Habitat use and growth strategies of amphidromous fish “ayu” throughout a river system

Abstract

Adult spawners are key contributors to fish population dynamics, and thus understanding their life history is essential for effective population management and conservation. The radiogenic Sr isotope ratios (⁸⁷Sr/⁸⁶Sr) and ring numbers of adult spawners’ sagitta otoliths were measured to determine the habitat use strategy of ayu Plecoglossus altivelis. Six groups of habitat use were identified based on profiles of the otolith ⁸⁷Sr/⁸⁶Sr ratio, indicating high contributions of wild ayu (87.2%) and upper-to-middle mainstem habitats for spawning populations. Otolith ring number analyses showed that an early hatch date led to an early river entry and a larger fish size. The upper- and middle-segment (mainly mainstem) residents had earlier hatching and river entry dates than the tributary residents, with middle-segment residents having the largest fish size. These results indicate that early hatched and early river-entering ayu preferentially occupied mainstem habitats and grew large, whereas later-hatched and later-river-entering ayu tended to avoid mainstem habitats and enter tributaries as growth sites. This study can contribute to future habitat conservation efforts and fishery resource management for ayu.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-02988-8.

Introduction

Generally, migratory fish undergo habitat shifts throughout their life stages. Throughout this journey, fish face multiple mortality risks, the survivors of which eventually join the spawning population. In rivers, success of reproduction by spawning adults (hereafter “adult spawners”) and subsequent recruitment of juveniles are strongly influenced by the annual environmental conditions, such as water temperature and flow discharge, resulting in annual variabilities in the population¹. Simultaneously, the number of adult spawners (i.e., spawning population size) would be a more fundamental factor in population persistence. Therefore, the habitat use history sustaining growth and survival of adult spawners is key to ensuring the population persistence through identifying critical habitats and will contribute to future habitat conservation and fishery resource management.

The ayu Plecoglossus altivelis is distributed throughout the Japanese Archipelago and along the continental coast of the Far East^2,3. In Japan, the ayu is a highly valuable freshwater species for commercial and recreational fisheries. Ayu is an amphidromous fish that spawns in rivers during autumn. The hatched larvae immediately drift to the sea and inhabit coastal marine waters (particularly the surf zone) during winter^4–6. Juveniles return to the rivers in spring, grow in summer, mature, spawn in autumn, and then die (i.e., annual fish). From spring to summer, ayu disperse throughout the river system, both in the mainstem as well as in its tributaries⁷, feeding on periphyton⁸ and growing rapidly. During this period, some large ayu have territories and feed while excluding other fish, whereas others of small ayu form schools without territories⁹. In autumn, when triggered by a decrease in water temperature and freshet, ayu begin downstream spawning migration and congregate in alluvial fan segments^7,10.

Ayu is closely related to the lives of the Japanese people, as described in the Nihon Shoki (Chronicles of Japan), the first official history of Japan compiled in 720 AD (Nara era), and in a law (taxation by products) of the same era¹¹. Aquaculture and fishery resource management studies^12,13 were first performed in the early 1900s, and several ecological and genetic studies have subsequently been conducted. In rivers and oceans, the behavior, habitat characteristics, spatiotemporal dynamics (phenology), and environmental triggers of each life stage of the ayu have been thoroughly studied^{5–7,9,10,14–17}. However, no spatial or quantitative assessments of the past habitats (i.e., habitat use history) used by spawning adults exist, and this topic remains a simple and fascinating mystery.

Early hatched ayu enter rivers earlier, grow larger, and spawn earlier than those that hatch later^14,18–20. Based on catch records, the middle mainstem of a river system is occupied earlier within a river system²¹. These observations led to the hypothesis that if early hatched and early river-entering ayu preferentially occupy the mainstem habitats, later-hatched and later-river-entering ayu may avoid the mainstem and move to the tributaries to find feeding sites with less competition.

