聲景生態學研究於近年蓬勃發展，逐漸成為評估生態變動、環境衝擊的監測工具之一。隨著遊憩需求的提高，溪頭自然教育園區已成為國內最熱門的森林遊憩場域，平均每年約有180萬遊憩人次。過去研究指出龐大的遊客量對於當地的生物多樣性可能產生衝擊。過往由於受限於人力、器材與時間等諸多因素之關係，多無法進行長期且全面的監測評估。為能全面且深入地探討遊憩壓力對於生物聲音多樣性組成之影響，本研究採用長期錄音的方式，深入探討溪頭自然教育園區內遊客壓力與生物聲景之相關性。 本研究採用被動式聲學監測工具，以排程錄音機SM3、SM4錄製3處遊客活動區域熱點及2處人煙罕至地點之聲音資料後，將原始錄音資料以非監督式聲源分離方法分為非生物音及生物音兩類，並以聚類分析計算生物聲音多樣性指數，再以ANOVA及廣義可加性模型分析時空變化趨勢。 研究結果共包含以下五點：（1）本研究的所有樣點，均存在日夜及季節差異，且生物聲音集中於春夏兩季。日間生物週期性聲音主要由鳥類繁殖季合唱組成，夜間主要由昆蟲合唱組成。偶發性聲音（例如：鳥類警戒叫聲、赤腹松鼠等哺乳類動物鳴叫）則主要集中於日間。（2）遊憩壓力較高區域的生物發聲活動較少。（3）新冠疫情三級警戒封園期間，生物發聲活動大致高於未封園期間（包含封園前後1個月，及涵蓋秋冬的全年資料）。（4）生物發聲活動與氣溫呈正相關。（5）非遊客活動區域的生物聲音與遊客人數呈負相關，但遊客活動區域內的生物發聲活動，未必隨遊客人次增加而下降。 基於本研究的分析結果，遊憩壓力與生物發聲活動之間確實存在著顯著的負相關性。然而部份遊客活動區域的生物發聲活動卻未必因遊客人次增加而減少，此結果可能源自於邊緣效應，及部分物種已適應人為干擾等因素所致。目前，本研究尚未針對個別物種對遊憩壓力的實際反應進行分析。未來結合更深入的聲源辨識與不同動物對遊憩壓力的個別適應行為研究，將是必要之趨勢。
In the recent decade, the research of soundscape ecology was booming and becoming a powerful tool for environmental impact monitoring. Because the necessity for leisure and recreation enhanced, the Xitou nature education area became the most popular forest recreation area where attracting 1.8 million tourists per year. Many studies indicate that high tourists number might impact local biodiversity. Because of the resource limit (e. g. investigator, tools, and time et al.), the long time and monitoring broadly study are still lacking. To clarify the impact of biophony by tourist noise in Xitou. The present study used a long-time continuous sound record method to reveal the correlation between tourist pressure and biophony. We use passive acoustic monitoring (PAM) tools (e. g. SM3 and SM4 recorders ) to record the sounds of three touristy and two peaceful sites. The original data were separated into non-biological and biological two types by an unsupervised classification method. We use K-means clustering to calculate acoustic diversity and statistics by ANOVA and generalized additive model (GAM) methods. The analysis results include the following five points: 1. All sites in the present study have diel and seasonal differences, and creature sounds are concentrated in spring and summer. The day periodic creature sounds are composed of birds' breeding season choruses, but insect choruses at night. The transient sounds are concentrated during the daytime (e. g. bird alert calls, red-bellied squirrels, and other mammals' calls). 2. The areas with higher recreational pressure have fewer creature sound activities. 3. During the period of the park closure under Level 3 epidemic alert, the creature sound activity is generally higher than the park is not closed (including one month before and after the closure of the park and the whole year covering autumn and winter). 4. There is a positive correlation between creature sound activity and air temperature. 5. The creature sound in the non-recreational area is negatively correlated with the tourist number but may not necessarily decrease when the tourist number increases in the recreational area. Based on the analysis results. There is a significant negative correlation between recreational stress and creature sound activity. However, the creature sound activities in some recreational areas may not decrease when the tourist number increases. It may be due to the edge effect, or some species have adapted to human disturbance. Currently, the present study has not analyzed the responses of each species to recreational stress. In the future, it will be a necessary trend to combine advanced sound source identification and adaptive behavior studies of different animals when under recreational stress.