A monogenean parasite reveals the widespread translocation of the African clawed frog in its native range

1. The management of bio-invasions relies upon the development of methods to trace their origin and expansion. Cointroduced parasites, especially mono - genean flatworms, are ideal tags for the movement of their hosts due to their short generations, direct life cycles and host specificity. However, they are yet to be applied to trace the intraspecific movement of host lineages


| INTRODUC TI ON
The Anthropocene is characterised by the unparalleled humanmediated passage of organisms from their natural ranges to ecosystems across the world (Capinha et al., 2015;Ricciardi, 2007).If these non-native organisms manage to disperse from their initial introduction point and seed new populations in their non-native range, they are regarded as invasive (Blackburn et al., 2011;Richardson et al., 2011).This reshuffling of the world's biota is a decisive agent of global change, as evidenced by the fact that invasive species are implicated as the cause of a third of recent animal extinctions (Blackburn et al., 2019).
Clearly, the development of methods to trace both the origin and expansion of biological invaders is an urgent task.Genomic tools can be informative for the reconstruction of invasion histories, the prediction of potential range expansion and invasion success and the development of control strategies (e.g. de Busschere et al., 2016;Dlugosch & Parker, 2008;Dufresnes et al., 2017;Hiller & Lessios, 2017;Hudson et al., 2022;Lee & Gelembiuk, 2008;Prentis et al., 2009;Stouthamer et al., 2017;Zahiri et al., 2019).
Nonetheless, in many cases the available molecular data or chosen molecular markers are of insufficient resolution by themselves to trace the movement of species (e.g.Ahyong et al., 2017;Criscione et al., 2006).Thus, despite the accessibility of genomic tools and the widespread use of phylogeographic analyses in invasion science, these methods certainly have their limitations (Fitzpatrick et al., 2012;Rius & Turon, 2020).
A more holistic approach to solving the origin and range shifts of invasive species involves molecular data from cointroduced parasites (Gagne et al., 2021).Such an integrative approach has been proposed to investigate invasions that are difficult to trace (Clavero et al., 2016).For example, the long-debated cryptogenic history of the European marine snail Littorina littorea (Linnaeus, 1758) was solved with the help of an associated digenean (Platyhelminthes: Digenea) parasite (Blakeslee et al., 2008).Likewise, intraspecific morphometric variation of monogenean flatworm (Platyhelminthes: Monogenea) parasites identified both the origin and introduced life stage of invasive clupeid fish in Africa (Kmentová et al., 2019), shed light on the history of an invasive population of odontobutid fish in Europe (Ondračková et al., 2012) and provided information on the introduction route of invasive gobiid fish in Europe (Huyse et al., 2015).
Moreover, the genetic variation among monogenean gill parasites was used to illuminate the introduction history of non-native cichlid fish in Africa (Geraerts et al., 2022;Jorissen et al., 2022).
Matters are complicated even further when we wish to trace movement of conspecific organisms between different populations.This phenomenon, also known as intraspecific cryptic invasion, describes the process where one lineage of a species spreads into parts of the species' native range where another lineage is already present, often with human help (Morais & Reichard, 2018;Saltonstall, 2002).Once again, cotranslocated parasites hold great promise to provide information on the movement of lineages of their hosts.Monogenean parasites in particular exhibit sufficient levels of phylogeographic structuring to assess intraspecific spatial patterns in their hosts (Baldwin et al., 2012;Hahn et al., 2015;Huyse et al., 2017;Kmentová et al., 2018).At the very least, the phylogeographic structuring of monogenean lineages can support evidence revealed by the host dataset or historic records, such as the well-documented anthropogenic introduction of salmonid fish host stocks across Europe that is reflected in the phylogeography of their gyrodactylid monogenean lineages (Hahn et al., 2015).Monogenean parasites are regarded as ideal tags of host genealogy due to their short generations, direct life cycles and generally high host specificity (Nieberding & Olivieri, 2007;Pariselle et al., 2011).Yet, the comparative phylogeographic analysis of host and parasite data has never been applied to trace intraspecific cryptic invasions within the native ranges of species.
Previous studies on the morphology and genetics of X. laevis have identified several differentiated populations in South Africa that are segregated by large mountain ranges and different climatic regions, with the most notable divergence between northern and southern populations (de Busschere et al., 2016;du Preez et al., 2009;Furman et al., 2015;Grohovaz et al., 1996;Measey & Channing, 2003).In opposition to the clear phylogeographic structuring stands the widespread occurrence of both natural and human-mediated translocation of X. laevis in the region.This movement is a consequence of naturally occurring overland migration, the domestic export of the frog for teaching and learning and the widely adopted practice of the translocation of the frog for use as live bait in recreational angling (de Villiers & Measey, 2017;Measey, 2016;Measey et al., 2017;Measey & Davies, 2011;van Sittert & Measey, 2016;Weldon et al., 2007).
One would expect these translocations to leave a genetic trace (van Sittert & Measey, 2016).Yet, the cryptic invasion of X. laevis over long distances and across major geographic barriers in the native parasite dynamics in both native and invasive populations.Therefore, we recommend the inclusion of parasite data as a more holistic approach to the invasion ecology of animals on the intraspecific level.
We posit that the bladder polystomatid flatworm Protopolystoma xenopodis (Price, 1943) Bychowsky, 1957 (Monogenea: Polystomatidae) is the ideal tool to trace the hitherto unconfirmed contact between northern and southern lineages of its host X. laevis across its native range of southern Africa.In natural populations across sub-Saharan Africa, P. xenopodis, the most widely distributed of the species of Protopolystoma, is restricted to a single host species group, the leavis group, even in sympatry with other Xenopus species (Tinsley & Jackson, 1998a, 1998b).Furthermore, in P. xenopodis, notable intraspecific morphological variation exists between populations of the polystomatid infecting X. laevis in southern Africa and those infecting more northerly host species (Tinsley & Jackson, 1998b).This pattern is echoed even on a smaller regional scale in South Africa, where notable morphological and phylogenetic divergence exists between northern and southern populations of P.
xenopodis that parasitise X. laevis (Schoeman et al., 2022).The combination of high host specificity and intraspecific divergence suggests that P. xenopodis could be useful in a holobiont approach to a bio-invasion study, since any inferred movement of the parasite has to be mediated by movement of the host.
Therefore, in the present study, we use the widely distributed and intimately associated X. laevis and P. xenopodis system as a proof of concept that intraspecific parasite phylogeography can provide information on the cryptic introductions of their host lineages.We hypothesise that the phylogeography of P. xenopodis in South Africa will magnify that of its host X. laevis, which will allow us to employ a comparative phylogeographic analysis of this host and parasite in the context of their domestic translocation.We postulate that the parasite phylogeography will reveal contact between distinct lineages of X. laevis where evidence for translocation events might be lacking in the frog phylogeography and provide new insights on invasion pathways in the native range and beyond.

