Hitchhiking Parasites: Unstructured Populations of Bat Ectoparasites Reveal Host‐Driven Dispersal

TL;DR

This study explores how bat ectoparasites disperse across populations, revealing that host movement strongly influences parasite spread.

Contribution

The study compares dispersal dynamics of two bat ectoparasites using population genetics, revealing insights into host-driven parasite spread.

Findings

Both parasite species show high genetic homogeneity across metapopulations, indicating strong dispersal driven by bat mobility.

Population-specific FIS and excess of low-frequency alleles suggest within-site expansion and life cycle insights.

Findings highlight the role of host movement in shaping parasite population structures and potential pathogen transmission.

Abstract

Understanding factors influencing the dynamics and distribution of parasites is essential to decipher the mechanism behind their spread and the identification of populations with elevated risks of infection. Bats—together with the diverse parasites they host and the influence of their social behaviour on parasitism—offer a suitable system. We investigated the extent to which differences in life history traits between parasite species found on the same host influence their dispersal dynamics across bat metapopulations. To do so, we compared the population genetic structure of two obligate ectoparasites of the same bat, the Daubenton's bat (
Myotis daubentonii
): the specialist wing mite Spinturnix andegavinus and the more generalist bat fly Nycteribia kolenatii, and we expected the bat fly to exhibit a higher connectivity than the wing mites. Using double‐digest restriction site‐associated DNA sequencing (ddRADseq), we genotyped 426 bat flies and 171 wing mites across 13 maternity colonies and foraging sites and 1 putative swarming site in Switzerland. We found high genetic homogeneity across the metapopulations for both species, highlighting the dispersal capacity of N. kolenatii and S. andegavinus, probably driven by the high mobility of their bat hosts. The positive population‐specific F
IS and the excess of low‐frequency alleles in both species suggest within‐site expansion and provide insights into the life cycle of these ectoparasites. Altogether, these findings enhance our understanding of the interplay between nycteribiid flies and spinturnicid mites and their host movement in shaping population structure in Europe, offering broader insights into their potential role in pathogen transmission across bat populations.

Figures

Map of the 14 sampling sites of the study. Number in bold depict the sampling site ID. Sampling sites are: 1—Boiron Nyon, 2—Bois de la râpe, 3—Clarens, 4—Cully, 5—Dorigny, 6—Embouchure Aubonne, 7—La Cascade, 8—Lac Morat, 9—Le Pont, 10—Maison de la Rivière, 11—Moudon, 12—Rossinière, 13—Vufflens la Ville and 14—Yvonand. Sampling sizes before and after filtering, type of site, date of capture and host sex ratio are explained in Table 1.

Population genetic analyses based on the RAD sequences of the bat flies. (A) Distribution of the pairwise kinship between individuals. (B) Principal component analysis (PCA). (C) Heatmap of the mean kinship and pairwise F ST over pairs of sampling sites. The upper matrix is the mean kinship between sites (FsM); high values represent pairs of populations where individuals are more related than with other populations. The diagonal represents the intra‐populational mean kinship (population‐specific F ST), and the lower matrix represents the pairwise F ST. (D) Allele frequency distribution.

Population genetic analyses based on the RAD sequences of the wing mites. (A) Distribution of the pairwise kinship between individuals. (B) Principal component analysis (PCA). (C) Heatmap of the mean kinship and pairwise F ST over pairs of sampling sites. The upper matrix is the mean kinship between sites (FsM); high values represent pairs of populations where individuals are more related than those with other populations. The diagonal represents the intra‐populational mean kinship (population‐specific F ST), and the lower matrix represents the pairwise F ST. (D) Allele frequency distribution.

Tables

TABLE 1. Sampling site characteristics.

Site	Capture date	Type of site	Host sex ratio	Number of bat flies captured	Number of bat flies after filtering	Number of wing mites captured	Number of wing mites after filtering
1	17‐07‐2023	Foraging	3.67	28	28	28	15
2	06‐07‐2023	Foraging	10.5	28	23	28	12
3	30‐08‐2023	Maternity colony	0.133	18	17	18	7
4	08‐08‐2023	Maternity colony	0.259	38	34	38	12
5	13‐07‐2023	Foraging	1.33	22	21	22	8
6	05‐07‐2023	Foraging	0.333	44	40	44	10
7	06‐09‐2023	Swarming site	3.5	10	9	10	6
8	28‐06‐2023	Foraging	0.421	28	27	28	15
9	16‐08‐2023	Foraging	1.29	34	32	34	7
10	14‐07‐2023	Foraging	0.783	44	41	44	19
11	18‐07‐2023	Maternity colony	0.611	58	58	58	30
12	09‐08‐2023	Foraging	2.36	38	37	38	15
13	10‐08‐2023	Foraging	1.18	28	24	28	8
14	15‐08‐2023	Foraging	1.33	38	35	38	7

