2026年7月15日，早上8點50分。國道3號林內段，一個監測員看到約120隻紫斑蝶飛越北上外側車道，立刻啟動程序。九點整，每分鐘過境數字爬到608隻。封道。
這段路，平日有卡車、聯結車、自小客，全速北上。那天早上，它讓給了蝴蝶。
這不是詩意的比喻。這是寫在作業手冊裡的門檻數字：每分鐘250隻，啟動「國道讓蝶道」機制，封閉外側車道，直到蝶流消退。7月15日當天，封到中午12點。
20年的資料，第一次出現「7月」
台灣紫斑蝶的遷徙曆，長期以來只有兩個關鍵詞：秋冬南下、春季北返。茂林、林內、觸口段，這些名字出現在每年3月至清明前後的報導裡，不在7月。
台灣紫斑蝶生態保育協會統計，7月15日單日過境總量達25,566隻，最高密度每分鐘608隻。這是監測20年以來，首次出現的7月紀錄。
協會理事長陳瑞祥的解釋具體而直接：這批蝶有八成是新羽化的第二代。5、6月梅雨充沛，古坑與林內山區的盤龍木嫩芽繁茂，幼蟲大量孵化，羽化周期約20餘天，趕上7月15日天晴，傾巢而出。另外兩成是舊蝶，部分可能來自7月初墾丁的另一波爆發後北飛的個體。
南投鹿谷同步出現萬蝶停駐的景況，不是孤例。
歷史對照更值得細看：2026年春季北返，因年初乾旱，全程僅統計到約4.1萬隻，陳瑞祥當時預估全年總量約5萬隻，與2014年乾旱年相近。結果梅雨一來，7月反轉，預估整個被推翻。
義守大學趙仁方教授追了20年
義守大學趙仁方教授追蹤紫斑蝶遷徙逾20年，早在2007年起的研究就記錄到一個趨勢：氣溫偏高的暖化現象愈來愈明顯，且與斑蝶提早北返飛行呈現正比之勢。他曾在海拔3,300公尺的嘉明湖目睹蝶群——那個高度，過去幾乎不在紫斑蝶的移動範圍內。
7月爆量不是浪漫的生態奇觀，是一筆氣候帳目裡新增的項目。物候位移：原本屬於秋冬的現象，往夏季移動。這類訊號在台灣本島同步出現的還有其他物種，但紫斑蝶因為有20年監測基線，是目前最清晰的一筆。
世界首創的門檻是每分鐘250隻
「國道讓蝶道」機制於2007年世界首創，由台灣蝴蝶保育學會調查發現觸口段單日遷移曾逾百萬隻後，與高公局共同建立的制度設計。運作邏輯：監測員在現場計數，每分鐘過境數達250隻，封閉國3北上251K至253K林內段外側車道。蝶流退去，恢復通行。
作為比較：墨西哥帝王斑蝶的保育以禁止砍伐越冬林為主軸，靠的是土地政策。台灣的讓蝶道是即時交通管制，每次啟動都是一個實際的封道決定，不是宣示。2024年春季最高峰曾達每分鐘1,000隻、單日7萬隻，同樣啟動過封道。
這套機制能運作，背後是高公局、台灣蝴蝶保育學會與林內在地社群近20年持續的監測協作。不是哪一個部會的政策宣示，是長期積累的操作慣例。
觸口段的蝶道，到10月都還在
雲林林內觸口段、南投鹿谷鹿芝谷園區、南投竹山鯉魚社區、古坑華山村文學步道——這條蝶道廊道目前估計可觀察至10月。若無颱風干擾，二代個體的後續移動仍在進行中。
最佳觀察時段是天晴上午。這不是導覽詞，是監測數據的結論：7月15日的封道從8點50分開始，9點達到峰值，10點已降到每分鐘60隻。峰值窗口短，但真實。
高速公路為蝴蝶讓道，每年都在發生。今年只是第一次發生在7月。
— 鍾雅筑
Taiwan Stopped the Highway for Butterflies
The Protocol Kicks In at 250 per Minute
At 8:50 a.m. on July 15, 2026, a monitor at National Freeway 3’s Linnei section counted roughly 120 Euploea butterflies crossing the northbound outer lane. The protocol started. By 9:00 a.m., the count hit 608 per minute. The lane closed.
