Differences in phytoplankton dynamics and community structure between a wet year and dry year in the Zhoucun Reservoir

Abstract

Rainfall-induced nutrient fluctuations in surface water affect phytoplankton communities and cause algal blooms. To examine this influence in the Zhoucun Reservoir, China, field surveys were performed weekly or biweekly from February 2013 (a wet year) to December 2014 (a dry year). Physical conditions, nutrient levels, and phytoplankton functional groups were investigated during monitoring. Nutrient levels increased in the summer of 2013 (TN: 0.85–2.66 mg/L; TP: 0.037–0.138 mg/L), while nitrogen deficiency occurred in the summer of 2014 (TN: 0.38–0.73 mg/L, TP: 0.044–0.079 mg/L). A cyanobacteria bloom that appeared in the summer of 2013 had the dominant phytoplankton functional groups M and Tc. In the summer of 2014, algal cell density and biomass were much lower than those during the corresponding period in 2013, and the co-dominant phytoplankton groups were F, L₀, Tc, and S2. Non-N-fixing cyanobacteria were not abundant in summer 2013 probably because of the low ratio of NO₃⁻ to NH₄⁺, although nitrogen was also deficient. In contrast to these summer trends, little difference was noted between phytoplankton dynamics and community structure in the two winters and two springs, regardless of nutrient conditions. Redundancy analysis indicated that rainfall-induced nitrogen fluctuations were the primary factor causing phytoplankton dynamics to differ between the wet and dry year. The effects of water temperature, stratification, and light on the phytoplankton community were independent of nutrient fluctuations. This study shows that rainfall fluctuations change summer nutrient conditions and that cyanobacteria blooms are likely to occur in wet years in the Zhoucun Reservoir.

Keywords:

• phytoplankton functional groups
• nutrients
• nitrogen
• cyanobacteria bloom
• rainfall

Introduction

Algal blooms are a global surface water problem and compromise drinking water quality and aquatic environments (Vonlanthen et al. 2012; Horst et al. 2014). Algal blooms require factors such as adequate nutrients, long water retention time, stable hydraulic condition, suitable water temperature, and adequate light intensity to exist (Dokulil and Teubner 2000, Jacoby et al. 2000). The key factors that result in algal blooms vary with phytoplankton community and environmental conditions (Paerl and Huisman 2008). Algal blooms often occur in lakes, reservoirs, and slow-moving rivers when sufficient nutrients are available. Anthropogenic activities have led to increased nutrient discharges, which in turn result in higher nitrogen (N) and phosphorous (P) concentrations in lakes and reservoirs. Numerous studies have reported increases in algal bloom frequency in lakes and reservoirs because of increasing nutrient availability (Duan et al. 2009; Winter et al. 2011; Bridgeman et al. 2013).

Although P was initially regarded as the key factor controlling algal bloom occurrence in surface water, N has been receiving attention recently (Elser et al. 2007). The content, structure, and ratio of nutrients in aquatic systems profoundly influence phytoplankton dynamics and community structure (Zhao and Quigg 2014). Because the response of algae to nutrient fluctuations varies by species (Donald et al. 2013), the effects of nutrients on phytoplankton communities are site specific (Xu et al. 2014). Most nutrient enrichment studies investigating the influence of nutrients on phytoplankton have been conducted in controlled environments, such as microcosms (Wever et al. 2008; Xu et al. 2013; Paerl et al. 2014) and mesocosms (Gardner et al. 2008; Zhang et al. 2014). However, specific ecological conditions in lakes and reservoirs (such as stratification, nutrient release, and hydraulic conditions) cannot be simulated entirely in such experiments (Carpenter 1996; Schindler et al. 2008). Whole-ecosystem studies or field surveys do not have such a drawback (Jansson et al. 2001; Vadeboncoeur et al. 2001; Schindler et al. 2008). Nevertheless, too few in situ studies have been conducted to understand the response of phytoplankton to nutrient fluctuations in different parts of the world.

