Sunday, May 18, 2008

How many Toucanets? (Part 1) Many species of birds have significant variation in regional plumage and morphology. Just think of the variation in the Junco, Song Sparrow, or Horned Lark. These example species have broad, continuous ranges across North America, and we can examine the assortative mating and intergradation between the various taxa where they come into contact to justify lumping them into a single polytypic species. In the neotropics, many putative species vary across different allopatric mountain ranges. Because the different taxa never come into contact, it is extremely difficult to test the Biological Species Concept. These groups are generally lumped into polytypic species similar to the contiguous species mentioned above. All too frequently, these broad lumps obscure actual diversity, and are thrown out when closely examined. New work published in the Auk examines the Emerald Toucanet (Aulacorhynchus prasinus) species complex with molecular evidence and suggests splitting it into several different species.

A. prasinus, or... ?

The Emerald Toucanet is a montane forest species with disjunct populations on mountain ranges from Mexico to Bolivia. The complex currently contains as many as 17 subspecies:

A.p. wagleri (Mexico: Pacific slope)
A.p. prasinus (Mexico: Atlantic slope)
A.p. warneri (Mexico: Sierra de los Tuxtlas)
A.p. chiapensis (Mexico: Chiapas and adjacent Guatemala)
A.p. virescens (Belize – Nicaragua)
A.p. stenorhabdus (Belize – Nicaragua)
A.p. volcanius (Belize – Nicaragua)
A.p. maxillaris (Costa Rica)
A.p. caeruleogularis (western Panama)
A.p. cognatus (eastern Panama: Darien)
A.p. lautus (Columbia: Sierra de Santa Marta)
A.p. phaeolaemus (Columbia: Western Andes)
A.p. griseigularis (Columbia: Central Andes)
A.p. albivitta (Columbia and Venezuela: Eastern Andes)
A.p. cyanolaemus (Ecuador and Peru)
A.p. dimidiatus (Bolivia)
A.p. atrogularis (Peru)

Puebla-Olivares et al. (2008) construct a phylogeny of the complex using three mitochondrial genes to follow up on the morphology and color analysis of Navarro-Siguenza et al. (2001). The tree was essentially congruent for all three loci, and using all three phylogenetic reconstructions (parsimony, likelihood, and Bayesian), indicating strong support for the given results. Here is the Bayesian tree - I'll be discussing each part of it in turn, so you don't have to grasp it all at once.

(click to view large)

The 'prasinus' species complex is a monophyletic group, when using the other species in the genus as outgroups, so the 'prasinus' complex taxa are each other's closest extant relatives. They aren't very 'closely related' however - the molecular data indicate substantial divergence within the group, averaging around 5% between the seven main lineages, which is on par with the divergence between other members of the genus, which average 6 to 7% between species. This level of divergence is often considered a species-level in birds, because it indicates that the different lineages have been evolving independently as separate lineages for a considerable amount of time - possibly millions of years (there is a standard molecular clock for birds of 2% mtDNA divergence per million years, but this isn't totally universal or accepted). Saying what level of divergence is "enough" for species is not something that can be answered, but the fact that the other members of the genus are equivalently divergent is pretty indicative that we are dealing with a species complex rather than an array of subspecies.

The various lineages in the 'prasinus' complex map out well geographically. The first and largest divergence in the group creates two clades – a Central American clade with four main lineages, and a South American clade with three main lineages. These two main clades average around 7% divergence from each other. I’ll dive into the details of the Central American clade first, and address the South American clade in the next post.

I've summarized the range and coloration of the populations of the Central American clade in the composite figure below, drawing from the map and face pattern diagrams of Navarro-Siguenza et al. (2001), and pictures from various field guides (see full citation below). The Central American populations can be broken down into two groups by throat color (blue vs. white) and can be further subdivided by bill and face pattern features. Based on this, one might be most tempted to split the group into two species, each with two main subspecies groups - the white-throated northern populations, and the blue-throated southern populations.

