Phytoplankton communities determine the spatio-temporal heterogeneity of alkaline phosphatase activity: evidence from a tributary of the Three Gorges Reservoir

In order to reveal the role of phytoplankton in the spatio-temporal distribution of alkaline phosphatase activity (APA), monthly investigations were conducted in the Xiaojiang River, a tributary of the Three Gorges Reservoir in China. Different APA fractions, environmental parameters, and phytoplankton communities were followed. High spatio-temporal variations of APA were observed, with the highest value in summer and the lowest in winter. The annual average APAT (total alkaline phosphatase activity) ranged from 7.78–14.03 nmol∙L−1∙min−1 with the highest in the midstream and the lowest in the estuary. The dominant phytoplankton phyla in summer and winter were Cyanophyta and Bacillariophyta, respectively. The mean cell density in the midstream and in the estuary was 5.2 × 107 cell∙L−1 and 1.4 × 107 cell∙L−1, respectively. That APA>3.0 μm was significantly higher than APA0.45-3 μm indicating phytoplankton was the main contributor to alkaline phosphatase. Correlation analysis indicated the dominant species and cell density could determine the distribution pattern of APA. Turbidity, total phosphorus, chemical oxygen demand, water temperature (WT), pH and chlorophyll a were proved to be positively correlated with APA; soluble reactive phosphorus, conductivity, transparency and water level(WL) were negatively correlated with APA. It was concluded that spatio-temporal heterogeneity of APA determined by phytoplankton communities was related to WT and WL.

Size-fractionation of APA. The average size-fractionated APA indicated that APA <0.45 μm accounted for the major portion of APA T , whereas the average APA >3.0 μm was higher than APA 0.45-3 μm (Fig. 2a). The average APA >3.0 μm accounted for 28.1% of APA T and APA 0.45-3 μm accounted for 16.7%. In addition, the size-fractionated APA (APA <0.45 μm , APA 0.45-3 μm and APA >3.0 μm ) in summer and autumn are higher than those in winter.
At spatial scales, the average APA T consisted of 30.2% APA >3.0 μm and 20.4% APA 0.45-3 μm in all sites. The APA <0.45 μm kept a relatively stable and high level. Both APA 0.45-3 μm and APA >3.0 μm in midstream (GY01) are higher than those in estuary (XJ).
Phytoplankton communities. Bacillariophyta was the dominant group in winter and spring (72.7% of total cell density on average, Fig. 3a) except cyanophyta are dominant in April. In summer and autumn, phytoplankton mainly consisted of Cyanophyta (65.6% of total cell density on average) except the Cryptophyta accounted for 88.4% in August. The mean algal cells density was the highest in July 2014 (1.27 × 10 8 cell•L −1 ), and the lowest in January 2014 (1.3 × 10 6 cell•L −1 ). The cell density was higher in summer and autumn than in spring and winter. Cyanophyta dominated the phytoplankton in upstream (QM01, QM02, GY01, GY02, 69.9% of total cell density on average). In the downstream (HS01, HS02, XJ), phytoplankton mainly consisted of Bacillariophyta (35.3% of total cell density on average, Fig. 3b). Spatio-temporal characteristics of chlorophyll a and environmental parameters. Spatio-temporal variations of chlorophyll a (Chl a) and environmental parameters could be observed in Fig. 4. The values of Chl a, total phosphorus (TP), chemical oxygen demand (COD) in spring were apparently higher than the values of other seasons, because the river suffered a Microcystis sp. bloom in May, which also resulted in the minimum values of SRP and transparency (SD) emerged. The levels of TP, COD, Chl a, water temperature (WT), turbidity (Turb), dissolved oxygen (DO) and pH stayed low in winter, contrary to the values of SRP and SD. The values of SRP fluctuated more frequently than other parameters in different seasons. At spatial scale, the concentrations of SRP and Chl a were higher in estuary than in upstream.
Relationships between APA and environmental parameters. SRP concentrations showed negative correlation to APA <0. 45 μm (Fig. 5a), APA 0.45-3 μm (Fig. 5b), APA >3.0 μm (Fig. 5c) and APA T (Fig. 5d). The Spearman correlations among environmental variables and APA <0.45 μm , APA 0.45-3 μm , APA >3.0 μm and APA T were presented in Table 1. Turb, TP, COD, WT and pH were positively correlated with APA fractions. Cond., SD and WL were negatively correlated with APA.  Redundancy analysis (RDA) was performed to analyze the relationship between environmental parameters and size-fractionated APA. The ordination diagrams of environmental variables and size-fractionated APA for axis 1 and axis 2 were shown in Fig. 6. The Monte Carlo test revealed that the first canonical axis and all canonical axes were significant (F = 25.932, P = 0.002; F = 3.086, P = 0.002; 499 random permutation). For environmental variables and size-fractionated APA, all canonical axes cumulatively explained 83.3% of the variance in APA-environment relationships, and the first two canonical axes accounted for 26.5% and 31.5% of the variance separately. The first axis was positively correlated with DO (0.57), COD (0.65) and negatively correlated with SRP   Table 1. Spearman correlations between APA and 10 environmental variables: water temperature (WT); transparency (SD); dissolved oxygen (DO); conductivity (Cond); pH, turbidity (Turb); total phosphorus (TP); chemical oxygen demand (COD); water level (WL) in the Xiaojiang River. *P < 0.05. **P < 0.01.

