In its original intent, the ecosystem engineers definition was "those organisms that directly or indirectly modulate the availability of resources (other than themselves) to other species, by causing physical state changes in biotic and abiotic materials" (Berkenbusch and Rowden 2003). Such a definition would seem to allow all sorts of ecological interactions to be included under the theme of ecosystem engineers (both the trellis formed by the corn plant and the nitrogen supplied by the bean plant, as much as the beetle attracted to the field, would be part of an engineered ecosystem, as would almost any other "effect" on the environment). More recently the idea has been more restrictive "distinguishing it from trophic interactions" (Berkenbush and Rowden 2003, Jones et al. 1997, Wilby et al. 2001), from keystone species, and allowing it to take on both positive and negative values (Jones et al. 1997). Furthermore, the idea of "niche construction" (Olding-Smee et al. 2003, Sterelny2005) is closely connected to the concept ofecological engineering— constructing a niche immediately conjures up images of engineering that niche.
While the intersection of these various lines of thought remains an interesting exercise that ought to be engaged, in the spirit of this volume we restrict our focus to ecological engineering as a subcomponent of species' effects on the environment, emphasizing the nontrophic effects
Ecosystem Engineers © 2007 Elsevier, Inc.
All rights reserved.
(which, formally, as we understand it, constitute cases of ecological engineering), but perhaps straying into semitrophic effects when they seem particularly interesting and engineering-like. In the case of agroecosys-tems the decision as to what falls within the limits of ecosystem engineers is sometimes difficult. For example, as discussed further in the following text, if the main effect of one part of the decomposition cycle is to alter soil structure (e.g., forming conglomerates of soil organic matter and clay particles to alter the cation exchange capacity of the soil), does this effect fall within the rubric of ecosystem engineering? We believe many such examples exist in agroecosystems and take the position that at this point in development of the concept of ecosystem engineering inclusion is probably the wisest strategy. Consequently, although we exclude effects that are obviously only trophic, some of our examples are certainly on the margin between trophic and engineering (to say nothing of niche construction).
Agroecosystems, and other managed systems, present a unique situation for focusing the issue of ecosystem engineering. Even though the principal engineer is without doubt Homo sapiens, a focus on the particular activities of that one species would be of little interest, considering contemporary literature on ecosystem engineering. Yet, because of the normative behavior of that species, there is an important dichotomy that immediately comes to light. Much as with the more general topic of biodiversity, ecosystem engineers fall in two very general categories, planned and associated (Swift et al. 1996, Vandermeer et al. 1998)—some engineers are the direct consequences of the farmer's planning, and others are indirectly associated with the organisms introduced by the farmer.
Planned engineers are the plants and animals purposefully incorporated into the system by the farmer. They clearly "engineer" the entire system since the very definition of the system—cornfield, coffee plantation, pasture—is based on their overwhelming dominance. They create the habitat where other organisms live. However, once the planned elements are in place, a host of other organisms become associated with them. Many of those associated organisms are also engineers, equivalent to the engineers normally associated with ecosystem engineering in the standard literature (e.g., Berkenbush and Rowden 2003; Jones et al. 1994, 1997).
In addition to the planned-associated framework, we find it useful to cast the problem in another well-known framework, the response-effect conceptualization of Goldberg (1990). Multidimensional agroecosystems have already been characterized as falling into this classical response-effect framework (e.g., Vandermeer 1989). Thus, for example, in an intercropping system of corn and beans, the beans purportedly modify the environment of the corn by adding to the total nitrogen pool (being able to harvest nitrogen from the air as well as soil), while the corn modifies the environment of the bean with respect to physical structure (providing a trellis on which the bean vine can grow) (Vandermeer 1984, 1989). Yet it is evident that the corn and beans are competitive with one another when intercropped, since they both use some of the same nutrients, the same source of light, and the same space.
The response-effect framework nicely captures both of these aspects, the positive or facilitative (supplying more nitrogen or making a convenient trellis) and the negative (frequently, but not always, competition). These effects range from the purely physical (the corn supplies a trellis), through the questionably trophic (a rich nitrogen environment is thought by some workers to be a physical aspect of the environment and the shading of one plant by the other is competitive, but not resource competitive), to clearly trophic (resource competition). Yet the "effect" part of the response-effect framework is frequently thought of as ecosystem engineering (contingent, of course, on one's preferred definition of the concept to begin with). Competitive production (Vandermeer 1981, 1989) can be advantageous (a phenomenon most recently referred to as complementarity, Loreau et al. 2001) or not depending on conditions of response and effect. However, facilitation can be strong or weak (never, by definition, negative) depending on conditions (Vandermeer 1984, 1989). The balance between competition and facilitation (or, more generally, negative and positive) in organisms' effect on the environment and how it is related to the fundamental idea of ecosystem engineers has been previously acknowledged (Jones et al. 1997 citing Callaway and Walker 1997).
However, the corn and beans are planned components of the system, which, for most agroecosystems, represents only a small fraction of what is interesting ecologically. For example, in the Americas the corn attracts a beetle, the corn rootworm (Diabrotica spp.), whose larvae burrow into its roots and the beans attract the same beetle, whose adults eat bean foliage. The beetle is certainly not planned by the farmer, but is part of the associated biodiversity that normally comes with the agroecosystem. The bean, then, has another effect on the corn by attracting the beetle, which eventually has a negative affect on the corn, and the corn attracts the beetle, which has a negative effect on the bean. The corn and beans thus are "apparent competitors" with the beetle acting as the agent that has the proximate effect on the environment. The beetle itself is clearly part of the "associated" biodiversity.
Although this tale of corn and beans is obviously a "just so" story (Berkenbush and Rowden 2003), it does focus attention on the question of what should be included in a review of ecosystem engineering in agroecosystems. Ecosystem engineering is part of the more general "effect" an organism has on the environment, to which that organism and others must "respond." And in the case of managed ecosystems, some of those engineers are placed in the system by the planner, the "planned engineers," while others arrive by means unrelated to human planning, the "associated engineers."
Was this article helpful?