Proposal (399) to South American Classification Committee

Waive “peer-reviewed publication” requirement for online vocal data

Recent proposals and relatively new on-line resources have precipitated the following proposal. Most committee members have followed the philosophy of having a peer-reviewed publication before implementing a taxonomic change. However, with the advent of readily accessible vocal data, through Xeno-canto and Macaulay Library of Natural Sounds, I submit that we should change our protocol for those taxa where the primary characters for making a taxonomic change are vocalizations. This would be restricted to those taxa that are named and thus have associated documented morphological differences. A good example of this was the recent Dubusia proposal (# 392), where plumage morphology and distribution have been described and the sole new information was vocal which was accessible on-line via the above web sites.

So, what are the advantages to making this change? Aside from the obvious of expediting taxonomic changes, this protocol has other advantages over paper publications: (1) more critical review. For most journals, outside review is limited to two or three people. Moreover, the process isn’t always transparent. This is in sharp contrast to our protocol that has input from a minimum of ten people and is as transparent as possible; (2) I think everyone would agree that it is far more elucidating to listen to a vocalization while viewing a spectrograph, as is the case with on-line vocal data; (3) space limitations in paper publications vs. unlimited sample sizes on-line. In a typical published taxonomic note one is limited by space to just a few spectrographs, whereas on-line resources there are no limitations. Excellent examples for contrasting these limitations/advantages are two recent proposals: # 387, splitting Poospiza cabanisi from P. lateralis and # 392, elevating Dubusia taeniata stictocephala to species level. In both cases, names and plumage morphology were previously described – vocalizations were the new data. In the Poospiza proposal a single spectrograph of each song and call note were presented whereas 30 vocalizations could be visualized *and* listened to, via Xeno-canto and MLNS, for the Dubusia proposal. The comparison speaks for itself. Moreover, on the above two websites the localities are often mapped, whereas one needs to consult a gazetteer for ascertaining localities in a paper publication. In other words, it doesn’t get any easier in having all the critical data available.

Thus, I see nothing but advantages to adopting this new protocol. Hence, I recommend that we adopt this protocol.

Mark Robbins, June 2009

Comments from Cadena: “NO. I see Mark's point regarding the accessibility of online recordings, which is a lot better than simple verbal descriptions of vocalizations that have often been used to justify changes in species level taxonomy (e.g. in field guides). However, I don't think that wide accessibility over the internet overturns the need for rigorous analyses that are passed through the peer-review process and published in scientific journals in order to justify changes to our classification. Online databases such as XenoCanto are an excellent complement for analyses presented in scientific papers, but are no substitute for them.”

Comments from Zimmer: “YES. I’m really torn on this one. First, I don’t remember that we, as a committee, had taken a hard-line stance requiring “peer-reviewed publication”, although that is certainly the way we have leaned in the majority of cases lacking such published analysis. I wholeheartedly echo Daniel’s point that there is no substitute for a rigorous, peer-reviewed, published analysis. However, there have been a number of cases involving proposals to split species that were originally described as distinct species, and then subsequently lumped without comment (particularly by Peters), with the result that our status quo is reflecting an unsupported, unanalyzed change that is not founded upon any published evidence – peer-reviewed or otherwise. In such cases, my usual position has been that any available published evidence (even in the form of field guide vocal descriptions) supporting the original species-limits treatment trumps a status quo that lacks any kind of published support other than inertia. To be clear, we are talking about named taxa, for which morphological differences are documented and published in peer-reviewed publications. So the information on vocal differences represents the only “new” or non-vetted data. As Mark points out, online vocal collections allow committee members to more readily evaluate anecdotal or published transcriptions of vocalizations. They should not be viewed as a substitute for a published analysis, which I would always consider to be preferable, but I do think that in overturning some unsupported “lumps” they should be considered adequate, in the interest of getting things right as opposed to rigidly following a precedent that all available evidence indicates is wrong. So, if in fact we do have a “requirement” for peer-reviewed publication of vocal differences, then I would vote YES for waiving that requirement for on-line vocal data, but only in cases involving previously named taxa in which our current species-limits are not based upon any published analysis. If previous published analysis is the basis for our current treatment, then I think any change demands counter-evidence subjected to the same level of scrutiny.”

Comments from Schulenberg: “YES. With enthusiasm. Daniel voices a concern about the lack of "rigorous analyses" in simply listening to recordings of various taxa from online, institutional sources (Macaulay Library, xeno-canto). My response is, there are plenty of cases in which two taxa really don't sound anything at all alike, the 'whole world' can tell this with their own ears, and we're just wasting our time (and not looking very sharp, either, cutting edge transparency notwithstanding) if we have to sit around for someone to write a paper that says 'these two taxa don't sound anything at all alike.' In other words, if you listen to the material yourself, and you are convinced of the differences just by listening, then ... why not go ahead and allow yourself to vote for a change that you believe in?”

