由生態代謝理論之觀點推估近熱帶山地森林枯落物產量

Abstract

枯落物在森林生態系中是重要的代謝產物，因此了解由冠層進入森林地表之枯落產量，將有助於掌握森林中的養分及碳循環動態。然而，單以現地收集枯落物的方式需要耗費大量人力及時間，且目前本來就為數不多的枯落物模型所需的冠層參數 (例如：冠層厚度、冠層直徑及冠層高)，也常不利大面積、頻繁規律地收集。在生態代謝理論中，代謝速率與體型大小比例 (scaling) 關係使我們得以由總體生物量的大小與淨生產量之間的關係推估枯落物產量。相對於直接地測量或間接的模型估算枯落物產量，遙測技術的進步在估森林總體生物量上有較高的效率及準確度，因此以遙測方法所得之森林總生物量後，再推估枯落物產量，更可能全面的了解大範圍森林覆蓋之區域。而透過遙測技術進行的生態系監測通常皆須品質良好準確的地面資料進行經驗模型建立，因此本研究選擇有悠久林業歷史及台灣長期生態樣區之一，且是目前台灣現存最大之檜木林之棲蘭山地區作為研究區，並以已有現成生物量推估公式 (allometric equation) 之檜木 (Chamaecyparis obtusa var. formosana and Chamaecyparis formosensis) 做為目標物種，使用不易於高生物量地區受到飽和作用影響之遙測感測器──光達 (lidar)，將其取得之三維樹高模型進行進一步的生物量推估。我們架設了15個樣區共60個0.5 m^2的枯落物網，並測量樣區中共1128顆檜木的胸高斷徑 (diameter at breast height, DBH)，依據現地生物量與枯落物的關係為基準，使用生物量進行兩期：非干擾季與颱風干擾季之枯落物推估。兩季之枯落物與樣區內生物量皆有顯著相關 (R2 = 0.75, p < 0.05)，且發現隨著樣區內生物量增加，枯落物產量在樣區生物量約大於200 Mg ha-1後反而隨生物量增加而下降。最後我們使用3組現地收集之枯落物資料驗證此枯落物產量地圖並有相當高的相關性 (R2 = 0.98, p < 0.05, RMSE = 23.12 Mg ha^-1 month^-1)。本研究除了可使未來對於枯落物動態不再受限於單點式的推估，能更有效率的了解生態系中物質、養分循環之動態，也提供實證資料討論生態代謝理論中縮放關係 (scaling relationship) 於生態系面臨干擾下之適用性。

Litterfall plays a crucial role in the carbon and nutrient cycles of forest ecosystems. The amount of litterfall governs the amount of carbon and nutrient to be returned in a forest ecosystem. However, when it comes to quantifying forest litterfall, collection of canopy characteristic parameters for existing litterfall models is usually time-consuming and labor-intensive. Recent studies indicated that, in metabolic scaling theory, there is a common relationship between terrestrial plant production and biomass; a major part of the production is contributed by litterfall. Therefore, there could be a relationship between litterfall and biomass, which could facilitate large spatial scale estimation of litterfall since biomass may be assessed using remote sensing. To investigate this relationship, we acquired monthly litterfall of a hinoki (Chamaecyparis spp.) dominant montane forest in the northeastern Taiwan (23.98 N, 120.97 E) across the elevation range of 1267–2080 m a.s.l. Monthly litterfall data were recorded from fifteen 0.09 ha plots and each plot consisted from four randomly arranged 0.5 m2 litterfall traps. In addition, diameter at breast height of each live hinoki tree (n = 1,128) within all plots was measured and total biomass was derived using an in-situ species-specific allometry. Species-specific hinoki litterfall regression models were developed for two seasons (growing season: March–June, typhoon season: July–October) across a wide range of biomass density. We found that the relationship between biomass and litterfall might depend on the amount of total biomass. In both seasons, increase of litter production with increase of biomass density could be found when biomass is less than about 200 Mg ha-1, after that, inverse relationship appears. With the aid of high spatial resolution airborne light detection and ranging (lidar) data, we may be able to provide a spatial layer of hinoki biomass and map monthly litterfall over a vast region based on this biomass-litterfall relationship. At the end, the litterfall map were verified by three ground truth litterfall measurements with satisfactory results (R2 = 0.98, p < 0.05, RMSE = 23.12 Mg ha^-1 month^-1). Furthermore, the study could facilitate our understanding of the mechanism governing the litter production and improve future prediction of metabolic scaling theory on ecosystem function.