Association of Seawater Nanoparticle Size Distribution With Diversity of Marine Plankton

Upon tangential ultrafiltration and asymmetric flow field analyses, seawater nanoparticle size distribution and the associations with the total number of bacteria and plankton diversity were evaluated. Of the nanoparticles in the Jiulong River Estuary, 79% were from 1 to 6 nm (C1). C2 (6–20 nm) was correlated with dissolved organic carbon, while C2 and C3 (>20 nm) were associated with dissolved inorganic nitrogen, suggesting that C2 and C3 were impacted by biogenic elements. The total number of bacteria was correlated with C1 and C3. The correlations of Shannon's diversity index (H′) and C3, richness (d), and C1 suggested a link of particle size with phytoplankton biodiversity. Significant correlations of the H′ of zooplankton and C3, and of Pielou's evenness index (J) and C3, suggested C3 as a primary digestion product of zooplankton. The negative correlations of nano-organic carbon (NOC) with d and J suggested NOC as a carbon source for zooplankton. Biodiversity was associated with seawater nanoparticle size distribution. Biological activities regulated the nanoparticle size distribution, which impacted the estuarine nutrient cycling, in turn affecting the stability and balance of biodiversity. Correlation analysis of the size distribution of seawater nanoparticles and the plankton diversity index provided a potential tool for evaluating ecological effects.


INTRODUCTION
Nanoparticles are tiny particles of sizes 1-100 nm, with unique specific surface area, surface charges, redox activity, and bioavailability. Seawater nanoparticles exert increasingly significant impacts on the marine environment and pose potential harm to marine life and human health via transfer through the food chain, which has caused widespread concerns (Brumfiel, 2003;Service, 2004). Previously, 0.45-or 0.7-µm filters were often used to separate dissolved and insoluble matter in seawater (Wilding et al., 2005;Belzile and Guo, 2006). With the development of ultrafiltration technology in the 1990s (Guo et al., 1995;Benner et al., 1997), ultrafiltration membrane that can separate particles of 1 kDa in seawater promoted the research on colloidal matters. Currently, with advanced nano-characterization technologies, the flow field analysis technology has been applied for separating nanoparticles in natural water (Clark et al., 2008;Yang et al., 2013). The successful isolation of nanomaterials of different particle sizes opens a door for studying the biogeochemical behavior of nanoparticles in natural seawater. Studies have shown that the particle size distribution of natural seawater is closely related to biogeochemical processes, and the biological effect is an important factor regulating the particle size of organic marine matter (Hannides et al., 2013;Benner and Amon, 2015). As a primary producer, phytoplankton are the main contributors to marine particulate organic matter (POM) and dissolved organic matter (DOM) (Swan et al., 2011;Herndl and Reinthaler, 2013). The size of the released DOM is related to the phytoplankton community structure and aquatic environment (Baines and Pace, 1991). Microparticles from phytoplankton are converted to nano-or subnano-sized ones through degradation and digestion by heterotrophic bacteria and zooplankton (Ogawa et al., 2001;Hannides et al., 2013). Therefore, it is of great significance to explore the particle size, composition, and ecological effects of nanoparticles in natural seawater.
Jiulong River, the second largest river in Fujian Province of China, is an important water source for residential, industrial, and agricultural applications in southwestern Fujian Province (cities of Longyan, Zhangzhou, and Xiamen). Jiulong River and its surrounding watersheds are very rich in fishery resources, rendering an important area for aquacultural and fishery products. The Jiulong River estuary, representative of a subtropical estuary, provides an ideal site for exploring the potential relationships between the particle size/composition of natural nanomaterials and biodiversity (Jiang et al., 2019).
This study aimed to analyze the particle size and composition of nanoparticles in the seawater, the plankton biodiversity index, followed by assessment of the correlation between the nanomaterials with different particle sizes and the nutrients, e.g., carbon, nitrogen, and phosphorus. The correlation between seawater nanoparticle size and the total number of bacteria, the plankton biodiversity index, is evaluated using statistical analysis by SPSS. The results will improve understanding of the biogeochemical cycling process and the environmental impacts of natural nanoparticles.

