A distinctive new species of flowerpecker ( Passeriformes : Dicaeidae ) from Borneo

The enigmatic ‘Spectacled Flowerpecker’—a probable new bird species from the island of Borneo—was first sighted in the Danum Valley of Sabah, Malaysia in 2009. However, the absence of a holotype specimen has prevented its formal scientific description. Since then only a handful of reports from widely disparate localities across the island have emerged, all from lowland sites and often in close association with fruiting mistletoe. Here, we report the long-awaited capture of a specimen of this putative new species and confirm its morphological and molecular distinctiveness as a novel species in the genus Dicaeum.


INTRODUCTION
The vast and ancient tropical forests of Borneo are a globally important center for biodiversity (De Bruyn et al. 2014). Situated within the Greater Sunda Archipelago in Southeast Asia, the island serves as an area of endemism for many taxonomic groups, with new taxa of plants and animals still routinely being discovered. One such group is the flowerpeckers (Dicaeidae)-a family of small, frugivorous passerine birds-whose range extends from the northernmost regions of Indomalaya to Australasia. Borneo sits within the core of the family's distribution, with 13 of the 48 currently known and accepted species occurring on the island, surpassed only by nearby Sumatra (16 species) and the Philippines (15 species) (Gill & Donsker 2019). With the exception of one montane endemic (Dicaeum monticolum), Borneo's flowerpeckers favor lowland elevations, inhabiting a wide variety of forest and edge habitats throughout the island where they are often closely associated with hemiparasitic aerial mistletoe (Cheke et al. 2019).
In 2009 DPE reported on the discovery of a potential new species of flowerpecker from the Danum Valley in the lowlands (180m asl) of Sabah in northeastern Malaysian Borneo ( Fig. 1) (Edwards et al. 2009). The bird appeared among several other species of flowerpecker foraging on a patch of fruiting mistletoe (likely Macrosolen cochinchinensis (Lour.) Tiegh. (Loranthaceae), as inferred from photos) in a tree along the canopy walkway of the Borneo Rainforest Lodge. Their report and photographs documented a bird species with the clear proportions and bill shape of a flowerpecker, uniform gray plumage save for a white stripe from throat to vent, and conspicuous white eye-arcs giving the bird a uniquely "spectacled" appearance. The authors felt confident that these characters distinguished it from any other species currently known from Borneo. Attempts to locate additional individuals were unsuccessful, and the species eluded observation for years thereafter. However, the prospect of an entirely new species of bird, likely the first for Borneo in over 100 years, led to considerable excitement in the scientific and birdwatching communities, and garned the attention of global media outlets (Braun 2010, Doyle 2010, Walton 2010. Unfortunately, the lack of a holotype specimen prevented the species from being scientifically described and formally named (Collar 2009, Peterson 2014.  Gaveau et al. (2016), and SRTM elevational dataset obtained from Jarvis et al. (2008). Non-forest shaded in white, the majority of which represents deforested areas converted to agriculture.
Following another gap in observations, at least one individual was spotted and photographed at the Belalong Canopy Walkway in the Temburong District of Brunei in May 2017, its appearance coinciding with a mass fruiting of nearby Loranthaceous mistletoe plants ( Fig. 1) (Ong 2018). From this handful of observations, the 'Spectacled Flowerpecker' was beginning to reveal itself as an elusive but widely distributed bird species that had somehow evaded scientific description. Its frequent association with hemiparasitic mistletoe has led some to speculate that the bird is a secretive canopy dweller with a highly specialized foraging ecology, potentially nomadic tendencies, and low population densities; but the real reasons for the bird's confounding scarcity remains a mystery (Edwards et al. 2009, Phillips & Phillips 2014, Sykes & Loseby 2015).
In early 2019 (24 March-7 April), an ornithological field expedition led by members of the Smithsonian Institution (CMM and JRS) and the Sarawak Forestry Corporation (RR) surveyed the Lanjak Entimau Wildlife Sanctuary (LEWS) of Sarawak as part of a series of field surveys to characterize and document avian biodiversity in the lowland forests of Sarawak. On 31 March, a female individual of this putative new species was captured in a mistnet and preserved as a scientific specimen. The bird was immediately discernible from other flowerpeckers in the study area and expressed all of the characteristic features described by previous observers and as depicted in field guide illustrations (Phillips & Phillips 2014, Myers 2016. In addition to a traditional museum skin specimen, various anatomical, diet, and genetic samples were saved for preservation and analysis.
