Nikon Monarch 7

aba events

ICYMI: Open Mic: Hoary Redpoll; A False Dichotomy?

The ABA Blog has been in existence for almost 7 years, and there’s a lot of good content back in the archives that deserves an audience now that it might not have received way back when. So, semi-regularly we will bring some of that stuff back. Here’s an Open Mic post about a redpoll lump from Andy Boyce that seems particularly prescient in light of a recent proposal submitted to the AOS.


At the Mic: Andy Boyce

Missoula, Montana, resident, Andy Boyce is a graduate student currently working on his PhD at the University of Montana.  His area of interest is the ecology of tropical birds in Borneo.


    We are currently experiencing  one of the largest southern movements of redpolls in the last 50 years and birders across the country are out and about, scrutinizing flocks of redpolls at Nyjer feeders, looking for that holy grail of the redpoll flock; Hoary Redpoll.

With the birds, there has been a surge of posts to local list-servs, blogs, etc. One group searching for advice — “just one dark streak on the undertail coverts, is this a Hoary?”, and others diligently going through all of the traits espoused by field guides and experts that are supposed to allow us to sort through these messy flocks of finches.

Doing my own research on this topic, I discovered a paper published in 2008 entitled “Low support for separate species within the redpoll complex from analyses of mtDNA and microsatellite markers” (Marthinsen et al. 2008). Understandably I was a little shocked. My first thought being, what the heck are we doing trying to sort these things out into species if there is no evidence that they are even species at all?  After a  thorough read I was totally convinced that attempting to sort redpolls into species, or even subspecies is a totally futile exercise.

HORE Schmoker
Colorado’s first Hoary Redpoll? photo by Bill Schmoker

The crux of the paper is this; based on museum specimens from across the old-world range of the 2-3 redpoll species commonly recognized, the authors found no evidence at all that what we humans are describing as species or subspecies corresponded to any sort of meaningful, genetically differentiable groups.  After reading through this paper, I looked around some more and found out that several previous studies have also failed to find any genetic structure among species or subspecies of Redpolls (Marten & Johnson 1986, Seutin et al. 1995, Ottvall et al. 2002, Hebert et  al. 2004, Kerr et al. 2007).

The phylogenetic work was done using both mitochondrial and nuclear DNA, which evolve/mutate at different rates. Thus, this is a pretty robust study in terms of methodology and is unlikely to be a false negative. Some have argued that they simply missed the appropriate genetic markers to differentiate the species/subspecies, but this is also unlikely given that they were able to show the monophyly of two subspecies of Twite (close relatives of Redpolls) using the same methods.

Another hypothesis suggested by some is that the Redpolls represent incipient species undergoing sympatric speciation. I suppose that it possible, but we would still expect genetic differentiation between groups if this was the case.

This finding leaves us with 2 interesting possibilities;

1)      All redpolls belong to one single panmictic gene pool. There is no Hoary Redpoll, there is no Common Redpoll, there is no  exilpes, there is no flammea…etc. They are just Redpolls.

2)      There are monophyletic (genetically distinct) groups somewhere in the redpoll complex, but we humans, in particular museum curators and collectors, cannot identify them based on morphology, plumage, or even range.

These are two very different hypotheses, but for birders they really mean the same thing. They mean that we cannot ID a Hoary Redpoll. You know why? Because we have absolutely no idea what (or even if) a Hoary Redpoll is! If there is a mysterious genetically distinct group out there, it appears that it isn’t linked to any convenient suite of external traits. That means that Sibley’s wonderful page describing how to score a pale redpoll on a scale from 1-Hoary is more or less meaningless. The paper that the page is based on even admits to having essentially no clue about what a Hoary is, they simply state that they assume that any individual falling in the “hoariest” third of the total distribution of variation is a Hoary. This is shaky to begin with, but when you take into account that there is actually no such thing as a 100% surefire Hoary Redpoll, it’s totally nuts!

CORE wiki
Common Redpoll, Quebec, photo from wikipedia

Ok, so you don’t care about genetics, let’s talk about probability. We know that Redpolls that breed in shrubby habitat in the far north, in areas without trees tend to be larger, smaller-billed and whiter in various nebulous ways. For the sake of argument let’s throw the latest science out the window (if the government can, why can’t we, right?). Let’s call those hulking, frosty seed-killers Hoary Redpolls. Unfortunately, we know that all of these traits that are “good” for Hoary Redpoll are extremely variable and they don’t always occur in concert within an individual.  That is to say, there are birds out there that have extremely white upperparts, reduced “poll”, very little flank streaking, but gosh-darn-it look at those monster undertail covert streaks! Blast!  Given that these traits are fallible, we have the following problem, straight from the Ted Floyd book of rhetoric.

Based on observed traits we can assign some sort of probability that a bird showing those traits is a Hoary Redpoll. We can also assign some probability of a Hoary Redpoll showing up in Colorado versus a Common Redpoll.  Based on that, we can get a probability of a pale redpoll in Colorado being a Hoary Redpoll.  Here are some numbers:

Probability of a pale redpoll being a Hoary: 99% (this is incredibly generous)

Probability of a redpoll in Colorado being a Hoary: .001% (numbers from the east coast during irruptions suggest ratios of around 1000:1).

Probability of a pale redpoll in Colorado being a Hoary Redpoll: ~1%

I should point out that the current ratio of Common Redpoll to Hoary Redpoll on Colorado Ebird checklists for 2013 is more like 100:1. This ratio is tremendously skewed because birders, understandably, chase and report rare birds more than common ones.  Even if we take that ratio  as gospel, the probability of any given pale redpoll being a hoary in Colorado is still only 50%.

So, what am I trying to say with all this? I guess the take home message is that if you believe in the current scientific evidence, you have to acknowledge that we have no idea what or if a Hoary Redpoll is. If you don’t believe in the scientific literature but you do believe in the power of statistics, then you have to acknowledge that even if you see a pale redpoll in Colorado, there is no reason to believe that it is a Hoary. In fact, the statistics tell us that it is almost certainly NOT a Hoary Redpoll.


Literature Cited:

Hebert, P.D.N., M.Y. Stoeckle, T.S. Zemlak, C.M. Francis. 2004. Identification of birds through DNA barcodes. Pub. Lib. Sci. Biol. 2; 1657-1663.

Kerr, K.C.R., M.Y. Stoeckle, C.J. Dove, L.A. Weigt, C.M. Francis, P.D.N. Hebert. 2007. Comprehensive DNA barcode coverage of North American Birds. Molecular Ecology Notes 4: 535-543.

Marten, J.A., and N.K. Johnson. 1986. Genetic relationships of North American Cardueline finches.  Condor 88: 409-420.

Marthinsen, G., L. Wennerberg, J.T. Lifjeld. 2008. Low support for separate species within the redpoll complex (Carduelis flammea-hornemanni-caberet) from analyses of mtDNA and microsatellite markers. Mol. Phylo. & Evolution 47: 1005-1017.

Ottvall, R., S. Bensch, G. Walinder, and J.T. Lifjeld. 2002. No evidence of genetic differentiation between lesser redpolls Carduelis flammea caberet, and common redpolls Carduelis f. flammea. Avian Sci. 2: 237-244.

Seutin, G., P.T. Boag, and L.M. Ratcliffe. 1995. Mitochondrial DNA homogeneity in the phenotypically diverse redpoll finch complex (aves: Carduelinae: Carduelis flammea-hornemanni). Evolution 49: 692-973.

The following two tabs change content below.


The ABA Blog's Open Mics offer an opportunity for members of the birding community to share their voice with the ABA audience. We accept all and any submissions. If you have something you'd like to share, please contact blog editor Nate Swick at [email protected]
  • Ted Floyd

    Good stuff, Andy. Great to have you back in our hemisphere… 🙂

    And nice to see that you got yourself some of that Ted Floyd rhetoric stuff during your all-too-short stint in Boulder.

