The yield in cereals is determined by the yield components, panicle number, spikelet number per panicle, percentage of ripened grain and 1000-grain weight, and the achievement for high yield in rice is necessary and effectively through the improvement in these yield components. However, the relations among the components always have the strains each other on the achieving high yields, especially the relation between the panicle number (tiller number) and grain (spikelet) number per panicle. And spikelet number per panicle would influence on the grain ripening and single grain weight. So it is essential that to clarify the relations among the yield components, specially the relations between the tiller number and the spikelet number per panicle before heading stage since the panicle number per m2 and spikelet number per panicle have been determined before anthesis. The spikelet number per panicle is the difference between differentiated spikelet number per panicle and aborted spikelet number per panicle before anthesis. The increase of differentiated spikelet number per panicle and decrease of aborted spikelet number per panicle are necessary to increase the spikelet number per panicle. So the response of differentiated or aborted spikelet number per panicle to panicle number should be studied in rice. On the other side, the variation of spikelet number per panicle within plant is another factor influencing on the average spikelet number per panicle, since the differences about the spikelet number per panicle among the tillers are exist. In this thesis, the determination of spikelet number per panicle in relation to panicle number was studied from (1) the response of spikelet number per panicle on each panicle within plant to panicle number variation; (2) the response of spikelet number per panicle on each panicle within plant to transplanting density; (3) the effects of spikelet number per panicle on the grain yield and yield components

Chapter 1 The spikelet number per panicle and its variation within plant in 16 rice cultivars

The average spikelet number per panicle (SPP), differentiated spikelet number per panicle (D-SPP), and preflowering aborted spikelet number per panicle (A-SPP) in a plant, and their variations within plant on the each panicle were investigated in 16 rice cultivars in 2005 and 2006. There was magnificent genetic difference about SPP, D-SPP, and A-SPP on both averages and on main stems. The cultivars with larger panicle on main stems always had larger panicle on average in SPP, D-SPP and A-SPP, indicating that there were closely positive relations between the main stem and tillers about SPP, D-SPP, and A-SPP. The negative relationships were observed in panicle number hill -1 and SPP, D-SPP, and A-SPP. According to the panicle order, relative SPP on tillers to that panicle on main stem, D-SPP were declined, however, the spikelet abortion percentages were increased. The SPP and D-SPP reduced on average to 35% and 46% of respective main stem panicles. Panicles on tillers showed greater variation in SPP than D-SPP, due to the high variation in A-SPP. Larger SPP on main stem increased the variation of SPP within hill, but not D-SPP, and the lower A-SPP on main stem reduced the variability of A-SPP in tillers. This indicated that the assimilates supplying relations among the tillers would affect on the spikelet number per panicle.

Chapter 2 The response of spikelet number per panicle to transplanting density and its influence on yield in rice

Spikelet number per panicle (SPP), differentiated spikelet number per panicle (D-SPP), and preflowering aborted spikelet number per panicle (A-SPP) were examined in five rice cultivars (Akihikari, IRAT109, Nipponbare, Akenohoshi, IR65564-44-51 (NPT65)) under three planting densities (HD; high, MD; medium, LD; low planting density) in the field condition. Rice plants at LD produced a higher panicle number per plant but lower panicle number per unit area, accompanied by higher D-SPP and SPP, on average. A-SPP and the ratio of A-SPP to D-SPP (A%) showed no consistent trends. There was a broader range of D-SPP values at LD than at HD because of higher D-SPP in lower order panicles. D-SPP declined as panicle order increased in all cultivars and years, whereas A% increased. D-SPP and SPP of each panicle were positively correlated with tiller size (tiller height, leaf area, and neck internode diameter).

Spikelet production efficiency for D-SPP or for SPP (D-SPP or SPP per leaf area) of each tiller was higher in NPT65 and Akihikari than other cultivars, indicating a greater capacity of tillers to produce spikelet or support spikelet growth. In each cultivar except NPT65, spikelet production efficiency for D-SPP increased as panicle order increased, whereas spikelet production efficiency for SPP remained constant or decreased. This finding indicates that, irrespective of planting density, higher order panicles produce more spikelets than they can afford physiologically, but they were regulated downward to a nearly constant value in four cultivars. In NPT65, different from other cultivars, spikelet production efficiency for D-SPP decreased with panicle order increase.