Discussion
APase has different sources, different kinds of bacteria, phytoplankton and zooplankton can excrete extracellular phosphatase 19 . Specific APA was related to different phosphatase producing organisms. In our investigation, APA >3.0 μm contributed in average 28.1% in the APA T , while bacterial APA accounted for 16.7%, APA in algal fraction (APA >3.0 μm ) was higher than that in the picoplankton/bacterial fraction (APA 0.45-3 μm ). Though many (not all) phytoplankton cells have a host heterotrophic bacteria inhabiting or in close association with cells, and the overlapping size on the filter also influenced the final data, which making it difficult to assign the different size fractionation by filtration to individual cells alone, it could be admitted that the coarser fraction (APA >3.0 μm ), mainly from algae, was conventionally defined as "algal APA" 20 due to phytoplankton was the main contributor according to its size, biomass and physiological activity. It was confirmed APA >3.0 μm accounted for the major portion of total APA (55-87.9%) than APA 0.45-3 μm

21
. It could be deduced that the phytoplankton was the major contributor of bulk APA based on the larger proportion of APA >3.0 μm (52.73%) than APA 0.45-3 μm (21.09%) 22 . Therefore, the phytoplankton contributed greatly to APA production. Meanwhile, the picoplankton/dissolved APA (APA <0.45 μm ) kept a relative stable and high level (53.4% of the APA T ). Some studies showed that the picoplankton/dissolved APA represents a significant part of the total activity. For example, Labry et al. reported that picoplankton/dissolved APA represented 13% to 44% of APA T in the Bay of Biscay (on the French Atlantic coast) 23 . Higher proportions were recorded in the northern Red Sea (42-74%) 24 . The dissolved APase can be liberated into the environment through the lysis of dead phytoplankton cells and from cells damaged by zooplankton grazing 25 . The high values may result from physical damage of cells by water current and zooplankton grazing on phytoplankton. Nevertheless, some study found that the dissolved APA might origin from bacteria 26 . In order to elucidate the origins of dissolved APA, the dissolved alkaline phosphatase was capsulated into reverse micellar media, and it was proved that the different behaviors of dissolved phosphatase of surface and overlying water might be due to the different origins, with the former being algae and the latter being bacterial 27 . The results of Song et al. (2005) couldn't identify the exact origins form, the fraction <0.45 µm contains some pico-bacteria and some pico-phytoplankton and can't be called as the dissolved fraction. Here we changed it as the picoplankton/dissolved fraction. Besides, the positive relationships between APA and the environmental parameters that have been treated as the indexes of the productivity and trophic status, such as Chl a, Turb and COD, and the negative relationship between APA and SD can also indicate that the phytoplankton is the main contributor of APA.  Different algal species showed significant different secreting ability of APase. In this study, phytoplankton communities were dominated by Bacillariophyta in winter. Pyrrophyta, Bacillariophyta and Chlorophyta can easily produce extracellular phosphatase as evidenced by ELFA labeling 21 . The low APA >3.0 μm during this period may result from the low algal cell density of phytoplankton and the increased concentrations of SRP. When the Pyrrophyta subdominated the phytoplankton community in May, the APA >3.0 μm peaked. Results in some shallow eutrophic lakes revealed that the species belonging to Pyrrophyta were regularly phosphatase-positive, while Bacillariophyceae were phosphatase negative except Aulacoseira sp 28 . Dinoflagellates were poor competitors for phosphate accumulation compared to diatoms; they have to excrete much more APase than diatom to hydrolyze DOP to satisfy their P demand, even when phosphate is adequate 29 . In nutrient addition experiments, a higher percentage of dinoflagellates were identified with cell-specific APA than diatoms 30 . It can explain why APA >3.0 μm peaked when Pyrrophyta subdominated the phytoplankton.It was consistent with the results in Monterey Bay that dinoflagellates comprised only 14% of all cells counted and accounted for 78% of APase-producing cells examined 16 . Microcystis aeruginosa was confirmed can also synthesize APase 31 . It can explain that as the cell density of Cyanophyta increased in summer and autumn, the APA >3.0 μm was also prompted. The dominating of Cyanophyta during the summer and autumn resulted in the high amount of APA. The synchronous pattern of alkaline phosphatase activity and algal cells amount can also be found in Jialing River 32 . The higher algal cell density in midstream than in estuary can also explain why the APA T was higher in midstream. It could be concluded that phytoplankton communities determined the level of APA >3.0 μm , which determined the significant seasonal and regional variations of APA T .
APA showed significant seasonal and regional variations, with lower value in inlet waters and higher value in the estuarine, and relatively low in winter and high in summer 7 . However, the distribution characteristics of APA in this study were not consistent strictly with the above mentioned. The APA T fluctuated frequently from spring to autumn. Relative stable level of APA T in winter can be seen in Fig. 1. This phenomenon may result from the fluctuant water level of the TGR. For the sake of flood control and hydropower, the water level in the TGR is subjected to the specific management of the TGD and is meant to seasonally fluctuate between 145 and 175 m a.s.l. It has been demonstrated that the turbulence promoted the phytoplanktonic APA and accelerated the biogeochemical cycle of P in Lake Taihu 33 . This was consistent with our results that the high APA was present during the significant water level fluctuated period from spring to autumn. Meanwhile, it has been proved that the APA increased with water temperature 34,35 . The positive relationship between WT and APA in this study (Table 1) supports the conclusion that WT determined the APA through its effects on the phytoplankton seasonally and the direct influences on APase.

