Stock discrimination and vertical migration of two gadoid species based on otolith stable isotope analysis

Abstract

Pacific cod (Gadus macrocephalus) and walleye pollock (Gadus chalcogrammus) are commercially important species in bottom trawl fisheries in Japanese waters. However, their first-year life history processes and whether current management units align with the scale of their biological processes remain unclear. To investigate this, we conducted otolith macrostructural analysis on age-1 Pacific cod and walleye pollock collected from the Pacific Ocean off Hokkaido, Aomori Prefecture, and Sendai Bay (Honshu). Stable isotope analysis (δ¹³C, δ¹⁸O) was performed on both the inside and outside of the otolith check using isotope ratio mass spectrometry (IRMS) and secondary ion mass spectrometry (SIMS). The δ¹⁸O shift across the check, combined with higher δ¹³C and δ¹⁸O values on the outside of the check compared to the inside, suggests that both species undergo a habitat transition during their first-year settlement. Stock discrimination was more successful for individuals from Hokkaido and Sendai Bay than for those from Aomori Prefecture. When isotope data from the two southern regions (Aomori Prefecture and Sendai Bay) were combined into a single Honshu stock, the overall classification success increased from 71 to 76%, with Cohen’s Kappa (κ) rising significantly from 0.31 to 0.60. These results indicate that managing stocks based on geographic distribution is both scientifically valid and practical though inter-stock admixture should be taken into account. Overall, otolith stable isotope analysis proves to be a valuable tool for identifying key life history events, stock structure, and migration patterns in these gadoid species.

Supplementary Information

The online version contains supplementary material available at 10.1007/s42995-025-00344-3.

Introduction

A stock is defined as an intra-specific, semi-discrete group of randomly mated individuals that maintain temporal or spatial integrity and possess characteristics of interest to managers (Ihssen et al. 1981). Understanding fishery stock structure is essential for rational and sustainable management (Secor 2014). While stocks based on spatial distribution offer management conveniences, effective sustainable fisheries management requires that interventions align with the scale of biological processes (Kerr et al. 2017; Reiss et al. 2009). Species from different sites can be categorized into geographically distinct groups based on genetic variation (Sakuma et al. 2019). However, even though the ranges of some marine species may appear continuous, they can exhibit locally distinct genetic populations (Riccioni et al. 2010). Therefore, fine-scale stock identification—rather than relying solely on spatial distribution—should be prioritized to define management units (MUs) effectively.

Pacific cod (Gadus macrocephalus) and walleye pollock (Gadus chalcogrammus) are among the most commercially important species targeted in northern Japan’s waters (Narimatsu et al. 2010; Tanaka et al. 2017). The management of Pacific cod in Japan centers on three primary stocks: the Sea of Japan stock (in the northern half of the Sea of Japan along Honshu Island), the Hokkaido stock (surrounding Hokkaido Island), and the Pacific stock (Sakuma et al. 2019; Fig. 1). For walleye pollock, there are four regional stocks: the northern Sea of Japan stock, the southern Okhotsk Sea stock, the Nemuro Strait stock, and the Japanese Pacific stock (JPS; Tsuji 1989). This study focuses on the Pacific Ocean side of Japan, particularly the coastal waters off Hokkaido and Honshu islands. While these regions host distinct stocks of Pacific cod, potential migration between them remains unclear (Suda et al. 2017). Notably, the detection of radiocesium in fish captured near Hokkaido implies possible movement through waters adjacent to the Fukushima Dai-ichi nuclear power plant on Honshu Island (Nakamura et al. 2018). In contrast, the JPS of walleye pollock represents the largest and most commercially critical fishery resource in Japan (Honda et al. 2004; Tsuji 1989). Despite its ecological and commercial significance, key aspects of its population connectivity and stock delineation in this region remain unresolved.

Fig. 1.

Sampling locations for Pacific cod and walleye pollock in the Pacific Ocean off Hokkaido (A), off Aomori Prefecture (B), and from Sendai Bay (C)

Pacific cod spawn on sandy mud bottoms at depths of 20−100 m from December to late March off the coasts of Hokkaido and Honshu (Fukuda 1985; Sakurai and Hattori 1996; Yoseda 1992). After hatching, the larvae swim to the surface, where they metamorphose into juveniles when they reach a total length of 25 mm (DiMaria et al. 2010; Takatsu et al. 1995). The juveniles then settle in demersal areas at depths of 40–120 m by the age of three or four months (Morioka and Kuwada 2002; Narimatsu et al. 2015b). In summer, Pacific cod migrate to the upper continental slope (200–400 m), where they are one of the most important species in this ecosystem (Kitagawa 2002; Narimatsu et al. 2015b).

