The split-beam hydro-acoustics surveys were conducted to learn about the diel migration and distribution characters of Toxabramis houdermeri and to assess its current abundance in Baise Reservoir of China in October 2013. The study revealed the average density of T. houdermeri was 258 individuals/1000 m3, and gradually increased from downstream, through middle to upstream (203 individuals/1000 m3, 252 individuals/1000 m3 and 318 individuals/1000 m3, respectively). The study also showed that the target species always distributed in the 2–42 m water layer during the daytime, before migrating upward around dusk which led the fish density to sit in the 2–12 m water layer, to reaching a peak value of 1740 individuals/1000 m3 during the night. Finally, the fish volume density was found to positively correlate with the concentration of chlorophyll, dissolved oxygen (DO) and pH value in their studied areas. In conclusion, the results presented that T. houdermeri preferred a relatively higher average chlorophyll concentration of 3.58 ug/L DO of 6.33 mg/L, and pH of 7.96 in the upstream and upper layers of Baise Reservoir.
Introduction
Over the past half century, hundreds of artificial reservoirs were constructed in China and the riverine ecosystems have been changed gradually at the same time. So far, these changes of the natural riverine ecosystems have aroused the attention of natural scientists and professional researchers internationally. There were many researches focusing on the comparison between the current state of fish communities when the reservoirs were built at first (Gido et al. 2000, Matthews et al. 2004). As the issue was highlighted, the research was conducted timely, concentrating on a dominating species of cyprinid fish Toxabramis houdermeri Pellegrin, 1932 in Baise Reservoir of China in 2013. This species is a kind of small and pelagic fish which has been reported to be distributed widely along the River Pearl in Guangxi province according to the investigation of Freshwater Fishes of Guangxi, China 2005. Additionally, many researchers have reported that there was a considerable amount of pelagic fish species preying on fish eggs and larvae (Bailey & Houde 1989, Swain & Sinclair 2000). Thus, such a huge amount of T. houdermeri was undoubtedly having a significance influence on the dynamics of fish populations and biodiversity in the freshwater ecosystems of Baise Reservoir. As little as 41 years ago, a few researchers (Zhu 1977) had taken measures to restrain the quantity of Hemiculter leucisculus and Toxabramis in Dashahe and Tongsha reservoirs of China (where the quantity reached 67 kg/h and 62 kg/h respectively) by floating lift net. So far, fishery resource surveys have never been conducted in Baise Reservoir on this special pelagic species. It was therefore meaningful to assess the stock of the species and to learn about its life habits. Hydro-acoustics equipment is an efficient and harmless tool, and has been used in fish stock assessment and fish behaviour research in various fresh water lakes and reservoirs for many years. In 2006 and 2007, there were surveys carried out in Tunisian man-made lakes and a sampling strategy for fish population studies using hydro-acoustics was employed (Djemali et al. 2008, 2010), which paid more attention to vertical fish distribution as shown elsewhere (Brandt et al. 1991, Shirakihara et al. 2001, Čech et al. 2005, Jarvalt et al. 2005). Over the past three decades, vertical (or down-looking) hydroacoustics has become increasingly important to the assessment of fish stocks in deep lakes and reservoirs (Thorne 1998, Cyterski et al. 2003). To make the sonar data referring to density, biomass or size distribution more credible, hydro-acoustic assessment method was combined with traditional methods, such as purse seines, mid-water trawl, electro fishing and angler surveys. Furthermore, it has been suggested that the distribution and the behaviour of fish species in lakes and reservoirs are influenced by a range of abiotic, biotic, and behavioural factors such as temperature, oxygen concentration and vertical distribution of predators/prey (Lucas et al. 2002). Thus, those parameters were always considered together with the sonar data to analyze fish distribution, along with spatial and temporal behaviour.
This study was conducted mainly with the objectives of presenting distribution characteristics, demonstrating the diel migration pattern, obtaining the abundance estimation of T. houdermeri, and the brief analysis of the correlation between fish distribution pattern and biotic/abiotic factors in Baise Reservoir, China, 2013.
Material and Methods
Study area
Baise Reservoir is located 22 km away from the city of Baise in Guangxi province in the south-western part of China, at 23°55′16″ N and 106°27′21″ E, at an altitude of 210 m (Fig. 1). The reservoir was established with a surface area of 136 square kilometers and a volume capacity of 5.66 billion cubic meters in 2006. The maximum depth of the three investigated regions were 70.39, 83.03 and 98.41 m in upstream, middle and downstream respectively, and the width ranged from 200 to 800 m.