Along with air, soil, and water pollution, rainfall-induced nutrient fluctuations are increasingly affecting phytoplankton communities. Field surveys have been conducted on the influence of short-term rainfall on algal blooms and the phytoplankton community. For example, Nõges et al. (2011) showed that precipitation increased nutrient levels and algal biomass in Lake Varese and Paidere et al. (2007) reported that floods increased phytoplankton diversity and decreased phytoplankton biomass in two floodplain lakes. In addition, the effects of nutrients on phytoplankton communities have been compared among different water bodies in the same area: Gikuma-Njuru et al. (2013) revealed that the Nyanza Gulf of Lake Victoria was P-deficient and dominated by non-nitrogen-fixing cyanobacteria, whereas Lake Victoria as a whole was N-deficient and occupied by diazotrophic cyanobacteria and diatoms. However, to our knowledge, only rarely have phytoplankton dynamics and communities in lakes or reservoirs been compared between dry and wet years.

In this study, we investigated the physical conditions, nutrient levels, and algae growth in the Zhoucun Reservoir in a wet year (2013) and a dry year (2014). The aims of the present study were to (1) compare the nutrient content, form, and ratios between the two years; (2) compare phytoplankton community dynamics and structure between the two years; and (3) determine the relationship between phytoplankton community structure and environmental factors in each of the two years. We conducted field surveys to address these goals. This study may shed light on the response of phytoplankton to rainfall fluctuation in temperate lakes and reservoirs in northern China. The data obtained in this study are essential for effectively managing eutrophication.

Methods

Study area

The Zhoucun Reservoir is a medium-sized reservoir whose functions are to control floods, irrigate, generate power, and supply drinking water for the city of Zaozhuang, located in eastern China. The reservoir receives run off from a 121 km² hilly catchment, which comprises villages, agricultural land, and forest. The reservoir's total storage capacity is 84.04 million m³ and the flood control capacity is 33.04 million m³. The reservoir has a surface area of 6.5 km² with a mean depth of 13 m and elevation range of 121–128 m. The region is located in the temperate zone with a sub-humid continental monsoon climate. The average annual air temperature is 11–14 °C, with the lowest and highest air temperatures of about −10 °C in January and about 35 °C in July. The mean annual precipitation is 847 mm. Three rivers flow into the reservoir: the Xuwa and Xiashi rivers, located northwest of the reservoir, and the Xijia river, located to the northeast. Of the three, the Xiashi and Xuwa rivers are relatively polluted due to the presence of chicken farms along their banks.

Sampling and pretreatment

Sampling was carried out weekly or biweekly in the reservoir and monthly in the inflow rivers from February 2013 to December 2014 (sampling was not conducted in January 2013, when the reservoir was frozen). The three sampling sites in the reservoir were located in the pelagic zone (34° 56′ 38.49″ N, 117° 41′ 14.03″ E; 34° 56′ 38.86″ N, 117° 41′ 4.52″ E; and 34° 56′ 44.05″ N, 117° 41′ 14.68″ E), while the sampling sites in the inflow rivers were near the entrance of the reservoir (Figure 1). The three sampling sites in the pelagic zone of the reservoir were considered replicates. Water samples were collected using a 2.5-L Van Dorn bottle at 0.5 m depth between 10:00 and 12:00 hrs. Water temperature (T), dissolved oxygen (DO), and water transparency (SD) were measured in situ. Vertical T and DO profiles were measured using Hydrolab™ DS5, while SD was measured using a Secchi disk. Samples for nutrient analyses were stored in acid-washed polyethylene bottles and preserved at 4 °C. Water samples for phytoplankton analysis were preserved using Lugol's solution.

Figure 1. Sampling sites in the Zhoucun Reservoir and the inflow rivers.

Sample analysis

Nutrient analyses such as total nitrogen (TN), dissolved total nitrogen (DTN), ammonium (NH₄⁺), nitrate (NO₃⁻), total phosphorus (TP), and soluble reactive phosphorus (SRP) were completed within 24 h of sample collection in the laboratory using standard water quality analysis methods (Administration 2002). Water was filtered through a cellulose acetate membrane for DTN and SRP analyses. TN, DTN, and NO₃⁻ were all measured in the same manner, although TN and DTN were first treated with a persulfate digestion to convert organic N to NO₃⁻ and all NO₃⁻ was then measured with the sulfanilamide method. NH₄⁺ was measured using the phenol--hypochlorite method. TP and SRP were analyzed with the molybdate-blue method. Rainfall data were collected from the Zaozhuang City online remote sensing database. The euphotic zone (Z_eu) was found to be 2.7 times as deep as the SD zone (Cole 1975). The mixed zone (Z_mix) is the depth where dO₂/dz reaches the maximum (Armengol et al. 1999). Z_eu/Z_mix can be used to evaluate the light availability in the mixed zone (Jensen et al. 1994).

