Nonetheless, in this article, we focus on what we think was the initial intent: Either herbivores overconsume their food source and are therefore “controlled,” or herbivores are consumed by their predators and are therefore “controlled.” Given this general notion that control must be from above, we are freed from the standard question and can begin exploring the nature of that control. Although advocates for a more ecologically rational agriculture persistently argue that the science of ecology should underlie the science of agriculture, we argue that the science itself is in a state of rapid evolution, especially with regard to acknowledging the important contextualization of complex systems. In the present article, we summarize much of the work we have done in the past 20 years, using the coffee agroecosystem as a model system, particularly focusing on the issue of pest control, which is to say control of the herbivore guild from above. Almost 10 years ago, we summarized what we then understood about this system , wherein we emphasized the importance of ecological complexity in the pursuit of efficient pest regulation. The current work is an update on this research program. Our emphasis then and now is on the irreducibility of the complexities involved in the species assemblage that forms the “above,” dutch bucket hydroponic perhaps reflecting the ideas of Robert Rosen: A complex system cannot be reduced to a collection of simple ones .
We describe the regulation of three coffee herbivores, a leaf herbivore, a seed predator, and a plant pathogen, by various players from above, emphasizing the remarkable complexity involved. The intersection of ecology with the burgeoning field of complex systems is evident, and this classical question of ecology is anointed with a well-deserved appreciation of irreducible complexity. The ecosystem engendering this control from above is a panoply of natural enemies, which is to say the predators and pathogens of the “pests.” They form three distinctly recognizable control-from-above subnetworks, all of which are representable by hypergraphs , all of which involve important spatial dynamics, and each of which involves some additional aspect of contemporary theory in the field of complex systems. We treat each of the networks in turn, focusing on the green coffee scale , the coffee berry borer , and the agent of the coffee leaf rust disease . As has been reported in numerous venues , coffee is an unusually important agricultural commodity. First, it is lauded in popular culture for its drugbased constituents and therefore occupies a special position in international commerce. Second, its role in creating the high-quality matrix that is necessary for biodiversity friendly landscapes is undeniable , given proper management. However, not all production styles are equivalent. Although sometimes categorized dichotomously into shade versus sun coffee, there is, in fact, a range of management styles, most of which include some kind of shade trees as part of the system . Third, it is well worth remembering the importance of trees in agriculture as part of the global agenda of containing the runaway climate change era we currently live in , especially relevant for shade coffee.
Fourth, it is the basis of livelihood for millions of small-scale farmers the world over and important economically for many poor nations. For the present article, however, the idea of ecosystem services is the normative background of our narrative, especially as regards the important service of pest regulation, or ecologically speaking, controlling herbivores from above. There are three herbivores that need to be regulated, which is to say for which a system of control from above is sought. A key feature of this control is a species of ant, Azteca sericeasur, that nests in the shade trees and is either directly or indirectly involved in the regulation of the three coffee herbivores. We begin with an essential feature of the system, the spatial distribution of that species of ant.The ant is oddly nonrandom in its spatial distribution: When you find a nest , you frequently find another nest nearby, but large sections of shaded farms have no nests at all. Quantitative sampling verifies this simple observable fact , an important feature of the regulation of all three of the herbivores. The question first arises as to where this pattern comes from. There is now substantial evidence that the spatial pattern of the ants is self-organized, which is to say that it emerges from the internal dynamics of the ant population itself, not from any underlying forces such as moisture or temperature or particular vegetation formations . The pattern is formed in a complicated fashion by a process similar to that described by Alan Turing in 1952. Turing was interested in chemicals, especially morphogens . In chemistry, a reaction is frequently assumed to be stabilized by the balance between an activation process and a repression process. However, a completely different form of chemical process occurs in a spatially constrained space, the process called diffusion. Inject a drop of black ink into a beaker of water, and the instability of the drop isolated from the water gradually turns into a beaker of grey—very stable indeed. It would be natural to think that these two stabilizing processes, activation or repression and diffusion, when combined, would also be stable. What Turing demonstrated was that if the repression force diffused at a rate greater than the activation force, a nonrandom pattern of some sort would develop.