“Obviously there will be cases, perhaps many of them, where the vocalizations of two or more taxa are more similar - different, but still recognizably similar. If you're not certain just by listening, then vote 'no' and wait for the published rigorous analysis. I don't see how allowing this option does any harm, and it could do us a great deal of good.

“I'd like to go one farther, and suggest that we take more advantage of institutional, digitized, online sound recordings as documentation for the presence of a species in a country for the SACC country lists. As worded, the criteria (never voted on by SACC? or am I forgetting?) are "Such evidence may consist of a specimen, a photograph or video, or an audio-recording, as long as the evidence is archived in an institutional collection, and its existence is published." I understand the importance of relying on documentation that is archived. But I don't understand why it is more important that someone mention in print that a sound recording is archived, than it is that anyone who cares to can listen to the sound themselves and verify the identification independently.”

Comments from Remsen: “NO. Why should vocal data and analyses be treated any differently from any other data? If we adopt this, then why not just dispense with publication requirements altogether? Why not just circulate some unpublished trees to show that a genus is polyphyletic, for example, and the heck with the formality of getting the data-set and its analyses and interpretations reviewed and into print. Hey, we all know supposedly how to tell from a tree whether a genus is polyphyletic from well-supported nodes on a tree, and so can the ‘whole world’, right? A parallel case can be made for almost any sort of data that we examine. I’m sure that the Peters Checklist-era taxonomists felt the same way – why bother to publish analyses and rationale when the conclusion is obvious? And now we are paying the price.

“Even sticking strictly to vocal data, the raw data, the sonograms, should be analyzed with respect to sample size, geography, context (e.g., playback response/not), season, etc., and the rationale for the change outlined specifically and quantitatively. To reduce decisions to “sounds different to me” (or not) represents retrogression in the direction of the realm of art appreciation. To go this route represents a threat to our credibility. I’m all for using Xeno-canto, Macaulay, and any online sources as supplementary material – just not as a substitute for an analysis.

“Finally, by the time someone writes a proposal with sufficient rigor to submit to SACC, they’ve already almost all the work necessary to submit the proposal as a short note for publication. The current proposal on Dubusia (392) is a good example – with slightly more effort, it is publishable in Cotinga or BBOC. Therefore, pragmatically, going the route of this proposal really doesn’t reduce the workload much, and the process of getting that proposal draft published as a short note will only improve the final product.”

Comments from Stotz: “YES.  I have never been as doctrinaire as some on requiring a peer-reviewed publication for a split or a lump. We have dozens of cases where the basis for our current treatment is weak to non-existent.  We treat them a certain way because Peters or Hellmayr or somebody treated them that way. I think that if somebody makes a clear case why the treatment we currently follow is wrong then we should adopt that new treatment. Mark makes a  good case here for the advantages of these on-line resources for vocalizations. I am not as convinced as Van is that that it is trivial to turn a SACC proposal into a paper that would be published although certainly it would be good to have that occur as much as possible.”

Additional comments from Remsen: “Although I couldn’t agree more strongly with Doug on the point that the status quo often has little or no established rationale, this is not an excuse, in my opinion, to revert to standards of a half-century ago. In fact, if this proposal passes, I could no longer in good faith support the mission of SACC and will resign. I also doubt that the American Ornithologists’ Union would continue its support.”

Comments from Stiles: “NO. Once again, I will vote NO here because I believe that the evidence in such cases should appear in a peer-reviewed publication. Especially when I am not familiar with the birds in question, I find a published analysis important in helping me reach a decision. Quite often the proponents of changes in status based (at least in part) on vocal evidence that already exists but has not been subjected to a careful analysis could perform the analyses with relatively little effort and the publication would almost surely follow. This might mean a bit of a delay by comparison to Mark’s proposal, but here I think it is worth it.”

Additional comments from Robbins: “Both Daniel and Tom make good points about the need for revisiting some earlier proposals. A good example is the very first proposal that we assessed, as it is germane to the current discussion on using online information to evaluate species limits for taxa that already have names. Below I paste in Van’s comments from that proposal. After reading this I pose the question, given the many shortcomings of the vocal data in this paper (all but one of us supported the proposal) is it any better than having a minimum of 10 committee members (remember outside members can comment as well) evaluate a proposal that has more complete sampling and allows one to analyze every song both aurally and spectrographically? Given that the Maijer article was published in the Auk, it is likely there were no more than two reviewers (there is no mention of any reviewer in the acknowledgments). Moreover, only a single spectrogram is presented for each proposed species. In comparison, there are 30 songs from > 25 localities for the Dubusia proposal. I’m using this not because I was part of the proposal, but because at present it is our only example of using online vocal data for making a taxonomic decision. This is just one example that illustrates the shortcomings of subscribing to a published account only philosophy of acceptance.

