Dispersal, Mimicry, and Geographic Variation in Northern Melanesian Birds1

Jared Diamond2

Abstract: I present new information about 34 of the 195 resident land and freshwater species of Northern , an area characterized by a rich avifauna, high endemism, and great geographic variation in morphology. There are many examples of geographic variation in voice, behavior, habitat prefer­ ence, altitudinal range, vertical stratum, abundance, and nest. Possible vocal convergence or mimicry between sympatric populations of different species is described between the goshawk Accipiter albogularis and the kingfisher Halcyon chloris, between the cuckoo-shrike Coracina [tenuirostris] and other species in its mixed-species foraging flocks, between the white-eyes Zosterops murphyi and Z. rendovae kulambang;rae, and between the starlings Aplonis g;randis and Mino dumontii. Hybridization is reported between the Bismarck and races of the cuckoo Eudynamys scolopacea on Long Island (described as a new ), between the whistlers pectoralis and P. melanura, and between the honey-eaters tristrami and M. cardinalis. Cyclones bring Australian species, some of which occasionally remain to breed. Over-water dispersal ability varies greatly, from species that can be seen flying over water any day to species that rarely or never cross water. For instance, a channel 12 km long and only 0.15-1 km wide divides Florida Island into two halves, one of which possesses and the other ofwhich lacks a resident population of the coucal Centropus milo.

NORTHERN MELANESIA comprises the hun­ Northern Melanesian avifauna includes many dreds ofislands of the Bismarck and Solomon little-known endemic populations (five en­ Archipelagoes east of New Guinea. Its rich demic genera, 30 endemic full species or su­ avifauna, their intensively studied distribu­ perspecies, 102 endemic allospecies, and 380 tions and , the marked geographic endemic subspecies). Rampant ongoing de­ variation in morphology of conspecific pop­ forestation of Northern Melanesia makes it ulations among different islands, and the urgent to study this treasure trove of diversity existence of bird species representing many before much of it disappears. different stages in the speciation process have Between 1969 and 1976 I made four all combined to make Northern Melanesian expeditions to Northern Melanesia: 27 June­ ideal material for studying speciation 24 July 1969, West ; 23 June­ (Mayr 1942 and many papers). However, little 1 September 1972, Umboi and Long and has been reported about geographic variation neighboring islands, and Bougainville; 18 in ecology, behavior, and vocalizations. The August-31 October 1974 and 5 September­ 15 October 1976, almost all ornithologically significant (except Buka,

1 Manuscript accepted 10 May 200l. Ysabel, , Ulawa, Ramos, Gower, and 2 Physiology Department, University of California the remote outliers), plus 94 small islets in Medical School, Los Angeles, California 90095-1751 (fax: and near Wana Wana Lagoon of the New 310-206-5661; E-mail: [email protected]). Georgia group. Previous publications have described the avifaunas of Rennell and Bel­ Pacific Science (2002), vol. 56, no. 1:1-22 lona (Diamond 1984), vocalizations of the © 2002 by University of Hawai'i Press. white-eye superspecies Zosterops [g;riseotinctus] All rights reserved (Diamond 1998), character displacement in 2 PACIFIC SCIENCE· January 2002 myzomelid honey-eaters (Diamond et al. SPECIES ACCOUNTS 1989), community assembly (Diamond 197Sa), selected species of Bougainville (Diamond 3. Australian Pelican, Pelecanus conspicillatus. 1975b) and West New Britain (Diamond Two sightings in Wana Wana Lagoon: one 1971, 1972) and Umboi and nearby islands soaring high on 16 September 1976, and one (Diamond 1974, 1976), species/area/distance resting on a sandbar on the following day. relations (Diamond and Mayr 1976, Dia­ Local people said that one or two pelicans mond et al. 1976, Gilpin and Diamond 1976, had been present for several weeks, and 1981), the montane avifauna (Mayr and Dia­ that cyclones (especially the cyclone of 1952) mond 1976), and two new taxa (Diamond occasionally brought small numbers, all of 1989a, 1991). which eventually died out or departed. I re­ In this paper I report significant field ceived such reports from 11 Solomon islands observations and distributional records­ (Ganonga, Gatukai, Honiavasa, Kulamban­ especially ones involving endemic popula­ gra, Mono, , , Rendova, tions and/or concerning geographic variation, Roviana, , Wana Wana), with vocalizations, overwater dispersal, and hy­ the earliest remembered arrival being around bridization. I also mention two records based 1918. There are several previous reports of on unreported specimens that I found in mu­ windblown pelicans reaching Northern Mel­ seums. anesia (e.g., Bradley and Wolff 1956, Cain Taxonomy and distributions are based on a and Galbraith 1956, Hadden 1981). comprehensive, just-published account of the 4. Little Pied Cormorant, Phalacrocorax Northern Melanesian avifauna by Mayr and melanoleucos. Previously known to be resident Diamond (2001), hence information in that on six of the larger Northern Melanesian is­ book is not repeated in this paper. To facili­ lands. In addition, Teu Zinghite told me that tate finding such information, I cite each res­ the species had been resident on the Kukundu ident species by its number used in Appendix River of Kulambangra during the 1960s and 1 of that book. To facilitate use of the that he had found a young bird not yet able taxonomic literature on Northern Melanesian to fly. Residents of Ganonga, Gatukai, New birds, most ofwhich predates 1960, the island Georgia, and Wana Wana reported several names and spellings used here and in that dozen as having been brought by the cyclone book are those then in use, differing in nine of 1952. cases from modern names and spellings 8. Great Egret, Ardea alba. Residents of (the older Florida, Ganonga, Gatukai, Gizo, numerous islands of the New Georgia group Gower, Kulambangra, New Hanover, San reported this heron to me as appearing or Cristobal, and Tetipari rather than the cur­ increasing in numbers after cyclones, and also rent Nggela, , Nggatokae, Ghizo, sometimes resident on exposed reefs, beaches, , , Lavongai, , and and mudflats, and along rivers. Tetepare, respectively). I frequently cite in­ 12. Sacred Ibis, Threskiornis moluccus. Pre­ formation provided by island residents (es­ viously known in Northern Melanesia only as pecially Teu Zinghite of Kulambangra and an endemic resident race on Rennell and Alisasa Bisili ofNew Georgia), many ofwhom Bellona. Residents of Mono told me in 1974 proved to be walking encyclopedias of field that a pair of this species arrived around the information. For example, Mr. Zinghite had same time as the cyclone-blown pelicans of discovered nests, reared nestlings, and re­ 1952 and had been resident since then, but corded diets, seasonal movements, and over­ that one ofthe pair disappeared around 1973, water dispersal frequencies of most bird leaving a single survivor. The description species of Kulambangra. On all islands I was of a slender tall bird similar to the "so" quizzed older men to obtain local-language (Mono name for Eastern Reef Egret, Egretta species names, according to methods de­ sacra) but taller, with a long, curved black bill, scribed elsewhere (Diamond 1966, 1989b,cj black head and tail, and white body. Diamond and Bishop 1999). 25. Pied Goshawk, Accipiter albogularis. Dispersal, Mimicry, and Geographic Variation in Northern Melanesian Birds . Diamond 3