    Anyhoo. The Bayesian approach advocated by Andy is spot-on, it seems to me–especially with regard to “Sibley Scores,” as I’m hearing them called (not David Sibley’s own moniker, to be fair). It’s not just that a bird with a “good” “Sibley Score” could conceivably be a Common Redpoll; in areas where Hoaries are rare (like, the Lower 48 and Minnesota), such a bird is probably a Common.

    The yin and yang of birding, I’ve come to realize, are rarity and probability. Really, the yin and yang of humanity are rarity and probability. We humans do a lousy job of realizing that rare events are, well, rare.

    Here are two excellent, user-friendly treatises on the problems of rarity and probability:

    A Mathematician Reads the Newspaper (Knopf Doubleday)

    Innumeracy: Mathematical Illiteracy and Its Consequences (Hill and Wang)

    Both are by the great mathematics and statistics educator, John Allen Paulos (b. 1945), a professor at Temple University. Although now well into their third decades, these two books are supremely contemporary–perhaps now more than ever (“if the government can, why can’t we, right?”).

  • Ted Floyd

    Here’s another thought. In order to assess the probability that a “pale” bird (pale, short-billed, unstreaked, etc.) is a Hoary, we are critically in need of a piece of information that nobody has access to:

      the known probability of Hoary (or “Hoary”) Redpolls in the Lower 48 and Minnesota

    Let’s say that that probability is 1 in 50 in 2012-2013, but that it was 1 in 500 in 2011-2012. I have no idea; those are totally made-up numbers. But they serve to make the essential point:

      A bird with Hoary/”Hoary” characters is less likely to be a Hoary/”Hoary” in 2011-2012 than the exact same bird in 2012-2013.

    I am aware that there are those who resist that truth. What can I say?–There are lies, there are damned lies, and there are statistics. One of these days, when birders get good at statistics, there will be a revolution in how we ID birds in the field.

  • Jane Alexander

    I’m going with the science. A Redpoll is a Redpoll is a Redpoll. Thanks for this Andy and Nate.

  • Richard Klim

    It’s notable that BirdLife International/IUCN lumped the redpolls in 2012, and now recognise only one species.

  • Ted Floyd

    Hey, all. Guess I’m in a chatty mood this morning. Anyhow, one more thought, if I may.

    I agree with this statement from Andy Boyce:

      “After a thorough read I was totally convinced that attempting to sort redpolls into species, or even subspecies is a totally futile exercise.”

    I guess I worry that that sentiment could be taken too far. I hope none of us is saying, in effect, “Attempting to document and ponder variation in redpolls is totally futile.” It is fascinating that there are plain, pale, small-billed redpolls–whatever they are. It is fascinating to ponder the relative roles of geographic variation, local adaptation, assortative mating, panmixis, etc. in the Holarctic redpoll complex. I sure hope folks will continue to photograph, audio-record, and write down field notes about redpolls during this remarkable irruption of 2012-2013.

  • Andy Boyce


    Yeh, I’m with you. Documenting the variation in these birds, and better yet, searching for the reasons behind this variation is both a fun and scientifically valuable exercise. My point being that trying to put these birds in neat little boxes is not something we can do.


  • Jesse Ellis

    Hold up there! Assigning the odds of a “pale” redpoll being a Hoary is one thing. Assigning the odds for a “pale, short-billed, barely-side-streaked, zero-undertail-covert-streaked, pinkless bird” is very different. The addition of each character should reduce the odds that that the bird is a Common, and increase the odds it’s a Hoary. And they may not even be straight multiplicative (which is difficult to determine if we can’t check the genes because there aren’t any underlying genetic patterns…). Andy’s example holds – it’s useful to realize that pale individuals are far more likely to be Commons in Colorado (or even Minnesota). But that’s predicated on the point that some folks just use “pale” to ID Hoary. If someone is all characters, their odds of finding a Hoary (for themselves) are very low, but the odds of them confusing a Common for a Hoary are also very low.

  • Jesse Ellis

    Ted, totally agree with you here, and had a half-written response that got eaten by the internet. There are clearly some birds at one end of the spectrum that irrupt some years and don’t others, and that’s a very interesting part of redpoll biology.

  • Kirk Roth

    Andy, it sounds like you need to send a proposal to AOU.

    You need to address breeding behavior and assortive mating, or lack thereof. I’m not sure what, if any, papers have been published regarding this. Beware that a lack of genetic difference is not, by itself, enough to split under the Biological Species Concept that ornithologists tend to employ. For example, see the large Larus gulls, Townsend’s/Hermit Warblers, etc.

    Also, the probability argument has a number of flaws. I ennumerated some of these in another post, which got “eaten by the internet,” much like Jesse’s post. So I’ll take that as a sign to leave it be – but I’ll state that I would avoid the probability argument altogether.

  • Good food for thought, Andy, and I certainly can’t argue redpoll genetics. The AOU & ABA checklist committees have been more in the splitting mode lately but I’m sure this is one they should consider lumping.

    I don’t have papers to cite but I also recall hearing anecdotally that species pairs like Cinnamon / Blue-winged Teal and Golden-winged / Blue-winged Warblers are genetically indistinguishable (can anyone chime in on this more specifically?)

    Genetics may also help to differentiate additional cryptic species in the ABA area such as Mono Basin Sage Grouse or Pacific/Atlantic Common & King Eiders.

    A few points I’d like to make:
    1) You seem to be singling out Colorado but eBird is showing a huge hit of Hoaries across the northern tier of states this year. Conversely, Daniel Gibson reports few if any Hoaries in central Alaska this year (I think eBird coverage in interior AK is too sparse to see this but compare eBird maps for the winter of 2012/2013 to last winter to see the hit of Hoaries in the lower 48.)
    2) I doubt if you mean it but one could imply from your post that all pale redpolls are being called Hoaries. I know there is a risk of over-eager ID here but I also know that many pale redpolls in Colorado, and presumably in other redpoll-blessed states, are being rejected for Hoary based on traits like undertail or rump streaking.
    3) I’m agreed that eBird checklists can be biased towards rarity. Another possible reason for eBird over-reporting of Hoaries may be that birds on the far end of the spectrum are recognizable as individuals and counted as such. But flocks coming to feeders are probably under-counted. For example, if someone counts and reports 75 redpolls at a feeding station, there well may be a pool of 200 or 300 birds coming and going. One stand-out bird in a flock like this will be disproportionately represented if counted as 1 to 75. Color-banding studies of rosy finches and hummingbirds have demonstrated that many more birds may be around feeders than can be counted at any time, as a given bird doesn’t usually just sit on a feeder all day.
    4) On the topic of numbers, other expert birders would put the ratio of Hoaries at more like 1 in 200 or even better during irruptive years. (Yes, clearly depends much on what someone considers a Hoary to be…) Regardless, even going with the conservative 1 in 1000 doesn’t mean that you have to sort through 1000 birds to come up with a single Hoary. One could look through 5000 “commons” without finding a Hoary candidate, or one might find a few Hoaries in a smaller flock (or even a small group of only Hoaries, as mentioned in Kim Eckert’s article on redpoll ID in The Loon: If something is indeed going on with pale, unstreaked, small-billed, redpolls as it seems to be this winter, then the “traditional” ratios might not fit anyway.
    5) I would say that instead of defaulting to Common Redpoll, any uncertain bird in this complex should be called a Redpoll sp.
    6) Thus, under the current regime that splits the birds into two species, Common Redpolls are being mis-identified as well, perhaps even more so than Hoaries.
    7) Ted beat me to it while I was typing, and I know you agree, but I think to ignore field-identifiable differences in birds, whether species, subspecies, sexual dimorphism, color morphs, plumage aberrations, hybrids, intergrades, etc. is weak. Whatever their taxonomic status, Hoary/Hoary-type Redpolls definitely deserve our scrutiny and attempts to seprate, as do Common/Common-types and the swath of intermediates in the middle of the bell curve.
    8) Bottom line is that taxonomy is subjective, whatever the species concept being applied. For now AOU/ABA has the redpolls split, a controversial decision perhaps but one not just randomly decided on. Many birds aren’t easily placed on either side of the split, but I feel that those with clear traits for one or the other can be labeled as such. I agree with you that many redpolls can’t be placed in neat little boxes, and that birders (myself included) should probably be much more liberal with the spuh designation!

