Print Insides

Insides 2/6/04 9:38 am Page 108

Whittaker et al. 108 Bull. B.O.C. 2004 124(2)

References: Brumfield, R. T., Swofford, D. L. & Braun M. J. 1997. Evolutionary relationships among the potoos (Nyctibiidae) based on isozymes. Pp. 129-145 in Remsen, J. V. (ed.) Studies in Neotropical ornithol- ogy honoring Ted Parker. Orn. Monogr. No. 48. Cleere, N. 1998. Nightjars. A guide to the nightjars and related nightbirds. Pica Press, Robertsbridge. Dickerman, R. W. & Phelps, W. H. 1982. An annotated list of the birds of Cerro Urutaní, on the border of Estado Bolívar, Venezuela and Território Roraima, Brazil. Amer. Mus. Novit. 273: 1–20. Hilty, S. L. & Brown W. L. 1986. A guide to the birds of Colombia. Princeton Univ. Press. del Hoyo, J., Elliott, A. & Sargatal, J. (eds.) 1999. Handbook of the birds of the world, vol. 5. Lynx Edicions, Barcelona. Olivares, A. 1964. Adiciónes a las aves de la Comisaría del Vaupés (Colombia), II. Caldasia 9: 379–393. Roca, R. L. 1994. Oilbirds of Venezuela: ecology and conservation. Publ. Nuttall Orn. Cl. 24. Cambridge, MA. Ridgely, R. S. & Greenfield, P. J. 2001. The birds of Ecuador. Cornell Univ. Press, Ithaca, NY. Sick, H. 1993. Birds in Brazil: a natural history. Princeton Univ. Press. Snow, D. W. 1962. Natural history of the Oilbird Steatornis caripensis, in Trinidad, W.I. Part II. Population, breeding, ecology and food. Zoologica 47: 199–221.

Addresses: Andrew Whittaker, Conjunto Acariqara, Rua Samaumas 214, Manaus 69085-410, Amazonas, Brazil, e-mail: [email protected]. Axel H. Antoine-Feill S. & Robin Scheiele Z., Apartado Aéreo (POB) 94438, Bogotá D.C., Colombia.

The small endemic West Indian emberizine known as the Lesser Antillean Bullfinch Loxigilla noctis is widespread in the Leeward and Windward Lesser Antilles (Fig. 1). It has usually been regarded as highly polytypic and nine subspecies are currently recognised (Table 1). With one exception, all are strongly sexually dichromatic: males are jet black with rufous throats, rufous or black undertail-coverts, and black bills; females are brownish-olive, paler ventrally, with horn-coloured bills. The exception is the population found only on the isolated and geologically discrete island of Barbados. Originally described by Cory (1886) as a separate species, Barbados Bullfinch Loxigilla barbadensis, nearly all subsequent workers, hewing to 20th-century taxonomic practice, have treated it as another subspecies of the Lesser Antillean Bullfinch, its scientific name then becoming Loxigilla noctis barbadensis. However, the bullfinches resident on Barbados differ in one striking way from the other eight named populations of the Lesser Antillean Bullfinch: they exhibit no sexual dichromatism. In the light of recent treatments of other allopatric West Indian bird populations, reconsideration of the taxonomic status of barbadensis is warranted. Insides 2/6/04 9:38 am Page 109

P. A. Buckley & Francine G. Buckley 109 Bull. B.O.C. 2004 124(2) Background Until the mid 1980s, essentially the only examinations of Lesser Antillean Bullfinch’s biology concerned its taxonomy (Hellmayr 1938, Bond 1956). Then, Bird (1983) compared the behaviour and ecology of barbadensis with that of several other Lesser Antillean Bullfinches, especially St Lucia’s sclateri and St Vincent’s crissalis, the two geographically closest to barbadensis. Regrettably, no portion of that thesis has ever been published and, despite intense efforts, we have been unable to contact its author. It nonetheless remains a seminal work bearing on the taxonomic status of barbadensis. Equally important information concerns the geological history and age of Barbados. Here, too, the picture has now been clarified, and Barbados’s striking geological history and youth, especially its relationship to islands in the main Lesser Antillean arc, are now well understood. The third piece of the puzzle finally fell into place with publication of a recent analysis of the molecular genetics of Barbados’s breeding landbirds, including barbadensis.

Geology, vegetative cover and climate Barbados is unique among the Lesser Antilles in being both very young and lacking any history of volcanism. It is also 150 km east of the main Windward Island chain, well out in the trade winds. Owing to its geological origins, it is relatively low in relief, its highest point, Mt Hillaby, reaching only 335 m, with about two-thirds of the island ranging from sea level to c.100 m, and almost another third to c.200 m. Whereas basal rocks in north-eastern Barbados have been dated to the Upper Eocene/Tertiary (50–70 My BP), these are crustal rocks pushed up as a tectonic mound along the Barbados Ridge. This feature originated (and is still rising) as the South American Plate slid west-northwestward under the Caribbean Plate, in the process heating crust that emerged as lava from volcanoes on the Lesser Antillean Ridge from Grenada to Saba, along the main Lesser Antillean arc. These present- day islands first appeared in the Miocene, 15–30 My BP, and are still rising (Speed 1988, Perfit & Williams 1989, Speed & Keller 1993). Their volcanic soils were doubtless vegetated quickly, so they would have had a long history of avian colonisation and evolution. ‘Proto-Barbados’ on the other hand only emerged in fits and starts at the end of the Pleistocene, c.700,000–1 My BP, and for much of its early history was a combination of coral reefs and low-lying small islands, periodically eroded, submerged and rebuilt in response to sea-level changes, oceanic and rainwater erosion, and sedimentation. On average, it has been rising c.0.3 m per 1,000 years. The best evidence (Mesolella 1967) suggests that the core of present-day Barbados emerged 600–700,000 y BP. Originally a circular dome (the centre of the uplift), perhaps two-thirds of this original island subsequently eroded away; then c.300,000 y BP, a new island arose to its south-west (near present-day Bridgetown), eventually fusing via crustal uplift with the now much smaller main island. Insides 2/6/04 9:38 am Page 110

