以被動式聲學探討中華白海豚在河口的活動模式

Abstract

鯨豚利用回聲定位來探測環境與覓食，中華白海豚 (又稱印太洋駝海豚，Sousa chinensis)主要分布於水深小於20公尺的淺水域，其會依賴聲音來進行探測與搜索，因此牠們回聲定位的聲音提供了重要的資訊來研究牠們的行為。被動式水下聲學監測系統已廣泛使用於鯨豚類的長期監測研究。 河口被認為是中華白海豚核心的活動範圍，白海豚在河口的分布梯度會隨著乾濕季而變化。為了瞭解白海豚的活動頻度與行為模式和河川流量之間的關係，從2009年七月起至2015年十一月間，三套被動式的水下監測儀器被架設在台灣雲林縣的新虎尾溪口來紀錄白海豚回聲定位的聲音，並將監測範圍從河口內至外分成四個區域。本研究的結果分成兩個部分，第一部分我調查並計算白海豚在不同程度河川流量下的活動頻度，並以太陽週期的日期與河川流量指標來建立預測模型。第二部分我比較海豚聲音行為在不同季節與河川流量指標下的變化。 第一部分採用重複取樣與樣條函數的方法，將中華白海豚的活動頻度與河川流量指標的關係建立模型。結果顯示海豚偵測率在河口內側與外側有不同的趨勢，在河口內側海豚聲音的偵測率高峰主要在乾季(每年的十二月至隔年四月)，但在河口外側，回聲定位則在每年四月至九月(濕季)達到高峰。此外，在河川流量較低時，海豚偵測率與河川流量呈正相關，但在河川流量較大時(大於5毫米)，偵測率則與河川流量成負相關，且雨量預測模型的解釋力在河口外側較低。基於被動式監測系統提供足夠的大數據資料，海豚活動頻度才得以被此數值模型有效地預測，此模型也是第一個利用聲音資料建構的白海豚活動頻度預測模型。 第二部分以中華白海豚的回聲定位的答聲型態組成與歧異度、聲源角度的變化與聲源範圍來與河川流量指標進行相關分析，結果顯示在河口內側，當河川流量較低時海豚多使用短距離的聲納（過去研究顯示短距離的聲納與追逐或覓食行為有關），然而隨著河川流量增加，其改用較多長距離的聲納。另外發現當河川流量較低時，海豚在河口內側的移動模式較為複雜，且其活動範圍相對較廣。另一方面，在河口外側，海豚聲音行為的指標與河川流量多寡沒有相當明顯的改變。總言之，海豚活動頻度與聲音行為的改變可能與食餌魚類豐度與分布有高度的關聯，而上游降雨量為影響海豚在河口活動頻度與行為的間接因子。

Odontocetes use echolocation clicks to navigate and forage. Primarily active in estuarine and shallow coastal waters of less than 20 m, Indo-Pacific humpback dolphins (Sousa chinensis) rely on echolocation for investigating their environment and for feeding purposes. It has also been shown that the spatial distribution of some humpback dolphin populations varies between dry and wet seasons. During the dry season, they prefer the inner estuary, whereas during the wet season and after heavy rainfall, they move into the outer estuary. Following previous research, this study used passive acoustic monitoring (PAM) to investigate the occurrence and behaviors of humpback dolphins in response to different quantities of river runoff during the dry and wet seasons. The monitoring area was divided into four sections based on their distance to the estuary. To detect the echolocation clicks of the cetaceans, three acoustic data loggers were deployed in the Xin Huwei River estuary, Yunlin County, Taiwan, from July 2009 to November 2015. This thesis presents the results of two separate studies. In the first study, I monitored and calculated the occurrence rate of humpback dolphins with different levels of runoff to generate predictive models using the Julian day and runoff index. In the second study, I compared the acoustic-behavioral changes of humpback dolphins in dry and wet seasons with various levels of runoff. The first study adopted the resampling approach and spline function was used to build predictive models of humpback dolphin occurrence and response to runoff. The dolphin-detecting rate showed distinct patterns between the inner and outer estuary. During the dry season (from December to April), the echolocation clicks of humpback dolphins were mostly detected in the inner estuary, whereas clicks peaked in the outer estuary during the wet season (from April to September). The dolphin detecting rate was shown to positively correlate with the lower levels of runoff but negatively correlate with the higher levels of runoff (>5mm). While the explanatory power of the rainfall model was lower for the outer estuary, the big data set obtained through PAM can be used to effectively predict the occurrence pattern of humpback dolphins by building a value model. The resultant predictive model offers the first reliable tool for predicting the occurring pattern of humpback dolphins based on acoustic data. In the second study, I analyzed the composition of humpback dolphin echolocation clicks and their detecting angles and ranges at different levels of runoff. The results showed that the short-range sonar (having been associated in previous studies with feeding and foraging behaviors) occurred primarily at lower levels of runoff in the inner estuary. By contrast, the long-range sonar occurred mainly at higher levels of runoff. Moreover, humpback dolphins exhibited more complex movements and relatively wider ranges at lower levels of runoff but showed no significant changes in movement type and range with increasing levels of runoff in the outer estuary. The variation in acoustic behaviors and spatial distribution indicates that the activities and behavioral patterns of humpback dolphins in the estuarine habitat are probably driven by river runoff which could impact prey abundance.