AMERICAN MUSEUM NOVITATES Published by Number 1360 THE AMERICAN MUSEUM OF NATURAL HISTORY October 31, 1947 New York City

A REVIEW OF THE DICAEIDAE BY ERNST MAYR AND DEAN AMADON

CONTENTS INTRODUCTION ...... 1 THE FAMILY DICAEIDAE...... 1 POSITION OF FAMILY...... 3 DISTRIBUTION OF THE DICAEIDAE...... 5 PLUMAGE AND SEXUAL DIMORPHISM...... 6 HABITS AND ADAPTATIONS...... 7 GENERA AND OF DICAEIDAE...... 13 REFERENCES ...... 30

INTRODUCTION The last comprehensive revision of the larly the puzzling relationship of certain entire family of (Dicaeidae) species in the Lesser Sunda Islands, are was published in 1885 (Sharpe). Not only discussed in the present paper. Finally, have numerous new species and subspecies the attempt has been made to form some been described during the past 60 years, general conclusions on the evolutionary but the whole concept of the species has history of this family. been revolutionized since then. A general Original quotations and full synonymies review of the family is therefore urgently are not given since they are easily available needed. Recent work by Mayr (1941) on in various ornithological lists and hand- the New Guinea flowerpeckers, by Mayr books. However, when possible, reference (1945) and by Delacour and Mayr (1946) has been made to the habits and the ecology on the flowerpeckers of the Philippines, of the various species. The senior author by Delacour (1946, 1947) on the flower- has had field experience with various peckers of Malaysia, and by Hindwood species of Dicaeidae in New Guinea and the and Mayr (1946) on the genus Pardalotus Solomon Islands, and the junior author in has cleared up some of the taxonomic dif- the Philippines. All the illustrations are ficulties in this family. Others, particu- by Alexander Seidel.

THE FAMILY DICAEIDAE This family of seven genera and 54 species (Paramythia) is crested. The species (in about 40 superspecies) of pas- edges of the bill are frequently serrated. serine is restricted to the Oriental and The tenth (outer) primary is visible, though Australian regions. Flowerpeckers are small, in the primitive genera and vestigial birds of small or, rarely, medium size, fre- in the others. The legs are short; the quently of bright and varied colors. The toes and claws stout; the tarsus is either plumage is often glossy but only rarely scutellated or almost booted; the nostrils metallic. Some have bristles around the are partly covered by an operculum. The gape and nostrils, and hair-like plumes on more typical genera Anaimos and the nape and occasionally on the flanks. which contain 41 of the 54 species feed They are without wattles or bare areas of about flowers on nectar and insects (hence skin on the head, and only one aberrant their name flowerpeckers), and the bill is 2 AMERICAN MUSEUM NOVITATES [No. 1360

Dio

Rhampho(haris

IMelano(haris

NECTARINIIDAE I

%% DDIC-AEIDAE --a

Fig. 1. Diagr'ammatic phylogeny of the genera of Dicaeidae. 19471 A REVIEW OF THE DICAEIDAE 3

sometimes decurved and the tongue tu- genera are rather diverse in habits, as bular in correlation with these habits. pointed out in the following section. These two genera also feed much on the The supposed phylogeny of the genera is berries of parasitic mistletoes. The other given in figure 1.

POSITION OF FAMILY Three families, the Meliphagidae (Aus- Nectariniidae and Zosteropidae are well tralian honeyeaters), Nectariniidae (sun- represented in the Old World tropics both birds), and Zosteropidae (white-eyes), are in Ethiopian and Oriental regions and have often thought toberelated to the Dicaeidae, a few stragglers in the Australian region, but the evidence is rather slight. All four especially the Zosteropidae which have families (at least in some genera) take nec- undergone considerable evolution in Poly-

K// /A! AII c D

Fig. 2. Tongues of three flowerpeckers (A-C) and a sunbird (D). A. Melanocharis versteri, X 51/2. B. Paramythia montium, X 51/2. C. Dicaeum trigonostigma (after Gadow). D. Nectarinia (Leptocoma) sp. (after Gardner). tar and small insects from flowers. The nesia and the Solomon Islands. The tongue and bill are variously modified for Dicaeidae seem to have been originally this habit, especially in the honeyeaters Papuan, but reached the East Indies and and sunbirds, less so in some flowerpeckers Philippines early, where many species have and white-eyes. The tenth (outer) pri- evolved. The Nectariniidae may well mary is always evident in the Meliphagidae represent an offshoot of stock similar to and Nectariniidae, present or absent in the Dicaeum. In both, the nest is a special- Dicaeidae from genus to genus, and always ized pensile bag with side entrance which, absent (vestigial) in the Zosteropidae. together with other resemblances, is prob- The Meliphagidae. are typical of the Aus- ably significant (see also Delacour, 1944). tralian region, extending even to Hawaii We have examined the tongues of two but barely reaching Bali and Celebes. The specimens of Paramythia (fig. 2). The 4 AMERICAN MUSEUM NOVITATES [No. 1360

tongue is triangular, somewhat concave they could readily have evolved. The dorsally and with a small notch at the tip. tongues of Melanocharis and Paramythia Ventrally there is a vaguely indicated de- do not differ significantly from those of pression or line extending back from the many birds. The tongue seems split at the tip. In two specimens of to be of little taxonomic value except in Melanocharis the tongue is similar but genera or families where it is much special- perhaps slightly more split at the tip ized. (fig. 2). The tongue of these two genera Two or three types of tongues occur in is of a generalized type, in keeping with the Nectariniidae (Delacour, 1944). In their frugivorous habits. In the two one type, found in Nectarinia and other species Dicaeum hirundinaceum and D. genera, the basic pattern is similar to that concolor examined by us the notch at the of Dicaeum hirundinaceum (or similar end of the tongue is deeper, and each of the species), but the tongueisgreatlyelongated, two tips resulting is semitubular. In D. very thin, and much rolled so that not only trigonostigma, as figured by Gadow and the tip but most of the tongue is conspic- reproduced in our figure 2, each of the two uously tubular. The tip is split as in tips is slightly split to form "four equally- Dicaeum, with two tubular projections. sized semitubular projections without the We see no reasons why the tongue of sun- slightest indication of laciniated or frayed- birds could not represent a greater modi- out margins. . . ." (Gadow, p. 235). fication of the type fouvid in Dicaeum. Gadow represents the edges of the tongue A tongue of Nectarinia figured by Gardner behind the split tip as meeting dorsally to is similar to one we have examined and is form a tube. In the two species we ex- reproduced in figure 2. amined only a semitube is indicated, but In the Meliphagidae the tongue is very perhaps in life the edges are brought to- brush-like at the tip and split into four gether, forming a complete tube. distinct tips and hence considerably Gardner (1925, p. 26) wrote as follows: different from that of any genus of Dicaei- "The Dicaeidae have small tongues that dae. Some New Guinea genera of meli- are flat posteriorly but at about the middle phagids resemble Melanocharis, but in view become abruptly narrower and begin to curl of their difference in habits, this is probably into a semitube which is deeply cleft at the parallelism. tip, the margins of which are smooth, Gadow considered the dorsal feather forming two' slender semitubular tips. tract of Dicaeum very similar to that of the This is found in Dicaeum cruentatum, swallows (Hirundinidae), but the families D. sanguinolentum, D. flammeum, and can scarcely be closely related in view of D. celebicum [and according to our findings the many great differences between them. in D. hirundinaceum and D. concolor]. The purse-shaped nests of Dicaeum are In Acmonorhynchus [=Dicaeum] aureo- quite similar to those of the penduline tit- limbatus the same holds true except that mice (Remiz). Delacour (1944, p. 19) the edges of each tube show a slight notch- has suggested that Cephalopyrrus flam- ing, with an attempt at the production of miceps of southeastern Asia may provide a four tips, while in [ = An- link between Remiz and Dicaeum. Cepha- aimos] these notches have deepened to lopyrrus, which is usually placed with actual splitting with the formation of four the titmice, nests in holes in trees. The semitubular fringeless projections." Gard- females are very similar to those of Remiz. ner mentions Gadow's description of D. It will require further anatomical and life trigonostigma as representing another type history studies to determine whether Remiz of modification. is really related to Dicaeum, or merely In summary, the tongues of Dicaeum builds a similar nest. It is also a moot and Anaimos are modified in a character- question whether Remiz is closely related to istic way for sucking nectar, and at the the true titmice (Parus). same time resemble the simpler tongues of Sharpe (1885, p. 2) associated the wax- the fruit-eating flowerpeckers from which wings (Bombycillidae) with the Dicaeidae. 1947] A REVIEW OF THE DICAEIDAE 5

Despite some morphological similarities, African genera, Pholidornis and Parmop- these families differ considerably in habits tila, placed by Sharpe with the Dicaeidae, and distribution. The relationship of the have been considered by Chapin to be Bombycillidae to the Neotropical Ptil- aberrant Ploceidae and by Bates (and more ogonatidae is more certain, and has led recently Delacour) perhaps to comprise, Stresemann to unite these families. It-is along with another monotypic genus, conceivable but unlikely that the Bomby- Hylia, a separate family. We have ex- cillidae provide a northern link between the amined all three and with Ptilogonatidae and Dicaeidae. The simi- agree Chapin larity to the Hirundinidae emphasized by and Bates that they have nothing to do Gadow is of equally doubtful significance. with the Dicaeidae. A few genera have been erroneously or From the above it is only too apparent doubtfully allied with the Dicaeidae. The that much anatomical work is needed be- Hawaiian Drepaniidae are now considered fore the relationships of the Dicaeidae to to be of American derivation. Two other families is placed on a sound basis.

DISTRIBUTION OF THE DICAEIDAE New Guinea and the Philippine Islands mythia). ThelattertworesemblePardalotus are the home of the greatest number of in some respects but are more like Melano- species. In the east the family extends as charis in habits. One superspecies of the far as the Solomon Islands, in the south to wide-ranging Dicaeum has also reached Australia and Tasmania, in the west and New Guinea via the Moluccas and ex- north to south China and northwest India. tended through New Guinea to the Bis- The number of genera and species of marck Archipelago and Solomon Islands. Dicaeidae in selected areas is as follows. In the Philippines, East Indies, and Malay Some species occur in several of the areas Peninsula there are two closely related listed. genera, Anaimos and Dicaeum, each con- Ceylon: Dicaeum, 3 taining a number of species. Anaimos Southern India: Dicaeum, 3 seems to be more primitive and has fewer Eastern and northern India, northern Burma: species. In tropical Asia only Dicaeum Dicaeum, 7 is known (except in the Malay Peninsula). Malay Peninsula: Anaimos, 3; Dicaeum, 6 Borneo: Anaimos, 4; Dicaeum, 6 The family does not occur beyond India Palawan: Anaimos, 1; Dicaeum, 2 and south China. Luzon and Mindanao: Anaimos, 1; Dicaeum, 8 These distributional facts contain some Celebes: Dicaeum, 3 clues as to the place and time of origin of Timor: Dicaeum, 3 Flores: Dicaeum, 4 the family. The general distribution Australia: Dicaeum, 1; Pardalotus, 6 pattern, as well as the prevalence both of Tasmania: Pardalotus, 3 primitive and of specialized genera and New Guinea: Melanocharis, 5; Rhamphocharis, species on New Guinea, makes it seem 1; Oreocharis, 1; Paramythia, 1; Dicaeum, 1 (superspecies) probable that the family was originally a Solomon Islands: Dicaeum, 1 (superspecies) Papuan group. From here the family spread south, producing the Pardalotus The generic distribution may be further group in Australia, and west into the Malay summarized as follows: In Australia and Archipelago, giving rise to Anaimos and Tasmania there is one aberrant genus Dicaeumn. (Pardalotus) containing seven species and It is difficult to decide when the Asiatic Dicaeum with one species which is a recent mainland was reached. The family is imnigrant from the Lesser Sunda Islands. fairly well represented in the humid tropical In New Guinea there are two primitive and subtropical parts of the Indo-Chinese genera (Melanocharis and Rhamphocharis) countries (Malay Peninsula, Burma, Indo- containing six species and two special- China), but is drastically reduced in abun- ized, monotypic genera (Oreocharis, Para- dance in the more arid parts of India. It 6 AMERICAN MUSEUM NOVITA TES [No. 1360 has not spread as far as the Palearctic, into Africa or whether there was always an Africa, or Madagascar. Although the environmental barrier between Africa and pronounced aridity of northwest India and Asia impervious to flowerpeckers. Arid Arabia is a comparatively recent phe- zones obviously are a considerable obstacle nomenon, it is an open question whether the to flowerpeckers, as indicated by the poor Dicaeidae arrived so recently in India that representation of the family in Australia, they have had no- opportunity to spread adjacent to New Guinea with its rich fauna.