An interesting range extension for this wide­ Museum of Natural History, mention on p. spread hawk is Borokna, an isolated, small, 226 of volume V that the Whitney collector steep extinct volcanic island halfway between Hannibal Hamlin observed one individual of and the New Georgia group and this species in the canebrakes of a mountain 59 km from the next island. A pair was resi­ creek on Choiseul sometime between Sep­ dent and calling frequently during my visit on tember and November 1929. Second, among 7-9 September 1974. the 42 bird species that residents of Sasa­ This goshawk is involved in a remarkable monga Village on Choiseul described to me vocal convergence or mimicry with the Col­ or identified with me in the field in 1974 was lared Kingfisher, Halcyon chloris. In the New one named the "ki," described as follows: Georgia and Bukida groups the common call similar to, but a different species from, the of both species is a moderately high-pitched "peyonga" (= Bush-hen, Amaurornis oliva­ nasal two-note or three-note element with a ceus); unmarked dark body; bill about 4 cm forced quality, repeated several times at the long; lives on the ground; calls "ki-ki-ki-ki­ rate of one per second. Each element consists . ..."; runs very quickly; and either does not fly of an upslur, followed by either one or two at all or else flies along the ground if pursued notes at lower pitch. Other New Georgia and by dogs. Bukida calls ofA. albogularis are a rapid series 54. Yellow-bibbed Fruit-Dove, Ptilinopus of 17-50 high-pitched nasal notes delivered solomonensis. This fruit-dove's ecological rela­ at about five per second, rising and then lev7 tions with congeners vary among islands. In eling off in pitch, and still with a quality very the Solomons on high islands of the New like the call of Halcyon chloris; a slower series, Georgia group (Kulambangra, Rendova, lasting 3 sec, of about five high-pitched up­ , Gatukai, Vella Lavella) I found it slurs; and a series of 5-10 nasal notes deliv­ from 400 m up to the summits, not in the ered as if the voice were cracking. The San lowlands. At middle elevations it coexisted Cristobal races of both A. albogularis and H. with Claret-breasted Fruit-Dove, P. viridis, chloris lack the first-named of these calls (the and Superb Fruit-Dove, P. superbus, but lived one with the repeated two- or three-note alone at higher elevations above the altitudi­ element). San Cristobal A. albogularis still nal ceilings of these other two species. The P. gives the rapid rising series and the slower solomonensis populations of Bougainville and series ofupslurs. Guadalcanal are confined to even higher ele­ Accipiter albogularis and its congener Grey vations, above 900 m (Diamond 1975b). On Goshawk, A. novaehollandiae, coexist on many low islands of the New Georgia group (New Solomon islands but differ strikingly in for­ Georgia, Kohinggo, and four lagoon islets) aging technique. The former hunts by soar­ and on Florida, P. solomonensis, P. viridis, and ing swiftly in circles without flapping, just P. superbus coexisted at sea level, P. solo­ above or high above the canopy. The latter monensis being by far the least common. On does not soar, but is a still hunter, searching San Cristobal, where P. superbus is absent, P. from a perch within the canopy and then fly­ solomonensis is common from coastal lowlands ing slowly with alternate flaps and glides be­ to the mountains, and the distinctive race P. low canopy height to its next perch. viridis eugeniae is confined to inland forest. 40. Woodford's Rail, Nesoclopeus woodfordi? On Ugi (P. superbus also absent) P. viridis is This endemic rail is known definitely only joined by Pink-spotted Fruit-Dove, P. rich­ from the three largest islands carved out by ardsii; both of these doves are common in high Holocene sea levels from the expanded open habitats, and P. solomonensis is confined Pleistocene island of Greater Bukida: Bou­ to forest. gainville, Ysabel, and Guadalcanal. There are In the Bismarcks P. solomonensis is instead a also two possible reports for the fourth-largest supertramp species (i.e., confined to species­ Bukida-derived island, Choiseul. First, the poor small, remote, or volcanically recently unpublished diaries of the Whitney South defaunated islands: Diamond 1974), excluded Sea Expedition, preserved in the American from large, central, species-rich islands by 4 PACIFIC SCIENCE· January 2002 the very closely related White-bibbed Fruit­ higher. Beyond 30 m one hears only the sec­ Dove, P. rivoli, a member of the same super­ ond note, which would be an utterly nonde­ species. The sole Bismarck island with estab­ script vocalization (a single "hoo") were it not lished populations of both species is Umboi, for the distinctive feature that the call is re­ where I found P. solomonensis only in one area peated ad nauseam, at regular intervals of 3-8 on the lower slopes of one mountain (Mt. sec depending on the individual bird, for up Birik) up to 500 m. Ptilinopus rivoli was on Mt. to 15 min. Birik and other mountains of Umboi, mostly 68. Crested Cuckoo-Dove, Reinwardtoena at elevations above P. solomonensis but over­ crassirostris. The distinctive, far-carrying call lapping marginally with it at Mt. Birik's base. of this Solomon endemic is often heard, but On New Hanover and possibly on New Ire­ the bird is very shy, inconspicuous, and rarely land and New Britain, where P. rivoli is also seen. It lives mainly in forest within the resident, P. solomonensis has been recorded shaded midstory and lower canopy at 8-18 m, only as a very rare vagrant. Ptilinopus solo­ in the mountains or near lowland rivers. monensis was the sole fruit-dove on recently Florida residents said that the population defaunated (A.D. 1888) Ritter near Umboi. there disappeared after Cyclone Ida destroyed On recently defaunated (ca. A.D. 1670) Long, much of the forest in 1971. I heard the call Crown, and Tolokiwa P. solomonensis was on Kulambangra, Choiseul, San Cristobal, joined by Red-knobbed Fruit-Dove, P. in­ Vangunu, and Gatukai: two notes at medium­ solitus, and on T olokiwa by P. superbus as well, high pitch, repeated a few times or else ad but P. solomonensis was by far the most abun­ nauseam for many minutes. The first note is dant and ubiquitous fruit-dove from the short, staccato, softer, and at lower pitch by coastal lowlands up to the summits of these about two whole tones; the second note is three islands. longer and downslurred. At a distance the first Because this dove is far more often heard note becomes inaudible, so one only hears the than seen, ability to distinguish the calls from downslur. The pair of notes is repeated at those of congeners is essential to field identi­ intervals averaging 5-6 sec, occasionally up to fication. There are two calls, a "hoo', series 8 sec. The time interval from the short first and a two-note call. The hoo series consists note to the longer second note is shorter (2 of 8-16 (race vukanorum of the New Georgia sec) than the interval from the second note to group) or 12-32 (race solomonensis of the San the next short note (3-6 sec). The quality of Cristobal group) high-pitched notes. The the call is so human that Choiseul islanders series accelerates and rises and then usually told me that the bird arose from a child who drops in pitch. Hoo series of solomonensis was left at home by its mother, kept calling (heard on San Cristobal and Ugi) are much for its mother, and was eventually changed faster and longer than those ofvulcanorum. In into the bird. the Solomons confusion is possible only with 68. Pied Cuckoo-Dove, Reinwardtoena the hoo series of P. superbus (Beehler et al. browni. Calls of this Bismarck endemic that 1986: 106), which differs in being slower, I heard on Umboi were more similar to somewhat shorter, and not accelerating nor those of R. crassirostris (described above) than dropping in pitch at the end. The hoo series to those of R. reinwardtii (Beehler et al. ofrace meyeri in the Bismarcks on Umboi and 1986:102). Like R. crassirostris's call, R. neighboring islands (Long, Crown, Tolo­ browni's is human in quality, medium-high­ kiwa, Sakar) may either accelerate greatly or pitched, far carrying, and begins with a short scarcely at all, and I found it difficult to dis­ note followed by a longer downslur at a tinguish from the hoo series of P. rivoli. higher pitch. But the R. browni call then has a All three of these races that I encountered third note, another downslur at a lower pitch. also have a two-note, high-pitched, hollow This three-note pattern is repeated at 4-sec call. The first note is very faint, short, and intervals for up to several minutes, as in the inaudible at a distance beyond 30 m; the sec­ Solomons. ond note is longer and pitched a half-tone 79. Meek's Lorikeet, Charmosyma meeki. Dispersal, Mimicry, and Geographic Variation in Northern Melanesian Birds . Diamond 5

This Solomon endemic is definitely known on Tolokiwa, the two islands nearest Long. only from the mountains of Bougainville, Since then, a forest dominated by rapidly Guadalcanal, Kulambangra, Malaita, and growing softwoods has regenerated. The avi­ Ysabel. However, the species should also be faunas of Long, Crown, and Tolokiwa are sought in the mountains of Vangunu, where largely shared and consist entirely of species at 650 m I twice glimpsed a single green par­ known to be good over-water colonists and rot that appeared to be this species and flew dominated by eight vagile supertramp species just above the canopy. specializing in.small, remote, or recently de­ 85. Solomon Cockatoo, Cacatua ducorpsi. It faunated islands (Diamond 1974, Diamond et has seemed surprising that this parrot, which al. 1989: table 1). Some populations are de­ can often be seen flying over water between rived from New Britain or other Bismarck nearby islands and is otherwise distributed islands, and others are derived from New throughout the central Solomons, is absent Guinea. from the San Cristobal group at the south­ Collections of birds on Long were made eastern end of the Solomon island chain. In by W. F. Coultas of the Whitney Expedition addition, there are no records for Ganonga in 1933 and by me in 1972. Previous taxo­ and at the southwestern end of the nomic studies had identified three species New Georgia group. Although I observed with distinctive populations on Long and its this very conspicuous and noisy species on neighbors: Pacific Swallow, Hirundo tahitica, every other island of at least modest size in whose population is intermediate in color the New Georgia group, I failed to observe it and size between the New Britain and New on Ganonga or Simbo, and residents of these Guinea populations (Mayr 1955); and the two islands insisted that not even a single large and small honey-eaters Bismarck Black individual had ever been recorded there. Myzomela, Myzomela pammelaena, and Red­ Although Ganonga is only 9 km from Vella bibbed Myzomela, M. sclateri, whose popu­ Lavella, and Simbo is only 7 km from lations are· respectively even larger and even Ganonga, these water gaps are evidently a smaller than conspecific source populations potent barrier to the cockatoo. Diamond et elsewhere in the Bismarcks (Diamond et al. al. (1976) identified 31 other Solomon bird 1989). These distinctions presumably arose in species that they termed "superior short­ the last three centuries, by hybridization in distance colonists," frequently crossing nar­ the case of the swallow and by character dis­ row but not wide water gaps. placement in the case of the honey-eaters. Oriental Cuckoo, Cuculus saturatus. There The Eudynamys scolopacea populations of are few records of this palaearctic species Long and its neighbors also prove to be of from the Solomons, which lie at the eastern recent origin, between the Bismarck limit of its winter range. I observed one indi­ race salvadorii and the New Guinea race ru­ vidual in a tree on a riverbank in forest on fiventer. Males of both races are uniformly Choiseulon 14 October 1974. black, but females and immatures of both 91. Common Koel, Eudynamys scolopacea. sexes have a complexly patterned ochraceous/ The populations of Long, Tolokiwa, and rufous, black, and white . These two Crown provide an interesting example of a source races differ in two respects: salvadorii's distinctive population of hybrid origins that larger size in both sexes, as reflected in wing developed and stabilized in less than three length and weight (Table 1); and, in the centuries. Around A.D. 1670, Long was de­ female/immature plumage, salvadorii's paler, faunated by one of the largest volcanic erup­ more whitish, less ochraceous ground color tions of recent millennia, covering the island of the underparts and its broader black tail with layers ofvolcanic ash at least 100 m thick bands. and creating a caldera now filled by a lake of In size the Long/Tolokiwa/Crown popu­ 86 km2 (Pain et al. 1981, Ball and Hughes lation agrees with the small New Guinea race 1982, Blong 1982). Matching ash layers imply rufiventer; it apparently weighs even less than heavy ashfall on Crown and probably also rufiventer (Table 1). In breadth ofblack bands 6 PACIFIC SCIENCE . January 2002

TABLE 1 Wing Length and Weight in the Cuckoo Eudynamys scolopacea

Character rnfiventer hybrida salvadorii cl'wing 194 (176-206) [32] 191 (183-201) [22] 210 (203-220) [17] 'i2wing 188 (182-193) [15] 191 (194-199) [4] 207 (204-214) [7] cl' weight 219 (188-254) [13] 173 (152-198) [8] 275 (205-330) [5] 'i2 weight 219 (202-234) [5]

Note: Specimens of rufiventer are from New Guinea (10 specimens) and adjacent islands (37); of hybrida, from Long (22), Tolokiwa (2), and Crown (6); and of salvatbrrii, from New Britain (20), Umboi (4), and Sakar (2). Only fully adult specimens were used. Wing lengths are in millimeters, weights in grams. Values given are the average, followed by the range in parentheses, followed by the number ofspecimens in brackets. All specimens are in the American Museum of Natural History. See text for discussion. in the female's tail it agrees with salvadorii. In may have required further time. Hence sub­ ventral coloration of the female it is interme­ speciation must have been achieved in some diate, being paler and less ochraceous than unknown time less than 263 years (1933 rufiventer, though not as white as salvadorii. minus 1670). There is no overlap in wing length or weight Because hybrida has the size of the New between the 31 available individuals of the Guinea population, the female tail pattern of Long/Tolokiwa/Crown population and the the New Britain population, and a female 26 available adults of salvadorii. All four ventral color intermediate between these two available females of the Long/Tolokiwa/ populations, it seems likely that colonist in­ Crown population can be distinguished from dividuals .of E. scolopacea arrived both from rufiventer in plumage. Hence this population New Guinea and from New Britain and hy­ is subspecifically distinct, and I propose the bridized to form an intermediate population name Eudynamys scolopacea hybrida ( spec­ now distinct from both ancestral populations, imen AMNH no. 422566, collector's field no. as also seems likely for the Long population 44762, adult female, collected 28 November of Hirundo tahitica. One hint that colonists 1933 by W. F. Coultas on Long Island). may still be arriving from the ancestral pop­ If one divides the available series of adult ulations is provided by a subadult male that male hybrida into four subseries by island and I collected on Crown in 1972 (now in the year of collection, the resulting small sub­ American Museum of Natural History await­ series appear not to differ in wing length: ing cataloging; my field number 1637). Its Long, 1933, 14 0', average 190 mm; Long, small size is compatible with either hybrida or 1972, 3 0', average 192 mm; Tolokiwa, 1972, rufiventer, and its plumage is largely the black 2 0', average 195 mm; Crown, 1972, 3 0', adult male plumage, which fails to distinguish average 190 mm. This type of comparison hybrida from rufiventer. However, female­ is not possible for adult females, because the plumaged areas of the tail and underparts four available were all taken on Long in 1933. both agree with New Guinea rufiventer, This stabilized hybrid population must rather than with hybrida or salvadorii. have formed by the year 1933, because the 92. Channel-billed Cuckoo, Scythrops 1933 Long series is already distinct in both novaehollandiae. I saw one individual twice on size and female plumage, and because the Savo on 24 August 1974 and saw and heard 1972 Long/Crown/Tolokiwa specimens (all one three times on Simbo on 11 October of them adult males) do not differ in size 1974. The sole other Solomon record is for from the 1933 Long males. Formation of the Rennel!. Thus, all three Solomon islands of population must have begun sometime after record are species-poor and isolated. Most 1670, because Eudynamys scolopacea is confined records of this species outside are of to forest, and some time must have elapsed wintering visitors from Australia; there are before forest regenerated after the erup­ three unproven reports of breeding in the tion. Recolonization of forest by E. scolopacea Bismarcks, but none in the Solomons. Dispersal, Mimicry, and Geographic Variation in Northern Melanesian Birds . Diamond 7

