衫木的地理變異，是由於生態環境多樣，加以地理及生殖隔離，及長達兩千年的人工栽培歷史所促成，為達成適地適種，提高育種成效，實應將種源區加以劃分，前人研究曾將大陸杉木種源劃分為九個生態區，本文就林業試驗所在蓮華池及六龜兩個試點所觀察之生長變異及遺傳變異評鑑杉木種源區劃分之得失，證實杉木九個生態區之劃分對六龜試點早年生長變異之分析確實有效，但對蓮華池晚年生長變異則殊不理想，對同功酵素基因頻率及異質性之變異分析則完全無效，由於大陸所作生態區之劃分是根據杉木幼年之生長特徵，並未考慮到晚年之生長狀況，所以在六龜早年生長資料而言，生態區劃分勝於省區劃分，但在蓮華池晚年生長資料，省區劃分反而優於生態區劃分，再就分區後選取最優的兩區而言，省區劃分可能有較高的選種增益，但生態區之劃分會有更穩定的入選者次序。
China fir shows significant geographic variation because it grows in a great diversity of ecological environments; it is subject to geographic and reproduction isolation; and it has been cultivated selectively and separately by many local farmers. In order to match provenances and planting sites for greater genetic gain, seed collection zones are to be delineated. Forest geneticists in mainland China have separated China-fir provenances into 9 ecotypes. Using growth and isozyme data obtained from 2 test plantations (Liukuei and Lienhuachih) we tested the effectiveness of such ecotype delineation. We found that the grouping of China-fir provenances into 9 ecotypes was most effective for the analysis of early growth at the Liukuei plantation. It was only moderately effective for analysis of later growth at the Lienhuachih plantation. The grouping of provenances was not useful for isozyme frequency or heterozygosity analysis. The different degrees of effectiveness may be possible because the construction of such ecotype classifications in China were based solely on early growth data; no isozyme data were used. Grouping provenances by ecotypes was superior to grouping by political provincial origins for early growth. On the other hand, provincial grouping seemed to be better than ecotype grouping for later growth. When the 2 best ecotypes or 2 best provinces were selected, provincial selection seemed to be able to yield a higher genetic gain than ecotype selection. However, the selection results and their rankings seemed to be more consistent in the ecotype division than in the provincial division.
