Anthropogenic Inputs Affect Phytoplankton Communities in a Subtropical Estuary

: In the past few decades, with the rapid economic development of China and Vietnam, the marine ecological environment of Beibu Gulf is facing increasing pressure. To understand the impact of anthropogenic inputs on marine ecology, high-performance liquid chromatography (HPLC) was used to study phytoplankton in this paper. We examined the inﬂuence of anthropogenic inputs on phytoplankton biomass and community structure in a subtropical estuary. Anthropogenic inputs had signiﬁcantly increased the nutrient concentration in the estuary between 2010 and 2015. We observed that phosphorus limitation has been greatly relieved in 2015. However, the biomass of dominant phytoplankton was lower in 2015 than in 2010, possibly due to the expansion of oyster farming in the estuary. The coverage of oyster rafts was estimated to be 26.3 km 2 . The presence of dense oysters may signiﬁcantly reduce the phytoplankton biomass. The proportion of Diatoms decreased while some nano- and pico-phytoplankton (like Cryptophytes and Prasinophytes) increased, which indicated that oysters changed not only the biomass but also the size of phytoplankton communities. This study improved our understanding of anthropogenic inputs on phytoplankton communities in subtropical estuary. J.F.; formal analysis, X.L.; X.L., W.L., and K.P.; visualization, X.L.; W.L. and T.L.; project administra-tion,


Introduction
Phytoplankton is the main primary producer in marine ecosystems, providing an abundant food source for fish larvae and zooplankton and playing an important role in the top-down effects on secondary consumers [1]. There is a close relationship between the output of fishery resources and phytoplankton. Phytoplankton individuals are small in size and large in number. They are sensitive to environmental changes that some groups can be used as indicators of specific pollutants or environmental changes [2,3]. Phytoplankton plays an important role in studying the impact of climate changes and human activities on marine ecosystem.
Due to the rapid development of China and Vietnam, the Beibu Gulf is a typical bay heavily affected by anthropogenic inputs. The most common example is the increased nutrient levels in bays, which often lead to eutrophication. This usually results in a significant increase in the phytoplankton biomass and even red tides [4]. Estuaries are also ideal for shellfish farming which strongly affects phytoplankton communities. Many scholars have recently studied the top-down regulation effect of shellfish farming on phytoplankton [5]. Feeding and digestion selectivity of shellfish on phytoplankton also attracted widespread attention [6][7][8][9][10]. Land reclamation is common in estuaries and bays. Such reclamation can increase suspended solids in the water reducing light transmittance, changing the hydrodynamic conditions, and increasing the spatial heterogeneity of plankton communities [11,12].
Studies have confirmed that shellfish feed on different phytoplankton not indiscriminately but selectively [13,14]. Phytoplankton with small particle size is difficult to be retained by the cilia of shellfish. Particle size is an important factor affecting selective ingestion [15]. Intensive shellfish culture can significantly decrease the biomass of phytoplankton in aquaculture waters, a phenomenon that is widespread around the world [16,17]. Studies have shown that species of Dinoflagellates are more easily eaten by shellfish than that of Diatoms [6,9,14]. However, a study in Japan showed that Heterocapsa Circularisquama (one species of Dinoflagellates) could lead to algal blooms by significantly reducing the filtration rate of shellfish [18]. In general, the mechanism underlying the selection of shellfish for specific groups of phytoplankton is unclear.
Many researchers have reported the effects of anthropogenic inputs on phytoplankton biomass [19][20][21], but few have evaluated the overall effects of multiple human activities. Our understanding of the overlapping effects of multiple human activities on the ecological environment is still incomplete [22,23]. It is unclear whether regulation of nutrients or shellfish on phytoplankton is greater in estuaries.
Qinzhou Bay is located along the coast of Beibu Gulf, which is one of the most important subtropical bays in the south of China. From 2008 to 2015, the growth in the population, economy, and mariculture here were dramatic. Eutrophication, shellfish farming, and land reclamation can all be observed here. In this paper, we tried to investigate how anthropogenic inputs affect a subtropical estuary and identify any correlations between these inputs with phytoplankton communities. Our finding provides a scientific basis for the sustainable development of a subtropical estuary.

