Marc Thibault et al. 83 Bull. B.O.C. 2013 133(2) New and interesting records for the Obi archipelago (north Maluku, Indonesia), including field observations and first description of the vocalisation of Moluccan Woodcock Scolopax rochussenii by Marc Thibault, Pierre Defos du Ran, Olivier Pineau & Wesley Pangimangen Received 6 October 2012 Summary. —The avifauna of the Obi archipelago is rather poorly studied and current understanding is essentially based on several historic collecting efforts and few recent visits by modern ornithologists, none of which reached the mountains above 750 m. Furthermore, the taxonomic position of many bird populations restricted to the archipelago or shared with nearby Bacan Island remains confused. We describe the results of a two-week avifaunal survey of Obi in March 2010. We provide first records since 1982 of the poorly known Moluccan Woodcock Scolopax rochussenii as well as the first description of its vocalisation and first information on its habitat, which, contrary to what was previously speculated, includes lowland forest. We also provide confirmed records of five taxa previously unknown on the island, including one that possibly represents a new subspecies (Sulawesi Myzomela Myzomela chloroptera). New elevational information is presented for 34 species. Comments on the taxonomy of several endemic taxa are made on the basis of new vocalisation or photographic material, suggesting that at least two deserve biological species status (Northern Golden Bulbul Thapsinillas longirostris lucasi, Dusky Myzomela Myzomela obscura rubrotincta) and offering further support to treat Cinnamon-breasted Whistler Pachycephala johni as a species. Finally, we emphasise the need for taxonomic reappraisal of several other endemic insular forms (Cinnamon-bellied Imperial Pigeon Dncula basilica obiensis, Violet¬ necked Lory Eos squamata obiensis, Red-cheeked Parrot Geoffroyus geoffroyi obiensis, Hair-crested Drongo Dicrurus hottentotus guillemardi, Northern Fantail Rhipidura rufiventris obiensis, Paradise Crow Lycocorax pyrrhopterus obiensis and Island Leaf Warbler Phylloscopus poliocephalns waterstradti). Although most of our observations await confirmation, they suggest that endemism on Obi at the species level is perhaps significantly under-estimated. Obi and its satellite islands of Bisa, Tapat, Obilatu, Gomumu and Tobala form a remote oceanic archipelago in Maluku, 33 km south of Bacan, 100 km east of Mangole (Sula Islands) and 125 km north of Seram (Fig. 1). Total surface area is 3,040 km2 with Obi (2,670 km2) the largest island, predominantly hilly and rising to at least 1,558 m. Together with the three large islands of Morotai, Halmahera and Bacan, and several smaller islands, the Obi archipelago forms part of the Northern Maluku Endemic Bird Area (EBA 171) distinguished by 32 species and 101 subspecies restricted to this region (following Gill & Donsker 2012). To date, the resident avifauna of the Obi islands group is known to support 16 Northern Maluku endemic species shared with both Bacan and Halmahera. Only two endemics currently recognised at species level are confined to Obi (Carunculated Fruit Dove Ptilinopus granulifrons and Cinnamon-breasted Whistler Pachycephala johni), but the archipelago also hosts 19 recognised endemic subspecies, e.g. Paradise Crow Lycocorax © 2013 The Authors; Journal compilation © 2013 British Ornithologists' Club Marc Thibault et al. 84 Bull. B.O.C. 2013 133(2) Figure 1. Location of Obi in eastern Indonesia and localities visited or mentioned in the text. pyrrhopterus obiensis, while the enigmatic Moluccan Woodcock Scolopax rochussenii and two subspecies (Chattering Lory Lorius garrulus flavopalliatus and Island Leaf Warbler Phylloscopus poliocephalus zvaterstradti) are shared only with Bacan and Kasiruta (White & Bruce 1986). Interestingly, the Obi islands also host endemic subspecies of several residents that, according to current knowledge, do not occur elsewhere in northern Maluku: Pale Cicadabird Coracina ceramensis hoogerwerfi, Northern Fantail Rhipidura rufiventris obiensis, Turquoise