P. A. Buckley & Francine G. Buckley 110 Bull. B.O.C. 2004 124(2)

Forest vegetation was probably slow to accumulate in such a low-relief, dynamic environment, and though there are as yet no known dates for establishment, forests most probably followed on the merging of Proto-Barbados and Christ Church islands, i.e. up to 300,000 y BP. Limestone soil generated by erosion (and, eventually, plant growth) was augmented by episodic infusions of volcanic ash from neighbouring St Lucia and St Vincent (e.g. an estimated 3.25 million tonnes [76 tonnes/ha] fell on Barbados following the irruption of St Vincent’s Soufrière Volcano in 1901: Gooding 1974). Many plants probably colonised Barbados from the south (but not the north or west) via ocean currents such as the Antilles Current, fed by waters draining the Amazon and Orinoco Rivers and even western Africa. Winds doubtless also brought plant propagules from South America (Urania moths still regularly arrive on Barbados with southerly winds, as do vagrant and colonis- ing birds), from West Africa, the source of airborne sediments as well as Desert Locusts Schistocerca gregaria (Prospero 1968, Rainey 1989, Savoie et al. 1989), and also from the Lesser Antilles north of Barbados. Birds might still be contribut- ing plant material and seeds from North America, the West Indies and South America in soil on their feet and in their excreta. One analysis (Gooding 1974) estimated that of 478 extant, non-introduced, higher plant species on Barbados, 12% came from the north, 19% came from northern South America, 43% occurred throughout the Caribbean/tropical continental America, and 26% were pantropical; only five (1%) were Barbados endemics. By the time of European colonisation in the early 1600s, Barbados was covered in forest, but unlike the volcanic Antillean-arc forests, it was not tropical rain forest, nor was it widespread. The high centre of the island, with good soil and 2,030 mm per year of rainfall, did support a rich mesic tropical hardwood woodland, whilst lower and drier elevations averaging 1,520 mm per year and sea coasts with only 1,015 mm per year grew only poorer hardwood and dry scrub forests, respectively. Especially on the windward side of Barbados, woody vegetation was conspicuously stunted and wind-shaped by salt spray. Although persistent east–north-easterly trade winds blow at 15–30 kph most of the year, Barbados’s low relief has eliminated most orographic precipitation, so the island is relatively dry: its mean annual rainfall of 520 mm is only about half that of St Lucia, whose tall mountains (nine in excess of 610 m, the highest 950 m) and strikingly vertical, montane topography (10% of St Lucia lies above 365 m) extract much more moisture from trade winds. Accordingly, St Lucia’s tall, lush, and highly diverse tropical rain forests and elfin woodlands are wholly lacking on Barbados. This is reflected in endemic avian species: Barbados currently has but one (a few others were probably extirpated shortly after European colonisation: Hutt et al. in press), whilst St Lucia has three (perhaps four), including two endemic genera (Keith 1997).

Geographic variation within Lesser Antillean Bullfinches Nine subspecies of Loxigilla noctis have been described (Table 1), many on only minor differences in colour, patterning or (not shown) measurements. All but one Insides 2/6/04 9:38 am Page 111

P. A. Buckley & Francine G. Buckley 111 Bull. B.O.C. 2004 124(2)

TABLE 1 Subspecies of Lesser Antillean Bullfinch Loxigilla noctis (from north to south), their occurrence, and their male coloration (from Ridgway 1901) Islands Taxon Male plumage Anguilla, St Martin, ridgwayi head and upper breast matte black, fading to ashy black St Barthélémy, Barbuda, on rump, belly and tail; rufous-chestnut supraloral Antigua spot, chin, throat and undertail-coverts; black bill Saba, St Eustatius, St Kitts, coryi jet black, with rufous-chestnut supraloral spot, chin, Nevis, Montserrat throat and undertail-coverts, black bill Guadeloupe, Îles des Saintes, dominicana jet black, with rufous-chestnut supraloral spot, chin, Marie-Galante, Dominica throat and undertail-coverts, black bill Désirade desiradensis head and upper breast matte black, fading to ashy black on rump, belly and tail; rufous-chestnut supraloral spot, chin, throat and undertail-coverts; black bill Martinique noctis jet black, with rufous-chestnut supraloral spot, chin and throat, black undertail-coverts, black bill St Lucia sclateri jet black, with rufous-chestnut supraloral spot, chin and throat, black undertail-coverts, black bill Barbados barbadensis as female: brownish olive dorsally, grey ventrally, with orange-rufous undertail-coverts, dusky-horn bill St Vincent crissalis jet black, with rufous-chestnut supraloral spot, chin, throat and undertail-coverts, black bill Grenada grenadensis jet black, with rufous-chestnut supraloral spot, chin, throat and undertail-coverts, black bill