| Host and parasite sampling
In the native range of southern Africa, 46 adult X. laevis were cap- All frogs underwent double euthanasia with anaesthesia in 6% ethyl-3-aminobenzoate methanesulfonate (MS222) (Sigma-Aldrich Co., USA) followed by pithing as required by the institutional ethics committee.During subsequent dissection, 59 P. xenopodis (one per frog) were retrieved from the bladders for molecular analyses.
Polystomatids and liver tissue from their corresponding frog hosts were preserved in 70% ethanol.
For purification and sequencing, PCR products were sent to a commercial company (Inqaba Biotec, RSA) that used the ExoSAP protocol (New England Biolabs Ltd.) for purification and obtained the sequences with BigDye® Terminator version 3.1 Cycle Sequencing, utilising the corresponding primer pairs used in the PCR reaction, on an ABI3500XL analyzer.Sequences were assembled and manually edited in Geneious Prime version 2021.0.3 (www.geneious.com).The 236 obtained sequences were deposited in GenBank (accession numbers OP038566-OP38624, OP046050-OP046169).

| Calculation of sequence statistics
An investigation of sequence diversity is an exploratory analysis that can reveal the relative differentiation within and between geographic clusters.It can also give an indication of the markers' applicability to subsequent analyses.For these analyses, we obtained alignments of each of the four gene sets: COX1 or rrnS of the frog and COX1 or rrnS of the polystomatid.The Clustal Omega algorithm under default settings (Sievers & Higgins, 2017) implemented in Geneious Prime was used for the alignment of all four datasets.
For the protein-coding COX1 gene, translation alignment was implemented under the vertebrate mitochondrial genetic code (table 2) for the frog and the echinoderm and flatworm mitochondrial genetic code (table 9) for the polystomatid.Genetic diversity indices were calculated for all specimens combined and for geographic clusters in the host and parasite samples.
Number of haplotypes and polymorphic sites, haplotype diversity, nucleotide diversity and mean percentage of pairwise differences were calculated in Arlequin version 3.5 (Excoffier & Lischer, 2010).Three groups were defined based upon collection locality (see Appendix S2 for full details), namely specimens collected in South Africa to the southwest of the Great Escarpment, the major geographical barrier to dispersal in the region (SA1, SA2, SA3, SA4 and SA7; n = 30), to the northeast of the Great Escarpment (SA5 and SA6; n = 16), or in their non-native range in France (n = 12).The specimen from Portugal was not included as a separate group in the analyses because of small sample size.

