In the Yellow Sea ( Figure 1 ), we sampled copepods at nine stations in April 2019 (spring), six stations each in August 2020 (summer) and November 2021 (autumn), and seven stations in early March 2022 (winter). In summer, typhoon Bavi (950 hPa, 40 m/s) passed through the study area a few days before sampling.

A flowmeter (Model 438 115, Hydro-Bios, Kiel, Germany) attached conical net (diameter, 60 cm; 115 µm mesh) was towed vertically from the bottom to the surface. The live sample was filtered through a 115 µm net to remove seawater, frozen in liquid nitrogen, transported to the laboratory, and stored at –70°C until measurement. Because the collection time differed among stations ( Table 1 ), no distinction was made between day and night.

Temperature and salinity were monitored at the sampling stations using a conductivity, temperature, and depth instrument (Sea-Bird 911 CTD; Sea-Bird Scientific, Bellevue, WA, USA). Temperature and salinity data from adjacent stations were included to better describe the vertical distribution. Seawater samples (500 mL) were filtered with GF/F filter paper (Whatman, UK), frozen, and transported to the laboratory. Chlorophyll a (chl-a) was extracted for 24 h in a cool, dark place (–20°C) and measured using a fluorometer (10 AU; Turner Designs, San Jose, CA, USA). Sampling was performed at four to five water depths (0, 5, 10, 20 m for stations less than 50 m and 0, 10, 20, 30, 50 for others); the resulting values were averaged and expressed as mean water column chl-a concentrations.

Experiments were conducted on copepodite stages I–V (CI–CV) and adult (CVI) males and females of C. sinicus, without considering naupliar stages. Then, 4–100 individuals of each stage (average of 48 individuals) were selected, washed with distilled water, and subjected to chlorophyll extraction in 90% acetone for 24 h in the dark at –20°C. Fluorescence before and after acidification was measured using a fluorometer (Turner Design, Trilogy, USA) to determine gut pigment values. During the experiment, the laboratory was set to a weak light condition at 18°C, and the temperature of the sample was reduced with an ice pack to minimize pigment destruction.

The gut pigment content (ng chl ind.⁻¹) of copepod individual (ind.) was calculated as the sum of the chl-a and phaeopigment (ng chl eq. ind.^–1) concentrations, as follows (Mackas and Bohrer, 1976; Dagg and Wyman, 1983):

where f₀ is the fluorescence before acidification, f_a is the fluorescence after acidification, k is the machine calibration constant, and R is the acidification ratio during calibration.

The ingestion rate (I, ng chl ind.⁻¹ d⁻¹) of copepods was calculated as follows (Dagg and Wyman, 1983):

where G is the gut pigment content and E is the gut evacuation rate (min^–1), which was obtained from Uye and Yamamoto (1995) and converted into a daily rate daily rate (d^‒1). The population ingestion rate (mg m^–2 d^–1) was determined by multiplying the ingestion rate by the number of individuals (individuals m^–2) for each developmental stage. Grazing impact was expressed as a percentage (%), obtained by dividing the population ingestion rate by the mean chl-a concentration of the water column.

The Kruskal–Wallis test with Dunn’s multiple comparison test was performed to determine the seasonal mean differences in C. sinicus gut pigment, ingestion rate, and grazing impact using GraphPad Prism v8 software (GraphPad Software, San Diego, CA, USA). To determine the relative effects of environmental factors, a redundancy analysis (RDA) was performed between the ingestion rate of the C. sinicus developmental stages and environmental factors, including surface (5 m) and bottom temperature and salinity, as well as surface (0 m) and water column mean chl-a concentrations. A preliminary RDA was performed including all environmental factors, and variables with low significance were excluded. The significance of the RDA results was checked using permutation tests. Canonical community ordination was performed using Canoco v5 (ter Braak and Smilauer, 2012).

In spring, surface temperatures were in the range of 8.1–12.7°C and bottom temperatures were 7.1–10.4°C ( Figure 2 ). Salinities ranged from 31.7 to 33.3 psu in the water column. In summer, vertical mixing caused by a typhoon disturbed the isotherms in coastal areas. Surface temperatures were in the range of 20.2–26.7°C and bottom temperatures were 9.1–17.8°C. 'Salinities ranged from 30.7 to 33.1 in the water column. In autumn, surface temperatures were in the range of 17.2–19.4°C and bottom temperatures were 8.8–16.7°C. Salinities ranged from 31.1 to 32.8 in the water column. In winter, both water temperature and salinity were vertically homogeneous. surface temperatures were in the range of 4.8–8.9°C and bottom temperatures were 4.8–9.9°C. Salinities ranged from 31.9 to 33.0 psu in the water column. The surface chl-a concentrations were 0.35–3.10, 0.27–0.90, 0.28–1.63, and 0.48–8.49 µg L^–1 in spring, summer, autumn, and winter, respectively, and the water column-averaged chl-a concentrations were 0.86–3.30, 0.42–0.92, 0.28–1.58, and 0.48–6.49 µg L^–1, respectively ( Figure 3 ).