Full text

Introduction

1

Parasites depend on their host for survival and spread (Combes 2004), making their spatial dynamics closely linked to host dispersal capacity (Mazé‐Guilmo et al. 2016). Although specialist parasites rely on a single host, others navigate complex life cycles involving multiple host species. In such systems, limited dispersal in one host can be offset by the high mobility of another, facilitating parasite spread (Louhi et al. 2010). These host–parasite interactions create a complex dynamic network that is shaped by the life‐history traits of all species involved. Thus, life cycle complexity, reproductive strategy, host specificity and dispersal behaviour are key traits known to significantly influence parasite evolutionary trajectories (Barrett et al. 2008; Olsen 1974). In parallel, the life history and spatial structure of the host species strongly affect parasite dispersal. For example, the degree to which hosts aggregate into stable social units can promote intense transmission within groups while limiting opportunities for spread between groups. Similarly, host philopatry, seasonal movements, or even spatial distribution of breeding sites can either restrict or enhance parasite connectivity across the landscape (Barrett et al. 2008; Jarne and Théron 2001). Yet, mismatches between the genetic structure of hosts and that of their parasites are frequently observed (Mazé‐Guilmo et al. 2016). For example, structured host populations may harbour genetically homogeneous parasite populations when parasite dispersal is not exclusively mediated by the host (Witsenburg et al. 2015). Conversely, parasites infecting genetically well‐connected host populations can still exhibit genetic differentiation at both local and regional scales (Jossart et al. 2017). Therefore, a comprehensive understanding of how parasites spread within populations is required to decipher host–parasite interactions and to determine how their respective life‐history traits shape parasite distribution and dynamics. Genetic and genomic approaches provide powerful tools for investigating the mechanisms driving parasite transmission within and among populations (Blanchong et al. 2016). Specifically, differences in spatial genetic structures can reveal important patterns highlighting the variety of parasite spreading strategies (Biek and Real 2010; Blanchong et al. 2016).

In this context, bats form a suitable model for the many parasites they host as well as their dynamic life cycle influencing their susceptibility to parasitism. The spatiotemporal heterogeneity of movements between different bat species, among populations or even between sexes affects the dispersal of their obligate parasites (Amman et al. 2012; Christe et al. 2000, 2007; Dietrich et al. 2015; Sándor et al. 2019; van Schaik and Kerth 2017; Webber and Willis 2016; Witsenburg et al. 2014). Among ectoparasites, bat flies and wing mites are two obligate hematophagous bat parasites that reproduce on their host. Their life cycles differ in the location of early developmental stages and in their degree of host specialisation. Mites of the family Spinturnicidae are viviparous; they complete their entire life cycle on the membranous part of their host and are strictly dependent on close contacts between individuals to disperse and infect new hosts (Dowling 2006; Rudnick 1960). Because most wing mite species are associated with one or a few bat species, coevolution between wing mites and bats is strong and highly specialised (some host switching can occasionally occur—Bruyndonckx, Dubey, et al. 2009). Given this strong host specificity, wing mite dispersal is expected to be entirely dependent on the host movements. On the other hand, bat flies are less specific; while they have a major host, they can be found on several bat species (Szentiványi et al. 2016). The eggs and the larval stages occur within the female, and the terminal larva (prepupa) is deposited by the female fly on the cavity wall of the roost of the bat (Dick and Patterson 2006). Upon emergence, adult flies seek out and colonise new bats, potentially generating greater gene flow in bat flies than in wing mites. This process depends on roosts being used sequentially by multiple bat individuals or species, enabling emerging flies to encounter new hosts. Consequently, bat fly connectivity is expected to be higher between host populations than that of the wing mite.

Studies investigating population genetics of bat flies and wing mites are scarce and give contrasting results (Bruyndonckx, Henry, et al. 2009; Pejić et al. 2022; van Schaik et al. 2014; Witsenburg et al. 2015). On the same bat species, Myotis bechsteinii , two different parasite have been shown to have different connectivity patterns: the wing mite (Spinturnix bechsteini) populations are strongly geographically and temporally differentiated (van Schaik et al. 2014), whereas populations of the bat fly (Basilia nana) have no genetic structure within and between sites (van Schaik et al. 2015). But these results are not consistent among all wing mite species. For example, the wing mite S. myoti, specific of Myotis myotis shows high levels of genetic diversity and very little genetic differentiation (van Schaik et al. 2014). These studies pointed out that genetic population differentiation in parasites is driven by the host life history traits such as bat population size, social structure, mating system and dispersal patterns (van Schaik et al. 2014). All these studies rely on genetic markers such as mtDNA gene fragments and/or a few numbers of microsatellites. Although such markers provide valuable insights, their resolution is often limited, especially for detecting fine‐scale population structure or recent demographic events (Cumer et al. 2022; Dufresnes et al. 2023; Sunde et al. 2020). To resolve this issue, restriction site associated DNA sequencing (RADseq) can be used to genotype thousands of Single Nucleotide Polymorphism (SNPs) across the populations (Andrews et al. 2016; Davey and Blaxter 2010; Peterson et al. 2012), as it allows to detect subtle deviations from panmixia, infer recent demographic changes such as post‐hibernate expansion, and test whether rare alleles reveal hidden population substructure where microsatellite cannot (Cumer et al. 2022; Dufresnes et al. 2023; Sunde et al. 2020) and thus gives insight into individual movements, population differentiation and even disease spread (Dufresnes et al. 2022, 2023; Wilder et al. 2015).

