Landscape‐scale genetic differentiation of a mycangial fungus associated with the ambrosia beetle, Xylosandrus germanus (Blandford) (Curculionidae:Scolytinae) in Japan

Abstract In this study, we examined the genetic structures of the ambrosia fungus isolated from mycangia of the scolytine beetle, Xylosandrus germanus to understand their co‐evolutionary relationships. We analyzed datasets of three ambrosia fungus loci (18S rDNA, 28S rDNA, and the β‐tubulin gene) and a X. germanus locus dataset (cytochrome c oxidase subunit 1 (COI) mitochondrial DNA). The ambrosia fungi were separated into three cultural morphptypes, and their haplotypes were distinguished by phylogenetic analysis on the basis of the three loci. The COI phylogenetic analysis revealed three distinct genetic lineages (clades A, B, and C) within X. germanus, each of which corresponded to specific ambrosia fungus cultural morphptypes. The fungal symbiont phylogeny was not concordant with that of the beetle. Our results suggest that X. germanus may be unable to exchange its mycangial fungi, but extraordinary horizontal transmission of symbiotic fungi between the beetle's lineages occurred at least once during the evolutionary history of this symbiosis.

ancestors may have diverged more than 170 million years ago (Farrell et al., 2001). Thus, ambrosia beetles strongly depend on polyphyletic fungal groups.
In this study, we investigated the genetic structure of an ambrosia fungus isolated from X. germanus mycangia and adult females used for fungal isolation in order to elucidate the differentiated fungal and beetle lineage patterns. We also discuss evolutionary events that may have influenced the diversification of their mutualistic system.

| Insect collection and fungi isolation
We collected X. germanus samples from 14 sites in Japan (Table 1).
To capture live adult females, in 2007, we set up Nagoya University (Meidai) traps (Ito & Kajimura, 2006)  all sites and used 1-31 mature females from each site (Table 1). We also trapped adult females of Xylosandrus brevis (Eichhoff) in Aichi Prefecture (AIT) and Scolytoplatypus mikado (Blandford) in AIT and Wakayama Prefecture (WKT). Two species of Xylosandrus beetles and S. mikado were identified according to Nobuchi (1981Nobuchi ( , 1980, respectively. We isolated fungal conidia from mycangia of X. germanus, X. brevis, and S. mikado living adult females. All collected beetles were preserved at −20°C in 99.5% ethanol after fungal isolation. Isolates from mycangia were directly placed on potato dextrose agar (PDA) plates in 90mm sterile Petri dishes and incubated at 20°C for 5 days in the dark.
The isolates were grouped by cultural characteristics and identified at the generic level using the ambrosia fungi keys of Batra (1967).

| Data analysis
Sequences were aligned using the BioEdit v.7.0.2 software (Hall, 1999). BLAST searches were performed with sequences of each isolate in the NCBI GenBank database (http://www.ncbi.nlm.nih.gov), and published sequences of relevant and related species were incorporated into the datasets (Tables 2 and 3). Calculations for the G-C composition were performed using the MEGA 4 software (Tamura, Dudley, Nei, & Kumar, 2007). For the phylogenetic analysis, we chose maximum-parsimony (MP) method (Nei & Kumar, 2000), using the MEGA 4 software. The MP analysis also used 1,000 bootstrap replications. A phylogenetic analysis was also performed for the three loci (18S, 28S, and β-tubulin) using MP methods. Concordance among the three different gene datasets was evaluated by the incongruence length difference (ILD) test (Farris, Källersjö, Kluge, & Bult, 1995) implemented with PAUP*4.0b10 (Swofford, 2003), using 1,000 replicates.

| Morphological characters of symbiotic fungi isolated from X. germanus mycangia
Based on the color and growth pattern of colonies (mycelia tuft), isolates obtained from X. germanus mycangia were separated into three cultural types (Types I, II, and III) ( Figure 1). Five days after inoculation, colony characteristics of Types I and II were similar to those of A. hartigii and A. grosmanniae shown in Batra (1967)  Phialophoropsis sp. CBS460.82 KR673890 Mayers et al. (2015) found in two northern populations (HKF and YMT; Figure 2). Type I fungi were distributed throughout Japan, but the other two types were located in Japan.

| DNA sequencing and phylogenetic analyses
The amplicons obtained from the 18S regions of ambrosia fungi sequenced in this study were 997 bp in length. These fragments had 45.5% G/C content. Two haplotypes were defined from 132 isolates. Haplotype XgF18S01 was detected in all cultural types ( Figure 3). Haplotype XgF18S02 was detected only in Type II. The ILD test indicated that the 18S, 28S, and β-tubulin datasets were concordant (p = .635). On the basis of the three combined loci, 11 multilocus haplotypes were defined from 129 isolates (Table 4).
Haplotypes 07 were detected only in Type I ( Figure 6). Haplotypes haplotypes were clustered in a subclade (subclade II) and five Type I haplotypes, XgF02, and XgF04-07, were clustered in a subclade.
We compared geographical distribution of seven Type I haplotypes using chi-square test. Seven Type I haplotypes were not uniformly distributed in Japan (χ 2 -test, p < .05; Figure 7). XgF01 was only found , and clade C one haplotype (XgCOI20). X. germanus clades A, B, and C were unexceptionally associated with symbiotic fungi Types I, II, and III, respectively (

