Proposal (#367) to South American Classification Committee


Split the Warbler Finches: Certhidea fusca from Certhidea olivacea


Effect on South American CL: this proposal would split the Warbler Finch into two species, Certhidea olivacea and Certhidea fusca

Background: The taxonomy and relationships of the Darwin's Finches is messy, muddied, and complicated. In a multitude of papers Peter and Rosemary Grant have documented very quick changes in morphology as a response to climatic changes (ENSO events), hybridization and its role in increasing genetic diversity and perhaps its facilitation of speciation in island taxa, as well as invasions of conspecifics from other populations with slightly different bill sizes and the retention of these bill size differences through generations (in other words possible sympatry and assortative mating solely on bill size within a species). It is messy, particularly in the genus Geospiza.

One of the most divergent of the Darwin's Finches is the Warbler Finch (Certhidea olivacea), a small bodied, thin billed and distinctly warbler-like member of the clan. It is the sole member of its genus, and is perhaps the most widespread of all Darwin's Finches being found on all major islands. Its status as a species has never been questioned; it is quite different from other Darwin's Finches. However, recent data show that a cryptic species appears to be involved here and that the Warbler Finch is in fact two species.

New information: Surprisingly in a phylogeny of the Darwin's Finches (Petren et al. 1999) Certhidea appeared to be comprised of two distinct and old lineages. This phylogeny included 13 of the 14 species of Darwin's Finches, and was based on allele length variation at 16 microsatellite loci. Certhidea is basal within the monophyletic lineage of Darwin's Finches. The divergence of the two lineages of Certhidea predates the radiation of all other Darwin's Finches. I should note that Yang and Patton (1981) in a study of 11 of the Darwin's Finches using starch-gel electrophoresis procedures on 27 presumptive loci found concordant patterns to this, but given the early period in which it was done (1981!), it was interpreted differently. In that work Certhidea is again clearly basal to all other Darwin's Finches, and they were surprised to find that it showed a markedly high (compared to the other species they sampled) interpopulation levels of genic differentiation. The pattern that caught their eye was that heterozygosity was positively and significantly correlated with island size. They interpreted this as an effect of differing population sizes, although they did not find the pattern in other Darwin's Finches.

Grant and Grant (2002) described vocal differences between the two lineages of Certhidea. They went on to conduct various playback experiments to see if song was the isolating mechanism, as it can be in other Darwin's Finches, and found that song did not act as a pre-mating isolating mechanism between these two rather old lineages.

Finally, Tonnis et al (2005) focused on Certhidea itself. Their research sampled 84 individuals from all 16 populations, using mtDNA (cytochrome b) sequence data (864 bp). In Certhidea, genetic and geographic distances are not correlated. In fact genetic variation over a large area of the archipelago (the small peripheral islands) is very small. Again, their data sorts out into two separate lineages, and genetic distance between these lineages averages 3%. The two lineages of Warbler Finch are the most genetically divergent lineages of all Darwin's Finches! Unlike the analysis of geographic distance, if genetic distance is related to islands of habitat similarity, there is a correlation. In other words islands of similar habitats, independent of how distant they are from each other, have genetically more similar populations. The hypothesis is that habitat choice itself is the isolating mechanism. One lineage prefers mesic habitats (olivacea), the other dry habitats (fusca). Neighboring islands do not always harbor genetically similar populations, if the neighboring islands differ in habitat. The fusca lineage is found throughout the smaller islands in the north, east and south of the archipelago, practically spanning the entire length of the archipelago, whereas olivacea is found on the larger central and western islands. The olivacea lineage is found in larger islands where upland wet forests are to be found, and it is restricted to these moist forests. In the range of olivacea, there are no Certhidea populations in the dry lowlands. On the other hand fusca is restricted to islands that are low and dry; presumably it could spread to the dry lowlands of the larger islands where olivacea is found - but it has not. This puzzling pattern suggests that these two populations have other biological differences that keep them separate. The genetic information suggests clearly that both olivacea and fusca have undergone long-distance dispersal between islands in the near past, so inability to disperse is not what is keeping them separate.