Sampling
In January 2018 (dry season), seven sampling sites (Figure 1, J1-J7) were selected along the direction of Jiulong River into the sea (salinity = 0-30 ppt) according to the method of five-interval salinity gradient of two stations. The study area covers rivers, estuaries, and nearshore waters: J1, J2-J5, and J6-J7 represented the riverine, estuarine, and nearshore sites, respectively. During the wet season, rich water and surface runoff introduce more land-based nanomaterials into the estuary area, especially the input of small-sized terrigenous humus; however, it is beyond the scope of this present study.
The zooplankton samples were collected using a type I plankton net vertical trawl, following which the samples were immediately fixed with formalin. The water samples for particle size spectrum, dissolved organic carbon (DOC), nutrients, heterotrophic bacteria, and phytoplankton were obtained with 8-L sunflower-shaped water collectors. Sample storage and transportation were performed under the specification of "GB17378.3-2007 Marine Monitoring Specification Part 3: Sample Collection, Storage and Transportation" .

Sample Analysis
DOC, inorganic nitrogen and phosphorus, and active silicate were analyzed based on the "Environmental Monitoring of Nearshore Waters" 1 . The ecological analysis and evaluation were performed following the "GB17378.7-2007 Marine Monitoring Standards: Ecological Investigation and Biological Monitoring of Offshore Pollution" . Nitrite (NO − 2 -N), nitrate (NO − 3 -N), ammonium (NH + 4 -N), active phosphate, and silicate were determined using a continuous flow analyzer (Skalar San++, Skalar Analytical B.V., Breda, the Netherlands). DOC was measured using total organic carbon (TOC) analyzer (TOC-VCPH analyzer, Shimadzu, Kyoto, Japan). Salinity was measured on site using a multiparameter analyzer (WTW Multi 3430, Weilheim, Germany). Dissolved inorganic nitrogen (DIN) is the sum of NO − 2 -N, NO − 3 -N, and NH + 4 -N. The DOC samples were pre-filtered with a 0.70-µm glass fiber filter (Whatman GF/F). The pretreated samples were passed through a tangential flow ultrafiltration system (Figure 2) including a membrane module (1 kDa, 0.5 m 2 ; Millipore Pellicon, Burlington, MA, USA) and a peristaltic pump (Cole-Parmer Masterflex, St. Neots, UK) for extracting nano-organic carbon (NOC). Thereafter, the pre-filtered liquid (C B ), the retentate (C R ), and the ultrafiltrate (C P ) were collected for on-machine testing. NOC (C n , in milligrams per liter) was calculated using Equations 1 and 2.
where CF is the concentration factor and V B and V R are the volumes (in milliliters) of the pre-filtered liquid and retentate, respectively. The particle size spectrum was measured using the asymmetric flow field method with an AF4 field flow meter (AF2000 MultiFlow FFF, Postnova, Landsberg am Lech, Germany) (Hong et al., 2005). We modified a few key technical parameters of the asymmetric flow field flow analyzer, as follows: polyethersulfone ultrafiltration membrane with a pore size of 1 kDa; channel membrane thickness, 350 µm; carrier liquid, 9‰ NaCl (similar to the electrolyte content of the sample); volume of the loop, 0.6 ml; and sample injection volume of 1.5 ml to ensure that there is no residual liquid in the loop. The flow rates in the pipe (Tip), focus flow (Focus), and vertical direction (Cross) were set to 0.5, 3.8, and 3.5 ml/min, respectively. The focus time and elution time were 10 and 20 min, respectively. During the elution, the tangential flow velocity was maintained constant. The selected wavelength of the UV-vis spectrophotometer was 254 nm.
Vitamin B 12 (VB 12 , 1.3 kDa), lysozyme (14.4 kDa), cytochrome C (12.0 kDa), ovalbumin (45 kDa), and bovine serum albumin (BSA, 66.5 kDa), with known molecular weights, were mixed to prepare the standard solution. Then, the standard  solution was measured using the field flow meter under the same conditions. The relationship between elution time and molecular weight is displayed in Supplementary Figure 1, showing a standard logarithmic curve between molecular weight and the peak retention time. Then, the peak time of the measured sample was substituted into the standard curve to obtain the relative molecular weight of each sample. Afterwards, the standard working curve was fitted in Equation (3). ln(MW) = 1.8994 × ln(T − 11) + 7, 767 where MW is the molecular weight (in kilodalton) and T is the elution time (in minutes); since the system elutes the sample from the 11th min, we take (T -11). The integration area is the area between the 11th and the 51st min. The samples from the Jiulong River estuary were measured with the method above, and the molecular weights of the samples were calculated using the linear regression equation of the working curve. The time was converted to hydrodynamic diameter using the conversion equation between the molecular weight and size (Equation 4).
where S is the total number of species and N is total number of individuals in the sample. ln represents the natural logarithm.
Shannon-Weiner Diversity Index (H ′ ) (Shannon, 1948) where H ′ is the Shannon-Weiner diversity index, s is the total number of species in the samples, i represents the ith component of the specified species, and P i is the proportion of the individual component of each species.
where H ′ is the Shannon-Wiener diversity index and S is the total number of species in the sample.
where N 1 and N 2 are the numbers of individuals of the first and second dominant species in the samples, respectively. N T is the total number of individuals in the samples.

