Characterizing the microbiome of ectoparasitic louse flies feeding on migratory raptors

Louse flies (Diptera: Hippoboscidae) are obligate ectoparasites that often cause behavioral, pathogenic, and evolutionary effects on their hosts. Interactions between ectoparasites and avian hosts, especially migrating taxa, may influence avian pathogen spread in tropical and temperate ecosystems and affect long-term survival, fitness and reproductive success. The purpose of this study was to characterize the vector-associated microbiome of ectoparasitic louse flies feeding on migrating raptors over the fall migration period. Surveys for louse flies occurred during fall migration (2015–2016) at a banding station in Pennsylvania, United States; flies were collected from seven species of migrating raptors, and we sequenced their microbial (bacteria and archaea) composition using high-throughput targeted amplicon sequencing of the 16S rRNA gene (V4 region). All louse flies collected belonged to the same species, Icosta americana. Our analysis revealed no difference in bacterial communities of louse flies retrieved from different avian host species. The louse fly microbiome was dominated by a primary endosymbiont, suggesting that louse flies maintain a core microbial structure despite receiving blood meals from different host species. Thus, our findings highlight the importance of characterizing both beneficial and potentially pathogenic endosymbionts when interpreting how vector-associated microbiomes may impact insect vectors and their avian hosts.


Introduction
Each autumn millions of raptors migrate from their breeding grounds in North America to overwintering areas in South America, following well-established routes created by geographical features and ideal weather conditions [1][2][3]. One of the major migratory routes in North America, the Atlantic Flyway, ranges from northeastern Canada to the southern United States. The Atlantic Flyway extends along the Atlantic Coast and the Appalachian Mountains and is Pennsylvania, United States (40˚48' N, 75˚32' W), as part of an annual effort to monitor migrating raptor populations. Birds were trapped by authorized personnel using either mist nets or bow traps and a harnessed Rock Pigeon (Columba livia) [29] and immediately removed from traps and secured for processing, according to the U.S. Geological Survey Bird Banding Laboratory. A 90-second visual and physical survey (parting feathers) of the dorsal and ventral surface of the bird was conducted in order to identify the presence or absence of louse flies. A federal permit for the trapping, handling, and processing of birds was provided by the U.S. Geological Survey Bird Banding Laboratory (Permit #21371-G. Lahr, Little Gap Banding Station), and carried out in strict accordance with The Bander's Code of Ethics, approved by the North American Banding Council. Being a non-invasive procedure, no special permission was needed for the collection of louse flies off of raptor hosts. All louse flies from an individual were collected by hand or with forceps, immediately submerged in 100% ethanol (one tube per individual bird), stored at room temperature, and later transported to the laboratory for DNA extraction. Louse flies were identified to species level prior to extraction [22]. The age, sex, weight, species of raptor, date, and time of capture were documented for all birds from which louse flies were collected (S1 Table). Not all individuals being banded were examined for flies and not all flies were captured from each individual due to time constraints.

DNA extraction and sequencing
To identify the impact of host species and temporal variation on internal bacterial communities, collected louse flies were pooled by bird species and collection date, resulting in 1-5 individuals per sample (n = 2 louse flies/bird). A sub-sample of flies were sequenced from each species throughout the fall migration of 2015-2016 (Table 1). For purposes of decontamination, prior to DNA extraction louse flies received a two-minute wash in a 10% bleach solution.