“The online proposal embraces technology that provides better access, analyses of more data, a more rigorous review, transparency, and expedites the process. I fail to understand how this is not a vast improvement over the opaque and often untraceable single-author approach (e.g., Peters, Hellmayr, Meyer de Schauensee), which the committee often bemoans. As one final comment, months ago I suggested that we create an online SACC journal. Naturally, this would eliminate any concerns about having something published before it would be considered as a proposal.

SACC proposal # 1:

New information: Maijer (1996: Auk 113: 695-697), however, provided data on primary vocalizations that suggests that this taxon merits species rank. His rationale was as follows: (1) songs of tinamou taxa currently treated as species are surprisingly uniform (citing Hardy, Veillard, and Straneck ARA cassette); (2) songs of the three lowland subspecies of R. rufescens differ in only minor ways, perhaps only individual variation (based on his limited experience); and (3) the difference between maculicollis and lowland rufescens is substantial; to quote Maijer: "the song differences between maculicollis and lowland populations of R. rufescens are as great as between closely related species in other tinamou genera (pers. obs.)". Maijer compared tape maculicollis from 5 individuals from 4 widely separated areas of Bolivia to single representative recordings of lowland rufescens from the three named subspecies (from Bahia, Huanchaca, and Entre Rios) to quantitatively demonstrate the differences. Maijer noted that the foothills and lowlands populations are probably allopatric, separated by unsuitable (forest) habitat.

Recommendation: We could easily fault Maijer for not having larger and broader samples, for not evaluating whether the closest rufescens populations approach maculicollis in voice, and for not quantifying his statements on within-species vocal variation in other tinamous. However, such criticisms could be leveled at almost any new data set, and if we set our standards that high, we might as well endorse 99% of the status quo and all go home. Furthermore, our current status-quo taxonomy is based on much less. For example, Peters or whoever started treating maculicollis as a subspecies probably lumped it into rufescens without so much as a comment. From my experience on the AOU CLC, this example will be a typical dilemma for us: retain a status quo often based on unstated rationale or opinions versus accept a novel change backed by data that is often far below what we'd hope for. Unfortunately, a dissertation-quality study cannot be undertaken on every taxonomic problem.”

Additional comments from Remsen: “My point is not that print-version data are necessarily better than online data, nor that being in print guarantees quality. In fact, we’ll soon be dealing with peer-reviewed online-only journals. My points are: (1) simply referring to online recordings or any other online raw data and asking committee members to interpret them as “similar” or “different” as the basis of a proposal is not the same as an analysis; and (2) as long as one has to go to the trouble of analyzing and interpreting online raw data for a proposal, it doesn’t take that much more effort to get that proposal into publication draft form, e.g., the current Dubusia proposal, which by using and referring to Xeno-canto and other online resources, which I strongly support, would be a better paper than Maijer’s published note. Further, before I am dismissed as a fuddy-duddy for not favoring the proposal, note that I hope I have established some degree of non-fuddy-duddyhood in terms of developing internet-based tools (e.g., SACC is entirely internet-based). However, I am still concerned about the fundamentally ephemeral nature of web pages and their content (thus, eager to get a print-version of SACC classification); the content and availability of any online pages are potentially fluid, whereas that in Maijer (1996: Auk 113: 695-697) are fixed.”

Comments from Nores: “NO. Yo no veo porqué tengamos que hacer excepciones con vocalizaciones a lo que ha sido la política desde el primer momento. Al contrario, pienso que con vocalizaciones y análisis moleculares hay que ser más exigente que con otros aspectos, ya que los principales cambios en la taxonomía clásica están basados en estos dos aspectos. Además, esto no significa que uno no pueda usar lo que aparece en Xeno-canto and Macaulay Library como complemento de los papers, una cosa no elimina la otra. Yo coincido plenamente con Daniel que los database online son excelentes complementos para análisis presentado en papers científicos, pero no son substituto de ellos. Aunque Douglas tiene razón de que nosotros tenemos docenas de casos donde la base del tratamiento es liviana o no existente, eso no significa para mí que tengamos que fomentar ese tipo de tratamiento. Considero que tendríamos que ser más exigentes no sólo para las vocalizaciones sino para todos los otros aspectos. Pienso que sería importante, como ya ha sido señalado por otros miembros de SACC, re-visitar alguna propuestas hechas al principio.”

Comments from Pacheco: “YES. Minha opinião é semelhante, ou possivelmente idêntica, à do Kevin. Eu tenho para mim que o SACC poderia utilizar desses bancos de vozes online nos casos nos quais o tratamento taxonômico vigente foi arbitrado sem considerar (sequer de forma rudimentar) informações oriundas do repertório vocal dos táxons implicados (e.g. Hellmayr, Peters, etc).  Em suma, esse uso não deve ser visto como um substituto de uma análise publicada, mas um recurso preferível em lugar da “inércia” referida por Kevin.”