Along the north and north-west coasts of New Guinea lie ten isolated mountain ranges rising from the lowlands, lower in elevation and poorer in species than New Guinea's Central Range that forms the island's west / east axis (Fig. 1). In previous papers we described the avifaunas, especially the upland avifaunas, of three of those ranges (Foja, Fakfak and Kumawa: Beehler et al. 2012, Diamond & Bishop 2015), and of one former outlier that is now an island (Yapen: Diamond & Bishop 2020). The present paper describes the avifauna of the fourth lowest and nearly the least isolated of the outliers, the Adelbert Mts.

Male (left) and female Fire-maned Bowerbird Sericulus bakeri, the endemic bowerbird of the Adelbert Mountains (William Cooper, reproduced from Cooper & Forshaw 1977, with the kind permission of Dr Wendy Cooper)

Figure 1.

New Guinea's principal mountain ranges: the Central Range forming New Guinea's backbone; and the ten outlying ranges along the north and north-west coasts, from Huon and Adelbert in the east, to Fakfak and Kumawa in the west.

Four features make the Adelberts of particular interest. First, the Adelberts are home to one of the most spectacular, surprising, distinctive and beautiful discoveries of modern New Guinea ornithology: the Fire-maned Bowerbird Sericulus bakeri (Gilliard 1969, Frith & Frith 2004; see painting). When its discovery was first recorded, its home range was erroneously recorded by its collector Rollo Beck as Madang. But Madang was one of New Guinea's largest towns, and was the ornithologically already well-explored former capital of German New Guinea (Chapin 1929). Ever since Gilliard proved in 1959 that the bowerbird's range is actually the Adelberts, the question has remained: why is such a distinctive species endemic to such an unlikely location, a low mountain range close to a much higher and larger mountain range (the Huon Mts.; Fig. 2)?

Second, the search for S. bakeri was motivated partly by the hope that the home of such a distinctive endemic would also prove to harbour other distinctive undiscovered endemics. Indeed, until LeCroy & Diamond (2017) rediscovered Beck's diary and specimen register, and found that Beck had labelled the bowerbird's collecting site as Madang for banal reasons, it was believed that Beck had intentionally mislabelled the locality in order to preserve for himself the option of returning and discovering other new species (Gilliard 1969, Frith & Frith 2004). In fact, subsequent explorations of the Adelberts yielded not only no further distinctive endemic species but just two endemic subspecies, both of them barely worth recognising (Gilliard & LeCroy 1967, Pratt 1982, Beehler & Pratt 2016). How could no other distinctive taxon have evolved in a mountain range that generated one endemic that is so distinctive?

Third, subsequent explorations did reveal that the Adelberts are home to non-endemic populations of some upland species that are otherwise restricted to much higher mountain ranges of New Guinea, such as Superb Bird of Paradise Lophorina superba and the Papuan Lorikeet superspecies Charmosyna [papou]. (Here and elsewhere, we adopt the usual convention of denoting a superspecies by square brackets.) How can the Adelberts support those populations despite the Adelberts' modest elevation?

Figure 2.

Presence and absence of Sericulus bowerbirds on New Guinea's northern watershed, modified from Diamond (1969). Each filled dot represents records by one observer. There are only seven documented areas of occurrence of the allospecies Masked Bowerbird S. aureus, numbered from west to east, with observer names in parentheses: 1 = Tamrau Mts. (E. T. Gilliard and B. M. Beehler) and Arfak Mts. (many observers) of the Vogelkop. 2 = Wandammen Mts. (E. Mayr). 3 = Weyland Mts. (F. Shaw-Mayer). 4 = Bernhard Camp (A. Rand). 5 = Foja Mts. (B. M. Beehler and J. Diamond). 6 = Mt. Nibo (J. Diamond). 7 = Mt. Turu (J. Diamond). The asterisk depicts the range of the allospecies Fire-maned Bowerbird S. bakeri in the Adelberts. X = areas surveyed intensively without finding any Sericulus. ? = an undocumented report of S. aureus from the Jimi River. Except for the latter, S. aureus is known mainly from outlying ranges (locations 1, 2, 5, 6 and 7). There are only two documented localities for it in the Central Range (3–4), and there is a large gap between the easternmost record of S. aureus (7) and the range of S. bakeri. The remaining Sericulus allospecies in New Guinea is Flame Bowerbird S. ardens of the southern watershed.

Finally, ten species are confined in the Adelberts to a narrow altitudinal band extending only c.100 m below the highest summit. In all other well-explored New Guinea outliers, the highest-elevation populations extend at least 265 m, in some cases even 1,500 m, below the summit. How can those Adelbert populations survive when crammed into such a narrow altitudinal band, presumably supporting only a small population doubtfully sustainable in isolation?

We shall address these and other questions in light of our four explorations of the Adelberts, including their summit, in the years 2001, 2002, 2004 and 2006. We review and summarise records of other observers in a table listing all species known to have been recorded in the Adelberts.

Background

Environment and people.—The environment and people of the Adelberts are important for understanding the origins of the Adelbert avifauna, and the history of its ornithological exploration. The Adelbert Mts. rise from the lowlands of New Guinea's north-east coast (Fig. 1). The nearest mountains are those of the Huon Peninsula to the south-east, separated from the south-eastern Adelberts by the narrow Gogol River valley. The narrow Ramu River valley in turn separates the Huon Mts. to the north from the Central Range to the south, but broadens to the west to constitute a wider lowland gap between the south slopes of the Adelberts and the north slopes of the Central Range. A much broader gap, formed by the Sepik River and its marshy lowland basin, separates the Adelberts from the nearest outlying range to the west, the North Coastal Range. Hence upland habitats in the Adelberts are closest to those of the Huon Mts., then to those of the Central Range, and further from those of the North Coastal Range (Fig. 3).

