For decades, secondary ambrosia beetle species have been increasingly observed to attack apparently healthy trees. Some of them cause mortality to living trees. In Japan, Platypus quercivorus that is widely distributed in Asia has caused mass mortality of trees belonging to the family Fagaceae by vectoring Raffaelea quercivora. Similar disease has been in epidemic in Korea by attacks of Platypus koryoensis. On the other hand, unexpected introduction of ambrosia beetles to non-indigenous area has been increasing because of globalization. Some has become major components of ambrosia beetle fauna in the non-indigenous area. The worst example among invasive ambrosia beetles is Xyleborus glabratus that causes laurel wilt in US by vectoring Raffaelea lauricola. The laurel wilt has caused enormous economic damage to avocado farmers in US. Therefore, ecological information on ambrosia beetles is needed from a view point of biodiversity conservation as well as economic importance. The purpose of this study was to determine ambrosia beetle guild, factors influencing the guild, and niche of each species attacking on an oak tree Quercus serrata. Risks of invasion and of vectoring tree-killing disease were evaluated for each species.

The study was conducted in three locations of the University of Tokyo Forests in the Central Japan. Timing of cutting trees (= TC) was controlled by preparing bolts monthly from March to September in Chichibu. They were covered by metal mesh to protect from insect attacks and left on the forest floor in the three locations: Aichi (LOC-A), Chiba (LOC-B), and Chichibu (LOC-C) (= LOC). The LOC-A had Japanese oak wilt (JOW). However, the LOC-B and the LOC-C had no historical records of JOW. Timing of exposure (= TE) was controlled by removing the metal mesh so that wood oldness (= WO) at the timing of exposure after cutting tree was also experimentally controlled. Bait bolts were exposed to insect attacks for one month in a normal experiment, while they were left until the end of September in a conditioning experiment. Twenty eight regimes, each consisted of four bolts, were set in each of a normal and a conditioning experiment so that the total number of bolts on each location was 224. Ambrosia beetles were collected by dissecting the bolts after the exposure and identified into species. The number of entry holes was used as an indicator for abundance of each species.

In the normal experiment, twenty morphospecies of Scolytinae and six species of Platipodinae were collected. In the conditioning experiment, six Scolytid and two Platypodid species disappeared, whereas one Scolytid species recruited so that fifteen species of Scolytinae and four species of Platypodinae were collected. In total, twenty one morphospecies of Scolytinae and six species of Platipodinae were collected.

Species richness and abundance peaked on bolts prepared in April-May, on bolts exposed in July, and on 2-3-month-old bolts. Eliminating greatest influences of LOC on abundance, results of hierarchical partitioning showed that TC had a strong influence on both species richness and abundance. LOC-A (Aichi), in which Japanese oak wilt disease (JOW) incidence occurred, showed the greatest species richness and the smallest value of Pielou's evenness. Abundance of the most abundant ("the major") species was more than twice that of the second major species, which was a likely cause of the smallest evenness in LOC-A. Trees killed by JOW may have increased the abundance of the major species. On the contrary, in LOC-C (Chichibu), Pielou's evenness and alpha and gamma diversity and the Shannon index were greatest among the three locations although species richness was smallest. High similarity between guilds in LOC-A and LOC-B (Chiba) was probably caused by similarity in vegetation. The LOC had the greatest effect on determining guild structure. Effect of TE was greater than TC. The effect of WO was negligible. A hierarchical structure among the three factors was a likely cause of their relative importance determining guild structure.

