Six adult chub mackerel (females: Mk-1 to Mk-3, males: Mk-4 to Mk-6) were collected by longline fishing around Tsushima Island, East China Sea, on 27 April 2022. Fish were frozen onboard at -30 °C, stored at -20 to -25 °C for eight days, and subsequently transported to the laboratory at -15 °C. Body weight and fork length were recorded prior to dissection. Total blood volume was estimated as 60 mLkg-1 according to teleost values (Brill et al., 1998; Itazawa et al., 1983). Each individual was dissected into eight tissues (red and white muscle, liver, gonad, spleen, heart, gills, and blood). All tissue samples were stored at -80 °C until the analysis. Wet weight was measured, and hepatosomatic index (HSI) and gonadosomatic index (GSI) were calculated as follows (Allaya et al., 2013; Shiraishi et al., 2008): 1HSI=Liver weightTotal body weight×1002GSI=Gonad weightSomatic weight×100 Tissues with high iron content (red muscle, liver, gonad, blood, and heart) were selected for Fe K-edge XANES analysis. Small frozen fragments were immediately sealed in oxygen-impermeable polycarbonate film (ASONE, Japan) after dissection and stored until measurement.

Analytical procedures followed Hasegawa et al. (2023). Tissues were freeze-dried, and lipids removed from liver and gonad using chloroform : methanol mixture (2:1, v/v). Tests with certified reference materials confirmed no significant isotopic alteration during lipid removal (<0.07 ‰, n=4). Dried tissues (0.3–2.5 g) were digested in perfluoroalkoxy (PFA) vials using 68 % HNO3 and 35 % H2O2 (both Tamapure AA-100, Tama Chemical Corp., Japan) at 120 °C. The digests were evaporated to dryness and re-dissolved in 7 M HCl containing 0.001 % H2O2. Each sample was split for iron concentration analysis by inductively-coupled plasma mass spectroscopy (ICP-MS; Agilent 7700) and another aliquot for isotope analysis. Iron concentrations were determined from a single aliquot of each tissue sample.

Iron purification was performed using anion-exchange chromatography following Maréchal et al. (1999). Interfering elements (Ni, Cr, Cu) were removed with 7 M HCl, and Fe was subsequently eluted with 2 M HCl. The purified Fe fraction was dissolved in 2 % HNO3 at a concentration of 1 µgmL-1, with Cu added for mass bias correction.

Stable isotope ratios were determined by multicollector-ICP-MS (Neptune Plus, ThermoFisher Scientific) at The University of Tokyo. Instrumental settings were: RF power 1200 W; Ar gas flow 15 Lmin-1 (cool), 0.7 Lmin-1 (auxiliary), 1.0 Lmin-1 (nebulizer). Signal intensity was 5–8 V for 1 µgmL-1 56Fe+, with blanks below 0.004 V. Each tissue sample was measured once for iron isotope composition with 60 analytical cycles. The δ56Fe value was expressed relative to IRMM-014 as: 3δ56Fe=(56Fe/54Fe)sample(56Fe/54Fe)IRMM-014-1×1000[‰] Whole-body δ56Fe (δWB) was calculated as: 4δWB=∑i(δ56Fei×Ti)=∑i(δ56Fei×Ci×Mi) where i represents the ith tissue, and Ti, Ci, and Mi represent the Fe burden, Fe concentration, and tissue mass, respectively.

Analytical accuracy was verified using certified reference materials (BCR-414, DORM-4, DOLT-5, ERM-CE464). Results were consistent with reported values: BCR-414 (-0.10 ‰ ± 0.03 ‰), DORM-4 (-0.24 ‰ ± 0.06 ‰), DOLT-5 (-2.32 ‰ ± 0.02 ‰), ERM-CE464 (-0.55 ‰ ± 0.02 ‰).

Iron K-edge XANES was analyzed at BL-12C, Photon Factory (KEK, Japan) following the protocol of Hasegawa et al. (2023). Frozen tissue fragments were analyzed in fluorescence mode using a Si detector, with energy calibration against hematite (7110.2 eV). The number of independent spectral components was determined by principal component analysis (PCA).

