Population divergence and demography
The admixture analyses of the population genomics data revealed that the isolates of S. lacrymans were divided into geographically structured groups and sub-groups (Figure 1). When enforcing a two-group structure (K = 2), isolates from Europe and New Zealand were clearly separated from Japanese isolates (the cross-validation error was clearly lowest for K = 2; Supplementary Figure S1). The Principal Component Analysis (PCA, Figure 2a) likewise divided the isolates into the same two main groups (Europe/New Zealand vs Japan) along the first axis, explaining 49.8% of the variation. FST and DXY estimates confirmed that the Japanese and European populations were strongly differentiated, both based on average FST of 0.572, DXY of 0.524 and the genome-wide plots (Figure 3).
When the number of groups in the admixture analysis was raised to four (K = 4), both the Japanese population and the European/New Zealand group split into two (Figure 1d). Four French individuals were separated from the rest of the European/New Zealand isolates and five Japanese isolates were separated from the rest of the Japanese isolates. One individual from New Zealand (ICMP18202) was separated from the European samples along axis 1 in the PCA plot and from the European population in the admixture analyses at K = 7 (Figure 1d).
Admixture and PCA analyses performed on the European and Japanese populations separately revealed further geographic sub-structuring within both populations (Figure 1 and 2). For K = 4, the European population was divided into a northern European group, two groups in eastern France, and a southern French group, also visible in the PCA plot (Figure 2). When splitting the Japanese population into K = 7, a distinct northern group was identified, three additional admixed groups and three southern groups. These groups were also partially recognized in the PCA (Figure 2). Demographic analyses based on a mutation rate of 10-7, indicated that the historic population that gave rise to the current European population split from the Japanese population around 3,000 generations ago (confidence interval 2012-6157; Figure 4). The estimated divergence time (i.e. generations) was highly sensitive to the selected mutation rate, which increased to 18,770 generations with a mutation rate of 10-8. The modelled demographic scenario where both the European and the Japanese population experienced a change in population size was more likely than the scenario with only a population size change in Europe (the difference in likelihood between the two models, ΔAIC = 58,899). The European population went through a strong bottleneck about 250 generations ago (confidence interval 198-650), and Fastsimcoal2 analyses indicated that only six haplotypes were involved (confidence interval: 4.18-19.40). This could be a slight underestimate, since some of the European isolates were sampled at earlier time points (i.e. from 1935 to 2014; Table 1), which may affect the models. Nevertheless, based on our model, the Japanese population experienced a reduction in population size considerably earlier and in the same time period as the ancestral population diverged (around 3,500 generations). The reduction in the ancestral Japanese population was less severe, with an estimated effective population size of 1,337. The current effective population sizes were estimated to 1,273 individuals in Europe and 12,500 individuals in Japan (Figure 4).