| Phylogenetic analyses
The phylogenetic relationships between the different populations of both the host X. laevis and the parasite P. xenopodis, represented in phylogenetic trees derived from the COX1 and rrnS genes, F I G U R E 1 Localities in southern Africa where 46 adult Xenopus laevis were collected.Localities are classified according to the geographic extent of the native phylogeographic lineages of X. laevis (de Busschere et al., 2016).Key localities are named according to the closest town.Jonkershoek ( ) is the source locality of the majority of historic X. laevis exports.Mountain ranges are indicated with shaded relief and rainfall seasonality is represented by the wettest month of each 25 km 2 grid cell.Monthly mean precipitation data, obtained from the WorldClim 2.1 database at a spatial resolution of 2.5 arc-min (Fick & Hijmans, 2017), were used to identify the months with the highest historical mean precipitation from 1970 to 2000 for each grid cell.Elevation data were obtained from the Natural Earth database.The map was constructed in QGIS version 3.10.2-ACoruña (QGIS Development Team, 2018) with the Mercator projection.
formed the basis of the subsequent comparative phylogeographic analysis.
The sequences were prepared for the phylogenetic analyses through removal of duplicates, addition of outgroups for rooting, re-alignment and concatenation.Duplicate sequences were identified with the help of CD-HIT (Fu et al., 2012;Li & Godzik, 2006) implemented online (Huang et al., 2010).Only specimens with unique COX1 or rrnS gene haplotypes were retained for the analysis, leading to 30 unique sequence combinations (23 unique haplotypes of COX1 and 10 of rrnS) in the case of X. laevis and 44 (34 of COX1 and 35 of rrnS) in the case of P. xenopodis.In order to root the trees, the COX1 and rrnS gene sequences of the congeneric Müller's clawed frog X. muelleri (Peters, 1844) (GenBank accession numbers EU599031, AY581657; Evans et al., 2004Evans et al., , 2008) ) and its polystomatid P. occidentalis Tinsley et Jackson, 1998 (GenBank accession numbers KR856179, KR856121;Héritier et al., 2015) were added as outgroups.This polystomatid only infects species from the muelleri group, even in sympatry with X. laevis (Tinsley & Jackson, 1998b).
Furthermore, its host X. muelleri reportedly reproduces sterile offspring when hybridisation occurs between this species and its sympatric congener X. laevis (Fischer et al., 2000).Therefore, the selected outgroups are applicable to an intraspecific analysis of the X. laevis and P. xenopodis host-parasite system.The resulting datasets, after removal of duplicates and addition of the outgroups, were re-aligned by gene as detailed above.The two aligned genes of each species were concatenated by individual in Geneious Prime.
Ultimately, 31 sequences of a total of 1,153 bp (717 bp COX1 and 436 bp rrnS), containing 189 sites with gaps and 183 variable sites, of which 88 were parsimony informative, were used for the phylogenetic analysis of the frog dataset.Likewise, the 45 aligned sequences consisting of a total of 934 bp (423 bp COX1 and 511 bp rrnS), with 799 complete sites, of which 225 polymorphic sites with a subset of 174 parsimony informative sites, were used to infer the phylogeny of the parasite.
Prior to phylogenetic inference, the most optimal models of molecular evolution of each gene partition were determined.Initially, the concatenated alignments comprised four partitions each, namely the three codon positions of the partial COX1 gene and the rrnS gene.The ModelFinder selection routine (Kalyaanamoorthy et al., 2017), in conjunction with the FreeRate heterogeneity model (Soubrier et al., 2012), implemented in IQ-TREE version 2.1.2(Minh et al., 2020), was employed to select the optimal model of molecular evolution for each partition.To reveal the optimal partitioning scheme and most suitable evolutionary models for each alignment, the initial partitions were sequentially merged (Chernomor et al., 2016) until model fit, as evaluated by the Bayesian information criterion (BIC), ceased to improve.
Finally, we obtained phylogenies of the host and the parasite using the maximum likelihood (ML) approach.The ML phylogeny of the host was inferred under the TIM2 model (Posada, 2003) with edge-unlinked FreeRate heterogeneity models (Soubrier et al., 2012), proportions of invariant sites (Gu et al., 1995) and discrete gamma models (Yang, 1994) for all partitions.Similarly, the ML tree of the polystomatid was inferred under the TPM3 model (Kimura, 1981) with a FreeRate heterogeneity model (Soubrier et al., 2012) for all partitions.The analyses were performed in IQ-TREE version 2.1.2(Minh et al., 2020), with the assessment of branch support through ultrafast bootstrapping (UFboot2; Hoang et al., 2018) and the Shimodaira-Hasegawa approximate likelihood ratio test (SH-aLRT; Guindon et al., 2010), each with 10,000 replicates.The two phylogenies were rooted with respect to the outgroups in TreeGraph2 version 2.15.0-877 beta (Stöver & Müller, 2010).Clades with UFboot2 support lower than 95% or SH-aLRT lower than 80% were collapsed to polytomies in the same programme.

| Comparative phylogeographic analyses
The assessment of the global fit between the two phylogenies at the level of lineages provided a framework to interpret the mismatches or incongruencies represented by certain individual host-parasite links as indications of long distance movement and subsequent parasite switching to different host lineages.The global congruence of the two phylogenies was assessed by the implementation of both Procrustean Approach to Co-phylogeny (PACo) (Balbuena et al., 2013) and the fourth-corner statistical approach of ParaFit (Legendre et al., 2002).Both these methods consider the contribution of individual specimen interactions to phylogenetic congruence (Balbuena et al., 2013;Legendre et al., 2002;Poisot & Stouffer, 2015).Moreover, both methods can accommodate trees that are not fully resolved and support instances of uneven numbers of terminal nodes of hosts and parasites.
In PACo, we assessed the dependency of the parasite phylogeny upon that of the host through asymmetrical Procrustean superimposition of host and parasite distance matrices, while ParaFit tested the dependence of both phylogenies upon one another (Balbuena et al., 2013;Legendre et al., 2002).Prior to analyses, the outgroups were removed from the ML phylogenies.The phylogenies were converted to matrices of patristic distances by the package ape (Paradis & Schliep, 2019) in R version 4.0.2(R Core Team, 2020).These matrices served as input for both analyses, along with the corresponding host-parasite association matrix, where the polystomatids were assigned to their original host individuals, or to a host with an identical gene sequence to the original host individual.
The principal coordinates for PACo were derived with the r package paco (Hutchinson et al., 2017) version 0.4.2 for the Procrustes superimposition in conjunction with the Cailliez correction (Cailliez, 1983) to avoid the production of negative eigenvalues of non-Euclidean phylogenetic distances.The PACo analysis was performed, using the asymmetric Procrustes statistic to test for phylogenetic tracking by the parasite of its host, in paco with 10,000 permutations, with conserved row sums and number of interactions.The dynamics of the interacting clades were further explored by identifying those interactions that most contribute to the possible concordance between the two phylogenies through a jackknife procedure in paco.Those interactions with the smallest residual distances were interpreted as stronger supporters of overall congruence, while those with larger residual distances represented likely mismatches between host and parasite phylogeny.
The ParaFit analysis (Legendre et al., 2002) was carried out with ape based upon the patristic distance matrices, using 10,000 permutations with the Cailliez correction (Cailliez, 1983).Both the ParaFitLink tests were employed to assess the significance of the individual links (Legendre et al., 2002), which also pointed out hostparasite links that signify incongruencies.
Visualisation was achieved through a tanglegram of the host and parasite ML phylogenies with a weighted interaction network, based upon the residuals yielded by the PACo jackknife procedure.It was constructed with the r package phytools (Revell, 2012) with the function cophylo(), which has an internal algorithm that rotates nodes to optimally match tips.Due to the removal of duplicate sequences, certain host and parasite terminal nodes were involved in multiple interactions.
Haplotype genealogies based upon bifurcarting phylogenetic trees and Fitch (1970) distances between sequences are an alternative way of visualising phylogenies while giving an indication of the number of individual specimens represented by each unique sequence in the dataset.For this reason, we also visualised the 59 X. laevis and the 59 P. xenopodis, that is with the inclusion of duplicates, in two haplotype genealogies.The sequences were aligned and concatenated as detailed above.Two ML phylogenies, one for the frog and one for the parasites, based on all 59 sequences of each and no outgroups were derived using the procedure described above.These trees, along with the re-aligned sequences, served as input for Fitchi (Matschiner, 2015), which produced two haplotype genealogy graphs.

