The effect of insect biological control agents on invasive plant populations can vary spatially, and spatial variation in climate may drive regional variation in herbivory, and thus biocontrol efficacy. Within host plant populations, local plant abundance can also be affected by the spatial distribution of herbivores, but whether local patterns persist at larger scales is less well understood. We examined how infestation and damage of the stem-mining weevil Mecinus janthiniformis, a specialist biocontrol agent of the invasive plant Linaria dalmatica, varied with and among populations across the northern edge of the range in North America. We quantified weevil and invasive plant densities, as well as plant fecundity, stem diameter and height across sites spanning an area of ~39,000 km² in British Columbia, Canada. We found that specialist weevils did not respond to host plant density within sites across the study region. Instead, weevil attack and load were most sensitive to among-site variability in climate, with stems at warmer sites have four times as many weevils compared to stems at cooler sites. Weevils also reduced plant fecundity more at warmer sites, when controlled for plant size (stem diameter) with larger effects of weevils in thicker stems, indicating that L. dalmatica suppression is highest in warmer locations where weevils are more abundant and environmental conditions more favorable. Our results suggest that, with climate change, the efficacy of biocontrol for L. dalmatica will improve across the northern edge of the range. We recommend that at cooler sites, where biocontrols are less prevalent and effects on plant fecundity are weaker, alternative management strategies are necessary at this time. Across invasive plant species more generally, future studies that establish the role of climate in variability in biocontrol efficacy long after introduction can improve management of invasive plants under climate change.

We surveyed L. dalmatica populations across a wide swath of British Columbia, on the eastern side of the Canadian Cascade mountains in the summer of 2017, encompassing an area of approximately 39,000 km² (Figure 1, Table A1). In consultation with local invasive plant managers and data from the British Columbia Invasive Alien Plant Program (B.C. Ministry of Forests, Lands and Natural Resource Operations, 2017), we selected sites where no active management, e.g. herbicide spray or mowing, was planned for 2017, with the goal of sampling as broadly as possible. Since L. dalmatica populations respond positively to disturbance (Sing et al., 2016), our goal was to choose sites of similar levels of disturbance, and all sites were located within 100 meters of primary or secondary roads. 39 sites were used in the analyses (those we were able to access in late summer to collect stems) and data from those sites are included here.

In the field, we surveyed stem density in 1x1 m² plots - and counted vegetative and flowering stems separately. Later in the summer, we collected at least 20 stems from each site (10 from low density and 10 from high density plots) to bring back to the lab to quantify size, seed production, and dissect to count Mecinus janthiniformis individuals (the biological control). Please see Methods for details of data collection and analyses.

No proprietary programs required (all data in the mansucript were viewed and analyzed in R).