I've been reading some recent papers on the classification of the Cypriniformes, the extremely speciose order containing minnows, carps, suckers, loaches, and their kin. There seems to be a trend towards splitting some of these big families, often by elevating particularly vexing taxa to family rank and then grouping the resultant monophyletic chunks into families as well.

Of particular interest in this case are recent papers by Slechtova et al. (2007) and Chen and Mayden (2009). The loaches, formerly classified into a variable but small number of families, are expanded to a full five families: Balitoridae, Botiidae, Cobitidae, Nemacheilidae, and Vaillantellidae, with the Chinese algae eaters (Gyrinocheilidae) and suckers (Catostomidae) as successive sister groups to the loaches. More topically for us, the cyprinoids are divided into ten families in place of the single monolithic Cyprinidae! The families are as follows:

Acheilognathidae: bitterlings
Cultridae: sharpbellies, Asian bream, Asian minnows
Cyprinidae: true carps and barbs
Gobionidae: gudgeons and relatives
Leptobarbidae: mad barbs
Leuciscidae: true minnows, dace, chubs, shiners, roaches, etc.
Psilorhynchidae: stone carps
Rasboridae: rasboras and danios
Tanichthyidae: White Cloud Mountain minnows
Tincidae: tench

The upshot is that all native North American cyprinids would be, under this classification, members of Leuciscidae. The introduced cyprinoids fall into various families: common carp and goldfish remain in Cyprinidae, tench in Tincidae, rudd in Leuciscidae. The positions of the various "Asian carps" are not addressed in this paper.

Any thoughts on these proposed changes? I'm particularly interested in the choice to treat these lineages as families rather than subfamilies; this seemingly arbitrary decision is not mentioned, let alone explained, in either paper.



Slechtova, Vendula, Jorg Bohlen, Heok Hui Tan. 2007. Families of Cobitoidea (Teleostei; Cypriniformes) as revealed from nuclear genetic data and the position of the mysterious genera Barbucca, Psilorhynchus, Serpenticobitis and Vaillantella. Molecular Phylogenetics and Evolution 44:1358–1365

Chen, Wei-Jen and Richard L. Mayden. 2009. Molecular systematics of the Cyprinoidea (Teleostei: Cypriniformes), the world’s largest clade of freshwater fishes: Further evidence from six nuclear genes. Molecular Phylogenetics and Evolution 52:544–549

What are mad barbs??

They are largeish fish, about 24" long, that look somewhat like mahseers or terete tinfoil barbs. They are also called sultan fish. There are only a few species, all confined to southeast Asia.

Very interesting, Nathan. I am not well read on the classification of the Cyprinidiformes, so I am glad someone is keeping up and posting the updates! It seems the family Cyprinidae is/was fairly broad, so this seems to be a much needed proposal. I am interested to hear what others have to say.

Blake

the closeness between loaches, suckers, and algea eaters surprises me. I can sort of see a similarity between an algea eater and a sucker but not a sucker and a loach. This is also of interest because suckers are mostly north american species with only 1 or 2 members in asia and algea eaters and loaches (if I remember right) are mainly found in asia. Were american suckers descended from the 1 or 2 species found in asia? did north america once have algea eaters and loaches that died out? Could the one sucker found on both continents (longnose sucker) be ancestral to most of the new worlds sucker diversity. Or is it all due to a non species (a mysterious common ancestor who is not to be treated like a real fish).

This also answers the question in my mind of "are barbs just a pet story term for tropical shiners?"

I also see that maybe I shouldn't insult golden shiners by calling them "pond roaches" anymore as their is an actual fish called a roach.

I'd like to see a taxonomic tree showing how close these groups are to each other and which ones branches from which.

Josh: Hopes to one day discover the worlds most destructive sucker Catastomus Catastrophus

Actually, "pond roach" isn't too far off. Golden shiners are more closely related to European leuciscids (such as roaches) than to the other North American species. They also are similar to roaches in overall appearance and ecology.

From what I have read, the whole Cypriniformes seems to be of Asian origin. The single endemic Old World sucker, Myxocyprinus asiaticus, is considered the most basal sucker. The longnose sucker is not ancestral to New World suckers but is embedded within them; it probably invaded Asia from North America across Beringia, just like horses and camels.

As far as I know, Chinese algae eaters are and have always been restricted to Asia. The loaches extend throughout Europe (at least the Cobitidae) but are not known from the New World. Only the Leuciscidae and Catostomidae ever managed to get over here, until the recent spate of human-assisted introductions.

Hello Newt, I think your post is really interesting. Let me add my opinion. In short, I think the revision of the cypriniform families in Chen and Mayden 2009 is premature, as it is based on a phylogenetic perspective that has a taxon sampling that falls far short of what is needed for this question. Also, I am suspicious of the motive. If, for example, what was defined as Cyprinidae was grossly not monophyletic (all the species share a common ancestor) then by all means revise the family-rank taxonomy. However, what was resulting in a non-monophyletic Cyprinidae was Psilorhynchus. In the phylogenies Psilorhynchus is nested in Cyprinidae. Because the "family" rank has an undefined and special status, Chen and Mayden propose retaining the monogeneric clade Psilorhynchus that contains approximately 20 species as a family and elevate the subfamilies of Cyprinidae to the special "family" status.