From December to March, the Pacific stock of walleye pollock spawns near the mouth of Funka Bay in northern Japan (Maeda 1986). The larvae and juveniles primarily inhabit the surface and subsurface waters of the bay from January to June (Honda et al. 2004; Nakatani 1988). By the end of summer, juvenile pollock (~ 7 cm in length) move to the bottom of the shelf outside the bay (200–350 m depth), where they join many adults (Maeda et al. 1983; Ueno et al. 1975). While Pacific cod and walleye pollock share nearly identical early life history culminating in the critical settlement process, comprehensive investigations of this pivotal ecological process remain surprisingly limited.

Otoliths in the inner ears of teleost fish are formed through the deposition of calcium carbonate onto a protein matrix (Hüssy et al. 2004; Popper et al. 2005). These acellular, metabolically inert structures serve as natural records of the environmental conditions experienced by the fish. As such, otoliths can be used to reveal stock structure and track individual movements (Campana 1999; Campana and Neilson 1985; Elsdon et al. 2008). The occurrence and visibility of different zones within otoliths are thought to be influenced by varying proportions of proteins (Degens et al. 1969; Panfili et al. 2002). The resulting variation in the width and density of daily rings within otoliths provides a record of physiological or environmental stress events, such as metamorphosis, settlement, and temperature changes (Campana 1990). The fractionation of oxygen stable isotopes (δ¹⁸O) in fish otoliths, as well as in other marine biogenic calcareous materials, is primarily temperature-dependent (Campana 1999; Darnaude et al. 2014; Thorrold et al. 1997). In contrast, the deposition of carbon stable isotopes (δ¹³C) is partly regulated by metabolism, with carbon being derived from both seawater bicarbonate and CO₂ produced during respiration (Chung et al. 2019; Kalish 1991). Therefore, coupled with traditional otolith macrostructure analysis, dual-isotope profiling (δ¹⁸O/δ¹³C) enables precise reconstruction of ontogenetic shifts and migratory transitions in marine species.

In this study, we employed a combination of conventional micromilling with isotope ratio mass spectrometry (IRMS) and high-resolution analysis of discrete otolith spots using secondary ion mass spectrometry (SIMS). The objectives of integrating otolith macrostructural data in the analysis of age-1 Pacific cod and walleye pollock from the Pacific coasts of Hokkaido and Honshu islands were: (1) to compare stable isotope differences among groups and assess their diagnostic utility for stock identification, and (2) to investigate whether variations in carbon and oxygen stable isotopes can reflect key life history events during fish ontogeny.

Materials and methods

Fish collection and otolith preparation

Age-1 Pacific cod and walleye pollock were collected from the Pacific Ocean off Aomori Prefecture and from Sendai Bay in April and June 2022, respectively. Additionally, age-1 Pacific cod were also collected off the coast of Hokkaido in June 2022 (Fig. 1). Standard or fork lengths (measured to the nearest 0.1 cm) were recorded on board the research vessel (Supplementary Table S1). Both sagittal otoliths were removed from each fish, cleaned, and stored dry in labeled tubes (Supplementary Fig. S1). One randomly selected otolith from each pair was placed in a mold and embedded in polyester resin (#105; Struers AS, Tokyo, Japan) for isotope ratio mass spectrometry (IRMS) analysis. For secondary ion mass spectrometry (SIMS) analysis, one otolith from each of three Pacific cod (collected from coastal waters off Hokkaido, Aomori, and Sendai Bay, respectively) and one walleye pollock (collected from Sendai Bay) were embedded in epoxy resin (Struers AS, Tokyo, Japan). Given the high cost of SIMS, the four samples with the best quality of otolith microstructural preparation were selected for analysis. The otoliths were then cut, ground, and polished following the procedures described by Wu et al. (2023). Images were captured at 100 × magnification under reflected light using a Leica DM2500 microscope (Leica, Germany) equipped with a Radeon R5 430 imaging system.

