INTRODUCTION
Stable carbon and nitrogen isotope ratios as trophic biomarkers have great advantages over traditional gut content analysis for both freshwater and marine fish trophic ecology studies by providing information on the long-term assimilation items (Sugisaki & Tsuda, Reference Sugisaki, Tsuda, Sakai and Nozaki1995; Gu et al., Reference Gu, Schelske and Hoyer1996; Vander Zanden & Vadeboncoeur, Reference Vander Zanden and Vadeboncoeur2002; Bardonnet & Riera, Reference Bardonnet and Riera2005; Harrod et al., Reference Harrod, Grey, McCarthy and Morrissey2005). The contributions of prey to a predator's diet can be estimated using a model if all potential prey are collected and significant differences exist among prey types in their stable isotope patterns (Post, Reference Post2002; Pitt et al., Reference Pitt, Connolly and Meziane2009). However, in most cases, it is difficult to collect all potential diet items, and there are also cases where the stable isotope patterns of different diet items overlap. Therefore, fatty acid composition has been used, together with stable isotopes, to trace the transportation of organic materials through the food web (Sugisaki & Tsuda, Reference Sugisaki, Tsuda, Sakai and Nozaki1995) and to identify trophic interactions within the food web (Budge et al., Reference Budge, Parrish and Mckenzie2001; Alfaro et al., Reference Alfaro, Thomas, Sergent and Duxbury2006; Hessen & Leu, Reference Hessen and Leu2006; Maazouzi et al., Reference Maazouzi, Masson, Izquierdo and Pihan2007; Alfaro, Reference Alfaro2008; Stowasser et al., Reference Stowasser, Mcaallen, Pierce, Collins, Moffat, Priede and Pond2009a, Reference Stowasser, Pond and Collinsb; Wan et al., Reference Wan, Wu, Huang, Zhang, Gao and Wang2010).
Anadromous and catadromous fish are characterized by their extensive migrations between the open sea and fresh water (Nybakken, Reference Nybakken1997). According to previous studies on wild individuals and feeding experiment in the laboratory, the biomarkers of migratory fish were plastic, and changes in the habitat or food were both followed by variation in the patterns of both stable isotopes and fatty acid composition (Bardonnet & Riera, Reference Bardonnet and Riera2005; Ciancio et al., Reference Ciancio, Pascual, Botto, Amaya-Santi, O'Neal, Riva Rossi and Iribarne2008; Prigge et al., Reference Prigge, Malzahn, Zumholz and Hanel2012). Generally, oceanic organic matter is characterized by enriched δ13C and terrigenous organic matter is comparatively depleted in δ13C (Bardonnet & Riera, Reference Bardonnet and Riera2005). The difference in the δ13C patterns of the habitats may make it possible to discriminate the diet source when they shift between the two kinds of habitats. It has been observed that the changes in δ13C of the migratory fish were in accord with the δ13C pattern of the habitat (Bardonnet & Riera, Reference Bardonnet and Riera2005; Ciancio et al., Reference Ciancio, Pascual, Botto, Amaya-Santi, O'Neal, Riva Rossi and Iribarne2008). For instance, increases were found in δ13C values of anadromous rainbow trout (Oncorhynchus mykiss) after migration (Ciancio et al., Reference Ciancio, Pascual, Botto, Amaya-Santi, O'Neal, Riva Rossi and Iribarne2008). The δ13C values of European eels (Anguilla anguilla) decreased during their migration to a river (Bardonnet & Riera, Reference Bardonnet and Riera2005). However, the δ15N values of both species increased during their respective migrations (Bardonnet & Riera, Reference Bardonnet and Riera2005; Ciancio et al., Reference Ciancio, Pascual, Botto, Amaya-Santi, O'Neal, Riva Rossi and Iribarne2008). The observed increase in the polyunsaturated fatty acids concentrations of salmon after their migration to the ocean could be due to their dietary changes or to physiological adaptability, because when fish migrate from fresh water to salt water, they will adjust their fatty acid composition as a physiological adaptation (Saddler et al., Reference Saddler, Koski and Cardwell1972). It seems that both stable isotope profiles and fatty acid composition of migratory fish are under the influences of multiple factors including diet composition, location and ontogeny of individuals (Saddler et al., Reference Saddler, Koski and Cardwell1972; Bardonnet & Riera, Reference Bardonnet and Riera2005; Harrod et al., Reference Harrod, Grey, McCarthy and Morrissey2005; Ciancio et al., Reference Ciancio, Pascual, Botto, Amaya-Santi, O'Neal, Riva Rossi and Iribarne2008), which will bring challenges in the applicability of the biomarkers to the trophic study of migration species. There are several studies that reported the variation in biomarkers of two well-known migratory fish, rainbow trout and European eels (Bardonnet & Riera, Reference Bardonnet and Riera2005; Harrod et al., Reference Harrod, Grey, McCarthy and Morrissey2005; Ciancio et al., Reference Ciancio, Pascual, Botto, Amaya-Santi, O'Neal, Riva Rossi and Iribarne2008), but in view of the combined influences from multiple factors, shift in the diet sources and trophic levels during their migration period are still not well identified.