    Miss you here in Colorado- hope you find your way back soon, either for a visit or for another residential stint!!

    -Bill Schmoker

  • Andy Boyce


    The two arguments are making very different points. I should have been clearer. There is a scientific argument based on genetics, and the lack of correlation between genetics and morphology that indicates that these birds are not separate species. That is one reason to believe that we cannot identify these critters to species.

    There is another argument, regarding probability that says, even if you still believe there are two species, here is another reason you cannot ID them in the field, especially not in the lower 48. This argument is not linked in any way to my assertion that there is likely only one species involved in this mess.

    I’d be interested to hear about the flaws in my probability argument since I am just using Bayes’s theorem, which is not terribly controversial. Please elaborate (by PM if you want).

    We all know that species limits are a slippery subject with no good answers. That said, the arguments about assortative mating…etc are the status quo, but assortative mating should lead to genetic differentiation if there is a genetic basis for the traits being “assorted”. Genetics are a great tool that (to some degree) allow us to bypass the long, costly, labor-intensive studies of years past. Just a thought.

  • Ted Floyd

    Good stuff from Bill Schmoker, especially this:

    “5) I would say that instead of defaulting to Common Redpoll, any uncertain bird in this complex should be called a Redpoll sp.”

  • W Thomas Manders

    I enjoyed this discussion. For me I would consider myself lucky just to see a Redpoll as I have tried several times and had no luck yet. I also am not the best birder and don’t really keep a list of the birds I have seen but do enjoy the difference in the morphs of hawks especially in perfect light. I think this discussion will make many birders look much more closely at field marks to at least ID the bird with a more accurate description even if the sp is of ?. I will just take this message, birds don’t necessarily have to be of a different sub species to be of great interest and to challenge out skills in the field. We can then use these notes to look over time to see if these changes may lead to a geneticly measureable difference.
    So Research is fun although sometimes the RE part of it can be tougher than the search part.

  • Andy, I understand your frustration, but I feel the need to defend the redpolls. I think you’re viewing a problem – the definition of Hoary Redpoll is fuzzy and not supported by DNA – and reacting too far in the other direction – there’s no such thing as Hoary Redpoll – as if it’s all just random variation in plumage color.

    The variation in color is not random. Redpolls breeding in the boreal forest are ALWAYS dark like Commons. Redpolls summering on the tundra are more variable, but mostly pale. And pale color is strongly correlated with stubby bill. And pale birds are often very fluffy. And are common in central Canada in winter, rare south and east of there. That doesn’t mean every bird is identifiable, or that they are species, but there’s a lot more going on than just a blended population that ranges from dark to pale.

    There will always be a bit of ambiguity in identifying redpolls, just as there is with Thayer’s and Iceland Gulls, subspecies of Red-tailed Hawks, etc, but there is still a lot to be learned by trying.

    Cheers, David

    • Mike Patterson

      I think this is a point worth banging on as it goes to quite a few of the species (and subspecies) we claim we can identify away from where they breed. I’m talking about Fox Sparrows and crossbills; white-cheeked geese and Iceland/Thayer’s Gulls, as well as redpolls. If I go to a place where there are very pale redpolls breeding, my probability of seeing what we might call a “hoary phenotype” is high and I can say with higher confidence that I am seeing something the ABA says is listable. Away from those locations, we lack ecological information and that ecological information informs the phenotype. The statistical noise of random variability clouds any claim we might make away from the breeding grounds.

      Most of us do not go looking for Arctic rarities on their breeding ground. We puzzle over anomalous birds in our yards based not on personal experience or statistically sound biology; instead, we use paintings and photographs found among a very small sample of idealized images in an equally small sample of references. This works when sorting hawks from handsaws, but redpolls… not so much.

      Away from that breeding area, Paulson’s rule kicks in: “Common species are more common than rare species.” We birders sometimes forget that…

  • Nick Komar

    Ted, I like to argue that, in birding, rare birds are in fact common. Here is my reasoning. If you have to look through a thousand birds to find the one rarity, then the prevalence of a rarity is 0.1%. I think most birders would agree with that assessment of the definition of a rare bird. A thousand birds (on average) would have to visit your feeder before a truly “rare” one turn up. If you are birding all day, you may have to look carefully at 1000 birds (on average) and identify every one of them before you stumble upon the “rarity.” In a flock of 1000 gulls, there is always a rare one among them. Many birders who are birding all day will critically identify many thousands of birds, and thus will report numerous “rarities” during the course of a day of birding. We all know of some intense or extreme birder that seems to find multiple rarities on every outing. Thus, if a birder see’s enough birds in a day, multiple rarities are to be expected. So, rarities will be reported daily (and in fact, they are). Hence, rarities are common. This would not apply to “mega-rarities”, which would be more like 1 in a million birds.

  • Andy Boyce


    Thanks so much for your comment. Again, I want to reiterate that I am not discouraging the study of Redpolls, far from it. If they didn’t fascinate me I never would have written the post I suppose. I don’t think most birders will become disinterested in studying Redpolls if they aren’t separate species, but that seems to be what some folks are suggesting. Anyway, non-random variation in a few traits (and totally random variation for the majority of measurable traits) across a habitat gradient does not two species make. Especially when phenotypically intermediate individuals are all over the place. Thus, to call identifying the two redpoll “species” ambiguous is an understatement of the problem, I think. We don’t even know if there are two species to identify. Species limits for a ton of different taxa were identified in a time when modern genetic techniques were not available. We birders have long espoused genetics to split species, so why can’t we accept the idea that things might go the other way sometimes as well?

    Again, thanks for the comment. You might have heard this before but that book you published a while back — is amazing. Taught me a lot.


  • Ted Floyd

    Hi, all. As David Sibley notes:

      “The variation in color is not random. Redpolls breeding in the boreal forest are ALWAYS dark like Commons. Redpolls summering on the tundra are more variable, but mostly pale. And pale color is strongly correlated with stubby bill. And pale birds are often very fluffy.”

    I don’t disagree with any of the preceding. (Let me more honest: I don’t really know! I have almost zero experience with redpolls on the breeding grounds–and, honestly, not even all that much on the wintering grounds.)

    But why isn’t the scenario laid out by David just an example of what the evolutionary biologists refer to as “local adaptation”? Jim Rising–a great birder, a great scientist, and, by the way, a member of the AOU “Check-list Committee” and an all-round great guy–has written some really helpful, no-nonsense stuff for birders on this very matter. See especially his note in Birding, July 2010, pp. 12-14.

    I’m not questioning–and I don’t think Andy Boyce is, either–the existence of redpolls that tend to show a certain suite of characters, e.g., pale overall, small-billed, not very streaky. But that sounds to me like the existence of some populations of Savannah Sparrows (to cite a taxon about which Jim Rising has written a great deal) that exhibit a suite of characters that are better explained by local adaptation (i.e., to islands) than by full-species rank. Could it be the same with redpolls–are we seeing local adaptation (i.e., to boreal forest vs. tundra)?

    To go back to what I take to be David’s key point:

      “The variation in color is not random.”

    That certainly seems to be the case! But I’m not persuaded that it proves the existence of Hoary Redpolls–no more than the non-random occurrence of, say, orange-variant House Finches proves that they are a taxon of any rank.

