A new paper confidently claims that there are four giraffe species rather than one, but I’m not so sure

The giraffe, Giraffa cameleopardalis, was first described by Linnaeus, and gets its species name from its fancied resemblance to a hybrid beast (as Wikipedia notes, the name comes from the Greek καμηλοπάρδαλις” meaning “giraffe”, from “κάμηλος” (kamēlos), “camel” + “πάρδαλις” (pardalis), “leopard”, due to its having a long neck like a camel and spots like a leopard). It’s always been considered a single species, but divided into about a half dozen subspecies that live in different areas and are distinguishable by different patterns of reticulation in their coats. Here’s an old subspecies designation and map; note that the populations included in each of the six subspecies live in different areas:


Here’s a classification of nine subspecies based on pattern (the number of named subspecies has been between four and about nine (I haven’t searched extensively).


Note that this classification is more or less arbitrary because the populations are geographically isolated and so one can’t use the classical “biological species definition” (BSC), in which members of the same species are able to interbreed in nature and produce fertile hybrids, while members of different species, when present in the same area, either do not mate with each other, or, if they do mate, do not produce hybrids that are fertile. Note that to use the BSC, putatively different (or identical) species have to be “tested” when living in the same area (“sympatric”). If they do not encounter each other in nature, there’s little you can do to apply the BSC.

One way around this is to hybridize them in zoos. If different “subspecies” do not mate with each other, or can’t produce fertile hybrids when they do mate in captivity, they’re almost certainly unlikely to do so in nature, and can be considered members of different species. However, if two different types do hybridize and produce fertile offspring in captivity, that doesn’t mean they’re members of the same species, for in nature other “isolating barriers”, like different breeding times or a genetically-based aversion for mating with other types, could keep them genetically separated even though barriers could break down in the artificial environment of zoos.

Ligers“, for instance, are hybrids between male lions and female tigers, made famous in the movie Napoleon Bonaparte. Lions and tigers will mate in captivity, and some liger hybrids are fertile. But we don’t consider lions and tigers to be members of the same biological species because, when they were living in the same place in nature—in India, though they no longer are “sympatric” there—no hybrids were produced.

Allen Orr and I discuss the rationale for using the BSC for investigating how species originate in Chapter 1 of our book Speciation (2004), and conclude that if you want to understand “the species problem”—why animals and plants in a given area divide into clumps and are not simply a schmear)—the BSC is the concept to use.  But if you merely want to name species, rather than understand how discrete cluster originate, then you’ll have to engage in a subjective classification if those groups don’t live in the same area.

This is the problem with  a new paper in Current Biology by Julian Fennessy and coauthors (reference and free link below). This paper tries to determine the number of giraffe species that really do exist, but since the subpopulations live in different places and don’t encounter each other (see below), they’re forced to use another criterion to determine the number of giraffe “species”. They use genetic divergence, as measured by the amount of sequence difference in the DNA.

Fennessy et al. sequenced eight genes: 7 nuclear genes and the mitochondrial DNA (effectively a single gene) in nine previously-described subspecies of the giraffe. Phylogenetic analysis of the sequences, shown below, showed that populations fell into four distinct clusters, which they decided to call “four species”. Here are the new species they distinguish and name:

  • southern giraffe (Giraffa giraffa),
  • Masai giraffe (G. tippelskirchi),
  • reticulated giraffe (G. reticulata)
  • northern giraffe (G. camelopardalis), which includes the Nubian giraffe (G. c. camelopardalis) as a distinct but related subspecies.

Here’s the phylogenetic analysis (turn sideways); the colors of the populations correspond to colors on the map below (note again that the subspecies do not inhabit the same areas, though they may have in the past; we just don’t know!) Note too that each species is “monophyletic”, containing all the individuals that descend from a single common ancestral population.


Here are some of the populations they sampled; the colors correspond to the dots in the phylogeny above.


(From the paper): Figure 1 Distribution and Sampling Locations of Different Giraffe Subspecies in Africa (A) Distribution ranges (colored shading) provided by the Giraffe Conservation Foundation [7], plotted on a map of Africa (http://www.naturalearthdata.com/). Circles represent sampling locations; for coding, see Figure 2. (B) Enlarged view of the South Sudan region. Note that the samples of the putative Nubian giraffe were taken west and east of the Nile River.

Finally, the authors do a genetic “STRUCTURE” analysis, which assumes the presence of different groups (varying how many groups they preconceive) and tries to see, using genetic differences, how distinct those groups.  The assumption of both three and four groups (the latter corresponding to the four species they name) were the most useful for giraffes. At the top of the figure below you can see how well the clusters are demarcated with a K (group number) of four, and how the demarcation becomes marginally worse when you assume five clusters.

Also, below the cluster diagram is a DNA-based estimate of giraffe “species” divergence times calibrated from the fossil record of other mammals. You can see that the four groups diverged from each other between 1 and 2 million years ago.


So everybody seems to be happy with the conclusion that we have four species instead of one.

Everybody, that is, but Matthew Cobb and I.  But more on that in a second.

The breathless conclusion that we have four species was accepted not only by the journal Current Biology, which published the paper (note the title), but by major news outlets, including the New York Times, the BBC, Science magazine, and so on. I haven’t found a word of criticism of this conclusion. So here’s some.

Dividing up geographically isolated groups into species is an exercise in naming and classification rather than understanding the problem of why nature is discontinuous—and by “discontinuous”, I mean the answer to the fascinating question “Why, in one area, are animals and plants divided into discrete groups rather than forming a continuum?” Using a genetic-distance measure, as the authors did in this paper, simply tells you how long the groups have been separated rather than whether they’d remain discrete were they to meet in nature. It tells you about evolutionary history but not about speciation.