The two clades.

olivacea includes: olivacea (Santiago, Rabida, Pinzon, Isabela, Fernandina, Santa Cruz islands).

fusca includes: fusca (Pinta and Marchena), becki (Darwin and Wolf Is), mentalis (Genovesa), bifasciata (Santa Fe), luteola (San Cristobal), cinarescens (EspaĖola), ridgwayi (Floreana).

Visually, the two are quite similar. However, males of olivacea are often brighter colored and develop a reddish throat patch not observed in fusca. Similarly, their general tendency is to have a slightly more colorful yellowish plumage overall, whereas fusca is grayish overall and tends to be duller than olivacea. Bright male olivacea are rather distinctive by the way, the warm throat color is not subtle.

English Names: The only ones I have seen mentioned are "Gray Warbler Finch" for fusca, and "Green Warbler Finch" for olivacea. This suggestion was made in Price (2008) Speciation in Birds. It is not a bad one given that it matches up somewhat to the Scientific Name and notes the general plumage differences between the two. The only other names I could come up with are Highland and Lowland Warbler Finches, but this suggests that they can be found in different elevations on the same island perhaps. So Gray and Green seems fine to me, unless I have missed other published English names for these clades.

Recommendation: I recommend a YES vote for this proposal. Various lines of evidence show a deep split in the Warbler Finch, one associated with habitat, vocal, and plumage differences. This is the most divergent split within any of the Darwin's Finches (estimated at nearly 2 million years ago), and certainly this situation seems much more clear cut than what is going on in newer lineages such as Geospiza, in which we may have species that are not monophyletic (i.e. some Geospiza difficilis populations may in fact be basal to all Tree/Ground finches, whereas others are clearly in Geospiza). But that is for another proposal, in this case I think that Certhidea is clearly two species.

Literature Cited.
GRANT, B. R., & P. R. GRANT. 2002. Lack of premating isolation at the base of a phylogenetic tree. American Naturalist 160: 1-19.
PETREN, K., B.R. GRANT, & P.R GRANT. 1999. A phylogeny of Darwin's finches based on microsatellite DNA length variation. Proc. Royal Soc. London B 266: 321-329.
TONNIS, B., P. R. GRANT, B. R. GRANT, & K. PETREN. 2005. Habitat selection and ecological speciation in Galápagos warbler finches (Certhidea olivacea and Certhidea fusca). Proceedings of the Royal Society: Biological Sciences 272:819-826.
YANG, S.Y. & J.L. PATTON. 1981. Genetic Variability and differentiation in the Galapagos Finches. Auk 98: 230-242.

COMMENTS FROM PETER GRANT - JULY 14, 2008 (via e-mail).
Dear Alvaro,
Thank you for sending me a copy of your proposal to the Checklist Committee to recognize 2 species of warbler finches on the Galapagos.

As you say, the situation is complex. Our views are most fully expressed in a book we published with Princeton University Press earlier this year: How and Why Species Multiply. The Radiation of Darwin's Finches (P. R. Grant & B. R. Grant). As we explain in chapter 9 we adopt the biological species concept, and from this perspective there is no biological evidence to separate the olivacea and fusca populations into two species. Our colleague Ken Petren is more inclined to adopt the phylogenetic species concept. Have you contacted him also? He is the senior author on a paper in Molecular Ecology (14:2943-2957, 2005), which is relevant to the Certhidea question. Also Peter Boag would be worth contacting for an opinion. He wrote a relevant paper with Joanna Freeland in Auk (116:577-588, 1999). These papers are cited in our book.

