Genetic diversity and parasite facilitated establishment of the invasive signal crayfish (Pacifastacus leniusculus) in Great Britain

Abstract Successful establishment of non‐native species is strongly influenced, among other factors, by the genetic variation of founding populations, which can be enhanced by multiple introductions through admixture. Coexisting pathogens can also facilitate the establishment of non‐native species by detrimentally impacting on the native fauna acting as novel weapons. The signal crayfish (Pacifastacus leniusculus) is a highly invasive species, which has caused mass declines of native crayfish in Europe through displacement and transmission of the oomycete Aphanomyces astaci (crayfish plague), which is typically lethal to native European crayfish. However, whether Aphanomyces astaci may have facilitated the invasion of the signal crayfish is not known. We estimated the genetic diversity at microsatellite DNA loci, effective population size, and potential origins of seven infected and noninfected signal crayfish populations in Europe and one founder population in North America. Approximate Bayesian computation analysis and population structuring suggested multiple host introductions from diverse source populations, as well as higher heterozygosity among infected than uninfected populations, which could reflect a fitness advantage. Low effective population size, moderate heterozygosity, and lack of isolation by distance suggest that some invasive signal crayfish populations may not be fully established or that their genetic diversity may have been reduced by eradication attempts.

In both cases, these novel pathogens threaten the survival of phylogenetically close indigenous species, like the red squirrel (Sciurus vulgaris) and two native crayfishes, the white-clawed crayfish, Austropotamobius pallipes and the noble crayfish, Astacus astacus, respectively. Yet, the extent to which novel pathogens drive invasion success is controversial (Blackburn & Ewen, 2017) and understanding why some species become established while others fail to do so remains a key question in invasion biology (Davis, 2009).
Signal crayfish appears to have dispersed rapidly across Great Britain over the last 30 years (James, Slater, & Cable, 2014;James et al., 2017) and carry a particularly virulent strain of A. astaci, which has caused mass mortalities of A. pallipes in several European populations (Collas et al., 2016;Grandjean et al., 2014Grandjean et al., , 2017. However, the extent to which its current distribution has been facilitated by multiple introductions (Filipová, Holdich, Lesobre, Grandjean, & Petrusek, 2009) and/or by the presence of A. astaci is unclear. In addition, some populations have been subjected to control measures, mainly through the mechanical removal of thousands of crayfish, but the impact of these control F I G U R E 1 Map of UK sampling sites for Pacifastacus leniusculus, infection status, and the three highly significant breaks in genetic continuity generated by BARRIER in relation to sample sites (1 = Sirhowy, 2 = Lugg, 3 = Dderw, 4 = Lea, 5 = Bachowey, 6 = Mochdre, 7 = Gavenny, 8 = Pant-y-Llyn) measures is difficult to assess (Freeman, Turnbull, Yeomans, & Bean, 2010). Here, we compared the genetic diversity, effective population size, and potential origin of seven signal crayfish populations with different plague infection status and assessed the relative roles of the crayfish plague and multiple introductions in the establishment and dispersal of invasive crayfish in Great Britain.

| Study sites and sample collection
American signal crayfish were collected using baited crayfish traps (checked every 24 hrs) and hand netting (James et al., 2017), from five sites in Wales (Sirhowy, Dderw, Bachowey, Mochdre, and Gavenny) and two sites in England (Lugg and Lea) between May and September 2014 and one site (Pant-y-Llyn) in 2016 ( Figure 1; Table 1).
In addition, 30 crayfish were collected from a native population with unknown infection status in Oregon (US) as a reference for genetic diversity. The crayfish plague pathogen had not been detected at sites 1 (Sirhowy), 2 (Lugg) and 3 (Dderw), but had been isolated from crayfish at the remaining sites (James et al., 2017). Crayfish were collected under NRW Permits NT/CW065-C-652/5706/01 and NT/ CW081-B-797/3888/02.