93. Buff-headed Coucal, Centropus milo. is in the Florida group, where I found coucals Striking reflections of the poor over-water on the largest island (Big Nggela) but not on colonizing ability of this Solomon endemic the second largest island (Small Nggela). The coucal are two examples of its absence on two "islands" are virtually halves of a single moderate-sized islands very close to islands larger island divided by a narrow channel 12 populated by the species. km long, only 150 m wide at its narrowest This coucal is a weak flier that I never saw and 1 km at its widest, and about 300 m wide in long, flapping flight. Instead, its usual for half of its length. Small N ggela is only mode of covering horizontal distances up to slightly smaller than Big Nggela (174 versus 2 about 30 m is to climb a tall tree by a series of 194 km ), nearly as high (1250 versus 1312 short hops and jumps, then glide from the top m), and ecologically similar, and supports of that tree to another tree while losing alti­ most of the same bird species; the sole fla­ tude and occasionally flapping its wings. Like grant discrepancy is the coucal. Florida resi­ many other weak-flying species, the coucal dents are of course thorougWy familiar with is largely confined to larger islands, and the this conspicuous and distinctive bird, for fraction of islands in a given size class that which the local name is "vulev:lu." They told it occupies decreased with decreasing island me that cyclones occasionally bring coucals area. This distributional pattern arises be­ from Big Nggela to Small Nggela, but that cause risk of population increases the birds then disappear or attempt to fly steeply with decreasing population size and back to Big N ggela across the channel, into hence with decreasing island area (Pimm et which most of them fall down. al. 1988). Occasional reimmigrations are re­ The Solomon coucal's closest relative is quired to offset those , but popu­ Pied Coucal, Centropus ateralbus of the Bis­ lations on smaller islands experience more marcks, related in turn to other coucals of the frequent extinctions and receive fewer im­ Moluccas and New Guinea. Because North­ migrants, so that smaller islands lack popu­ ern Melanesian islands arose from the ocean lations an increasing fraction of the time and have never had a land connection to con­ (Diamond and Marshall 1977, Gilpin and tinental land sources, coucals must have Diamond 1981). reached the Solomons over water. Presum­ The first example is in the New Georgia ably, as true of so many island bird pop­ group, where the coucal is present on the ulations; the ancestor of the Solomon coucal 2 nine largest islands (areas 95-2044 km ), ab­ must have been a better over-water disperser sent on the tenth largest island (Wana Wana, and then undergone evolutionary loss of dis­ 2 69 km ), present on the eleventh, twelfth, and persal ability on reaching the Solomons. This 2 thirteenth largest islands (13-35 km ), and evolutionary loss has involved actual flight­ then absent on all smaller islands except for lessness in a few Northern Melanesian species 2 three (0.15-1.6 km ) lying very close (20-150 (perhaps Nesoclopeus woodfordi), weak flying m) to large coucal-occupied islands. Coucals ability in others (such as the coucal), and are easy to detect because of their loud calls strong flying ability but behavioral aversion like a lion roaring, and they are well known to to flying over water in many species. residents of the New Georgia group (name 95. Barn Owl, Tyto alba. Although barn "nao" for adults, "sengege" for juveniles, in owls have been observed or collected by the Roviana language). I failed to find coucals Western ornithologists on 13 Solomon and on Wana Wana Island in two visits there (30 two Bismarck islands outside the New Geor­ September 1974 and 18 September 1976), gia group, only two and Wana Wana residents insisted that it (Vella Lavella and New Georgia itself) have never occurred there. The nearest island to such records. However, residents of eight Wana Wana supporting coucals is Kohinggo, other major islands (Ganonga, Gatukai, Gizo, from which Wana Wana is separated by a Kohinggo, Kulambangra, Rendova, Vangunu, strait 12 km long and only 2 km wide. Wana Wana) and five islets in the group de-' The second, even more striking, example scribed to me a night bird, with the Roviana 8 PACIFIC SCIENCE· January 2002

name of "dunduru," that is evidently this to September and undoubtedly represents species. Residents of New Georgia and Wana nonbreeding winter visitors from Australia. Wana also described a second, smaller, brown However, in the Solomons a few have been re­ species of "dunduru," but no other owl spe­ ported to remain throughout the year (Stevens cies has been recorded from the New Georgia and Tedder 1973), and there are breeding group. A possible candidate is a boobook owl: records from Guadalcanal (Cain and Galbraith the Solomon Boobook, Ninox jacquinoti (see 1956) and Three Sisters (French 1957). next treatment) is known from 12 Solomon I observed Sacred Kingfishers on most islands outside the New Georgia group. Solomon islands that I visited and throughout 97. Solomon Boobook, Ninox jacquinoti. dates offield observations on both ofmy visits The Mono race N j. mono is known only (22 August-28 October 1974 and 7-30 Sep­ from the type series of five specimens col­ tember 1976). Numbers declined markedly lected by the Whitney South Sea Expedition. toward October, and at late dates most of the This was presumably the author of a call that individuals that I saw were on small islets in I heard at 5 A.M. on Mono, a series of seven Roviana and Wana Wana Lagoons and on high notes somewhat similar to calls of the larger outlying islands (Rennell: Diamond races on other Solomon islands (Diamond 1984) rather than on large central islands 1975b). Mono residents attributed the call such as Guadalcanal, Choiseul, and New to a hole-nesting nocturnal bird named the Georgia. Preferred habitats are villages, gar­ "kuru," the same as the name for N jacquinoti dens, the coast, and of large is­ on Choiseul and similar to the name for N lands, and coconut plantations and forest of jacquinoti ("nduru") on Florida and for owls small islands, whereas its somewhat larger (see previous treatment: "dunduru") in the close relative Collared Kingfisher, Halcyon New Georgia group. Residents of Shortland, chloris, prefers forest edge and open forest. an island near Mono and Choiseul from The two species also differ in preferred perch which N jacquinoti had not yet been recorded, heights: H. sancta usually is found below 4 m described to me as present but not common a and H. chloris up to the tops of tall trees. small, brown, nocturnal, hole-dwelling bird At my campsite on Lola Island in Wana with a catlike face and with the same name Wana Lagoon in the period 2-23 September ("kuru") applied to N jacquinoti on Mono and 1976 I repeatedly observed a pair trring to Choiseul. excavate a nest hole in a termite nest 1.5 m 100. White-throated Nightjar, Eurostopodus high on the side of a tree trunk. After aban­ mystacalis. In Northern Melanesia this nightjar doning that site as too small, they excavated is confined to the New Georgia group (nine in the top of a coconut stump 1 m high, until of the larger islands plus at least four lagoon they abandoned that site as well. islets) and the northern Bukida group (Bou­ My most detailed information about nest­ gainville, Shortland group, and Ysabel). My ing and habits came from Teu Zinghite of field experience confirms what I was re­ Kulambangra. According to him, H. sancta peatedly told by Solomon residents: the bird can be found on Kulambangra in any month lives mainly on small islets, and on the sea ofthe year, and he has found numerous nests, side but not the lagoon side of the coast of close to the ground on predator-free small larger islands. The call is a series of 13-25 islands (as I observed on Lola), at heights of staccato notes at a rate of four to five notes 3-6 m on large islands but still lower than the per second. The series initially rises in pitch preferred nest height for H. chloris. Nests are and then levels out to constant pitch. The excavated in rotten trees or in termites' nests quality is slightly musical, as of a log being on the sides oftrees, in sunnier, more exposed, struck with a blunt object. less concealed locations than those used by H. 112. Sacred Kingfisher, Halcyon sancta. chloris. The clutch size on Kulambangra is two Recorded from almost all Northern Mela­ to three for H. sancta, two to four for H. chloris. nesian islands, where the vast majority ofrec­ Eggs ofthe two species are similar except that ords falls in the austral winter from April those ofH. chloris are slightly larger. Dispersal, Mimicry, and Geographic Variation in Northern Melanesian Birds' Diamond 9

