The influence of host dispersal on the gene flow and genetic diversity of generalist and specialist ectoparasites

The dispersal and subsequent gene flow within parasite species is the result of a complex interaction between parasite life history, host life history and abiotic environmental factors. To gain more insights into the drivers responsible for parasite dispersal, COI mtDNA genetic data derived from six southern African generalist parasite species, including fleas, mites and ticks were compared with four specialist species from the same geographic region. Generalist tick species represented by Amblyomma hebraeum, Hyalomma truncatum and H. rufipes all occur temporarily on highly mobile ungulate hosts and showed high levels of haplotypic genetic diversity and high levels of dispersal with an average intraspecific global F st (population differentiation index) value of 0.27 (±0.13). Generalist parasites, such as fleas, Listropsylla agrippinae and Chiastopsylla rossi, and one mite species, Laelaps muricola, that are all semi-permanent on the host and restricted to less mobile hosts species, showed a similar high level of genetic diversity, but an intermediate average F st value of 0.67 (±0.11). Highly specialised semi-permanent parasites, such as the mite L. giganteus and the permanent lice Polyplax praomydis, Hoplopleura patersoni and P. arvicanthis recorded the lowest level of genetic diversity and a low level of gene flow among geographic sampling localities with an average F st value of 0.95 (±0.05). This study provides strong support for the Specialist Generalist Variation Hypothesis (SGVH) and highlights the role that host dispersal and host specialisation by parasites play in the dispersal and evolution of ectoparasites.

Connectivity and subsequent gene flow among natural populations of species is one of the main factors maintaining conspecificity across a geographic range of a species. In the absence of sufficient gene flow among populations, genetic drift and selection can result in genetic divergence among populations that can ultimately contribute towards an allopatric speciation process (Schneider 2000;Huyse et al. 2005;Clegg and Phillimore 2010). For parasites, the factors affecting the dispersal and connectivity among populations are not well established (Barrett et al. 2008;van Schaik et al. 2014;DiBlasi et al. 2018). Through comparative phylogeographic studies between parasites and their hosts, it is evident that differences in parasite abundances on the hosts (Engelbrecht et al. 2016), ecological interactions between parasites (Cangi et al. 2013), and biogeographic conditions (Dick and Patterson 2007;du Toit et al. 2013;van der Mescht et al. 2015;Martinů et al. 2018;Stefan et al. 2018), can synergistically or separately all play a role in the dispersal and subsequent evolution of parasites. Given the close association between parasites and their hosts, and the fact that most terrestrial ectoparasite species are intrinsically not highly mobile, it is reasonable to suggest that host vagility will be a major contributing factor to facilitate parasite gene flow in terrestrial environments (Blasco-Costa, Waters and Poulin 2012;Araya-Anchetta et al. 2015;Matthee et al. 2018). In addition to host dispersal abilities, the life history of parasites and specifically also the host range (generalist or specialist) is probably equally important from an evolutionary point of view. Specialist parasites (those that can only survive on members of a single host species) often show more impediments to gene flow (high levels of genetic structure) and also often low levels of genetic diversity, a scenario best explained by parasites), and lower levels of genetic diversity in specialist species are expected, because these taxa are more prone to local stochastic extinction events when members of their single host species experience local stochasticity (Li et al. 2014;Matthee et al. 2018).
To estimate the rate of dispersal and subsequent gene flow among populations of species is no trivial task (Huyse et al. 2005;Cayuela et al. 2018;Duranton et al. 2019). Furthermore, when studying inconspicuous and small-bodied taxa, such as the majority of parasites, dispersal rates have to be inferred and cannot rely on traditional methods, such as mark recapture and or telemetry. One commonly used indirect measure of migration among populations is to use the population differentiation index expressed as F st (Wright 1949;Weir and Cockerham 1984). From the F st equation it is possible to infer the number of migrants per breeding generation and F st values range between zero (indicating complete panmixia) to one (indicating the complete absence of any movement inferred from the absence of any shared alleles). Although F st values are useful measures to compare population genetic structure across species (Bohonak 1999;Medina et al. 2018), it is important to acknowledge that this population differentiation measure also relies on assumptions that can bias the results (Waples 1998;Whitlock and McCauley 1999;Huyse et al. 2005;Bhatia et al. 2013;Spies et al. 2018). Typically factors, such as differences in effective population size between species (Waples 1998;Huyse et al. 2005;Sovic et al. 2019), geographic range of the sampling localities (Steele et al. 2014), evolutionary rate of different markers (Willing et al. 2012;Zhou et al. 2013), and small sample sizes (Willing et al. 2012) can all make direct comparisons of F st values among species problematic.
