Volume 47, Issue 3 p. 420-429
Original Article
Open Access

Hitchhiking into the future on a fly: Toward a better understanding of phoresy and avian louse evolution (Phthiraptera) by screening bird carcasses for phoretic lice on hippoboscid flies (Diptera)

Leshon Lee,

Leshon Lee

Department of Biological Sciences, National University of Singapore, Singapore

Search for more papers by this author
David J. X. Tan,

David J. X. Tan

Department of Biology and Museum of Southwestern Biology, University of New Mexico, Albuquerque, New Mexico, USA

Search for more papers by this author
Jozef Oboňa,

Jozef Oboňa

Department of Ecology, University of Prešov, Prešov, Slovakia

Search for more papers by this author
Daniel R. Gustafsson,

Daniel R. Gustafsson

Guangdong Key Laboratory of Animal Conservation and Resource Utilization, Guangdong Public Laboratory of Wild Animal Conservation and Utilization, Institute of Zoology, Guangdong Academy of Sciences, Guangzhou, China

Search for more papers by this author
Yuchen Ang,

Yuchen Ang

Lee Kong Chian Natural History Museum, National University of Singapore, Singapore

Search for more papers by this author
Rudolf Meier,

Corresponding Author

Rudolf Meier

Department of Biological Sciences, National University of Singapore, Singapore

Museum für Naturkunde, Leibniz-Institut für Evolutions- und Biodiversitätsforschung, Berlin, Germany

Correspondence

Rudolf Meier, Museum für Naturkunde Leibniz-Institut für Evolutions- und Biodiversitätsforschung, Berlin, Germany.

Email: rudolf.meier@mfn.berlin

Search for more papers by this author
First published: 18 February 2022

Funding information: Pearl River Talent Recruitment Program of Guangdong Province, Grant/Award Number: 2019QN01N968; Ministry of Education, Grant/Award Number: R-154-000-A22-112

Abstract

Many phoretic relationships between arthropods are understudied because of taxonomic impediments. We here illustrate for avian lice riding on hippoboscid flies how new natural history data on phoretic relationships can be acquired quickly with modern and cost-effective barcoding techniques. Most avian lice are host-specific, but some can arrive on new hosts by hitchhiking on hippoboscid flies that feed on bird blood. Our summary of the literature yielded 254 published records which we here show to belong to two large and 13 small interaction networks for birds, flies, and lice. In order to generate new records, we developed a protocol based on screening bird carcasses sourced with the help of citizen scientists. The inspection of 131 carcasses from Singapore led to the first record of a Guimaraesiella Eichler (Ischnocera: Philopteridae) louse species riding on Ornithoica momiyamai Kishida flies collected from a pitta carcass. Phoresy may explain why this louse species is now known from three phylogenetically disparate hosts (Pitta moluccensis (Müller), Ficedula zanthopygia (Hay); Pardaliparus elegans Lesson). A second new case of phoresy enhanced a large interaction network dominated by Ornithophila metallica (Schiner), a cosmopolitan and polyphagous hippoboscid fly species. Overall, we argue that many two- and three-way phoretic relationships between arthropods (e.g., mites, pseudoscorpions, beetles, flies) can be resolved with cost-effective large-scale NGS barcoding, which can be used to partially overcome taxonomic impediments by pre-sorting specimens for taxonomic revision.

INTRODUCTION

Phoretic relationships between phylogenetically disparate species are common in insects. Small arthropod phoronts regularly use larger arthropod and vertebrate species for transportation. Phoresy can be an important precursor for parasitism/symbioses and often involve species that deliver important ecosystem services (e.g., decomposition of dung and carrion) (Bartlow & Agosta, 2021). Yet, comparatively little is known because of taxonomic impediments. These are particularly serious for understanding phoresy because resolving phoretic relationships (1) require taxonomic expertise for at least two groups (host and phoront), (2) the phoronts tends to be small and thus have a high chance of belonging to a taxonomically poorly known clade (e.g., mites, lice), and (3) the phoronts are often incidentally discovered as part of more detailed investigations into host biology. This means that most casual observations of phoretic arthropods are either never published or hidden within the host-specific literature.