    In some sense, we’re all saying the same thing: Redpolls showing a certain suite of traits (pale overall, small-billed, not very streaky) surely exist, and sure are interesting. But the existence of birds showing a certain suite of traits doesn’t necessarily lead us into the realm of taxonomy. This may be a non-taxonomic problem; whatever it is, though, it is fascinating.

    [David mentions Thayer’s/Iceland and Red-tailed Hawk subspecies. I definitely agree that they involve ambiguity, but the situation isn’t quite parallel to that of the redpolls. Thayer’s/Iceland are mainly allopatric, although of course with some overlap and evidently a fair bit of clinal variation; same thing, more or less, with geographic variation in Red-tailed Hawk subspecies. But the breeding ranges of Hoary and Common redpolls are heavily sympatric. I realize, by the way, that David didn’t say that redpolls are parallel to Thayer’s/Iceland and Red-tails; I’m just providing a point of clarification.]

  • Nick Komar

    Andy, thank you for bringing this issue to light. I think it is worth quoting and summarizing some of the points in Alaskan biologist Declan Troy’s 1985 paper published in the Auk. This is the paper that provides the material for David Sibley’s scoring index for Redpolls.

    The paper quotes an older study by Jehl and Smith (1970), which states that redpoll taxonomy is “one of the most perplexing problems in North American ornithology.”

    In trying to determine whether morphological features may be used to distinguish “flammea” Common Redpolls from “exilipes” Hoary Redpolls, Troy recognizes that “it has been common practice to accept species distinctiveness if hybridization is limited”. This statement makes a lot of sense thinking about Thayer’s/Iceland Gulls, Cinnamon/Blue-winged Teal, Blue-winged/Golden-winged Warbler, Western/Clark’s Grebes and other closely related species pairs. Hybrids in these pairs are not super common, and where both species coexist (sympatry), the great majority of them are in one camp or the other.

    While Troy recognizes that there are stong geographic patterns of distribution of very pale redpolls and very dark redpolls, the key observation of his study of hundreds of specimens across North America is that where and when the darkest birds and the palest birds are breeding together, the biphasic pattern of redpoll morphology (with very few hybridization-derived intermediates) disintegrates. In other words, the distribution of any one morphological character is fairly evenly distributed across the dark-intermediate-pale spectrum.

    In his conclusion, Troy states that he found “no evidence to support the specific distinction of flammea and exilipes,” and asserts that “these two taxa represent the ends of a continuum of plumage and skeletal variability. If they were once distinct species, hybridization has resulted in introgression to the extent that any species delimitation is now arbitrary.”

    These conclusions make the most sense to me, with respect to the variability I have observed among redpolls this winter here in northern Colorado. In some individuals, it is clear that some the traits are from the exilipes end of the spectrum and some traits are from the flammea end of the spectrum (in the same individual). I would even venture to claim that a majority of the redpolls here fit that description. Check out this series of photographed redpolls at Following Troy’s observations, I would speculate that the local redpolls in Fort Collins invaded from an area where both taxa (i.e. both extremes) have been breeding sympatrically.

    Troy touches on how this situation may have arisen, and why assortative mating among these taxa may still be occurring. One reasonable explanation is that climate conditions had supported a process of speciation among redpoll populations, but that changes in climate reversed this process such that now the complex is highly variable, and the remnants of behavioral differences among the populations (e.g. Hoaries breeding in tundra) lead to confusion among us simple-minded birders.

    Finally, Troy cites studies of Old World and Greenland redpolls that came to the same conclusion as his study, and thus concludes that all redpolls should be considered a single species.

  • Ted Floyd

    This is a mildly nerdy reply, but so be it. Nick says:

      “In a flock of 1000 gulls, there is always a rare one among them.”

    Under what probability distribution?… 🙂

    Assuming poisson or negative binomial (reasonable starting points, although we’d want to do some tweaking) there isn’t always a rare one among them. Indeed, there often isn’t.

    [And even if there is, there’s the entirely different problem of recognizing it as such. The bird that merely looks like a rarity isn’t necessarily the rarity; conversely, the bird that looks like a common species may be the actual rarity.]

  • Ted Floyd

    Hi, Jesse. You say:

      “Hold up there! Assigning the odds of a “pale” redpoll being a Hoary is one thing. Assigning the odds for a ‘pale, short-billed, barely-side-streaked, zero-undertail-covert-streaked, pinkless bird’ is very different. The addition of each character should reduce the odds that that the bird is a Common, and increase the odds it’s a Hoary.”

    Or maybe we’re looking at a suite of characters that reflect local adaptation, like Jim Rising’s Savannah Sparrows that nest on islands.

    There’s also Andy Boyce’s broader point that we’re sorta putting the cart before the horse. We’re saying that some birds exhibit a certain suite of characters, whereas others exhibit a different suite of characters. That’s fine, and I think it’s true. But what’s taxonomy got to do with it?

    Answer: Maybe something, maybe nothing. There are non-taxonomic explanations for certain patterns of biological variation, just as surely as there are taxonomic explanations for other patterns of biological variation.

    By the way, I’m totally fine with calling these pale, small-billed, weakly streaked birds Hoary Redpolls. It’s a useful name. But you know what I think about names and “species”… Or, if you don’t, check out Berkeley biologist Brent D. Mishler’s provocative essay, “Species Are Not Uniquely Real Biological Entities”:

  • Ted Floyd

    Hey, Kirk. Bummer about getting eaten by the internet. Wish I’d seen your longer response; I bet it was interesting.

    Realizing that I haven’t seen it, I’m nevertheless rubbed the wrong way, just a wee bit, by this:

      “Also, the probability argument has a number of flaws…but I’ll state that I would avoid the probability argument altogether.”

    Needless to say, you may be talking specifically about Andy’s argument and rhetoric, or, perish the thought, even mine… 🙂

    That’s totally fine.

    But I trust you’re not denigrating “the probability argument” in the broad sense. Probability, like evolution, is once of the central organizing principles of life on Earth. I find that there are those who resist explanations based on probability, just as there are persons who resist explanations based on evolution.

    • Kirk Roth

      Saw this comment about four years too late – but yes, of course I accept that probability is real and useful as an explanation! However, in the use of probability we need to be careful with our assumptions – and if I recall correctly, this is what Andy and I discussed a bit. To choose a single relevant example here – the argument above seems to initially define Hoary Redpolls by their genetics, yet research increasingly shows that “Hoary” is a physical (phenotypic) not genetic (genotypic) trait (ie, the argument David Sibley makes above) – indeed this is why the blog post was written in the first place. Also there is more to a Hoary than paleness or not paleness (ie the argument Jesse Ellis makes above). If, conversely, we define a Hoary Redpoll as simply a pale redpoll – then 100% of pale Colorado R(r)edpolls are H(h)oary by definition. As you can see, I’ve rather simplistically turned the Bayes Theorem on its head, just by tinkering with assumptive definitions.

      Sorry Ted, for not seeing your response earlier – but I’m happy to have relieved this undoubtedly difficult period of uncertainty for you!

      Joking, of course,

  • Andy Boyce


    Thanks for the great comment. As far as the Blue-winged/Golden-winged Warbler the latest genetic work certainly separates the two species, although we do know that the hybridize regularly. Check out:

    Lovette et al. 2010. A comprehensive multilocus phylogeny for the wood-warblers and a revised classification of the Parulidae (Aves)

    As for the teal, a quick search shows that the lasted work was done in 1999 with relatively rudimentary techniques. The got a really weird result. There was divergence between North American Blue-winged and Cinnamon Teal, BUT, South American Blue-winged Teal fell out as identical to the North American Cinnamon Teal. Weird. Very weird. Personally I would put that down to the very basic techniques they used, but who knows, I guess we have to wait on that one. Check it out here:

    Johnson, K.P. and Sorenson, M.D.. 1999. Phylogeny and biology of dabbling ducks (Genus: Anas). A comparison of molecular and morphological evidence.