Flycatcher Eumyias panayensis obiensis and Hair-crested Drongo Dicrurus hottentotus guillemardi. Despite its distinctive avifauna but as a result of its remote situation, Obi has received little attention from modern ornithologists. Bernstein was the first bird collector to visit the island in 1862 (Jansen 2008) followed by Guillemard in 1883, Doherty in 1897 (the latter also visiting Bisa), Lucas in 1898 and Waterstradt in 1902 (Hartert 1903a). Following a gap of 50 years, an expedition from the Museum Zoologicum Bogoriense was conducted under the leadership of Wegner in July-November 1953 (Mees 1982) and its results published by Jany (1954, 1955). Further specimens were collected in 1981-83 by Momou and Tatu, during expeditions organised by the Smithonian Institution with the assistance of the Indonesian Nature Conservancy, as part of ethnobiological research conducted by Taylor (1990). These specimens were deposited at the National Museum for Natural History (NMNH), Washington DC. In December 1989 Linsley (1995) conducted a bird survey as part of a collaborative project between Manchester Metropolitan University, the Indonesian Institute of Sciences (LIPI) and the Directorate-General for Forest Protection and Nature Conservation (PHPA). He observed nine species not previously reported on Obi and provided the first ecological information for many taxa and descriptions of the vocalisations of several endemic forms. A brief exploration of Bisa was conducted on 11 June 1990 by Bishop (1992). Further ornithological observations were made in February 1992 by Lambert (1994) during a one-month survey of parrots under the auspices of IUCN; eight species were recorded on Obi for the first time. While maximum elevations reached by the earlier collectors are undocumented, Linsley (1995) and Lambert (1994) managed to reach c.500 m and 730 m, respectively. The basic ecology of a large suite of species remains poorly known and the rate of new island records hints that the avifauna is still poorly documented. This, together with the paucity of records for the enigmatic Moluccan Woodcock, encouraged us to undertake a short expedition to the island with the aim of improving knowledge of its avifauna. © 2013 The Authors; Journal compilation © 2013 British Ornithologists' Club Marc Thibault et al. 85 Bull. B.O.C. 2013 133(2) Study area and Methods On 7-21 March 2010, observations were made at ten sites on Obi. Details of these, survey dates and habitats are provided in Table 1 and Fig. 1. As many days as possible were spent above 800 m (10-16 March) and we eventually reached forest between 1,000 and 1,220 m on 13-16 March. Elevations and geographical coordinates were measured using a Garmin eTrex Vista HCx GPS. Subsequently, elevations and geographical data collected in the field were compared with satellite data available via Google Earth and some minor elevation corrections were made. It is of note that the highest elevation of 1,558 m given for Obi by Google Earth satellite data contradicts both the US Army map SA-52 1:1,000,000 (1968) and that in Coates & Bishop (1997), where 1,611 m is given as the highest elevation. To support identifications and descriptions, MT took photographs using a Canon 40D digital camera with a 100-400-mm lens. Sound-recordings were made by MT using an Edirol R09-HR digital recorder with inbuilt microphone. Sonograms were prepared using Syrinx 2.6h by John Burt (available at www.syrinxpc.com). Recordings of Moluccan Woodcock have been uploaded to www.xeno-canto.org and further recordings will be uploaded to this online database. Recent state of natural habitats on Obi F. G. Rozendaal in Collar & Andrew (1988), Linsley (1995) and Vetter (2009) already reported that much of the lower lying forest on Obi was logged during the 1980s and early 1990s, while much of the remainder is under logging concession. According to Vetter (2009), Obi has suffered some of the largest reductions of lowland forest within north Maluku as a result of plantation development. In addition, illegal gold mining was reported to be destroying some areas of forest (F. G. Rozendaal