Methods
Samples area and sites. Xiaojiang River, a tributary of the TGR, originates from Kaixian, Chongqing Municipality with a length of 180 km and watershed area of 5172.5 km 2 . It flows from north to south; entering into the TGR in Yunyang County. The distance from the estuary to the TGD is 248 km.
Water temperature (WT), pH, dissolved oxygen (DO) and conductivity (Cond.) were measured using a YSI model Professional Plus multiparameter probe (USA); Transparency (SD) was measured with a Secchi disk; and turbidity (Turb.) was measured with a WGZ-B turbidmeter (XinRui, Shanghai). Water level (WL) was recorded by GPS in situ. Surface water samples (0.5 m) were collected with a Van Dorn sampler at seven sampling sites (XJ, HS02, HS01, GY02, GY01, QM02, QM01) (Fig. 8) monthly from October 2013 to September 2014. All samples were run in triplicate. In order to avoid the physiological and biological parameters changed dramatically, the water samples for APA test were filtered immediately after collection and strong oscillation in situ, the filters were put into a portable refrigerator at 0 °C and analyzed within 24 h. All water samples for the other parameters measurement were also stored in a portable refrigerator at 0 °C after collected and tested within 24 h. Concentrations of chlorophyll a (Chl a), total phosphorus (TP), soluble reactive phosphorus (SRP), chemical oxygen demand (COD) were analyzed after samples collected within 24 h. Samples for quantitative phytoplankton analyses were fixed with neutral Lugol's solution, and concentrated after 48 h sedimentation 36 . Measurement of APA. APA was measured using a modified procedure 37,38 . A total of 2 ml water samples were incubated at 37 °C for 4 h in the presence of 1 ml 0.05 M Tris-HCl buffer (pH = 8.5) and 2 ml 0.3 mM p-nitronphenylphosphate (p-NPP) as substrate, subsequently, 0.1 ml 0.1 M NaOH was added into the mixture after 4 h. The release of p-nitrophenol from p-nitronphenylphosphate was determined by absorbance at 410 nm using a spectrophotometer (TU-1810), and APA was calculated in nM·L −1 ·min −1 . APA was determined in unfiltered water (APA T ) and water samples filtered through 0.45 (the picoplankton/dissolved alkaline phosphatase activity, APA <0.45 μm ) and 3. Measurement of SRP, Chl a, TP, COD and phytoplankton quantification. Water samples used for the Chl a measurement were filtered with Whatman GF/C filter, then the residuals on the filter were extracted using 90% acetone solution in the darkroom for 24 h at 4 °C, and Chl a was analyzed spectrophotometrically. The concentrations of SRP were measured after all water samples were filtered through pre-washed filters (Whatman GF/C, glass microfiber filters).The concentrations of SRP, total phosphorus (TP) and chemical oxygen demand (COD) were analyzed according to the standard methods 40 . Phytoplankton was quantified at 400 × magnification with a light microscope (OLYMPUS BX41). The identification of phytoplankton species is according to Hu and Wei 41 .
Statistical analysis. Statistical analysis was carried out using the SPSS 13.0 package. Variance analysis (one-way ANOVA) was used to compare the means of APA in different seasons and sampling sites. Normal distribution of the data was ensured by visual inspection of Q-Q plots, and Levene's test was used to check for homogeneity of variances before ANOVA. Non-parametric correlation (Spearman) analyses were employed for determining relationships among APA <0.45 μm , APA 0.45-3 μm , APA >3.0 μm , APA T and the environmental factors. Detrended correspondence analysis (DCA) of the size-fractionated APA and environmental data was performed using CANOCO version 4.5 to determine whether linear or unimodal ordination methods should be applied 42 . Before the analysis, the abiotic and biological data were transformed by log(x + 1). We calculated an unconstrained ordination through the DCA. Using detrending by segments and Hill's scaling, the length of the longest axis estimated the beta diversity in the data set. Moreover, the value of the data set implies that it is appropriate to use the Redundancy analysis (RDA) method. RDA was performed to get an approximate ordering of the size-fractionated APA's optima for environmental variables. The significance of canonical axes and environmental variables to explain the variance of the size-fractionated APA was tested using Monte Carlo simulations with 499 permutations 43,44 . Data Availability. All data analyzed during this study are included in this published article.