The dominant Fe species in animal tissues include heme (Fe–porphyrin), ferritin (storage), and transferrin (Fe transport). Transferrin-bound Fe was excluded from further analysis due to minor abundance and negligible isotope fractionation (Walczyk and von Blanckenburg, 2005). Thus, subsequent analyses focused on ferritin- and heme-bound Fe species. Linear combination fitting (LCF) was performed to quantify their relative proportions in each tissue.

To account for possible post-mortem oxidation of hemoproteins, spectra of oxyHb, deoxyHb, and metHb were included in the fitting. Horse spleen ferritin (Hs-Ft) and bovine blood metHb (Bb-metHb) were used as reference standards, while oxyHb and deoxyHb were prepared from metHb following Di Iorio (1981) and Wilson et al. (2013). Pre-edge features were extracted and fitted using pseudo-Voigt functions implemented in Python. The quality of spectra fits was evaluated using the R-factor, defined as: 5R=∑{χobs(E)-χcal(E)}2∑{χobs(E)}2

Statistical analyses were conducted in Python (ver. 3.9.19). Sex differences in physiological traits, Fe concentration, and isotope ratios were tested using the Mann–Whitney U test.

The six chub mackerel analyzed had fork lengths (mean ± 2 S.D.) of 37.7 ± 2.41 cm in females and 40.9 ± 4.00 cm in males, and corresponding body weights of 0.71 ± 0.30 kg and 0.89 ± 0.24 kg, respectively (Table 1). Although females tended to be slightly smaller, the difference in body length was not significant (p=0.06). Based on the growth curve of Shiraishi et al. (2008), all individuals were estimated to be at least four years old and sexually mature. The hepatosomatic index (HSI) ranged from 1.05–2.14 in females and 0.74–1.26 in males, and the gonadosomatic index (GSI) ranged from 4.26–10.1 in females and 7.25–10.0 in males, consistent with the spawning season in the East China Sea (Shiraishi et al., 2008). Among tissues, white muscle represented the largest biomass fraction (31 %–52 % of total body weight; Fig. S1 in the Supplement), followed by gonads (3.8 %–8.9 %), red muscle (4.5 %–7.7 %), gills (2.1 %–3.9 %), liver (0.4 %–1.9 %), heart (0.2 %–0.6 %), and spleen (0.2 %–0.4 %). No significant sex-related differences were observed (p>0.05). Red and white muscles, as well as gills, exhibited strong positive correlations with body weight (r>0.82, p<0.04), whereas gonads showed a positive but non-significant trend (r=0.77, p=0.07; Fig. S2).

Mean iron concentrations (±2 S.D.) were highest in the spleen (3100 ± 860 µg(gd.w.)-1), followed by blood (1300 ± 480 µg(gd.w.)-1), heart (720 ± 580 µg(gd.w.)-1), liver (630 ± 510 µg(gd.w.)-1), gills (290 ± 90 µg(gd.w.)-1), red muscle (250 ± 84 µg(gd.w.)-1), gonads (37 ± 24 µg(gd.w.)-1), and white muscle (15 ± 10 µg(gd.w.)-1) (Fig. 1A). Male spleens tended to have higher Fe concentrations than those of females, and liver and gonads also tended to be higher in males.

Assuming a blood volume of 30 mLkg-1 and 1.05 as the specific gravity of fish blood (Davison, 2011), the total iron content in blood was estimated to be between 3.2 and 10.3 mg (Table S1 in the Supplement). Red muscle contributed the next largest Fe pool (2.3–6.1 mg), followed by gills (1.3–3.7 mg), white muscle (0.9–2.0 mg), spleen (0.8–2.6 mg), liver (0.5–2.0 mg), gonads (0.3–1.0 mg), and heart (0.2–0.9 mg). The total body iron inventory was therefore estimated to be approximately 17–26 mgkg-1. Although the liver is the principal iron storage tissue, it accounted for only 4 %–8 % of total body iron. In contrast, blood and red muscle together comprised 51 %–67 % of the total iron burden.