The chain of the Adelberts lies along a south-east / north-west axis, rising towards the north-west. The highest peak is Mt. Mengam near the chain's north-west end, whose elevation we determined as 1,675 m by ascending it and measuring the elevation repeatedly with our altimeters. Because Mt. Mengam is at the end of the chain furthest from the coastal town of Madang and was recognised only in the 1970s to be the tallest Adelbert peak (T. K. Pratt pers. comm.), it was not explored ornithologically until 1974 by Pratt and colleagues and in 2004 by us. Earlier explorations of the Adelberts by Beck, Gilliard and Ziegler were of lower but more accessible mountains to the south-east.

Rainfall in the Adelberts is highest in December–March and lowest in August–October (Brookfield & Hart 1966, McAlpine et al. 1975, Tupper 2012). However, seasonality is modest, and rainfall in the wettest months is barely double that in the driest months. Annual rainfall at the two recording stations closest to our field sites, Wanuma and Saruga, is 320–340 cm.

The Adelbert terrain is sandstone formed into steep, narrow, razorback ridges separated by deep narrow valleys. Even by the standards of montane New Guinea's generally rugged terrain with which we are familiar, the Adelberts rank as especially difficult for travel. In our helicopter flights over the Adelberts, we saw no large level-ground area.

That terrain has several consequences relevant to ornithologists. First, there is no motor vehicle road at higher elevations within the Adelberts, and only one landing strip for small fixed-wing aircraft (at Wanuma, 22 and 29 km south-east of our two study sites). Second, that lack of transport impedes commercial logging, and the terrain's steepness compels villagers to seek small favourable patches for their gardens, so that the Adelberts are still largely covered by rainforest. Third, the terrain's difficulty, and the chronic warfare and fierce reputation of its inhabitants, explain why the Adelberts remained unknown to Europeans for so long, despite their proximity to the German colonial capital of Madang (Friedrich-Wilhelmshafen)—and why the discovery of Sericulus bakeri at ‘Madang’ (see below) occurred so late and caused such astonishment. Only one additional bowerbird species (Archbold's Bowerbird Archboldia papuensis) and one bird of paradise (Ribbon-tailed Astrapia Astrapia mayeri) were discovered after S. bakeri, both of them in remote areas of the Central Range.

Information about European contact in our two study sites comes from our conversations with older villagers, and from the linguist Ian Tupper (2012). Stone tools were abandoned, and steel tools began to be acquired by trade from the coast, only in the 1940s. There was never any contact with a foreign patrol during German colonial times, which ended in 1914. Australian patrols did not visit the Adelberts until 1944. The first missionary visit to our study site of Kangarangate was in 1959, and the first mention (by a missionary linguist) of the Pamosu language spoken at our other study site of Munggur was in 1975. Chronic warfare was late in disappearing: our Kangarangate hosts told us that warfare compelled them to abandon their homeland in 1949, and that they could not make peace and return until 1982. That chronic warfare, plus the rugged Adelbert terrain, contribute to explaining why Beck and Gilliard collected only up to 760 m and about 1,220 m, respectively, and depended on native collectors to go further: because it would have been too dangerous for a European to do so. Even at the time of our 2004 and 2006 visits, few of the villagers at our two study sites had travelled, worked or lived outside their language area.

As mentioned above, arable land occurs in only small patches, and gardens are small and scattered. For example, we encountered an isolated garden c.1 km from the summit of Mt. Mengam, more than an hour's walk from Munggur village even for the briskly trotting villagers. Hence people traditionally spent most of their time in garden houses and scattered hamlets. While Australian patrols and then the Papua New Guinea government tried to induce people to gather in villages, Australian patrol reports mention finding few people in the villages, because they were instead mostly in their garden houses. In such small-scale societies there were no hereditary chiefs.

Subsistence traditionally depended on gardens producing yams as the major crop, plus taro, bananas, sago, Pandanus, and the introduced crop cassava. The only domestic food animals, pigs and chickens, yield little meat: pigs are saved to be slaughtered at a big feast held every year or two. Instead, protein is obtained mainly by hunting wild mammals.

Languages.—While New Guinea is famous for having the world's highest language diversity, the Adelberts are diverse even by New Guinea standards. Approximately 60 languages, all of them confined to the Adelberts, are spoken in an area of about 7,000 km² (Lewis 2009). Hence the area occupied by the average language is c.120 km², and the average cross-section of a language area is only about 11 km. (These numbers refer to mutually unintelligible languages, not mere dialects.)

That diversity is a consequence of New Guinea's and the Adelberts' rugged terrain, chronic warfare impeding travel, long history of human occupation (c.60,000 years) and lack of political unification until colonial times (Foley 1986). With one exception, New Guinea languages fall into dozens of language families that are strictly or almost entirely confined to New Guinea, and that have no demonstrable relationship to each other or to any other language family in the world. Most Adelbert languages belong to the largest of those families, the Trans-New Guinea family, whose extent is attributed to population expansions associated with New Guinea's independent development of agriculture within the last 10,000 years. The sole exception is the Austronesian language family that spread from Taiwan to Polynesia, reaching New Guinea about 3,500 years ago, and that is represented mainly in the lowlands of northern New Guinea. Five Austronesian languages, presumed to have arrived with recent invaders, are spoken in small Adelbert coastal enclaves and south of the Gogol River near the Adelberts.

