AMERICAN MUSEUM NOVITATES Published by Number 1237 THE AMERICAN MUSEUM OF NATURAL HISTORY June 30, 1943 New York City BIRDS COLLECTED DURING THE WHITNEY SOUTH SEA BIRDS COLLECTED DURING THE WHITNEY SOUTH SEA EXPEDITION. 521 NOTES ON SOME NON-PASSERINE GENERA, 3 BY DEAN AMADON

The present taxonomic notes on some of Dr. E. Mayr gave me much valuable the non-passerine birds of Polynesia col- assistance and advice in the preparation of lected by the Whitney Expedition are a the manuscript, and in other ways. For continuation of those contained in Ameri- the loan of rare specimens or for other can Museum Novitates Nos. 1175 and courtesies I am greatly indebted to Drs. 1176. In this paper species belonging to H. Friedmnann, A. Wetmore, J. T. Zimmer one genus of the Anatidae and to several and Mr. J. L. Peters. genera of the Columbidae are discussed.

A REVISION OF ANAS SUPERCILIOSA Anas superciliosa is a typical member spread of superciliosa into the Pacific of the central, mallard-like group of the islands has given it a range more or less genus Anas. Its distribution suggests complementary, insofar as that of an island that this species is an Australian endemism species can be, with those of oustaleti and which later spread to New Zealand, the luzonica; this is believed to be secondary East Indies and much of Polynesia. Anas and fortuitous. superciliosa resembles A. luzonica of the In all, 273 specimens of superciliosa, not Philippines sufficiently to suggest that the including downy young, were studied. two may have been subspecies at some Dr. Herbert Friedmann of the U. S. Na- remote period. Other species of this tional Museum generously lent six Celebes genus native to the islands of the western skins from the series from which Riley Pacific are wyvilliana of Hawaii, laysanen- described the race percna. sis of Laysan and oustaleti of the Marianne VARIABLE CHARACTERS.-New Zealand Islands. The latter three species are specimens are paler than those from other evidently rather recent derivatives of the localities; this is the only geographical mallard, A. platyrhynchos, as shown by the color variation that was found. Non-. presence of curled tail feathers in some geographical variation is extensive and in males of all three and by numerous other small samples deceptive. This has re- characters. Presumably wyvilliana, like sulted in great over-emphasis on the im- Nesochen sandwichensis, evolved from portance of color as a racial character in North American migrants or stragglers this duck. Fading and bleaching, proc- which remained in the Hawaiian Islands. esses which are accentuated by the glare of Whether laysanensis and oustaleti repre- sunlight upon water and perhaps by the sent independent colonizations by mallards frequent contact of water with the plum- or secondary extensions of range from age, greatly affect its coloration. Many Hawaii is not at once apparent. The skins are discolored by the ferruginous stains so common on waterfowl. The 1 The preceding ten papers in this series are Ameri- light areas of the throat and neck are the can Museum Novitates, Nos. 1057, 1091, 1116, 1133, 1144, 1152, 1166, 1175, 1176 and 1192. first to become stained, sometimes so uni- 2 AMERICAN MUSEUM NOVITATES [No. 1237

formly that this buffy coloration appears urements given below average consider- natural. Misinterpretation of such stains, ably larger than those given by many other in my opinion, led Riley to state that writers for this species. For example, Celebes birds have the throat "deeper buff, wing measurements of males of the race more pinkish . . . " than Australian ones rogersi, as recorded in the literature, usu- (1919, Proc. Biol. Soc. Wash., XXXII, p. ally vary from 230-275 mm.; actually very 94). few adult males of this form with the pri- Sexual differences in color are restricted maries in good condition have a wing length to the elongated inner secondaries. In qf less than 258 or 260 mm. females these (always?) have buffy longi- tudinal streaks in addition to their narrow Anas superciliosa superciliosa Gmelin buffy margins. Unfortunately these TYPE LoCALITY.-New Zealand. streaks often disappear completely with SUBSPECIFIC CHARACTERS.-Like rogersi of wear, but when present they are diagnostic Australia but with the feather margins pale of females. Females also tend to have less grayish or buffy white, rather than buffy or brownish white; hence a paler bird with more of the metallic purple-green of the specu- conspicuous feather margins; size the same as lum continued on the inner vanes of these in roaersi. longer secondaries, but this is not constant. RANGE.-New Zealand region: "Kermadec Size variation in Anas superciliosa is Islands, North, South and Steward Islands, Great Barrier Island, Mayor Island, Kapiti, better reflected in the wing length than in Chatham, Auckland, Campbell and Macquarie any other "skin" measurement. Wing Islands. Abundant everywhere except on the length is greatly affected by wear. That southern islands where small numbers only have part of the tip of each of the longer pri- been seen" (Oliver, 1930, Birds New Zealand, p. 217). maries which is not covered by the one WING.-New ZEALAND: e 256+, 262; 9 above becomes bleached and weakened 246, 246+, 255+ (sex ?). by the sun until it breaks off as though cut CULMEN.-New Zealand: e 50, 53.5; 9 with a knife. The wing is thus shortened 47.5,49,52.5. by several millimeters in a way that is not REMARKS.-A. s. superciliosa is a rather evident without careful inspection. An- poorly differentiated race, but specimens other difficulty results from the anseri- in unworn plumage can probably be sepa- form molt in which all the primaries are rated from rogersi with few exceptions. renewed together. Primaries which ap- Eight old specimens from New Zealand pear newly molted and in good condition and a single juvenal female collected by for measuring may lack several millimeters the Whitney Expedition on Pitt Island, of attaining full length, even though the Chathams, in 1926 were available. The sheath has disappeared from their shafts. latter specimen is very pale, with the Such partially grown primaries are best feathers broadly margined; hence it rep- detected by the fact that they do not resents extreme development of the char- project beyond the ends of the secondaries acters of superciliosa. This is probably as much as is normal. The absence of any to be attributed to its immaturity or pos- characters by which immature birds may sibly to individual variation rather than be recognized as such is a further difficulty. to geographical variation. In general their plumage is softer and According to Oliver, several partial or more subject to wear and staining, and complete albinos of this race have been they are smaller than adults. Ma'es are recorded; he suggests that this may be the considerably larger than females, rnd result of crossing with feral domestic among adults only a few of the extremes ducks. One of the eight New Zealand overlap in wing length. birds examined is partially albinistic. It Since so many factors affect wing length is in wretched condition but shows no indi- in this species, the elimination of all ex- cation of the larger size or color characters cept geographical variation is difficult, b-4, which might be expected in a cross with every attempt to do so has been ma(le '.i domestic or mallard stock. Probably this the present study. As a result the mean- race has acquired an inherent albinistic 1943] BIRDS OF THE WHITNEY SOUTH SEA EXPEDITION. 52 3 tendency, as have so many other New Zea- occur on the mainland of south New land birds. Guinea, although no measurements seem to have been recorded. A female from Anas superciliosa rogersi Mathews Misima is rather small; perhaps it is im- Anas superciliosa rogersi MATHEWS, 1912, mature, but it is also possible that the Austr. Av. Rec., I, p. 33, Augusta, southwestern smaller race which occurs in north New Australia. Guinea has infiltrated to some extent Ana8s superciliosa percna RILEY, 1919, Proc. Biol. Soc. Wash., XXXII, p. 93, Celebes. around the eastern end of New Guinea. SUBSPECIFIC CHARACTERS.-Differing from In the Oranje, Saruwaged and perhaps superciliosa as noted above; in color similar to other mountains of New Guinea popula- the third race, pelewensi., but larger, with the tions occur intermediate between rogersi wing averaging about 20 mm. longer. RANGE.-Tasmania-0, Australia-37, Su- and pelewensis but distinctly nearer to the matra-3, Java-{, Kangean-3, Lesser Sunda latter and referred to it in this paper (see Ids. (Lombok-1, Timor-2, Savu-3, Sumba- measurements below), although some pre- 1, Sumbawa-0, Flores-0), Celebes-7, Moluc- vious authors have referred them to cas (Buru-1), Arfak Mts., Vogelkop, New Guinea-1, southern coast of New Guinea-0, rogersi. On the other hand, a male taken Louisiade Archipelago (Misima or St. Aignan by Mayr in the Arfak Mountains has the Is.-2). (Figure after localities indicates num- wing length of 260 mm. and represents ber of specimens examined.) rogersi, which has perhaps reached the WING.-Southern (temperate) half of Aus- Vogelkop from Buru. tralia: e 260 (type rogersi), 267, 267, 268; 9 This male is con- 250, 251. Northern (tropical) half of Australia: siderably larger than the maximum of a e 256+, 260, 260, 260?, 261, 264, 266?, 275; 9 series from higher altitudes in the Oranje 249?, 250, 250?, 252, 252, 253, 254. Sumatra: Mountains. e 258; 9 243?, 250. Lombok: e 263? Savu: e 257; 9 250. Sumba: cP 266+. Celebes: ce 260+, 265?, 270; 9 246, 248?- Anas suaperciliosa pelewensis Buru: eP260. Arfak Mts.: o' 260. Misima: Hartlaub and Finsch e 262; 9 234(?). CULMEN.-Australia: 14 e 49.5-55 (52.36); TYPE LOCALITY.-Palau (Pelew) Ids. 6 9 48-53 (50.0). Misima: cl 53.5; 9 45. SUBSPECIFIC CHARACTERS.-Like rogersi in Arfak: 6' 52. Celebes: ce 50.5, 52, 52, 53 color, but markedly smaller, wing averaging (51.9); 9 48, 50 (49). Other East Indies Ids.: about 20 mm. shorter, except in one or two local- 5 e 49-51.5 (50.4); 5 9 46-51 (48.4). ities as noted below. RANGE.-Society Ids. (Tahiti-2, Moorea- REMARKS.-There seems to be no geo- 16, Huaheine, Raiatea-1), Austral Ids. (Rimi- graphical variation within Australia. The tara-7, Tubuai-9, Rapa-14), Cook Ids. populations of the various East Indian (Rarotonga-1), Tonga Ids. (Fanua Lai-4, islands other than New Guinea may have Vavua, -3, Tongatabu), Samoa (Anuu-1, Tutuila-1, Upolu-3), (Ono a slightly shorter average wing length than Ilau-4, Mothe-2, Oneata-4, Mango-6, those of Australia, but a very large series Kanathea, Taviuni-1, Ngau-2, , Viti and statistical treatment would be needed Levu, Matathoni-1, Kandavu-3), Niuafoo to be sure of The culmen of Id., New Caledonia-2, Loyalty Ids. (Uvea-2, this. length Lifu), New Hebrides (Aneiteum, -1, the Australian birds averages slightly Aoba-1 duckling, ), Banks Ids. longer, but it is possible that more material (-3), Santa Cruz Ids. (Tucopia-13), would not confirm this variation. Since the Solomon Ids. (Rennell-7, Guadalcanal-8, Celebes and other East Indian birds are Malaita-1, Tetipari-3, Rendova-4, New Georgia-1, Ysabel-5, Choiseul-5, Bougain- identical with Australian ones in color and ville 6), Bismarck Archipelago (Feni Is., east differ in size, if at all, very slightly indeed, of New Ireland-7, New Hanover-3, Manus, it seems best to consider percna a synonym Admiralty Group-5), Palau-3, New Guinea of (northern lowlands-12, Balim R., 1600 meters- rogersi. 3, Oranje Mts., 3225-3600 meters-21). (Figure This race seems to have extended its after localities indicates number of specimens range into New Guinea along the south- examined; where none is given locality was eastern coast in the Louisiades and in the taken from the literature.) Peninsula. A male from Misima WING.-Moorea: e 232, 236, 238, 242, 243, Vogelkop 246; 9 226, 226, 227, 228, 228, 229, 231. Aus- Island is of average size for rogersi, and the tral Ids.: e 233, 235, 236, 237, 240, 241, 245; same is probably true of the birds which 9 223, 224, 225, 225, 226, 227, 227, 227, 228, 4 AMERICAN MUSEUM NOVITATES [NO. 1237

