The largest and highest expanse of montane habitat in New Guinea is the Central Range, which rises to an elevation of nearly 5,000 m. It extends without interruption for 1,800 km from west to east. Not surprisingly, the Central Range supports the most species-rich upland avifauna in New Guinea: about 193 full species or superspecies (Mayr 1941, Rand & Gilliard 1967, Beehler & Pratt 2016).

In addition, along New Guinea's north and north-west coasts lie ten lower and much smaller outlying mountain ranges, isolated from the Central Range and from each other by intervening lowlands (Fig. 1). Almost all of the upland bird populations (defined below) of these outliers belong to species or superspecies shared with the Central Range. The outliers do support many endemic subspecies and c.20 endemic allospecies, but only two endemic full species (Vogelkop Whistler Pachycephala meyeri in the Vogelkop and Emperor Bird of Paradise Paradisaea guilielmi on the Huon Peninsula), both of which are sympatric with and closely related to wide-ranging species.

Figure 1.

Mountains of New Guinea (fig. 1 in Diamond & Bishop 2021b). The Central Range is cross-hatched. In solid black and named are ten outlying mountain ranges, detached from the Central Range and rising from the lowlands along New Guinea's north and north-west coasts. One of those ten ranges, the Cyclops Mts., is the subject of this paper. During Pleistocene cold phases of low sea level, the current Arafura Sea was a large exposed platform joining southern New Guinea and its Fly River bulge to northern Australia. The Aru Islands remain above sea level today, as a surviving modern fragment of that platform. Similarly, Yapen Island remains above sea level as a surviving modern fragment of a platform exposed during the Pleistocene off north-west New Guinea.

The number of upland species reported for each outlier ranges from 37 to 129, in almost perfect correlation with outlier elevation ranging from 1,262 to 4,121 m (table 3 of Diamond & Bishop 2021b). The glaring exception to this correlation is the Cyclops Mts., which reportedly have the fourth-highest elevation (2,158 m) but only the second-lowest upland species number among the outliers. Ernst Mayr, who in 1928 made the first extensive bird collection in the Cyclops, wrote, ‘The first collecting day proved to me that the bird life of the Cyclops Mountains was very poor. I did not hear the notes of any of the interesting species which were so well known to me from the Arfak and Wandammen Mountains, and especially listened in vain for those of the high mountain Birds of Paradise’ (Mayr 1930: 25).

Subsequent surveys supported Mayr's first-day impression: until our own survey of the Cyclops in 1990, only 40 upland species had been reported there. That number was barely higher than the 37 upland species recorded in the lowest and most species-poor outlier, the Van Rees Mts. (elevation only 1,262 m, reportedly 42% lower in elevation than the Cyclops). That Cyclops reported upland species total was much lower than the totals of 77 and 95 species found in the two outliers most similar to the Cyclops in reported elevation (Wandammen and Foja, 2,075 and 2,218 m respectively: Diamond & Bishop 2021b). In particular, Cyclops records were lacking for many common, noisy and conspicuous upland species (such as Fan-tailed Monarch Symposiachrus axillaris and Pygmy Drongo Chaetorhynchus papuensis) recorded on every other outlier, including all six that are reportedly lower in elevation than the Cyclops.

These two linked apparent paradoxes—of an anomalously low upland species total, and no records of otherwise widespread and conspicuous upland species—were what motivated us to resurvey the Cyclops Mts.

Background

Geography.—Fig. 2 is a photograph of the south (inland) face of the Cyclops Mts. rising to the summit. Fig. 3 is a contour map. The mountains are small in area: at the 200-m contour, the max. extent is only 37 km west to east and 6 km north to south. They are also steep, especially on the north (coastal) face. Our impression of the steepness of our trail ascending to the summit from the south foot is confirmed by measurements on the contour map. The average slope of a straight-line ascent to the summit from the nearest southern point on the 200-m contour is 35%. The coastal slope is even steeper, varying between 38% and 62% from the 200-m contour to the 500-m contour. Not surprisingly, we are unaware of any ascent of the Cyclops summit from the north: all recorded ascents have been from the south, e.g., the first recorded ascent by a Dutch military team in 1911 (Anon. 1920), and Dijkstra's 1961 ascent (van Royen 1965).

Figure 2.

The summit of the Cyclops Mts. from the south; note the steepness (K. David Bishop)

Figure 3.

Katherine Rinehart's contour map of the Cyclops Mts. from Google Earth, showing the 200-m, 500-m, 1,000-m and 1,500-m contours. The black dot within the 1,500-m contour is the location of Mt. Rafeni, the highest summit. Google Earth elevation estimates for the Cyclops may be overestimates by up to 8%, at least at high elevations (cf. Rochmadi 2023). Note that the north slope is mostly steeper than the south slope, and that the summit area above the 1,500-m contour is small.

Climate.—Annual rainfall in the lowlands around the Cyclops Mts. varies from c.150 cm on the southern watershed, lying in the rainshadow of the mountains, to c.390 cm on the north coast (Brookfield & Hart 1966). The wetter months are December through April; May–November is drier, with half of the monthly rainfall of the wetter months. Except for one waterfall at 365 m, we encountered no flowing streams along our route in the Cyclops during the dry season (July).

As elsewhere in New Guinea, rainfall evidently increases with elevation, but there are no Cyclops rainfall stations above 400 m. Anecdotally, JD was struck by the impression that temperatures at our Cyclops 1,305-m camp were colder than temperatures he had experienced at similar elevations in the Wandammen Mts., another of the outlying mountain ranges similar to the Cyclops in elevation and coastal location.

Vegetation.—Cyclops forest on a route close to ours was described by van Royen (1965). Lowland forest up to 600 m elevation is dominated by Pometia pinnata trees up to 30 m tall, plus the usual New Guinea lowland tree genera such as Anisoptera, Dillenia, Intsia, Myristica and Pandanus. Above 600 m oaks (Castanopsis and Quercus) become dominant, with occasional emergent Araucaria cunninghamii. Mossing increases above 800 m. At elevations above 1,000 m beeches (Nothofagus) and Myrtaceae dominate, joined still higher by conifers (Dacrydium, Papuacedrus and Phyllocladus) and Rapanea. The highest elevations near the summit support a stunted, wet, heavily mossed forest dominated by Nothofagus, with bamboo, ferns and lichens. Forests on ultrabasic soils east of our route are somewhat similar floristically but stunted, with narrower and more crowded trunks.

Studies of Cyclops birds by others.—Six studies of Cyclops Mts. birds have been reported prior to our visit. Four of those studies collected specimens, and three (by Gjellerup, Mayr and Beehler) reached or probably reached the summit from the south, apparently by similar routes to each other and to us. In this paper, we do not consider collections and observations near sea level in the lowlands near Jayapura (formerly called Hollandia), Humboldt Bay and Lake Sentani by Beaufort, Goodfellow, Mayr, the Third Archbold Expedition (Gilliard 1969: 430–432) and recent observers. Taxonomic nomenclature follows that of Beehler & Pratt (2016), except for our treatment of the perplexing Sericornis virgatus complex.