The languages at both of our study sites belong to the so-called Madang subgroup of the Trans-New Guinea family. The Pamosu language, with about 1,500 speakers, is that spoken at Munggur and four nearby mountain villages. The Aiti language, with about 3,300 speakers, is spoken at Kangarangate and some neighbouring villages. (Kangarangate villagers insisted to us that Aiti is the name of their language, although it appears to be referred to as Mum or Katiati in Lewis 2009.) These are the two languages whose names for bird species we used in our conversations with villagers, and which we provide in Appendix 1 for the convenience of future ornithologists wishing to find Sericulus bakeri and other Adelbert bird species. It will be apparent from Appendix 1 that the two sets of names are almost entirely different, illustrating the mutual unintelligibility of those two related languages spoken only 9 km apart.

Previous ornithological studies.—Although numerous collectors obtained bird specimens near the German colony coastal capital of Madang from the 1880s onwards, the first collection in the Adelberts just inland of Madang was not made until 1928–29, by the professional collector Rollo Beck. On Beck's return from eight years as leader of the American Museum of Natural History's (AMNH) Whitney South Sea Expedition, the museum paid Beck to conduct further collecting in New Guinea, thereby helping Beck deal with personal financial difficulties (LeCroy & Diamond 2017). Beck evidently received no specific instructions about where in New Guinea to collect, and he had no opportunity to familiarise himself in advance with its birds. Instead, it seems likely that, after Beck landed in Madang, he selected the nearby Adelberts as his first field site merely because missionaries could arrange for him to stay at their stations there.

Beck's Adelbert itinerary, including his collecting locality for Sericulus bakeri, remained unknown for many decades, because Beck initially labelled the sites of all of his Adelbert specimens simply as ‘Madang'. He adopted this practice because he had previously been collecting on smaller Pacific islands where differences between sites on the same island were unimportant. Only recently did Mary LeCroy (in LeCroy & Diamond 2017) reconstruct Beck's itinerary from his diary and specimen registers. LeCroy showed that Beck collected at four sites in and near the Adelberts between August 1928 and January 1929: Madang itself, on the coast; Nobonob, a lowland mission station near Madang at the foot of the Adelberts; Maban, at an elevation of a few hundred feet (Beck's estimates are in feet, not metres) on the Gogol River in the Adelberts’ southern foothills; and Meganum only c.9 km from Maban at an elevation of 1,200 feet. From Meganum, Beck himself collected up to approximately 2,500 feet, and his local collectors spent three days further inland and probably at somewhat higher elevations. Beck himself could go no further, because the Adelberts were still lethally dangerous for a European at that time. (Another of Beck's sites, Keku, was in the western foothills of the Huon Mts., not in the Adelberts, and is not to be confused with Keki Lodge now frequented by birdwatchers in the Adelbert foothills.)

Beck's Adelbert collections, housed at AMNH, comprised 502 specimens of 129 species (Table 1 of LeCroy & Diamond 2017). Only 11 of those species were upland taxa (records listed in Appendix 1), almost all of them obtained while Beck was based at Meganum—including Sericulus bakeri, probably taken by Beck's local collector. All of those species are ones that we and other observers have found at elevations below 1,000 m, supporting the evidence from Beck's itinerary that he and his collector remained at modest elevations.

The next collection was made in 1959 by E. Thomas Gilliard, who was searching specifically for Sericulus bakeri. In the course of two months Gilliard and his New Guinean hunters reached elevations of c.1,220 m on Mt. Memenga in the south-eastern Adelberts and collected 385 specimens of 130 species, thereby adding 26 species to the list of Adelbert upland species (Gillard & LeCroy 1967). That collection is also at AMNH.

In 1967 Alan Ziegler collected up to 1,100 m at Atitau and obtained 32 species, including 11 upland species, two of them (Black-bellied Cicadabird Edolisoma montanum and Russet-tailed Thrush Zoothera heinei) new records for the Adelbert upland avifauna. His collection, also housed at AMNH, was examined by one of us (JD) and by Pratt (1982).

In 1974 an expedition of the Wau Ecology Institute including Thane Pratt collected for two months and became the first collectors to reach the highest summit (Mt. Mengam, not to be confused with the lower Mt. Memenga reached by Gilliard). They added 25 upland species (Pratt 1982) including all but one of the high-elevation populations now known from the Adelberts. Those specimens are mostly housed at the Bishop Museum, Honolulu.

In 1985 C. B. Frith, D. W. Frith and Roy Mackay visited Mt. Mengam to seek Sericulus bakeri and to photograph MacGregor's Bowerbird Amblyornis macgregoriae at its bowers. Frith & Frith (1988) described the courtship display of Superb Bird of Paradise Lophorina superba. From 1985 to 1988 Mackay and others made five additional Adelbert visits, from which Mackay (1991) reported eight species not found by any previous or subsequent visitor to the Adelberts: Wattled Brushturkey Aepypodius arfakianus, Mountain Kingfisher Syma megarhyncha, Dimorphic Fantail Rhipidura brachyrhyncha, Friendly Fantail R. albolimbata, Brown-backed Whistler Pachycephala modesta, Papuan Sittella Daphoenositta papuensis, Elfin Myzomela Myzomela adolphinae and Varied Honeyeater Gavicalis versicolor. Those records were queried or not accepted by Beehler & Pratt (2016). While an Adelbert population of Aepypodius arfakianus appears to us quite possible, and Daphoenositta papuensis and Myzomela adolphinae unlikely but not impossible, the other five records are in our opinion improbable. The two Rhipidura species and Pachycephala modesta are behaviourally or vocally conspicuous, are unlikely to have escaped Pratt and us if they had been present on Mt. Mengam's summit as claimed, and probably involved misidentified sightings of congeners that are common there. Especially unlikely is the report of Gavicalis versicolor, a strictly coastal species, reported at 700 m in the interior of the Adelberts. Out of caution we have not included these records in the Appendix but mention them here for the consideration of future observers.