Tapertail anchovy (Coilia mystus) is a well-known estuarine migratory species in China and widespread in Chinese coastal waters (He et al., Reference He, Li, Liu, Li, Murphy and Xie2008). In the Changjiang Estuary fishery in China, it provides approximately 48.6% of the total fish and shrimp catch and became one of the most important commercial species in this area when the total fishery yield decreased in recent years (Liu et al., Reference Liu, Zhang, Xu and Shi2004; He et al., Reference He, Li, Liu, Li, Murphy and Xie2008). Besides the commercial value in the coastal region, C. mystus also occupies a crucial ecological position. It preys predominantly on zooplankton, and is preyed upon by predators such as hairtail (Trichiurus lepturus), little yellow croaker (Pseudosciaena polyactis) and white spotted conger (Conger myriaster) (Zhou et al., Reference Zhou, Xu and Xu2004). The Changjiang Estuary and Hangzhou Bay are two important spawning sites for C. mystus inhabiting around the Changjiang Estuary, and the Zhoushan Islands offshore are its primary overwintering site (Zhou et al., Reference Zhou, Xu and Xu2004), as shown in Figure 1. In spring, adult fish gather and migrate to estuarine brackish water to spawn (Yuan & Qin, Reference Yuan and Qin1984; Zeng & Dong, Reference Zeng and Dong1993; He et al., Reference He, Li, Liu, Li, Murphy and Xie2008). Fish larvae graze and grow in the Changjiang Estuary and north Hangzhou Bay (Ni, Reference Ni1999). In October and November, when the water becomes cold, the adult population and the juvenile fish start to move offshore and pass the winter in deep waters from December to April of the next year (He et al., Reference He, Li, Liu, Li, Murphy and Xie2008). In May, offshore overwintering C. mystus gather and migrate back into the estuary area to spawn (Ni, Reference Ni1999).
Fig. 1. Sampling locations for Coilia mystus around the Changjiang Estuary in China. Migration routes are labelled with an arrowed line. Solid and dotted arrowed lines imply spawning and overwintering migration routes, respectively.
Coilia mystus is different from other anadromous migratory species (e.g. salmon) in the migration distance and the habitat. The migration distance of C. mystus around the estuary is approximately 200 km, maybe ten times shorter than that of salmon. Further, unlike salmon, C. mystus never goes upstream to the freshwater, only spawns in the estuary, so the differences between the biomarker patterns of each habitat may not be as distinct as those of salmon. On the other hand, the estuarine area is characterized by complex carbon sources, making the application of biomarkers in this area very challenging and interesting. Additionally, study on the biomarkers of the short distance migratory fish is very limited. Information about the diet composition of C. mystus in different migration stages was not well documented, but it is important for understanding the population dynamics of C. mystus. It is unknown how the stable isotopes and fatty acid composition of the fish change during migration and whether dietary influence can be reflected on the biomarker variation. To answer these questions, C. mystus individuals were collected at different migration stages and their stable carbon and nitrogen isotope patterns and fatty acid composition were analysed. Then the diet sources among different migration stages of C. mystus around the Changjiang Estuary were inferred from variation in the two biomarkers.
MATERIALS AND METHODS
Sample collection
Coilia mystus in various migration stages were captured from three habitats (i.e. the Changjiang Estuary, Hangzhou Bay and the offshore Zhoushan Islands) in 2008 and 2009, by the fixed net (25 m width, 5 m height, mesh size 2.5–12 cm) and bottom trawl (7.5 m width, 3.5 m height, mesh size 2.5–12 cm), whose mesh size gradually narrowed from the opening to the end. Samples in four different migration stages, which were labelled M1 to M4, denoting growth stage, overwintering stage, pre-spawning stage and spawning stage, were collected from five stations (HZB1, DS, JT, HZB2, CXI), as shown in Figure 1. The samples were identified by their migration stage and the station where they were collected. The sampling station, date, and migration stages of the collected C. mystus are shown in Table 1. Juvenile samples in growth stage (M1-HZB1) of C. mystus were collected from nearshore waters of the Hangzhou Bay in September 2008. Overwintering stage samples (M2-DSa) were collected north-east offshore from the Zhoushan Islands in March 2009. Pre-spawning stage C. mystus were collected from the waters offshore north-east (M3-DSb) and the south-west (M3-JT) of the Zhoushan Islands in May 2009. Coilia mystus in the spawning stage were sampled near the Changxing Islands in the Changjiang Estuary (M4-CXI) and the north of Hangzhou Bay (M4-HZB2) in June 2009. Standard length and body weight were measured following collection (Table 1). Pre-spawning stage C. mystus collected from DS station (M3-DSb) were separated into four groups according to their standard length, and numbered from I to IV, to investigate whether individuals of different body lengths within one migration stage exhibited stable isotope and fatty acid composition variation.
Table 1. Sample name, date, station, migration stage, number of fish (N), standard length and weight of Coilia mystus collected in 2008 and 2009. SD, standard deviation
The diet composition of C. mystus has previously been investigated (Yuan & Qin, Reference Yuan and Qin1984; Ni et al., Reference Ni, Wang, Jiang and Chen1999; Zhou et al., Reference Zhou, Xu and Xu2004). Copepods, euphausiaceans, Acetes chinensis, decapods (e.g. Leptochela gracilis and Squilla juveniles), and some small juvenile fish (e.g. Pseudosciaena polyactis and Harpadon nehereus) have been identified in the gut of C. mystus collected offshore from the Zhoushan Islands; euphausiaceans, copepods and A. chinensis were the predominant species (Zhou et al., Reference Zhou, Xu and Xu2004). Accordingly, seston, copepod, and fish larvae samples were collected at the CXI station, and A. chinensis was collected at the HZB1 station in the present study. Approximately 1–2 l of water was filtered through a 200 µm net and then filtered on a pre-combusted GFF 47 mm filter (0.7 µm, Whatman, 500°C, 5 h) to obtain the seston sample dominated by the phytoplankton (Pond et al., Reference Pond, Bell, Harris and Sargent1998). Zooplankton were sampled by plankton trawl (diameter 50 cm, mesh size 505 µm), washed with filtered seawater and filtered on a pre-combusted GFF 47 mm filter (0.7 µm, Whatman, 500°C, 5 h). Fish larvae and A. chinensis were then removed manually and placed in plastic bags. All samples were frozen immediately and stored at −20°C. In the laboratory, the white dorsal muscle of the fish was cut from the body. White dorsal muscle was used to study the fish's feeding ecology because it was found to be less variable in δ13C and δ15N than red muscle and other organs such as the liver and heart (Pinnegar & Polunin, Reference Pinnegar and Polunin1999). All samples were lyophilized in a freeze-dryer (LOC-1, Christ, Germany) and stored at −40°C until further analysis. The dorsal muscle was ground into powder before analysis.
