Resemblance between animal taxa may be due to convergence rather than to recent common ancestry. Constraints on biological materials and adaptation to particular habits or habitats will produce widespread convergence. How may we distinguish the two causes of resemblance? The relationship between convergence and taxonomy is discussed, demonstrating that the choice of taxonomic method will itself determine the extent to which convergence is perceived. In particular, cladistic analysis based on parsimony will tend to minimise and thus conceal convergence: neither the resulting cladogram nor a consistency index derived from it can be used to assess the prevalence of convergence. With any taxonomic system, there can be no substitute for evaluation of the morphological characters used. Complementary use of molecular characters shows promise: we wait further understanding of constraints in genetic evolution and of the possibilities of convergence at this level also.

These general principles are illustrated with a range of examples from within and between invertebrate phyla: the phylogeny of Cnidaria and Platyhelminthes cannot be traced with certainty, but where the fossil record allows clear rooting, as for the echinoderms and in particular the echinoids, combination of morphological and molecular methods has made much progress. Sub-groups within a phylum, for example opisthobranch molluscs and the dipteran Phoridae, may show an uncontested phylogeny, and here studies have precisely identified convergence and shown that it may be the commoner cause of resemblance. Adaptation to exacting environments shown by terrestial and freshwater nemertines may also result in a predominance of convergent resemblance.

Traditional grouping of phyla breaks down on re-examination of supposedly key characters, such as segmentation, body cavities, germ layers and symmetry, each of which must have had multiple origins: nor are developmental stages (especially not larvae) a reliable guide to relationships. Demarcation of phyla may be difficult, as with arthropods, and location of phyla is even more difficult, due to their early and rapid radiation. Over-simplified definition of characters has bedevilled invertebrate classification and the use of molecular data has not yet resolved the major controversies.

The question ‘How common is convergence?’ remains unanswered and may be unanswerable. Our examples indicate that even the minimum detectable levels of convergence are often high, and we conclude that at all levels convergence has been greatly underestimated.