DISCUSSION
We confirmed that the two populations of Serpula lacrymans var.lacrymans from the built environment in Japan and Europe are highly divergent, as previously revealed by Kauserud et al.(2007), and that the divergence between the two populations predated human influence. The results from the demographic modelling analyses strongly support this hypothesis, indicating that the ancestral population split more than 3,000 generations ago and led to the current divergent populations in Japan and Europe. Based on genomic similarity of the wild Serpula lacrymans var. shastensis and the building occupant Serpula lacrymans var. lacrymans , we have previously suggested that this species is one of few wood-decay fungi that was highly pre-adapted to the built environment (Balasundaramet al. 2018). As a result, it established extensive populations in temperate regions worldwide vectored by human translocation (Kauserudet al . 2007).
The split between the two invasive populations predates human influence. The natural habitat of S. lacrymans is high alpine regions with scattered large conifer logs (Kauserud et al. 2012). From there, Serpula lacrymans has established independently at least twice in the indoor habitat (Kauserud et al . 2007). However, our dating of this split is influenced by the implemented mutation rate, illustrated by an increase in number of generations from 3,067 to 18,770 when using 10-8 as alternative mutation rate (to 10-7). Assuming a split approximately 3,000 generations ago and a generation time of one year under optimal fungal growth conditions, human-made habitats will have existed, i.e. 3,000 years before present. Indeed, there are dated wooden constructions circa 7,000 years ago from Europe (Tegel et al.2012). However, a longer generation time than one year (for establishment, vegetative growth and fruiting) is plausible and the mutation rate is uncertain, which suggest that the split of the ancestral population probably predated the availability of the built environment. Central and eastern Asia, where S. lacrymans can be found in nature, were not covered by ice sheets during the last glaciation 115 – 11 K BP (Tian & Jiang 2016). The dry and cold areas may have driven S. lacrymans into refugia, thus splittingS. lacrymans into several sub-populations. Although speculative, such a scenario also fits well with the estimated reduction in population size at almost the same time as the split. Consequently, the Himalayas, where the fungus has it westernmost natural distribution (Kauserud et al. 2012), and another north Eastern Asian population, might have functioned as separate source populations of the current indoor populations in Europe and Japan. From previous population studies of S . lacrymans , it is known that the European population shares genetic material with specimens from the Himalayas, and that the Japanese population is geographically and genetically closer to the natural population in East Asia (Kauserud et al . 2012, 2007).
The genome-wide SNP data confirmed that the Japanese population is genetically more variable than the European population and that more genetic variation was retained during its founder event(s). This is mirrored by the higher number of vegetative incompatibility (vic ) alleles and diversity of mating types present in the Japanese population (Engh et al. 2010a). The two genetic regions in scaffold 1 and 3 with lower genetic diversity, cover two large genomic regions containing many genes with various functions. One possible interpretation is that these two genomic regions harbor inversions and therefore show lower rates of recombination in the Japanese population (and in one New Zealand isolate). However, better genome assemblies, which can be achieved with long read sequence data is required to test this. Currently, the function and effect of these possible genome rearrangements are unknown.
The European population invaded the built environment about 250 generations ago. As noted above, genetic similarity between Himalayan specimens collected in forests and the European indoor population has previously been observed (Singh et al. 1993; Kauserud et al. 2007). Our demographic modelling suggests that the colonization of Europe happened between 200 and 400 generations ago. Assuming a generation time of one year, this estimate of 200-400 generations fits well with the species description of S. lacrymans in 1781 (Wulfen 1781), based on a collection from Europe (Austria). During this period, and also somewhat earlier, there were considerable trade activities between Europe and Asia which could easily have vectored the colonization of S. lacrymans to Europe, e.g. through the trade and transport of timber. Human activity might have further spreadS. lacrymans , from Europe to North America and Australia (Kauserud et al. 2007) and is also considered the main agent for dispersal of several plant pathogens, such as Microbotryum lychnidis-dioicae and Ophiostoma ulmi (Brasier 1991; Fontaineet al. 2013). To better understand the colonization event ofS . lacrymans in Europe, the relatedness of European isolates to isolates from the natural range in the Himalayas should be further investigated.