One particularly fascinating, three-way phoretic relationship involves some species of avian lice (mostly Philopteridae) that use blood-sucking hippoboscid ‘louse’ flies (Diptera: Hippoboscidae) to travel between avian hosts. All avian lice are flightless obligate ectoparasites that live on birds throughout their life (Johnson et al., 2003). Louse transmission between hosts usually requires direct physical contact such as parent-offspring interaction (e.g., Clayton & Tompkins, 1994; de L Brooke, 2010). However, some lice can arrive on new hosts via phoretic associations with flying insects (Keirans, 1975a) or brood parasitism (Hahn et al., 2000, but see Balakrishnan & Sorenson, 2007). One of the most important facilitators of indirect avian louse transmission is hippoboscid flies, which are highly mobile hematophages that can transport avian lice attached to the flies' legs or abdomens (Figure 3a–c). Hippoboscidae is a moderately-sized family of ca. 200 described species of which >80% are bird parasites belonging to two clades (Dick, 2006; Petersen et al., 2007). Because many hippoboscids feed on a wide variety of bird hosts and have very extensive geographic distributions (Bequaert, 1953), they have the potential to transfer lice between host species from different bird orders. The fact that many birds are migratory further increases the chance that phoretic lice jump continents and distantly related bird species. In comparison, there are only a few records of phoretic interactions between avian lice and other insects such as butterflies and bees (Keirans, 1975a), so these interactions are less likely to be important for host switching.

Not all avian louse species are involved in phoresy. Phoresy is mostly found among chewing lice (Ischnocera) feeding on feathers and skin detritus. In contrast, there is only one known case of phoresy involving a blood-feeding avian louse species [Amblycera: phoresy of Hohorstiella gigantea (Denny) on an unidentified hippoboscid fly (Hopkins, 1946)]. Even within Ischnocera, phoresy appears to be concentrated in certain ecotypes. For example, ischnocerans specializing in wing feathers of pigeons are better at phoretic attachment than those feeding on body feathers because the latter are more likely to fall off when attempting to ride on a moving fly (Bartlow et al., 2016; Harbison et al., 2008, 2009). This may explain higher levels of genetic structure among pigeon body lice (non-phoretic) as opposed to wing lice (DiBlasi et al., 2018; Johnson, Williams, et al., 2002). Additionally, co-phylogenetic analyses revealed that wing lice on pigeons have higher levels of host-switching compared to pigeon body lice, which generally coevolve with their hosts (Clayton & Johnson, 2003). On pigeons, body lice were also found to have higher genetic codivergence than wing lice, possibly due to the availability of (or rather lack of) hippoboscid flies for phoretic dispersal (Sweet & Johnson, 2018).

However, despite its potential importance, phoresy remains poorly documented in the wild, so that its importance is hard to assess although recent experimental work has yielded many important insights (Bartlow et al., 2016; Harbison et al., 2008, 2009). We here use species interaction networks to summarize the literature on louse phoresy. This revealed three major issues. Firstly, the number of observations that provide species-level taxonomic resolution for all species (i.e., bird, fly, and louse) is very limited because this requires extensive taxonomic knowledge for three very different animal groups. Secondly, the number of new reports has been in steep decline with the majority of published records concentrated in the first half of the 20th century (see time-lapse video: https://youtu.be/IiiLCZdZbYI; Keirans, 1975b; Bartlow et al., 2016). This is likely correlated with the overall decline in the number of natural history publications (Tewksbury et al., 2014). Lastly, a large proportion of the known phoresy records come from temperate Europe and North America despite the relatively species-poor Holarctic avian fauna (Figure 1).