The seasonal mean gut pigment content of C. sinicus CI was 0.24–0.49 ng chl ind.⁻¹, with higher values in winter and lower values in spring ( Figure 4 ). CII was 0.39–0.46 ng chl ind.⁻¹, with small seasonal differences. The seasonal mean gut pigment content of CIII-V was 0.45–0.81, 0.47–1.70, and 0.52–2.79 ng chl ind.⁻¹, and that of females was 1.00–6.41 ng chl ind.⁻¹, and that of males was 0.74–1.94 ng chl ind.⁻¹. The values for CIII to adults were higher in spring and lower in summer. The observed seasonal differences were nonsignificant for all stages (p > 0.05).

The seasonal mean ingestion rate of C. sinicus CI was 16.6–42.1 ng chl ind.⁻¹ d⁻¹, with higher values in summer and lower values in spring ( Figure 5 ). CII was 24.1–41.8 and CIII was 32.7–58.5 ng chl ind.⁻¹ d⁻¹, with higher values in autumn and lower values in winter. The seasonal mean ingestion rate was 45.8–114.4 for CIV, 50.2–189.3 for CV, 96.9–438.9 for females, and 69.8–132.3 ng chl ind.⁻¹ d⁻¹ for males. The values for CIV to adults were higher in spring and lower in summer. Seasonal differences were nonsignificant for all stages.

The seasonal mean population ingestion rate of C. sinicus CI was 0.01–0.10 mg chl m^–2 d^–1, with higher values in spring and lower values in winter ( Figure 6 ). The values for CII–V were 0.01–0.09, 0.01–0.15, 0.02–0.27, and 0.19–0.81 mg chl m^–2 d^–1; it was 0.03–0.15 in males and 0.20–1.42 mg chl m^–2 d^–1 in females, being higher in spring and lower in winter. The value for CIII was significantly higher in spring than summer (p = 0.043). The combined population ingestion rates of C. sinicus from the CI to adult stages were 2.98, 0.58, 0.95, 0.60 mg chl m^–2 d^–1 in spring, summer, autumn, and winter, respectively. That of females was the highest among stages, followed by CV, while those of the remaining stages were low ( Table 2 ).

The seasonal mean grazing impact of C. sinicus at CI was 0.02–0.06%, with higher values in spring and lower values in winter ( Figure 7 ). The impact of CII was 0.02–0.06%, and was highest in autumn and lowest in winter. The impact of CIII was 0.02–0.10% and that of CIV was 0.05–0.22%, with higher values in autumn and lower values in summer for both stages. The impact of CV was 0.26–0.49% and those of females and males were 0.18–1.01% and 0.07–0.16%, being higher in autumn and lower in winter. The combined grazing impacts of C. sinicus from the CI to adult stages were 1.66%, 1.43%, 2.04%, 0.65% in spring, summer, autumn, and winter, respectively. The impact of females was highest, at 45.9% of the total impact, followed by CV, at 27.4% ( Table 2 ). Seasonal differences were nonsignificant at all stages.

RDA results for the ingestion rate of C. sinicus and environmental factors in all seasons showed that CIV to adults had positive relationships with the mean chl-a concentrations in the water column ( Figure 8 ). The early developmental stage CI-III were rather related to mean temperature in the water column. The eigenvalues of axes 1 and 2 were 0.34 and 0.06, respectively, and the cumulative percentage variance of axes 1 and 2 was 39.9% ( Table 3 ). Environmental factors explained 97% of the variation in the ingestion rate on axes 1 and 2. The correlation coefficients for axes 1 and 2 were 0.67 and 0.73, respectively. Axis 1 and 2 were significantly related to mean chl-a concentrations (r = 0.55) and temperature (r = 0.67) in the water column, respectively.

Because water temperature affects all phases of the copepod life cycle, including respiration, survival, production and development (Halsband-Lenk et al., 2002; Heine et al., 2019), it may also affect their feeding. In this study, the ingestion rate of C. sinicus on phytoplankton was particularly high in spring and decreased sharply during the summer ( Figures 4 – 6 ). This is likely due to the ecological characteristics of this species, which is sensitive to water temperature. The temperature tolerance range for the survival of C. sinicus is known to be 1–27°C in the Yellow Sea (Wang et al., 2003). The optimal water temperature for C. sinicus was 20°C in the Inland Sea of Japan, and the lower and upper thermal limits for C. sinicus embryonic development were 5°C and 23°C, respectively (Uye, 1988; Uye, 2000). Therefore, high or low water temperatures in summer or winter are unfavorable to the feeding of C. sinicus. In addition to water temperature, food composition in the fields are also important factors affecting copepod feeding. C. sinicus is an omnivorous copepod that can feed on a variety of prey, including not only phytoplankton but also ciliates, flagellates, metazoan eggs and larvae, bacteria, and fungi (Huo et al., 2008; Ho et al., 2017; Yi et al., 2017; Hirai et al., 2018). In the Yellow Sea, seasonal fluctuations in phytoplankton communities have been reported, with diatom blooms occurring in spring and increased proportions of flagella and cyanobacteria in summer and autumn (Gao et al., 2013; Liu et al., 2015). Therefore, the diet of C. sinicus may vary with seasonal environmental changes in its habitat. In a previous study, herbivorous feeding supported more than 82.5% of C. sinicus egg production in spring but only 47.8% in autumn in Bohai Bay, likely driven by the prevalence of large diatoms and sufficient non-phytoplankton food sources during autumn (Zhang et al., 2006).