Relative water column stability (RWCS) was calculated to describe the thermal stratification of the Zhoucun Reservoir (Xiao et al. 2011):where D_B is the water density at the bottom, D_s is the water density at the surface, D₄ is that at 4 °C, and D₅ at 5 °C. Calculations of loads of inflow nutrients were made by taking a one-month period as the basic time interval. The nutrient load was calculated as follows:where i = January to December, j = 1 to 3, with the numbers representing the inflow rivers, q_ij is the runoff of river j in month i, c_ij is the average nutrient concentration of river j in month i. Discharge data of the three rivers were supplied by the Bureau of Water Resources of Zaozhuang Municipality.

Phytoplankton identification and counts were performed using an Olympus microscope at 400× magnification according to Hu and Wei (2006). Each sample was counted until approximately 200 cells of the dominant species were recorded, giving a precision of ±20% within a 95% confidence limit. Algal biomass was estimated from biovolumes (Hillebrand et al. 1999), assuming the algal cell density as 1 mg/mm³. The average biovolume of each species was calculated based on at least 10 specimens. Phytoplankton assemblages were divided into functional groups in accordance with the scheme proposed by Reynolds et al. (2002). The functional groups that contributed more than 5% to the total algal biomass were designated as the dominant functional groups (Xiao et al. 2011). The Jaccard index (JI) was applied for estimating the similarity in terms of the phytoplankton species composition between the same period in each of the two years (Jaccard 1908). The Jaccard formula iswhere a is the number of species present in both samples (the same period of the two years), b is the number of species present in sample 1, but absent from sample 2, and c is the number of species present in sample 2, but absent from sample 1.

The relationship between phytoplankton functional groups and environmental variables was analyzed using CANOCO version 4.5. Phytoplankton functional group data were transformed by log(x + 1) before a detrended correspondence analysis (DCA). Redundancy analysis (RDA) was then performed using the biological and environmental data.

Results

Physical conditions

Rainfall patterns differed between 2013 and 2014 in the Zhoucun Reservoir (Figure 2(a)). The long-term (1995–2015) average annual rainfall is 847 mm. The maximum annual rainfall was 1103 mm (in 1996) while the minimum annual rainfall was 698 mm (in 2003). The total rainfall in 2013 was 904 mm, which is above the average; hence we designated 2013 as a wet year. Rainfall in 2013 mainly occurred in the summer with the maximum in July, accounting for 44.2% of the full year. Rainfall in August and September was 162 and 65.5 mm, respectively. The total precipitation in 2014 was 758.5 mm, which was in the middle of the mean annual value and minimum value; therefore, we designated 2014 as a dry year. In 2014, rainfall was relatively evenly distributed from May to September.

Figure 2. Physical conditions in 2013 and 2014 in the pelagic zone of Zhoucun Reservoir. (a) Precipitation, (b) water temperature and RWCS, (c) the mixing zone (Z_mix), euphotic zone (Z_eu), and Z_eu/Z_mix.

The thermal structure of the Zhoucun Reservoir in 2013 was similar to that in 2014 (Figure 2(b)). The stratified period of the Zhoucun Reservoir was from May to October 2013, while the mixing period was from November 2013 to April 2014. Water temperature slightly differed between the two years. The seasonal changes of RWCS and water temperature in the Zhoucun Reservoir were unimodal. The highest monthly average water temperature occurred in July (about 28–29 °C) and the lowest in January (about 0–1 °C).

Light availability in the mixing layer in 2013 was very similar to that in 2014 (Figure 2(c)). Z_eu/Z_mix was higher in summer and lower in winter and spring in the Zhoucun Reservoir.

Nutrient dynamics

Nitrogen levels varied seasonally in the Zhoucun Reservoir (Figure 3(a)). Nitrate was the most abundant nitrogen species in the Zhoucun Reservoir most of the time and, thus, it heavily influenced TN. Ammonium was the second most abundant source of nitrogen, while NO₂⁻ was least abundant. Nitrogen levels were higher during the summer of 2013 (TN: 0.85 2.66 mg/L) than 2014 (TN: 0.38–0.73 mg/L). The proportion of each nitrogen species likewise differed between the two summers. In the summer of 2013, NO₃⁻ accounted for 28.4%–71.3% of TN while NH₄⁺ contributed 14.3%–39.8%. The ratio of NO₃⁻ to NH₄⁺ was 1.34–4.17. In the summer of 2014, NO₃⁻ made up only 5.1%–30% while NH₄⁺ contributed 32.6%–58% of TN. The ratio of NO₃⁻ to NH₄⁺ was only 0.08–0.69.