So yes, we know the four groups they call species have been separated for 1-2 million years, and have developed coat color differences. But human “races” have also developed coat-color differences (and other genetic differences) over a period of about 60,000-100,000 years, also largely among geographically isolated groups. Now human populations cannot be distinguished from their DNA sequences nearly as unambiguously as can these giraffe subspecies, but if you gave them another million years of geographic isolation (now impossible because of travel), they might well have separated genetically to the same extent as these giraffes. Would you then be happy to call human populations different “species”? What if a human population became marooned on an island for a million years, and changed its hair color and became genetically distinguishable using sequencing of largely neutral sites. Would you be happy to say, unambiguously, that here was have a new species: Homo islandensis?

Also note that genetic STRUCTURE analysis of human populations shows a pretty neat division into six groups (Rosenberg et al. 2002), corresponding well with 5 geographic regions. (Using other assumptions of group numbers, with K =  2, 3, and 4, doesn’t give as clean a resolution, as shown by areas with mixed colors). The groups are from Africa, East Africa, Europe + Middle East, Central/South Asia, Oceania, and the Americas.


Do we have five human species here, based on genetic clustering? How much genetic difference would be sufficient to call them different human species? (These problems are all discussed in the Appendix of Speciation).

The correlation of genetic differences with geographic isolation, of course, reflects the fact that populations that don’t get a chance to exchange genes become more and more different through time via natural selection and—probably the major force in this case—genetic drift. But how much difference do we need to diagnose species? It’s arbitrary, since populations can become “reciprocally” monophyletic for even a single gene. Okay, so how many genes do you need?

So the designation that there are four genetic clusters of giraffes is sound, but do they correspond to what evolutionary biologists call “species”? Who knows? We don’t know whether these groups are reproductively isolated. I can find no evidence that these groups ever lived in the same area of Africa at breeding time—or any time. Further, they can apparently hybridize in zoos (see reference 20 for THE GIRAFFE STUD BOOK; the fertility of hybrids isn’t mentioned). So while these are “genetic clusters”, we have no idea whether they’re biological species.

Matthew in fact was quoted in the BBC piece from an email he sent them, but they left out his caveat about reproductive isolation. Here’s what the article said:

Matthew Cobb, professor of zoology at the University of Manchester explained that the “four groups of giraffes had “been separated for 1-2 million years, with no evidence of genes being exchanged between them”.

Here’s what the BBC chose not to quote from Matthew’s email (reproduced with permission):

Defining species is difficult, and neither number of genetic differences between two groups, nor the time they have been separated, are necessarily relevant. The key point is what is called the biological species concept – does mating between these groups produce fertile offspring? If not, they are species. If they can produce fertile offspring, then no matter how long they have not done so in the wild, they are still not true species. However, this technical aspect is probably the least interesting point about this study. Instead, we should be focusing on developing appropriate conservation strategies for the four groups identified in this study. Whatever we call them, these four groups have distinct genes and have had distinct histories for over a million years.

I agree almost completely with Matthew, though I have one quibble. Yes, Fennessy et al. have identified groups that we can use for conservation purposes—if we want to conserve the phenotypic and genetic diversity in what was once considered a single species. But do we want to do that? This question is never really discussed. ‘

If we’re trying to save the coat color genes, well, they’re probably segregating at low frequency in other “species” of giraffes, and we could reconstruct any extinct pattern through artificial selection. If we’re trying to save the DNA diversity, most of which probably involves nucleotide bases whose differences among the “species” is of no biological consequence, why are we trying to do that? My point here is that—and I may be missing some literature—is I have never seen a detailed and critical discussion of what exactly conservationists are trying to accomplish when they decide to preserve populations X, Y, and Z.  As Dick Lewontin pointed out long ago, a single fertilized female within a species contains half of all the “additive genetic variation” (selectable genetic variation for various traits) of the entire species, and that kind of latent variation is probably present in several or all of the four “distinct” giraffe species. When we say “we have to save all four giraffe species”, we don’t consider what we’re trying to save? The populations themselves? The phenotypic differences that distinguish them? Or the genes that distinguish them? (My own view is that we should save everything just out of respect for animals and plants that were here before us.)

I’ve been hard on this study because I study species using the BSC, but I want to finish by saying out that the study of Fennessy et al. is indeed very good, except for its certainty that we have four “species”. They don’t discuss the difficult problem of what, exactly, a “species” is—something that Matthew summarized in one paragraph. That aside, Fennessy et al. have done great work. They have identified four distinct genetic clusters, which tells us about the evolutionary history of these groups and may ultimately give a key to their degree of reproductive isolation if the giraffes move around (perhaps due to climate change). It also gives us an idea of which groups were geographically isolated for long enough to allow such genetic differences to accumulate, and thus raises questions about biogeography of these populations.

But four species for sure? I can’t say. And I’m surprised that the major science journals and newspaper sections have accepted the authors’ conclusions uncritically.


Fennessy, J. et al. 2016. Multi-locus analyses reveal four giraffe species instead of one.  Current Biology 26:1-7, online doi.org/10.1016/j.cub2016.07.036



  1. colnago80
    Posted September 11, 2016 at 10:17 am | Permalink

    It is my information that female ligers are fertile and can mate with either male tigers to produce viable ti-ligers or male lions to produce viable li-ligers.. However, as I understand it, male ligers are sterile and cannot mate with female ligers to produce viable offspring.