PLUMAGE AND SEXUAL DIMORPHISM In the primitive genera Melanocharis blackish coloration has extended around to and Rhamphocharis, the prevailing color- the underparts. In the brightly colored ation of the females is greenish olive. D. papuense-haematostictum superspecies Usually the males are more or less glossed of the Philippines, the female has acquired with bluish black above, and they may even the bright coloration of the male. (in Melanocharis versteri) have a metallic In Melanocharis versteri and longicauda luster. Males of M. striativentris are and in Rhamphocharis there are white spots olive green like the females. In Rham- on the tail feathers. It is a moot question phocharis the male is only slightly glossed; whether these are homologous with those the female is brownish spotted above occurring in three species of Dicaeum and below with white and thus very differ- (vincens, melanoxanthum, and agile), as in ent in appearance. This spotting is perhaps Dicaeum the spots are nearer the tips of foreshadowed in the obscure ventral streak- the feathers. These marks are among ing of M. striativentris. In most species of many indicating that melanoxanthum and these two genera the tail is spotted with vincens are related; agile, on the other hand, white; otherwise the pattern is very simple. is a rather aberrant species. Small round The coloration varies greatly in the nu- white spots occur on the tail feathers of merous species of Dicaeum and Anaimos. some species of Pardalotus, but again it is D. bicolor is whitish below and bluish uncertain whether they indicate relation- black above, thus resembling some species ship with other flowerpeckers or are merely of Melanocharis (e.g., nigra). This re- associated with the extensive spotting semblance is probably indicative of characteristic of the pattern in this genus. relationship. Females of Dicaeum are Such spots on the tail do not occur in An- usually simply colored, often brownish aimos, Oreocharis, or Paramythia. olive above and yellowish green below, There is little in the coloration of the even in the more brightly colored species aberrant Oreocharis to suggest the more of the genus. In a few species such as typical genera of flowerpeckers, although D. concolor, D. erythrorhynchos, and D. the black head and throat may imply dis- tristrami the males as well as the females tant relationship to such forms as Anaimos are dull colored, but in some of these, thoracicus. Paramythia is still more especially tristrami, there is little doubt specialized but with unmistakable re- that secondary simplification in coloration semblances to Oreocharis. Yellow is has taken place. A few species of Dicaeum conspicuous in the remaining genus, Par- and Anaimos are streaked ventrally, but dalotus, and probably indicates distant the degree of streaking varies greatly even affinity with Oreocharis and more remotely among the races of one species (agile). with the Anaimos-Dicaeum group. The In many species bright patches of red or white spots on the outer webs of the sec- orange feathers occur on the crown, breast, ondaries in some of the species of Par- rump, or back, and in one species (D. dalotus are especially suggestive of Oreo- annae) there is a peculiar half-concealed charis. Various combinations of spots and patch of orange feathers on the lower back. streakings are distinctive of Pardalotus, The back in the males is usually dark and though one species (quadragintus) has more or less glossy. In some species the reverted to a somewhat simpler pattern. 1947] A REVIEW OF THE DICAEIDAE 7

In Oreocharis the females are much generic separation on this basis is less con- simpler in coloration than males and not vincing in Dicaeum. Thus the peculiar greatly dissimilar to those of some species D. annae looks like Anaimos, but the of Melanocharis and Dicaeum. In Par- tenth primary is vestigial, so it must be amythia and some species of Pardalotus left in Dicaeum. The same seems to be females have acquired the brighter color- true of D. anthonyi (the only species in ation of the males. the family not examined). On the other hand, in D. melanoxanthum the tenth TENTH PRIMARY primary is visible, though smaller than in Anaimos. This species is obviously close The tenth primary is evident, though re- to D. vincens, and we have left it in Di- duced, in Melanocharis and Rhamphocharis, caeum. two genera which, on the basis of other The tenth primary is vestigial in Oreo- characters as well as this, are probably charis, Paramythia, and Pardalotus; these primitive. A few species of typical flower- are specialized and rather aberrant genera peckers have an evident tenth primary. in other respects. It is somewhat sur- These we have segregated in Anaimos, prising that the four endemic frugivorous leaving the others in a presumably slightly New Guinea genera, Melanocharis and more advanced genus, Dicaeum. While Rhamphocharis on the one side and Oreo- there is no dQubt that the six species of charis and Paramythia on the other, should Anaimos, wit! the doubtful exception of be so different as regards the tenth primary olivaceus, belong to a natural group, the and most other characters.

HABITS AND ADAPTATIONS Rand (1942, p. 511) wrote of Melano- that P. melanocephalus uropygialis hunts charis nigra unicolor, "this species was insects in the flowers of a Grevillea and euca- usually found solitary, low in the substage lypts, becoming smeared with honey and and undergrowth, but once at the 1,200- pollen about the throat and forehead while meter Camp I found numbers of them doing so. This is suggestive of the habits feeding on fruits high in a forest tree." of some species of Dicaeum. He also notes that Melanocharis longicauda Chasen (1939, p. 403) summarized the frequents both undergrowth and tall fruit habits of the species of Dicaeum as follows: trees. Of Melanocharis versteri meeki he "Usually they are seen alone; tiny, fat, adds: "This fed in the sub- but very active little birds, darting across stage and commonly in low second growth. the path from one tall flowering bush to It was usually alone, and an active another with a sharp chit-chit-chit, a sound with quick movements. Three stomachs like two pebbles being tapped together, examined contained fruit." but often they congregate at the tops of Rand (1937, p. 239) found Oreocharis very tall flowering trees and then, through usually in parties of up to 20 or more in the binoculars, the busy birds can be seen lower fruit trees of the New Guinea forest. fitting like a swarm of hymenoptera about Sixteen stomachs contained only fruit of the blossoms on the exposed surface of the several varieties. Usually the fruit had forest's canopy.. . . I have known cases in been swallowed whole. The related Par- which natives employed for insect-catching amythia, as observed by Rand (p. 241), is in the Botanic Gardens, Singapore, have taken the tiny species of Dicaeum in their also a fruit eater. butterfly nets. They eat small spiders, Diamond birds (Pardalotus) are almost insects and small berries and are especially entirely insectivorous. They seek food fond of the berries of plants allied to the among twigs and foliage, actively and with mistletoe. There is a marked partiality for a variety of postures similar to those of tit- the glutinous fruit of Loranthus. Epi- mice. Hill (1911, p. 283), however, states phytic plants of tall trees seem to attract 8 AMERICAN MUSEUM NOVITATES [No. 1360 flowerpeckers. The nectar of flowers is well as Peepal and Banyan figs. Bombax also taken as food." and Erythrina flowers are regularly resorted Ali (1931) has made a special study of the to for the nectar, and spiders likewise role of flowerpeckers in India in fertilizing form part [of] the dietary. It utters chik, mistletoes (Loranthus) and in spreading chik as it restlessly hops about the foliage. their seeds. The flowers of Loranthus ... While thus engaged, it spreads out its longiflorus are entirely dependent on nec- stumpy square little tail and screws it tar feeding birds for fertilization, but sun- nervously edgewise from side to side." birds are much more important in this than Lawrence and Littlejohns (1916) wrote of flowerpeckers. During the season when Dicaeum hirundinaceum, ". . . the ripe they are available, Loranthus berries form berry [Loranthus] was taken crosswise in the principal food of Dicaeum erythro- the bird's bill, and the soft case split in rhynchos; at other times it takes the berries halves by pressure. The free portion of the of Phyllanthus, Lantana, and Viscum case was then dropped, leaving the white (another mistletoe), and occasionally small berry protruding from the half still at- spiders. "While on a clump [of mistletoe] tached to the branch. By pressure of this the bird hops restlessly from bunch to remaining half between the mandibles the bunch uttering an almost incessant chik, seed was forced out sufficiently to allow chik, chik which is occasionally varied by of its being easily taken in the bill." Nest- a series of twittering notes which might lings were fed insects when stnall but before be termed its song. Each berry is first they left the nest were receiving chiefly tested between the mandibles; if ripe it is mistletoe berries. This flowerpecker also plucked and swallowed, broad end (i.e. eats other types of berries, such as wild where the stalk attaches) first. After it cherries. Insects are secured both from has bolted down three or four berries, one flowers and by search among foliage. This after another the bird retires to the ex- species is very restless and usually stays in tremity of some bare branch at the top the tree tops. of the host on an adjoining tree and sits W. K. Dammerman (quoted by Chasen, quiet for a few moments with the feathers 1939, p. 404, without a reference) wrote: partly fluffed out. . . presently one of the "Bartels found in Java the glutinous seeds viscous seeds is excreted." The seeds ad- of these parasitic plants [Loranthus] only here to the branches, thus spreading the in the intestinal tract of the bird, the large parasite. In a later publication Ali (1936 seeds being unable to enter the very small p. 778) mentions finding as many as 27 stomach, which is found always filled with Loranthus berries in the intestines of Di- minute spiders, the chief food of the birds." caeum erythrorhynchos. Porsch (1929) in his exhaustive study of Although D. erythrorhynchos rarely strips flower-visiting birds and associated flowers, the fleshy outer covering from the devotes 15 pages to the Dicaeidae, which is, berries without swallowing them, this so fait as the birds are concerned, a com- method is usual in the thicker-billed D. pilation. Dicaeum concolor sollicitans has agile. "The berry is plucked and invar- been found with its head smeared with iably revolved between the mandibles nectar and pollen. Porsch found refer- which being thicker and stouter, appear ences to at least five species of Dicaeum better adapted to this method of eating. visiting flowers. The following genera The flesh is soon detached and the seed of plants are mentioned: Loranthus, wiped on to a neighboring twig" (Ali, 1931, Elytranthe, Bauhinia, Bombax. pp. 148-149). In a later publication Ali The habits of the species of Anaimos are and Abdulali (1938, p. 160) wrote of the similar to those of Dicaeum so far as known. habits of Dicaeum a. agile: "Its food con- sists largely of the berries of the tree par- FEEDING ADAPTATIONS (INTESTINAL asites, Loranthus and Viscum. . . . Other TRACT) berries such as those of Lantana camara and Desselberger (1931) distinguishes two Bridelia retusa are also greedily taken, as functional categories of intestines among 1947] A REVIEW OF THE DICAEIDAE 9

the Dicaeidae: a more primitive one, ex- species consists exclusively of fruit. The emplified by Melanocharis, and a more anatomy of the related, also frugivorous specialized one, as found in Dicaeum. genus, Oreocharis, is unknown. The intestines of Melanocharis are short Desselberger studied several species of and wide, a condition typical for fruit- Dicaeum. He found that the muscular -eating birds. The stomach consists of a stomach (gizzard) has become a blind sac small, glandular proventriculus and a well- or diverticulum with a sphincter at its developed muscular gizzard. The stomach opening. By this means berries (Loran- of Paramythia is considerably larger, but thus and others) are shunted directly from Oes. Int. Oes.

I #tAN NO

I M.F

.01 .

. . . . I

. .I

Fig. 3. Stomach of Paramythia cut open. Abbreviations: Oes., esophagus; int., entrance to the intestines. One of the two horny pads was cut when the stomach was opened. it is not clear in the examined specimen the esophagus to the intestine without (Archbold-Rand collection) whether or not entering the gizzard. Insects and spiders, the lower end of the esophagus is modified on the other hand, enter the gizzard where to a proventriculus; it does not appear so. they are comminuted and the digestible The gizzard is very muscular, with two components are removed. This special- well-defined pads (fig. 3). The pylorus, ization permits large numbers of Lor- which is near the esophageal entrance of the anthus berries to pass rapidly through the stomach,, leads into a wide intestine, the intestine, where the presumably slight mucous wall of which is modified into amount of nutriment in their gelatinous innumerable villae. The food of this capsules is removed. It also avoids the 10 AMERICAN MUSEUM NOVlTATES [No. 1360

N\ 15 12 \

ii I\ /\ 13

t~~~~~~~~~4_<>- \ 9 10 \- 8

0~~ ;>E

\ 7 / \ 4

3 /3 \5 2

\ I Fig. 4. Rapid evolutionary change of bill shape in the Dicaeidae. 1. Melanocharis nigra. 2. Rhamphocharis. 3. Oreocharis. 4. Paramythia. 5. Pardalotus melanocephalus. 6. Anaimos olivaceus. 7. Dicaeum agile. 8. D. aureolimbatum. 9. D. nigrilore. 10. D. bicolor. 11. D. papuense. 12. D. retrocinctum. 13. D. nehrkorni. 14. D. eximium. 15. D. tristrami. Compare particularly the closely related forms 1 and 2, 8 and 9, 11 and 12, and 13 to 15. A REVIEW OF THE DICAEIDAE 11 laxative effect on other food that these might postulate a gradual adaptation of the berries would have if held in the gizzard. heavier bill of fruit- and insect-eating an- Since chitinous parts are not found in the cestors to permit greater utilization of intestines, it appears that they are re- nectar. This seems to be, in general, true, gurgitated as pellets. This elimination but there have been frequent reversals of protects the mucous villae of the intestines this trend, and both types occur in closely from injury by sharp-edged chitinous related species. D. tristrami, with a particles. The intestinal tract of Dicaeum heavy, finch-like bill, is a geographical rep- tristrami studied by us is exactly like that resentative of and derived from species in of the other species described and figured which the bill is thinner. D. nigrilore by Desselberger. One examined stomach is thin billed but undoubtedly closely re- contained 15 spiders, 14 of which were lated to the heavy-billed D. aureolimbatum. Argiopidae, and one a salticid. The. in- In the papuense-retrocinctum group of the testines contained remains of berries with Philippines there is considerable variation very minute seeds (we counted more than within a superspecies (fig. 4). 90). It is possible that some but not all of In the tanagers (Thraupidae) of the these variations among species of Dicaeum genus Tangara (Euphonia), which are may be eventually proved to be correlated thought to feed almost entirely on mistletoe with specific differences in food habits. berries of Phoradendron and related genera, The variation is undoubtedly correlated a different and greater specialization has with the fact that both the seeds of mis- led to a vestigial condition of the gizzard tletoe and also tiny insects and nectar from as a section of the intestinal tract (Stein- flowers are important foods in this genus. bacher, 1935). Utilization of the bill for opening the pods In Dicaeum and Anaimos the tongues of Loranthus and (at least in D. agile) for are, in varying degrees, tubular as in other stripping the gelatinous capsule from the birds in which nectar is an important item seeds may favor a heavier type of bill. of the diet, as described above. Desselberger, however, found that the alimentary system of the stout-billed D. BILL aureolimbatum is modified for the passage of In the four fruit-eating genera of New Loranthus seeds, so we cannot assume that Guinea, the bill is unspecialized, slightly the stout-billed species are those in which curved, and rather broad basally. In seeds are not swallowed. Apparently con- Rhamphocharis it is moderately attenuated, sumption of nectar and tiny insects found but we do not know if this is adaptive. in blossoms favors a thin, decurved bill The bill in Pardalotus is stout, arched, (and tubular tongue). On the other hand, and short. Probably it is adapted for utilization of mistletoe berries is probably digging nesting burrows or enlarging easier for birds with a stout bill. We may cavities in trees. In feeding, the bill of suppose that now one and now the other Pardalotus isutilized fortakingsmall insects, of these two factors has dominated in the their larvae and eggs much in the manner of evolution of the various species of the various titmice, vireos, or other insec- genus, leading to the great variation that tivorous birds that have comparatively exists from species to species. So far as short, heavy bills. the family as a whole is concerned, the thin, In the genera Anaimos and Dicaeum nectar-sucking type of bill and tubular the variations in the bill are greater and tongue must be considered the more spe- often quite puzzling either from a func- cialized. This is exclusive in the tional or phylogenetic point of view. In type pos- some species the bill is very stout and short, sibly related specialized sunbirds (Nectar- scarcely if at all decurved. In others the iniidae) which feed on nectar and tiny bill is rather thin and decurved, approach- insects but not berries. ing the condition found in more typically The bills of the tanagers (Tangara) nectar-eating birds such as sunbirds. We that feed on mistletoe berries are quite 12 AMERICAN MUSEUM NOVITATES [No. 1360 similar to those of the thick-billed species were thick so that the chamber scarcely of Dicaeum. accommodated the two young. Diamond birds (Pardalotus) nest in holes NESTS in trees or in banks in burrows which they The nest of Dicaeum cruentatum, which often dig themselves. They 'even dig is typical for the genus, has been described nesting burrows in level sandy ground. by Baker (1934, p. 237) as follows: "They They build elaborate nests inside these seem to be invariably made of the beau- hollows which are often more or less domed tifully soft seed-down of the Simul-tree but sometimes cup-shaped. These nesting (Bombax malabarica), very little comnpressed habits are'aberrant, but such birds as the or felted but kept in shape and position House Sparrow (Passer domesticus) or by a few shreds of grass, fungoid mycelae Common Mynah (Acridotheres tristis) or by very fine hair-like roots. The nest show that the transition from hole nests to is egg-shaped and the materials, other than domed nests in the open, or vice versa, is the down, are used to work round the twig easily effected, sometimes even locally from which the nest is pendent, and from within a species. this they are brought down and round the The aberrant Paramythia builds a bulky, nest itself, a few of the longer strands cup-shaped nest among the twigs of thick coming under the nest and up again on the bushes (Rand, 1942, p. 513). The nests 6f far side. Round the rim of the entrance, the other New Guinea fruit-eating genera which is rather large and at the upper side are unknown. of the nest, a few grasses are twisted, making it firm enough to withstand the EGGS constant passing in and out of the parent The eggs of Anaimos percussus (the bird. Here, too, as on the outside of the only species of the genus for which we nest, cobwebs and silk are sometimes used have information) are whitish and un- to strengthen its structure. The lining is spotted, and the same is true of the major- of the same cotton-seed down and quite ity of species of Dicaeum. D. agile and soft when first put in, but soon becoming D. papuense haematostictum are known to more or less felted when the eggs are laid lay spotted eggs. These species are not and the birds begin to sit." closely related. Nevertheless, when the A nest of Dicaeum eximium from New color of the eggs of all species of Dicaeum is Britain, which we have examined, is coarser known it may give some clues to the affin- in construction, being made of fine grass ities of the species. It would be interesting ornamented on the outside with pieces of to know if. D. bicolor and D. anthonyi, dead leaves. The nest of Dicaeum vincens which seem distantly related to papuense, (for a photograph see Lushington, 1940) lay spotted eggs. Most forms of Dicaeum, has "a pronounced hood over the some- so far as known, lay two or three eggs. Nests what large entrance." One nest of this of D. trigonostigma in the Malay Peninsula species was 75 feet and another 125 feet contained but one nestling (Chasen, 1939, high in, Dipterocarpus trees. The nest of p. 407). Dicaeum concolor raises two Dicaeum agile is unusually finely woven broods (Whistler, 1934, p. 285), and the and elastic so that it will regain its shape same has been inferred for other species. after being rolled up. The species of Pardalotus also lay whitish A nest of Anaimos percussus ignicapil- unspotted eggs, the clutch varying from lus from Borneo was similar to those of two 'to five according to species. Par- amythia,lays one egg which is white spotted Dicaeum (Baker, 1934, p. 249). Coomans with brown (Rand, 1942). de Ruiter (1936) has published a photo- graph and description of a nest of Anaimos NESTING HABITS xanthopygius. It was only about a meter In Dicaeum vincens the female builds above the ground but well protected and the nest and incubates; the male helps concealed by large leaves. 'The walls feed the young (Lushington, 1940). Chand- 1947] A REVIEW OF THE DICAEIDAE 13

ler (1912, p. 132) found that in Dicaeum In Paramythia it was observed that only hirundinaceum only the female works on the female incubates, but the male ac- the nest. The male sometimes accom- companies her when off the nest and aids panies her while gathering nesting material, in feeding the young (Rand, 1942). and the male utters a distinctive note in the vicinity of the nest. Both sexes feed the young, at first insects, later mistletoe ber- VOICE ries. The male drives other males from the vicinity of the nest (Lawrence and Little- Typical flowerpeckers have a variety of johns, 1916). rather sharp, almost metallic call notes. In Pavdalotus both sexes participate in D. hirundinaceum also has a sweet song and, digging the nest tunnel and probably in a variety of twittering notes. We noticed building the nest. The male is also said to that Dicaeum ignipectus luzoniense has a aid in incubating as well as in caring for the rather loud warbled song, and the same is young (Chandler, 1910). Diamond birds probably true of many species of the are said to be somewhat colonial at times genus. Pardalotes (Pardalotus) have rather in suitable nesting banks. Roberts (1937), monotonous unmusical call notes and also however, noted that the male of a pair of a trill, but little in the way of song. Rand P. ornatus feeding young in a hole in a (1937) wrote that the calls of Paramythia bank repeatedly drove away another par- are low and short but mentions no song. dalote that was hunting a nest site in other Individuals of Oreocharis, when flying in a holes in the bank. flock, call a sparrow-like chip-chip.

GENERA AND SPECIES OF DICAEIDAE The type species of genera are given in indistinctly streaked (striativentris); upper- parentheses following the name of the parts olive green or blue black, often glossy describer of the genus. In general, we or even metallic (versteri); rectrices often have not attempted to revise subspecies, with white marks. Edges of bill well and where we have listed them, it is serrated. Bill rather short and heavy; primarily to indicate to what species we culmen moderately decurved. Tarsus believe'various forms should be allocated. with only one or two scutes near base in front, thus almost booted (specimens of MELANOCHARIS P. L. SCLATER a.GRA) M. ver8teri in fluid). SYNONYMS: Neneba De Vis (striativen- This genus includes the most primitive tris); Pristorhamphus Finsch (versteri); flowerpeckers. Sexual dimorphism is well Urocharis Salvadori (longicauda). developed in several species, with the DIAGNOSIS: Tenth primary well devel- males more brighly colored but the females oped, the exposed portion being 15 mm. larger. Data on the latter point, as given or more in length and almost half the by Mayr (1931, pp. 668-670), are:

WEIGHT (GMS.) WVING (MM.) SPECIES 61 ~~9 Q, 9 Melanochari8 nigra bicolor 13,14 15, 15.5, 17 67,67.5 66,68,68.5 Melanochari8 longicauda captata 13-15 (13.9) 14-16 (15.4) 64-67 (65.9) 64-67 (65.5) Melanocharis versteri maculiceps 12.5-15.5 (14.0) 16.5-20 (18.5) 59-64 (61.8) 66-71 (69) length of the ninth. Tail relatively long, The data for captata and maculiceps are varying from about 75 per cent (nigra) based on series of about six of each sex. to 140 per cent (versteri) of the wing length. The figures for Melanocharis striativentris Color pattern simple: underparts ranging indicate no sexual size dimorphism. from grayish white through various shades The genus Urocharis has not been re- of yellowish or greenish brown, sometimes cognized by recent authors. Mayr (1931, 14 AMERICAN MUSEUM NOVITATES [NO. 1360 p. 668, footnote) gave reasons for uniting DIAGNOSIS: Anaimos has a relatively Pristorhamphus with Melanocharis. At large tenth primary and serrated bill, as that time he retained the genus Neneba for do the two preceding genera. It differs striativentris because of its somewhat differ- from Melanocharis and Rhamphocharis as ent proportions and body build, but we now follows: tail much shorter (about 30 mm., feel it is best to associate this species with 50-60 per cent of wing); general size much its close relatives in the genus Melanocharis. smaller (total length about 80-90 mm. The five species here included in Mel- against 105-107 mm.); color pattern more anocharis comprise a monophyletic group varied, usually with some red and yellow in whose species intergrade in all variable plumage; no white in tail; axillaries and characters. The total range of variation under wing coverts snow white,~females is not greater than that often included by not larger than males; tongue more or less recent workers within the limits of a single tubular. genus. While Melanocharis is in most Anaimos is very closely related to Di- respects primitive, some of the species are caeum, the only constant difference being the only ones in the family with metallic the noticeable tenth primary of Anaimos. plumage. The bill in the latter is heavier than in all INCLUDED SPECIES: 1, arfakiana; 2, nigra; but a few species of Dicaeum, and has a 3, longicauda; 4, versteri; 5, striativentris. more or less characteristic shape. The All the species are restricted to New Guinea. crown patch and white malar stripe of Their ranges and subspecies are given by Anaimos (except olivaceus) are rare in Mayr (1941). Melanocharis arfakiana, Dicaeum, but occur in a few species. So far the plainest colored and perhaps most as known the two genera are identical in primitive species of the genus, is known nidification and habits. from only two specimens collected almost Anaimos is, with the doubtful exception at the opposite ends of New Guinea. of A. olivaceus, a closely knit, natural The genus is in some respects RHAMPHOCHARIS SALVADORI group. (CRASSIROSTRIS) primitive, as indicated by the large tenth DIAGNOSIS: Closely related to Melano- primary and the heavy bill. In coloration, and thin habits, and nidification it belongs with the charis but bill elongated rather more specialized flowerpeckers. The six 18 mm. as against 12 mm. (culmen about species are confined to Malaysia and the in the larger species of Melanocharis). Philippines and have apparently originated green with a very slight Male above olive in this area. gloss and below gray. Female brown spot- The suggested interrelationships of the ted with white. Female very noticeably of Anaimos and Dicaeum are in- larger than male (wing in male 66 mm., in species dif- dicated diagrammatically in figure 5. female 73 mm.). The larger size and SPECIES: 1. A. olivaceus. This Philip- ferent coloration of the female are re- pine species is the simplest of the genus in markable. The male is not unlike some a of simpler species of Melanocharis in coloration and the only one without the crown patch and without any bright colors. color. The single species of this genus is New Guinea. It has neither a malar stripe nor a rump restricted to the mountains of patch. There are two races, olivaceus Rhamphocharis and Melanocharis comprise and parsoni. a rather old and primitive section of the 2. A. maculatus. This is the only spe- Dicaeidae, which is surprisingly absent cies that is streaked ventrally. It resem- from tropical north Australia. bles olivaceus to some extent in coloration SPECIES: 1, crassirostris. but has a crown patch, which links it with ANAiMoS REICHENBACH (THORACICUS) the following species. It occurs in pen- SYNONYM: Charitociris Oberholser insular Siam, Malaya, Sumatra, Borneo, (percussus). The name Prionochilus and some of the nearby smaller islands formerly used for this genus is preoccupied (four races). in the insects. 3. A. percussus. This species is slaty 19471 A REVIEW OF THE DICAEIDAE 15 hirundinaceum

thoracicus is

ANAIMOS Fig. 5. Diagrammatic phylogeny of the species of Anaimos and Dicaeum. blue above with a crimson crown patch 4. A. plateni (johannae auct.). Like and yellow below with a red pectoral spot. percussus but with the rump yellow as in A white malar stripe is present. The the following species, xanthopygius. Since races are: ignicapillus (Malaya, Sumatra, plateni, which is endemic to Palawan, is Rhio Archipelago, Billiton, Banka, North intermediate between percussus and xan- Natuna Islands, Borneo), regulus (Tana thopygius, which coexist on Borneo, it Massa, Batu Islands), and percussus (Java). cannot be safely united with either species 16 AMERICAN MUSEUM NOVITA TES .[No. 1360 as a race (Mayr, 1945, p. 115). We may species as we have pointed out above. assume that the widespread percussus, or Mayr (1945, p. 114) has shown in more its ancestor, spread to Palawan, giving rise detail the drawbacks of further subdividing to plateni which has reentered Borneo to the genus Dicaeum. produce the third species, xanthopygius. As it has long been known that the tenth According to a recent letter from Hach- primary is vestigial in many flowerpeckers, isuka, the name plateni (Braunschw. it is surprising that Oberholser (1923) con- Anzeiger, no. 52, 1888) antedates johannae, fused the ninth primary with the tenth in published in the same year. the species he studied (quadricolor, vincens, 5. A. xanthopygius. This species is some races of agile) and stated that the much like plateni but lacks the white malar tenth primary is not reduced in these stripes. It is found on Borneo and the species. Actually it is reduced to the North Natuna Islands. vanishing point. 6. A. thoracicus. A rather rare, mono- Although some of the species of Dicaeum typic species of Malaya, Borneo, and Bil- are easily arranged into species groups, the liton. The color pattern is distinctive: head natural arrangement of all species of the and breast black, mid-breast and crown genus is difficult. As already noted, the crimson, rump yellow, back citrine olive. bill tends to become thinner in the more This pattern suggests remote relationship advanced species, but this trend has been with Dicaeum trigonostigma or D. quadri- reversed several times. The coloration of color. most species of the presumably more prim- itive Anaimos is bright and varied, and DICAEUM CUVIER (CRUENTATUM) hence the more simply colored species of SYNONYMS: Acmonorhynchus Oates (vin- Dicaeum are not necessarily primitive. cens); Chilociris Oberholser (aeruginosus, a D. concolor, D. erythrorhynchos, and D. race of agile); Chromatociris Oberholser tristrami probably owe their plain colors to (quadricolor); Cryptociris Oberholser (ob- secondary simplification. All three are soletus, a race of agile); Pachyglossa Blyth closely linked with bright-colored species. (melanoxantha); Piprisoma Blyth (agile); Certain groups of forms that still are geo- Polisornis McGregor (anthonyi). graphical representatives (superspecies) Polisornis was proposed as a subgenus. show great variation in shape of bill and in McGregor also proposed a subgenus Bourn- coloration, e.g., the papuense-retrocinctum sia with the same type (aeruginosus) as the group, the bicolor-quadricolor group, and earlier Chilociris of Oberholser. the erythrothorax group (including tris- DIAGNOSIS: Similar to Anaimos but trami). Some of the more unusual color tenth primary vestigial (in D. melano- patterns in Dicaeum, such as those of xanthum, the only species of Dicaeum with chrysorrheum, agile, anthonyi, and annae, a visible tenth primary, this primary is suggest that Dicaeum may not be mono- short and very narrow, only abouit 10 mm. phyletic in the narrowest sense of the term, long and 1 mm. wide, compared with but that various branches may have de- about 13 mm. by 2.5 mm. in Anaimos); veloped from different species of Anaimos crown patch and malar stripe usually lack- or Anaimos-like ancestors. Thus Dicaeum ing; bill often thinner than in Anaimos with annae resembles Anaimos olivaceus in some the lower mandible less upturned. In speci- respects, and the streaked D. chrysorrheum mens preserved in fluid the tarsus is seen (and perhaps agile) may be directly related to be scutellate in front (probably the same to the streaked A. maculatus. Some of is true of Anaimos) and hence more "prim- these resemblances are doubtless deceptive, itive" than in the other genera (Parda- and we cannot suggest any further sub- lotus?) in which it is virtually booted. division of the genus that would not ob- The various genera that we synonymize scure rather than elucidate relationships. were based primarily on differences in the The only question is whether it is profitable thickness and shape of the bill, but these are to separate Anaimos and Dicaeum. We not of a nature to indicate natural groups of do so only tentatively. 19471 A REVIEW OF THE DICAEIDAE 17

The phylogeny of Dicaeum suggested in Most races of agile have white spots on the figure 5 and explained in the following inner vanes of the outer tail-feathers account of the species attempts to rec- (absent in aeruginosum). This suggests oncile the above difficulties, but will relationships to vincens and melanoxanthum doubtless be improved as further knowl- the only other species of Dicaeum with such edge of life histories, eggs, and anatomy marks. accumulates. The first eight species listed The races of agile are: zeylonense (Cey- in Dicaeum are the most primitive and lon); agile (India); modestum (? eastern most like Anaimos in the shape of the bill India, Burma, Malay Peninsula; see and in coloration. Deignan, loc. cit.); pallescens (Siam, Indo- SPECIES: 1. D. annae. This peculiar China); atjehense (Sumatra); finschii species occurs in the Lesser Sunda Islands, (Java); tinctum (Sumba, Flores, Alor); the nominate race on Flores and the race obsoletum (Timor); bungurense (North sumbavensis on Sumbawa. Records from Natuna Islands); everetti (Borneo, other islands seem to be due to early con- Labuan); affine (Palawan); aeruginosum fusion with D. agile. In D. annae the (Philippine Islands). males have a half-concealed patch of The name atjehense was proposed by bright yellow feathers on the lower back. Delacour (1946, p. 4) for Piprisoma This feature is unique in the family but sumatranus Chasen, preoccupied in Di- possibly homologous with the yellow rumps caeum. The older records of this species or lower backs of such species as Dicae- from Sumbawa are based on a mistake of um quadricolor or Anaimos olivaceus. It Hartert's in identifying immatures of D. equally resembles Dicaeum agile, which also annae as the present species. We have occurs in the Lesser Sundas, and since found no definite records for Sumbawa. both of these species lack the tenth primary 3. D. chrysorrheum. This species the relation may be closer than with contains three races which range from east- olivaceus. ern and northern India through the Indo- 2. D. agile. Many of the races of this Chinese countries and Malaysia. This and widespread species (or superspecies) are agile (some races) are the only species of rare in collections. It is a dull species, Dicaeum having streaked underparts. grayish above and whitish below, usually D. chrysorrheum has a thin bill, very unlike more or less streaked. The bill is thick, that of agile. The yellow under tail short, and blunt. The eggs are spotted, coverts of chrysorrheum apparently link and the nest, while in general like those of it with the following group of yellow-bellied the genus, is said to be especially finely species (nos. 4-8). The bright olive woven, so as to regain its shape even after upperparts of chrysorrheum are also partly being rolled up or crushed. The'species duplicated in aureolimbatum and nigrilore must be considered rather aberrant but not of the group just mentioned. We thus necessarily'primitive. Two forms of the tentatively place chrysorrheum as a link agile group have been thought to occur between the streaked agile and the species together in the Malay Peninsula, but this with partly yellow underparts but without seems doubtful (Deignan, 1945, p. 550). conviction that the similarity to agile The heavily streaked Philippine form (aeru- actually indicates close relationship. ginosum) is linked with the others by affine 4. D. anthonyi. The few known speci- of Palawan, while the unstreaked obsoletum mens of this species, which McGregor of Timor is linked with the striated forms described from the moss forests of the by tinctum of the nearby islands. We northern Luzon mountains, were in the believe it will prove correct to unite all the destroyed collection of the Manila Bureau forms listed below as races of agile. Ober- of Science (for a colored plate see McGregor, holser (1923) divided them among three 1914). It is the only species of Dicaeum genera and five species, with two other with a yellow crown patch (male only) like species (vincens, quadricolor) sandwiched in that usual in Anaimos. The back is glossy between. black in the male, olive green in the female. 18 AMERICAN MUSEUM NOVITATES [No. 1360

The throat is white; the breast and abdo- caeum isag. The original description does men are yellow. This coloration plus the not mention a single character not ap- heavy bill, mountain range, and apparent plicable to nigrilore, and the colored plate rarityof anthonyi suggest that it isprimitive. looks like nigrilore. McGregor (1927, p. 522) clearly states 9. D. hypoleucum. This Philippine that anthonyi has the tenth primary ves- species contains three races: hypoleucum tigial, so we must place it in Dicaeum. In of Basilan and Mindanao, pontifex of the color pattern anthonyi is rather similar to Samar-Leyte group, and obscurum of the four species that follow (nos. 5-8), all Luzon. Although there is a considerable of which perhaps have a relict distribution. variation in color of plumage and soft parts 5. D. melanoxanthum. This Himalayan from north to south, pontifex is intermediate species is the largest of the genus and, as are between hypoleucum and obscurum, so that many mountain species, is rather long it seems justifiable to unite the three forms winged. It is the only Dicaeum in which in a single species (Mayr, 1945, p. 115). the tenth primary is visible, though very Better material may show that Tweeddale's small. Like anthonyi it is bluish black race mindanense is separable from typical above and yellow below with white throat hypoleucum. Mayr (1946) proposed the and breast, but it lacks a crown patch, and name pontifex as a substitute for Dicaeum the black has encroached on the sides of the everetti Tweeddale, which is preoccupied by breast. There are white spots on the outer Prionochilus everetti Sharpe, once the latter tail feathers. The bill is broad and short. is transferred to Dicaeum as a race of D. It was formerly separated in the genus agile. The rather attenuated bill and to Pachyglossa. some extent the coloration of hypoleucum 6. D. vincens. This species resembles resemble nigilore, although the relationship D. melanoxanthum in so many details of may not be very close. coloration as to leave little doubt they are 10, 11. Dicaeum bicolor, D. quadricolor related, as P. L. Sclater (1874) pointed out (superspecies). Males of D. bicolor are long ago when figuring them together in the bluish black above and gray below with a Ibis. D. vincens is found in the lowlands whitish mid-ventral area. Two races occur and foothills of Ceylon. The tail feathers in the Philippines. On Cebu it is replaced have large white spots near their tips that by the little known quadricolor (colored are undoubtedly homologous with those of plate: Proc. Zool. Soc. London, 1877, melanoxanthum. pl. 77, fig. 2), similar but with the upper 7. D. aureolimbatum. This Celebes back scarlet and the lower back ochraceous species resembles the three preceding ones, green. The two species agree in all struc- but the white of the breast extends back tural details and seem to be geographical with yellow restricted to the sides. Above representatives despite these striking dif- only the wings, tail, 'and sides of the head ferences in coloration. Similar variation are blackish; the rest is yellowish olive. occurs in the following superspecies (pap- The bill is short and thick. A race occurs uense). Females and young of both species on Sangir. are dull and scarcely separable. 8. D. nigrilore. A Mindanao species Dicaeum bicolor is like D. h. hypoleucum rather rare in collections (colored plate: in color and perhaps related, but the bill Novitates Zool., 1920, vol. 26, pl. 6). is heavy as in anthonyi or melanoxanthum Very similar to aureolimbatum but the back and (together with papuense) it links these is brownish, only the crown and flight rather primitive, heavy-billed species with feathers olive. The bill is thin and de- the more widespread and apparently ad- curved, but the coloration and distribution vanced, thin-billed species that follow. leave little doubt that nigrilore is a close Both bicolor and hypoleucum resemble relative of aureolimbatum. It helps link some species of Melanocharis in their style the heavy-billed species with those with of coloration. thin bills. The species was apparently 12, 13. D. papuense, D. retrocinctum redescribed by Hachisuka (1941) as Di- (superspecies). The ranges of the three 1947] A REVIEW OF THE DICAEIDAE 19 forms in this Philippine superspecies are as colored, thin-billed species similar to follows: D. retrocinctum, Mindoro; D. erythrorhynchos, though they occur to- papuense haematostictum, Guimaras, gether in some parts of India. It is much Negros, Panay; D. p. papuense, most of the more widespread, with the following races: other Philippines from Luzon to Mindanao concolor (southern India), ? subflavum (cen- and Basilan. D. p. papuense is similar to tral India), olivaceum (Himalayas to Yun- D. bicolor bicolor, but the bill is thinner nan and Kwangtung, south through the and longer, the back is duller black, and the Indo-Chinese countries), ? sinense (Sze- mid-ventral area is crimson, not white. chwan), uchidai (Formosa), virescens D. papuense haematostictum is similar, but (Andaman Islands), minullum (Hainan), the back is glossier and the ventral red area borneanum (Malaya, Sumatra, Borneo, is partly encircled with black at the breast. North Natuna Islands), sollicitans (Java, In D. retrocinctum the bill is much thinner Bali). and the entire breast is black, meeting the 17. D. pygmaeum. This Philippine and black of the top of the head and neck. Palawan species may be a representative of There is a patch of lanceolate red feathers concolor, a species which it resembles in its on the throat and also a transverse crimson dull coloration, thin bill, whitish spot patch of red in the scapular region. Thus, before the eye, and small size. On the a red dorsal patch is a variable character in other hand, males of pygmaeum have a this superspecies, as it is in D. bicolor- slight bluish gloss on the wings, a whitish quadric&kor. mid-ventral area, and a yellowish rump, This superspecies provides another in- suggesting the following cruentatum group stance of the small taxonomic value of of species, also lacking in the Philippines variation in the bill in flowerpeckers. It is unless represented by pygmaeum. Thus the only brightly colored species of Di- the latter seems to link concolor loosely with caeum in which females have acquired the the cruentatum group. male type of coloration. Dicaeum jlavi- venter Meyer was probably based on a speci- THE cruentatum SPECIES GROUP men of D. papuense discolored by alcohol The ranges of the 13 species comprising (Mayr, 1945). this species group (species 18-30, below) 14. D. trigonostigma. This species are mapped in figure 6, and their supposed occurs in the Philippines, Malaysia, and the phylogeny is shown in figure 8. southern Indo-Chinese countries. There Before the evolution of this group is are eight races in the Philippines alone and discussed, the species will be listed and about an equal number in the rest of the briefly described. range. This is a thin-billed species with 18. D. nehrkorni. In the male of this a red or orange patch on the upper back and Celebes species, the crown, forehead, rump, often with a yellow rump patch, which and a small spot on the mid-breast are sometimes is joined with the back patch. crimson, the back and wings glossy bluish Throat and breast are grayish and the rest black, the breast and sides gray, the belly of the underparts bright yellow. Females and under tail coverts white with a blackish are dull greenish. This is one of the most area medially. The female is similar but specialized species of the genus; it is duller, with no crimson on the breast and probably of Philippine origin for the color only a suggestion of it on the head; the pattern suggests D. retrocinctum. rump is red as in the male. 15. D. erythrorhynchos. This dull- 19. D. vulneratum. This monotypic spe- colored, thin-billed, tiny species is found cies is found on Amboina, Seran, and Goram in peninsular India, with a race on Ceylon. in the southeastern Moluccas. Like nehr- The dull coloration is probably secondary. korni it is very grayish below, lighter in the It is a close relative of concolor and so female, with a red pectoral spot in the male. similar to it that occasional specimens may This red mark is much larger in vulneratum not be identifiable. than in nehrkorni. Dorsally both sexes of 16. D. concolor. This is another dull- vulneratum resemble the female of nehr- 20 AMERICAN MUSEUM NOVITATES [No. 1360 1947] A REVIEW OF THE DICAEIDAE 21 korni: rump red; crown grayish like the ences of degree, and we do not think it back (not red!); no dark blue gloss as in necessary to recognize nitidum as a species. the male of nehrkorni. Male without In this species the rump is red as is usual black ventral stripe. in the cruentatum group and not gray as in 20. D. erythrothorax. This species re- pectorale and erythrothorax from which sembles vulneratum, but flanks and belly geelvinkianum was presumably derived. are washed with yellowish olive. The 23. D. eximium. This species is found rump is not red, but the back is washed on the larger islands (New Britain, New with olivaceous which becomes yellowish Hanover, New Ireland) in the Bismarck on the rump in the male (probably the Archipelago. The male is white below final stage in the loss of the red rump). with a red breast spot bordered with gray- The male has a large red breast patch. ish brown which extends back more or less The race erythrothorax with white throat on the sides and mid-ventral area. Above occurs on Buru in the southern Moluccas, it is dull brownish olive with red rump, a and schistaceiceps with gray throat on maroon wash on the crown, and a faint Obi, Batjan, and Morotai in the northern blue gloss on the tail. The female is Moluccas. Its apparent absence on Hal- similar but entirely whitish below except mahera is surprising. along the flanks and sides. It is thus quite The characters of the two Moluccan different in appearance from the neigh- species leave little doubt that vulneratum boring species. Eye-ring, loral and malar is closest to nehrkorni, although the Buru stripes are white. In the latter character race of erythrothorax is now more or less it resembles Anaimos. intermediate geographically. Probably 24. D. aeneum. In this Solomon erythrothorax (schistaceiceps) was originally Islands species there is a return to the limited to the northern Moluccas where it brighter coloration usual in this group. was an offshoot of vulneratum. Recently The male is remarkably similar to D. e. it spread south to Buru where vulneratum erythrothorax, but the gray upperparts are for some reason was absent. more brightly glossed with greenish, with- 21. D. pectorale. This species is found out noticeable yellow wash even on the on the islands of Misol, Salawati, Batanta, rump. There are other minor differences. and Waigeu west of New Guinea and on the The female is similar to that of erythro- coasts of western New Guinea. These thorax. islands are close to the northern Moluccas, 25. D. tristrami. This very aberrant and pectorale is evidently a descendant of species is found on San Cristobal in the erythrothorax. It is similar to erythro- Solomons (where aeneum is lacking). It is thorax but more olivaceous, less grayish entirely without lipochromes: white below, throughout, even becoming greenish on the washed with gray on the sides, brownish crown and back. There is no trace of a across the chest. The back and sides of red rump patch. It is a smaller species. the head are brown, more or less mottled 22. D. geelvinkianum. The range is with whitish; darker and slightly glossy on New Guinea (except the part occupied by wings, tail, and malar stripes. The sexes pectorale), some of the nearby islands, and are alike. The bill is unusually heavy and the Louisiade Archipelago. In the male large. D. tristrami carries much further the upperparts are almost as in nehrkorni, tendencies apparent in the rather dull but glossed with purplish blue. Below it eximium of the Bismarck Archipelago. It is very similar to pectorale. The female is is possible that tristrami was derived from eximium, while aeneum descended more like the male but duller with the crown and recently and directly from geelvinkianum rump reddish. The two races of the Louis- (probably nitidum). On the other hand, iades (nitidum and rosseli) are somewhat it may be merely another example of the different: back glossed with greenish; parallelism that is common in the species tone of underparts ochraceous; crown group (e.g., the similarity of aeneum to duller, almost maroon. These are differ- erythrothorax). 22 AMERICAN MUSEUM NOVITATES [No. 1360

All of the species of the cruentatum group overlapped with the less closely related just listed, with the possible exception of trochileum. nehrkorni, may be considered members of Rensch (1931, p. 616) mentions finding a single superspecies, and all (except tris- trochileum in the lowlands of Lombok up trami) are obviously closely related geo- to 200 meters. D. maugei neglectum graphical representatives, though many occurs in the lowlands also but ranges are so distinct we think it better to consider up to 2000 meters and is much more com- them species. We now return to a group mon. We lack material of the Lombok of more specialized species which are linked race of trochileum and cannot judge which through nehrkorni to those listed above. of these subspecies is more differentiated 26. D. igniferum. The upperparts of (and hence perhaps older). Other evi- the male are similar to those of nehrkorni, dence suggests that trochileum is the more but the back is washed with red. Wings recent arrival on Lombok. D. maugei has and tail are black glossed with dark purpl- colonized the small island of Penida, but ish blue. The throat is white, the breast not adjacent Bali, suggesting that the red. The sides of the head are black, and presence of trochileum may have excluded this extends around as a border below the it on Bali though not on Lombok. red breast and then back as an indefinite 28. D. cruentatum. Whereas D. mau- line separating the white abdomen. The gei appears to be a relative of igniferum in pattern below is thus similar to the geo- which the red of the upperparts (except the graphically distant eximiunm. The female rump) has been lost, in cruentatum the red is whitish below, duller on the breast; the has been intensified so that the entire back is similar to that of the male but crown, back, and rump are scarlet. The duller. This species occurs in the north wings and tail are deep bluish black, the central part of the Lesser Sunda Islands, sides of the head and breast duller black, the nominate race on Sumbawa and Flores, and a broad mid-ventral stripe is white. and cretum on the islets Pantar and Alor. The female is duller, with only the rump 27. D. maugei. The underparts of the red. The tone of the plumage is soft, male in this species are almost the same suggesting nehrkorni, which some authors as in its close relative igniferum. Above, have considered the closest relative of only the rump is red; the rest of the plum- cruentatum. It is possible, therefore, that age is black glossed with dark purplish the red back was independently acquired in blue as are the wings and tail of igniferum. igniferum and cruentatum, although ig- The females of the two species are similar niferum is intermediate as regards the pres- ventrally, while dorsally red occurs only ence of red on the back. on the rump in maugei. The races of cruentatum are: nigri- The races of maugei are: maugei (Timor, mentum (Borneo); niasense (Nias); batuense Samau, Savu); salvadorii (Moa, Babar); (Batu Islands); sumatranum (Sumatra); romae (Roma, Damar); splendidum (Djam- ignitum (Malay Peninsula, most of Burma); pea and Saleyer, between Celebes and the siamense (Siam, Indo-China); erythronotum Lesser Sundas); neglectum (Lombok); (coccineum auct.; see Deignan, 1943; and mariae (islet of Penida, south of Bali). southern China); hainanum (Hainan); The validity of Neumann's race mariae cruentatum (eastern India, northern Chin is doubtful, as Meise intimates (1941). Hills). The ranges of igniferum and 29. D. trochileum (flammeum auct.). maugei The entire upperparts, head, throat, and suggest that the former evolved in the breast are light crimson in the male. The western and the latter in the eastern part of underparts are uniform gray; the wings the Lesser Sunda group. Recently maugei, and tail bluish black similar to cruentatum. which has been able to colonize small The female is light grayish with a red rump. islands, extended its range to more or less This is a very specialized species-the only encircle that of igniferum but without over- one of the genus with the head entirely red. lapping. On Lombok, however, it has For this reason we place it after cruentatum, 19471 A REVIEW OF THE DICAEIDAE 23 although on geographical evidence, if these korni of Celebes does furnish an excellent species were derived from igniferum, pre- transition from the Sunda to the Moluccan sumably this stock reached Java before species, as it is obviously related to both. Sumatra or Borneo. As noted above, an Perhaps the Sunda species are older than alternative theory is that cruentatum and the Moluccan ones, but nehrkorni, though later trochileum were independently de- a member of the Sunda group, never ac- rived from nehrkorni, the dispersal then quired the red back and other specialized being from Celebes to Borneo, Sumatra, characters of that group. More recently, we and Java. may assume, nehrkorni stock radiated east- Dicaeum trochileum occurs on Java and ward, giving rise to the series of allopatric Bali, with a race stresemanni on Lombok, Moluccan and Papuan species. Another where it overlaps with D. maugei neglectum. possiblity is that the Moluccan and Papuan This is the only overlap of range in the species are as old as, or older than, the cruentatum species group. D. trochileum Sunda ones, but that the former have been is so distinct there could be little doubt of kept more or less similar by occasional its specific status, even if this overlap stragglers from island to island with re- did not exist. sultant gene exchange. Whichever of these hypotheses is favored, nehrkorni may be EvOLUTION OF THE cruentatum taken as a central point in the cruentatum SPECIES GROUP species group (figs. 6, 8). It has already been pointed out that Dicaeum maugei is a relative of the red- pygmaeum may be a plainly colored mem- backed igniferum in which the back is black. ber of this group or it may represent the Its resemblance to the type of pattern of still more dully marked D. concolor in the the Moluccan-Papuan species is apparently Philippines. It thus serves to link the parallelism. cruentatum group with concolor. The latter, along with erythrorhynchos, is CHARACTER GEOGRAPHY IN THE cruentatum perhaps an Indian equivalent or repre- SPECIES GROUP sentative of some of the more brightly In the brightly colored species of Di- colored island species. The cruentatum caeum, the colorpattern oftenchanges rather group is basically one with a bright and abruptly in closely related forms. This varied color pattern. A suitable type )f is true both of the cruentatum species group ancestral stock among existing species is and of others already mentioned. Perhaps provided to some extent by the papuense- this means that some of these color char- retrocinctum superspecies of the Philippines acters are controlled by a few or single which supplies a transition from the heav- genetic factors. In several instances, ier-billed Anaimos type of species to the however, it is obvious that red marks are more advanced ones, although the re- only partially or dully indicated, suggesting lationship is not close. a more complex genetic basis. This type Turning to the interrelationships of the of abrupt variation is favored by the in- members of the cruentatum group, the four sular distribution of many of these forms. species of the Sunda Islands (cruentatum, The bright varied coloration of the an- trochileum, igniferum, maugei) are more cestral Anaimos demonstrates that such differentiated than the others, and it is patterns were already present when the here that the only instance of overlapping species of Dicaeum, such as the cruentatum range in the group occurs (Lombok). This group, evolved. Some of the plainer spe- suggests that these species are older than cies of Dicaeum are not primitive in this the Moluccan-Papuan ones. On the respect, but have lost the bright coloration other hand, the rather specialized char- of their relatives secondarily. This is acters of the Sunda species (such as the obviously true of tristrami and probably red back) do not make them likely im- of pygmaeum and erythrorhynchos. mediate ancestors of the somewhat simpler In the above list of species some use Moluccan and Papuan species. D. nehr- was made of character analysis, as in point- 24 AMERICAN MUSEUM NOVITATES [No. 1360 ing out the closer relationship of vulneratum breast is characteristic of all the closely than of erythrothorax to nehrkorni. The related species extending from Celebes geographical variation in color characters (nehrkorni) eastward through the Mol- is more fully analyzed here (table 1). uccas and the New Guinea region to the RUMP: In most species of the cruentatum Solomon Islands, except for the aberrant group the rump is red (in both sexes), but tristrami. This red breast spot may be a four species erythrothorax, pectorale, aeneum transmutation of the red abdominal stripe and tristrami, are exceptions. The rump found in the superspecies D. papuense. is still yellowish in erythrothorax, suggesting This assumption is strengthened by the recent loss. In pectorale, which seems to fact that the red spot is continued in nehr- to have been derived from it, the rump is korni and eximium by a dark abdominal no longer yellow. In aeneum of the Sol- stripe. A red spot on the lower throat is omon Islands the loss of the red rump was also found in some species of the Malay- probably independent of that in pectorale, sian branch of the cruentatum group (ig- since in the intervening area two red- niferum-maugei), but is lost in the more rumped species, geelvinkianum and eximium, specialized species cruentatum and tro- occur. The fact that the rump is not red chileum. The red throat of trochileum seems in tristrami might be interpreted to mean not to be homologous with the red breast that it was derived from the other Solomons spot of the other species, but rather due to species, aeneum, but other features of the the encroachment of the red of the crown coloration indicate that tristrami may have around the sides of the head onto the been derived from the Bismarck Archi- throat. pelago (eximium) and aeneum from New The accompanying chart (table 1) sum- Guinea or the Louisiades (geelvinkianum). marizes the variation of the more im- CROWN AND BACK: A red crown is less portant color characters in the group.

TABLE 1 CHARACTER VARIATION IN Dicaeum nehrkorni AND RELATIVES Males Belly Stripe Red Breast Patch Crown Back Rump nehrkorni Black Small Red Black, blue gloss Red vulneratum - Medium Gray Gray Red erythrothorax - Large Gray Olive Olive pectorale Medium Olive Olive Olive geelvinkianum Small Red Blue gloss Red nitidum Medium Red Green gloss Red aeneum Large Gray Gray, green gloss Gray eximium Brown Medium Reddish brown Brown Red tristrami Mottled brown Brown Brown characteristic of the cruentatum group than THE SUPERSPECIES a red rump, occurring in fewer species and Dicaeum hirundinaceum being absent in the female when present in The species of this superspecies are very the male. It occurs in igniferum, cruen- similar to some of the species of the cru- tatum, and trochileum, and in these the back entatum group, and there can be no ques- is also more or less red. They are undoubt- tion of a very close relationship. However, edly closely allied. The crown is also both groups occur together on Celebes and red in males of nehrkorni. The red crown in most of the Greater and Lesser Sunda has been lost in the two Moluccan species Islands, although they are usually altitudi- and in pectorale of the western New Guinea nally separated. The greatest similarity Islands but reappears in geelvinkianum and exists between the species maugei and san- eximium, though reduced in the latter. guinolentum which represent the two groups- It is entirely lost again in the two Solomon on the Lesser Sunda Islands, and scarcely Islands species. two authors have agreed as to how to asso- BREAST AND THORAX: A red spot on the ciate these species with the species of the 1947] A REVIEW OF THE DICAEIDAE 25

CSi

r

- -'p- -v 49 i

0

el U,

0. 26 AMERICAN MUSEUM NOVITATES [No. 1360 neighboring islands. A close study suggests Bali), rhodopygiale (Flores), wilhelminae that maugei or the form ancestral to maugei (Sumba), hanieli (Timor). Robinson and gave rise to the hirundinaceum group, but Kloss described a race ablutum from eastern the exact stages in which this speciation oc- Java and Bali, of which the most striking curred were subsequently obscured by character was the supposed absence of the reinvasions and range changes. The most red rump in the female. Others (Chasen, probable hypothesis is that the hirundin- 1940, p. 265) surmised that an immature aceum group originated by the colonization male may have been mistakenly considered of the western Lesser Sunda Islands (giving a female. This seems to be the case, as the rise to sanguinolentum) by a maugei-like only female from the range of ablutum stock coming from the Timor group. examined by us (Bali, Stresemann col- Subsequent range expansions of sanguino- lector) has the rump as red as in any race of lentum and celebicum and of descendants of sanguinolentum. Chasen thought ablutum these forms has led to the present dis- valid on the basis of other characters, but tribution. The cruentatum and hirundin- our material, which includes four males aceum groups still replace each other geo- from Bali, makes this appear doubtful. graphically on most smaller islands, espe- The upperparts of the male are glossed cially in the area from Celebes and the with dark bluish purple varying slightly Lesser Sundas eastward. The hirundin- in tone among the races. The underparts aceum group is considered to be the younger are buffy white with a large red pectoral spot becauseofthe lesser degree of differentiation in s. sanguinolentum. D. s. rhodopygiale is among its members and, in part, because similar, but the breast spot is paler. In of its distribution (fig. 7). hanieli the red areais smaller, and the under- The only constant character separating parts are whiter. In all these races there the hirundinaceum and cruentatum groups is a small indistinct blackish area posterior is the fact that the rump is never red in to the red breast. In wilhelminae, the males of the former group. It is some- breast spot is large, extending to the chin, times red in the female, especially in san- and the mid-belly and sides are extensively guinolentum which is nearest to the cruen- blackish. The females of all races are tatum group. Even this is not a very grayish or buffy white below and olive gray satisfactory distinction, as the red rump above with a red rump. The bill is has been lost in a few Moluccan and Papuan small in sanguinolentum and rhodopygiale, species of the cruentatum group. The mediu-m in hanieli, and stout in wilhel- upperparts in males of hirundinaceum and minae. The Flores race is nearer to the allies are in general more glossy than in Javan one in all particulars than is that of those of the other group, and there are Sumba. This is a mountain species. other minor points which help in separating 31. D. hirundinaceum. This we con- them. Here, too, the difference is least in sider a separate species because the under the Lesser Sunda forms of maugei and san- tail coverts are red or pink in both sexes guinolentum. in all four races. The rump is no longer It is difficult to divide the hirundinaceum red in the female of h. hirundinaceum of group satisfactorily into species. Deignan Australia, but is still tinged with red in the (1945, p. 548) has recently listed them three other races: ignicolle (Aru Islands), all as races of hirundinaceum, although keiense (Kei Islafids), fulgidum (Tenimber he confusingly uses the term superspecies. Islands). The back in the male is similar Since, in several other instances, forms of to that of sanguinolentum, but the gloss is Dicaeum as similar as sanguinolentum, a little more bluish. Ventrally there is a hirundinaceum, and celebicum occur to- great resemblance to D. s. wilhelminae of gether as good species, we prefer to recog- Sumba, from which hirundinaceum was nize the following four groups of forms as apparently derived, though the breast is species: redder, less orange. The bill is quite 30. D. sanguinolentum. There are four heavy as in wilhelminae. In this instance races, as follows: sanguinolentum (Java and there is little doubt that Australia was 19471 A REVIEW OF THE DICAEIDAE 27

HIRUNDINACEUM GROUP

celebicum

CRUENTATUM GROUP

papdense Fig. 8. Diagrammatic phylogeny of the cruentatum and hirundinaceum groups. 28 AMERICAN MUSEUM NOVITA TES [No. 1360 colonized from Sumba, not from Timor being the most distinct member of the which is more usual (Mayr, 1944). superspecies in this respect. The rump is Dicaeum h. hirundinaceum is the com- not red in the female. The derivation of mon Mistletoe Bird of Australia, generally this species is somewhat puzzling. The and commonly distributed, and, as would be Philippine races bear a strong resem- expected, is mainly not a mountain bird. blance ventrally to monticola, but it seems Mathews has described a number of Aus- rather unlikely that the species originated tralian races now in need of scouting, of in the Philippines from monticola (Borneo) which perhaps yorki of northern Australia stock, then spread north through Formosa is valid. to China and back through southeastern 32. D. celebicum. This is kept spe- Asia to Sumatra. On the other hand, the cifically distinct from sanguinolentum and Sumatran and Malayan races resemble hirundinaceum because the rump is not red D. sanguinolentum ventrally. There is no in the female and the under tail coverts trace of the red rump of females of sangui- are never red. Although the back is dark nolentum in any form of ignipectus, while purple in the Celebes race, some of the the color of the back is also very different. others on islands near Celebes are much The Philippine and Formosan sub- bluer, almost like monticola of Borneo. species have an extensive red area on the As the latter form does not differ in any breast and throat; in ignipectus and doli- significant respect, we keep it in celebicum. chorhynchum it is smaller and in cambo- Most of the races occur both in moun- dianum (not seen) and beccarii it has faded tains and lowland, but monticola only in to cinnamon buff. Females are like those mountains. of monticola but even greener and lighter Dicaeum celebicum is closely related to above. This species ranges as high as sanguinolentum. The distance between 12,000 feet in parts of Asia, but on Luzon the races of the former occurring in the ranges down almost to the lowlands. Lesser Sundas and celebicum is not great. D. celebicum kuhni of the Tukang Besi SUMMARY OF hirundinaceum GROUP Islands is very much lrke D. sanguino- This group of four closely related allo- lentum wilhelminae in color (except for the patric species probably was derived from red rump of the female in the latter), but the cruentatum'species group in the general the shape of the bill is more as in D. s. region of the Lesser Sunda Islands, where rhodopygiale of Flores. The females of D. sanguinolentum bears a strong resem- most races of celebicum are glossier than in blance to D. maugei of the cruentatumgroup. the allied species, while the males tend to The hirundinaceum group is characterized be darker and suffused with black or gray by the absence of a red rump in the males ventrally. and by a tendency to lose it in the females. The races of celebicum are: kiuhni (Tu- This mark still exists in the females of D. kang Besi Islands), sulaense (Sula Islands), sanguinolentum, it is either absent or celebicum (Celebes, Bditon, Muna), san- present, but pale, in the various subspecies ghirense (Sangir Islands), talautense of D. hirundinaceum of Australia and the (Talaut Islands), and monticola (Mts. nearby islands, and is absent in the females Kinabalu and Dulit, Borneo). of the other two species. D. hirundin- 33. D. ignipectus. The races and dis- aceum resulted from colonization of Aus- tributtion are as follows: beccarii (northern tralia by sanguinolentum stock from the Sumatra); cambodianum (Cambodia, Indo- Lesser Sundas, probably Sumba. D. China); dolichorhynchum (peninsular Siam celebicum of Celebes, Borneo, and ad- and Malaya); ignipectus (Indo-Chinese joining islands also seems to have been countries, southern China, Himalayas); derived from sanguinolentum of the Lesser formosanum (Formosa); luzoniense (Lu- Sundas. D. ignipectus is more distinct, but zon); bonga (Samar); apo (Mt. Apo, seems closest to sanguinolentum, though Mindanao). In this species the upper- in some races it is similar to D. celebicum parts are glossed with greenish blue, it monticola. 1947] A REVIEW OF THE DICAEIDAE 29

Aside from the disappearance of the red out chestnut and with the feathers mar- rump, the other color variation within gined with black. the superspecies does not exhibit any very The tail is rather short and the general pronounced trends, but several of the appearance like that of a titmouse (Parus) characters are useful in trying to deter- in which genus the species was first de- mine the interrelationships of the species, scribed. There are hair-like filoplumes on as shown above. The bill varies con- the napes and flanks, a character tending siderably within the species sanguinolentum. to ally it with the Dicaeidae, as similar In the eastern part of their ranges, the feathers are found in some species of An- cruentatum and hirundinaceum groups re- aimos and Dicaeum. Tarsus booted except place each other (except in the Lesser basally. Sundas). D. hirundinaceum has reached Australia but avoided the Moluccas and PARAMYTHIA DE VIS (MONTIUM) Papua where the other group is present. As in Oreocharis there is only one species. The two groups coexist in the larger west- It is endemic to New Guinea. The ern islands and on the mainland-the systematic position of Paramythia had hirundinaceum group usually at higher been uncertain until Mayr (1933) showed altitudes. In Celebes this is reversed, that it resembles Oreocharis in so many and nehrkorni of the cruentatum group respects that there can be no doubt it is a ranges at higher levels, while celebicum specialized relative of that genus. Simi- occurs both in the lowlands and mountains. larity between the two had been casually indicated by one or two earlier authors. ABERRANT GENERA DIAGNOSIS: Similar to Oreocharis but The three remaining genera of the Dicaei- much larger and with a full nuchal crest of dae are aberrant. They appear to be special- pointed feathers; tail relatively longer. ized offshoots of the family and seem prop- In both sexes, which are colored alike, the erly assigned to it. In all three the tenth head is black as in Oreocharis, but the primary is vestigial and the bill is not bases of the crown feathers are white (a serrated. Two of them, Oreocharis -and character suggested in the crown feathers Paramythia, are monotypic and restricted of Oreocharis). The dull bluish coloration to New Guinea. They seem to be strictly of the tail of Oreocharis has extended in fruit eaters which suggests that they are Paramythia until it occupies the upper back distantly related to the fruit-eating primi- and most of the underparts except the tive New Guinea genera, Melanocharis black throat and a small area on the belly and Rhamphocharis. The third genus, and under tail coverts which are yellow Pardalotus, of Australia and Tasmania, as in Oreocharis. The back is similar to though insectivorous, appears distantly Oreocharis but duller. In immatures of related to Oreocharis. Paramythia the breast feathers are tipped with yellow, suggesting a previous more OREOCHARIS SALVADORI (ARFAKI) extensive distribution of that color. There DIAGNOSIS: Tenth primary vestigial. are long filoplumes on nape and flanks. Bill much as in Melanocharis but not Tarsus booted except basally. serrated. The male has the head and throat glossy black with bright yellow PARDALOTUS VIEILLOT (PUNCTATUS) cheek patches; underparts bright yellow SYNONYMS: Dipardalotus Mathews with a medial chestnut brown area; back (rubricatus yorki); Nesopardalotus Math- olive green; flight feathers dark with yel- ews (quadragintus); Pardalotinus Math- low patches near the tips of the secondaries ews (striatus). Mathews later synony- on the outer vanes; tail dark washed with mized these genera himself. blue. In the female the throat and cheeks The genus Pardalotus is rather aberrant are gray; the upperparts similar to the both structurally and in habits. It agrees male but the crown also green; the under- with Oreocharis in lacking serrations on the parts duller yellow than in the male, with- bill and in the vestigial tenth primary. 30 AMERICAN MUSEUM NOVITATES [NSo. 1360

Some of the species, such as P. punctatus, wing. The proportions are the same; there have terminal spots on the outer vanes of is a tendency for a light-colored rump and the secondaries, very suggestive of those of for white tail tips. Oreocharis. In addition they have many 1. Pardalotus punctatus. Eastern, other spots and streaks, producing a rather south, and southwest Australia, Tasmania. variegated pattern and suggesting the The arid country form xanthopygius names Pardalotus and Diamond Bird. appears to be a race of this species. Male They are stumpy, short-tailed, and short- and female are strikingly distinct in this billed little birds and thus resemble to some species. The' superciliary is indistinct. extent their rather distant relatives, 2. Pardalotus quadragintus. Tasmania. Dicaeum. The internal anatomy has ap- Male and female are hen feathered and parently not yet been studied. rather similar to the female of punctatus. DIAGNOSIS: Bill not serrated; short, This is probably another case of double deep, with strongly arched culmen and invasion on Tasmania, with quadragintus sometimes slightly compressed (modified the first, and punctatus the second coloni- for burrowing). Tenth primary vestigial. zation. The primitive color of this species Small, short-tailed, rather long-winged appears to be a secondary simplification. birds, similar to Dicaeum in proportions 3. Pardalotus rubricatus. North and but averaging somewhat larger. Sandy- central Australia to New South Wales. colored birds, often with bright yellow, This arid country species leads in some especially on the throat and sometimes respects from the spotted to the striped tiny yellow or red marks on the wing pardalotes. The superciliary with the coverts; crown, wings, and' tail usually loral spot (reddish in this species) is, how- black, much spotted and streaked with ever, typical for the striped pardalotes. white; feet and claws stout. The tarsus There is no sexual dimorphism. appears booted in dried skins, but this 4. Pardalotus striatus may be deceptive. 5. Pardalotus ornatus. 6. Pardalotus substriatus. The differ- ences, the geographical distribution, and THE SPECIES OF Pardalotus the reasons for the specific distinctness of The seven species of Pardalotus can be these three species have been discussed divided into two groups, the spotted-wing recently by Hindwood and Mayr (1946). group (punctatus and quadragintus), and 7. Pardalotus melanocephalus. This the striped-wing group (the other five more or less tropical species with the plain species). Since the spotted-wing pattern black crown is very- close to the other is very similar to that of Oreocharis arfaki three striped pardalotes, particularly to (New Guinea) it seems likely that this is substriatus. It ranges from northwest the more primitive pattern. The two Australia through northern Australia into groups are quite similar, except for the northern New South Wales. REFERENCES ALI, SALIM A. BAKER, E. C. STUART 1931. The role of sunbirds and flowerpeckers 1934. The nidification of birds of the Indian in the propagation and distribution of Empire. London, Taylor and Francis, the tree-parasite, Loranthus longiflorus vol. 3, vii + 568 pp. Dest., in the Konkan (W. India). CHANDLER, L. G. Jour. Bombay Nat. Hist. Soc., vol. 35, 1910. Notes on pardalotes. Emu, vol. 10, pp. 114-149; pp. 113-118. 1936. The ornithology of Travancore and 1912. Notes on the mistletoe-bird (Dicaeum Cochin. Part 5. Jour. Bombay Nat. hirundinaceum). Ibid., vol. 12, pp. Hist. Soc., vol. 38, pp. 759-790. 130-133. ALI, SALIM, AND HUMAYUN ABDULALI 1938. The birds of Bombay and Salsette. CHASEN, F. N. Part 4. Jour. Bombay Nat. Hist. 1939. In Robinson, H. C., and F. N. Chasen, Soc., vol. 40, pp. 148-173. The birds of the Malay Peninsula. 19471 A REVIEW OF THE DICAEIDAE 31

London, H. F. and G. Witherby, Ltd., nochilus from the highlands of Luzon. vol. 4, xiii + 485 pp. Philippine Jour. Sci., vol. 9, ser. D, p. 1940. Notes on some Javan birds. Treubia, 531. vol. 17, pp. 263-266. 1927. New or noteworthy Philippine birds, 5. COOMANS DR RUITER, L. Philippine Jour. Sci., vol. 32, pp. 513- 1936. Het goed verborgen nest van den 525. Geelromp Bastaard-Honingvogel MAYR, ERNST (Anaimos xanthopygius [Salv.]). De 1931. Die V6gel des Saruwaged- und Herzog- Tropische Natuur: Jubileum- gebirges (NO-Neuguinea). Mitt. Uitgave, pp. 132-133. Zool. Mus. Berlin, vol. 17, pp. 639-723. DEIGNAN, HERBERT G. 1933. Zur systematischen Stellung von Para- 1943. Two preoccupied names of Oriental mythia De Vis. Ornith. Monatsber., birds. Proc. Biol. Soc. Washington, vol. 41, pp. 112-113. vol. 56, p. 70. 1941. List of New Guinea birds. New York, 1945. The birds of northern Thailand. Bull. American Museum of Natural History, U. S. Natl. Mus., no. 186, v + 616 pp. xi + 260 pp. DELACOUR, JEAN 1944. Timor and the colonization of Aus- 1944. A revision of the family Nectariniidae tralia by birds. Emu, vol. 44, pp. (sunbirds). Zoologica, vol. 29, pp. 113-130. 17-38. * 1945. In Delacour, Jean, and Ernst Mayr, 1946. Notes on the of the birds of Notes on the taxonomy of the birds of Malaysia. Ibid., vol. 31, pp. 1-8. the Philippines. Zoologica, vol. 30, 1947. Birds of Malaysia. New York, The pp. 105-117. Macmillan Co., xvi + 381 pp. 1946. A new name for a Philippine flower- DELACOUR, JEAN, AND ERNST MAYR pecker. Ibid., vol. 31, p. 8. 1946. Birds of the Philippines. New York, MEISE, WILHELM The Macmillan Co., xv + 309 pp. 1941. Ueber die Vogelwelt von Noesa Penida DESSELBERGER, HERMANN bei Bali. Jour. fulr Ornith., vol. 89, 1931. Der Verdauungskanal der Dicaeiden pp. 345-376. nach Gestalt and Funktion. Jour. fur OBERHOLSER, HARRY C. Ornith., vol. 79, pp. 353-370. 1923. A review of the genus Prionochilus GADOW, HANS Strickland and its closest allies. Ohio 1890-1899. In Wilson, Scott B., and A. H. Jour. Sci., vol. 23, pp. 287-294. Evans, Aves Hawaiiensis: The birds PORSCH, OTrO of the Sandwich Islands. London, A. 1929. Kritische Quellenstudien uber Blumen- H. Porter, xxv + 257 pp. besuch durch Vogel, 4. Biologia GARDNER, LEON L. Generalis, vol. 5, pp. 157-210. 1925. The adaptive modifications and the RAND, A. L. taxonomic value of the tongue in birds. 1937. In Mayr, Ernst, and A. L. Rand, Re- Proc. U. S. Natl. Mus., vol. 67, pp. 1- sults of the Archbold expeditions. No. 49. 14. Birds of the 1933-1934 Papuan ex- HACHISUKA, MASAUJI pedition. Bull. Amer. Mus. Nat. 1941. Description of a new species of bird Hist., vol. 73, pp. 1-248. from the Philippine Islands. Bull. 1942. Results of the Archbold expeditions. Biogeogr. Soc. Japan, vol. 11, p. 1. No. 43. Birds of the 1938-1939 New HILL, G. F. Guinea expedition. Ibid., vol. 79, pp. 1911. Field notes on the birds of Kimberley, 425-516. north-west Australia. Emu, vol. 10, RENSCH, BERNHARD pp. 258-290. 1931. Die Vogelwelt von Lombok, Sumbawa HINDWOOD, K. A., AND ERNST MAYR and Flores. Mitt. Zool. Mus. Berlin, 1946. A revision of the striped-crowned vol. 17, pp. 451-637. pardalotes. Emu, vol. 46, pp. 49-67. ROBERTS, N. L. LAWRENCE, S. A.,AND R. T. LITTLEJOHNS 1937. The red-tipped pardalote. Emu, vol. 1916. Nesting habits of the mistle-toe bird 37, p. 102. (Dicaeum hirundinaceum). Emu, vol. SCLATER, P. L. 15, pp. 166-169. 1874. On the Prionochili of British India. LUSHINGTON, CICELY Ibis, pp. 1-3, pl. 1. 1940. Notes on the nesting of Legge's flower- SHARPE, R. BOWDLER pecker (Acmonorhynchus vincens 1885. Catalogue of the birds in the British [Sclater]). Jour. Bombay Nat. Hist. Museum. London, vol. 10, xiii + 682 Soc., vol. 42, pp. 186--187. PP. MCGREGOR, RICHARD C. STEINBACHER, GEoRG 1914. Description of a new species of Prio- 1935. Ueber den Bau des Magens von 32 AMERICAN MUSEUM NOVITATES [No. 1360

EBuphonia. Ornith. Monatsb., vol. 43, WHISTLER, HUGH pp. 41-45. 1934. The Vernay scientific survey of the WETMAPE, ALEXANDER Eastern Ghats. Part 8. Jour. Bom- 1914. 'the development of the stomach in the bay Nat. Hist. Soc., vol. 37, pp. 281- euphonias. Auk, vol. 31, pp. 458-461. 297.