The specimen in Liverpool known as Forbes' Lemon Dove, collected pre-1844 purportedly in Cayenne (French Guiana), was catalogued by Forbes & Robinson in 1900 as Haplopelia principalis, despite this species having been described from the island of Príncipe in the Gulf of Guinea. As a result of the discrepancy in localities, the Liverpool specimen was subsequently described as a new species (Haplopelia forbesi) by Salvadori and suggested to be from West Africa. Over the course of the next century, the new taxon was subject to a variety of taxonomic treatments. To investigate the status and provenance of Forbes' Lemon Dove, we obtained a 472 bp cyt-b sequence from the specimen. This possessed 100% similarity with a Lemon Dove Columba (Aplopelia) larvata sequence from Príncipe and 99.79% similarity with a sequence of the same species from São Tomé. This suggests that Forbes & Robinson were correct that the specimen represents A. larvata principalis and was thus probably collected on Príncipe. However, more sequencing from across the Lemon Dove's range is required to resolve the taxonomy of this complex group and place Forbes' Lemon Dove more definitively.
The islands of the Gulf of Guinea are rich in endemic bird species and are covered by several Key Biodiversity Areas and Important Bird and Biodiversity Areas. De Lima & Melo (2021) recently compiled an updated checklist of birds known from these islands but predicted it would rapidly become outdated due to ongoing taxonomic changes, especially as a result of molecular work. More specifically, the taxonomy of Lemon Doves Columba (Aplopelia) on the islands, and their mainland allies, has been in constant flux. A case in point is Haplopelia forbesi Salvadori, 1904, the type specimen of which has been subject to a variety of taxonomic treatments since the mid 1800s (Table 1).
The holotype of what is now known as Forbes' Lemon Dove was purchased by the 13th Earl of Derby from the London dealers, Leadbeater, in September 1844. The specimen was unsexed, thought to be fully adult, and suspected of having been in captivity as its right wing was clipped and its tail feathers, except the central two, appeared worn at the tips. It bore the label ‘interior of Cayenne’ (French Guiana) but apparently no other information.
In the 13th Earl of Derby's museum stock books, a multi-volume register of specimens, compiled by his curators Louis Fraser and Thomas Moore in the late 1840s (Largen 1987) and now held at National Museums Liverpool, specimen NML-VZ D3567b was grouped (with a question mark) as a female (series no. ‘3567’) with a specimen of Caribbean Dove Leptotila jamaicensis. A single specimen of Caribbean Dove (NML-VZ D3567a) had been in the Earl’s menagerie before it was accessioned into the museum at Knowsley Hall on its death in August 1811. Leptotila jamaicensis is a very different species from Forbes’ Lemon Dove (Fig. 1) but given their proximity in G. R. Gray’s (1844–49) Genera of birds, and the locality on the label, this tentative grouping was understandable. These pigeons were subsequently bequeathed to the people of Liverpool upon the 13th Earl of Derby’s death in 1851, along with most of his substantial natural history collection (World Museum 2021), founding what became the Derby Museum, Liverpool Museums (Morgan 1978) and now World Museum, National Museums Liverpool.
Scientific names used for Forbes' Lemon Dove Haplopelia forbesi Salvadori.
Henry Ogg Forbes, Director of the Liverpool Museums and his assistant, Herbert Christopher Robinson, published a catalogue of the pigeons in the Liverpool Museums in 1900. Identification of NML-VZ D3567b proved especially difficult. Firstly, Forbes & Robinson (1900) must have concluded that it was congeneric but not conspecific with other specimens grouped under Haplopelia as defined by Salvadori (1893). These included the taxa larvata (type locality, modern-day South Africa), simplex (São Tomé), bronzina (Ethiopia) and johnstoni (Malawi) (Fig. 2). Forbes & Robinson (1900) also noted the specimen's close agreement with Columba erythrotorax [sic] Temminck, 1811, said to be from Suriname, but with a key difference: ‘under tail coverts are hoary grey, not cinnamon as in that species’. C. erythrotorax had already been synonymised under H. larvata from South Africa by Gray (1844–49) and Salvadori (1893). ‘With assistance of the British Museum’, Forbes & Robinson (1900) ultimately catalogued NML-VZ D3567b as Haplopelia principalis (Hartlaub 1866), noting that it differed from H. simplex by ‘its more rufous breast’ and from all other species of Haplopelia by ‘its white under tail-coverts’ (Forbes & Robinson 1900).