Isotope ratio mass spectrometry

A distinct check was observed on the otoliths of both Pacific cod and walleye pollock, typically appearing as a shape that generally follows the outline of the otolith (Fig. 2A, B). This check is a structural feature in otoliths, typically associated with settlement-related changes in feeding behavior and environmental conditions (Casas and Piñeiro 2000). To ensure accurate identification, only age-1 individuals (based on standard length) were selected, and the distance from the primordium was used to confirm that the first prominent opaque zone observed in the otolith corresponds to the settlement-related check rather than the first annual ring (Wu et al. 2021, 2023). Two samples were selected from each otolith for micromilling: one from inside the check, including the core, and another from outside the check, including the otolith edge. For each sample, approximately 0.6 mg of powder was collected using a 400-μm diameter carbide bit (Minitor, Japan) and transferred to a glass vial. To prevent cross-contamination between samples, the drill bit, microspatula, and other tools were cleaned with 100% ethanol. The powdered material was then analyzed for δ¹⁸O and δ¹³C using a dual-inlet isotope ratio mass spectrometer (Delta V Advantage; Thermo Fisher Scientific) coupled with an automated carbonate reaction device (GasBench II; Thermo Fisher Scientific) at the MS Laboratory of Tohoku University (Sendai, Japan). The atmosphere in the glass vial was replaced with helium, and the powdered material was reacted with dehydrated phosphoric acid at 70 °C under vacuum. The CO₂ liberated during the reaction was analyzed for stable isotope compositions. All isotope values were reported in delta notation relative to Vienna Pee Dee Belemnite (VPDB). The calcium carbonate standard (δ¹⁸O_V-PDB =  − 14.2‰; δ¹³C_V-PDB =  − 40.7‰) used to monitor machine drift was run before sampling, between sample blocks, and after each set of analyses each day.

Fig. 2.

Schematic diagram showing the areas inside and outside the check on the otolith sampled for isotope ratio mass spectrometry analysis for A Pacific cod and B walleye pollock. The yellow and white arrows indicate the check and the annual ring, respectively. C Polished sagittal transverse section of a Pacific cod otolith following secondary ion mass spectrometry analysis along the anti-sulcus, from the core to the edge. Inset: gold-coated indium mount containing two otolith thin sections and two pieces of UWC-3 calcite standard

Secondary ion mass spectrometry

High-resolution oxygen stable isotope (δ¹⁸O) analysis requires meticulous sample preparation to ensure precision. To achieve a smoother and flatter surface, the four samples used for SIMS analysis were polished with 9- and 3-μm lapping films, as described by Hane et al. (2020, 2022). A mirror-polished surface was obtained through a final polish using 0.3-μm alumina suspension (Baikalox 0.3 CR, Baikowski). Pre-polished UWC-3 calcite standards were placed at the center of an indium mount with a 1 cm diameter, and the two otoliths, with excess resin removed, were positioned at the periphery of the calcite standards (Fig. 2C). A constant force of 0.2–0.3 kN was applied to ensure the sample was fully embedded in the indium mount. The samples were then subjected to a force of 10 kN for 2–3 h to achieve flatness and remove any excess indium. Afterward, the samples were re-polished with 0.3-μm alumina suspension, cleaned using an ultrasonic cleaner, dried in a vacuum oven at 40 °C for two hours, and sputter-coated with ~ 60 nm of gold. The oxygen stable isotope content from the otolith core to edge along the anti-sulcus axis (Fig. 2C) was analyzed using a CAMECA IMS-1280 large-radius, multi-collector ion microprobe (SIMS) at the Kochi Institute for Core Sample Research, Japan Agency for Marine-Earth Science and Technology. All spots were spaced 50 μm apart, with the first seven spots at 50 μm intervals, followed by 100 μm spacing, resulting in a total of 17–25 spots, depending on the sample diameter. Each spot had a pit depth of approximately 1 μm after being sputtered with a ¹³³Cs + primary ion beam (20 kV, 1.9 − 2.1 nA) focused to a 10-μm diameter. Three Faraday cup detectors simultaneously measured the secondary ions (¹⁶O⁻, ¹⁸O⁻, and ¹⁶OH⁻) during acceleration at 10 kV. The UWC-3 calcite standard was analyzed after every 10–15 sample measurements to correct for mass fractionation changes in the instrument. Isotope values, reported in delta notation relative to the Vienna Pee Dee Belemnite (VPDB) reference, were used for subsequent analyses.

