2007-B-1: Recognize Larus smithsonianus as a species

NO. Neither the genetic nor the vocal data seems sufficient for splitting these two taxa at the species level. Although the mitochondrial data are weakly suggestive of differentiation, the nuclear data (de Kniff et al 2001, Crochet et al 2003) do not support recognition of argentatus and smithsonianus as distinct lineages. The vocal data consist solely of differences in response to alarm calls played during the non-breeding season, rather than of vocalizations that might be used for species discrimination while breeding. While interesting, these results cannot be viewed as definitive.

NO. I agree with others that mtDNA by itself is insufficient to elevata taxa to species, a concern that is heightened in this case by the propensity for hybridization in gulls. Furthermore, Crochet et al. (2002) only include smithsonianus from two locations (9 from Manitoba, 1 from Quebec), which is insufficient to definitively state that the taxon has a unique haplotype, and the genetic differences in both markers (control region, cytochrome b) are only 1 base pair. There does not seem to be any strong additional (i.e., independent) evidence, so for now I don’t think a split is warranted. If we were to recognize smithsonianus, I prefer Coues’ Gull to Smithsonian Gull.

YES. I can see that this motion like the one for splitting Vega Gull won’t pass, but I support it. They are separable in all plumages and from my limited understanding the genetic differences support a split. They certainly seem better differentiated than say Larus glaucescens from L. occidentalis, which we continue to maintain as two species. Regarding the taxa from Europe, both argentatus and argenteus, Boertmann (1994) mentions specimens of argentatus/argenteus from Greenland and Mactavish has photographed birds from St. John’s, Newfoundland. He believes the adults with slightly darker mantles have been nominate argentatus. They would be considered casual there. Additionally there is a specimen record of a first year bird from Ontario that is either argentatus or argenteus. I believe the specimen was examined by Devillers. There is a paper in Ontario Birds and I could include the reference if needed (somewhere in my files). So the European Herring Gull would need a full account.

NO. I am having a hard time coming to grips with what is happening or has happened in the genetics of the large white-headed gulls, especially in regards to present/past hybridization. The most genetically distinct species (occidentalis) hybridizes nearly freely with an embedded species (glaucescens). Two taxa with nearly identical adults (argentatus and smithsonianus) are not at all closely related. Does the phylogeny represent history or episodes of past gene flow between taxa? For a group that seems to hybridize so freely, I would think that hybridization has probably overridden much of the historical signature showing which taxa are closely related. For instance, in the Liebers et al. paper, separate populations of argentatus and hyperboreus both show up in two different clades. They don’t argue for separate species status for these, of course, and instead invoke hybridization. So it seems possible that all smithsonianus may now have mtDNA of some populations that it has hybridized with in the past (in the Pons et al. it groups with mainly American taxa) and/or that argentatus has a Euro mtDNA (it groups with mainly European gull taxa in the Pons paper, though with several other taxa –marinus, fuscus, heuglini – in the Liebers et al paper). Genetic differences (in the Liebers et al paper) between smithsonianus and argentatus are minor (as close at 3 base pairs apart).

I think it is just as likely that argentatus and smithsonianus are each others closest relative, it’s just that episodes of gene flow from other taxa has erased the mtDNA signature of the close relationship. The nearly identical adults would mean that there would not be reproductive isolation. The vocal data in Frings et al. (1958) on response of Euro Herring Gulls to smithsonianus vocalizations has no bearing on their reproductive isolation: the tests for difference in responses were done at different times of the year and by different people, they seemed to only test repellent alarm calls, and they were not done on breeding grounds.

NO, somewhat emphatically. As far as I can see, the only real evidence for this split comes from the mtDNA separation of smithsonianus from the Old World taxa and its apparent mtDNA clustering with other NA Larus species. The acoustic evidence seems dubious to me, as it comes from an era (1958) before people really understood how to do playback experiments with the appropriate controls (but I didn’t check that paper myself). The mtDNA data are bothersome because all of the relevant papers make a fairly big deal out of the known propensity for mtDNA introgression among these and other gull taxa; some of those papers go further and actually suggest that maternal-line introgression will actually be lower than autosomal introgression. In any case, the observed separation of smithsonianus mtDNA is consistent with hybridization and introgression at the time of smithsonianus colonization of NA. This scenario of mtDNA replacement appears to have happened in some of the old world Herring Gull subspecies, and is not far-fetched in a situation in which a (rare) taxon prone to hybridization was colonizing an area already occupied by a more demographically numerous mtDNA-source taxon.