    • Posted September 11, 2016 at 10:19 am | Permalink

      Since one sex is fertile, that allows for gene exchange between the species that, if the groups were sympatric, would probably merge them, but also make them considered members of the same species.

      • Posted September 11, 2016 at 10:34 am | Permalink

        Today, there is only a small region in India where their ranges overlap. I suppose, however, that the hybrids have reduced fitness (even if this is not evident in captivity) and that the absence of reported wild ligers reflects prezygotic isolation mechansms already in place.

  2. Posted September 11, 2016 at 10:40 am | Permalink

    Matthew said “…Does mating between these groups produce fertile offspring? If not, they are species. If they can produce fertile offspring, then no matter how long they have not done so in the wild, they are still not true species.”

    I think this needs to be stated more carefully to avoid confusing the public. Would mating actually occur regularly when the groups are in contact? It may well be that IF they mated, the offspring would be fertile, but they may never naturally mate, as Jerry mentioned.

    I work with orchids, which produce perfectly fertile offspring even between genera, if they are forced to mate. Mating doesn’t occur in nature because the species use different pollinators.

    Also, if there are occasional fertile hybrids, this is not necessarily enough to destroy the integrity of a species. There is an entrenched belief among population geneticists that even one migrant per generation is enough to homogenize two populations, but this belief is based on some mathematical misunderstandings in pop gen.

    From the range maps, it does look to me like there are zones where three of the putative giraffe species are in close proximity. I don’t know the geography of the area but it does seem like there might have been recent opportunities for cross-breeding. If so, the distinctiveness of the clusters argues for absence of significant interbreeding. It would have been nice to see the authors of the giraffe article discuss the paleogeography of the area; if they could convincingly argue that that these populations have been in contact in the recent past, this would have strengthened their argument.

    • Posted September 11, 2016 at 11:08 am | Permalink

      What you have is pollinator isolation, the subject of a large section in Speciation. That itself is a genetically based form of reproductive isolation. And even if fertile hybrids are formed, as in some birds, the intermediate phenotypes of those hybrids themselves cause reproductive isolation, in the case of orchids probably because they don’t attract any pollinators.

      And yes, we talk about “intermediate” levels of gene flow in that book as well. This post was simply not long enough for me to cover all those bases. But given that we don’t know anything about the “mating opportunities” of giraffes in the wild, it’s not kosher to speculate, as you have, that “this argues for absence of significant interbreeding when they’ve had the opportunity.” And if they had that evidence, then yes, they’d be on firmer ground.

      I can’t write a chapter on the Biological Species Concept in one website post; readers who want to hear the arguments in full are urged to read Chapter 1 and the Appendix of Speciation by Coyne and Orr.

      • John Harshman
        Posted September 11, 2016 at 2:15 pm | Permalink

        So what’s really needed in order for two populations to be two species is that either hybrids never happen or that they have greatly reduced fitness compared to non-hybrids of either population. And the definition of “greatly reduced fitness” would be whatever would fail, in complete sympatry, to result in a merging of the two populations.

        Incidentally, there are, at least in some groups, ways to test species status other than direct contact. In passerine birds, for example, it’s considered good evidence if birds of one population do not respond to song playbacks of the other. I would suppose that genital shapes in some groups of flies would be another.

        • Achrachno
          Posted September 11, 2016 at 6:37 pm | Permalink

          Or perhaps coat pattern in giraffes. Biological isolating mechanisms can be almost anything.

          • John Harshman
            Posted September 11, 2016 at 7:52 pm | Permalink

            Perhaps, but I doubt it. It would help if there were good evidence that some feature is important as an isolationg mechanism in related species before assuming it in other species.

            • Posted September 12, 2016 at 8:32 am | Permalink

              I do that when describing new orchids. Between- population variation in parts not related to pollinization count much less than variation in the size and structure of the column and lip (the “genitalia” of orchids).

  3. µ
    Posted September 11, 2016 at 10:48 am | Permalink

    To address natural interbreeding, it will of course help to sample transects across contact zones (I would guess this is discussed somewhere in Fennessy et al).

    Note that two of the new species (reticulata, tippelskirchi) are sampled each at only one locality, and tippelskirchi is sampled in only one apparently isolated population within its fragmented range (per map distribution). The apparent monophylies/STRUCTURE could perhaps be due to local inbreeding; and consequently, some admixture may emerge with more comprehensive, geographically representative sampling.

    All good information to request more research funding.

  4. Christopher
    Posted September 11, 2016 at 11:19 am | Permalink

    Just putting my two cents in so PCC can know that posts like this are appreciated, if not fully understood (which is not an indictment of the writer’s ability to explain but rather of the reader’s intellect, that is to say, me not too smrt). I do enjoy these posts more than any other, as they are the reason I started reading this site in the first place (come for the science, stay for the cat videos).

    I would like to know more about what current conservation biology hopes to accomplish. The goals have changed over time, for sure, but then at one time some biologists were certain we should gun down predators to protect their prey. I can see the logic in trying to protect as much genetic variability as possible, in light of climate change, habitat destruction, poaching…all the usual issues, plus a bigger reservoir for future evolution, even if lumping the 9 into 4 species or splitting them into others isn’t the right method. At least if we protect as much as we can, then in the future, with better understanding, have more to work with. Better safe than sorry, I suppose, as it’s not so easy to un-extinct an species or subspecies. I guess that would be my highly uneducated, primarily emotionally-based opinion/answer to the question posed by PCC above:

    “…Fennessy et al. have identified groups that we can use for conservation purposes—if we want to conserve the phenotypic and genetic diversity in what was once considered a single species. But do we want to do that? This question is never really discussed.”

