Fig 1.
Hypothesized biological control efficiency (BCE) dynamics.
As pest density increases, the overall efficiency of natural enemies to control pest infestation will likely vary, resulting in zones with different management implications. Here we define biocontrol efficiency (BCE) as the proportion of pests that are prevented from infesting or damaging crops by a predator. In our study we expect BCE to initially increase with increasing pest density, perhaps as ants switch to the more abundant prey. This could result in a potential buffer zone where ants help to buffer the pest outbreak. Eventually, however, BCE may drop off, as ants satiate or become overwhelmed at higher pest densities. This could create a potential pest outbreak threshold, which would result in an insurance zone where farmers would need to rely on the insurance of other natural enemy species to compensate for the decreased biocontrol efficiency.
Fig 2.
Exclosure experimental design.
Shows paired-branch treatment design and calculation of biocontrol efficiency (BCE). Infestation was measured as the number of bored coffee fruits on branches 24 hrs after coffee berry borer placement. Total coffee berries per branch, leaf platforms, and berry clusters near leaf platforms are illustrated. This design was repeated on four different branch pairs within each coffee bush replicate, at different times, using four levels of CBB density. Asterisks indicate data is hypothetical and is only intended for the purpose of demonstrating the BCE calculation.
Fig 3.
Coffee berry borer (CBB) infestation.
Bars show the mean number of bored berries per branch (± SE) in the presence and absence of Azteca sericeasur at each CBB density treatment after 24 hours. Statistically significant differences in infestation between branches with and without ants are marked, where * = p < 0.001
Fig 4.
Biocontrol efficiency (BCE) curve.
Shows mean BCE (± SE) of Azteca sericeasur at each coffee berry borer (CBB) density treatment (10, 20, 40, and 80 individuals). There was no statistical difference in mean BCE across the treatments. The curve illustrates that BCE is maintained at a high level; however, trends between 0 and 10 CBB were not tested. The curve was produced using the “loess” smoothing function in the ggplot package in R.
Fig 5.
Shows the mean number of coffee berry borer (CBB) individuals attacked by Azteca sericeasur (± SE) across CBB density treatments (10, 20, 40, and 80 individuals). A simple linear regression model, shown by the line, suggests a Type I functional response (R2 = .9948, F = 761.4, p < .001).