I would be much happier if additional, unlinked molecular markers supported this split. But while the AFLP study (Knijff et al. 2001) found rampant gene flow among Eurasian forms, it didn’t include smithsonianus. Allozyme data suggest no differentiation (Snell 1991 summarized in Crochet et al. 2003). Microsat (also Crochet et al. 2003) patterns are not definitive owing to taxon sampling limitations, but would seem to argue against this split: notably smithsonianus-argentatus has the LOWEST msat differentiation among all pairwise comparisons involving those two taxa plus fuscus, michahellis, and marinus.

Given the very, very low nuclear differentiation in this species group, and the known problems with hybridization and mtDNA introgression among these gulls, I doubt that there is a phylogenetic magic bullet that can resolve this. It might be better to fall back on a BSC-based guess: if smithsonianus and argentatus occurred in broad sympatry, my money would be on a lack of reproductive isolation leading to rampant hybridization and genetic homogenization.

YES, hesitantly. But I basically give up on species limits in this group. RE common names: Coues’ Gull would be my first serious choice, American Gull not a bad choice either. It would mean the masses would very often be right in gull identification, even without having to wade through all those odious papers and websites.

NO. Crochet et al.’s data represents a gene tree, not necessarily a population tree, and the way mtDNA floats through gull populations, this is insufficient data for ranking as a species. Therefore, for example, whether two species are each other’s sister in terms of mtDNA refers to that gene, but not necessarily the populations as a whole. The vocalization data are interesting, but should provide direction for thorough analyses and studies. When Olson & Banks appears, I will re-evaluate, of course. What is needed, in my opinion, if there are no direct tests of gene flow, is a comparative analysis of the kinds of phenotypic characters associated with assortative mating or nonassortative mating at the many contact zones among large larids; with all the variation in bill spot color, leg color, iris and eyering color, mantle color, and wing tip pattern, there is potential for some clear patterns to emerge. If smithsonianus and argentatus differ to the same degree as the many parapatric taxa that mate assortatively (with L. occidentalis and L. glaucescens the coming the closest in phenotypic similarity and (barely) assortative mating, then I’ll change my vote.

NO. Let’s put this off. If ok, I don’t like the “Smithsonian” English Common Name.

NO, based on comments from other committee members.

NO. In reviewing the primary literature I am not convinced that species limits in this group are yet well known or defined. The mtDNA evidence on gull species limits is not particularly strong, and with evidence of hybridization and reticulation occurring in this group it would be premature to change current taxonomy. Nuclear genetic markers and deeper population sampling are required to resolve this issue. (Another problem in gull research is the apparently poor representation of voucher specimens.) On the English name of this taxon, I found myself swayed more by Dan Gibson’s suggestion of “Sea Gull” (not part of this proposal) than by using the specific epithet to create “Smithsonian Gull.”

2007-B-2: Recognize Larus vegae as a species


NO. The same problems as above apply to using mtDNA alone for species limits. Furthermore, Crochet et al. (2002) only had 6 samples of vegae in their analysis, and the Liebers et al. (2004) paper showed that vegae is not distinctive in its mtDNA. The differences in eye ring and mantle color are probably important, but like Van, I would like to see a formal analysis.

YES. If this motion were to pass (seems unlikely) then I would assume that L. v. birulai (often not recognized as a valid subspecies from vegae) would be merged. Vega Gulls are not casual in Alaska. They are probably best termed as uncommon in the Bering Sea region, and indeed they are a common breeder at St. Lawrence Island.

Other work within the large Larus that needs to be done is to split off heuglini from L. fuscus. There is a recent specimen supported record of heuglini from Shemya Island.

YESVegae differs genetically from argentatus to about the same degree as do smithsonianus and argentatusVegae and smithsonianus, however, do not group apart in the Libers et al. network tree (in group H). More importantly though, vegae differs from smithsonianus (and presumably argentatus) in characters that are probably important in species recognition (eye-ring color and mantle color). They also differ in wing-tip pattern. I therefore think that we should split vegae fromargentatus/smithsonianus. We will have to know where to draw the line between argentatus and vegae in central Asia; the proposal is unclear. By the way, vegae is much more common than “causal” in western Alaska; it breeds on St. Lawrence Island.

NO, for the same reasons outlined above, but less much emphatically because the morphological evidence for differentiation is stronger here.


NO. Crochet et al.’s data represents a gene tree, not necessarily a population tree, and the way mtDNA floats through gull populations, this is insufficient data for ranking as a species. Defining species-level taxa by distribution patterns of maternally inherited haplotype distributions is a misinterpretation of genetic data in my opinion. Further, I can’t think of a worse group than the large Larus in which to use any mtDNA – the frequency of hybridization to varying degrees at contact zones and among overlapping species is perhaps higher than in any other group of birds. Other comments suggest that indeed vegae should be ranked as a species based solely on phenotypic criteria, but I would have to see a formal analysis. See also my comments under 2007-B-01.