Understanding how parasite life history influences local genetic structure can provide key insights into their dispersal dynamics across bat metapopulations. Both ectoparasites carry a great variety of pathogens and act as vectors within bat populations (Jaenson and Wilhelmsson 2021; Leulmi et al. 2016; McKee et al. 2019; Tendu et al. 2022) and can transmit agents associated with human zoonoses (Lack et al. 2011; Nováková et al. 2009; Reeves et al. 2006; Sándor et al. 2018, 2019, 2021, 2024; Szentiványi et al. 2019, 2022; Szubert‐Kruszyńska et al. 2019; Thiévent et al. 2020). Because of their distinct life‐history traits, obligate bat ectoparasites such as bat flies and wing mites are likely to differ in their role as potential disease transmitters. Examining bat flies and wing mites' genetic diversity and structure can reveal the degree of connectivity among parasites' populations and provide an initial insight into pathogen spread in bat populations.

Here, we investigated the genetic diversity and structure of bat flies (Nycteribia kolenatii) and wing mites (Spinturnix andegavinus) of the Daubenton's bat ( Myotis daubentonii ) in Western Switzerland. This vespertilionid bat is present in almost all countries in Europe and its habitat ranged up to central Asia (Krättli and Moeschler 2021). They usually form small colonies along watercourses and, unlike most temperate bat species, do not only rely on swarming to locate mates and reproduce (Bogdanowicz 1994; Encarnaçao and Becker 2023). Whereas males of most bat species are solitary throughout the summer, Daubenton's bats show high sexual segregation, with female‐dominated roosts, bachelor roosts and mixed groups (Angell et al. 2013; Linton and Macdonald 2019; Russo 2002). In addition to mating at swarming sites, mating also occurs at maternity colonies (Encarnação 2012a), with higher mating success in males that are spatially closer to females, either through proximity of bachelor roosts to maternity colonies or through co‐roosting in mixed colonies (Angell et al. 2013; Linton and Macdonald 2019). Consequently, the Daubenton's bat exhibit an absence of genetic structure at both local and continental scales (Atterby et al. 2010; Laine et al. 2013; Ngamprasertwong et al. 2008). This social system increases the opportunities for parasite transmission in Daubenton's bats. We hypothesised that if the life‐history of the parasite influences their dispersal opportunities then the bat fly should exhibit less population structure than the wing mite due to its time spent off‐host and its lower degree of specialisation. However, the host life‐history may also shape the parasite population genetics and therefore, the lack of spatial structure in the Daubenton's bat could facilitate gene flow between bat flies and wing mites metapopulations. To test these hypotheses, we used population genetic analyses based on RADseq data on parasites collected across maternity colonies, foraging sites and a putative swarming site.

Materials and Methods

2

Data Collection

2.1

Three hundred eighty‐nine M. daubentonii were captured in 14 sampling sites (bat maternity colonies, foraging sites and a putative swarming site) across the Vaud canton in Switzerland, between June and September 2023, under the licence authorization VD3679 (Figure 1, Table 1). For each bat, all bat flies and wing mites were collected, put in absolute ethanol and conserved in −20°C freezers. In total, 1197 bat flies and 1307 wing mites were collected. The species and sex of each ectoparasite were identified. To compare genetic diversity within bat, between bats and between sampling sites while maintaining the number of ectoparasites sequenced reasonable, we randomly selected two bat flies (N. kolenatii) and two wing mites (S. andegavinus) per bat. Since not all bats had two or more ectoparasites, 228 bats were retained for a total of 456 bat flies (N. kolenatii) and 456 wing mites (S. andegavinus).

Map of the 14 sampling sites of the study. Number in bold depict the sampling site ID. Sampling sites are: 1—Boiron Nyon, 2—Bois de la râpe, 3—Clarens, 4—Cully, 5—Dorigny, 6—Embouchure Aubonne, 7—La Cascade, 8—Lac Morat, 9—Le Pont, 10—Maison de la Rivière, 11—Moudon, 12—Rossinière, 13—Vufflens la Ville and 14—Yvonand. Sampling sizes before and after filtering, type of site, date of capture and host sex ratio are explained in Table 1.