The European population of S. lacrymans holds low genetic diversity, and the founding members were estimated to consist of only four or five haplotypes (depending on mutation rate). Hence, it is possible that only two or three heterokaryotic isolates colonized and founded the entire European population. This assumption is strongly supported by few mating (MAT) and vegetative incompatibility (vic ) alleles detected in the European population where three heterokaryotic isolates may account for the entire allelic diversity (Kauserud et al. 2006; Engh et al. 2010b; Skrede et al. 2013). Similar to S. lacrymans , the European population of the invasive ash dieback fungus H. fraxineus is genetically extremely homogenous and was established by only two haploid individuals (McMullan et al. 2018). Two different subspecies of the Dutch Elm disease parasite Ophiostoma novo-ulmi established as clonal parasites in Europe. Later, sexual reproduction between these populations produced a hybrid with high pathogenicity (Brasier & Kirk 2010). Hence, there are now multiple lines of evidence that invasive fungi may establish on other continents through very tight founder events followed by extensive population growth.
The small founder event establishing the European population is also reflected in the slow LD decay and high levels of Fis observed in the European population. A high degree of selfing or clonal dispersal may cause such abnormal linkage decay curves (Nieuwenhuis & James 2016). Similar patterns of linkage disequilibrium were observed for isolated and highly biparentally inbred wolf populations in Italy and the Iberian peninsula (Pilot et al. 2014). The very low number of isolates founding the European population of S. lacrymans has probably also led to high levels of biparental inbreeding. The limited number of mating alleles available will necessarily lead to a reduction in mating opportunities between individuals, and the tetrapolar mating system of S. lacrymans still allows 25% of the spores from the same fruit body to mate. No putative clonal isolates were found in our or previous studies (Kauserud et al. 2007; Engh et al.2010b; Maurice et al. 2014), supporting that selfing or clonal dispersal is not the main explanation of the slow linkage decay.
Signatures of selection in the built environment. In both populations, we observed signals indicative of balancing or purifying selection acting on two loci. For these two loci we observed significantly lower FST values than expected (DXY was not different from the expectations). Genomic regions of low differentiation could be caused by pleiotropic effects (one locus having an effect on two or more phenotypes), where these phenotypes have different selection pressure in the two populations. The differences in Tajima’s D and π between the populations at these two loci indicate that pleiotropic effects may be involved, i.e., the SNF1 complex at scaffold 8 show signals of purifying selection in the European populations, but balancing selection in the Japanese populations. The SNF1 complex is involved in gene expression regulation as a response to starvation in yeast (Sanz 2003), but is also shown to be involved in responses to several different environmental stressors (summarized in Shashkova et al . 2015). The homolog of the main SNF1 kinase was differentially expressed in response to varying substrate composition during wood decay in S. lacrymans,supporting a role in substrate-dependent gene regulation (Hess et al . 2021).
Another gene with an oxysterol-binding domain (as was found in the locus at scaffold 27 in our analyses) evolved at a specifically rapid pace (significantly higher dN/dS) in S. lacrymans compared to the sister taxon S. himantioides (Balasundaram et al. 2018). Oxysterol-binding proteins are involved in transportation of lipids in eukaryotic cells (revised in Qui & Zeng 2019), and have shown to be important to polarized hyphal formation and growth in Candida andAspergillus (Ghugtyal et al . 2015; Bühler et al . 2015). Maintaining diversity for both the oxysterol-binding and the SNF1 complex could therefore be important for S. lacrymans’ unique ability for rapid growth and decay of large substrates.
In the European population, selective sweeps were found in genes related to DNA replication (a DNA helicase with a Pif1 domain) and protein modification (O-Fuct domain involved in glycosylation). In parallel to this, associations between helicases and the environment have been found in several landscape genomic studies, i.e. the wood decay fungusPhellopilus nigrolimitatus (Sønstebø et al . in prep.), the ectomycorrhizal Suillus brevipes (Branco et al. 2017) and the red bread mold Neurospora crassa (Ellison et al . 2011). Suggesting that modification of DNA replication may be the first signals of local adaptation for recently founded populations. Thus, in the European population we may be observing recent events, which corresponds well with the demographic history of the European population, with its small effective population size and the recent bottleneck.