Niche center and niche breadth were estimated for the three niche contexts. Relative competitive ability (RCA) was calculated by subtracting log-transformed abundance after adding 1 in normal experiment from that in the conditioning experiment. The relationship between the RCA and niche center of WO was determined to test the hypothesis that there are trade-offs between these values. No significant relationship between the two values was found for all species. However, significant positive relationship was obtained if thirteen minor species we excluded indicating that poor competitors with negative RCA value tended to attack fresh wood to avoid competition. Species that have been reported as an alien species tended to have broader niche. All Scolytid species collected in my study have a habit of haplodyploidy and sibling mating so that Scolytid have a higher risk of invasion than Platypodid. Platypodid species tended to attack fresh bolts compared to Scolytid so that Platypodid have higher risk to vector tree-killing disease than Scolytinae. Among Platypodinae, Crossotarsus niponicus and Platypus calamus significantly increased in the conditioning experiment and had great value of the RCA so that these are good competitors and relatively high risk species of invasion. However, they did not show strong preference to fresh bolts compared to other Platypodid species. Xylosandrus germanus and Xylosandrus crassiusculus, which attack living trees in non-indigenous area, also did not show a strong preference to fresh bolts so that there are no guarantees that C. niponicus and P. calamus are safe enough in non-indigenous area. Positive RCA values of X. germanus and of X. crassiusculus supported that they became invasive alien species. Euwallacea validus significantly decreased in the conditioning experiment compared to the normal so that this species seemed a poor competitor, whereas this species tended to attack fresh bolts. Xyleborus ganshoensis was considered as the highest potential to establish population in non-indigenous area because of its broad niche. X. ganshoensis was also likely to attack living trees since many individuals attacked fresh bolts but unlikely an invasive alien species because of its negative RCA value. Platypus quercivorus, a vector of the Japanese oak wilt, showed the highest mean abundance on 0-month-old bolts. On the other hand, Xyleborus seiryorensis had the smallest value of WO niche center, indicating that this species could be aggressive enough to attack living trees. The method developed in this study would be applicable to similar risk assessment of ambrosia beetles on other host species and in other countries.

アンブロシアキクイムシは、アンブロシア菌と共生関係を持つ昆虫で、近年、共生菌が木を枯らすことから世界中で問題になっている。日本でもナラ類に寄生するカシノナガキクイムシが引き起こすナラ枯れが大きな問題になっている。

秩父で伐採したコナラ丸太を、千葉演習林・秩父演習林・愛知演習林の3箇所で曝露し、寄生するアンブロシアキクイムシを採集した。伐採時期、曝露時期を人為的に制御して、丸太の鮮度も制御した。種数、アバンダンス、そしてギルドの構造に及ぼすこれらの要因の影響を解析した。種数は、伐採時期が4~5月、曝露時期が7月、伐採後2-3ヶ月でピークを示した。ギルドの構造を決定するうえで場所が最大の効果を及ぼしており、物理的な距離よりも、標高の違いが影響したものと考えられた。標高は、気温を通してあるいはさらに食性の違いを通して作用したものと推測された。残りの要因では、曝露時期、伐採時期、丸太の鮮度の順に影響が大きかった。4つの要因間にみられる階層構造が、ギルドの構造を決定するうえでこれら要因の相対的な重要性を決定していると考えられた。

次に、場所以外の3要因について、それぞれの種のニッチの中心とニッチ幅を求め、侵入生物としてのリスクと生立木を枯らすリスクを評価した。キクイムシ亜科のアンブロシアキクイムシの方がナガキクイムシ亜科のアンブロシアキクイムシよりもニッチ幅が広い傾向が認められた。ナラ枯れの媒介昆虫であるカシノナガキクイムシや生立木に加害することが報告されている種は、伐倒後まもない丸太に選好性が認められた。丸太の鮮度について広いニッチ幅を持つ種の中にも、伐倒後まもない丸太に多く穿孔する種も認められた。ガンショキクイムシは、3要因すべてのニッチ幅が広く、また新鮮な丸太にも加害する性質があるため、侵入生物としても生立木を枯らすリスクも高いものと考えられた。ナラ枯れの媒介昆虫であるカシノナガキクイムシは、伐採当月の丸太に穿孔した個体が最も多く、生立木に加害する性質を裏付けた結果が得られた。セイリョウキクイムシは、丸太の鮮度に対するニッチの中心がもっとも小さくやはり萎凋病媒介のリスクが高い種と考えられた。今回の方法は、今後の温暖化と国際化に伴う、侵入生物と萎凋病発生のリスク評価を行ううえで、他の樹種や他の地域のアンブロシアキクイムシに対しても有効な方法と考えられた。ただし、ニッチ幅が広い種が侵入の危険性が高いという前提については、メタ解析などを使ったさらなる研究・証明が必要という指摘が審査委員からあり、今後の研究のさらなる発展を期待する。また、生立木を枯らすためには、アンブロシアキクイムシだけではなく共生菌の寄主植物に対する病原力も検討する必要があり、この点も今後の研究の課題である。

以上の通り、近年開発されてきた新しい統計手法を取り入れながら、古典的なニッチ理論に基づく解析を行い、侵入生物や森林被害という応用的な側面の評価を行った点で、審査委員一同は本論文が博士(農学)の学位論文として価値あるものと認めた。