Spikelet number per panicle was larger in LD than in HD. This was because of the larger tiller size in leaf area, shoot dry matter in LD than in HD. There was clear compensation of spikelet number per panicle increase to the panicle number m-2 reduction. So the spikelet number per m2 and grain number per m2 kept stable on varying transplanting density. The filled grain percentage was constant. There was a little higher yield production in HD than in LD, because of higher 1000 grain weight. The yield and its components showed clearly higher on primary rachis branch than on secondary rachis branches. As a conclusion, although the grain number per m2 kept constant, the more grain number on the secondary rachis branch should be responsible for the lower yield in low transplanting density.

Chapter 3 The response of spikelet number per panicle and yield to transplanting density with root restriction in rice

As the indication of Chapter 2, the smaller tiller size in terms of tiller height, leaf area per tiller and dry matter per tiller due to the dense planting density resulted into the smaller spikelet number per panicle in high transplanting density (HD) than in low transplanting density (LD), so the possible root function in this response was studied by the root restriction treatment (RRT), a horizontal restriction treatment on root rhizosphere size. The spikelet number per panicle, yield and nitrogen accumulation were examined under two planting density with RRT using two Japonica cultivars, Akihikari and IRAT109, in 2007 and 2008. The above ground dry matter, yield, nitrogen accumulation per unit area were not differed in both planting density without RRT. However, RRT reduced above ground dry matter, yield, nitrogen accumulation, and panicle number per unit area evidently in LD, though the reductions were very weak in HD. Spikelet number per panicle was decreased significantly by RRT in LD than in HD, through the reduction in differentiated spikelet number per panicle (D-SPP). These reductions of above ground dry matter, yield and spikelet number per panicle were elucidated by the amount of nitrogen accumulation. Therefore, it is indicated that the effects of planting density on spikelet number per panicle and yield were by the nitrogen accumulation with below ground parts than the light shading on aerial parts. The different responses were observed in two cultivars in yield related characteristics in LD with RRT indicated that the strategies to limited nitrogen availability were dependent on cultivars.

Chapter 4 QTLs analysis for spikelet number per panicle under two nitrogen conditions

As the indications from Chapter 2 and 3, the planting density had effects on the spikelet per panicle, mainly through effecting on the nitrogen accumulations from soil. And also, in Chapter 3, it was verified that there was genetic difference in the response of spikelet per panicle to nitrogen condition in the soil. So in this Chapter, by using BC1F8 and BC1F9 of a 105BILs, inbred lines from a cross of Akihikari (temperate Japonica) × IRAT109 (tropical Japonica), QTLs for spikelet number per panicle (SPP), and their components: primary rachis branch per panicle (BPP), spikelet number per primary rachis branch (SPB), and spikelet abortion percentage before flowering (A%) were identified under two nitrogen application conditions, low nitrogen (LN) and high nitrogen (HN), in 2006 and 2007. The results showed that 11 QTLs and 12 QTLs totally were detected in LN and HN in two years. Among them, QTLs for SPP(3), BPP(7), SPB(2) and A% (6) were identified. Stable QTLs for SPP(2), BPP(1), and A% (2) were detected in two years under both nitrogen conditions, therefore, control mechanism of them were common under different nitrogen conditions. Co-location of QTLs on chromosome 6 was observed which could indicate strong relation between SPP and SPB under high nitrogen applied condition.

Therefore, in this study, it was shown clearly that (1) The spikelet number per panicle on each panicle within plant were declined with the increase of panicle orders. This is because of the decline of the leaf area, neck internode diameters and tiller height: (2) The spikelet production efficiency (spikelet number per leaf area) of each tiller within plant for D-SPP increased as panicle order increased, whereas spikelet production efficiency for SPP remained constant or decreased, with irrespective variation of tiller number per hill variation due to the planting density. Higher order panicles produce more spikelets than they can afford physiologically, but they were regulated downward to a nearly constant value: (3) Panicle number per hill had positive effects on the spikelet number per panicle through the differentiated spikelet number per panicle. This is mainly through the influence of the nitrogen accumulation by roots from soil: (3) The genetic control mechanism of spikelet number per panicle response to nitrogen were common under different nitrogen application conditions.

