| 摘要: | 全球生物多樣性正在迅速的變化，造成原因與人類活動及棲息地改變有關。其中因人為活動促成的外來種與入侵種的議題，為影響生物多樣性的重要因素之一並受到廣泛討論與研究。台灣地區的地理環境與氣候孕育出豐富的鳥類多樣性，近年來隨著貿易與交通工具的便利普及，外來種引入引發的環境問題越趨嚴重，並需予以重視。多數台灣防制外來種報告與特生中心的研究皆提到台灣的入侵鳥類中椋鳥科是種類、數量最高且分布最廣泛的一個入侵類群。本研究以冠八哥以及與其競爭之外來種八哥作為研究對象，因冠八哥為代表椋鳥科的台灣特有亞種，族群數量和生存環境卻受到外來種八哥威脅而下降。因此，本研究之目的，為了解原生種冠八哥與外來種白尾八哥和家八哥長期的變化趨勢，同時企圖利用環境資料與不同時期的土地利用資料，理解三種八哥在棲地需求上是否有異同，以及可能影響其族群變遷的土地利用變遷因素。
本研究選取中華鳥會鳥類調查資料較豐富的時段1986-2010年作為鳥類研究資料，並將鳥類資料分為兩種資料，第一種資料為單次調查中三種八哥在樣區出現與否的資料、第二種資料為彙整計算五年中三種八哥在每個樣區的出現頻率，作為代表冠八哥、白尾八哥、家八哥等椋鳥科鳥種在台灣地區鳥類變遷的趨勢。土地變遷部分，則使用兩種解析度的景觀變遷資料：1km*1km臺灣地區生態與環境因子地理資訊資料庫的資料，以及兩期(1992-1995年及2006-2008年)國土土地利用調查資料(1/5000)。而兩期調查時間分別在1992-1995年以及2006-2008年.鳥類資料為了配合兩期國土調查時間，以五年為一單位選取在1991-1995年以及2006-2010年間，進行棲地分析。首先將景觀變遷資料利用主成份分析幫助精簡數據，從各主成份軸篩選出25項與三種八哥關聯性較高且重疊性較低的的變項作為環境因子變項。最後，將選出的環境因子變項與三種八哥的兩種鳥類資料各別進行卡方檢定、T檢定、ANOVA分析、分類樹、Logit概化線性模式、概化線性模式和線性回歸分析，驗證本研究一共提出11種棲地特性相關假說是否成立。
趨勢分析結果發現，冠八哥確實與其他兩種外來八哥有相反的變遷趨勢，兩種外來八哥有增加的趨勢，而冠八哥卻呈下降趨勢。其中又以白尾八哥增長趨勢最為明顯，增長速度較家八哥更為迅速。假說驗證結果發現：1）三種八哥偏好出現在都市環境。2）三種八哥偏好出現低海拔的區域，白尾八哥以及家八哥主要分布低於250M，冠八哥分布主要低於125M。這兩點假說驗證了在大尺度中三種八哥偏好低海拔且又是都市的區域。3）三種八哥都偏好綠地，但對於溼地、裸露地、及農耕地偏好程度較低。4）三種八哥對於不同的綠地類型的偏好程度有所差異，冠八哥最為偏好較自然的綠地空間如草生地等，而白尾八哥略為偏好較自然的綠地空間，家八哥則偏好人工的綠地空間如公園綠地。5）兩期最佳棲地模式比較結果顯示三種八哥在挑選棲地時，可能受到族群勢力大小左右，當冠八哥或外來種八哥的族群較大時在競爭力微棲地時較具有優勢，得以佔領較偏好的微棲地，而族群勢力較小的八哥則放寬條件退居至符合巨棲地條件即可，無法競爭較偏好之微棲地。6）若將土地變遷情況與鳥類變遷狀況利用線性回歸建立棲地模式，則發現三種八哥的族群變遷確實受到土地變遷的影響，變遷模式顯示水利用地轉作為其他用途如：公園綠地或是綠地組成較高之遊憩用地，都會提高外來種家八哥的出現頻率、若減少相對自然的草生地時冠八哥的出現頻率則會下降。7) 從景觀組成分析，則發現當自然草地在整體土地覆蓋中超過2%時，冠八哥的出現頻率較高。
棲地模式、變遷模式以及相關假說結果，皆顯示相對自然的草生地為冠八哥族群影響最大的棲地類型，因此未來規劃及設計綠地空間時，整體土地覆蓋上至少要設置2%的自然草地，此外，對於現有冠八哥族群數量較多的地區應該要保持較高的天然草地，並對於在這些區域的外來種八哥族群數量進行監控，並進一步減少外來八哥的勢力，以確保冠八哥族群在這些區域具有優勢。本研究建議優先以下列地點重點實行：台北關渡自然保留區、台北華江橋、南投埔里公田溝、台北社子、台北石門老梅、台北華中橋、台北淡大后山、台北田寮洋、嘉義布袋、台北金山等地區，讓冠八哥族群生存的機率藉由良好的環境管理一步步提升。
Human activities and habitat change have caused dramatic changes to global biodiversity. Biological introductions and invasions aggravated by human activities is one of the most important factors affecting biodiversity globally, and is widely researched and discussed. Taiwan's geographical environment and climate nurtures a rich diversity of birds. However, with the international trade and convenient transportation system in recent years, the introduction of exotic species has caused serious environmental problems. Of the exotic birds in Taiwan, starlings, especially white-vented myna (Acridotheres javanicus) and common myna (Acridotheres tristis) were found to be the most abundant and widely distributed. The native crested myna (Acridotheres cristatellus) of Taiwan has been negatively affected by exotic mynas. The purpose of this thesis was to study the long term trends of the native myna and the two exotic mynas in Taiwan, and through analyzing the relationship between trend variation of the three mynas and land use change data, understand if there were similarities and differences in their habitat requirements, and how land use change affected their communities.
