Predicting non-target impacts from a biocontrol weevil on a federally threatened thistle of the upper Great Lakes

Abstract

Better understanding of the ecological and evolutionary consequences of plant-insect interactions has important applications in ecology, agriculture, and pest management. Larinus planus Fabricius (Coleoptera: Curculionidae), a Eurasian seed predator, was distributed in the U.S. and Canada as a biocontrol for weedy thistles. It now has non-target impacts on natives, including the federally listed shoreline endemic, Cirsium pitcheri [Torr. ex Eaton] Torr. and A. Gray (Pitcher's thistle). Females weevils lay eggs (oviposit) in floral heads; larvae then consume ovules and seeds as they develop. I asked: 1) What factors put some C. pitcheri plants at risk from L. planus? and 2) How might non-target impacts affect population dynamics? Damage from Larinus planus on flowering C. pitcheri occurred mostly in the northern and shoreward dunes. Impact of L. planus on C. pitcheri can be influenced by: 1) abundance of C. pitcheri as host plant and 2) other plant associates in the community. I measured these factors in 77 randomly distributed 100 m2 plots in two state parks in northern lower Michigan, including a long-term study site at Sturgeon Bay, Wilderness State Park. Larinus planus damage was not significantly associated with abundance of flowering C. pitcheri or percent cover of other plant associates. I also used discriminant function analysis (DFA) to ask which C. pitcheri plants were at risk of damage by L. planus including the factors of northing and easting coordinates, host abundance, and percent vegetation cover. The DFA forecasted 73.1% of plants that would have received damage from L. planus. We rely on the use of models to predict population size and viability for the conservation of rare species. I compared L. planus impacts on a C. pitcheri population using two population models: stage-based matrix (SBM) and integral projection (IPM) models. SBM has some biases; IPM is offered as an alternative model. The models showed similar trends in yearly population growth rates and predicted similar time to extinction for C. pitcheri populations. Seed predation by weevils decreased population time to extinction in both models, but SBM showed a greater decrease than IPM. My elasticity analysis also confirms earlier suggestions that survival of the smallest individuals most affects population growth and should be the focus of conservation. Effective pest management, including unintended damage from biocontrol agents, requires extensive biological knowledge of plant-insect interactions. For L. planus, host plant abundance and other plant associates do not appear to directly influence host choice; however, sample sizes and use of non-parametric analyses may limit the strength of my inferences. Further study is needed on behavior, population dynamics, and overwintering of this weevil. For C. pitcheri, less intensive stage-based modelling offers outcomes similar to integral projection approaches and remains a viable tool for conservation. Further development of appropriate assessment of Pitcher's thistle fecundity, relative to plant size and seedling recruitment, is a needed next step. Population-level impacts from seed predators are significant, no matter what the model, and are greater for rare plants. Assessment of new threats from regular monitoring and adaptive management are critical for effective conservation, particularly in a time of global change.