are strongly sexually dichromatic: black-billed males are jet black (or, on shrubby, xeric islands, black fading to ashy black posteriorly), with rufous-chestnut supraloral spots, chins, throats and undertail-coverts (Fig. 1). Two taxa have wholly black undertail-coverts, and individuals in a few others sometimes show black feathers intermingled with rufous. Females vary in more subtle ways through the Lesser Antilles and not always concordantly with males. This local differentiation indicates that the Lesser Antillean Bullfinch is morphologically labile, probably in response to environmental conditions. Such eco-typical adaptation in plumage colour is hardly unknown in birds, so its appear- ance in the West Indies is unsurprising, even if its manifestation on islands only 10 km apart might be (jet black dominicana on Guadeloupe vs. ashy-black desiradensis on Désirade). Analysis of similar inter-island variation in Darwin’s Finches (Geospizinae; Grant 1986) suggests that Lesser Antillean Bullfinch variation probably has a genetic basis. However variable, all Lesser Antillean Bullfinch populations still maintain strong sexual dichromatism—except barbadensis (Fig. 2), which has uniquely become sexually monochromatic, males and females Insides 2/6/04 9:38 am Page 112

P. A. Buckley & Francine G. Buckley 112 Bull. B.O.C. 2004 124(2)

Figure 1. The West Indies, indicating all islands supporting forms of Lesser Antillean Bullfinch Loxigilla noctis, sensu lato, discussed in this paper.

resembling females of all other populations, a condition also called ‘hen-feathering’ (Mayr 1942). In addition, we were immediately struck by how different Lesser Antillean Bullfinches were on Barbados and St Lucia, hen-feathering aside. Where sclateri is a solitary, generally shy, usually within-vegetation forager, barbadensis is a bold, open-country and woodland, ground-feeder that often occurs in small feeding groups with Black-faced Grassquits Tiaris bicolor, especially during the dry season. The question immediately posing itself is whether morphological, ecological and behavioural changes in barbadensis have been accompanied by changes in its biological-species status, because we suspect that this unique switch from sexual dichromatism to sexual monochromatism was a profound, not trivial, evolutionary step. Results Ecology and behaviour of Barbados and St Lucia bullfinches Bird (1983) questioned whether the change in morphology in barbadensis was attended by parallel behavioural and ecological changes and, if so, how these might be explained in terms of selective pressures and evolutionary history. Although she was not especially concerned with taxonomy, her results nonetheless have taxonomic implications. She compared barbadensis to its nearest geographic (and Insides 2/6/04 9:38 am Page 113

P. A. Buckley & Francine G. Buckley 113 Bull. B.O.C. 2004 124(2)

Figure 2. Male Lesser Antillean Bullfinch Loxigilla noctis sclateri from St Lucia, Lesser Antilles (Wayne Arendt)

Figure 3. Male Barbados Bullfinch Loxigilla barbadensis (P. A. Buckley) Insides 2/6/04 9:38 am Page 114

P. A. Buckley & Francine G. Buckley 114 Bull. B.O.C. 2004 124(2)

TABLE 2 Ecological and morphological comparison of the resident forms of 'Lesser Antillean Bullfinch' on Barbados (barbadensis) and St. Lucia (sclateri); data adapted from Bird (1983) and Ridgway (1901). All quantitative differences mentioned are statistically significant; see text and (Bird 1983) for additional information.

Character barbadensis sclateri Male plumage brownish olive dorsally, jet black with rufous- grey ventrally with orange- chestnut throat and pre- rufous undertail-coverts; orbital spot; bill glossy dusky-horn lower mandible black Female plumage brownish olive dorsally, brownish olive dorsally, grey ventrally with orangish grey ventrally with undertail-coverts; pale horn orangish undertail-coverts; lower mandible pale horn lower mandible Diet fruit 10% 32% flowers 29% 40% seeds 36% 10% insects 20% 8% Foraging behaviour pecking 59.5% 45.5% insect gleaning 13.0% 7.5% foot-holding 11.0% 27.5% flower-base chewing 10.0% 4.0% nectar-feeding 4.5% 14.5% ground-feeding frequency 50% 2% foraging between 1–12 m 49% 92% modal foraging height 0 m 4-5 m mean foraging height 2.4 m 6.6 m

Breeding biology male nest-building effort (as % of female's) 114.0 60.5 mean incubation bout length 25.7 min 56.6 min mean non-incubation interval 6.9 min 14.6 min mean times male feeds incubating female 9.0 3.7 male nestling feeding trips/hr 1.5 1.1 female nestling feeding trips/hr 1.7 1.0 mean time male at nest/feeding trip 38.5 sec 77.4 sec mean time female at nest/feeding trip 58.6 sec 101.7 sec response to nest predators attack, vocalise flee, remain silent advertising/courtship displays/vocalisations undescribed undescribed time males spent within 3 m of nest during AM: nest building 22–30% 4–6% incubation 28–46% 4–6% brood rearing 14–22% 4% Insides 2/6/04 9:38 am Page 115

P. A. Buckley & Francine G. Buckley 115 Bull. B.O.C. 2004 124(2)

mean intruder chases by males/2-hr period during: nest building 8.0 2.0 incubation 16.0 1.7 brood-rearing 13.7 2.0 mean number of male 'songs'/2-hour period during: nest building 9.7 3.3 incubation 14.0 1.0 brood rearing 7.0 2.0

dichromatic) neighbours, St Lucia’s sclateri and St Vincent’s crissalis, seeking insight into the geographic source of Barbados birds. Because we now know that barbadensis colonisers came from St Lucia and not St Vincent (see below), for the purposes of this paper we will only compare her findings for sclateri and barbadensis (in Table 2); readers are referred to Bird (1983) for additional information. All quantitative comparisons discussed here were statistically significant (P = 0.05).

