珊瑚礁立體結構的量化研究： 以人為擾動對萬里桐珊瑚礁的影響為例

Abstract

珊瑚礁有極複雜的立體結構，提供各種生態功能，並造就豐富的生物多樣性；然而，人為活動的干擾往往造成珊瑚礁立體結構的破壞，使其結構趨於單純。本研究運用近年來興起的運動恢復結構(Structure from Motion, SfM)方法，確立水下攝影、三維建模與結構複雜度的量化分析流程，並以人為活動密集的恆春萬里桐珊瑚礁為例，探討不同人為活動頻度對珊瑚群聚與結構複雜度的衝擊。 本研究首先探討不同底質及底棲生物對珊瑚礁結構複雜度的貢獻，結果顯示葉片形與分枝形珊瑚有較高的複雜度(斜度 = 3000~5000、VRM = 0.3~0.4、表面粗糙度 = 3~8)，表覆形和團塊形珊瑚的複雜度則較低(斜度 < 1500、VRM < 0.15、表面粗糙度 < 2)，藻類與礁岩則介於兩者之間。斜度以比率(tanθ x 100)為單位，可區別不同底質特徵，而以角度(θ)為單位則否；VRM相較於表面粗糙度，更能區別低結構複雜度的底質特徵。結果亦顯示形態較脆弱的珊瑚體是結構複雜度的主要貢獻者。 人為活動對萬里桐珊瑚礁立體結構的影響調查，分別在距離入水點40~70 m (近入水點)與170~200 m (遠入水點)，共選定10個深度介於2~5 m，且為礁岩底質的5 x 5 m樣區，每樣區平均拍攝810張照片，以SfM方法建立的數值高程模型解析度均可達毫米等級。結果顯示，近入水點礁區的結構複雜度 (n= 6, VRM = 0.09 ± 0.01) 比遠入水點 (n=4, VRM = 0.16 ± 0.02) 來的低；但線性粗糙度則無明顯差異 (近入水點= 1.52 ± 0.15；遠入水點=1.69 ± 0.16)，主要係因其受到底質高程起伏的影響(r = 0.58)；珊瑚覆蓋率在近入水點樣區(15 ± 7%)與遠入水點樣區(22 ± 6%)之間無顯著差異，珊瑚覆蓋率與結構複雜度也沒有明顯的正相關。 萬里桐不同樣區的結構複雜度差異，主要係與珊瑚體形態組成有關；在人為活動較密集的近入水點礁區，有較多團塊形與表覆形珊瑚(耐衝擊型)，而在人為活動較少的遠入水點礁區則多亞團塊形與分枝形珊瑚(脆弱型)。本研究顯示，運用三維建模與量化分析方法能具體呈現珊瑚體和珊瑚礁結構複雜度的差異，若透過監測累積長時間的資料，將能更深入了解珊瑚礁在不同環境或極端事件衝擊下的變化，在海洋生態評估、保育和復育方面都有很高的應用價值。

The structural complexity of coral reefs provides a variety of ecosystem functions and lead to high biodiversity. However, human activities may cause damage to coral reefs that reduce their structural complexity. In this study, I used the emerging Structure from Motion (SfM) to quantify the 3D models of coral reefs. I set up a standard protocol for underwater filming, 3D modelling and quantification of structural complexity. Then I applied it to study the possible human impacts on structural complexity of coral reefs at Wanlitong, a marine activity hotspot on the west coast of Hengchun Peninsula. The structural complexities of foliose and branching corals were higher (slope = 3000~5000, VRM = 0.3~0.4, surface rugosity = 3~8), while those of encrusting and massive corals were lower (slope < 1500, VRM < 0.15, surface rugosity < 2). The structural complexities of macroalgae and rocks were in the middle. For the unit of slope, ratio (tanθ x 100) is a better index than degree (θ) to reveal the structural complexity. Moreover, VRM has a higher resolution than surface rugosity in distinguishing the structural complexity of benthic features. The results also showed that branching and foliose corals are major contributors to the structural complexity of coral reefs. Ten quadrats (each 5 m x 5 m) at 2~5 m depth on reef substrate at Wanlitong were surveyed. These quadrats could be divided into heavily trafficked sites (40~70 m from entrance) and lightly trafficked sites (170~200 m from entrance) based on the distance from entry point. An average of 810 photos were filmed in each quadrat for building 3D models by SfM method and the resolution was at millimeter level. The results showed that the VRM values were lower at heavily trafficked sites (VRM = 0.09 ±0.01, n = 4) than those of lightly trafficked sites (VRM = 0.16 ± 0.02, n = 6). However, linear rugosity was similar between heavily trafficked sites (1.52 ± 0.15) and lightly trafficked sites (1.69 ± 0.16). Coral coverage was also similar between heavily trafficked sites (15±7%) and lightly trafficked sites (22±6%). Coral coverage is not significantly correlated with structural complexity. The differences of structural complexity among sites could be explained by the growth forms of coral colonies. Massive and encrusting corals were more abundant at heavily trafficked sites, while submassive and branching corals were common at lightly trafficked sites. This study shows that SfM can be applied to distinguish the difference of structural complexity among coral colonies and coral reefs. With the improvement of hardware and data accumulation in the future, the SfM methods will have high potential in marine ecological survey, ecosystem assessment, conservation and restoration.