Stable carbon and nitrogen isotope analysis
Seston and zooplankton samples were digested with acid (1 M HCl) to remove carbonates and dried at 50°C for 12 h. Coilia mystus dorsal muscle was analysed without acidification. Organic carbon and total nitrogen contents were analysed using an elemental analyser (Vario EL III, Elementar, Germany). Stable carbon and nitrogen isotopes were measured with an isotope-ratio mass spectrometer (Finnegan Delt plus XP, Thermo, Germany). Lipids were not removed prior to measurements. The results were normalized to Vienna Pee Dee Belemnite standard (PDB) for δ13C and the atmospheric N2 standard (AIR) for δ15N (Overman & Parrish, Reference Overman and Parrish2001); results were expressed in δ notation as δX (‰) = ((R sample/R standard) − 1) × 1000, where X = 13C or 15N and R = 13C:12C or 15N:14N. The precision of δ13C and δ15N analyses was ±0.1‰.
Coilia mystus trophic levels were calculated according to the method described in Vander Zanden & Fetzer (Reference Vander Zanden and Fetzer2007), and 3.4‰ was used as the δ15N accumulation coefficient of the marine food web across trophic levels. Acetes chinensis, a primary consumer collected from HZB1, was set as the baseline because primary consumers tended to have more steady isotopic ratios than primary producers (Vander Zanden & Fetzer, Reference Vander Zanden and Fetzer2007). The trophic level of primary consumer was set at 2.
Lipid extraction and fatty acid composition analysis
Methods of lipid extraction and fatty acid composition analysis were referred to Cui et al. (Reference Cui, Wu, Zhang and Wang2012). For dorsal muscle samples, approximately 5 ml of a chloroform and methanol (2:1) solvent was added to each 100 mg sample. For the seston and zooplankton samples, a similar procedure was used except that the weight of the samples used for extraction was adjusted according to their total lipid concentration. Fatty acid data are expressed as a mass percentage of the total fatty acids. The recovery rate of the whole analysis procedure for fatty acids was in the range of 83.3–98.7%. The standard deviations (SDs) of individual fatty acid proportions in replicate analyses were in the range of 0.0–1.5%. Fatty acids were named using a shorthand notation of CA:B n-X, where A indicates the number of carbon atoms, B is the number of double bonds and X indicates the position of the first double bond relative to the terminal methyl group (Budge et al., Reference Budge, Iverson and Koopman2006). If not specifically stated, concentrations of C20:1 and C22:1 are the sum of their n-11 and n-9 isomers.
Statistical analysis
PRIMER 5.0 (Primer-E) and SPSS17.0 (SPSS Inc.) were used to perform the data analysis (Budge et al., Reference Budge, Wooller, Springer, Iverson, McRoy and Divokyet2008; Wan et al., Reference Wan, Wu, Huang, Zhang, Gao and Wang2010; Prigge et al., Reference Prigge, Malzahn, Zumholz and Hanel2012). Values of fatty acids reported in the form of mass % of total fatty acids were root-square transformed to achieve normalization in multivariate analysis (Iverson, Reference Iverson, Arts, Brett and Kainz2008). Bray–Curtis similarity matrices were calculated and principal component analysis (PCA) was performed to investigate the difference in fatty acid compositions of C. mystus among stations and determine which fatty acids accounted for this difference (Loseto et al., Reference Loseto, Stern, Connelly, Deibel, Gemmill, Prokopowicz, Fortier and Ferguson2009). A cluster analysis was performed based on the PCA scores of samples from various stations to separate the samples into several groups. Then, similarity analysis (one-way ANOSIM) was performed to determine whether the difference in stable isotopes and fatty acid composition among the sample groups was significant (Clarke, Reference Clarke1993, Budge et al., Reference Budge, Wooller, Springer, Iverson, McRoy and Divokyet2008). Similarity of percentages (SIMPER) was performed to identify the fatty acids responsible for the difference among sample groups. Relationships between the biomarkers and the body lengths of C. mystus were analysed with correlation analysis. One-way ANOVA was performed to determine whether the specific fatty acids of individuals in different stations were different significantly.
RESULTS
Stable isotope signatures of C. mystus in various migration stages
The carbon and nitrogen contents of dorsal muscles of C. mystus were 44.2 ± 1.7% and 13.7 ± 0.7%, respectively. The C/N ratio ranged from 3.6 to 4.2 (average: 3.8 ± 0.2). The δ13C values of C. mystus were in the range of −21.5 to −15.4‰ (average: −17.5 ± 1.6‰). Values of δ15N were in the range of 6.9–15.8‰ (average: 9.5 ± 2.2‰). ANOSIM was performed based on the δ13C and δ15N values of C. mystus samples using PRIMER 5.0. The results (ANOSIM, R = 0.605, P = 0.001) suggested a significant difference in the stable isotope signatures among stations. Increases were found in both the δ13C and δ15N values across the migration stages (Figure 2). The average δ13C value of A. chinensis was similar with that of C. mystus at M1-HZB1. Coilia mystus at M1-HZB1 were significantly different from those of C. mystus at M2-DSa (ANOSIM, R = 0.809, P = 0.003) and other stations (M3-DSb, M3-JT and M4-HZB2, M4-CXI) (ANOSIM, R = 0.995, P = 0.01) in δ13C values. Coilia mystus can be separated into three groups according to the δ15N values. The values of individuals at M1-HZB1, M2-DSa, and M3-JT were the lowest, values for M3-DSb, M4-HZB2 were intermediate, and values of M4-CXI were the highest. Significant differences existed among the three groups (ANOSIM, R = 0.671, P = 0.001). The average trophic level of C. mystus at M1-HZB1 was 3 (Figure 2).
Fig. 2. Dual isotope plot of Coilia mystus and zooplankton (Acetes chinensis) with trophic levels. Data are presented as the average ± standard deviation.