NO. Let’s see what the molecular people say.

NO, based on comments from other committee members.

NO, for the same reasons given in proposal 1. Both of these taxa may very well be full species, but greater sampling depth and nuclear markers will be required to confidently make this determination.

2007-B-3: Adopt reclassification of Laridae by Pons et al. (2004), incl. generic limits (also see SACC #250)

Partial YES on adopting the order suggested here. The first part of this (through ChroicephalusHydrocoloeus, PagophilaXema, and Rissa) seems reasonable, but I would suggest re-arranging the Larus and Leucophaeus species somewhat. The phylogeny of Pons et al. (Figs. 2, 3) is not well resolved in many places, but the bulk of the evidence would suggest that Larus and Leucophaeus form a clade with the Leucophaeus species positioned basally. There is some resolution within Larus and I think this should be reflected in the sequence, but there is no resolution within the “white-headed group.” I think we should keep the current order within the white-headed gulls, which reflects the evidence of relationships based on phenotype. The suggested order for the Leucophaeus/Larus group would therefore make changes only where there is some reasonable evidence for relationships (either sister relationships or basal vs derived positions) based on the molecular data, as follows:

      Genus Leucophaeus Bruch, 1853
Leucophaeus atricilla Laughing Gull
Leucophaeus pipixcan Franklin’s Gull
Leucophaeus modestus Gray Gull (A)

     Genus Larus
Larus crassirostris Black-tailed Gull (A)
Larus belcheri Belcher’s Gull (A)
Larus heermanni Heerman’s Gull
Larus canus Mew Gull
Larus delawarensis Ring-billed Gull
Larus occidentalis Western Gull
Larus livens Yellow-footed Gull
Larus californicus California Gull
Larus argentatus Herring Gull
Larus vegae Vega Gull
Larus smithsonianus Smithsonian Gull
Larus michahellis Yellow-legged Gull (A)
Larus thayeri Thayer’s Gull
Larus glaucoides Iceland Gull
Larus fuscus Lesser Black-backed Gull (N)
Larus schistisagus Slaty-backed Gull
Larus glaucescens Glaucous-winged Gull
Larus hyperboreus Glaucous Gull
Larus marinus Great Black-backed Gull
Larus dominicanus Kelp Gull

YES. There is strong support at most nodes in the Baysian tree (although some nodes are not well-supported), and the clades nicely match the major phenotypic groupings. I agree with others that we should retain Rhodostethia for Ross’s Gull. We need to think carefully about the linear sequence, however, as suggested by others.

YES. I would add that Little and Ross’s Gulls are very similar to one another in juvinal plumage.

YES, except to continue to recognize Rhodostethia. I think that the genetic dataset of Pons et al. is much more useful for higher level classification. The results look fairly robust. I think, however, that we should continue keep Rhodostethia separate from Hydrocoleus, especially if we keep Pagophila separate from Xema.

NO. This one is borderline for me and I probably could be convinced easily to change my vote to the positive. I’m not quite as worried about the mtDNA issue at this deeper level of the tree (though it is certainly still there as a potential problem). I’m more concerned about resolution at several critical nodes at the base of their revised Larus, where I note the Pons et al. ‘consensus’ tree used to present the revised classification actually conflicts with the Bayesian tree (the only other tree shown) on the placement of crassirostris. That general region of the tree is also comprised of short branches with negligible support. My general feeling is that if genera such as Larus are going to be based on monophyletic groups (which I strongly support philosophically), the evidence for that monophyly should be reasonably strong, and that is not universally the case here. It is unfortunate that the study did not include additional mtDNA sequencing and an unlinked marker or two, as the former might well have added the necessary resolution and the latter provided an independent corroboration. I would guess that this topology is generally accurate but would prefer to await a stronger corroboration before making large generic nomenclature changes.

I agree that the proposed generic-name handling of the various well-differentiated sister-species pairs seems arbitrary. I lean strongly toward using a single genus name for each pair if the genus taxonomy is to be revised.

YES, although like others I think Little Gull and Ross’ Gull should be in separate genera. Incidentally, Little Gull does have some distinctive osteological characters.