In this study, we spatially and quantitatively assessed the habitat use history of ayu at the segment scale throughout a river system through analyzing the sagitta otoliths of adult spawners that congregated at a spawning site and examined their relationships with their life history traits. For these purposes, we combined otolith daily ring counting²² with radiogenic strontium isotope ratios (⁸⁷Sr/⁸⁶Sr) as signatures for habitat use^23–25. In the Nagara River system, an isoscape of the ⁸⁷Sr/⁸⁶Sr ratio was constructed based on thorough sampling of river water throughout the river system²⁶ (Fig. 1).

Fig. 1.

(a) Isoscape of ⁸⁷Sr/⁸⁶Sr ratios and (b) its distribution in each of the different segments of the Nagara River system: Yoshida River (Yo), upper mainstem (UM), middle mainstem (MM), Kibijima River (Ki), Itadori River (Ita), Tsubo River (Tsu), and Mugi River (Mu). Colored and gray circles indicate the habitats available and non-available for wild ayu, respectively, due to presence/absence of a dam (bold black line) intercepting upstream migration of wild ayu. The distances (km) from the river mouth are also shown. The alluvial fan Sect. (42.0–53.0 km), including the main spawning sites, is surrounded by a dashed line. Isoscape visualization was conducted using QGIS version 3.10.6 (https://ftp.osuosl.org/pub/osgeo/download/qgis/windows/).

Strontium deposited in otoliths is derived from ambient water and food, with ambient water contributing 83% and 88% to juvenile mummichogs (Fundulus heteroclitus)²⁷ and Nile tilapia (Oreochromis niloticus)²⁸, respectively. Therefore, the ⁸⁷Sr/⁸⁶Sr ratios of the river water and otoliths generally do not match perfectly. However, the ⁸⁷Sr/⁸⁶Sr ratio of ayu otoliths should be extremely close to that of river water because the diet of ayu in rivers is almost exclusively composed of periphyton, with an ⁸⁷Sr/⁸⁶Sr ratio almost identical to that of river water²⁹. Additionally, the ⁸⁷Sr/⁸⁶Sr ratio of river water at each location is stable over time, except for sudden events such as floods^30,31. Hence, the habitat use history of ayu during freshwater life can be predicted with a high accuracy based on the correspondence between ⁸⁷Sr/⁸⁶Sr ratios of river water and otoliths.

Results

An ⁸⁷Sr/⁸⁶Sr isoscape of the Nagara River system (Fig. 1a) was drawn based on the results of Ida et al.²⁶ using QGIS. In the current study, ⁸⁷Sr/⁸⁶Sr values for the mainstem and five main tributaries, which are potential habitats for ayu, were used. In Fig. 1a, the colored reaches are available habitats for wild ayu migrating from the ocean, and some reaches with transparent gray circles are unavailable habitats for wild ayu due to check dams which intercept their upstream migration. Figure 1b shows the ⁸⁷Sr/⁸⁶Sr values separately for each segment and tributary, corresponding to the coloring in Fig. 1a. The ⁸⁷Sr/⁸⁶Sr values were low (generally < 0.71) in the upper mainstem (UM) and Yoshida River (Yo), which originated from Pleistocene volcanic rocks, including felsic and mafic rocks, whereas in the tributaries (Kibijima (Ki), Itadori (Ita), and Tsubo Rivers (Tsu)) originating from Mesozoic sedimentary rocks, the ⁸⁷Sr/⁸⁶Sr values were high (generally > 0.713). Although the upper reaches of the Mugi River (Mu) also had high ⁸⁷Sr/⁸⁶Sr values (generally > 0.716), the ⁸⁷Sr/⁸⁶Sr values in the reaches available to wild ayu were moderate (approximately 0.7125) because of a mixture of tributary water with low ⁸⁷Sr/⁸⁶Sr values (< 0.71) originating from limestone. The ⁸⁷Sr/⁸⁶Sr values of the mainstem were higher downstream because of the confluence of tributaries with higher ⁸⁷Sr/⁸⁶Sr values, ranging from approximately 0.7075 to 0.713 in the upper (UM) and middle (MM) mainstems, respectively.