Peristera principalis Hartlaub, 1866, was described (in Dohne 1866) from a bird collected on Príncipe in 1865 (Jones & Tye 2006). Hartlaub noted that it was nearly allied to his own nomen, simplex Hartlaub, 1849, described from the nearby island of São Tomé. According to Salvadori (1893, 1904), there were no specimens of principalis at the then British Museum (Natural History) (BMNH), however, apparently unbeknownst to Salvadori (1904) a principalis (NHMUK 18220.127.116.11) was accessioned there in 1896. This specimen (Fig. 3) said to be from ‘W. Africa’ and ‘Presented [to BMNH] by the Committee of the Manchester Museum, The Owens College, Manchester’, bears a strong similarity to Forbes’ Lemon Dove. According to the register in Manchester, the specimen was given to Manchester Museum by ‘A. Yarquar’ in 1893. There is no indication of who determined the specimen to be principalis or when.
The wildly incongruous localities of both Príncipe and Cayenne for H. principalis had drawn the attention of Salvadori (1903, 1904) who described Forbes' Lemon Dove as a new species, Haplopelia forbesi Salvadori, 1904, and stated that the type (NML-VZ D3567b) must have originated from West Africa, not South America. A year earlier Haplopelia poensis Alexander, 1903, had been described from Bioko, while earlier in the same year Haplopelia plumbescens Sharpe, 1904, was described from Efoulen, Cameroon as ‘a species very similar to H. principalis’ (Sharpe 1904).
Bannerman (1916) revised Haplopelia in 1916, making a distinction between a larvata ‘section’ and a simplex ‘section’; larvata being found in South and East Africa, and simplex centred on the islands and adjacent mainland of the Gulf of Guinea. Taxa listed under simplex were: s. simplex restricted to São Tomé; s. inornata from Cameroon; s. poensis from Bioko; s. hypoleuca from Annobón; s. plumbescens from southern Cameroon; and s. jacksoni from Uganda. Taxon forbesi (represented by NML-VZ D3567b and NHMUK 1818.104.22.168) from West Africa, was retained as a species and principalis, confined to Príncipe, and being pinkish instead of grey, was kept as a separate ‘well differentiated race’, i.e. not considered conspecific with H. simplex but part of the ‘section’.
Fifteen years later, Bannerman (1931) still treated principalis as a species but was ready to include forbesi as a subspecies under simplex and, although hesitant to ‘ally forbesi with simplex plumbescens due to the rusty-red tint to the plumage’, mentioned the latter as a potential synonym. Peters (1937) tentatively did synonymise forbesi, whilst retaining all of Bannerman’s other subspecies, but placed principalis (still treated as a species by Bannerman 1931) under simplex.
Until the 1950s, Bannerman's two ‘sections’, simplex and larvata, were generally followed, with the simplex group having greyish males and much paler brown females. However, a population with apparently overlapping characteristics was found by White (1948) in modern-day Zambia. From females, White (1948) had thought birds in this region were akin to larvata (Amadon 1953), however on obtaining a male described it as Aplopelia simplex samaliyae (see White 1948), but subsequently (in a letter to J. Chapin seen by Amadon 1953) he concluded that this demonstrated that larvata and simplex were conspecific. Amadon (1953) recognised only two taxa in the Gulf of Guinea: A. larvata simplex (São Tomé) and A. l. principalis (Príncipe). Meanwhile Bannerman (1953) retained four ‘races’ of simplex, plus principalis as a species, in the Birds of West and Equatorial Africa, but made no mention of forbesi, presumably following Peters (1937) in lumping it with plumbescens. On the basis of overlapping plumage characters among specimens at BMNH, Serle (1959) made a strong case for synonymising inornata, plumbescens and poensis with simplex as the only subspecies under larvata (following Amadon 1953), but thought they were usefully retained as designators of geographic origin. Due to lack of precise locality information, forbesi was not included in his assessment, while the validity of principalis was not questioned.
By the 1960s the conspecificity of larvata and simplex was well accepted, and Goodwin (1967) treated all of the taxa in Bannerman's (1916) simplex ‘section’ under larvata, without mentioning forbesi. According to Wagstaffe (1978), Derek Goodwin examined NML-VZ D3567b in 1975 and had no doubt the specimen was an immature female of larvata sensu lato, but with subspecies indeterminate. To our knowledge, the type specimen of Forbes’ Lemon Dove has not been taxonomically reassessed since then. Jones & Tye (2006) and de Lima and Melo (2021) listed C. l. principalis on Príncipe and C. l. simplex on São Tomé.
