Understanding recruitment sources is essential for stock assessments of marine fish populations. In 2014 and 2019, schools of sardine in the Sea of Japan and East China Sea (SJ-ECS), which come to spawn in Japanese coastal areas every spring, were shockingly sparse. Abundances of eggs and juveniles also declined abruptly, indicating a sharp decline in reproduction in the SJ-ECS. However, in spring of 2015 and 2020, age-1 fish appeared as usual in the coastal areas, challenging the current assumption that sardine in the system is a self-recruiting subpopulation. To test the self-recruiting hypothesis, we analysed the stable oxygen and carbon isotopes (δ¹⁸O, δ¹³C) for otolith areas formed during the first spring and summer in otoliths of age-0 and age-1 sardines in 2010 and 2013–2015 year-classes captured in the SJ-ECS, as indices of temperature and metabolic trajectories. Age-0 sardines generally showed a significant decrease in otolith δ¹⁸O from spring to summer, reflecting seasonal warming in the SJ-ECS. However, the majority of age-1 captured in spring 2011, 2015, and 2016 showed non-decreasing profiles. The δ¹⁸O for summer thus revealed different migration groups: the “locals" growing up off the Japanese coast and the migrating “nonlocals", the former not being the main source of recruitment, contrary to previous assumptions. The isotope values of the “nonlocals” overlapped with those of age-0 captured in the subarctic Pacific, suggesting that the “nonlocals” may be migrants from the Pacific, or an unobserved northward migration group in the SJ-ECS. Only in 2014 did the majority of age-1 consist of the “locals”, suggesting that the abrupt decline in catches was caused by the absence of the “nonlocals” and accompanying adults. Our results highlight the considerable uncertainty in the population structure assumed in current stock assessment models for Japanese sardine, thereby requiring focused investigations on their migrations for sustainable fisheries.

Collected fish were frozen after landing or on board at -20 °C, and thawed at a laboratory. After measurements of length and weight, the sagittal otoliths were extracted. The otoliths were cleaned using a thin brush and rinsed with fresh water. These otolith samples were processed in two different protocols, namely low-resolution analysis similar to Sakamoto et al. (2020) and high-resolution analysis similar to Aono et al. (2023). The majority of the otoliths were analysed using a low-resolution protocol. In addition to these, high-resolution data from age-0 fish captured in 2015 around Kyushu and the Noto Peninsula (Aono et al., 2023) were also included in analyses to allow a more comprehensive comparison.

For low-resolution analysis, otoliths were embedded in Petropoxy 154 (Burnham Petrographics LLC) resin and kept at 80 °C for 12 h to cure. Otoliths were ground and polished until the core was revealed using sandpaper and alumina suspension (BAIKOWSKI International Corporation). Using an otolith measurement system (RATOC System Engineering Co. Ltd.), daily increments were examined along the axis in the post-rostrum from the core as far as possible. Daily increments could be identified until the edge for most otoliths of age-0 fish but not for those of age-1 fish, likely because the otolith growths became significantly slower during winter. The otolith portion formed during 0–60 dph, representing spring, was identified and milled out using a high-precision micro-milling system, Geomill 326 (Izumo-web, Japan). For approximately half of the samples, the portion of the otolith formed during 106–120 dph was additionally milled out to represent values for summer. The milling depth for the spring and summer portions was 50 and 100 μm, respectively. The δ¹⁸O and δ¹³C of powdered samples were measured using an isotope ratio mass spectrometer (Delta V plus, Thermo Fisher Scientific) equipped with an automated carbonate reaction device (GasBench II, Thermo Fisher Scientific), and installed at the Atmosphere and Ocean Research Institute, the University of Tokyo, Chiba. Detailed analytical conditions have been reported elsewhere (e.g., Shirai et al. 2018), with minor modifications where 4.5-ml glass vials were used (Breitenbach & Bernasconi, 2011). All isotope values are reported using delta notation relative to the Pee Dee Belemnite. No correction was applied for the acid fractionation factor between calcite and aragonite [phosphoric acid–calcium carbonate reaction temperature 72 °C (Kim et al. 2007)]. Analytical precisions of δ¹⁸O and δ¹³C for international standards (NBS-19) were 0.06–0.13 (1σ) and 0.05–0.11 ‰, respectively. Because the difference between the acid fractionation factor of calcite and aragonite is temperature dependent (Kim et al., 2007), 0.09 ‰ was subtracted from the δ¹⁸O value to allow comparison with data analysed using another analysing system operating at 25 °C.

High-resolution analysis was performed for 17 otoliths of age-0 and age-1 fish captured in 2015. Otoliths were embedded in epoxy resin (p-resin, Nichika Inc.) and kept in a dryer at room temperature for more than a day to cure. Otoliths were ground and polished until the core was revealed using sandpaper and alumina suspension (BAIKOWSKI International Corporation). Unfortunately, microstructure analysis was not performed for some of these otoliths. The otolith portions that were formed every 5–30 days or 30–160 μm were milled sequentially from the edge to the core using GEOMILL326. The δ¹⁸O of collected otolith powders was determined by a customized continuous-flow isotope ratio mass spectrometry system (MICAL3c with IsoPrime100) at the National Institute of Technology, Ibaraki College, Hitachinaka, Japan (Ishimura et al., 2004; 2008; Nishida & Ishimura, 2017). Otolith powders were reacted with phosphoric acid at 25 °C, and the evolved CO₂ was purified and introduced into the mass spectrometry system. δ¹⁸O values of each sample were reported in standard δ notation (‰) relative to the Vienna Pee Dee Belemnite (VPDB) standard. Analytical precisions were ± 0.1 ‰ for both δ¹⁸O and δ¹³C. For otoliths for which microstructure analysis was not performed, the corresponding age range for each milling area was later estimated from the distance from the core using the mean relationship between otolith radius and age of other fish captured in the same year, season, and region. For comparison with the low-resolution data, the high-resolution data and data from Aono et al. (in revision) needed to be rescaled to spring (0–60 dph) and summer (106–120 dph) resolution. Therefore, δ¹⁸O and δ¹³C data for which the median of the corresponding age range falls in 0–60 and 106–120 dph were averaged, linearly weighted by the width of the milling area.