Foraging theory for individual-based models of pest control by birds: Tests in a model of coffee borer suppression by warblers in Jamaica

Background/Question/Methods

The coffee berry borer (CBB, a beetle) is the only globally important pest on coffee and has few natural enemies. Field studies in Jamaica confirm that overwintering North American warblers consume economically significant quantities of CBB. However, CBB are a very small part of the bird diet; birds depend mainly on arthropod prey in forest and trees in or near coffee farms. This observation raises important and complex questions: how much forest/tree habitat is necessary to support enough birds to suppress CBB outbreaks? Would too much arthropod food reduce consumption of CBB? Such questions are too complex to answer with field studies alone so we developed an individual-based model in which birds decide, from minute to minute, where to forage and whether to eat arthropods or CBB. While there is a long history of foraging theory focused on birds, it was not clear how useful traditional theory is for a relatively realistic simulation model. We therefore identified four alternative theories and tested them by how well they caused the model to reproduce nine patterns, observed on Jamaican coffee farms at the individual and population levels, that emerge from foraging behavior.

Results/Conclusions

Our first result was that much traditional foraging theory is not useful simply because it does not address the question of where birds forage (e.g., the “marginal value” theory addresses only when a bird leaves its current patch or food type, not where it goes next). A “null” theory, that birds select randomly from among their current and surrounding habitat cells, reproduced four of nine observed patterns, indicating that those patterns do not provide strong tests of theory. A variation of traditional theory—assuming birds move randomly when food in their current cell is depleted below a threshold—reproduced five patterns. However, that theory produced the opposite of one important pattern, that bird densities increase in patches experiencing irruptions of high arthropod availability. Two theories that assume birds pick the cell providing highest food intake, over either short or long distances, were more successful. The experiments showed that useful theory must assume birds have some ability to sense food availability, over short, not long, distances. The resulting model indicates that landscape-wide availability of forest/tree habitat strongly affects bird survival and, less strongly, pest-control services. The spatial arrangement of habitat types is not predicted to have strong effects.