Proposal (190) to South American Classification Committee


Merge Oceanodroma into Hydrobates Boie 1822



The current SACC Checklist recognises seven species within the genus Oceanodroma Reichenbach 1853.



Penhallurick and Wink carried out an analysis using Bayesian Inference (Mr Bayes3.1) of 95 taxa within the Procellariiformes.


We used PAUP (Swofford 1998) and ModelTest (Posada and Crandall 1998) to analyse the substitution patterns in a cytochrome-b data set against 56 different models of DNA evolution. We used the Akaike Information Criterion (AIC). The model selected was GTR+I+G.


We used Bayesian inference and Markov Chain Monte Carlo (mcmc) with Metropolis coupling for estimating phylogenetic hypotheses from DNA data. The model of nucleotide substitution used was the GTR model, following the results of ModelTest mentioned above. Thus Nst was set to 6 and Rates to Invgamma. In relation to prior distributions, we used a flat Dirichlet for both Revmatpr and Statefreqpr. Ronquist et al. (2005) stated that use of the default settings is appropriate if we want to estimate these parameters from the data assuming no prior knowledge about their values. After three million generations, the average standard deviation of split frequencies was 0.008673. The mean of the 13 Potential Scale Reduction Factor (PSRF) figures was 1.002 (SD 0.003). We give, as Figure 1, a plot of the generation versus the log probability of the data (the log likelihood values). The plot approximates "white noise". All the data just cited support the validity of the analysis.


Figure 1 Plot of generation versus the log probability of the data (the log likelihood values) for the MrBayes3.1 analysis of the data


Figure 2 Phylogeny of all Procellariiformes

Bayesian Inference Tree. This represents a 50% majority rule consensus tree. Posterior Probability values are indicated, usually to the right of the nodes, but occasionally, for reasons of space, to the left.


Figure 3 displays that section of Figure 2 containing the storm-petrels


Note: All three figures are available at the author's website.

Click on Enter, and then on "SACC Documents" at the bottom of the menu on the left.


Figure 1 can be found under "Mr Bayes Plot"

Figure 2 under "Mr Bayes tree for Procellariiformes"

Figure 3 under "Mr Bayes subtree for Storm-Petrels"


Figure 2 shows two subgroups: Oceanitini and Hydrobatini, both with posterior probability values of 1.00 in Figure 2. Clearly in Figure 2, the genus Oceanodroma Reichenbach 1853 is paraphyletic. The citation for Oceanodroma Reichenbach 1853 is:


Oceanodroma Reichenbach, 1853["1852"], Avium systema naturale: das natürliche System der Vogel mit hundert Tafeln grosstentheils Original-Abbildungen der bis jetzt entdecken fast zw_lfhundert typischen Formen, p.iv. Type, by original designation, Procellaria furcata J. F. Gmelin, 1789.


Oceanodroma furcata (J. F. Gmelin 1789) Fork-tailed Storm-Petrel aligns with Hydrobates pelagicus (Linnaeus 1758) European Storm-Petrel. That node has 1.00 posterior probability support. 


Since Hydrobates Boie 1822 predates Oceanodroma Reichenbach 1853, Oceanodroma must become a junior synonym of Hydrobates.



That Oceanodroma Reichenbach 1853 become a junior synonym of Hydrobates Boie 1822.


Note: a further proposal considers the application of generic names within the Hydrobatini.


John Penhallurick, November 2005




Comments from Stiles: "[190 and 191] are linked, such that a YES vote on 190 implies a YES on 191 as well. So far as it goes, the genetic evidence appears to favor the proposals, so a tentative YES to both. My main caveat is that only one gene was sequenced: is this sufficient?"


Comments from Remsen: "NO. I strongly suspect that the Penhallurick-Wink data reveal the true phylogeny of the Hydrobatidae. Nonetheless, my personal policy on changes in classification is ... or tries to be ... based on congruence of two or more "data sets", e.g., two genes show same signal within the same paper, or two separate papers show same genetic results using different genes, or two sets of independent phenotypic characters shows the same thing, or any partial combination of the above. The guiding principal is that congruence among independent data sets gives me confidence that the results reveal true phylogeny, whereas I treat a single character as a phylogenetic hypothesis that should spawn additional testing. I doubt that I've been consistent in the application of this "policy", but I think when I've voted for a change based on 1 character, other characters that may not have been formerly analyzed are nonetheless consistent with the change. In this case, we're dealing with placing in synonymy the largest and most familiar genus name in the family, and so before making such a 'drastic' change to the classification, I eagerly await a second data set that supports this change."