The first collections of Cyclops birds were made in 1903 by L. F. de Beaufort (1909) in the course of four months spent at Humboldt Bay. While he did not describe his itinerary in the Cyclops Mts., his specimen records show that he collected 22 specimens of 14 species, nine of them upland taxa, on 13–15 April at elevations of 900–1,270 m.

The next collection, reported by Hartert (1932), was made in 1911 by K. Gjellerup, the medical officer with a Dutch military team that reached the Cyclops summit (Anon. 1920). Labels of Gjellerup's specimens that Hartert examined in the Bogor Museum show that, in addition to making extensive lowland collections at Hollandia, Humboldt Bay and Lake Sentani, Gjellerup collected specimens in the Cyclops Mts. during 16–23 June. At elevations that his labels list as ± 1,000 to ± 2,000 m and denote as ‘Cyclopengebirge,’ he collected 27 specimens of 15 species, 13 of them upland species.

By far the largest collection of Cyclops birds was made by Ernst Mayr (1930) in 1928. Setting out from Ifar on Lake Sentani on 18 August, he ascended to a camp at 800 m, then to a higher camp at 1,400 m, from which he climbed several times to the summit, before returning on 14 September to Ifar. His c.400 Cyclops specimens, again described by Hartert (1930), include 35 upland species.

The Third Archbold Expedition of 1938–39 (Archbold et al. 1942), while best known for its discovery and exploration of the Baliem Valley, used Hollandia as its base. In addition to specimens of many species labelled ‘Hollandia' or ‘Lake Sentani,' the expedition also collected 74 species labelled ‘Cyclops Mts.' (Rand 1942), a few of them noted as having been taken at 150 m, but the others without indication of elevation. Most of the ‘Cyclops Mts.’ specimens were collected on dates (25 March, 5 May, and 13 October–17 December 1938) when the expedition ornithologist A. L. Rand was working far from the Cyclops Mts., in or near the Baliem Valley. Those ‘Cyclops Mts.’ specimens were instead probably collected by a Mr Ebeli, who operated a sawmill on the lower south-west slopes of the Cyclops Mts. (Archbold et al. 1942: 209), and who is mentioned as having collected many grassland and marsh birds for the expedition near Lake Sentani. Most by far of the Archbold Expedition’s 74 species labelled ‘Cyclops Mts.’ are actually lowland species, and about 14 (such as Tree Martin Petrochelidon nigricans and Blue-tailed Bee-eater Merops philippinus) are open-country species that could not have been taken in Cyclops Mts. forest. However, the specimens labelled ‘Cyclops Mts.’ also include four upland forest species, all of them also collected or observed in the Cyclops Mts. by other observers. Hence our list of Cyclops Mts. records (Table 1) will include those four Archbold Expedition records, but not the expedition’s other ‘Cyclops Mts.’ records, all of which are lowland species probably taken in the lowlands or very low in the foothills.

TABLE 1

Cyclops upland species. For all 44 of the ‘upland species’ (as defined in the text) recorded in the Cyclops Mts., the column ‘observers’ lists the observers named in Table 3 who observed that species (J = Diamond & Bishop, M = Mayr, B = Beaufort, G = Gjellerup, A = Archbold Expedition, L = Beehler, D = Bishop, N = RAP). ‘Range’ is the elevational range in metres, and ‘abundance’ is the abundance rank (1 = just one or two records, 2 = ≥3–12 records, 3 = common [14–28 records], 4 = the most abundant species with 15–31 records), as observed by us on the summit transect (--- = not observed by us).

Continued

On 4–8 April 1980 Bruce Beehler ascended from Lake Sentani to the Cyclops Mts. summit, observed 66 species and captured 11 of them in mist-nets. His unpublished list, which he kindly provided to us, includes 25 upland species, three of them not previously recorded for the Cyclops Mts.

In September and December 1983 one of us (KDB) ascended the Cyclops Mts. to 365 m above Angkasa and observed 46 species, including three upland species, one of which (Red-fronted Lorikeet Charmosyna rubronotata) was a first record for the Cyclops Mts.

Finally, in August 2000, after our 1990 study (see below), a multi-institutional group conducted a Rapid Assessment Program (RAP) training course up to an elevation of 70 m near Yongsu on the Cyclops Mts. north coastal slope (Setio et al. 2002). To date, this is the only study of birds on the Cyclops north slope, which rises steeply from the sea without an intervening coastal plain. The group recorded five upland forest species, all of them previously observed in the Cyclops Mts.

In short, of those seven studies of Cyclops Mts. birds, five ascended from the southeast, which offers the most direct (and to date the only) route to the summit. The Archbold Expedition collection was made in the foothills and/or lowlands at an unknown site or sites on the southern watershed. Only the Rapid Assessment Program surveyed the northern watershed, at low elevations.

Our study.—On 1 July 1990 we arrived in Jayapura. On 3 July we made a trial ascent of the Cyclops, from Pos Social at 527 m at the mountains' south-eastern base, along a trail up to a helipad that had been constructed in a clearing at 1,305 m in order to salvage a crashed Second World War airplane. The trail went initially through gardens, then through second-growth woods from 332 m, merging into forest. We then flew to a different New Guinea site, while our Dani field associates prepared the helipad as our campsite and cleared trails. Returning to Jayapura, we devoted 22 July to ascending from Pos Social to the waterfall at 365 m, where we found the expected Torrent Flycatcher Monachella muelleriana.

On 23 July we climbed to our 1,305-m mountain camp at the helipad, where we remained until returning to Jayapura on 28 July. Each day from 24 July through 27 July, we alternated one of us climbing to the summit while the other descended through forest to c.670 m. The whole trail above about 314 m was steep.

We repeatedly measured elevations of trail reference points and of significant bird sightings, using Thommen 2000 altimeters. These are the altimeters that we used in our other field studies in New Guinea (e.g., Diamond & Bishop 2015, 2021a,b). We confirmed their accuracy by many readings at sea level and at airstrips of known elevation. Our six measurements of the elevation of the Cyclops summit (Mt. Rafeni, 02.52^oS, 145.50^oE) on 24, 25, 26 and 27 July yielded a value of 1,906 ± 9 m (mean ± standard error of the mean). Looking along the summit ridge, we saw that there was no higher elevation. Our value is in reasonable agreement with the only other two measurements known to us as having been obtained by people standing on Mt. Rafeni's summit and holding an altimeter: 1,880 m, measured by J. Dijkstra of the Dutch Forestry Department on 4 July 1961 (van Royen 1965: 455); and 1,950 m, measured by J. ten Klooster, P. Hubrecht and K. Gjellerup of a Dutch New Guinea Military Exploration team in June 1911 (Anon. 1920: 57, 190, 401 and map 2).¹

On 29 July, our last day of field work, we drove west to Doyo Baru village (02.55^oS, 140.47^oE) at the south-west foot of the Cyclops Mts. From the village at 113 m, we climbed to enter forest at 152 m and continued uphill to 421 m, where further progress was impeded by a ravine. Our purpose was to determine whether the different location at Doyo Baru was associated with different bird species from those we had encountered on our 22–27 July transect from the mountains' south-east foot. In fact, we encountered no upland species and only four lowland forest species above Doyo Baru that we had not already encountered on our main transect.