Statistical analysis

Statistical analysis was performed using R 4.2.1 software (R Core Team 2022). Isotope data obtained from the IRMS analysis were tested for normality and homogeneity of variance using the Shapiro–Wilk and Levene’s tests, respectively. As not all data met parametric assumptions, a permutational multivariate analysis of variance (PERMANOVA) was used to assess capture region and otolith position (inside and outside the check) differences in carbon and oxygen stable isotopes for both Pacific cod and walleye pollock (Anderson 2001; Artetxe-Arrate et al. 2021b). In the full factorial model, capture region and otolith position were treated as fixed factors. Although age-1 individuals were analyzed in this study, whole otolith weight was included as a covariate to minimize variability (Bastow et al. 2002; Currey et al. 2014). The resemblance matrix was calculated using Euclidean distance dissimilarities, as this metric is appropriate for continuous and commensurate variables, such as δ¹³C and δ¹⁸O values. Unlike Bray–Curtis or Jaccard distances, which are more suitable for species abundance or presence/absence data, Euclidean distance provides a direct and interpretable measure of dissimilarity in stable isotope ratios (Legendre and Legendre 1998). A total of 999 unrestricted permutations were performed using the adonis2 function from the “vegan” package. Post hoc comparisons were conducted to identify pairwise differences in element ratios among regions using the lincon function in the “WRS2” package. Modified P values, adjusted using the Benjamini–Hochberg correction, were used to assess statistical significance (Benjamini and Hochberg 1995).

Random forest (RF) classification has been shown to outperform conventional algorithms when applied to otolith chemistry data, as it does not rely on assumptions of normality or within-group homogeneity (Mercier et al. 2011). Consequently, we employed RF to evaluate the accuracy of carbon and oxygen stable isotopes for stock discrimination. In this approach, 500 binary classification trees were generated. The baseline data were randomly split into training and testing sets, with 70% assigned to the training set and the remaining 30% to the testing set. This process was repeated 1000 times (Artetxe-Arrate et al. 2021a). Classification success, defined as the correct assignment of fish to their capture regions, was calculated, and mean values were derived for each iteration. The performance of RF in distinguishing between groups was assessed using the weighted Cohen’s Kappa (κ), which accounts for the possibility of agreement occurring by chance (Titus et al. 1984). Two models were developed for each species, incorporating combinations of carbon and oxygen stable isotopes from both inside and outside the check.

To quantify changes across life history stages, we employed sequential t-test analysis of regime shifts (STARS) on oxygen isotope values obtained through SIMS analysis. STARS is a method designed to detect statistically significant shifts in a sequentially ordered data series, identifying points where the mean of the series changes markedly (Rodionov 2004). The STARS algorithm is implemented as an add-in for Microsoft Excel, available at https://www.beringclimate.noaa.gov/regimes/index.html. Prior to using the STARS algorithm, a “pre-whitening” procedure was applied to remove the effect of autocorrelation. All analyses used a 95% confidence interval, with cut-off lengths for proposed regimes (L) and Huber weight parameters (H) set to four and one, respectively.

Results

Spatial variability

For Pacific cod, PERMANOVA indicated significant effects of region for δ¹³C and a significant effect of otolith position for both δ¹³C and δ¹⁸O, but no significant region × position interaction (Table 1). Pairwise post hoc comparisons further showed that δ¹³C values differed among regions in an additive manner across otolith positions. Specifically, samples from Hokkaido had higher δ¹³C than those from Aomori and Sendai Bay inside the check, and δ¹³C values outside the check also differed among all three regions, showing a general decrease from north to south (Fig. 3). For δ¹⁸O, differences were primarily associated with otolith position. Pairwise comparisons indicated that Aomori samples had lower δ¹⁸O than the other regions inside the check, while Hokkaido samples had higher δ¹⁸O than the other regions outside the check, although the overall regional effect was not significant in the PERMANOVA.

Table 1.

PERMANOVA analysis of the effects of region and otolith position on δ¹³C and δ¹⁸O values in Pacific cod and walleye pollock otoliths

Species	Isotope	Source	df	F	P value
Pacific cod	δ13C	Region	2	3.59	0.030*
		Position	1	389.14	< 0.001***
		Region × Position	2	1.35	0.269
	δ18O	Region	2	0.00	0.996
		Position	1	114.45	< 0.001***
		Region × Position	2	0.05	0.944
Walleye pollock	δ13C	Region	1	4.49	0.032*
		Position	1	245.87	< 0.001***
		Region × Position	1	0.25	0.620
	δ18O	Region	1	22.60	< 0.001***
		Position	1	291.81	< 0.001***
		Region × Position	1	0.14	0.703

Significant effects were observed: *P < 0.05; ***P < 0.001

Fig. 3.