In May 1999 Bruce Beehler and Kevin Vang observed 90 species at elevations of 500–900 m, including 17 upland species. Two of those (Red-fronted Lorikeet Charmosyna rubronotata and Banded Yellow Robin Gennaeodryas placens) had not been previously reported for the Adelberts but were subsequently observed by us. In 2010 Beehler observed the rare Obscure Berrypecker Melanocharis arfakiana at Keki Lodge in the Adelberts (Beehler & Pratt 2016: 345).

We are aware of two other Adelbert collections that we cannot discuss because they have not been published and we have not examined them: that by William Peckover in 1969 (mentioned by Pratt 1982), housed at Yale's Peabody Museum; and the other by Brett Benz in 2007, housed at the Univ. of Kansas. A list of Benz's specimens that he kindly sent us reports one specimen of an upland species not otherwise known from the Adelberts, New Guinea White-eye Zosterops novaeguineae. That identification awaits confirmation because of the close resemblance in north-east New Guinea between that species and the abundant Black-fronted White-eye Z. atrifrons of the Adelberts.

Our study

We observed (but did not collect) Adelbert birds in 2001, 2002, 2004 and 2006. Our principal study was from 26 July to 15 August 2004 at two Adelbert sites that KDB selected during a previous helicopter visit: Munggur village (04°41.38′S, 145°14.64′E) at an elevation of 1,223 m, directly below the Adelberts' highest peak, Mt. Mengam; and Kangarangate village (04°39.34′S, 145°10.24′E) at 875 m, and c.8 km west-northwest of Munggur. The habitat around both villages consisted largely of forest, interrupted by scattered or regenerating gardens.

We arrived at Munggur by helicopter from Madang on the morning of 26 July and divided our time between two campsites: the village itself, and a mountain camp at 1,655 m on Mt. Mengam's summit ridge (04°42.13′S, 145°13.70′E), near Mt. Mengam's summit of 1,675 m (04°42.20′S, 145°13.83′E). We occupied the mountain camp on 30 July–2 August, and devoted those days plus 27–28 July to surveying the ridge for c.1 km west and east of the camp. From Munggur we descended the Ululu trail to 1,150 m, surveyed a trail west from and at the same altitude as the village, and ascended steeply to our mountain camp.

On the morning of 4 August 2004 we transferred by helicopter from Munggur to Kangarangate, where we again divided our time between two campsites: the village itself, and a mountain camp at 1,191 m (04°38.63′S, 145°94.67′E). From the mountain camp we surveyed up to 1,294 m above Musiamunat village. From Kangarangate we surveyed down to a pond at 835 m and a river at 639 m, and steeply up to Makokapi hamlet on a ridge crest at 994 m, and up to the 1,260-m crest of the trail from Kangarangate to the coast. On the morning of 15 August we returned by helicopter to Madang.

In 2006 we again travelled by helicopter on 2 May from Madang to Munggur, and then on 4 May from Munggur to Kangarangate, before returning to Madang on 7 May. During those visits we remained within 1 km of each village. We observed five lowland species and one upland species that we had not observed in 2004.

On 12 August and 1 September 2001 and 27 July 2002 KDB made single-day trips by car to Keki Lodge (700–1,080 m, 04°41.49′S, 145°24.22′E). He observed 90 species, including nine lowland species and two migrant visitors that we did not subsequently see in 2004 or 2006.

Methods.—Our methods were similar to those that we described for our studies elsewhere in New Guinea (Diamond & Bishop 2015, 2020). Briefly, except KDB's visits in 2001 and 2002, all of our observations were made on foot trails, mostly within forest, in areas with no motor vehicle roads. We devoted much effort to recording vocalizations with Sony TCM 5000 EV tape recorders, playing back unidentified vocalisations in the field to attract and identify the singers, and re-listening to recordings in camp each day because our directional microphones often captured vocalisations that we had not noticed in the field. We stopped at fruiting and flowering trees where birds gathered. We began observing by 05.00 h to detect nocturnal birds. Elevations of all significant observations were measured using Thommen altimeters or a Garmin GPS. We did no collecting.

We were constantly accompanied by Munggur or Kangarangate villagers, who pointed out and identified birds seen and heard, informed us about their experience of each species, and described to us other species that lived in the area of their village, but which we did not encounter together. These conversations were undertaken in the language Tok Pisin, and the bird names used were in the local Pamosu or Aiti languages spoken at Munggur or Kangarangate. As we routinely do elsewhere in New Guinea, we went to much effort to identify these names, for several reasons: local people thereby helped us to identify birds seen or heard; they guided us to find species that we particularly wished to observe; they shared with us their lifelong knowledge of bird species in their environment; and they provided clear identifiable descriptions of 11 species that we did not encounter. Those 11 are denoted by square brackets in Appendix 1; all have been recorded by other European visitors to the Adelberts except the Palearctic winter visitor Grey Wagtail Motacilla cinerea, for which there is no other local record, but it is widespread in New Guinea's mountains during the boreal winter.