233. Rarotonga: 6e 242. Tonga Ids.: e 240, from the Solomons is almost identical with 241, 242; 9 222, 223, 226. Upolu: e 236, 240?; 9 230. Fiji Ids.: e 231, 235, 239, 242, that of the central Polynesian series. 243, 244, 246, 246; 9 222, 227, 227, 230, 230, When better specimens are available, the 231, 233, 233, 235. New Caledonia: 9 240, 243. birds of northern New Guinea and the Loyalty Ids.: e 251?; 9 241. New Hebrides: Solomons may prove to be almost if not e 248. Banks Ids.: e 241, 243. Santa Cruz Ids.: e 243, 246, 253, 253; 9 230, 232, 237, quite as large as those of Fiji and other 239, 240, 242, 243. Solomon Ids.: ci 227, central and eastern Polynesian localities. 227+, 228, 230, 230+, 230?, 232?, 235, 236, 237, This leads to the important question of 238, 240; 9 219, 220?, 221, 222+, 223?, 226, the size of the birds of Palau, the type 226. Bismarck Arch.: d 239+, 241, 245; 9 220?, 226? Northern lowlands of New Guinea: locality of pelewensis. Unfortunately, of i 224, 224, 227, 227, 230?, 236. Balim R., N. the three skins secured there by the Whit- G.: dc 235, 236; 9 223+. Oranje Mts., ney Expedition two are only half-grown N. G.: e 238, 238, 242, 242, 243, 246, 246, 250, and the other, although a male with gonads 250, 250; 9 218?, 221, 226, 227, 228, 232, 236, 243. Saruwaged Mts., N. G.: 9 "237" (Mayr, indicated as large, has the primaries not 1931, Mitt. Zool. Mus. Berlin, XVII, p. 704). fully grown. Its wing measures 223 mm. CULMEN.-Eastern and central Polynesia: plus. The bill of this bird measures 44 16 e 42-50 (45.6). Southern Melanesia mm., well over the minimum found in (Santa Cruz to New Hebrides): 5 ce 46-51 48.4). Solomon Ids.: 19 ci 42.5-48 (45.4). males from the Solomons and Polynesia. On the basis of the wing lengths listed Palau birds are probably of the same size above, two major and two or three minor as those of the Solomons, or very close to size groups may be recognized whose it. To be sure, Finsch (1875, Jour. average measurements are as follows. Mus. Goddefroy, III, p. 40) has published For comparison rogersi has been included. measurements which suggest that they may New Caledonia has been included with be even smaller. Changed to millimeters, southern Melanesia. his measurements of the wing for seven

WING LENGTHS OF Anas superciliosa petewensi8 Palau 1 , 223+ Solomons, Bismarck and northern lowlands of New Guinea 18 0' 224-245 (232.6) 8 9 219-226 (222.6) Central and eastern Polynesia 27 e 231-246 (239.7) 30 9 222-235 (227.6) Oranje and Saruwaged Mts., New Guinea 10 c 238-250 (244.5) 8 9 221-243 (231.3) Southern Melanesia 8a 241-253 (247.3) 10 9 230-243 (238.7) rogersi 19 e 257-275 (263.3) 14 9 243-254 (249.8) The populations of eastern and central unsexed Palau skins are: 207,212, 212, 214, Polynesia, from the Society Islands to Fiji, 223, 225, 230. These measurements are composed of birds which are very uni- indicate that the Palau birds are rather form in size. It has been difficult to de- small, but it is not necessary to conclude cide whether the series from the Solomons that they are smaller than those of the and the northern lowlands of New Guinea Solomon Islands. As already noted, most actually are smaller or whether this im- authors have not been critical enough in pression is the result of the large number of eliminating non-geographical variation; molting and badly worn specimens from furthermore in other species where ma- these localities. The few specimens from terial has been available, Finsch's meas- the Bismarck Archipelago are as large as urements average considerably smaller ones from central Polynesia, and it is very than mine. doubtful if smaller birds would occur to the Specimens of Anas superciliosa from north and south of the Bismarcks in the southern Melanesia and New Caledonia Solomons and New Guinea. Furthermore, are distinctly larger than those already the average culmen length of the series discussed. Despite the considerable north- 1943] BIRDS OF THE WHITNEY SOUTH SEA EXPEDITION. 52 5 south range involved, this is a uniform size these populations merely reflect a general group from Tucopia in the Santa Cruz size variation, it is to be expected that Group to New Caledonia. In the Saru- weights will show an even greater relative waged, Oranje and perhaps other mountain difference. This is true of the few avail- ranges of New Guinea altitudinal in- able: crease in size occurs, but even from the highest altitude (Lake Habbema) speci- A. s. pelewensi8 mens do not average quite so large as those Solomons e 650,690,700, from southern Melanesia. Furthermore North coast 700 gms. occur at in New Guinea cP 700, 700, 700 9 655 gms. some small birds high altitudes Loyalty Ids. c? 835 9 751 New Guinea, and there is a gradual de- Saruwaged crease in size until one reaches the small Mts...... 9 800 birds of the north coast, as Rand has al- A. s. roger8i ready pointed out (1942, Bull. Amer. Mus. Arfak Mts. c 950 Nat. Hist., LXXIX, p. 429). Comparison of the above measurements makes it evi- If Palau birds do prove to average ap- dent that the populations of the Oranje preciably smaller than those from the and (probably) the Saruwaged Mountains Solomons and central Polynesia, they are distinctly closer to pelewensis than to probably will not overlap at all in wing rogersi. The southern Melanesian birds length with those of southern Melanesia, are almost intermediate, yet closer to and it might be justifiable to describe the pelewensis. For geographical reasons also latter as a new race, to which some of the it is more satisfactory to refer them to higher mountain populations of New pelewensis. Guinea would unfortunately have to be Since the differences in wing length of referred also.

Columbidae NOTES ON FOUR SPECIES OF PTILINOPUS Ptitinopus SWAINSON, 1825, Zool. Journ., I, forms were known. Although the females p. 473. Chrysoena BONAPARTE, 1854, Compt. Rend. of victor, luteovirens and layardi might ap- Acad. Sci., Paris, XXXIX, p. 879. pear to be only subspecifically distinct, Among the most beautiful native birds the males are very different from one an- of Fiji are three geographically representa- other and have specialized plumages. The tive species of golden fruit pigeons, victor, male of victor is reddish orange in colora- luteovirens and layardi (viridis auct.), tion, and its plumage is diffuse and hair- which have their centers of distribution on like; that of luteovirens is yellowish green, the three largest islands of the group, and the feathers are lanceolate and thick- , and Kandavu. ened and have bifid tips. The male of The females are plainly colored, greenish layardi is less specialized; it is greenish birds washed with yellowish white on the with a ring of thickened feathers with bifid under tail coverts and abdomen. This is tips around the breast and upper back. the commonest and apparently primitive In all three the head is greenish yellow. coloration of the female in the widespread These specializations of coloration and genus Ptilinopus. Even such specialized plumage in the males have prompted most species, as regards male coloration, as authors to place these species in a separate nanus, superbus and solomonensis have genus, Chrysoena. Wetmore (1925, Ibis, females very similar in coloration to those p. 853) went a step further and segregated of the Fijian species; indeed solomonensis victor in a subgenus, Chrysophaps. Peters might be considered conspecific with the (1934, Proc. 8th Int. Orn. Cong., p. 382) latter if only the females of the various accepted this subdivision and stated that 6 AMERICAN MUSEUM NOVITATES [No. 1237 he did not elevate Chrysophaps to generic same stock that gave rise to Ptilinopus level only because "the peculiar structure tannensis." After comparison of most of of the contour feathers is found only in the the species in the genus, I have reached a males." This is true of all three species, different conclusion. Although the luteo- however, and is an equally valid reason for virens superspecies and P. tannensis both not recognizing Chrysoena as distinct from have yellow heads, this is apparently Ptilinopus. Furthermore, the male of parallelism. Among .the important dif- layardi is also quite different, and to be ferences between the two are: (1) the consistent each of the three species should luteotvirens superspecies belongs to that be put in a subgenus, or none of them. In section of Ptilinopus in which bifid breast view of the great similarity of the females feathers occur; tannensis does not; (2) and the complementary ranges of these tannensis is obviously an offshoot of the doves, the second alternative seems prefer- perlatus-ornatus group of New Guinea,with able, and it is justifiable to consider them which it agrees in large size, presence of as comprising a superspecies. spots on the scapulars and in other char- As regards the question of maintaining acters lacking in the luteovirens group; Chrysoena distinct from Ptilinopus, it has (3) the luteovirens group has marked sexual already been pointed out that the females dimorphism and the males tend to be of some forms in the two groups scarcely orange or yellow; neither is true of tannen- appear to be specifically, much less generi- s5s. cally, distinct. The peculiar plumage of Evidently the luteovirens group was de- the males of Chrysoena represents only a rived from the central group of Ptilinopus, further development of two characters of which such species as coronulatus and common in various species of Ptilinopus, regina are examples. This group agrees namely, (1) a tendency toward diffuse and with the luteovirens group in the following hair-like plumage which is noticeable in characters, among others: small size, pres- all the species of eastern Polynesia, such ence of bifid breast feathers, bright colora- as huttoni; (2) the presence of feathers in tion of the males (and sometimes of both which the barbs are closely appressed, giv- sexes). Many of them have orange colora- ing a thickened appearance, and the tip is tion ventrally which is very reminiscent of forked. In other species, however, such P. victor. In several the crown patch is bifid feathers are found only on the breast, bordered by yellow, and in some (e.g., coro- and the thickened portion does not extend nulatus) the throat is yellow too. An ex- along most of the feather as in luteovirens. tension of this would produce the yellow- From these considerations it seems justi- headed Fijian species. All the eastern Poly- fiable to conclude that the peculiarities nesian species of the genus have been de- of plumage found in the males of these rived from this stock (1942, Ripley and Fijian species, especially since all three Birckhead, Amer. Mus. Novitates, No. differ from one another, are of specific but 1192). Perhaps the three species of the not generic importance. D. G. Elliot luteovirens group are the only descendants reached the same conclusion many years ofan earlier arrival of this typical Ptilinopus. ago (1878, Proc. Zool. Soc. London, p. stock in Polynesia. 511). If Chrysoena becomes a synonym of Ptilinopus, the species usually known as Ptilinopus layardi Elliot Chrysoena viridis Layard requires a new name. Elliot supplied the appropriate Chrysoena viridis LAYARD, 1875, Proc. Zool. Soc. London, p. 151, Kandavu, Fiji. alternative, layardi. Ptilopus layardi ELLIOT, 1878, Proc. Zool. INTRAGENERIC RELATIONSHIPS.-For Soc. London, p. 567, new name for Chrysoena convenience in the present discussion, the viridis Layard, preoccupied by Columba viridis under consideration are re- Linnaeus, 1766. three species TYPF LOCALITY.-Kandavu, Fiji. ferred to as the luteovirens superspecies. RANGE.-Kandavu Group (Kandavu and Peters (ibid., p. 383) has stated that this Ono), Fiji. group was "doubtless derived from the WING.-Kandavu: e116, 116, 118, 118, 118, 1943] BIRDS OF THE WHITNEY SOUTH SEA EXPEDITION. 52 7