| DISCUSSION
Symbiotic fungi isolated from X. germanus mycangia in Japan had all three cultural types (Figures 1 and 2). The three types formed one clade with A. grosmanniae, A. roeperi, A. hartigii and A.xylebori in 18S and 28S (Figures 3 and 4). In β-tubulin gene, X. germanus fungi clustered as a monophyletic group together with A. beaveri, A. hartigii and A. xylebori clade ( Figure 5). In combined three loci, X. germanus fungi clustered as a monophyletic group together with A. hartigii and A. xylebori clade ( Figure 6). Therefore, all fungal isolates obtained in this study were identified as closely related species to four species, A. grosmanniae, A. roeperi, A. hartigii, and A. xylebori. Phylogenetic analyses based on 28S rDNA, β-tubulin, and the combined three loci revealed that Types I and II haplotypes formed subclade within X. germanus fungi clade with high bootstrap values (Figures 4-6). These results suggest that three types of X. germanus fungi, which are distinct from each other as per morphological and phylogenetic characters,  are distributed in Japan. These results also suggest that Type III first differentiated from ancestral members, common to all three types, and subsequently, ancestral members of Types I and II have differentiated into Types I and II.
The COI haplotypes of X. germanus were divided into three distinct lineages (Figure 8). This result was the same that of Ito et al. (2008). The beetles had a specific type of symbiotic fungi for each clade (Figure 8 and Table 4). These results suggest that X. germanus are unable to exchange mycangial fungi between clades of beetles.
However, horizontal transmission of mycangial fungi may occur within the same beetle lineage, because no specific relationships were found between beetle and fungal haplotypes within same clade (Table 4).
Some bark beetles likely exchange their fungi between neighboring nests in the same host tree (Six, 2003;Six & Bentz, 2007). Each Type I haplotype showed a nonrandom distribution on the Japanese archipelago (Figures 7 and 8). These distributions may be formed by X. germanus, because fungal dispersion depends on beetle migration.

X. germanus cannot migrate between Hokkaido and other regions in
Japan because of the Tsugaru Strait geographical barrier (Ito et al., 2008). However, scolytine beetles have a flying range of 10-15 km (Gries, 1985;Wood, 1982). Thus, the migration ability of the beetle may regulate fungal dispersion, resulting in the lack of random distribution in the Type I haplotypes. Additionally, COI clades of X. germanus can be distinguished by the cultural types of its symbiotic fungi, because the clades have strong correlations with the cultural types ( Figure 8 and Table 4).
The phylogenetic divergence patterns of the symbiotic fungi did not coincide with those of X. germanus (Figures 6 and 8). In mycangial fungi, Type III lineage was sister to a clade containing Type I and Type II lineages (Figure 6). In contrast, ancestral members of X. germanus

Ambrosiella roeperi KF646767
Ambrosiella grosmanniae LC175288  (Ito et al., 2008). After colonization, clade A and B beetles secondarily came into contact during the last glacial epoch in Japan (Ito et al., 2008). Type II was not differentiated within subclade I-II ( Figure 6), suggesting that clade B ancestors may have symbiotically associated with Type II when clade B occurred and contacted to clade A.
We obtained two important results related to the phylogeny of X. germanus and its symbiotic fungi: a single beetle lineage is consistently associated with a single fungal type in the X. germanus fungal symbiont system, although more than two types of symbiotic fungi were found in northern populations (Figure 2), and exceptional horizontal transmission in symbiotic fungi between beetles lineages occurred at least once, sustaining novel beetle-fungus symbiotic relationships. Why are the beetles unable to exchange symbiotic fungi from the existing type to other types? In ambrosia beetles, glandular secretions into the mycangium can facilitate the growth of specific ambrosia fungi (Harrington, 2005;Norris, 1979).
Some bark beetles such as the southern pine beetle (Dendroctonus frontalis Zimmermann) also have glandular cells in their mycangia and carry one specific fungal symbiont (Bridges, 1985). Thus, it is possible that specific ambrosia fungi lineages in X. germanus are selected by mycangia secretion. Colony growth rate on PDA and the competitive race of each fungal type vary according to thermal conditions (Ito & Kajimura, 2011). Some symbiotic fungi of bark beetles also have thermal traits in the field (Six & Bentz, 2007;Six & Paine, 1997;Solheim & Krokene, 1998). The fitness level of ambrosia and bark beetles decreases or increases depending on the symbiotic fungal species used for nutrition (Harrington, 2005;Kajimura, 2000;Six & Bentz, 2007). Therefore, X. germanus and their mycangial fungi mutual systems may experience constant F I G U R E 6 Phylogram obtained from about 2,000 bp of the combined loci (the 18S, the 28S rDNA, and the β-tubulin gene) of symbiotic fungi isolated from mycangia of Xylosandrus germanus and related fungi species. One of 10 maximum-parsimony (MP) trees. CI = 0.863, RI = 0.913, length = 519 steps. Bootstrap values (left) and branch support values (right) (>50%) are given above the branches. Bold letters indicate sequences obtained in this study. Cultural types (Type I-III) defined in Figure 1 are shown in bracket after haplotype codes (XgF01-11). XbF and SmF represent symbiotic fungi isolated from mycangia of Xylosandrus brevis and Scolytoplatypus mikado, respectively

CONFLICT OF INTEREST
None declared.

AUTHOR CONTRIBUTIONS
M. Ito designed the study, wrote the initial draft of the manuscript, and analyzed and interpreted data in the study. H. Kajimura contributed to interpretation of data, assisted in the preparation of the manuscript, and critically reviewed the manuscript. All authors approved the final version of the manuscript, and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.