Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
, 7 (9), e45508

Ecological Complexity in a Coffee Agroecosystem: Spatial Heterogeneity, Population Persistence and Biological Control


Ecological Complexity in a Coffee Agroecosystem: Spatial Heterogeneity, Population Persistence and Biological Control

Heidi Liere et al. PLoS One.


Background: Spatial heterogeneity is essential for the persistence of many inherently unstable systems such as predator-prey and parasitoid-host interactions. Since biological interactions themselves can create heterogeneity in space, the heterogeneity necessary for the persistence of an unstable system could be the result of local interactions involving elements of the unstable system itself.

Methodology/principal findings: Here we report on a predatory ladybird beetle whose natural history suggests that the beetle requires the patchy distribution of the mutualism between its prey, the green coffee scale, and the arboreal ant, Azteca instabilis. Based on known ecological interactions and the natural history of the system, we constructed a spatially-explicit model and showed that the clustered spatial pattern of ant nests facilitates the persistence of the beetle populations. Furthermore, we show that the dynamics of the beetle consuming the scale insects can cause the clustered distribution of the mutualistic ants in the first place.

Conclusions/significance: From a theoretical point of view, our model represents a novel situation in which a predator indirectly causes a spatial pattern of an organism other than its prey, and in doing so facilitates its own persistence. From a practical point of view, it is noteworthy that one of the elements in the system is a persistent pest of coffee, an important world commodity. This pest, we argue, is kept within limits of control through a complex web of ecological interactions that involves the emergent spatial pattern.

Conflict of interest statement

Competing Interests: The authors have declared that no competing interests exist.


Figure 1
Figure 1. Field vs. simulation snapshot of the distribution of ant nests.
A) From the 120×90 lattice from the theoretical model. B) From a 45 ha plot field survey conducted in the summer of 2006, qualitatively similar to the model results (for methods see Vandermeer et al. 2008). We use this size lattice as in their model, since the study system in nature is approximately that size and contains about 11,000 potential ant-nesting sites (shade trees).
Figure 2
Figure 2. Log of cumulative frequency vs. log cluster sizes of ant nests (for easier interpretation, the axes have arithmetic scales).
The open circles represent field data from 2004 (rainy and dry seasons) and the dashed line is the combined power law fit to both field surveys. Cluster size is based on a minimum distance of 20 m between nests, when these are judged to belong to the same cluster (see Vandermeer et al. 2008). The blue lines represent the fitted power law lines to each of 200 runs for the Vandermeer et al. 2008 model where a parasitic phorid fly is the cause of density dependent ant nest mortality. The red lines represent the power law fits to each of 200 model runs, where the coccinellid beetle is the indirect cause of ant nest mortality.
Figure 3
Figure 3. Ripley’s K index versus size of sampling circle around each tree.
The blue color represents the results from the phorid model (Vandermeer et al. 2008) (the dashed line is the median and the shaded area shows the 95% confidence limits calculated from 200 realizations of the model); the slate blue color represents the results from the beetle model (200 realizations with the GA-fitted parameters); the dark gray shaded region represents the 95% confidence limits for 200 random distributions generated using the 2004 rainy season field data. The solid black lines are the results from rainy season and dry season field surveys performed in 2004. Deviations above the zero line indicate the data are more clustered than expected by random.
Figure 4
Figure 4. Simulation time series of the four populations (ants, scales, larvae beetles and adult beetles).
Population sizes on the y-axis are on a log+1 scale. A) Simulation in which the ant population goes extinct (ant mortality parameters: do = 0.85; d1 = 0). B) Simulation in which the ant population occupies the whole lattice (ant mortality parameters: do = 0.2; d1 = 0.3). C) Simulation in which the ant population is similar to the one observed in the field, the beetles however, are not able to ‘perceive’ the ant clusters (local migration parameter: 0; global migration parameter = 1). D) Simulation in which the ant spatial pattern emerges from the model interactions and is similar to the one found in the field.
Figure 5
Figure 5. Diagrammatic representation of the proposed mechanisms allowing beetle population persistence (A.) and the formation of clusters of ant nests (B.).
Arrowheads indicate positive effects, closed circles indicate negative effects.

Similar articles

See all similar articles

Cited by 5 PubMed Central articles


    1. Hassell MP, Comins HN, May RM (1991) Spatial structure and chaos in insect population dynamics. Nature 353: 255–258.
    1. Rohani P, Godfray HCJ, Hassell MP (1994) Aggregation and the dynamics of host-parasitoid systems: a discrete-generation model with within-generation redistribution. American Naturalist 144: 491–509.
    1. Rohani P, Lewis TJ, Grünbaum D, Ruxton GD (1997) Spatial self-organization in ecology: pretty patterns or robust reality? Trends in Ecology & Evolution 12: 70–74. - PubMed
    1. Comins H, Hassell M, May R (1992) The spatial dynamics of host–parasitoid systems. Journal of Animal Ecology 61: 735–748.
    1. Rohani P, Miramontes O (1995) Host-parasitoid metapopulations: the consequences of parasitoid aggregation on spatial dynamics and searching efficiency. Proceedings: Biological Sciences 260: 335–342.

Publication types

Grant support

U.S. National Science Foundation, DEB 0349388 to JV and Ivette Perfecto; Graham Environmental Sustainability Institute fellowship, University of Michigan, to HL and DJ; and Helen Olsen Brower Fellowship, Department of Ecology and Evolutionary Biology, University of Michigan, to HL. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.