  5. BobTerrace
    Posted September 11, 2016 at 11:26 am | Permalink


  6. Posted September 11, 2016 at 11:34 am | Permalink

    Not just you and Matthew, Jerry— me too. I saw the media coverage and thought immediately, “piffle”. Others have argued that giraffes are several species on the same basis– that a certain amount of genetic difference=species. There was even an unfortunate passage in Carl Zimmer and Doug Emlen’s admirable evolution textbook extolling the alleged species richness of giraffes (I don’t know if this survived into the second edition– I hope not). This is just a reversion to the old typological species concept, except now genes rather than morphology are used. Adopting such a species concept would make the study of the origin of intrinsic reproductive isolating barriers not the study of speciation, and we’d have to call it something else I guess. As you rightly note, under this species concept there would be several species of extant Homo. But making Homo several species would completely obscure the actual population genetic dynamics of human populations.

  7. jaxkayaker
    Posted September 11, 2016 at 11:37 am | Permalink

    Perhaps the authors and journal editors felt the claim was justified by a phylogenetic species concept, not the BSC, which is out of favor among many biologists. They should have been explicit about the premises of and justification for the claim.

    Good critique, Jerry. Thanks for taking the time to write it.

    • Posted September 11, 2016 at 11:50 am | Permalink

      The phylogenetic species concept is no longer in favor by any biologist that I know of, as it’s ridden with problems. The alternative that’s being used by many is the ‘evolutionary’ species concept, which lacks any explicit connection to either reproductive isolation or phylogenies. But again, I refer you to Chapter 1 and the Appendix of Speciation, which talks about this.

      In fact, as I say in that book, every biologist who is setting out to study the PROCESS of speciation relies on the BSC and reproductive isolation.

      And the BSC is not, as far as I know, out of favor among evolutionary biologists. It’s only out of favor–and has always been–among phylogeneticists, some of whom have explicitly said that “We cannot allow species concepts to get into the hands of those in other areas of biology.”

      • daveyc
        Posted September 11, 2016 at 12:42 pm | Permalink

        Would you mind providing a reference for that quote so that we can assess the context?

  8. Hempenstein
    Posted September 11, 2016 at 11:46 am | Permalink

    This is a long post that I don’t have time to fully process now, but I want to throw this out since a cursory scan suggests it hasn’t been covered:

    Looking at the map, it looks like the green blue populations are close to the equator. And green & blue seem to be the earliest. Is there any thought that these represent descendants of refugia during an Ice Age, and that the others radiated out during an Interglacial, with habitats changing in-between the two climatic extremes and forcing the blue/green into isolation from the red/magenta (with some other explanation for yellow)?

  9. MATT
    Posted September 11, 2016 at 11:55 am | Permalink

    Fascinating…my only contribution is that the movie is Napoleon Dynamite.

  10. Avis James
    Posted September 11, 2016 at 11:59 am | Permalink

    Jerry, thanks for the wonderful morning read. This will definitely go into my evolution lectures!

    • Posted September 11, 2016 at 12:24 pm | Permalink

      Don’t tell them there are four giraffe species!!!!

      • Avis James
        Posted September 11, 2016 at 1:14 pm | Permalink

        Jerry- Don’t worry! I will tell them this whole story! An example of mis-interpreted science getting wrapped up in popular press. BTW- they will already know the flaw in defining species by genetic clusters from my lectures. It is a good story to add on. I a sure they have read the giraffe story, now they can apply critical thinking.

        • ladyatheist
          Posted September 11, 2016 at 2:44 pm | Permalink

          It’s also an excellent example of selective quotation from a ‘source.’ This is why I advice students to go to the bibliography of an article that they consider authoritative and really read the sources that have been quoted to be sure the author isn’t so bent on telling a story that they tell the wrong story.

  11. Posted September 11, 2016 at 12:03 pm | Permalink

    Something is off with this:

    “northern giraffe (G. camelopardalis), which includes the Nubian giraffe (G. c. camelopardalis) as a distinct but related subspecies.”

    G. camelopardalis ssp. camelopardalis is the subspecies that goes with the type of the species. If you’re trying to refer to “a distinct but related subspecies,” (an unclear statement when made with reference to the species as a whole) presumably you should be listing the one that is NOT G. c. camelopardalis.

    (Pedantic plant taxonomist here.)

    • gravelinspector-Aidan
      Posted September 11, 2016 at 2:56 pm | Permalink

      What was that story I heard recently about a taxonomist re-entering a burning museum and coming out, clutching an armful of type psecimens?

  12. Posted September 11, 2016 at 12:15 pm | Permalink

    I’d say the fertility of the captive hybrids is a key and actually easily testable point here. The authors should have looked that up.

    There is an error in some side details of this article. You write: “The groups are from Africa, East Africa, Europe + Middle East, Central/South Asia, Oceania, and the Americas.”
    No, there is no Central/South Asia component at K=5. There is a rather clear East Asian component however. (Indian and Central Asian groups mostly have the same color (teal?) in the picture at K=5 as Europeans and Middle Easters and actually forming a cluster that called West Eurasian in human population genetic talks, while the next component that involves Chinese and Japanese among South East Asian groups is usually called East Asian.)