DNA Extraction and Library Preparation

2.2

To maximise individual DNA and avoid contamination from the intestinal tracts, parasite legs were dissected under binocular magnifier for every individual. DNA of bat flies and wing mites was then extracted from the legs using the DNAeasy Blood & Tissue kit (Qiagen, Hilden, Germany) following the instructions of the manufacturer. Extracted DNA samples were kept at −20°C until library preparation.

ddRAD sequencing and library preparation was done in plates, with the parasites from the different sampling sites present in each plate. The protocol followed Brelsford et al. (2016), adapted from Parchman et al. (2012) and was modified for the Aviti sequencer. First, DNA was digested using the EcoRI‐HF and MseI restriction enzymes (New England Biolab, Ipswich, MA, USA). Then, enzyme adapters were ligated to the fragments. A unique barcode was included in the EcoRI adaptors for library tagging. After purification, RAD fragments were amplified with i5‐ and i7‐indexed adaptors (Unique Dual Indexes UDI) combined with PCR primers for individual tagging. Samples were then pooled by libraries and ectoparasites. DNA fragments were size selected in between 300 and 500 base pairs (bp) using BluePippin 2% DF Marker V2 cassette (Sage Science, MA, USA). Species‐specific size‐selected libraries were then quantified, pooled, and sequenced on two separate lanes of the SP flowcell in the Aviti sequencer at the Lausanne Genomic Technologies Facility (GTF, University of Lausanne) using paired‐end 150 bp sequencing. For the bat flies, a total of 495,854,653 reads were sequenced. The number of reads per individual ranged from 181,596 to 4,807,453, with an average of 1,087,401 reads per individual (Figure S1). For the wing mites, two sequencing runs were necessary because of the low coverage initially obtained (Appendix S1). After the resequencing, a total of 1,025,309,625 were obtained with an average number of reads per wing mite of 2,204,967 (Figure S1). The number of reads per individual ranged from 7967 to 11,194,013.

ddRAD Genotyping and SNPs Calling

2.3

Bat flies and wing mites RAD sequences were processed separately but with the same pipeline, an adaption of the ProcessMyRad pipeline from Cumer et al. (2021). First, reads were trimmed to remove the Illumina adapter sequences, using the bbduk function from bbmap v39.1 (Bushnell 2014). Reads were then processed with Stacks v2.53 (Catchen et al. 2013). Individual sequences were demultiplexed using process‐radtags. The filters used had to be adapted to each species‐specific dataset (Table S1). Using the de novo pipeline of Stacks v2.53, we tested different ranges of thresholds to call loci and decided on filters that gave the best coverage and number of SNPs. We set a mismatch threshold within individuals (M) of 3 and 5 for the bat flies and wing mites, respectively, a coverage threshold within individuals (m) of 4 and 3 for the bat flies and wing mites, respectively and a mismatch threshold between individuals (n) of 4 and 6 for the bat flies and wing mites, respectively. SNPs were identified with gstacks and first filters were done using populations. For both species, SNPs were filtered to be present in 60% of the individuals within and among sampling sites including markers of heterozygosity of 0.5 (‐‐max‐obs‐het 0.5). After quality control on each VCF file, correlation between heterozygosity and coverage was checked.

Individuals with more than 25% missing data were excluded from the analyses (30 individuals out of 456 for the bat flies and 285 individuals out of 456 for the wing mites). Further filtering was performed using vcftools v0.1.16 (Danecek et al. 2011), with parameters values adjusted for each dataset. To minimise errors in heterozygous genotype calls caused by insufficient sequencing depth, we applied a minimum coverage threshold (‐‐minDP) of 5X for the bat flies and 10X for the wing mites. SNPs present in at least 80% (bat flies) and 70% (wing mites) of individuals were retained (‐‐max‐missing). A mean minimum coverage (‐‐min‐meanDP) of 10X and 20X and a mean maximum coverage (‐‐max‐meanDP) of 65 and 60X for the bat flies and the wing mites, respectively was set up. Finally, only SNPs under Hardy–Weinberg equilibrium were retained with a p‐value threshold of 0.05 (‐‐hwe 0.05) and SNPs with less than 5 minor allele count (mac) were discarded (‐‐mac) for both species. A total of 1544 SNPs were retained for 426 bat flies with a mean coverage per individual of 32.22X (ranging between 11.82X and 129.04X) and 9525 SNPs for 171 wing mites with a mean coverage of 28.42X (ranging from 17.77X to 77.53X).

In both species, some individuals showed high kinship with many others, likely reflecting an excess of reference alleles (Appendices S2 and S3). Although including these individuals did not clearly affect the results (Appendices S2 and S3), we excluded them from subsequent analyses. Consequently, 354 bat flies and 163 wing mites were retained for the subsequent analyses.