In the present study, global F st values and haplotypic diversity of ten ectoparasite species, that differ in their life history and also sometimes utilise the same host species, were compared with each other. To limit some of the tribulations associated with F st comparisons among species (Willing et al. 2012;Sovic et al. 2019), all these parasite taxa were represented by a reasonably large number of individuals (97-278; Table 1) and populations (6-25; Table 1), they were sampled in the same geographic region, and the COI and COII mtDNA molecular markers used show no dramatic intergene variability (Lopez et al. 1997). Three generalist (parasites completing their life cycle on multiple host species) tick species (Amblyomma hebraeum, Hyalomma truncatum, Hyalomma rufipes) that represent temporary parasites (parasites that are free-living during part of their life cycle) and whose adult stages commonly also occur on large mobile antelope and domestic ungulates (Cangi et al. 2013;Sands et al. 2017) are compared with two semi-permanent (parasites that utilise a host at one or more life stages, but some life stages are free living) generalist flea species (Listropsylla agrippinae, Chiastopsylla rossi) that exclusively occur on small rodents, shrews, and elephant shrews (Table 1; van der Mescht et al. 2015). Compared with the hosts of the tick taxa, the hosts of the flea species are less mobile; if host vagility is an important factor in parasite dispersion, then larger population differentiation values are expected for the fleas that occur on less mobile hosts. These five generalist parasites are compared with another generalist semi-permanent mite (Laelaps muricola) that is also restricted to rodents as hosts (Matthee et al. 2018) and to one host specific semi-permanent mite (L. giganteus), which is restricted to Rhabdomys spp. (Table 1; Engelbrecht et al. 2016). These parasitic taxa were then also compared with three permanent (a parasite that spends its entire life on its host), highly host specific, lice species (Polyplax praomydis and Hoplopleura patersoni that occur on Michaleamys namaquensis (Bothma et al. in review) and P. arvicanthis that occur exclusively on Rhabdomys spp. (Table 1; du Toit et al. 2013)). By drawing on comparative codivergence studies between parasites and their hosts (du Toit et al. 2013;Engelbrecht et al. 2016;Matthee et al. 2018;Bothma et al. in review) it is envisaged that the present comparative study can provide support for the generalisations that parasite host range (generalist versus specialist) and host dispersal ability are two of the main factors influencing parasite gene flow and diversity.
I hypothesise that host specific permanent parasites (such as lice) and those that are host specific, but not permanent (such as mites occurring mainly in the nest of the host), but which all occur on rodent hosts with limited dispersal capabilities, will show high levels of population differentiation and lower levels of genetic diversity. Generalist temporary parasites (such as ticks) occurring on both small mammals and large ungulates will show low levels of population differentiation and higher levels of genetic diversity (Li et al. 2014;Matthee et al. 2018). If the vagility of the host species is an important driver of parasite structure across the landscape, then it is reasonable to also hypothesise that generalist temporary species, such as fleas and some mites, only occurring on small mammals (rodents, shrews and elephant shrews) should show more genetic structure than ticks where some life stages also occur on highly mobile hosts (such as ticks and some mites; also see Araya-Anchetta et al. 2015).

Data
Published mitochondrial DNA cytochrome oxidase I (COI) data (or in the case of the fleas COII), derived from 1 515 ectoparasites representing 10 different species were included in the present study (Table 1). All samples were collected from hosts occurring in southern Africa and, where relevant, ethical clearance and sampling permits were obtained and reported in the various publications (du Toit et al. 2013;van der Mescht et al. 2015;Engelbrecht et al. 2016;Bothma et al. in review). All sequence data were included from the published studies except in two instances. Hyalomma truncatum is a widely distributed species throughout Africa and show considerable population differentiation at the continental level (Sands et al. 2017) and in this instance only individuals in the southern African clade were included. In the case of Polyplax arvicanthis, two cryptic species with similar genetic patterns were detected and only lineage 2 (the most comprehensively sampled clade) was included (du Toit et al. 2013).