122; 9 114, 115, 116, 121. Ono: e 116, 117, in the wing- but not enough to make the 117, 117, 117, 118, 120; 9 116. measurement valueless for comparison. Birds from the two islands appear identi- WING.-Vanua Levu: e 114, 117, 119, 120, cal. 120; juv. ce 120, 121; 9 122. Kio: e 120; 9 117. Rambi: e 115, 119; 9 120. Tavi- Ptilinopus luteovirens uni: e118, 119, 119,1>20,1>21, 122, 128, 123,1]25; Hombron and Jacqu:not1 9 118, 120, 120, 121, 122. TAIL.-(e only) Vanua Levu: 60, 61, 61, 61, TYPE LoCALITY.-Ovalau, Fiji. 62, 64. Kio: 60. Rambi: 61, 63. Taviuni: RANGE.-Viti Levu Group (Viti Levu, Ngau, 60, 61, 62, 63, 64, 65, 66. Ovalau, Mbenga, Waia), Fiji. Levu, Kio and Rambi: 11 WING.-Ovalau: e 118, 118, 119, 120, 120, CULMEN.-Vanua 121, 124; 9 116, 117, 117, 120, 120, 121. e 13-15 (14.05). Ngau: e 123. Viti Levu: e 119, 121, 123, 123; 9 117, 120, 120, 123, 123. Mbenga: Ptilinopus victor aureus, cl 117, 120. Waia: cl 125. new subspecies No geographical variation is apparent in TYPE.-No. 249,168, Amer. Mus. Nat. Hist.; the material examined. e ad.; Ngamea, Fiji Ids.; November 25, 1924; Whitney South Sea Expedition (J. G. Correia). Ptilinopus victor victor Gould SUBSPECIFIC CHARACTERS.-Sizelarger; colora- TYPE LOCALITY.-M'Bua, Vanua Levu Is., tion of the upper parts, wing and tail lighter Fiji. than in P. v. victor. SUBSPECIFIC CHARACTERS.-Size smaller; In adult males the upper parts are bright coloration of the upper parts and wing and orange, not appreciably darker than the under tail, especially in adult males, darker than in parts. The primaries are yellowish, only slightly the other (following) race. dusky, and have much broader yellowish orange In adult males of P. v. victor the bright orange edgings on the outer vanes than in victor. In of the under parts extends around as a collar aureus even the shafts of the outer primaries fringing the greenish head. The remainder of become light and yellowish near their tips. the upper parts, however, are noticeably darker, Females and juvenal males are a shade paler with a dull reddish cast. The primaries are below, especially as regards the green of the dusky gray, washed with yellowish, with a nar- breast, than the corresponding stages of the row yellowish orange edge on the outer vane, nominate form. and a wider concealed edge of the same color RANGE.-Ngamea, Fiji Ids. The specimens on the inner vane. The tail feathers have in the British Museum from the small island of large dusky or even blackish areas, which tend Lauthala (Lanthala, Laucala), a few miles east to form a poorly defined subterminal band. of Ngamea, probably belong to the race aureus Specimens from Taviuni tend to vary both in also. color and size in the direction of the following WING.-Ngamea: e 124, 127, 128, 128, 128, race, but are much nearer to P. v. victor. 130; juv. ce 126, 126, 131; 9 123, 125. RANGE.-Vanua Levu, Kio Rambi and Tavi- TAIL.-Ngamea: dI 64, 64, 65, 66, 67, 68. uni, northern Fiji Ids. CULMEN.-Ngamea: 8 cP 14-15.5 (14.88). Wetmore (1925, Ibis, p. 832) recorded a specimen of P. victor from "Lambasa, Viti Ptilinopus tannensis Latham Levu." Dr. Wetmore has been kind TYPE LoCALITY.-Tanna, New Hebrides. enough to look up this specimen and writes RANGE.-Specimens were examined from the me that Lambasa is on Vanua Levu. It following islands, which seem to include all those from which this species was previously known, was first said to be on Viti Levu, and as well as some others. New Hebrides: Tanna, through an oversight this error was not Erromanga, Efate, Mau, , Mai, Tongar- corrected. iki, , Epi, , Malekula, Espiritu The smaller size of the present race is Santo, Pentecost, Aurora; Banks Ids.: Gaua, . noticeable even from a gross comparison WING.-Tanna: e 153, 159; 9 161. Erro- of skins. Most of our series have the pri- manga: e 157; 9 154. Efate: &I 155, 155, 156, maries in molt. The italicized measure- 156, 157, 158, 158, 162, 162, 163; 9 149, 154, ments (both of this and the following sub- 154, 155, 157. Mau: e 154, 155. Nguna: 9 154. Mai: e 165. Epi: e 157. Tongoa: species) are of specimens with some molt e 155. Ambrym: 9 152. Aurora: ep158. 1 The parentheses have been omitted from the Pentecost: e 155. Gaua: cl 157. Vanua describers' names here and throughout this paper in Lava: cP 157, 158. agreement with W. H. Osgood's suggestion that this and practice has outworn its usefulness (1939, Science, TAIL.-Tanna Erromanga: e 83, 84, LXXXIX, p. 9). 89; 9 80, 90. Other New Hebrides Ids.: 23 cl 8 AMERICAN MUSEUM NOVITATES [No. 1237

82-91 (86.43); 12 9 79-86 (83.33). Banks Ids.: distinct barring, which is entirely lacking e 80, 82, 84. in adult males. PLUMAGES.-Only the males of this In juvenals of both sexes the yellow of species have white spots on the lesser wing the under parts is as extensive as in the coverts. In P. perlatus and other related adult female, or more so, and tends to be species similar spots or shoulder bars, when deeper and less whitish. Juvenal males present, occur in both sexes. Males of have fewer white spots on the wing coverts tannensis have the under tail coverts than adults, and the spots tend to be gray- mostly and to a variable extent the ish, b'ut they can be distinguished from yellow, juvenal females by this character. Juve- lower abdomen is yellowish or yellowish nals have the primaries narrowly tipped white. In females the yellowish extends and edged on the outer vane with yellow. much further forward over the entire Most of the back feathers and wing coverts posterior half of the under parts. On the are also tipped with yellow. upper abdomen the yellow is restricted to No geographical variation either in size the tips of the feathers and produces an in- or color was found in Ptilinopus tannensis.

A REVISION OF DUCULA PACIFICA This pigeon occurs on a great many the bill, which is absent or very small in islands scattered over a very large area in immature birds. The latter are also the south Pacific. Inevitably specimens smaller; they have the gray of the hind from several localities were described as neck darker, and the plumage of the under "new species" by early authors who lacked parts is tinged with brownish and dusky comparative material. Later it was real- as compared with the clear bright vina- ized that all belonged to the same species, ceous of adults. Males are somewhat larger Ducula pacifica Gmelin, described from the than females but are otherwise the same Tonga Islands, but museum material was externally. Measurements of this species still too insufficient and scattered to deter- are summarized in the table. Undoubt- mine how many distinct subspecies exist. edly a few mis-sexed specimens have The present study is based on a splendid affected some of the means given. The series of 315 skins from 71 different islands. bill was measured from the nostril to the It includes specimens from the type locali- tip. The number of specimens measured ties of all the names which have been ap- is indicated in parentheses. plied to this species except Sikaiana The following maximum and minimum Island. Skins from other islands in the measurements of the various populations Solomons make this single deficiency less should be considered in connection with important. Careful study of this long the table: series of specimens shows the nominate WING.-Solomon Ids.: e 239-255; 9 226- race to have a very extensive range. All 237. Santa Cruz Ids.: e 241, 243, 252; 9 other names that have been applied to this 231-242. Duff Ids.: e 245; 9 242. Reef are of Ids.: c' 241, 243. Banks Ids.: e 241, 245; 9 species best considered synonyms 233, 233, 238. New Hebrides: e 240-250; 9 D. p. pacifica. Nevertheless the isolated 229?-242. Loyalty Ids.: c246. Fiji Ids.: e population found in the Bismarck Archi- 241-260; 9 234-256. Tonga Ids.: ce 247?; pelago and the coastal islands of north 9 232, 242. Alofa Is.: e 250. Boscawen Is.: 9 236, 237, 242. Danger Ids.: e 243, 245, central New Guinea is distinctly different 251; 9 237. Samoa Ids.: c 240-256; 9 232- in size and to a lesser extent in color, and 243. Cook Ids.: e 245?, 253; 9 233?, 241, is here described as new. The evidence 244. prompting this rather drastic treatment TAIL.-Solomon Ids.: e 137-159; 9 132- 150. Santa Cruz Ids.: e 139-157; 9 135- is summarized below. 153; Duff Ids.: e 145; 9 148. Reef Ids.: e This species, like several other species of 143, 150; 9 139, 150. Banks Ids.: e 140-149; Ducula, has a horny knob at the base of $9 140-145. New Hebrides: e 140-155; 9 1943] BIRDS OF THE WHITNEY SOUTH SEA EXPEDITION. 52

135-150. Loyalty Ids.: e 137, 145, 153. Fiji the same color distinctions in his series Ids.: e 146-161; 9 144-160. TongaIds.: e from Tarawai Island that were found in 145, 145; 9 146, 146. A]ofa Is.: e 146. Bos- Cawen Is.: 9 150, 152, 155. Danger Ids.: e the Bismarck Archipelago specimens. Be- 149, 150, 158; 9 144. Union Ids.: 9 135. tween the range of D. p. sejuncta and the Samoa Ids.: e 148-160; 9 145-152. Cook westernmost colonies of D. p. pacifica in Ids.: c' 153, 158; 9 155, 157. the Solomon Islands and Louisiade Archi- It will be seen from the table that the pelago (subsp.?) there is a gap of several four birds from the Bismarck Archipelago hundred miles. This species would have and the seven from Tarawai (D'Urville) been collected before now if it occurred on Island, New Guinea, are distinctly smaller the intervening islands such as New than the others. The measurements of Britain. It is not surprising that this the Tarawai specimens are quoted from isolated population has become subspeci-