To facilitate homogenization of the louse flies prior to extraction, louse flies were dissected (i.e. body cut into pieces) with flame-sterilized scissors. DNA extractions were performed following the DNA PowerSoil DNA Isolation Kit (Qiagen) manufacturer's recommended protocol. Individual DNA yields were quantified using the Qubit 1 dsDNA High Sensitivity assay and Qubit 2.0, and resulting DNA was stored at -20˚C prior to library preparation. Library construction and sequencing (Illumina MiSeq, 2 x 250 bp paired-end reads) was performed as described by [30] at the Michigan State University Genomics Core Facility. Dual indexed primers 515f/806r (5 0 -GTGCCAGCMGCCGCGG-3 0 , 5 0 -TACNVGGGTATC TAATCC-3 0 ) were used to amplify the V4 region of the 16S rRNA gene [30,31]. Prior to sequencing, the PCR products were normalized using SequalPrep 1 normalization plates

Statistical analyses
We tested for effects of year on diversity (Faith's PD, Shannon, P > 0.05) and taxonomic composition (Kruskal-Wallis, P > 0.05), and found no statistically significant difference, thus 2015 and 2016 were pooled for further analyses. To determine if sampling date or raptor host species impacted the diversity of the internal microbiome present in louse flies, differences in alpha-diversity metrics were tested using Kruskal-Wallis tests in R (v 3.5.2) with a False Discovery Rate (FDR) correction for multiple samples [38]. Pairwise comparisons among species were conducted using a Mann-Whitney test with an FDR correction in R. Similarly, differences in the relative abundance of bacterial phyla and genera among host species were tested with Kruskal-Wallis tests. To limit the potential for spurious results from taxa with very low abundance, only taxa which comprised greater than 1% of the total bacterial community were compared at the genus level. To test if temporal effects impacted the ectoparasite bacterial communities [i.e., early (September) or late (November) migrating raptors], beta-diversity metrics were analyzed using PERmutational Multivariate Analysis Of Variance (PERMA-NOVA) after checking for homogeneity of variances between samples groups in the vegan package (v 2.5-2) [39]. Figures were created using a combination of the vegan, phyloseq (v 1.24.1), and ggplot2 (v 3.0.0) libraries in R [40,41]. All code used in this analysis is available at: https://github.com/BenbowLab/MUHippoboscid.

Bacterial communities
A total of 40 samples comprising 64 louse flies were sequenced (Table 1) resulting in 2,490,039 reads and 339 sequence variants after filtering. To limit bias due to differing read library sizes [42], samples were rarified to 5,000 reads based on alpha rarefaction plots. We tested for effects of year on alpha-diversity (Faith's PD, Shannon Diversity, P > 0.05) and taxonomic composition (Kruskal-Wallis, P > 0.05), and found no statistical difference, thus 2015 and 2016 were pooled for further analyses. The phylum Proteobacteria dominated the internal microbiome of all louse flies but one sample, and comprised greater than 97% of the total bacterial community regardless of which raptor host species they were collected from ( Fig 1A). Firmicutes comprised 2% of the community, while the remaining 1% of phyla included a combination of Bacteroidetes, Actinobacteria, and Acidobacteria. The family Enterobacteriaceae comprised greater than 94% of the bacterial communities of I. americana, with a specific taxon classified as "primary endosymbiont of the pigeon louse fly (Pseudolynchia canariensis)" representing 68.13% (+/-5.34 SEM) of the total community ( Fig 1B). In addition to the endosymbiont of Ps. canariensis, bacterial genera detected in lower relative abundances included: Pectobacterium (19.31% +/-3.99 SEM), Arsenophonus (6.96% +/-3.74 SEM), Ignatzschineria (1.74% +/-1.72 SEM), and Exiguobacterium (1.69% +/-1.69 SEM). Among louse flies from different host species there were no statistical differences in the relative abundance of any bacterial taxa at the phylum, family or genus level (Kruskal-Wallis, P > 0.70). When testing for differences in beta-diversity, weighted UniFrac distance had significantly different homogeneity of variances between host species (PERMANOVA, P < 0.01) while Jaccard distance did not (P > 0.05), so Jaccard distance was used for subsequent analysis as it did not violate the assumptions of PERMANOVA. Similarly, there were no differences in alpha-diversity (Faith's PD, Shannon, Kruskal-Wallis, P > 0.5, Fig 1C) or beta-diversity (Jaccard, PERMANOVA, P > 0.90) metrics among louse flies on different avian host species. There were no statistically significant temporal differences [early (September) vs. late (November)] in taxonomic composition (Kruskal-Wallis, P > 0.05, Fig 1D), alpha-diversity metrics (Kruskal-Wallis, P > 0.05) or beta-diversity (PERMANOVA, P = 0.54, Fig 2).

Discussion
Few studies have examined the microbiome of raptor hosts, much less the microbiomes of their ectoparasites [43][44][45]. To our knowledge, this study is the first to characterize the bacterial communities of louse flies collected off of various species of migrating raptors in North America. Surprisingly, only one species of louse fly, I. americana, was found across seven raptor host species during fall migration. The relative bacterial abundance of I. americana was dominated by the phylum Proteobacteria (97%) and to a lesser extent Firmicutes (2%), regardless of which raptor host species they were feeding on. Proteobacteria and Firmicutes have also been identified as abundant bacteria in other microbiome studies of avian tick ectoparasites [46], carrion beetles [47], as well as other birds [44,48]. Similarly, 80% of the bacterial sequences detected from cloacal swabs of Bald Eagles captured from the coastal beaches of Washington and Oregon, were Firmicutes and Proteobacteria [49], suggesting these prominent phyla can be found across a range of host species (e.g., from insects to birds).
