Different ways of sorting organisms into groups that serve the basic parts of ecosystem models can produce different results. Changing the taxonomic level at which organisms are sorted for an ecological analysis, e.g. from species to Genera or Genera to Family, changes the number of objects in the analysis and the composition of those objects. This can have both statistical and biological consequences that alter the results ecologists get. A variety of constraints often determine the taxonomic resolution of a study, not the ecologist, hence what is known in ecology as the problem of taxonomic resolution. This problem has led some ecologists to worry that their results could be more artifacts of taxonomic resolution than features of the system.

This complex of problems ultimately results in multiple representations of a system. The conventional nature of supra-specific taxonomic categories makes it difficult to justify one model (e.g., at Generic level of resolution) over another (e.g., Family level of resolution). I examine three possible solutions, (1) using species level resolution, (2) defaulting to the finest possible taxonomic resolution, and (3) using functional rather than evolutionary classes. In all cases, I show that the problem of taxonomic resolution is not directly solved. Different species concepts change the number and membership of objects in a system, exactly the problem species level analysis was intended to solve. On the other hand, I show how defaulting to the finest resolution possible does not handle the problem for a both analytical and empirical reasons. Finally, functional analyses take a different approach, directly relating group definitions to ecosystem processes. This promising approach, however, also produces problems much like those of taxonomic resolution.

My cases against the universality of these solutions suggests that the problem of taxonomic resolution needs to be re-conceptualized, particularly in order to handle the irreducible plurality of ways of sorting organisms into the parts of ecosystems. I end with a preliminary account of how scientific pluralism makes the best sense of this problem complex by embracing multiple representations as a virtue of ecological analyses.