Split Diglossopis from Diglossa

Proposal (417) to South American Classification Committee

The recently published analysis of the phylogeny of the flower-piercers by Mauck et al. (2009) demonstrated a deep division in this group between two monophyletic clades: one that includes the species caerulescens, cyanea and glauca (genus Diglossopis of Bock 1985), and a larger clade containing the remaining species including indigotica. Because this species was placed in Diglossopis by Sibley & Monroe (1990), Mauck et al. argued that the genetic data rendered this genus polyphyletic and that it should not be recognized. The original recommendation for recognition of Diglossopis was made by Bock (1985) based upon a morphological study of the bill, skull and especially the tongue. Bock presented detailed evidence for differences between Diglossopis and Diglossa that he considered sufficiently pronounced to warrant generic separation, and doubted that they could have derived their adaptations for nectarivory from a recent common ancestor. (These differences seem to me to be of the same order of magnitude as those used by Lanyon in a series of studies to delimit genera in the Tyrannidae). Bock stated clearly that, while he had not seen anatomical material of indigotica, its bill morphology placed it in Diglossa, not Diglossopis, which he restricted to precisely the three species in the aforementioned clade found by Mauck et al. The inclusion of indigotica in Diglossopis was made by Sibley & Monroe apparently following the superficial division of Diglossa into “species-groups” by Vuilleumier (1969), who made a distinction between the “black” vs. “blue” groups and included indigotica with caerulescens, cyanea and glauca on the basis of its blue color and red eye, a feature shared with cyanea in particular – but evidently he did not examine the morphology in any detail. Bock argued that the coloration of indigotica represented a convergence with his Diglossopis, perhaps related to social behavior in mixed flocks. While Sibley & Monroe credited Bock with recognition of Diglossopis, they evidently did not read his conclusion regarding indigotica but in effect followed the “black vs. blue” groups of Vuilleumier – and the same applies to the discussion of Mauck et al. in this respect, since their genetic data in reality present a strong argument for recognition of Diglossopis as defined by Bock. Just how closely related are Diglossopis and Diglossa (specifically, whether or not they are sister groups; Bock argued on morphological grounds that they are not) is perhaps debatable since Mauck et al. did not examine in detail the relationships of these two with respect to a larger set of outgroups within the Thraupidae.

It is worth noting that although members of Diglossopis as well as Diglossa pierce flowers, they also take much fruit, which members of Diglossa do not. I examined specimens in the ICN collection with data on stomach contents and found that of 48 specimens of Diglossopis cyanea, 33 had annotated fruit in stomachs (vs. 38 with insects); stomachs of 29 of 34 caerulescens contained fruit, vs.21 with insects; these data contrast with those from Diglossa: for humeralis 58 of 60 contained insects vs. 3 with fruit; for 26 lafresnayi, all contained insects and one, fruit; for albilatera, 49 stomachs all contained insects vs. two with fruit. An examination of fecal samples by L. Rosselli produced similar results, as did detailed observations of foraging behavior by Rojas-Nossa (2005) which also showed that flower-piercing was also much more frequent in the same Diglossa species vs. those of Diglossopis. This correlates nicely with the depth of the hook at the tip of the bill: much deeper in Diglossa (and deeper in Diglossopis cyanea, which visits flowers more frequently and does not pierce in about half of the visits, than in caerulescens, which visits flowers comparatively rarely). By contrast, 80% or more of the much more frequent visits to flowers by species of Diglossa involve piercing in the manner described by Skutch (1954). My observations of indigotica agree with its placement in Diglossa – it pierces flowers as do others of this genus and I never saw it take fruit, nor was fruit present in the three stomachs I examined. It has a very deeply hooked bill like that of albilatera, found by Mauck et al. to be its closest relative in Diglossa.

In conclusion, I submit that the combination of genetic, morphological and behavioral data strongly support recognition of Diglossopis as circumscribed by Bock, and recommend a YES on this proposal. I note that recognition of Diglossopis would not affect the ordering of species in Proposal 413 (it would merely place the last three in a different genus).