DIET Proportional diet composition differed strikingly between the islands. Seeds obtained on the ground and from flower heads accounted for 65% of barbadensis’s diet, but nearly that same fraction of sclateri’s came from fruit and flowers. Conversely, barbadensis consumed only c.10% fruit, sclateri only 8% seeds. (Diet composition as a function of food-type availability was not examined on either island.) Differences in wet–dry season diet composition were statistically significant on Barbados (much more fruit in the wet season, more seeds in the dry) but perhaps unsurprisingly were not on St Vincent (the wet island surrogate for St Lucia, from where Bird was unable to obtain dietary data).

FORAGING BEHAVIOUR Unsurprisingly, the two populations’ different diets were obtained in different fashions: barbadensis spent fully half its time feeding on the ground, which sclateri almost never did; barbadensis foraged only half the time at 1–12 m, where sclateri foraged nearly exclusively. Correspondingly, feeding postures and gleaning techniques also differed (Table 2). In the dry season barbadensis increased seed eating, insect eating and ground foraging (the latter almost doubling); dry-season foraging data were unavailable for sclateri. Data on flock sizes were not collected, as sclateri is essentially solitary, and barbadensis more gregarious although not properly ‘flocking’.

NEST BUILDING, PARENTAL CARE AND AGGRESSION The most obvious difference between barbadensis and all other Lesser Antillean Bullfinches is male coloration. Not unexpectedly, relatively cryptically coloured male barbadensis behaved quite differently from non-cryptic sclateri. Male Insides 2/6/04 9:38 am Page 116

P. A. Buckley & Francine G. Buckley 116 Bull. B.O.C. 2004 124(2)

TABLE 3 Results of MANOVA contrasting seven sclateri and barbadensis measurements (mm) of males: wing chord, tail, tarsal width, exposed culmen, bill width @ proximal end of nares, middle-toe claw, and hind-toe claw. MANOVA pairwise comparisons shown for the only two features differing significantly (bill width and tarsal thickness).

Variable Mean Std Err Lambda F df P bill width barbadensis 5.500 0.129 4.127 4.127 1,14 .0616 sclateri 5.083 0.154 tarsal thickness barbadensis 2.850 0.076 13.443 13.443 1,14 .0025 sclateri 2.417 0.083

Omnibus MANOVA Value F df P Wilks's Lambda 0.788 4.237 7,8 0.0301

barbadensis expended more than twice as much nest-building effort as sclateri and fed incubating females twice as frequently, and their incubation bouts were only half as long as those of sclateri, as were their inter-incubation intervals. Once young hatched, male barbadensis made 50% more feeding trips per unit time, and females 70% more, than sclateri. Conversely, both male and female barbadensis spent only about half the time at the nest during each feeding bout than did sclateri males and females. Cryptically coloured male barbadensis remained in the vicinity of their nests five times longer than did sclateri from nest building through brood rearing; were strikingly (4–8 times) more likely to chase intruders (especially during incubation); and ‘sang’ 3–14 times more frequently than did sclateri males. Aggression towards conspecifics was not significantly different under high and low densities on Barbados, but was higher than on St Lucia (sclateri), where bullfinch densities in ‘semi-residential lowlands’ were essentially identical to those of barbadensis. Categorical responses to both avian (especially Green Heron Butorides virescens,

TABLE 4 Results of mtDNA analyses sequencing 842 bp of overlapping ATPase-6 and ATPase-8 genes in haplotypes of barbadensis and sclateri. Cf. Lovette et al. (1999) for additional information.

Variable barbadensis sclateri Sample size 21 36 Range of pairwise nucleotide differences within 0–21–7 Range of pairwise nucleotide differences between - 3–7 Minimum pairwise divergence between - 0.36% Insides 2/6/04 9:38 am Page 117

P. A. Buckley & Francine G. Buckley 117 Bull. B.O.C. 2004 124(2)

Broad-winged Hawk Buteo platypterus, American Kestrel Falco sparverius, Carib Grackle Quiscalus lugubris, Grey Kingbird Tyrannus dominicensis, Tropical Mockingbird Mimus gilvus) and non-avian (typically Anolis spp., domestic cat Felis sylvestris) predators on various islands were even more dramatic: male barbadensis generally aggressively attacked and vocalised, when male sclateri remained completely silent and/or fled.