| Use of frogs by anglers
As an addition to the information provided by (Weldon et al., 2007) on the use of X. laevis as live bait by anglers in South Africa, we obtained anecdotal information on fishing practices from several stakeholders while conducting our fieldwork across the region.We asked about the practices surrounding the use of X. laevis as live bait during fishing and the translocation and release of X. laevis during fishing activities.Nine recreational and subsistence anglers in the Eastern Cape, Limpopo, Northwest, KwaZulu-Natal and Gauteng Provinces provided information on the use of X. laevis as live bait by their communities in the vicinity of some of our sampling localities.

| Sequence diversity
The genetic diversity of the two parasite genes was generally higher than the two orthologous host genes, with greater genetic diversity among the samples collected in South Africa than in France in both host and parasite (Table 1).Genetic diversity was generally higher in the hosts collected to the southwest of the Great Escarpment than in those collected from the northeast, which was not the case for the parasites (Table 1).

| Phylogenetic relationships of host and parasite
The phylogenetic analyses confirmed marked phylogeographic structuring in both the host and parasite in their native range, which provided the necessary resolution to assess spatial patterns.Tree reconstruction showed that frogs from the northeastern SA5 and SA6 localities in the native range form a well-supported clade, along with some of the frogs from the SA4 locality (clade NE in Figure 2 with UFboot2 = 95%/SH-aLRT = 95.8%).In addition, two clades with unresolved relationships to one another and to NE could be identified among sampled frogs from the southwestern localities in the native range of X. laevis (clades SW coast with UFboot2 = 81%/SH-aLRT = 86.1% and SW inland with UFboot2 = 98%/SH-aLRT = 99.7% in Figure 2).On the whole, members of the first clade originate from localities closer to the coastline (clade SW coast in Figure 2

| Congruence of host and parasite phylogeographies
The phylogeny of X. laevis exhibited significant overall congru- The haplotype genealogies of the frog host and the polystomatid parasite (Figure 3) unsurprisingly demonstrated the same phylogeographic structuring as those revealed by the phylogenetic analysis of X. laevis (Figure 2).The frog haplotype genealogy clearly revealed that all northeastern specimens clustered together, except for frog TA B L E 1 Genetic diversity of the bladder flatworm Protopolystoma xenopodis and its frog host Xenopus laevis, grouped according to sampling locality-collected in South Africa to the southwest or to the northeast of the great escarpment, or in France.Indices were inferred separately by gene for each group and for all samples combined from the aligned partial mitochondrial COX1 and rrnS genes

| Translocation of frogs by anglers
Our informal discussions with anglers confirmed the widespread use of X. laevis as live bait in five of the nine South African provinces, with all nine anglers confirming the existence of the practice in their com-

| DISCUSS ION
In this study, the widespread, yet previously hidden, domestic expansion of X. laevis lineages in South Africa is brought to light through a comparison with the phylogeography of its coinvasive parasite, P. xenopodis.The use of parasite data to reveal host connectivity is hardly novel (reviewed in Gagne et al., 2021).Yet, the present study is the first application of host-parasite phylogeographic analyses to the detection of intraspecific cryptic invasions.Furthermore, it demonstrates that even a comparatively modest dataset of less than 60 host-parasite pairs can reveal compelling evidence for historic translocation.

| Parasite clustering mirrors the intraspecific phylogeography of the host
If the phylogeographic structuring of a parasite mirrors that of its host, a phenomenon which is not necessarily a consequence of coevolutionary processes (de Vienne et al., 2013), the genealogy of the parasite can provide insight on the translocation of the host (Nieberding & Olivieri, 2007).Based on our findings, there is phylogenetic congruence between the African Clawed Frog and its monogenean parasite.
This means that the observed phylogeographic structuring in the parasite, which is supported by both molecular and morphological evidence (Schoeman et al., 2022), can corroborate and expound upon the phylogeography of the host X. laevis, which also shows notable phylogeographic structuring across its native range (de Busschere et al., 2016;du Preez et al., 2009;Furman et al., 2015;Grohovaz et al., 1996;Measey & Channing, 2003).After a brief free-swimming stage, P. xenopodis completes the rest of its life cycle in the excretory system of X. laevis (Theunissen et al., 2014), which translates to a very specific and intimate host-parasite association that can faithfully portray the phylogeographic structure of the frog.However, the parasite dataset provides an independent line of evidence and not a mere echo of the phylogeographic structuring of the frog, since each parasite represents a separate infection event from a pool of free-swimming larvae.
The genetically distinct host-parasite clusters display distributions that roughly follow the two climatic regimes in the region, namely the summer rainfall region to the northeast and the winter rainfall region to the southwest, with the plateau edge of the Great Escarpment and the many ridges of the Cape Fold Mountains as assumed geographical barriers to natural dispersal (Furman et al., 2015).
The more pronounced clustering among hosts and parasites and the higher haplotype diversity in hosts that hail from the southwest of the Great Escarpment is likely brought about by the presence of the many ridges of the Cape Fold Mountains (de Busschere et al., 2016;du Preez et al., 2009;Furman et al., 2015;Grohovaz et al., 1996;Measey & Channing, 2003).On the other hand, well-documented human-mediated translocation of X. laevis in the area to the south-

