Samples were collected during winter and summer of 2009 in the nSCS (18°30'–23°00'N, 111°00'–117°00'E). The winter cruise was from 11 to 23 February 2009 and comprised 22 stations (two transects perpendicular to the coast line and two transects parallel to the coast line, Fig. 1). The summer cruise was from 12 to 23 August 2009, and also comprised 22 stations (same configuration as the winter cruise, Fig. 1). To understand the spatial variation of the different variables, the nSCS was categorized into the coastal zone (<50 m), continental shelf region (50–200 m), and open sea (>200 m) according to depth.

Samples of seawater were collected for the determination of temperature (T), dissolved oxygen concentration (DO), chlorophyll a concentration (Chl a), and bacterial abundance (BAC) using 5 dm³ Niskin bottles at the following depths: surface layer (0 m), 10 m, 25 m, 50 m, 100 m, 150 m, and 200 m. There were seven and five sample stations for measuring CR_eu in winter and summer, respectively. Sampling depths corresponded to light levels of 100%, 10%, and 1% of the surface irradiance. Water samples in each layer were pre-screened through 200 μm mesh to remove larger zooplankton, and were then incubated in parallel.

The integrated average values of environmental parameters measured within the water column (<200 m) in the nSCS are presented in Table 1. The results show that the distribution of environmental parameters in the nSCS exhibit significant seasonal differences and spatial variation. In the winter, T, DO, Chl a and BAC were (21.8±2.1)°C, (484.55±64.52) μmol/dm³, (0.73±1.58) mg/m³, and (0.9±0.4)×10⁶ cell/cm³, respectively. In terms of horizontal distribution, T increased with increasing distance from the coastline, while DO, Chl a and BAC gradually decreased from the coastal zone to the open sea. The contributions of net-, nano- and pico-fractions to Chl a were 49%, 27% and 24%, respectively. In terms of vertical distribution, T decreased with increasing water depth. Coastal water was uniform owing to vertical mixing, and thermocline intensity was weak in the nSCS (Fig. 2).

In comparison with winter, the entire study area was characterized by high T ((24.8±4.8)°C), low DO ((419.46±77.71) μmol/dm³), and low biomass (including Chl a ((0.50±1.23) mg/m³) and BAC ((0.8±0.4)×10⁶ cell/cm³) in summer. In terms of horizontal distribution, T, Chl a and BAC gradually decreased from the coastal zone to the open sea; however, the highest value of DO was encountered in the continental shelf area. Net-, nano and pico-fractions contributed 50%, 29% and 21% to Chl a, respectively, which was similar in comparison with that in winter. In terms of vertical distribution, the intensity of the thermocline in the summer was relatively strong. The thermocline in the coastal zone appeared at a depth of 20–30 m, while the thermocline of the open sea appeared at 50–75 m layer (Fig. 2). Both Chl a and BAC presented the phenomenon of subsurface maximum, significantly (Le et al., 2015; Zeng et al., 2017).

In the South China Sea, in general, there is a significant concomitant relationship between physical, chemical and biological process at the large- and mesoscales (Ning et al., 2004), such as biological processes in the upper layers were greatly affected by monsoon-driven oceanic movements. In summer, estuary and coastal upwellings are induced by the prevailing southwest monsoon, but the upper water of the whole sea area generally presents as oligotrophic, with a lower plankton community biomass owing to the influence of strong anti-cyclonic circulation on the basin scale (Liu et al., 2002). However, in the winter when the northeast monsoon prevails, the temperature decreases and the vertical mixing effect becomes strong. Coastal waters, driven by the northeast monsoon, move towards the southwest, and bring abundant nutrients into the South China Sea. Meanwhile, a large and strong cyclonic western boundary current forms in the western South China Sea, thereby causing an upwelling of lower layers, and thus supplements nutrients in the euphotic zone (Behrenfeld and Falkowski, 1997). Hence, the primary productivity in winter increases, which supports the higher standing stock of phytoplankton during this time. In addition, the dissolved organic matter, either released from the phytoplankta or from the process of zooplankta consuming phytoplankta, can be used by bacteria for growth and propagation to enhance bacterial abundance (Liu et al., 2007). High levels of nutrients would also lead to higher BAC, for bacteria can only effectively use dissolved organic matter when sufficient quantities of inorganic nutrients are present (Tupas and Koike, 1990; Cotner et al., 1997; Donachie et al., 2001). This study also demonstrates that the intensity of the thermocline was relatively weak in the winter (Fig. 2), and the biomass of the plankton community (both Chl a ((0.73±1.58) mg/m³) and BAC ((0.9±0.4)×10⁶ cell/cm³)) in the water column in winter was significantly higher than that in the summer ((0.50±1.23) mg/m³ and (0.8±0.4)×10⁶ cell/cm³, respectively). In addition, owing to the dual influences of the infusion of eutrophic rivers and offshore upwelling, high levels of nutrients and a high plankton community biomass are commonly reported in the coastal zone as opposed to in the open sea (Smith, 2006; Liu et al., 2007; Legrand et al., 2015). Similarly, high values of both Chl a and BAC encountered in the coastal zone were observed in the present study, along with a relatively high nutrient concentration, while the lowest values appeared in the open sea (Table 1).

