Pinus merkusii is the only pine that distributes across the equator and has a feature of year round reproduction. It is one of the important tree species in Indonesia for wood materials and resin production. Its genetic improvement was started in the mid 1970th with the establishment of seedling seed orchard (SSO) by planting selected families from artificial plantations in thirteen districts of Java Island, Indonesia.

Panmictic equilibrium is a key factor to make a breeding strategy in success by transforming parental genetic diversity to the filial generation. Moreover, synchronization of flowering among parent trees affects on quantity and quality of produced seed by influencing mating system. Thus, study on flowering characteristics of P. merkusii is important to ensure the sustained production of high quality seed crops under its distinguishing flowering character. However, there was no previous study dealing with the relationship between mating system and genetic diversity of P. merkusii.

This study aimed (1) to assess the status of mating, the level of genetic diversities maintained in the parental population and in the offspring population were determined and (2) to clarify the seasonal variations in the mating system that may influence on the productivity and the genetic diversity of the seed crops, seasonal differences in flowering and mating were assessed in the SSO. All the researches in this thesis were based on the data and materials from the SSO at Jember, East Java, Indonesia.

This thesis was composed by six chapters. The first chapter is a general introduction, where background of this study, taxonomic and biological features of P. merkusii, reviews on genetic diversity and mating system in seed orchards, and purposes of this study were described. A breeding plan for P. merkusii in Indonesia and the SSO used in this thesis were explained in Chapter 2.

In Chapter 3, development of microsatellite markers for P. merkusii to study genetic diversity and mating system was described. Ten microsatellite markers were successfully isolated by a dual-suppression-PCR technique. Of ten microsatellite loci isolated, five were codominant and polymorphic, two were monomorphic, two were multiband and one locus was not amplified. The number of alleles observed for each locus ranged from 3 to 6. The value of HE ranged from 0.389 to 0.728. Three loci were deviated significantly from HWE, due to an excess of homozygotes. Null-alleles were expected to be present at these loci. On the other hand, closely related individuals present in SSO may also be responsible for these HWE deviations. No evidence for linkage disequilibrium was detected for these loci.

In Chapter 4, the level of genetic diversity maintained in the SSO and in the produced offspring was determined using five microsatellite markers. And the mating status in the SSO was discussed based on the results.

The genetic diversity parameters were quite similar among districts where mother trees in the SSO were selected, and FST value among districts was very low (0.008). These results indicate that there was a high degree of genetic uniformity among districts of parent sources. Our data support the suggestion in a previous study that the trees in the SSO originate from materials with limited genetic diversity.

Allelic richness was higher in the offspring populations (mean=3.90) than in the parental populations (mean=3.48) in every block. HO and HE were higher in the offspring (mean=0.601 and 0.545, respectively) than in the parental populations (mean=0.471 and 0.489, respectively) in most of the blocks. The FIS values did not significantly deviate from zero both in parental and offspring population. Significant linkage disequilibrium was only observed in the parental populations (mean=2.00). The effective population size calculated was higher in the offspring populations (mean=606.22) than in the parental populations (mean=486.53). The FST values among parental populations and offspring populations were 0.038 and 0.029 respectively. These findings suggest that seed production in the SSO may be nearly panmictic and that genetic exchange freely occurs among the parent trees. Thus, high genetic diversity is still maintained in the offspring generation, in spite of the genetic similarity of trees in the SSO.

Seasonal variations in mating system in the SSO were studied in Chapter 5, which was composed by two sub chapters.

In the first sub chapter, number of female and male flowers were assessed in the three different seasons, abundant female flowers but few male flowers in August (62.2 vs. 4.5), similar numbers of female and male flowers were observed in November (37.9 vs. 28.4) and low numbers of both female and male flowers in March (35.3 vs. 5.4). The numbers of pollen grains observed were low in August (1.2±0.7) and March (1.3 ±1.5) and much higher in November (3.4±6.1). These results show that the abundance of pollen roughly corresponded to the frequencies of male flowers. The number of seeds per cone was higher in March than in both November and August. The number of filled seeds per cone was high in both November and March, but low in August. Since disproportionate flowerings can lead to insufficient pollen, and thus adversely affect the number of filled seeds by increasing the selfing rate. Insufficient pollen lead poor quality pollen grains can make substantial contributions to fertilization, thereby yielding weak progeny.