(click to view large)

However, a two-species split in this clade based on throat color is not supported by the genetic evidence. As I mentioned before, the phylogeny of this complex maps out well geographically. The basal members of the group are in Panama, as expected by the main Central vs. South American divergence, and the group branches out in a northward progression. However, if there were a two-species split in the clade, we'd expect to see a tree with two main branches, subdivided into subspecies. Instead, the basal member of the clade is the population in the Darien of eastern Panama, sister to the remaining Central American populations. The Costa Rican populations fall out next, and the northern Central America populations fall out last. So, the two blue-throated populations are not closely related sister taxa, but are actually among the more divergent of the group. Based on this, Puebla-Olivares et al. (2008) end up treating the Central American populations of the 'prasinus' complex as four distinct species, broken down as follows:

A summary of the Central American toucanet tree with approx. mtDNA divergences
(click to view large)

The basal member of the Central American clade is A.p. cognatus from the Darien mountains of eastern Panama. The Darien population is divergent from the rest of the Central American lineages by about 6%, and the authors suggest it be split as Goldman’s Blue-throated Toucanet (A. cognatus). This form is not depicted in any field guide that I could find. It is very similar to, yet shows subtle fixed characters distinct from, the Costa Rican form based on the color patterns described in Navarro-Siguenza et al. (2001) - compare the face pattern diagrams in the figure above.

The next group of the remaining Central American lineages is the population in Costa Rica and western Panama – A.p. caeruleogularis and A.p. maxillaris. This group is split from the remaining lineages in Central America by an average of 5% divergence, and is split as the Blue-Throated Toucanet (A. caeruleogularis). The two subspecies in this group show essentially nil divergence (0.09%). I observed this species on my trip to Costa Rica last winter break, and managed a few poor images. Here is a Blue-throated Toucanet in mid-hop:

The remaining Central American populations are separated from the Costa Rica/Panama group by the lowlands of Lake Nicaragua, and encompass the mountains from Nicaragua up through Mexico. These populations are characterized by their white throats, as opposed to the blue throats of the previous two groups. This northern Central America group is split into two lineages. One lineage is found in the Sierra Madre del Sur, the mountains of the Mexican states of Guerrero and Oaxaca on the Pacific slope. This group is separated by about 3.7% divergence from the other northern Central America populations, and is split as Wagler’s Toucanet (A. wagleri). Although this population only represented one subspecies in the 'prasinus' complex, it showed a surprised amount of divergence – 1% - between the Guerrero and Oaxaca populations. The authors identify this as a region for further study - it also supports a number of other phylogeographic splits and is a region of high endemism in Mexico. See the Howell and Webb (1995) field guide to Mexico for more examples.

The last Central American lineage encompasses six subspecies (prasinus, warneri, chiapensis, virescens, stenorhabdus, volcanius), and shows weak genetic structuring that supports some level of incipient divergence within the group. The whole group is split (or rather, retained) as Emerald Toucanet (A. prasinus) because it contains the type subspecies of the 'prasinus' complex. The most divergent population is A.p. chiapensis of the Mexican state of Chiapas and adjacent Guatemala, split by 0.96% divergence from the other populations. The remaining populations on the Sierra Madre Oriental, Sierra Madre de los Tuxtlas, and from Belize through Nicaragua are very weakly split. Navarro et al. (2001) identify little variation within this group, despite it containing a number of subspecies populations.

All four of these species identified by Puebla-Olivares et al. (2008) have unique combinations of face and bill pattern and color, and show substantial genetic divergence. Perhaps the weakest of these splits in morphology and color is Wagler's Toucanet, but 3.7% divergence is still substantial , and Navarro-Siguenza et al. (2001) show Wagler's to be fixed for color traits that separate them from other similar Emerald Toucanet populations, including bill pattern, superciliary color, and color at the base of the mandible. In noting the fairly large divergences within populations of A. wagleri and A. prasinus, one must wonder again where the limit is when it comes to splitting up lineages - how much divergence does it take to be species-level? If we look at the variation within A. prasinus, up to 1% divergence amongst populations in Mexico, Guatemala, and down to Nicaragua - this variation is not congruent with distinctive, fixed coloration traits, as Wagler's Toucanet is, although there is some variation discussed in Winker (2000). These various populations are not split up at the species level, while Wagler's is split from the group.