Data analysis
Population differentiation values or global F st values were calculated in Arlequin v. 3.5.2.2 (Excoffier and Lischer 2010). Although expressed as F st in this paper, it is important to know that these values also take into account allelic evolutionary history (distances among different alleles) and strictly speaking therefore equate to Phi st (φ) values (Excoffier et al. 1992). Nucleotide diversity (π) and haplotype diversity (h) values were obtained using DNASP v. 5 (Librado and Rozas 2009). To additionally depict dispersal across the landscape, five localities of each of the 10 species were selected based on their distances from one another. In an attempt to standardise the sampling, distances among populations range from a minimum of 50 km to a maximum of 1 079 km, and in instances where localities overlap, the same comparisons were included for multiple species. All comparisons were done among localities within previously detected monophyletic geographic clades to minimise differentiation, because of natural barriers to dispersal. The latter specifically pertains to P. arvicanthis and L. giganteus where only samples collected from R. bechuanae were used (du Toit et al. 2013;Engelbrecht et al. 2016). In the case of H. truncatum only samples from South Africa and Namibia were considered (Sands et al. 2017). Distances among sampling sites followed straight lines and were determined by plotting and measuring the sampling localities for each species in Google-Maps. Pairwise F st values among sampling sites were obtained through an AMOVA analyses in Arlequin v. 3.5.2.2 (Excoffier and Lischer 2010). The natural log of the distance among these sites was then plotted against the pairwise F st values between the sampling sites for each species, respectively. A regression analyses was performed in Microsoft Excel to visualise the effect of isolation by distance for the different parasite species.

Results
Global F st values ranged from a low of 0.2 (p < 0.05) in the generalist temporary ticks A. hebraeum and H. truncatum to a high of 0.99 (p < 0.05) in the host specific permanent louse H. patersoni (   Figure 1). Average haplotypic diversity for generalist parasites was 0.91 (±0.12). Excluding A. hebraeum, which seemed to be an outlier, this average value for generalist parasites increased to h = 0.95 (±0.02). Specialists showed a lower level of heterozygosity with an average h = 0.84 (±0.06). Nucleotide diversity values did not show any marked trend that could be associated with the generalist or specialist parasites sampled in this study ( Table 2). The generalist tick species included herein showed weak relationships between geographic distance and F st values (cf. R 2 = 0.05 for H. truncatum and R 2 = 0.29 for H. rufipes), whereas the permanent host specific louse H. patersoni occurring on M. namaquensis showed the steepest isolation by distance slope and a much better linear relationship between F st and geographic distance (R 2 = 0.90; Figure 2). Generalist species, such as the fleas that only utilise small mammals during their life cycle and do not occur permanently on the host, showed more congruence with isolation by distance (R 2 = 0.60 and 0.87, respectively) than the generalist ticks that also utilise larger ungulates as part of their life cycle. In the case of the nest generalist, L. muricola, the relationship between geographic distance and genetic divergence was weak (R 2 = 0.16; Figure 2).

Discussion
Gene flow in ectoparasites is regarded as the result of a complex interaction between host dispersal abilities and host specialisation (Appelgren et al. 2018;Wessels et al. 2019), landscape effects (Nieberding et al. 2008;du Toit et al. 2013;Talbot et al. 2017) Matthee et al. 2018), it is reasonable to suggest that although gene flow in ectoparasites can be strongly influenced by host vagility, other factors, such as abiotic vicariance processes and parasite life history, also play a role (also see Engelbrecht et al. 2016). In the present study, both host vagility and host range (level of host specificity) are strongly correlated with the amount of gene flow among geographic populations. There was a strong correlation between host range and host mobility, and the F st values among the 10 parasites compared herein. Low F st values (ranging from 0.20 to 0.43) were present in the three tick species with a wide host range and which all utilise large ungulates for their adult stages. Intermediate F st values (ranging from 0.56 to 0.77) were detected in other generalist species, such as the fleas and L. muricola that mainly utilise small rodents and shrews, whereas high F st values (> 0.87) were present in the specialist mite (L. giganteus) and the three specialist lice all occurring on single rodent species (Table 2; Figure 1). Effective population sizes, however, can influence the F st values obtained (Huyse et al. 2005). Generally species with small effective population sizes tend to have elevated F st values, whereas species with larger effective population sizes will have comparatively lower F st values (Nadler 1995). Unfortunately, effective population size estimates for the species sampled herein do not readily exist, but it is interesting to note that in other studies, ticks of the genus Amblyomma showed low effective population sizes locally (Huber et al. 2019), whereas fleas showed higher effective population sizes (Harimalala et al. 2017). If this holds true, the differences in F st estimates among the 10 parasite species sampled herein, unambiguously point to a strong influence of host vagility and host range in the dispersal and subsequent gene flow in parasites (also see Araya-Anchetta et al. 2015).