Figure 3. Seasonal variations of nutrients concentration from 2013 to 2014 in the Zhoucun Reservoir. (a) Nitrogen, (b) TP and SRP, (c) TN:TP.

The concentrations of TP and SRP were high in the summer of 2013 and relatively low in the summer of 2014 (Figure 3(b)). SRP was similar to TP in 2013, but the two varied in 2014.

TN:TP in 2014 was much lower than that in 2013 (Figure 3(c)). This ratio ranged from 9.8 to 198 (on July 9 and April 16, respectively) in 2013, while it was only 5.9 to 92 (on September 13 and February 19, respectively) in 2014. In the summer and autumn of 2014, TN:TP fluctuated but remained low (5.9–16.7).

The inflow water of the Zhoucun Reservoir brought large amounts of nutrients (Table 1). Cao et al (2013) indicated that the release of NH₄⁺ and TP from sediment in the Zhoucun Reservoir were 7.6 and 7.3 t/year, respectively. The inflow nitrogen load was much higher than the internal nitrogen load, while the inflow phosphorus load was lower than the internal phosphorus load. Analysis of inflow nitrogen load revealed that nitrate was the primary nitrogen species, followed by ammonium. The inflow TN load in the wet year (2013) was 2.4 times as much as that in the dry year (2014), which was responsible for the higher nitrogen level in the wet year (2013).

Table 1. Annual ranges of nutrient concentrations, runoff volume, and nutrient loads of inflows to the Zhoucun Reservoir, China.

Year	Index	Concentration	Runoff	Load
		mg/L	10^6 m^3/yr	t/yr
2013	TN	2.44–8.49	78.37	591.4
	NO3^−	1.85–8.34		516.8
	NH4^+	0.04–2.92		34.7
	TP	0.03–2.79		13.7
	SRP	0.01–2.55		8.45
2014	TN	4.36–13.81	48.13	244.7
	NO3^−	0.22–8.15		175.7
	NH4^+	0.02–12.2		44.8
	TP	0.01–4.17		12.6
	SRP	0.01–3.77		9.2

Phytoplankton dynamics and community structure

A total of 124 species of phytoplankton belonging to 7 phyla and 21 phytoplankton functional groups were detected during monitoring. There were 13 dominant phytoplankton functional groups in the Zhoucun Reservoir (Table 2). Algal cell density and algal biomass were correlated in 2013 and 2014 (Figure 4). In 2013, algal cell density reached its maximum (7.384 × 10⁷ cells/L) on September 8 and the minimum (1.188 × 10⁶ cells/L) on April 16. In 2014, the maximum (4.875 × 10⁷ cells/L) was seen on July 13 and the minimum (420 × 10⁴ cells/L) on April 24. Algal cell density was about 2000 × 10⁴ cells/L in the summer of 2014, which was much lower than that in the summer of 2013.

Table 2. Common phytoplankton functional groups and representative species in the Zhoucun Reservoir.

Group code	Representative species	Taxonomic group
C	Cyclotella spp.; Asterionella sp.;	Bacillariophyta
D	Synedra spp.;	Bacillariophyta
F	Nephrocytium spp.; Oocystis spp.; Chlorococcum sp.;	Chlorophyta
G	Eudorina sp.; Pandorina sp.; Asterococcus sp.;	Chlorophyta
J	Tetraedron sp.; Scenedesmus spp.;	Chlorophyta
LO	Chroococcus spp.; Merismopedia spp.;	Cyanophyta
M	Microcystis spp.;	Cyanophyta
MP	Cocconeis sp.; Navicula spp.;	Bacillariophyta
P	Melosira spp.; Closterium sp.	Chlorophyta
S2	Spirulina spp.;	Cyanophyta
Tc	Phormidium spp.; Lyngbya spp.; Oscillatoria sp.;	Cyanophyta
X1	Chlorella spp.;	Chlorophyta
Y	Cryptomonas spp.;	Cryptophyta

Figure 4. Seasonal variations in algal cell density and biomass from 2013 to 2014 in the Zhoucun Reservoir.