Mean iron isotope compositions (δ56Fe ± 2 S.D.) of each tissue across individuals were as follows: red muscle, -1.54 ‰ ± 0.20 ‰; white muscle, -1.46 ‰ ± 0.21 ‰; liver, -1.19 ‰ ± 0.16 ‰; gonads, -1.23 ‰ ± 0.42 ‰; spleen, -1.37 ‰ ± 0.15 ‰; heart, -1.43 ‰ ± 0.19 ‰; gills, -1.37 ‰ ± 0.22 ‰; and blood, -1.39 ‰ ± 0.20 ‰ (Fig. 2). The net δ56Fe in whole mackerel bodies weighted by tissue Fe contents ranged from -1.50 ‰ to -1.35 ‰ with the liver in all individuals showing higher values than the net values. No sex differences were detected, although ovarian δ56Fe was lower than testicular values.

Iron K-edge XANES spectra identified the dominant chemical forms of iron in each tissue (Wilke et al., 2001). Pre-edge peak positions for all samples were distributed between ferritin (most oxidized) and hemoglobin (reduced) reference standards (Fig. 3A). We note that the XANES spectrum of MK-2 liver was unavailable because its quality was not good.

Linear combination fitting (LCF) using Hs-Ft and Bb-Hb derivatives revealed that red muscle and heart were dominated by heme-bound Fe, whereas ferritin accounted for 8 %–31 % of total Fe in the liver (Fig. 3B). One individual exhibited extremely low liver ferritin iron (8 %). The liver ferritin fraction was comparable to those reported for chub mackerel from other regions (28 %; Hasegawa et al., 2023) but substantially lower than those of mouse liver (66 %; Chen and Chen, 2018).

Sex-related differences were also apparent: females exhibited lower ferritin-bound Fe proportions in the liver, while higher proportions in the red muscle and gonads than males. In ovaries, ferritin represented the predominant Fe form (70 %–80 %), whereas testes showed highly variable proportions (3 %–59 %).

First, we focused on the liver because it is a central tissue for iron storage in vertebrates and is expected to exhibit iron isotope fractionation through the synthesis of ferritin. All the chub mackerel individuals showed higher δ56Fe values in the liver than δWB across eight tissues. Because ferritin preferentially incorporates heavier Fe isotopes during the oxidation in the storage process according to human studies (Albarède et al., 2011), this enrichment most likely reflects the contribution of ferritin-bound iron within the liver. It should be noted that human ferritin consists of H and L subunits responsible for iron oxidation and mineralization (Plays et al., 2021). While teleost ferritin includes an additional M-type subunit with intermediate functional properties, its effect on iron oxidation and isotope fractionation remains unclear and is not considered further here (Dickey et al., 1987; Andersen et al., 1995).

The proportion of ferritin-bound Fe in mackerel liver (8 %–31 %) was notably lower than that reported for mammalian liver (∼60 %; Chakrabarti et al., 2015; Chen and Chen, 2018), suggesting that a smaller fraction of hepatic Fe is stored as ferritin in teleost. Consequently, the δ56Fe enrichment in the mackerel liver appears more moderate than in mammals, possibly due to a greater influence of isotopically lighter Fe pools such as circulating hemoglobin iron.

Because blood contains little ferritin-bound iron, the elevated liver δ56Fe cannot be explained by residual blood contamination. Given that iron cycling within the body is a semi-closed system, in which the liver acts as the central exchange site between storage (ferritin) and transport (hemoglobin and transferrin) pools (Slusarczyk and Mleczko-Sanecka, 2021), hepatic δ56Fe can be interpreted as a mixture of these isotopically distinct components. This relationship can be expressed as: 6δLiv=pδFt+(1-p)δHm where p is the fraction of ferritin-bound Fe in the liver, and δFt and δHm represent the isotopic compositions of ferritin- and heme-bound Fe, respectively.