Fatty acid compositions of organisms around the Changjiang Estuary
A large difference existed in the total fatty acid content (TFA) of organisms in the region (Table 2). The seston samples contained the lowest TFA, only 0.2 ± 0.1 mg g−1. The highest TFA value was found in fish larvae (46.1 mg g−1). Of all C. mystus, those collected at M1-HZB1 contained the lowest TFA, and those from M2-DSa contained the highest TFA. There was no significant difference in the TFA contents of C. mystus among different stations except for those at M1-HZB1. The lipid content (mg g−1 of dry weight) of C. mystus was significantly positively correlated to the TFA (r = 0.516, P = 0.006).
Table 2. Fatty acid composition of Coilia mystus and potential prey items (mass % total fatty acids). Values are shown in average ± standard deviation if more than one sample was analysed (n.d., not detected). N, number of samples used in the analysis; SFA, saturated fatty acids; MUFA, monounsaturated fatty acids; PUFA, polyunsaturated fatty acids; n-3, sum on unsaturated fatty acids with the first double bond starting after the third carbon atom relative to the terminal methyl group; n-6, sum of unsaturated fatty acids with the first double bond starting after the sixth carbon atom relative to the terminal methyl group; TFA, total fatty acids (unit: mg g−1). Copepods, fish larvae and Acetes chinensis are pooled samples with more than 10 individuals, respectively, and are labelled with p.s. in the table.
For seston samples, saturated fatty acids (SFA) were the dominant fatty acids, and the concentration of polyunsaturated fatty acids (PUFA) was extremely low (Table 2). For copepods, fish larvae, A. chinensis and C. mystus, PUFA was the dominant fatty acid. The seston samples were characterized by higher C16:1n-7, C18:1n-7 and C18:2n-6 + C18:3n-3 concentrations compared with zooplankton. In the zooplankton samples, copepods contained the highest C20:1 + C22:1 (>5%) and C22:6n-3 concentrations; fish larvae and A. chinensis were characterised by high C20:4n-6 and C18:1n-9 concentrations. For C. mystus in all stations, C20:4n-6 (Arachidonic, ARA), C20:5n-3 (eicosapentaenoic, EPA) and C22:6n-3 (docosahexaenoic, DHA) were the dominant essential fatty acids, and they accounted for almost 50% of the total fatty acids. DHA was especially high and accounted for over 30% of the total fatty acids. C16:0 and C18:0 were the dominant SFA, and C16:0 accounted for over 20% of the TFA. The predominant MUFA were C16:1n-7 and C18:1 (n-9, n-7). For C. mystus at M3-DSb and M3-JT, the C20:1 + C22:1 concentration was greater than 2%. Regarding the C22:6n-3/C20:5n-3 (DHA/EPA) ratio, those of copepods, fish larvae and C. mystus were higher than 1, but those of seston and the A. chinensis samples were lower than 1.
Fatty acid composition variation of C. mystus in different migration stages
A PCA was performed based on the fatty acid compositions of C. mystus using PRIMER 5.0. The first two PCAs accounted for 63.0% of the variance (Figure 3). The first PC axis on the score plot separated the fishes by placing C. mystus from HZB1 on the negative side of PC1 and the samples from other stations on the positive side (Figure 3A, PC1 48.9% of variance explained). The second PC separated DSa from the other stations and accounted for 14.1% of the variance (Figure 3A). According to the results of the cluster analysis based on the PCA scores of the samples from various stations, the samples were separated into three groups: M1-HZB1, M2-DSa, and a group composed of M3-DSb, M3-JT, M4-HZB2, M4-CXI. ANOSIM analysis was performed on the fatty acid compositions of C. mystus samples from the three groups. The results (R = 0.683, P = 0.001) of the ANOSIM suggested that the difference among the three groups was significant.
Fig. 3. Principal component analysis (PCA) of fatty acids in Coilia mystus (PC1: 48.9%, PC2: 14.1% of the total variance): (A) PCA scores plot showing separation among prey groups; (B) PCA factor loadings showing individual fatty acids contributing to the separation among prey groups. Those fatty acids with highest factor loadings have the greatest influence on discrimination among prey items.
The PCA variable loading plot showed that the fatty acids determined the positions of C. mystus on the score plot (Figure 3B). C18:2n-6, C18:3n-3, C20:4n-6, C16:0, C17:0, C20:3n-6 and C18:1n-7 were on the negative side of the PC1 axis, indicating the contribution of benthic detritus biomarkers to the variance (Stowasser et al., Reference Stowasser, Mcaallen, Pierce, Collins, Moffat, Priede and Pond2009a, Reference Stowasser, Pond and Collinsb). However, C18:1n-9, C18:4n-3, C20:1, C22:1, C16:1n-7 and C22:6n-3 were on the positive side of the PC1 axis, indicating the contribution of pelagic zooplankton biomarkers to the variance (Kattner & Hagen, Reference Kattner, Hagen, Arts, Brett and Kainz2008). C18:1n-7, C15:0 and C17:0 were on the negative side of the PC2 axis, indicating the contribution of bacteria (Parrish et al., Reference Parrish, Abrajano, Budge, Helleur, Hudson, Pulchan, Ramos and Wangersky2000; Stowasser et al., Reference Stowasser, Mcaallen, Pierce, Collins, Moffat, Priede and Pond2009a, Reference Stowasser, Pond and Collinsb). SIMPER analysis showed that C18:1n-9, C22:6n-3, C20:5n-3, C20:4n-6, C16:0, C16:1n-7, C18:1n-7, C18:2n-6, C20:1, C22:1 and C18:0 contributed the most to the dissimilarity among the groups.