We elicited most of those names when we and our guide saw or heard a bird together. If there was any doubt as to which individual bird in the vicinity was being referred to, we confirmed the identity by asking our guide to describe the species named. Once we had accumulated many such names securely identified in the field, we elicited more by asking our guides to name and describe for us additional species that we and they had not yet encountered together, e.g., we asked them to describe nocturnal species, or ground-dwelling species, or species found near water, or species similar to (‘brata bilong') species that we and our guide had already encountered. For example, after we and our Munggur guides had found several species of lories that they named (‘kirikirik' = Dusky Lory Pseudeos fuscata, ‘ororovion’ = Black-capped Lory Lorius lory), we asked them to name and describe other very similar birds that proved to be other species of lories, then other somewhat similar birds with similar bills that proved to be other parrots. The Appendix gives the Pamosu and Aiti names that we identified in this way, so that future visitors to these villages can use the names in order to find particular species.

Results and Discussion

Species number.—The total number of species recorded from the Adelberts is 235 (Appendix). Of those, we observed 197 ourselves. Other visitors observed 37 species that we did not record. As already mentioned, one additional species (Motacilla cinerea) was reported to us by Adelbert residents, but not observed by us or by other visiting ornithologists.

Our focus here is on upland species, defined as those largely confined to sloping elevated terrain, and absent from the flat lowlands at or near sea level. We have found this definition more useful and less arbitrary than defining ‘montane species’ as species largely confined to elevations above some arbitrary specified elevation, such as 1,000 m or 1,700 m. We have discussed in more detail elsewhere (Diamond & Bishop 2015, 2020) the advantages of this definition, and the ambiguities and practical issues in applying it. By this definition, we recognise 71 Adelbert species as upland species (abbreviated S_up), denoted by an asterisk in the Appendix.

Let us place this number in context by comparing it with S_up values for New Guinea's nine other outlying mountain ranges. One of those ranges (Van Rees) is much lower (1,262 m) and much poorer in upland species (S_up = 37) than the Adelberts. Two of them (Vogelkop and Huon) are much higher (2,954 and 4,121 m respectively) and richer in upland species (S_up = 129 and 127 respectively) than the Adelberts. The remaining six are more comparable to the Adelberts, with elevations of 1,400–2,218 m, and S_up values from 44 to 95 species.

Table 1 summarises, for all ten mountain ranges, their elevations, S_up values, and numbers of upland species shared with the Adelberts. The following conclusions emerge from Table 1.

First, the number of upland species in a mountain range increases with the range's elevation, from 37 to 129 species. That is, as one expects, in accordance with experience gained from mountains elsewhere in the world: greater elevation translates into more ‘niches’, i.e. more opportunities for elevationally specialised species. The same trend is more weakly evident for the seven comparable ranges (the Adelberts and the other six), partly because their span of elevations is modest (only 1,400–2,218 m). The other disturbance of the trend arises from the flagrantly low S_up value of only 44 species in the Cyclops, although they are second-highest of the seven comparable ranges. Evidently, elevation is not the only factor influencing S_up values. Some others include area of upland habitats, and extensive flat lowlands isolating a mountain range from other ranges. The Cyclops, although they are high, are small in area.

TABLE 1

Upland avifaunas of the Adelberts and the other outliers.

TABLE 2

Upland species distributions on the Adelberts and other outliers.

Second, although New Guinea has approximately 193 upland species or superspecies (Diamond & Bishop 2020), the Adelberts share most of their upland species (59–79%) with five of the other six comparable ranges. (The outlier is again the species-poor Cyclops.) The percentage of Adelbert species shared increases with the S_up value of the range compared, from 41–42% for the most species-poor (Van Rees and Cyclops) to 92% for the most species-rich (Vogelkop). That is as one would expect if richer ranges tended to contain those of poorer ranges plus additional species.

If each range contained a random sample of New Guinea's 193 upland species, one would not expect such high sharing among seven samples of only 44–95 species each. That outcome suggests that some species are disproportionately good colonists and succeed in establishing themselves on many isolated mountain ranges. The next section explores this interpretation systematically.

Species identity.—Table 2 provides a systematic test of the suggestion that some upland species are disproportionately able colonists of outlying ranges, and that the upland avifauna of the Adelberts (as well as of the other outliers) is enriched in such species. For each of New Guinea's approximately 193 upland species or superspecies, we calculated on how many of the ten outliers that species occurs. Those calculations are summarised in columns 1–2 of Table 2. That number ranges from zero for species of the Central Ranges present on no outlier (e.g., Sooty Honeyeater Melionyx fuscus), to one for species present on just a single outlier (e.g., Papuan Treecreeper Cormobates placens, present only on the Vogelkop), to ten for species present on all ten outliers (e.g., White-eared Bronze Cuckoo Chalcites meyerii).

TABLE 3

Number of ‘restricted’ upland species on each outlier.

As we noted previously (Diamond & Bishop 2020: Table 3), many upland species occur on no outlier (35 species) or on only a few outliers (e.g., 28 species confined to a single range). Few species (14, ten and ten) occur on an intermediate number of outliers (five, six or seven outliers, respectively), but somewhat more species (20, 17 or nine) occur on most or all outliers (eight, nine or ten outliers, respectively). That is, New Guinea upland species are not randomly distributed in colonising ability: there is instead a large excess of unsuccessful colonists, and a smaller excess of very successful ones.

Consider the Adelberts from this perspective (column 3 of Table 2). Inevitably, all nine species that occur on all ten of the outliers occur on the Adelberts. Unsurprisingly, most species present on the majority but not all outliers also occur on the Adelberts (e.g., 16 of the 17 species on nine outliers). Also inevitably, the 35 species absent on outliers do not occur on the Adelberts. Most of those 35 are high-elevation species of the tall (5,000 m) Central Range, for which the outliers provide no or little high-elevation habitat. The 28 species confined to a single outlier are also mostly confined to elevations above 1,500 m, with the result that all of them are confined either to the highest (Huon, 4,121 m) or second-highest outliers (Vogelkop, 2,954 m) (Table 3, column 3).