Here, we formally confirm and describe this bird as a distinctive new species of flowerpecker, for which we propose the following name:

REMARKS
Phylogenetic relationships.-To infer phylogenetic placement of D. dayakorum within the Dicaeidae we sequenced DNA from the holotype to place it in the framework of a previous phylogenetic study of the family by Nyári et al. (2009). We downloaded sequence data from that study posted to GenBank, which included the nicotinamide adenine dinucleotide dehydrogenase subunit 2 (ND2, 1034 bp), ND subunit 3 (ND3, 351 bp), and the transforming growth factor beta 2 (TGFb2, 542 bp) for 30 species of flowerpecker and 11 outgroup species (10 sunbirds and 1 leafbird; Appendix Table 2). Genomic DNA was isolated from DMSO preserved tissues of the holotype specimen (USNM 663246). The DNA extraction was performed at the Smithsonian Laboratories of Analytical Biology (LAB) using a fully automated phenol-chloroform genomic extraction machine (Gene Prep, AutoGen, Holliston, MA). PCR products were amplified for all three coding regions (ND2, ND3, and TGFb2) using GoTaq® G2 Hot Start Master Mix (Promega Corporation) with additional BSA (New England BioLabs, 50 mM stock solution) and magnesium chloride (Bioline, Biolase DNA Polymerase, London, UK). The annealing temperature for ND2 was 57°C, and 59°C for ND3 and TGFb2. PCR products were cleaned and visualized on a 1% agarose gel. Cycle-sequencing was performed using ABI Prism® BigDyeTM Terminator (v3.1 chemistry, ThermoFisher Scientific, Waltham, MA) for both forward and reverse directions. The products were purified using SephadexTM (GE Healthcare Bio-Sciences Corp., Pittsburgh, PA) spin columns before submitting for sequencing on an ABI 3730 xl DNA analyzer at the LAB facilities. Raw sequences were visualized and quality-trimmed using Sequencher (version 5.4.6; Gene Codes Corporation, Ann Arbor, MI). Sequences of the holotype were aligned with the rest of the dataset (42 taxa in total) using the MAFFT plugin (version 7.388; Katoh & Standley 2013) implemented in Geneious Prime 2019.1.3 (https://www.geneious.com, accessed on 28 June 2019). The multiple sequence alignments for each of the three loci were visually inspected and manually curated when necessary. As ND2 and ND3 represent portions of the same mitochondrial gene, both matrices were concatenated into a dataset of 1,383 bp. The individual datasets for both ND and TGFb2 genes were trimmed to remove characters not suited for phylogenetic inference using BMGE (Criscuolo & Gribaldo 2010) with a sliding window size of three, a maximum entropy threshold and gap rate cut-off of 0.5, and a minimum block size of five. We estimated nucleotide substitution models for individual genes using ModelFinder (Kalyaanamoorthy et al. 2017) implemented in IQ-TREE (version 1.6.11 Nguyen et al., 2014). We then concatenated the multiple sequence alignments for both genes, resulting in a matrix of 1,909 bp, and performed a Maximum Likelihood analysis with partitions using IQ-TREE, allowing the program to find the best-fit partitioning scheme possible to merge partitions (Chernomor et al. 2016). The Maximum Likelihood analysis was replicated with 1000 ultrafast bootstrap iterations (Hoang et al. 2017) by resampling partitions and then sites within resampled partitions, as recommended by Gadagkar et al. (2005). To detect incongruences in the topology of trees built from different organelles (i.e., nuclear vs. mitochondrial sequences), we performed ML analyses using only the concatenated ND2 and ND3 loci representing mitochondrial DNA, and the TGFb2 representing nuclear DNA. We estimated models of nucleotide substitutions for each of these loci and conducted a ML analysis using IQ-TREE with 1000 bootstrap replications as described above. Bayesian Inference (BI) analyses were conducted in MrBayes (version 3.2.6; Ronquist et al. 2012) using the CIPRES Science Gateway web server (Miller et al. 2010). The BI was conducted using two separate four Markov Chain Monte Carlo (MCMC) analyses with 10,000,000 generations and sampling every 1000 generations. The first 25% of sample trees ("burn-in period") were removed and a majority rule (> 50%) consensus tree was built. We estimated pairwise sequence divergences of raw mtDNA from the ND2-ND3 concatenated matrix using the Kimura 2-parameter distance model (Kimura, 1980) with a Gamma distribution (Gamma parameter: 5) in MEGAX (version 10.1.0 Kumar et al. 2018). We also conducted a Poisson tree processes (PTP) species delimitation test with a subset of the taxa including all species of Dicaeum and Prionochilus with Chloropsis as outgroup, using a MCMC analysis with 100,000 generations with a burn-in period of 50% and a thinning of 100 (Zhang et al. 2013).