イネの収量を決める収量構成要素(単位面積あたり株数、株あたり穂数、一穂穎花数、登熟歩合、粒重)は、一方が増加すると他方が減少するといった相補的な関係にあることが知られているが、品種や栽培条件の差異によって相互の関係が異なることが予想される。高収量を得るためにはこの相互関係を明らかにし、最適条件を整えることが必須である。本研究では、収量構成要素のうち一穂穎花数と穂数を、栽培条件として移植密度を取り上げ、始めに同じ栽培条件での品種による一穂穎花数と穂数との相互関係の違い(第1章)を明らかにした上で、栽植密度の変更に伴うこの相互関係の変動(第2章)を解析した。さらに栽植密度は根域の大きさを変えるため、根域制限がこれらの関係に及ぼす影響について解析した(第3章)。その中で、窒素吸収量がそれらの関係を規定することが示唆されたため、窒素施肥量の異なる条件で栽培した組換え近交系を用いて量的形質遺伝子座(QTL)解析を行った(第4章)。一穂穎花数については、一穂分化穎花数=一穂生存穎花数+一穂退化穎化数、および、退化穎化率=一穂退化穎化数/一穂分化穎花数として、分化と退化を分けて検討を行った。

第1章では、16品種を供試し、個体内の全穂について調査した。その結果、一穂生存穎花数には大きな品種間差異があり、一穂生存穎花数の多い品種は概ね穂数が少なかった。また、一穂生存穎花数が多く穂数の少ない品種ほど個体内における一穂生存穎花数の変動係数が大きく、すなわち穂数が少ないほど個体内の大きい穂と小さい穂の一穂生存穎花数の差が大きかった。また、いずれの品種においても、個体内の一穂分化穎花数が多い大きい穂ほど退化穎化率が低く、小さい穂ほど穎花退化率が高かった。

第2章では、5品種を供試し、3段階の栽植密度(高密度15×15cm、中密度15×30cm、低密度30×30cm)で3年間栽培して一穂穎花数と穂数の相互関係について調査した。その結果、いずれの品種においても、栽植密度が低いほど個体あたり穂数の増加と共に、一穂生存穎花数の増加が認められた。各々の増加程度には品種間差異があり、穂数の増加が顕著な品種と一穂生存頴花数の増加が顕著な品種とがあった。一穂生存穎花数の増加は、特に個体内の大きい穂で顕著であった。一穂分化穎花数は茎の大きさ(草丈、葉面積、穂首かん径)と正の相関関係にあることから、穎花生産効率(一穂分化穎花数/葉面積または一穂生存穎花数/葉面積)を算出したところ、同一品種内では、小さい穂ほど一穂分化穎花数の穎花生産効率が高いが、一穂生存穎花数に対する穎花生産効率は穂の大きさに関係なくほぼ一定であることが示された。このことは各品種内の小さい茎が、その後の成長を保証できる穎花数よりも多くの穎花を分化し、その結果退化穎花率が高くなることを示している。また、品種間で比較すると、一穂生存穎花数の穎花生産効率には品種間差異があった。したがって、第1、2章の結果から、低密度で一穂生存穎花数を増やすためには、穂数が多く、低密度で穂数が増加しやすく、穎花生産効率が高いという特性が有効であることが示された。

栽植密度の差異は、個体の根域の違いに反映するため、第3章では根域を制限して2品種(アキヒカリとIRAT109)を栽培し、地下部の影響を解析した。その結果、低密度区で根域のみを高密度区と同様にすると、穂数、一穂分化穎花数、地上部重、収量が根域制限をしない区より著しく低下し、それは、窒素吸収量が低下したことによっていた。この根域制限に対する反応には品種間差異が認められ、アキヒカリでは穂数の減少が、IRAT109では一穂分化穎花数の減少が大きかった。

第2、3章で、栽植密度に対する一穂分化穎花数と穂数の反応が品種間で異なること、およびそれらが窒素吸収量を介していることが示されたため、第4章では窒素施肥量の異なる2条件で栽培したアキヒカリ×IRAT109戻し交雑由来組換え近交系105系統を用いてQTL解析を行った。その結果、一穂穎花数について3つのQTLが、またそのうち2つは窒素条件・年次に関わらず検出され、2つとも既知の穂数のQTLと同一の領域であった。したがって、このゲノム領域は窒素条件に関わらず一穂穎花数と穂数を制御していることが示された。窒素条件に特異的なQTLは検出されなかった。

以上の通り、本論文は一穂穎花数と収量およびその穂数との関係における品種間差異および栽植密度の影響について、分化穎花と退化穎花に分けて解析し、学術上の新たな知見を得た。これらは、一穂穎花数と穂数を基にした品種ごとの栽培条件の最適化において応用上の価値が高いと判断される。したがって、審査委員一同は、本論文は学術上および応用上価値が高く、よって、博士(農学)の学位論文として価値あるものと認めた。