| DNA extraction and amplification
Total genomic DNA was extracted from each crayfish from a section of tail fan, soft abdominal cuticle and walking leg tissue using the DNeasy Tissue Kit (Qiagen, Sussex, UK) following the manufacturer's instructions (James et al., 2017). A total of 214 crayfish were analyzed using nine microsatellites (Table 2), in three separate multiplex reactions (Azuma, Usio, Korenaga, Koizum, & Takamura, 2011;Froufe et al., 2015). Extracted DNA was analyzed for quantity and quality using a Nanodrop 2000 (Thermo Fisher Scientific Inc., USA) and approximately 8 μg were used for amplification using the Qiagen Multiplex PCR kit, following the Qiagen multiplex reaction protocol (Qiagen) in a total volume of 12 μl. Each reaction consisted of the concentrations of primers detailed in (Froufe et al., 2015; Supporting InformationTable S1), with the exception of Scop31 (forward and reverse), which was reoptimized at 1 μM.
Amplification conditions consisted of a single-cycle initial activation step of 15 min at 95°C followed by a touchdown PCR of eight cycles with a 30 s denaturation step at 94°C, a 90 s annealing step starting at 64°C and descending in 2-cycle steps of 2°C (64, 62, 60, 58 and 56°C) and 90 s of extension at 72°C. Twenty-four additional cycles of PCR were then run as above at an annealing temperature of 56°C followed by a single final extension cycle of 30 min at 60°C. Microsatellites were resolved on an Applied Biosystems ABI3130xl Genetic Analyser (Applied Biosystems, Sussex, UK), and fragment length was determined using the GeneScan 500-LIZ size standard and scored using GeneMapper v45.0 (Applied Biosystems). MICRO-CHECKER v2.2.3 (Van Oosterhout, William, Hutchinson, Wills, & Shipley, 2004) was used to assess presence of null alleles, large allele drop-outs and scoring errors due to stuttering. GENALEX v6.5 (Peakall & Smouse, 2006) was used to estimate effective number of alleles (N EF ) and the populations' expected (H E ) and observed heterozygosities (H O ) respectively. Deviations from Hardy-Weinberg equilibrium and tests for linkage disequilibrium were investigated using GENEPOP online v4.0.10 (Rousset, 2008).

| Genetic analyses
Pairwise F ST values and heterozygosity per locus were calculated using FSTAT v1.2 (Goudet, 1995). Analysis of Molecular Variance (AMOVA) among populations, among individuals and within individuals was calculated in ARLEQUIN v3.1 (Excoffier, Laval, & Schneider, 2005). Homozygosity by locus (H L ), which weighs the contribution of each locus to the homozygosity index depending on their allelic variability, was estimated for each individual in Cernicalin v1.0 (Aparicio, Ortego, & Cordero, 2006). Effective population size was estimated using N e Estimator v2.01 (Do et al., 2014)  . The number of clusters tested ranged from K = 1 to 9, with 20 repetitions for each K value, and 60,000 MCMC steps discarding the first 10,000 as burn-in (Pritchard et al., 2000).
The best fitting K value was estimated using StructureSelector (Li & Liu, 2017), which utilizes four alternative statistics (MEDMEDK, MEDMEAK, MAXMEDK, and MAXMEAK) to produce more accurate results for populations with uneven sample size. BARRIER v2.2 (Manni, Guérard, & Heyer, 2004) was used to detect discontinuities in allelic frequencies between British crayfish populations based on genetic distance and geographical distance values using the Monmonier's maximum difference algorithm (Monmonier, 1973). Initially one data matrix containing pairwise F ST values was imported in BARRIER to detect genetic barriers across all populations.
Eight data matrices were then imported into BARRIER containing pairwise F ST values per locus to assess the number of loci supporting each barrier and test for barrier robustness' (Manni et al., 2004).
The most likely scenario of colonization for UK populations was estimated using and Approximate Bayesian Computation approach implemented in the software DIYABC v2.1.0 (Cornuet et al., 2014).
For this analysis the Lea, Mochdre, and Gavenny populations were grouped into one genetic group (pool 1) based on F ST values and similarity of genetic clusters from the STRUCTURE analysis and the remaining populations were analyzed as separate populations.
Three scenarios of colonization were tested ( Figure 3): Scenario 1 -simultaneous divergence (null hypothesis), Scenario 2 -simultaneous divergence of Sirhowy, pool 1, Lugg, Bachowey and Dderw followed by divergence of Pant-y-Llyn from Bachowey, Scenario 3 -simultaneous divergence of Sirhowy, pool 1, Lugg, Bachowey and Dderw followed by admixture of pool 1 with Bachowey to produce the Pant-y-Llyn population. Default settings were used for mutation rates (generalized stepwise mutation model (Estoup, Jarne, & Cornuet, 2002) with a uniform prior distribution of mean mutation rate between 10 −4 and 10 −3 , priors were set uniformly distributed, prior distribution of individual locus mutation rates were set between 10 −5 and 10 −2 following a Gamma distribution with mean determined by the mean mutation rate across loci. Effective population sizes were set between 10 and 2,500 for all populations. A total of 1,000,000 simulations per scenario (1,2,3) were generated from the parameters prior distributions. Mean gene diversity across loci and mean M index diversity across loci (one sample summary statistics) were calculated for each population. Pre-evaluation of each scenario was carried out by generating Principal Component Analysis (PCA) plots based on summary statistics using 30,000 (1%) simulated data sets and the posterior distribution of the parameters was estimated using the logit function (Cornuet et al., 2014). For model checking, we performed a PCA using new simulated datasets (1,000,000 per scenario) drawn from the posterior distribution of parameters, which are also represented on the PCA. Two sample summary statistics were used in model checking (mean number of alleles, mean genic diversity, mean size variance, F ST , classification index, shared allele distance and (dμ) 2 distance) to assess whether the observed data was included within the distribution of the predictive posterior parameters of the simulated data. Confidence in each scenario was obtained from the highest posterior probability using logistic regression, estimated by comparing the summary statistics from simulated and observed results, and from calculating type I and type II errors using 1000 simulated datasets (Cornuet et al., 2014).
Population heterozygosity and effective population size were compared between infected and noninfected populations using a Welch t test for unequal variances. We also modeled infection status (yes/no) and plague intensity (measured as density of plaque-forming units, PFU) in individual crayfish using population of origin as a random factor and individual homozygosity (H L ) as a predictor with either a binomial logit link (infection status) or a Gaussian link (plague intensity, measured as log(PFU+0.5) with the lme4 package in R, version 3.3.2.