Body architecture Structural differences between male barbadensis and sclateri are not immediately obvious, but our own measurements of males of both taxa in the collections of the Museum of Comparative Zoology (MCZ), Cambridge, Massachusetts, USA, revealed that barbadensis males had almost 50% thicker (stouter) tarsometatarsi and 10% broader bills than sclateri males. Overall, males of the two populations differed significantly when all measurements were analysed using multivariate techniques (Table 3), although these results may signify only multivariate mean differences, not individual diagnosability. Bird (1983) measured several structural variables in barbadensis and sclateri, searching for sexual as well as population differences. Two measures of bill length (exposed culmen, nares to tip) were inconclusive or contradictory, owing perhaps to small sample sizes, although she did demonstrate that bill depth, flattened wing and tail lengths of both barbadensis and sclateri males exceeded those of females. She also measured ‘tarsal length’ but in two ways that were not so unequivocally described as to allow confident replication. The first she called ‘tarsus 1: length of tarsometatarsus,’ and the second ‘tarsus 2: distance from proximal end of tarsometatarsus to distal margin of last undivided scute.’ ‘Tarsus 2’ (which she believed ‘accurate only to 2 mm’) would seem the conventional tarsus measurement but she did not so state, despite citing Baldwin et al. (1931). Moreover, she referred to ‘tarsus 1’ as ‘length of tarsus,’ a term Baldwin et al. (1931) would restrict to her ‘tarsus 2.’ In any case, she failed to find sexual differences in either barbadensis or sclateri using ‘tarsus 2’ measurements, but did with ‘tarsus 1.’ Finally, although she did compare ‘tarsus 1’ and ‘tarsus 2’ measurements for male and for female barbadensis to male and to female sclateri, respectively, she presented no raw data or sample sizes, opting for a table that nonetheless indicated that by both methods barbadensis had highly significantly longer tarsi than sclateri.

Mitochondrial DNA Lovette et al. (1999) analysed mtDNA from eight resident passerines on Barbados, among them Lesser Antillean Bullfinches from Barbados (barbadensis), St Lucia (sclateri), and St Vincent (crissalis), to ascertain patterns and timing of colonisation on Barbados. Initially the entire 894–bp mtDNA genomes for ATPase 6 and ATPase 8 for five barbadensis and sclateri were sequenced to confirm and characterise inter-island differentiation, followed for an additional 21 barbadensis and 32 sclateri by either Restriction Fragment Length Polymorphism (RFLP) digestion of Insides 2/6/04 9:38 am Page 118

P. A. Buckley & Francine G. Buckley 118 Bull. B.O.C. 2004 124(2)

PCR-amplified products, or by additional sequencing. To examine individual island monophyly, enzyme cleavage and RFLP gels were compared among islands, and sample sizes used (Table 4) were regarded as adequate for detection of low-frequency haplotypes. They found (1) no evidence of any founder-effect on the three islands; (2) only a single clade of lineages on each of the three islands (i.e., recip- rocal monophyly between islands); (3) that the three taxa had accumulated 3–5 sub- stitutions among the three islands; (4) that although barbadensis and crissalis had 0.60% nucleotide divergence, barbadensis and sclateri differed by only 0.36%, implying St Lucia to be Barbados’s source population; and (5) assuming a uniform molecular clock, consistently low levels of mtDNA differentiation on Barbados indicating relatively recent colonisation events by at least four species, the Lesser Antillean Bullfinch among them.

Discussion Bird’s (1983) ecological and behavioural data demonstrate that barbadensis differs from sclateri in diet (particularly during Barbados’s stressful dry season, to which it seems well adapted), foraging behaviour, nest attentiveness, and aggression. At all stages in its breeding cycle, barbadensis biology is in many respects (including overall coloration but excluding sexual dichromatism) more like that of a grassquit Tiaris spp. than all other Lesser Antillean Bullfinches. Granivory and insectivory are well-known avian adaptations to xeric environments (Lack 1976), and Barbados is among the driest of Lesser Antillean islands. Food pressures are greatest in dry seasons, when both insects and seeds are more buffered from immediate climatic vicissitudes than is fruit, so facultative granivory/insectivory assumes high selective value under such conditions. Ground foraging is obviously associated with granivory, and pale and/or cryptic coloration is another classic avian adaptation for terrestrial foraging. This is not to say that barbadensis is not an arboreal forager or frugivore; it most certainly is during the wet season, as one would expect of a generalist. Higher feeding rates of barbadensis males are probably a function of the low energy content of small seeds, and increased feeding in turn leads to more frequent encounters with conspecifics, etc. There are no resident hawks or owls on Barbados, so such predation pressure on adults as now exists probably comes largely from feral quadrupeds and perhaps grackles. Even though these were surely unimportant in the evolution of barbadensis monochromatism, it remains wholly conjectural what might have been. There have been occasional references in the literature to partially or wholly black male bullfinches on Barbados (e.g. Hellmayr 1938, Bird 1983), usually attrib- uted to new St Lucia or St Vincent immigrants or to residual ancestral variation within barbadensis. We believe that, owing to the distance between St Lucia and Barbados and prevailing easterly trade winds, it is likely that few if any bullfinches continue to arrive unassisted on Barbados from St Lucia (with no evidence of any ever having arrived from St Vincent). However, this does not preclude occasional Insides 2/6/04 9:38 am Page 119

