Leptomys ernstmayri Rümmler, 1932

Musser, Guy G., Helgen, Kristofer M. & Lunde, Darrin P., 2008, Systematic Review of New Guinea Leptomys (Muridae, Murinae) with Descriptions of Two New Species, American Museum Novitates 3624 (1), pp. 1-60 : 34-40

Leptomys ernstmayri Rümmler, 1932

HOLOTYPE AND TYPE LOCALITY: The holotype of ernstmayri is ZMB 42400, adult male,

study skin and skull, from Ogeramnang (06 ° 269S, 147 ° 229E), 1785 m, Huon Peninsula, collected February 20, 1929, by Ernst Mayr; measurements are listed in table 1. Rümmler (1932) placed Ogeramnang in the Saruwaged Mountains, but the locality is not in that range. H. M. Van Deusen (partial manuscript in AMNH Mammalogy archives) noted that Ogeramnang ‘‘lies on high ground above and just north of the lower Bulum River, and looks south to the Rawlinson Range. The Saruwaged Mountains lie far to the northwest’’. The geographic position of Ogeramnang is clearly portrayed in the published map showing collection sites of the Seventh Archbold Expedition to New Guinea undertaken in 1964 ( Van Deusen, 1978: 19).

DIAGNOSIS AND CONTRASTS: Leptomys ernstmayri is a distinctive species that has usually been misidentified as L. elegans in museum collections and reports published prior to 1993 ( Tate, 1951, and Flannery, 1990, are examples). It is distinguished from L. elegans by the diagnostic morphological traits and altitudinal associations that were carefully enumerated by Rümmler (1932) and illuminated in greater detail by data presented here. Leptomys ernstmayri is smaller in body size (as indicated by length of head and body, and mass; contrast measurements in tables 2 and 3) than L. elegans , but has a much longer tail relative to length of body. This smaller size of L. ernstmayri is also reflected in the lesser cranial and dental dimensions compared with those characteristic of L. elegans (contrast values in tables 4 and 5). An anterolabial cingulum is present in more than half of the total available samples of L. ernstmayri , but is generally absent in our material representing L. elegans (table 6). In an ecological context, L. ernstmayri replaces L. elegans at high altitudes in tropical lower and upper montane rain forests.

The morphological distinctions setting L. ernstmayri apart from L. elegans are generally similar to those separating the former from L. signatus and L. arfakensis , n. sp. (see those accounts). Leptomys signatus has the additional combination of short, grayish upperparts broken by a white blaze on the head, which contrasts it with any specimen of L. ernstmayri we have studied.

Leptomys ernstmayri , in body size and ecology is more similar to a species we describe below that lives in mountain forests covering the Owen Stanley Range, but differs significantly by its larger dimensions (tables 3, 5; figs. 5, 7, 8, 16, 20) and a suite of other morphological traits that are chronicled in the account of L. paulus , n. sp.

GEOGRAPHIC DISTRIBUTION: The geographic distribution of Leptomys ernstmayri is documented by voucher specimens collected from four discrete montane populations, one in western New Guinea, all the others in eastern New Guinea (see gazetteer and map in fig. 3): (1) the Foja Mountains, an isolated north-coast range in western New Guinea, where the species is represented by one specimen collected at 1150 m; (2) the mountain ranges of the Huon Peninsula (including the Saruwaged and Rawlinson ranges), where it is recorded from 1340 m to 1785 m; (3) the isolated Adelbert Range on the northern coast, where it is documented from 1200 m to 1500 m; and (4) the eastern body of New Guinea’s extensive Central Dividing Ranges (Central Cordillera), where it is been collected from Mt. Karimui (on the southern slopes) and the Schrader Range (on the northern slopes) in the west to the headwaters of the Aroa River (on the southern slopes) and Mount Missim and Lake Trist (on the northern slopes) in the east, at altitudes between 1200 m and 2200 m.

Rümmler (1932, 1938) and Musser and Carleton (1993) included the Arfak Mountains within the range of L. ernstmayri , but samples from there represent a distinct species ( L. arfakensis , n. sp.) that we describe in this report. Flannery (1995) and Cole et al. (1997) identified specimens from the Maneau Range of southeastern New Guinea as L. ernstmayri , but these likewise represent another new species ( L. paulus , n. sp.), also described here.

