宜蘭地區陸貝的空間分布模式

Abstract

研究生物空間分布的型態，建立生物的分布資料庫與分布模式，在物種多樣性的熱點、棲地保育、物種的經營管理等延伸課題中，將可以提供極大的幫助。過去台灣在陸棲貝類方面，曾有針對南亞蝸牛科在台灣分布的研究，但並未進一步進行分布模式的建立與潛在棲地的預測。因此本研究針對宜蘭地區的陸棲貝類進行： (一)全區的陸棲貝類空間分布調查；(二)將整理調查所得之出現紀錄與現有之環境因子資料庫整合；(三)再將生態模型與地理資訊系統整合運用，建立宜蘭地區陸棲貝類的空間分布模式，並預測宜蘭地區之陸棲貝類各物種之出現機率，找出各物種的潛在棲地。本研究在2004-2008四年的調查期間，一共調查了226個調查點，記錄到了3,252個陸貝個體，其中有1,515個活體、1,737個死殼，分屬於24個科，共89種的陸貝物種。再將陸貝出現紀錄與環境因子結合，針對出現記錄最多的前11種陸貝套用邏輯迴歸 (Logistic regression)與生態棲位因子分析 (Ecological-Niche Factors Analysis, ENFA)兩種生態模式，預測宜蘭地區這11種陸貝的出現機率。結果顯示：(一)11種陸貝中，多數物種之分布熱點都位在蘭陽平原與周邊丘陵地交界之地帶。(二)外來種及廣布種則主要分布在平原地區，縣境內之山地區域各物種之出現機率皆低。(三)微小型的陸貝物種則偏向於出現在中海拔之地區，而不是低海拔之丘陵與平原交界地帶。兩種模式之預測結果、選擇之重要環境因子與模式準確率都存有差異，邏輯迴歸之整體預測準確率雖然良好，但是敏感性卻稍偏低；而生態棲位因子分析之準確率普遍較低。出現紀錄越多的物種，兩種模式預測的結果與熱點越接近，而且準確率也越高。顯示模式之預測結果與預測準確率可能與出現紀錄之數量有關。造成兩種模式之預測不同的可能原因有(一)調查點分布不均；(二)物種出現紀錄不足；(三)生態習性不滿足模式假設前提；及(四)使用的環境因子有修正的空間。

Studying the spatial distribution of organisms is one essential topic for ecological research. By using data of field investigations, spatial distribution database of species and prediction models that depict environmental relationships of species distribution can be constructed and will provide tremendous helps for relevant issues such as identifying biodiversity hotspot, habitat conservation, and species management. The spatial distribution of Camaenidae in Taiwan has been reported, however no study has constructed spatial distribution models and predicted potential habitats of land snails in Taiwan. This study was aimed to (1) investigate the spatial distribution of land snails in I-Lan County ; (2) link the species presence records with environmental factors by Geographic Information System; and (3) construct spatial distribution models that predict occurrence probabilities and potential habitats of species. During the four-year field investigations, 226 sites were sampled, and 3,252 individuals (1,515 living snails and 1,737 emptied shells) were recorded, including 24 families and 89 species. Logistic regression and Ecological-Niche Factor Analysis (ENFA) were employed to link the records of land snails with environmental factors and predict the occurrence probability of land snail species in I-Lan County. Results showed that distribution hotspots of most species were located in the borders of I-Lan Plain with surrounding foothills. Non-native species and common species were mainly distributed in the plains. The occurrence probabilities were low in mountain areas for most species. Nevertheless, many tiny land snail species were tended to distribute in mid-elevation areas in I-Lan County, instead of lower-elevation foothills and plains. Differences in predicted distribution, model predictor variables, and model accuracy were found between logistic regression and ENFA. Although the overall accuracy of logistic regression was satisfying, its sensitivity was low, and the accuracy of ENFA was generally low. For species with more presence records, logistic regression and ENFA provided better match in predicted distribution and hotspots and had higher model accuracy, indicating the predicted results and model accuracy might be affected by the number of presence records. The discrepancy in the predictions of logistic regression and ENFA might result from (1) uneven distribution of sampling sites, (2) insufficient presence records of species, (3) not match the model presumptions and prerequisites, and (4) inadequate environmental factors.