Composition of δ¹³C and δ¹⁸O inside (A, C, E, G) and outside (B, D, F, H) the check for age-1 Pacific cod (A, B, C, D) and walleye pollock (E, F, G, H) from different regions. Letters indicate significant differences among stock means. The 75th and 25th percentiles correspond to the upper and lower edges of the boxes, respectively, while the bars represent the 10th (lower) and 90th (upper) percentiles

For walleye pollock, PERMANOVA indicated significant effects of both otolith position and region on δ¹³C and δ¹⁸O (Table 1). However, pairwise post hoc comparisons showed no significant differences in δ¹³C among regions at either inside or outside positions, indicating that the overall regional effect detected by PERMANOVA does not correspond to significant differences at the individual position level. For δ¹⁸O, values in walleye pollock from Sendai Bay were significantly higher than those from Aomori Prefecture, a pattern consistent both inside and outside the check (Fig. 3).

Variability in δ¹⁸O and δ¹³C values in otoliths

The δ¹⁸O and δ¹³C values obtained from IRMS analysis for both Pacific cod and walleye pollock were consistently higher outside the check than inside (P < 0.001), regardless of the region from which the samples were collected (Fig. 3). No interaction between region and otolith position was observed (Table 1). For both species, we plotted the otolith δ¹⁸O content against the corresponding temperature variation curves, based on the relationship established by Kastelle et al. (2017; 2022). SIMS analysis data show that for Pacific cod, δ¹⁸O values are consistently higher during the later phase of the life history compared to the earlier phase, with the shift corresponding exactly to the check on the otolith (Fig. 4). The mean differences in δ¹⁸O values before and after the shift are 0.56‰ (Fig. 4A), 0.94‰ (Fig. 4B), and 0.67‰ (Fig. 4C) for three specimens. While one sample shows fluctuating δ¹⁸O values (Fig. 4B), the other two exhibit a sharp increase in δ¹⁸O, with changes of 1.73‰ (Fig. 4A) and 1.16‰ (Fig. 4C), respectively. This shift corresponds with a decrease in habitat water temperature from > 7 °C to 2–3°C, with absolute values of 8.2, 4.7 and 5.5 °C. Additionally, two Pacific cod specimens show higher δ¹⁸O values in the otolith primordium than in adjacent regions (Fig. 4A, C). For walleye pollock, the δ¹⁸O values show a sharp increase from − 0.44 to 1.44‰ at the check, after which the value stabilized around 1.5‰ (Fig. 4D). The projected water temperature dropped from 10.4 to 0.5 °C and subsequently fluctuated around 0.2 °C.

Fig. 4.

Transverse sections of otoliths from age-1 Pacific cod (A: Pacific Ocean off Sendai Bay; B: Pacific Ocean off Hokkaido; C: Pacific Ocean off Aomori Prefecture) and walleye pollock (D: Pacific Ocean off Sendai Bay), with superimposed graphs showing variations in δ¹⁸O (red) from the core to the anti-sulcus edge (measured using SIMS) and temperature (yellow) derived from Kastelle et al. (2022; 2017). The black lines indicate the ion microprobe trajectory, and the blue shading highlights the location of the δ¹⁸O shift, identified through sequential t-test analysis of regime shifts (cut-off length = 4, Huber = 1, P = 0.05), corresponding to the check (white arrow)

Classification success of δ¹³C and δ¹⁸O

For Pacific cod, the classification success based on carbon and oxygen stable isotopes inside the check was the highest for samples from Hokkaido (70%) and the lowest for those from Sendai Bay (Table 2). The classification success rate for δ¹³C and δ¹⁸O outside the check was over 80% for samples from Hokkaido and Sendai Bay, but only 46% for samples from the Pacific Ocean off Aomori. Overall, the classification success was higher outside the check (71%, κ = 0.31) than inside (56%, κ = 0.23).

Table 2.

