Nearest-Neighbor Analysis of Trilobite Morphospace

Abstract

-Previous observations indicate that morphologic distinctness of suprageneric taxa (superfamilies, suborders, and orders) of trilobites increased during their Cambro-Ordovician diversification. Analysis of nearest-neighbor distances (NNDs) among trilobites in morphospace is used to test the robustness of this pattern while avoiding the overprint of higher-level taxonomy. The properties of the uniform distribution are used to determine the bounds of the morphospace, and computer simulation of random points is used to circumvent edge effects in the analysis of NNDs. Observed NNDs became significantly smaller relative to expected distances, supporting the hypothesis that morphospace became more discontinuously occupied from the Cambrian to the Ordovician. However, absolute NNDs (not considered relative to expected NNDs) actually increased. Therefore, the increase in clustering intensity reflects the expansion of morphospace that accompanied the diversification, because the expected NND is proportional to the morphologic range. The increases in clustering and in absolute NND hold even if Cambrian genera are treated as equivalent to Ordovician species. [Morphometrics; spatial analysis; trilobites; diversification; computer simulation; taxon-free analysis; early Paleozoic.] Morphology provides the fundamental information upon which evolutionary analysis of the fossil record is based, but macroevolutionary questions are often discussed primarily in taxonomic terms (e.g., Valentine, 1969; Yochelson, 1979; Runnegar, 1987). Taxonomy imposes structure on macroevolutionary analysis not only in cases where conclusions are based on taxon counting. For example, it has been observed that suprageneric taxa of trilobites are morphologically more distinct from each other in the Ordovician than in the Cambrian (Whittington, 1966; Foote, 1989b, in press). The validity of this conclusion rests implicitly on the assumptions that forms are accurately assigned to higher taxa and that these higher taxa are comparable entities. If such assumptions are not met, the reliability of macroevolutionary studies based on higher taxa is limited. (In discussing higher taxa in macroevolutionary analysis of the fossil record, one commonly implies a level between that of the phylum and that of the family. By higher taxa of trilobites I mean superfamilies, suborders, and orders. Discussion of the rationale behind the use of these categories can be found elsewhere [Foote, in press], but is not immediately relevant to the present discussion.) One obvious approach to circumventing problems with the use of higher taxa is to improve the taxonomic structure so that higher taxa are increasingly accurately and consistently defined, by whatever criteria one deems relevant. It is clearly desirable to work within an improved taxonomic framework, but inaccuracies and inconsistencies will always remain. More importantly, even an improved taxonomy still imposes structure on macroevolutionary