsy moth Lymantria dispar, which causes widespread damage in some years but is rare in other years. We studied the effects of two commonly employed biocontrol agents on gypsy moth dynamics using a series of field-tested and empirically parameterized mathematical models, which allowed us to examine various potential control strategies and assess long-term effects. In a non-spatial model, addition of either a manufactured version of the same baculovirus involved in natural epizootics, or a general bioinsecticide Bacillus thuringiensis var. kurstaki (Btk), which directly kills a fraction of the population, decreases the amplitude between boom and bust portions of the cycle. However, ill-planned biocontrol applications can result in increased gypsy moth densities over the long term. Thus, control efforts may maintain pest populations at unexpectedly high numbers, which could result in constant forest defoliation. In a spatial two-patch model, where one patch is sprayed and the other is left untreated, there is also considerable danger that migration between patches may drive the unsprayed population to levels that could result in constant forest defoliation. Synthesis and applications. Perturbations to host-pathogen systems may have unexpected results, driving and maintaining populations at multiple levels including those far from desired management goals. It is often assumed that any control strategy that decreases pest populations in the short term is beneficial, but our results show that undesirable outcomes may often occur. The mechanisms we describe apply to many systems that undergo population cycles or outbreaks regulated by density-dependent processes, and in which disease or pesticide application is used for pest control. We suggest that successful management strategies should closely monitor population responses immediately following the control application to ensure that pest populations are not being maintained at artificially high levels compared with historic data.