Tailoring biocontrol to maximize top-down effects: on the importance of underlying site fertility.

The degree to which biocontrol agents impact invasive plants varies widely across landscapes, often for unknown reasons. Understanding this variability can help optimize invasive species management while also informing our understanding of trophic linkages. To address these issues, we tested three hypotheses with contrasting predictions regarding the likelihood of biocontrol success. (1) The biocontrol effort hypothesis: invasive populations are regulated primarily by top-down effects, predicting that increased biocontrol efforts alone (e.g., more individuals of a given biocontrol agent or more time since agent release) will enhance biocontrol success. (2) The relative fertility hypothesis: invasive populations are regulated primarily by bottom-up effects, predicting that nutrient enrichment will increase dominance by invasives and thus reduce biocontrol success, regardless of biocontrol efforts. (3) The fertility-dependent biocontrol effort hypothesis: top-down effects will only regulate invasive populations if bottom-up effects are weak. It predicts that greater biocontrol efforts will increase biocontrol success, but only in low-nutrient sites. To test these hypotheses, we surveyed 46 sites across three states with prior releases of Galerucella beetles, the most common biocontrol agents used against invasive purple loosestrife (Lythrum salicaria). We found strong support for the fertility-dependent biocontrol effort hypothesis, as biocontrol success occurred most often with greater biocontrol efforts, but only in low-fertility sites. This result held for early stage metrics of biocontrol success (higher Galerucella abundance) and ultimate biocontrol outcomes (decreased loosestrife plant size and abundance). Presence of the invasive grass Phalaris arundinacea was also inversely related to loosestrife abundance, suggesting that biocontrol-based reductions in loosestrife made secondary invasion by P. arundinacea more likely. Our data suggest that low-nutrient sites be prioritized for loosestrife biocontrol and that future monitoring account for variation in site fertility or work to mitigate it. We introduce a new framework that integrates our findings with conflicting patterns previously reported from other biocontrol systems, proposing a unimodal relationship whereby nutrient availability enhances biocontrol success in low-nutrient sites but hampers it in high-nutrient sites. Our results represent one of the first examples of biocontrol success depending on site fertility, which has the potential to inform biocontrol-based management decisions across entire regions and among contrasting systems.