The distribution of CR_eu, which also exhibited high temporal and spatial variability in the nSCS, was consistent with data for Chl a and BAC. Seasonally, CR_eu was substantially higher in the winter ((0.53±0.27) μmol/(L·h)) than in summer ((0.26±0.20) μmol/(L·h)). In terms of horizontal distribution, the CR_eu observed in the coastal zone and continental shelf ((0.82±0.20) and (0.50±0.25) μmol/(L·h), respectively) were quite higher than that observed in the open sea ((0.25±0.17) μmol/(L·h)). The highest CR_eu was observed in the continental shelf during winter, while it was observed in the coastal zone during summer. In terms of vertical distribution, the highest CR_eu occurred in the layers where the Chl a and BAC maxima appeared. Previous studies indicate that plankton respiration is positively correlated with the biomass of the plankton community (Robinson et al., 2002; Chen et al., 2003). However, as the composition and biomass of the plankton community change concomitantly with the nutritional conditions and physicochemical environment, there is also a change in the respiration rate. In the East China Sea, bacteria were shown to be the main contributor of plankton respiration when nitrate concentrations were greater than 0.3 mmol/m³; however, when nitrate concentration was not greater than 0.3 mmol/m³, protozoans governed the respiration of the plankton community (Chen et al., 2003). In addition, a study on plankton respiration in the eastern Atlantic found that BAC and Chl a could explain the variability in the respiration of the plankton community by 70% and 61%, respectively (Robinson et al., 2002). In comparison with the aforementioned ocean regions, nutrients in the nSCS were relatively abundant, and the nitrate concentration at most stations was greater than 0.3 mmol/m³. Thus, it is unlikely that phytoplankton and bacterial biomass decreased, and that protozoans became the major component of plankton respiration as a result of nutrient limitation. The results of the present study indicate that CR_eu in the nSCS was significantly and positively correlated with BAC and Chl a, where Chl a could explain 36% of the variability in CR_eu, and BAC could explain 55% of the variability in CR_eu, suggesting that heterotrophic bacteria are the main contributors to plankton respiration in the nSCS, and that phytoplankton and heterotrophic bacteria affect the majority of CR_eu in a cooperative manner.