  • 10-4, thanks!

  • Andy concludes this article with “In fact, the statistics tell us that it is almost certainly NOT a Hoary Redpoll”. While many pale redpolls may not be Hoaries and Hoary Redpolls are certainly less abundant than Commons in the United States, I do not believe that these facts can be used to draw a conclusion regarding the Colorado bird. If we do consider the Hoary Redpoll to be a distinct species, does this logic tell us that none of the Hoaries that show up anywhere are indeed Hoaries? Here in Pennsylvania, the ratio of reported Hoary to Common Redpolls is still quite low; is Andy saying that these redpolls are “almost certainly NOT” Hoaries? Rare birds do show up, even against what we might believe to be statistically expected…

  • Andy, your dichotomy (“leaves us with two interesting possibilities. . . 1) All redpolls belong to one single panmictic gene pool. . . . or 2) There are monophyletic (genetically distinct) groups somewhere in the redpoll complex, but we . . . cannot identify them based on morphology, plumage, or even range”) is critically flawed and ignores the most likely explanation, the one that the authors argue for in that paper: 3) Genetic lineages have yet to completely sort. Gene trees are not equivalent to species trees. After two lineages diverge, even to the species level, they may each carry genes from the ancestral species. This is called “incomplete lineage sorting,” and it is a typical pattern of recently evolved species. Incomplete lineage sorting does not mean that two species are not real. As the paper cites, Hoary and Common Redpolls have been shown to differ in physiology, timing of migration, nest habitats, diets, vocalizations, and behavior (in addition to the sometimes tricky morphological differences). If you look carefully at the results in that paper, the genetic differences even between the redpolls (collectively) and the linnets are not consistent – redpolls and linnets share some mitochondrial haplotypes and also cluster together in a “STRUCTURE” analysis of microsatellite allele frequencies. In other analyses in this paper, Common and Hoary Redpolls did differ significantly: for example, in the AMOVA tests these two taxa differed at p=0.0086 (traditionally any value less than 0.05 is considered a significant difference). These results are completely consistent with two valid, biologically meaningful species that have recently diverged from one another. The paper itself represents this situation honestly, and says that “molecular data alone should not be used for designating species status and we acknowledge that there might have been missed the relevant genetic markers to distinguish gene pools.” I’m afraid your summary here oversimplifies things and ignores the (quite likely) possibility of incomplete lineage sorting. This misleads people like Jane who read this and conclude that Hoary and Common Redpolls are not good species. But that is not what the data say, and not what the authors say.

    In your response to David Sibley, you said, “We birders have long espoused genetics to split species, so why can’t we accept the idea that things might go the other way sometimes as well?”” This also misrepresents what genetics can and can’t tell us about species, because the case for splitting based on genetics is not symmetric. If genetics tell you that two groups of birds do not interbreed, and have not interbreed for millenia, then that is a good case for them being separate species. But if genetics (especially neutral markers like those used in this study) do not show a difference, there might still be one. If physiology, behavior, morphology, vocalizations, and studies of mate choice show that two species are different, then genetics might not be necessary. And in fact, genes can be shared by two species. An excellent example is Polar Bears and Brown Bears: they are clearly morphologically, ecologically, and behaviorally different species, yet they are polyphyletic according to mitochondrial DNA. Closer to home, phylogenetically speaking, Gadwalls and Eurasian Wigeon share mitochondrial haplotypes – would you argue that they are the same species, as you have for Common and Hoary Redpolls?

    Finally, I don’t follow your math at all. You say that 1/1000 = 0.001%? No, 1/1000 = 0.1%. Then, you say that 99% x 0.001% = 1%? No, it is about 0.001%. But you started with the wrong numbers. It should be 99% x 0.1%, which is (no surprise) about the same as our original number, 1/1000. That is, if the odds of seeing a Hoary Redpoll in a flock of Commons is about 1/1000, and our odds of correctly identifying a Hoary are around 100%, then the odds of having a Hoary in Colorado are about 1/1000. Therefore it seems pretty likely that if there are a couple thousand redpolls in Colorado, at least a few are Hoaries. Then you somehow figure that the odds of detecting a Hoary Redpoll in Colorado is 50%, but conclude that “In fact, the statistics tell us that it is almost certainly NOT a Hoary Redpoll.” What? No, if your odds of correctly identifying a Hoary Redpoll are 99%, then your odds of correctly identifying a Hoary Redpoll in Colorado are 99%. No Bayesian statistics necessary.

  • Andy Boyce


    Thanks, your grasp of phylogenetics is certainly better than mine and I don’t fully understand what AMOVAs are actually doing, but I do understand incomplete lineage sorting and the complexities it introduces. The problem I see with the argument that these are incipient species that simply haven’t sorted is that there doesn’t seem to be any compelling evidence for it. Why does it seem likely that we have caught the Redpolls at this (relatively speaking) brief period in their evolutionary history where they are diverging, but not yet enough to detect through most means? These are birds that breed sympatrically, show a huge amount of morphological overlap and variation, and regularly shift their breeding grounds large distances in accordance with change in local climate. To me, the argument that they simply haven’t had the time to diverge doesn’t seem like a good one. It might be more believable in the context of geographically, or sexually isolates species but for Redpolls it doesn’t make much sense. As for their statement that they might have missed the relevant genetic markers…sure, they might have, but that is always the case with this sort of work, and short of doing a full genomic, or better yet, transcriptomic study of these guys, we won’t be able to know for sure. However, that would be an atypically high standard to hold a study to and I don’t think it is appropriate to discount the findings based on that.

    You state;

    “If physiology, behavior, morphology, vocalizations, and studies of mate choice show that two species are different, then genetics might not be necessary.”

    That is true, but it’s certainly not appropriate for this case, where to my knowledge (I think Nathan Pieplow at will be talking about vocal differences soon) there are not concrete differences in these things you mention, so genetics does give us meaningful information.

    Regarding the probability stuff. I was sloppy with my use of the % sign, my apologies. But I believe the math still stands.

    If 99% (.99) of pale redpolls are Hoary Redpolls

    and .1% (.001) of redpolls in colorado are Hoary Redpolls Bayes’s theorem (not really Bayesian stats as we usually think of them), approximately 1% of pale Redpolls in Colorado will be Hoary Redpolls. This is because that 1% of pale redpolls overall that are actually Commons end up being a much larger percentage of the total population in a place like colorado where they vastly outnumber Hoaries. Does that make sense? I admit I may be doing this wrong, but I don’t think I am. You can’t round to 100% from 99% because that’s going from a probability to a certainty, which totally changes things and misrepresents the real situation which is that Hoaries cannot be readily identified in the field anywhere close to 100% of the time.

    It’s basically like saying that 99% of the plutonium in the universe is from mars, but 1% is from Earth. If you find some on Earth, the chance that is is from Earth, and not say, an asteroid, is way higher than 1% because a much more improbable event has to occur for any of that plutonium from Mars to get to Earth.

    I may have just confused myself and everyone more. Sorry, hope that makes sense.

    Again, thanks for bringing a deeper knowledge of phylogenetics to the discussion, much appreciated.


  • Andy Boyce

    Rare birds like Hoaries (if they exist) do show up. But what probability tells us is that it may happen far less frequently than we think. Especially when we cannot separate them perfectly from a more common species. When I say “almost certainly”, I mean that literally, as in close to, but not 0% of the time. Again, sorry if I’m being unclear.


