摘要

 本計畫調查太魯閣國家公園合歡山區外來種的分布，外來種大多分布在公路及步道兩旁。從合歡山公路昆陽到落鷹山莊間，發現昆陽以東至武嶺、松雪樓東縣界至克難關南縣界，及農業部生物多樣性研究所高海拔試驗站至合歡北峰登山口較多外來種分布，共發現二十種。這些外來種常伴隨一起出現在停車場、施工區域邊坡塌陷等人為破壞地。大羊蹄是最主要的外來種佔 32%，其次是白花三葉草、貓兒菊、及葦狀羊茅，分別占所有外來種分布總面積(1.3243 公頃)的15%、14%及 13%。相較距今 16 年前本高海拔區域的調查有 4 歸化種(孫麗珠 2008)，這十幾年來，外來種在種類和數目上都明顯的增加。

外來種除佔據原生植物棲地外，本研究也發現在選取之合歡尖山溪谷樣區高山植物傳粉者生態網絡，已受到入侵外來種明顯干擾。貓兒菊是最長驅直入原生植物棲地的外來入侵種，開花時會吸引大量蜂類及蠅類傳粉者拜訪，且以蠅類拜訪為主，在五個時期中，大量散播花粉佔據他種原生植物柱頭上總花粉量0.4%~35%不等，傳粉網絡中16%~90%原生植物種類都受其沾汙。尤其7 月中花季第三時期起貓兒菊的開花數激增，在網絡中其能廣泛性地吸引各類傳粉者，並顯著與其他原生植物競爭共享傳粉者 (網絡性質分析特指為insect niche overlap - 傳粉者棲位重疊程度高)。且貓兒菊顯著已融入本樣區高山植物傳粉網絡中佔據棲位成為優勢物種(dominant species)，搶佔生態系主要資源分配，建議應該在每年開花前立即剪除花序及地上葉。

本委託案也是臺灣首次科學研究亞熱帶高山生態系中，授粉昆蟲與蜜源植物之間的交互作用網絡是否健全。本研究為第一篇以昆蟲實際攜帶的花粉種類及量，來建構分析交互作用網絡性質，發現合歡山的昆蟲與植物交互網絡雖緊密但穩健性不夠，蜂期綜合網絡的穩健性指數 0.46，蠅期為 0.43 都偏小，顯示合歡山昆蟲及植物交互網絡的穩健性不足，原因在網絡中依賴優勢物種支撐，倘若失去優勢物種，則網絡就會崩解。依滅絕曲線模擬分析，發現網絡僅依賴少數優勢授粉昆蟲及蜜源植物支撐，如信義熊蜂、高山薔薇、一枝黃花；一旦這些優勢物種因氣候暖化和外來種競爭喪失棲位而滅絕，整個生態網絡就崩解。由於本研究中交互網絡是以昆蟲身上的花粉重建出實際交互作用訪花關係，清楚鑑定昆蟲身上攜帶花粉種類及數量之量化交互作用關係，比起實驗室之前發表僅以授粉昆蟲拜訪花朵頻率，但無法確定該拜訪是否真的有攜帶花粉及花粉的數量的間接觀測改進的更精確。

此外也為更了解高山植物集中夏季開花、因彼此之間共享僅有的昆蟲傳粉者，是否會彼此競爭、還是促進更多傳粉者拜訪，以增加授粉生殖成功率? 透過檢驗各植物柱頭上同種(CP)與異種(HP)花粉的數量及種類，以廣義線性混合1模型分析(GLMM)， 整理而言發現亞熱帶高山生態系植物一同開花會有一定程度促進彼此授粉成功的效果(β=CP/HP >0)；而在重建出昆蟲傳遞花粉共沾網絡上發現，花季中期(7 月)由蜂類主要傳粉時，花粉傳遞網絡互動較緊密，代表蜂類廣傳花粉，而共同開花植物彼此授粉成功率促進競爭互見，但大多未偏出中性共棲(β~0)；反之，花季後期(8 月)蠅類為主傳粉時，網絡分化出較多子群體，且蠅類身上普遍攜帶較專一的花粉種類，代表此時開花的植物種類各自仰賴不同蠅類為其授粉，呈現促進彼此授粉成功效果(β>0)。

從這些結果看來，在植物多樣性高之合歡山亞熱帶高山生態系，植物與授粉昆蟲的交互作用分化出花季早中期由蜂類廣傳花粉，而晚期由蠅類相對專一對植物授粉。但在國家公園外來種入侵區域，授粉昆蟲與植物交互網絡受到外來種搶佔資源優勢物種地位，使得網絡穩健性(穩定度)不足交互作用崩解物種滅絕，外來種入侵顯然已產生磁吸昆蟲傳粉者作用，人為將貓兒菊花粉添加在其他原生植物柱頭上可明顯發現減低結子率，嚴重干擾降低本土植物生殖成功。