In the field, KDB recorded bird vocalisations with a Sony TCM 5000EV cassette tape recorder and Sennheiser directional microphone. He began recording before dawn, continued while on the trail, and recorded occasionally at night from camp (cf. Papuan Hawk-Owl Uroglaux dimorpha and Mountain Owlet-nightjar Aegotheles albertisii). Back in camp each day, and again in Jayapura at the conclusion of our field work, we repeatedly listened to these tapes together. That served several purposes. One was that it often enabled us to notice and identify birds whose vocalisations had escaped our attention in the field. Another was that we could concentrate on fine points of species discrimination by voice, e.g., the confusingly similar calls of Ptilinopus fruit doves, of which ten species occur sympatrically in the Cyclops Mts. Still a third purpose was to give us practice in identifying Cyclops bird populations with distinctive local dialects. Finally, KDB recorded and played back in the field puzzling vocalisations, thereby sometimes attracting the singer so that it could be identified by sight.

Species observed

We dichotomise species observed in the Cyclops Mts. as either ‘upland species’ or ‘lowland species’. As in our other papers about birds of New Guinea mountains (Diamond & Bishop 2015, 2020, 2021a,b), we define upland species as species ‘largely confined to sloping elevated terrain, and absent from the flat lowlands at or near sea level’ (Diamond & Bishop 2021b: 454). That definition proves more useful and less arbitrary than defining ‘montane species’ as those largely confined to elevations above some completely arbitrary value (e.g., 1,500 m). Our previous papers discuss in detail the advantages and the ambiguities of this definition. We define ‘lowland species’ as those regularly occurring in the flat lowlands at or near sea level, irrespective of how high is their elevational ceiling.

By this definition, Table 1 recognises 44 upland species as having been recorded in the Cyclops Mts. All 44 have been recorded on the forested part (i.e., excluding gardens and cleared areas up to c.330 m) of the summit transect used by us and by four other observers (Beaufort, Gjellerup, Mayr, Beehler). Table 1 also notes the few records of upland species on two other transects (the Third Archbold Expedition ‘Cyclops Mts.’ collection, and the RAP north slope study)—but all of the upland species recorded in those studies were also recorded on the summit transect. Similarly, the four upland species encountered on our Doyo Baru transect of 28 July (Claret-breasted Fruit Dove Ptilinopus viridis, Papuan Cicadabird Edolisoma incertum, Piping Bellbird Ornorectes cristatus and Hooded Pitohui Pitohui dichrous) had all been encountered on our summit transect.

Table 2 lists lowland species recorded by one or more of the five observers on the forested summit transect. Because the focus of this paper is on Cyclops upland species, we do not tabulate the several dozen other lowland forest species recorded elsewhere in the Cyclops area by the Archbold Expedition, Mayr, ourselves, or many other observers around Lake Sentani and its marshes and other locations near sea level at the foot of the Cyclops Mts.

TABLE 2

Lowland species of the Cyclops summit transect. All lowland species (as defined in the text) recorded by us or by one or more of the four other observers on the Cyclops forested summit transect. ‘Ceiling’ is the elevational ceiling in metres (floors of all of these lowland species are at or near sea level), and ‘abundance’ is the abundance rank in the four categories of Table 1 (1 = least abundant, 4 = most abundant), as observed by us on the summit transect. Ceiling and abundance are not available for the six lowland species recorded by another observer but not by us.

Continued

Continued

TABLE 3

Accumulation of Cyclops upland species records. For each of the eight observers who recorded upland bird species in the Cyclops Mts., the table lists their year of observation, the elevation that they reached, their month(s) of observation, the number of upland species recorded, the number of upland species added (i.e., not recorded by previous observers), and the cumulative number of species recorded by all observers up to that time. Note that only nine upland species have been added since Mayr's 1928 study, although (i) the Archbold Expedition and the RAP team surveyed sites different from the five other observers who shared the same summit transect; (ii) Beehler and the RAP team had the advantage of using mist-nets; (iii) the four observers from Beehler onwards had the advantage of knowledge of New Guinea bird vocalisations; and (iv) all months of the year except January and February have now been studied.

Table 1 confirms Mayr's 1928 impression of the poverty of the Cyclops upland avifauna. The Cyclops are still the second-poorest outlier in upland species: only the much lower Van Rees Mts. (1,262 m) are poorer, with 37 upland species (Diamond & Bishop 2021b). Table 3 shows the accumulation of knowledge of Cyclops upland species. Subsequent observers have added only nine species to Mayr's 35, despite Beehler and both of us having had the advantage of modern familiarity with New Guinea bird vocalisations, Beehler and the RAP team the advantage of mist-nets, and us the advantage of being able to devote all of our field time to observing and to tape recording. (Mayr devoted much of his time to collecting and preparing specimens.) Reasons why the nine added upland species had escaped the attention of Mayr and the three other early collectors include that they are very local (Monachella muelleriana only at the waterfall), or nocturnal (Aegotheles albertisii), or difficult to collect (the soaring Pygmy Eagle Hieraaetus weiskei, the high-flying Blue-collared Parrot Geoffroyus simplex), or cryptic (Bronze Ground Dove Alopecoenas beccarii), or inconspicuous (Red-breasted Pygmy Parrot Micropsitta bruijnii), or difficult to identify without modern knowledge of vocalisations (Aegotheles albertisii, Barred Cuckooshrike Coracina lineata).

TABLE 4

Species or superspecies widespread on other outliers, but unrecorded from the Cyclops. Of New Guinea's ten outlying mountain ranges (Fig. 1), the Cyclops stand out in having no records for many otherwise-widespread species present on all nine of the other outliers, or on most (seven or eight) of them. The table lists these apparently missing in the Cyclops species, most of which are so easily detectable that they are unlikely to be present but overlooked.

Apparently missing upland species

Of New Guinea's c.193 upland species or superspecies, nine have been recorded from all ten of New Guinea's outlying mountain ranges, including the Cyclops (Diamond & Bishop 2021b). But 22 upland species that are widespread on the other outliers have not been recorded from Cyclops (Table 4). Of those 22 species, seven have been recorded on all nine other outliers and are unrecorded only on Cyclops; eight have been recorded on eight of the nine other outliers; and seven have been recorded on seven of the nine other outliers. The only other outlier that approaches Cyclops in its number of unrecorded otherwise-widespread upland species is the lowest and most species-poor outlier, Van Rees, with 18 unrecorded widespread upland species (Diamond & Bishop 2021b).