  • Andy, you said, “The problem I see with the argument that these are incipient species that simply haven’t sorted is that there doesn’t seem to be any compelling evidence for it.” If that’s the case, and it may be so, then it has nothing to do with the genetics. But many other people have looked at the evidence and concluded that there IS compelling evidence for a specific difference between these two, in terms of plumage characters and biometric measures (reviewed in Knox 1988); physiology (Brooks 1968); and timing of migration, nest habitats, diets, vocalizations, and behavior (Molau 1985, Herremans 1989). Most importantly, even the most recently split species in the group (Lesser Redpoll and Common Redpoll) do not form mixed pairs where the breed sympatrically (Lifjeld and Bjerke 1996). And although there are occasional individuals that are difficult to identify, there are no conclusive observations of hybrid offspring (Molau 1985, Knox 1988). So, perhaps none of that evidence is compelling to you, but I think it is overreaching by far to say that “we have no idea what (or even if) a Hoary Redpoll is” or that “there is actually no such thing as a 100% surefire Hoary Redpoll.”

    You say that the argument that they simply haven’t had time to diverge genetically doesn’t seem like a good one. How many Common Redpolls are there in the world? It’s obviously hard to say, but they have a huge range and are circumboreal. The thing about genetic drift is that it happens more slowly the larger a population is. So a species with a large range and accordingly large total population size will take longer to reach divergence. Redpolls are the perfect case study for a situation where incomplete lineage sorting could last a long time; millions and millions of years.

    You said that you don’t think it is appropriate to discount the findings based on not having selected relevant markers, but I’m not doing that. They did an excellent study, and their conclusions are appropriate for their data. But I’m not discounting their findings when I propose that you interpreted them incorrectly. The authors interpreted their results the same way I did.

    (A minor point: you said a transcriptomic study would be ideal, but analyzing the full genome of these redpolls would be more useful than analyzing the transcriptome because the transcriptome will be directly affected by the habitat, diet, temperature, etc., whereas the genome will not.)

    I still think your math is incorrect. If 99% of pale redpolls are Hoary Redpolls, why do you think the proportion of pale redpolls that are Hoary Redpolls should be lower in Colorado than elsewhere? Somehow I think you’re trying to mathematically show that the proportion of redpolls in Colorado that are Hoaries is lower than the proportion elsewhere. That might be the case, but I don’t see how your “Bayesian” stats lead to that conclusion. The plutonium example doesn’t fit because it requires (as the rest of your math does, I think) that we cannot identify the source of the plutonium by looking at it. If plutonium from Mars is pale and fluffy with a shorter bill, and plutonium from Earth is darker and slimmer with a longer bill, then our chance of correctly identifying the origin of a pale fluffy piece of plutonium with a short bill, regardless of what planet it is found on, is much higher than random. Probabilities become irrelevant when we have a priori knowledge to use to make the assessment.

  • I’m not nearly qualified to jump in on this, even if I will, but generally it seems that when animals – mammals, reptiles, birds, insects, probably flatworms, as a species range from open areas to increasingly dense forests they become smaller and darker. Lots of good reasons to suspect for this. Maybe the big picture helps to illuminate the smaller.

  • Jesse Ellis

    Hi Ted- I was only referring to Andy’s scenario that was “taxonomy/genetics aside”, and your response to it (second response, which I replied to). All of this could be true regardless of taxonomy, if we decide that a Hoary-type is a bird with characters x1, y1, and z1 and a Common is a bird with x2, y2, and z2.

    I am very curious about why some redpolls show x1 y1 and z1 and others don’t, and whether that reflects some particular aspect of their ecology or natural history.

  • Ted Floyd

    Hi, Ryan. Usual disclaimer: I’m wary of attempts, mine or anybody’s, at mind-reading. That said, I wonder if you’re misreading Andy’s, er, Ninety-nine Percent. The redpoll situation is analogous to the following:

      * 99% of people with a certain disease are green, and 1% of them are blue.

      * 98% of people who don’t have the disease are blue, and 2% of them are green.

      * You see a person who is green.

    What is the probability that that person has the disease?

    Answer: You need more information.

    You need to know what percentage of people have the disease, and what percentage do not. You cannot answer the question without information about the percentage of disease-carriers in the population.

    If the percentage of people with the disease is low (say, 1 in 1,000), then the chances are quite high that a green person is healthy, not diseased–even though 99% of people with the disease are green, and only 2% of healthy people are green.

    It’s potentially the same thing with Hoary/Common Redpolls, Thayer’s/Herring Gulls, and other “species” pairs where there is overlap in characters. Have you ever caught yourself saying, “If this were California, I’d call that a Thayer’s” or “If this were Barrow, I’d call that a Hoary”?

    If so, then Bayesian statistics are, in fact, required for the ID process.

  • But then I wonder how can we even determine a probability of a redpoll being Hoary if we are not always certain about the identification and/or taxonomic status!

  • Ted Floyd

    “I am very curious about why some redpolls show x1 y1 and z1 and others don’t, and whether that reflects some particular aspect of their ecology or natural history.”

    Me, too. The following is a bit oblique, but it serves our general point: For quite some time, the male and female Williamson’s Sapsuckers were described as separate species. Of course, some other factor (namely, sex) accounts for why certain sapsuckers show characters x1, x2, and x3, whereas other sapsuckers show characters y1, y2, and y3. We now that those characters correspond to “male” and “female,” not to “Species X” and “Species Y.”

    In the case of redpolls, I wonder what the x’s and y’s really are. Maybe they’re different taxa, maybe they’re not. They’re interesting, whatever they are. But I don’t think we should assume they’re different taxa, let alone different species.

  • Ted Floyd

    Indeed, and that’s the crux of the problem, if you ask me.

  • Derek

    I was unaware that there was debate over Red-tailed Hawk subspecies. Does anybody have a link to a primer on that topic? Also, I just wanted to say that, as someone who has never seen a redpoll buts hopes to next week, I’ve immensely enjoyed reading this article and its comments and this is a big reason as to why I follow this blog.

  • Are we bird watchers, or are we DNA watchers? Do we pursue birds to analyze them, or to experience their beauty? Are we scientists, or are we philosophers? I will speak as the latter.

    If you live in the west and you have seen a Red-shafted Flicker, I strongly suggest you add Yellow-shafted Flicker to your bucket list. Position yourself with the early morning sun — still heavy in its devinely golden hues — at your back, perhaps near a feeder, and wait for a Yellow-shafted Flicker to fly in head on, and watch as it flutters its wings as it brakes itself for its landing. You will under these conditions witness a shade of yellow that will vibrate your heart stings. It may cause you to drop your coffee cup out of your hands onto the ground.

    If you live in the east and are dismissive of the ubiquitous Slate-colored Junco, please put Pink-backed Junco on your bucket list. You will know, in the pure and innocent way that a child knows, that you are seeing a different bird. Watch as it forages for seed in the snow. Your socks may roll up on you!

    I suggest that somebody out there prepare a bucket list of North American birds, a list of birds that every person should be lucky enough to see before they leave this planet, of birds so beautful that should you be fortuneate enough to ever see one, you will know instantly that, if there is a God, you have at arms length just stared Him strait in the eyes.

    A Hoary Redpoll produces an entirely different human experience to the observer, and DNA aside, I believe every birder needs to see this creation of nature for himself, and count it on their life list.

  • MartinW

    Could Brent Mischler be regarded an an extremophile?

  • Thanks Andy, Ted and others for the replies,

    I wasn’t addressing the species question, just saying that a population of paler redpolls does exist at the northern edge of the range, and whether they have a name or not they can be identified. I enjoy watching and identifying them the same way I study variation in other birds, and I feel like I get meaningful information about the bird’s origins from that. As I said before, that doesn’t mean they are species (and I would probably vote for lumping if pressed) but it’s wrong to say Hoary “doesn’t exist”.

    And since the probability math keeps coming up, the idea that a pale redpoll in Colorado is statistically unlikely to be a Hoary… makes my head spin. When is a pale redpoll NOT a “Hoary”? The probability of any pale redpoll being a Hoary should be 100%, since Hoary is defined as a pale redpoll, and Troy Corman reported that ALL of the redpolls breeding in the boreal forest are dark, there is no evidence that real Commons can ever be pale. The problem is that there are lots of intermediates, so it depends where we draw the line for “pale”, but even the intermediate birds come from more northern places.

