Extinction space--a method for the quantification and classification of changes in morphospace across extinction boundaries.

Three main modes of extinction are responsible for reductions in morphological disparity: (1) random (caused by a nonselective extinction event); (2) marginal (a symmetric, selective extinction event trimming the margin of morphospace); and (3) lateral (an asymmetric, selective extinction event eliminating one side of the morphospace). These three types of extinction event can be distinguished from one another by comparing changes in three measures of morphospace occupation: (1) the sum of range along the main axes; (2) the sum of variance; and (3) the position of the centroid. Computer simulations of various extinction events demonstrate that the pre-extinction distribution of taxa (random or normal) in the morphospace has little influence on the quantification of disparity changes, whereas the modes of the extinction events play the major role. Together, the three disparity metrics define an "extinction-space" in which different extinction events can be directly compared with one another. Application of this method to selected extinction events (Frasnian-Famennian, Devonian-Carboniferous, and Permian-Triassic) of the Ammonoidea demonstrate the similarity of the Devonian events (selective extinctions) but the striking difference from the end-Permian event (nonselective extinction). These events differ in their mode of extinction despite decreases in taxonomic diversity of similar magnitude.