Three of the world's four main avian checklists, two recent monographs of the Cuculiformes (Payne 2005, Erritzøe et al. 2012) and a handbook to the birds of the Malagasy region (Safford & Hawkins 2013) all treat the Crested Coua Coua cristata of Madagascar as consisting of four subspecies, arranged anti-clockwise as nominate cristata in the east, north and north-west, dumonti in the centre-west, pyropyga in the south-west and south, and maxima in the south-east (Dickinson & Remsen 2013, Clements et al. 2023, Gill et al. 2024). The fourth list splits pyropyga as Chestnut-vented Coua but treats the remaining three subspecies as conspecific (del Hoyo & Collar 2014, HBW & BirdLife International 2024). It is certainly the case that pyropyga is well differentiated from nominate cristata and subspecies dumonti (these two latter being very similar to each other, to the point where we speculate if they might form components of a cline), but it is the taxon maxima that in several ways is the most distinctive of the four. However, this last is known from a single specimen, which inevitably reduces the confidence with which a defensible position on its taxonomic status can be taken. Consequently it has been left as an unaddressed issue for many years, with checklists quietly parking it until further light can be shed. Here we attempt to provide a little more context and clarity to this interesting case, prior to and preparatory for the obviously needed step of a molecular analysis.

Three examinations of the holotype

Coua cristata maxima was established by Lt. Col. Philippe Milon based on a male (inexplicably both Erritzøe et al. 2012 and Goodman 2013 state the specimen is unsexed) which he collected at Fort Dauphin (now Taolagnaro or Tolagnaro), in far south-east Madagascar, on 18 February 1948 (Milon 1950); he later mentioned that he found it in humid forest (Milon 1952). He diagnosed it (in French; all quotations from the original description are our translations) on the basis of both size and colour. As its name indicates, the holotype in the Muséum national d'Histoire naturelle, Paris (MNHN-ZO-1950-392; Figs. 1–5), proved larger than the largest of the known subspecies of Crested Coua C. c. pyropyga (in the following sequence of measurements, in millimetres, maxima is first vs. pyropyga, with the latter's values expressed as means of 14 specimens): bill (from commissure) 30.0 vs. 26.7, tarsus 45.0 vs. 41.2, wing 175 vs. 162, tail 232.5 vs. 212.0 (Milon 1950). None of the highest values of the pyropyga sample were as high as those for maxima (this is also true in the independent sampling of the former undertaken by Benson et al. 1976–77); moreover, the width of the central rectrices of maxima proved notably greater than those of all three other subspecies (45 vs. <40), and various other, unquantified indications of its greater size involved ‘the thickness of the tarsi, the width of the back, etc.’ (Milon 1950). In plumage, the holotype was deemed to have four basic points of difference from C. cristata (Table 1).

Figure 1.

Holotype of Coua cristata maxima (MNHN-ZO-1950-392) in ventral view; note the loss of all undertail-coverts (Guy M. Kirwan)

Figure 2.

Holotype of Coua cristata maxima (MNHN-ZO-1950-392) in lateral view (Guy M. Kirwan)

Figure 3.

Holotype of Coua cristata maxima (MNHN-ZO-1950-392) in dorsal view (Guy M. Kirwan)

Figure 4

(left). Holotype of Coua cristata maxima (MNHN-ZO-1950-392), view of right side of head (Guy M. Kirwan)

Figure 5

(right). Holotype of Coua cristata maxima (MNHN-ZO-1950-392), view of left side of head (Guy M. Kirwan)

In the field the absence of white on the underparts and the presence of blue in the upperparts was striking, such that when Milon (1950) first saw the bird (when it must have been facing away from him) he thought it was a Blue Coua C. caerulea. No other specimen exists, but Milon (1950) mentioned that when the holotype was collected ‘other individuals of the same form were seen in the vicinity’.