Furthermore, several specific fatty acids were chosen according to the results of the PCA and SIMPER to analyse the variations in the fatty acid composition of C. mystus during their migration (Figure 4). Significance of difference was labelled in the figure. The C20:1 + C22:1 concentration was high in C. mystus at the M3-DSb, M3-JT and M4-HZB2 samples but low in M1-HZB1, M2-DSa, and M4-CXI (Figure 4A). The change in C18:1n-7 was consistent with that of C20:4n-6, which was high in M1-HZB1 and M2-DSa, then decreased in M3-DSb, M3-JT, but increased in M4-HZB2, M4-CXI (Figure 4A). The C18:2n-6 + C18:3n-3 concentration was as high as 2% of total fatty acids in M1-HZB1, but only approximately 1% in other migration stages (Figure 4A). C15:0 + C17:0 and the C18:1n-7/n-9 ratios were slightly higher in M1-HZB1 and M2-DSa than in the other migration stages (Figure 4A, B).
Fig. 4. Station-related changes in the fatty acid biomarkers of Coilia mystus: (A) specific fatty acid concentrations; (B) C18:1n-7 and PUFA/SFA ratios; (C) PUFA ratios. Data are shown as the average ± standard deviation. ‘*’ indicates that the difference in the specific fatty acids of individuals in different stations was significant.
Relationship between the biomarkers and body length of C. mystus
For total C. mystus samples, both δ13C and δ15N were positively correlated with the body length (δ13C r = 0.648, P = 0.003, δ15N r = 0.629, P = 0.004). Linear regression results showed that the body length explained 41.5% and 40.1% variation of δ13C and δ15N in C. mystus, respectively (Figure 5). There was no significant correlation between the specific fatty acids and the body length for the C. mystus total samples (P > 0.05). If only those collected from M3-DSb were considered, δ13C was still positively correlated with the body length (r = 0.881, P = 0.004), but δ15N was not significantly correlated with the body length (P > 0.05). There also was no significant difference in the C20:1 + C22:1 and C20:4n-6 concentrations and C18:1n-7/n-9 ratio among the four groups of individuals in M3-DSb (ANOVA, P > 0.05).
Fig. 5. Relationships between stable isotopes and body length of Coilia mystus: (A) δ13C; (B) δ15N.
DISCUSSION
Potential dietary influence on the stable isotopes of C. mystus
δ13C and δ15N values are often used as indicators of the carbon source and trophic level, respectively, of organisms within the food web (Gu et al., Reference Gu, Schelske and Hoyer1996; Overman & Parrish, Reference Overman and Parrish2001; Post, Reference Post2002). In the present study, both δ13C and δ15N values of C. mystus were enriched during the migration. Generally, the changes in the δ13C and δ15N values of the fish may suggest variation in the diet source and trophic level. But influences from the factors other than the diet source on the δ13C and δ15N values should be born in mind, especially for the migration species undergoing habitat shift. First, large individuals tend to accumulate heavy isotopes during the metabolic process, so the metabolic effects associated with life stage cannot be totally ignored when performing stable isotope analysis (Malej et al., Reference Malej, Faganeli and Pezdifi1993; Overman & Parrish, Reference Overman and Parrish2001). This accumulation increases the risk of a bias in prey source identification when long term variation is studied. Correlation between stable isotopes and body length has been identified in C. mystus. Even the individuals in the same migration stage exhibited enriched δ13C with increasing body length. The body lengths of individuals in stage M4 were higher than those in stage M1. Thus, it is possible that the increases in the average δ13C and δ15N values of C. mystus during migration are partly attributed to the metabolic accumulation related to the body length. But in view of the fact that the inter-individual δ13C variation for animals having a similar food source usually does not exceed 2‰ (Bardonnet & Riera, Reference Bardonnet and Riera2005), it is very likely that the dietary influence contributed to the high increase (6.1‰) in C. mystus to a certain degree.
Second, it has been shown that both overwintering and spawning could lead to increased δ13C and δ15N values due to the reduction in lipid content caused by the usage of lipid as fuel during fasting or migration, and the increase in δ13C was normally higher than that of δ15N (Ciancio et al., Reference Ciancio, Pascual, Botto, Amaya-Santi, O'Neal, Riva Rossi and Iribarne2008). Because the lipid content was positively related to the TFA, it can be inferred that the lipid content of C. mystus increased from M1 to M4 according to the corresponding TFA content. The increase in lipid content should lead to depleted δ13C values because lipid is depleted in δ13C (Bardonnet & Riera, Reference Bardonnet and Riera2005). However, the average δ13C and δ15N values of C. mystus increased 6.1 and 8.9‰ from M1 to M4, respectively. So, influences of overwintering and spawning on the δ13C and δ15N values were not observed.
Third, human impact is a potential contributor to the high δ15N value in coastal regions. The biota in the aquatic ecosystem with high anthropogenic nutrient inputs tends to have highly elevated δ15N value (Steffy & Kilham, Reference Steffy and Kilham2004; Mercado-Silva et al., Reference Mercado-Silva, Helmus and Vander Zanden2009; Herbeck, Reference Herbeck2011). In a previous study, the average δ15N value of organisms in highly human-impacted region was 10‰ enriched than those in less impacted system in Mexico (Mercado-Silva et al., Reference Mercado-Silva, Helmus and Vander Zanden2009). Samples at Stations CXI and HZB2 were all at stage M4, the effect from the metabolic accumulation is supposed to be similar. Thus, a factor other than the physiological accumulation may lead to the significantly higher δ15N values of those in M4-CXI. In addition, the average δ15N value of seston collected in Station CXI was as high as 6.9 ± 0.8‰, even higher than that of A. chinensis collected from Hangzhou Bay. The Station CXI is near the large city of Shanghai and may be extensively influenced by anthropogenic run-off. So, it can be suggested that anthropogenic run-off may contribute to the high δ15N values in C. mystus at M4-CXI.