MEANS OF MEASUREMENTS OF DUCULA PACIFICA (SEE TEXT) LOCALITY WING TAIL BILL Q 9 a" 9 a" 9 E. Bismarck Arch. 229.5 (2) 220.5 (2) 133.0 (2) 126.0 (2) 17.0 (2) 16.3 (2) Tarawai Is., N. G. e 9 "230-235" (7) e 9 "130-135" (7) Solomon Ids. 243.6 (9) 232.0 (10) 146.6 (14) 140.5 (10) 18.5 (13) 17.9 (10) Santa Cruz, Duff, Reef Ids. 244.2 (6) 238.0 (6) 147.4 (7) 144.3 (9) 18.0 (4) 17.8 (6) Banks, New Hebrides, Loyalty Ids. 245.9 (8) 234.-8 (9) 145.8 (15) 142.9 (9) 18.8 (4) 18.0 (1) Fiji Ids. 250.4 (19) 244.5 (19) 150.0 (15) 149.4 (17) 18.8 (5) 18.8 (4) Tonga Ids. 247.0 (1) 237.0 (2) 145.0 (2) 146.0 (2) 18.3 (2) 19.2 (3) Samoa Ids. 251.5 (10) 237.6 (12) 153.5 (10) 147.9 (8) 18.1 (8) 17.8 (4) Other Central Polynesian Ids. 247.3 (4) 238.0 (4) 150.8 (4) 147.2 (5) 19.2 (3) 17.7 (3) Cook Ids. 249.0 (2) 239.3 (3) 155.5 (2) 156.0 (2) 17.5 (3) 17.2 (3) Salvadori (1880-2, Ornit. Pap. Mol., III, fically distinct, a process which may have p. 73). He remarked that the birds from been accelerated by the acquisition of more this island are uniformly smaller than sedentary habits. In sejuncta the wing is Polynesian ones. Since the Tarawai-Bis- somewhat rounded, with the third pri- marck population differs also in color, it is mary longest. In the larger race, pacifica, described as new. which is known to be of vagrant, roving habits, the wing is more pointed, with the Ducula pacifica sejuncta, second primary longest. new subspecies The restricted and irregular distribution TYPE.-NO. 336,321, Amer. Mus. Nat. Hist.; of this species in the Papuan part of its e ad.; Loof Is., Hermit Group, Bismarck range may be the result of competition Archipelago; May 24, 1934; Whitney South of Sea Expedition (W. F. Coultas). with closely related species similar SUBSPECIFIC CHARACTERS.-Smaller than ecological requirements, such as D. pistri- typical pacifica; top of head and hind neck pale naria and perhaps D. rubricera. Indeed, gray, lighter than in pacifica; under parts slightly D. pacifica may have extended its range paler than in that race. RANGE.-Recorded from the Ninigo Group into the Papuan region secondarily from (Pihun Is.) and Hermit Group (Loof Is.) in the Polynesia However, the presence of western Bismarck Archipelago, and from two closely related forms to the west (D. islands (Seleo and Tarawai) off the north central coast of New Guinea. myristicivora and D. concinna) as well as to WING.-Loof: e 229 (type); 9 218, 223. the east (D. p. pacifica, D. oceanica, etc.) Pihun: 230? would suggest that the range of this group TAIL.-Loof: ce 131 (type); 125, 127. once continuous on suitable Pihun: 135. of species was BILL.-c 17, 17; 9 16, 16.5. islands from the Moluccas to eastern REMAtRKS.-Salvadori (loc. cit.) noted Polynesia. 10 AMERICAN MUSEUM NOVITATES [No. 1237

Ducula pacifica pacifica Gmelin edly not complete. Mr. L. Macmillan, Columba pacifica GMELIN, 1789, Syst. Nat., I, while collecting for the American Museum, pt. 2, p. 777, Insulis amicis (= Tonga Ids.). saw one individual of this species on Globicera tarrali BONAPARTE, 1854, Compt. Uvea Island, Loyalties, following a hurri- Rend. Acad. Sci., Paris, XXXIX, p. 1073, Vani- koro, Santa Cruz Ids. cane, but states that native persecution, Globicera sundevalli BONAPARTE, 1854, Consp. if not other reasons, prevents this species Av., III, p. 32, "ab Arnoux. ex Ins. Tonga-tabou from becoming established there. At et Wallis." Peters (1937, Check-List, III, p. least at the present time, Mr. Macmillan 44) gives the Loyalty Ids. as the type locality of considers Ducula pacifica as accidental on this name, for reasons unknown to me. Speci- also. mens from the Loyalty Ids. and from Tonga New Caledonia appear identical. REMARKS.-Within the wide range of Globicera microcera BONAPARTE, 1855, Compt. D. p. pacifica only slight geographical vari- Rend. Acad. Sci., Paris, XL, p. 215, "ab Astro- ation exists. Even extremes from other labe ex Ins. Vavao, a Zelee, ex Samoa." I am not localities cannot be distinguished with aware on what grounds Peters (loc. cit.) restricts from this name to Samoa, instead of Vavao, Tonga. certainty topotypical Tongan speci- In any event, the birds of Tonga and of Samoa mens. The trends of variation, insofar as appear to belong to the same race. can be judged from the various samples Carpophaga frauenfeldi PELZELN, 1865, Reise available are as follows. Size: central Novara Voeg., p. 106, Stewart (Sikaiana), Solo- Polynesian birds are large. Those from mon Ids. the Cook Islands are fully as large, but Globicera farquhari SHARPE, 1900, Ibis, 249, their bills tend to be slightly shorter. The Erromanga, New Hebrides. of Ducula pacifica intensitincta "Neumann" birds the Solomon Islands average STRESEMANN, 1923, Arch. f. Naturg., VIII, p. slightly smaller in all dimensions, as shown 76, Fiji Ids. in the table. Those from southern Mela- SUESPECIFIC CHARACTERS.-Larger than se- nesia are intermediate in size, as well as in juncta. Wing in males averaging about 248 geographical position, between the Solomon mm., in females about 238 mm.; head and hind Islands and central Polynesian groups. neck usually darker gray; coloration of under parts perhaps slightly deeper than in 8ejuncta. Perceptible geographical variation in RANGE.-Cook Ids. (Rarotonga); Samoa color seems restricted in this race to the (Savaii, Upolu, Tutuila, Olosinga, Ofu, Tau); gray of the hind neck and top of the head. Tonga Ids. (Tongatabu,* Ata, Tofua, Late); Some authors have referred to color vari- Union Ids. (Atafu,* Fakaafo); Ellice Ids.* ation in the under wing coverts. I have (Funafuti); Fiji Ids. (Turtle, Ongea Levu, Ful- anga, Yangasa cluster, Marambo, Kambara, been unable to detect any in the occasional Wangava, Tavunasithi, Namuka Ilau, Olorua, specimens in which these feathers are not Oneata, Aiwa, Naiau, Tavutha, Thikombia, grease-stained. The specimens examined Avea, Kimbombo,, Naitamba, Vatu Vara, from the Cook Islands were collected al- Yathata, Vatauua, Ngele Levu, Nakumbasanga, Nukumbalate, Namena, Sovu Rocks, Ovalau,* most fifty years ago. They are very Wakaia,* Mokongai,* Viti Levu,* Kandavu,* greasy but appear to average slightly Kanathia*); other central Polynesian Ids. (Ni- darker on the head and neck than Tonga uafoo,* Danger, Boscawen, Alofa, Wallis, Niue); birds. Samoan specimens tend to have Santa Cruz Ids. (Duff Group: Disappointment, Treasurers; ReefGroup: Nupani, Lomlom; Santa the gray of the hind neck a shade paler. Cruz Group: Santa Cruz, Anuda, UJtupua, Vani- Those from the various islands in southern koro, Tinakula, Tucopia); Banks Ids. (Valua, Melanesia agree closely with topotypes. Vanua Lava, Bligh, Gaua); New Hebrides (Es- Sharpe's type of farquhari from Erro- piritu Santo, Pentecost, Epi,* Malo,* Malekula, Mai, Efate, Erromanga, Aniwa, Tanna, Anei- manga must be abnormally dark, unless teum*); Loyalty Ids. (Lifu, Uvea*); New Cale- it is immature or grease-stained. In the donia; Solomon Ids. (Sikaiana,* Rennell, Solomon Islands a tendency toward paler Gower, Buena Vista, Ontong Java, Ramos); coloration can be detected, but the great Louisiade Archipelago* (Duchateau, Teste, majority of specimens cannot be separated. Suau). Mayr (1931, Amer. Mus. Novitates, No. This species wanders to so many small 486, p. 11) reached the same conclusion, islets that even the above list is undoubt- using specimens from the Santa Cruz * No specimens were examined from these localities. Islands -for -comparison. Hence Solomon 1943] BIRDS OF THE WHITNEY SOUTH SEA EXPEDITION. 52 11

Islands birds vary slightly towards se- CLOSER TO Topo- CLOSER juncta in both size and color. On geo- TYPICAL TO graphical ground the records from the LOCALITY pacifica sejuncta Louisiade Archipelago probably belong to Solomons 26 (a few pacifica also, but this population may be doubtful) 9 intermediate if it continues the trends Southern Melanesia 64 4 Fiji 86 3 noted in the Solomon Islands. The follow- Samoa and nearby ing tabulation is based on variation in the islands 71 9 shade of gray of the head and hind neck: Cook Ids. 4 0

NOTES ON THE RACES OF DUCULA OCEANICA This Micronesian pigeon has a more BILL.-c 16, 16.5, 16.5, 16.5; 9 16, 16.5, rounded wing than D. pacifica, with the 16.5, 17, 17. third rather than the second primary Ducula oceanica ?townsendi Wetmore and is more seden- longest, undoubtedly SUBSPECIFIC CHARACTES.--Gray of the head tary. Although its range is much smaller and hind neck, and to a lesser degree of the than that of pacifica, subspeciation has breast, much darker than in the two preceding proceded further. races; bill slightly longer. Differs from the following race, oceanica, only by being larger, but ?monacha adequate material may reveal that the differ- Ducula oceanica ence is too slight to make it profitable to recog- Momiyama nize townsendi. In describing townsendi Wet- SUBSPECIFIC CHARACTERS.-Doubtfully dis- more mentioned only two adults of the new form tinct from the following race. Gray of the head, and one from Kusaie. He described townsendi neck and breast perhaps a shade paler; back as similar to oceanica from Kusaie but " . . . less bluish, and more noticeably tinted with darker on upper breast, foreneek, hindneek, and olive and green than in teraokai. upper back; no whitish line indicated on lower TYPE LOCALITY.-Yap, Caroline Ids. eyelid beneath eye; and under tail coverts RANGE.-Palau Ids. and Yap, Caroline Ids. paler." Careful comparison of a long series Kuroda described a race from Palau which was taken at about the same time from the two local- later synonymized with monacha by other Japa- ities has not revealed any differences in color. nese authors. I have seen no birds from Yap Those noted by Wetmore apparently represented Is., but since those of Palau are only doubtfully individuzl variation. distinct from specimens from Truk, east of Yap, TYPE LOCALITY AND RANGE.-Ponape Is., the Palau race very probably cannot be upheld, Carolines. as variable characters in this species change more WING.-c 236; 9 228,231,233. or less gradually from east to west. TAIL.-c 157, 158, 159, 160, 163; 9 157. WING-.c 225, 230, 232; 9 229. BILL.-c 17, 17, 17.5, 18; 9 16.5, 17.5, 18. TAIL.-c 158, 160; 9 157. BILL.- e17, 17.5; 9 15.5. Ducula oceanica oceanica All measurements are from Palau specimens. Lesson and Garnot SUBSPECIFIC CHARACTERS.-Like the pre- Ducula oceanica teraokai Momiyama ceding race, townsendi, but slightly smaller. SUBSPECIFIC CHARACTERS.-Possible differ- TYPE LOCALITY.-Kusaie, Caroline Ids. ences between this and the preceding race are RANGE.-Kusaie, Caroline Ids. According to given above. Our series of teraokai differs from Japanese authors, this is the form occurring on all other material of this species examined in Jaluit and Elmore Ids. in the Marshall Group. having the back purer blue, without tints of olive Probably specimens from these islands will be and green. This series was collected in the intermediate between oceanica and the following 1890's, while all the other material was collected race, in agreement with the cline toward diminu- after 1930. If this distinction in the color of the tion of size from west to east which is noticeable back proves to be the result of age, monacha may in this pigeon. Ducula oceanica was secured be united with teraokai, as the other differences at an early date on the Gilbert Ids., but nothing are negligible. is known of the racial affinities of that popula- TYPE LOCALITY AND RANGE.-Islands of the tion. It is best referred tentatively to D. o. Truk Atoll, Caroline Ids. oceanica until a comparison becomes possible. WING.-c? 227, 227, 234, 236, 237; 9 222, WING.-c 226,229,232,232; 9 220,220,223? 228,232,233,235,237,241,242. TAIL.-c? 154, 154, 157, 158, 159; 9 148, 148, TAIL.- 156, 159, 160; 9 148,150, 156, 156. 149, 149, 151, 154. 12 AMERICAN MUSEUM NOVITATES [NO. 1237

