Volvocine taxonomy is in a sorry state. Most nominal genera, and some nominal species, are almost certainly polyphyletic. More than once, I’ve been asked during a talk, “Why is Volvox scattered all over the tree?”

Fig. 2A from Herron et al. 2010. The traits characteristic of the genus Volvox—asexual forms with >500 cells, only a few of which are reproductive, and oogamy in sexual reproduction—have arisen at least three times independently: once in the section Volvox (represented by V. globator, V. barberi, and V. rousseletii), once in V. gigas, and once or possibly twice in the remaining Volvox species. Branch shading indicates maximum-parsimony reconstruction (white = absent, black = present, dashed = ambiguous). Pie charts indicate Bayesian posterior probabilities at selected nodes. Numbers to the left of cladograms indicate log-Bayes factors at selected nodes: positive = support for trait presence, negative = support for trait absence. Interpretation of log-Bayes factors is based on Kass and Raftery’s (1995) modification of Jeffreys (1961, Theory of probability. 3rd edn. Oxford Univ. Press, Oxford, UK.): 0 to 2, barely worth mentioning; 2 to 6, positive; 6 to 10, strong; >10, very strong. Boldface numbers following species names indicate Volvox developmental programs following Desnitski (1995).

Nor is this an easy problem to fix*. Unlike, say, mammals or mollusks, volvocine algae are difficult to preserve, and we don’t have museum collections of skins or shells. Volvocine algae are well-represented in culture collections, notably UTEX and NIES, but there are problems here as well. Unlike many microbes, we have no means to cryopreserve most species, and even for the few that we can cryopreserve, the process requires liquid nitrogen, making it relatively difficult and expensive. Strains are generally propagated asexually and serially transferred on the order of days or weeks. This means that they undergo frequent bottlenecks, making them subject to genetic drift and Muller’s ratchet. Sexual competence is routinely lost over a period of years or decades under these conditions. Worse, strains are often lost due to mishaps, spontaneous induction of the sexual cycle, or without explanation.

Most strains collected before Richard Starr founded what is now the UTEX Culture Collection in 1953, and many collected since then, are lost forever. This makes taxonomic revision problematic. For example, the nominal genus Volvox probably has at least three independent origins: the clade that includes V. globator, V. gigas, and the clade that includes V. carteri (V. aureus may represent a fourth origin or may share a Volvox-like ancestor with V. carteri). Ideally, we would split Volvox up into three (or four) genera, each of which is monophyletic. The rules of taxonomic nomenclature require that the first species described in a genus must retain that genus name. For Volvox, this is V. globator, first described by Linnaeus in 1758, so presumably the clade including V. globator (sometimes referred to as Euvolvox or section Volvox) should remain Volvox. But there’s a problem: how do we know that what we call V. globator today is actually the same species as what Linnaeus described? We know that convergence is common in this group, and we know that strains that are described as one species based on morphology often turn out to belong to two or more (and some strains described as separate species are surely synonymous). We don’t have a type specimen to go back to (or even a type culture), so there’s no chance of doing any molecular diagnostics, testing for sexual compatibility, etc. And this or similar problems will exist for all but the recently described species.

*This difficulty was first explained to me by Hisayoshi Nozaki; any errors in the explanation are mine.