Your summary under New Information covers the background well, except for the last couple of sentences. You write "This puzzling pattern suggests that these two populations have other biological differences which keep them separate." So far none have been identified, and this would be a weak basis for separating them into two species. The next sentence says that because they dispersed in the past, "inability to disperse is not what is keeping them separate." The logic is flawed. There is a well-developed theory of taxon cycles in which the ability to disperse to islands weakens with time. As far as I know only one warbler finch individual has ever been collected on an island that it does not belong to. In 35 years of field research on the island of Daphne Major we have recorded only one warbler finch, a short time visitor. Rare dispersal between islands can be inferred from patterns of private alleles. There are inevitable uncertainties in the estimates. Nevertheless, at face value this suggests the populations of the two lineages are not kept apart by intrinsic reproductive mechanisms.

Your summary describes the distribution of the lineages according to reports in the literature. I should emphasize that the distinction between highland and lowland populations is not quite as clear-cut as presented. The fusca lineage is to be found at moderate to high elevations on San Cristóbal, Floreana (apparently extinct) and Pinta. The olivacea lineage is found on Pinzón and Rábida, two islands that are not very high.

The mitochondrial data (cytb) imply a time of separation of fusca and olivacea lineages in the range 1.5 to 2.0 MYA. According to a broad survey of birds by Price and Bouvier the first signs of genetic incompatibility between closely related populations do not appear in this relatively short time frame. If the two lineages are reproductively isolated they are so by virtue of their behavior, and yet our playback experiments show they respond to each other's songs.

Finally the name Gray Warbler was introduced by Salvin, I believe, in 1876 for the Pinta and Marchena populations. I suggested it to Trevor. I used the names Gray Warbler and Green Warbler in the Afterward to the 2nd edition of my book Ecology and Evolution of Darwin's Finches (1999).
With best wishes,
Peter Grant

COMMENTS FROM KEN PETREN - AUGUST 8, 2008 (via e-mail).

Thank you for the chance to offer my input. Like Peter, I also favor the Biological species concept over the phylogenetic one, but in the following way. Evidence of shallow molecular phylogenetic distinctness alone I regard as weak evidence of species, unless it can be combined with other evidence. Importantly, I firmly believe that a lack of phylogenetic evidence, based on the arbitrary criterion of a single marker, is not sufficient grounds for concluding that two populations are not distinct species, IF there is other evidence showing they are distinct biological entities. Thus, even though most Darwin's finches are not phylogenetically distinct, they represent different lineages with different adaptations that respond differently via natural selection to environmental changes. To use species concepts to conceal such obviously important genetic variation and biodiversity would be a mistake.

In the case of Certhidea, the phylogenetic evidence is backed by the geographical evidence, which implies some kind of barrier to interbreeding. The Grants assayed males for song response at a single point in time and found no supporting evidence. This does not imply that reproductive isolation does not occur by song; it simply means the particular assay did not detect a difference. I also think that equating the Biological species concept to differences in male response to song is too narrow. Further work has indeed revealed eco-morphological differences among the species, which should be added to your proposal:

[1. Grant, P.R., and Grant, B.R. (2003) Reversed sexual size dimorphism in the beak of a finch. Ibis 145, 341-343].

There are a host of other behavioral or ecological mechanisms, some of which you alluded to from the Tonnis et al paper, that could be underlying the geographic pattern. In a nutshell, the geographic pattern of interspersed haplotypes associated with distinct habitat differences, to me and to most others, constitutes sufficient evidence of biological distinctness to support the phylogenetic evidence. Recognition of two warbler finch species is warranted.

Your proposal seems to be very reasonable, well-reasoned and well-written. One minor comment. The following statement is generally accurate, but superfluous: it does not pertain to the revision and can be omitted: “The pattern that caught their eye was that heterozygosity was positively and significantly correlated with island size. They interpreted this as an effect of differing population sizes, although they did not find the pattern in other Darwin's Finches.”

The proposal has my support, with the caveat I mention in the first paragraph.


Kenneth Petren, Ph.D.
Associate Professor
Department of Biological Sciences
University of Cincinnati
Cincinnati, OH 45221-0006



Alvaro Jaramillo, September 2008



Comments from Nores: "YES. Por las razones dadas en la propuesta."

Comments from Zimmer: "YES, for reasons nicely summarized in the proposal."