119. Blyth's Hornbill, Rhyticeros plicatus. and they are replaced outside forest by The hornbill underwent a major range ex­ Yellow-eyed Cuckoo-Shrike, C. lineata, and pansion in the Solomons several decades ago. White-bellied Cuckoo-Shrike, C. papuensis. When the Whitney Expedition was collecting On high islands (Kulambangra, New Britain, in the Solomons (1927-1930), the hornbill Umboi) I found C. tenuirostris up to about was largely confined to the Bukida chain of 1000 m. islands (Buka to Guadalcanal) plus Malaita. Vocalizations show greater geographic Whitney collectors observed it in the Russell variation than do vocalizations of any other group only on the small island of Leru, and Northern Melanesian species except Golden in the New Georgia group only on Kicha, Whistler, Pachycephala pectoralis. I found the Gatukai, and Vangunu at the southeastern vocalizations recognizably distinct on every end of the group (sight records in volume S island without exception, and distinct be­ of the expedition diaries). When I was in the tween different sites on Umboi. There are Solomons in 1974 and 1976, I saw hornbills three types of vocalizations. First, all pop­ on Borokua halfway between the Russells and ulations share a flight call and contact call, the New Georgia group, and on 11 islands of which is a low-pitched, unmusical, spitted, the New Georgia group (Gatukai, Vangunu, and staccato "whk" repeated rapidly up to New Georgia, Rendova, Tetipari, Kohinggo, five times. Second, the populations of at least Wana Wana, Kulambangra, and three smaller four Bukida islands (Fauro, Choiseul, Florida, islands in Wana Wana Lagoon). Islanders in Guadalcanal) give a series of several slow the New Georgia group told me that the nasal whistled slurs. Related to these Bukida hornbill seasonally visits, but is not resident slurs are the calls of the morphologically very on, Vella Lavella and Giza, that its status on distinct San Cristobal population, C. salamo­ Wana Wana is only as a seasonal visitor, and nis, considered a distinct allospecies: a slightly that it never occurs on Simbo or Ganonga. decelerating series of 7-14 upslurred or Many people told me that this range ex­ downslurred nasal whistles at the rate of 0.7 pansion in the New Georgia group took place sec per slur. after World War II. The year can be pin­ The third and most widespread call is a pointed more closely as falling between 1945, long crescendoing series of clear, ringing, when Sibley (1951) lived in the New Georgia musical notes at or near the same pitch. (This group without encountering hornbills, and is the song replaced in South New Guinea 1949-1954, when Alisasa Bisili lived on Vella and Australia by the series ofunmusical buzzes Lavella and used to hunt them there. I did not familiar to Australian ornithologists and re­ observe hornbills on the Florida Islands (one sponsible for the vernacular name "Cicada­ of the smaller island groups in the Bukida bird"). I heard this song from C. [tenuirostris] chain) nor on Savo between Florida and Gua­ populations on islands of the Papuan Region dalcanal, but residents of both Florida and (e.g., Karkar) and on all occupied islands of Savo said that hornbills were brought by Northern Melanesia except Florida, the Rus­ Cyclone Ida in 1952 but did not remain. sell group, and San Cristobal. Geographic 127. Superspecies Coracina [tenuirostris]: variation in this song involves five features Cicadabird, C. [t] tenuirostris, and San Cristo­ summarized in Table 2. The number ofnotes bal Cicadabird, C. [t.] salamonis. Cicadabird varies from the shortest series of 10-16 on populations of Northern Melanesia are nota­ Shortland and Mono, to the longest Northern ble for their habitat preference and extreme Melanesian series of 41-46 on Vella Lavella, geographic variability in vocalizations (plate 7 to as many as 72 notes on Karkar (Diamond ofMayr and Diamond 2001). and LeCroy 1979). The speed or initial speed The ancestral New Guinea population is varies from the slow songs of Mono, New strictly a nonforest species replaced in forest Britain, Gatukai, Shortland, and Vangunu by five congeners. In contrast, Northern Mel­ (one to two notes per second) to the fast songs anesian populations live mainly in groups of of Guadalcanal, Rendova, and Vella Lavella two to three in the forest canopy above 6 m, (five to six notes per second). The series is 10 PACIFIC SCIENCE· January 2002

TABLE 2 Geographic Variation in the Third Song of the Cuckoo-shrike Coracina tenuirostris

2. 3. 4. 5. 6. 7. 8. 1. Subspecies Subspecies Number Initial speed Change of Change of Change of Island group ofnotes (notes/sec) speed series pitch note pitch

Karkar muellerii muellerii 30-72 D Long muelleni muellerii A start No D Umboi: Arot muellerii rooki No No U Umboi: Lablab muellerii rooki 30 D D U New Britain muellerii heinrothi 1.5 D U Shortland remota saturatior 10-16 2 D++ start U Mono remota saturatior 10-12 1 No D Fauro remota saturatior 13-27 D++ No U Choiseul remota saturatior 12-30 No A D Vella Lavella remota saturatior 41-46 6 D++ end D Gizo remota saturatior D Kulambangra remota saturatior 24-35 3 D No Wana Wana remota saturatior 23-27 3 D++ A W.W.lagoon remota saturatior 18-35 No A,D Vangunu remota saturatior 2 D Dend Gatukai remota saturatior 10-30 1.8 D Rendova remota saturatior 5 No Guadalcanal remota erythropygia 31-33 5 D++ D Uend Savo remota erythropygia 4 A U

Note: Column 1: Island (W.W. lagoon = Wana Wana Lagoon; songs of the Arot and Lablab Districts of Umboi Island differ and are characterized separately). Column 2: Subspecies group (megasubspecies). Column 3: Subspecies (from Appendix 1 of Mayr and Diamond 2001). Column 4: Number ofnotes in the whole song. Column 5: Speed of the song, in notes per second (speed of the whole song if its speed is constant, or initial speed if the speed changes). Column 6: "No," speed ofsong remains constant from start to end; "A, D, D++," accelerates, decelerates, decelerates gready; "start" or "end," accelerates or decelerates especially at start or end ofsong. Column 7: "No, A, D," or "A, D," pitch remains constant, ascends, descends, or ascends then descends from start to end. Column 8: "No, U," or "D," each note remains at one pitch, is an upslur, or is a downslur ("end," each note is at one pitch for most of the song but becomes an upslur or downslur toward the end of the song). Blank entries, I did not establish that characteristic on that island. See text for discussion. delivered at a constant rate on some islands This geographic variability in song corre­ (Choiseul, Mono, Umboi's Arot District, lates imperfectly with geographic variation in Wana Wana Lagoon), but on other islands plumage. On the one hand, the morphologi­ it decelerates slightly (e.g., Kulambangra) cally very distinctive populations of San Cris­ or greatly near the start (e.g., Shortland) or tobal (C. salamonis) and of the Russell group greatly near the end (e.g., Vella Lavella), or (G. t. nisoria) are vocally distinctive in lacking else it accelerates (Long). The series remains the otherwise widespread third song. On the at constant pitch (Fauro, Kulambangra, Long, other hand, the distinctive second song is Umboi's Arot District), descends in pitch given on Bukida islands but not by birds (Umboi's Lablab District, Guadalcanal, Vella on islands of the New Georgia group, all of Lavella), ascends in pitch (Choiseul, Savo, which belong to the race G. t. saturatior; and Wana Wana), or ascends and then descends there is much quantitative variation in the (Wana Wana Lagoon). Finally, the individual third song between populations belonging to notes are either at constant pitch (Rendova), the same subspecies. downslurred (e.g., Gizo, Gatukai), upslurred Finally, there appear to be some parallel (e.g., Fauro, Shortland), initially at constant geographic variation and convergence in the pitch but becoming downslurred at the end of Solomons between songs of G. [tenuirostris] the series (Vangunu), or initially at constant and of the satin flycatcher superspecies Myia­ pitch but becoming upslurred at the end of gra [rubecula], some ofwhose vocalizations on the series (Guadalcanal). Bukida islands, the New Georgia group, and Dispersal, Mimicry, and Geographic Variation in Northern Melanesian Birds . Diamond 11

San Cristobal are surprisingly difficult to dis­ Britain and two of its satellite islands (Long tinguish from the local versions of the second and Vuatom). The Los Angeles County Mu­ and third songs of C. [tenuirostris] on that is­ seum of Natural History contains a specimen land. collected by James Smith on 19 July 1979 at 128. Solomon Cuckoo-shrike, Coracina Hilalom on , where there is also holopolia. This endemic, unobtrusive, little­ one sight record (Finch 1985). known, taxonomically isolated species had 142. Kulambangra Leaf-warbler, Phyllosco­ previously been recorded from nine of the pus amoenus. Nothing has been recorded of larger Solomon islands. On 24 September this warbler, endemic to Kulambangra, since 1976 I observed it twice at sites several kilo­ the collection of the type and one other meters apart on New Georgia, whence records specimen without field notes by the Whitney had been lacking. The species is confined Expedition in 1927 (Hartert 1929). In 7 days to the crowns of tall trees, mainly at perch that I spent on Kulambangra, including 5 heights above 7 m but once down to 4 m, and days camped at high elevation, I observed it mainly in forests on hilly terrain, occasionally only twice (5 and 6 October 1974): single in­ in isolated trees and on flat terrain. Most of dividuals gleaning at 3-5 m in forest 12 m tall my observations were of individuals or pairs near the summit (1595 and 1620 m). The (once a trio) in small mixed flocks with its birds appeared dull-colored above, dull yel­ congeners C. lineata, C. papuensis, and C. ten­ low with a slight olive tinge below, and with a uirostris, and once with Solomon Satin Fly­ suggestion of an indistinctly pale superciliary. catcher, ferrocyanea. I saw one glean The song is a fast, high-pitched, formless and capture an insect 3 cm long, hold it with warble. The widespread Island Leaf-Warbler, one leg, and tear it apart. P. poliocephala pallescens, which I observed The vocalizations of this species have not nearby at the same elevation and in the same been previously described. I found that the habitat, differed in its pale gray breast and subspecies C. h. holopolia of the Bukida group whitish belly without a trace ofyellow and in and C. h. pygmaea ofthe New Georgia group, the much more distinct pale superciliary. which are very distinct morphologically, are These two warbler species differed ecologi­ also distinct vocally. On Guadalcanal (c. h. cally in that all my observations of P. polio­ holopolia) the song is a regularly spaced series cephala were at heights of 6-12 m above the of4-15 identical upslurs delivered at a rate of ground, considerably above the midstory 1 slur/1.5-1.7 sec. Each slur begins slightly heights of P. amoenus, and that I observed P. hoarsely and becomes more nasal. The series amoenus only at the summit (1595-1620 m) initially increases in volume. This song is at but P. poliocephala from the summit down to first easy to confuse with the upslur series of 1020 m. The population of P. amoenus, evi­ the Common Koel, Eudynamys scolopacea, but dently uncommon and confined to the sum­ differs in being delivered faster and compris­ mit of one island, must be small. ing more notes, with the notes spaced at equal 144. Solomon Pied , Rhipidura cock­ rather than irregular time intervals, and with erelli. This Solomon endemic belongs to the the series composed of more notes. On Ku­ same superspecies as, and represents, North­ lambangra and New Georgia (c. h. pygmaea) ern Fantail, R. rufiventris, of the Bismarcks, the song differs in consisting of downslurs New Guinea, Australia, and the Lesser Sun­ rather than upslurs, decelerating slightly in das. It is virtually confined to Solomon islands 2 speed through the series, and dropping of area ~ 95 km • It lives mainly in the slightly in pitch at the end of the series, like shaded open middle story at 3-20 m under a long-playing record played on a turntable the canopy of tall forest, never inside dense that is slowing down. The frequently heard vegetation. Most of my observations were of call of C. h. pygmaea is a single faint nasal solitary individuals separate from the mixed­ note. species insectivorous foraging flocks of other 131. Pied Chat, Saxicola caprata. Previously flycatcher species. known in Northern Melanesia only from New As for its foraging technique, it is the only 12 PACIFIC SCIENCE· January 2002