Are those 22 unrecorded species really absent from Cyclops, at least from the well-studied summit transect, or might they be present but overlooked? Five of the 22 might have been overlooked. We daily observed a group of brown Aerodramus swiftlets flying over our 1,305-m camp in the afternoon, as did Beehler (pers. comm.) at his 1,000-m camp. However, Mountain Swiftlet A. hirundinaceus is very difficult to distinguish from the lowland Uniform Swiftlet A. vanikorensis in the field. The elevations of 1,305 and 1,000 m are compatible with either swiftlet species. Great Woodswallow Artamus maximus and Blue-faced Parrotfinch Erythrura trichroa are nomadic species: sometimes present and recorded at a site, sometimes unrecorded there. The nocturnal Feline Owlet-nightjar Aegotheles insignis is detectable in the field almost exclusively by its calls at night, but the elevation of our mountain camp where we remained at night was below the species' elevational range. Spotted Honeyeater Xanthotis polygrammus is uncommon and inconspicuous; it was collected by Mayr and by the Archbold Expedition at the foot of the Cyclops Mts., so it presumably occurs in the mountains but has not yet been recorded there.

With those five possible exceptions, we consider that the other 17 widespread upland species unrecorded from the Cyclops are unlikely to be present but overlooked, at least on the summit transect. All of them would have been easily detected, because 15 of them are common or abundant, 12 of the 15 possess distinctive, frequently given and loud vocalisations, MacGregor's Bowerbird Amblyornis macgregoriae is detectable by its bowers, and Pheasant Pigeon Otidiphaps nobilis and New Guinea Vulturine Parrot Psittrichas fulgidus occur in low numbers but have far-carrying calls. (Psittrichas fulgidus has been hunted to extinction throughout much of its range; perhaps it has similarly been hunted out in the Cyclops Mts., although that would be difficult on the extremely steep, uninhabited, unclimbed Cyclops north slope.) Of course, those 17 upland species and others might occur elsewhere in the Cyclops Mts. but might for some unknown reason be absent from the summit transect used by all five observers who reached Cyclops elevations above 150 m. For example, Beehler & Prawiradilaga (2010: 284) mentioned a local Cyclops bird-hunter's description of shooting, in the Cyclops northern foothills, a bird that the hunter considered most similar to a male Lawes's Parotia Parotia lawesii. It remains for future studies to confirm whether those 17 unrecorded upland species really are absent from the Cyclops Mts.

Upland vs. lowland species at different elevations

Fig. 4 depicts the changes, with elevation, in the numbers of lowland and upland species. As expected, lowland species greatly predominate at low elevations. They decrease steadily in number until they disappear by around 1,550 m. Even by 1,400 m, lowland species have become rare: there were only three lowland species (Chestnut-breasted Cuckoo Cacomantis castaneiventris, Red-bellied Pitta Erythropitta erythrogaster and Little Shrikethrush Colluricincla megarhyncha) of which we encountered more than a single individual at or above 1,400 m.

In contrast, we encountered upland species over our entire forested transect, from 330 m to the summit. They reached their greatest diversity (22 species) around 1,200 m, above which they decreased in diversity, even above the elevation at which lowland species disappeared. On the summit itself (1,908 m), where we spent an hour on each of four days, we encountered a total of just eight species discussed below: all of them abundant, six of them small passerines, and six of them with elevational floors of 1,512–1,585 m. The two summit species with lower floors were Fan-tailed Berrypecker Melanocharis versteri and Mountain Fruit Dove Ptilinopus bellus, with floors of 1,287 and 674 m respectively.

Figure 4.

Number of upland species (¤), lowland species (x) and their sum (all species •) that we observed on our Cyclops summit transect, as a function of elevation. As expected, lowland species decrease in number with elevation and disappear several hundred metres below the summit. Also as expected, upland species initially increase in number with elevation, but they then decline above 1,200 m to only eight species at the summit.

Upland species impoverishment vs. elevation

The previous sections confirmed Mayr's impression that the Cyclops Mts. really are poor in upland species and lack many otherwise-widespread upland species. Is there any pattern to this impoverishment?

We addressed this question by an analysis similar to that we had carried out for the Van Rees Mts., the only outlying mountain range that is even poorer in upland species than the Cyclops (Diamond & Bishop 2021b). The Van Rees Mts. lie to the west of the much higher, more extensive and more species-rich Foja Mts., which lie in turn to the west of the Cyclops Mts. (Fig. 1). We had divided Foja Mts. upland species into sets according to their elevational floor in the Foja Mts. For each set, we calculated the percentage of that set's species also present in the Van Rees Mts. (fig. 4 in Diamond & Bishop 2021b). We found that the percentage of Foja upland species present in the Van Rees Mts. decreased regularly with elevation, from 88% for the species set with the lowest Foja floors (only 266 m) to virtually zero for species with Foja floors above 1,200 m. That is understandable, because the elevation of the Van Rees Mts. is only 1,262 m, so of course the Van Rees Mts. are too low to support species with floors above 1,200 m.

Figure 5.

The upland species recorded in the Foja Mts. were divided into sets based on their Foja elevational floors (abscissa, in metres, from Beehler et al. 2012). For each set, the ordinate gives the percentage of the Foja upland species in that set that has also been recorded in the Van Rees Mts. (curve from fig. 4 of Diamond & Bishop 2021b) or from the Cyclops Mts. (points • from our observations in this paper). See text for discussion.

As depicted in Fig. 5, a similar analysis for the Cyclops Mts. yields results that are somewhat similar, but with two differences. In the Cyclops Mts., as in the Van Rees Mts., Foja upland species with low Foja floors (<800 m) are better represented than are Foja upland species with high Foja floors (>1,200 m). But: even in the species set with the highest Foja floors (≥1,450 m), 22% are present in the Cyclops Mts., which too is predictable, because (unlike the Van Rees Mts.) the Cyclops summit of 1,906 m is much higher than that Foja floor. And: even in the species set with the lowest Foja floors (294 m), only 57% of the species are present in the Cyclops Mts., compared to 88% in the Van Rees Mts.

Why, then, are the Cyclops Mts. impoverished at low elevations as well as at high elevations? We shall return to this question.

The summit avifauna

Table 5 lists characteristics of the eight species that we observed on the Cyclops summit. All eight are abundant there: five are in our highest abundance category (category 4) and the other three are in our second-highest category (3). All are widespread on New Guinea's ten outlying mountains: one species occurs on all ten outliers, five on eight outliers, one is present on six outliers, and one species (Mayr's Streaked Honeyeater Ptiloprora mayri) is on four outliers but is replaced by another similar-sized and closely related allospecies or congener on four other outliers.