Details are in the caption following the image
Map of bird species richness (blue shading across countries) and phoresy records (yellow bars) by country illustrating the large number of Holarctic phoresy observations. Numbers indicate approximate bird species richness of each biogeographical realm and the number of phoresy records in parentheses. Bird species richness map: Clements et al. (2019), Species richness figures: Newton and Dale (2001), Basemap from thematicmapping.org, biogeographical realm shapefile from UNEP-WCMC (2011), generated by QGIS v2.18 (Las Palmas)

We here address these problems by developing a protocol for obtaining new phoresy records. Although it is developed for avian lice, the same molecular techniques are also applicable for resolving other phoretic relationships. In our study of avian lice, we screened bird carcasses reported by members of the public in Singapore. This yielded 131 dead birds (54 species) from which lice, hippoboscid flies, and mites were collected. For the lice and flies, we use NGS barcoding (Srivathsan et al., 2021; Srivathsan, Hartop, et al., 2019; Srivathsan, Nagarajan, & Meier, 2019; Wong et al., 2017; Yeo et al., 2018) to address the taxonomic impediments that interfere with so much natural history research. The specimens were sorted into putative species based on NGS barcodes which can now be obtained at low cost for thousands of specimens within days (Srivathsan et al., 2018, 2021; Srivathsan, Hartop, et al., 2019; Srivathsan, Nagarajan, & Meier, 2019; Wang et al., 2018; Yeo et al., 2018). This yielded two new cases of phoresy which are here added to the species interaction networks identified based on literature data.

MATERIALS AND METHODS

Literature review

Using the records in Bartlow et al. (2016) as starting point, we checked the literature on louse-hippoboscid phoresy for overlooked or misattributed records. In addition, we conducted a literature search using the keywords ‘Hippoboscid*’, ‘Mallophaga*’, ‘Phthiraptera*’, ‘Louse’, ‘Lice’, and ‘Phore*’ in Web of Science, Biodiversity Heritage Library, Phthiraptera.info database, and Google Scholar.

Collection of avian ectoparasites

We collected bird carcasses reported by citizen scientists in Singapore as part of a long-term project to monitor avian mortality due to window-collisions or road accidents (Low et al., 2017; Tan et al., 2017). Carcasses collected between 2013 and 2019 were identified to species based on the Clements et al. (2019) checklist, bagged separately, and stored at −20°C. We brushed 131 bird specimens (54 species) for lice and hippoboscid flies using a toothbrush and preserved all ectoparasites in 95% ethanol at −20°C. We then identified hippoboscid flies carrying lice using a Nikon SMZ460 stereomicroscope (Maa, 1966a, 1966b, 1969a, 1969c, 1969d).

Ectoparasite DNA barcoding

For all the lice and hippoboscids collected, we determined the number of species-level units using NGS barcodes (Baloğlu et al., 2018; Meier et al., 2016; Wang et al., 2018; Yeo et al., 2020). We extracted genomic DNA from hippoboscid flies and lice using a modified hotSHOT protocol (Truett et al., 2000). For lice, we used 10 μl of alkaline lysis buffer and neutralizing reagent per specimen, while the quantities were increased to 15 μl for hippoboscids. A 313-bp Cytochrome Oxidase 1(COI) minibarcode was amplified using modified primers published in Geller et al. (2013) and Leray et al. (2013) [m1COlintF: 5′-GGWACWGGWTGAACWGTWTAYCCYCC-3′ (Leray et al., 2013) and modified jgHCO2198: 5′-TANACYTCNGGRTGNCCRAARAAYCA-3′ (Geller et al., 2013)]. In addition, we amplified another 379-bp COI fragment for one specimen for all putative louse species to ensure overlap with the barcoding region used in previous louse barcoding studies (L6625F: 5′-CCGGATCCTTYTGRTTYTTYGGNCAYCC-3′ and H7005R: 5′-CCGGATCCACNACRTARTANGTRTCRTG-3′ (Hafner et al., 1994)). For all amplicons, the forward and reverse primers were tagged with a 9-bp oligonucleotide tag at the 5′-end. Unique tag combinations could then be used to distinguish the amplicons for each specimen (Meier et al., 2016; Wang et al., 2018). Successful amplification was checked for a subsample of all PCR reactions on a 1.5% agarose gel. Subsequently, all PCR products were pooled and purified using Sera-mag SpeedBeads (Fisher Scientific) as per Rohland and Reich (2012). The pooled and cleaned amplicons used in our report on phoresy were sequenced at the Genome Institute of Singapore on a partial Illumina Hiseq2500 Rapid Run lane (251-bp paired-end).

