The de-coupling via morphology and habitat of Z. (palpebrosus) auriventer from its Malacca Straits paratype population (re-named Z. p. erwini) (Wells 2017) invites a re-opening of the century-old question of where, at species level, taxon auriventer actually belongs. A. O. Hume's type description of Z. auriventer (Hume & Davison 1878) emphasised a bold yellow stripe running from vent to mid-breast level, fronted and flanked by grey. Additional to erwini, however, at least six other white-eye taxa breeding in continental South-East Asia show some evidence of this character. Z. p. williamsoni and Z. p. buxtoni, respectively, of mangrove forest from the north-east (north?) Thai-Malay Peninsula to Cambodia, and inland, mainly slope-land forests of Sumatra and far western Java, are both more similar to erwini than to auriventer, especially in tones of upperparts green. They also show a yellow supra-loral stripe, which is actually broader and more prominent than in erwini, whereas the cap of auriventer is uniformly green with no stripe. In addition, buxtoni has uppertail-coverts, merging onto rump, conspicuously yellow, and an all-black tail (lacking any green on outer webs). These characters reappear in an unidentified Bornean coastal white-eye (Wells 2017) but otherwise are unshared.

The third contender (Figs. 1–2) comprises intermediates between the yellow- and grey-bellied morphs of Z. p. siamensis of more northerly, mainly upland forests, e.g., USNM 534535 and 535703 from montane forest edge on Doi Inthanon, north-west Thailand. Mees (1957) may have had such grey-flanked individuals in mind when he maintained the link between auriventer and species Z. palpebrosus. Over most of its range, however, siamensis shows contrasting yellow above the lores and bill base and, irrespective of underparts colour and pattern, aligns with other northern and western subspecies of Z. palpebrosus in being at least as bright Citrine Green (Smithe 1975) above as erwini, williamsoni and buxtoni, contra the more olive-toned, mid-green auriventer. It also differs from auriventer in wingtip shape (Table 1), and the outer webs of the rectrices of siamensis are broadly green-edged virtually to their tips, as in northern nominate palpebrosus, whereas green edging on the tail of auriventer is no more than a fine fringe reaching not more than halfway towards the feather tips.

TABLE 1

Morphometrics (mm; range and mean) of the Zosterops taxa discussed in this paper. Wing, tail and shortfall of p5 and p9 (descendant) behind wingtip measured as max. chord; tarsus from tarso-tarsometatarsal notch to third toe flexure point; bill from anterior edge nasal groove to tip. Sexes combined (label determinations discounted).

The museum record offers a further, distributional clue to relatedness, centred on Mount Mulayit (2,100 m) in the Dawna Range of Tenasserim (south-east Myanmar), c.200 km north of the Tavoy district type locality of auriventer. L. Fea collected the more northerly of the two accepted specimens of auriventer at 300–400 m elevation just south-west of this peak, whereas on the mountain itself he, and a few years before him Hume's curator W. R. Davison, collected yellow-bellied siamensis, and only siamensis, from the summit zone down to 1,200 m (Hume & Davison 1878, Salvadori 1888). Based on March and April collection dates, potentially, in this area at least, the breeding season ranges of auriventer and siamensis, two different-looking birds, might have approached one another closely. E. C. S. Baker (NHMUK egg collection records) implied actual overlap, arising from nests of siamensis and auriventer both reported at Tavoy itself, but this must be rejected as Baker had acquired his clutches from other collectors without full data. Their identity has not been verified independently below genus level (D. G. D. Russell pers. comm.).

Other, as far as is known entirely allopatric, yellow-striped candidates form what Mees (1954) chose to lump with Everett's White-eye Zosterops everetti Tweeddale, 1878, widespread in the Philippines and (see below) accepted by him as represented by subspecies tahanensis in inland forests of Borneo and the southern Thai-Malay Peninsula, and wetmorei Deignan, 1943, in the northern half, north to 11°40′N on the Thailand / Tenasserim upland divide (Meyer de Schauensee 1946). During mid-March 1966, B. F. King collected three additional wetmorei (USNM 535707-709) in montane forest on Khao Soi Dao Tai, Chanthaburi province, south-east Thailand, and in July 2004 (Pierce & Round 2006) two Z. everetti were mist-netted at 14°26′N in upland Khao Yai National Park, east-central Thailand, on a latitude with confirmed auriventer localities (Fig. 3), but a published photograph of the wingtip of one of the two captures showing disposition of the tips of its outer primaries indicates wetmorei (see below).

