合齒魚科之分子系統分類

Abstract

合齒魚科（Synodontidae）包含四個屬與八十三個有效種，其下可再分為兩個亞科：鐮齒魚亞科（Harpadontinae）內包含了鐮齒魚（龍頭魚）屬（Harpadon）與蛇鯔屬（Saurida）；狗母魚亞科(Synodontinae)內則包含了狗母魚屬（Synodus）及大頭狗母魚屬（Trachinocephalus）。本科魚類為底棲性掠食者，體橢圓形或延長，具有極寬的口與尖利的牙齒，多棲息於熱帶及副熱帶海域的沙泥地。自從此類群之分類於西元1861年被Gill建立以來，其地位一直存在爭議，各屬下之物種更常發生因外型相似而有誤鑑之情形，近一步分類上的修訂仍需仰賴完善的親緣關係建構。鑑此，本研究利用分子系統分類學的方法，針對合齒魚科的物種探討以下的問題：(1)建構較完整的合齒魚科及其相對於其他仙女魚目物種的親緣關係 (2)檢驗合齒魚科及其下四屬是否皆為單系群 (3)在已建立的完善親緣關係下推估合齒魚的分化時間 (4)鑑驗目前合齒魚科下各物種的種名有效性及探索其物種之多樣性。本研究中建立了兩種型態的資料矩陣，一為包含了四個核基因（RH，RAG1，ZIC1和ENC1）與三個粒線體基因（COI, 12S和16S）的多基因資料矩陣，用以建構合齒魚科及其相對於仙女魚目物種間之親緣關係；另一種則為搭配種類界定軟體 (ABGD與 bPTP) 用以進行物種界定的粒線體COI基因資料矩陣，其中包含了975條序列（202條為本研究提供，773條則由資料庫或私人資源取得）與52個已知種。根據結果顯示，合齒魚科並不是一個單系群，然而其下之兩個亞科卻各自為單系群。因此，本研究結果支持將現今分類的龍頭魚亞科及狗母魚亞科提升至科別的等級。而在原先分類下的四個屬中，僅龍頭魚屬與大頭狗母魚屬為單系群，蛇鯔屬與狗母魚屬則為多系群，分別具三及四個分支。其中龍頭魚屬與蛇鯔屬的其中兩個分支為姊妹群，而大頭狗母魚屬則與狗母魚屬中的其中兩個分支形成的單系群為姊妹群。分化時間推估顯示出狗母魚科的最近共同祖先出現於白堊紀晚期，而龍頭魚科的最近共同祖先則是出現於始新世。此外，物種界定分析的結果中，本研究定義約87個假定物種（演化操作單元），此數目遠高於本研究中所包含的已知種，顯示出此類群的生物多樣性可能有被低估的情形。本研究對於合齒魚的系統分類提供了一個更深入的觀點，並對於後續分類學及演化學的相關研究提供了一個完善的架構。

Lizardfishes are a group of benthic carnivorous fishes that have cylindrical bodies and relatively large mouths, live mostly in the sandy bottom of tropic and sub-tropic seas. Belong to the family Synodontidae, the lizardfishes include 83 currently recognized species in 4 genera, Synodus, Harpadon, Saurida, and Trachinocephalus. However, the taxonomy of the family has been controversial since it had been established by Gill (1861). A robust phylogeny of this family is required for further taxonomy revision. In this study, we aimed to: (1) reconstruct a comprehensive phylogeny of the family Synodontidae along with its aulopiform allies; (2) test the monophyly of the family and the four recognized genera; (3) estimate the divergence times of synodontids under a robust phylogeny; (4) investigate the validity of the existing species within the family with an integrated approach in taxonomy (i.e. species delimitation). To infer the phylogeny of the family, a multi-gene dataset was composed of four nuclear genes (RH, RAG1, ZIC1 and ENC1) and three mitochondrial (COI, 12S, 16S) genes. The datasets used for the species delimitation analysis with two programs (ABGD and bPTP) included 975 mitochondrial COI gene sequences, of which 202 were newly generated in this study and 773 were retrieved from public and private sources. In the results, the Synodontidae was not monophyletic while each of the two subfamilies of the Synodontidae was resolved as a monophyletic group. Therefore, the elevation of the two subfamilies Synodontinae Gill 1861 and Harpadontinae Bleeker 1875 is suggested, i.e., from subfamily level to family level. Within the family, only two out of the four currently recognized genera are monophyletic, which are Harpadon and Trachinocephalus. The monophyletic Harpadon is sister to the clade composed of two of the four well-resolved Saurida lineages, while the monophyletic genus Trachinocephalus is sister to the clade composed of two of the three well-resolved Synodus lineages. The divergence time estimations recovered that the most recent common ancestor of the family Synodontidae arose during the Late Cretaceous, while that of the family Harpadontidae occurred during the Eocene. From the species delimitation analyses, 87 putative species (OTUs) were observed out of the 52 recognized species included in the study, indicating an underestimated biodiversity of the family. This study offers an insight into the systematics of the family Synodontidae, and provides a framework for further taxonomic and evolutionary studies.