Random Forest discriminant function analysis of classification success (%) in assigning age-1 Pacific cod and walleye pollock to their respective source regions in the Pacific Ocean

Species	Position	Region	Classification success
Pacific cod	Inside the check	Pacific Ocean off Hokkaido	70
		Pacific Ocean off Aomori	46
		Sendai Bay	0
		Overall	56
		κ	0.23
	Outside the check	Pacific Ocean off Hokkaido	82
		Pacific Ocean off Aomori	46
		Sendai Bay	80
		Overall	71
		κ	0.31
Walleye pollock	Inside the check	Pacific Ocean off Aomori	27
		Sendai Bay	76
		Overall	59
		κ	0.55
	Outside the check	Pacific Ocean off Aomori	36
		Sendai Bay	86
		Overall	69
		κ	0.41

Bold values indicate overall classification success and the corresponding Cohen’s κ statistics derived from the random forest models

For walleye pollock, the classification success was much higher for samples from Sendai Bay (76 and 86%) than for those from the coast of Aomori Prefecture (27 and 36%), regardless of whether the isotopes were inside or outside the check (Table 2). Using δ¹³C and δ¹⁸O values from outside the check increased the classification success from 59 to 69% for both regions compared to the data from inside the check. However, the κ values showed a slight decrease from 0.55 to 0.41.

Discussion

Ontogenetic vertical migration

The settlement process is a common trait among many species of the Gadidae family (Supplementary Table S2). Similarly, the transition of juvenile Pacific cod and walleye pollock from pelagic to benthic habitats has also been confirmed, but accurate descriptions of the relationship between settlement and the presence of the check on otoliths remain scarce, with some being speculative and unverified (Li et al. 2021; Wu et al. 2021). The settlement check has been documented in Atlantic cod (Rehberg-Haas et al. 2012), and the presence of similar non-annual translucent zones in Pacific cod has been proposed by Kastelle et al. (2017). Step changes in the Sr:Ca ratio, observed ~ 480 μm from the core, and the corresponding check in the otoliths of Pacific cod from the Yellow Sea, may indicate settlement at 80–120 days (Wu et al. 2021). Additionally, increased water temperatures and reduced prey abundance during summer lead juvenile fish to move to deeper waters (Takatsu et al. 2001), which may contribute to the formation of an additional translucent zone, referred to as the “summer ring” in otoliths (McQueen et al. 2019; Weidman and Millner 2000).

Otolith δ¹⁸O is incorporated into aragonite at or near equilibrium with environmental δ¹⁸O, with temperature-dependent fractionation (Campana 1999; Thorrold et al. 1997). The isotope composition of seawater is influenced by evaporation and is largely regulated by salinity (LeGrande and Schmidt 2006). Consequently, both temperature and salinity of the water column control the oxygen stable isotope content in fish otoliths. Previous studies have demonstrated negative correlations between otolith oxygen stable isotope values and temperature (Darnaude et al. 2014; Kim and O’Neil 1997; Martino et al. 2020; Thorrold et al. 1997), as well as positive correlations with salinity (Campana 1999; Kalish 1991), including in Atlantic cod (Høie et al. 2004b; Jones and Campana 2009; Schade et al. 2019; Stanley et al. 2015) and Pacific cod (Kastelle et al. 2017). We hypothesize that the higher carbon and oxygen stable isotope values found outside the check, compared to those inside, may indicate the movement of Pacific cod and walleye pollock to cooler or more saline waters as they mature. Although bottom salinity was higher than surface salinity (Fig. 5A), salinity in offshore water masses tends to remain relatively constant over small horizontal distances (Wright 2000). Given that these fish are winter-hatched and subject to increasing water temperatures as they grow, we infer that changes in water depth correspond with shifts in oxygen stable isotope values. The difference in mean oxygen stable isotope values inside and outside the check for Pacific cod was 0.46, 0.48 and 0.43‰ for samples from the Pacific Ocean off Hokkaido, Aomori Prefecture, and Sendai Bay, respectively. Using empirical fractionation equations relating aragonite δ¹⁸O to seawater temperature, as established by Kastelle et al. (2017) and Helser et al. (2018a) for Pacific cod in the North Pacific (with slopes of − 0.21 and − 0.25, respectively), the temperature difference between the water layers inhabited by the fish before and after settlement indicated by these isotope values is approximately 2 °C.

Fig. 5.