We used default parameters in PEAR 0.9.6 (Zhang et al., 2014) to merge paired-end reads and demultiplex the reads via a Python script utilizing the unique F and R primer tag combinations for each specimen (Meier et al., 2016; Wang et al., 2018). Read counts and variants were processed according to the quality control pipeline described in Meier et al. (2016). The read variants with the highest and second-highest counts for each specimen were checked against GenBank using BLAST, and reads with similarity scores >97% to non-Phthiraptera taxa were removed. Next, we aligned all 313 bp COI barcode sequences using Mafft v7 (Katoh & Standley, 2013). We then clustered the barcodes using Assemble Species by Automatic Partitioning (ASAP: p-distances model) (Puillandre et al., 2021) and obtained the partition with the lowest ASAP score. We also clustered the barcodes using objective clustering at 2%, 3%, 4% and 5% distance thresholds (Meier, 2008; Meier et al., 2006; Meier et al., 2008). This allowed for identifying stable Molecular Operational Taxonomic Units (MOTUs).

Identification of specimens

We used BLAST to establish whether there are species-level matches for the louse and fly barcodes in NCBI (>97%). In addition, we used morphology to confirm species limits and to identify the specimens based on keys and checklists. As recommended by Meier (2017), taxonomic experts of avian lice and hippobosicid flies are co-authors and we here list the identification literature: Maa, 1966a, 1966b, 1969a, 1969c, 1969d; Gustafsson & Bush, 2017. We documented phoresy and the morphology of the ectoparasites by obtaining high-resolution dorsal and ventral views imaged at different focal lengths with a Dun Inc. Passport II Imaging system (Canon 7D Mk II with MPE-65 lens). Images were then focus-stacked using Zerene Stacker (Zerene Systems LLC) and prepared for publication using Adobe Photoshop CS5. All voucher specimens have been deposited in the Zoological Research Collections (ZRC) in the Lee Kong Chian Natural History Museum (LKCNHM).

RESULTS

Literature review

We found 254 literature records (1857–2021) of louse-hippoboscid phoresy with at least a genus-level identification for either lice or hippoboscids (Table S1). Three records reported in Bartlow et al. (2016) were misattributions and two records could not be verified based on primary sources (see Table S1). All louse-hippoboscid phoresy records are shown in Figure 2. Note that the species interaction network based on literature data omits records that do not have a species-level resolution for the louse, fly, or bird. The only exceptions are the newly discovered louse-bird interactions from Genbank related to the newly discovered phoresy interactions.

Details are in the caption following the image
Species interaction network of phoretic avian lice records (red) on hippoboscid flies (yellow), and associated birds (light blue = Passeriformes; dark blue = Coraciiformes; grey = remaining bird orders). Size of nodes corresponds to interaction number. Genus abbreviations for birds: C. = Corvus, G. = Garrulax, H. = Hypocryptadius, S. = Sturnus, T. = Turdus; for flies: O. = Ornithoica, S. = Stilbometopa; lice: G. = Guimaraesiella, S. = Sturnidoecus, T. = Trogoninirmus). Thickness of black arrows corresponds to the number of records for a particular interaction. Green dashed arrows refer to the here newly described interactions including louse-bird interactions from Genbank