Gary Stiles, October 2009

Comments from William Mauck: “The first thing I would like to comment on is the statement “Mauck et al. did not examine in detail the relationships of these two (clades of flowerpiercers) with respect to a larger set of outgroups within the Thraupidae”. Although a phylogeny including outgroups is not presented as a figure, these analyses were conducted and described in the paper (see Table1, methods, and results). Twenty-seven species, representing 23 genera, were used as outgroups to root the phylogeny. These representatives were chosen specifically to represent major lineages within Thraupidae (Barker et al. Temporal and phylogenetic patterns of diversity in New World tanagers, cardinals, sparrows, blackbirds, and warblers 2007 AOU, Wyoming) and taxa thought to be closely related to flowerpiercers. The latter were chosen based on molecular and morphological data to test monophyly and elucidate what taxa are sister to flowerpiercers. Therefore, a concerted effort was made to test the monophyly of flowerpiercers and to determine the relationship of flowerpiercers to other Thraupidae. As described in the results, support for the monophyly of flowerpiercers was high, with 98% bootstrap support and 1.0 posterior probability. Thus, the data reject Bock’s idea that the flowerpiercers are not monophyletic.

“In regards to the proposal to split flowerpiercers into two genera, the molecular and relative hook size results are concordant with the internal cranial morphological and distinctiveness presented by Bock 1985 (as stated in Mauck & Burns 2009). Despite this, our paper recommended that all flowerpiercers belong to a single genus, Diglossa. Below is an expanded list of reasons why all flowerpiercers might be retained in a single genus:

“1. Flowerpiercers represent a monophyletic group of nectarivorous birds that evolved from a common ancestor (Mauck & Burns 2009) with an overall bill shape unique from other Thraupidae (Vuilleumier 1969).

“2. Separating flowerpiercers into two genera would emphasize the uniqueness of a specific (and important difference in) feeding morphology of this group, but would also obscure the evolution of other morphological characters and their homoplastic nature, which are currently being investigated (i.e., evolution of the blue plumage and red eye of D. indigotica as compared to D. cyanea, D. glauca, and D. caerulescens; small relative hook size evolving in D. major and D. cyanea, D. glauca, and D. caerulescens; leapfrog plumage pattern; sexual dimorphism, etc).

“3. Finally, the emphasis of a ‘deep division’ between D. cyanea, D. glauca, and D. caerulescens and all other flowerpiercers is a subjective description and was never described as such in Mauck & Burns 2009.  The average uncorrected sequence divergence between D. cyanea, D. glauca, and D. caerulescens and all other flowerpiercers was ~11.6%, whereas the average uncorrected sequence divergence of the clade containing D. indigotica and the baritula superspecies complex to all other flowerpiercers (excluding D. cyanea, D. glauca, and D. caerulescens) is only ~11.3%. Thus, the divergence between the proposed species of Diglossopis and all other flowerpiercers is only slightly larger than that found between other clades of flowerpiercers. This is also illustrated by comparing branch lengths on Fig. 2. Note that the branches involved in the initial split in flowerpiercers are not any longer than other branches in the phylogeny; there isn’t a ‘deep division’ here.

“In conclusion, flowerpiercers are a clade of tanagers with a distinct bill morphology which has evolved into species with different degrees of specialization for feeding on nectar versus fruit. Despite the degree of specialization of bill morphology, nectar thieving is a feasible and documented feeding behavior for many flowerpiercers, even Diglossa cyanea (Moynihan 1963), which is a member of the proposed Diglossopis genus. Fruit is also a food source for flowerpiercers, even those with a large relative hook specialized for thieving nectar (Schondube & Martinez del Rio, 2003). Thus, retaining the genus Diglossa for all flowerpiercers would preserve the shared evolutionary history found among these species.

“The research by Dr. Stiles and colleges explicitly details and quantifies the behavior, feeding preferences, and food sources of flowerpiercers, which is invaluable to understand the ecological role and impact flowerpiercers have in Andean ecosystems. No matter the decision of the SACC it will bring further attention towards understanding the evolution and feeding ecology to this interesting group of birds.”

Response from Stiles: “I thank Mauck and Burns for clarifying the outgroup question regarding the evolution of the flower-piercers: I am willing to accept their argument that they do indeed constitute a clade. However, this does not preclude the division of the clade into two monophyletic genera that are well characterized with respect to morphology, ecology and behavior.  In fact, I feel much more comfortable with this split than I do with the splitting up of Pheugopedius in this respect. It seems worth noting that the sampling of the genome with the molecular markers we have so far is still pretty restricted: one could argue that a considerably larger part of the genome is involved in producing the series of phenotypic characters separating the two groups of flowerpiercers. Perhaps this is a stick-in-the-mud criterion but I do feel that some kind of phenotypic diagnosis of genera is important (and in the case of Cantorchilus with song learning, I am a bit leery of using ONLY song as such a criterion.)”