Data Processing
SPSS 20.0 was used for data analysis, assuming a linear relationship between the dependent variable and the independent variable. A linear regression model was used to fit the data of the dependent variables and the independent variables.

RESULTS AND DISCUSSION
Size Distribution of Nanoparticles Table 1 shows that the mean abundance of C 1 ranged from 1.56 to 3.03, with an average of 2.46. The highest values appeared at J2 and J4, while the lowest value appeared in the seaward J7. The mean size of C 2 spanned from 0.13 to 0.62, with a mean of 0.36. The high and low values appeared at J2 and J3, respectively. The mean size of C 3 spanned from 0.062 to 0.57 nm, with an average size of 0.33. The highest and lowest values were 0.57 (J7) and 0.062 (J4), respectively. Single-factor analysis of variance (F = 3.04, p = 0.086) showed that there was no significant difference in the particle size between the surface and bottom layers of the Jiulong River estuary along the seaward direction. This may be due to the intensive water exchange between the surface and bottom layers upon the combined effects of runoff and tidal movement along the Jiulong River. Table 1 shows that C 1 dominated the composition of nanoparticles, accounting for 78% of the total. The outflow time of nanoparticles (Figure 3) was mainly concentrated at 10-16 min, corresponding to the outflow interval of C 1 , which further verified that seawater nanomaterials were mainly composed of small-sized particles. This is consistent with an earlier study which showed that, in the Chukchi Sea, the colloidal materials were primarily dominated by small-sized particles (55.4%) (Lin et al., 2016). Stolpe et al. (2014) and Zhou et al. (2016) obtained similar findings in the Mississippi River and in small rivers in the Gulf of Mexico, which they attributed to the important contribution of small-sized terrestrial humus to river nanoparticles.
The spatial distributions of C 1 , C 2 , and C 3 are shown in Figure 4. No clear distribution pattern of the three components was observed. There was no obvious trend along the Jiulong River into the sea, indicating minute nanoparticles inputs from terrestrial sources. The composition had a poor correlation with salinity. Overall, C 1 , C 2 , and C 3 were all abundant in J2 at the estuary, but low contents appeared in J3. J7 contained a rather low content in C 1 , but high levels in C 2 and C 3 . C 1 was unaffected by the composition of nutrients, but the C 2 component was inversely correlated with DOC (C 2 = −0.081 × [DOC] + 0.61, R 2 = −0.52, n = 14, p = 0.05), suggesting the same origin of C 2 and DOC. Multiple regression analysis of C 2 , C 3 , and nitrite, nitrate, and ammonium salt was performed with Equations 9 and 10, as follows: The results showed that ammonium was proportional to the particle sizes of C 2 and C 3 , whereas nitrite was the opposite. The particle size of the ultra-miniature phytoplankton (Pico) was 0.2-3 µm, which was slightly larger than that of C 3 (<0.1 µm), suggesting that C 3 may contain partially broken phytoplankton. Phytoplankton preferentially uptake ammonium, suggesting that the C 3 component may contain phytoplankton.