Among progeny populations, NA values were similar in November (21) and March (22) and highest in August (27). HO was higher in August (0.384) than in both November (0.349) and March (0.314), while HE was similar in August (0.402) and November (0.397), but low in March (0.361). The FIS was lower in August (0.046) than in November (0.120) and March (0.131). The L-D was highest in August (6) and similar between November (2) and March (2). Insufficient pollen may cause substantial rates of pollination from distant pollen, which may explain higher NA and the HE value in the August was higher than for both November and March. Another caution in insufficient pollen population was an opportunity of selfing. Then we suggested that high value of significant L-D in August might indicate the occurrences of non-random mating. In November, near-panmictic equilibrium and minimum selfing rates could be expected due to balance in number female vs. male flowers and sufficient pollen. Thus, value of HE in November was high. In March, limitations in both female and male flowering may lead restricted pollen dispersal contributed to the lowest level of HE and hence high FIS. Moreover, the pairwise FST values between parent population and offspring populations were significantly different; the values were 0.034, 0.031 and 0.054 in August, November and March, respectively. Low values of FST may confirm the occurrences of extent gene flow mating in August and nearly panmictic in November.

In the next sub chapter, seasonal patterns of mating system and gene flow in each season were confirmed using three models: a radius-based analysis, mixed-mating model and neighborhood model.

A radius-based analysis revealed the FIS values were insignificant at radii up to 108 m, 36m and 12m for the mating in August, November and March, respectively. The L-D values were high in August (1-6), and low in November and March (1-2 in both cases). Changes in genetic differentiation (pairwise FST) between offspring population in radius 12m and parental populations in different radius were high in August and March, but low in November. The occurrence of substantial fertilization from distant pollen that suggested for the mating in August could be confirmed by distant radius size to get significant FIS. High selfing could be confirmed by high value of L-D. Mating among surrounding trees in November could be confirmed by moderate radius size to get significant FIS. Low value of FST could explain panmictic equilibrium in November. Restriction on gene flow for March mating could be confirmed by lowest distant to obtain significant value of FIS and high value of FST.

Mixed-mating model showed that the mean multilocus outcrossing rate (tm) was higher in November (0.962) than in both August (0.941) and March (0.956). Correlation of paternity (rp) was highest for the August progeny (0.304) and moderate for both the November (0.205) and March progenies (0.193). Insufficient pollen in August may cause low tm value and high rp, in contrast sufficient pollen in November and March might cause high tm and low rp. Lowest value of tm indicated high selfing in August. Indeed, highest value of rp showed the insufficient pollen. High tm and low rp in November and March represented the sufficient pollen in both seasons.

One of the advantages of Neighborhood model is describing effect of migratory pollen (m). Estimated selfing rates obtained by the mixed-mating model and neighborhood model were comparable. Selfing was higher in August (s=0.06) than in November (0.02) and March (0.04). Substantial fertilization from distant pollen in August could be explained by highest value of immigration pollen rate (m=0.03) than November (0.01) and March (0.02). Panmictic equilibrium in November could be confirmed by higher value of effective population size (71.42%) than August (54.08%) and March (24.68%).

Using these three methods, we succeeded in describing the real seasonal mating system in the SSO which depend on their flowering characteristics.

In Chapter 6, general discussion was conducted based on the results and some implications for the genetic improvement of P. merkusii were proposed.

While the parent trees in the SSO were genetically similar among seed sources, they kept moderate level of diversity. The genetic diversity in the SSO was maintained in offspring populations by nearly panmictic mating and comparatively high gene flows. However, temporal fluctuation of flowering greatly affected seed production and gene diversity of offsprings. Disproportionate flowerings might lead insufficient pollens that make weak pollen competition. In contrast, balanced number of female and male flowers might promote panmictic equilibrium. Differences in flowering characteristics influenced the mating system and patterns gene flow in spatial area in the SSO that caused differences in seed production and their genetic diversity among seasons. However, since parent trees in the SSO tend to have high outcrossing rate, so high genetic diversity of the seed crops from the SSO population may be expected for all mating season such had been examined in Chapter 4. The results in this study indicated that flowering and mating status should be considered for good seed production. In addition, reproductive phenology should be included in the criteria to consider in the layout and design of future seed orchard to enhance panmixia. The study on reproductive process of P. merkusii should be employed more frequent to obtain more accurate estimations on mating system in the SSO.

