- 期刊
孟宗竹與银合歡人工林生物量之綜合關係式
Generalized Biomass Equations for Moso Bamboo and Leucaena Plantations
摘要
就作者己發表之生物量資料予以併合後，導出銀合歡與孟宗竹單株或林分之各部位生物量之關係式，圖供為生物量資源調查之依據。結果顯示此二森林作物單株各部位乾重均與胸徑具極顯著相關，銀合歡冠部乾重綜合預測式之變異係數高達70-73%，幹部及地上部乾重之變異性則較低，孟宗竹各部位之變異性均低於銀合歡。為了解此種變異之發生，各資料來源之生物量關係遂被詳盡解析。導入樹高為迥歸模式之自變數，並未降低關係式之變異係數，亦未提高決定係數或適合指數，故欲建立此2樹種之綜合預测式，僅以胸徑為自變數即已足夠。對林分之生物量關係而言，林分底面積雖與各部位生物量具顯著之相關，惟以林分平均樹高為第2自變數所獲致之統計利益相當顯著，銀合歡枝部生物量之估算則屬例外。基本上，銀合歡幹部及地上部生物量可由此2變數精確估算，葉部及枝部之變異性雖低於單株之生物量關係，然其適合指數及決定係數仍未甚滿意；孟宗竹稈部及地上部生物量之變異性雖略大於銀合歡，然各部位生物量關係式之變異係數均在9%-14%間，適合指數則位於0.81-0.94间。就實用之觀點而言，上述之生物量關係，可用於其他地區具類似林分結構者，幹部及地上部生物量概略估算之用，然對任一特定生育地應用此一般性關係式，必須慮及潛在之偏誤，尤以葉部及枝部為然。
並列摘要
Prediction equations for the biomass of individual tress and stands of Phyllostachys pubescens and Leucaena leucocephala were developed and compared. The data upon which the equations are drawn were from published biomass studies which mostly conducted by the authors. Three precision statistics including the coefficient of variation (CV), coefficient of determination (R^2) and fit index (FI) were used to evaluate the accuracy of the regression models. Results indicated that the individual stem and above-ground portion weight can be accurately predicted by diameter at breast height alone for both species. Equations developed from the combined data were less reliable for branch and leaf weight, especially for leucaena. To reveal the source of variation, relationships established from the original data for each sites were compared. The addition of tree height improved very little on the accuracy of the regression equations for all components for both species. This marginal effect of tree height in the prediction of individual biomass were disscussed. The biomass stand components were also related either to the stand basal area alone or in combination with the stand height From the comparisons among CV, FI and R^2, indicated that the statistical benefits by adding stand height as the second predictor were highly significant, except for the branch portion of leucaena. For the stem and above-ground biomass estimation, advantages of adding stand height to the prediction models were more obvious for leucaena than those of the moso bamboo. From the pratical viewpoint, biomass production is more likely to be estimated from stand variables without the laborious computation. The generalized biomass equations presented here might be applied on a regional level for stands with similar structure defined in this paper. However, the potential bias attributed to the site specification must be carefully evaluated before applying to a specific target population, especially for the estimation of leaf and branch biomass.