INFOWEB 13 : Rivulus revised and proposed split, still temporary

 

From Jean H. Huber
Private address: 7 Bd Flandrin, 75116 Paris, France
M.N.H.N., Ichthyology, PARIS, France.

 

Paris, December 18. 2011 [updated on December 27. 2012, July 1. 2015, November 30. 2019… Rivulus, Plesiolebias].

 

Dear Colleague, dear Aquarist!

Wilson Costa has published an important paper on the phylogeny of the South and Central American genus. It is a milestone in his series of research papers on that genus (the last one being in 2008 with Amazonian Rivulus), but it is not the end of the line for this morphologically very homogeneous group of fishes. The study is based on the morphology of 15 species included in Rivulus (in the old sense, but without Kryptolebias). The method used is a maximum parsimony analysis of a combined set of morphological data (by Costa himself) and mitochondrial data available in the literature (from outside researchers, notably Murphy et al. published in 1999), for 33 rivuline taxa (15 Rivulus species and 18 non-Rivulus species (in the related groups) plus 3 outgroups.

The resulting matrix induces Costa to propose a new generic classification on rivuline relationships, with the following proposed taxonomic changes : Rivulus, restricted to two species endemic to Cuba, is hypothesized to be the most basal rivuline lineage, distinguished from all other non-annual rivulines by having all hypurals fused into a single plate, neural prezygapophysis of caudal vertebrae rudimentary, fourth ceratobranchial teeth absent, about 50% of the anterior portion of the caudal fin covered by scales, four neuromasts on the anterior supraorbital series, and a black round spot with white margin on the dorso-posterior portion of the caudal peduncle in females; Anablepsoides, Atlantirivulus, Laimosemion, Melanorivulus and Cynodonichthys, previously classified as subgenera of Rivulus, are considered as valid genera; Laimosemion, including 24 species from northern South America, constitutes the sister group to a clade comprising Melanorivulus, Cynodonichthys, Anablepsoides, Atlantirivulus, and all annual rivuline genera, which is supported by a well-developed dorsal process of the urohyal and an expanded lateral articular facet of the first hypobranchial; Melanorivulus, comprising 34 species from central and northeastern South America, Cynodonichthys with 27 species from Central America and Trans-Andean South America, Anablepsoides with 42 species from northern and northeastern South America and Smaller Antilles, and Atlantirivulus, with 11 species endemic to the eastern Brazilian coastal plains are diagnosed by combinations of morphological characters, including osteology, cephalic laterosensory system and colour patterns.

From an analytical and opportunistic point of view this additional work in the series of researches by Wilson Costa on Rivulus is welcome ; the first diagnosis to separate Rivulus s.s. from the other groups is important if the characters are indeed clear-cut (and at first sight it may not be so clear for the character of all hypurals fused into a single plate or for the distinction of the supracaudal ocellus, among others) ; however -and this is by no means criticism- the separation between Rivulus s.s. and the other subgenera may be one thing, but the separation between each of the other subgenera, Anablepsoides, Atlantirivulus, Laimosemion, Melanorivulus and Cynodonichthys, that Costa considers as valid, may be another story (without mentioning other described subgenera, such as Owiyeye, Vomerivulus, Oditichthys, Benirivulus that he synonymizes… but in another subsequent paper he already says it might be temporary, at least of Oditichthys).

 

What are the impacts of Costa's reshuffling of the genus Rivulus? 

Here is a table that summerizes the changes (they are not followed in Killi-Data presently, see conclusion and addendum).

 

Will Killi-Data online follow the impacts of Costa's reshuffling of the genus Rivulus? 

 

As discussed in details in a preceding newsletter INFOWEB 11, Killi-Data will not automatically align to generic moves derived from research, notably when diagnoses remain weak.
In the present case the matrix appears preliminary with only 15 species studied (over a total, today, of nearly 150 hypothesized as valid in Rivulus in the traditional sense, i.e. about 10%… but the concept study to merge morphological and molecular data pushes to that major limit) and the diagnoses of the proposed upgraded genera are still not definitive (with several characters being not clear-cut). Basically in the present paper, Costa splits the genus Rivulus into 6 new genera (Rivulus, Anablepsoides, Atlantirivulus, Cynodonichthys, Laimosemion, Melanorivulus, not to mention Kryptolebias, Millerichthys, Prorivulus, already separated in "old" times), but he includes in his matrix very few species for each group (except for Atlantirivulus, Melanorivulus, 2 groups he knows best) and this might appear as somewhat insufficient.
A final issue is that the herein refusal to follow may be seen as a personal matter (the present author having published the reference book on Rivulus in 1992, still on sale)… let's face frankly that suspecting issue : (1) Costa himself in a different paper published at about the same time mentions that he considers his synonymisation of Oditichthys into Anablepsoides in this paper may be premature (and that Oditichthys could be a valid unit) ; (2) the present author had, parallely to Costa, another publication in the pipeline with distinct results from Costa's and had to withdraw it (obviously in order to take into account and evaluate Costa's new characters) ; (3) even the use of the 2 diferent techniques (osteology and molecular biology) may be at odd (not to speak about fusing the 2, like in Costa's paper) : for example, Rivulus mahdiaensis has been described as related to lyricauda based on molecular data (then in Laimosemion), but Costa & Souza in 2009 published a short paper stating it is not the case based osteology and that the species should be placed in Owiyeye (another species group including rectocaudatus and tecminae !).
In total, a very very difficult group of fishes to study for the 10 years in future (by molecularists, by morphologists, by collectors -knowing that most of the Amazon basin is virgin of collections- simply to get additional and sufficient data, and also to have at hands competitive views and objective results.
This moves with numerous names changes appears then more preliminary (tree-naming driven and splitting-driven) with little added value to the nomenclature which promotes stability and reason.
This move will then not be followed by Killi-Data, until strong and solid diagnoses are proposed which show that the split situation reflects better the philogeny than previously.
In practical terms, the names prevailing in Killi-Data until further evidence are the previous ones, with only 1 genus Rivulus and and several subgenera.