Solomon bird species that specializes ex­ flock with Rhipidura rufifrons and Pachycephala clusively in sallying. Unlike Rufous Fantail, pectoralis. Unlike R. cockerelli, R. tenebrosa is Rhipidura rufifrons, and other typical a typical fantail in its behavior, though not but like R. rufiventris, R. cockerelli perches with as extreme as R. rufifrons: it often hops and its body held upright rather than horizontally, turns, droops its wings, flares its tail, thereby and it never fans its tail, droops its wings, presenting a maximum cross-sectional area in or crashes into foliage to beat out insects. its line of advance, and moves thus through Holding a perch for up to 10 min, but more dense vegetation while using its large profile typically 10-30 sec (depending on insect to stir up insects. abundance), it turns its head but keeps its 151. Superspecies [melanopsis]: body stationary. From that solitary perch it Spot-eyed Monarch, M. erythrostietus; sallies out to catch an insect in midair with a Chestnut-bellied Monarch, M. castaneiventris; loud snap of the bill, or less often sallies to White-capped Monarch, M. richardsii. pluck an insect from vegetation while flying Throughout the Solomons this is the most past without stopping to hover. Sallying and common lowland flycatcher, a core member sallying/plucking were the sole foraging ofmixed-species insectivorous foraging flocks, modes that I observed for it; I never saw it specializing in capturing prey while hover­ glean, hover-glean, probe, or hop along a ing, but ranging from the understory to the branch. After a sally, it either returns to the canopy. Other flock members are (usually) same perch or else goes to a new perch up to pied monarchs Monarcha [manadensis]; Rufous 10 m away. However, most sallies consist of Fantail, Rhipidura rufifrons; and satin fly­ a large arc or a sweep toward the ground, so catchers Myiagra [rubecula]; often, Solomon that the distance flown in a sally is several Pied Fantail, Rhipidura cockerelli; cicadabirds times the straight-line distance between suc­ Coracina [tenuirostris]; white-eyes New Geor­ cessive perches. gia White-eye, Zosterops rendovae; or Bukida The song is faint and consists of various White-eye, Z. metcalfii; and Golden Whis­ three-note or four-note patterns within a tler, Pachycephala pectoralis; and occasionally, small pitch range, delivered in an unpre­ honey-eaters Myzomela [pammelaena]. Only dictably varying tempo as in a Chopin waltz on San Cristobal were the flocks joined by played rubato. The notes variously consist of Long-tailed Triller, leucopyga, and San voiced, short, slightly hoarse, whistled mono­ Cristobal Starling, Aplonis dichroa, as well as by syllables at a single pitch or else unvoiced di­ San Cristobal Cicadabird, Coracina salamonis, syllables. Each note begins with a sharp attack and San Cristobal Satin Flycatcher, Myiagra like the notes of a keyboard instrument, es­ cervinicauda; the association of these four pecially like a flute stop of a baroque organ. species with each other and with M. castanei­ Songs that I heard from the races septen­ ventris in mixed flocks may be a biological trionalis (heard on Shortland), interposita factor driving the surprising vocal conver­ (Choiseul), cockerelli (Guadalcanal), lavellae gence among these five species on San Cris­ (Ganonga, Vella Lavella), and albina (Ku­ tobal (see discussion at the end of the section lambangra, New Georgia, Vangunu, Tetipari, on Species Accounts). Rendova) were all quite similar. The song Solomon populations of the superspecies pattern is repeated ad nauseam every 2-4 sec (plate 5 of Mayr and Diamond 2001) use at for up to several minutes. I found it very easy least four types of vocalizations, all of which to call up individuals of R. cockerelli by imi­ vary geographically. The first, which is given tating the song. long before dawn as well as during the day 145. San Cristobal Fantail, Rhipidura tene­ and which I take to be the song, is a weak, brosa. I encountered this little-known species tremulous, eerie, high-pitched, whistled note endemic to San Cristobal only once, in the repeated two to five times at intervals of 2-3 shaded middle story and lower crown of 18­ sec. The pied monarchs Monarcha [manadensis] m-tall forest at an elevation of 600 m and at manadensis and M. guttula ofNew Guinea give 6-12 m above the ground. It was in a mixed a very similar song, but (strangely) the North- Dispersal, Mimicry, and Geographic Variation in Northern Melanesian Birds . Diamond 13 ern Melanesian representatives of M. [mana­ cally is only a weakly differentiated subspecies, densis] do not. As for geographic variation, M. is vocally the most distinct Solomon popu­ erythrosticta's version of this song (heard on lation of the M. [melanopsis] superspecies in Bougainville, Fauro, Shortland, and nearby possessing this unique call and in lacking the islets) is either not tremulous or else tremu­ otherwise ubiquitous first vocalization (weak lous just at the end of each note, and is either tremulous repeated note). It is also distinct at constant pitch or downslurred. This song in its foraging mode of hopping rather than of M. richardsii (New Georgia group) is at hovering. constant pitch or slightly upslurred, and cre­ 152. Superspecies Monarcha [manadensis]: scendoes on most islands, but crescendoes Bukida Pied Monarch, Monarcha barbatus; very little on Vangunu and Gatukai and not at New Georgia Pied Monarch, M. browni; San all on Rendova. In M. castaneiventris this song Cristobal Pied Monarch, M. viduus. It was is tremulous and downslurred on most is­ well known previously that the Solomon al­ lands, but varies considerably among islands: lospecies differ strikingly in plumage and size the Choiseul song is very high pitched, not (see plate 3 ofMayr and Diamond 2001). For tremulous, and with long notes; the Florida example, M. viduus is a much smaller bird, songis oftengiven as asimultaneous duet; notes weighing 20% less than the other two allo­ of the Pavuvu song are short and louder and species. I noted equally striking behavioral sound like a police whistle; and the San Cris­ differences in the field: M. viduus is much tobal population does not give this song at all. more nervous and Rhipidura-like in its quick The second vocalization is a nontremulous, movements. On San Cristobal M. [mana­ louder, downslurred, repeated, whistled note densis] viduus is more common and forages at a medium pitch considerably lower than the on the average higher than its congener M. first song. I heard this song from M. eryth­ [melanopsis] castaneiventris, but the relations rosticta (islets near Fauro) and M. castanei­ are reversed in the Bukida and New Georgia ventris (Choiseul, San Cristobal), but not groups, where M. [manadensis] barbatus and from M. richardsii. M. [manadensis] browni, respectively, are less The third vocalization is a rapid, energetic common and forage on the average lower than series of somewhat musical scolding notes, M. [melanopsis] castaneiventris and M. [mela­ reminiscent of the chatter of a squirrel, with nopsis] richardsii, respectively. the sense of a wound-up toy discharging, and 159. Golden Whistler, Pachycephala pectora­ very similar to calls of the closely related Is­ lis. In 1962 the Noona Dan Expedition added land Monarch, M. cinerascens. I heard this call Dyaul to the Northern Melanesian islands in­ from all populations except M. erythrosticta. habited by this hypervariable species, but the The last type of vocalization, heard from Dyaul specimens were not studied taxonomi­ all populations, consists of rasps: either a cally. I examined five adult males and seven single, big, dry rasp like that of Shining Fly­ adult females from Dyaul in the Zoological catcher, Myiagra alecto, or else several ex­ Museum of the University of Copenhagen. tremely harsh rasped notes like New Guinea's Compared with specimens of the race P. p. Black Monarch, M. axillaris. citreogaster that the Noona Dan Expedition Unique to the San Cristobal population is collected on the nearest Bismarck islands a medium-high-pitched, cheerful, fairly loud, (New Ireland, New Hanover, New Britain), slightly hoarse upslur, downslur, or note Dyaul specimens agree well, differing only in slurred up and then down, repeated three to the yellower, less olive, back of males. Males six times, suggestive of calls ofM. cinerascens, of the closely related Golden and convergent on calls ofLong-tailed Triller, Whistler, P. melanura, differ in the same re­ Lalage leucopyga, and other San Cristobal spe­ spect from male P. p. citreogaster, making one cies that join the monarch in mixed flocks (see wonder whether the Dyaul population is of earlier under Coracina salamonis). It is curious hybrid origin (see discussion of hybrid pop­ that the San Cristobal population, M. casta­ ulations below). However, I cannot detect neiventris megarhynchus, which morphologi- other P. melanura-like traits in the Dyaul 14 PACIFIC SCIENCE· January 2002 population: for example, Dyaul males have As for vertical stratum preference, most the olive primary edges and narrow breast populations that I observed prefer the middle band of P. pectoralis. story and lower crown, but the Tetipari and This is the most geographically variable New Georgia populations live mostly high bird species in the world, with 66 recognized in the crown, and the Rennell population re­ subspecies (Galbraith 1956, Mayr 1967), mains mostly below 5 m, never gets above 12 many ofthem very distinctive in plumage and m, and is unique in often feeding thrushlike sometimes considered separate allospecies. on the ground. In addition, males forage Sixteen of these subspecies live in Northern markedly higher than females on some is­ Melanesia, and speciation has proceeded to lands: I observed heights for males and fe­ the point of two sister-species living sympat­ males, respectively, of 8-18 and 0.6-9 m on rically beside P. pectoralis (P. melanura and Kulambangra and 6-24 and 0.6-16 m on Solomon Mountain Whistler, P. implicata). Pavuvu, and Cain and Galbraith (1956) ob­ (See plate 2 of Mayr and Diamond 2001 for served even more extreme vertical segrega­ paintings of seven P. pectoralis races and both tion of the sexes on Guadalcanal. However, sister-species.) Although this geographic vari­ there is much less vertical segregation on San ation in plumage is famous and regularly cited Cristobal, where I found both sexes at 5-15 in textbooks ofbiology (e.g., Mayr 1942), little m, in agreement with the observations there has been reported about variation in behavior of Cain and Galbraith (1956), who noted that and life history. My observations indicate interisland variation in vertical segregation equally marked variation in at least three may relate to interisland variation in sexual characteristics: altitudinal range, vertical stra­ dimorphism of cryptic plumage. tum preferences, and song. I shall describe interisland variation in As for altitudinal distribution, of the 34 song, proceeding from the islands with the Northern Melanesian island populations for simplest songs to those with the most com­ which I have information, 26 occur at sea plex songs. Within each population, each in­ level and extend up to various elevations, and dividual has a repertoire of multiple songs, eight are confined to mountains and absent and individuals differ in their songs. The sim­ from sea level: those on New Britain, New plest and weakest songs are on Tolokiwa (race Ireland, Bougainville, Ganonga, Kulamban­ citreogaster) and Bougainville (bougainvillei) gra, Rendova, Vangunu, and Gatukai. For and consist of only three to seven notes, of example, the populations of Kulambangra, which the first one to four are high pitched Vangunu, and Gatukai are confined to ele­ and constant pitched (not slurred) and tin­ vations above about 150 m; the populations of kling, the next one to two notes lower pitched Rendova and New Britain's Mt. Talawe, to and fuller and still constant pitched, and the above about 300 m; and the populations of last note a single upslur or downslur. Umboi Bougainville's Kieta area, to above about 600 songs (also citreogaster) are almost as simple, m. Altitudinal distribution varies within sub­ differing only in that the first note is some­ species: there are both montane and lowland times disyllabic. All of these songs are similar populations ofthe subspecies citreogaster (New to those of New Guinea's Rusty Whistler, Britain and New Ireland versus Umboi and Pachycephala hyperythra. other islands, respectively), bougainvillei (Bou­ Still weak, and only slightly more complex, gainville versus Buka and Shortland), centralis is the song on Choiseul (orioloides): a rapid (Kulambangra, Vangunu, and Gatukai versus series of 3-10 notes, most of them upslurs or New Georgia and Kohinggo), and melanoptera downslurs, and few ornone ofthem at constant (Rendova versus Tetipari). Furthermore, pitch. TheFloridasong(also orioloides) also con­ there is also variation within the same island: sists ofa rapid series ofupslurs and downslurs, on Bougainville P. pectoralis reaches sea level is louder than on Choiseul, and stands out by at Cape Torokina but not at Kieta or Mt. being very long, often lasting more than 3 sec. Balbi, and on New Britain at Ralum but not Guadalcanal songs (cinnamonea) are similar to on Mt. Talawe. those on Florida but shorter. Dispersal, Mimicry, and Geographic Variation in Northern Melanesian Birds . Diamond 15