Comments from Stiles: "YES, tentatively. This one is decidedly complicated (as might be expected from the Galápagos..). As I read it, the two Certhidea differ in plumage and morphology to a degree consonant with species status in other genera like Geospiza; they also differ in song and in preferred habitats, albeit not completely. Genetically, they comprise two monophyletic groups of populations that differ to an extent consonant with species status as well (although this is not necessarily conclusive). The most telling evidence against species status is that they apparently respond (how?) to each other's songs (but this assumes that song is the only or most important vocal isolating mechanism, which is not always the case) and the fact that they never co-occur on any island, even when their different habitats are available. All in all, I feel that the evidence as it stands is weighted in favor of assigning species status to both olivacea and fusca. However, the cards are not yet all in and further study might be called for to remove certain doubts."

Comments from Stotz: “YES.  Seems like a pretty clear story, although the failure of isolation by song is an interesting subplot.”

Comments from Pacheco: “YES.  A partir das evidźncias aqui apresentadas, ainda que considerando este caso bastante complexo.”


Comments from Remsen: “NO.  Once again, the Galapagos “finches” provide an intriguing glimpse into the mechanisms of speciation.


“Differences in habitat, elevation, and plumage are insufficient criteria for species-level separation – we have hundreds, perhaps thousands, of such examples among subspecies known to intergrade at contact.  As Grant pointed out, the differences are not clear-cut.  Everyone knows example of populations of the same species that have different habitats and elevational ranges in different areas.  Yes, they may be the two oldest separate lineages in the group, but the phenotypic differences between them are not yet consistent with species-level taxon ranks.  Many sedentary tropical populations that are barely distinguishable phenotypically differ by far more than 2%.  The biological differences between them are already recognized by separate subspecies names.  No, the N on the playback trials is small, but nonetheless inconsistent with species-level differences in this group.”


“The strongest point in favor of species-level rank is the intermingling, geographically, of the lineages, with retention of genetic integrity.  However, the following scenario could explain this.  The Certhidea lineage, the oldest lineage relative all other GFs, splits before any significance splits and diversification in the non-Certhidea branch; thus, these early Certhidea branches, olivacea and fusca, accumulate more genetic differences between them than do any two branches on the non-Certhidea side, with one found on a more mesic island than the other.  A dispersal phase produces a splattering of olivacea and fusca types in the archipelago; but in spite of (or because of?) the tendency in habitat differences, no two types co-occur.  That brings us to the present, with two separate lineages, without sympatry and with ecological, vocal, and plumage differences.  However, the divergence in these characters is not yet to the level associated with species-level differences in GFs.


”Qualitatively, the world’s expert on the GFs, Peter Grant, does not think that the two groups have diverged to the level associated with species differences in GFS – I find that compelling evidence to retain subspecies rank.”


Comments from Cadena: “YES. I see Van's and Dr. Grant's points, but I am more compelled by the arguments described by Alvaro, and especially by Ken Petren. There are two different mtDNA lineages that remain distinct despite being geographically intermingled likely because they are associated with different habitats, and this strongly suggests species status. Van argues that the species do not exhibit true sympatry so they have no opportunity to interbreed, but I would like to argue that this is a case of "mosaic sympatry" as defined by James Mallet in a recent (2008) Phil. Trans. R. Soc. paper: " important intermediate situation which neither corresponds to large-scale geographical parapatry, nor is it truly panmictic. We might call this ‘mosaic sympatry’, characterized by patchy, interdigitated distribution of the two forms and their resources, with the patch size not much larger than the scale of dispersal". In situations like this, where dispersal between habitats is likely in terms of distance (as indicated by the mtDNA data) but does not happen, I think we can consider populations effectively sympatric, yet isolated as a result of habitat selection, hence different species. That an experiment demonstrated that the two forms respond to each other's songs says little about taxonomic status, considering all the known cases of interspecific territoriality, etc. It may be habitat selection, not song, the behavioral barrier between the two.”