New louse-hippoboscid phoresy records

We collected 32 hippoboscid flies from 22 of the 131 bird carcasses (13 bird species). Of the 32, three carried phoretic lice (Figure 2 and Table S1). Two hippoboscid specimens (ZRC_BDP0273056, ZRC_BDP0273057) that were collected from a carcass of a Blue-winged pitta (Pitta moluccensis (Müller); specimen CR465) carried three phoretic lice (Louse specimens ZRC_BDP0298043 and ZRC_BDP0298044 attached to hippoboscid specimen ZRC_BDP0273056. Louse specimen ZRC_BDP0298045 attached to hippoboscid specimen ZRC_BDP0273057). The third hippoboscid specimen (ZRC_BDP0273050) carried four phoretic lice (ZRC_BDP0298039, ZRC_BDP0298040, ZRC_BDP0298041, ZRC_BDP0298042) and was obtained from a Black-naped oriole (Oriolus chinensis Linnaeus; specimen CR619). We also barcoded the free-roaming lice on the bodies of the Blue-winged pitta and the Black-naped oriole to determine whether the lice belonged to the same species that were also found on the hippoboscid flies.

NGS barcoding of lice and hippoboscids

A total of 603 louse specimens (including the phoretic lice) and all 32 hippoboscid specimens were successfully barcoded (Genbank Accession numbers for specimens involved in phoresy: MT762409-MT762417). Clustering the 603 louse barcodes using Objective Clustering showed that the number of louse MOTUs was stable at 56 MOTUs for pairwise distance (p-distance) thresholds between 2% and 5%. For hippoboscids, the number of MOTUs obtained using Objective Clustering was 12 at 2%–3%, and 11 at 4%–5%. The thresholds were chosen based on the literature (Meier et al., 2006). Using ASAP (p-distances model) (Puillandre et al., 2021), the top partition clustered the louse and hippoboscid sequences into 57 (ASAP score = 4) and 12 MOTUs (ASAP score = 1.5) respectively. The phoretic lice belonged to two louse species, which clustered with other free-roaming lice from the same bird carcass. The phoretic louse species from the Blue-winged pitta is also known from a Yellow-rumped flycatcher [Ficedula zanthopygia (Hay); Table S1] carcass when compared to other data from Singapore (Lee, 2019). However, the same louse species was not detected on two additional Blue-winged pitta carcasses with lice (Lee, 2019). The second phoretic louse species reported here was found on one Black-naped oriole carcass.

Identification of flies and lice

Although none of the 313 bp COI minibarcodes for lice had species-level matches, BLASTing the 379-bp sequence of a Blue-winged pitta louse resulted in a 99.74% match with Brueelia sensu lato sp. in Genbank (now assigned to genus Guimaraesiella Eichler; Gustafsson & Bush, 2017) sampled from an Elegant tit (Pardaliparus elegans Lesson), a species endemic to the Philippines (Johnson, Adams, & Clayton, 2002; Bush et al., 2016; GenBank Accession number: AY149382). The Black-naped oriole louse had five Genbank matches >97% to a 379-bp sequence of a different louse species originally identified as Brueelia sensu lato (now Guimaraesiella), all of which were found on Australasian passerine hosts. One match (98.17%) was obtained from a louse collected from a Brown treecreeper (Climacteris picumnus Temminck & Laugier), another (97.38%) from a louse from a Green catbird [Ailuroedus crassirostris (Paykull)], and three matches (97.13%) from lice from a Crested shriketit [Falcunculus frontatus (Latham)], a Grey shrikethrush [Colluricincla harmonica (Latham)], and a Great bowerbird (Chlamydera nuchalis Jardine & Selby). With regard to the hippoboscid flies, only five 313 bp COI barcodes had species-level (>97%) matches to Pseudolynchia canariensis (Macquart). The hippoboscids on the Blue-winged pitta were thus identified using morphological keys (Maa, 1966a, 1966b, 1969a, 1969c, 1969d) as Ornithoica momiyamai Kishida (Figure 3a,b) and the one on the Black-naped oriole as Ornithophila metallica (Schiner) (Figure 3c).