Distribution of Total Number of Heterotrophic Bacteria
In January 2018, the total number of bacteria in the surface water of the Jiulong River ranged from 9.3 × 10 3 to 4.3 × 10 4 colony forming unit (CFU)/ml, with an average value of 2.3 × 10 4 CFU/ml. The total number of bacteria in each station was almost in the same order of magnitude. This was consistent with a previous report of 2.0 × 10 4 -8.6 × 10 5 CFU/ml (Hong et al., 2017). The total numbers of bacteria at J4 and J5 at the estuary end were higher, followed by the nearshore sites (J6 and J7). Those of J1 and J2 at the river end were lower. The total number of bacteria at the estuary was four times that at the river. Therefore, the overall distribution of total bacteria was estuary > nearshore > river. The total number of heterotrophic bacteria was not greatly related to salinity, suggesting its negligible source from terrestrial input, whereas it was significantly related to inorganic nitrogen and soluble reactive phosphate (SRP). The result is shown in Equation 11.
The correlation coefficient indicated that nitrite and active phosphate were the promoting factors for the growth of heterotrophic bacterial communities, while nitrate and ammonium were negative factors. In contrast, Hong et al. (2017) showed a different finding in that the distribution of the heterotrophic bacterial community in the estuaries was mainly affected by the input of terrestrial sources, such as salinity and human activities, which may be related to variations in the sampling times and locations.

Phytoplankton Diversity Index
In this work, a total of 89 species of 45 genera and four phyla were identified, of which diatom was the dominant phylum.
The dominant species were Skeletonema costatum, with a mean number of cells of 1.98 × 10 5 /L, accounting for 71.7% of the total cells. From the perspective of ecological group analysis, the widetemperature species were dominant in the number of species and cells. The biodiversity indices of phytoplankton are shown in Table 2. The statistical results indicated that the diversity index (H ′ ) of phytoplankton ranged from 0.903 to 2.950, with an average of 1.715. The diversity indices at stations J1 and J5 were the highest and the lowest, respectively. Abundance (d) ranged from 1.392 to 2.542, with a mean of 1.898, of which stations J7 and J5 had the highest and lowest abundance, respectively. Evenness (J) ranged from 0.197 to 0.580, with an average of 0.340, of which J1 and J5 had the highest and lowest uniformity, respectively. Dominance (D 2 ) was in the range 0.664-0.931, with an average of 0.840, of which stations J5 and J1 had the highest and lowest dominance, respectively. As a whole, J1 had high richness, diversity, and uniformity, but low dominance and a relatively stable phytoplankton community. In contrast, the richness, diversity, and evenness of J5 stations were low, while the dominance was high, and the stability of the phytoplankton community was relatively low.
The multiple regression equations of H, J, and D 2 and inorganic nitrogen, active phosphate, and DOC in the Equations (12)(13)(14) showed that the H ′ and J of phytoplankton were proportional to DOC and DIN and inversely proportional to SRP, suggesting that DOC and DIN were the promoting factors for phytoplankton growth and that the active phosphate may be the limiting factor. Yet, D 2 was directly proportional to   (Fisher et al., 1992), and the relative restriction law can be derived from the relative concentration ratios of the nutrients, according to the stoichiometric threshold limit standards for nutrients proposed by Justić et al. (1995) and Dortch and Whitledge (1992): if Si/P and N/P were both >22, then P was the limiting factor; if N/P < 10 and Si/N > 1, then N was the limiting factor; if Si/P < 10 and Si/N < 1, then Si was the limiting factor. To accurately identify the roles of the nutrients in the estuary of Jiulong River during the dry season, the nutrients were characterized, as shown in Supplementary Table 1. The nitrogen, phosphorus, and silicon contents were all greater than the absolute limit threshold, while Si/P and N/P were 63.8 and 95.6, respectively, which were both >22. Therefore, the limiting nutrient for phytoplankton growth was phosphorus, in line with the results from multiple regressions. In the present study, we focused on the effects of limiting nutrients on phytoplankton diversity. Salinity was not considered, but it is discussed in the correlation with zooplankton diversity.