A surprising result involves species present on just two or three outliers (Table 3, columns 4–5). Because species restricted to a single outlier prove to be confined to either of the two highest outliers, one might guess by extrapolation that, for species restricted to 2–3 outliers, the number of populations would just increase with outlier elevation: the highest outliers would have the most such ‘restricted’ populations, and the next highest would have the next most. In partial accord with this expectation, the two outliers that are by far the highest (Huon and Vogelkop) far exceed all other outliers in their numbers of restricted populations (46 and 39 species, respectively), while the next highest outlier (Foja) has the third-largest number (eight species). But the next two-highest outliers, Cyclops and Wandammen, have zero and one restricted populations, respectively. Instead, the next-highest number of restricted populations, seven, is on the Adelberts, despite their being fourth from last in elevation among the ten outliers!

Those seven Adelbert populations (Table 4) are shared only with the much higher Vogelkop or Huon (five shared populations each). For three of those seven restricted species in Table 4, we should not attribute significance to their absence from low outliers other than the Adelberts, because two of the species (Melanocharis arfakiana and Papuan Parrotfinch Erythrura papuana) are very rare and cryptic, and the third (Dimorphic Jewel-babbler Ptilorrhoa geislerorum) is almost confined to New Guinea's north-east corner. But three others of those seven Adelbert species absent from other low outliers (Stella's Lorikeet Charmosyna [papou] stellae, Marbled Honeyeater Pycnopygius cinereus and Ornate Melidectes Melidectes torquatus) are common vocal species confined in the Adelberts to the highest elevations. Why do those high-elevation species succeed in maintaining small populations at the top of the Adelberts, but not on the other low outliers, four of which are higher than the Adelberts?

Table 5 lists all ten species for which we have sufficient observations to suggest that their Adelbert populations were confined during our study to within 150 vertical metres of the Adelbert summit (1,675 m). Among the outliers, the Adelberts are unique in this respect. For the seven other outliers for which we have sufficient information concerning elevational ranges, Table 6 summarises how far below the summit is the highest elevational floor of any species. For example, in Kumawa, whose summit is at 1,654 m, the highest floors are of a smoky honeyeater Melipotes sp. and Regent Whistler Pachycephala schlegelii, both at 1,389 m. That is, those species are compressed into an elevational band extending 265 m below the summit. That minimum elevational range of high-elevation populations is between 265 and 299 m for the four lowest mountains of Table 6 other than the Adelberts, and 541, 954 and 1,521 m for the three highest mountains in order of elevation. As an explanation for that increase in minimum elevational range on the highest mountains, we note that ambient temperature, hence productivity, decreases with elevation. Therefore, populations confined to the summits of the highest mountains require a larger area of habitat, and so a greater elevational range, to sustain some minimum population size than those confined to the summits of lower mountains.

TABLE 4

Adelbert populations of ‘restricted' upland species. This table names the ‘restricted’ upland species that occur in the Adelberts, and that constitute the seven entries for the Adelberts in the right-hand column and row 7 of Table 3.

TABLE 5

Adelbert species with high elevational floors.

Thus, Table 5 suggests that in the Adelberts the highest-elevation populations are compressed into a narrower elevational band, and presumably have smaller populations, than on other outliers of similar elevation. But there is a caveat; of the ten species listed in Table 5, Pratt (1982; pers. comm.) reported nine as extending to lower elevations than we do. Probable contributory factors for this difference include the following. (1) Pratt et al. spent more time (22 days) at elevations of 1,400–1,600 m than did we (six days), potentially allowing them to pick up more low-elevation records. (2) We recorded precise elevations for every significant observation (e.g., 1,545 m), while Pratt's reported elevation ranges are general (e.g., ‘1,400–1,600 m’, or ‘down to 1,400 m'). (3) Pratt collected not only on Mt. Mengam at or somewhere near our highest camp, but also at another site, Kowat, that provided lower elevations of 1,200–1,400 m. Hence the three of the ten species in Table 5 that Pratt recorded at Kowat certainly reached elevations lower—during Pratt's study—than the floors of 1,525 m or higher that we found on Mt. Mengam in Table 5. However, for the other seven species in Table 5, Pratt's generalised elevation ranges are compatible with the precise apparent floors of 1,525 m or higher that we measured. (4) Pratt's study was in 1974, but ours was in 2004. In the intervening three decades, global warming caused bird elevational ranges to shift upwards on New Guinea's mountains (Freeman & Class Freeman 2014) and on tropical mountains elsewhere in the world. On p. 91 we shall discuss further consequences of those shifts for the Adelbert avifauna.

TABLE 6

Summit elevations and highest elevational floors.

Why are the Adelberts unusual among New Guinea's outliers (perhaps more so in 2004 than in 1974) in the high floors, narrow elevational bands, and small populations of their highest-elevation populations? We suggest two contributing factors: proximity to much higher and larger mountain ranges, permitting immigration from the latter to augment small Adelbert populations that would otherwise not be viable; and climate fluctuations.