By assuming that the lowest δ56Fe observed here in red muscle represents δHm (-1.66 ‰), the isotopic offset between ferritin- and heme-bound Fe in the liver (Δ=δFt-δHm) was estimated to be 2.72 ‰ ± 3.03 ‰ (2 S.D.) in all specimens, whereas it was 4.15 ‰ ± 2.84 ‰ (2 S.D.) in females and 1.76 ‰ ± 0.88 ‰ in males. These estimated values are higher than previous observations in skipjack tuna (Katsuwonus pelamis, female, Δ=1.52 ‰, n=1; Hasegawa et al., 2023) and another chub mackerel specimen (Δ=1.41‰, male, n=1; Hasegawa et al., 2023), and, in females, appear higher than the equilibrium isotope fractionation between Fe(II) and Fe(III) (∼2.8 ‰, Johnson et al., 2002; Welch et al., 2003). The two females exhibited a lower ferritin-bound iron fraction than the three males, while their liver δ56Fe values were comparable to those of the males, resulting in a tendency for their estimated Δ values to be higher. Because of the limited sample size, it cannot be definitively concluded that a sex difference exists. However, one possible explanation is that the release of isotopically lighter Fe from hepatic ferritin following Rayleigh fractionation leads to significant enrichment of heavy Fe in the residual ferritin and results in Δ values larger than those expected under isotopic equilibrium. As the sample consists of mature individuals during the spawning season, it cannot be ruled out that physiological changes relating to spawning may have temporarily altered iron metabolism, particularly in females.

Outside the liver, some tissues did not exhibit elevated δ56Fe values despite high ferritin content. In particular, ovaries and red muscle in females exhibited high ferritin-to-heme ratios but δ56Fe values comparable to other tissues. This pattern suggests that ferritin-bound Fe in these tissues is not retained over long periods but is actively mobilized and recycled in response to metabolic activity and oxygen consumption. For example, ferritin in skeletal muscle can be quickly mobilized to meet oxygen demand (Robach et al., 2007; Ryan et al., 2021), preventing the accumulation of a heavy isotope signature and maintaining δ56Fe values close to those of the whole-body iron pool dominated by heme iron. Moreover, δ56Fe has been associated with oxidative stress markers including superoxide dismutase and glutathione peroxidase (Sauzéat et al., 2025), indicating that tissues prone to oxidative stress, such as gonads, may display isotopic variability due to redox-related iron turnover.

These observations suggest that ferritin-bound (storage) iron in chub mackerel exhibits a high turnover rate, resulting in small δ56Fe differences among tissues. In contrast, heme iron likely turns over more slowly. Fish erythrocyte lifespans range from ∼80 to 500 d (Götting and Nikinmaa, 2017; Avery et al., 1992), comparable or longer than those of mice (∼60 d, Dholakia et al., 2015) and humans (∼120 d; Shemin and Rittenberg, 1946). Marine mammals, by contrast, display extended erythrocyte lifespans that scale with diving capacity (Pearson et al., 2024), highlighting the link between oxygen storage and iron utilization. In migratory fish such as mackerel, the high and continuous demand for heme to support aerobic activity may promote rapid redistribution of heme iron, which may result in reduced δ56Fe variability among organs.

The range of δ56Fe variation among tissues in chub mackerel was clearly narrower than that reported for mammals. For example, the differences in δ56Fe values between muscle and liver range from 0.72 ‰ to 1.71 ‰ in mice (Balter et al., 2013), 1.64 ‰ to 2.72 ‰ in sheep (Balter et al., 2013), and 0.83 ‰ to 1.75 ‰ in humans (Walczyk and von Blanckenburg, 2002), whereas the maximum isotopic difference among tissues in the present chub mackerel specimens was 0.73 ‰. Despite this relatively small internal fractionation, δ56Fe values in marine organisms often show species-specific ranges (Fig. 4). For example, muscular δ56Fe values in skipjack tuna are high regardless of regional differences (-1.46 ‰ to -0.71 ‰; Hasegawa et al., 2022), whereas mammals generally show lower values (-3.79 ‰ to -1.5 ‰; Walczyk and von Blanckenburg, 2002; Balter et al., 2013). This pattern does not appear to arise from the enrichment of lighter isotopes in specific tissues, but rather from the limited influence of internal isotopic fractionation on whole-body δ56Fe. This interpretation is consistent with our previous study on a single individual (Hasegawa et al., 2023), and the limited inter-tissue δ56Fe variation observed across multiple individuals in the present work supports that this characteristic is generally applicable to chub mackerel.

