頁籤選單縮合
題名 | 藉品種與環境交感作用之大小修訂臺灣水稻育種程序之研究=Studies on the Variety Environment Interaction of Rice Regional Trial in Taiwan and Their Inplications on Breeding Program |
---|---|
作者 | 張魯智; Chang, Lu-chih; |
期刊 | 中華農學會報 |
出版日期 | 19680900 |
卷期 | 63 民57.09 |
頁次 | 頁1-34 |
語文 | chi |
關鍵詞 | 品種環境交感作用; 水稻; 育種; |
中文摘要 | (一)本研究之目的:在利用臺灣歷年水稻品種比較區域試驗結果記錄之大量資料,經若干特殊之統計分析及估算推計,顯示此類試驗之年度數、區域數及重複次數不同組合,對試驗結論準確度之影響,指出最佳之組合,檢討現行組合之得失,供本省今後水稻育種工作計畫改進之參考。 (二)本研究之方法:首先向全省各有關機關徵集歷年區域試驗之結果記錄,分組合併進行綜合變方分析,利用下列各式估算多種成分變方: σ2αλ=S2αλ=1/SLn(MSYLV-MSYLV)……(1) σ2ωλ=S2ωλ=1/YSn(MSLV-MSYV)……(2) σ2αωλ=S2αωλ=1/Sn(MSYLV-MSE)……(3) σ2=S2=MSE……(4) 然後再以下式推算各種不同組合之品種總平均值的均方: S2x=S2αλ/Y+S2ωλ/L+S2αωλ/YL+S2/YSLn……(5) 根據上式求得之多個S2x值批評各種組合之好壞,及討論本省現行組合之得失。 (三)經徵集結果,得六組合併資料各項情形及求得之多種均方列如下表17。觀表可知六組試驗前後經13年之久,每組兩年兩期作,第六組則三年六期作,同時舉行試驗之區域數由L=8至L=13不等,平均為10強,六組包括264個單獨試驗,參試品種數V=8至V==16不等,重複次數n=6或n=5,由六組綜合變方分析結果求得MSE及MSYV最小,MSYLV次之,MSLV最大。 (四)將各種均方代入(1)(2)(3)(4)(5)五式,求得各種組合之S2x值列為附表3,由該表中所列之各個S2x值比觀,可知區域數有最大之影響力,年度數次之,重複次數最小,至於年度數、區域數及重複次數三者對吾人經濟生活之關係,作者個人意見,以年度數最重要,區域數及重複次數較不重要,二者想仿,似此,則年度數最少(一年)區域數最多(L=18-25)及重複次數亦少(2-4次)之組合方式佔優勢。 (五)作者另將六組合併資料每兩個品種總平均值比較得448個相差,並以此等相差作一頻度分配表,求得累積頻度百分率,倘顯著機率及累積頻度百分率為試驗者所指定?可查表得理論t值然後代入下式求得夠小之S2x,作挑選適當組合方式之標準:S2x=d2/2t2……(6) (六)由研究實際資料所曾獲得之提示:臺灣現行之組合方式,一因二或三年嫌太費時,且有浪費人力物力之弊,二因區域數太少,以致試驗結論之準確度不高,三因重複次數太多。每一單獨試驗之試區數太多,易生錯誤,亦有徒勞無功之嫌,故應有所修訂。作者建議:以空間換時間在顯著機率固定為5%之場合作,設顯著相差之累積頻度為40%,則YSLn=R為1×2×21×4=168 (S2x=3,603.65),1×2×23×3=138 (S2x=3,524.90),1×2×22×4=176 (S2x=3,451.54),1×2×24×3=144 (S2x=3,388.74)等四個組合均甚優,另設顯著相差之累積頻度為30%,1×2×15×3=90 (S2x=5,267.71)及1×2×14×3=84 (S2x=5,625.62)二者均佳,倘提高顯著相差之累積頻度為50%,則1×2×35×6=420 (S2x=2,131.50)為較好組合,以此等組合與現行者比觀,則相異之處甚多(a)年度數減少1或2年,(b)區域數平均增加一倍,(c)重複次數減少1至3次,(d)總重複次數減少一半左右,現行組合經此修訂不但可以加速新品種之育成,並得節省人力物力及提高試驗結論之準確度。 (七)本省對水稻育種之概括範圍,係以全省環島所有稻田為同一空間整體,以今後十數年之水稻耕作為同一時間整體,且第一及第二期作品種不分為原則,本研究之概括範圍亦然。 |
英文摘要 | 1. The object of this paper is to find out different combinations of years locations and replications that affect the precision in the regional trials of rice. Some appropriate combinations are derived from all data (from fall 1950 to summer 1965) of rice varieties comparison experiments, through the aid of special statistical analysis and inference, and set out a reference for the improvement of rice breeding program in Taiwan. 2. The procedure to investigate this subject is to collect the data of rice varieties regional trials from different agricultural experiment stations for the application of combined analysis of variance. The following formulae are used: σ2αλ=S2αλ=1/SLn(MSYLV-MSYLV)……(1) σ2ωλ=S2ωλ=1/YLn(MSLV-MSYV)……(2) σ2αωλ=S2αωλ=1/Sn(MSYLV-MSE)……(3) σ2=S2=MSE……(4) where S2αλ and S2αλ are the estimates of the component variance of the first order interaction of year × variety (σ2αλ), location × variety (σ2ωλ) respectively; S2αωλ is the estimate of the component variance of second order interaction of year × location ×variety (σ2ωλ) S2 is the estimate of variance of error (σ2), and Y, S, L, n, represent the numbers of years, the number of rice cropings within a year, locations, and replications in the combined data respectively. Applying the above four equations and the formula below, the mean square of a variety mean of various combinations can be found easily S2x=S2αλ/Y+S2ωλ/L+S2αωλ/YL+S2/YSLn……(5) where S2x is the mean square of a variety mean and the other symbols are the same as above four equations, except that the bar stands for the mean. The value of every S2x so obtained can he applied for the justification of fitness of various combinations and thus, it is used to criticize whether the present breeding program is suitable. 3. The data are appropriately combined and divided into six groups. The required mean squares are tabulated as follow: From the above table, we know that the six groups of experimental data cover the period of thirteen years. Besides the sixth group contains three years six crops, each group contains two years (two crops per year). The number of locations in the trials varies from 8 to 13, and its mean is over 10. There are 264 individual experiments in these six groups, the number of participated varieties varies from 8 to 16, and the replication is either 5 or 6. By using the combined analysis of variance, the final results are obtained orderly that MSE and MSYV are the smallest, MSLV the largest and MSYLV in between. 