In ventral view (Fig. 1), on colour pattern—allowing for some plumage soiling—the two Tenasserim auriventer, Soi Dao Tai specimens, representative wetmorei from the Thai-Malay Peninsula including topotypes, representative tahanensis from the Peninsula and Borneo including the types of its synonyms medius (south-west Sarawak) and parvus (Mount Kinabalu, Sabah) are inseparable. All have white lower wing-coverts, one common tone and distribution of lead grey on breast and flanks, a common extent of ventral yellow, and similar dorso-ventral merging of colours behind the jaw. Dors ally (Fig. 2), all show the same fine green outer-web fringing at the base of rectrices and, except for one yellow-based supra-loral feather on one of the Soi Dao Tai specimens, green of the cap continuous to the lores and bill base, with no interposed yellow.

Figures 1–2.

The Zosterops auriventer grouping and neighbours, in ventral and dorsal views. Top row in each, left to right: auriventer (Tavoy, Tenasserim; holotype); auriventer (Mulayit, Dawna Range, Tenasserim); two wetmorei (Soi Dao Tai peak, south-east Thailand); two wetmorei (Trang, Peninsular Thailand); bottom row: two tahanensis (Main Range, Peninsular Malaysia); medius (Sadong peak, south-west Sarawak, Borneo; holotype); medius (Mount Kinabalu, Sabah, Borneo; holotype of parvus); Z. palpebrosus siamensis (Doi Inthanon peak, north-west Thailand); Z. palpebrosus williamsoni (Ra Island, Peninsular Thailand) (Harry Taylor, © The Natural History Museum, Tring)

Figure 3.

Currently understood geographical range relations of the Zosterops auriventer subspecies and neighbouring Philippine Z. everetti mandibularis and Z. e. basilanicus. Key: 1 —Mulayit peak; 2 — Tavoy; 3 — Khao Yai National Park; 4—Soi Da Tai peak; 5—Mount Tahan; 6—Sadong peak; 7—Mount Kinabalu.

Within this series, auriventer and wetmorei specimens show identical tones and distribution of upperparts green, and measurements demonstrating that the supposedly diagnostically long bill of wetmorei, described from the Peninsula, is replicated north as well as east—making it difficult to accept that these two could be anything except one another's geographical representative. Apart from wingtip shape—shortfall of the outermost large primary (p9 descendant) behind tip <2 mm in auriventer, 3–5 times less than that of p5, vs. p9 and p5 sub-equal in wetmorei (Table 1)—indeed, nothing obvious from morphology stands in the way of linking them at an even lower taxonomic level. As such, it appears that auriventer should be added to Mees' Z. everetti.

The mainland tahanensis sample differs from wetmorei only by its shorter bill (Table 1) and, in most individuals, including all of three from the type locality, Mount Tahan in Taman Negara National Park, Malaysia, by being subtly darker olive-green above (Smithe colour Greenish Olive, no. 49), with even less uppertail-coverts contrast. Mees (1954) included Borneo in the range of tahanensis, but upperparts greens and level of tail-coverts contrast in most Bornean specimens, from east and north-west Sabah and south-west Sarawak, more closely resemble auriventer and wetmorei, only one (of three from Mount Kinabalu) being as dark above as topotypical tahanensis. They also average shorter winged than tahanensis from the Peninsula (Table 1), with a generally finer bill. No good reason has been found not to re-recognise their separate subspecies status, as medius Robinson & Kloss, 1923 (senior to Hachisuka's Kinabalu parvus by three years), type locality Mount ‘Sidong’ (= Sadong), Samarahan Division, south-west Sarawak.

Completing the continental round-up (cf. Stresemann 1939), a link with Black-capped White-eye Z. atricapilla, prominently yellow-striped and almost as similar in its tones and distribution of green, grey and yellow, is ruled out of contention mainly by a combination of black (rather than green) forecrown and face in adults, and, in montane forests of Borneo, up to 800 m of altitudinal range overlap with medius (Mann 2008).