Acoustic surveys and water quality test
The surveys were conducted in the main channel of Baise Reservoir, almost from the Yintun dam to Boai estuary during October 23th to 29th in 2013. Three typical regions (downstream, middle and upstream) were selected based on the hydrology regime and geographic feature of the reservoir to perform diel acoustic surveys (Fig. 1). The whole system was calibrated before the regular diel acoustic surveys were conducted according to the standard process (Foote et al. 1987) by a copper calibration sphere of 13.7 mm in diameter and nominal target strength of -45 dB re 1 m2. All acoustics surveys were performed on a mobile scale vertically by a split-beam scientific echo-sounder Simrad EK60 operated at a frequency of 200 kHz (ES200-7c) with a beam opening angle 7*7° at 3-dB down points and the pulse length at 0.256 ms. The transducer was mounted on a 11 m long power vessel by a bracket and submersed to a depth of 0.6 m. At each investigated region diel acoustic surveys were carried out and repeated the next day with a zig-zag trajectory at a mean speed of 4.5 knots. Acoustic surveys were performed at three time period each day respectively 9:00 a.m., 4:00 and 7:00 p.m. Each survey conducted took about one hour. Generally, surveys at night were launched about one hour after sunset. The degree of coverage values were calculated for each survey using the formula of Aglen (1983). Dc = L/√ where L = distance covered by the boat (in m), A = survey area (in m2).
Each investigated region of the reservoir was divided into seven layers from the water surface with a 5-meter interval, with a few water chemistry factors tested by a calibrated YSI 6920V2-2 MPS probe in each layer. Six typical points ranging from inshore and offshore were picked out at each region to perform detections on water chemistry factors. These factors, including temperature (Temp), dissolved oxygen (DO), chlorophyll, pH and ammonium N were focused in this study.
Data processing
Raw acoustics data was converted and analyzed by Sonar5 (Balk & Lindem 2005) post-processing software. Vertically oriented transducers were subject to a hydro-acoustics blind-zone. This effect was due to the relatively small acoustics sampling volume near the transducer (Lindem & Sandlund 1984) and the fishes' natural fear of the boat on which the equipment was mounted (Olsen 1979). In this research acoustics data within 2 m of the transducer were excluded. A bottom line of each transect was automatically detected using image analysis and was manually defined with the upper 0.5 m relative to the bottom line excluded from analysis. Each acoustics survey transect was cleared of noise (bubbles, debris, bottom structures etc.) by setting appropriate threshold and manual deletion, so mainly only target echoes remained for further analysis. The target species Toxabramis houdermeri which dominated the reservoir in quantitative terms was highlighted in this research. The body length of the target species collected by fishermen ranged from 5 to 12 cm and the corresponding target strength (TS) values from -61.8 to -50.3 dB were tested by a tethered method in situ which was introduced in the research of Zhang (2013) to measure the acoustic target strength of two economically important species.
Table 1.
Weather condition, degree of acoustic coverage (distance covered/square root of surface) and diel volume density of Toxabramis houdermeri in three investigated regions.
Fig. 2.
Spatial and temporal volume density (mean + SD) variation of Toxabramis houdermeri in Baise Reservoir in October 2013. Capital letters (A, B, C) represent the diference among regions (upstream, middle and downstream) and lowercase letters (a, b, c) represent the difference in diel (a.m., p.m., night).
These are the banded grouper Epinephelus awoara (Temming & Schlegel, 1842) and threadsial filefish Stephanolepis cirrhifer (Temming & Schlegel, 1850) in coastal areas of the South China Sea. Furthermore, the study calculated the abundance of T. houdermeri using “sv/ts scaling” approach and the TS derived from single echo detections that was scaled from -62 to -50 dB and separated into six groups with 2 dB interval that excluded the signals weaker than -62 dB and stronger than -50 dB from total acoustical abundance assessment.
Fish sampling
To obtain the accurate fish composition of Baise Reservoir, light net (length*width: 16*18 m, mesh: 0.5 cm) and gill net (length*height: 200*10 m, mesh size 6 cm) methods were integrated and the daily fish captures were recorded by three native veteran fishermen using other methods such as set net and electric fishing. Light nets and gill nets were positioned one hour after sunset and harvested the next morning before daybreak.