Although there are simultaneous influences from metabolic accumulation and anthropogenic effect, the dietary influence on the δ13C and δ15N pattern of C. mystus can be indicated. The habitat of the migration species is variable during their life stages. Considering the possible difference in the baseline of the food web, only the trophic level (TL) of C. mystus in the stage M1 was calculated with A. Chinensis collected in HZB1 as the baseline. Individuals in stage M1 were secondary consumers, roughly at the same TL as the planktivore fish Engraulis japonicus, and both preyed predominantly on zooplankton (Wang, Reference Wang2008). Even if the extraordinarily high values of M4-CXI were supposed due to the anthropogenic effect, the individuals in M3-DSb and M4-HZB2 also exhibited more enriched δ15N values than those of stage M1. The enrichment in δ15N values has been attributed to changes in prey type and the size of prey (Jennings et al., Reference Jennings, Pinnegar, Polunin and Warr2002; Xu et al., Reference Xu, Zhang and Xie2007; Stowasser et al., Reference Stowasser, Mcaallen, Pierce, Collins, Moffat, Priede and Pond2009a). So the increase in the δ15N values of C. mystus during migration may suggest that the diet compositions of C. mystus at various migration stages may be different, and this species may prey at different TLs when it is at different life stages. Previous gut content analyses of C. mystus indicate that although zooplankton account for over 60% of the C. mystus diet, items such as small shrimp (e.g. Leptochela gracilis) and small fish (e.g. goby) in higher TLs than zooplankton are also included (Zhou et al., Reference Zhou, Xu and Xu2004; Zhuang et al., Reference Zhuang, Luo, Zhang, Zhang, Liu, Feng and Hou2010). Diet shift of C. mystus was also documented in previous studies. It was determined that C. mystus smaller than 60 mm feed primarily on copepods, cladocerans and amphipods. Those individuals 60–150 mm feed predominantly on polychaetes, decapods, chaetognathans and small fish. Decapods, mysidaceans, and small fish were the dominant diet items of individuals larger than 150 mm (Yuan & Qin, Reference Yuan and Qin1984; Ni et al., Reference Ni, Wang, Jiang and Chen1999). Variation in the TL of a species has been found in many fish (Zhang, Reference Zhang2004; Xu et al., Reference Xu, Zhang and Xie2007; Stowasser et al., Reference Stowasser, Mcaallen, Pierce, Collins, Moffat, Priede and Pond2009a; Wan et al., Reference Wan, Wu, Huang, Zhang, Gao and Wang2010). For instance, the TL of the large lake anchovy (>130 mm) was 0.6 higher than that of small individuals (<130 mm). The difference between individuals >180 and <180 mm was even as high as one level for Coryphaenoides armatus (Stowasser et al., Reference Stowasser, Mcaallen, Pierce, Collins, Moffat, Priede and Pond2009a).
Potential dietary influence on fatty acid composition of C. mystus
The variation in the δ13C and δ15N patterns of C. mystus during migration suggested that their diet sources may vary during different life stages. However, due to the potential influence of metabolic accumulation and the environment change, the dietary influence on the stable isotopes requires further confirmation with fatty acid composition. The biosynthesis and storage of the fatty acids in organisms usually follow several regular patterns, making them useful tracers of diet source (Iverson, Reference Iverson, Arts, Brett and Kainz2008). Both specific fatty acid and overall fatty acid signature can be used as tracers (Budge et al., Reference Budge, Iverson and Koopman2006). The correlation between body size and fatty acid composition was not supported by the correlation analysis. Thus, it can be supposed that variations in the fatty acid composition are mainly due to the shift of diet source.
The significant difference in the fatty acid compositions of C. mystus in stage M1, stage M2 and stages M3, M4 may confirm that variation in diet composition occurred during migration. The dietary influence can also be reflected in the variation in the specific fatty acids. Generally, high C18:1n-9 content and low C18:1n-7/n-9 ratio may indicate carnivorous feeding behaviour of zooplankton (Petursdottir et al., Reference Petursdottir, Gislason, Falk-Petersen, Hop and Svavarsson2008; Brett et al., Reference Brett, Müller-Navarra, Persson, Arts, Brett and Kainz2009). In addition, a high PUFA/SFA ratio also indicates carnivorous behaviour of zooplankton. Herbivorous zooplankton exhibits a comparatively lower PUFA/SFA ratio because seston usually contains a high SFA content and is depleted in PUFA (Rossi et al., Reference Rossi, Youngbluth, Jacoby, Pagès and Garrofé2008). And there is a general trend that PUFA accumulates from low (phytoplankton and detritus) to high TLs (anchovy larvae) (Rossi et al., Reference Rossi, Sabatés, Latasa and Reyes2006). In light of the limited ability of the marine fish to biosynthesize the fatty acids de novo (Iverson, Reference Iverson, Arts, Brett and Kainz2008), it can be supposed that the fatty acid pattern of the zooplankton can be reflected in the fatty acid pattern of their predator (Rossi et al., Reference Rossi, Sabatés, Latasa and Reyes2006). Coilia mystus in stages M1 and M2 preyed on lower TL than those in other stages, as indicated by the higher C18:1n-7/n-9 ratios compared to other stages, which was consistent with the result that the δ15N values of C. mystus in stages M1 and M2 were lower than those in stages M3 and M4. Low PUFA/SFA ratios in stage M1 samples may also suggest the low carnivorous level of C. mystus in that stage. The PUFA/SFA ratios of C. mystus in stage M2 were elevated, possibly because high PUFA concentrations were needed for C. mystus to resist low temperatures in March, considering their physiological function related to the membrane fluidity at low temperatures (Hall et al., Reference Hall, Parrish and Thompson2002). Our results also showed that the C18:1n-7/n-9 ratio was a more reliable biomarker of carnivorous activity than the PUFA/SFA ratio.