BILL.-d 17.5, 18, 18, 18, 18.5, 19; 9 17.5, tinguishable from Kusaie specimens in 17.5, 18.5, 19. coloration. The primaries of Ducula oceanica become Ducula oceanica ratakensis greatly abraded. This is very noticeable Takatsukasa and Yamashina in our series of typical oceanica from Kusaie SUBSPECIFIC CHARACTERS.-Like oceanica, but and, together with the poor condition of the smaller. TYPE LOCALITY.-Arno, Marshall Ids. two specimens of ratakensis examined, RANGE.-Arno and Wotje, Marshall Ids. makes it impossible to reach a final conclu- MEASUREMENTS.-Arno: e wing: 203? (juv. sion as to whether it is necessary to recog- quills); tail: 138 (ad. rectrices?); bill: 17.5. nize three size races, townsendi, oceanica Wotje: 9 wing: 218 (much worn, ad. ?); tail: 137; bill: 18.5. From original description and ratakensis. (1932, Dobuts. Zasshi, XLIV, p. 221): "Wing REMARKS.-The four species, Ducula 6 e, ad., 211-217; 3 9, ad., 208-213." pacifica, oceanica, aurorae (Society Islands) I am greatly indebted to Mr. J. L. and galeata (Marquesas Islands) form a Peters for the loan of two specimens of this superspecies (together, perhaps, with other race from the collection of the Museum of Papuan and East Indian species). Mayr Comparative Zoology. As will be noted (1940, Amer. Nat., LXXIV, p. 270) has from the qualifications above, they are not mapped the distribution of these four in very good condition for measuring. species. D. galeata, although specialized However, since the measurements of the as regards gigantism and hypertrophy of original describers also indicate a small the bill excrescence, is in coloration closer bird, ratakensis must be maintained. It to the more primitive pacifica-oceanica is possible, however, that further material group than is aurorae. The latter differs may show this impression of small size to from the other three species by lacking have resulted from the measurement of the chestnut under tail coverts, by having birds in badly worn plumage or subadult. a distinct immature plumage and by the The original describers also attributed deep blue rather than greenish blue color certain peculiarities of color to ratakensis. of the upper parts. It is therefore better Since these seemed at variance with the to consider galeata as independently de- 'type of geographical variation occurring in rived from the oceanica and pacifica groups this species, it was not surprising to find rather than directly from aurorae, as was that the specimens examined are indis- done in the map referred to.

NOTES ON DUCULA GOLIATH, D. BAKERI AND D. LATRANS Ducula goliath G. R. Gray that the female in goliath may actually be RANGE.-New Caledonia and the Isle of Pines. larger. WING.-d 292, 295, 300; 9 294, 295, 301, All the examined specimens are from New 301, 307. Caledonia. TAIL.- e 210, 214, 218; 9 206, 210, 212, 220,236. of the four birds Ducula bakeri Kinnear WEIGHT.-The weights TYPE LOCALITY.- Espiritu Santo, New whose wing lengths are italicized, given in Hebrides. the same order, were: e 697, 680.6; 9 RANGE.-Larger islands of the northern New 711.2, 716 gms. Mr. L. Macmillan, who Hebrides (Espiritu Santo, Pentecost, Ambrym, collected these specimens, states that the Aurora) and Banks Ids. (Vanua Lava, Gaua, Bligh). It will not be surprising if this species is females were in laying condition; the eventually found to occur on Malekula, New heavier one had a shelled egg in the ovi- Hebrides, also. duct. Since the weight of female birds is WING.-New Hebrides: Espiritu Santo: cl increase 224, 225?, "230" (type, fide Kinnear), 235; 9 known to considerably during 224? Ambrym: 9 218. Aurora: cl 221. Banks the laying season, perhaps the female is Ids.: Vanua Lava: ci 218, 222, 222, 223, 223, normally no heavier than the male, al- 224, 224, 224; 9 216, 218. Bligh: e 224. though the other measurements suggest Gaua: e 219. 1943] BIRDS OF THE WHITNEY SOUTH SEA EXPEDITION. 52 13

TAIL.-New Hebrides: Espiritu Santo: Ce The small samples available from each "162" (type), 175, 175; 9 167. Ambrym: 9 169. Aurora: diP 159, 162. Pentecost: 9 162. island tend to give to individual variation Banks Ids.: Vanua Lava: c? 159, 161, 162, 166, the appearance of geographical variation. 167, 167, 170; 9 151, 153. Bligh: aI 158. The sexing is from the labels and may in- Gaua: c? 164. clude a few errors. It will be noted that REMARKS.-The above measurements one male from Vanua Mbalavu is very make it quite certain that large; this specimen is a giant, visibly specimens average slightly smaller than larger than all other specimens. Although those of the New Hebrides. Perhaps this the longest primaries are in molt, its wing difference, which is paralleled in many length is considerably greater than in other species, is correlated with the smaller other skins. One of the females from this size and lower elevation of the islands in island, if correctly sexed, is unusually the Banks Group. I can detect no other large, but the other pair is of normal size. difference between the two populations. REMARKS.-Ducula latrans varies con- Juvenals with down still clinging to siderably in the shade of gray of the head their feathers scarcely differ in coloration and hind neck. In lighter birds this area from adults. They are duller, however, is set off from the back; in darker ones the with the gray of the head washed with two blend. No geographical variation in blackish and reddish brown, and with the color could be detected. maroon areas, especially the band across RELATIONSHIPS OF THE SPECIES.-D. the shoulders, duller. Juvenals also lack goliath has been thought to differ from all the bluish gray bloom on the wings, and its congeners by having forked tips on the their rectrices are narrower and become feathers of the breast and upper back. somewhat pointed with wear. Most specimens of D. bakeri, however, have a few such feathers, but I have been Ducula latrans Peale unable to find any in brenchleyi or latrans. RANGE.-Fiji Ids. Specimens were ex- amined from the following islands: Matuku, This, as well as general coloration, shows , Moala, Naiau, Thithia, Tavutha, bakeri to be most closely related to goliath Mango, Avea, Vanua Mbalavu, Naitamba, and not to latrans as stated by Kinnear in Vatu Vara, Thikombia, Tavinui, Rambi, Vanua Levu, Koro, Ngau, Viti Levu, Ovalau the original description. These four spe- and Kandavu. The species has been recorded cies are in some respects rather unspecial- also from Kanathia, Wakaia and Mokongai. ized, and it is difficult to say to which of WING.-Matuku: e 243; 9 227. Moala: the found on the islands to e 223, 230, 233, 235, 237; 9 225, 227, 229. species larger Naiau: e 232. Thithia: 9 229? Tavutha: the west they are most closely related. e 241, 245; 9 236. Mango: e 240?, 245. They have a resemblance to the unspecial- Avea: o' 243. Vanua Mbalavu: e 252+ ized group of species which includes (!), 236+; 9 231, 239. Naitamba: cl 242. Thikombia: ci 240, 241. Taviuni: 9 238. rosacea, pickeringii and perhaps cineracea, Vanua Levu: 9 232. Ovalau: 9 237. Viti but whether this indicates close relation- Levu: e 230,235?, 235?; 9230,231,232,233+. ship or chance retention of more or less Kandavu: e 242. TAIL.-14 e 167-186 (176.7); 7 9 163-178 primitive characters in two long separated (169.4). groups is uncertain.

