Fine-scale invasion genetics of the quarantine pest, Anoplophora glabripennis, reconstructed in single outbreaks

The xylophagous cerambycid Anoplophora glabripennis, the Asian long-horned beetle (ALB), is highly polyphagous and can colonize a wide range of broadleaved host trees causing significant economic damage. For this reason, it is considered a quarantine pest in Europe and North America. Although the global spread of ALB has been depicted recently, no comprehensive studies exist on the genetic pattern of populations’ establishment and dynamics at fine-scale (i.e. within invasive outbreaks), before eradication measures are applied. This information may, however, be particularly important for an efficient management and control of invasive pests. Here, we characterized population genetic diversity and patterns of spread of ALB within and among the four outbreaks detected in Switzerland between 2011 and 2015. For this, we genotyped 223 specimens at 15 nuclear microsatellite loci and conducted specific population-based analyses. Our study shows: (1) At least three independent introductions and a, human-mediated, secondary dispersal event leading to the four outbreaks in the country; (2) An overall low intra-population genetic diversity in the viable and several years active invasive populations; (3) A colonization of single trees by homogeneous ALB genotypes; And (4) an establishment of populations several generations prior to its official discovery.

. STRUCTURE bar plots using 13 SSRs of 192 ALB genotypes in four Swiss outbreaks and 7 single findings. Each bar represents the average estimated individual membership probability (ordinate) of an individual to belong to a specific cluster (indicated by specific color). In the lower left part (below), the barplots are supported by a scatterplot with mean log-likelihood values (± standard deviation) for different numbers of clusters (K) and in the lower right side with curve of ∆K .  Figure S4. STRUCTURE bar plots using 13 SSRs of 101 ALB multilocus genotypes Marly and Brünisried outbreaks, canton Fribourg, Switzerland. Each bar represents the average estimated individual membership probability (ordinate) of an individual to belong to a specific cluster (indicated by specific color). In the lower left part (below), the barplots are supported by a scatterplot with mean log-likelihood values (± standard deviation) for different numbers of clusters (K) and in the lower right side with curve of ∆K .

Details on the ABC analysis of Asian longhorned beetles outbreaks in Marly and Brünisried, Switzerland
Based on the ratio FST_Ma#1Br < FST_Ma#2Ma#1 < FST_Ma#2Br and on the STRUCTURE results, next 6 competing scenario of the demographic events were assumed ( Figure S1).
In a first step, to define likelihood scenario of demographic events in Brünisried and Marly we simulated pseudo-observed datasets (PODs) under uniform distribution of the parameters in a broad ranges: 10 -10000 of population effective size, from 1 to 10^4 generations of occurrence of the events according to the condition t3 > t2 > db2 > t1 > db1. In a second step, in order to infer the time in generation ago when particular demographic event happened in the selected in a first step the likelihood scenario we used more narrow time frame from 1 to 10 generation, consider information from the field observation and prior study of Forster & Wermelinger 2012. In this, second step we compared two scenario of the selected topology but excluding or including third change of effective size of initial population due to degradation ( Figure S2). In addition, we estimated the posterior distributions of each demographic parameter for the best demographic model, by carrying out local linear regressions on the 1000 closest of simulated data sets, after a logit transformation to parameter values. In both steps, for each simulation, a value for each parameter was drawn from prior distribution and performed coalescent simulations with the same number of alleles and loci per population as in the observed dataset. A set of 15 summary statistics describing within and among population genetic diversity were calculated for each POD and the observed data. Population specific statistics included allele size variance (VAR), Garza -Williamson mean across loci (MGW). Between population statistics included FST , mean variance of the absolute allelic size (V2P), Goldstein's genetic distances between populations based on the variance in the repeats number (DM2). We used for simulation generalized stepwise-mutation model and default parameters of the DIYABC v.2.1.0.
The posterior probabilities of each competing scenario were estimated using a logistic regression on the 1% of 10^6 simulated datasets closest to the observed dataset. The best-fitting scenario was selected based on the highest posterior probability with a non-overlapping 95% confidence interval. We evaluated the ability of the ABC analysis to discriminate between the competing scenarios by analysing subset of 500 of closest to observed simulated data sets. Therefore, we estimated the Type-I error rate as the proportion of occurrences the best-supported scenario did not show the highest posterior probability among the competing scenarios and we estimated the Type-II error rate, by calculating the mean proportion of occurrences in which the best-supported model was incorrectly supported instead of selected for simulation one of the competing scenario. * Posterior probabilities inferred directly from summary statistics of the closest to observed simulated data sets; ** Posterior probabilities inferred from linear discriminants of the summary statistics of the simulated data sets For the best supported scenario, the posterior probability distribution of time and demographic parameters were estimated, after a logit transformation, by local linear regression on the 1% of simulations closest to the observed data. Finally, for model checking we simulated 10000 PODs from the posterior under the bestsupported scenario in order to evaluate whether this model could successfully reproduce the observed data.    The initial ALB populations derived from ancestral NA population from South Korea and inevitably experienced bottleneck and reduction of the genetic diversity due to founder effect in time t3. Then in time t2 from population stepwise population Ma2_b was established, further developed to sampled Ma2. In the time t1 from Marly was transported fire wood with beetles to Brünisried and population Br_b was established, further on developed to sampled population Br.
The initial ALB population Ma2 appeared from ancestral NA population, inevitably experienced bottleneck and reduction of the genetic diversity due to founder effect in time t3. Then in time t2 from population Ma2 was transported fire wood with beetles to Brünisried and population Br_b was established, further on developed to sampled population Br. In the time t1 from Ma2 stepwise population Ma1_b was established, further developed to sampled Ma1.
The infested wood from unsampled ancestral population NA were immediately transported to Brünisried and initial population Br was established in the time t3. Later on, in time t2 from population Br beetles were transported to Marly and population Ma2_b was established, which developed to sampled population Ma2. Then in t1 from Brünisried stepwise population Ma1_b was established, further developed to sampled Ma1. The initial ALB Ma1_b population derived from ancestral population NA to Marly in time t3. Inevitably experienced bottleneck and reduction of the genetic diversity due to founder effect. Then in time t2 from Ma1_b were infested one more location in Marly and Ma2_b founder population was established, later since time t2-db2 developed to sampled Ma2. In time t1 from Ma1_b was transported fire wood with beetles to Brünisried and initial bottlenecked population Br_b was established, further on developed to sampled population Br, whereas genetic drift and extinction of some lineages resulted in sampled population Ma1.
The initial NA population derived from ancestral population to Marly in time t2, experienced bottleneck and reduction of the genetic diversity due to founder effect and established as Ma1_b, with a time also experienced genetic drift and due to other natural demographic processes developed to sampled population Ma1. In time t1 from Ma1_b population was infested one more location in Marly, that is Ma2_b founder population, which developed to sampled population Ma2. Meanwhile, from Ma1_b was transported fire wood with beetles to Brünisried and initial founder population Br_b was established, further on developed to sampled population Br.
All three populations of ALB derived from ancestral population NA to both locations in Marly and to Brünisried approximately in the same time t1. Inevitably all experienced bottleneck due to founder effect and established initial populations Ma1_b, Ma2_b and Br_b. Then populations independently developed to sampled Ma1, Ma2 and Br.  Figure S6. Second step of ABC evaluation of the two assumed scenario. First scenario was supported in the first step of the analysis, whereas second scenario exclude population effective size change in Marly #1 in time t1-db. Supported scenario marked with green rectangle.