GEOGRAPHIC VARIATION: Each of three geographically isolated populations that are represented by moderately large samples and identified here as L. ernstmayri has a combination of morphological attributes that indicate a history of diversification in isolation. This is reflected especially by divergence among populations in external proportions (table 3), qualitative dental traits (table 6), and cranial and dental morphometrics (figs. 16, 17).

Our sample from the Huon Peninsula is the most phenetically distinctive of the three in external proportions and molar architecture. It is characterized by a very long tail relative to head and body length (table 3), and the fixed presence of a cusp t7 and an anterolabial cingulum on the second upper molar (table 6).

Phenetic contrasts in external and dental traits between samples from the Adelbert Range and Central Cordillera are present but not as notable. Differences in proportions of external traits and qualitative molar structures exist but are average, not exclusive. Compared with the sample from the Adelbert Range, the tail averages longer relative to length of head and body in the Cordilleran sample, and exhibits a relatively shorter white segment (table 3). Cusp t7 on the second upper molar occurs in 15 of the 18 specimens (83%) forming the Cordilleran sample but in only 2 of the 10 animals (20%) from the Adelbert series, and the anterolabial cingulum is found in 4 of the 18 Cordilleran specimens (22%) as compared with 7 of the 10 (70%) from the Adelbert Range (table 6). In the samples from the Adelbert Range, the fur between ears and shoulders is thinner producing a more conspicuous balding pattern than seen in examples from the either the Huon Peninsula or Central Cordillera.

The morphometric structure derived from discriminant-function analyses of cranial and

TABLE 8 Morphometric Separation (Principal Components Analysis) of 39 Adult Leptomys with Small Body Size, Leptomys ernstmayri and Leptomys paulus (Principal components are extracted from a covariance matrix of 17 log-transformed cranial and dental variables; see table 5, fig. 16.)

dental variables in samples of L. ernstmayri reflect a somewhat different pattern of geographic variation (fig. 17). The spread of specimen scores along the first canonical axis forms three clusters isolated from one another, each representing the three separate geographic regions sampled: Huon Peninsula, Adelbert Range, and Central Cordillera. Covariation in interorbital breadth, length of rostrum (as indexed by length of nasals), breadth of braincase, and length of bony palate are primarily responsible for the dispersion of scores along the first axis; lengths of diastema and first upper molar along with breadth of the incisive foramina are also influential but carry less weight (table 9).

The scattering of specimen scores along the second canonical axis reveals a different pattern of covariation in morphometric space that is mostly influenced by size as indicated by the many negative correlations (fig. 17). There the scores for specimens from the Huon Peninsula and Central Cordillera are assembled into overlapping clusters, but those representing the sample from the Adelbert Range form a group isolated from the other two. Variation in breadths of interorbit and rostrum, length of bony palate, and spaciousness of the incisive foramina primarily influence the spread of scores along the second axis that results in the peripheral position of the Adelbert cluster; lengths of skull and molar row contribute but with less force (table 9). Proportionally, these dimensions are greater in the Adelbert sample compared to the other two geographic series.

Distribution of scores along the second axis in both the principal components and canonical variate ordinations reveals another aspect of variation among the three geographic samples (figs. 16, 17). Compared with the tight clusters of scores representing samples from the Huon Peninsula and Adelbert Range, the points within the cloud for the sample from the Central Cordillera are more widely scattered, especially within the principal components scatter plot. This pattern is not surprising. Samples from the most expansive geographic subdivision of L. ernstmayri , stretching from the Owen Stanley Range in the east to Mount Karimui and the Schrader Range in the west, also exhibits the greatest morphometric variability.