Topologies of the ML and BI phylogenetic trees largely mirror those found in Nyári et al. (2009), and place D. dayakorum at a basal position of the "core" Dicaeum clade (Clade B sensu from Nyári et al. 2009) with high support (99% bootstrap support and 1.00 posterior probability in the in the ML and BI analyses, respectively; Figure 3). The topology from the concatenated ML tree is largely congruent with the trees built using the two individual genes. However, as each of these individual datasets have fewer phylogenetically informative sites, the resultant trees show several unresolved nodes with low support values, especially at deep branches. The only notable discrepancy is the weak placement of D. dayakorum as sister to D. australe (74% bootstrap support) in the tree constructed from the TGFb2 gene.
Morphology and voice.-We took standard morphological measurements from the holotype, along with several of the most closely related species (as per our molecular analyses), and similar/co-occurring species of Dicaeum for comparison (Appendix Table 1).
Classic taxonomic treatments of the Dicaeidae focused on morphological characters such as bill shape, tongue morphology, and relative length of the outermost primary. D. dayakorum displays a relatively small and slender bill similar to that of D. monticolum and D. cruentatum, and distal bill length is at the low end of the range for the genus. The culmen is moderately arched downward and slight serrations are present along the edge of the distal half of both maxilla and mandible. The base is relatively wide compared to the tip and rictal bristles are reduced.
As opposed to the triangular, non-tubular tongues shown by members of Prionochilus, 1841, many members of Dicaeum have specialized tongues with tubular structures and deeply bifid tips (Morioka 1992). Tongue morphologies in this genus are often distinctive, reflecting varying degrees of specialization between frugivory and nectarivory (Mayr & Amadon 1947, Morioka 1992. The tongue and hyoid of the holotype were preserved in aqueous formaldehyde and later transferred to ethanol. Examination revealed an overall structure that is not atypical for Dicaeum. The tongue has a triangular, fleshy base extending approximately halfway up the length of the structure giving way abruptly to a narrow, straight, and deeply concave anterior portion. This latter half is horny and semitubular at the base, becoming increasingly translucent distally, and ending in a strongly bifid tip (Fig. 4).
One of the most reliable characters differentiating flowerpecker genera has been the length of the 10 th or outermost primary, being relatively long in Prionochilus compared to Dicaeum (Mayr & Amadon 1947, Salomonsen 1960a. Evaluation of this character in our holotype specimen revealed it to be greatly reduced (5.5 mm), fitting with the vestigial outer primaries possessed by other members of genus Dicaeum.
