PS 78-85
Protecting coffee: Exogenous and endogenous factors behind spatial patterns of biological control processes in an agroecosystem

Friday, August 9, 2013
Exhibit Hall B, Minneapolis Convention Center
Kevin Li, School of Natural Resources and Environment, University of Michigan, Göttingen, MI
Ivette Perfecto, School for Environment and Sustainability, University of Michigan, Ann Arbor, MI
John Vandermeer, Ecology and Evolutionary Biology, University of Michigan, Ann Arbor, MI
Background/Question/Methods

Ecological spatial patterns are commonly thought to reflect underlying environmental patterns (exogenously determined).  Alternatively, they may be organized through local interactions that do not reflect an underlying pattern (endogenously determined).   In a Mexican coffee agroecosystem, the status of the agroforestry matrix as well as an arboreal ant species was censused in a 45-ha plot from 2004 to 2012.  The ant, Azteca instabilis, plays a role in maintaining and regulating important biological control agents and pests in the coffee ecosystem.   Ant nest development and survival may be dependent on factors that are external to the Azteca ecosystem, such as host tree population or site topography (exogenous factors); or they may be influenced by factors which change dynamically over time through feedback from a network of ecological relationships (endogenous factors).  Previous research suggests that Azteca population is strongly linked with the latter through density-dependent negative relationships.  During the census period, farm intensification eliminated 30% of trees, which preceded an increase in ant nest population.  Survival analysis was used to model ant nest colonization and survival within the 45-ha plot to identify significant exogenous and endogenous covariates that may explain nest distribution and the changes in population following intensification.

Results/Conclusions

Survival analysis modeling of Azteca nest formation and collapse indicate that site topography variables have no significance in predicting either event.  Instead, nest formation and collapse are both predicted by the existent surrounding nest population in non-linear relationships, where intermediate nest densities correlate with the highest probability of either event.  This result gives support to a density-dependent relationship between Azteca and endogenous covariates, but the interaction is more complex than simply higher nest densities incurring higher survival cost.  Some exogenous variables appear to play a weak role in determining nest distribution.  Larger trees hosting ant nests are less likely to see their nests collapse.  Applying these relationships as parameters to a spatially-explicit time series simulation based on: 1) the layout of a nearby intensified coffee agroforestry plantation and 2) a randomly-generated forest will validate whether this set of relationships can recreate similar distribution patterns in novel landscapes.   These results have implications on coffee agroforestry management, demonstrating that maintaining certain tree densities promotes a more stable Azteca ecosystem and subsequently the pest-control ecosystem service that it provides to coffee agroecosystems.