NOTES ON COLUMBA VITIENSIS This pigeon is one of a group of closely nesia) and norfolciensis (eastern Australia). related species, most of which still replace Stresemann (1939, Jour. f. Orn., each other geographically as follows: LXXXVII, p. 351) has published a map janthina (Japan and nearby islands), showing the distribution of some of these versicolor (Bonin Islands), jouyi (Riu Kiu species. C. pallidiceps, vitiensis and norfol- Islands), pallidiceps (Bismarck Archipelago ciensis probably evolved in New Britain, and Solomon Islands), vitiensis (Moluccas, New Guinea and Australia, respectively. New Guinea, Solomon Islands and Poly- Later, pallidiceps and vitiensis spread to 14 AMERICAN MUSEUM NOVITATES [No. 1237 the Solomon Islands, but in neither instance to halmaheira but duller, and as a result the dark gray grqund color of the under parts is much less has this population become subspecifically concealed by iridescent green and amethyst distinct. C. vitiensis is unknown from tints, especially in the female; size medium, New Britain, although that island is, of wing averaging about 7 mm. shorter than in course, closer to New Guinea than are halmaheira. the Solomons. The interesting probability RANGE.-New Hebrides (Aneiteum, Tanna, Aniwa, Erromanga, Efate, Nguna, Makura, thus exists that vitiensis has been unable Mai, Epi, , Pauuma, Ambrym, Male- to colonize New Britain, the home island kula, Malo, Espiritu Santo) and Torres Ids. (Lo, of pallidiceps, though both species have Hiw). It is surprising to find this species in the secondarily colonized the Solomons, where Torres Ids. but not in the Banks Ids.; yet the Whitney Expedition collected extensively in the they occur side by side. latter group without securing it. Although vitiensis appears to have been WING.-New Hebrides: Tanna: e 249. once one in a chain of subspecies which Aniwa: e 237. Efate: e 229. Makura: 9 eventually extended from Japan to Aus- 223+. Mai: 9 214. Malekula: e 230; 9 tralia, it has been distinct long enough to 224, 225? Malo: 9 228?; Espiritu Santo: e divide secondarily into eight subspecies. 224, 227, 232, 233. Torres Id8.: Hiw: e 242. TAIL.-New Hebrides: Tanna: eP 171. Efate: Two of these, metallica of the Lesser Sunda e 157. Makura: 9 162. Mai: 9 150. Male- Islands and griseogularis of the Philippines, kula: e 160; 9 155. Malo: 9 152. Espiritu both characterized, among other things, by Santo: c? 151, 159, 163, 164; 9 164. Torres having the throat gray in both sexes, do not Ids.: Hiw: e 169. come within the scope of the present notes. NOTES ON PLUMAGE.-Females varv in The widespread race halmaheira occurs in the amount of gray present on the throat the Moluccas, New Guinea and the Solo- and cheeks, but I believe it is never entirely mons. Since the Polynesian races are absent. This is difficult to determine from closely related to halmaheira, it is desirable the specimens at hand, most of which have to state the racial characters of this race. the head feathers soiled and saturated with They are: throat and cheeks white in both plaster. The pigmented malar spot and sexes; scapulars and wing coverts sooty line beneath the eye are reduced or absent black with conspicuous green and ame- in a few individuals. It is interesting to thyst margins; plumage highly iridescent; note that such a spot, or a suggestion of size large. Of the four Polynesian races, one, occurs as an individual variation in that of the New Hebrides is most like about ten per cent of skins of halmaheira. halmaheira, while the Fijian form is also Females in fresh plumage are dark slate similar but continues a trend towards duller gray below; when viewed in certain lights, coloration and smaller size. The other tints of green and amethyst are conspicu- two races, one native to New Caledonia ous. This is the plumage described by and one to Samoa, are much more distinct Salvadori (1893, Cat. Birds, XXI, p. 318) and do not continue the cline just de- from a skin from Aneiteum. In worn scribed. Presumably their greater dis- plumage the breast feathers in the female tinctness is to be attributed to the genetic become noticeably reddish, and the gray effects of more complete isolation upon one feathers of the abdomen acquire dull chest- or a few islands, rather than to longer iso- nut tips. The coloration of males is similar lation. Columba vitiensis probably reached but brighter and more iridescent, with the the New Hebrides from New Guinea and breast always noticeably reddish; the spread to the other Polynesian localities changes with wear are as in the female. from there. Males of leopoldi are thus more like halma- heira in coloration than are females but Columba vitiensis leopoldi Tristram average duller and grayer than that race. SUBSPECIFIC CHARACTERS.-Throat and Juvenals of the New Hebrides race are like cheeks white in the male, washed with gray in dull-colored females. Their color changes the female; dark coloration of crown extending greatly as the result of fading, bleaching below the lores as a malar spot and narrow line below the eyes; scapulars and wing coverts dull and wear and becomes very dull and brown- black with inconspicuous margins; color similar ish, tinged with rufous ventrally. 19431 BIRDS OF THE WHITNEY SOUTH SEA EXPEDITION. 52 15

Columba vitiensis vitiensis constant in this race and noticeable even Quoy and Gaimard in juvenals. SUBSPECIFIC CHARACTERS.-Throat and cheeks white in males, washed with grayish in Columba vitiensis castaneiceps Peale females; upper parts similar to leopoldi, but lighter and with a grayish bloom which reduces SUBSPECIFIC CHARACTERS.-Throat and the iridescence and makes the coverts less black- cheeks white in the male, tending to be gray in ish; under parts prevailingly vinaceous or dull the female; plumage slaty gray with a con- chestnut (worn plumage); gray of under parts, spicuous greenish and slight reddish iridescence; as compared with leopoldi, paler and much re- crown purple-chestnut contrasting with neck duced, being in evidence only on the flanks and and back; a few feathers of the same color usuaUy under tail coverts; size small. extend around the margin of the white throat RANGE.-Fiji Ids. Specimens were examined patch. from Ongea Levu, Fulanga, Marambo, Kam- RANGE.-Western Samoan Ids. (Savaii, bara, Namuka Ilau, Mothe, Aiwa, Totoya, Upolu, Manono, Apolima). Moala, Vanua Vatu, Thithia, Tuvatha, Kata- WING.-Savaii; e 237; 9 220?, 222, 225, vanga, Mango, Munia, Thikombia Ilau, Avea, 226, 228, 230. UIpolu: e 228, 231?, 232, 233; Vanua Mbalavu, Kimbombo, Naitamba, Vatu 9 222. Vara, Yathata, Taviuni, Kio, Namena, Makon- TAIL.-Savaii: ci 162; 9 148, 155, 156, 157, gai, Wakaya, Mbatiki, Nairai, Yanutha, 160, 165. Upolu: e 155, 156, 159, 161; 9 Ovalau, Viti Levu, Vatu Leile, Ngualito, Waia, 147,153. Mathatoni, Nathoulla, Yasawa, Kandavu, NOTES ON PLUMAGE.-In cistaneiceps Ono, Yankuve and Vanua Kula. Others have the reddish coloration and recorded the species from Vanua Levu and iridescence Ngau. noticeable in the other Polynesian races of MEASUREMENTS.-Single specimens or small this species are almost entirely lacking, series from each of no less than forty-two islands except on the crown. Even in worn plu- as just listed were available. There is no indi- mage the feathers of the under parts do not cation of geographical variation within this uni- form race. The measurements for C. v. vitiensis acquire rufous tips, but remain gray. It is have therefore been combified. That this pro- interesting to find that in the only juvenal cedure is justified is further indicated by the in the series the ventral feathers are tipped low values obtained for the Standard Devia- with dull rufous and the specimen is tion (ar) and Coefficient of Variation (V). If populations differing appreciably in average scarcely separable from immatures of wing and tail lengths had been mixed, these vitiensis. values would very likely have been larger. In females the white throat patch tends WING.-27 e 214-239 (226.15); o- = 5.20; to be smaller, and there is not such a dis- V = 2.3. 26 9 208-226 (217.5); o- = 4.81; V= 2.21. tinct break where it meets the gray of the TAIL.-27 e 152-172 (160.59); a = 5.19; breast as in males. Some of the females V = 3.23. 28 9 146-163 (155.11); a = 4.79; have the throat and cheeks washed with V = 3.09. grayish, but as in leopoldi the condition NOTES ON PLUMAGE.-AS Salvadori of the specimens hinders a study of this noted (op. cit., p. 317), females of vitiensis variation. In this race the malar spot are duller and have the under parts less and line are usually abs2nt; only one of vinaceous than males. As in leopoldi, a fifteen skins has them well developed. pigmented malar spot and line beneath the eye are almost always present; in Columba vitiensis hypoenochroa Gould females they blend with the gray of the SUBSPECIFIC CHARACTERS.-(Male only.) cheeks. Females also have the scapulars Throat and cheeks white; crown, hind neck and wing coverts slightly more brownish and under parts except under tail coverts purple- chestnut, glossed with amethyst; upper parts, and less grayish. In juvenals the breast except crown and hind neck, like leopoldi; size feathers are gray, tipped with rufous to large. produce a barred effect; this appearance RANGE-.New Caledonia, Isle of Pines, is not present in the fluffy feathers of the Loyalty Ids. (Mare, Lifu, Uvea). WING.-New Caledonia: e 243, 247. Mare: abdomen, which are more broadly tipped e 248. Lifu: e 240; 9 235. Uvea: e 241. with rufous. In juvenals in worn plumage TAIL.-cI 167, 170, 170, 176, 177; 9 170. the under parts become dull rufous washed WEIGHT.-Loyalty Ids.: e 373, 426, 441,459; with grayish on the flanks and breast. 9 456. The sexual difference in throat color is REMARKS-We have only one female 16 AMERICAN MUSEUM NOVITATES [No. 1237

of this race. This specimen is bleached the description is based upon a painting and worn but agrees quite well with Salva- by Raper. Peters (1937, Check-List, III, dori's description. Females of hypoeno- p. 70) suggested that the locality attributed chroa are evidently much like those of to the bird portrayed by Raper was prob- leopoldi, but adequate material of the ably erroneous and considered the above former will probably show that some dif- name a synonym of halmaheira. Hind- *ferences in addition to size exist. Sexual wood, however, in his recent paper on the color dimorphism is much greater in hyp- birds of Lord Howe Island (1940, Emu, oenochroa than in other races of vitiensis. XL, p. 10, footnote) has shown that Raper Males of this race usually have a pig- visited Norfolk Island several times and mented malar spot and line beneath the probably stopped at Lord Howe Island or eye. Mayr (1940, Amer. Mus. Novi- obtained birds from there. He also quotes tates, No. 1057, p. 2) has already pointed many early accounts which leave no doubt out that the birds of the Loyalty Islands that a large pigeon, which became extinct and New Caledonia are identical and that at an early date, inhabited this island. It 'Sarasin's race uveaensis cannot be upheld. thus appears probable, though by no It may be mentioned here that it now means certain, that Raper's painting por- seems probable that Lord Howe Island trays this extinct bird. The painting, was formerly inhabited by a race of this which Hindwood has had reproduced, ob- pigeon, now extinct, which Mathews has viously represents some form of Columba named. vitiensis which might be halmaheira or the male of leopoldi. Lord Howe Island is so Columba vitiensis godmanae Mathews isolated, however, that if this species oc- Raperia godmanae MATHEWS, 1915, Austr. curred there, it was probably racially dis- Av. Rec., III, p. 24, Lord Howe Is. tinct; this race may be tentatively listed No specimen of this bird is known, and under the name godmanae.

SOUTHERN MELANESIAN RACES OF MACROPYGIA MACKLINAYI The following tabulation of measure- Macropygia rufa RAMSAY, 1878, ibid., p. 287, Sandwich (Efate) Is. (rufous phase). ments of adults of this dove indicates that SUBSPECIFIC CHARACTERS.-Larger and two races may be recognized in southern longer-tailed than the other races; coloration Melanesia, a small one from the Santa slightly paler than that of the following race, Cruz Islands and a large one from the New much paler than that of the non-Polynesian races; occurs in a gray phase unknown in the Hebrides and Banks Islands. other races of the species. RANGE.-New Hebrides: Aneiteum-1, LOCALITY WING LENGTH Tanna-5, Erromanga-1, Efate-4, Mau-1, New Hebrides 19 ci 153-166 9 9 150-161 Mai-2, Tongoa-2, -2, Epi-3, (158.53) (154.44) Lopevi-1, Pauuma-2, Ambrym-6, Male- Banks Ids. 1 157 3 9 148-155 kula-3, Pentecost-1, Aurora-3, Aoba-18, (151.67) Malo-2, Espiritu Santo-5, Dolphin-1. Banks SantaCruzIds. 9 e 148-154 4 9 139-153 Id8.: Meralav-3, Gaua-3, Vanua Lava-3, (150.22) (148.25) Valua-3. (The figures indicate the number of TAIL LENGTH specimens examined.) NewHebrides 11 e 166-194 10 9 160-180 WING.-New Hebrides: Aneiteum: o 158. (178.36) (171.4) Tanna: e 159, 161; 9 160, 161. Erromanga: Banks Ids. 0 2 9 165, 165 e 166. Efate: e 153, 156, 159, 162; 9 150, 150, SantaCruzIds. 4 e 157-167 3 9 164-171 155. Mau: e 161. Mai: 9 157. Tongariki: (164.0) (167.67) c* 155. Epi: e 158. Lopevi: e 160. Aurora: * 158. Aoba: 6 156, 157, 158; 9 150, 153, 154. Espiritu Santo: e 156, 157, 162. Banks Macropygia mackinlayi mackinlayi Ids.: Meralav: 9 148. Gaua: d' 157. Vanua Ramsay Lava: 9 152. Valua: 9 155. Macropygia mackinlayi RAMSAY, 1878, Proc. TAIL.-New Hebrides: Tanna: " 172, 181; Linn. Soc. New South Wales, II, p. 286, Tanna 9 170+, 177. Erromanga: " 194. Efate: " Is. (gray phase). 176, 178, 180; 9 160, 167, 169. Mau: e 190. 1943]. BIRDS OF THE WHITNEY SOUTH SEA EXPEDITION. 52 17