P. A. Buckley & Francine G. Buckley 119 Bull. B.O.C. 2004 124(2)

ship-borne transport by inter-island ferries, although we are also aware of no shipboard observations of Lesser Antillean Bullfinches near Barbados. At any rate, cock-feathered males are excessively rare on Barbados, and several active observers with 120+ years of combined field experience on Barbados (E. B. Massiah, M. D. Frost, the late M. Hutt and the late H. F. Hutt pers. comms.) have never seen or heard of a single one. Whether from natural colonisers, ship-assisted waifs or intra- Barbados recombination, genes of such individuals would quickly be swamped on Barbados. Why has barbadensis diverged so greatly from all other Lesser Antillean Bullfinches? There are several not mutually exclusive possibilities: (1) the potential for eco-typical differentiation within the Lesser Antillean Bullfinch complex is well established (see above); (2) molecular analyses found Barbadian Black-faced Grassquit’s mtDNA nucleotide divergence from St Lucia birds to be 0% (Lovette et al. 1999), indicating recent arrival on Barbados. Thus, in the absence of past competition from its most likely ground-feeding, seed-eating emberizine competi- tor, ‘proto-barbadensis’ could have quickly moved into an important vacant niche; (3) body-structure differences (especially stouter tarsi and wider bills) are clearly associated with, respectively, terrestrial habits and increased seed-eating. (In our own analyses, length of hind claw, another cursorial adaptation, approached statistical significance); (4) Barbados is radically different from almost all other Lesser Antillean islands in geological history, soil, climate and vegetation; and, (5) Barbados is sufficiently far east of the main Lesser Antillean arc to preclude repeated reinvasion from source populations. Hellmayr (1938) assumed that monochromatism in barbadensis was a primitive or ancestral condition and that sexual dichromatism in all other Lesser Antillean Bullfinch populations derived from it, but we now know that barbadensis derived from a surely dichromatic St Lucia population, not the reverse. Bird (1983) on the other hand did not specifically address this question, although suspecting (based on a single Barbados male with some blackish body and rufous throat feathers) that the Barbados founding population was dichromatic and thus that barbadensis was secondarily monochromatic. We believe it is taxonomically significant that despite plumage and structural changes within the nine currently recognised Lesser Antillean Bullfinch subspecies, eight have nonetheless retained sexual dichromatism. We are unaware of any published studies of the Lesser Antillean Bullfinch nuclear genome (including micro-satellites), so all inferences concerning the age of various populations rely on mtDNA data. These offer several possible sources of error, some of which are amenable to study, some not: (1) it is possible, although quite unlikely, that rare barbadensis haplotypes in sclateri were not detected; (2) the ‘mean rate’ of mtDNA accumulation customarily used for passerines (2% per My) surely varies among genera and species, yet it is exceedingly difficult to calibrate the absolute rate for a single species, or to calculate error terms; (3) the duration of time required for each taxon to complete the biological speciation process is unique; Insides 2/6/04 9:38 am Page 120

P. A. Buckley & Francine G. Buckley 120 Bull. B.O.C. 2004 124(2)

(4) statistical separation of small among-species differences in percent nucleotide divergence requires much larger samples than have been available; and, (5) the approach used by Lovette et al. (1999) would fail to detect earlier invasions masked by complete replacement of mtDNA lineages, although the probability of an invading haplotype wholly replacing a resident one would be very small indeed (1/population size). Nonetheless, given that geological data strongly support Barbados’s emergence =700,000 y BP; assuming a molecular clock with a nucleotide substitution rate of 2% per My; accepting that barbadensis is monophyletic in origin; and agreeing that the 0.36% nucleotide substitution divergence between barbadensis and sclateri is approximately correct, we reason, after consideration of the potentially confound- ing factors noted above, that proto-barbadensis probably invaded Barbados from St Lucia sometime 700,000–180,000 y BP—and considerably closer to the latter than the former considering mtDNA data from four Barbados taxa. Although courtship, pair formation and attendant displays (postural and vocal) have never been studied in Loxigilla, we suggest that females choose males and non-randomly, given that female choice is customary in strongly sexually dichromatic passerines. We further propose that male advertising song plays a major role, as does distinctive male plumage (and thus that sexual selection operates to some degree within the dichromatic species), but that the importance of both has been severely curtailed in sexually monochromatic barbadensis. We further propose that barbadensis females choose their males using behavioural or fitness characters such as frequency of song, aggression toward conspecifics and predators, and frequency and quality of courtship feeding. Finally, we consider that this complex of factors, coupled with other behavioural and ecological features, especially loss of distinctive male plumage, has resulted in barbadensis having diverged so far from noctis stock as to have achieved biological-species status. Collectively these notions offer fertile ground for vocal playback and plumage-alteration experiments designed to elucidate the exact operation and relative roles of isolating mechanisms, whose existence we do not doubt. Peterson (1996) accumulated many instances of geographic variation in avian sexual dichromatism. He determined that whilst variation in coloration of either sex was about equally likely, males were about five times more likely to lose than to gain bright plumage (reflecting initial incidence), and although males of island populations were more likely to lose than gain bright colour, they were not necessar- ily more likely to do so than their continental counterparts. He suggested that sexual dichromatism characters are likely to have a simple Mendelian or sex-linked basis whose alleles may be segregating in wild populations. Granted that plumage aberra- tions in wild birds often follow Mendelian inheritance (reviewed in Buckley 1987), the evidence for widespread Mendelian inheritance of species-specific characters is less strong. Further, Peterson (1996) asserted that loss and/or gain of bright or dull plumage (especially on islands) is more frequently due to genetic drift than to selection. He did acknowledge, though, that loss of bright male plumage might play Insides 2/6/04 9:38 am Page 121

