金門鳥類多樣性的長期趨勢分析

物種多樣性隨時間的變化，是管理生物多樣性的重要一環。在分析物種多樣性長期趨勢時，如何應用物種多樣性指數，及處理資料內固有的缺失值，是二個愈來愈受重視的議題。物種複合指數概念的相對豐度幾何平均值（G-value），及缺失值插補技術的生態監測數據的趨勢及指數（TRIM），近年在國外已被廣為應用，但臺灣仍甚少相關研究。本研究透過分析不同的指數特性，及缺失填補技術在不同資料缺失率的限制與應用，以更能全面地探討生物多樣性不同面向與提升資料的應用價值。本研究以1999年、2004年、2010年和2018年的金門鳥類相普查資料，在重複的6個月份和21個樣點，計算其總豐度、物種豐富度、物種歧異度、皮洛均勻度和G-value，以探討所有鳥種和5個鳥類類群的物種多樣性長期趨勢，以及在3種資料缺失率經TRIM建立所有鳥種、5個鳥類類群和2種遷徙習性的物種多樣性長期趨勢。研究結果顯示，金門在1999年至2018年間所有鳥種的總豐度、物種豐富度和G-value上升，物種歧異度和皮洛均勻度下降。不同類群間的物種多樣性趨勢互異，雁鴨科、鷸鴴類、秧雞科的趨勢下降，可能與金門棲地環境變遷或東亞-澳洲遷徙線的水鳥族群受棲地劣化、狩獵壓力等影響有關，而八哥科、鳩鴿科的趨勢上升，可能與金門環境和農作物變更有關。大部分類群在不同資料缺失率的物種多樣性趨勢無明顯差異，但經TRIM插補後，所有鳥種、雁鴨科、鷸鴴類的多樣性有顯著改變。本研究建議可應用G-value在國家公園或保護區的生物監測計畫，除了輔以不同層面探討多樣性，也可探討生物多樣性的長時間趨勢，以評估經營成效。然而，當數據集內具有缺失值資料時，應根據研究目的和資料特性，藉由TRIM的插補，以降低缺失值資料對物種多樣性指數的影響。

關鍵字

生物多樣性指數；相對豐度幾何平均值；缺失值；生態監測數據的趨勢及指數

並列摘要

Assessing temporal changes of species diversity is a critical part in biodiversity management. However, in analyzing long-term trends of species diversity, choosing various species diversity indices and missing values in the data are two major problems that have received increasing attention. In recent decades, geometric mean of relative abundance indices (G-value) of multi-species indices, and TRends and Indices for Monitoring data (TRIM) of missing value imputation application have been widely used to evaluate long-term trends in biodiversity but remained understudied in Taiwan. This study was aimed to (1) examine the characteristics of different species diversity indices to investigate different aspects of species diversity, and (2) evaluate the applications and limitations of missing value imputation in different levels of missing values. I used bird survey datasets that were conducted in 21 sampling points in Kinmen in the exactly same 6 months of 1999, 2004, 2010, and 2018 to calculate and compare temporal trends of 5 diversity indices (total abundance, species richness, Shannon-Wiener index, Pielou’s evenness index, and G-value) of all species, 5 avian taxa and 2 migratory statuses. The results revealed that abundance, species richness, and G-value of birds in Kinmen increased from 1999 to 2018, while Shannon-Wiener index and Pielou’s evenness index decreased in the same time span. The diversity indices differed among the 5 avian taxa. The declining trends of diversity indices in Anatidae, Rallidae, and shorebirds might be related to environmental changes of Kinmen or the habitat degradation, global climate change, hunting pressure for waterbird populations along the East Asian-Australasian Flyway. The increasing trends of diversity indices in Sturnidae and Columbidae might be related to environmental changes and the transition of major agricultural crops in Kinmen. The long-term diversity trends of most avian taxa showed no significant difference across different levels of missing data ratio. After missing values imputation by TRIM, the diversity trends of some taxa had significant changes. I suggest that the G-value can be applied in biodiversity monitoring programs of national parks and protected areas as an alternative index to evaluate long-term diversity trends and assess management effectiveness. Also, depending on research objectives and characteristics of datasets, missing value imputation by TRIM can be applied in survey data with missing value to reduce the impacts of missing values on species diversity indices.