Mai: 9 170. Tongoa: 9 172, 180. Tongar- to a limited extent. Macmillan reports iki: ci 166. Epi: e 173. Ambrym: 9 174. Aoba: c? 175; 9 175. Espiritu Santo: c? 177. that of fifty birds observed on Erromanga Banks. Id8.: Vanua Lava: 9 165. Valua: 9 only one was gray. Regarding Aneiteum 165. we know only that the single specimen col- COLOR PHASES.-This race occurs in a lected there is brown. normal brown phase and in a rarer gray NOTES ON PLUMAGE.-Several species phase. In the latter, brown coloration is of Macropygia have a patch of specialized entirely lacking, except perhaps on the feathers on the breast. In M. mackinlayi under tail coverts, which are pale buffy these feathers are forked at the tip and white. Intermediates between the two have black bases which in the females ex- phases do not occur. In worn plumage, tend nearer to the tips of the feathers and females of the brown phase become some- produce distinct black spots on the breast. what grayish on the breast, but examina- In the males this black is scarcely or not tion of the less faded parts of their plum- at all visible, This is the principal color age shows at once that they are not inter- difference between the sexes. The ventral mediate. A gray juvenal from Aoba is as surface is slightly paler chestnut in the completely gray as adults of this phase. female and with wear becomes pale rufous The above suggests that the gray phase is stippled with grayish white on the breast. controlled by a single alternative genetic The coloration of males seems to be less factor. Phases are unknown in other affected by plumage wear. races of this species or in other species of In juvenals of both sexes the breast the genus. feathers are not bifid; they are more black The two phases of AMacropygia mackin- than those of adult females, only the tips layi were at first believed to be distinct being brown. Hence the blackish breast species, while Peters (1937, Check-List, contrasts with the remainder of the under III, p. 81) considered them to represent side more than in adults. The dorsal the sexes. That the brown and gray birds feathers in juvenals have blackish sub- represent color phases which occur inde- terminal areas followed by clearly definoed pendently of age or sex became apparent brown tips. Their wing feathers are also as soon as adequate material of this race brown tipped. was collected for the first time by. the The only sexual color difference in the Whitney Expedition. Five of fifteen speci- gray phase is the more extensive black mens taken on in the southern mottlings on the breast in the female. The New Hebrides by the American Museum plumage of this phase has a peculiar ap- collector, L. Macmillan, were in the gray pearance, as though a white powder had phase, and this proportion is probably a been sifted over it. As noted above, fair indication of their proportions on this brown phase birds in worn plumage acquire island. He believed that the gray phase this stippled appearance to some extent. is more common in the higher parts of These markings are apparently equally Tanna, but this requires confirmation. Of present in both phases, but mostly con- seventy-eight skins from other islands, one cealed by the brown coloration of the com- from Mau, an island off the north coast of mon phase. Gray juvenals differ from Efate in the south central New Hebrides, adults as do those of the brown phase, and one from Aoba in the northern part of except that all brown markings are re- the group are in the gray phase. This is a placed by gray. small proportion, but it is interesting that the mutation has been carried the length Macropygia mackinlayi troughtoni of the New Hebrides. With the exception Kinghorn of these two skins, the gray phase has ap- Macropygia rufa troughtoni KINGHORN, 1937, taken on Tanna. It Proc. Zool. Soc. London, p. 177, Vanikoro, Santa parently been, only Cruz Ids. might be expected that gray birds would SUBSPECIFIC CHARACTERS.-AS compared with be commoner in the other southern islands the large pale race, mackinlayi, of the New near Tanna; this may be true but only Hebrides and the small, deep chestnut race, 18 AMERICAN MUSEUM NOVITATES [No. 1237 arossi, of the Solomons, troughtoni agrees in REMARKS.-The observations on plum- coloration with mackinlayi, though slightly more the race apply to chestnut, and in size with arossi, though con- age made for preceding siderably larger. Kinghorn compared the Santa troughtoni, except that the gray phase is Cruz birds only with arossi and not with mackin- unknown. The long, graduated tail is layi, to which they are much more closely allied, rarely in perfect condition for measuring, and it is quite by accident that troughtoni may be but there is little doubt that the tail is accepted as a somewhat poorly differentiated race. relatively, as well as absolutely, longer in RANGE.-Santa Cruz Ids. (Vanikoro, Utupua, mackinl&yi than in troughtoni and the other Santa Cruz, Tinakula, Lomlom). races of this dove. Banks Islands birds are WING-.Vanikoro: e 150; 9 139, 152. closest to mackinlayi, but they are more or Utupua: e 149, 149, 149. Santa Cruz: e 148, less intermediate. The two races are very 152, 153; 9 149, 153. Tinakula: e 148. Lom- lom: e 154; 9 147+. similar in coloration; individuals cannot be TAIL.-Vanikoro: e 157?; 9 168. Utupua: identified, but in a series of troughtoni the e 166,166? SantaCruz: 9 171. Lomlom: chestnut coloration averages perhaps a 167?; 9 164. shade deeper.

THE POLYNESIAN RACES OF CHALCOPHAPS INDICA The southeastern part of the range of RANGE.-Santa Cruz Ids. (Disappointment, this dove is occupied by four subspecies, Lomlom, Fenualoa, Nupani, Tinakula, Utupua, all of which lack the white forehead and Vanikoro), Torres Ids. (Lo, Hiw), Banks Ids. (Bligh, Vanua Lava, Gaua), New Hebrides superciliary stripes found in the other (Espiritu Santo, Malo, Malekula, Ambrym, races. Pauuma, Epi, Efate, Erromanga, Tanna, These four subspecies may be separated Futuna-near Tanna, Aneiteum and probably as follows: others), Loyalty Ids. (Mare, Lifu, Uvea). The A.-Shoulder patch white in adults of both sexes. 1.-Size medium (wing 12 9 : 151-160; 14 d: 155-170 ...... timoriensis. 2.-Size large (wing 1 "9 ": 160; 3 cl: 168?, 169?, 172...... lo1ngirostris. B.-Shoulder patch white in males, brownish gray or grayish white in females. 1.-Size small (wing 16 9: 135-147; 24 c<: 136-153) ...... sandwichensis. 2.-Size medium (wing 6 9: 149-153; 9 d: 153-163) ...... chrysochlora. The only specimen of longirostris sexed as species was observed but not collected on Mare on by Macmillan. female is violaceous the breast instead WING.-Santa Cruz Ids.: Disappointment: of brownish, as are the females of other 142. Lomlom: e 142, 142, 144, 144, 144; 9 139. races, and may be a male despite the small Nupani: e 139. Tinakula: 6" 140; 9 141, size. Regardless of what the color char- 142. Utupua: " 136, 140; 9 138. Torres acters to is valid be- Ids.: Lo: ci 141, 147, 147; 9 136. Banks Ids.: prove be, longirostris Bligh: 9 136. Vanua Lava: c" 143. Gaua: 9 cause of its large size. Only Australian 135. New Hebrides: Espiritu Santo: ci 147. specimens are included in the above meas- Malo: 147, 147. Malekula: e 153. Am- urements of chrysochlora. brym: 143, 149. Pauuma: e 153; 9 142. Epi: c 148; S9 147. Efate: 9 141, 142, 144, 145, Erromanga: ci 144? Tanna: e 146+; 9 Chalcophaps indica sandwichensis 144. Loyalty Ids.: Lifu: 9 145+. Uvea: 9 Ramsay 145? TYPF LoCALITY.-Efate, New Hebrides. TAIL.-Santa Cruz Ids.: Lomlom: o' 81, 83; SUBSPECIFIC CHARACTERS.-Much like chryso- 9 81. Fenualoa: o' 83. Nupani: ce 86. Tina- chlora, but smaller; breast, crown and upper kula: 9 83. TorresIds.: Lo: e 77, 87; 9 82. back in males vinaceous cinnamon; in males Banks Ids.: Bligh: 9 85. Vanua Lava: e 84. of chrysochlora in fresh plumage these regions Gaua: 9 81. New Hebrides: Espiritu Santo: are perceptibly lighter, more violaceous, but c 88. Malo: e 90, 92. Malekula: e 90. many worn specimens are inseparable; the wing Ambrym: e 87, 91. Pauuma: ce 92; 9 86. patches are slightly less conspicuous in both Epi: 9 92. Efate: 9 85, 87, 89, 90. Erro- sexes in sandwichensis; females of the two races manga: c 89. Tanna: i 88; 9 91. Loyalty are otherwise identical in coloration. Ids.: Lifu: 9 93. 19431 BIRDS OF THE WHITNEY SOUTH SEA EXPEDITION. 52 19

There is a slight increase in size from where: Lord Howe Is.; eastern Australia; east- ern New Guinea westward to the Oriomo R. and north to south but no perceptible difference Astrolabe Bay; Manam; D'Entrecasteaux in color. From the Loyalty Islands we Archipelago; Trobriand Ids.; WoodlarkGroup; have only two specimens, both females; in Bonvouloir Group; Louisiade Archipelago. neither of these can the wing length be WING.-New Caledonia: &I 154, 154; 9 149+; New South Wales: c~ 155, 156, 156, determined with certainty. Brasil (1916, 163; 9 149, 151, 151. Rev. Franc. d' Orn., IV, p. 195) records the TAIL.-New Caledonia: e 96, 100, 101; 9 wing length of a female from Lifu as 147 97. New South Wales: ci 88, 92, 95, 97; 9 mm. The measurements of these three 92,94,97. birds are nearest to those of scndwichensis, New Caledonian specimens are distinctly to which the Loyalty Islands population is larger than those of the New Hebrides and here tentatively referred. It is possible nearer both in wing and tail measurements that a better series would show them to be to chrysochlora, even when comparison is closer to chrysochlora. limited to topotypical specimens from New South Wales. They are fully as large as Chalcophaps indica chrysochlora Wagler those of many of the Papuan islands and other tropical parts of the range of chryso- Columba chrysochlora WAGLER, 1827, Syst. Av., Columba, sp. 79. New South Wales designated as chlora. Brasil based disjuncta on supposed type locality by Mathews. size differences, but the difference in size Chalcophaps chrysochlora disjuncta BRASIL, between sandwichensis and chrysochlora is 1916, Rev. Franc. d'Orn., IV, p. 195, New certainly not sufficient to justify an at- Caledonia. tempt to maintain an intermediate race SUBSPECIFIC CHARACTERS.-This race is nearest to sandwichensis but is larger; there are between them. Brasil and others, perhaps also slight color differences as noted above; from influenced by Gould's name longirostris, longirostris, chry8ochlora differs by being smaller, have emphasized variation in bill length and perhaps adequate material of the former will scarcely reveal color differences. C. i. chrysochlora is also in this dove. Aside from slight, very similar to timoriensis, and some males are demonstrable differences correlated with indistinguishable. Those of chrysochlora aver- variation in general size (of which wing age slightly grayer and less violaceous ventrally, length is the best indicator available) I and the white patch on the bend of the wing is no variation in bill length in the in many individuals less pronounced, a smaller have found portion of each feather being white; they usually races studied. The difference in color be- have the neck and upper back less suffused with tween sandwichensis and chrysochlora is so slate color. Females of chrysochlora have the slight that it was of no aid in allocating the wing patch grayish, while in adult females of few available from New Cale- timoriensis it is white as in the male; chrysochlora specimens also averages slightly smaller than timoriensis. donia and the Loyalty Islands, all of which RANGE.-In Polynesia: New Caledonia. Else- are in rather poor plumage.