Pinus merkusii は唯一南半球にまで分布するマツであり、通年開花という特徴的な繁殖様式を持つ。本種は、インドネシアでは木材および樹脂生産において最も重要な樹種の一つであり、現在、その種子のほとんどが1970年代の半ばから設定された実生採種園より生産されている。育種の成否は、採種園における任意交配による親世代の遺伝的多様性を次世代へと伝達させることができるかにかっている。また、採種木間の開花の同調性は、交配様式に影響を与えることで、生産される種子の質と量を大きく左右する。従って、P. merkusiiの開花特性およびそれによってもたらされる交配様式を明らかにすることは、継続的な育種と高い質の種子の生産のために極めて重要な課題である。しかし、これまで本種の開花特性および交配様式と遺伝的多様性の関係について明らかにした例はない。そこで、実生採種園における交配の現状および種子の遺伝的多様性と生産性に開花および交配の季節変動が与える影響について研究した。

本論文は6章から構成されている。第I章は序論として、本研究の背景、P. merkusiiの分類学及び生物学的特徴、実生採種園における遺伝的多様性及び交配様式についてレビューし、本研究の目的を述べた。第II章ではインドネシアにおける、実生採種園を利用した育種計画について述べた。第III章ではマイクロサテライトマーカーの開発を行った。第IV章では、 実生採種園内の成木及び実生の遺伝的多様性を明らかにし、実生採種園における交配実態の概況を明らかにした。第V章では、実生採種園の交配様式の季節変動が遺伝的多様性に及ぼす影響について考察した。第VI章では、研究の結果から導き出されるP. merkusiiの特徴および実生採種園管理についての提案を述べた。以下に研究の概要を述べる。

第III章では、dual-suppression-PCR法を用いて10のマイクロサテライト遺伝子座の分離に成功した。そのうち十分な多型性が確認できた5座をマーカー化して以下の研究に供した。

第IV章では、まず、実生採種園内の親木集団の遺伝的多様性について検討した結果、遺伝的多様性は中庸であり必ずしも低くなかった。次に、親木の選抜地間の遺伝的分化程度は非常に低く、選抜対象となった人工林が同一の天然林由来であるという、既往の報告と矛盾しなかった。次に、親木集団と種子集団との比較を行った結果、親木集団よりも種子集団で遺伝的多様性は高く、調査ブロック間の遺伝的分化度は低かった。これらは、採種園内では概ね任意交配が生じていることを示唆している。

第V章では、まず、3つの異なる季節における雌花、雄花の数および飛散花粉量を調査した結果、雌雄花量および雌雄花の比率は季節により大きく異なること、また飛散花粉量は概ね雄花量に対応して増減することが認められた。次に、それぞれの季節に開花した雌花が結実して出来た球果について種子の品質を比較した結果、球果あたりの種子数はおよび有胚種子数は季節により変動し、雌花および雄花数の変動が、種子の品質に大きく影響することが明らかになった。さらに、季節毎の種子の遺伝的多様性も大きく変動し、花粉量の不足が、遠距離からの花粉の飛散を誘導し、かつ自殖率にも影響すること、バランスの良い雌雄花の着生は、低い自殖率と任意交配を導くこと、少ない雌雄花が花粉飛散の制限を導くことなどが示唆された。

次いで、各シーズンにおける交配様式と花粉流動について3つのモデル(a Radius-based analysis, Mixed-mating model and Neighborhood model)から検証した。独自に考案したRadius-based analysisでは、遺伝的多様性の直接的な比較から任意交配範囲を推定したが、各季節の着花傾向を反映してその大きさは季節ごとに大きく異なった。Mixed-mating modelを用いて、多遺伝子座における他殖率(tm)および花粉親の同一度(rp)を推定した結果も、季節ごとの着花傾向を反映して変化し、花粉量の不足は低いtm及び高いrpを、十分な花粉は高い及tmび低いrpを導いた。さらに、Neighborhood modelにより範囲外からの花粉の流入を検出した結果、雌花過多の季節における遠距離からの花粉流入が確認された。すなわち、これらの3方法を用いることで、開花の季節変動に伴った交配様式の変動について詳述することができた。

以上、本論文は、P. merkusii実生採種園における季節間の開花特性の違いは、交配様式と花粉流動に影響を与え、種子の生産性及び遺伝的多様性に大きな影響を与えていることを実証し、また、その機構の一端を解明した。さらに、さまざまな開花時期に生じた多様な交配から産み出された種子が混合することによって、種子集団における遺伝的多様性が維持されることを示した。これらの知見は、P. merkusiiの種特性の解明および育種計画の推進に大きく貢献するものであり、学術上、応用上資するところが少なくない。よって審査委員一同は、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。