in Collar & Andrew 1988). Three logging companies, PT Poleko Yurbarson Trad., PT Telaga Bhakti Persada and PT Pusaka Agro Sejahtera, currently run six large logging licences covering two-thirds of Obi. These licences generally permit selective logging up to c.100 m3 / ha and for c.20-30 years (N. Brickie pers. comm.). Along the 70-km stretch of coast between Jikotamo and the TABLE 1 Study site details, survey dates and habitat. See also Fig. 1. Location (elevation) Coordinates Dates Habitat Soligi (sea level) 01°39'S, 127°25'E 7-8, 20 March 2010 Village; coconut plantation and other tree crops; mangrove PT Telaga Bhakti Persada 01°42'S, 127°29'E 19 March 2010 Village; coconut plantation; grassland logging settlement (sea level) Site A (200-300 m) 01°38'S, 127°30'E 19 March 2010 Logged forest Site B (360-450 m) 01°36'S, 127°33'E 8,17-19 March 2010 Logged forest Site C (400-700 m) 01°33'S, 127°34'E 8, 9,16-17 March 2010 Logged forest Site D (700-850 m) 01°32'S, 127°35'E 9-13, 16 March 2010 Logged and primary forest Site E (850-1,220 m) 01°32'S, 127°37'E 13-16 March 2010 Logged and primary forest including montane forest Laiwui (sea level) 01°19'S, 127°38'E 20-21 March 2010 Village; coconut plantation; beach Jikotamo (sea level) 01°20'S, 127°39'E 20-21 March 2010 Village Site F (50-150 m) 01°21'S, 127°40'E 20-21 March 2010 Logged forest; regrowth forest; coconut plantation; grassland © 2013 The Authors; Journal compilation © 2013 British Ornithologists' Club Marc Thibault et al. 86 Bull. B.O.C. 2013 133(2) PT Telaga Bhakti Persada logging settlement south of Soligi, hills were mainly covered by plantations (including cloves, coconut palm, nutmeg, ambarella and cacao) and fragmented secondary forest. Hills separating Manis Lake from the west coast were barren as a result of ongoing open-cast nickel mining. Remnant primary forest was mainly restricted to the most rugged and inaccessible terrain. Alluvial flats were extensively covered by coconut plantations, with the notable exception of a relatively large tract of coastal forest bordered by old-growth mangrove immediately north of Soligi. Within the PT Telaga Bhakti Persada logging concession south and east of Soligi, forest was very degraded below 200 m, large areas had already been selectively logged up to 950 m and the lowest patch of primary forest was at 500-700 m, although some large trees there were marked for logging. At 700-950 m, forest was in better condition, with fragments of primary forest still covering some of the steeper, less accessible slopes, while the more accessible areas were mainly covered by selectively logged forest. A small area of tropical forest explored at 1,100-1,220 m showed no evidence of logging. Soligi villagers reported that tree-plantation operations were planned by the logging company and a tree nursery was seen in the concession. Several drill holes excavated during recent mining exploration were found well inside the logging concession up to 600 m. According to some villagers, these have to date not resulted in any further mining. It is of note that in the north of the island, a large-scale drilling project was withdrawn in September 2011. It was initially set to target a large epithermal gold / silver / base metals system under a license covering 7,700 ha, which is already the subject of extensive artisanal gold mining (Ashburton Minerals Ltd. 2012). To our knowledge, Obi lacks any protected area, although a nature reserve of 450 km2 covering highlands in the centre of the island and further protection forests on steep terrain have been proposed (FAO 1981-82, BirdLife International 2013). Species accounts Overall we recorded 74 bird species, of which 66 are resident or presumed so and eight are Palearctic migrants. Three additional taxa are either unconfirmed or remain unidentified to species. Tire following details new island records, new elevational or noteworthy records from a distributional, ecological or taxonomic perspective. Other species encountered during our survey are listed in Table 2. Taxonomy and nomenclature follow Gill & Donsker (2012) except