Taxonomic revision of francolins – reflections from a central African perspective

Ostrich is co-published by NISC (Pty) Ltd and Informa UK Limited (trading as Taylor & Francis Group) The recent publications by Mandiwana-Neudani et al. (2019a, 2019b) have re-evaluated the relationships of several francolin species, which occur across Africa. They have elevated some forms to full species, but the distribution maps they present are different to those in recently published citizen science projects (Harrison et al. 1997; DowsettLemaire and Dowsett 2006; Dowsett et al. 2008). There does not appear to be a single taxon that is represented by more than one genetic sample, which is well outside the norm for robust studies and does not allow for mistakes in the sample labelling or analysis to be detected (Hunter et al. 2021). Hunter et al. (2021) also commented extensively on their interpretations, mainly from an E African perspective, to which Mandiwana-Neudani et al. (2021) have replied. In this commentary, the focus is on the central Africa forms of Coqui Peliperdix coqui, Chestnut-breasted Pe. dewittei, Shelley’s Scleroptila shelleyi, Whyte’s S. whytei and Red-winged Francolins S. levaillantii and Red-necked Spurfowl Pternistis afer. Specimen data provided by Mandiwana-Neudani et al. (2019a, 2019b) were checked with the online Genbank database and the specimen databases of the American (AMNH) and British Museums of Natural History (BM), respectively. This revealed multiple inconsistencies for these taxa, which are outlined below (Table 1), confirming the concerns raised by Hunter et al. (2021). The lowest denominator of the ‘currency’ of conservation (Frankham et al. 2012) is the type specimen. This is the originally described specimen of the taxon, often taken in a series, where it occurred and how it is phenotypically different to similar forms and are not likely to be available for destructive tissue sampling. To get around this, a topotype (a specimen from the type series or another obtained at the type locality or as close as possible to it) should have been used for the CYTB analyses. Plotting the localities of type specimens sets a baseline, against which the geographical variation of plumage across the current distributional range of the taxa under consideration can be made. Type localities cannot be used to determine the distributions of the various taxa in isolation, but it is reasonable to expect that distribution maps should contain the type locality unless significant habitat change has occurred subsequent to the discovery of the species to preclude this. Mandiwana-Neudani et al. (2019a, 2019b) do not appear to have consulted the type specimens, their descriptions or localities, because their approximate distribution maps do not include the type localities of some taxa. Material from topotypes, with one exception were not sequenced either. This was unexpected, because these baseline specimens should be the starting point for any major taxonomic study. Mandiwana-Neudani et al. (2021) express confidence in their organismal matrix as a taxonomically useful set of attributes, irrespective of the numbers investigated per taxon. The type specimens set this baseline and without reference to these any comparisons are subjective. There is much uncertainty about the morpho-vocalisation scores as a result. Up to 10 specimens were examined initially (Mandiwana-Neudani et al. 2019a, 2019b), but in the response to Hunter et al. (2021) they state that more than 10 specimens of their putative terminal taxa were examined. No specimens of Peliperdix coqui stuhlmanni were located in an online search of databases of the AMNH, BM, Chicago Field Museum and Belgian Museum of African History at Tervuren; all museums with extensive African bird collections and it is uncertain whether more than ten reliably identified specimens of this taxon exist. The defining phenotypic identification criterion used to identify Pe. c. stuhlmanni by Mandiwana-Neudani et al. (2019a) is the reduction in breast patterning, but this is not mentioned in the type description, which states that it is very similar to nominate Pe. coqui (Reichenow 1889). A Coqui Francolin specimen identified as Pe. c. stuhlmanni collected in Malawi in the Zimbabwe Museum of Natural History, shows more breast banding, than Pe. coqui from miombo woodland in Zimbabwe, confirming that the breast patterning of central African male Pe. coqui subspecies can be quite variable (Hall 1963, Irwin 1981) and that it is an unreliable taxonomic character (Irwin 1981). This opinion, from an experienced African field ornithologist and taxonomist, has been overlooked. Mandiwana-Neudani et al. (2019a) incorrectly recorded/nominated Pe. coqui as having an unpatterned breast in their morpho-vocalisation scores (character 8; Table 6) suggesting confusion of the specimens examined and/or errors in the data handling. This confusion extends to the CYTB data, where the sequence of Pe. c. stuhlmanni saved in Genbank (Mandiwana-Neudani et al. 2019a, and their Appendix 2, Genbank FR694152) is registered as Pe. c. coqui (Table 1) with stuhlmanni Perspective