The 13 dominant phytoplankton functional groups underwent seasonal succession in the Zhoucun Reservoir (Figure 5). In 2013, groups M and Tc belonging to Cyanophyta accounted for more than 50% of the algal biomass in summer. Groups Tc, F, X1, and D prevailed in autumn. In contrast, the co-dominant groups were F, L₀, Tc, and S2 in the summer of 2014 and C, F, and X1 in the autumn of 2014. Groups C, D, and F dominated during both winters and early spring, while groups F and J were dominant in the two late spring seasons. According to the Jaccard index, the similarities of the phytoplankton community were highest in spring, with the maximum of 66.7% for May 2013. In the summer of 2013, the low similarity with the range of 18.2% (July) to 28.6% (September).

Figure 5. Seasonal succession of phytoplankton functional groups in the Zhoucun Reservoir. (a) The wet year (2013) and (b) the dry year (2014).

Redundancy analysis

The maximum gradient length for the 13 dominant phytoplankton functional groups was 2.25 in 2013 and 1.64 in 2014. Therefore, RDA was carried out to test the relationship between the phytoplankton community and environmental variables (Figure 6; Lepš and Šmilauer 2003). In the 2013 RDA biplot, the first two axes explained 63.4% of the phytoplankton–environment correlations (axis 1: 38.8%; axis 2: 24.6%). RWCS, Z_eu/Z_mix, and T had marked effects on phytoplankton community, while TN and NO₃⁻ had little influence. Groups L₀, Tc, and D showed a strong positive correlation with NH₄⁺. Groups C and P negatively correlated with RWCS, Z_eu/Z_mix, and T. In the 2014 RDA biplot, the first two axes accounted for 74.2% of the variance in the relationship between phytoplankton and the environment (axis 1: 46.8%; axis 2: 27.6%). TN, TP, NO₃⁻, T, RWCS, and Z_eu/Z_mix were all important factors for explaining phytoplankton community succession. Rainfall, RWCS, T, and Z_eu/Z_mix were positively related to groups L₀, X1, and S2. TP showed a positive relationship with groups F and J. TN, NO₃⁻, and SD were positively associated with group P.

Figure 6. RDA biplot of dominant phytoplankton functional groups and environmental factors. (a) The wet year and (b) the dry year. TN = total nitrogen; NO₃-N = nitrate; TP = total phosphorus; NH₄-N = ammonia nitrogen; T = water temperature at 0.5 m; RAIN = rainfall; SD = water transparency; N:P = TN/TP.

Discussion

Nitrate differences between the wet year (2013) and the dry year (2014) were more significant than interannual differences of other environmental variables in the Zhoucun Reservoir in those two years. There are two ways by which nutrients can enter the reservoir: exogenous input and endogenous release. Exogenous input mainly comes from rainfall-induced runoff, whose primary nitrogen source is NO₃⁻ (Rabalais 2002; Lehman et al. 2015), whereas NH₄⁺ is the primary form of nitrogen if released from the sediment under anaerobic conditions (Reddy et al. 1996). Given the similar thermal stratification and sediment properties of the Zhoucun Reservoir in 2013 and 2014, the amount of nutrients discharged from the sediment probably varied little between the two years. Therefore, rainfall fluctuation likely altered the concentration of NO₃⁻ in the Zhoucun Reservoir, affecting nitrogen species and the mass ratio of N to P. This in turn may have influenced phytoplankton dynamics and community structure (Horst et al. 2014, Zhao and Quigg 2014).

Phytoplankton community structure is thought to be determined by its limiting resource (Danger et al. 2008; Becker et al. 2010). It has been reported that the eutrophication threshold of N for lakes and reservoirs is about 1 mg/L (Lin et al. 2008). The concentration of TN ranged from 0.85 to 2.74 mg/L in the Zhoucun Reservoir in 2013, indicating that N was sufficient for phytoplankton growth. In contrast, the range of TN was 0.38–0.73 mg/L in the summer of 2014, implying N deficiency during this period. Consequently, the N concentration difference, which may be influenced by rainfall, was one of the key factors responsible for the difference in phytoplankton growth between 2013 and 2014.