that we use Common Golden Whistler Pachycephala pectoralis instead of Black- chinned Whistler P. mentalis. MOLUCCAN WOODCOCK Scolopax rochussenii The following records were obtained at site B at c.420 m, from a logging track bisecting an extensive area of logged forest on flat and more graded sandstone soils: at c.19.00 h on 17 March 2010, calls reminiscent to those given in display-flight by Bukidnon Woodcock S. bukidnensis were briefly heard (OP, MT & PDR). Between 02.00 h and 04.00 h on 18 March 2010, similar calls were repeatedly heard by OP and subsequently at 05.55 h and 06.07 h by the three observers, when sound-recordings (www.xeno-canto.org/109282, www.xeno- canto.org/109283) and a glimpse of a fast-flying bird were eventually obtained at dawn by MT. At 07.00-07.15 h (before dusk) on the same day, what was presumably the same bird was observed three times by MT, when it called in roding (i.e. courtship) display-flight at the same location. It was flying rather straight, rapidly and low above the canopy, sometimes <100 m from the observer, and was briefly seen opening its bill when calling. Rock Dove Columba livia-like size, overall compact stocky shape, with a rounded bulky chest, short tail, broad pointed wings and a very long straight bill were noted. These characters, together © 2013 The Authors; Journal compilation © 2013 British Ornithologists' Club Marc Thibault et al. 87 Bull. B.O.C. 2013 133(2) TABLE 2 Species encountered on Obi not mentioned in the main text. Species name Comments Dusky Megapode Megapodius fmjcinet 1 at c.150 m Pacific Reef Heron Egretta sacra sacra 1 on the coast Striated Heron Butorides striata 1 on the coast Eastern Cattle Egret Bubulcus coromandus few seen on the coastal plain Eastern Osprey Pandion cristatus singles seen twice near sea level; presumably resident Pacific Baza Aviceda subcristata rufa 1 near sea level Brahminy Kite Haliastur indus quite common from sea level to 200 m, with additional records to 1,200 m White-bellied Sea Eagle Haliaeetus leucogaster 4 on coast Oriental Hobby Falco severus 1 at Soligi Spotted Kestrel Falco moluccensis common in coastal plain Common Sandpiper Actitis hypoleucos 3 near Soligi Red-necked Phalarope Phalaropus lobatus common in coastal seas, with 600-1,000 from boat between Soligi and Jikotamo Greater Crested Tern Sterna bergii c.10 seen inshore between Soligi and Jikotamo Pied Imperial Pigeon Ducula bicolor 2 in coconut / coastal forest near Soligi Great-billed Parrot Tanygnathus megalorhynchos 2 near sea level in degraded forest and coconut plantation near Soligi, 1 heard near Laiwui Barking Owl Ninox connivens rufostrigata 1 photographed at site F at c.100 m Large-tailed Nightjar Caprimulgus macrurus schlegelii seen twice at 400-500 m Glossy Swiftlet Collocalia esculenta from sea level to 940 m Moustached Treeswift Hemiprocne mystacea confirmata three records at 300 m, 480 m and 600 m. Common Paradise Kingfisher Tanysiptera galatea obiensis 1 seen at site F Blue-and-white Kingfisher Todiramphus diops Beach Kingfisher Todiramphus saurophagus common on coast around Soligi with, e.g. 9 from a boat along 8 km of coast Common Kingfisher Alcedo atthis 1 (A. a. hispidoides) at 480 m; 1 (possibly migrant A. a. bengalensis) on coast White-bellied Cuckooshrike Coracina papuensis 6 records from sea level to 420 m Willie Wagtail Rhipidura leucophrys Torresian Crow Corvus orru 1 near sea level near Jikotamo Arctic / Kamchatka Leaf / Japanese Leaf Warbler several encounters from sea level to 1,000 m Phylloscopus borealis / examinandus / xanthodryas Gray's Grasshopper Warbler Locustella fasciolata 3 at 750-840 m Moluccan Starling Aplonis mysolensis fairly common to 450 m Grey-streaked Flycatcher Muscicapa griseistricta quite common migrant, from lowlands to 990 m Black-faced Munia Lonchura molucca flock of 30 at 220 m and 2 more at 360 m Eastern Yellow Wagtail Motacilla tschuschensis ssp. 1 at site F Grey Wagtail Motacilla cinerea 5 records © 2013 The Authors; Journal compilation © 2013 British Ornithologists' Club Marc Thibault et al. 88 Bull. B.O.C. 2013 133(2) with the flight behaviour, are