Finally, Mees (1954) came to his view of the geographical range of Z. everetti impressed by similarity of the colour patterns of Bornean ‘tahanensis’ (= medius) and Z. e. basilanicus of Mindanao west to Basilan, at the east end of the Sulu Island chain (Dickinson et al. 1991); also by equivalence of their habitats. Though widely adopted, his merger proposal faced certain difficulties. One, noted by Mees himself, being that Z. e. mandibularis Stresemann, 1931, which occupies the Sulu archipelago range gap between basilanicus and medius, rather than intergrading, diverges from both of these forms in the brightness of its upperparts green and paleness of its grey flanks. Another, not previously raised but apparent from the distribution map, Fig. 8 in Mees (1957), is that no representative has been found on Palawan, a biogeographical ‘stepping stone’ between the Sunda region and oceanic Philippines, or on any of its satellites; nor is one known to occur in the Philippines archipelago anywhere near the eastern end of this dispersal route. Among inland forest passerines accepted as being shared by Borneo and the Philippines such a gap is exceptional, as far as is known being shown only by Rufous-tailed Jungle Flycatcher Cyornis ruficauda and Chestnut-capped Thrush Geokichla interpres, the latter in any case reaching east only as far as the Sulu archipelago itself (Kennedy et al. 2000). Adding in a re-assigned auriventer and additional Thai populations also gives Z. everetti sensu lato a far larger continental Asian range than envisaged by Mees when he proposed the hypothesis of everetti stock having ‘recently’ invaded westward from the Philippines.

Any of three scenarios might apply: (a) Asian continental forms and basilanicus have separately retained hypothetical ancestor-group plumage features lost in other Philippine representatives of Z. everetti, (b) basilanicus does not form a part of Philippine Z. everetti but is a misclassified vicariant population of an otherwise wholly continental species; or (c) in comparable ecological space, similarities are due to convergence. Other tools are needed to tease these apart, but note is taken of size divergence between basilanicus and continental taxa (Table 1) that actually peaks against nearest neighbouring medius. Contra options (a) and (b), basilanicus also retains characters found only among other Philippine populations of Z. everetti (including mandibularis): dusky brownish vs., in continental adults, black lores; a relatively stout bill, especially different from that of medius; pale brown vs. slate-grey feet; and extensive green fringes to the rectrices. In fact, the tail pattern of Philippine Z. everetti is much more like that of siamensis and other northern Z. palpebrosus subspecies than it is of any presumed representative of everetti on the continent. In addition, some Mindanao basilanicus show yellow above the lores (Mees 1957). These populations might be related, but the strictly morphological rationale for species lumping has surely been stretched.

Returning Z. everetti to the status of Philippine endemic would leave nomenclature on the archipelago undisturbed. As first accepted (apparently unwittingly) for an inland forest white-eye by Sharpe (1887a) and recognised as such at least temporarily by several subsequent investigators (e.g., Stresemann 1931), by seniority the name of its continental counterparts would revert to being Zosterops auriventer Hume. Rather than an oceanic island species with a continental bridgehead, this proves to be widespread in inland mixed evergreen forests of mainland continental South-East Asia but has penetrated and / or persisted in only a part of the latter's island sector (absence particularly from Sumatra as yet unexplained). Provisionally, four subspecies must be recognised, the first two with uncertain range limits: nominate auriventer Hume, 1878, known only from central Tenasserim; wetmorei Deignan, 1943, in southern Thailand (northern and eastern range limits as yet unknown) and the northern Thai-Malay Peninsula; tahanensis Ogilvie-Grant, 1906, in the southern Thai-Malay Peninsula; and medius Robinson & Kloss, 1923, on Borneo. ‘Hume's White-eye’, the only already published English name appropriate to an actual species auriventer, is proposed here in recognition of A. O. Hume's original description.

Conclusion

Having cut auriventer adrift from species-level moorings on a combination of morphology and habitat selection (Wells 2017), the search here has been for a best fit of characters among alternative regional contenders, sorted by showing some degree of development of the main character of the auriventer type description—a mid-ventral yellow stripe. Among these, tightness of the fit of a range mainly of plumage features backed by apparently unique habitat equivalence permits a strong presumption that the true relatives of auriventer, at not above subspecies level, are the continental, as opposed to Philippine, forms of Z. ‘everetti’. This and the taxonomic realignments that follow nevertheless still require the support of both molecular genetics and field data such as on vocalisations, especially song. Sampling should be sufficiently broad to address at least the following: (1) the relatedness of nominate auriventer (for which, given the age of museum material, fresh tissue collection probably will be required) and siamensis, testing the proposition that these two are not conspecific; (2) relatedness of nominate auriventer and the erwini / williamsoni pair, testing the proposition that they too are not conspecific; (3) range-wide relatedness within Z. auriventer as re-defined; and (4) the relatedness of medius, mandibularis and basilanicus, testing the Mees hypothesis. A wider investigation of Z. everetti within the Philippines would be expected to follow.