P. A. Buckley & Francine G. Buckley 121 Bull. B.O.C. 2004 124(2)

a significant role in biological speciation, although allopatry would prevent the majority of island examples from being tested in nature. Doherty et al. (2003) considered what effect sexual selection might have on local extinction rates and turnover in avian communities. Extending earlier work indicating that dichromatic waterfowl (Promislow et al. 1994) and passerines (Promislow et al. 1992) seemed to suffer higher mortality rates than sexually monochromatic congeners, Doherty et al. (2003) modelled 21 years of data from the North American Breeding Bird Survey (BBS), finding that dichromatic species showed a mean 23% increase in annual local extinction rates over monochromatic species per BBS survey route. But they unexpectedly also showed that this increase in local extinction was almost exactly offset by local immigration and recolonisa- tion, leading to relative community stability. Whereas this offset may occur regularly in continental settings, it is demonstrably rare on oceanic islands (MacArthur & Wilson 1967), which may explain the disproportionately large number of monochromatic bird species on such islands. We believe it most likely that ancestors of the present-day West Indian endemic emberizine assemblage (Loxigilla, Melopyrrha, Melanospiza), irrespective of number of invasions, were dichromatic, possibly akin to present-day Volatinia, and that there has been strong stabilising (sexual?) selection to maintain dichromatism. We consider that the plumage, behavioural and structural changes that have occurred in Lesser Antillean Bullfinch populations in general (and Barbados Bullfinch in particular) have a nuclear genetic basis that has been selected for. Therefore, Barbados Bullfinch would show both nuclear and mitochondrial evolution, the package comprising sufficient divergence from the main line of Lesser Antillean Bullfinch evolution as to have developed concomitant prezygotic reproductive isolating mechanisms. Thus, we suggest that the form of ‘Lesser Antillean Bullfinch’ on Barbados is best treated as a separate biological species, Barbados Bullfinch Loxigilla barbadensis Cory 1886, which probably colonised Barbados from St Lucia around 180,000–700,000 y BP, and which has achieved specific status since its arrival. As such this represents one of the more rapid examples of avian speciation, but because this happened in an oceanic island context, it is perhaps less surprising; e.g. Klicka & Zink (1997) have also noted, on the basis of mtDNA chronology, that a few avian species do appear to have arisen unexpectedly rapidly. And what of the low level of mitochondrial nucleotide divergence found by Lovette et al. (1999) for barbadensis? Accepted practice, developed from divergence levels of known species-pairs, indicates that values below c.3% indicate only subspecific differentiation. Yet it is already known that birds may display pronounced, nuclear-genetic-based geographic variation that is not mirrored in mtDNA (e.g. Brawn et al. 1996). We believe Barbados Bullfinch is another example, and suggest that this phenomenon may be more common on oceanic islands than continents. We would also argue that whenever possible, it is better science to infer taxonomic status from non-molecular evidence, and then to use Insides 2/6/04 9:38 am Page 122

P. A. Buckley & Francine G. Buckley 122 Bull. B.O.C. 2004 124(2)

molecular data to derive time since colonisation, hence timing and duration of speciation. Increasingly, genetic and ethological studies are suggesting that island bird populations formerly lumped into polytypic species-complexes can be considered biological species just as parsimoniously. In the West Indies alone, the four endemic Myiarchus flycatchers were split into seven (Lanyon 1967); each of the four allopatric subspecies of ‘Stripe-headed Tanager, Spindalis zena’, is now considered a species (Garrido et al. 1997); each of the three isolated subspecies of ‘Dendroica adelaidae’ is now considered a species (Adelaide, Barbuda and St Lucia Warblers, respectively; Lovette et al. 1998); the isolated (and still extant?) Puerto Rican Bullfinch Loxigilla portoricensis on St Kitts has just been recognised as St Kitts Bullfinch L. grandis (Garrido & Wiley 2003); and the taxonomic status of the two allopatric Cuban Bullfinch Melopyrrha nigra populations is also under evaluation (J. Wiley, pers. comm.). To this group we now add Barbados Bullfinch.

Acknowledgements Our work on the avifauna of Barbados has been supported by the US National Park Service and the US Geological Survey. We thank Douglas Causey and Alison Pirie for access to specimens in the MCZ collections. For hospitality, support and field companionship on Barbados, we are indebted to our Barbados Checklist co-authors: the late Maurice and the late Hazel Hutt, Edward Massiah and Martin Frost. Wayne Arendt kindly supplied the photo of Loxigilla noctis sclateri, and Dennis Skidds, Environmental Data Center, URI, prepared Fig. 1. For discussion of their ongoing work on the deriva- tion of the West Indian avifauna, for reprints and preprints, and for information of various sorts, we thank Eldredege Bermingham, the late Nedra Klein, Irby Lovette, Herbert Raffaele, Robert Ricklefs and James Wiley. And for constructive comments on various versions of this manuscript we acknowledge Chris Feare, Martin Frost, Douglas Futuyma, Allan Keith, Edward Massiah, Michael Patten, Townsend Peterson, Robert Ricklefs and James Wiley. We dedicate this paper to Ernst Mayr on the occasion of his 100th birthday.