Pavuvu songs (pavuvu) are the loudest less often in Vangunu songs) is fast, accel­ songs that I have heard from any Pachycephala erating runs of many notes in a series either species, rivaling the related Grey Shrike­ ascending in pitch, descending in pitch, or , Colluricincla harmonica, in volume. ascending and then descending several times The song is long (up to 18 notes), consists at breakneck speed. As on Vangunu, Rendova, mostly of slurs rather than constant-pitched and Pavuvu, the elements are diverse and in­ notes, includes diverse qualities ofnotes (slurs, clude high soft squeaks, explosive notes, con­ smacking notes, whiplike notes), and crescen­ stant-pitched notes, and slurs. Dawn songs, does. Noteworthy on Pavuvu is song struc­ but not those later in the day, are antiphonal, turing: the song begins with an introductory with the male giving most of the song and the note or slur repeated up to eight times. The female ending the song with one or several song of the distinctive sexually monomorphic explosive and especially loud notes. On Rennell race (feminina) is also loud and di­ Kulambangra and Vangunu the whistler songs verse, thereby reminding one of an Acroce­ completely dominate the dawn chorus, over­ phalus warbler or New World mockingbird whelm the songs of other bird species, and (Mimus) (Diamond 1984). Elements on Ren­ the loud volume of individual singers multi­ nell include many constant-pitched notes, plus plied by the species' abundance makes the . squeaks, mellow notes, impure tones, rapid overall volume deafening. doublets and triplets (a very rapid pair or trio 160. Mangrove Golden Whistler, Pachyce­ of notes), and some slurs. Characteristic on phala melanura. Pachycephala melanura and its Rennell is the repetition ofa note two to eight very close relative P. pectoralis occur together times on one pitch, then on another pitch, and in a fine-grained checkerboard over most of again for a total of up to four pitches, with P. melanura's range from coastal northern each successive pitch set being louder, faster, Australia through coastal eastern New Guinea and less pure toned. The diverse call notes and the Bismarcks to the northwestern Solo­ include extremely harsh notes, a staccato mons (Galbraith 1967, Diamond 1976). On froglike note, and a rattled high-pitched note, the whole, they segregate as a result ofP. mel­ and they are convergent on Rennell calls of anura occupying smaller or more remote or the Island Thrush, Turdus poliocephalus, on volcanically recently defaunated islands, and which the Rennell whistler population also P. pectoralis occupying larger and more central converges in its ground-feeding habits. islands. There is no known case of overlap of The Rendova song (melanoptera) is loud breeding territories, but P. melanura vagrants and diverse in its sounds, which include un­ have been found on islands with resident P. voiced staccato notes (lacking on the pre­ pectoralis populations. The eastern limit of P. viously named islands), bell-like notes, melanura's range is reached in the Shortland squeaky notes, and many slurs. group of the Solomons, just south of Bou­ When I first heard the song on Vangunu gainville. On three small islands there (Akiki (centralis), I considered it the wildest, most at 155 0 38' E, 70 OD'S, Momalufu at 155 0 3D' spectacular Pachycephala song known to me­ E, 60 57' S, and Whitney somewhere nearby) but I had not yet been to Kulambangra (next the Whitney Expedition in 1927 collected a paragraph). Elements include fast runs; re­ population that Hartert (1929) described as a peated slurs that become louder, higher P. pectoralis race P. p. whitneyi, but that Mayr pitched, and more excited with each repeti­ (1932b) showed to constitute a very variable tion; explosive sounds; ringing notes at high hybrid population between P. melanura dahli consistent pitch; and notes variously resem­ (the P. melanura race occupying the Bis­ bling in quality those of the Singing Parrot, marcks and islets in the Northwest Solomons) Geoffroyus heteroclitus; Collared Kingfisher, and P. pectoralis bougainvillei (the P. pectoralis Halcyon chloris; and thrush alarm calls. race ofBougainville, Buka, and Shortland, the The most brilliant P. pectoralis song that I three largest islands in the Northwest Solo­ heard was on Kulambangra (also centralis). A mons). notable, almost unique feature (present much In October 1974 I visited the three largest 16 PACIFIC SCIENCE· January 2002 islands (Shortland, Fauro, Mono) and 10 from P. m. dahli populations established on small islands in the Shortland group. Short­ small islets fringing Bougainville. This is a land itself supports P. pectoralis bougainvillei. typical example of hybridization between two I found no whistlers on Fauro, Mono, and species that maintain reproductive isolation seven of the 10 small islands (from east to in the geographic core of the range, but that north to west going counterclockwise around hybridize at the periphery ofthe range ofone Fauro: Bagora, Samarai, Piru, New, Kanasata, species (P. melanura), where potential con­ Kukuvulu, and Nusave). On the remaining specific mates are hard to find. three islets (Tapanu and Nameless at 1560 08' 168. Kulambangra Mountain White-eye, E, 60 50' S, and Elo at 155 0 53' E, 60 53' S) I Zosterops murphyi. This white-eye, endemic to saw and heard, and collected four males and Kulambangra, was previously known only one female specimen of, evidently hybrid from the type series offive specimens with no populations. Sizes were intermediate between field notes. It proved to be the most abundant the larger P. p. bougainvillei and smaller P. m. bird in the mountains of Kulambangra, living dahli: weights ~ 25, 31, 32, 32, ~ 33 g; wings in large flocks ofup to 100 or more individuals ~ 93, 95, 95, 96, ~ 90 mm; tail ~ 67, 69, 69, that move downhill in the morning and back 70 mm, ~-. The male's throat was either uphill in the midafternoon. I found it from the pure white as in P. m. dahli or else had in­ summit (1620 m) down to 950 m, once even conspicuous lemon tips on white, but was at 660 m, so there is some altitudinal overlap never the pure yellow of P. p. bougainvelli. with its famous lowland congener New Male axillaries were white with variably lemon Georgia White-eye, Zosterops [griseotinctus] edges, intermediate between the two parent rendovae, which I found up to 1130 m. How­ taxa. Male primary edges were gray as in P. ever, the two species overlap commonly only m. dahli or else olive-gray, but never the olive at 950-1070 m, with higher and lower ele­ of P. p. bougainvillei. The female specimen vations occupied mainly just by Z. murphyi was closest to P. m. dahli in its dark gray to and Z. rendovae, respectively. In the field Z. gray-brown (not brown) crown, dark gray murphyi and Z. rendovae kulambangrae are primaries with dull olive-gray edges (not most quickly distinguished by the former's brown), dull olive rather than olive-brown very broad white eye-ring and gray legs, the back, and gray-tipped whitish throat grading latter's very thin white eye-ring and orange posteriorly through pale gray and dirty lemon legs (see plate 1 ofMayr and Diamond 2001). to pale lemon on the belly. Zosterops murphyi is a canopy species, re­ The Eli:> population consisted of at least maining in tall forest mainly above 5 m, but seven individuals, ofwhich at least three were descending to 1 m in the low « 9 m) vegeta­ singing simultaneously. Songs that I heard tion on Kulambangra's summit. Its diet and on all three islets were relatively weak and foraging modes are diverse: I saw it take in­ short and consisted of three to five introduc­ sects, pick at fruit 3 mm in diameter, and tory repeated notes, upslurs, or pairs of slurs, probe at flowers in Eugenia trees shared with ending in a final upslur. These songs are simi­ Meek's Lorikeet, Charmosyna meeki, and New lar to P. m. dahli songs of the Bismarcks and Georgia Myzomela, Myzomela eichhorni. P. p. bougainvillei songs of Bougainville, which A nest found on 5 October 1974 at a are in turn fairly similar to each other. They height of 1.5 m in a 1.8-m-high bush on Ku­ differ from P. p. orioloides songs of Choiseul lambangra's summit was a woven cup 5 cm and songs of other Solomon P. pectoralis pop­ deep and 5 cm in diameter, made of tiny dry ulations (see preceding treatment). twigs and containing two naked chicks. Evidently, small islands of the Shortland The song is a high-pitched warble distinct group still supported hybrid populations in among Zosterops songs by being very long (up 1974, as they did in 1927. The populations to 5 sec) and not distinctly broken into sylla­ are evidently maintained by dispersing va­ bles as is the "telegraph" song of Z. rendovae grants from P. p. bougainvillei populations es­ (Diamond 1998). The quality is very sweet tablished on Bougainville and Shortland, and and sometimes penetrating. Many notes have Dispersal, Mimicry, and Geographic Variation in Northern Melanesian Birds . Diamond 17 a sharp attack, like the sound ofa coin dropped "kikite-marao," meaning "the kikite from on a hard surface. The overall effect resembles Marao Oocal name for Three Sisters]" be­ a small version of the song of Europe's Com­ cause these M. cardinalis vagrants are believed mon Blackbird, Turdus merula, as also true for to originate from Three Sisters, as may in­ the race Z. rendovae kulambangrae, so there deed be true. The sole bird collections may be vocal convergence between the two obtained on Santa Anna, by the Whitney Ex­ Kulambangra white-eye populations. The call pedition in 1927 and 1930, included 17 is either a constantly given peeping note with specimens of "pure" M. tristrami, no "pure" a sharp attack, faint from each individual bird M. cardinalis, and two evident hybrids closest but cumulatively loud from the whole flock, to M. tristrami (black except for a few red­ or else a descending whinny. edged feathers on the rump: Mayr 1932a). 175. , Myzomela cardi­ As for San Cristobal itself, I found (as did nalis, and 177. San Cristobal Myzomela, My­ Cain and Galbraith [1956] in 1953) M. tris­ zomela tristrami. The San Cristobal group of trami abundant everywhere from the coast islands at the southeastern end of the Solo­ inland to the highest elevation that I reached mon chain consists of the large island of San at 635 m. Myzomela cardinalis was abundant in Cristobal, plus four nearby (8-20 km) smaller the Kira Kira area on the coast and in coastal islands (Santa Anna, Santa Catalina, Ugi, coconut plantations, but the farthest from the Three Sisters). The current distributions of coast that I saw even a single individual was these two honey-eaters in the group are only a few hundred meters inland along the seemingly straightforward: both species occur Magoha River. Both species fed in the same on San Cristobal, M. tristrami alone on Santa flowering trees at the government station at Anna and Santa Catalina, and M. cardinalis Kira Kira, where M. cardinalis chased M. tris­ alone on Ugi and Three Sisters. However, trami, as expected from the former's some­ new information and reexamination of speci­ what larger size. mens suggest a more complex and interesting According to San Cristobal residents, M. story of a recent invasion, ongoing dispersal, cardinalis was formerly absent, arrived re­ and hybridization (see plate 4 of Mayr and cently, is still spreading over the island, and Diamond 2001). hybridizes with M. tristrami at the front ofits The facts are as follows. On the most re­ expanding range where the two species have mote of the four smaller islands, Three Sis­ just come into contact, but hybridization ters (20 km from San Cristobal), M. cardinalis ceases in areas of longer contact. This ac­ is resident, and M. tristrami has never been count is supported by my examination of recorded (French 1957). On Ugi (8 km from specimens and of the literature (Figure 1). San Cristobal) as well, M. cardinalis is resi­ Nineteenth-century collectors on San Cris­ dent, but Ugi islanders told me that M. tris­ tobal obtained only M. tristrami, neither M. trami occasionally visits from San Cristobal, cardinalis nor hybrids. A. S. Meek in 1908 and Dutson (2001) saw two individuals. Ugi obtained M. tristrami and hybrids, but no names are "kikito" for M. cardinalis and pure M. cardinalis; the Whitney Expedition "kikito-ni-hanuato" (meaning "kikito from in 1927-1930 found M. cardinalis nearly as the big land [San Cristobal]") for M. tris­ abundant as M. tristrami and also took hy­ trami. On Santa Anna and Santa Catalina brids; and Cain and Galbraith (1956) found (also 8-9 km from San Cristobal) M. tristrami M. cardinalis more abundant than M. tristrami is the resident species, but G. Dennis and G. and took no hybrids. In many other cases of Kuper of Santa Anna and W. Ma-onga of range expansions as well, hybridization occurs Santa Catalina told me that M. cardinalis ar­ in early stages or at the geographic front, rives infrequently in small numbers at both where the expanding species is greatly out­ islands (e.g., after a December 1971 cyclone) numbered by the resident species and has and remains for some time near the coast. difficulty finding conspecific mates (Mayr The name given to M. tristrami on Santa 1963; cf. my discussion of Pachycephala mel­ Anna is "kikite," but M. cardinalis is called anura x P. pectoralis earlier). 18 PACIFIC SCIENCE· January 2002