4. Substituting all the mean squares into equations (1), (2), (3), (4) and (5) the mean squares of mean (S2x) are obtained and stated in appendix 3. If all S2x are compared in this appendix, it is obvious that the number of locations affects it most, the number of years the second, and the number of replications the least. In the author's openion, it is said, the number of years is the most important relation on our daily economical lives; the number of locations, the seeond/and the number of replications is not very important when compared to the first and second. As a matter of fact, in rice breeding program I wonld suggest a combination with least number of years (1 is enough) and replications (2-4 is enough), but a rather great many location number (18-25) is enough. The above combination of the three factors, years, location, and replication, seemed to me, the best of all. 5. The author have also compared the means of each pair varieties and got 448 differences from the six group data. These differences have been used to build up a frequency table. The accumulative percentage of the frequency has been calculated. If the probability of significance and accumulative frequency are assigned by the experimenter one can get the t-value from Fisher's table and the least values of mean squares of a variety mean (S2x) can be calculated by using the equation below. In refering to all the S2x, the appropriate combinations can he found. S2x=d2/2t2……(6) Where d is the difference of the two means of each pair of varieties. 6. Frorm the final result we obtained, it is evident that the rice breeding program recently in use is not good, for the combination of the three factors is not appropriatly combined, as in the first place, 2 or 3 years will not only waste the time and work, but also unnecessary expenses. Secondly, the number of location is too little to gain a accurate result. Thirdly, the number of replications is too large, since within an individual experiment, too many plots will cause the difficulty to handle and may in sometimes get a confused result. The author, thus, suggests that the "time" be in place of "space". Limit the significant probability at the point of 5% and suppose the difference of the accumulative frequency to be 40%, then from Table12. 2, the combinations YSLn=R: 1×2×21×4=168 (S2x=3603.65),1×2×23×3=138 (S2x=3524.90),1×2×22×4=176 (S2x=3451.54),1×2×24×3=144 (S2x=3388.74) are the best of combinations. Further more, let the significant difference of the accumulative frequency be 30%, the best combinations obtained from Table 12.5 are 1×2×15×3=90 (S2x=5267.71) and 1×2×14×3=84 (S2x=5625.62). If the significant difference of the accumulative frequency is increased to 50%; then, 1×2×35×6=420 (S2x=2,131.50) is the best combination. The above combinations as compared to the current ones, there have great differences: (a) The number of years is decreased to 1 or 2; (b) the number of locations is doubled; (c) the number of replications is reduced to 1 to 3; and (d) the total replications are reduced to about one-half. So if the current combination employed, could be adjustified, it would not only accelerate breed out of new varieties, but also diminish labour, financial expenses as well as precising the experimental results. 7. The rule of thumb in the rice breeding program in Taiwan is that the paddy fields in Taiwan are taken as a whole in the same "space", the same "time" the several decades that follow, and the unseparation of the first and second crops. The present studies are done under the same consideration. |
本系統之摘要資訊系依該期刊論文摘要之資訊為主。