Proximity of the Adelberts to higher and larger mountain ranges.—Examination of a map and of contours reveals that the Adelberts are the least isolated, or one of the least isolated, of New Guinea's ten outliers. Only c.15 km of low-elevation forest in the narrow Gogol River valley separates the southern foothills of the Adelberts from the western foothills of the mountains of the Huon Peninsula, which in turn are separated from the foothills of the Central Range by the even narrower upper Ramu River valley (Fig. 3). The distance between the Adelbert Mts. and the Huon Mts. across the Gogol River is 18 km or 12 km at the 500-m or 300-m contours, respectively. The corresponding distance between the Huon Mts. and the Central Range across the Ramu River is only 5 km at the 500-m contour; and these ranges are connected at the 300-m contour. (We are grateful to I. Woxvold for these measurements and for Fig. 3.) In contrast, the North Coastal Range is separated along most of its length from the Central Range by the broad Sepik Basin, c.100 km wide; the Foja and Van Rees Mts. are separated from the Central Range by the Lake Plains, c.50 km wide; the Kumawa and Fakfak Mts. are both isolated by 70–100 km of flat lowlands; and the Cyclops Mts. are isolated by undulating but low terrain. Perhaps the Adelberts are sufficiently accessible to the Huon and the Central Range that the small populations of the Adelberts' ten high-elevation species listed in Table 5 are not truly isolated, but are augmented by immigrants from high elevations in the Huon and Central Range. Seemingly contra this admittedly speculative interpretation stands the feature for which the Adelbert avifauna is famous: its distinctive endemic bowerbird allospecies Sericulus bakeri.

Figure 3.

Terrain map of north-east New Guinea. Areas above 500 m are coloured grey; areas between 300 and 500 m blue; and areas below 300 m green. 1 = the Gogol River valley, between the Adelbert Mts. to the north and the Finisterre Mts. of the Huon Peninsula to the south. 2 = the Ramu River valley, between the Adelberts to the north-east and the Central Range to the south-west. 3 = the upper Ramu River valley and its headwaters, between the Finisterre Mts. of the Huon Peninsula to the north and the Central Range to the south. 4 = the Sepik River basin, between the Adelbert Mts. to the east and Prince Alexander Mts. of the North Coastal Range to the west. Map kindly produced by Iain Woxvold.

Adelbert endemism and climate fluctuations.—Other than Sericulus bakeri, the Adelbert avifauna is very poor in endemics. Beehler & Pratt (2016) recognised just two endemic subspecies, both of which are weakly defined: Forbes's Forest Rail Rallicula forbesi parva and Large Scrubwren Sericornis nouhuysi adelberti (described by Pratt 1982). Just two other endemic subspecies have been proposed for the Adelberts—Trumpet Manucode Phonygammus keraudrenii adelberti and MacGregor's Bowerbird Amblyornis macgregoriae amati described by Gilliard & LeCroy (1967) and Pratt (1982), respectively—but Beehler & Pratt (2016) considered both too undistinctive to recognise. All other Adelbert populations belong to the same subspecies as their conspecifics on the Huon or Central Range or both, or in one or more cases as their conspecifics on the North Coastal Range (Pratt 1982, Beehler & Pratt 2016).

That Adelbert upland populations are near-identical to those on the nearest other mountains is expected given two factors. One is the already discussed narrowness of the lowland barriers separating the Adelberts from those sources (Fig. 3). That prompted our speculation that the high elevational floors, narrow elevational ranges, and small population sizes of the Adelberts' highest upland populations (Table 5) might be due in part to their being augmented by immigrants, and not completely isolated.

The other factor is natural climate fluctuations of the mid-Holocene. During the so-called hypsithermal period from about 8,000–4,000 years ago, global temperatures were warmer than at present, with peaks c.1°C higher than present around 7,000 and again 4,000 years ago. But even the more modest temperature increases of the last half-century have produced measurable upward shifts in avian elevational ranges on tropical mountains worldwide. On two New Guinea peaks, Mt. Karimui and Karkar, Freeman & Class Freeman (2014) found that elevational floors shifted upslope by an average of 100 m between 1965 or 1969 and 2012 or 2013, respectively, associated with a 0.4°C rise in annual mean temperature over that period. Assuming that these floor shifts are linearly related to temperature changes, then the recent 0.4°C rise that shifted floors 100 m upslope suggests that the 1°C increase at the peak of the hypsithermal would have shifted floors upwards by c.250 m. But today's Adelbert floors of the ten species in Table 5 are already just 85–150 m below the Adelbert summit, so if during the hypsithermal the floors of these species had risen 250 m they would have effectively disappeared. Consequently, today's populations probably became established or re-established only within the last 4,000 years. That inferred young age of the Adelbert high-elevation avifauna is a second reason for its non-existent or weak endemism.

We have suggested two factors—weak isolation and young age—contributing both to the lack of endemism in almost all Adelbert upland birds, and to the small population sizes and high floors of the highest-elevation populations. But how should one then interpret the distinctness of Sericulus bakeri, which seemingly implies strong isolation and considerable age? Read on!

The three easternmost outliers.—Of the ten outlying mountain ranges along New Guinea's north and north-west coasts, the three easternmost are the North Coastal Range (NCR), Adelberts and Huon (Figs. 1 and 3). The lowland gap between the eastern foothills of the NCR (the Prince Alexander Mts.) and the western foothills of the Adelberts is c.200 km wide at the 500-m contour (I. Woxvold pers. comm.). But the lowland gap between the Adelberts' southern foothills and the Huon's western foothills is only c.15 km wide (as mentioned above). Hence the NCR / Adelbert gap is a barrier for bird distributions more than ten times wider than the Adelbert / Huon gap.

