Fish species composition, distribution and community structure in the lower reaches of Ganjiang River, Jiangxi, China

The Ganjiang River (length: 823 km; drainage area: 82,809 km2) is the largest river that flows into Poyang Lake and an important tributary of the Yangtze River. In this study, fish fauna were collected from 10 stations in the lower reaches of the river (YC: Yichun, XY: Xinyu, SG: Shanggao, GA: Ganan, ZS: Zhangshu, FC: Fengcheng, NC: Nanchang, QS: Qiaoshe, NX: Nanxin, CC: Chucha) from March 2017 to February 2018. The species composition and distribution as well as spatio-temporal variation in biodiversity and abundance were then examined. Overall, 12,680 samples comprising15 families and 84 species were collected, the majority of which belonged to the Order Cypriniformes (69.05% of the total species collected) and Cyprinidae (64.29%). Moreover, of these 84 species, 36 (42.86%) were endemic to China. Dominant species were Cyprinus carpio (index of relative importance (IRI): 17.19%), Pseudobrama simoni (IRI: 10.81%) and Xenocypris argentea (IRI: 10.20%). Subsequent cluster analysis divided the samples into three significantly different groups by sample site. Meanwhile, Margalef species richness and Shannon−Wiener diversity indices were both low, and along with analyses of abundance-biomass curves suggested moderate disturbance. Current threats to the conservation of fish biodiversity in the lower reaches were also reviewed and management solutions suggested. The results will help form the basis for reasonable exploitation and protection of freshwater fish in the lower reaches of the Ganjiang River.


Results
Species composition. Eighty-four species in total were collected from the 10 sampling sites (12,680 individuals weighing 354.09 kg), and categorized into seven orders comprising 15 families (Table 1). Of these, the most species-rich order was Cypriniformes (two families, 58 species), followed by Perciformes (five families, 10 species), Siluriformes (three families, 10 species) and Synbranchiformes (two families, three species). Meanwhile, Clupeiformes, Osmeriformes and Beloniformes were represented by only one family and one species each ( Table 1).
A total of 36 endemic species (42.86% of the total species collected) in eight families were identified. Cyprinidae was the dominant family (22 species), followed by Bagridae (five species), Cobitidae, Percichthyidae and Odontobutidae (two species each). Meanwhile, Amblycipitidae, Salangidae and Osphronemidea were each represented by one species (Table 1).
Cluster analysis divided the fish species into three significantly different groups (Fig. 2 Spatial-temporal variation in the fish assemblage. The highest fish species richness, species abundance and species biomass as well as Margalef species richness index and Shannon−Wiener diversity index occurred in the summer, while lowest values observed in the winter. In contrast, the highest Pielou evenness index was observed in the winter, with lowest values in the summer ( Table 2).
Highest species richness was observed in sites GA and NC (41 fish species, respectively), while lowest values (20 species) were observed in site XY. Highest abundance (2840 individuals) was observed in site NC, while lowest (237 individuals) was observed in site SG. Highest biomass (63058 g) was observed in site ZS, while lowest (3762 g) was observed in site YC. The species diversity indices also differed among sampling sites. The highest Margalef species richness index (6.90) was observed in site GA, while the lowest (3.20) was observed in site FC. The highest Shannon−Wiener diversity index (2.93) and Pielou evenness index (0.78) were observed in site GA, while the lowest values were observed in site NX (1.54 and 0.44, respectively) ( Table 3).  Table 2. Temporal variations in fish species, individuals, biomass and diversity indexes in the lower reaches of the Ganjiang River.   www.nature.com/scientificreports www.nature.com/scientificreports/ Non-metric multidimensional scaling ordination also clearly separated the samples into three groups by sampling sites (Fig. 3). Meanwhile, one-way ANOSIM also revealed a highly significant effect of sampling site (Global test R = 0.448, P = 0.001) based on fish abundance data.
Characteristics of the ABC. Abundance curves of fish communities at sites YC, FC, QS, NX and CC lay either above the biomass curve, or the two curves crossed, and the W-statistics were negative. For the remaining five sampling sites, the biomass curves lay above the abundance curves in some places and had small positive W values (Fig. 4). These findings suggest moderate disturbance.

