Plants provide a variety of mutualistic bodyguarding insects with various provisions: extra-floral nectar, food bodies, domatia, and dead insects. These systems have been long-studied experimentally; generally the mutualists or the provisions are removed and the effects on herbivory and plant fitness measured. This strongly demonstrates the mutualists’ protective functions to the plant. Less often investigated is the selective environment around investment in the provisions. We first made theoretical models starting with predator functional responses and incorporating negative interactions (e.g. intraguild predation) and positive interactions (e.g. aggregations). We then tested the predictions of the model in a pair of field experiments. We titrated the provisions available to mutualists in two previously-studied systems: serpentine columbine (Aquilegia eximia) and tarweed (Hemizonia congesta). With these experiments, we asked: Do plants provide optimal amounts of provisioned insect carrion to mutualistic true bugs and spiders? Does excessive provisioning cause satiation and/or intra-guild predation (IGP) and reduce the efficacy of the defense? How well do the modelled predictions match the experimental data?
Our models based on predator functional response and predator interactions suggest the maximum benefit to the plant will be found at intermediate provision levels. We found no evidence of an intermediate provisioning level being optimal for plant fitness, predator number, or herbivore suppression, in columbine, but the finding that increasing provision levels increase IGP in tarweeds supports the model. Using this experimental data and several years of observational data, we suggest that the divergence from model predictions is due to two factors (1) year-to-year variation in mutualistic partner identity, and (2) year-to-year variation in plant provisioning traits. These experiments are among the first to test the function of a provisioning trait across a positive and negative adjustment of trait values and both demonstrate effective methodology and add to our knowledge of how protective mutualisms operate in the field.