More detailed information on the diet items can be suggested by several specific fatty acids. C20:1 + C22:1 has been used as biomarker of copepods in food web study (Kattner & Hagen, Reference Kattner, Hagen, Arts, Brett and Kainz2008; Rossi et al., Reference Rossi, Youngbluth, Jacoby, Pagès and Garrofé2008). High level of C20:1 + C22:1 concentration in fish generally indicates diet containing copepods (Graeve et al., Reference Graeve, Lundberg, Böer, Kattner, Hop and Petersen2008; Stowasser et al., Reference Stowasser, Mcaallen, Pierce, Collins, Moffat, Priede and Pond2009a, Reference Stowasser, Pond and Collinsb). Copepods were important diet items for stage M3 C. mystus collected from DSb and JT as indicated by the high C20:1 + C22:1 concentration in their muscle. The gut content analysis of C. mystus collected from HZB2 and CXI found that copepods and mysidaceans were the dominant diet items for these fish (Liu & Xu, Reference Liu and Xu2011). In addition to the pelagic copepods mentioned above, the contribution of benthic food items can also be confirmed by the specific fatty acids found in C. mystus. C18:2n-6 and C18:3n-3 have been used as indicators of the contribution of terrestrial organic matter because they are typical fatty acids in freshwater and terrestrial primary producers (Parrish et al., Reference Parrish, Abrajano, Budge, Helleur, Hudson, Pulchan, Ramos and Wangersky2000; Iverson, Reference Iverson, Arts, Brett and Kainz2008). However, some marine macroalgae also contain high levels of C18:2n-6 + C18:3n-3 and C20:4n-6 (Li et al., Reference Li, Fan, Han, Yan and Lou2002; Koussoroplis et al., Reference Koussoroplis, Bec, Perga, Koutrakis, Bourdier and Desvilettes2011). Whether from terrestrial or marine primary producers, the most likely mechanism for their entrance into the food web is in the form of detritus, because few organisms can graze on macrophytes directly. The presence of C18:2n-6 + C18:3n-3 in the seston samples might reflect a significant contribution of detritus in the phytoplankton fraction (Graeve et al., Reference Graeve, Kattner, Wiencke and Karsten2002). C18:1n-7 can be biosynthesized by bacteria and diatoms, and also in high concentration in some macroalgae species (Volkman et al., Reference Volkman, Barrett, Blackburn, Mansour, Sikes and Gelin1998; Li et al., Reference Li, Fan, Han, Yan and Lou2002). It has been used as a biomarker of bacteria (Alfaro, Reference Alfaro2008). C20:4n-6 is found in high concentrations in benthic organisms such as amphipods, echinoderms, crabs and deep-sea polychaetes (Copeman & Parrish, Reference Copeman and Parrish2003; Hall et al., Reference Hall, Lee and Meziane2006; Maazouzi et al., Reference Maazouzi, Masson, Izquierdo and Pihan2007; Würzberg et al., Reference Würzberg, Peters, Schüller and Brandt2011). Generally, C20:4n-6 is found in low concentrations in most marine fish, but fish that prey on exclusively benthic foods (e.g. benthic shrimps) tend to contain high C20:4n-6 concentrations (Stowasser et al., Reference Stowasser, Mcaallen, Pierce, Collins, Moffat, Priede and Pond2009a, Reference Stowasser, Pond and Collinsb, Wan et al., Reference Wan, Wu, Huang, Zhang, Gao and Wang2010; Koussoroplis et al., Reference Koussoroplis, Bec, Perga, Koutrakis, Bourdier and Desvilettes2011). Thus, C18:1n-7 and C20:4n-6 can be used as indicators of benthic food sources within the food chain (Stowasser et al., Reference Stowasser, Mcaallen, Pierce, Collins, Moffat, Priede and Pond2009a, Reference Stowasser, Pond and Collinsb). In this study, the C20:4n-6 concentration was high in fish larvae and A. chinensis. The high concentrations of C18:1n-7 and C20:4n-6 in C. mystus collected from HZB1 indicated the contribution of benthic-origin organic matter to their diet, which may be transferred through grazing on benthic organisms directly or through preying on those who consume benthic organisms (e.g. mysidaceans, fish larvae and A. chinensis). Benthic food has been found in the diet of C. mystus in previous studies (Luo et al., Reference Luo, Wei and Dou1997; Li et al., Reference Li, Zuo, Dai, Jin and Zhuang2009). According to our results, the concentrations of benthic biomarkers and pelagic biomarkers in C. mystus in different stages were changing. Therefore, it can be suggested that the ratio of pelagic to benthic food sources is variable in C. mystus during migration.
The higher C18:2n-6 + C18:3n-3, C18:1n-7, C20:4n-6 and C15:0 + C17:0 concentrations and lower C20:1 + C22:1 concentration in M1 C. mystus suggest that the detritus-derived benthic food chain contributed more to their diet than copepods. Individuals in stage M2 were also characterized by high C18:1n-7, C20:4n-6 and C15:0 + C17:0 concentrations and a lower C20:1 + C22:1 concentration. It appears that the benthic food web, especially bacteria, contributed more than the pelagic food web to the diet composition of stage M2 C. mystus. However, compared with individuals from stage M1, the C18:2n-6 + C18:3n-3 concentration was lower in stage M2 C. mystus. Using C18:2n-6 + C18:3n-3 as an indicator of the contribution of terrestrial matter, it appears that terrestrial organic matter contributed more to stage M1 C. mystus than to those in stage M2. This difference may due to the difference in the geographic locations of the stations where they were collected. The station where C. mystus in stage M1 were collected was closer to the land, where terrestrial matter was more dominant, compared to the station where C. mystus in stage M2 were collected. Conversely, stage M3 C. mystus and stage M4 individuals collected at the HZB2 station contained high concentrations of the copepod biomarker C20:1 + C22:1 and low levels of benthic biomarkers (i.e. C18:2n-6 + C18:3n-3, C18:1n-7, C15:0 + C17:0, C20:4n-6), suggesting that pelagic copepods dominated the C. mystus diet over foods from benthic sources. Compared with stage M3 C. mystus, the C20:1 + C22:1 concentration of stage M4 individuals decreased and concentrations of C18:1n-7 and C20:4n-6 increased, whereas the concentration of C15:0 + C17:0 did not change. This variation in the fatty acid biomarkers suggests that the dietary contribution of copepods decreased in stage M4 and the contribution of benthic food increased, but that from bacteria remained the same. This also indicates that the benthic food of M4 stage C. mystus was at a higher trophic level than that of stage M1 or M2 C. mystus.