Table 7 compares the effects of those two gaps on upland bird distributions at the species or allospecies level. At the broad gap between the NCR and Adelberts, five species reach their eastern limit (present in the NCR and further west, absent in the Adelberts): Claret-breasted Fruit Dove Ptilinopus viridis, Josephine's Lorikeet Charmosyna josefinae, Tropical Scrubwren Sericornis beccarii, Piping Bellbird Ornorectes cristatus and Capped White-eye Zosterops fuscicapilla. No species reaches its western limit there, i.e. is present in the Adelberts and further east, but absent in the NCR and further west. (Ptilorrhoa geislerorum, formerly believed to reach its western limit there, was recently discovered far to the west on Yapen Island: Verhelst & Pottier 2020.) Three superspecies are represented by different allospecies on opposite sides of the NCR / Adelbert gap: the Rallicula [leucospila] forest rail superspecies, the Sericulus [aureus] bowerbird superspecies, and the Zosterops [atrifrons] white-eye superspecies.

In contrast, at the narrow gap between the Adelberts and Huon, the only species-level limits are that Charmosyna rubronotata and the Sericulus superspecies reach their eastern limit there, and that the Amblyornis [inornata] bowerbird superspecies and perhaps the Ailuroedus [crassirostris] catbird superspecies may be represented by different allospecies either side of the gap. Even that limit in the Amblyornis superspecies can be questioned, because the distinctiveness of the two Amblyornis populations is based on a very slight size difference, different bower form and display, and an unpublished molecular analysis (summarised in Frith & Frith 2004: 278, and Beehler & Pratt 2016: 278). But bower design is in part culturally transmitted, as shown by the drastic differences in bower form and decorations between A. inornata populations only 8 km apart, although the populations constructing those bowers are near-identical morphologically and exhibit only slight molecular differences (Diamond & Bishop 2015: 314–315). Four other upland species present in the Adelberts are absent from the Huon, but do not reach an eastern limit between the Adelberts and Huon, because they reappear immediately east and south of the Huon (Blue-collared Parrot Geoffroyus simplex, Yellow-legged Flyrobin Kempiella griseoceps, Papuan Scrub Robin Drymodes beccarii and Phonygammus keraudrenii).

TABLE 7

Species-level differences between the Adelberts and neighbouring upland avifaunas

Thus, the distinctiveness of Sericulus [aureus] bakeri may be partly due to the fact that it is unique in being absent east of the Adelberts, and in being separated by a broad geographic barrier from the nearest allospecies to the west, Masked Bowerbird S. [a.] aureus (Fig. 2). All other Adelbert upland populations are identical or similar to upland populations nearby on the Huon and the Central Range. We acknowledge that Mackay reported two sightings of S. aureus from the Jimi River at 144°25′E on the northern slopes of the Central Range (Frith & Frith 2004: 334). However, there is no other record of S. aureus from the Central Range east of the Third Archbold Expedition's Bernhard Camp at 129°22′E, which is 1,600 km west of the Jimi River. KDB now discounts his own possible sighting in the Jimi Valley reported by Coates (1990: 400), and made at a time when KDB had only recently arrived in New Guinea. Five experienced collectors and observers undertook extensive studies of the Central Range's northern watershed in Papua New Guinea, three of them within 25 km of the Jimi River, without encountering S. aureus: E. T. Gilliard at Telefolmin (141°63′E), J. Bürgers on four mountains south of the upper Sepik River (Mäanderberg, Hunsteinspitze, Lordberg, Schraderberg: 141°68′E, 142°82′E, 143 00′E and 144°22′E, respectively), Gilliard in the Schrader Mts. (144°40′E), and I. Majnep and R. Bulmer in the Kaironk Valley (144°47′E). The easternmost record of S. aureus of which we are confident is not from the Central Range but from Mt. Turu at 143°34′E in the NCR (Diamond 1969). The range map of S. aureus in Pratt & Beehler (2015: 172) shows it as distributed continuously along the north slopes of the Central Range, from the Weyland Mts., in the far west, east to the Jimi River. However, the only firm records known to us on those slopes are from the Weyland Mts. and Bernhard Camp. Instead, in northern New Guinea the S. [aureus] superspecies, like Barred Cuckooshrike Coracina lineata, is much more frequently recorded on the outlying mountains than on the north slopes of the Central Range itself (Fig. 2, p. 78).

The other likely reason for the far greater distinctiveness of Sericulus bakeri than of other Adelbert upland populations is its low elevational floor: only 700 m. Upwards shifts of 250 m in elevational floors during the hypsithermal, potentially sufficient to eliminate Adelbert high-altitude populations, would have left S. bakeri secure in an elevational range from 950 m to the summit at 1,675 m. Thus, the floor of Sericulus has been sufficiently low to protect it against the risk of extinction during mid-Holocene warm climates, but sufficiently high to impede its dispersal via the Sepik lowlands between the North Coastal Range and the Adelberts during Pleistocene cold climates.

Outlook

What additions to the Adelbert upland avifauna are possible? Beck's initial exploration at low elevations in 1928–29 recorded 11 upland species. Gilliard in 1959 and Ziegler in 1967 reached middle elevations and added 26 and two species, respectively. Pratt was the first to reach the Adelberts' summit in 1974, adding 25 upland species. Beehler in 1999 and 2010 added three species at low elevations. Finally, our visit in 2004 reached the summit and added four more species.

For three other species there are uncertain reports: Aepypodius arfakianus, Black-billed Sicklebill Drepanornis albertisi and Zosterops novaeguineae. We consider it likely that the first two of those will be found. Other possibilities are Meyer's Goshawk Accipiter meyerianus, Charmosyna josefinae, Wallace's Owlet-nightjar Aegotheles wallacii, Grey Thornbill Acanthiza cinerea and Obscure Honeyeater Caligavis obscura. Unlikely but not impossible are Bronze Ground Dove Alopecoenas beccarii, Rusty Whistler Pachycephala hyperythra and Daphoenositta papuensis.

Selected species accounts

We provide brief details of significant observations and species for which there were few or no previous Adelbert records.