Previous studies have indicated that N greatly affects phytoplankton community structure in lakes and reservoirs in the temperate zone (Huszar and Caraco 1998; Palsson and Graneli 2004), subtropics (Lv et al. 2011), and tropics (Leon Lopez et al. 2012). The amount of energy required by algae to absorb N varies depending on the algal species and N form (Donald et al. 2013); thus, phytoplankton communities respond differently to the addition of different types of N. Therefore, the different nutrient forms also may be responsible for the different phytoplankton community between the two years.

Nitrogen deficiency occurred in the Zhoucun Reservoir in the summer of 2014 and N-fixing cyanobacteria did not dominate the phytoplankton community in terms of biomass. Some studies have indicated that N-fixing cyanobacteria become dominant in summer with a nitrogen shortage (Mugidde et al. 2003; Marcarelli and Wurtsbaugh 2009; Paterson et al. 2011). However, N-fixing cyanobacteria did not dominate the phytoplankton community in several additional studies under analogous circumstances (Havens et al. 2003; Scott and McCarthy 2010; Gikuma-Njuru et al. 2013). The common N-fixing cyanobacteria in temperate lakes are Oscillatoria spp., Nostoc spp., and Anabaena spp. These species are capable of converting dinitrogen (N₂) into NH₄⁺ (Markou et al. 2014) and prefer eutrophic lakes with low nitrogen content (Padisák et al. 2009). Non-N-fixing cyanobacteria are better able to assimilate NH₄⁺ than N-fixing cyanobacteria (Blomqvist et al. 1994). During the stratification of 2014, NH₄⁺ was released from the sediment and was maintained at 0.19–0.37 mg/L in the epilimnion of the Zhoucun Reservoir, when exogenous N inputs were much lower than in the wet year (2013). Hence, the relatively adequate levels of NH₄⁺ may be at the most influence of non-N-fixing cyanobacteria. Previous studies have suggested that N-fixing cyanobacteria can outcompete non-N-fixing cyanobacteria only under sufficient P levels (Piehler et al. 2009; Vahtera et al. 2010). The concentration of SRP was 2–8 μg/L in the summer of 2014 in the Zhoucun Reservoir and may be inadequate for an N-fixing cyanobacteria bloom (Vrede et al. 2009).

Nutrient fluctuations which may be induced by rainfall had a large influence on the dominance of Microcystis blooms in the Zhoucun Reservoir. Microcystis spp. belong to group M and usually appear in eutrophic and standing water (Padisák et al. 2009). Compared to July 2014, the N and P introduced by rainfall in July 2013 provided suitable conditions for Microcystis blooms. Additional nitrogen has previously been found to facilitate Microcystis blooms (Downing et al. 2005; Moisander et al. 2009). A similar phenomenon was reported in a lake in eastern USA (Davis et al. 2010). However, the Microcystis bloom in our study disappeared in August 2013 despite high nutrient levels (dissolved inorganic nitrogen (DIN) = 2.05 mg/L, SRP = 0.03 mg/L). Previous studies have suggested that the content and structure of N significantly influence Microcystis blooms. Liu et al. (2011) found that the biomass of Microcystis spp. was positively related to the ratio of NH₄⁺ to NO₃⁻. Jacoby et al. (2000) suggested that Microcystis spp. dominance was positively associated with a low ratio of NO₃⁻ to NH₄⁺ and a low ratio of N to P. With increasing NO₃⁻ introduced by runoff, the ratio of NH₄⁺ to NO₃⁻ declined from 0.75 to 0.24 in the Zhoucun Reservoir in August 2013. Afterward, group Tc, which is well adapted to eutrophic standing waters with low light intensity (Padisák et al. 2009), replaced group M. The addition of N and P, as occurred in the Zhoucun Reservoir in August 2013 (Figure 3), is known to promote the growth of group Tc (E-wen 2010). However, the concentration of NO₃⁻ in August 2013 (about 2.5 mg/L) was above the reported nitrate–N saturation concentration for Lyngbya wollei, which belongs to group Tc (Cowell and Dawes 2004). This indicates that group Tc could have obtained a high growth rate in August 2013.