Data Analysis

2.4

To determine the genetic diversity and differentiation of the vectors at both the individual and population levels, individual heterozygosity was determined based on the number of heterozygous sites relative to the total number of genotyped loci. Values were then averaged across individuals within each sampling site and across all sites. We used a Kruskal‐Wallis test to assess whether heterozygosity differed significantly among sampling sites. When significant, a Tukey's HSD post hoc test was used to identify populations that differed significantly from the others. To analyse the population genetic structure of bat flies and wing mites, the following analyses were performed using the hierfstat v0.5.11 package (Goudet et al. 2022) unless otherwise indicated. First, Principal component analyses (PCA) were performed to look for genetic clustering among sampling sites using indpca function. To assess the genetic differentiation among sampling sites, population‐specific F ST and F IS, population mean pairwise kinship (FsM matrix) and F ST (F st 2 × 2 matrix) were calculated using the fs.dosage function. These F statistics were bootstrapped 100 times across loci, with all loci considered independent since loci could not be mapped to a reference genome. To test whether rare alleles showed evidence of population structure, fs.dosage function was applied to a down‐sampled dataset retaining only rare alleles (mac ≤ 10). The pairwise kinship between pairs of individuals was measured using the beta.dosage function. To test for isolation by distance, the correlation between the pairwise F ST and the geographical distances between sampling sites was performed using a mantel test with 1000 permutations with the mantel function from the vegan package (Dixon 2003). Sex‐biased dispersal between bat flies and wing mites was tested using the sexbias.test function with the mAIc method as described in Goudet et al. (2002). Finally, to understand the vectors population dynamics, allele frequency distribution was drawn and the Tajima's D were estimated for each SNPs using the ‐‐TajimaD function from vcftools v0.1.16 before any minor allele filters (Danecek et al. 2011). Contemporary effective population sizes (N e) of both species were estimated using the linkage disequilibrium method from the NeEstimator v2.1 software (Do et al. 2014) and implemented in the gl.LDNe function from the dartR package (Gruber et al. 2018), assuming non‐overlapping generations and random mating. A minor allele frequency threshold of 0.05 was applied for each calculation.

Results

3

Population Genetics of the Bat Flies

3.1

At the individual level, the mean heterozygosity per individual ranged from 0.067 to 0.101 with an average of 0.084. After removing spurious individuals, pairwise kinship between individuals ranged from −0.194 to 0.142 with a mean of 3.263e^−17^ (Figure 2A; Appendix S2).

Population genetic analyses based on the RAD sequences of the bat flies. (A) Distribution of the pairwise kinship between individuals. (B) Principal component analysis (PCA). (C) Heatmap of the mean kinship and pairwise F ST over pairs of sampling sites. The upper matrix is the mean kinship between sites (FsM); high values represent pairs of populations where individuals are more related than with other populations. The diagonal represents the intra‐populational mean kinship (population‐specific F ST), and the lower matrix represents the pairwise F ST. (D) Allele frequency distribution.

The distribution of the diversity was homogeneous across sampling sites but revealed some non‐random mating in the population. Mean heterozygosity per sampling site ranged between 0.081 and 0.088, with a consistent level of individual heterozygosity across sampling sites (Kruskal‐Wallis test: χ ^2^ = 21.676, p = 0.061). However, the overall inbreeding coefficients (F IS) of 0.028 (95% CI = [0.022; 0.033]) were significantly different from 0. This trend was similar for population‐specific F IS, ranging from −0.031 to 0.075, with confidence intervals of 12 out of 14 sampling sites not overlapping 0 (Figure S2; Table S2).

Population genetic analyses revealed high levels of connectivity among bat fly sampling sites. The overall fixation index was extremely low (F ST = −7.923e^−05^; 95% CI [−0.001; 0.001]), indicating little to no genetic differentiation across sampling sites. Consistently, a principal component analysis (PCA) of individual genotypes showed no clustering by sampling location (Figure 2B). Pairwise F ST analysis between sampling sites ranged from −0.003 to 0.003, with all confidence intervals overlapping zero (Figure 2C; Table S3). Estimates of average genetic kinship between sites ranged from −0.021 to 0.026 (Figure 2C; Table S3), further supporting the absence of genetic structure. There was no significant correlation between the pairwise F ST between sampling sites and the geographic distances (Figure S3A; mantel test: r = 0.06586; p = 0.35365), suggesting a lack of isolation by distance. Down sampling the dataset to keep only loci with mac ≤ 10 yielded the same pattern and revealed the high connectivity across bat fly's populations (Figure S4).

The dynamic of the population was further assessed using the allele frequency distribution along with Tajima's D statistic (Figure 2D). The Tajima's D was significantly negative indicating an excess of rare alleles (mean Tajima's D = −0.708; Wilcox test: V = 1,004,951, p < 2.2e^−16^). Our results also showed that there was no significant sex‐biased dispersal between bat flies (test mAIc: t = −1.155, p = 0.249). Finally, the N e estimate of the bat flies was 13261.9 and ranged from 4508.7 to infinite (Jackknife 95% confidence intervals).

Population Genetics of the Wing Mites

3.2

At the individual level, the mean heterozygosity per individual ranged from 0.102 to 0.175 with an average of 0.158. After removing spurious individuals, pairwise kinship between individuals ranged from −0.065 to 0.131 with a mean of 9.795e^−17^ (Figure 3A; Appendix S3).

