Recently described vocal variation within the monotypic Yellow-spotted Barbet Buccanodon duchaillui has been used to suggest the presence of two allopatric species separated by the Dahomey Gap in western Africa. Using mitochondrial and nuclear DNA sequences from two genes, we investigated molecular patterns of divergence across the species' range, in light of the published vocal variation. We found support for a genetic break at the Dahomey Gap, but also identified much deeper divergence among other populations in the eastern part of the species' range. Deep genetic divergence, and geographic variation in the species' vocalisations, suggest a greater degree of diversity in this species than currently recognised.
Yellow-spotted Barbet Buccanodon duchaillui occurs in forested regions of tropical Africa, from Sierra Leone east across the Congo Basin to Kenya (Short et al. 2020). The western and eastern populations are separated by the Dahomey Gap, a dry forest-savanna break within otherwise contiguous lowland tropical rainforest (e.g., Salzmann & Hoelsmann 2005, Demenou et al. 2016, Dowsett-Lemaire & Dowsett 2019). The species was described by Cassin in 1855 based on specimens taken along the Mondah (Moonda) River in Gabon. Subsequently, subspecies ugandae was described from the western base of the Ruwenzori Mountains in Uganda based on its lack of yellow spotting on the back (fide Chapin 1939; Reichenow, 1911, Wiss. Ergebn. Deutsche Zentral-Afr. Exped. III: 278); subspecies gabriellae was described from specimens taken in Pangala, ‘French Congo', c.80 miles north-west of Brazzaville, based on multiple plumage differences including ‘the feathers of the forehead bright scarlet-vermilion instead of crimson' compared to the nominate (Bannerman 1924); and subspecies bannermani was described by Serle (1949: 52) from the ‘Highlands of the Bamenda Division, British Cameroons’ and differentiated by its ‘larger size’ vs. the nominate. See Fig. 1 for mapped type localities of these subspecies. Chapin (1939: 507) considered ugandae invalid ‘as yellow spots are not always wanting on the upper back of Uganda birds’, but affirmed that subspecies gabriellae was valid due to the light red coloration of the crown patch. White (1965) considered bannermani to be invalid and Short & Horne (1988, 2001) treated the species as monotypic for no given reason, thereby subsuming gabriellae, but noted that ‘Birds at higher elevations are larger than lowland birds’ (Short & Horne 1988: 442). The species is currently usually treated as monotypic (e.g., Dickinson & Remsen 2013, Gill et al. 2020, Short et al. 2020). Differences in the vocalisations of the western and eastern populations were first noted by Borrow & Demey (2001). Boesman & Collar (2019) investigated this variation using the number of notes, length of longest note, pace of notes, and acceleration. Following criteria published by Tobias et al. (2010), they concluded that western and eastern populations should be recognised as separate species: Western Yellow-spotted Barbet B. dowsetti, occurring west of the Dahomey Gap, and Eastern Yellow-spotted Barbet B. duchaillui, to the east of it. Gill et al. (2020) did not accept the newly proposed species B. dowsetti, citing the need for further work, including genetic analysis.
Using DNA sequence data, we investigated patterns of genetic divergence within the Yellow-spotted Barbet to determine if these patterns matched those in vocal variation outlined by Boesman & Collar (2019). Based on Boesman & Collar's (2019) conclusions and previously recognised biogeographic patterns across the Dahomey Gap, we hypothesised that molecular evidence would support differentiation between western and eastern populations.
We used 11 specimen-vouchered tissue samples of B. duchaillui housed at the Univ. of Kansas Natural History Museum, Lawrence, USA (KU) and the Field Museum of Natural History, Chicago, USA (FMNH) from across the species' distribution: one sample from Sierra Leone, two from Ghana, four from Cameroon, three from Equatorial Guinea, and one from Uganda (Table 1). Samples from Sierra Leone and Ghana came from the range of the proposed western species and the eight remaining samples from that of the proposed eastern species (following Boesman & Collar 2019; Table 1). We used a White-eared Barbet Stactolaema leucotis (blood sample, FMNH A92024, GenBank AY279277.1) from Kenya as an outgroup sample.
We extracted genomic DNA using a manual magnetic bead-based protocol ( https://github.com/phyletica/lab-protocols/blob/master/extraction-spri.md) based on Rohland & Reich (2012), and eluted DNA from beads using 1X TE buffer. We amplified the mitochondrial gene cytochrome b (cytb) using primers L14841 (Kocher et al. 1989), H4a (Harshman 1996), barbCBL (Moyle 2004) and barbCBH (Moyle 2004). We also amplified the nuclear region Beta Fibrinogen intron 7 (β-fibint7) using the primers FIB-B17L and FIB-B17U (Prychitko & Moore 1997). We amplified both genes using a touch-down type polymerase chain reaction protocol (DeCicco et al. 2020). Amplified DNA was sequenced by Genewiz. Consensus sequences have been uploaded to GenBank (Table 1).