One or more of the authors of Goodman et al. (1997), reconsidering the case of maxima, examined the holotype and compared it with other material of cristata. This involved plumage descriptions that extended but also slightly modified the diagnosis in Milon (1950), focusing solely on details of the upperparts and underparts (Table 1), and referring to a particularly frustrating feature of the specimen, not mentioned by Milon himself, which is that it had ‘lost most of the undertail coverts during preparation’. With this very unusual and unfortunate circumstance, a crucial piece of evidence concerning maxima's relationships to the other taxa in the C. cristata complex—chestnut/rufous or white/buffy undertail-coverts—has been lost (witness Fig. 1). Goodman et al. (1997) made the ambiguous remark that Milon (1950) had ‘noted that the ventrum of the maxima specimen was tawny cinnamon with no reddish coloration at the base of the tail’, which must be the source of the mistaken assertion in Payne (2005) that the ‘under tail coverts lack rufous’. It is true that Milon (1950) described the belly (‘ventre’) of maxima as ‘cinnamon fawn’, but he made no comment about the colour at the base of the (under)tail. However, an illustration in Milon (1952) of all four subspecies of C. cristata (Fig. 6) shows the undertail-coverts of maxima the same colour as in pyropyga (and indeed of the belly of maxima). It is tempting to assume that this was based directly on the recently collected holotype, but the loss of the undertail-coverts after preservation in Paris seems less likely than their loss, as Goodman et al. (1997) said, during preparation or when the bird was shot. At any rate, this image must be assumed to have given Goodman (2013), in his account of maxima’s characters (condensing those in Goodman et al. 1997), the confidence to report ‘lower breast to vent tawny-brown (not whitish) with increasing colour saturation’.

TABLE 1

Schematised diagnoses of Coua cristata maxima based on three detailed reviews; comparisons throughout are with other C. cristata taxa.

From our own examinations of the holotype (NJC in 2013, NJC & GMK in 2023)—when we could find no undertail-coverts—we confirm the diagnostic characters identified by Milon (1950) and Goodman et al. (1997), sometimes with slight modifications (Table 1; we use ‘cinnamon-tawny’ for the rather beautiful colour of the breast). However, we were and remain struck by the rather different-looking configuration of plumage and pattern around the face of maxima. Milon (1950) stated that ‘the colours of the … bare skin around the eye’ were as in ‘the other subspecies’, but the holotype itself is hardly supportive: (1) it possesses a relatively weak and diffuse black superciliary line dividing the crested crown from the naked head-sides, and (2) it appears to lack a large area of bare skin around (and especially above) the eye (Figs. 4–5). In C. cristata (s. l.) the bold superciliary line and wide bare periocular skin are as obvious in museum specimens (see Figs. 7–8) as they are in photographs of live individuals (e.g. Fig. 9). They are, by contrast, so little apparent in the holotype of maxima that we must query its illustration in Safford & Hawkins (2013) and Hawkins et al. (2015) (where, incidentally, the undertail-coverts are shown as rufous).

Figure 6.

Illustration of the four subspecies of Coua cristata (left to right: nominate cristata, dumonti, pyropyga and maxima) in Milon (1952, plate IV).

Figure 7

(left). Three heads of specimens of Coua cristata cristata held in the Natural History Museum, Tring, top to bottom: NHMUK 1931.8.18.515, 1931.8.18.513, 1931.8.18.510 (N. J. Collar, © Trustees of the Natural History Museum, London)

Figure 8

(right). Three heads of specimens of Coua [cristata] pyropyga held in the Natural History Museum, Tring, top to bottom: NHMUK 1931.8.18.536, NHMUK 1931.8.18.541, NHMUK 1931.8.18.542 (N. J. Collar, © Trustees of the Natural History Museum, London)

Figure 9

(left). Crested Coua Coua cristata cristata, Ankarafantsika National Park, October 2019 (detail) (Paul van Giersbergen)

Figure 10

(right). Blue Coua Coua caerulea, Ranomafana National Park, November 2019 (detail) (Paul van Giersbergen)

Options for classification

What, then, are the options for the classification of Coua cristata maxima? Del Hoyo & Collar (2014) exposed the issue when they split pyropyga from cristata and dumonti on the basis of its pale chestnut or strong rufous vs. whitish or buffy-white undertail-coverts, paler upperparts (especially tertials), much longer white tail tips and larger size (see Figs. 11–12). This arrangement offered a suite of options for the treatment of maxima, arguably the least coherent of which was to leave it, as Del Hoyo & Collar (2014) did, as an outlying subspecies of the rather dissimilar C. cristata. Three alternatives were suggested, to which we here add a fourth (the first in the following list): a subspecies of the more similar-sized C. pyropyga; a taxon requiring species rank of its own; an ‘aberrant individual of a known form’ (Goodman 2013); or a hybrid (a possibility first raised by Goodman et al. 1997). We consider these five options here, dwelling longest on the last, after reviewing the issue of the age of the bird at the time of collection.