    I suspect we’re getting hung up on names and the technicality of whether we put them in two species or one. Maybe we could agree if we talked about a “pale tundra-breeding form of Redpoll” instead of calling them Hoary?

    Best, David

  • One follow-up thought on the probability just occurred to me. Here are two slightly different questions:
    1) If a person feels hot to the touch, what is the probability that they have a fever?
    Who knows? It depends a lot on the reliability of the person reporting, the range of possible errors, alternative explanations, etc.

    2) If a person has a temperature of 101 F, what is the probability that they have a fever?
    Essentially 100%.

    I think you are posing the “Pale Redpoll” probability question like the first example, and I’m interpreting it like the second example.

  • I find the argument that because something cannot be defined to be 100%, and therefore it doesn’t really exist, to be very black and white and misleading…..this is how this reads to me. This is to “throw the baby out with the bathwater” . As David says, “when is a pale redpoll NOT a Hoary”. As Kenn Kaufman has written about this complex, we might very well be underreporting this species….or something along those lines. With the information we have, there are VERY clearly pale redpolls that easily fit what a Hoary is…..and there are intermediates that can be argued over, and are probably best left as unknowns.

  • Great way of putting.

    Matt Young

  • Just a minor quibble, Kirk. The Townsend’s/Hermit example is not comparable to redpolls. Larus gulls are a much better comparison. There are distinct genetic differences between the “pure” populations of Townsend’s and Hermit Warblers. The genetics are only mixed up in the genetic “wake” of a hybrid zone that has slowly been moving south. It’s pretty clear what’s going on with the two species, but we have no idea what the scenario is with the redpolls because we cannot yet discern between panmixia, secondary contact, etc. Thankfully, the redpolls and other complexes of very young species with unclear genetics will be much easier to understand genetically with next-generation DNA sequencing techniques. It’s only a matter of time until someone gets the money to look at these groups with this approach…

  • Ted Floyd

    As it turns out, Brent Mishler (no “c”) and I were in touch just this morning. I should probably let Prof. Mishler speak for himself (Brent, you out there?), but he reminded me (in a context distinct from this redpoll debate) that, taking the broad view, it is the ornithologists who are, well, the extremists about Ernst Mayr’s Biological Species Concept.

    Talk to folks in botany, entomology, and bacteriology (you know, organisms that play massively dominant and pervasive roles in life on earth), and you soon start to realize that lots of smart and informed people regard debates about species to be akin to debates about unicorns.

    The concept of the “species” is powerfully entrenched in our western culture, no question about that. But so are unicorns; just got 71.7 million Google hits.

    I totally agree with David Sibley and others that pale redpolls are worthy of our study. And I don’t mind calling them Hoary Redpolls.

  • Hi, Ryan. Your interpretation of the AMOVA results is not correct. The only statistically significant difference in the AMOVA results was between Lesser and Common Redpoll, and only for mtDNA. The 0.0086 value is the FST value for the Common/Hoary microsatellite comparison, not a p-value.

    Also, the clustering of Linnets and a few redpolls in the STRUCTURE analysis occurred because they used the ‘no admixture’ model. This is a huge assumption and is the reason most researchers always use the ‘admixture’ model when doing STRUCTURE analyses. The ‘admixture’ model analysis supported K=1, meaning there was no significant population clustering among the redpolls. I’m really surprised they didn’t show this graphically. To be frank, I think their STRUCTURE analyses of the microsatellites were not conducted well, and the results could have been made much clearer.

    Your point about incomplete lineage sorting is absolute correct, of course. If redpolls are just very young, incipient species, it’s certainly possible for microsatellites to not yet show population clustering. However, in general, microsatellites have an excellent ability to separate populations that are extremely young, and I don’t see a strong reason to doubt that there is a large amount of gene flow occurring among the redpoll “species.” But perhaps it is speciation with gene flow? This has certainly be shown to occur between young, ecologically distinct species. Maybe that’s what is happening with the redpolls. That’s where further study in the areas of sympatry is needed (and some next-gen DNA sequencing to go along with it wouldn’t hurt).

    Anyway, I think your main point about Andy’s dichotomy being flawed is correct because it doesn’t account for incipient species or speciation with gene flow. I prefer the dichotomy presented by the paper’s authors:

    Two major alternative interpretations exist. Either redpolls form a single gene pool with geographical polymorphisms possibly explained by Bergmann’s and Gloger’s rules, or there are separate gene pools of recent origin but with too little time elapsed for genetic differentiation to have evolved in the investigated markers. Future studies should therefore examine whether reproductive isolation mechanisms and barriers to gene flow exist in areas with sympatric breeding.

    (Although they also don’t acknowledge the possibility of speciation with gene flow.)

  • Ted Floyd

    Hi, David.

    “When is a pale redpoll NOT a ‘Hoary’? The probability of any pale redpoll being a Hoary should be 100%, since Hoary is defined as a pale redpoll[.]”

    I didn’t know that. Honestly, I don’t know a lot about redpolls. (Am enjoying the learning experience–here at The ABA Blog, as well as on the listservs, Facebook group pages, etc., with great photos and discussions.)

    Anyhow, wouldn’t you say that your question can be answered in the affirmative for, say, Blue-headed and Cassin’s vireos? Haven’t you ever found yourself in situations like this?–

      “Wow. That vireo is fairly bright, colorful, and contrasting. If I were in New York, I would call that a Blue-headed. But I’m in California, so the standard of proof is higher. Maybe it’s just an unusually bright, fresh Cassin’s.”

      “Wow. That vireo is surprisingly dull, drab, and fuzzy. If I were in California, I would call that a Cassin’s. But I’m in New York, so the standard of proof is higher. Maybe it’s just an unusually dull, worn Blue-headed.”

    Such assessments, which I think are reasonable and defensible, are “Bayesian.” We have prior knowledge about relative probabilities, even if they are coarse (“Cassin’s is much more common in California than Blue-headed” or “Cassin’s is exceedingly rare in New York”), and we base our IDs, in part, on those relatively probabilities.

    To go back to David’s actual question, I am, again, at a disadvantage here because I don’t have much of a feel for the relative abundance of Hoary vs. Common in, say, northern Alaska vs. Pennsylvania. But if (a) there is a difference in relative probability of occurrence and (b) there is some (non-trivial) overlap in characters, then the Bayesian approach is correct. However, I realize that David is pretty much putting the nix on (b) because “The probability of any pale redpoll being a Hoary should be 100%, since Hoary is defined as a pale redpoll[.]”

    Nevertheless, there are situations (“Solitary” Vireos in Calif. vs. New York, Thayer’s/Iceland Gulls in Calif. vs. New York, silent wood-pewees in Calif. vs. New York) where the Bayesian approach is correct.

  • Ted Floyd

    These are good examples, and they’re not merely semantic.

    An example of your #2 would be, say, adult male breeding-plumage American Redstart vs. adult male breeding-plumage Black-throated Blue Warbler. They’re in the same genus (unless Irby Lovette is up to more mischief…), but the probability that they are what they appear to be is essentially 100%. Even if you happen to see one in Holland or Korea. I totally get that.

    An example of your #1 would be, say, a Cassin’s Vireo in New York or a Blue-headed Vireo in California. They’re in the same genus, but it’s not as easy as the two warblers in the same genus. Our assessments depend on molt, wear, age, and, importantly, relatively probability of occurrence.

    You’re right. I’ve been thinking of “pale redpoll” as an example of your #1, not your #2. But I am, as always, open to new ways of thinking.

  • Cool post, Andy, and interesting discussion, everyone.

    At the risk of being uncomfortably anthropocentric:

    1. There are darker-feathered redpolls and lighter-feathered redpolls. They tend to live in different regions and usually (but not always?) mate with their own kind. But there are a lot of intermediates, and they can’t be easily genetically separated. We call them different species.