Our results also indicate that the abiotic factors T and DO were weakly correlated with CR_eu, with relatively lower R², when compared with the correlations between CR_eu and biotic factors. In the nSCS, CR_eu was negatively correlated with T (R²=0.20, p<0.05), which was inconsistent with the reports of a previous study conducted in the Chesapeake Bay, where the seasonal variation in CR_eu was positively correlated with temperature (Sampou and Kemp, 1994). Generally, plankton respiration demonstrates a significant sensitivity to fluctuations in temperature, where increasing temperature would accelerate the respiration rate (Ganf, 1974). In general, Q₁₀—the temperature sensitivity coefficient—of plankton respiration has been shown to be approximately 2.4 (Jankowski et al., 2014). In other words, with every 10°C increase in temperature, plankton respiration would also increase 2.4 times the original value. The temperature-dependent nature of plankton respiration has also been demonstrated in the North Atlantic Ocean Gyre (García-Corral et al., 2014). However, the respiration rate of the plankton community in this case was also controlled by the biomass of the plankton community (García-Corral et al., 2014), where this was mainly dependent on nutrient level and the concentration of dissolved organic carbon. In the nSCS, the increasing temperature would enhance the thermocline, thereby preventing the high-nutrient water under the thermocline to be transported into the euphotic zone, which restrains the growth of phytoplankton. Therefore, increased temperature indirectly reduces primary production in the water, and consequently lowers the biomass of phytoplankton and bacteria (Liu et al., 2007; Le et al., 2015; Hao et al., 2016), thereby decreasing the respiration of the plankton community. According to the results of the present study, we believe that in the nSCS, the effect of biomass on CR_eu is more pronounced than temperature. This could be the main reason for the negative correlation between CR_eu and T observed in the nSCS. In addition, the CR_eu in the nSCS was positively correlated with DO (R² = 0.45, p<0.05). The results of a previously published study indicated that in waters with relatively abundant levels of nutrients, low levels of DO (31.25–62.50 μmol/dm³) would limit the respiration of the plankton community, and thus the influence of DO above the limited value on respiration was not very significant (Sampou and Kemp, 1994). In the present study, the minimum values of DO in winter and summer in the nSCS were 286.07 μmol/dm³ and 177.55 μmol/dm³, respectively, which were both substantially higher than the DO concentration that is considered to be limiting in plankton community respiration. Thus, the DO in the study area would not impose a restriction on CR_eu. The positive correlation observed between DO and CR_eu could be a result of the following scenarios: (1) DO increased with increasing primary productivity, where approximately 10% of the primary productivity was released in the form of soluble inorganic carbon, thereby accelerating the growth of bacteria; (2) DO was generally positively correlated with phytoplankton biomass, which was also the main contributor of respiration under conditions of darkness (Markager et al., 1992; Zheng et al., 1992). Therefore, taken together, the aforementioned information indicates that abiotic factors do not directly influence the variability of CR_eu in a significant manner, but they mainly exert indirect effects on the variability of CR_eu via their effects on biotic factors.

In comparison with the published findings regarding respiration in subtropical ocean waters, the distribution of CR_eu in the nSCS presented consistency in a general trend, and also preserved the peculiarity of the investigated area (Table 2). On one hand, the CR_eu in the study area decreased sharply from the coastal zone to the open sea, which was consistent with trends of plankton respiration gradually decreasing from the inner to the outer southeastern continental shelf in the USA (Pomeroy et al., 2000) and the continental shelf regions of the Mediterranean and the East China Sea (Le Ferla et al., 2006; Chen et al., 2003). Affected by factors such as the input of terrestrial organic matter and near shore upwelling, nutrients in the coastal zone and the continental shelf are relatively rich, and effectively support the high plankton community biomass observed in these regions, which causes the relatively higher respiration of plankton communities in the coastal zone and inner continental shelf (Iriarte et al., 1991; Chen et al., 2003). On the other hand, plankton community respiration in the winter in the nSCS was significantly higher than that in the summer, in contrast to the findings of a higher-latitude continental shelf near Georgia, USA (Jiang et al., 2010). Bearing in mind the Q₁₀ value of plankton (2.4) (Jankowski et al., 2014), the temperature difference between winter and summer in the continental shelf of Georgia was between 10–20°C (which exerted a significant influence on CR_eu), while the temperature difference between winter and summer in the present study was approximately 1–6°C, and appears to have had little influence on the respiration of the plankton community.

A comparison between the CR_eu in the nSCS and that of open sea regions of similar latitudes can be seen in Table 2. This comparison shows that the CR_eu in the nSCS in summer was higher than the values observed in oligotrophic regions (e.g., Central North Pacific Ocean Gyre, North Atlantic Subtropical Gyre, Eastern Tropical Atlantic Gyre) (Williams and Purdie, 1991; Gonzále et al., 2001; Robinson et al., 2002), and was slightly lower than that in the open sea areas of upwelling with relatively abundant nutrients (North West African Upwelling) (Robinson et al., 2002).