    • Posted September 11, 2016 at 2:07 pm | Permalink

      Also you say K=5 is the best fit, then you list 6 groups. One of the list is “East African” and that is probably supposed to be East Asian (explaining why you do not mention that component), but there is still no Central+South Asian common group at K=5 and neither at K=6, because even at K=6 only the Kalash form an unique group by themselves and their South/Central Asian neighbors do not cluster witzh them (they still cluster with Europeans and Middle Easterns at K=6, like at K=5.

      The components at K=5:
      1. Sub-Saharan African (the only listed North-of-the-Sahara African population, Mozabite, clusters more with Europe-Middle East!)
      2. West Eurasian (this includes Central Asia and India, with two of the Central Asian populations (Hazara and Uygur) having considerable East Asian blood)
      3. East Asian
      4. Papua-Melanesian (You can call it Oceanian based on just this analysis, but most of Oceania would possibly cluster more with East Asia if they were included. Papuas are special, with a populatuion history very different from of the Pacifix Islands.)
      5. Amerindian

      At K=6 it is the same, except the Kalash (an isolated tribe in the Pakistan-Afghanistan border mountains that is kind of a last Mahican of an ancient group) form their own group that shows up in lesser extent in some neighbor tribes. I think this is a bit exaggerated, other similar tests do not show the Kalash this much different as to make them the sixth component coming up.

      I am sorry for ranting this long about something not core for the article, but the error is still real.

      • John Harshman
        Posted September 11, 2016 at 2:20 pm | Permalink

        Yeah, the sample misses a lot of spots, including many of those along which we would like to see a transect sampled. Is it coincidence that in the major case in which there’s a good geographic sample (the Silk Road, more or less), we see what looks like a smooth cline in the N=5 plot?

        A transect along the Nile would be nice, and across the Bering Strait. And through India, Burma, Malaysia, etc. And where is Australia?

  13. Posted September 11, 2016 at 12:15 pm | Permalink

    There is a literature about what we’re trying to save when we preserve species or populations in plants. Consensus seems to be that we want to preserve as much genetic diversity as practical in population that are large enough that they are not highly susceptible to extinction. (One pregnant female would not be enough.)

    A problem is that relevant laws are based on names, not genetic diversity per se. Giving the giraffes names (as species or subspecies) would make them visible to these laws.

    • Posted September 11, 2016 at 12:26 pm | Permalink

      But what kind of genetic diversity? That was my point? Neutral diversity? Surely not? Diversity affecting phenotype? Do you want every allele, including very rare ones? And populations are a lot more resistant to size-based extinction than we think.

      The solution has been to save subspecies to conserve “diversity” but never have I seen it explicitly specified how much diversity we want to save, and of what type? Do we want to save diversity for tail length in beach mice, for instance?

      • Posted September 11, 2016 at 12:41 pm | Permalink

        We don’t know what diversity is going to be neutral in the future, and what won’t. Therefore, the assumption is that we need to save diversity of all kinds, including diversity that we don’t see. (The question is often addressed using presumed neutral isozyme diversity, though most people I’ve worked with are using them as markers and are more interested in adaptive variation.)

        There are too many kinds of diversity to go measure them all and say, “Got to keep this one, she has a longer tail.” Save some representatives of all the different groups we know about. Keep representatives of the different geographically isolated groups, for example. Value more distantly related individuals when establishing a captive breeding program.

        And if at all possible, keep reasonably large populations so they are less likely to go extinct and so that genetic diversity will remain, not be lost in inbred populations.

        • Posted September 11, 2016 at 6:17 pm | Permalink

          This is a tough question. I think neutral diversity is often a reasonable proxy for the total, hidden molecular diversity of a population, and that it is worthwhile to apportion conservation resources to maximize the conservation this total molecular diversity in a species of interest.

          But nowadays we try to take an ecosystem approach to conservation rather than a species-specific approach. We do this without knowing how ecosystems work in detail. So we are working in the dark, doing the best we can.

          And we are biased by our focus on macroscopic species. I suspect that the genetic diversity of the macroscopic component of ecosystems is dwarfed by the diversity of fungi, bacteria, nematodes, etc. In terms of genetic diversity, the most important ecosystems in the world are probably not places like the Serengeti but extremophile habitats like Mammoth Hot Springs, with their deeply divergent microscopic life forms.

  14. Heather Hastie
    Posted September 11, 2016 at 12:35 pm | Permalink

    It’s my understanding that giraffes are not threatened currently and that at the moment there’s no need for any special conservation measures. The cynic in me can’t help wondering if framing this study as being useful for conservation is playing into a particular public mood to increase notice or interest. Surely all knowledge of any species is useful if there’s ever a hint we need to worry about its preservation?

    • Posted September 11, 2016 at 12:49 pm | Permalink

      I’ve read articles by people working with giraffes who have said that nobody worries about giraffes so they’re not looking, but populations are declining. (Possibly being lost in a few cases, but my memory isn’t that sharp.) A point in one of the articles was that one or two of the geographic variants(subspecies? species?) have become very rare without generating interest in the conservation community because some giraffes are more common. I suspect this concern is behind this study.

      • Heather Hastie
        Posted September 11, 2016 at 1:10 pm | Permalink

        Interesting. Thanks for the info.🙂

  15. Gregory Kusnick
    Posted September 11, 2016 at 1:03 pm | Permalink

    Thanks for this write-up.

    Suggestion: You might consider reproducing the caption for Figure 3A, which is fairly opaque without it (at least for those of us who have not seen that sort of diagram before).

  16. Posted September 11, 2016 at 1:12 pm | Permalink

    I agree that the BSC should be the “gold standard” for determining what is a species and what isn’t. What we do we when the populations are allopatric and the BSC can’t be applied? We can’t prove if they’re biological species or not. We have to use taxonomic judgement, a.k.a. arbitrary choices. That’s just real.

