To explore how biogeography, parasite life history and host vagility influences evolutionary codivergences, we followed a comparative phylogeography approach using a host-specific nonpermanent mite, Laelaps giganteus, that occurs on four rodent species within the genus Rhabdomys. A mtDNA COI haplotype network derived for 278 parasite specimens showed marked phylogeographic congruence with host distributions. Analysis of the less variable nuclear intron Tropomyosin was in part consistent with these results. Although distance-based cophylogenetic analyses in axparafit failed to support significant mtDNA codivergences (P ≥ 0.02), event-based analyses revealed significant cophylogeny between sampling localities of Rhabdomys and Laelaps using core-pa (P = 0.046) and jane (P = 0.026; P = 0.00). These findings, in conjunction with the weak congruence previously reported among the permanent ectoparasitic lice Polyplax and Rhabdomys, suggest that host-parasite intimacy is not the most important driver of significant codivergence in our study system. Instead, the more restricted dispersal ability of L. giganteus, when compared to Polyplax, resulted in stronger spatial structuring and this could have resulted in significant codivergence. Host switching occurred predominantly on the edges of host distributions and was probably facilitated by climate-induced range shifts. When host ranges shift, the phylogeographic structure of L. giganteus is not reflecting the host movements as most of the nest bound parasites do not disperse with the host (they miss the boat) and the genetic contribution of the few dispersing mite individuals is often overwhelmed by the large number of individuals already present in nests within the new environment (causing them to drown on arrival).

Laelaps giganteus and Laelaps muricola (Mesostigmata; Laelapidae) are widespread and locally abundant host generalists on small mammals in southern Africa. The large host range and complex life history of these ectoparasites may allude to possible intraspecific cryptic diversity in these taxa. To assess genetic and morphological diversity in L. giganteus and L. muricola, we sampled 228 rodents at eight localities in South Africa. This sample included nine previously recorded host species and on these, L. muricola was only recorded from Mastomys natalensis and Micaelamys namaquensis while L. giganteus was found on Rhabdomys dilectus and Lemniscomys rosalia. Phylogenetic analyses of partial mtDNA cytochrome oxidase subunit I (COI) and nuclear ITS1 data strongly supported the recognition of L. giganteus and L. muricola, a scenario partly supported by the Tropomyosin intron. Strong support for evolutionary distinct lineages within L. giganteus is found: L. giganteus lineage 1 is confined to R. dilectus and L. giganteus lineage 2 is confined to L. rosalia. These host specific monophyletic lineages were also separated by 9.84% mtDNA sequence divergence and 3.44% nuclear DNA sequence divergence. Since quantitative morphometric analyses were not congruent with these findings, these two lineages more than likely represent cryptic species.

No abstract

BackgroundThe specialist-generalist variation hypothesis (SGVH) in parasites suggests that, due to patchiness in habitat (host availability), specialist species will show more subdivided population structure when compared to generalist species. In addition, since specialist species are more prone to local stochastic extinction events with their hosts, they will show lower levels of intraspecific genetic diversity when compared to more generalist.ResultsTo test the wider applicability of the SGVH we compared 337 cytochrome oxidase I mitochondrial DNA and 268 nuclear tropomyosin DNA sequenced fragments derived from two co-distributed Laelaps mite species and compared the data to 294 COI mtDNA sequences derived from the respective hosts Rhabdomys dilectus, R. bechuanae, Mastomys coucha and M. natalensis. In support of the SGVH, the generalist L. muricola was characterized by a high mtDNA haplotypic diversity of 0.97 (±0.00) and a low level of population differentiation (mtDNA Fst = 0.56, p < 0.05; nuDNA Fst = 0.33, P < 0.05) while the specialist L. giganteus was overall characterized by a lower haplotypic diversity of 0.77 (±0.03) and comparatively higher levels of population differentiation (mtDNA Fst = 0.87, P < 0.05; nuDNA Fst = 0.48, P < 0.05). When the two specialist L. giganteus lineages, which occur on two different Rhabdomys species, are respectively compared to the generalist parasite, L. muricola, the SGVH is not fully supported. One of the specialist L. giganteus species occurring on R. dilectus shows similar low levels of population differentiation (mtDNA Fst = 0.53, P < 0.05; nuDNA Fst = 0.12, P < 0.05) than that found for the generalist L. muricola. This finding can be correlated to differences in host dispersal: R. bechuanae populations are characterized by a differentiated mtDNA Fst of 0.79 (P < 0.05) while R. dilectus populations are less structured with a mtDNA Fst = 0.18 (P < 0.05).ConclusionsThese findings suggest that in ectoparasites, host specificity and the vagility of the host are both important drivers for parasite dispersal. It is proposed that the SGHV hypothesis should also incorporate reference to host dispersal since in our case only the specialist species who occur on less mobile hosts showed more subdivided population structure when compared to generalist species.Electronic supplementary materialThe online version of this article (10.1186/s12862-018-1245-7) contains supplementary material, which is available to authorized users.