Wild ayu, which have a constant seawater ⁸⁷Sr/⁸⁶Sr ratio (0.70918) in the front half of the core-to-edge otolith transect (i.e., earlier life stages), constituted 116 of 133 individuals (87.2%) (Fig. 2a–e, Groups 1–5); only 17 (12.8%) showed no sign of marine life (Fig. 2f, Group 6). Five patterns of freshwater habitat use for wild ayu were visually and comprehensively determined based on Sr isotopic profile in the back half of the otolith transect, its numerical value, and isoscape. Group 1 (N = 39, 29.3%) comprised the “upper segment migrants” with a low ⁸⁷Sr/⁸⁶Sr ratio (< 0.71) (Fig. 2a), which generally corresponded to that of the mainstem and a tributary (Yoshida) in the upper basin (Fig. 1). Group 2 (N = 38, 28.6%) constituted the “middle segment settler” individuals with a phase of an almost constant ⁸⁷Sr/⁸⁶Sr ratio of 0.711–0.713 (Fig. 2b), which corresponded to that of the middle mainstem and a tributary (Mugi) (Fig. 1). Group 3 (N = 20, 15.0%) comprised “middle segment wanderers,” having fluctuations of ⁸⁷Sr/⁸⁶Sr ratio within the range (0.711–0.713) of Group 2 (Fig. 2c). Group 4 (N = 13, 9.7%) constituted “tributary settlers” with almost constant and elevated ⁸⁷Sr/⁸⁶Sr ratio (over 0.714) (Fig. 2d), which generally corresponded to that of the tributaries (Itadori and Tsubo) of the middle basin (Fig. 1). Group 5 (N = 6, 4.5%) comprised “tributary-mainstem wanderers” whose ⁸⁷Sr/⁸⁶Sr ratios stair-stepped from the middle mainstem (0.711–0.713) to the middle tributaries (> 0.714) or ultimately decreased from the middle tributaries to the middle mainstem (Fig. 2e).

Fig. 2.

Habitat use patterns of wild ayu (a-e, Groups 1–5) and others (f, Group 6) based on profiles of ⁸⁷Sr/⁸⁶Sr ratios along core-to-edge otolith transects of adult spawners (N = 133). The shaded area represents the 95% confidence interval of each profile. The dotted line indicates the ⁸⁷Sr/⁸⁶Sr ratio of seawater (0.70918). The x-axis is the relative distance on the core-to-edge otolith transects, ranging from 0 to 1.

The ages of wild ayu (N = 82), which had countable, clear otolith rings, were estimated to be 250–338 d; hence, the estimated hatch dates ranged from October 29, 2021, to March 27, 2022. The dates of spring river entry, as estimated based on the relative position of the change point deviating from the seawater ⁸⁷Sr/⁸⁶Sr ratio (0.70918) on the otolith transect, ranged from February 23, 2022, to August 17, 2022 (Fig. 3a). Linear regression analyses showed that an earlier hatch date, led to an earlier river entry (P < 0.001) (Fig. 3a), a slightly longer ocean residence time (P = 0.03) (Fig. 3b), and a larger fish size (P < 0.001) (Fig. 3c).

Fig. 3.

Relationships between the hatch date and three ecological traits of wild ayu (N = 82): (a) river entry date, (b) ocean residence time, and (c) standard length. The wild ayu that were judged to be accurately counted as based on a clear otolith ring presence were used. The solid lines indicate the regression models, and the shaded area represents the 95% confidence interval of the regression model.

Considering the spatial overlaps, the five habitat-use patterns of wild ayu (Groups 1–5) were classified into three resident types: Type I was “upper segment residents (Group 1),” Type II was “middle segment residents (Groups 2 and 3),” and Type III was “middle tributary residents (Groups 4 and 5).” Two-sample Kolmogorov-Smirnov (K-S) tests showed differences in hatch and river entry dates among the resident types, with earlier hatch for Type II and earlier river entry dates for Types I and II than for Type III (Fig. 4). Kruskal-Wallis (K-W, P = 0.004) and subsequent Dunn tests (Types I–II: P = 0.005, Types II–III: P = 0.016) revealed that fish sizes (standard length) were largest in the “middle segment residents (Type II)” (Fig. 5).