The relationship between host dispersal and parasite dispersal is also reflected by the isolation by distance comparisons among the 10 parasite species. The four species that showed the weakest isolation by distance, as shown by the R 2 values (ranging from 0.04 to 0.42 in H. rufipes, H. truncatum, A. hebraeum and L. muricola), are generalists occurring on either highly mobile hosts, such as ungulates, or in the case of the mite, L. muricola, on several small rodents, but particularly also on Mastomys natalensis and M. coucha (Sands et al. 2014;Matthee et al. 2018). The majority of the hosts of these four generalists show panmictic or shallow population structures with admixture among geographic areas within their southern African range (Lorenzen et al. 2012;Sands et al. 2014). In the case of the flea taxa included herein, that are also generalists, there is a stronger pattern of isolation by distance (R 2 = 0.6 − 0.86). In the latter case, however, the common hosts of these parasites are all rodents documented to show phylogeographic structure throughout their range (host species for the fleas include  Smit et al. 2007). In line with the trend, the four specialist species also occur on geographically structured hosts Rhabdomys spp. and Micaelamys spp. and again they show a stronger pattern of isolation by distance (R 2 range from 0.65 to 0.91). Overall,  In small isolated populations inbreeding and bottlenecks will reduce intraspecific genetic variability (Nei et al. 1975;Charlesworth 2003;Huyse et al. 2005). A high level of migration and subsequent gene flow throughout the landscape will conversely maintain intraspecific genetic diversity (Slatkin 1985;Song et al. 2013). In parasites, the effective population sizes are generally smaller than those of the hosts (Huyse et al. 2005;Harimalala et al. 2017) and intraspecific genetic diversity is therefore also predicted to be more severely reduced by population fluctuations. Along these lines, host specific parasites will show the lowest genetic diversity, because they depend on a single host for survival, whereas generalists will show higher genetic diversity, because their population numbers are most likely more stable across the landscape (Li et al. 2014). With the exception of A. hebraeum, all generalists included herein have higher inferred genetic diversity (h ≥ 0.93) than found in the specialist (h ≤ 0.91). The reason(s) for the comparatively low haplotypic diversity in the generalist tick, A. hebraeum, is not clear. In fact, this tick species is more of a generalist when compared with H. rufipes (Cangi et al. 2003). The most plausible speculative explanation for the low genetic diversity in A. hebraeum may be linked to a comparatively small occupancy range of the species and also interspecific competition with the more widely distributed A. variegatum (Bournez et al. 2015). If A. hebraeum is exposed to more severe population fluctuations than the other two generalist tick species sampled in the present study, their genetic diversity will be comparatively lower. Irrespective of this discrepancy, when population structure and genetic diversity for the 10 parasites studied herein are considered, there is nearly 100% (with the exception of the genetic diversity in A. hebraeum) support for the SGVH (Li et al. 2014;Matthee et al. 2018).
The present study contributes to our understanding of some of the uncertainties surrounding the mechanisms involved in the dispersal and subsequent gene flow of ectoparasites. Although it is well documented that a variety of factors influence the dispersal of parasites (Nieberding et al. 2008;Cangi et al. 2013;du Toit et al. 2013;van der Mescht et al. 2015;Engelbrecht et al. 2016;Talbot et al. 2017;Appelgren et al. 2018;Wessels et al. 2019), the present comparative study highlight the important roles that host dispersal abilities and host specialisetion can play in the evolution of ectoparasites. Generalist temporary parasites occurring on highly mobile hosts showed a low level of population structure and a high level of genetic diversity, whereas specialist permanent parasites occurring on hosts that are also genetically structured throughout the landscape showed a high level of population differentiation and a lower level of genetic diversity.