Seven of the eight summit species are passerines similar to each other in five respects: all belong to species, genera or families endemic to mainland New Guinea; all are absent from all offshore islands in the New Guinea region; six are small (body mass 11–30 g, the seventh passerine 56 g); all of them have high elevational floors (six at 1,512–1,585 m, one at 1,287 m), hence their populations are crammed into a small area extending only a short distance below the Cyclops summit; and all seven forage regularly in the forest understorey and hence are regularly caught in mist-nets on other New Guinea mountains where we have used mist-nets. (We did not use mist-nets in the Cyclops Mts.)

TABLE 5

Summit species. For each of the eight species that we encountered on the Cyclops summit, the columns from left to right give: 1. On how many of New Guinea's ten outlying mountain ranges is the species present? 2. At what level is the group to which the species belongs endemic to New Guinea? (For example, Ptilinopus bellus is a New Guinea endemic allospecies, Ptiloprora is a New Guinea endemic genus, Melanocharis versteri belongs to a New Guinea endemic family, and the genus Peneothello is endemic to New Guinea except that one of its species extends into northern Australia). 3. At which taxonomic level, if any, is the Cyclops population endemic to the Cyclops? (Three of the populations constitute Cyclops endemic subspecies). 4. Mass (g), mostly from specimens collected in the Cyclops and weighed by Ernst Mayr. 5. Foraging stratum (grd = ground, ls = lower storey, ms = midstorey, cr = crown). 6. Abundance, using the categories of Table 1 (3 = common, 4 = the most abundant species). 7. Elevational floor (m).

The partial exception among these eight summit species is the only non-passerine, Ptilinopus bellus. Like the seven summit passerines, it is abundant (category 4) and it is widespread on the outliers (present on all ten). It differs from the seven summit passerines in being much larger (body mass 152 g), descending in the Cyclops to a much lower elevation (674 m), being a canopy species rarely caught in mist-nets, reaching an offshore island near New Guinea (Karkar Island) and otherwise being endemic to New Guinea only at the allospecies level (about five other allospecies are distributed from the Moluccas to the Solomons).

One may object that we made only four visits of one hour each to the Cyclops summit. Would we have reached different conclusions about the summit avifauna if we had spent more time and accumulated more species there? Presumably, additional summit species would have been drawn mainly from species that we observed just below the summit but not on the summit itself. At elevations down to 1,768 m, 140 m below the summit, we encountered six species that we did not observe on the summit itself: Papuan Mountain Pigeon Gymnophaps albertisii, the parrot Geoffroyus simplex, Black-bellied Cicadabird Edolisoma montanum, Island Leaf Warbler Seicercus poliocephalus, Black Fantail Rhipidura atra and Pitohui dichrous.

Two of these six almost-summit-species (the pigeon and the parrot) resemble Ptilinopus bellus in being large canopy-dwelling non-passerines that occupy many outliers (ten and eight respectively) and that descend in the Cyclops to 610 m and 994 m respectively, but they are less abundant than P. bellus (only abundance rank 2). Three others of the six almost-summit-species—the warbler, the fantail and the pitohui—resemble the summit's seven passerine species in being small abundant species often found in the understorey, and endemic to New Guinea at the allospecies, species and genus level, respectively. The last of the six—Edolisoma montanum—is an abundant passerine endemic to New Guinea and widespread on the outliers (on nine of them) but lives in the canopy, unlike the seven species of summit passerines.

In a later section, we shall review these characteristics of the Cyclops summit avifauna in the wider context of its origins.

Endemics

The only large collection of bird specimens from the Cyclops Mts. is that made by Mayr in 1928. From that collection, Hartert (1930) described eight subspecies endemic to the Cyclops Mts., and Mayr (1931) described a ninth (Eupetes = Ptilorrhoa leucosticta sibilans) (Table 6). No new subspecies had been described from the earlier and much smaller Cyclops collections made by Beaufort and Gjellerup.

Hartert's and Mayr's diagnoses of nine Cyclops endemics have been reassessed by Mayr (1941), Rand & Gilliard (1967) and Beehler & Pratt (2016). Several of the Cyclops endemics were also reassessed by Diamond (1969, unpubl.) and by Beehler & Prawiradilaga (2010) in light of post-1960 explorations of the North Coastal Range and Foja Mts. respectively, and of the discoveries there of populations related to or identical to several of Hartert's described Cyclops endemics.

Table 6 summarises our view (close but not identical to that of Beehler & Pratt 2016) of the current taxonomic status of the nine claimed Cyclops endemics. As do Beehler & Pratt (2016), we consider one of Hartert's taxa (Piping Bellbird Pitohui cristatus arthuri) insufficiently distinctive to recognise taxonomically. Of the other eight claimed endemics, two (the populations of the two Charmosyna species) are no longer considered endemic to the Cyclops, because the latter's populations proved similar to subsequently discovered conspecific populations in the North Coastal Range and (Fairy Lorikeet C. pulchella rothschildi) in the Foja Mts. Also subsequently discovered in the North Coastal Range and Foja Mts. were Rallicula and Ptiloprora populations related to but diagnosably different from their counterparts in the Cyclops. Hence the Rallicula and Ptiloprora populations of the Cyclops, Foja, and North Coastal Range are now considered to be separate allospecies of superspecies widespread in the Central Range and on other outliers, but with the Cyclops populations constituting distinct subspecies. Paradoxically, in both cases the western (Foja) and eastern (North Coastal Range) populations are more similar to each other than either is to the geographically intermediate Cyclops population.

TABLE 6

Taxa described as endemic to the Cyclops. Left column: original description as a taxon endemic to the Cyclops, either by H = Hartert (1930) or by M = Mayr (1931). Middle column: current taxonomic status as judged by recent reviewers, especially Beehler & Pratt (2016) and/or us. (Abbreviations: NCR = North Coastal Range; CR = Central Range). Right column: current name in Beehler & Pratt (2016), except that they sink Pachycephala schlegelii cyclopum in P. s. obscurior of the Central Range.

Finally, we mention that while we, Mayr (1941), Rand & Gilliard (1967) and Beehler & Pratt (2016) are mostly in agreement that six Cyclops endemics deserve subspecific recognition, none of them is very distinctive. That reinforces the distributional paradox of Cyclops ornithogeography. On the one hand, most New Guinea upland bird species have been unable to establish themselves in the Cyclops Mts., suggesting that those mountains have been difficult to reach or else to colonise. On the other hand, the Cyclops populations of those species that did establish themselves are not very distinct, suggesting the opposite conclusion: that the Cyclops Mts. have been easy for those species to reach. We shall suggest a resolution to this paradox in a later section ‘Why are the Cyclops Mts. poor in upland bird species?’.

How have upland species reached the Cyclops Mts.?