typical of Scolopax. MT also noted that the bird appeared to be significantly larger than congenerics with which he had previous experience (Eurasian Woodcock S. rusticola and S. bukidnonensis), but light conditions were too poor to observe any plumage characters. At 06.02-06.20 h on 19 March 2010, still in poor light conditions, a bird (probably the same) was seen again once and heard twice more (PDR & MT). Further observations and sound-recordings (www.xeno-canto.org/109284) were made at site F, at c.100 m: at c.06.45 h on 20 March 2010, a woodcock was seen in flight and heard twice roding low over the canopy (MT & WP). The bird was seen in better light conditions and flew just overhead at a distance of >30 m. Typical Scolopax shape and large size were noted and WP also noticed yellowish, mainly unstreaked underparts. At 06.02 h the following morning, what was presumably the same roding bird was seen and heard again twice at the same location (MT & PDR). The observations were made in a coconut plantation at the edge of a patch of disturbed, fragmented primary forest, with evidence of ongoing small-scale logging. The undergrowth included a marshy gully with bare ground bisected by a low-flowing stream. It was adjacent to a wet grazed pasture and surrounded by steep hills covered by degraded forest interspersed by scattered clove and nutmeg plantations. Our observations are the first documented records of Moluccan Woodcock since two females were collected in August-September 1982 at 'Galala', said to be on Obi (http:// collections.si.edu). The latter locality is considered untraced by BirdLife International (2011b) and indeed we failed to find a locality of this name on Obi, although a remote village called Galala exists on the island of Bisa (Fig. 1). The species is known from only six other specimens, of which three were also collected on Obi (two in the 19th century and one in 1953, the latter at Wayaloar on the south coast), one on Bacan (in 1902) and two of unknown origin (ITartert 1903b, BirdLife International 2011b). During their field work on Obi in 1989 and on Obi and Bacan in 1992, respectively, Linsley (1995) and Lambert (1994) did not observe the species, but the latter obtained a convincing report from a guide at Kelo, Obi, who 'occasionally flushed this bird from ridgetops above ca. 500 m'. Our sound-recordings permit the vocalisation of S. rochussenii to be described. In display, it gives a loud and distinctive call, comprising a hard, metallic, rattled or staccato phrase, with 8-11 motifs, each motif given at intervals of 0.04-0.05 seconds (Fig. 2). This rattle phrase can be transcribed ti'ti'ti'ti'ti'ti'ti'ti'ti. Each motif has a similar structure, comprising a distinctive pulse (c.4 kHz rising to c.5-5.5 kHz) connecting with a concurrent, higher piched and slightly variable curled syllable (c.6-6.8 kHz rising to 6.5-7.3 kHz and falling to 5.3-6 kHz). The rattle phrase is given at intervals of 1.8-3.3 seconds. Rattling calls of Moluccan Woodcock are strikingly similar in structure and frequency to those of S. bukidnonensis, but the latter have only five motifs and the pulse begins at a lower pitch (c.3 kHz) (Kennedy et al. 2001, www.xeno-canto.org/species/Scolopax-bukidnonensis). Rattles delivered by S. bukidnonensis are frequently interspersed by lower pitched, growling or grunting phrases, which were not heard from S. rochussenii, although this might just reflect the small sample of recordings. The calls of S. rochussenii are very different from those of New Guinea Woodcock S. rosenbergii (Kennedy et al. 2001; G. Wagner, www.xeno-canto. org/23274) and from those attributed to Javan Woodcock S. saturata (G. Wagner, www. xeno-canto.org/42307; F. Ducry & D. Marques, www.xeno-canto.org/56931), but we have been unable to locate sound-recordings of Sulawesi Woodcock S. celebensis, which are undescribed, the latter being closest to S. bukidnonensis by morphology. Our results suggest a close evolutionary relationship between S. bukidnonensis and S. rochussenii. Based on the very scant information available, Moluccan Woodcock was previously assumed to be a bird of hill and montane forest (Hartert 1903b, White & Bruce 1986, Coates & Bishop 1997, Vetter 2009). However, the only traceable locality (Wayaloar, southern © 2013 The Authors; Journal compilation © 2013 British Ornithologists' Club Marc Thibault et al. 89 Bull. B.O.C. 2013 133(2) A 111 j|,! ih|ii] nifti .i i'll 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 Time (sec) B ) | II | 1. 