References: Baldwin, S., Oberholser H. & Worley, L. 1931. Measurements of birds. Sci. Publ. Cleveland Mus. Nat. Hist. 2: 1–165. Bird, J. 1983. Behavioral and ecological comparisons of Lesser Antillean Bullfinches: a study of the evolution of sexual dimorphism and monomorphism. Ph.D. thesis. Univ. of Montana, Missoula. Bond, J. 1956 et seq. Checklist of the birds of the West Indies, fourth edn, and supplements 1–27. Academy of Natural Sciences, Philadelphia, PA. Brawn, J., Collins, T., Medina, M. & Bermingham, E. 1996. Associations between physical isolation and geographic variation within three species of Neotropical birds. Molecular Ecol. 5: 33–46. Buckley, P. A. 1987. Mendelian genes. Pp. 1–44 in Cooke, F. & Buckley, P. A. (eds.) Avian genetics: a population and ecological approach. Academic Press, London. Cory, C. 1886. The birds of the West Indies, including the Bahama Islands, the Greater Antilles, and the Lesser Antilles, excepting the islands of Tobago and Trinidad. Auk 3: 1–59, 108–119, 223–231, 311–327. Doherty, P., Sorci, G., Royle, J., Hines, J., Nichols, J. & Boulinier, T. 2003. Sexual selection affects local extinction and turnover in bird communities. Proc. Natl. Acad. Sci. USA 100: 5858–5862. Garrido, O. H., Parkes, K. C., Reynard, G. B., Kirkconnell, A. & Sutton. R. 1997. Taxonomy of the Stripe-headed Tanager, Spindalis zena (Aves: Thraupinae), in the West Indies. Wilson Bull. 109: 561–594. Garrido, O. H. & Wiley, J. W. 2003. The taxonomic status of the Puerto Rican Bullfinch (Loxigilla portoricensis) (Emberizidae) in Puerto Rico and on St Kitts. Orn. Neotrop. 14: 91–98. Insides 2/6/04 9:38 am Page 123

P. A. Buckley & Francine G. Buckley 123 Bull. B.O.C. 2004 124(2)

Gooding, J. 1974. The plant communities of Barbados. Barbados Ministry of Education, Bridgetown. Grant, P. R. 1986. Ecology and evolution of Darwin’s Finches. Princeton Univ. Press. Hellmayr, C. E. 1938. Catalogue of birds of the Americas and adjacent islands in the Field Museum of Natural History. Part XI. Ploceidae—Catamblyrhynchidae—Fringillidae. Field Mus. Nat. Hist. Zool. Ser. XIII, Part XI: 1–662. Hutt, M. B., Hutt, H. F., Buckley, P. A., Massiah, E. B., Frost, M. D. & Buckley, F. G. In press. The birds of Barbados, West Indies: an annotated checklist. British Ornithologists’ Union & British Ornithologists' Club, Tring. Keith, A. R. 1997. The birds of St Lucia: an annotated check-list. BOU Check-list No. 15. British Ornithologists’ Union, Tring. Klicka, J. & Zink, R. 1997. The importance of recent ice ages in speciation: a failed paradigm. Science 277: 1666–1669. Lack, D. 1976. Island biology. Blackwell, Oxford. Lanyon, W. E. 1967. Revision and probable evolution of the Myiarchus flycatchers of the West Indies. Bull. Amer. Mus. Nat. Hist. 161: 329–370. Lovette, I., Bermingham, E., Seutin, G. & Ricklefs, R. 1998. Evolutionary differentiation in three endemic West Indian warblers. Auk 115: 890–903. Lovette, I., Seutin, G., Ricklefs, R. & Bermingham, E. 1999. The assembly of an island fauna by natural invasions: sources and temporal patterns in the avian colonization of Barbados. Biological Invasions 1: 33–41. MacArthur, R. & Wilson, E. O. 1967. The theory of island biogeography. Princeton Univ. Press. Mayr, E. 1942. Systematics and the origin of species. Columbia Univ. Press, New York. Mesolella, K. 1967. Zonation of uplifted Pleistocene coral reefs on Barbados, West Indies. Science 156: 638–640. Perfit, M. & Williams, R. 1989. Geological constraints and biological retrodictions in the evolution of the Caribbean Sea and its islands. In Woods, C. & Baker, P. (eds.) Biogeography of the West Indies: past, present, and future. Sandhill Crane Press, Gainesville, FL. Peterson, A. T. 1996. Geographic variation in sexual dichromatism in birds. Bull. Brit. Orn. Cl. 116: 156–172. Promislow, D., Montgomerie, R. & Thomas, T. 1992. Mortality costs of sexual dimorphism in birds. Proc. Roy. Soc., Lond., Ser. B 250: 143–150. Promislow, D., Montgomerie, R. & Martin, T. 1994. Sexual selection and survival in North American waterfowl. Evolution 48: 2045–2050. Prospero, J. 1968. Atmospheric dust studies on Barbados. Bull. Amer. Meteorological Soc. 49: 645–652. Rainey, R. 1989. Migration and meteorology: flight behavior and the atmospheric environment of locusts and other environmental pests. Clarendon Press, Oxford. Ridgway, R. 1901. The birds of North and Middle America. Part I. Family Fringillidae. Bull. US Natl. Mus. 50: 1–715. Savoie, S., Prospero, J. & Saltzman, E. 1989. Non-sea salt sulfate and nitrate in trade wind aerosols at Barbados: evidence for long-range transport. J. Geophysical Res. 94: 5069–5080. Speed, R. 1988. Geological history of Barbados: a preliminary synthesis. Trans. 11th Carib. Geol. Congr. 29: 1–11. Speed, R. & Keller, C. 1993. Synopsis of the geological evolution of Barbados. J. Barbados Mus. & Historical Soc. 41: 113–139.

Addresses: P. A. Buckley, USGS-Patuxent Wildlife Research Center, Box 8 @ Graduate School of Oceanography, University of Rhode Island, Narragansett RI 02882 USA, e-mail: [email protected]. Francine G. Buckley, Dept. of Natural Resource Sciences, Coastal Institute-Kingston, University of Rhode Island, Kingston RI 02881 USA, e-mail: [email protected]