We have a sample from a fourth montane area. The filled circle identified by an arrow in the principal component and canonical variate scatter plots (figs. 16, 17) is the score for a young animal in adult coat, the sole specimen collected in the Foja Mountains, which is a northern outlier of the Central Cordillera in western New Guinea, far from the sources of our larger samples to the east. The specimen score is associated with the cloud of scores for the Cordilleran samples along the first and second axes in both the principal components and canonical variate (where the specimen was entered as an unknown) ordinations. Other traits we examined are unhelpful in assessing the affinities of this individual with other montane samples. It lacks a cusp t7 and an anterolabial cingulum on the second upper molar, but so do some specimens from both the Adelbert Range and Central Cordillera (table 6). Length of head and body is less than the averages for the Adelbert and Cordilleran samples, but is within the ranges of variation observed for those series (table 3). The tail is shorter and outside the scope of recorded variation, but its length relative to length of head and body is not. All cranial and dental measurements are within the latitude of variation recorded for these measurements in the Cordilleran sample (table 5). The external contrasts may simply reflect the young age of the specimen. Its weight of 70 g, however, might be significant. That figure is outside the range, although not by much, of values recorded for samples from the Adelbert Range and Central Cordillera, and even the Huon Peninsula (table 3). If that body mass is typical of young adults, older animals may weigh more and the population on the Foja Mountains could be characterized by, among other traits, significantly larger body size than found in populations occurring elsewhere.

Different combinations of qualitative and quantitative morphological traits indicate phenetic divergence among samples from the Central Cordillera, Adelbert Range, and Huon Peninsula, each a discrete geographic entity. Examining morphological attributes in larger series from more places in all three regions may help confirm or dispel the pattern of geographic variation we have documented here. Additional material from now unsampled portions of the Central Cordillera, the most expansive of all highland areas surveyed, would be of special interest for providing insight into the significance of the morphometric variation we described among our specimens from Cordilleran localities, including the Foja Mountains, which are isolated from the Central Cordillera. The addition of molecular data would be especially helpful. That critical component would help us assess whether the distinctive discriminating phenetic traits associated with samples from the Huon Peninsula, Adelbert Range, and Central Cordillera represent the outcome of prolonged genetic and ecomorphological divergence rather than a product of genetic drift among recently subdivided montane populations of a single species. For now we regard the available samples from the three montane regions to represent populations of L. ernstmayri based on their equivalence in overall size and similarity in pelage coloration and qualitative cranial features.

To us, the single specimen from the Foja Mountains also represents another population of L. ernstmayri . In overall size, texture, and coloration of pelage, and cranial traits, it is

TABLE 9 Results of Discriminant-Function Analysis Performed on 31 Adult Leptomys ernstmayri from Four Geographic Regions (Correlations, eigenvalues, and cumulative variance are explained for two canonical roots; see table 5, fig. 17.)

closely similar to material in all our other samples that we identify as that species. We need a larger sample from the Foja region that contains a greater range in age to evaluate the genetic and zoogeographic status of what appears to be an extremely remote montane isolate in western New Guinea. We note, however, that the Foja Mountains are emerging as an area of striking montane endemism within New Guinea, especially in birds ( Diamond, 1982, 1985; Beehler et al., 2007).

HABITAT AND BIOLOGY: The specimens of L. ernstmayri we can tie to reliable descriptions of collection sites are all from tropical lower montane and ‘‘midmontane’’ rain forest formations. The AMNH series from Gang Creek (1340 and 1370 m) in the Rawlinson Range of the Huon Peninsula is the largest of this species so far collected. They were obtained by members of the Seventh Archbold Expedition to New Guinea in 1964. Hobart M. Van Deusen (partial manuscript in AMNH archival files) wrote that the credit for the fine material was entirely due to Adion, a young native boy from the village of Zengaren, and the patience of Grierson [a member of the Expedition from the United States] who encouraged him to trap for us. A simple deadfall trap was constructed; the top was loaded with soil, and several large and active beetles, tied together with bush string, were attached to the complicated trigger. This was the only trap and trigger of this kind that we saw in use on the Huon Peninsula. The only other specimen obtained on the trip came to one of my hanging meat sets that had concealed steel traps buried beneath the bait. This method also took Leptomys on previous expeditions.

Except for a close view (fig. 6 in Van Deusen, 1978: 11) of the forest at the Gang Creek camp, which shows ‘‘dense, tangled vegetation with heavily mossed tree trunks’’, clearly indicating tropical lower montane forest (as does the altitude), no other information, either in field journals or publications, can be tied to the Gang Creek sample.