Differences in bottom and surface salinity (A), and surface and bottom temperature (B) in the Pacific Ocean off Hokkaido (0 m and 250 m), off Aomori Prefecture (0 m and 150 m), and in Sendai Bay (0 m and 100 m). Gray shading indicates periods of habitat depth transition for Pacific cod (April to May) and walleye pollock (May to June). Salinity and temperature data were derived from Japan Oceanographic Data Center https://www.jodc.go.jp/jodcweb/JDOSS/index_j.html

For walleye pollock, the oxygen stable isotope difference between the inside and outside of the check was 0.74 and 0.71‰ for samples from the Pacific Ocean off Aomori Prefecture and Sendai Bay, respectively. Based on the slope of − 0.19 derived from controlled experiments (Kastelle et al. 2022), the temperature difference across the check for walleye pollock is estimated at ~ 4 °C. These inferred temperature differences align closely with surface-to-depth temperature gradients for the regions studied (Fig. 5B) and are consistent with the timing of settlement for Pacific cod (April to May) and walleye pollock (May to June; Funamoto et al. 2014; Morioka and Kuwada 2002). Although oxygen stable isotope differences measured using SIMS analysis were slightly higher than those obtained using IRMS, the discrepancy can be attributed to the differing methodologies and is within acceptable limits (Helser et al. 2018b). Given the small sample size for SIMS, more samples are needed for a robust comparison between these methods.

The carbon stable isotope composition in otoliths reflects a combination of dissolved inorganic carbon (DIC) from the surrounding water, absorbed through gills or intestines, and metabolic carbon derived from food (Chung et al. 2019; Kalish 1991), with relative contributions of 1 and 16%, respectively (Sherwood and Rose 2005; Tagliabue and Bopp 2008). A negative correlation between otolith carbon stable isotope values and both temperature and metabolism has been documented (Kalish 1991; Martino et al. 2020), with metabolism largely influenced by temperature (Høie et al. 2004a). This study supports a positive correlation between δ¹³C and δ¹⁸O values. Since oceanic DIC is generally uniform, significant dietary changes are required to noticeably affect otolith isotope values (Shephard et al. 2007; Trueman et al. 2016). Thus, changes in otolith δ¹³C values likely reflect shifts in the proportional contributions of carbon sources (Dufour et al. 2007; Kalish 1991). Dietary shifts linked to habitat depth changes in Pacific cod and walleye pollock likely have a larger impact on isotope values than vertical variations in DIC (~ 1‰; Becker et al. 2016; Kroopnick 1985; Park et al. 2018; Yamamura et al. 1993, 2013). We observed significant differences in δ¹³C values inside and outside the check for Pacific cod (1.45, 1.5 and 1.08‰ for samples from the Pacific Ocean off Hokkaido, Aomori Prefecture, and Sendai Bay) and walleye pollock (1.21 and 1.3‰ for samples from the Pacific Ocean off Aomori Prefecture and Sendai Bay; Table 1; Fig. 3). The variations in carbon stable isotope values further substantiate the settlement process in both Pacific cod and walleye pollock.

Stock discrimination

The spawning grounds for Pacific cod in the fishing areas off Hokkaido are distributed throughout the species’ range (Mishima 1989). In the Tohoku region, Sendai Bay serves as one of the primary spawning and nursery grounds, where larval and juvenile cod grow and develop for several months (Hashimoto 1974; Narimatsu et al. 2015a). For walleye pollock, the main spawning grounds of the Japanese Pacific stock are located in the waters around Funka Bay (Madea et al. 1981; Yoon 1981). The relatively restricted spawning areas and the broad distribution of these species likely explain why classification success based on carbon and oxygen stable isotope analysis was higher for samples taken from outside the check compared to those taken from inside the check, for both Pacific cod and walleye pollock. Therefore, we focused our interpretation on the δ¹³C and δ¹⁸O values from the areas outside the check.

The carbon stable isotopes in Pacific cod otoliths showed a decreasing trend from north to south across the three regions, which may be linked to a negative correlation with temperature (Kalish 1991; Martino et al. 2020), as well as geographic differences in carbon stable isotope composition (Becker et al. 2016; Tagliabue and Bopp 2008). Samples from Hokkaido exhibited significantly higher δ¹⁸O values than those from the southern regions. This is likely due to the overriding effect of temperature, which dominated over the latitude-related influence, despite the generally lower oxygen stable isotope content of high-latitude seawater (Campana and Thorrold 2001; Nelson et al. 1989; Trueman et al. 2012).