Statistics analysis
In order to test the diel distribution of Toxabramis houdermeri in the longitudinal (upstream, middle and downstream) and vertical gradient (different water layers), an ANO VA analysis was performed with Statistic software SPSS 21.0 with a set at 0.05. The variation of water chemistry factors in spatial profile was also tested by one-way ANO VA analysis and the correlation between fish distribution and water chemistry parameters was analyzed by Pearson correlation (2-tailed) with SPSS 21.0. The whole graphs in the paper were figured by Origin 8.0 and Arc View GIS software.
Results
Abundance
The main species composition in the investigated reservoir were Toxabramis houdermeri, Tilapia niloticus, Cyprinus carpio, Hypophthalmichthys nobilis and Hemiculter leucisculus. Depending on the analysis of fish captured, the amount of T. houdermeri was fairly spectacular, reaching 100 kg per light net (length*width: 16*18 m, mesh: 0.5 cm) every single night. The main captures of gill net (length*height: 200* 10 m, mesh size 6 cm) were Tilapia and Cyprinus, about 10–20 kg and 2–3 kg respectively per night. The captures recorded by three native veteran fishermen presented that the weight percentage of T. houdermeri among the whole captures was 44.6 %, 50.6 %, 56.8 % and the quantity percentage of T. houdermeri was 99.3 %, 99.3 % and 99.4 % respectively in the three investigated areas (downstream, middle and upstream). Additionally, the average fish abundance of T. houdermeri in Baise Reservoir obtained by hydro-acoustics method was presented in Table 1. The average density of T. houdermeri was 253 individuals/1000 m3, which means a total quantity of 1431.98 million individuals existed in the whole reservoir at that time. As the target individuals were generally weighed 7 g, the total stock of T. houdermeri in the studied reservoir was assessed to be approximately 10020 t.
Fish distribution in longitudinal transect
Responding to the acoustic surveys, the volume density estimation of T. houdermeri varied in temporal and spatial distribution in Baise Reservoir of China in October 2013 (Fig. 2).
Firstly, the average volume density variation of T. houdermeri was temporally consistent in the three investigated regions. It was higher during the night-time. Secondly, the average volume density of T. houdermeri continuously increased from downstream to upstream regions during the whole day. The result from one-way ANOVA statistics analysis indicated that the volume density variation of T. houdermeri in the investigated regions was significantly different in spatial (F = 10.005, p < 0.01, df = 2) and temporal scale (F = 4.654, p = 0.01, df = 2). Further comparison of the fish density variation performed by the LSD test presented no significant difference between a.m. and p.m. (p = 0.393 > 0.05).
Vertical distribution and migration
The vertical distribution of target individuals in three investigated regions presented similar characteristics; that the volume density of T. houdermeri was mainly concentrated among the upper four water layers which accounted for 98.07 % of the density of the whole investigated layers (Fig. 3). The concentration degree at night was higher than the day. The volume density of T. houdermeri in the 2–22 m water layer accounted for 84.1 % of the density of the 2–82 m water layer during day, increasing to 95.54 % at night which indicated a remarkable diel rhythm of vertical migration. Based on the one-way ANOVA statistics analysis, the research revealed that the fish density of T. houdermeri was significantly different (F = 131.823, p < 0.01, df = 7) in the defined water layers. Further comparison of the fish density variation performed by the LSD test presented that the 2–12 m layer was significantly different from other defined layers (p < 0.01), the 12–22 m layer was significantly different again from other defined layers, except the 22–33 m layer (p = 0.137), while there was no significant difference among the 22–82 m layers.
Table 2.
The correlation analysis between water chemistry factors and volume density of Toxabramis houdermeri in different regions and different water layers at 0.01 level.