DPE heard the putative male sing with a series of ~12 high-pitched see notes, rising and then falling in pitch (Edwards et al. 2009), but no recoding was obtained. Boyd et al. (2016) subsequently obtained the first and only known voice recordings of this species during their sightings of the birds in Central Kalimantan in 2015 (xeno-canto.org recordings XC239907, XC301233). The recordings reveal a simple call typical of other flowerpecker species and an accelerating trilled song of 18 notes lasting 1.5 seconds, starting with two notes at 4.8 kHz, then immediately accelerating to a trill that rises from 5.3 kHz to a maximum centered at 6.1 kHz (5.8-6.4 kHz), and falls for the last three notes to 5.7 kHz. The song thus bears similarities, but is distinctive, to other known flowerpecker songs. Taxonomy.-Clear phenotypic distinctiveness from known species of flowerpeckers, and high confidence values from phylogenetic and species delimitation analyses of our molecular dataset, strongly support Dicaeum dayakorum as a new species. Nyári et al. (2009) recognized two predominant molecular clades in the Dicaeidae: generally thicker-billed species with a longer outermost primary (Clade A); and generally more slender-billed species with shorter outer primaries (Clade B). These clades are roughly concordant with the two established genera of flowerpeckers, Prionochilus and Dicaeum, respectively, with the exception of four traditionally Dicaeum species that tend to group with Clade A (Nyári et al. 2009). Our independent analyses of the same molecular dataset (with the addition of USNM 663246) mirrors this arrangement, with strong support for D. dayakorum as sister to the core Dicaeum (Clade B; Fig. 3). With the added consideration of morphological characters (e.g. outer primary length and tongue morphology) that are closely (though not exclusively) associated with the core Dicaeum lineage, we strongly support the view of Edwards et al. (2009) that the new species be placed in this genus.
We recovered no clear sister taxon to D. dayakorum from our phylogenetic analyses (Fig. 3). Although D. dayakorum seems to be sister to the rest of Clade B with high support (99% ML; 1.0 BI), branches separating this and related taxa are short and support values are low (Fig. 3), indicating that more molecular data may be needed to define this relationship satisfactorily. Results from the pairwise distance analysis suggest that genetic distances between D. dayakorum and other basal members of Clade B are high, ranging from 14-16% (D. anthonyi: 16.1%; D. auerolimbatum: 16.8%; D. australe: 16.8%; D. bicolor: 15.6%; and D. trigonostigma: 14.5%). Species-level distinctiveness was further supported by the species delimitation test with a posterior probability support value of 0.99. In fact, the PTP analysis suggest that D. dayakorum is more distinctive at the molecular level than the rest of the 30 other flowerpecker taxa included in the Nyári et al. (2009) study. Although further taxon sampling is warranted, no unsampled species is a realistic candidate for conspecific status, and thus the proposed status of D. dayakorum as a new species will not be affected.
Ecology and distribution.-The type locality for Dicaeum dayakorum (USNM 663246) was characterized by tall, mixed dipterocarp forest upslope from the Nanga Segerak field station (LEWS, 350m asl). This section of forest lies at the edge of an expansive sanctuary dominated by primary lowland forest that has previously been subjected to light selective logging by local communities (Chai 1996). The immediate area of capture was a relatively open and steep section of ridgeline, possibly the site of a previous treefall or landslide in the advanced stages of succession. The location seemed to be a crossing point for birds cresting the ridge and several bird species typical of canopy or upperstory flocks were observed here near ground level (e.g Irena puella, Psilopogon chrysopogon, Coracina fimbriata). Documented observations depict a species with affinities for mature forest at lower elevations (30-350 m) (Edwards et al. 2009, Sykes & Loseby 2015. The elevation of the sightings in the Maliau Basin are unclear (likely between 260-1100 m), but could potentially represent higher elevation limits for the species A DISTINCTIVE NEW SPECIES OF FLOWERPECKER Zootaxa 4686 (4) © 2019 Magnolia Press · 459 (Sykes & Loseby 2015). The Spectacled Flowerpecker has not yet been observed in areas with significant anthropogenic disturbance and we suspect it may be very rare or absent in heavily selectively logged forest and will abandon largely deforested areas. FIGURE 5. Sketch of a Loranthaceous mistletoe seed recovered from the lower gut of the holotype (USNM 663246). The sides of the seed show two longitudinal ribs which end in horn-like projections while the base depicts the uncoiled viscin aril which helps the seed attach to tree branches.