1 ! hi j '1 V' v v i V i I I I 1 1 1. 1 i i i f t l JLT.iL ! ; : i ii . ‘ - . . • . 1 .5 1.5 Time (sec) [III 1 ill, c \\X\ M l 1 1 n h n p 1 | I 1 ‘ 1 : j ! • 1 I i a 11 i t. i. 1 - V- ' • ,1 ij i '• v< I J "(cid:9632) ' : . 1 * .5 1 1.5 2 2.5 3 3.5 Time (sec) 11 10 D 9 ! ! ‘ 1 j ^ ' M U p l ; * I t i t i f Time (sec) Figure 2. Sonograms of the roding display call of Moluccan Woodcock Scolopax rochussenii from (A-C) east of Soligi, Obi, 18 March 2010 and (D) near Jikotamo, Obi, 21 March 2010 (M. Thibault). X-axis = time (one second per tick. Fig. 2A; 0.5 seconds per tick. Figs. 2B-D), y-axis = frequency (1 kHz per rick). © 2013 The Authors; Journal compilation © 2013 British Ornithologists' Club Marc Thibault et al. 90 Bull. B.O.C. 2013 133(2) Obi, August 1953) is on the coast, raising 'the possibility that its optimal (or seasonally optimal) habitat is to be found in lower-lying regions of islands where the largest, dampest substrates may perhaps occur' (BirdLife International 2011b). Our records were obtained at two of three localities visited below 500 m that we surveyed at suitable times, i.e. dusk and dawn. These records confirm that the species holds territory (and presumably breeds) in lowland forest, hence supporting the latter hypothesis and that it may not be uncommon in suitable habitat. Woodcocks were not recorded at the two localities we surveyed at 600-1,200 m despite dedicated searching. Thus, we provide evidence that the species is not only, if at all, associated with montane forest. Our records also suggest that, to some extent, Moluccan Woodcock tolerates habitat degradation and fragmentation, and that wet habitats in or close to forest (i.e. marshes or wet gullies around low-flowing streams) are perhaps important for the species. However, much remains to be learned concerning population size and ecological requirements. In the near future, direct habitat destruction resulting from mining, logging, plantations and agriculture are likely to be extended in lowland and montane areas on Obi and Bacan. Therefore, its status as Endangered (BirdLife International 2011b) should be maintained, at least until further data are collected. METALLIC PIGEON Colwnba vitiensis halmaheira On 13-15 March 2010, a calling bird was seen and two more heard at site E at c. 1,200 m in primary forest (MT, PDR & WP). On 14 March 2010 a bird was heard at the same site at 1,100 m (MT). Photographs and sound-recordings were made. No conclusive record was obtained at lower elevation. The species is widespread in Wallacea. White & Bruce (1986) and Coates & Bishop (1997) reported its presence on Gomumu, a small satellite island 9 km south of Obi, but they did not mention it from mainland Obi. These statements contradict Hartert (1903a), who mentioned that a specimen of 'Columba albigularis' was sent by Lucas from 'Obi Major'. Furthermore, during a visit to Tring museum in 2011, MT & PDR were able to check five (three females and two males) C. v. halmaheira collected by Waterstradt on Obi in January 1903. These specimens might have been collected too late for inclusion in Hartert's review of the birds of Obi. In any case our investigations clarify the status of this taxon on mainland Obi and also describe its altitudinal range on this island. SLENDER-BILLED CUCKOO-DOVE Macropygia amboinensis albiceps Common in forest and forest edge from sea level to 1,210 m, with photographs made by MT at c.1,200 m. Our records considerably extend the altitudinal range on Obi as previous published records are below 500 m (Coates & Bishop 1997) and are in accordance with its altitudinal range from elsewhere in Wallacea (Coates & Bishop 1997). SCARLET-BREASTED FRUIT DOVE Ptilinopus