Research Periods and Stations
Comprehensive surveys were conducted during six voyages in Qinzhou Bay Thirteen stations in the bay were examined, including five stations in the inner bay (S1-S5) and eight stations in the outer bay (S6-S13). We also examined two sections of the Qinjiang River (its eastern and western branches, respectively) and one section of the Maoling River. Monthly monitoring was performed at the same time ( Figure 1).

Site Sampling
At the 13 marine stations, we measured the temperature, salinity, and dissolved oxygen (DO) in situ. We also collected surface seawater samples to analyze the dissolved inorganic phosphate (DIP), dissolved silicate (DSi), dissolved inorganic nitrogen (DIN: including nitrate, nitrite, and ammonium), chlorophyll-a (Chl-a), and suspended solids (SS) in the laboratory. The sampling, storage, transportation, and analysis of these samples were performed in strict accordance with the Specification for Marine Monitoring of China [24].
In the 3 river sections, we monitored the flow rate, ammonium, nitrite, nitrate, particulate inorganic nitrogen, particulate inorganic phosphorus, and dissolved inorganic phosphate monthly. The total amounts of nitrogen and phosphorus input into Qinzhou Bay were calculated. The entire process was carried out according to The Surface Water Environmental Quality Standard of China [25].

Site Sampling
At the 13 marine stations, we measured the temperature, salinity, and dissolved oxygen (DO) in situ. We also collected surface seawater samples to analyze the dissolved inorganic phosphate (DIP), dissolved silicate (DSi), dissolved inorganic nitrogen (DIN: including nitrate, nitrite, and ammonium), chlorophyll-a (Chl-a), and suspended solids (SS) in the laboratory. The sampling, storage, transportation, and analysis of these samples were performed in strict accordance with the Specification for Marine Monitoring of China [24].
In the 3 river sections, we monitored the flow rate, ammonium, nitrite, nitrate, particulate inorganic nitrogen, particulate inorganic phosphorus, and dissolved inorganic phosphate monthly. The total amounts of nitrogen and phosphorus input into Qinzhou Bay were calculated. The entire process was carried out according to The Surface Water Environmental Quality Standard of China [25].

Photosynthetic Pigments and CHEMTAX
Surface seawater samples (2L) were collected on-site as photosynthetic pigment samples. The water was rapidly transported to the laboratory under light-protected chilled conditions for membrane filtration. After filtration, each membrane was folded in half and placed in liquid nitrogen for storage. During analysis, we defrosted the filter membrane on the filter paper to absorb excess water. N, N-Dimethylformamide (2 mL) was used as the extraction agent. Extraction was carried out in the dark at −20 °C for 2 h in order to fully extract the pigment. The extraction solution was centrifuged after full mixing (5 min, −4 °C), and the supernatant was collected. The entire process was performed at a low light intensity and low temperature to reduce the degradation of the photosynthetic pigment.

Photosynthetic Pigments and CHEMTAX
Surface seawater samples (2L) were collected on-site as photosynthetic pigment samples. The water was rapidly transported to the laboratory under light-protected chilled conditions for membrane filtration. After filtration, each membrane was folded in half and placed in liquid nitrogen for storage. During analysis, we defrosted the filter membrane on the filter paper to absorb excess water. N, N-Dimethylformamide (2 mL) was used as the extraction agent. Extraction was carried out in the dark at −20 • C for 2 h in order to fully extract the pigment. The extraction solution was centrifuged after full mixing (5 min, −4 • C), and the supernatant was collected. The entire process was performed at a low light intensity and low temperature to reduce the degradation of the photosynthetic pigment.
Thirteen characteristic photosynthetic pigments were separated and their contents in all samples determined by High Performance Liquid Chromatography (HPLC). The chemical taxonomy program, CHEMTAX, was applied under MATLAB platform to acquire the relative contributions of taxa to TChl a. The contributions of various phytoplankton groups to Chl a were determined from the data for the 13 characteristic photosynthetic pigments with a CHEMTAX factor analysis ( Table 1). All phytoplankton groups were represented as Chl a biomass (µg·m −3 ). The phytoplankton were divided into eight major groups: Diatoms (Dia), Chlorophytes (Chl), Dinoflagellates (Din), Haptophytes (Hap), Prasinophytes (Pra), Cryptophytes (Cry), Synechococcus (Syn), and Prochlorococcus (Pro). The ratios of initial inputting pigment to Chl a followed the processes used in previous studies [26].

Data Processing and Analysis
We calculated the total input of nitrogen and phosphorus into the rivers based on the monthly flows and nutrient concentration as follows: where TN is the total N input from the river section throughout the whole year; CiN is the average N concentration in the month i (i ranges from 1 to 12); and Fi is the flow in the river section in the month i. The same variables are presented in the formula for TP.
Clear remote sensing images from September 2010 and September 2015 were used to analyze the shellfish mariculture area. GIS software was used to map the distribution of shellfish rafts and reclamation activity, including calculating the area. A field survey was adopted for shellfish rafts which were under the water in the inner bay. The survey was conducted at the lowest water level of the spring tide, when the positions and boundaries of the aquacultural rafts were recorded.
SigmaPlot 10.0 was used to create all histograms and graphs. The photosynthetic pigment biomass data were transformed via log (x + 1). The similarity level between different communities was determined with resemblance analysis and cluster analysis in Primer 6.0. If the level of similarity between two communities was greater than 93%, they were classified as similar. Surfer 11.0 was used to draw the station maps and horizontal distribution maps. SPSS 16.0 was used for independent sample t-test of each environmental parameter between two seasons. The CANOCO software package was used for a speciesenvironment redundancy analysis (RDA). If the result of Monte Carlo test is significant (p < 0.050), this sorting result was deemed to be reliable and the sorting analysis result was accepted [27]. Table 2 lists the annual fluxes of nitrogen and phosphorus into Qinzhou Bay. In 2015, there were significant increases in the nutrient input from both Qinjiang River and Maoling River. The inputs of nitrogen and phosphorus were higher in 2015 than in 2010. Shellfish farming activity is common in Qinzhou Bay. The main mariculture species was a single oyster (Crassostrea hongkongensis). In 2010, the farming rafts were mainly distributed in the inner bay and the neck area of Qinzhou Bay, in an area of about 40.5 km 2 . In 2015, the oyster-raft-containing area was about 66.8 km 2 . Compared with 2010, the area of oyster rafts had decreased about 1.2 km 2 . However, many new oyster rafts were installed in the inner and outer bay ( Figure 2) at this time. The area of new oyster rafts was about 26.3 km 2 . In general, the scale of oyster mariculture in 2015 significantly increased. From 2010 to 2015, the proportion of oyster rafts relative to the entire Qinzhou Bay increased from 11.9% to 21.0%.

Influence of Human Activity in Qinzhou Bay
Land reclamation was widespread in Qinzhou Bay between 2010 and 2015, and an area of about 21.2 km 2 was reclaimed. Reclamation activities thus reduced the area of Qinzhou Bay from about 339.7 km 2 to 317.4 km 2 . Moreover, the eastern area of the outer bay was divided into two parts by the reclamation. The western area of reclamation in the outer bay separated the sea into three different parts ( Figure 2).

Interannual Variation of Environmental Parameters
The water temperature, pH, and DIP in 2015 were significantly higher than that in 2010 (p ≤ 0.01). N/P and Chl-a in 2015 were significantly lower than that in 2010. The differences of other parameters between two years were not significant (p ≥ 0.10, Table 3).

Interannual Variation of Environmental Parameters
The water temperature, pH, and DIP in 2015 were significantly higher than that in 2010 (p ≤ 0.01). N/P and Chl-a in 2015 were significantly lower than that in 2010. The differences of other parameters between two years were not significant (p ≥ 0.10, Table 3).

The Relationship between Phytoplankton and Environmental Factors
A redundancy analysis (RDA) and sorting were used to analyze the data on phytoplankton biomasses and environmental factors in 2010 and 2015 (Figure 3). Water temperature (correlation weight (CW) = 0.5533), Chl-a (CW = 0.4963), DO (CW = 0.2957), and DIP (CW = 0.2945) were the major factors indicating the distribution of the phytoplankton groups. Dia and Pra, Dia, Din, and Pro showed good positive correlation with Chl-a, N/P and salinity, and negative correlation with DIP, DIN, SS, and Dsi. These correlations were opposite for Cry. The Syn biomass correlated well with water temperature. The phytoplankton communities also showed an obvious succession. From 2010 to 2015, the phytoplankton communities gradually evolved from the first quadrant to the third quadrant in Figure 3.
Water 2022, 14, x FOR PEER REVIEW 9 of 16

The Relationship between Phytoplankton and Environmental Factors
A redundancy analysis (RDA) and sorting were used to analyze the data on phytoplankton biomasses and environmental factors in 2010 and 2015 (Figure 3). Water temperature (correlation weight (CW) = 0.5533), Chl-a (CW = 0.4963), DO (CW = 0.2957), and DIP (CW = 0.2945) were the major factors indicating the distribution of the phytoplankton groups. Dia and Pra, Dia, Din, and Pro showed good positive correlation with Chl-a, N/P and salinity, and negative correlation with DIP, DIN, SS, and Dsi. These correlations were opposite for Cry. The Syn biomass correlated well with water temperature. The phytoplankton communities also showed an obvious succession. From 2010 to 2015, the phytoplankton communities gradually evolved from the first quadrant to the third quadrant in Figure 3.

Interannual Variation of Phytoplankton Biomass
In the inner bay, the average phytoplankton biomass (represented as the concentration of Chl a) at each station in 2010 ranged from 2.4 × 10 3 to 3.6 × 10 3 g•m −3 in 2010 and from 0.7 × 10 3 to 2.0 × 10 3 g•m −3 in 2015. Thus, the biomass was clearly higher in 2010 than in 2015. In the outer bay, the average biomass of phytoplankton at each station ranged from 2.4 × 10 3 to 3.6 × 10 3 g•m −3 in 2010 and from 0.7 × 10 3 to 2.0 × 10 3 g•m −3 in 2015.
Among the eight main phytoplankton groups, the biomasses of Dia, Chl, Hap, Din, Pra, and Pro were significantly lower in 2015 than in 2010, but the biomass of Syn basically remained the same. The biomass of Cry was significantly higher in 2015 than in 2010 (Table 2, Figure 4). Therefore, the biomass of Dia was significantly lower in the inner bay than in the outer bay, whereas the biomass of Pra was significantly higher in the inner bay than in the outer bay.

Interannual Variation of Phytoplankton Biomass
In the inner bay, the average phytoplankton biomass (represented as the concentration of Chl a) at each station in 2010 ranged from 2.4 × 10 3 to 3.6 × 10 3 g·m −3 in 2010 and from 0.7 × 10 3 to 2.0 × 10 3 g·m −3 in 2015. Thus, the biomass was clearly higher in 2010 than in 2015. In the outer bay, the average biomass of phytoplankton at each station ranged from 2.4 × 10 3 to 3.6 × 10 3 g·m −3 in 2010 and from 0.7 × 10 3 to 2.0 × 10 3 g·m −3 in 2015.
Among the eight main phytoplankton groups, the biomasses of Dia, Chl, Hap, Din, Pra, and Pro were significantly lower in 2015 than in 2010, but the biomass of Syn basically remained the same. The biomass of Cry was significantly higher in 2015 than in 2010 (Table 2, Figure 4). Therefore, the biomass of Dia was significantly lower in the inner bay than in the outer bay, whereas the biomass of Pra was significantly higher in the inner bay than in the outer bay.  Dia and Pra were the two most important groups of phytoplankton in Qinzhou Bay. Their proportions ranged from 64.5% to 93.1%. The proportions of Dia in the inner bay varied from 33.8% to 62.2% and that of Pra varied from 7.9% to 48.2%. In the outer bay, the proportion of Dia varied from 44.9% to 84.8% and that of Pra varied from 3.6% to  Dia and Pra were the two most important groups of phytoplankton in Qinzhou Bay. Their proportions ranged from 64.5% to 93.1%. The proportions of Dia in the inner bay varied from 33.8% to 62.2% and that of Pra varied from 7.9% to 48.2%. In the outer bay, the proportion of Dia varied from 44.9% to 84.8% and that of Pra varied from 3.6% to 29.4%. The proportion of Dia was significantly lower in the inner bay than in the outer bay, whereas that of Pra showed the opposite trend (Figures 5 and 6). the proportion of Dia varied from 44.9% to 84.8% and that of Pra varied from 3.6% to 29.4%. The proportion of Dia was significantly lower in the inner bay than in the outer bay, whereas that of Pra showed the opposite trend (Figures 5 and 6).
There was a large difference in the phytoplankton group structure between 2010 and 2015. During this period, the proportion of Diatoms decreased from 63.9% to 50.0% and that of Syn remained basically stable (23.0% and 24.3% in 2010 and 2015, respectively). The proportion of Syn increased from 4.8% and to 10.0% and that of Cry increased from 1.0% to 10.4%. There was also a significant difference in the phytoplankton group structure between the inner and outer bay. The proportion of Dia was significantly higher in the outer bay than in the inner bay in 2010, and the same phenomenon was seen in 2015 ( Figures 5 and 6). The proportion of Syn and Cry was high in spring and autumn but low in summer, whereas those of Dia and Syn showed the opposite trend ( Figure 5). The proportion of Dia and Cry increased gradually from the inner bay to the outer bay, whereas those of Pra and Syn showed the opposite trend ( Figure 6).

Spatial and Temporal Differences in Phytoplankton Community Structure
In 2010, most phytoplankton communities (including eight stations) could be grouped into similar groups: Group A, Group B, and Group C (with similarities of 95.4%, 96.7%, and 94.5% respectively). The communities at the remaining five stations were significantly different from those at the other stations (similarity <93.0%) (Figure 7, 2010). In 2015, most communities could be sorted into groups that share high-level similarity: Group D and Group E (with similarities of 95.6% and 94.4%, respectively). From 2010 to 2015, land reclamation separated S7 from S9 and S8 from S11 (Figure 7, 2015). Sites S7 and S9 could not be clustered into a similar group; the similarity between S8 and S11 also decreased by 4.2% as well in 2015. There was a large difference in the phytoplankton group structure between 2010 and 2015. During this period, the proportion of Diatoms decreased from 63.9% to 50.0% and that of Syn remained basically stable (23.0% and 24.3% in 2010 and 2015, respectively). The proportion of Syn increased from 4.8% and to 10.0% and that of Cry increased from 1.0% to 10.4%. There was also a significant difference in the phytoplankton group structure between the inner and outer bay. The proportion of Dia was significantly higher in the outer bay than in the inner bay in 2010, and the same phenomenon was seen in 2015 ( Figures 5  and 6). The proportion of Syn and Cry was high in spring and autumn but low in summer, whereas those of Dia and Syn showed the opposite trend ( Figure 5). The proportion of Dia and Cry increased gradually from the inner bay to the outer bay, whereas those of Pra and Syn showed the opposite trend ( Figure 6).

Spatial and Temporal Differences in Phytoplankton Community Structure
In 2010, most phytoplankton communities (including eight stations) could be grouped into similar groups: Group A, Group B, and Group C (with similarities of 95.4%, 96.7%, and 94.5% respectively). The communities at the remaining five stations were significantly different from those at the other stations (similarity <93.0%) (Figure 7, 2010). In 2015, most communities could be sorted into groups that share high-level similarity: Group D and Group E (with similarities of 95.6% and 94.4%, respectively). From 2010 to 2015, land reclamation separated S7 from S9 and S8 from S11 (Figure 7, 2015). Sites S7 and S9 could not be clustered into a similar group; the similarity between S8 and S11 also decreased by 4.2% as well in 2015.

Spatial and Temporal Differences in Phytoplankton Community Structure
In 2010, most phytoplankton communities (including eight stations) could be grouped into similar groups: Group A, Group B, and Group C (with similarities of 95.4%, 96.7%, and 94.5% respectively). The communities at the remaining five stations were significantly different from those at the other stations (similarity <93.0%) (Figure 7, 2010). In 2015, most communities could be sorted into groups that share high-level similarity: Group D and Group E (with similarities of 95.6% and 94.4%, respectively). From 2010 to 2015, land reclamation separated S7 from S9 and S8 from S11 (Figure 7, 2015). Sites S7 and S9 could not be clustered into a similar group; the similarity between S8 and S11 also decreased by 4.2% as well in 2015. Group B: S7 and S9 communities; Group C: S6, S8, S10, S11, and S13 communities; Group D: S1, S2, S4, and S5 communities; Group E: S8-S11 communities). S7 and S9 communities; Group C: S6, S8, S10, S11, and S13 communities; Group D: S1, S2, S4, and S5 communities; Group E: S8-S11 communities).

Effects of Multiple Human Activities on Marine Environment
The marine ecosystem is significantly affected by many human activities. In this study, the increase in the nitrogen and phosphorus fluxes were 18.6% and 18.4%, respectively ( Table 2). The scales of oyster farming and land reclamation were higher in 2015 than in 2010. The area of oyster farming doubled and the area of reclaimed land increased by 21.2 km 2 ( Figure 2). These changes indicate that 2010-2015 was a period of rapid development around the estuary and that human activities were intense.
Consequently, the marine environment changed significantly during this period. Rivers are the main sources of nutrients taken in by estuaries and bays [28]. The concentration of DIP increased significantly in 2015, with the average concentration increasing from 8.8 × 10 −3 to 18.1 × 10 −3 mg·L −1 . However, there was no significant change in the DIN concentration in this period. The different changes in the nitrogen and phosphorus led to a change in the nutrient structure. The N/P ratio in the bay decreased significantly (Table 3). Reclamation can increase suspended solids [29], whereas the filter-feeding activities of oysters can significantly reduce the suspended solids in water [30]. Reclamation was finished around 2014. In 2015, the reduced reclamation and the expansion of oyster mariculture led to a significant reduction in suspended solids compared with the level in 2010, and the average concentration dropped from 19.7 mg·L −1 to 8.9 mg·L −1 . Changes in these marine environmental factors were bound to affect the phytoplankton communities.

Effects of Multiple Human Activities on Phytoplankton Community
The increase of nutrients and the decrease of suspended solids did not promote the growth of phytoplankton. On one hand, nutrients are the basis for phytoplankton growth. An increase in the nutrient concentration and the alleviation of nutrient limitation significantly increased the biomass of phytoplankton [31,32]. On the other hand, suspended solids can significantly affect the light intensity in the water and consequently affect phytoplankton growth [33]. According to the phytoplankton growth theory, the environment in 2015 should have been more suitable for phytoplankton growth. However, the phytoplankton biomass actually declined and was significantly lower in both the inner and outer bays in 2015 than in 2010 ( Figure 2). Therefore, other variables must have offset the effect of nutrients on phytoplankton.
The large-scale oyster farming in Qinzhou Bay significantly inhibited the phytoplankton growth and reduced the phytoplankton biomass. Shellfish eat phytoplankton and reduce phytoplankton populations [6,10,34,35]. In bays with intensive shellfish cultivation, the intensity of phytoplankton consumption by the cultured shellfish can account for 90% of the total consumption intensity in the marine ecosystem [36]. The inner bay contained an area of intensive oyster mariculture. Although the concentration of nutrients was high, the phytoplankton biomass was significantly lower in the inner bay than outer bay. This suggests that the top-down regulation of the oysters offset or even overtook the bottom-up regulation effect of nutrients on phytoplankton.
The expansion of oyster mariculture possibly caused the phytoplankton community to be dominated by species with smaller size. Oysters tend to eat large phytoplankton, especially in the absence of food [6]. Most species of Prasinophytes and Syn are relatively smaller than those of Dia. Therefore, Dia is usually the most important food source for oysters in the ocean [36,37]. Jiang et al. found that Prasinophytes became the dominant phytoplankton group in the oyster mariculture area in Daya Bay [6]. In this paper, Pra was much more dominant in inner bay than that in outer bay ( Figure 5). When the scale of oyster mariculture increased in outer bay, the proportion of Dia decreased significantly. Although Cry and Syn were rare in previous surveys based on microscopic analyses [38], the proportion of both increased from 2010 to 2015. As the scale of oyster farming increased, the phytoplankton community may have become dominated by small species. This may cause blooms of non-traditional species to occur more frequently. As we knew, Phaeocystis globosa blooms usually occur around March-May [39,40] and pose threats to mariculture and the cooling system of the nuclear power plant. These threats require constant attention.
The RDA results showed that water temperature, Chl-a, DO, and DIP were the most important factors indicating the distribution of phytoplankton groups ( Figure 3). Water temperature, salinity, and nutrients are three of the most important factors affecting phytoplankton [41]. Dia and Din in Qinzhou Bay were positively correlated with N/P and salinity and negatively correlated with DIP, DIN, and Dsi. These results contradict the general understanding of the relationships between phytoplankton and nutrients [42].
Changes in the community structure differed in various groups. Most species of Dia are thought to be better adapted to estuarine environments [43]. They often occupy an absolute advantage in coastal seas [44]. From 2010 to 2015, the proportion of Dia in phytoplankton declined significantly. Cry constituted much larger proportions of the phytoplankton in 2015. The growth of Cry is inhibited at environmental phosphorus concentrations below 31.0 × 10 −3 mg·L −1 . As the concentration of phosphorus gradually increases, the growth rate of Cry gradually accelerates and its generation time becomes shorter [45]. From 2010 to 2015, the concentration of phosphorus was basically below 31.0 × 10 −3 mg·L −1 , but the concentration had doubled (Table 2). This might be the main reason for the significant increase in the proportion of Cry in the phytoplankton community.

Spatial Heterogeneity of the Phytoplankton Community
Human activities had obviously effects not only on the community structure but also on the spatial distribution of the phytoplankton. Under the influence of both rivers and oyster farming, the inner is characterized by low salinity, rich nutrients, and high-intensity oyster farming. The cluster analysis showed that the phytoplankton communities in the inner bay were difficult to cluster into similar groups with those in the outer bay ( Figure 5).
The spatial fragmentation caused by land reclamation also increases the spatial heterogeneity of the phytoplankton communities. From 2010 to 2015, land was reclaimed on both the east and west coasts of the outer bay. Reclamations separated Stations S8 from S7 and S11 from S9, respectively. Artificial separation causes communities to develop in different directions. Reclamation can reduce community similarity and increase spatial heterogeneity [46,47]. In this paper, the similarity between the stations (between S8 and S7, S11, and S9) decreased from 2010 to 2015, which confirms this theory. Oyster farming could intensify the spatial heterogeneity. More intensive oyster farming resulted in a smaller proportion of Dia and a larger proportion of Pra ( Figure 5).

1.
The ecological environment of subtropical estuaries will be greatly affected by human activities such as nutrient input, oyster farming, and reclamation.

2.
The intensity of phytoplankton consumption by the cultured oyster overtook the bottom-up regulation effect of nutrients on phytoplankton, which will lead to low phytoplankton biomass.

3.
The expansion of oyster farming caused the phytoplankton community to be dominated by species with smaller size, like species of Cryptophytes and Cyanophyta.

4.
Oyster farming and reclamation would divide the estuary into different areas, which would increase the spatial heterogeneity of phytoplankton communities. Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The data that support the findings of this study are available from the corresponding author upon reasonable request.