For Pacific cod, classification success based on carbon and oxygen stable isotopes was higher for samples from Hokkaido and Sendai Bay than for those from the Aomori coast. We combined the Aomori and Sendai Bay samples into a single Tohoku stock and performed a random forest analysis comparing this stock with the Hokkaido stock. The classification success rates were 78% for the Tohoku stock and 72% for the Hokkaido stock, resulting in an overall accuracy of 76% (κ = 0.60). These results support the current management division of Pacific cod into Hokkaido and Honshu units based on stable isotope analysis, although a genetic study did not differentiate between the two stocks (Suda et al. 2017). Due to the small sample size from Sendai Bay, we could not combine the Hokkaido and Aomori samples for comparison. Nonetheless, the 46% classification success rate warrants further attention (Table 2). The detection of nuclear-derived radionuclides in Pacific cod near Hokkaido (Nakamura et al. 2018), the documented feeding migrations of Mutsu Bay-origin individuals to southeastern Hokkaido waters (Shimada 1994), combined with our otolith-derived geospatial data, collectively provide robust evidence for latitudinal migration within the species’ distributional range along Honshu Island. While the fine-scale migratory connectivity of Hokkaido stocks requires further elucidation, a non-negligible proportion of Pacific cod exhibit trans-regional migratory behavior between northeastern Honshu and Hokkaido (Fig. 6A). Current genetic evidence shows no detectable genotypic differentiation among Pacific cod populations within northeastern Japan’s waters. Such genetic homogeneity, however, does not preclude phenotypic plasticity, as demonstrated by Brophy et al. (2020) in cases where conspecifics occupy ecologically distinct habitats during critical life stages. Critically, while sporadic reproductive connectivity may transiently counteract genetic divergence, contemporary management frameworks must prioritize these demographically independent units given their temporal relevance to fisheries sustainability (Kritzer and Liu 2014).

Fig. 6.

Migration routes for Pacific cod (A) and Walleye pollock (B) along the Pacific coast of northern Japan. Dashed arrows indicate hypothesized migration routes (unconfirmed)

The ambient temperatures experienced by walleye pollock in Sendai Bay during their first year were lower than those of fish sampled from Aomori Prefecture. However, no definitive explanation for this observation has been found although Yang et al. (2008) indicated that oxygen stable isotope values do not always show a negative correlation with temperature. Similar to Pacific cod, the discrimination success for walleye pollock samples from Sendai Bay was significantly higher than for samples from Aomori Prefecture. Funka Bay serves as a principal spawning ground for the Japanese Pacific stock of walleye pollock, where spawning aggregations occur predominantly at the bay mouth (Nakatani 2016). Eggs hatch during the process of being transported from the mouth part of Funka Bay into the bay by the wind-driven current and Coastal Oyashio Water (Shimizu and Isoda 1997). This critical habitat supports the walleye pollock Pacific stock distributed from Hokkaido’s Pacific coast to the Tohoku region, which is targeted by fishermen using bottom gill nets and set nets operated from the mouth of Funka Bay to the continental shelf outside the bay (Nakatani and Maeda 1981). This evidence indicates unidirectional migration of walleye pollock from Hokkaido to Honshu, with demographic mixing observed among individuals reaching Honshu waters based on our otolith-derived geospatial data. However, the migratory patterns within Hokkaido’s distribution range remain unclear (Fig. 6B). Given the limited number of sampling sites in the current study, future investigations will incorporate expanded geographical coverage to facilitate more robust ecological interpretations.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

RW conceived the ideas and designed methodology; RW and SK collected the data; RW and QZ analyzed the data; RW, QZ, SK, JY, and YT conducted writing of the manuscript. All authors contributed critically to the drafts and gave final approval for publication.

Data availability

All sample information and otolith δ¹³C and δ¹⁸O values are provided in the main text. Environmental data used in this study are available for download via the links included in the article.

Declarations

Conflicts of interest

The authors declare that they have no conflict of interest. Yongjun Tian is a member of the Editorial Board, but he was not involved in the journal’s review of, or decision related to this manuscript.

Animal and human rights statement

This study was conducted in compliance with the guidelines and approval of the Animal Research and Ethics Committees of Ocean University of China (Permit Number: 20141201), and does not involve any studies with human participants.