crowded with stem group organisms. Predictably, these appear bizarre
to us because stem group organisms just are aberrant early forms of modern taxa. The Cambrian forms may well have been uniquely dis- parate. But we cannot use traditional taxonomy—counting the orders, classes, and phyla present in some biota—to capture the morphological disparity of that biota. The Cambrian example makes the problem of capturing disparity vivid. But the point it illustrates about the limits of the Linnaean system for capturing phenotype variation is general. Un- like species, higher taxa (genera, families, orders, classes, and phyla) are not objective features of the natural world. How might we do better? In his 1991 article for Paleobiology, Gould suggests an alternative approach: we should capture the extent of, and changes in, morphological disparity as changes in the occupation of a “morphospace.” A morphospace represents the disparity of a biota by defining a dimension for each morphological characteristic of the or- ganisms in that biota. If there are, say, three characteristics that matter (as there were in a famous application of this idea to shell morphology), then we would get a three-dimensional space. The actual trait values would then determine how much of that space was actually occupied by the biota under consideration. Spatial metaphors have often appealed to those thinking about biodi- versities of various kinds. Richard Dawkins’s “genetic space” (1986, 73) and Daniel Dennett’s Library of Mendel (1995) were attempts to repre- sent phenotypic and genetic possibility. But these are thought experi- ments, as George McGhee (1999) noted. They are conceptual models, probing the scope and limits of evolution, rather than attempts at modeling actual biological organisms or formulating testable empiri- cal hypotheses about particular biological systems. In this chapter, we will consider attempts to put spatial representations of morphological diversity to real empirical work. We think this (still young) tradition is impressive. In this chapter we explain why, but we also discuss the limits on these representations of phenotypic diversity.
4.2 morphological diversity
The simplest and most direct approach to phenotype disparity is simply
to measure the traits of interest and compare the results. If we were interested in the variation in length in two clades of fish, it would be perverse to count the species in each clade. We need a phylogeny to know which fish belong to each clade, but once we know that we should sample each clade, measure the fish, and analyze the raw length data.1 Such simple measures of length extent allow us to make genuine