YES and NO. YES to recognizing multiple genera in Larus. At this level, mtDNA differences are useful, and I see no flaws in the interpretations of Pons et al. The congruence with plumage similarities is comforting. Like others, however, my YES vote does not included considering minutus and rosea congeneric. This seems inconsistent with treatment of sabini and eburnea in separate genera (XemaPagophila), both in terms of genetic and phenotypic differences. As far as the overall classification (the proposal includes more than just generic limits), NO – any changes at species level should be dealt with individually as in the proposals above. Also, concerning the specific proposed sequence of genera and species, I have lots of problems that are best outlined separately, but include overall structure (e.g., I would start PagophilaXemaRissa and end with Larus sensu stricto to keep the sequence as close as possible to an inversion of the traditional one as well as recognize the hierarchy of the better supported nodes in Pons et al.) to specific points (e.g., the internal nodes in Pons et al. have no support, and glaucescens and occidentalis barely differ phenotypically, are exchanging genes at a level that is as close to nonassortative as it gets among North American taxa ranked as species – they need to place as close together as the occidentalis-livens relationship will allow).

YES. If accepted I want to keep Ross’s Gull in its own genus.

YES. This looks good. My one objection is to placing Little Gull and Ross’s Gull into the same genus. I would prefer to maintain Ross’s Gull in a monophyletic genus, Rhodostethia.

YES, in part. Pons et al. clearly state that the phylogeny is not fully resolved. Their tree has a number of poorly supported nodes shown that should have been collapsed (and some with no support given at all!). Also, the apparent lack of vouchers in many cases is a concern. Multiple specimens per species currently represented by just one individual and nuclear loci will be good to see added to this work to further resolve relationships. I do not support merging Little & Ross’ gulls into Hydrocoloeus. I have field and museum experience with both and, while the generic name of the Little Gull would have to be changed (to Hydrocoloeus Kaup 1829) given their phylogeny (Fig. 1), Rhodostethia should remain as a monotypic genus. The differences between these birds are sufficient to warrant such, in my view. The “masked” group as Chroicocephalus, the “hooded” group as Leucophaeus, and the “black-headed” group as Ichthyaetus all have good support and could, as they suggest, work well as separate genera. The remaining members as the genus Larus itself, however, rest upon one of those rather poorly supported nodes; i.e., its monophyly is not assured. Given the preponderance of evidence, however, especially how well key larid plumage features generally corroborate the phylogenetic findings (though there are surprises), I can accept all of these changes. Where Larus saundersi falls in the phylogeny is unresolved. Nevertheless, it is not (according to these data) a close relative of the other groups, and so I also accept that the monotypic genus Saundersilarus should be resurrected as they suggest for Saunders’ Gull. I do applaud the authors for providing in their paper detailed taxonomic analyses and recommendations based on their work.

2007-B-4: Recognize Chondrohierax of Cuba as a species

NO. I don’t see enough evidence that the Cuban population is distinctive enough to warrant species status.

NO. This could go either way, but I don’t think that the evidence is strong enough as it stands to justify a change in our treatment.

YES. Unlike the case of Cuban Black Hawk, there is no vocal data to back up this split, but they certainly do look different (e.g. bluish rather than yellow facial skin) and I think the most prudent course is to split this nearly extinct species.

YES. This is a total judgment call, because there is no way to determine if the bill color differences would be important enough for reproductive isolation. Given the rather deep split between wilsoni and the rest of uncinatus (incl. Grenada), that many of the closely related taxa that share the distribution are recognized at the species level (ButeogallusIcterus dominicensis), and that the taxa differ in a fairly good morphological character, I think we should go for the split.

NO. Both mtDNA and morphological evidence seem very borderline at best, and I am more comfortable with just sticking with the status quo in situations like this where a change is not clearly warranted.

YES. Obviously a borderline case but the yellow bill in combination with the genetic divergence seems to bring it above the level of differentiation typical of subspecies. And I do think it matters whether an extinct taxon is considered a species or not just as much as for an extant one. How else can we document and analyze levels and patterns of extinction?

NO. That the Cuban and Grenadan populations differ somewhat in % sequence divergence of two mitochondrial genes is unsurprising given their insular distributions. The only surprise would be if no genetic difference were detected (which would reflect recency of colonization or some sort of constraint on differentiation). Minor genetic differentiation is irrelevant to establishing species or even subspecies limits (under BSC); further, if these insular populations are indeed diagnosable phenotypically, then with the reasonable assumption that those characters are genetically controlled, we already know that the populations are “genetically distinct” even if we can’t isolate the responsible genes. Even using a (fatally flawed) % sequence divergence yardstick, 1.8-2% really isn’t that much for tropical and insular bird taxa ranked as species.

What is needed is a study of characters associated with gene flow or lack of it in related raptors (e.g., voice), or a more general analysis of species limits in parapatric/sympatric taxa treated as either species or subspecies in raptors and the ways in which they differ.

YES, but this one is a crap shoot. MIGHT be a good place for subspecies.

YES. Data are not outstanding, but this is a pretty distinctive taxon compared to the widespread Chondrohierax. It think splitting it is consistent with our treatment of other isolated West Indian taxa.