The ratio of N to P (TN:TP) in the Zhoucun Reservoir varied with rainfall, leading to the transformation from nitrogen deficiency to phosphorus deficiency. The classic ratio (7:1 by mass) proposed by Redfield (Redfield 1958) cannot be applied to all ecological conditions. Sterner et al. (2008) suggested that a power function should be used instead of the constant ratio. Klausmeier et al. (2004) suggested that the optimal ratio varied with species and ranged from 8.2 to 45. Although the optimal ratio is still controversial, Klausmeier et al. (2004) found that the species whose structural N:P ratio was less than 30 accounted for 89% of the 29 phytoplankton species. Therefore, it can be concluded that the Zhoucun Reservoir had a tendency for P limitation during the monitoring period except during the summer of 2014 (TN:TP = 5.9–16.7). The N decrease induced by reduced rainfall was primarily responsible for the decline of TN:TP and the transformation from P limitation to N limitation in the summer of 2014.

Groups C and D dominated the Zhoucun Reservoir in the winter and spring of both years. The difference of phytoplankton community structure between the two winters and between the two springs was not significant despite different nutrient levels. Deng et al. (2014) concluded that nutrient concentrations had a greater influence on phytoplankton community succession at higher temperatures. For this reason, the low water temperature in winter and spring may have mitigated the effects of nutrients on phytoplankton in the Zhoucun Reservoir. Group C is sensitive to stratification and silicate and is usually abundant in spring, while group D is sensitive to nutrient depletion and usually inhabits eutrophic, shallow, and turbid waters (Reynolds et al. 2002). The abundance of groups C and D was negatively correlated with temperature (Figure 6), which was consistent with the results of previous studies (Zhang and Prepas 1996; Calijuri et al. 2002; Fernandez et al. 2014). Group C was positively related to NO₃⁻, which is consistent with Donald et al. (2013).

T, RWCS, and Z_mix/Z_eu were important factors affecting phytoplankton community succession in the wet year and in the dry year in the Zhoucun Reservoir. Reynolds (2006) suggested that temperature, stratification, and light were crucial for algal growth, which was also true in this study. The optimal temperature for algal growth varies with the species (Winder and Schindler 2004; Reynolds 2006; Trolle et al. 2011). In both 2013 and 2014, seasonal phytoplankton succession fits the plankton ecology group (PEG) model, which was based on seasonal variations in temperature (Sommer et al. 1986). RWCS represents thermal structure, which strongly influenced phytoplankton community structure. Some groups (such as C, D, and F) that prefer the mixed conditions accounted for a relatively high proportion of the phytoplankton community in winter and spring in the Zhoucun Reservoir. In contrast, other groups (such as M and L₀) prevailed during stratification (Reynolds et al. 2002; Padisák et al. 2009). Group C, which is adapted to low light conditions, was negatively related to Z_mix/Z_eu. Groups S2 and L₀, which can tolerate high light intensity, were positively associated with Z_mix/Z_eu.

Fluctuations in annual rainfall between the wet year and the dry year have inconsistent effects on phytoplankton community in lakes and reservoirs due to the different hydrological condition and pollution near the region. The phytoplankton community of Liuxihe Reservoir, which is subject to a subtropical monsoon climate, was significantly affected by the changing physical and chemical environment in an exceptionally wet year (Xiao et al. 2011), while the phytoplankton succession in Tenkiller Reservoir (Cooke et al. 2011) and Río Seco Lake (Garcia-Jurado et al. 2011) was influenced by the altering nutrient condition. For Zhoucun Reservoir, the rainfall in 2013 (wet year) was not strong enough to obviously change the physical conditions as Liuxihe Reservoir (Figure 2). This study can provide guidance to deep lakes and reservoirs which belong to a sub-humid continental monsoon climate.

In the Zhoucun Reservoir, nitrate differences between the wet year (2013) and the dry year (2014) were greater than the differences in other environmental factors in these two years. Nitrate variations, which may be induced by rainfall, affected the content and structure of N and the ratio of N to P, thus influencing phytoplankton growth. Phytoplankton dynamics and community structure differed between the two summers and between the two autumns, but were not significantly different between the two winters and between the two springs. This study provides basic data that could be used to create and justify nutrient management strategies to prevent future eutrophication.

Acknowledgments

This study received the support of the National Science and Technology Pillar Program (No. 2012BAC04B02). We acknowledge the anonymous reviewers for their useful comments on the early version of the manuscript. We also wish to thank the agencies involved for sampling assistance.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

The National Science and Technology Pillar Program [grant number 2012BAC04B02]