This four-species split of the Central American populations of Aulacorhynchus 'prasinus' are the same as those identified with color and morphology analysis in Navarro-Siguenza et al. (2001). As Navarro-Siguenza et al. discuss, nowhere in Central America do these various groups come into contact, so direct tests of the Biological Species Concept can't occur. They do note that bill and face patterns are important social cues in toucans, and they reason that these could serve as reproductive barriers and thus qualify as Biological species in the absence of any evidence to the contrary. The groups outlined by Navarro-Siguenza et al. and Puebla-Olivares et al. are congruent in morphology, color, and genetics, and also qualify as species under the Phylogenetic Species Concept (they are diagnosable populations) and the Evolutionary Species Concept (they are evolutionarily distinct, independently evolving groups). To me, this combined evidence provides a very clear picture for splitting the Central American members of the 'prasinus' complex into four species.

As you'll see, the picture in South America is a lot less clear. See Part 2...

References and Citations

Image 1 - Source
Bayesian Tree - Puebla-Olivares et al. (2008)
Tree summary - redrawn from the Bayesian tree of Puebla-Olivares et al. 2008

Central American clade figure:
Map - Navarro-Siguenza et al. 2001
Face pattern figures - Navarro-Siguenza et al. 2001
A. prasinus - Howell and Webb 1995
A. wagleri, A. caeruleogularis - Short, Horne, and Gilbert 2002

Howell, SNG, and S Webb. 1995. A Guide to the Birds of Mexico and Northern Central America. Oxford University Press, USA.

Navarro-Siguenza, A.G., A.T. Peterson, E. lo Pez-Medrano, and H. Benitez-Diaz. 2001. Species limits in mesoamerican Aulacorhynchus toucanets. Wilson Bulletin. 113(4): 363-372.

Puebla-Olivares, F., Bonaccorso, E., de los Monteros, A.E., Omland, K.E., Llorente-Bousquets, J.E., Peterson, A.T., Navarro-Siguenza, A.G. (2008). SPECIATION IN THE EMERALD TOUCANET (AULACORHYNCHUS PRASINUS) COMPLEX. The Auk, 125(1), 39-50. DOI: 10.1525/auk.2008.125.1.39

Short, L, J Horne, and AE Gilbert. 2002. Toucans, Barbets, and Honeyguides. Oxford University Press, USA.

Winker, K. 2000. A new subspecies of Toucanet (Aulacorhynchus prasinus) from Veracruz, Mexico. Ornitologia Neotropical. 11:253–257. SORA.


  1. I began wondering about this group when I returned from CR last year and couldn't find my new life bird, Emerald Toucanet, in some lists. i don't know whether is' Howard and Moore or the new Clements that splits them. Certainly the new IOC list does.

    Really nice breakdown Nick.

  2. Thanks, Nate. That reminds me, I should look up the various checklists and see what else people are calling these splits. I know I've seen different common names than the ones suggested in these papers.

    Part 2, South America will be up sometime soon, its a lot more complicated and unresolved than Central America.

    ~ Nick

  3. The idea of "official" common names is a somewhat bemusing concept to me, I have to admit. Perhaps because I generally work with invertebrates rather than vertebrates, I'm actually often more familiar with the taxonomic names than with the common names. This actually came up at one point last year when I had no idea what a spinifexbird was, but suddenly understood when it became clear a spinifexbird was Eremiornis.

    I'm also going to add that toucans are one of those organisms that are so colourful that it's always kind of surprising to consider that they actually exist. You just don't expect colours like that in nature.

  4. Great post. The images, story, and taxonomy of this bird is really interesting.

  5. These guys are such cool birds! I hope one day to see one in the wild!

  6. Glad you like it guys! Part 2 is now up...

  7. i have four emerald toucanets. they have their own twitter page.

  8. My friend was in Costa Rica, he traveled thorough last minute travel. I saw his photos from there, it is amazing place. I would like to go there in the future.

  9. I love birds, they are so majestic.

    You did a really good job with this article, Nick. It was very informative!

    Keep up the good work!

    Director, SEO Services
    Top Rank Consultants

  10. A friendly reminder that--no matter which definition you use--Mexico N of the Isthmus of Tehanatepc is not part of Central America. So it's not quite correct to call wagleri a "Central American lineage". In most cases above, substituting "Middle America" for "Central American" would alleviate the issues.