Fig. 4.

Frequency distributions of (a) hatch and (b) river entry dates for each habitat-use type of wild ayu (N = 82). Type I: upper segment residents (Group 1, N = 29), Type II: middle segment residents (Groups 2 and 3, N = 36), and Type III: middle tributary residents (Groups 4 and 5, N = 17). The significant differences based on the K-S test are also represented.

Fig. 5.

Standard lengths of ayu (N = 82) for each habitat-use type. Type I: upper segment residents (group 1, N = 29), Type II: middle segment residents (groups 2 and 3, N = 36), and Type III: middle tributary residents (groups 4 and 5, N = 17). The top and bottom of the box indicate the 75th (Q3) and 25th (Q1) percentiles, respectively, whereas a horizontal line inside the box indicates the median (Q2). The whiskers extend from the box to the minimum and maximum values within 1.5 times the interquartile range from Q1 and Q3, respectively. Dots are outliers. Different letters above the box indicate the statistical significances based on the Dunn tests after the K-W test.

Discussion

Based on an analysis of the otoliths of adult ayu spawners, this study provides the first vivid picture of the history of habitat use throughout their lives. Hatch timing influenced the timing of subsequent life stages as well as the habitat-use strategy throughout the river system. This study provides the first quantitative evidence of segment-scale habitat use by spawning ayu populations throughout a river system. These results can contribute to efficient habitat and resource management of wild ayu populations.

Profiles of ⁸⁷Sr/⁸⁶Sr ratios in otoliths indicated that wild ayu exhibited five patterns of habitat use during their freshwater life (Fig. 2a–e). Habitat use patterns revealed that ayu used a wide range of river systems consisting of tributaries and mainstems, with spatial and quantitative variability. The upper segment migrants (N = 39, 29.3%), middle segment settlers (N = 38, 28.6%), and middle segment wanderers (N = 20, 15.0%) were the major groups of adult spawners, indicating their substantial contribution to the spawning population. The Sr isotopic ratios did not discriminate between mainstem and tributary users within each of these groups. In addition to individuals from the mainstem, the upper segment migrants included individuals from the Yoshida River, and the middle segment settlers and wanderers included individuals from the Mugi River. However, our earlier study based on environmental DNA analysis showed that the DNA concentrations of ayu were considerably lower in these tributaries than those in the mainstem from August to November, indicating lower fish density⁷. Additionally, the carrying capacity, which is strongly influenced by habitat area and primary production (i.e., periphyton), should be also lower in the tributaries than that in the mainstem. The channel widths of the Mugi and Yoshida Rivers were approximately 30 m, while those of the mainstem were approximately 80 m and 60 m, respectively, immediately upstream of each confluence (check with Google Earth). Primary production was lower in the lower-order river segments (i.e., tributaries)³². Therefore, upper-segment migrants and middle-segment settlers and wanderers would be predominantly mainstem users, which contribute strongly to the spawning populations. The other two groups were in the minority: tributary settlers (N = 13, 9.7%) and tributary-mainstem wanderers (N = 6, 4.5%). Although the contribution of tributary habitats to spawning populations was low, tributaries may also be important for increasing the number of adult spawners and the diversity of ecological traits.

The percentage of wild ayu was as high as 87.2% (Fig. 2a-e), indicating a high contribution of wild ayu to the spawning population. Previous studies have also reported a high percentage of wild ayu during spawning season in the Nagara River: 71.4–86.1% in October^17,33 and 89% in late September to late November³⁴. The remaining 12.8% of ayu without a sign of marine life stage on otoliths (Fig. 2f) would consist of hatchery-reared ayu. However, the ⁸⁷Sr/⁸⁶Sr ratios in the front half of the otolith transect (i.e., earlier life stages) differed among individuals. This discrepancy may have occurred because ayu originated from multiple hatcheries. Another possibility is that wild ayu reared in estuaries (brackish water) were also included³⁵. Further studies are required to determine the origins of these fish.