Flowerpeckers are known to feed on a wide variety of fruits, flowers, and invertebrates (Cheke et al. 2019). However, specialization on mistletoe fruit is one of the characteristic traits of the family, and is likely reciprocal due to their role as canopy seed dispersers (Ali 1931, Kannan 1966, Cheke et al. 2019. Most, if not all, flowerpecker species possess gut adaptations thought to be specifically evolved for processing mistletoe, in which the rapidly digestible berries are able to bypass the muscular ventriculus so that the seeds can be expelled intact (Desselberger 1931, Cheke et al. 2019. Many of the reported observations of D. dayakorum show it foraging in or closely associated with various species of mistletoe, often engaging in pulp-predation behavior . A review of available photographs from these observations revealed several mistletoe species from at least two families (Loranthaceae: Macrosolen cochinchinensis; Santalaceae: Ginalloa arnottiana Korth. and Viscum ovalifolium DC). Examination of the holotype's gut contents found a partially digested jumping spider (Salticidae sp.) in the gizzard as well as mistletoe seeds in the large intestine (Fig. 5). These seeds were identified as a species of Loranthaceae upon morphological examination. However, a BLAST search of its chloroplast DNA was inconclusive. Further investigation is needed to determine which of the over 16 genera of Loranthaceae reported for Borneo can be a suitable match.
The degree to which D. dayakorum is dependent on mistletoe is an important question that remains to be evaluated. The partially tubular tongue structure (Fig. 3), the presence of arthropod material in the stomach, and observations of the species foraging on Medinella (Melastomataceae) berries suggest a diet similar to that of other flowerpeckers in which a mistletoe-dominated diet is supplemented to some extent by nectarivory, general frugivory, and arthropod predation. The role of diet specialization on habitat preference and distribution is poorly known (Reif et al. 2016). The various species of mistletoe that the bird has been observed to consume are ubiquitous species of Southeast Asia that likely occur widely throughout Borneo where mature forests persist (Beaman et al. 2001(Beaman et al. , 2004. However, the patchy distributions of mature mistletoe plants (Aukema 2004), and seasonality of fruit (Barea & Watson 2007) suggest that the distribution of this resource is likely sparse and ephemeral. These factors, along with competition for fruit resources from other mistletoe specialists-especially other species of flowerpeckers as noted in the initial report by Edwards et al. (2009)-may have a limiting effect on population density. All other aspects of the bird's life-history, including breeding biology, phenology, and vagility, remain undocumented.
Conservation.-Extrapolation of range based on documented localities of occurrence (Fig. 1) and the bird's currently known foraging ecology suggest that D. dayakorum has a widespread distribution, with the potential to occur anywhere in the extensive lowlands of Borneo where suitable habitat exists. The pattern of occurrences suggests a bird that is patchily distributed, possibly even nomadic while following shifts in an ephemeral food source. It is also possible that the core distribution of the species has yet to be discovered. However, the considerable scarcity of these observations, despite documented occurrence in relatively well-explored areas (e.g. Danum Valley) is more difficult to explain and may be compounded by low population densities and difficulty of detection owing to the bird's small size, dull coloration, and canopy dwelling tendencies.
Given our current lack of information on processes governing the distribution of this species, the potential effects of habitat fragmentation and disturbance should not be underestimated. However, our discovery of the species in Lanjak Entimau Wildlife Sanctuary-an approximately 187,000 ha. tract of protected lowland/hill forest managed by the Sarawak Forestry Corporation-is a reason to be hopeful (Guntavid et al. 1997). We see no reason why D. dayakorum would not occur throughout the sanctuary's pristine forests, or the contiguous Batang Ai National Park to the south. Although further surveys of the area are needed to assess habitat and mistletoe prevalence. Lowland forests such as this are the heart of Borneo's ecosystems. Sadly, these sanctuaries are under increasing threat with the encroachment of intensive selective logging and unsustainable agricultural practices, including widespread conversion of lowland forest to oil palm (Wilcove et al. 2013). It should also be noted that the indigenous Iban Dayak inhabitants of the region are the primary facilitators of conservation upkeep and protection. The imperilment and continued diminishment of these traditional longhouse communities is an underappreciated threat to the protection of these vital areas.
Many questions remain regarding the distribution, population dynamics, and ecology of the Spectacled Flowerpecker. Though the species may have once enjoyed a historical distribution of relative continuity throughout the lowland forests of Borneo, its current distribution has almost certainly become increasingly fragmented and diminished. We are hopeful that the formal scientific description of this exciting new species will help to underscore the importance of Borneo's lowland forests as an area of ecological significance and future discovery.