Water chemistry factors
The water chemistry factors were tested temporally and spatially immediately following each acoustics survey. The water chemistry characters varied among layers and regions (Fig. 4). In vertical direction from water surface to 30 m water layer, the mean ±SD values of water Temp, DO, pH, chlorophyll and ammonium N concentration were 24.85 ± 0.45 °C, 5.76 ± 0.52 mg/L, 7.94 ± 0.07, 2.73 ± 1.88 ug/L and 0.09 ± 0.013 mg/L respectively. The water Temp and chlorophyll concentration presented the basically consistent trend of continuous declination from 25.41 to 23.92 °C and 5.31ug/L to 0.24ug/L respectively while the DO values initially declined from 6.48 mg/L to 5.07 mg/L and then increased to 5.55 mg/L which was similar to pH values variation from 8.05 to 7.89 and then increased to 7.93. Furthermore, the ammonium N values first increased from 0.08 mg/L to 0.1 mg/L and then declined to 0.08 mg/L. In contrast, the average water Temp (24.64 ± 0.38 °C) and ammonium N value (0.086 ± 0.008 mg/L) of the upstream was the lowest compared with the values in middle (24.81 ± 0.38 °C, 0.088 ± 0.011 mg/L) and downstream (25.09 ± 0.47 °C, 0.096 ± 0.016 mg/L). The average pH value (7.96 ± 0.06) and chlorophyll concentration (3.59 ± 2.25 ug/L) of upstream was the highest compared with the values in middle (7.95 ± 0.06, 2.63 ± 1.65 ug/L) and downstream (7.9 ± 0.07, 1.96 ± 1.24 ug/L). The average DO values in upstream, middle and downstream were 5.78 ± 0.6 mg/L, 5.81 ± 0.46 mg/L and 5.69 ± 0.51 mg/L respectively. Furthermore, the statistic results revealed that the concentration of chlorophyll and pH values were significantly different among downstream, middle and upstream (ANOVA p < 0.01, p < 0.01 respectively), while the concentration of DO had no significant difference (ANOVA p = 0.185). In addition, the concentration of chlorophyll, DO, and pH in different water layers were significantly different (ANOVA p <0.01, p < 0.01, p < 0.01 respectively).
Correlation between fish distribution and biotic/ abiotic factors
To confirm the correlation between the distribution characteristics of T. houdermeri and the water chemistry factors, the statistic correlation analysis Pearson two-tail test was performed by SPSS 21.0. The result presented that the chlorophyll concentration, DO and pH value were positively correlated with the volume density distribution characteristics of T. houdermeri in the investigated regions (Table 2). Specifically, a relatively higher volume density (318 individuals/1000 m3) of T. houdermeri in the upstream of Baise Reservoir corresponded to higher chlorophyll concentration (3.59 ug/L), DO (5.78 mg/L) and pH values (7.96).
Discussion
Abundance evaluation
The hydro-acoustics method has been proven to be an environmental, scientific and cost-effective way for fisheries to conduct resource assessment and management (Burczynski et al. 1987, Argyle 1992). Baise Reservoir is believed to be an especially valuable location for acoustical studies as it is capacious, calm with little boat traffic, and low human activity. Although it is claimed to be impossible to distinguish fish species by acoustic (Draštík et al. 2008), the community structure of fish in Baise Reservoir was confirmed by net captures and T. houdermeri was proved to be the absolutely dominant species, making the assessment of fish abundance by hydro-acoustics method more reliable. Since Baise Reservoir was established in 2006 the riverine ecosystem has changed and T. houdermeri has gradually come to dominate the freshwater area. Such evolution could be rightly illustrated by the classical theory of r-selection (Jennings et al. 1998) which states that the small size species with strong reproductive and rapid growth capacities will be dominant and replace the larger sized species with long life history in a disturbed environment. In this study, the abundance assessment of T. houdermeri was inconsistent with temporal findings. According to the former studies, it has been reported that behavioural factors can complicate acoustic estimates of fish abundance. Specifically, fish migration in vertical (different water layers) and horizontal (near-shore and off-shore) orientation between day and night can complicate the estimation of fish abundance. The target species was proved to hide in deeper water layers and shelters near-shore during day time which made the estimation of fish abundance inaccurate. The target species preferred the upper water layers and off-shore regions at night which provided a more appropriate condition for fish abundance evaluation. Thus, the abundance variation of T. houdermeri in time can be partially illustrated by the different fish behavioural factors.
Firstly, there is less disturbance from human activity at night than the day. In addition, it has previously been suggested that fish would be better avoiding vessels during the day than at night (Olsen et al. 1983b, Soria & Fréon 1991). Secondly, fish are dispersed in the open water at night, contributing to a less solid acoustics shadow (Eckmann 1991, Kubečka & Wittingerová 1998, Malinen et al. 2005) which makes the assessment more accessible. Furthermore, the target species that stay closer to the near-shore during daylight can be undetectable by hydro-acoustics (Guillard 1998, Guillard et al. 2004, Simmonds & MacLennan 2005). In this study, the target species was mainly distributed offshore in the upper 42 m at night. In conclusion, the abundance estimation during night tends to be more accessible and accurate.