in forest understory near the summit of 100 • Kulambangra. ~mi 188. SuperspeciesAplonis [grandil']: Brown­

75 winged Starling, Aplonis [grandil'] grandis, and .~ ,0 San Cristobal Starling, A. [grandis] dichroa. .x /cardinalis This superspecies, endemic to the Solomons, % 50 is represented on the Bukida and New Geor­ gia groups by the large A. g. grandil', and on cf' "'" Guadalcanal by the very similar A. g. macrura; 25 hybrids~~ CJ-_.. / • • g. "" '')( on Malaita, by the intermediate-sized A. " / '" malaitae; and on San Cristobal, by the small a (J,;0- -- -0'" CJ..------CJ A. dichroa. My field observations comparing A. grandis and A. dichroa supplement previous 1855-1880 1908 1927 1953 notes by Cain and Galbraith (1956) and by Finch (1986). My comments about A. grandis FIGURE 1. Relative frequencies (as percentages) ofMyzo­ mela tristrami, M. cardinalis, and hybrids between them, in apply equally to the races grandil' and ma­ collections from coastal areas of San Cristobal at four crura, between which I did not note differ­ different periods. 1855-1880, various collectors; 1908, ences; I did not visit Malaita. A. S. Meek; 1927, Whitney Expedition; 1953, Cain and Aplonil' grandis is widely distributed. I ob­ Galbraith. Note that the first collectors found only M. served it on every island of the New Georgia tristrami, later collectors found hybrids and rising num­ 2 bers of M. cardinalis, and the latest collectors found M. group exceeding 35 km in area, and on 14 cardinalis outnumbering M. tristrami and did not find small islands in lagoons. It has also been re­ hybrids. This appearance and subsequent disappearance corded on every moderate-sized island of the of hybridization is a frequent phenomenon when one Bukida group. Hence it must colonize across species invades the range of a close relative. See text for discussion. water frequently. Although I never saw it fly over water, Teu Zinghite has seen it cross between islands up to 2 km apart. In contrast, The invasion of San Cristobal by M. car­ A. dichroa is known only from San Cristobal, dinalis must have originated from the resident has never been reported on any ofthe smaller populations of the same subspecies (M. c. islands of the San Cristobal group, and evi­ pulcherrima) on Three Sisters and Ugi. In ac­ dently does not cross water now. cord with this interpretation, the first known Aplonis grandil' occurs mainly outside forest established M. cardinalis population on San or in very open forest, much less often in Cristobal was that encountered by the Whit­ deep forest, mainly in the lowlands (including ney Expedition in 1927 at Kira Kira and on the coast), and only uncommonly in the Wainoni Bay on the north coast opposite mountains. I recorded it a few times in mon­ Three Sisters and Ugi. The Whitney collec­ tane forest of Kulambangra up to 1000 m, tors did not find the species during their stay never in the mountains of Vangunu or Ren­ at Star Harbor at San Cristobal's east end dova. In contrast, A. dichroa occurs regularly opposite Santa Anna and Santa Catalina Is­ in forest as well as in more open habitats; lands, where M. cardinalis is still not estab­ I found it inland but not immediately on the lished (volumes D, P, and V of the Whitney coast; and Cain and Galbraith (1956) ob­ Expedition diaries). It would be of great in­ served it most often in the mountains. Both terest today to survey the coasts of San Cris­ allospecies prefer the canopy, but A. grandis is tobal to determine how far along the coastline more extreme in its preference for the top of M. cardinalis has spread, and where hybrid­ the tallest trees. ization is still occurring. Aplonis grandil' occurs mainly in pairs. Out 183. Blue-faced Parrot-Finch, Erythrura of49 cases in which I tallied the number seen, trichroa. Previously known in the Solomons there was one quartet, three trios, 44 pairs, only from Bougainville and Guadalcanal. I and one individual. However, A. dichroa oc­ heard an individual twice and saw it twice curs mainly in groups of three to eight. Dispersal, Mimicry, and Geographic Variation in Northern Melanesian Birds . Diamond 19