From the family Enterobacteriaceae, the most predominant genus found in the louse fly microbiome collected from all seven raptor host species was classified as a "primary endosymbiont of the pigeon louse fly (Ps. canariensis)". Another study characterized an endosymbiotic bacterium related to the Arsenophonus genus, Candidatus Arsenophonus arthropodicus, from the tissues of the pigeon louse fly [50]. In our study, we found a higher abundance (i.e., > 50% across species) of the primary endosymbiont from Ps. canariensis, thus suggesting a similar finding [50] with the presence of Arsenophonus sp. observed in our samples. Other arthropod studies have identified Arsenophonus, which forms a distinct monophyletic clade of arthropod endosymbionts in the gamma subdivision of Proteobacteria [50], in ticks [51], whiteflies [52], aphids [53,54], psyllids [55], and parasitoid wasps [56]. Primary endosymbionts have a long evolutionary history with their hosts insects and may play an unknown beneficial role in the host [50].
Louse flies and other closely related hematophagous insects maintain an obligate symbiosis with bacterial species (e.g., Candidatus Arsenophonus arthropodicus [50], Candidatus Arsenophonus lipopteni [57]). In many species, these interactions are highly specialized with a single to a few symbionts [57]. It has been hypothesized that this relationship is a result of the need for insects to obtain vitamins (e.g., vitamin B) that they do not acquire from their blood meals, but instead are produced by the symbiotic bacteria [58,57], and thus compounds in the host blood may have an effect on the community composition of the louse fly microbiome. Enterobacteriaceae are pervasive and considered part of the autochthonous intestinal flora of some species of birds, but not normally found in raptor species including hawks and vultures (Accipitriformes), falcons (Falconiformes), or owls (Strigiformes) [59]. Though we did not sample the microbiota of raptor host species in this study, the consistent dominance of this primary endosymbiont in all the louse flies regardless of which raptor host species they were collected from suggests this bacterium is specific to louse fly hosts and not acquired through blood meals.
Pectobacterium (Phylum: Proteobacteria, Family: Enterobacteriaceae) was the second most abundant bacterial genus found in louse flies in this study, but with < 1% found in louse flies collected from Northern Goshawks. These bacteria are related to soft-rot disease in plants [60] where they infect a range of hosts by consuming cellular nutrients and degrading plant tissues [61]. Pectobacterium contributes to its insect host's nutrition by fixing atmospheric nitrogen, an essential element for physiological activities such as reproduction and development [62]. Pectobacterium sp. have been associated with bark beetles (Dendroctonus spp.) in the United States and Mexico [63], Mediterranean fruit flies (Tephritidae: Ceratitis capitate) [64], Brazilian fruit flies (Drosophilidae: Drosophila spp.) [60], and wood-boring beetles in Italy [65]. Associations with nitrogen fixing bacteria, such as Pectobacterium, among tephritid and drosophilid Diptera are critical for fly development and reproduction [64]. Because louse fly larvae feed on nourishing "milk" glands from within female louse flies [23], it is possible that this endosymbiont may be beneficial to louse fly larvae in the processing of nitrogen-rich blood as a result of adults feeding on migrating raptors. If this is true, this is the first time a bacterial endosymbiont has been associated with a hematophagic (fluid-feeding) vector of potentially harmful endoparasites to birds.
The bacterial genera Exiguobacteria (Firmicutes) and Ignatzschineria (Firmicutes) comprised the remaining genera (< 4%) found within I. americana in this study. The genus Exiguobacterium is a diverse group of pigmented gram-positive bacteria [66] that possess stressresponsive genes allowing them to occupy and thrive in diverse ecological niches and can cause bacteremia in immunocompromised hosts [59,67]. This bacterium has been found in the microbiomes of herbivorous insects such as leafhoppers (Macrosteles sexnotatus) [68] and grasshoppers (Sathrophyllia femorata) [69], as well as in the midguts of the filter-feeding larval mosquito species Aedes japonicus, A. triseriatus, and Culex pipiens [70]. Interestingly, we document the presence of Exiguobacterium in hematophagic louse flies collected from Broad-winged Hawks (13.5%) only and not in the other six avian host species. Broad-winged Hawks are forest nesting raptors [71] and perhaps this association with louse flies may be connected to the breeding habitat differences between Broad-winged Hawks and the other raptor host species in this study. The other Firmicute bacterium found within louse flies in this study was Ignatzschineria sp. This gram-negative bacterium is strongly associated with carrion-breeding flies, especially the obligate parasitic fly, Wohlfahrtia magnifica (Diptera: Sarcophagidae) responsible for myiasis [72,73], black soldier flies (Hermetia illucens) (Diptera: Stratiomyidae) [74], and blow flies (Diptera: Calliphoridae) [75]. We found a higher percent of Ignatzschineria in louse flies of Cooper's Hawks (9.83%). Thus, further research is necessary to determine if the presence of these Firmicutes in ectoparasitic louse flies is related to the louse fly larval habitat, and if this factor influences the ability for the bacterium to be acquired by louse flies from host blood meals.