The specimen from Purosa (1970 m) and that from nearby Kamila (1900 m), collected by members of the Sixth Archbold Expedition to New Guinea in 1959, were taken in primary tropical lower montane rain forest. Brass (1964: 199–200) described this as ‘‘mid-mountain forest’’ (p. 208) and chronicled the rich diversity of canopy trees near camp, the great variety of subcanopy species, and the lush floristic wealth of the undergrowth, especially in ravines and shallow gullies. Purosa is at the head of a grassy valley in deep forest covering a divide over which about 2 km away was the Kamila site, also in deep forest and at the head of a smaller grassy valley. This is a common vegetational pattern in the New Guinea highlands: remnant primary forest on high divides and ridges giving way at lower altitudes to anthropogenic grasslands ( Paijmans, 1976; Grubb and Stevens, 1985).

Kassam (1350 m), another camp made by the 1959 expedition but far to the northeast of Purosa and Kamila, also yielded a single example of L. ernstmayri . The surrounding vegetation consisted of primary tropical lower montane forest (with oak, Castanopsis , Engelhardia , and a variety of laurels dominating) interdigitating with second-growth formations and anthropogenic grasslands ( Brass, 1964: 203–204).

The specimen from the Foja Mountains was trapped by Helgen along the forested bank of a rushing stream at about 1200 m (fig. 18). The trap had been set on the ground 6–10 ft above the streambed. We asked Wayne Takeuchi of Harvard University, the botanist on the expedition to the Foja Mountains, to characterize the habitat at the exact site where Helgen took the rat. ‘‘The point you speak of’’, wrote Takeuchi,

is where the stream starts to level off a bit and you get real alluvial debris (stream abraded rounded rocks instead of the angular colluvium higher up the bed). The forest there looked to me like a mid-stage succession; Ficus , Maesa , Cypholophus , stuff I would not normally associate with primary growth. However, the higher slopes bordering the bed are mainly tall growth Myrtaceae (Syzygium) , Lauraceae ( Cryptocarya , Litsea ), and would count as mature growth. Like most places with unstable slopes, it’s a community mosaic, but the streambed itself is successional, and transitional to older, more shade-adapt- ed communities.

The terrestrial habitus observed by Van Deusen was also indicated by Rudd and Stevens (1994: 45) who trapped one individual ‘‘at ground surface’’ at 1980 m on the southern flank of Mount Missim (then identified as ‘‘ L. elegans ’’, but corresponding to BBM-NG 160978, a specimen of L. ernstmayri ); this specimen was trapped ‘‘near the transition zone from mid-montane rain-forest to upper montane rain-forest’’ ( Willett et al., 1989). Two specimens collected at Kassam Pass (1400 m) in the Kratke Range (BBM-NG 54966 and 54995) bear label notations indicating they were taken in a ‘‘live trap on the ground in forest’’ and a ‘‘snap trap on the ground in forest’’, respectively. The series at BBM from the Adelbert Range (see gazetteer) and another from Mt. Karimui (BBM-NG 105313) were likewise snap-trapped, all in ‘‘forest’’, according to their labels. Another specimen (AM M14682) was taken on Mt. Karimui at 1750 m in a trap baited with sweet potato, placed below a fallen tree trunk ‘‘in lower montane forest with open understory interspersed with dense stands of climbing bamboo’’ (K. P. Aplin, personal commun.). A small series of specimens from the Schrader Range (at UPNG) were dug out of an underground nest by Hagahai women (Yarao, in litt., 2005).

As with all the other species of Leptomys , a carnivorous diet has been attributed to L. ernstmayri , which is certainly corroborated by Van Deusen’s observations at Gang Creek. Yarao (in litt., 2005) found that the stomach contents of animals trapped at Hagahai in the Schrader Range included only arthropods (including termites, spiders, cockroaches, and chalcid beetles) and fungal hyphae, with cockroaches and beetles predominating.

Reproductive information is meager. One of the two specimens from Lake Trist is a pregnant female that bore a single fetus (10 mm in crown-rump length) in the left horn of the uterus. Another specimen from the Schrader Range was a pregnant female carrying a single fetus (crown-rump length 5 34 mm).