All 44 upland species that have Cyclops populations are also represented on the Central Range, and on many or all of the other nine outliers, by populations that are taxonomically identical in most cases, or by recognisably distinct but similar subspecies or allospecies in eight cases. (See the section above and Table 6, on endemism.) How do upland populations disperse and become established on the Cyclops Mts. and on other outliers?

In well-studied areas elsewhere in the world, bird dispersal has been documented directly by banding or radio-tagging individual birds, by flying ultra-light aircraft among migrating birds, and by radar and sound tracking of migrants at night (Lovette & Fitzpatrick 2016). Such direct evidence is unavailable for New Guinea. But eight other types of evidence (1a–3b, Table 7) suggest the possibility of at least three mechanisms of dispersal between New Guinea mountains: 1. Direct flights from mountain to mountain. 2. Re-ascending a different mountain after descent to the lowlands. 3. Shifts of species elevational ranges associated with Pleistocene climatic cycles.

1a. Flights high over the canopy.—Some species—especially some nectarivorous and frugivorous parrots, frugivorous pigeons, soaring hawks, and swifts—fly over the canopy for long distances in the course of their daily foraging. Such flights permit direct colonisation of upland habitats mountain to mountain. Among Cyclops upland species, we observed such flights (column 1a of Table 7) by flocks of one frugivorous parrot species (Geoffroyus simplex), three nectarivorous parrot species (three Charmosyna lories), and two frugivorous pigeons (Ornate Fruit Dove Ptilinopus ornatus and Gymnophaps albertisii); by one individual of a soaring hawk (Pygmy Eagle Hieraaetus morphnoides); and possibly by one swiftlet species, if the swiftlets that regularly flew over our mountain camp were the upland Aerodramus hirundinaceus.

1b. Overwater colonisation of oceanic islands.—Among the islands surrounding New Guinea are numerous ‘oceanic’ islands—i.e., islands that rise from deep water beyond New Guinea’s continental shelf, and that therefore were not connected to New Guinea even at Pleistocene times of low sea level. Bird populations on such islands must have reached them by direct overwater flights. For example, three Cyclops upland species have populations belonging to the New Guinea mainland subspecies on Karkar, a deep-water volcanic island off New Guinea’s north-east coast, probably defaunated by volcanic eruptions then recolonised several times during recent millennia (Diamond & LeCroy 1979, Johnson 1981). In all, six Cyclops upland species (denoted in column 1b of Table 7) have colonised Karkar and/or other New Guinea oceanic islands (Goodenough, Ferguson, and Gam) that could have been reached only by direct flights, and occur there as a New Guinea mainland subspecies, suggesting recent overwater colonisation. Seven other Cyclops upland species (denoted (√) in column 1b) occur on oceanic islands as insular subspecies distinct from the New Guinea mainland subspecies, suggesting overwater colonisation less recently.

TABLE 7

Correlates of means of dispersal by Cyclops upland species. Correlates of four means of dispersal—flights above the canopy today, dispersal through the lowlands today, dispersal in the Pleistocene, and means uncertain or speculative—by which New Guinea upland species may have reached the Cyclops Mts. For each of these four, we denote by a checkmark which of those plausible correlates (1a–d, 2a, 2b, 3a, 3b and 4) of that means of dispersal characterise that species. See text for discussion. In column 1b, = present on an oceanic island as a New Guinea mainland subspecies, suggesting recent overwater colonisation; () = present on an oceanic island as a distinct insular subspecies, suggesting less recent overwater colonisation. Note that all 44 Cyclops upland populations possess at least one correlate of at least one means of dispersal. Right-hand column (no. of outliers): of New Guinea's ten outlying mountain ranges, how many ranges does that species or superspecies occupy? Note that the 44 Cyclops upland species occupy on average 8.1 ± 0.2 (mean ± standard error of the mean) of the ten outliers, and that 37 of the 44 species occupy at least seven of the ten outliers.

Continued

1c. Living in, flying through or flying just over the canopy.—Many species of New Guinea birds, including at least 14 Cyclops upland species, forage predominantly or frequently in the canopy (column 1c of Table 7). Some of these are among the species mentioned previously as also flying high above the canopy, but most have not been observed doing so. For example, Edolisoma montanum forages predominantly in the canopy, in groups of up to four individuals. At intervals of a fraction of an hour, the whole group flies to another canopy site within a few hundred metres, giving a distinctive flight call, and flying up to 20 m above the canopy, but not higher.

The previously mentioned upland species (column 1a) that routinely fly high above the canopy seem more likely to fly between mountains than are the currently discussed upland species that live in and fly through or just over the canopy but do not routinely fly high above it. However, the latter species are still candidates to colonise mountains directly from other mountains, either voluntarily or involuntarily at times of high winds.

1d. Frugivores and flower feeders (column 1d of Table 7).—Our group of upland species that routinely fly high above the canopy consists mainly of large non-passerine frugivores and flower-feeders. But there are many other frugivores and flower-feeders that have never been observed to fly high above the canopy: small passerines, especially honeyeaters (Meliphagidae) and white-eyes (Zosteropidae). (The flower-feeding passerines visit flowers to obtain both nectar and insects.) Fruit and flower sources in the forest are ephemeral, so frugivores and flower-feeders must shift location to track shifts in plant fruiting and flowering. Among Myzomela honeyeaters and Zosterops white-eyes, those shifts include elevational migrations. We know nothing about whether and how those local movements might translate into movements between mountain ranges. Somehow, they evidently do become so translated, because all four Cyclops species of upland Myzomela and Zosterops occur on many outliers (on average 8.5 outliers per species). But not all the frugivores in column 1d fly above the canopy: notably not the Common Smoky Honeyeater Melipotes fumigatus.

2a. Seasonal commuting between uplands and lowlands.—Three New Guinea species of frugivorous pigeons (column 2a of Table 7)—Gymnophaps albertisii, Ptilinopus ornatus and Macropygia nigrirostris—nest and vocalise on New Guinea mountains. After breeding, they descend to the lowlands, where they are silent, and where Ptilinopus ornatus and Gymnophaps albertisii each form wide-ranging flocks. From the lowlands, those birds re-ascend mountains to breed—potentially, either the same mountain from which they just descended, or a different mountain. Not surprisingly, those three pigeon species occupy respectively all ten, nine, and all ten of New Guinea's outliers. Equally unsurprisingly, they exhibit no geographic variation within New Guinea, or in the case of Ptilinopus ornatus have just one geographic isolate besides one wide-ranging subspecies—because, we speculate, they apparently have the opportunity to colonise a new mountain every year.