NOTES ON GALLICOLUMBA STAIRI This ground dove belongs to a super- more distinct from beccarii than is cani- species containing from west to east, the frons, and I see no reason to follow Sharpe following species: hoedti, Wetar Island in assigning it to a separate genus or sub- (north of Timor); beccarii, New Guinea, genus as Peters has done (1937, Check-List, Bismarck Archipelago and Solomon III, p. 137). The three species beccarii, Islands; sanctaecrucis, Santa Cruz Islands sanctaecrucis and stairi are more closely re- and Espiritu Santo Island, New Hebrides lated to each other than to hoedti or cani- (specimen in British Museum, fide Mayr); frons but are so distinct that after a general stairi, central Polynesia; and canifrons, consideration of this genus it seems best to Palau Islands. Dr. Mayr has pointed out consider them species. The relationships to me that canifrons is a member of this of the rare G. salamonis are uncertain, but superspecies and not of the jobiensis group it is probably closer to the jobiensis group as he at first thought (1936, Amer. Mus. (only one specimen, an immature female, Novitates, No. 828, p. 4). G. hoedti is no was seen). 20 AMERICAN MUSEUM NOVITATES [No. 1237

G. b. beccarii exhibits striking sexual that of the male. The other Samoan skin dimorphism in color. Males have a is a female in the retarded type of plumage. sharply defined light gray breast shield and Some related species such as sanctaecrucis the bend of the wing is deep red; females are known from so few specimens that it is lack the red, and the shield is cinnamon still possible, though improbable, that they and contrasts little with the general brown- have two types of female plumage also. ish tone of the plumage. In other races of The normal juvenal plumage of the niem- beccarii the females have acquired a some- bers of this superspecies is similar to the what male-like, advanced type of plumage retarded female plumage but somewhat in which the breast shield is grayish, but darker, and the back and wing feathers are not so light and contrasting as that of the tipped with cinnamon-brown. In most male. In Gallicolumba stairi, the series forms, for example, G. b. beccarii, this juve- collected by the Whitney Expedition re- nal plumage is retained for a considerable veals the interesting fact that this trend has period and full-grown males in this plumage continued and produced a female plumage or just molting over to adult plumage are which is indistinguishable from that of the common in collections. G. stairi is peculiar adult male, though perhaps averaging a in that the juvenal plumage is largely sup- shade more olive and less brown on the pressed. Two or three partly grown indi- back. More remarkable is the fact that a viduals with down still clinging to some of second type of female plumage of the re- the feathers already had red feathers on tarded, feminine type also occurs in stairi. the bend of the wing, and the breast shield The latter has hitherto been considered the was aquiring a white border. Such speci- only female plumage of the species and is mens have juvenal back feathers, but the one described in the "Catalogue of these too apparently are replaced by adult Birds" and elsewhere. The existence of feathers much sooner than in related spe- the male type of plumage in females of cies. Evidence is insufficient to determine stairi has been overlooked because of the if females in the retarded type of plumage rarity of this species in collections and the also assume adult plumage immediately improbability of two types of female plum- upon leaving the nest, but presumably they age existing. G. stairi seems to be the do. only dove or pigeon in which both ad- NOMENCLATURE.-Gray (1856, Proc. vanced and retarded plumages occur in the Zool. Soc. London, p. 7) described Galli- same sex. Surprisingly, none of the speci- columba stairi from a bird in the London mens in our series is intermediate, and the Zoo; the specimen was later added to the type of plumage appearing in each female British Museum collection. Concerning must be controlled by a mechanism oper- the origin of the type he wrote: "I suppose ating on the "all or none" principle. The was brought from the Samoan or Navi- following numbers of females have the ad- gators' Islands, as the British Museum was vanced, male type of coloration: 18 of 30 previously in possession of a skin given by from Fiji, 6 of 6 from Tonga, 0 of 1 from the Rev. J.S tair as from that locality .... " Alofa Island. Sexual size dimorphism is Finsch (1872, Jour. f. Orn., XX, p. 49), greater in this species than in any other without seeing the type and with very pigeon treated in this paper and permits scanty material, concluded tentatively from the sexing of the collectors to be evaluated. the plate published by Gray that the type Eight of nine specimens from Samoa are had come from the Tongas; he proposed of male type plumage. They are battered names for both the Samoan and Fijian skins of "missionary" make and cannot be populations, should they prove to be dis- measured accurately enough to determine tinct. Salvadori (1893, Cat. Birds, XXI, how many, if any, of the eight are females, p. 596), after comparing Gray's type and especially since the normal measurements plate with two Samoan specimens, con- of the Samoan population are not known. cluded that Gray's name was based on a Most are sexed as male, but perhaps only bird from Fiji or possibly Tonga. How- because this plumage was considered to be ever, the two birds from Samoa which he 1943] BIRDS OF THE WHITNEY SOUTH SEA EXPEDITION. 52 21 had for comparison must have been im- series, poor though it is. There would seem mature or abnormal for he said they differ to be little doubt that the Samoan and from Fiji skins by the absence of the white Fijian populations are racially distinct, as border on the breast shield and gray on the would be expected in a land bird such as occiput and nape. All of the eight Samoan Gallicolumba, but the following diagnosis birds in this pluniage now available have will undoubtedly be modified when good both of these characters. At least one of material of the present race becomes avail- the eight agrees better with Gray's plate able. than do most Fijian skins. Salvadori SUBSPECIFIC CHARACTERS.-Differs from the evidently found the type to agree with other race, vitien8is, as follows: size smaller, wing Stair's supposed Samoan bird, as he ques- averaging about 8-10 mm. shorter in each sex; breast shield in male plumage lighter, more tioned the locality of the latter also and vinaceous, less brownish and with a narrower, lists it under the Fiji form (p. 596). less well-defined white border; under wing per- We now know that the three populations haps more extensively rufous; gray patch on of this species are so similar that it is im- back of head restricted to occiput and not ex- tending onto upper neck; the single specimen in possible absolutely to identify the published female plumage has the upper parts more brown- plate with any of them. Salvadori's action ish, less olive and greenish than in vitiensis (fad- was based on a misunderstanding of the ing?). characters of the Samoan form, and he did RANGE.-Upolu and Savaii Ids., Samoa. what was ap- WING.-Upolu: 148, 154, 154+. "Samoa" find the type to agree with 143+, 150+. All these skins are sexed as parently the only adult he had from Samoa. males, but perhaps only because it was believed Under the circumstances the changing of that this plumage is restricted to that sex. The the type locality designated by Gray ap- one undoubted female is from Upolu and has the and I am wing in very poor condition; perhaps its true pears to have been unjustified length was about 145 mm. using Samoa as the type locality of stairi. TAIL.-"c ": 84, 88, 88, 88, 91, 95. Dr. J. T. Zimmer has been kind enough to examine the evidence and reached the same conclusion. Re-comparison of the Gallicolumba stairi vitiensis Finsch type with good specimens from Fiji and SUBSPECIFIc CHARACTERS.-Differs from 8. Samoa is needed. stairi by being larger and by the color characters described above. stairi stairi Gray TYPE LOCALITY.-Fiji. Gallicolumba RANGE.-Alofa Is. (subsp.?), Fiji Ids. (Yang- Caloenas (Phlegoenas) Stairi G. R. GRAY, asa cluster, Olorua, Aiwa, Tavutha, Vanua 1856, Proc. Zool. Soc. London, p. 7, P1. cxv, Mbalavu, Vatu Vara, Ngamea, Taviuni, Rambi, probably from Samoa. Kio, Vanua Levu, Koro, Makongai, Wakaya, Phlegoenas samoensis FINSCH, 1872, Jour. f. Ovalau, Viti Levu, Mbenga), Tonga Ids. (Late, Orn., XX, p. 50, Samoa. Nomukaiki, Levuka, Telekitonga, Tonumela, Unfortunately the Whitney Expedition Hongatonga, Hongahapai, Kuakafa, Kao) and two specimens of this dove probably other islands in these groups. The obtained only only specimen from the rather isolated island of in Samoa, both immatures. We also have Alofa is a juvenal female which in size seems to seven adults collected by Woodford in agree best with this race, as would be expected on 1895 and another undated but more recent geographical premises. skin lent by the U. S. National Museum. WING.-Fiji: Yangasa cluster: e 166. Olorua: ce 160+, 162; 9 157. Aiwa: e 163, All these specimens show evidence of 164, 167, 171; 9 152, 153, 153, 154, 155, "foxing," and most of them have the wing 156. Tavutha: ci 163, 164. Vanua Mbalavu: in molt. Specimens available for com- e' 166, 166; 9 152. Vatu Vara: 9 152. from and Fiji were all col- Taviuni: 6 160. Rambi: e 161; 9 155. parison Tonga Kio: 9 148. Koro: 9 148. Makongai: 9 lected more recently, so it is difficult to 149. Wakaya: ci 162+; 9 146. Ovalau: 9 say to what extent the differences visible 143. Viti Levu: e 161; 9 153. Mbenga: 9 may be the result of age. It is significant 147. Tonga: Late: ce 161. Nomukaiki: c 163+. Hongatonga: e 161+; 9 150+, that Finsch and especially Wiglesworth 151. Hongahapai: " 162. Summary: 15 c (1891, Aves Polynesiae, p. 57) found much 160-171 (163.8); 18 9 143-157 (151.3). the same characters in Samoan specimens TAIL.-Fiji: Yangasa cluster: e 108. which seem to be indicated in the present Olorua: e 107, 107; 9 104. Aiwa: 6' 105, 22 AMERICAN MUSEUM NOVITATES [No. 1237

106, 108, 108; 9 96, 96, 97, 98, 99. Tavutha: Late: c' 105. Nomukaiki: e 103. Teleki- e' 107. Vanua Mbalavu: e 104; 9 99. Vatu tonga: 9 96. Hongatonga: e' 109; 9 97, 98, Vara: 9 99. Ngamea: e 107 Taviuni: 98. Hongahapai: e 105. Summary: 21 e ce 102. Rambi: e 104; 9 101. Vanua 98-109 (105.3); 18 9 93-104 (98.2). Levu: ce 104. Makongai: 9 98 Wakaya: e 105; 9 93. Ovalau- 9 100. Viti Levu: e 98, No differences could be found between 103: 9 98. Mbenga: e 107; 9 100. Tonga: Fijian and Tongan birds.