| Parasite spillover is an indication of contact between divergent host lineages
Parasite phylogeography offers a tool to recover historical insights of host dispersal pathways, since host lineage switches reflect past interactions between host organisms at various time-scales (e.g.Barbosa et al., 2012;Criscione et al., 2006;Galbreath et al., 2020;Galbreath & Hoberg, 2015;Nieberding et al., 2004).The usefulness of parasite species in reflecting the introduction pathways of their hosts has been demonstrated for fish hosts and their monogenean parasites on a few occasions (Geraerts et al., 2022;Huyse et al., 2015;Jorissen et al., 2022;Kmentová et al., 2019;Ondračková et al., 2012;Reyda et al., 2020).However, these studies focused on species-level invasions from the native to the non-native range, rather than intraspecific invasions of different lineages within the native range.Our study, on the other hand, demonstrates the use of parasites as tags to trace intraspecific cryptic invasions.Multiple instances of host lineage switches revealed by the phylogeographic analyses of P. xenopodis act as an indicator for historic contact between distinct X. laevis lineages.This provides us with the means of recognising such contact even in cases where the locally non-native lineage of X. laevis may have gone extinct after the introduction event or where they are not detected at a certain sampling effort.
In fact, thus far, genetic evidence for the admixture of the lineages of X. laevis with native ranges on opposite sides of the Great Escarpment has not been found, despite historical records pointing towards large scale exports from the southwest to the rest of southern Africa (van Sittert & Measey, 2016;Weldon et al., 2007).In our study, the host phylogeographic analyses suggest the presence of contact zones between the northeastern (NE) and southwestern lineages (SW inland and SW coast ) at two localities to the northeast of the Great Escarpment where these lineages co-occur.
It is this sympatry of multiple host clades that makes host lineage switching by parasite lineages a possibility.In turn, the observation of host switches from one lineage to another provides more evidence for the translocation of X. laevis across southern Africa.
Indeed, our study confirmed the presence of southwestern parasite lineages (sw1, sw2 and sw3) at seven localities to the northeast of the Great Escarpment, where one would expect the northeastern parasite lineage (ne) to occur.In these instances, southwestern parasite lineages that are translocated along with their respective host lineages, thus locally non-native at these new localities, spill over to the locally native northwestern host lineage when these divergent, previously allopatric host lineages come into contact (Galbreath & Hoberg, 2015).
Differential survival between locally native and non-native X.
laevis lineages in southern Africa can partially explain the fewer observed instances of establishment by non-native X. laevis lineages when compared to its cointroduced polystomatid.In translocation experiments of the northern and southern lineages in southern Africa, X. laevis tadpoles and metamorphs exhibited remarkable phenotypic plasticity when translocated between rainfall regimes at the cost of long-term survival (Kruger, 2022).Under these circumstances, the locally native X. laevis lineage would outcompete the non-native lineage, preventing successful local invasion.Thus, the employment of host phylogeography alone to trace long range translocation of X. laevis in southern Africa would mask a number of introduction events, as our findings clearly demonstrate.
Upon arrival in the new area, the cointroduced P. xenopodis lineage may rapidly spread in a population of locally native and therefore naive X. laevis.It has been shown that primary infections by P.
xenopodis in previously uninfected X. laevis elicit strong long-term immunological responses against secondary infections by the same parasite species (Jackson & Tinsley, 2001).Correspondingly, a crossinfectivity experiment between northeastern and southwestern lineages of X. laevis and P. xenopodis further suggested that lineages of X. laevis in their native ranges might be more susceptible to infection by the locally non-native lineage of P. xenopodis than to their corresponding locally native parasite lineage (Jackson & Tinsley, 2005).
In turn, this effect will boost the nascent spread of non-native P.
xenopodis lineages in the locally native lineage of X. laevis.
There are two possible explanations for the observed contact between divergent lineages of this host-parasite system in its na- It is clear from the long branch lengths between native southwestern polystomatids in the haplotype genealogy that there is considerable differentiation among the southwestern populations of P.
xenopodis.It is reasonable to assume that greater sampling effort will yield more southwestern haplotypes that could be very similar to the translocated southwestern haplotypes now found at northeastern localities.Furthermore, it is worth noting that the number of domestic exports from the southwestern region of South Africa greatly dwarfs the number of the international exports (van Sittert & Measey, 2016), which translates to more introduction events of southwestern X. laevis to areas across South Africa, thus greater haplotype diversity among the domestically introduced frogs, than to France.
However, there is another reason affecting the plausibility of the scenario of earlier natural expansion of this host-parasite system.
Of course, hosts and parasites can experience shared histories of geographic expansion and contact between isolated populations in response to historical environmental shifts, such as climate change (Galbreath & Hoberg, 2015;Hahn et al., 2015;Huyse et al., 2017;Kmentová et al., 2018).Likewise, postexpansion, the non-native host lineages may be rapidly lost from the population due to genetic drift or differential survival, while the non-native parasite lineages may persist and become widely distributed across the geographic range of its new host lineage, as has been demonstrated before on the species level for tapeworm parasites in Beringia (Galbreath et al., 2020).
Regardless, the formation of the Great Escarpment around 130 million years ago (Burke & Gunnel, 2008) predates the evolution of X. laevis over the last 15 million years (Furman et al., 2015).The Great Escarpment is a dominant barrier to dispersal that has played a defining role in shaping the distribution and evolution of southern African fauna (e.g.Barlow et al., 2013;Makokha et al., 2007;Mynhardt et al., 2015;Nielsen et al., 2018;Predel et al., 2012), not least among which X. laevis (Furman et al., 2015).When considering the evidence from the phylogeographic structuring of insects, frogs, lizards, snakes and small mammals in southern Africa, where the Great Escarpment is a definite barrier to natural dispersal (Barlow et al., 2013;Furman et al., 2015;Makokha et al., 2007;Mynhardt et al., 2015;Nielsen et al., 2018;Predel et al., 2012), it seems highly unlikely that climatic fluctuations could have given rise to natural expansion of X. laevis beyond the plateau edge of the escarpment that may shape the distribution of the southwestern lineages of P.
xenopodis to the current day.
Rather, we consider recent human-mediated translocation a more plausible explanation.In fact, we know that the bulk of exports of X. laevis since the 1930s from the southwest of Southern Africa, which amounts to over 400,000 animals, was sent to domestic destinations, especially to urban centres inland where X. laevis was used in research and teaching (van Sittert & Measey, 2016;Weldon et al., 2007).Subsequent escapees from these centres of higher learning would lead to the presence of locally non-native X. laevis and P. xenopodis near these urban areas.Moreover, these escapees can spread even further through the ability of X. laevis to move overland, a mode of dispersal which is complemented by expansion via networks of artificial waterways (de Villiers & Measey, 2017; Measey, 2016).The rural locations of the parasite spillover events revealed by our analyses of polystomatid phylogeography suggest that smaller scale human-mediated dispersal events also have a role to play in the dispersal of X. laevis.The role of recreational fishing across South Africa in the dispersal of X. laevis, and by extension, P. xenopodis, has been recognised before (Weldon et al., 2007).
Previously documented interviews with local anglers and suppliers exposed the use of X. laevis, frequently bought from pet shops and local breeders, as live bait for predominantly catfish angling (Weldon et al., 2007).Our own informal discussions with local anglers at the localities where we collected X. laevis affirm the widespread use of this frog, sometimes obtained from pet shops, as live bait for the capture of both black bass (Micropterus spp.) and catfish Clarias gariepinus.
In terms of sheer numbers, the impact of these translocation events on the domestic dispersal of X. laevis has been deemed negligible when compared to exports from Jonkershoek, the most prolific exporter, and other official suppliers to the domestic market (van Sittert & Measey, 2016; Weldon et al., 2007).Nonetheless, previous studies on other aquatic species have pointed out that the use of live bait in angling is an important pathway for introductions, even if the market might be small in economic terms (Kalous et al., 2013;Kilian et al., 2012).Therefore, the impact of anglers on the spread of locally non-native X. laevis lineages to more remote areas, as opposed to the bigger centra that received the documented official consignments, should not be underestimated.Our informal interviews identified a myriad of ways for X. laevis to be released in new environments by anglers, from escapees en route, to the deliberate release of surplus bait, to the intentional stocking of water bodies for future use.
Therefore, the combination of overland movement, dispersal along natural and artificial waterways and translocation by the angling community could form a widespread undocumented dispersal network of X. laevis in South Africa.