 

Addendum dated December 27. 2012 : Huber publishes a review of Rivulinae + Kryptolebiinae, including all lineages of Rivulus in the large sense, using computerized systematics [source : Killi-Data Series 2012, 9-25, 3 figs., 2 tabs. (December 21.). Reappraisal of the Phylogeny of Rivulus and its Allied focused on External Characters]

In that study, the author starts by mentioning that his first manuscript had been withdrawn in Autumn 2011 because of Costa's unexpected publication and in the presently published study all known to date phylogenetic lineages encompassing the genus Rivulus in the large sense and its relatives, 44 in total, are benchmarked against a very long list of external characters involving global and detailed morphology of each, plus some characters related to biology, behavior and ecology, plus a few characters concerning the vertebrae counts and their structure. The parsimony analysis for these 144 characters results in 3 trees only, very much in line with the global stability of major lineages and the consensus tree produces bootstrap values that are much higher than previously released studies with morphological and osteological characters (while a re-running of the matrix by Costa without the molecular characters produces a very unstable set of trees, inferring that molecular data overwheighted Costa's global matrix). In the resulting consensus tree by Huber, outside Kryptolebias already separated in 2004, the genus Rivulus in the large sense (even including Prorivulus) is shown as a monophyletic assemblage like in the molecular study by Murphy, Thomerson & Collier (1999), with annual and non-annual species separated, but unlike Costa (2011) who used both morpho-osteological and molecular combined characters. In conclusion, Huber suggests that the split of the genus Rivulus in the large sense should be considered as premature and much more research is needed to ascertain (if necessary) that split, notably in the lineages inhabiting the Amazon basin for which large uncertainties could be detected probably linked to our very poor knowledge.

 

[addendum (July 1. 2015), : a new work with molecular data processed through the maximum likelihood method by Andrew Furness {ref. Furness, A.I. 2015. The Evolution of an annual Life Cycle in killifish: Adaptation to ephemeral aquatic environments through embryonic Diapause. Biological Reviews of the Cambridge Philosophical Society (Biol. Rev.)} with taxonomic notes on Rivulidae quotes: <Costa's revision of Rivulus and elevation of subgenera to generic level status has proven contentious. Based upon the larger molecular data set analyzed here, Laimosemion, Cynodonichthys, Atlantirivulus, Melanorivulus, and Anablepsoides represent well-supported monophyletic clades. However, the relationship between these clades and the placement of the Plesiolebi(at)ini clade of annual killifish (represented in the tree by the species Maratecoara formosa, Maratecoara lacortei, Pituna poranga, Papiliolebias bitteri, Plesiolebias aruana) remains largely unresolved. Here, the Plesiolebi(at)ini clade is sister to Atlantirivulus but this relationship has very weak bootstrap support and in general this clade is unstable with respect to position, as in the original phylogenetic analyses including these species. Given additional molecular data it is possible the Plesiolebi(at)ini clade could shift position and become sister to the well-supported annual clade composed of the genera (Austrofundulus, Rachovia, Micromoema, Gnatholebias, Llanolebias, Renova, Terranatos, Neofundulus, Pterolebias, Trigonectes, Aphyolebias, Moema)< (end quote) ; this is opposite to Costa's original tree ; hence it is more parsimonious not to consider Laimosemion, Cynodonichthys, Atlantirivulus, Melanorivulus, and Anablepsoides as valid genera (in Killi-Data, they are conservatively kept as valid subgenera, only)]. 

 

[addendum (November 30. 2019), : in the above reference of 2011 [Costa, W.J.E.M. 2011. Phylogenetic position and taxonomic status of Anablepsoides, Atlantirivulus, Cynodonichthys, Laimosemion and Melanorivulus (Cyprinodontiformes: Rivulidae). Ichthyological Exploration of Freshwaters, 22 (3): 233-249, 1 fig., 1 tab.], there is a minor nomenclatural error that Costa (and reviewers) did not notice : while most previous subgenera of Rivulus are of masculine gender like genus Rivulus, then the species moved in upleveled subgenera do not change their names, it is not the case of subgenus Laimosemion that is of neuter gender, then in that case some species names must change endings ; for examples, Rivulus xiphidius becomes Laimosemion xiphidium (not xiphidius) and the same for cladophorus, dibaphus, frenatus, rectocaudatus, strigatus ending with -um, and for altivelis, mahdiaensis ending with -e (not -is). 

 

In total a very important and sensitive newsletter !
Hopefully a boost to our community and a spur to speed up knowledge progress on our (beloved) fishes !

 

Take care and enjoy the scientific or aquaristic complexity of killifish !

Do not hesitate to ask questions for future Newsletters.

Visit frequently the website www.killi-data.org!

Thank you for your support over the years.

With my kindest regards.

Jean

Literature cited:
Costa, W.J.E.M. 2011. Phylogenetic position and taxonomic status of Anablepsoides, Atlantirivulus, Cynodonichthys, Laimosemion and Melanorivulus (Cyprinodontiformes: Rivulidae). Ichthyol. Explor. Freshwaters, 22 (3): 233-249, 1 fig., 1 tab.

---

I am interested in reading other Newsletters, click INFOWEB.