Louse fly reproduction is tied to the host nest where the larva pupates almost immediately after being deposited into the nest [23], thus infestations are common among hatch year (juvenile) birds as they provide a food source for newly emerged louse flies of that year [23, 24]. The infestation occurs in the nest, but the louse flies will remain and feed on the young birds throughout the year [24]. This relationship may explain why louse flies in our study were predominately collected off of hatch year host raptor species (S1 Table) during fall migration, although this warrants further investigation. Host switching was detected in Olfersia spinifera, a species of louse fly that parasitizes the Great Frigatebird (Fregata minor), when researchers discovered that the blood meal from the louse fly did not match the blood from the host [76]. The unknown life span of louse flies (i.e., probably surviving for several months) [23] in combination with dispersal of hatch year birds from their natal sites and potential for host-switching warrants further investigation of vector traits, such as its microbiome, that help facilitate the movement of louse flies among hosts and across geographical boundaries.
Vector-associated bacterial taxa can be beneficial to their insect hosts in multiple ways, including dietary specialization to an exclusive blood diet [77], development [78], and immunity [79]. Recent studies have also highlighted the importance of the microbiome in the context of host pathogen transmission processes [80][81][82]. The genetic ability for an insect to transmit pathogens is measured in terms of its vector competence and this competence depends on the proficiency of the host insect's immune responses, which may be facilitated through host immune priming by the host-associated microbiota [80]. For example, natural gut microbiota stimulated the same innate immune pathway that regulates viral resistance in mosquitoes [83]. The gut represents the primary pathogen entry point during each blood meal, therefore gut microbiota may directly mediate pathogen colonization and extrinsic incubation time (i.e., time necessary for pathogens to development inside the insect vector) [84], as observed in ticks fed on antibiotic-treated mice that exhibited modified gut microbiota composition and lower Borrelia burgdorferi colonization rates [85].
Previous studies have hypothesized the current consensus of global climate change may be promoting increased numbers of avian blood parasites and pathogens [86,87], which depend on the abundance and distribution of vectors, leading to an increase in documented ectoparasitic loads on avian hosts [88,89]. The prevalence of louse flies collected from nearly half (i.e., seven out of the 16) of the migrating raptor species in Pennsylvania demonstrates the potential for widespread distribution of louse flies. The seven host species for the studied louse flies migrate from Canada down to Mexico (e.g., Bald Eagle, Northern Goshawk, Red-shouldered Hawk), Central America (e.g., Red-tailed Hawk, Sharp-shinned Hawk, Cooper's Hawk), and as far south as the northern tip of Argentina, South America (e.g., Broad-winged Hawk). The geographical distribution of I. americana ranges from 49˚N to 30˚S [90], suggesting the intercontinental dispersal of louse flies via their migrating hosts. The first record of I. americana on a breeding American Kestrel (Falco sparverius) in central Argentina (i.e., 36˚S) indicates southern range expansion for this dipteran species [7].
In conclusion, I. americana was the only species of ectoparasitic louse fly found feeding on the seven species of raptor hosts sampled. We found no difference in the bacterial communities of louse flies regardless of which raptor species they were collected from. The louse fly microbiome was dominated by a primary endosymbiont, suggesting that louse flies may maintain a core microbial structure despite receiving blood meals from different host species. Based on this finding, we suspect this primary endosymbiont plays an unknown beneficial function for its insect host, such as increased vector competence. We suggest that future work should include isolating the primary endosymbiont from I. americana and test for strains of specific pathogens found within ectoparasite hosts. Future studies should sample for louse flies on raptors in South America, both residents and migrants, to determine if there is an overlap in vector-associated bacterial communities between continents.