As discussed in the previous section, hepatic ferritin iron likely undergoes isotopic fractionation on the order of approximately 1.5 ‰–3 ‰, based on the Δ values estimated in this study and Hasegawa et al. (2023). However, because ferritin constitutes only a small fraction of total iron, its contribution to whole-body δ56Fe variation is limited. Consequently, heterogeneous distribution of δ56Fe across tissues is likely to have only a minor influence on the δ56Fe of indicator tissues representing whole-body δ56Fe values (e.g. muscle with a large Fe pool size). The primary determinant of tissue δ56Fe in chub mackerel therefore appears to be iron uptake process from external sources rather than internal redistribution.

The relatively high mean δ56Fe values in mackerel tissues may primarily reflect more efficient intestinal iron absorption than in mammals, resulting in a stronger reflection of environmental δ56Fe values. Additionally, selective ingestion of prey with intrinsically higher δ56Fe values could also contribute to their isotopic composition. Various prey taxa of marine fish, including zooplankton, squid, and crustaceans, are known to exhibit relatively high δ56Fe values (-1.00 ‰ to -0.03 ‰; von Blanckenburg et al., 2013; Hasegawa et al., 2022, 2023). The δ56Fe differences observed among fish groups such as tuna and mackerel (-1.58 ‰ to -0.71 ‰) versus sardine and herring (-2.64 ‰ to -1.73 ‰; Hasegawa et al., 2022) seemed difficult to explain solely by isotopic differences in their prey since all prey species analyzed so far showed higher δ56Fe values than sardine and herring (Hasegawa et al., 2022). However, recent studies have suggested species-specific δ56Fe variability even among zooplankton (Hasegawa et al., 2026) and therefore prey-derived δ56Fe may also partially contribute to the overall isotopic composition of fish such as sardines and herrings that feed selectively on specific plankton groups. Further expansion of marine biological δ56Fe databases may enable a more comprehensive understanding of Fe physiology in wild marine organisms.

In contrast, terrestrial mammals such as humans, mice, and sheep consistently show large δ56Fe differences between liver and muscle (Walczyk and von Blanckenburg, 2002; Balter et al., 2013), reflecting a well-developed iron recycling system based on the long-term ferritin storage. This mechanism reduces daily iron requirements and promotes preferential uptake of lighter isotopes, thereby lowering the overall δ56Fe values. In contrast to mammals that rely on ferritin-based iron storage and recycling, the isotopic pattern observed in fish may indicate that long-term iron storage plays a less dominant role in iron homeostasis, despite inhabiting chronically iron-limited environments. They may instead have evolved a “recycle dominant” mode of iron homeostasis that relies on intensive recycling of iron to sustain metabolic demands.

The results of this study suggest that the δ56Fe variability in marine fish reflects dietary characteristics and uptake efficiency more strongly than internal physiological variations. The distinctly lighter isotopic signatures of animals compared to seawater may ultimately influence the isotopic composition of iron recycled through marine food webs. Therefore, whole-body δ56Fe values could serve as an integrated tracer of biologically available iron within marine ecosystems. Our analyses also indicate that in fish, δ56Fe values in non-destructive tissues such as blood can represent the whole-body isotopic composition, highlighting their potential as key species for assessing iron supply and cycling in higher marine ecosystems.

In the future, applying a Rayleigh-type model may allow quantitative evaluation of the relative contributions of absorption and retention to isotopic fractionation in fish. Combined with XAFS analyses that provide species-level insights into ferritin- and heme-bound iron pools, these approaches could offer a powerful framework for linking biological iron metabolism to marine biogeochemical iron cycling. In addition, further biochemical studies examining Fe isotope fractionation during enzymatic incorporation into iron-binding proteins such as ferritin or heme in fish would help to better constrain the mechanisms controlling Fe isotope distribution in marine organisms.