The primary focus in ecological networks is on diffuse coevolution. The first issue is then to clarify what exactly is meant by that term. Strauss et al. (2005) have come up with three criteria, which have been modified from those developed first by Hougen-Eitzman and Rausher (1994), to determine whether coevolution is diffuse. First, the traits that are under selection by more than one species must be genetically correlated. This criterion is clearly fulfilled in the extreme example where multiple species exert a selective force on the same trait. Second, the addition of species to the interaction must change either the strength or direction of selection. In other words, the strength and direction of selection cannot be determined by a single species (tight coevolution). The third criterion proposed by Strauss et al. (2005) is that the presence of the other species must alter the G-matrix. The second species must alter the expression of genetic variance (focal trait) or covariance (between focal and other traits) essentially creating a genotype X genotype X environment interaction (Thompson 2005).
An important null model in the arena of diffuse coevolution is an extension of Hubbell’s Unified Neutral Theory and the hypothesis of ecological equivalence (Hubbell 2006). Hubbell (2006) notes that when species rich communities are characterized by low dispersal rates they should tend to converge on similar life history strategies based on the optimum for the most frequently encountered conditions. In a species rich community any given individual is more likely to co-occur with an unrelated individual (of another species). This results in diffuse selective forces causing selective pressure in multiple directions. The net result of this diffused selective regime is a lack of directional selective pressure on any given species by its neighbors. Thus the result is that the organisms should adapt to the most frequently encountered environment, resulting in the highest individual fitness. Since these organisms are dispersal limited, the local environment should be more or less the same for each individual and therefore the fitness optima becomes the same for each species.
Hubbell’s (2006) model works well for organisms that are extremely dispersal limited, such as trees on Barro Colorado Island. Nonetheless this model does not work as well for organisms whose dispersal capabilities are not severely limited. It is difficult to imagine the scenario established by Hubbell to apply to species that are constantly moving through a changing environment. Species experience heterogeneous environments, which generate a mosaic of selective forces (Thompson 2005). Differences in selective forces across a landscape can be caused by local environmental conditions, or the presence (or absence) of certain species. The variable presence or absence of certain species in a local area alters composition of the ecological network of interacting species. These networks can be viewed as diffuse multispecific interactions among species, as in Hubbell’s (2006) ecological equivalency, or they can be viewed as coevolved structures that are shaped in predictable ways (Thompson 2005). Research on networks has shown that there are universal properties common to all networks (Jordano et al. 2003), and many of these properties can develop in ecological networks, such as those of mutualisms through coevolution in multispecific interactions.