Population genetic analyses based on the RAD sequences of the wing mites. (A) Distribution of the pairwise kinship between individuals. (B) Principal component analysis (PCA). (C) Heatmap of the mean kinship and pairwise F ST over pairs of sampling sites. The upper matrix is the mean kinship between sites (FsM); high values represent pairs of populations where individuals are more related than those with other populations. The diagonal represents the intra‐populational mean kinship (population‐specific F ST), and the lower matrix represents the pairwise F ST. (D) Allele frequency distribution.

Heterozygosity of the wing mites showed little variation between sampling sites, ranging from 0.148 to 0.161 (Kruskal Wallis: χ ^2^ = 20.98, p = 0.073). Population‐specific F ST was consistent with all sampling sites having low values. However, five sampling sites (sites 4, 7, 8, 10 and 11) had confidence intervals that did not overlap 0 (Figure S5; Table S2). As in bat flies, the population‐specific F IS was low (ranging from 0.043 to 0.098 with an overall population‐specific F IS of 0.063), yet confidence intervals of all 14 sampling sites did not overlap 0, suggesting substructure in the metapopulation (Figure S5; Table S2).

Population genetic analyses revealed low levels of structure among the wing mites sampling sites. The overall fixation index was low (F ST = 0.001; 95% CI [0.0006; 0.003]) and individuals were scattered in the PCA without any clear clustering according to the sampling site (Figure 3B). Consistently, pairwise F ST analysis between sampling sites revealed low values, ranging from −0.008 to 0.006 (Figure 3C; Table S4). The pairwise kinship between sampling sites was low and ranged from −0.013 to 0.030 (Figure 3C; Table S4), supporting a high connectivity between sampling sites. A lack of isolation by distance was identified with an absence of significant correlation between the pairwise F ST between sampling sites and the geographic distances (Figure S3B; mantel test: r = −0.1827; p = 0.83516). The high connectivity was also present in the dataset including only rare alleles, in line with estimates based on the full dataset (Figure S6).

The dynamic of the wing mites population was assessed with the allele frequency distribution and Tajima's D statistic (Figure 3D). Tajima's D was significantly negative indicating an excess of rare alleles (mean Tajima's D = −0.539; Wilcox test: V = 54,992,098, p < 2.2e^−16^). There was no significant sex‐biased dispersal between wing mite individuals (test mAIC: t = 0.231, p = 0.818). Finally, the N e estimate was 3837.6 but the upper limit to the Jackknife 95% confidence intervals was infinite.

Discussion

4

In this study, we explored how different life‐history traits and level of specialisation influence patterns of genetic structure and dispersal in bat ectoparasites. We used Restriction site Associated DNA sequencing (RADseq) to sequence the genome of two ectoparasites of the Daubenton's bat ( M. daubentonii ), the generalist bat fly N. kolenatii and the specialist wing mite S. andegavinus. We assessed how their life‐history traits and degree of specialisation shape their spatial genetic patterns within populations. Contrary to our initial prediction that wing mites would exhibit stronger genetic structure due to strict host dependence, we found that both ectoparasite species are genetically homogeneous across bat metapopulations and exhibit an excess of rare alleles and negative Tajima's D.

High Connectivity Between Sampling Sites in Both Parasites

4.1

Spatial Homogeneity

4.1.1

Our aim was to test if different parasite life‐history traits would influence dispersal potential in two bat ectoparasites and would show different patterns of genetic structure. Because of the higher degree of specialisation of the wing mite, we predicted that the Daubenton's bat wing mite S. andegavinus would exhibit higher population structure than the bat fly N. kolenatii. In terms of diversity, our results showed lower heterozygosity in bat flies (0.062–0.101) than in wing mites (0.102–0.175). However, regarding the genetic structure of the populations, in both mites and flies, the low overall and pairwise F ST, the homogeneous heterozygosity across sampling sites and low kinships between pairs of individuals independently of their locality of origin suggested strong gene flow across sampling sites (Figures 2 and 3). This pattern was further supported by the absence of isolation by distance (Figure S3).

The strong connectivity of bat flies was consistent with previous studies investigating the population genetic structure of different species of bat flies in several bat species (Olival et al. 2013; van Schaik et al. 2015; Witsenburg et al. 2015). However, studies of wing mite population genetic structure have yielded contrasting results (Bruyndonckx, Henry, et al. 2009; van Schaik et al. 2014; Zamora‐Mejías et al. 2022), highlighting the importance of host social systems, in particular in terms of home range size, population size, mating system and dispersal patterns (van Schaik et al. 2014). In our system, the combination of swarming behaviour, within‐colony mating, seasonal roost switching, and mixed hibernation groups in Daubenton's bats likely homogenises parasite gene pools across colonies, illustrating how host life‐history strongly shapes parasite genetic structure (Biek and Real 2010; Blanchong et al. 2016; Mazé‐Guilmo et al. 2016).