    2. There are darker-skinned humans and lighter-skinned humans. They tend to live in different regions and usually (but not always) mate with their own kind. But there are a lot of intermediates, and they can’t be easily genetically separated. Nobody would dispute that we are all the same species…

    If we apply the same standards to redpolls that we use to define ourselves, I don’t see the logic of splitting them – even if some individuals are recognizably hoarier than others.

  • Good call about the p-values vs. PhiST; you’re right that I misread that table. I should add that I appreciate Andy’s bringing genetics into the discussion, and his very provocative post. Thanks, Andy.

  • Olivier Barden

    I have three points to make on this topic:

    1- The fact that phenotypically distinguishable taxons don’t show a proportional amount of genetic divergence is not new (think large white-headed gulls). I believe these are two good biological species in the sense that they mate assortatively in zones of sympatry, only this behaviour is too recent be reflected in their genetic heritage.

    2- Your assertion that a ton of intermediate individuals are being seen isn’t accurate. Redpolls are tough. Like many other species, there is likely a bias towards young/females being more frequent at the southern end of the distribution, with adults/males being more frequent at the northern end. So right now, lots of birders inexperienced with redpolls are getting a couple of Hoaries, most of them being the tough, rather “intermediate” types, while the textbook pale birds are extremely rare. I can imagine it must be hard to learn in such conditions. I see Hoary Redpolls every winter in south-central Quebec, usually in a ratio of 1:100. If I didn’t know how to pick out the tough birds, this ratio would be more like 1:400. The redpolls have moved so far this winter that the ratio of HORE to CORE around here is now more like 1:25, and more Greenland Common and Hoary Redpolls (taxons rostrata and hornemanni) than I have ever seen.

    I’ve now seen several hundred hoaries over the past 18 years, so I feel quite comfortable in IDing them, regardless of plumage. Also, I have banded both species of redpolls, so I now look to age and sex them in the field whenever possible. While HY/SY birds may ressemble Commons quite a bit, I have never seen a truly intermediate looking adult redpoll (look for relatively rounded and fresh tips to the rectrices, with bold white edging).

    3- I was fortunate enough to conduct surveys in a zone of sympatry between the two species in northern Québec in July, where the ratio was 2:3 in favor of Common. Again, no intermediates, and mating was evidently assortative despite the widespread overlap in habitat use. Some family groups involved a mix of both species, the adults being easy to identify (the juveniles, less so, though hoaries were consistently shorter-billed and fluffier despite the warm weather).

    Olivier Barden

  • Most interesting, Oliver- good to hear from someone in redpoll central!! Thought-provoking data to know that your Hoary ratio is way up this winter.

  • Given what Olivier said above (good to see you chime in bud), I’m sure he’s seen all 4 NA subspecies in one flock or on the same day several times this winter? I had all 4 NA subspecies in my yard this past Sunday. Between my house and work at the Lab of O (central NY), I again saw all 4 NA subspecies Wednesday. As much as the big redpoll story is about what’s going on in Colorado (more records this year than in the past 25 years combined?), this redpoll invasion is huge here in the Northeast and Great Lakes……seeing several Hoaries in a day is becoming common — the same appears to be happening in the Great Lakes as well. As for data here in central NY, I had 3 Hoaries in a flock of 40 redpolls. At the Lab of O there’s 2 Hoaries in a flock of ~100 redpolls. Small sample size, but that’s 5 Hoaries in ~140 birds. 1:28

    Matt Young

  • Roger that, Matt, excellent to hear.  'Twill be a most interesting tally to review when the irruption finally ebbs!!

  • Ron Pittaway

    The papers cited leading to this discussion are not new. The most recent was published in 2008. Regardless, the topic of revising redpoll taxonomy is not on the AOU’s agenda which is posted on their website. So we will continue to have two species, Common Redpoll and Hoary Redpoll, each with both a “southern” and a “northern” subspecies. The two larger northern subspecies, the “Greater” Common Redpoll (rostrata) and the “Hornemann’s” Hoary Redpoll (nominate hornemanni), are quite distinct from each other with very few known hybrids, even though their ranges overlap. The distinctiveness of these two northern taxa is another reason not to lump redpolls.

    Having watched redpolls for many years in Ontario, I agree with Olivier Barden of Quebec and Matt Young of New York State that identifying Hoary Redpolls is not that difficult given some experience. You also will see more Hoaries and fewer “intermediates” once first year birds are recognized. For new birders wanting to see a Hoary I suggest studying them at feeders because seeing redpolls well in natural habitats is usually difficult. They are flighty and constantly moving while feeding in trees and weedy fields. For a countable Hoary, the combination of a paler bird, white rump, lightly streaked sides, very lightly marked to immaculate undertail coverts, in adult males the paler and less extensive pink on the breast, and most importantly a stubbier more obtuse bill shape (pushed in face appearance) should make it a Hoary.

    As for seeing all four subspecies (The Redpoll Challenge) on the same day or at the same time, Jean Iron and I had this experience several times at our feeders in Toronto in March 2011. So far this winter we have a regular Hoary (exilipes) with about 30 Commons and no intermediates. Numbers at feeders should increase through February and March as natural food supplies diminish and as more redpolls move south and as they become more dependent on nyger seeds at feeders.

    Ron Pittaway

  • Just to defend “genetics” a bit. Well I’m a birder, and was, as you did, looking for some literature about this redpoll “complex” 😉 I’ve seen this paper you cite here, but to be honest there is nothing surprising is these paper, especially the one you cite at the beginning. If these differences between redpolls species are quite recent, then the actual genetic markers used by the scientific community will not detect anything. The actual genetic markers do need a complete reproductive isolation of both species/subspecies for quite a long time (we are speaking about thousand years at least) and everything depends on the size of populations and species characteristics!!!
    So unless people are going into genomics, genetic is not gonna show anything…. Unfortunately, this does not mean that they are actually not different subspecies….

  • Hi, Chris. This isn’t quite correct. If there’s enough variation present in microsatellite loci, which they used, you can see genetic clustering in much less than thousands of years. Some studies have actually shown genetic separation (and merger) in only a handful of generations. But I agree that next-generation genomic sequencing is what will be needed to reach a conclusion that people might actually all agree on. These examples of extremely recent speciation are perfect for exploring with next-gen sequencing.

  • Andree Dubreuil

    I do not know much about Common Redpolls, but in 2011, I had an irruption…
    no one cared,
    Today after 7 days of Irruptions,, the galup was more than 1200, very well documented on amateur video and pictures..
    It was wonderful..
    Thanks for being there,, I guess, I just wanted to share
    Andree Dubreuil
    Trois Rivières, Quebec
    Andree birds carbonari dubreuil on Facebook

  • andree birds carbonari dubreuil

    And,, I am not a PHD.. but a Redpoll is just about the only bird I know…. and they remembered where I lived,, after two years.. and stayed for brunch and diner for 10 days,,, and not one of them got stunned on my patio windows…. and guess how many came to the snacks..

  • andree

    did anyone see the Photo of the day on National Geographic,, on Redpolls in Alaska, at the feeder…

  • john lofgreen

    What about the Lesser Redpoll in Europe? Separate species or not?

American Birding Podcast
Birders know well that the healthiest, most dynamic choruses contain many different voices. The birding community encompasses a wide variety of interests, talents, and convictions. All are welcome.
If you like birding, we want to hear from you.
Read More »

Recent Comments




ABA's FREE Birder's Guide

If you live nearby, or are travelling in the area, come visit the ABA Headquarters in Delaware City.

Beginning this spring we will be having bird walks, heron watches and evening cruises, right from our front porch! Click here to view the full calender, and register for events >>

via email

Enter your email address to subscribe to this blog and receive notifications of new posts by email.

Follow ABA on Twitter