Details are in the caption following the image
(a and b): Ornithoica momiyamai (ZRC_BDP0273056) with Guimaraesiella specimens attached to the abdomen. (c): Ornithophila metallica (ZRC_BDP0273050) with four Guimaraesiella specimens attached to its abdomen. (d): Habitus of Guimaraesiella ZRC_BDP0298043. (e): Habitus of Guimaraesiella ZRC_BDP0298039

New cases of Phoresy

Based on morphology, we identified the phoretic lice found on the flies (Ornithoica momiyamai) obtained from the Blue-winged pitta carcass as belonging to the ‘core’ group of the genus Guimaraesiella ZRC_BDP0298043 (sensu Gustafsson, Malysheva, et al., 2019). This genus was not previously known to infest pittas (Figure 3d; Somadder & Tandan, 1977; Gustafsson & Bush, 2017), and is now known to infest Blue-winged pittas, Yellow-rumped flycatchers, and the Philippine endemic Elegant tit. The louse specimen from the Elegant tit is genetically very similar to lice from at least 24 other host species (Bush et al., 2016), and morphologically indistinguishable from specimens collected on additional species (D.R. Gustafsson, unpublished data). Based on morphology, we identify the phoretic louse on the fly Ornithophila metallica from the Black-naped oriole carcass as also belonging to the genus Guimaraesiella ZRC_BDP0298039 (Figure 3e), which matched on Genbank with lice collected from five Australasian passerine species.

DISCUSSION

Phoresy remains an understudied and underappreciated phenomenon because taxonomic impediments make it particularly unlikely that new observations are communicated. We here propose to address the taxonomic impediments by sorting specimens to species-level based on NGS barcodes. We believe that NGS barcodes are a good choice (Srivathsan et al., 2018; Srivathsan, Hartop, et al., 2019; Srivathsan, Nagarajan, & Meier, 2019; Wang et al., 2018; Yeo et al., 2018), because they are now sufficiently cost-effective (<US$0.10 per specimen: Srivathsan et al., 2021; Yeo et al., 2021) and can be obtained with very basic laboratory equipment within days (Srivathsan et al., 2018, 2021; Srivathsan, Hartop, et al., 2019; Srivathsan, Nagarajan, & Meier, 2019; Wang et al., 2018). Barcodes can be matched easily across samples collected at different times and places so that host and phoront species repeatedly involved in a phoretic relationship can be prioritized for identification and/or description. In addition, specimens, especially in their early instar stages, can be assigned to putative species via barcode databases (Yeo et al., 2018). This may be especially useful for identifying immature avian chewing lice, as these stages are not often described, and their morphology may differ substantially from that of conspecific adults (Mey, 1994). Few records of phoresy of nymphs have been published, but Allingham (1987) reported a nymphal Haematopinus Leach on a buffalo fly, indicating that the ability for phoresy may be developed in nymphs as well. We here used NGS barcodes to find new three-way species interactions between louse, fly, and bird, but the same techniques are also valuable for ornithologists who are interested in two-way relationships between birds and their flies, lice, or mites.

Some biologists may object that using barcodes for delimiting putative species is unsatisfactory, but we would argue that it is a step in the right direction given that fewer than a third of the published phoresy cases involved lice identified to species. An additional ~20% of lice were only identified to genus and improving the taxonomic resolution of these records would require locating the relevant specimens in numerous collections. We predict that many would not be found and/or not sufficiently well preserved for identification based on morphology. Such species-level matching is more straightforward with barcodes. For example, matching the Guimaraesiella record from Singapore to a louse record from the Philippines obtained almost 20 years ago took minutes via NCBI although the genus attribution of both the louse and host species in question had changed (Del Hoyo et al., 1992; Gustafsson & Bush, 2017).

Note, however, that such assignments of specimens to putative species via barcodes are no endorsement for describing species based on barcodes only. As argued by numerous authors (Ahrens et al., 2021; Engel et al., 2021; Meier et al., 2021), barcode clusters are only first-pass grouping statements that require additional testing before the taxa can be described as species. Such groupings are, however, an important first step toward taxonomic revision and description. Fortunately, NGS barcoding is now sufficiently fast and cost-effective that it can replace morphological pre-sorting. If all specimens are barcoded, it also yields approximate abundance information and morphological testing of species boundaries can be carried out at the haplotype level (Hartop et al., 2021).