Length of crest and degree of feathering around the eye are gauges of age in couas (Benson et al. 1976–77, Goodman 2013; R. B. Payne in litt. 2024), so it may be that maxima's comparatively short crest and apparently largely feathered periocular area are signs of its immaturity. However, Milon (1950) explicitly indicated that the ‘colour of… the bare skin around the eye’ of maxima was ‘as in the other races’, and indeed it was illustrated as such in Milon (1952; see Fig. 6). The current condition of the head-sides in maxima may therefore simply be attributable to shrinkage of the skin over time (H. van Grouw and R. B. Payne in litt. 2024). So is the holotype of maxima adult or immature? Bare skin was visible when the feathers on the head-sides were moved aside, but the extent of it was extremely hard to gauge. We are therefore frankly unsure, as we have examined many specimens in which such a contraction has not occurred over comparable or longer periods of time: every one of the 21 specimens of nominate cristata, ten full-grown dumonti and 11 pyropyga held in the Natural History Museum, Tring, UK (NHMUK), was taken years before the holotype of maxima but has unreduced bare periocular skin (examples in Figs. 7–8).

Figure 11.

Ventral view of three Coua [cristata] pyropyga and three C. c. cristata specimens in the Natural History Museum, Tring, top to bottom: NHMUK 1931.8.18.542, NHMUK 1931.8.18.541, NHMUK 1931.8.18.536, NHMUK 1931.8.18.510, NHMUK 1931.8.18.513, NHMUK 1931.8.18.515 (N. J. Collar, © Trustees of the Natural History Museum, London)

Subspecies of C. cristata.—This is the status quo, whether C. pyropyga is accepted as a separate species or not. Because the elevation of C. pyropyga to species rank isolates maxima from the remaining subspecies of C. cristata, del Hoyo & Collar (2014) judged that the case for also treating maxima as a species was compelling. However, this was to miss the consideration raised by R. B. Payne (in litt. 2024) that in its darker upperparts and mid-sized white tail tips maxima is taxonomically closer to nominate cristata than it is to pyropyga, and that nominate cristata, known to occur 200 km north of Taolagnaro (R. J. Safford in litt. 2024), might extend in undetected pockets of secondary habitat south throughout Madagascar's eastern humid forests (as vaguely speculated in Rand 1936, Milon 1952), making it as close geographically to maxima as pyropyga seems to be. Nevertheless, we are sceptical, for at least three reasons: (1) the size difference of maxima vs. nominate cristata is greater than it is vs. pyropyga (Benson et al. 1976–77 offer independent means: nominate cristata wing 136, tail 185 and pyropyga 162, 207 vs. maxima's 172, 224—C. W. Benson's own measurements); (2) the likelihood that maxima shared rufous undertail-coverts with pyropyga appears rather greater than that it shared white ones with nominate cristata; and (3) the known distance between the type locality of maxima and the easternmost occurrence of pyropyga is far smaller.

Figure 12.

Dorsal view of the same three Coua [cristata] pyropyga and three C. c. cristata specimens, as in Fig. 11, in the Natural History Museum, Tring, top to bottom: NHMUK 1931.8.18.542, NHMUK 1931.8.18.541, NHMUK 1931.8.18.536, NHMUK 1931.8.18.510, NHMUK 1931.8.18.513, NHMUK 1931.8.18.515 (N. J. Collar, © Trustees of the Natural History Museum, London)

Subspecies of C. pyropyga.—The last three points—relatively close in size, probably identical in colour of the undertail-coverts, and known geographical proximity—make a fair case for treating maxima as a form of pyropyga, if the latter is treated specifically and despite its overall paler coloration.