| Parasite phylogeography as an underutilised tool in invasion biology
As shown above, the present study acts as a proof of concept for the potential of the comparative phylogeographic approach to the clarification of patterns in intraspecific host divergence and dispersal.In turn, this independent line of evidence can provide additional insight on invasion pathways and inform management strategies.Here, we highlight some of the potential implications to demonstrate the broader applicability of this integrative approach.
Our approach was able to provide a new perspective on the origin of the relatively high haplotype diversity of the invasive population of X. laevis in France, recovered our analysis and others, when compared to other invasive populations in Portugal, Sicily and Chile (de Busschere et al., 2016;Lillo et al., 2013;Lobos et al., 2014).The idea of multiple introduction events from various source populations in South Africa has been offered before as the most parsimonious explanation for this genetic diversity in the invasive population of X.  Measey, 2016;Weldon et al., 2007).
The novel insights provided by the addition of the parasite dataset have implications for both the conservation of X. laevis and its host-specific parasites.The widespread anthropogenic translocation of this host-parasite system disrupts interactions between X. laevis lineages and the P. xenopodis lineages that tracked their phylogenetic divergence, ancient interactions that deserve to be targets of conservation (Gómez & Nichols, 2013).Subsequent host switches between closely related lineages set the stage for new diversity to arise and for novel ecological interactions to be established (Galbreath et al., 2020;Sures et al., 2017).In the case of host-specific parasites, translocation of lineages within the native range of a species may actually be of more concern than translocation to far-flung non-native ecosystems, since these ecosystems will probably not harbour close relatives of the invasive host to facilitate host switches (Blackburn & Ewen, 2017).Host-parasite interactions are also something to keep in mind when conducting conservation re-introduction programmes, where captive-bred or locally non-native individuals of a species are used to re-stock locally extinct populations of endangered species (Britt et al., 2004).Our results highlight that the evaluation of the genetic and parasitological suitability of a source population for translocation should also encompass the potential history of local introductions that may have influenced the intraspecific parasitism of the host population (Jørgensen, 2015;Northover et al., 2018).
This approach further holds great promise in the management of invasive species.Prior to invasion, translocations in the native range may have already altered the source populations of invasive species, as is the case of X. laevis.We need to take altered phylogeographic structure into account to adequately monitor the course of an invasion or the novel interactions that may arise.Parasites as passengers or invaders themselves easily slip under the radar in invasion biology (Blackburn & Ewen, 2017;Solarz & Najberek, 2017).Yet, the role of parasites on the move in emerging infectious diseases and habitat transformation via indirect density-mediated effects cannot be overemphasised (Amundsen et al., 2013;Bobadilla-Suarez et al., 2017;Daszak, 2000;Dunn et al., 2012;Goedknegt et al., 2016;Lymbery et al., 2014;Roy & Lawson Handley, 2012;Sures et al., 2017).
Furthermore, the identification of the local angling community as a player in the translocation of X. laevis in the native range allows policy makers to broaden their intervention efforts in the management of X. laevis as a domestic exotic.