2b. Non-breeding stragglers in the lowlands.—For at least nine other Cyclops upland species (column 2b of Table 7), there are records of non-breeding individuals, usually immatures, in the lowlands far below the species' usual range where it vocalises and breeds (Stein 1936, Diamond 1972, Pruett-Jones & Pruett-Jones 1986, Beehler & Pratt 2016). As in the preceding paragraph about seasonal commuting, descents to the lowlands by such individuals offer the potential for dispersal through the lowlands, followed by ascent of a mountain different from the natal mountain. We guess that the cases discussed in this paragraph involve once-in-a-lifetime dispersal of immatures, whereas the cases discussed in the preceding paragraph involve annual commuting by adult breeders. But in both cases, the result is dispersal through the lowlands, rather than direct dispersal from mountain to mountain.

3. Dispersal driven by Pleistocene climate shifts.—In this section, our discussion so far has concerned dispersal mechanisms operating today, either directly between mountains or via the lowlands. In the past, an additional mechanism contributed to dispersal via the lowlands. During the Pleistocene, alternate phases of warmer and colder climate caused zones of climate and vegetation to shift up and down New Guinea mountains. Evidence of those shifts is especially visible today as glacial moraines and other signs of glaciation more than 1,000 m below New Guinea's current snowline.

Modern global warming is causing such shifts to occur even more rapidly today. For example, when in 1993 we surveyed birds in the Star Mts., near modern New Guinea's easternmost ice field on Mt. Mandala, the snowline that we saw lay at about 4,300 m. Today, that snowline has shifted at least 400 m upwards, because Mt. Mandela is now snow-free up to its summit (4,700 m). On Mt. Karimui in Papua New Guinea's Eastern Highlands, where one of us (Diamond 1972) measured elevational ranges of bird species in 1965, Freeman & Class Freeman (2014) found in 2012 that species elevational ceilings and floors had shifted upwards by an average of 113 m and 95 m respectively in the intervening 47 years.

Rather than the outliers being completely isolated from each other and from the Central Range by lowlands at sea level, each outlier is connected to other mountains by ‘bridges’ of raised terrain: bridges of just c.50 m elevation for the Kumawa Mts. and Fakfak Mts., c.150 m for the Cyclops Mts., and c.200 m for the Van Rees Mts. Hence outlier upland populations whose elevational floor is currently above the elevation of an outlier’s highest bridge might have descended during Pleistocene cool phases to the bridge’s elevation, and might thereby have exchanged colonists with mountains at the opposite end of the bridge.

In the absence of precise knowledge of how far Cyclops climate zones descended during the coolest recent phase of the Pleistocene, we suggest two methods for assessing those descents.

3a. One consists of recognising that two areas now close to sea level—southern New Guinea's Fly River bulge, and the Aru Islands that were formerly part of southern New Guinea at Pleistocene times of low sea level (Fig. 1)—today support sea-level populations of 17 species that are largely confined to mountains elsewhere in New Guinea (table 6 in Diamond & Bishop 2020: 441). The Fly River bulge and the Aru Islands would have been c.200 m above sea level at the coldest and lowest-sea-level phases of the Pleistocene. Those 17 species include eight Cyclops upland species (column 3a of Table 7), whose current elevational floor is on average 920 m (range, 546–1,372 m) above sea level. That suggests that at least those eight species could have arrived in the Cyclops during the Pleistocene, via the lowlands or surrounding hills that are currently several hundred to 1,000 m below the elevational floors of all eight of those species.

3b. The other method is to consider which Cyclops upland species have elevational floors low enough that they might conceivably have colonised the Cyclops via the lowlands during Pleistocene cold phases. While we know something about how far New Guinea glacier lower limits dropped during the Pleistocene, we don't know how far climate zones at lower elevations descended. Suppose we assume that low-elevation climate zones dropped by 500, 750 or 1,000 m. Six, ten and 19 Cyclops upland species currently have Cyclops floors at or below 500, 750 and 1,000 m, respectively (column 3b of Table 7). In the Pleistocene, those species would have descended to the sea-level lowlands, could then have colonised the Cyclops through the lowlands, and would have retreated upslope to become Cyclops upland populations as Pleistocene climates became warmer. Six of those 19 species still have sea-level populations today in the Fly River bulge or Aru Islands.

4. Explanation uncertain.—We have thus identified eight factors that might be correlated with dispersal of upland species by three mechanisms, for 32 of the 44 upland species in the Cyclops. However, there remain 12 Cyclops upland species to which we do not recognise any of the possible correlates of dispersal listed in columns 1a–3b of Table 7 as unequivocally applying.

All 12 are forest interior species. All are widespread on New Guinea's outliers: on 6–9 of the ten outliers, average 7.8. All live at least partly in the understorey and can be caught in understorey mist-nets; four are terrestrial, and three occur mainly in the understorey. All with well-established Cyclops elevational floors are high-elevation species: the lowest well-established Cyclops floor among the 13 species is 1,070 m; five species have floors above 1,510 m; and four of those five species plus two other species occur on the summit itself. All 12 are endemic to New Guinea: six, five and one at the level of endemic species, endemic genus and endemic family, respectively.

These attributes of forest-interior, understorey, high-elevation New Guinea endemics are attributes that we associate with New Guinea's most sedentary upland species. Nevertheless, those species are widespread on the outliers and Central Range. How can this paradox be explained? How do they disperse between mountaintops?

It is unlikely that they disperse through the lowlands today: we are unaware of even a single record in the lowlands for any of the 12 species. It's uncertain whether any could have dispersed via the lowlands even during cold phases of the Pleistocene, because the Cyclops floors today of those with well-established Cyclops floors are so high above the lowlands. However, some of the 12 species have lower floors on other New Guinea mountains today, which might have permitted them to descend to the lowlands during the Pleistocene.

A speculative alternative explanation is immature dispersal by flight above the canopy. Most of the 12 species, except perhaps Wattled Brushturkey Aepypodius arfakianus, Spotted Jewel-babbler Ptilorrhoa leucosticta and Crateroscelis robusta, give the impression of possessing normal flying ability. Hence some of these 12 species may be ones that disperse once in their lifetimes, as immatures, possibly at night.

Why are the Cyclops Mts. poor in upland bird species?

Let us now apply what we have learned about the Cyclops Mts. to the question that motivated this study, and that struck Ernst Mayr already on his first day in the Cyclops Mts. Why, among New Guinea's ten outlying mountain ranges, are the Cyclops the second poorest in number of upland bird species? Why do the Cyclops lack records of at least 17 and possibly as many as 22 upland species present on all or most of the other outliers? Why, compared to the Foja and Van Rees outliers, are the Cyclops impoverished in upland species not only at high elevations, but also at low elevations of 300–700 m (Fig. 5)?

A full answer to these questions must await a quantitative multi-factorial analysis of upland species number and its potential determinants on all ten outliers. Pending the necessary data and such an analysis, we already can recognise two geographic contributory factors: elevation and area.

First, the Cyclops are lower in elevation than is widely assumed. They are not 2,160 m high, as shown on current maps based on old aerial photography. Instead, our altimeter readings standing on the summit were only 1,906 ± 9 m, in reasonable agreement with the sole readings obtained by other people standing on the summit: 1,880 and 1,950 m, measured by Dijkstra and by the Dutch Military Exploration team, respectively.