Species.—The possibility than maxima could be a ‘distinct species’ was glancingly mentioned by Goodman et al. (1997), rather strongly endorsed (‘may well represent a separate species’) by Goodman & Wilmé (2003), and echoed by Goodman (2013) and Safford et al. (2022). Under the Tobias criteria (Tobias et al. 2010), where a cumulative score of 7 is required for species rank, we allow the larger size (always conceding n = 1) a score of 2, the cinnamon-fawn lower breast and belly 3, and the blue vs. green inner remiges and tertials 2, thus a total of 7. If the shorter crest (allow 2) and more feathered head-sides/weaker superciliary line (allow 2) are not discounted as signs of immaturity, the total score could rise to 11.

Aberrant.—The concept of the ‘aberrant’ specimen rose to prominence with the steady trawl of described taxa by the Peters (1931–87) checklist, whenever a plausible treatment of forms known only by one or two specimens was needed. Over time, a few such taxa have been shown to be based on genuinely ‘aberrant’ specimens, e.g. Hooded Seedeater Sporophila melanops (Areta et al. 2016), but the majority of cases remain unresolved. However, the ascription of ‘aberrant’ to a taxon almost invariably involves colour variation and rarely if ever size; even the phenomenon of ‘runt’ (small, seemingly malformed) birds has rarely if ever been invoked as a confident explanation of an anomalous taxon. Certainly the circumstance of an aberrant bird that is larger than the form to which it is judged to belong, as well as differently coloured, appears simply to be undocumented in wild birds. The fact that its discoverer saw at least one other similar bird only diminishes the aberration hypothesis further, to the point where we elect to discard it.

Hybrid.—The possibility that the holotype of maxima is a hybrid, first suggested by Goodman et al. (1997), then repeated by Payne (2005) and Erritzøe et al. (2012), needs exploration. The position of the type locality, Taolagnaro (25°01′S, 46°59′E; elevation 0–40 m), allows for the following members of the genus Coua, having ranges which are known or thought to approach the city within several tens of kilometres, to be considered as potential parents: Crested Coua C. [cristata] pyropyga, Blue Coua C. caerulea, Red-fronted Coua C. reynaudii, Olive-capped Coua C. [ruficeps] olivaceiceps (also split in del Hoyo & Collar 2014), Running Coua C. cursor and Giant Coua C. gigas. (Even C. cristata cristata might be considered—see above under ‘Subspecies of C. cristata’.) Goodman et al. (1997) took the view that C. [cristata] pyropyga had to be one parent, given the overall resemblance of maxima to the phenotypes represented within the cristata complex that led Milon to treat it trinomially; and they suggested that the other might be C. caerulea or C. reynaudii. These two are certainly stronger candidates than the other three on the list, mainly because they are, like C. cristata (s. l.), arboreal rather than terrestrial couas: the prospect of a successful interbreeding of two species with such different ecological adaptations seems remote.

Judging hybrid status involves the scrutiny of organisms for morphological intermediacy. Such intermediacy is not necessarily exact, but it is typically present in certain characters such that a diagnosis can be made with some confidence (Wilson 1990, Estabrook et al. 1996, Hennache et al. 2003, Gholamhosseini et al. 2023). In this respect plumage considerations further reduce the chances of a terrestrial coua being involved in the pedigree of maxima: all three species have heavy facial markings like cristata (s. l.), so would hardly be likely to produce an offspring without them, and olivaceiceps has a creamy throat and lilac-grey breast while cursor has a buffy-rufous throat and a grey breast. Size is another consideration: assuming the intermediacy of the male maxima between the smaller pyropyga and a larger other parent, and using wing length (no difference between the sexes) as an index (mean male data in mm from Goodman 2013), we have 135 for cursor, 137 for reynaudii, 162 for pyropyga, 169 for olivaceiceps, 175 for maxima, 194 for caerulea and 215 for gigas. Clearly all three terrestrial birds fail this test, cursor and olivaceiceps by being far too small and gigas far too large.