Fish distribution
A typical pattern of fish distribution was demonstrated in this study. In Baise Reservoir, estimates of fish abundance were highest in upstream (tributary areas) and decreased towards downstream (dam areas). Similar distribution pattern has been reported by many other authors from European and American reservoirs (Vondracek et al. 1989, Fernando & Holčík 1991, Brosse et al. 1999, Świerzowski et al. 2000, Gido et al. 2002b, Vašek et al. 2003), with fish sampled using hydro-acoustics, gill nets or other fishing gear. Generally, it was clear that the fish community has preferences and selectivity along environmental gradients in terms of density (Prchalová et al. 2008). Fish were seldomly distributed homogeneously or randomly within their environments. The distribution of fish community was always determined by the physical habitat, chemical, and historical constraints of environmental conditions (Benson & Magnuson 1992, Borcard et al. 1992). All the elements referred above were supposed to contribute to the distribution characteristics of T. houdermeri in Baise Reservoir and the correlation between fish distribution and water chemistry factors was confirmed. In the upstream of Baise Reservoir, the abundance of T. houdermeri, concentration of chlorophyll, pH values and DO values were relatively higher than the value of other investigated areas (middle and downstream). The abundance of T. houdermeri was tested to be positively correlated with the concentration of chlorophyll, pH values and DO values. Furthermore, variation of the three water chemistry factors showed that only the concentration of chlorophyll and pH values were significantly different. While the DO value was not significantly different in longitudinal orientation, it had a high enough value to account for the physiological activity of the target species among the three investigated regions. Consequently, the distribution characteristics of T. houdermeri in horizontal directions was not directly affected by the DO value of the studied regions and the target species preferred a relatively higher average chlorophyll concentration of 3.58 ug/L and pH value of 7.96 in the upstream of Baise Reservoir in October 2013. Similar studies had been conducted in 40 lakes and showed that pH contributed significantly to fish species distribution (Öhman & Buffam 2006). The species (roach, bream and crucian carp) recorded in these lakes were similar to T. houdermeri, which dominated in Baise Reservoir. According (Öhman & Buffam 2006), roach was a species known to need high pH to survive and Degerman & Lingdell (1993) found roach to be missing from lakes with pH under 5.5–5.9. This would be a proper interpretation to the distribution of T. houdermeri in the studied area with high pH environment.
Fish migration
Although the chlorophyll concentration, DO and pH gradient proved to be positively correlated to the volume density distribution of Toxabramis houdermeri, the diel migration pattern of T. houdermeri was not directly affected by these water chemistry factors. The chlorophyll concentration, DO and pH gradient were not significantly different between day and night (ANOVA p = 0.537, p = 0.439, p = 0.369, respectively).
Thus, there must be some other factors that guided the diel migration pattern. Former research reported that biotic interactions such as competition of a water body can highly modify the spatial distribution of fish (Diehl & Eklov 1995, Mehner et al. 2005). The distribution and variation of food (i.e. plankton) can also account for fish density and distribution patterns. As many researchers (George & Jones 1987, George & Edwards 1976, Benson & Magnuson 1992, Öhman & Buffam 2006) mentioned before, there were many factors such as prey pressure, food and spatial competition and water chemistry impacting the fish distribution. In our research, the potential prey group (-70 dB < TSc < -62 dB) was always aggregated simultaneously in the same water layer of 2–42 m during day and in 2–12 m during night, which was supposed to induce the diel migration of T. houdermeri. In addition, prey pressure from predaceous groups also existed according to the TS composition of acoustic surveys and fish captured by multi-net approach. Based on the results mentioned above, the conclusion can be drawn that the prey pressure and food accessibility must contribute to the diel vertical migration pattern of T. houdermeri while the chemistry factors considered in this study did not guide the migration. There are other factors associated with the fish migration such as light intensity (Mackinson et al. 1999, Mowbray 2002, Cardinale et al. 2003) and competition of space, and these were not considered in this study.
Riverine ecosystems have been changing gradually, and to learn about the newly formed characteristics of fish composition and distribution, and the correlation to water chemistry factors are essential projects for ecology restoration and sustainable development of fishery resources. This has been the purpose of this work, and the primary focus of this study. Further studies on the specific interactions between T. houdermeri and these chemistry factors in the investigated reservoir are expected.
Acknowledgements
The study was specially funded by Agro-scientific Research in the Public Interest, China (N201303048). We would like to thank the fisheries bureau and fishery administration of the River YouJiang and the professional fisherman Liang Zhao and his colleagues for their help on the research. Additional thanks was given to Brooke Hazelgrove, Long Fei, Huang Xi, Chang Tan and Xi Xi Xiao who helped to revise the manuscript.