    In giraffes, we have differences in pattern (possibly adaptive) and gene sequences. We have groups with limited gene exchange. Do we call them four (or six or nine) species? Or subspecies? Or metapopulations within one species? Biologically, it doesn’t really matter. (It may influence conservation, though.)

  17. Jim Mattila
    Posted September 11, 2016 at 1:16 pm | Permalink

    What I’m curious about is whether figure C is actually accurate, for indeed it shows divisions within ancestral Giraffes which may or may not have ever existed historically speaking… I speak of the vertical lines representing the branching of one type into two. Might not ALL the ancestral populations indeed have been of but the one original species (on the left margin) with just allele frequencies differing across it’s range? Meaning the inference in the figure that some type of clear division had occurred in the past, has no historical validity? Is the conclusion that branching had truly occurred exactly correct? And if not, and all modern types came from but ONE biologically valid species, what does that say about how Giraffes populations may be classified today although indeed disjunctive. But regardless, if due to say ecological fitness and breeding preferences the four extant groups each are absolutely a valid species, do they not in total represent a genus after all and so we are looking at geniation in all of this as much as speciation?

    • John Harshman
      Posted September 11, 2016 at 2:30 pm | Permalink

      Wouldn’t every clade have arisen from but one biologically valid species? There must at one point have been a single species, for example, from which every extant primate descends. And yet Homo sapiens and Pan troglogydtes are still separate species.

      What Fig. C shows (if by that you mean the tree), is that the groups in question have not been exchanging genes for a very long time. It doesn’t say how many biological species there are, because simple geographic isolation could be the reason, rather than any genetic isolating mechanisms. Based on the tree, there could be four species, or three, or two, or one. Or even five or six, because lack of clear genetic separation might just mean that speciation happened recently. The tree doesn’t constrain the number of giraffe species very much at all, though it could give us clues about what to look for.

      • Jim Mattila
        Posted September 13, 2016 at 6:12 am | Permalink

        Of course John all clades arise from a common ancestral species, but as I see it genetic relationship trees typically infer divisions along the way which to my mind may not have occurred in reality.

        Again, what if Figure C (yes the Giraffe tree) shows 4 species on the right as claimed, and yet all come from but ONE species on the right with each arising roughly at the same time in disjunctive locations.

        Might the divisions going back to the single ancestral species just represent its original allele frequencies in different parts of its range and so say nothing in the temporal sense, or even as to relatedness of the 4 species altogether other than that to the ancestor species in general?

        I guess what I’m saying is to me the easily observable grouping of animal life into clades from Phylum down to Genus has to be real, and therefore speciation can and does occurs in genus nodes at times (which can be logically explained)rather than just dichotomous branching alone.

        I mean seriously, if every event giving rise to a new species was always just one budding off another, it then follows that the groupings we see as obvious clades simply would not exist.

        Moreover it’s rather easy to envision how a new clade may come about via common descent from but one species although lol I’m sure few would agree that higher levels of classification above species are valid even if indeed they are…

        • John Harshman
          Posted September 13, 2016 at 2:18 pm | Permalink

          I don’t see how one can infer divisions that never happened. What we see here is clearly a genetic divergence among populations, whether or not it’s speciation. The trees are constructed based not on allele frequencies but on divergence of particular sequences.

          Could there be such great divergence within a single population? Generally, no. The exception would be for strong local selection, very unlikely across multiple, unrelated loci.

          Certainly it’s possible for several populations of a species to become geographically isolated at the same time, but this tree doesn’t show that. It shows successive bifurcations. Multiple loci are unlikely all to sort out in the same way if isolation is simultaneous.

          Molecularly, there is no such thing as “one budding off another”, implying that only one population changes. There is no stasis in sequence evolution, and even if we consider one modern population the “original” one, it will have diverged from the ancestral state, and about as much as the other populations.

          • jim mattila
            Posted September 15, 2016 at 12:30 am | Permalink

            I understand what you are saying John, its conventional thinking on this topic no doubt, obviously however I definitely (and with full respect of course) firmly disagree.

            In particular I argue that the current description of molecular genetics as it may conventionally apply to evolution is wholly inadequate to explain what we see in nature.

            So when you say “Multiple loci are unlikely all to sort out in the same way if isolation is simultaneous.” Or “There is no stasis in sequence evolution, and even if we consider one modern population the “original” one, it will have diverged from the ancestral state, and about as much as the other populations.” I think you may be missing something rather profound, being that in all likelihood these concepts are wrong at the level in which divergence leading to speciation occurs upon the landscape.

            What I’m trying to point out John, is that given what we observe in the real world, a way must exist wherein one ancestral species CAN give rise to several new populations all but at once, each disjunctive from the other, with all sharing a unique genes setting them on a path to becoming new species separate from their common ancestor.. AND as a group the new species must form all in kind without themselves interbreeding.

            And I easily see a mechanism by which a discrete clade may thus come into being, all from but one species, and be in a group which is valid at all scales down to that which may be molecular. Meaning as much as the dichotomous branches conventionally depicted in evolutionary relationships (regardless of type or scale) there is a speciation process which validly creates nodes which we often recognize as genus period.

            I’d be happy John to illustrate just what this process is and how it works to you or any other serious student of evolution (Jerry in particular) only I’d like to maintain ownership over my intellectual property and so any suggestion you may have about how best to go about it would be welcome…

            • notesfromberingia
              Posted September 15, 2016 at 12:27 pm | Permalink

              Write it up and publish it. bioRxiv or other preprint outlet could be used prior to submission elsewhere if desired.

