Understanding and predicting variability in the stable isotope ratios of nitrogen and carbon (δ15N and δ13C, respectively) of small pelagic fish is crucial to enable isotopic studies of a variety of marine predators that feed on them. However, because the isotope ratios reflect plastic feeding habits and fish migration in addition to baseline variation, their predictions require a mechanistic understanding of how each factor contributes. Here, we investigated the habitat-wide variability of δ¹⁵N and δ¹³C of the Japanese sardine Sardinops melanostictus in the western North Pacific and its marginal seas (the East China Sea and the Sea of Japan). By combining this with the archived particulate organic matter (POM) dataset as a baseline, we aimed to understand how ecological processes and baseline fluctuations affect isotope ratios of the sardine. Both δ¹⁵N and δ¹³C of sardine showed significant geographical and seasonal trends, with higher values in southern nearshore areas, including the Seto Inland Sea, intermediate values in marginal seas and lower values in Pacific offshore areas. As the variations were largely consistent with the geographic and temporally integrated seasonal trends of isotope ratios of POM, respectively, the baseline variations are the main determinant of sardine isotope composition. The trophic levels of sardine are therefore not significantly different between regions, with possible minor increases in the southern nearshore area. Adults showed less geographic variation than larvae and juveniles, likely due to slower turnover periods and wider migration ranges. Although larval and juvenile isotope ratios in marginal seas mostly reflected the local baseline, those in the Pacific offshore often reflected the baseline in the neighbouring southern region, suggesting contrasting juvenile movements between regions. Our results suggest that the δ¹⁵N and δ¹³C of Japanese sardine strongly reflect baseline variations, but can also be influenced by life-stage- and region-dependent fish movements, thereby demonstrating both the possibility and difficulty of mechanistically modelling the isoscapes of lower trophic level species.

Tissues were freeze-dried and ground into powder. Lipids were extracted from all samples using a 2:1 chloroform:methanol solution, freeze-dried again, and 800 μg of a subsample was extracted for isotope analysis. The δ¹⁵N and δ¹³C values of the samples were determined at Fisheries Resources Institute (Yokohama, Japan) or GeoScience Laboratory (Nagoya, Japan) using a continuous-flow stable isotope ratio mass spectrometer (IsoPrime100, Elementar, Stockton, UK; Delta Plus Advantage, Thermo Fisher Scientific, Waltham, Massachusetts, USA) coupled to an elemental analyser (vario MICRO cube, Elementar; FLASH2000, Thermo Fisher Scientific, Yokohama Japan). The δ¹⁵N and δ¹³C values were reported in δ-notation against the atmospheric N2 standard and the VPDB reference standard (Vienna Pee Dee Belemnite), respectively, and given as a ‰ value. Analytical accuracies were ± 0.2‰ for δ¹⁵N and δ¹³C in both laboratories. The agreement of the reported values between the two laboratories was tested using a blind standard (powder of fish eye lens) where the differences in the reported values for both δ¹⁵N and δ¹³C were less than the analytical precisions.