The basic idea is that the activating chemical starts the reaction at a specific point in the space but begins its diffusion away from that point immediately. The repressive chemical is eventually produced by the reaction and cancels the effect of the activator, but, because it diffuses at a rate that is greater than that of the activating chemical, it eventually occupies a space where the activator had not yet arrived, therefore canceling the effect of the activator at that point. The results could be spots or stripes or some other more complicated form, but the point is simply the qualitative one that a nonrandom pattern would spontaneously develop when these two stabilizing forces were combined. The two stabilizing forces, reaction and diffusion, combine to form an instability; the whole system is therefore referred to as diffusive instability or sometimes Turing instability. Something very similar happens in ecological systems. Evident at a qualitative level but also explored several times mathematically , a predator–prey system distributed in space is such a system. The prey as activator and the predator as repressor is the reaction, and the migration or dispersal of predator and prey is the diffusion. However, an important difference between the ecological and chemical metaphor is that, in the predator–prey situation, it is frequently the case that, at a very local level, the predator and prey form an unstable relationship , whereas adding diffusion to the mix may result in stabilizing the system. The classic experiments of Huffaker illustrated this point many years ago , with two species of mites, one a forager on the surface of oranges, the other a predator on that forager. Huffaker devised a spatial system in which a part of the surface of an orange provided a substrate for the two mites, and the oranges could be arranged in an array to represent a spatial matrix. As frequently cited in ecology textbooks, when predator and prey were isolated on a single orange, the predator would inevitably overeat, first driving the prey locally extinct and then dying of starvation. In contrast, dutch buckets system if a spatial pattern of a number of oranges was presented to the mites, with the possibility of dispersal from one orange to another, a seeming stabilization of the predator–prey system over the whole space occurred. Although the take-home message of Huffaker’s experiment is that environmental heterogeneity can stabilize an inherently unstable system, what is less frequently discussed is Huffaker’s observations of the resulting spatial distribution of the mites. In figure 1, we display one cycle of the three-cycle predator–prey system he observed in the spatially distributed system, along with the spatial pattern of the foraging mites. Note the emergence and disappearance of clusters. In the spatial pattern in figure 1a, one can visualize that a single cluster seems to be forming. Figure 1b shows that initial cluster growing and the possible formation of a second cluster. In figure 1c, the two clusters have grown considerably.
In figure 1d, the largest cluster seems to have split into two smaller clusters, whereas the other cluster seems to have expanded slightly. In figure e, the two smaller clusters that came from the first cluster seem to be disappearing and the second cluster seems to have split in two. In figure f, two of the clusters remain. Although this interpretation of Huffaker’s data is qualitative , it serves to illustrate the basic process of cluster dynamics that is expected from a spatial system emerging from the Turing mechanism . It is furthermore worth noting that, in the present example, as well as in Turing dynamics in general, the spatial pattern occurs even though the physical space in which the processes occur is uniform or, at least, distinct from the patterns formed. In the coffee agroecosystem, we have argued that the Azteca nest pattern is formed in a similar fashion. The repression agent is thought to be a parasitic fly, Pseudacteon spp., in the family Phoridae . The fly oviposits on the back of the ant’s head, and its larva penetrates the head capsule, there developing to the point that the head of the ant falls off of the body and a new fly emerges to mate and repeat the cycle. As a local population of ant nests builds up from single queens taking a part of the colony to a new shade tree , spatial clusters are formed. As the clusters become larger, they are targets for the phorid flies, either because the flies are attracted from far away or they build up local populations within the area of the nest cluster. Either way, the flies act as the repressor in the system. The result is a patchy distribution of ants. In figure 2, we show the nest distribution for 3 of the 12 years of the study and enlarge one section of the data to illustrate how clusters form and dissipate over time. We note that the general qualitative dynamics of the formation and dissolution of patches, such as the Huffaker mite example above, resonates quite well with the basic expectations of the Turing process. We further note that the dynamic details associated with the cluster scalings that we discussed earlier seem to repeat themselves in subsequent years, following a special pattern of self-organization .These and other early ecologists certainly recognized patchiness in the environmental scaffolding on which ecological communities sit. Historically, it is likely the case that the massive vegetative surveys that energized the development of ecology in the early twentieth century tacitly assumed that ubiquitous edaphic and climatologically produced patches determined regular combinations of plant species. These physical patches clearly made intelligible many particulars of plant distributions, and to the present day, there is a tendency to ask what the underlying edaphic factors are that prejudice one species as opposed to another or one species complex over another. However, the prescient observations of Tansley and Chipp bring to the table the possibility that the biological interactions themselves create the heterogeneity, what we have referred to as a pilot pattern , similar to the important idea of niche construction but with a key interspecific and spatial element: Species group X creates the spatially explicit conditions for species group Y to exist. This framing is convenient for the present summary. The clustered distribution of the Azteca ants indeed does provide an essential environmental heterogeneity on which the other systems operate. However, we add to this narrative the idea that the species group providing the pilot pattern must operate in a distinct time frame from the other systems operative within it, a restriction that seems to be relevant at least for the systems we report in the present article. The pilot pattern formed by the Azteca–phorid complex must be experienced by the rest of the relevant organisms as parallel to the old idea of edaphic or climatological factors—relatively permanent compared with the dynamical changes of the systems living within the pattern .