Population Expansion

4.1.2

Significant negative Tajima's D values were observed in both bat flies and wing mites, indicating an excess of rare alleles and suggesting that populations are expanding (Aris‐Brosou and Excoffier 1996; Tajima 1989). Genotype‐based SNP calling may increase the number of rare variants, and Stacks can produce lower Tajima's D at low coverage (< 20X; Crawford and Lazzaro 2012; Han et al. 2013; Heller et al. 2021). Although low coverage can bias heterozygote calls and accentuate negative Tajima's D, our datasets had high mean coverage (bat flies: 32.22X; wing mites: 28.42X), which limits these effects. Missing data were also filtered below the threshold at which VCFtools begins to overestimate Tajima's D (> 30%; Bailey et al. 2025), thus giving confidence in our estimates.

In bat flies, one nycteribiid species has been reported to reproduce year‐round, whereas wing mites do not reproduce during winter (Lourenço and Palmeirim 2008). Both populations have been described to experience seasonal bottlenecks during the bat hibernation period, with a drop in the number of gravid bat flies and wing mites from late autumn and throughout hibernation (Lourenço and Palmeirim 2008). Because of this annual bottleneck, we expected to observe a reduced effective population size and increased genetic structure between sampling sites. Significant positive F IS (Figures S2A and S5A) showed that bat flies and wing mites reproduce more with individuals from the same sampling site. It appears that, after the winter bottleneck in all populations, the population size increases when bat flies and wing mites reproduce in maternity colonies (Lourenço and Palmeirim 2008), consistent with an excess of rare alleles in the populations. It would be interesting to study the mutation rate of bat flies and wing mites to investigate if the population expansion found from the Tajima's D results is due to the winter bottleneck or not.

Population Substructure at Swarming Site

4.1.3

Despite the high connectivity in both species, significant population‐specific F IS suggested substructure in the metapopulation (Figures S2A and S5A). As F IS measures the degree of inbreeding within populations, significant F IS reflects that both ectoparasites mate more with individuals from the same sampling site than with individuals from different sites. This is consistent with the biology of the species since they spend all or most of their time on the bat. Their reproduction is synchronised with that of the host, with more gravid bat flies and wing mites during the pregnancy and the lactation of bats (Christe et al. 2000; Lourenço and Palmeirim 2008). The generation time of bat flies (two generations per year; Reckardt and Kerth 2006) indicates that several generations occur. The generation time of wing mites is not well known, but a removal experiment on S. bechsteini showed that there is no barrier to mite dispersal within a bat colony (Bruyndonckx, Henry, et al. 2009), increasing the likelihood that wing mites mate more with other mites from the same site.

The substructure was more apparent in wing mites, as shown by the mean kinship and pairwise F ST, with site 7 appearing more closely related and less differentiated from other sites (Figure 3C). However, this pattern may reflect limited statistical power due to the small sample size at this site. Initially 10 individuals were sampled in this site, but because of the filtering only 6 individuals were kept for the analyses. Even though most sampled sites are maternity colonies or foraging sites, site 7 is a presumed mating site for Daubenton's bats. The higher mean kinship between sampling sites could mean that individuals sampled in this site are genetically closer to individuals from the other sites because bats sampled in site 7 (and therefore wing mites) come from different maternity colonies, as it is usually the case in swarming sites (Johnson et al. 2015). The lack of genetic structure in both bat flies and wing mites is likely driven by frequent exchanges of individuals among bats at swarming sites, which homogenise allele frequencies across all sampling sites. These findings present a strong basis to test the swarming potential of this site, through capture‐mark‐recapture (CMR) of bats or by investigating the population genetic structure of the Daubenton's bats in the Vaud canton.

Bat‐Induced High Mobility of the Parasites

4.2

Despite their different level of specificity, Daubenton's bat flies and wing mites exhibit high connectivity between sampling sites and other factors could help increase the dispersion of these two ectoparasites. Bat life history has been highlighted as shaping the population genetic structure of its bat flies and wing mites in different ways (Olival et al. 2013; van Schaik et al. 2014, 2015; Witsenburg et al. 2015; Zamora‐Mejías et al. 2022). Globally, bat flies are expected to be less impacted by their host life‐history traits and usually show higher connectivity and less genetic structure than their host, as it is the case for N. schmidlii parasitizing the bent‐winged bat Miniopterus schreibersii (Witsenburg et al. 2015). The high mobility of the bat flies may result from their lower specificity, which allows them to parasitize multiple bat species (Szentiványi et al. 2016), thereby increasing their effective population size and reducing genetic drift. This hypothesis is consistent with the infinite upper bounds of the effective population size estimates for bat flies. Estimates of N e are highly sensitive to migration (Waples 2025; Waples and Do 2010). Because bat flies move from their host to the walls of the roosting cavity to deposit the prepupal stage, this could increase their dispersal and the connectivity potential (Marshall 1982). We assumed random mating and non‐overlapping generations, as the mating systems of bat flies and wing mites are poorly characterised. However, overlapping generations are unlikely to have substantially affected the estimates of N e (Waples 2025). Wing mites are more strongly influenced by their host's life‐history, but the effect varies depending on the social context and dispersal behaviours of the bats. For instance, Spinturnix bechsteini, which parasitises the highly philopatric Bechstein's bat, shows clear genetic structure, whereas S. myotis, the wing mite associated with the highly mobile Greater mouse‐eared bat does not (van Schaik et al. 2014). In our study, S. andegavinus N e estimates likely reflect the weak genetic drift driven by the high dispersal capacity and genetic connectivity of the Daubenton's bat, facilitating parasite exchange among colonies.