While that modest summit elevation of the Cyclops would obviously contribute to the impoverished number of upland species whose floors elsewhere lie near or above 1,906 m, it doesn't explain the Cyclops' impoverishment at lower elevations. There, the role of area appears important.

The Cyclops are the smallest of the ten outliers in area. Table 8 compares the Cyclops with the next two smallest outliers (Kumawa and Wandammen) with respect to areas above the 200-, 500-, 1,000- and 1,500-m contours. Above each contour, with one exception, Kumawa and Wandammen both have 3.4–6.4 times more area than the Cyclops. (The exception is the low relative area of Kumawa above 1,500 m; the highest elevation of 1,654 m consists of a single small peak.)

Think of the outliers as mountain islands for upland species. Species number S on islands or in island-like habitat patches varies with island or patch area A as A^z, where z values in a wide range of studies lie around 0.3 (MacArthur & Wilson 1967). For example, if A₁/A₂ = 3.9, as is true for the area ratio of Kumawa to Cyclops above 200 m (Table 8), the expected ratio of species number is 3.9^0.3 = 1.50; the actual Kumawa-to-Cyclops ratio of upland species numbers is (72 species/44 species) = 1.64. If A₁/A₂ = 4.5, as is true for the area ratio of Wandammen to Cyclops above 200 m, the expected ratio of species number is 4.5^0.3 = 1.57; the actual Wandammen-to-Cyclops ratio is (79 species/44 species) = 1.80 (see row 2 of Table 8 for the actual upland species numbers).

TABLE 8

Areas, elevations, and upland species of the three smallest outliers. Summit elevations were measured using the same altimeters for all three mountains by both of us (Cyclops and Kumawa) or by JD alone (Wandammen), standing on the summits. Upland species numbers are from this paper for Cyclops, from Diamond & Bishop (2015) for Kumawa, and from surveys by Ernst Mayr (Hartert 1930) and JD for Wandammen. Areas above the stated contour lines were measured by Katherine Rinehart from Google Earth, whose absolute values for New Guinea mountains are subject to small sources of error (Rochmadi 2003), but whose ratios for Kumawa and Wandammen relative to Cyclops appear to be credible approximations. See text for discussion.

Thus, differences in area above the 200-m contour may contribute much of the explanation for species poverty in the Cyclops compared to Kumawa and Wandammen, the next two smallest outliers. Small summit area above 1,500 m (just 7.8 km²) may also contribute to the Cyclops' small summit avifauna of just seven species with floors above 1,500 m. The even smaller summit area of Kumawa above 1,500 m, just 2.3 km², surely helps explain why Kumawa has only three upland species confined to elevations above 1,500 m (Diamond & Bishop 2015).

As for a further likely role of area, from Table 8 the area of Cyclops between the 200-m and 500-m contours is only 169 km², compared to 422 km² for Kumawa and 825 km² for Wandammen. Hence small area even at low elevations is probably the main reason why the Cyclops are impoverished in upland species at low as well as at high elevations. It is striking to see in Fig. 5 that the Cyclops at elevations of 300–700 m are impoverished in upland species compared not only to the larger and higher Foja Mts., but also compared to the larger albeit much lower Van Rees Mts.: only 1,262 m high, vs. 1,906 m for Cyclops!

Why do smaller islands, smaller habitat patches and smaller mountains have fewer species than do larger islands, patches or mountains? A major reason is that island species number approaches a balance between species immigration rates and species extinction rates (MacArthur & Wilson 1967). Species number increases with immigration rates, which decrease with distance from source populations on other islands. Species number decreases with extinction rates, which decrease with increasing population size and hence with increasing island area. That is, smaller islands support smaller populations, which suffer a higher probability per year of extinction, than do the larger populations on larger islands. As New Guinea's smallest outlying mountain range, the Cyclops have the highest extinction rates and hence the second-lowest upland species number.

This interpretation readily explains why the Cyclops upland avifauna has such low endemism. Of the 44 currently known Cyclops upland populations, only six belong to Cyclops-endemic subspecies (Table 6), all of them not very distinctive. In contrast, of New Guinea's nine other outliers, two have endemic full species (e.g., the Huon's Paradisaea guilielmi), five have endemic allospecies (e.g., Foja's Golden-fronted Bowerbird Amblyornis flavifrons) and most have distinctive endemic subspecies (e.g., Kumawa's Smoky Robin Peneothello cryptoleuca maxima). One can think of the Cyclops upland avifauna as just the current frame of a rapidly turning over species kaleidoscope: some currently absent species were formerly present, but those former Cyclops populations are now extinct; most currently present populations are recent arrivals and still undifferentiated; only six populations have survived long enough to evolve into weak subspecies; and none has survived long enough to evolve into a distinctive species or allospecies.

Future studies

We mention five of the many problems for which future studies in the Cyclops would be rewarding.

Of the surveys reporting upland bird species in the Cyclops, five used essentially the same route to the highest peak: the surveys by Beaufort, Gjellerup, Mayr, Beehler and us. The four surveys by other routes (by the Archbold Expedition, Bishop, the RAP team and our one-day ascent from Doyo Baru village) encountered only four, three, five and four upland species respectively, all except one of those species also recorded on the summit transect. (That one species, Charmosyna rubronotata, was probably present but not identified on the summit transect among the many Charmosyna lories that we heard daily in flight.) Might the 22 upland species expected but not recorded for the Cyclops occur elsewhere in the Cyclops Mts.? For most of them, we doubt it. But a convincing answer requires surveying other Cyclops upland areas: especially the slightly lower Mt. Dafonsero in the western Cyclops (van Royen 1965), the north coastal slope (Setio et al. 2002) and the ultrabasic terrain in the north-east (Ratcliffe 1984).

Other possible explanations for the Cyclops' upland species deficit warrant evaluation. They include steep terrain and locally ultrabasic soils.

Better understanding of how upland species reach the Cyclops Mts. might be gained from studies seeking to detect individual birds in the act of dispersing. Examples of such possible studies include: mist-netting in lowland forest near the base of the Cyclops Mts.; mist-netting in the mountains at the lower edge of a species' breeding range, where non-breeding immatures have been reported (e.g., Stein 1936, Diamond 1972, Pruett-Jones & Pruett-Jones 1986); and radio-tracking individuals of the three pigeon species that are known to breed at higher elevations but to descend seasonally to the lowlands after breeding (p. 80).

Subspecific morphological characters suggest that some Cyclops populations arrived from the nearest two outliers (Foja and North Coastal Range), while others arrived instead from the Central Range (Table 6). Modern genetic methods could offer more and better markers of the sources of Cyclops populations.

Selected species accounts

We provide brief details of our observations of three upland species lacking published Cyclops records, two Cyclops endemic subspecies, and five other species of particular interest.