Taxonomy should reflect relationships. Stability is often a guide, but not a rule. In other words, a taxonomy that accurately reflects evolutionary (i.e. phylogenetic) relationships will trump stability if the previous classification does not reflect the evolutionary genealogy. In this case one has to ask what is the difference in making Psilorhynchus a subclade of Cyprinidae (e.g. Psilorhynchinae) and changing the taxonomy by elevating eight additional families? In this case, both taxonomies reflect the same evolutionary relationships. The difference is that in the first scenario Psilorhynchus loses it's special status as a family, which has absolutely no evolutionary meaning, and the taxonomy is changed for approximately 20 species and a single genus. In the other scenario, the taxonomy is changed for approximately 2,400 species and 225 genera. Remember, both taxonomies reflect the same evolutionary relationships. I would argue that having a taxomony that reflects Psilorhynchus as a subclade in Cyprinidae communicates much more about the evolutionary relationships than listing it along with 9 or 10 other named clades that end in -idae.

Fitting taxonomies to reflect evolutionary relationships can be unstable. For example, Reptilia is no longer used as a taxonomic rank name without acknowledging that Aves is nested in this group. In other words, birds are reptiles and crocodillians are reptiles that are more closely related to birds than they are to squamates (lizards and snakes) or turtles. Loosing reptilia was a hard pill to swallow, but it has gone down.

Also, the taxon sampling in the paper is not what would be considered appropriate for the changes proposed. By my count only 49 cyprinid species were sampled in this study. Chen and Mayden (2009) are proposing broad taxonomic changes for a clade that contains approximately 2400 species based on an analysis that sampled less than 50 species. For the group that most of us are interested, the North American cyprinids (Leuciscinae), only five species were sampled (two European and three NA species). I am not sure of the exact number of species, but there are more than 350 leuciscine cyprinids in North America, and a lot of species in Europe. The monophyly of each of the elevated families is hardly tested in this analysis. Also, the support for some of the nodes in the tree, particularly the ones relating the Leuciscinae with other cyprinid clades, are not marked with high bootstrap or significant Bayesian posteriors. This indicates some degree of uncertainty in the evolutionary relationships inferred from the genes they sampled.

This revision of cyprinid higher level taxonomy is unnecessary and has a cost of instability that is not commensurate with any gain in the knowledge of evolutionary relationships between the two alternative taxonomic arrangements. This is a "ranks gone wild" exercise where the undefined and special status of the family rank leads to instability, without any additional light on evolutionary relationships.

The paper is attached as a PDF.

Edited by TomNear, 24 March 2010 - 09:42 PM.

Here is Chen and Mayden (2009).

Thanks, Tom, I was hoping you would have something to say.

Your comments on familial status are close to the fuzzier thoughts I had on reading it. Psilorhynchus is a rather obscure lineage that has already been treated as a cyprinid subfamily by some authors in any case. Hardly comparable to Aves.

I could think of two other reasons that might make sense of the move: (1) the monophyly of the "cyprinoid" assemblage is questionable, so elevating known monophyletic groups allows future revisers to shuffle good families around rather than worrying about submerging or elevating well-supported monophyletic groups into or out of this single dubious family; or (2) the authors simply felt Cyprinidae was too large and unwieldy a clade to be treated as a single family. Neither was mentioned. I feared there was some obvious reason I was overlooking or too ignorant to notice.

Mayden's lab has produced some recent papers on North American leuciscines; in fact, it was seeing a talk by Angelo Bufalino that got me started reading up on this topic. However, his paper seems to take the monophyly of the Leuciscidae as given. Three outgroups (Aphyocypris chinensis, Cyprinus carpio, Gobio gobio) are used; of these, only Gobio is from the crown group that contains Leuciscidae in the Chen and Mayden paper. The terms "phoxinin" and "leuciscin" are used in the paper; I assume these refer to tribes within Leuciscinae (shouldn't they be elevated to subfamilies if Leuciscinae is elevated to family?). Their content is not made specific but the Materials and Methods suggests that all North American species are considered phoxinins and the European species sampled (Rutilus rutilus and Abramis bjoerkna) are leuciscins. The fact that their trees found Notemigonus crysoleucas grouped with the "leuciscins" and sometimes placed the Notemigonus-leuciscin clade embedded within the North American "phoxinins" impies to me that the tribes or subfamilies should be dropped or modified in some way.

I'm abysmally ignorant on the subject of phylogenetic theory and methods and am just now starting to really think seriously about the papers I'm reading, instead of just looking at the trees, so I appreciate any guidance you can give me on this stuff.


Bufalino, Angelo P. and Richard L. Mayden. 2009. Molecular phylogenetics of North American phoxinins (Actinopterygii:Cypriniformes:Leuciscidae) based on RAG1 and S7 nuclear DNA sequence data. Molecular Phylogenetics and Evolution 55:274–283

Newt, if that was meant to make the rest of us feel smart, it didnt work. |;>)

I'm abysmally ignorant on the subject of phylogenetic theory and methods ...

There are two papers in that same issue of Molecular Phylogenetics and Evolution about phoxinins, attached below.

Bufalino, Angelo P. and Richard L. Mayden. 2010. Phylogenetic relationships of North American phoxinins (Actinopterygii:Cypriniformes:Leuciscidae) as inferred from S7 nuclear DNA sequences. Molecular Phylogenetics and Evolution 55:143–152
 Bufalino&Mayden2010a.pdf   809.51KB   62 downloads

Bufalino, Angelo P. and Richard L. Mayden. 2010. Molecular phylogenetics of North American phoxinins (Actinopterygii:Cypriniformes:Leuciscidae) based on RAG1 and S7 nuclear DNA sequence data. Molecular Phylogenetics and Evolution 55:274–283
 Bufalino&Mayden2010b.pdf   1.78MB   55 downloads

Thanks Ben, I'd missed the other paper.

Thanks Ben, I'd missed the other paper.

Newt, not hard to miss as these are both essentially the same paper, a nice example of publication double dipping...