  18. Alan
    Posted September 11, 2016 at 1:35 pm | Permalink

    I think taking an absolutist stand (no fertile offspring, ever!) on the BSC is also problematic.

    Thought experiment related to BSC and Lewontin and Birch 1966:

    2 “species” A and B are ecologically similar;

    Species A has a large range that overlaps and extends over the smaller range of species B. The overlapping populations are in sympatry and RI (e.g. through reinforcement)

    Now climate warming.

    The species B undergoes range expansion, the colonizer on the wave-front hybridize and become introgressed by naive populations of the species A.

    So if two species are reproductively isolated throughout most of their sympatric distribution except on the wave-front, are they one or two species? is it two species in the core, but one in the wave-front?

    I think two biological entities can under certain natural scenarios produce fertile offspring and still be two “species”

    • Posted September 11, 2016 at 3:24 pm | Permalink

      If you read Speciation, you’d see that I don’t take an absolutist stand; things are “more or less” biological species depending on the degree of hybridization. So you’re mischaracterizing my views on this. But that’s not a criticism, for I can’t reproduce everything I’ve written about this issue in a website post. If people want to argue about the BSC and gene flow, I think they need to go read Chapter 1 and the Appendix of Speciation first.

      The problem with your scenario is that they act like good biological species in the area of overlap because of reinforcement, but other parts of the range of both species haven’t evolved genes yielding reproductive isolation. So the status of the entities as biological species is initially problematic, unless the geographic isolation was true “habitat isolation” due to genes.

      • Alan
        Posted September 11, 2016 at 3:44 pm | Permalink

        I didn’t mean to mischaracterize your views.

        I do remember in Chp 1 you saying that hybridization can sometimes occur, although I don’t think a natural scenario was given.

        It does mention how RI can break down between sympatric species under confinement, like in zoos (e.g. Leopons).

        I was just pointing out a way in which two real species can hybridize under natural conditions, maybe more often than previously thought.

  19. ladyatheist
    Posted September 11, 2016 at 2:47 pm | Permalink

    Excellent read! Giraffes are fascinating in so many ways. As I was reading the beginning my burning question was “how does this compare to human genetic diversity” and you answered it! That made my day. Thank you!

  20. Posted September 11, 2016 at 3:46 pm | Permalink

    Great post! To my reading, I interpret this type of phylogenetic analysis as placing an upper bound on the number of species under consideration. Because isn’t it necessarily true that two different species will always be genetically distinct? Just how distinct is unknown, but surely substantial differences must exist, at least to ensure the unviability of hybrids. So the way I am interpreting things is that there are very likely no more than 4 different species of giraffe (but there could be as little as one).

    • Posted September 11, 2016 at 6:32 pm | Permalink

      Actually, I don’t think it places an ypper bound.. Maybe if we know about genetically based forms of isolation (e.g., is there a preference for different habitats?), there could be even more than four species. The fact is that, from a BSC point of view, we don’t know squat.

    • John Harshman
      Posted September 11, 2016 at 7:58 pm | Permalink

      There could be several as long as some of the species haven’t been separate for very long, i.e. long enough for genetic differentiation to have happened.

    • Gregory Kusnick
      Posted September 11, 2016 at 8:02 pm | Permalink

      I’m hardly an expert, but my guess is that speciation doesn’t necessarily require substantial genetic differences. One can imagine a scenario in which a single mutation in (say) color vision or pheromone production could abruptly fragment a population into distinct breeding subsets with different mate preferences, so that hybrids, although viable, rarely occur.

      • ladyatheist
        Posted September 12, 2016 at 1:17 pm | Permalink

        How about chromosome splitting? At some point in the past, chimps & humans’ ancestors developed a different number of chromosomes. That first mutant with the wrong chromosomes shouldn’t have been able to mate with normal ones, but unless there were two like that in the same group, it must have happened.

        • John Harshman
          Posted September 12, 2016 at 10:12 pm | Permalink

          Actually, difference in chromosome number does not, by itself, greatly interfere with reproduction. There are a number of populations of various species that are polymorphic for differences in chromosome number (usually due to Robertsonian fusions).

    • Posted September 12, 2016 at 9:03 am | Permalink

      At a rough guess, Fennessy et al have enough data to detect a speciation if it occurred 0.1My ago or earlier. A priori, a speciation is unlikely during that time, so I think this type of analysis can let us say things along the lines of “we are 90% sure there are no more than 4 species”.

      • John Harshman
        Posted September 12, 2016 at 9:30 am | Permalink

        I think the main problem with your reasoning is “a priori, a speciation is unlikely during that time”. Why?

        • Posted September 12, 2016 at 10:44 am | Permalink

          Because speciation rates are typically much less than one per 0.1My.

          • John Harshman
            Posted September 12, 2016 at 12:20 pm | Permalink

            I don’t think that’s a valid inference. It’s quite possible for multiple speciations to happen roughly simultaneously, especially if a single cause (say climate change) fragments one habitat into multiple islands. And the rate of speciation is not uniform over time, if adaptive radiations mean anything.

            • Posted September 12, 2016 at 7:06 pm | Permalink

              There are good orchid species that are less than 300k years old.