Correlation between environmental factors and fish composition. Water temperature (WT) and
the dissolved oxygen (DO) content differed by seasons; however, no obvious spatial or temporal differences in pH were observed. Maximum WT and DO values were 32.7 °C (site NX in summer) and 12.7 mg/L (site CC in autumn), while lowest values were 8.8 °C (site FC in winter) and 3.9 mg/L (site GA in summer). In contrast, the mean WT and DO did not differ significantly within sampling sites. The site average WT ranged between 19.7 °C and 22.1 °C (mean ± SD, 20.8 ± 0.7 °C), while the mean DO (±SD) was 7.2 ± 0.5 mg/L, and ranged between 6.3 and 8.0 mg/L. The maximum and minimum pH were 8.1 and 6.8, respectively, and the site average pH ranged from 7.0 to 7.8 (mean ± SD, 7.5 ± 0.3) ( Table 4).
RDA was subsequently used to generate bi-plots after extracting and integrating the data from the fish community indices with the environmental data (WT, DO and pH) (Fig. 5). The first axis of the cumulative percentage of variance of the species-environmental relationship was 26.85%, with four axes accounting for 67.34%. Overall, these findings suggest that WT, DO and pH had a significantly effect on fish distribution and assemblage composition in the study area (P < 0.05).

Discussion
General characteristics of the fish fauna and assemblage. Eighty-four fish species belonging to seven orders and 15 families were collected from the lower reaches of the Ganjiang River, the majority belonging to Cypriniformes and Cyprinidae. The most species-rich order and family also represented the richest order and family in the Yangtze River Basin and China 2,4 . Overall, 36 species were endemic to China. The total number of species and endemic species represented 38 and 27% of the total for Jiangxi Province 5 , and 20 and 11% of these in the Yangtze River Basin, respectively 2,4 . Fish biodiversity in the study area is therefore also important in the larger context of Jiangxi Province and the Yangtze River Basin.
Further analyses revealed that most species were omnivorous, pelagic or settled fish that lay pelagic eggs. However, the Margalef species richness index and Shannon−Wiener diversity index were low compared with adjacent watershed areas, such as middle reaches of Ganjiang River 15,17,20 and Poyang Lake [21][22][23][24] .
The number of species with an average body weight of less than 200 g per individual represented approximately 90% of the total species richness. Meanwhile, smaller individuals (<200 g) represented 98.5% of the total abundance. Moreover, ABC analyses suggested moderate disturbance of fish population in this area. The findings also revealed a small proportion of large commercial fish species and larger numbers of small species in the sample catches, indicating an obvious low age trend. Taken together, these findings further suggest that fish resources in the lower reaches of the Ganjiang River are reaching exhaustion. Current threats. Disturbances resulting from dam construction, sand excavation and overfishing are the most significant threats to fish biodiversity in this area. For example, construction of Wan'an dam has affected connectivity in the river, altering the overall hydrology, and disrupting migration of fish species from connecting lakes as well as the velocity and floating distance of drifting eggs. A decline in M. reevesii yield has also been observed in the Ganjiang River 6 , due to reduced water inflow, which in turn is destroying breeding habitats 25 . Historically, 12 spawning grounds of four domestic fish species once existed in the middle reaches of Ganjiang River 6 . However, construction of Wan'an, Shihutang, Xiajiang and Xingan dams has caused flooding of many of these spawning areas 7 . www.nature.com/scientificreports www.nature.com/scientificreports/ Meanwhile, overfishing is threatening traditional fisheries, with pressure to adopt modern harvesting activities. This process is occurring throughout all major river systems in China, especially the Yangtze River Basin 2 . Overfishing of spawning fish is the main cause of the decline in fish stock 26 , leading to the so-called of 'fishing down the food web' phenomenon, whereby larger elements of a multispecies fish assemblage are successively removed and replaced by smaller elements, which typically represent lower trophic levels 27 . Illegal fishing, the effects of which are relatively small-scale but wide-ranging, is also inflicting considerable damage on freshwater fisheries. For example, electrofishing has increased throughout China, providing short-term profits and an efficient collecting method 28 . Although this practice is illegal, it remains rampant, having a huge detrimental effect on fish populations across China. In the Ganjiang River, a large number of fishing methods are employed, including traps, gill nets and; electrofishing, all of which are leading to overfishing and a dramatic decrease in fish biodiversity 29,30 .
Sand extraction is also having damaging effects on fish feeding, migration and reproduction grounds 29,31,32 . For example, a 50-flod increase in water turbidity occurred from 1998 to 2004 as a result of sand excavation, www.nature.com/scientificreports www.nature.com/scientificreports/ causing a 0.3 km 2 grass island to slide into Poyang Lake in 2004 33 . The combined effects of high temperatures and precipitation, mountainous and hilly terrain, and large-scale construction has also resulted in severe soil erosion and water losses in Jiangxi Province 34,35 . As a resulted, widespread alterations and loss of habitat have occurred, with subsequent decreases in fish resources 31,36-39 . Conservation recommendations. Although a variety of measures aimed at conserving fish biodiversity in the Ganjiang River have been implemented at a local and national level, these efforts remain inadequate and further conservation strategies are essential. Protection of freshwater fish should be carried out based on a comprehensive understanding of large-scale species richness patterns as well as patterns of endemism 40 . This approach would provide a platform for evaluating the current status of freshwater fish resources. This updated status information is essential in determining appropriate strategies of conservation management. Such data would also help highlight the overall status of freshwater fish in Jiangxi Province. Government-sponsored, national-scale screening of freshwater fish species in Jiangxi Province is therefore recommended. Furthermore, a system that responds to real-time threats such as dam construction would also help the implementation of conservation strategies prior to disruption.
The development of protected areas is also essential in preventing habitat loss and degradation; however, few such areas have been created for freshwater habitats. Instead, freshwater habitats tend to be protected incidentally due to inclusion within terrestrial reserves. Saunders et al. 41 suggested that freshwater species and habitats be directly conserved through the creation of freshwater protected areas. Recently, such areas have been established globally, playing an important role in conserving freshwater fish diversity [42][43][44] . In Jiangxi Province, approximately 156 protected areas have been established for the conservation of plants, animals and wetlands; however, no freshwater areas or fish passage facilities in Ganjiang River have yet to be included 5 . Protection of fish biodiversity therefore also requires the immediate establishment of freshwater protected areas.  Table 4. Physic-chemical parameters of the survey sites. www.nature.com/scientificreports www.nature.com/scientificreports/ According to Chinese Fishery Laws implemented in 1987, fishing is annually prohibited for two months from June 1st to July 31st along the Xiajiang, Xingan, Jishui and Ji'an reaches of the Ganjiang River. This initiative has played an important role in restoring populations of M. reevesii in the Yangtze River 6 . Fishing is also forbidden during the breeding season in mainstream rivers and tributaries in Jiangxi Province. However, this law is not widely respected, and many local residents fish throughout the year using traps and electro-fishing techniques 5 . To more effectively protect fish diversity and resources, enforcement of these laws is therefore required.
In addition to the above, the following conservation measures are also recommended: (1) restocking of economically important fish species; (2) enforced implementation of a close season; and (3) active development of sustainable aquaculture 4,5,45,46 .

Methods
Ethics statement. The study was approved by the Institutional Animal Care and Use Committee (IACUC) of Nanchang University, Jiangxi, China. All necessary permits were obtained for the described field studies from the IACUC of Nanchang University and the Yangtze River Fishery Administration of China. The handling of fish was also conducted in accordance with the guidelines on the care and use of animals for scientific purposes set by IACUC of Nanchang University, Jiangxi, China. All methods were carried out in accordance with relevant guidelines and regulations.

Study area.
Jiangxi Province which is located in the middle and lower reaches of the Yangtze River (24°29′14′′-30°04′41′′N to 113°34′36′′-118°28′36′′E) covers an area of approximately 166,900 km 2 . Its northern areas are relatively flat, while remaining areas are surrounded by mountains. Main rivers in Jiangxi Province are Ganjiang, Xinjiang, Fuhe, Raohe and Xiuhe, all of which flow into Poyang Lake and drain into the Yangtze River 6 .
Of these 47 , Ganjiang River is the longest, spanning more than 823 km and with a drainage basin of 82,809 km 2 . The riverhead is located in Shiliaodong (Yangdi Town: 116°22′E and 25°57′N), while its estuary is located in Wangjiangting (Wucheng Town: 116°01′E and 29°11′N). Its basin presents a mid-subtropical humid climate, with annual average precipitation of approximately 1580.8 mm and mean annual river runoff of about 2125 m 3 /s 47 .
The lower reaches of the Ganjiang River run from Xingan to the mouth of the river, covering approximately 208 km. This segment of river meanders among plains and hills, and has two tributaries (Yuanhe and Jinjiang River) draining into it. These lower reaches flow through Nanchang then divide into three branches: northern, middle and southern branches (Fig. 1). Yuanhe River, which originates to the west of the Wukong Mountain Range (114°10′E and 27°27′N), with its estuary located in Hehuguan (Zhangjiashan Town: (Fig. 1). Physico-chemical parameters (water temperature, the dissolved oxygen content and pH) were measured at each sampling site using a hand-held YSI multi-meter.
Sampling method. Fish samples were obtained quarterly, in April, July, October and January, 2017 to 2018.
The water depth in the study area was more than 1 m, and therefore, local fishermen were hired to catch the fish samples. Each site was sampled using a ground cage (5 m long × 0.5 m × 0.5 m, 5 mm mesh) and a gillnet (50 × 3 m) comprised of five panels (1.5, 3, 4.5, 6 and 7.5-cm bar mesh, respectively). Fishing was carried out overnight for approximately 10 h. Additional collection using a trawling net (100 m × 2 m, 5 cm mesh) was also performed at each site for about 1 h.
Where possible, fish were identified on collection then released. Those that were not were preserved in 10% formalin solution and taken to the laboratory for identification. Identification was carried out according to Zhu 48 , Chen 49 , Chu et al. 50 and Yue 51 . All fish specimens were deposited in the museum specimens of the fish existed at School of Life Sciences, Nanchang University.

Statistics analyses.
Fish dominance in each catch was determined by the index of relative importance (IRI) based on the number percentage, weight percentage and frequency of occurrence 52 : where %N i and %W i represent the percentage number and percentage weight of species i in the total catch, respectively, and %F i is the occurrence frequency of species i. When IRI i was greater than 10%, it suggested that the species i was dominant, while 1% < IRI i < 10% suggested that species i was common. Species diversity can be defined as the species richness in a certain area in a certain period. In this study, diversity indices were used to measure the spatial-temporal variation in fish species diversity as follows 53 where S is the number of species, N is the sum of individual number of species in the community, N i is the individual number of species i, and P i is the ratio of N i to N. A dataset covering all species collected at each site was then constructed, and similarity analyses were carried out based on the presence (1) or absence (0) of each species at each site 5 . Pairwise similarities among sites were then computed in order to create a similarity coefficient matrix. The hierarchical cluster, furthest-neighbour method with squared Euclidean distance was then used for cluster analysis based on the matrix. All analyses were performed using SPSS 13.0 software.
One-way analysis of similarities (ANOSIM) was used to determine significant differences in under non-metric multidimensional scaling (NMDS) ordination. First, a global R statistic was calculated to determine significant differences between all groups (analogous to the global F test in ANOVA). Significant differences at a global level were then determined using pairwise comparisons between sample groups to test for differences between pairs. In the global test, significance was set at P < 0.05 55 . All multivariate analyses were performed using Plymouth Routines in the Multivariate Ecological Research (PRIMER v5.0) software.
Abundance-biomass curves (ABC) were also obtained to determine differences in the community disturbance level. The ABC method, which is based on r and k-selection, was first proposed by Warwick 56 as a variant of K-dominance curves 57 to determine the effect of disturbances to invertebrate communities 58 . Communities in a stable state are dominated by k-selected species with slow growth rates, large body sizes, late maturation, and population sizes close to the environmental carrying capacity. In contrast, opportunistic species with fast growth rates, small body sizes and highly variable population sizes (r-selected) dominate systems where disturbance is recurrent or has occurred recently. These characteristics translate into two distinct patterns: a biomass curve lying on top of the abundance curve in undisturbed environments and the opposite in disturbed environments 59 . W-Statistics vary between −1 and +1 representing the difference between abundance and biomass curves. Values close to +1 represent a higher biomass than abundance curve (i.e. a stable environment), while values close to −1 suggest a reverse pattern (i.e. a disturbed environment). Meanwhile, values close to 0 represent a moderately disturbed community 60 .
Redundancy analysis (RDA) was also carried out to analyse the correlations between fish species composition and the measured environmental factors. Species composition and environmental data were log 10(X + 1)-transformed to meet the assumptions of multivariate normality and limit the effect of extreme data. All ordinations were carried out using CANOCO 5.0 61 .