DNA Barcoding Of The Endemic Fishes From The Upper Yangtze River Basin

The Upper Yangtze River basin is a natural gene pool for freshwater fishes in China. It contains261fish species, including107endemic fishes in this basin. Among them, Cypriniformes is the most abundant order with93endemic fishes accounting for86.92%of the total endemic fishes in this basin, thus cypriniform fishes play an important role in the fish diversity of the Upper Yangtze River. On a global scale, Cypriniformes is consisted of6families,321genera and about3,268species. They are mainly distributed in the Southeast Asia. On the one hand, climate change and anthropologic activities have caused a massive decline in the diversity of fish species. On the other hand, it’s still difficult to identify some fish species using traditional taxonomic methods due to phenotypic plasticity and genetic variability in some fishes. Therefore, it is of great importance to quickly acquire the genetic information and further build a complete and accurate gene sequence library to facilitate the preservation of the fish genetic diversity.DNA barcoding is a rapid and accurate method for species identification through sequence comparison. It has been widely applied in many fields, such as ecology, environmental science, food safety monitoring, diseases control and biodiversity protection. In previous study, CAOS software was frequently used to predict nucleotide-diagnostic sites in DNA barcoding, which is involed in a relatively complex procedure and requires sophisticated sequence format. However, self-organizing maps (SOM) is an artificial neural network model. It is based on a principle of non-linear data prediction that can lead stronger prediction ability, results of higher reliability and more intuitive plots when compared with linear models.In the present study, a partial mitochondrial cytochrome C oxidase subunit I (COI) gene was employed to barcode the endemic fishes from the Upper Yangtze River. In addition, we tested the effectiveness of COI gene for barcoding of cypriniform fishes combined our data with the data downloaded from online database (GenBank and BOLD). Furthermore, SOM model was used to validate its feasibility of predicting nucleotide-diagnostic sites in DNA barcoding research.The main results are shown as following:1. In the present study, three hundred and one individuals of the endemic fishes from the Upper Yangtze River were barcoded using partial DNA sequence of the COI gene. These301individuals belong to45species,29genera,8families and4orders. Of which242individuals were sequenced in the present study and the rest were downloaded from GenBank and BOLD. The results showed that the average Kimura’s2-parameter (K2P) distance within species, within genus, within family and within order was0.47%,2.60%,16.32%and21.76%, respectively. Totally,88.89%species formed barcoding gap between the maximum intraspecific and minimum interspecific genetic distance. Among the29genera, species in28genera clustered respectively in monophyletic clades on the neighbor-joining tree, which consisted of96.55%of the total cases. However, species in the following four genera clustered non-specific monophyletic clades. These species were (1) Schizothorax chongi, S. cryptolepis, S. davidi, S. kozlovi, S. graham, S. prenanti, S. sinensis and S. wangchiachii in genus Schizothorax;(2) Sinibotia superciliaris and S. reevesae;(3) Jinshaia. abbreviate and J. sinensis;(4) Pareuchiloglanis sinensis and P. anteanalis. In addition, Sinibrama taeniatus and Acipenser dabryanus did not form a monophyletic specific group, respectively. Within these taxa, we suspected the reasons might be rapid speciation, introgressive hybridization and/or incomplete lineage sorting. To sum up, the results indicated that DNA barcodes based on COI gene were effective for rapid and accurate identification of the majority of the endemic fish species from the Upper Yangtze River and demonstrated the validity of SOM mothod for predicting the nucleotide-diagnostic sites in family level of these groups of fishes.2. Partial sequences of COI gene from GenBank and BOLD were used to barcode3,434individuals, representing989taxonomic species in274genera and6families in Cypriniformes. The K2P distance within species was0-13.85%, in which90.20%of cases were below2%. In the threshold of2%, the average K2P distance within species, genus, family and order were0.21%,11.34%,16.15%and18.44%, respectively. Divergence between congeneric species was54-fold higher than between conspecific sequences. The levels of genetic divergence between conspecific sequences, congeneric species, and confamilial genra across the five major families of the cypriniforme fishes were compared in the present study. Results from Kruskal-Wallis nonparametric rank sums tests indicated that the genetic divergence of congeneric species and confamilial genera were significantly different among the major families (P<0.0001), showing that the surveyed cyprinid fishes on average showed the highest genetic divergence within species, while the mean congeneric distance within catostomid fishes was the lowest one. Barcoding gaps between intra-and interspecific distance was formed in Balitoridae, Cobitidae and Nemacheilidae. Results from SOM method showed that nucleotide-diagnostic sites in levels of species and genus could be successfully predicted, while failed in level of family. The vast amount of COI sequence of cyprinid samples and large number of variable sites were considered that might account for this pattern. The present study indicated that SOM method was feasible for predicting the nucleotide-diagnostic sites, thus could be widely applied in DNA barcoding research.