Our literature review reveals how taxonomic impediments impact the understanding of phoresy. Many records could not be used because they lacked taxonomic resolution – mostly for lice. This meant that a comparatively low number of reports cover louse-hippoboscid phoresy in sufficient detail to assess the importance of phoresy for louse-host specificity and co-speciation between lice and bird hosts. This is unfortunate given that bird lice are species-rich (Price et al., 2003; Gustafsson, Zou, et al., 2019) with many bird species hosting several louse species. This makes it likely that the number of louse species will eventually exceed the number of bird species. Despite the small number of published records, we were able to reconstruct two large species interaction networks that connect 87 species of birds, 16 species of flies, and 18 species of lice. The networks are so sizable because many hippoboscid flies visit a very large number of hosts and have wide geographic distributions (Bequaert, 1953). In addition, many bird hosts are migratory, which increases the chance for transferring lice from birds breeding in the temperate region to birds resident in the tropics and vice versa.

It should be noted that the networks would be even larger if we had mapped all two-way species interactions (bird-louse and fly-bird). For example, the Guimaraesiella specimen obtained from the Blue-winged pitta belongs to an undescribed species that has been found on 24 other host species (see Bush et al., 2016). Based on morphology, specimens likely belonging to the same species are furthermore known from another 30+ host species (D.R. Gustafsson, unpublished data). The geographic range of these specimens spans from New Guinea and Australia over China, Thailand and India to Malawi (Bush et al., 2016). This illustrates that screening a sufficiently large number of bird carcasses and barcoding the phoronts would likely yield very large species interaction networks.

How species interaction networks expand with the addition of only a few new records is illustrated by the new phoresy cases reported here. One involves two hippoboscid flies belonging to the same species (Ornithoica momiyamai) obtained from the same Blue-winged pitta carcass. Both carried the same louse species (Guimaraesiella sp.) that we had previously found on a Yellow-rumped flycatcher carcass from Singapore. As a long-range migrant that breeds in East Asia and overwinters in western Sundaland (Del Hoyo et al., 1992), the broad geographical range of the Yellow-rumped flycatcher likely brings the species in contact with a large number of hippoboscid flies throughout its range. This may facilitate the phoresy-mediated transmission of Guimaraesiella sp. lice to numerous other bird species in the region (see Bush et al., 2016). The combination of highly vagile avian and hippoboscid hosts as vectors of louse dispersal may also explain why this Guimaraesiella species was previously genotyped from a Philippine endemic host, the Elegant tit (Del Hoyo et al., 1992).

The unexpected discovery of a putatively identical Guimaraesiella species on a pitta in Singapore may represent a case of an incipient phoresy-mediated host-switch given that no other Guimaraesiella species has ever been reported from a pitta species (Gustafsson & Bush, 2017) and we did not find Guimaraesiella specimens on two additional pitta carcasses with lice. Pittas are typically parasitized by lice of the genus Picicola (Somadder & Tandan, 1977) which are only distantly related to the Brueelia-complex, of which Guimaraesiella is part (Johnson, Weckstein, et al., 2002). However, additional records will be needed to better assess whether the Guimaraesiella-pitta association will lead to host-switching, since it is possible that the Guimaraesiella sp. species may not be able to establish itself on pittas.

In contrast to Ornithoica momiyamai, which is found throughout Asia and is known to feed on 11 bird species (Maa, 1966a,  1969a; Suh et al., 2012), our second novel phoresy record involves Ornithophila metallica, which has an exceptionally wide distribution across all biogeographic regions (except Antarctica) and is known to feed on bird species belonging to 134 genera (Maa, 1969a, 1969b, 1969c, 1969d; Suh et al., 2012). It is thus surprising that the Guimaraesiella louse species associated with Ornithophila metallica does not have a wider host distribution. This may be due to poor sampling in the Old World tropics. Our louse specimen is most similar to specimens from Australasia (see Bush et al., 2016) and our record is the first outside of this region although an association with Ornithophila metallica should open up many additional opportunities for host switching. Within Australasia, this louse species is known from non-migrating hosts belonging to at least four different families, suggesting that phoresy may have been important for the distribution.