We used the avian survey data (from 1986 to 2010) of 102 study sites in Taiwan from the bird record database managed by the Chinese Wild Bird Federation (CWBF). We used two types of compilations of the bird data: The first data set was the presence/absence of the three mynas in each survey, and the second data set was the frequency of occurrence (%) at each study site. Landscape was analyzed at two scales: a 1km*1km ecological and environmental GIS database, and a 1/5000 land use survey data. The land use survey data included two periods, once during 1992-1995, and once during 2006-2008. To match the land survey data, the bird data was also selected for two five-year periods 1991-1995 and 2006-2010 during habitat analysis. First we used principal component analysis to help downsize the data set, and finally chose 25 variables. Chi-square test, T-test, ANOVA analysis, Classification regression tree, Logit generalized linear model, Generalized linear model and Linear regression models were used to test 11 proposed hypotheses of myna habitat preferences. Trends across the twenty five years showed that the frequency in which the native crested myna was observed decreased while the other two exotic mynas (white-vented myna and common myna) increased. White-vented myna, in particular, increased faster than the common myna. The results of hypothesis testing found: 1) The three mynas all preferred urban environment. 2) The three mynas all preferred lowlands. White-vented myna and common myna were distributed primarily below elevation 250M; native crested myna was distributed primarily below elevation 125M. These two results confirmed that the three mynas preferred urban environments in lowlands. 3) The three mynas preferred green space, but showed less preference for wetlands, barren land, and agricultural lands. 4) The three mynas also showed preferences for specific types of green spaces. Crested mynas showed a stronger preference for more natural green spaces; white-vented myna showed slight preference for natural green spaces; and the common myna showed preference for artificial green spaces. 5) The best habitat models for the three mynas changed during the two periods in ways that suggested that habitat selection may be affected by population size: the myna with the largest population were more strongly associated with their favorite microhabitats; the mynas with smaller populations relaxed their habitat choices to include less favorable microhabitats within preferred macrohabitats. 6) Changes in myna frequency were analyzed using changes in land use to form transition models to see if changes in myna populations were affected by land use changes. Results suggested that riparian land converted to other uses such as parks will increase the frequency of exotic species (common myna), and the reduction in natural grassland will decrease the frequency of native crested myna. 7) Further analysis of landscape composition suggested that landscapes with more than 2% natural grasslands were found to have higher crested myna frequency.
Habitat models, transition models and hypotheses testing results all suggested that relatively natural grasslands was the most important microhabitat affecting native crested myna. Green space planning should allocate at least 2% of the overall landscape to relatively natural grassland to improve native myna habitats. Also, locations currently with higher native crested myna populations should be maintained with higher natural grass cover, and measures taken to reduce the exotic mynas populations in these locations to assure dominance of the native mynas in these areas. Areas of importance identified by this thesis includes: Guandu Nature Reserve (Taipei), Hua jiang bridge (Taipei), Puli Kung Tin ditch (Nantou), Shezi (Taipei), Shimen Laomei (Taipei), Huazhong Bridge (Taipei), Danda mountain(Taipei), Tianliaoyang(Taipei), Budai (Chiayi), Jinshan(Taipei). Through such measures, native crested myna populations may have a higher chance of survival. |