Comments solicited by Remsen from Dr. Chris Witt, MVZ, UC Berkeley: "In a cytochrome b phylogeny, Oceanodroma is paraphyletic with respect to Hydrobates pelagicus. The generic name Hydrobates has priority. If the cytochrome b phylogeny reflects the true species relationships, a generic revision is warranted. Penhallurick is proposing that O. furcata and H. pelagicus should be in the same genus, although he doesn't state that explicitly. It seems like a reasonable idea given the close mtDNA relationship of these two species.... but shouldn't morphology, behavior, life-history, or biogeography also enter into this decision? Making the argument based on molecular data alone might be acceptable if there were data from multiple loci, but this case involves only cytochrome b sequence data. Available effort should be spent on generating more data or examining patterns of phenotypic divergence with respect to the cytochrome b phylogeny. After all, O. furcata and H. pelagicus look totally different!!!


"Certainly, if furcata and pelagicus are to be considered congeneric, they should both be in Hydrobates.... but I think the best course of action would be to hold off on any changes until nuclear loci corroborate the phylogeny and/or a reasoned argument based on levels of phenotypic divergence is made for the splitting or lumping of storm-petrel genera within the Hydrobates/Oceanodroma cytochrome b clade."


Comments sent to Remsen from someone who wished to remain anonymous: "It is my understanding that the SACC bases its decisions entirely on published material, is that right? If so, I'm not sure if you guys have realized that the results on which Penhallurick is partly basing his arguments for these new proposals have not been published, as the Bayesian analyses he refers to were not included in the paper he co-authored with Wink.


"In that paper they only presented a distance tree, a parsimony tree, and a maximum-likelihood tree, and they only presented bootstrap values for the distance tree, which in general means that the support for particular clades on their other phylogenies cannot be assessed. With regards to proposal 190, yes, all analyses put Hydrobates in a clade with Oceanodroma but relationships within the clade relevant to the position of Hydrobates are not well-supported in the only tree that has bootstrap values on it, which implies that based on the results presented in the paper one cannot reject the possibility that Hydrobates and Oceanodroma are sister genera in favor of his proposal that Hydrobates is nested within Oceanodroma.


"To me, it seems that Penhallurick may have a good case here (at least based only on cyt-b and I see the jury is still out on whether that's enough), but only if one accepts that Hydrobates is sister to one of the species of Oceanodroma with a high posterior probability. However, that piece of information (i.e. the strong support for that sister relationship in a Bayesian analysis) has not been published. This may be only a minor technical issue, but I thought I'd bring it up just in case you think it may matter... "


Comments from Robbins: "NO. Given that the published data for support of this proposal are based on only a single gene coupled with comments by Witt and an anonymous reviewer I vote "NO". I note during the past two years that the vast majority of molecular derived avian phylogenies are based on at least two genes and depending on the level of the question usually at least a mtDNA and a nuclear gene are used. Today, one would have a difficult time getting a phylogeny published (at least in most journals) based on only a single gene."


Comments from Silva: "NO. I would prefer to see the paper published as well as evidence from at least one more gene."


Comments from Stiles: "I originally voted YES on this, but with the caveat that I was not sure that a single gene (cyt-b) was enough to justify such a change. The cogent arguments of Witt and the anonymous reviewer (and I hate anonymous reviewers!!..) have convinced me that one gene is NOT enough; more evidence is required, and pending that evidence I will change my vote to NO."


Comments from Pacheco: "NO. Diante dos argumentos bastante convincentes empilhados acima, sou de opiničo que os resultados de Penhallurick and Wink (2004) precisam ser corroborados por evidźncias extraídas de um conjunto diferente de genes."


Comments from Zimmer: "NO, for all of the reasons cited by others, particularly as summed up in the last paragraph by Chris Witt."


Comments from Jaramillo: "NO - Being one of the last to vote on this, well, most of the good stuff has been said. I think that when results from a single gene analysis are so surprising, well if I was the researcher I would want to back this up with more data to feel comfortable about it myself. Even a good summary of vocal, behavioral, or other data that may bear on this result would be good to have. I don't doubt that it could very well be the "truth" in this case, but for the reasons noted by others I am hesitant to accept this to be the case.