It is also important to note that variations in sampling and incubation time could lead to slight differences in the resulting measurements of plankton respiration. As aforementioned, it is ideal to collect water samples before dawn to ensure consistency between metabolic level and the circadian rhythm of plankton communities (Gasol et al., 1998). However, some studies have indicated that the activities of bacteria and the abundance of heterotrophic flagellates were the highest at midnight or during early morning (Kuipers et al., 2000). Therefore, collecting water samples before dawn could result in the estimated respiration rate of plankton communities to seem high, while collecting water samples during the daytime could result in an underestimation. Thus, in the present study, the later collection times may have resulted in an underestimation of the respiration rate. Likewise, differences in incubation may also result in differences in respiration rate estimation. In general, the incubation time is 24 h. Under the assumption that the respiration rate increases linearly with time, a 24 h incubation is ideal to record the complete diurnal variation of plankton community respiration. However, Bender et al. (1999) points out that the accelerated growth, grazing, and death of microzooplankton in the absence of large grazers during the incubation period in bottles may result in underestimations of plankton community respiration rates, and that using short incubation times can, at least partially, alleviate this problem. In the present study, the incubation time was 6 h, which could have caused a slight overestimation of CR_eu, even though it weakened the bottle effect. Despite the sampling and incubation times in the present study, the differences in CR_eu between this study and other similar studies are not substantial, and trends in the large-scale distribution of CR_eu tendency are comparable. Therefore, we suggest that the error from methodological variation did not exert a significant influence on large-scale seasonal variations in respiration and spatial tendencies.

The correlations between plankton community respiration and environmental parameters have been previously used to estimate plankton community respiration, either partially or entirely (Jensen et al., 1990; Chen et al., 2003). In addition, it has been shown that the correlation between plankton community respiration and a single parameter was significantly lower than the correlation between plankton community respiration and multiple parameters (Chen et al., 2003). In the present study, the biomass, physical environment, and community structure were considered in the multiple linear regression (since DO in the investigated area was much higher than the DO concentration limiting plankton respiration, it was not considered here). As indicated in Table 3, in the regression group with two variables, the combination of BAC and Chl a was the closest predictor of CR_eu, and could explain 82% of the variation in CR_eu, indicating that bacteria and chlorophyll were the main factors influencing plankton respiration in the nSCS. After adding the variable of temperature, and using the size-fractioned chlorophyll concentration instead of Chl a, R² did not significantly increased, indicating that temperature and community structure have little predictive effect on CR_eu. Thus, the combination of BAC and Chl a could be the most effective predictor of the distribution of CR_eu in the nSCS.

Based on the multiple linear regression equation of CR_eu, BAC and Chl a, and the measurements of Chl a and BAC in the water column at each station, plankton respiration in the water column (CR_int) within 100 m of the study area was estimated. The average CR_int in winter and summer in the nSCS was (17.32± 20.27) mmol/(m³·d) and (11.81±12.18) mmol/(m³·d), respectively, which is consistent with the distribution of CR_eu. However, CR_int below the euphotic zone of the open ocean—estimated from the multiple linear regression equation of CR_eu, BAC and Chl a—contained negative values. The reason might be the inconsistencies in plankton biomass and respiration inside and outside the euphotic zone. The changes in bacterial metabolism, which was caused by the sharply decreasing temperature from the upper euphotic zone to deeper layers, might have resulted in the above-mentioned inconsistency. It has been found in controlled experiments on respiration that when the controlled-environmental temperature is 5°C lower than the in situ temperature, plankton respiration dropped by about 20%–50%, and bacterial respiration also presented significant temperature dependency (Wang, 2014). Our observations indicated that the average temperature in the 100–200 m deep waters was 7.6°C lower than the average temperature in the waters above 100 m. The rapid drop in temperature possibly caused the decrease in bacterial metabolism below the euphotic zone, thereby causing a greater decrease in plankton respiration. Moreover, there were differences in the quality and quantity of organic substances available for consumption by the microbial communities in the upper and deeper layers of the ocean (Jiao et al., 2014). The dissolved organic matter brought from the outside and the dissolved organic carbon generated by primary productivity mainly exist in the euphotic zone, and nutrient concentration in the euphotic zone is relatively high and easily utilized by bacteria; below the euphotic zone, the growth of bacteria is slow because of the inefficient degradation of dissolved organic matter and the low availability of nutrient substances. Hence, owing to the changes in conditions such as temperature and nutrient availability in the water column, there may have been a nonlinear relationship between plankton biomass and respiration outside the euphotic zone. Using the empirical relationship obtained from the parameters in the euphotic zone may lead to an overestimation of plankton respiration in the water column.