台灣森林的葉全緣率與年均溫、降水之關聯

Leaf Margins and Their Relationships with Annual Temperature and Precipitation in Forests of Taiwan

田婕(Chieh Tien)

指導教授 ： 丁宗蘇

國立臺灣大學/生物資源暨農學院/森林環境暨資源學研究所/碩士(2011年)

https://doi.org/10.6342/NTU.2011.02659

摘要

葉緣分析是探討植物相的物種全緣率與年均溫之正相關的單變量分析，常用於推估古氣溫，或延伸探討葉緣形態與其他環境條件的關聯。本研究利用台灣26篇已發表的植群文獻調查之844個原生闊葉林樣區，配合年均溫、年降水、最冷季降水以及最暖季降水資料，進行全台灣及其六個地理氣候區的葉全緣率與年均溫及降水的迴歸分析，並比較在樣區尺度、平均樣區尺度以及區域尺度下的葉緣分析結果，結果顯示：(1)在樣區尺度下，全緣率與年均溫皆呈極顯著正相關；全緣率與部份降水因子亦有顯著相關，惟相關度並不高，對於多變量迴歸結果貢獻亦不大。(2)各空間尺度的葉緣分析均顯示全緣率與年均溫成正相關。然而在樣區尺度時，由於樣區面積太小，分布也不甚平均，導致迴歸式之相關度並不高。以集成樣區計算平均全緣率可以避免前述問題，但全緣率有隨集成樣區擴大而遞減的趨勢，須至少以15個文獻樣區為集成樣區（面積約0.5 ha），迴歸分析結果才趨於穩定。(3)僅以物種分布極端值來推估之區域尺度資料，其葉緣分析結果無異於大規模集成樣區，因此葉緣分析應可借助標本資料，增加分析物種數，同時解決樣區不足的問題。(4)由於台灣具特殊的生態環境與植物播遷歷史，台灣各尺度葉緣分析迴歸式都顯著異於國外其他已發表文獻。(5)除資料不足的東南區外，各分區的全緣率與年均溫均有顯著正相關，且東北、西北、中西區的迴歸式斜率顯著小於西南區及東區，然而分區樣區數不足及群落分布易干擾分析結果，實際的生態差異仍需進一步分析。綜合以上，台灣森林的植物葉全緣率與年均溫及降水因子的關聯深具研究潛力，可提供目前還十分缺乏的東亞葉緣分析之相關資訊。

關鍵字

迴歸分析 ； 溫度梯度 ； 海拔 ； 鋸齒葉 ； 台灣

並列摘要

Leaf-margin analysis (LMA) is an univariate regression analysis based on the positive correlation between the ratio of woody dicotyledonous species with entire leaf margins and mean annual temperature, which is often applied to estimate paleo-climate. I used data of 844 plots from 26 native vegetation investigation reports in Taiwan and four climatic variables, mean annual temperature (MAT), annual precipitation, and precipitation of coldest and warmest season, to analyze the ratio of entire leave species (REL) and its correlation with MAT and precipitation in Taiwan and its six biogeoclimatic regions at habitat, mean habitat, and regional scales. The results indicate that: (1) at habitat scale, REL had significantly positive correlation with MAT and some precipitation factors, but the correlations with precipitation factors were much weaker. (2) REL was positively correlated with MAT at all scales. At habitat scale, because the plots were with small area and uneven distribution, REL of plots showed great variation and had weak correlation with MAT. Compiling data from multiple plots, preferably at least 15 plots (about 0.5 ha in total area), could yield consistent results. (3) The LMA regression model from larger multiple plots data was similar with regional scale LMA, indicating that using extreme distribution data could be a reliable data source for LMA. Therefore, using specimen information could compensate the shortcomings caused by insufficient and uneven sampling. (4) LMA regression models in this study all significantly differed from results in other countries. This may be due to the unique ecological environment and plant dispersal history in Taiwan. (5) At habitat scale, MAT was significantly positive correlated with REL in all biogeoclimatic regions. But the Northeast, Northwest and Central-west regions had significantly lower slopes of regression than Southwest and East regions. Due to uneven and limited plots in each region, further studies are needed to examine the detail differences among the biogeoclimatic regions. In conclusion, LMA has high potential in Taiwan and results will fill the current information void in East Asia.

並列關鍵字

Regression analysis ； temperature gradient ； altitude ； toothed leaves ； Taiwan

參考文獻

丁憲中、應紹舜 (2005) 西勢水庫集水區植群分析之研究，國立臺灣大學生物資源暨農學院實驗林研究報告，19(1):1–16。

林建融 ( 2009) 台灣植群多樣性組成及分布之探討，國立台灣大學森林學研究所碩士論文，127頁。

邱清安、林鴻志、廖敏君、曾彥學、歐辰雄、呂金誠、曾喜育 (2008) 臺灣潛在植群形相分類方案，林業研究季刊，30(4):89–112。

陳和田、李金玲、鄭忠財、呂福原 (2009) 沙里仙溪北玉山植群之研究，中華林學季刊，42:1–22。

陳俊銘 (2004) 臺灣東北部北勢溪上游之植群分析，國立台灣大學森林學研究所碩士論文，92頁。