| Host genetic diversity and population structuring
MICRO-CHECKER results indicated that four microsatellites had significant evidence of null alleles (p = 0.001), however results of repeated analyses (F ST , STRUCTURE) removing the affected microsatellites showed no obvious deviations from the results including all nine microsatellites (Supporting Information Table S2; Figure S1), therefore we carried out all subsequent analyses with all of them (Van Oosterhout et al., 2004). The nine microsatellite loci displayed moderate to high levels of polymorphism (H E between 0.5 and 0.7) across all the sites. All populations displayed a degree of deviation from Hardy-Weinberg equilibrium (HWE) across various loci due to lower than expected H E . Of 81 Chi-square tests conducted (one per locus) 37 showed a significant deviation from HWE (Supporting Information Table S3)  Effective population size (N e; ) ranged between 12.9 (Sirhowy) and 90.4 (Gavenny; Supporting Information Table S2) and, probably due to small sample size, confidence intervals were relatively large (3.9-28.6).
The STRUCTURE and StructureSelector analyses indicated that K = 4 (Supporting Information Figure S2; Table S6) is the most likely number of clusters in the dataset for British populations only and K = 5 (Supporting Information Figure S3; Table S7)    that the most likely number of discontinuities in genetic connectivity was due to three barriers (Supporting Information Figure S4), the strongest division occurring between Sirhowy and all the other sites (Barrier a; Figure 1). The next largest discontinuity was observed between Dderw and surrounding populations (Barrier b; Figure 1), while the third barrier separated Lugg from the Welsh populations (Barrier c; Figure 1; site 2). All barriers were supported by seven of nine loci.   (Figure 4;  Information Table S5).

| D ISCUSS I ON
Signal crayfish represents an ideal species to test the roles of genetic diversity and pathogens as novel weapons on invasion success, as the species is highly invasive throughout most of Europe, and Great Britain in particular. Its success has been attributed to preadaptation, aggressive behavior, niche plasticity, and the presence of the highly infectious A. astaci (Becking et al., 2015;Holdich, James, Jackson, & Peay, 2014;Hudina, Galić, Roessink, & Hock, 2011;James et al., 2014). Admixture between lineages could have also facilitated the establishment of this species, allowing populations to overcome founder effects and loss of genetic diversity (Kolbe et al., 2004;Rius & Darling, 2014), particularly when combined with high propagule pressure (Consuegra et al., 2011), but this had not been considered before. In Britain, the species has continued to spread despite management and control measures (Holdich et al., 2014).
The presence of infected and uninfected signal crayfish in close proximity (i.e., Dderw and Bachowey) could be a consequence of physical barriers and is important in relation to the conservation of endangered native crayfish populations, as invasive signal crayfish and native European crayfish can coexist in the absence of plague (Bubb, Thom, & Lucas, 2005;Diéguez-Uribeondo, 2006;Filipová, Petrusek, Matasová, Delaunay, & Grandjean, 2013). Native crayfish tend to inhabit refugia in the headwaters of numerous catchments within Britain, some of which have tested positive for A. astaci downstream (Bubb et al., 2005;Filipová et al., 2013).
In summary, it is likely that human-mediated dispersal has contributed to the numerous colonization events from a minimum of four genetic origins and further facilitated population expansion and succession of signal crayfish. Populations with A. astaci displayed higher heterozygosity, which could potentially be an indication of fitness benefits or a consequence of the absence of suggests that genetic monitoring before and after physical removal of crayfish can provide measures of genetic diversity and effective population size that could be used to assess the population consequences of removal actions. Usk Foundation), who provided advice, access to field sites, trapping equipment, and significant time monitoring traps throughout the duration of the research.

CO N FLI C T O F I NTE R E S T
None declared.

AUTH O R CO NTR I B UTI O N S
SC & CVR designed the study; CVR performed the genetic analyses with advice from SC and POTW; JC & JJ contributed samples and information; CGL performed statistical analyses; and CVR, SC, and CGL wrote the manuscript, which was revised by of all the authors.

DATA ACCE SS I B I LIT Y
Raw data (microsatellite genotypes per individual, parasite loads) will be archived in Dryad upon acceptance.