We obtained the new records by screening >130 bird carcasses sourced with the help of citizen scientists. We found that ~73% of the carcasses had lice, ~16% had hippoboscid flies, and ~1.5% had hippoboscid flies carrying lice. Louse phoresy on hippoboscid flies is thus not particularly common, but may nevertheless be a significant phenomenon over evolutionary time given that recent studies of chewing lice have shown that successful establishment of louse populations on distantly related hosts may be rare, but not impossible (Bush et al., 2016; Gustafsson & Bush, 2017; Gustafsson, Lei, et al., 2019; Sychra et al., 2014). Obtaining the new phoresy records was not time-consuming because finding hippoboscid flies on bird carcasses is fast and the number of birds that are killed annually is vast. Loss et al. (2014) estimate that between 365 and 988 million birds die from window collisions in the United States alone and Grilo et al. (2020) estimate 194 million annual bird road kills in Europe. In Singapore, Tan et al. (2017) collected carcasses of 104 non-migratory and 204 migratory birds from 2013 to 2017 despite avoiding common species. We predict that hundreds of new phoresy records could be obtained via carcass screening within a short time period, especially if additional fly specimens were sourced from bird ringing initiatives. Screening bird carcasses for ectoparasites is thus a cost-effective complement to field sampling. It is non-invasive and requires very little specialist knowledge or field expertise. Note, however, that many flies abandon birds once the carcass cools to room temperature (Bequaert, 1953), so that screening has to be quick and be made part of carcass salvage protocols.

The widespread distributions and polyxenous nature of many hippoboscid flies explain why two of our species interaction networks are large although most published records lack taxonomic resolution for lice (Table S1 and Figure 2). Taxonomic impediments interfere with the publication of many natural history observations (Srivathsan, Hartop, et al., 2019; Srivathsan, Nagarajan, & Meier, 2019) and are particularly severe here because it is difficult to assemble a team consisting of an ornithologist and two entomologists with complementary taxonomic expertise. Even if such a collaboration can be arranged – as was the case in this study – the lack of comprehensive keys and a large number of undescribed louse species are major obstacles. Yet, obtaining accurate species-level data is crucial for reconstructing species interaction networks and understanding how the relationships between bird hosts, avian lice, and hippoboscid flies shape louse evolution.

CONCLUSIONS

Our study identifies significant gaps in our understanding of the phoretic relationships among avian lice, hippoboscids, and birds. The gaps are particularly large for many areas with rich avian faunas. At this point, we do not know whether these gaps reflect biology or sampling bias, but this question could be addressed quickly via screening bird carcasses from different biogeographic regions. This would not only facilitate the study of bird-louse coevolution as mediated by hippoboscid flies but also help with understanding why certain species/genera of flies and lice are frequently involved in phoretic relationships. It is important to restart natural history research (Tewksbury et al., 2014). Fortunately, combining traditional techniques such as carcass recovery and screening with new molecular techniques greatly facilitate and improve our ability to obtain new species-level results (Srivathsan, Hartop, et al., 2019; Srivathsan, Nagarajan, & Meier, 2019; Wong et al., 2017; Yeo et al., 2018).

ACKNOWLEDGEMENTS

We thank Kim Mi Jin for her help with the translation of Japanese and Korean papers and Tommy Tan, Goh Poh Moi, Morgany D/O Thamgavelu, and Ismail Bin Arshad for providing us access to NUS Lab 7's equipment and space for ectoparasite collection. The research was supported by a Ministry of Education grant on biodiversity discovery (R-154-000-A22-112) and the Pearl River Talent Recruitment Program of Guangdong Province (2019QN01N968).

Open access funding enabled and organized by Projekt DEAL.

    CONFLICT OF INTEREST

    The authors declare no conflict of interest.

    DATA AVAILABILITY STATEMENT

    The data from the literature review of this study are openly available in figshare at https://doi.org/10.6084/m9.figshare.12671711. The DNA barcodes of specimens involved in phoresy are openly available on NCBI GenBank (Genbank Accession numbers: MT762409-MT762417).