短期建議上，包括貓兒菊等外來種，依報告書結論防治建議以訓練國家公園志工以人為移除: 1.路旁及破壞地之外來植物族群，在開花前割草去除。2.在生態敏感區中與原生植物雜處棲地中，以人工剪除花序避免結子繁殖。中期防治建議則應透過:1.公務機關間溝通合作，勸說鄰近高山農場停止栽植這些百大入侵種作為吸引觀光客用途；2. 辦理環境教育工作坊，為清境或梨山地區當地民眾解說辨認外來種對生態的危害。3. 建議公路總局埔里及太魯閣工務段，避免使用外來種做護坡植物。長期而言，氣候暖化有利於這些原本在中海拔分布的入侵植物往高山移動，故而增加了與高山原生植物在競爭棲地及授粉者的種間競爭機會。除非全球暖化程度減緩，必須考慮將合歡山區高達50%的台灣特有種植物，以原地及移地復育、採種育苗方式保存。待原生植物育苗技術建立後，亦可提供中高海拔工務段邊坡養護護坡之用。

關鍵字

網絡穩健性(robustness); 植物傳粉者生態網絡(pollinator-plant interaction network); 入侵種(invasive species); 生殖成功(reproductive success); 優勢種(dominant species); 滅絕曲線(extinction curve)

--------------------------------------

Abstract

This project investigates the distribution of invasive species in the Hehuan Mountain area of Taroko National Park. Most invasive species are found along roads and trails. Between the Kunyang section of Hehuan Mountain Road and Luoying Mountain Lodge, a notable presence of invasive species was discovered from Kunyang eastward to Wuling, from Songxuelou’s eastern county boundary to the southern county boundary of Kannan Pass, and from the Ministry of Agriculture’s high-altitude experimental station to the Beifeng trailhead, totaling twenty species. Among them, the most prevalent invasive species is the Rumex obtusifolius (broad- leaved dock), accounting for 32% of the distribution. It was followed by Trifolium repens (white clover), Hypochaeris radicata (annual fleabane), and Festuca arundinacea (tall fescue), which made up 15%, 14%, and 13% of the total distribution area, respectively. In comparison to previous most recent survey conducted 16 years ago, which only recorded four naturalized species, the types and numbers of invasive species have significantly increased over the past decade.

In addition to occupying the habitats of native plants, this study is the first in Taiwan to discover that the pollination network of alpine plants which has been significantly disrupted by invasive species. H. radicata is the most aggressive invasive species that has invaded into native plant habitats, attracting a large number of bee and fly pollinators during its flowering period, with fly visits predominating. Throughout five sampling periods of the entire flowering season, the amount of pollen dispersed by H. radicata accounted for between 0.4% and 35% of the total pollen on the stigmas of other native plants, interfering 16% to 90% of the native plant species in the pollination network. Particularly, the flowering number of H. radicata surged in mid-July, allowing it to attract a wide range of pollinators and significantly competes with other native plants for shared pollinators (high niche overlap). H. radicata has notably integrated into the alpine plant pollination network, becoming a dominant invasive species. It is recommended that flower buds and ground leaves be removed immediately before the flowering season.

This study is also the first scientific survey into whether the pollination network in subtropical alpine ecosystems is healthy or not. This network property estimation revealed that this interaction network in Hehuan Shan lacks robustness. The network relies heavily on a few dominant pollinators and nectar plants, such as the Bombus formosellus, Muscidae, Rosa transmorrisonensis, Primula miyabeana and Solidago virgaurea var. leiocarpa. According to extinction curve simulations, if these dominant species were to lose their habitats due to climate warming and competition from invasive species such as H. radicata, the entire pollination network would collapse. By counting pollens species and numbers from each individual pollinator, this analysis provides more reliable data than previous studies that merely observed whether pollinators visited flowers.

Additionally, the study aims to better understand whether the simultaneous blooming of alpine plants in the summer leads to competition among them for the limited insect pollinators or it promote more visits from pollinators due to flowering together, thereby increasing reproductive success. By examining the number and types of pollen from the same species (CP) and different species (HP) on the stigmas of various plants, and analyzing the data using generalized linear mixed models (GLMM), it was found that co-blooming in subtropical alpine ecosystems can promote mutual pollination success to a certain extent (β=CP/HP >0). During mid- season (July), when bees were the primary pollinators, the pollen transfer network interactions were tighter, indicating that bees effectively spread many plants’ pollens altogether. Therefore, the pollination success among co-blooming plants were mostly neutral (β~0). In contrast, in late season (August), when flies dominated, the network diversified into more subgroups, with flies generally carrying more specialized types of pollen, indicating that the blooming plants each relied on different fly group species for pollination, promoting mutual pollination success with less pollen contaimination because lower degree of pollinator sharing (β>0).

These results suggest that in the alpine ecosystem of Hehuan Shan, the interaction between plants and pollinators evolves from broad pollen transfer by bees in the early to more specialized pollination by flies toward the later period. However, in areas of the national park affected by invasive species, the interaction network between pollinators and plants is compromised as invasive species occupy the dominant species position, leading to insufficient network stability and higher risk of species extinction. The extensive invasion of H. radicata has evidently attracted insect pollinators, disrupting and diminishing the reproductive success of native plants.

When adding pollens of H. radicata on to the stigma of native species, it caused significantly reduction of seed set. It therefore essential to remove these invasive species immediately.

Keywords:

Network robustness; pollinator-plant interaction network; invasive species; reproductive success; dominant species; extinction curve