  21. notesfromberingia
    Posted September 11, 2016 at 6:34 pm | Permalink

    I agree, Jerry — this is a case of oversplitting given present evidence. I also agree with your points equating human genetics with species limits in other taxa. We’re in a period of rapidly growing amounts of genomic data without an adequate understanding of how they relate to species limits (and an overwhelming bias for neutral processes over the actions of selection). However, regarding conservation, I make the point in a forthcoming ms (
    http://www.kevinwinker.org/Winker_systematics_popgen_taxonomy_and_global_change_Revised_final.pdf ) that extant diversity, especially among subspecies, should be preserved to maximize the potential of the species lineage to effectively thread the needle of rapid environmental change. Also included are my own observations about genomics (including human) and what they portend for taxonomic decisions. I sum it up as “Evolutionary independence is not equivalent to speciation.” (pp.6-7) Anyway, it’s good to see such lively discussion here, and thanks for pressing on with excellent points made over a decade ago in Speciation.

    Kevin Winker

  22. George
    Posted September 11, 2016 at 8:21 pm | Permalink

    I am skeptical (but by nature). But I am curious if we need a tighter definition of species. If these animals can breed without any issues then I have a hard time considering them distinct species.

    • daveyc
      Posted September 11, 2016 at 8:55 pm | Permalink

      There are literally dozens of species concepts, many of which are more useful in certain contexts than they are in others (some seem to be not be useful in any context). There is an EXTREMELY extensive literature on the subject of species concepts that you can dig into if you’re interested. This paper (De Queiroz 2007) is not a bad place to start (though I suspect PCC is not a huge fan of it):

  23. eric
    Posted September 12, 2016 at 7:41 am | Permalink

    Now human populations cannot be distinguished from their DNA sequences nearly as unambiguously as can these giraffe subspecies, but if you gave them another million years of geographic isolation (now impossible because of travel), they might well have separated genetically to the same extent as these giraffes. Would you then be happy to call human populations different “species”?

    Haven’t we done exactly that? Less than one million years separates Homo Sapiens from Homo Heidelbergensis, Denisovans, and even Homo Erectus. There are also the Neanderthals, though AIUI the most up to date thinking on them is that they were a subspecies of modern humans rather than a different species.

    In any event, I’m not suggesting that the only thing separating these groups was coat pattern.🙂 My point is that a million years is plenty of time for the homo line to diverge into groups that we would define as separate species based on other, more important phenotypic differences, because our understanding of history is that its diverged into separate species in that amount of time before.

    Applying that thought to giraffes, if there are a number of phenotypic differences between the groups in addition to coat pattern (the analog to homo would be skeletal differences), then I’d be perfectly happy calling them different species. Because if we don’t, then we’re being somewhat inconsistent, drawing very fine distinctions between ‘species’ in our own line that we don’t bother to draw when it comes to other animals.

    • eric
      Posted September 12, 2016 at 9:27 am | Permalink

      Post script to my last sentence; I think Darwin himself pointed out this sort of inconsistency, noting that florists will see important distinctions between flower types that other people might consider nothing but ‘varieties’ of the same type. So maybe giraffe-fanciers see the distinctions as more important than laypeople, and us being hominims means we see relatively subtle distinctions between hominims as more important than, say, a giraffe might.🙂

  24. Posted September 12, 2016 at 11:38 am | Permalink

    Is there anything about the time involved that should make one skeptical as well? A four way divergence in a million years? Is that what is being proposed?

  25. Erik
    Posted September 12, 2016 at 1:51 pm | Permalink

    First, I want to thank you again for taking the time and effort to do these critical analyses of recent papers. I share your frustration that the same critical analysis isn’t done by science journalists. The coverage will certainly translate into the “fact” that there are 4 species of giraffes. I think the critical point is that there are distinguishable genetic clusters in need of a more focused approach to conservation. (P. S. the paper and your analysis will be required reading for my students the next time I teach my Biodiversity course)

  26. Posted September 12, 2016 at 7:36 pm | Permalink

    Reblogged this on Wild Equus and commented:
    It is likely we have all heard the news ‘DNA studies have revealed there are actually 4 extant Giraffe species instead of one.’

    Jerry Coyne, explains why this might not be so. It is a great article and the arguments he puts forth are quite solid. Enjoy!

  27. Posted September 13, 2016 at 12:17 am | Permalink

    I appreciated the alternate perspectives on this. Also, I believe the Liger reference was from the movie ‘Napolean Dynnamite’, unless there is also a Liger reference in a movie called ‘Napolean Bonaparte’.

  28. Henri Burger
    Posted September 14, 2016 at 4:00 pm | Permalink

    I like your critique, but would like to take up the point of Ligers and their fertile offspring.

    Do we for sure know that this never happened under natural circumstances? That’s besides the point that lions only occur in small numbers in India, and in far as I’m aware in a relatively small reserve.

    However, in Southern Africa one gets two different acacia species (or whatever they’re called these days!), A. erioloba and A. haematoxylon, that produce natural hybrids where they occur in close proximity and these hybrids are also fertile. So, one doesn’t need a forced relationship as you would seem to suggest with your Liger example?

    Now, I have no knowledge of the genetic similarities of the two mentioned acacia species, but the alternative would then be to reclassify them as one species, although morphologically they are quite distinct.

3 Trackbacks/Pingbacks

  1. […] the African elephant as a single species. With regards to the giraffe study, evolutionary biologist Jerry Coyne wrote a critical note on his blog in response to Janke’s article and subsequent media coverage. […]

  2. […] … there are four species of giraffe. Or not. […]

  3. […] features such as pigmentation, behavior, or size. But recently, a group of researchers as set off a flurry of excitement following claims that these isolated populations represent distinct species, not just one unified […]

Post a Comment

Required fields are marked *

%d bloggers like this: