The Kulekhani watershed is comprised of 43.6% forest, 34% sloping agricultural land, 9.2% shrub, 5.7% level and valley terraces, and 7.5% other land uses [21] . Total population of Kulekhani Watershed was 31,562, with more than 80% of them depending on agriculture for their livelihood [22] . Among them, 239 households have been involved in fish culture in 1630 cages in 2009 [10] . The climate of Kulekhani watershed varies from subtropical at low land to temperate at higher elevation [4] . The average summer temperature of Kulekhani watershed is 15˚C to 25˚C whereas the winter average temperature is 10˚C to 15˚C (2) It has four distinct seasons namely; pre-monsoon (March to May), monsoon (June to September), post monsoon (October to November) and winter (December to February). The reservoir receives ~1400 mm of rainfall in the monsoon season and ~60 mm during other dry seasons, and water level in the reservoir varies with season and power generation [15] .

Water samples for the analyses of water quality parameters and planktons were collected from three sites, namely Site A (upper reach of the reservoir), Site B (middle reach), and Site C (lower reach around cage culture) in the Kulekhani Reservoir during eight months (February, March, May, June, July, August, September and October) in 2007 (Figure 1). Site A represents a natural river inflow environment with minimal reservoir impacts, and the water quality and plankton’s community composition before and sometimes immediately after (depending upon the season) the impoundment of the river water. While the Site A turned into a river channel during dry and high-power demand season, it was impounded during the high-flow low-demand seasons. Site B, approximately at the middle of the reservoir, was flooded/impounded throughout the sampling seasons and was not directly affected by the cage culture activities. Site C, near the dam, was also always impounded and was represents an area heavily affected by aquaculture activities (cage culture).

The Kolmogorov-Smirnov test and Levene’s test were used to check normality and homoscedasticity assumptions prior to analysis. Parameters with unequal variances were either Log10(x + 1) or square root transformed. The water quality and plankton data were then analyzed by two-way repeated measure ANOVAs (site [three levels], month [eight levels]) with Tukey’s 1 degree-of freedom test for no additivity [30] using PROC MIXED in SAS (SAS institute Inc. Cary, NC, USA). Tukey HSD post hoc tests were applied in the presence of significant main effect. One of the zooplanktons (Filina) was excluded from statistically analyses because of its rare presence. Multivariate statistical models, principal component analysis (PCA) and canonical correlation analysis (CCA), were also used to analyze the data. All the statistical analyses were done in JMP (SAS Institute, Inc. Cary, NC) and R. Level of significance was set at α = 0.05 for all the tests.

The surface water temperature, pH and transparency varied between 13.3˚C - 25.5˚C, 7.3 - 9.5, and 32 - 246 cm, respectively. Similarly, DO, CO₂, alkalinity, total hardness, nitrate, and phosphate varied between 6.3 - 11.1 mg/L, 0.01 - 1.1 mg/L, 38 - 112 mg/L, 36 - 88 mg/L, 0.01 - 0.19 mg/L, and 0.1 - 0.26 mg/L, respectively. The water quality parameters varied considerably among the sampling stations in the reservoir throughout the sampling periods (Figure 2). However, there was no significant interaction effects (P > 0.05) of sampling periods and sampling locations (i.e. no months*sampling sites interactions effects) on variation in these water quality parameters, except for transparency and alkalinity (Table S1, Figure 2). These results indicate that the variation in water quality parameters in the reservoir is a unique effect i.e., the effect of seasons on water quality parameters is not significantly different among the sampling stations and vice versa. As the interaction effects were not significant, the unique effects of months and sampling sites were evaluated for each water quality parameters (Tables S1-3). The results showed that temperature, pH, DO, and free CO₂ varied significantly (P < 0.05) among the sampling periods, while nitrate

and phosphate varied significantly among the sampling stations (Figure 2, Table S2, Table S3).

Water temperature is one of the most important limnologic parameters which influences all the other parameters directly or indirectly. The surface water temperature in the Kulekhani Reservoir followed similar pattern of air temperature throughout the sampling periods. It supports that the seasonal variation in surface water temperature of a reservoir or lake is strongly affected by the atmospheric temperature of the catchment area [9] [31] . Like the water temperature, DO is another critically important water quality parameter which is often used as a pollution indicator of an aquatic ecosystem. The surface water DO in the Kulekhani Reservoir was always >6.3 mg/L, which showed that the reservoir water had sufficient oxygen levels for the survival of aquatic life, and also indicates the level of organic pollution in the reservoir is considerably low. The average DO in the reservoir water was found to be highest (11.0 ± 2.2 mg/L) in August and lowest (6.5 ± 0.2 mg/L) in October (Figure 2). Comparatively, higher concentrations of DO in August can be attributed to higher photosynthetic activities at favorable temperature. In contrast, the low DO content in the surface water during July, and immediately after the rainy season, was probably due to the flooded water containing high amount of silt loaded in the catchment area of the reservoir [9] [20] . Similar to the DO levels, slightly alkaline reservoir water with pH 8.8 ± 0.5 (always >8.2, except for Site A in February, Figure 2) was found to be most suitable for fisheries, and for the growth and development of various aquatic organisms [32] . As expected, CO₂ in this study was found to be relatively low throughout the study period, which was due primarily to relatively higher concentrations of DO, and low level of organic pollution in the reservoir water throughout the study period.

Unlike other water quality parameters, significant months*sampling sites interaction effects were observed for transparency and alkalinity (Table S1, Figure 2). Water transparency had an average value of 0.85, 1.22 and 1.60 m in Site A, Site B and Site C, respectively, and was higher in February-March but dropped to lowest in July-September (Figure 2). The lower transparency of the reservoir water in June-September was primarily due to the runoffs of sediment laden rain water in to the reservoir, and also due to the increase in phytoplankton density in the reservoir. The transparency of Site C was always higher than Site A and Site B. The lower transparency at Site A than other sites was mainly attributed to its location in the upper reach of the reservoir which is largely affected by the sediment load carried by the feeding streams [19] . The observed spatial and temporal variation in transparency can also be due to other factors including the growth of planktons, material derived from the shore by wave action, erosion, the amount of wind induced turbulence keeping material in suspension, the settling of suspended particles introduced by the river inflow [19] [33] .

The mean annual alkalinity observed in the present study (66.6 ± 18.7 mg/L) was within the range of “very productive” 51 - 67 mg/L [34] , and high productivity > 40 mg/L [35] water bodies. Alkalinity showed seasonal variations in all sampling sites, higher in February-June followed by a decrease from July-September (Figure 2), which is a typical trend in this region [10] . In contrast to alkalinity, total hardness (52.4 ± 13.3 mg/L) in the reservoir water was relatively constant among sampling sites, and throughout the sampling periods (Figure 2). Calcium hardness attributed to more than 80% of total hardness, which is similar to the previously reported hardness distribution in the Kulekhani Reservoir [19] . Based on the observed concentrations of total hardness in the reservoir water, the Kulekhani Reservoir can be classified as a soft water body (i.e., with total hardness 0.00 - 75 mg/L). Thus, the reservoir water is suitable for fisheries, with total hardness ≥ 15.0 mg/L [32] .

In contrast to the majority of the other water quality parameters, no significant seasonal variations were observed in phosphate and nitrate concentrations in the reservoir water. However, the phosphate and nitrate concentrations were significantly higher at Site A than both Sites B and Site C (Figure 2, Table S3). The productivity of an aquatic ecosystem largely depends on the concentration of these two macronutrients, phosphorus (P) and nitrogen (N). In number of lakes and reservoirs, the phytoplankton’s productivity is primarily limited by available P, thus, an increase in P concentration can result in eutrophication of these water bodies [36] . Traditionally, water bodies are N rich and P limited, thus, P is generally regarded as the major nutrient controlling primarily productivity [37] . However, N can also be either primary or co-limiting nutrient for phytoplankton production in some of the lakes and reservoirs [38] [39] . Unlike many other P limited lakes and reservoirs, the observed concentrations of phosphate in the Kulekhani Reservoir was much higher than the normal range (0.1 - 0.2 mg/L) for maintaining phytoplankton’s life cycle [40] . A large amount of nitrate is potentially accumulated in the Kulekhani Reservoir via floodwater run-off, atmospheric depositions, inflow of organic matters and the bacterial decomposition of both allochthonous and autochthonous organic matters. Interestingly, the reservoir water had relatively higher phosphate levels during the study period which is likely due to the preferential microbial activities as they change organic forms of P into inorganic soluble orthophosphate. Moreover, the reservoir receives water from the large catchment area and water flows through the settlement and heavily agricultural areas, thus, the P from the soap and other animals’ wastes could be main reason for such high phosphate levels in the reservoirs. It is also likely that the higher phosphate concentrations could be due to the use of agricultural fertilizers loaded with P in the watershed [41] followed by agricultural runoffs into the reservoir.

Three groups of zooplanktons (Copepoda, Cladocera and Rotifers), with eleven genera and Nauplius larvae, and four classes of phytoplankton (Chrysophyta, Cyanophyta, Chlorophyta and Pyrrophyta) with twelve genera were identified and quantified in the Kulekhani Reservoir water. The list of both the zooplankton and phytoplankton identified in this study are presented in supplemental document (Table S4).

The average zooplankton density in the reservoir water was 494 ind./L, and varied from 235 in February to 743 ind./L in July (Figure 3). The group Rotifers (48.19%) was most the abundant zooplanktons in the reservoir, followed by Copepoda (36.38%), while Cladocera had the lowest abundance (15.57%) in the reservoir water. Among the eleven genera of the zooplanktons, Cyclops was most abundant, with 25% contribution in all three sites, which was followed by Asplanchana (19.94%) and Keratella (13.17%) while Filina (0.025%) was least abundant followed by Alona (0.45) (Table 1, Figure 4,). The zooplankton communities showed different patterns of temporal and spatial variations during the whole study period, however, zooplankton’s density was generally lower in a downriver direction, except in October, (Table 1 and Table 2, Figure 4). Among all the eleven zooplanktons’ genera, Cyclops, Keratella, and Polyanthra

showed a significant months*sampling sites interaction effects (Figure 3). In contrast, seasonal variations were observed for Nauplius, Daphnia, Bosmina, Alona, Moina, Pompholyx, Asplachana, and Brachionus (Table 1 and Table S5). Similarly, a significant variation was observed among the sampling stations for the zooplankton Nauplius, and Daphnia, Pompholyx, Bosmina, Asplachana, Alona and Brachionus (Table 2 and Table S5).

The average phytoplankton density of the reservoir water was 2014 ind./mL. Like the zooplanktons, the abundance of the total phytoplankton also varied significantly among the sampling stations, Site A and Site B were significantly different than Site C in terms of total phytoplankton (Figure 3). The total phytoplankton density was significantly higher in May-August than the other sampling months (Figure 3). Among the four groups of the phytoplanktons, Chryssophyta was found to be with highest number of genera as well as individuals comprising 72.29% of the total phytoplankton in the reservoir. Among the twelve

genera of phytoplankton, Synedra (34.85%) was most dominant phytoplankton followed by Melosira (24.56%), and Fragilaria (0.475%) was the rarest genera followed by Amphora (1.09%). Similar to the zooplankton, significant interactions were also observed for phytoplankton genera Navicula, Melosira, Amphora, Chroococcus, Staurastrum and Scendesmus (Table S6, Figure 4). Density of Synedra and Anabaena, varied among the months and study sites (Table 1, and Table 2), while Anabaena, Peridinium and Synedra differed significantly among sampling station (Table 1, and Table S6).

Among all the three sites, Site A had the highest number of zooplanktons (39.3%) and phytoplankton (40.8%), followed by Site B (respectively, 34.2% and 35.2%), and with lowest percentage in Site C (respectively, 26.5% and 24.0%). The higher phytoplankton’s density at Site A was likely due to large amount of

nutrients carried by the feeding streams, which, consequently, increase zooplanktons’ food availability, increasing zooplanktons’ abundance. In contrast, the cage culture at Site C likely decreases the amount of food materials for zooplanktons feeding on phytoplanktons, as well as the fishes feed on zoo- and phytoplankton reducing their numbers. Moreover, the concentration of nutrients was lower as we move further down in the reservoir, towards the cage culture area, which couples with lower abundance of both the phytoplankton and the zooplankton. The higher abundance of total planktons in the months of summer (June, July, August, and September) and the lower abundance in February and March (Figure 3), is mainly attributed to the temperature followed by lower primary production.

Planktons’ population and their dynamic in aquatic systems are governed by complex interactions between biotic and abiotic factors. Among them, the physico-chemical condition of a water body is the major factor controlling species diver-

sity of planktons [11] . The observed variation of planktons in the Kulekhani Reservoir was primarily due to the variations in temperature, availability of food, DO and pH. For instances, temperature is a major determining factor in planktons’ development and in controlling of zooplankton’s hatching, growth and filtering rates. In general, zooplanktons have lower generation times, and their growth is delayed by low temperature compared to that of phytoplankton. The observed zooplanktons’ diversity in the present study couples with the average surface water temperature of the reservoir, with a significant correlation coefficient (R = 0.62). Similar, associations between temperature and zooplanktons have been previously reported in several water bodies [42] [43] [44] . In addition to the reservoir temperature, transparency, pH and alkalinity also showed positive correlation with the planktons’ diversity and abundance in the reservoir water.

While P is considered as the prime nutrient that limits primary productivity in lakes and reservoirs [45] [46] , numerous studies have shown that the biomass of phytoplanktons can be nitrogen-limited [39] [47] . The relatively higher concentrations of phosphate than nitrate, observed in the present study, indicates that the plankton’s density in the reservoir was mainly the function of N. In general, inorganic forms of N and P were usually depleted in spring very rapidly and their concentrations remain at the low level till late autumn. Under such conditions, planktonic organisms can grow much more slowly than their maximum growth rates, thus growth limitation can occur. This result can be supported by the low level of N in the months of February, March, May and June (Figure 2).

The population of the zooplanktons and phytoplanktons did not have peak at the same months; zooplanktons were highest in October while phytoplanktons were highest in July (Figure 3). The low population of the phytoplanktons in September and October when there was higher population of the zooplanktons might be due to depletion of the nutrients, low light penetration due to low transparency, and grazing by zooplanktons and fishes. It is well known that density of zooplankton controls the phytoplankton’s biomass as the zooplankton grazing is a strong modulator of phytoplankton in lakes and reservoirs [48] [49] . Thus, the higher population of the zooplanktons could have reduced the population of the phytoplanktons in September and October.

All factors described above might be of secondary importance on planktons’ biomass in the reservoir, if considerable plankton standing crop is removed by flushing [50] . If plankton community structure is affected by such flushing and thus residence time of water in the reservoir, we expect that both phytoplankton and zooplankton biomass will decrease during periods of high water discharge and that plankton species with high generation time will be replaced during flooding periods by species with shorter generation time [51] [52] . But this effect seemed to be less important in the Kulekhani Reservoir, because it was made for hydropower generation and flushing is done from the bottom which probably does not affect the planktons residing on the surface.

Shannon-Wiener index values for zooplankton were highest at Site A, followed by Site B and then Site C during the study period, however, it remained similar among the sites for phytoplankton (Figure 5). Shannon-Wiener index for both zooplankton and phytoplankton were relatively lower earlier in the season followed by a slight increase in May then remained similar through September (Figure 5). While its interpretation remained controversial in term of scale of pollution, the Shannon-Wiener diversity index has been used as a suitable indicator for water quality [53] [54] . The Shannon-Wiener diversity index values measured in this study for both zooplankton (1.03 - 2.09) and phytoplankton (1.40 - 2.01) indicate that the Kulekhani Reservoir water as moderately polluted water body [55] [56] . The observed general pattern of low pollutants with decrease in values for the zooplankton and phytoplankton diversity as we move downstream form Site A to Site C is similar to previously reported trends [57] . However, this diversity index greatly varies from study to another study, which makes it very difficult to make comparison among the studies and draw a common conclusion because of environmental variabilities among the studies [58] .

Principal component analysis (PCA) was performed on correlation matrix on

nine environmental and biological variables and showed that the first and second principle components (PCs) respectively explained 34.64% and 20.52% of the variability in the data (Figure 6(a)). Given the foregoing component coefficients, the first PC appears to be essentially a weighted difference between transparence, temperature, total zooplankton, total phytoplankton, pH, and CO₂. Moreover, the second PC appears to be the sum of nitrate, CO₂, alkanility, total hardness, phosphate, total zooplankton and total phytoplankton. It was also noticed that the total zooplankton and phytoplankton strongly positively correlated with nitrate and phosphate, and negatively correlated with transparency (Figure 6(a)). Similarly, total zooplanktons were highly positively correlated with total phytoplanktons (Figure 6(a)). The PCA biplot for sites and samples distinguished the environmental and biological difference among sites and samples (Figure 6(b)). As a whole, Site A was clearly identified by being fresher and richer in nutrients (phosphate and nitrate), low transparency, high zooplankton

and phytoplankton density. The water quality and biological community in Site B differed than those in the Site A and Site C. Site B was categorized by medium nutrients, medium transparency, medium zooplankton and phytoplankton density, and Site C was characterized by lower nutrients, higher water transparency, low zooplankton and phytoplankton density. High absolute correlation values between PCs and the original variables, also referred as the loading of the variables on PCs, are useful to identify which parameters accounted for most variation in the data set [59] . PCA results in this study indicated that physical- chemical factors transparence, temperature, pH, CO₂, nitrate and phosphate played important role in overall environmental variation. As identified by PCA for zooplankton community, zooplankton genera Cyclops, Daphnia, Alona, Brachionus, Moina, Keratella, Polyuanthra, Asplachana, and Nauplius have greatest contributions to the variation of the zooplankton community. Similarly, Synedra, Melosira, Fragilaria, Amphora, Anabaena, Peridinium, Oscillatoria and Scendesmus played an important role in the variation of phytoplankton community composition (Figure 6).

The canonical correlation analysis (CCA) between environmental variables and zooplankton showed that two pairs of significant canonical variables, with correlations of 0.98 and 0.97, which explained 77.9% of the variation (Table S7). The CCA showed strong association of environmental variables (transparency, DO, temperature, and CO₂) with select zooplankton genera (Cyclops, Moina, and Polyanthra), and Nauplius. In contrast, one pair of canonical variable between environmental variables and phytoplankton, with correlation of 0.99, accounted for 77.5% of the variation. Strong positive correlations were observed for select environmental variables (DO and alkalinity) with select phytoplankton genera (Melosira, Anabaena, and Scendesmus) (Table S7).

The significant canonical correlation between water quality and planktons, and between zooplankton and phytoplankton suggest that planktons’ distribution is strongly associated with reservoir water quality and other planktonic community in the reservoir [11] . The high correlation between the planktons and temperature can be described as the effect of temperature on growth and development of planktons. The negative correlation between transparency and planktons in this study indicates decrease in light penetration capacity and transparency with higher density of planktons in the reservoir. Moreover, a light limitation is expected to prevail for phytoplanktons [60] . Thus, these phytoplanktons can be limited either by the ability to convert excess nutrients into biomass or by growth rate [61] which is probably the reason for the rapid and irregular fluctuation of planktons in the reservoir. In addition, nitrate and phosphate are major limiting nutrients for the growths of phytoplankton, and ultimately the growth of zooplankton. The results also showed that, inorganic of N and P were very rapidly depleted in spring and their concentrations remain at low levels till until late autumn. Such variation in nutrient concentrations and unfavorable temperature were probably the reasons for lower planktons’ abundance in February, March, and October (Table 2, Figure 3 and Figure 4).

The population of the phytoplankton is dependent upon the population of the zooplankton, as zooplankton grazing was identified as a strong modulator of the phytoplankton yield in lakes and reservoirs [62] . The results from the present study showed a pattern of low or inverse correlation between phytoplankton and zooplankton population density. It has been previously reported that the grazing effect of the herbivorous zooplankton is sufficient to limit the phytoplankton population in time as well as in quantity. However, some studies suggest that zooplankton avoid living in the area with dense phytoplankton, may be due to unfavorable chemical condition produced by phytoplankton populations’ i.e. exclusion by phytoplankton [63] . While a strong correlation between zooplankton and phytoplankton was observed in this study, whether such relationship was due to grazing or unfavorable chemicals remains to be tested.