Samples of Yellow-spotted Barbet Buccanodon duchaillui used in this research. All specimens are from the Univ. of Kansas Natural History Museum, Lawrence, except for the specimen from Uganda which is housed at the Field Museum of Natural History, Chicago. GenBank numbers refer to archived sequence data for the mitochondrial gene cytochrome b.
Average pair-wise molecular distances among sampled populations of Yellow-spotted Barbet Buccanodon duchaillui.
We used Geneious (Kearse et al. 2012) to trim, align, and create consensus sequences. Multi-sequence alignments were made using MAFFT (Katoh et al. 2002) in Geneious. We identified codon partitions and models of evolution using Partition Finder 2 (Lanfear et al. 2016) based on AICc scores. We estimated phylogenetic relationships using maximum likelihood methods in RAxML (Stamatakis 2014) run for 1,000 bootstrap replicates with previously identified by-codon partitions and the General Time Reversible + Gamma model of sequence evolution. We also used MrBayes (Huelsenbeck & Ronquist 2001) running four chains for one million generations, sampling every 1,000 generations with a burn-in of 0.25 using previously identified optimal partitions and models of sequence evolution. We calculated uncorrected pair-wise molecular distances among clades identified in our phylogenetic analysis in PAUP* (Swofford 2003).
We obtained complete gene sequences for both cytb and β-fibint7 for all 12 samples used. Because the β-fibint7 DNA sequence data provided almost no informative signal for phylogenetic analysis or in a haplotype network, we present results only from our cytb data. Using Stactolaema leucotis as the root, phylogenetic analyses placed the Ugandan sample of B. duchaillui as sister to all other populations, and the Equatorial Guinea samples in a clade sister to the Cameroon, Ghana and Sierra Leone samples. The Cameroon samples were in turn sister to the Ghana and Sierra Leone birds (Fig. 1). Bootstrap support was moderate to high (≥75%) for all nodes in the phylogeny. Genetic divergence in cytb was generally low within labelled clades (<1%) but substantial between clades. For example, the single sample from Uganda was 8–10% divergent from all other samples (Table 2). Divergence between clade C and clades A and B was c.6.5%. Divergence across the Dahomey Gap, the putative geographic division between B. duchaillui and B. dowsetti, was 4.2%.
Our results, based on the mitochondrial cytb gene, highlight a genetic break congruent with the vocal differences noted by Boesman & Collar (2019), consistent with their taxonomic suggestion to treat these populations as two species. However, our results also suggest a more complex evolutionary history for the Yellow-spotted Barbet than simply a Dahomey Gap split and a more complex pattern of molecular divergence than indicated by vocal variation alone, despite largely congruent sampling of vocal and genetic data. Genetic and vocal divergence across the Dahomey Gap has been reported in other bird species, but this pattern is variable among species (e.g., Fuchs & Bowie 2015, Kirschel et al. 2020).
Given this complexity, it is difficult to align our results directly with the simple Dahomey Gap split in vocal variation. We find it noteworthy that Boesman & Collar (2019) found the same vocal dialect in all sampled populations east of the Dahomey Gap, populations among which we found up to 10% average pair-wise molecular divergence. This clearly suggests that vocal and genetic variation in this species are decoupled. Denser genetic sampling east of the Dahomey Gap would be valuable to determine more precisely where genetic breaks occur in an otherwise apparently continuous distribution. Such sampling would also provide the ability to assess if this system follows expectations under Pleistocene rainforest refugia hypotheses (see Diamond & Hamilton 1980, Mayr & O'Hara 1986); however, the sampling to date suggests that this system may align with patterns expected under isolation in the three proposed Pleistocene refugia.
Both the vocal analysis provided by Boesman & Collar (2019) and our results suggest greater diversity within this species than previously thought. Discordance between the geographic patterns presented by vocal variation and that of genetic variation are not unexpected (e.g., Nwankwo et al. 2018). The complexities of this system presented jointly by the vocal (Boesman & Collar 2019) and molecular variation suggest that this taxon merits further research. How the vocal and genetic variation in a broader sense fit with the described, but not recognised subspecies, is beyond the scope of this note. Additional, denser genetic sampling is required to fully address this question. Clearly, due to the described plumage variation, particularly in subspecies gabriellae, there is probably cause to recognise more geographic forms, especially if genetic variation supports some of the described patterns in plumage or vocal variation. We believe a more thorough analysis of taxonomic history, plumage variation and genetic variation, the latter with denser geographic screening, is required to make adequate taxonomic suggestions. We hope that the information presented here, in conjunction with that in Boesman & Collar (2019), provides some insight into the previously unrecognised diversity within the Yellow-spotted Barbet.
We thank the field researchers from KU and FMNH along with local collaborators that provided the samples used in this research. John Bates, Lincoln Fishpool, an anonymous reviewer, and Guy Kirwan all provided helpful reviews of this manuscript, and we appreciate their input. Portions of this research were supported by the National Science Foundation (DEB-1557053 to RGM and the Graduate Research Fellowship Program to ENO).