| Recommendations to make parasite datasets work for invasion biologists
All in all, our findings agree with the necessity of a more holistic approach to data collection in invasion biology (Clavero et al., 2016), especially with regard to parasites.The idea of the 'holistic specimen' has recently gained some traction (Cook et al., 2017;Galbreath et al., 2019).This approach to the collection and study of organisms advocates for the integrated specimen-based studies of animals and their associated parasites and pathogens.The optimisation of sampling effort is crucial to improve the allocation of resources, both in terms of funding and specimen use, especially when working with endangered species or specimens that require considerable effort to obtain, or in view of other ethical concerns regarding invasive sampling.
The inclusion of parasite data during the data collection phase also ensures that the host and parasite datasets are comprised of directly comparable individuals which can improve the feasibility and value of comparative phylogeographic analyses.For example, the nonoverlap between host and parasite datasets was cited as a limitation in the cophylogeographic analyses conducted by both Barbosa et al. (2012) and Huyse et al. (2017).Overlap between datasets can be established through the sampling of host-parasite pairs from the same localities and time frames, while sampling of the corresponding parasite of each host specimen will not contribute to dataset overlap, especially in the case of parasites with a free-living stage.Even in cases where parasites are not the targets of the current project, the deposition of voucher material of both host and parasite from the sampling event into curated institutional collections, in other words overlapping datasets, can be vital for later studies (Galbreath et al., 2019;Thompson et al., 2021).Later on, DNA sequences could be linked to material deposited in these museum collections.
Similarly, another aspect to keep in mind is the choice of markers for our cophylogeographic analyses.More often than not, investigators rely on markers for which primers are already available and entire analyses may even be based on a single marker (de Vienne et al., 2013;Gutiérrez-García & Vásquez-Domínguez, 2011).Yet, successful comparative phylogeographic analyses rely on a combination of multiple markers with a sufficient degree of genetic divergence evident among populations (de Vienne et al., 2013;Gutiérrez-García & Vásquez-Domínguez, 2011), which is often higher in the parasite than the host (Nieberding & Olivieri, 2007).Thus, the independent yet comparable DNA signature provided by the parasite dataset may provide higher resolution than the host dataset, where spatial patterns may be blurred by the recent mixing of previously isolated genotypes in invasive populations (Rius & Turon, 2020).As a next step to improve confidence in phylogeographic inferences, we also propose the addition of morphological data from the parasite specimens, as demonstrated by several recent studies on monogenean parasites (Huyse et al., 2015;Kmentová et al., 2019;Ondračková et al., 2012;Schoeman et al., 2022).
Ultimately, this study serves as a proof of concept that the use of parasite datasets will add value to investigations in invasion biology, especially in the case of intraspecific cryptic invasions.We show that the utility of parasite datasets can go beyond the provision of supporting evidence and provide greater resolution that may reveal patterns that went undetected in the host dataset.These findings should provide an impetus for more comprehensive investigations with a wider geographic scope conducted over longer time periods to systematically assess the position of parasitology in bio-invasion research.In this, our study echoes the sentiments of Hoberg et al. (2015), who argued for an integrated approach to parasitology as a discipline that holds the keys to understanding and managing global change in the Anthropocene.
tured from March 2017 to April 2019 in liver baited funnel traps from 30 localities (Figure 1; see Appendix S1 for locality details).The localities were representative of the phylogeography of the frog, based upon the lineage assignment of de Busschere et al. (2016) (see Appendix S2 for locality classification).In the invasive range, 13 adult X. laevis were collected from 12 localities in western France and one in Portugal during the summers of 2017, 2018 and 2019 (see Appendix S1 for locality details).In France, frogs were captured in baited traps by members of the EU LIFE CROAA team as part of the regional eradication programme for invasive amphibians.Frogs in Portugal were obtained from the locally captured laboratory stock of the University of Lisbon.
tomatids.The DNA was extracted from host and parasite tissue with the PCRBIO Rapid Extract PCR kit (PCR Biosystems Ltd., UK) or the Quick-DNATM™ Miniprep Plus kit (Zymo Research, USA).Subsequently, amplification reactions for the mitochondrial COX1 and rrnS genes of both frogs and polystomatids were prepared with 2-5 μl genomic DNA, 1.25 μl forward primer [0.2 μm], 1.25 μl reverse primer [0.2 μm], 12.5 μl master mix and PCR-grade water to the final volume of 25 μl.The 2× PCRBIO HS Taq Mix Red (PCR Biosystems Ltd., UK) or the OneTaq® 2× Master Mix with Standard Buffer (New England Biolabs Inc., USA) were used with the temperature of the elongation steps as per supplier instructions (68 or 72°C).
Figure 2 with UFboot2 = 53%/SH-alrt = 90.7%)and another clade of polystomatids from southwestern SA2 and SA3 localities and some northeastern SA5 and SA6 localities (clade sw3 in Figure 2 with UFboot2 = 61%/SH-aLRT = 86.6%).Another well-supported clade, the sister clade to ne, contained polystomatids from SA1, France and Portugal (clade sw4 in Figure 2 with UFboot2 = 100%/ SH-aLRT = 100%).However, overall low support values of many internal nodes in the parasite ML tree, especially the nodes basal to the sw3, sw4 and ne lineages, hampered the interpretation of phylogenetic clustering among some of the polystomatids collected from ence with that of P. xenopodis (PACo m 2 xy = 0.1034, p = 0.0018, n = 10,000; ParaFitGlobal = 0.0202, p < 0.0001, n = 10,000), with the PACo analysis suggesting that the phylogeny of P. xenopodis tracked that of X. laevis.According to PACo, links between frogs from SW coast and their corresponding polystomatids that fall within sw1 in the parasite phylogeny, generally contributed the most to the overall phylogenetic congruence.The establishment of overall congruence between these two phylogenies allowed us to hone in on mismatches between the two phylogenies, which signified past contact between divergent clades.Based upon visual inspection of the tanglegram (Figure 2), a few links represented mismatches between the host and parasite phylogenies that could be explained by long distance translocation of frogs from the SW host lineages to the northeastern localities followed by spill-over of sw1, sw2 and sw3 polystomatids to NE hosts.These indications of host switching were identified in frog-polystomatid pairs from SA5 and SA6 localities (Dullstroom, Hopetown, Johannesburg, Verkykerskop, Adelaide, Ugie and Wakefield), where NE hosts harboured parasites that clustered with their southwestern sw1, sw2 and sw4 counterparts.The PACo jackknife procedure further identified the interactions between frog-polystomatid pairs from the invasive range in France and Portugal as links conveying low support to congruence.Moreover, interactions between frogs and polystomatids from the contact zone near Three Sisters (SA4) and Wakefield (SA6) demonstrate low support to overall congruence.Finally, the ParaFitLink tests reported 30 of 48 links as significant at a level of 0.05.Out of the 18 links that did not significantly support congruence, 11 signified host switches by parasites to another host lineage that occurred in the invasive range of France and Portugal and at SA4, SA5 and SA6 localities in the native range (Johannesburg, Three Sisters, Ugie and Wakefield).
munities and surrounding areas.The frog was usually employed in the capture of black bass Micropterus Lacepède, 1802 and African sharptoothed catfish Clarias gariepinus(Burchell, 1822).Black bass are non-native inhabitants of water bodies across South Africa(Hargrove et al., 2015;Khosa et al., 2019) and African sharp-toothed catfish are non-native to all areas to the south of the Great Escarpment(Weyl et al., 2016).Seven of the nine anglers indicated that specimens of X. laevis are generally caught and used locally, with short range translocation of the captured specimens.However, two of the anglers F I G U R E 2 The interaction network and phylogenies of the frog Xenopus laevis (left) and its polystomatid Protopolystoma xenopodis (right) with X. muelleri and P. occidentalis as outgroups.Topologies are based on maximum likelihood analyses of concatenated COX1 and rrnS gene alignments of 31 frogs and 45 polystomatids with low support clades (UFboot2 < 95% or SH-alrt < 80%) collapsed.Links indicate observed host-parasite associations, weighted according to the contribution of each interaction to the overall phylogenetic congruence with thicker lines corresponding to greater support.Colours refer to geographic origin to the southwest (SW) and northeast (NE) of the great escarpment, following the classification of de Busschere et al. (2016): SA1 (dark blue), SA2 (light blue), SA3 (red), SA4 (green), SA5 (yellow), SA6 (brown) and SA7 (purple) (see Figure1) with main lineages named and highlighted.Some parasite specimens from southwestern SA1, SA2, SA3, SA4 and SA7 localities are not assigned to any of the main lineages due to their paraphyletic grouping basal to lineages sw3, sw4 and ne.Invasive X. laevis and cointroduced P. xenopodis from France (grey) and Portugal (pink) are also included.Key localities are Adelaide (A), Dullstroom (D), Hopetown (H), Johannesburg (J), Ugie (U), Verkykerskop (V) and Wakefield (W).Jonkershoek ( ) is the origin of the majority of X. laevis exports.Scale bars indicate 0.1 nucleotide substitutions per site X. laevis from breeders and pet shops in urban areas (Potchefstroom and Johannesburg) and transporting them to more rural areas for use as live bait.These fishermen were active near some of our sampling localities, namely Potchefstroom, Johannesburg, Dullstroom and Verkykerskop.All fishermen confirmed that surplus live X. laevis are routinely released on site and that escapees are common.Notably, a fisherman in Ugie described the practice of acquiring live X. laevis from water bodies in the vicinity and translocating specimens to stock local dams for future use.