The selective sweep in the Japanese population was in a non-coding region. Thus, we cannot conclude about the function of this sweep. However, within a 10 kb window of this sweep there are three predicted genes, of which one is a Cytochrome P450. Cytochrome P450s are monooxygenase enzymes known to be involved in the detoxification of polyphenols and other defense molecules secreted by other organisms or encountered during wood decay (Cresnar and Petric 2011; Ichinose 2013; Xu et al . 2015). For fungi, cytochrome P450s were recently suggested to be involved in heavy metal tolerance in Suillus luteus (Bazzicalupo et al. 2019) and during pathogenic interactions between Heterobasidion annum and conifer trees (Karlsson et al. 2008). Wood type dependent difference in gene expression of cytochrome P450s have previously been found for two wood decay species (Wymelenberg et al . 2011). Further, in our recent study of S. lacrymans (Hess et al . 2021), a larger repertoire of differentially expressed cytochrome P450s was associated with significantly faster decay of recalcitrant pine wood in a Japanese isolate compared to an isolate from the European population. However, in the experiments in the current study we did not detect systematic differences in the ability to decay different substrates between the Japanese and the European populations, rather that there is high strain to strain variation in this trait. More detailed functional studies are therefore required to better understand the functions of individual cytochrome P450s.
The New Zealand population is admixed between European and Asian populations. Inclusion of two isolates from New Zealand enabled us to test whether this population has an admixed ancestry, as hypothesized in Kauserud et al . (2007). Indeed, isolate ICMP18202 possessed a combination of European and Japanese alleles, in support of this hypothesis. Similarly, in other invasive fungi, like B. dendrobatidis , O. ulmi/novo-ulmi, H. fraxineus and the oomycete pathogen Phytophthora alni , secondary admixture between independently evolved lineages have been observed (Ioos et al.2006; Brasier & Kirk 2010; McMullan et al. 2018; O’Hanlonet al. 2018). The merging of genetic lineages in founder areas may lead to novel and more aggressive genotypes, as shown for most of these species (Ioos et al. 2006; Stukenbrock & McDonald 2008; Brasier & Kirk 2010; McMullan et al. 2018; O’Hanlon et al. 2018). In our wood decay experiment, the New Zealand isolate ICMP18202 decayed pine wood rapidly (Figure 8), though more isolates are needed to confirm this pattern. Since only one of the two isolates showed signals of admixture, the admixture could be a recent event. Further, the admixture was heterogeneously distributed in the genome (mainly scaffolds 1 and 3), indicating limited cycles of recombination. It could also be that PCAdmix assigned genomic windows more readily to homogenous genomic regions with low diversity, and hence, more readily detects admixture in scaffolds 1 and 3 with lower diversity as compared to the rest of the genomes of the Japanese isolates. The reasons for reduced diversity and admixture in large genomic regions in the admixed isolate from New Zealand are currently unknown. This could be a consequence of genomic areas of low recombination, resulting from chromosomal rearrangements. In the future, more thorough analyses of genomic synteny, based on assembled genomes, among these three populations may shed light on their effect on population divergence. Furthermore, the isolate(s) that contributed to the admixture were not necessarily of Japanese origin, but could originate from other locations in Asia. Thus, including more isolates from the full geographic distribution of S. lacrymans may contribute to better understand the admixture event in New Zealand.
Concluding remarks. By using full genome data in combination with growth experiments, we revealed genomic differences among isolates of S. lacrymans from Europe, Japan and New Zealand. We estimated that the source population of the European population split from the Japanese population 3,000 to 19,000 generations ago, and that the European population established about 200 – 400 generations ago through a tight bottleneck event. Two genes related to mycelial growth seemed particularly important for survival of both indoor populations and probably evolved before the species became invasive in the built environment. However, population specific selective sweeps identified more recent events, which show that decay of various substrates in Japan and rapid adaptation of DNA replication and protein modification in Europe influence population survival. We were able to confirm that one isolate from New Zealand held genomic signatures of admixed ancestry. However, more samples from New Zealand and the native Asian populations are needed to infer, with better certainty the source populations ofS. lacrymans and for in-depth analyses of the evolutionary history of the populations in the built environment.