Of the two arboreal options, however, reynaudii can be rapidly eliminated on the basis of characters that find no evidence of expression or intermediacy in maxima, namely its overall dark green plumage, rufous crown, heavy black supercilium and relatively short bill and wings. This then leaves C. caerulea as the only plausible candidate to be the second parent of maxima. Blue Coua is coloured throughout a rather deep, soft blue, muted by the slightest suffusion of grey on the head and breast, while the wings and tail are somewhat glossy, the former with a violet tinge that can appear greenish at some angles, the latter shaded rich violet and lacking white tips; the skin above, below and behind the dark eye is a paler, brighter blue, circumscribed by a feathered black line. The notion that such a uniform bird, much larger than any extant congener except Giant Coua, could be one parent of maxima might seem outlandish at first. Nevertheless, it bears mention that molecular analysis has found Blue and Crested Couas more closely related to each other (except perhaps the unsampled Verreaux's: see genus account in Goodman 2013) than to other couas (Johnson et al. 2000), and the list of (six) characters in maxima which are consistent with caerulea as a progenitor is striking:

The retention of white tail tips in maxima and the fact that these tips are intermediate in size between pyropyga and dumonti is beyond our capacity to explain.

Extant, extinct, illusion?

If Coua [cristata/pyropyga] maxima is a valid taxon, the question arises whether a population representing it might still be extant. Goodman et al. (1997) ominously reported that ‘Most of the natural lowland forest in the immediate vicinity of Tolagnaro has been destroyed’, without inferring a consequence. Six years later Goodman & Wilmé (2003) were more categorical, even while conceding that some forest still stood: ‘The remaining forest blocks surrounding Tolagnaro are ornithologically well known, and it is certain that this form is extinct’. Even so, after another ten years Goodman (2013), who attributed the bird’s disappearance from Tolagnaro to hunting (because, after all, ‘seemingly appropriate forested habitat remains in this region’), reflected that the form might still be found in forest stretching 200 km to the north: ‘The [hinterland] forest from N of Manafiafy [35 km northeast of Taolagnaro north] to at least the Manantenina area or perhaps even Vangaindrano, which is ornithologically poorly known, holds potential for finding a remnant population of maxima and such exploration should be given high priority’. If that possibility is to be seriously entertained—and we are unaware that anyone has yet done so, or anywhere repeated the idea—our own inclination would be to consider the population as representing a full species, which would increase the urgency with which a new search should be undertaken.

For reasons stated above, we find the option that Coua cristata maxima is based on an aberrant individual unconvincing. On the other hand, the possibility that it is a hybrid C. cristata × C. caerulea appears plausible to us, but we emphatically do not consider the idea to be incontrovertible. One objection is that Milon (1950) saw other birds in the vicinity similar to the one he collected; a possible counter is that he simply saw a second bird that made more than one appearance while he was stalking the first, in which case (Crested Couas laying two eggs: Goodman 2013) he might have been in the presence of two full-grown offspring (a possibility consistent with the feathered head-sides) of an unseen pair (not necessarily mixed: avian hybrids are commonly the product of heterospecific rape by the larger-bodied male parent, which plays no part in rearing the offspring: Rohwer et al. 2014). Another objection is that Milon found maxima in ‘humid forest’ (Milon 1952), a habitat that pyropyga, at least, tends to shun (Goodman 2013): a possible counter here is that niche overlap between C. caerulea and C. cristata (s. l.) is greater than expected (Chiatante 2022), and that the two species approach each other in south-eastern Madagascar to ‘a ground distance of less than 2 km’ (Goodman et al. 1997: 49), which surely suggests a circumstance for at least occasional direct encounters between them.

Even so, the only realistic position to take in this interesting case is to urge a molecular study of the holotype of Coua cristata maxima and its putative parents and relatives. The possibility that it does, after all, represent a genuine taxon—and therefore in our assessment a distinct species, inevitably highly threatened (unless already extinct)—demands that the matter be resolved as soon as possible. It is disappointing that one attempt to extract DNA from the specimen failed (R. Davion in Goodman 2013), but perhaps modern techniques (e.g. Tsai et al. 2020) can now be applied to satisfactory effect.