In general, the copepod ingestion rate increases with increasing water temperature, body weight, and food concentration (Saiz and Calbet, 2011). The positive relationships between ingestion rate and body size ( Figure 5 ) and chl-a concentration ( Figure 8 ) were confirmed for C. sinicus in the Yellow Sea in this study. However, when developmental stage was considered, the relationship between ingestion rate in early stages (CI–CIII) and chl-a concentration were nonsignificant ( Figure 8 ). Feeding habits and prey preferences of copepods differ with developmental stage and sex (Calbet et al., 2007; Brucet et al., 2008; Saiz et al., 2014; Meunier et al., 2016; Ismar et al., 2018). In general, the size of the ingested prey increases with increasing copepod size. Therefore, relatively large and chlorophyll-rich diatoms are more likely to be consumed by adults or late copepodites than by nauplii or early copepodites. This preference explains why chl-a concentrations were significantly correlated with C. sinicus ingestion rates only at CIV to adults in this study ( Figure 8 ). The preferred food type may differ in association with sexual dimorphism. In laboratory experiments with three types of prey, namely, cryptophytes, haptophytes, and diatoms, females of the copepod Acartia tonsa preferred diatoms, while males preferred cryptophytes (Ismar et al., 2018). For C. sinicus, an analysis of gut contents reported that females consumed more diatoms than males (Chen et al., 2010). Methods other than chl-a measurement may produce higher ingestion rates for C. sinicus males on potential prey than are reported in this paper.

A large proportion of grazing impact is associated with CV and female individuals in the C. sinicus population ( Table 2 ). In the Yellow Sea, spring is the main spawning season of C. sinicus (Zhang et al., 2005). As copepod egg production is related to water temperature and prey concentration expressed as chl-a concentration (Richardson and Verheye, 1998; Wang et al., 2009; Kang and Kim, 2021), copepod feeding may increase during the spawning season. On the other hand, males had ingestion rates around three times lower than females ( Figure 5 ). Previous studies of several copepods have reported lower ingestion rates of males compared to females (van Someren Gréve et al., 2017), primarily due to the smaller size of males than females (Hirst and Kiørboe, 2014). Furthermore, during the mating season, females exhibit feeding-focused behaviors, while males engage in both mate searching and feeding behaviors (Kiørboe, 2008; Moison et al., 2013), which may increase female ingestion rates relative to male ingestion rates.

The gut pigment contents and ingestion rates of C. sinicus measured in this study were similar to those reported in a study of the southern Yellow Sea in summer, but higher than observations in inshore such as Taean and Asan Bay ( Table 4 ). Li et al. (2004) noted that their results were lower than local observations in the western bay of the Yellow Sea, indicating regional differences in gut pigment contents and ingestion rates. The seasonal difference in feeding impact observed in this study was less than 1%, which is smaller than in previous studies ( Table 4 ). This finding suggests that C. sinicus likely did not experience food limitation during any season in this study area, at least in terms of phytoplankton prey. The feeding impact of C. sinicus in this study was relatively small compared to other studies conducted in open seas ( Table 4 ). Because the feeding impact of copepods varies greatly depending on the environment and prey conditions of the sea area, it ranges from less than 1% to 40% of phytoplankton stock (Li et al., 2003; Schnetzer and Caron, 2005; Tseng et al., 2008; Saiz and Calbet, 2011; Lee et al., 2012; Jagadeesan et al., 2017). Therefore, caution is needed when comparing the feeding impact values obtained in this study with other studies, even for the same species.

Copepods have endogenous diel feeding rhythms that can be affected by external factors such as food availability, risk for predation and light intensity (Bollens and Stearns, 1992; Calbet et al., 1999; Olsen et al., 2000; Olivares et al., 2020). Diel feeding rhythm amplitudes and the time of active feeding vary among copepod species (Peterson et al., 1990; Atkinson et al., 1996; Yamaguchi et al., 2010; Lee et al., 2012). C. sinicus appears to have diel feeding rhythms of large amplitude, as the difference in gut pigment content over time was 4.20 ng chl ind^–1 for adult C. sinicus in the southern Yellow Sea (Li et al., 2004). In coastal waters of Taiwan, the gut fullness of female C. sinicus was 5% greater in the morning and at midnight than in the afternoon (Chen et al., 2010). This study did not assess the impact of collection time on feeding because sampling times varied ( Table 1 ), but it is likely that there are differences in feeding during the day and night.