Zooplankton Diversity Index
A total of three phyla and 49 species of zooplankton were identified in this work, of which arthropods were dominant. The dominant species was Acartiella sinensis. The average biomass and individual density were 56.4 mg/m 3 and 3,159.4 ind./m 3 , respectively. The community was mainly composed of estuarine low-salt groups and wide-temperature and wide-salt groups. Table 3 shows that the zooplankton diversity index (H ′ ) ranged from 1.68 to 3.66, with an average of 2.64. The evenness of zooplankton was in the range 0.40-0.77, with a mean of 0.63. The abundance of the zooplankton ranged 0.75-2.77, with an average of 1.64. Dominance of zooplankton (D 2 ) ranged from 0.29 to 0.79, with an average of 0.53. Overall, the levels of stability of the zooplankton community at J6 and J4 were relatively high and low, respectively. The species diversity and the evenness of the zooplankton in the studied area were at a medium level. No significant changes in the diversity and evenness of zooplankton were noted. The stability of the zooplankton community structure was weakened and the sensitivity was increased.
The H ′ value of zooplankton was significantly negatively correlated with the active phosphate content (R 2 = 0.72, n = 7, p = 0.05), but weakly correlated with salinity (S) (R 2 = 0.67, n = 7, p = 0.10). This may be due to the extensive salinity of the phytoplankton community in the estuarine area and the limited influence of salinity. Meanwhile, active phosphorus was the limiting nutrient for the phytoplankton community structure in the sea area, which affected the distribution of the phytoplankton community to a certain extent, thus impacting the community composition of the predators. It ultimately led to a significant negative correlation between the zooplankton H ′ and SRP. For the multivariate regression analysis of d of zooplankton, salinity, DIN, and SRP, the equation was: It indicated that the zooplankton abundance was mainly affected by salinity and nutrients, of which salinity and inorganic nitrogen were the promoting factors, suggesting that the land source input played a role in zooplankton abundance in the estuarine area. The significant negative correlation between d and SRP may be related to the limitation of phytoplankton diversity by phosphorus.

Correlation Between Total Number of Heterotrophic Bacteria and Nanoparticles
The total number of heterotrophic bacteria was significantly negatively correlated with C 1 (R 2 = −0.71, n = 7, p = 0.05) and C 3 (R 2 = −0.79, n = 7, p < 0.05). This may be due to heterotrophic bacteria being able to uptake easily degradable small molecular organics, especially amino acids and carbohydrates, to form large molecular organics in their cells (Ogawa et al., 2001). C 1 and C 3 in nanoparticles may be significant components of biodegradable small molecular matter.

Correlation Between Phytoplankton and Nanoparticles
Phytoplankton abundance (d) characterized the number of phytoplankton species, which was weakly negatively correlated with small-sized nanoparticles, C 1 (R 2 = −0.53, n = 7, p = 0.22). It may be related to C 1 being a required nutrient component for the phytoplankton growth site, while H ′ and J represented the maximum species diversity and diversity index, respectively. The large particle size component (C 3 ) in the nanoparticles was strongly positive with phytoplankton d, H ′ , and J, with correlation coefficients, R 2 , of 0.72 (n = 7, p = 0.05), 0.64 (n = 7, p = 0.12), and 0.63 (n = 7, p = 0.12), respectively. The above correlation may be related to the following three factors: (1) the particle size of the ultra-micro phytoplankton (Pico, 0.2-3 µm) decreased due to decomposition of the ultrafiltration process, generating particles of 20-100 nm to be an important group of C 3 ; (2) nutrients are needed for phytoplankton growth, and thus small-sized particles, C 1 , may be an important source of nutrients for its growth; and (3) phytoplankton, as a primary producer, was an important contributor of DOM in the ecosystem. The particle size of the DOM was related to the population structure, physiological state, and water environment (Baines and Pace, 1991;Benner and Amon, 2015). Additionally, large-sized particles, C 3 , may be a key component of DOM. The results indicated that phytoplankton, as a primary producer, may be an important provider of C 3 , and C 1 as a nutrient source for phytoplankton growth may affect the stability of biodiversity.

Correlation Between Zooplankton and Nanoparticles
The large-sized molecules (C 3 ) in the nanoparticles were significantly positively correlated with H ′ and J, with R 2 values of 0.75 (n = 7, p = 0.05) and 0.88 (n = 7, p < 0.05), respectively, which may be associated with zooplankton feeding activities. Some studies have shown that zooplankton can convert micron-sized particles into nanoscale or sub-nanoscale particles when feeding on phytoplankton. Meanwhile, due to the vertical movement of zooplankton, the vertical distribution of particle sizes was disturbed (Hannides et al., 2013). C 3 , as an important part of nanoscale matter, further corroborated the above conclusion. The dominance variable (D 2 ) represented the dominant species of zooplankton in the Jiulong River estuary. It was significantly negatively correlated with C 1 (R 2 = −0.73, n = 7, p = 0.06) and C 3 (R 2 = −0.74, n = 7, p = 0.06) in nanoparticles, which may be related to the following reasons: (1) the mirroring relationship between H ′ and D 2 , and (2) zooplankton may ingest the DOM in C 3 . NOC had a strong negative correlation with zooplankton abundance (d) and evenness (J); the R 2 values were −0.63 (n = 7, p = 0.13) and −0.71 (n = 7, p = 0.06), respectively, which may be related to NOC as an important part of POM, preferentially ingested by zooplankton. NOC may be an important carbon source for zooplankton in the sea area. The positive correlation of NOC and dominance (D 2 ) (R 2 = 0.56, n = 7, p = 0.18) may be related to the mirror relationship between D 2 and J. This indicated that the large-sized molecules in the nanoparticles may impact zooplankton diversity.

Correlation Between Nanoparticle Size, Total Bacteria, and Phytoplankton
The correlations of the extracted nanoparticles of different sizes with the number of bacteria, the diversity indices of phytoplankton and animals, are presented in Supplementary Table 2. The total number of bacteria was negatively correlated with the nanoparticles of C 1 and C 3 . The abundance of phytoplankton (d) was weakly negatively correlated with C 1 , while C 3 was positively associated with d and evenness, indicating that C 3 will affect the stability of phytoplankton diversity. C 3 was significantly positively correlated with the zooplankton H ′ and J, while C 1 and C 3 were strongly negatively correlated with D 2 . This suggested that C 1 and C 3 played a role in the balance of zooplankton diversity. NOC was negatively correlated with the d and J of zooplankton and positively correlated with D 2 , suggesting that NOC may impact the stability of zooplankton diversity.

Implications
The indices of plankton, such as the H ′ , d, J, and D 2 , explain the characteristics of the community structure from the macro level, yet they fail to yield the internal driving force of the structure of the community. The correlations between nanoparticles and H ′ , d, J, and D 2 provide insights for an in-depth exploration of the endogenous force of planktonic community. In future studies, isotopic tracing techniques may be employed to examine the fate and transport of nutrients at nanoscale, particularly how nanoscale nutrients affect the energy flow transformation process of bacterial → phytoplankton → plankton. Overall, the results of the present study provide a new approach for environmental protection in estuarine areas.

CONCLUSION
In the dry season, nanoparticles in the Jiulong River estuary were mainly composed of small nanoparticles, C 1 , accounting for 79% of the total, and no significant distribution differences were found between the surface and bottom layers. C 2 and C 3 were positively correlated with NH + 4 -N. C 1 , C 2 , and C 3 were involved in the energy flow transformation of heterotrophic bacteria → phytoplankton → zooplankton in the ecosystem. Heterotrophic bacteria may ingest C 1 and C 3 , while phytoplankton growth mainly absorbed the nutrients of C 3 and was the killer of nanoparticles; phytoplankton was the producer of C 3 in the nano-component. The results showed that biodiversity and the size of nanomaterials were mutually influencing processes. Firstly, biological activities would affect the particle size distribution of marine nanomaterials, which played an important role in the transformation of the size of DOM. Meanwhile, the change of particle size composition would also affect the nutrient element cycle process in the estuarine area, thereby affecting the stability and balance of biodiversity.