Aplonis dichroa appears to be considerably Cristobal), whose song is louder and includes more abundant than A. grandis. many fewer clicks. Aplonis grandis and M. du­ Aplonis grandis, but never A. dichroa, has montii are also similar in their typically loud the distinctive habit of sitting for a long time wing beats (not true of A. dichroa). The vo­ at midday with its black plumage fully exposed calizations of A. dichroa are instead similar to the hot sun, like a cold-blooded to those of three San Cristobal endemics: trying to warm itself. Why doesn't A. grandis San Cristobal Cicadabird, Coracina salomonis; overheat? Long-tailed Triller, Lalage leucopyga; and the I observed both allospecies to eat mainly San Cristobal race megarhynchus of Chestnut­ fruit, mostly tiny fruits of diameter 1-2 mm, bellied Monarch, Monarcha castaneiventris. but occasionally up to 5 mm in the case of the These vocal similarities may have some bio­ larger A. grandis. logical significance, because the vocally similar The nest of A. grandis is a conspicuous species in each case except for the insec­ large mass of sticks, vines, moss, and lichens tivorous monarch overlap in frugivorous diet. 20-60 cm in diameter in the crown of a tall tree, often an isolated dead tree. According to DISCUSSION Teu Zinghite, on Kulambangra one clutch per year of two (less often, three) eggs is laid Northern Melanesia's rich avifauna, its hun­ in the period September to November, and dreds ofendemic bird taxa, and its geographic the usual stick nest is much less often re­ fragmentation into hundreds of ornithologi­ placed by a hole in a branch. On San Cristo­ cally significant islands combine to make it bal neither I nor anyone else has noted the a premier study area for the field biologist. conspicuous stick nests, and one resident told This paper can only begin to indicate some of me that A. dichroa instead nests in a hole in a the innumerable problems awaiting further tree. study. The "song" of A. grandis is a medley of It is clear that the geographic variation in high-pitched soft notes of diverse, unmusical, morphology for which Northern Melanesian peculiar, unbirdlike qualities, delivered at birds are famous is paralleled by geographic a rate of about 0.7 sec per note. Common variation in other biological attributes. These elements are staccato, very short, slightly attributes include voice (e.g., Accipiter albogu­ sucked-in, unvoiced clicks at various pitches; laris, Ptilinopus solomonensis, Reinwardtoena staccato, short, snapping notes; gurgles; short [reinwardtii], Ninox [novaeseelandiae], Halcyon bell-like notes; downslurred mellow whistles; chloris, Coracina [tenuirostris], C. holopolia, short crescendoing whistles at constant pitch; Monarcha [melanopsis], M. [manadensis], Pa­ and a very high-pitched creaking upslur that chycephala pectoralis, and Aplonis [grandisJ), seems to strain the bird's throat and to rise in behavior (Monarcha [melanopsis] , M. [mana­ pitch beyond the human auditory range. The densis] , Aplonis [grandisJ), habitat preference "song" ofA. dichroa is also soft and varied but (Coracina [tenuirostris], Aplonis [grandisJ), alti­ otherwise entirely different: a whistled upslur tudinal range (Ptilinopus solomonensis, Pachyce­ repeated every few seconds; a mewing upslur phala pectoralis), vertical stratum (Monarcha at medium-high pitch; a high-pitched nasal [manadensis] , Pachycephala pectoralis, Aplonis note; a high-pitched piercing downslur; and [grandisJ), abundance (Ptilinopus solomonensis), a medium-high-pitched musical complex slur and nest (Aplonis [grandisJ). (downslurred then upslurred, or else vice There are many cases of sympatric pop­ versa) repeated 8-10 times at about 0.6 sec per ulations whose vocalizations are much more slur. similar to each other than are the vocalizations Both allospecies appear to be involved of populations of the same species from dif­ in vocal mimicry or convergence. The vocal­ ferent islands (e.g., Accipiter albogularis and izations of A. grandis are confusingly similar Halcyon chloris, Coracina [tenuirostris] and to those of the slightly larger starling Mino other species in its mixed-species foraging dumontii (absent on A. dichroa's island of San flocks, and Zosterops murphyi and Z. rendovae 20 PACIFIC SCIENCE . January 2002 kulambang;rae). These cases pose the questions help with fieldwork and for sharing their of whether the resemblance involves vocal knowledge; the staff of the Bird Department mimicry, vocal convergence, or just chance, of the American Museum of Natural History and-if the explanation is not chance-what for allowing me to study their collections for might be the selective forces underlying the the past 37 years; Jon Fjeldsa and Kimball mimicry or convergence. Garrett for allowing me to study bird collec­ I have reported three cases of hybridiza­ tions in the Zoological Museum of the Uni­ tion: between two colonist populations arriv­ versity of Copenhagen and the Los Angeles ing at the same unoccupied island (Eudynamys County Museum of Natural History, respec­ scolopacea), between an established species and tively; and the National Geographic Society an invading relative (Myzomela tristami and for support. M. cardinalis), and between a species in the center and another species at the periphery of Literature Cited its geographic range (Pachycephala pectoralis and P. melanura). Careful study is likely to de­ Ball, E. E., and I. M. Hughes. 1982. Long tect many more cases in which vagrants dis­ Island, : People, re­ persing over water become incorporated by sources, and culture. Rec. Aust. Mus. hybridization into a related resident popula­ 34:463-525. tion on an island that they reach. Beehler, B., T. Pratt, D. Zimmerman, H. This paper illustrates the enormous range Bell, B. Finch, J. M. Diamond, and J. Coe. of over-water dispersal ability among North­ 1986. Birds of New Guinea. Princeton ern Melanesian bird populations. Some are University Press, Princeton, NewJersey. maintained by Australian winter visitors (e.g., Blong, R. J. 1982. The time of darkness. Scythrops novaehollandiae, Halcyon sancta) or Australian National University Press, Australian wind-blown vagrants (Pelecanus Canberra. conspicillatus, Phalacrocorax melanoleucos, Ardea Bradley, D., and T. Wolff. 1956. The birds of alba, possibly Threskiornis moluccus) that only . Nat. Hist. Rennell lsI., Br. occasionally remain to breed. Northern Mel­ Solomon lsI. 1:85-120. anesian resident species themselves include Cain, A. J., and I. C. J. Galbraith. 1956. Field species that can be seen daily flying over wa­ notes on birds of eastern Solomon Islands. ter (Haliaeetus sanfordi, Accipiter albo!fUlaris), Ibis 98:100-134, 262-295. species that readily colonize defaunated vol­ Diamond, J. M. 1966. Zoological classifica­ canic islands (Ptilinopus solomonensis, Eudyn­ tion system of a primitive people. Science amys scolopacea), species that fly short but not (Washington, D.C.) 151:1102-1104. long distances over water (Cacatua ducorpsi), ---. 1971. Bird records for western New strong fliers that refuse to fly over water, weak Britain. Condor 73:481-483. fliers (Centropus milo), and flightless species ---. 1972. Further examples of dual sing­ (Nesoclopeus woodfordi). ing by Southwest Pacific birds. Auk Even within the 32 years since I first visited 89:180-183. Northern Melanesia, deforestation has re­ ---. 1974. Colonization of exploded vol­ duced opportunities to study these biological canic islands by birds: The supertramp phenomena. The greatly accelerated rate of strategy. Science (Washington, D.C.) deforestation today lends urgency to further 183:803-806. studies. ---. 1975a. Assembly of species com­ munities. Pages 342-444 in M. L. Cody and J. M. Diamond, eds. Ecology and ACKNOWLEDGMENTS evolution of communities. Harvard Uni­ I thank for 47 years offriendship versity Press, Cambridge, Massachusetts. and inspiration about Northern Melanesian ---. 1975b. Distributional ecology and birds; Teu Zinghite, Alisasa Bisili, and many habits of some Bougainville birds (Solo­ other residents of Northern Melanesia for mon Islands). Condor 77:14-23. Dispersal, Mimicry, and Geographic Variation in Northern Melanesian Birds . Diamond 21

---. 1976. Preliminary results of an orni­ Finch, B. W. 1985. Noteworthy observations thological exploration of the islands of in Papua New Guinea and Solomons. Vitiaz and Dampier Straits, Papua New Papua New Guinea Bird Soc. Newsl. Guinea. Emu 76:1-7. 215:6-12. ---. 1984. The avifaunas of Rennell and --. 1986. The Aplonis starlings of the Bellona Islands. Nat. Hist. Rennell lsI., Br. Solomon Islands. Muruk 1:3-16. Solomon lsI. 8:127-168. French, W. 1957. Birds of the Solomon ---. 1989a. A new subspecies of the Islands. Ibis 99:126-127. Island Thrush Turdus poliocephalus from Galbraith, I. c.]. 1956. Variation, relation­ Tolokiwa Island in the Bismarch Archi­ ships, and evolution in the Pachycephala pelago. Emu 89:58-60. pectoralis superspecies (Aves, Muscicapi­ 1989b. This-fellow frog, name dae). Bull. Br. Mus. (Nat. Rist.) Zool. belong-him dakwo. Nat. Hist. 98 (4): 16-23. 4:133-222. ---. 1989c. The ethnobiologist's di­ --. 1967. The Black-tailed and Robust lemma. Nat. Hist. 98 (6): 26-30. Whistlers Pachycephala melanura as a spe­ ---. 1991. A new species of rail from the cies distinct from the Golden Whistler P. Solomon Islands and convergent evolution pectoralis. Emu 66:289-294. of insular flightlessness. Auk 108:461-470. Gilpin, M. E., and ]. M. Diamond. 1976. ---. 1998. Geographic variation in vocal­ Calculation of immigration and extinction izations of the white-eye superspecies Zos­ curves from the species-distance relation. terops [griseotinetus] in the New Georgia Proc. Natl. Acad. Sci. U.S.A. 73:4130­ group. Emu 98:70-74. 4134. Diamond, ]. M., and K. D. Bishop. 1999. ---. 1981. Immigration and extinction Ethno-ornithology of the Ketengban peo­ probabilities for individual species: Rela­ ple, New Guinea. Pages 17-45 tion to incidence functions and species in S. Atran and D. Medin, eds. Folk biol­ colonization curves. Proc. Natl. Acad. Sci. ogy. MIT Press, Cambridge, Massachu­ U.S.A. 78:392-396. setts. Hadden, D. 1981. Birds of the North Solo­ Diamond,]. M., and M. LeCroy. 1979. Birds mons. Wau Ecology Institute, Wau, Papua of Karkar and Bagabag Islands, New New Guinea. Guinea. Bull. Am. Mus. Nat. Hist. Hartert, E. 1929. Birds collected during the 164:469-531. Whitney South Sea Expedition. 8. Notes Diamond, ]. M., and A. G. Marshall. 1977. on birds from the Solomon Islands. Am. Distributional ecology of New Hebridean Mus. Novit. 364. birds: A species kaleidoscope. ]. Anim. Mayr, E. 1932a. Birds collected during the Ecol. 46:703-727. Whitney South Sea Expedition. 18. Notes Diamond,]. M., and E. Mayr. 1976. Species­ on Meliphagidae from Polynesia and the area relation for birds of the Solomon Solomon Islands. Am. Mus. Novit. 516. Archipelago. Proc. Natl. Acad. Sci. U.S.A. ---. 1932b. Birds collected during the 73:262-266. Whitney South Sea Expedition. 20. Notes Diamond, ]. M., M. E. Gilpin, and E. Mayr. on thickheads (Pachycephala) from the Sol­ 1976. The species-distance relation for omon Islands. Am. Mus. Novit. 522. birds of the Solomon Archipelago, and the ---. 1942. Systematics and the origin of paradox of the great speciators. Proc. Natl. species. Columbia University Press, New Acad. Sci. U.S.A. 73:2160-2164. York. Diamond,]. M., S. L. Pimm, M. Gilpin, and ---. 1955. Notes on the birds of North­ M. LeCroy. 1989. Rapid evolution of ern Melanesia. 3. Passeres. Am. Mus. character displacement in myzomelid Novit. 1707. . Am. Nat. 134:675-708. ---. 1963. species and evolution. Dutson, G. 2001. New distributional ranges Harvard University Press, Cambridge, for Melanesian birds. Emu (in press). Massachusetts. 22 PACIFIC SCIENCE· January 2002

--. 1967. Subfamily Pachycephalinae, 1981. Pyroclastic deposits and eruptive Whistlers or Thickheads. Pages 3-51 in R. sequences on Long Island, Papua New A. Paynter, ed. Check-list of birds of the Guinea. Papua New Guinea Geol. Surv. world. Vol. 12. Museum of Comparative Men. 10:101-113. Zoology, Cambridge, Massachusetts. Pimm, S. L., H. L.Jones, and]. M. Diamond. Mayr, E., and]. M. Diamond. 1976. Birds on 1988. On the risk of extinction. Am. Nat. islands in the sky: Origin of the montane 132:757-785. avifauna of Northern Melanesia. Proc. Sibley, C. G. 1951. Notes on the birds of Natl. Acad. Sci. U.S.A. 73:1765-1769. New Georgia, central Solomon Islands. --. 2001. The birds of Northern Mela­ Condor 53:81-92. nesia: Speciation, ecology, and biogeogra­ Stevens, G. W., and]. L. O. Tedder. 1973. A phy. Oxford University Press, New York. Honiara bird guide. British Solomon Is­ Pain, C. F., R. ]. Blong, and C. O. McKee. lands Scout Association, Honiara.