This is not a metaphor. It is a written threshold: when butterfly density reaches 250 per minute, the “freeway butterfly passage” mechanism activates, closing the outer lane of the 251K–253K Linnei stretch of National Freeway 3. The lane stayed shut until noon.
Taiwan’s Butterfly Conservation Society recorded a total of 25,566 butterflies crossing that single day — the first July entry in 20 years of monitoring data.
Why July Has Never Appeared Before
Taiwan’s purple crow butterflies migrate on a predictable calendar: south in autumn and winter, north in spring. Linnei, Chukou, and the surrounding mountain passes appear in news every March through early April. Not July.
Chen Jui-hsiang, chair of the Taiwan Purple Crow Butterfly Ecological Conservation Association, traced this year’s July surge to a specific chain of events: unusually heavy plum rains in May and June produced abundant new growth in the host plant Macaranga tanarius across the Gukeng and Linnei mountains. The larvae hatched, the pupae cycled through in roughly 20 days, and on the first clear morning in mid-July, approximately 80% of the crossing butterflies were newly eclosed second-generation adults. The remaining 20% were older individuals, some possibly originating from a separate population surge at Kenting in early July that then flew north.
Nantou’s Lugu area saw tens of thousands roosting simultaneously. The event was not isolated to a single site.
The Spring Numbers Had Already Been Written Off
The reversal is striking when placed against the spring record. The 2026 spring northward migration, hampered by early-year drought, produced only around 41,000 counted crossings. At the time, Chen estimated a full-year total of roughly 50,000 — comparable to the drought year of 2014. Then the rains came. July erased that estimate entirely.
Professor Chao Jen-fang of I-Shou University has tracked purple crow butterfly migration for over 20 years. Since his research began in 2007, he has documented a consistent pattern: as temperatures rise, butterflies begin their northward journey earlier. He once observed a swarm at Jiaming Lake at an elevation of 3,300 meters — an altitude that had not previously been part of the species’ recorded range. His conclusion is direct: warming and phenological shifts move in proportion.
The July surge is a data point in that same record, not an anomaly to be explained away.
A Mechanism Built Over Two Decades
The freeway butterfly passage program was first established in 2007, the first of its kind globally. The Taiwan Butterfly Conservation Society had documented single-day migration counts exceeding one million butterflies at the Chukou section, and worked with the Freeway Bureau to build an operational response: real-time monitoring, a density threshold, and an actual lane closure decision made in the field.
For comparison, Mexico’s monarch butterfly conservation relies primarily on land protection — banning logging in overwintering forests. Taiwan’s mechanism is live traffic management, triggered by a field counter, executed by a road crew, every time it happens. In spring 2024, the peak hit 1,000 butterflies per minute and 70,000 in a single day; closures were activated then too.
This works because the Freeway Bureau, the Taiwan Butterfly Conservation Society, and the Linnei local community have maintained monitoring cooperation for close to two decades. It is not a policy announcement. It is a practiced routine.
The Window Runs Until October
The Linnei Chukou section, Lugu’s Luzhigu Garden, Zhushan’s Liyu community, and the Gukeng Huashan literary trail form a corridor that observers estimate will remain active into October. No typhoon interference assumed.
Peak hours are clear mornings. The July 15 data says everything: the gate opened at 8:50, peaked at 9:00, and dropped to 60 per minute by 10:00. The window is narrow, but the road stops for it every time.
— 鍾雅筑
延伸閱讀
https://justfly.idv.tw/s/F1EY2sF