bernsteinii micrus Regularly seen and commonly heard at 150-1,210 m in primary and logged forest. A nest attended by at least one adult, containing a single egg, was found on 13 March 2010 at 1,210 m in primary forest (WP, PDR & MT). The nest was a flimsy construction on a fern c.80 cm above the ground, comprising a small moss cup on a larger assemblage of small branches. P. bernsteini is endemic to North Maluku, with the nominate subspecies found from Halmahera to Bacan and micrus restricted to Obi (Baptista et al. 1997). Previously known at 180-600 m on Obi (Lambert 1994). Our records include the first description of a nest on Obi, whose characteristics and situation are similar to those described from Halmahera (Lansley & Fames 2006). © 2013 The Authors; Journal compilation © 2013 British Ornithologists' Club Marc Thibault et al. 91 Bull. B.O.C. 2013 133(2) SUPERB FRUIT DOVE Ptilinopus superbus Common at 150-1,000 m, with further records of two separate birds heard at 1,200 m in primary montane forest. Our records significantly extend the altitudinal range for the species on Obi, where it was previously unrecorded above 700 m (Lambert 1994, Coates & Bishop 1997). CARUNCULATED FRUIT DOVE Ptilinopus granulifrons Only recorded once, at site A on 19 March 2010, when at least eight were observed through a telescope in the upper canopy of a large fruiting Ficus, in logged forest at c.220 m (PDR & MT). This single record and the species' apparent absence at higher elevations appear to corroborate that this is an uncommon species closely associated with lowland forest (Coates & Bishop 1997, BirdLife International 2011a). SPECTACLED IMPERIAL PIGEON Ducula perspicillata Commonly heard and occasionally seen from sea level to 1,200 m in forest edge, logged forest and primary forest. Our records significantly extend the species' altitudinal range on Obi (Lambert 1994). They also significantly extend the altitudinal range for the species in the northern Maluku where it was previously recorded to 900 m on Halmahera (Poulsen & Lambert 2000). CINNAMON-BELLIED IMPERIAL PIGEON Ducula basilica obiensis This northern Maluku endemic was fairly commonly seen and heard in forest edge, logged forest and primary forest at 150-1,210 m (WP, PDR, MT) and it was photographed at c.1,100 m (MT). Our records significantly extend the species' altitudinal range on Obi, where it was previously unrecorded above 730 m (Lambert 1994) and match the upper limit of 1,230 m for the nominate subspecies on Halmahera (Poulsen & Lambert 2000). Sound-recordings of D. b. obiensis were obtained by MT (Fig. 3), enabling comparisons with vocalisations of D. b. basilica from Halmahera (www.xeno-canto.org/species/Ducula- basilica). Both forms give similar vocalisations: songs usually comprise a series of 1-2 (less frequently three) very deep, throaty growls, roozv roooorw-ooow, at 0.4-0.5 kHz. When the first call is a short note (lasting 0.2 seconds), the second call is often longer (2.0-2.5 seconds), whereas when a first, longer call (0.8-1.3 seconds) is given, there is no second call or it is frequently shorter (i.e. 1.3-1.7 seconds). Longer calls typically rise very slightly and end abruptly. It was recently proposed to split obiensis (which is confined to Obi) from basilica (on Morotai to Bacan; www.birdlife.org/globally-threatened-bird-forums/2012/) following application of the Tobias et al. (2010) criteria. Although justification for this taxonomic nn :r ) m f rmr ; p (cid:9632)mu (cid:9632)riMnii' 12 3 4 5 6 7 Time (sec] Figure 3. Sonogram of Cinnamon-bellied Imperial Pigeon Ducula basilica obiensis, Obi, 12 March 2010 (M. Thibault). X-axis = time (one second per tick), y-axis = frequency (1 kHz per tick). © 2013 Tire Authors; Journal compilation © 2013 British Ornithologists' Club Marc Thibault et al. 92 Bull. B.O.C. 2013 133(2) Figure 4. Cinnamon-bellied Imperial Pigeon Ducula basilica, with (A) basilica (a pair), Halmahera, 3 October 2011 (© F. Steinhauser), (B) obiensis (a pair). Obi, 15 March 2010 (M. Thibault). © 2013 The Authors; Journal compilation © 2013 British Ornithologists' Club