Recently, DNA analysis has been applied to the question of Lemon Dove taxonomy. Pereira (2013) sequenced three mitochondrial (NADH dehydrogenase subunit 2, cyt-b, cytoxidase I) and one nuclear gene (β-fibrinogen intron 7) from two individuals of Lemon Dove (Aplopelia larvata sensu lato) from São Tomé, two from Príncipe, and a museum specimen from Malawi. The sequences are unpublished, but the figured (Bayesian Inference) mitochondrial tree reveals each island population as monophyletic but highly similar to each other, and 4.3% distant (uncorrected ‘p’) from the mainland sequence.
The only sequences currently available on NCBI GenBank for Lemon Doves were generated as part of a global survey examining patterns of diversity in island bird communities by Valente et al. (2020). Cyt-b sequences were obtained from samples taken from three Lemon Doves on São Tomé (MH307437–439) and one on Príncipe (MH307436). The published tree showed the single sequence from Príncipe nested within those from São Tomé (note, sequence MH307437 from São Tomé is incorrectly labelled principalis on the published tree; see supplemental material in Valente et al. 2020).
To investigate the status and provenance of Forbes' Lemon Dove further, we obtained a 472 bp cyto-b sequence from NML-VZ D3567b. We designed four new pairs of primers specifically for Haplopelia (HapF1-3; HapR1-3; registered in the BOLD Primer Database www.boldsystems.org), each amplifying c.150 bp then concatenated the resulting sequences. Our molecular methods otherwise followed those of Senfeld et al. (2019). The sequence is published in NCBI GenBank (Sayers et al. 2021) under accession no. MZ438312. The Forbes' Lemon Dove sequence was ‘blasted’ against GenBank and showed 100% similarity with MH307436, from a Lemon Dove at Oquê Daniel, Príncipe. The next most similar sequence, MH307438, from a Lemon Dove at Contador Dam, São Tomé, showed 99.79% similarity (1 bp difference) to the sequence of forbesi.
As in the published tree of Valente et al. (2020), the cyt-b Neighbor Joining tree generated using the BLAST web app of the 100 sequences producing significant alignments (Zhang et al. 2000) showed the sequence from Príncipe (principalis) nested among sequences from São Tomé (simplex) (Fig. 4). The distance between the outlying simplex sequence (MH307437) and the cluster of the other four samples is relatively large (>2%) and of a magnitude generally seen between species, whilst the distance between principalis + forbesi, and the two other clustered simplex samples is relatively short (>1.5%) and more indicative of intraspecific variation at the mitochondrial locus (e.g. Johnsen et al. 2010). Although the 100% similarity of forbesi to the principalis sample tends to confirm its correct placement with the Príncipe population, the nesting of the morphologically distinct principalis within simplex is biogeographically counter-intuitive. More sequencing from across the range of larvata and simplex is ultimately required to resolve the taxonomy of this complex group.
Our result is, however, consistent with Forbes & Robinson's (1900) original determination that NML-VZ D3567b is a specimen of principalis. If so, given that principalis is restricted to Príncipe, it too would almost certainly have been collected there. Forbes' Lemon Dove matches females of principalis fairly well in size and colour, and some specimens of principalis have white/hoary grey undertail-coverts like those of NML-VZ D3567b (Bannerman 1931, Wagstaffe 1978). Based on the collection date we speculate very tentatively that NML-VZ D3567b was collected during the British Government Expedition to the River Niger 1841–42, which used the Gulf of Guinea islands as a base (Tye & Jones 2006). The provenances of many birds collected during this expedition were incorrectly recorded (Tye & Jones 2006) although how the specimen found its way to Leadbeater labelled ‘Cayenne’ is a mystery. De Lima and Melo’s (2021) prediction of further taxonomic changes to the checklist of birds from the islands of the Gulf of Guinea seems well founded.
Alex Bond (NHMUK) and Rachel Petts (Manchester Museum) provided information about the only other specimen known to have been identified as Haplopelia forbesi. We are grateful to Martim Melo and Luís Lima Valente for access to Hugo José Eira Pereira's M.Sc. thesis and information about recent Lemon Dove samples collected in the Gulf of Guinea. We are indebted to Peter Jones for invaluable comments on the manuscript. Robert Prŷs-Jones, Alan Tye and an anonymous reviewer provided very helpful suggestions on the submitted draft.