Marine organisms often contain high levels of n-3 PUFA, and terrestrial organisms often contain high n-6 PUFA. Thus, the n-3/n-6 PUFA ratio can be used as an indicator of the amount of marine-derived vs terrestrial-derived food in an organism's diet (Hebert et al., Reference Hebert, Arts and Weseloh2006, Ahlgren et al., Reference Ahlgren, Vrede, Goedkoop, Arts, Brett and Kainz2009). The n-3/n-6 ratio was lower in stage M1 C. mystus and very high in stage M3 individuals. This finding is consistent with the above conclusion that terrestrial-derived organic matter contributed more to the diet of stage M1 C. mystus, and marine-derived organic matter was the dominant food source for individuals in stage M3. The n-3/n-6 ratio of stage M4 C. mystus decreased compared to M3, which indicates that the proportion of terrestrial-derived food sources in their diet increased again.
The trophic strategy of C. mystus during migration as indicated by the stable isotopes and fatty acid composition can be described as follows: C. mystus in stages M1 and M2 prey on lower TL compared with those in stages M3 and M4. When they are in stage M1, items derived from benthic and terrestrial based food webs dominate their diet. When they migrate to the Zhoushan Islands to pass the winter, benthic organisms remain their dominant food source. However, in May the contribution of pelagic copepods to the diet of stage M3 C. mystus dramatically increases, and copepods become their dominant dietary source. For C. mystus in stage M4, the contribution of copepods decreases and the contributions of benthic food sources increase again. Notably, the benthic food source for stage M4 individuals is in a higher TL than those in stages M1 and M2. Bacteria contribute more to the diet when individuals are in stages M1 and M2 than when they are in stages M3 and M4.
Seasonal and spatial variation in the diet composition of fish is a common phenomenon (Letourneur et al., Reference Letourneur, Galzin and Harmelin-Vivien1997; Mandima, Reference Mandima2000; Schafer et al., Reference Schafer, Platell, Valesini and Potter2002; Xue et al., Reference Xue, Jin, Zhang and Liang2004; Zhang et al., Reference Zhang, Jin and Dai2008; Bacha & Amara, Reference Bacha and Amara2009; Stowasser et al., Reference Stowasser, Mcaallen, Pierce, Collins, Moffat, Priede and Pond2009a; Xue et al., Reference Xue, Xu, Gao, Xu and Lin2010). Diet variation may be closely related to the seasonal or spatial availability of the dominant prey types (Mandima, Reference Mandima2000; Zhang et al., Reference Zhang, Jin and Dai2008). Zooplankton in the Changjiang Estuary exhibited seasonal and spatial variation in both abundance and species dominance (Zhu, Reference Zhu1988; Xu et al., Reference Xu, Wang, Chen and Shen1995; Xu & Shen, Reference Xu and Shen2005; Zhu et al., Reference Zhu, Liu, Zheng and Wang2011). The abundance of zooplankton was very low in March but high in May (Zhu, Reference Zhu1988; Xu & Shen, Reference Xu and Shen2005). This may explain why the diet compositions of M2 and M3 C. mystus, which were both collected from the Daishan Islands, were different. In March, when there were not sufficient copepods for food, organic matter originating from the benthic detritus food web became the main dietary source for M2 C. mystus. However, M3 individuals could prey predominantly on copepods (e.g. Calanus sinicus) because the abundance of zooplankton was very high in May. Both Stations CXI and HZB1 are located in regions where zooplankton abundance is low (Zhu, Reference Zhu1988; Xu & Shen, Reference Xu and Shen2005; Ji & Ye, Reference Ji and Ye2006; Zhu et al., Reference Zhu, Liu, Zheng and Wang2011). However, Stations DS, JT and HZB2 were close to the Zhoushan Islands, which are characterized by high zooplankton abundance and high fishery catch (Zhu, Reference Zhu1988; Xu & Shen, Reference Xu and Shen2005; Ji & Ye, Reference Ji and Ye2006; Zhu et al., Reference Zhu, Liu, Zheng and Wang2011). The high availability of zooplankton in the waters around the Zhoushan Islands may result in the high contribution of zooplankton to the diets of M3 and M4-HZB2 C. mystus. Therefore, the dietary influence on the fatty acid composition of C. mystus may be related to the food availability in these different regions and seasons.
CONCLUSION
Though it may be covered by other factors, such as metabolism accumulation and environment shift, the influence of dietary source on stable isotopes of C. mystus during migration can be confirmed by the fatty acid composition information. According to the variation in the biomarkers, individuals at different migration stages seemed to have distinct diet sources. The pelagic (e.g. copepods) and benthic (e.g. mysidaceans) food sources alternatively dominated the diet items of C. mystus during migration, as suggested by the specific fatty acids. This study demonstrated the usefulness and feasibility of stable isotopes and fatty acid composition in determining the feeding habits of migratory fish in an estuary. Fatty acid composition was more sensitive in reflecting the dietary variation and can provide more detailed dietary information than bulk stable carbon and nitrogen isotopes.
ACKNOWLEDGEMENTS
We thank Y. Ni for identifying Coilia mystus. We are grateful to L.Y. Yang, H. Chen, J.J. Chen, S.H. Liu, A.L. Shen and J. Zhou for helping us to collect the samples and to G.S. Zhang for analysing the stable isotopic ratios of carbon and nitrogen.
FINANCIAL SUPPORT
This work was supported by the National Key Program for Basic Research (973 program, Grant No. 2011CB409801, 2011CB409802), the Natural Science Foundation of China (41276081), and the International Cooperation Project (2010DFA24590).