F
Unrooted haplotype genealogy graphs of the partial COX1 and rrnS genes of (a) 59 Xenopus laevis specimens and (b) their corresponding 59 Protopolystoma xenopodis parasites.For frogs and their polystomatids collected in the native southern Africa (see inset map), colours refer to the phylogeographic lineages of the host (de Busschere et al., 2016).Non-native X. laevis and P. xenopodis are from France (grey) and Portugal (pink).Highlighted haplotype groups correspond to the clades revealed by the phylogenetic analyses (Figure2): Northeastern host (NE) and parasite (ne) and southwestern host (SW) and parasite (sw) clades.Key localities with host lineage switches are named, as well as Jonkershoek, the source locality of most of the X. laevis exports.Sizes of circles are proportional to haplotype frequencies (see scale).The number of mutational steps is marked on the connecting branches with dots.Dashed lines are extensions of dotted sections for improved visualisation.
of the Great Escarpment(van Sittert & Measey, 2016;Weldon et al., 2007), in conjunction with widespread overland migration of the frog with the help of the many interconnected artificial water bodies in the region (de Villiers& Measey, 2017;Measey, 2016), ensure some gene flow-and exchange of parasites-between the divergent frog populations in this area.
tive range: natural geographic expansion over the evolutionary history of the host-parasite system in response to past environmental fluctuations, or more recent human-mediated translocation of the host along with its corresponding parasite lineage.At first glance, it is noticeable that the non-native P. xenopodis specimens collected at northeastern localities present with haplotypes that were generally highly differentiated from their counterparts from the southwest, despite belonging to the same sw lineages.In contrast, the haplotypes of the cointroduced French and Portuguese polystomatids are identical or nearly identical to the source South African haplotypes, indicating low divergence after the introduction event.The apparently deep divergence between native southwestern polystomatids and domestically translocated southwestern polystomatids suggests a scenario where parasites that underwent natural geographic expansion in deep time persist to the present day, yet with plenty of time to accrue genetic polymorphisms that differentiate them from their close relatives in the southwest.However, at our level of sampling effort, such comparisons of haplotype diversity are premature.