Daubenton's bat and Bechstein's bat share key life‐history traits including usually small colony size and swarming mating systems (van Schaik and Kerth 2017), which limits inter‐individual contact and consequently, parasite transmission outside of the mating season. We expected N. kolenatii and S. andegavinus to exhibit similar patterns of population structure to those observed in parasites of the Bechstein's bat (van Schaik et al. 2014, 2015). However, unlike the Bechstein's bat, Daubenton's bat populations exhibit high genetic diversity and low population differentiation at both the European and local scales, as shown using mitochondrial gene or microsatellite markers (Atterby et al. 2010; Laine et al. 2013; Ngamprasertwong et al. 2008). The social system of the Daubenton's bat increases the opportunities for parasite transmission. Indeed, while swarming is one mating strategy, Daubenton's male bats frequent female‐dominated roosts in the late summer and genetic analysis suggested that mating occurs within these roosts as well (Angell et al. 2013; Encarnação 2012b, 2012a; Encarnação et al. 2004; Linton and Macdonald 2019). Contrary to the highly philopatric Bechstein's bat, all Daubenton's bats, male and female, provide a viable host pool that can be consistently infected and exploited. Additionally, Daubenton's bats show fidelity to summer roosting areas, although they switch between maternity roosts (Ngamprasertwong et al. 2014), and hibernate in small or large groups from different maternity colonies. These behaviours enhance inter‐roost connectivity and likely promote parasite gene flow of both the bat flies and the wing mites (Bogdanowicz 1994; Johnson et al. 2015). Therefore, the high connectivity among Daubenton's bat roosts offers a compelling explanation for the observed low genetic differentiation in their ectoparasites. Indeed, a lack of genetic structure may reflect large and well‐connected host populations size and increasing host density is known to facilitate greater parasite prevalence and diversity (Morand and Poulin 1998).

Conclusion

5

Our study highlights the importance of host movement in shaping parasite population structure. In this system, Daubenton's bat parasite dispersals are largely shaped by the movement and social organisation of the bat rather than by parasite‐specific traits. When hosts are highly mobile and form dynamic social groups, parasites can achieve extensive gene flow regardless of their own dispersal abilities. The absence of spatial structure in Daubenton's bats, combined with seasonal aggregation in maternity colonies and frequent roost switching, appears to be the main driver of parasite connectivity at the regional scale. By comparing two parasites sharing the same host, we show that host behaviour can override differences in parasite life history and lead to similar population genetic outcomes. These findings emphasise the central role of host movement in structuring parasite populations and provide a framework for understanding how host ecology shapes parasite dispersal in other systems with obligate, host‐dependent transmission.

Author Contributions

C.C., J.G. and P.C. conceived and planned the study. C.C. did the field work and C.C. ran the analyses with the help of T.C. All authors contributed to the interpretation of the results. C.C. wrote the first draft of the manuscript with valuable inputs of T.C., and all authors provided critical feedback and approved the final manuscript.

Funding

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Table S1: Filters used per species and number of SNPs at each filtering step. Table S2: Bat flies and wing mites 95% confidence intervals for population‐specific F IS and F ST without the spurious individuals. Table S3: Bat flies 95% confidence intervals (CI) of F statistics among sampling sites without the spurious individuals. Table S4: Wing mites 95% confidence intervals (CI) of F statistics among sampling sites without the spurious individuals. Figure S1: Distribution of the number of reads per individuals, for (A) the bat flies and (B) the wing mites. Figure S2: Bat flies population‐specific F statistics based on 100 bootstraps on loci considered independent without the spurious individuals. Figure S3: Isolation by distance. Figure S4: Results on the dataset of bat flies including only rare alleles (mac ≤ 10). Figure S5: Wing mites population‐specific F statistics based on 100 bootstraps on loci considered independent without the spurious individuals. Figure S6: Results on the dataset of wing mites including only rare alleles (mac ≤ 10). Appendix S1: Wing mite's resequencing. Appendix S2: Population genetics of the bat flies on the entire dataset. Appendix S3: Population genetics of the wing mites on the entire dataset.