Numerous hatchery-reared ayu (approximately 4 million juveniles) have been released annually throughout the Nagara River system in recent years¹⁰, and the number exceeds that of wild ayu in some years. Nevertheless, most of the spawners were wild (87.2%); this is presumably because hatchery-reared ayu are selectively removed by “tomozuri” angling¹⁷, which captures large and dominant territorial ayu until the spawning season. Hatchery-reared ayu are generally released at a larger size (ca. 10 cm) than that of wild ayu in spring and can then retain a territory due to a competitive advantage, which causes them to be selectively fished. Interestingly, this dynamic implies that the mass release of hatchery-reared ayu may help protect wild ayu from fishing pressure in the study river.

Based on the previous study²², we should potentially involve errors in estimating the hatch date, river entry date, and ocean residence time of ayu due to potential errors (generally underestimations) in the daily otolith ring counts (i.e., age). The error did not exceed 5 d (2%) until 300 days of age, but was approximately 20 d (5–6%) on average after 300 d²². Nevertheless, the hatch and river entry dates of ayu in this study generally coincided with previous records for the Nagara River. Muto³⁶ estimated the hatch period of ayu to be at least mid-October to early February based on the daily otolith ring counts of juveniles (up to approximately 170 days old), which have subsequently been confirmed to be highly accurate²². The past 30 years of ayu spring run records at the Nagaragawa Estuary Barrage (see the study site) clearly show that upstream migration began in February, peaked from April to May, and then declined from June onward. This result suggests that the hatch and river entry dates in the present study were slightly overestimated owing to an underestimation of the otolith ring counts of adult ayu. Therefore, the present estimated data can be used to identify relative trends along the hatch and river entry dates, although the data are not suitable for specifying the hatch and river entry periods or analyzing behavior at a daily resolution.

The result of an earlier hatching date leading to an earlier river entry date (Fig. 3a) is consistent with those of previous studies^18,33. Previous studies have also shown that early hatched and early river-entering ayu are larger when they enter rivers in spring than later-hatched and later-river-entering ayu^18,19,37. This dynamic may be related to the slightly longer marine life of early hatched and early river-entering ayu (Fig. 3b). Primary production is high in the surf zone³⁸ which is the main coastal habitat of ayu^4–6. High primary production can promote the reproduction of copepods that feed on ayu³⁹, attracting the juveniles of several fish species, including ayu⁴⁰. Therefore, a long marine life is favorable for growth. However, the early river-entering ayu were not necessarily older with a longer marine life than the later river-entering ayu³⁶. Under the food satiation condition in artificial seawater, a higher growth rate of juvenile ayu was observed at higher water temperatures (15–25 °C)⁴¹. This suggests that the high growth of early hatched and early river-entering ayu may be explained by the higher water temperatures due to the early onset of marine life rather than the duration of marine life. Further studies are needed to elucidate the relationship between early growth and the duration and timing of the onset of marine life.

We also found that early hatched and early river-entering ayu became larger spawning adults (Fig. 3c). This result is consistent with the findings of Iguchi et al.²⁰, who determined that the weight of ayu caught after September can be explained by the duration of their freshwater life: the longer the freshwater life, the greater the weight. The early onset of freshwater life may promote ayu growth through a longer feeding period on periphyton. In addition, ayu within a territory have a relative growth advantage over non-territorial ayu⁹. Early river-entering ayu are large at the beginning of their spring run^18,19,37. Therefore, they continue to maintain competitive advantage and retain their territory until the spawning period, which may allow them a high growth rate.

Furthermore, the present study highlights the habitat use and growth strategies of ayu throughout the river system, depending on the timing of hatching and river entry. The hatching and river entry dates of the middle tributary residents (Type III) tended to be later than those of the upper (Type I) and middle segment residents (Type II) (Fig. 4). The size of the tributary residents (Type III) was smaller than that of the middle-segment residents (Type II) (Fig. 5), which could be attributed to the shorter freshwater life due to late river entry²⁰. In addition, the later-river-entering ayu were smaller, even during their spring runs^18,19,37. Considering that the mainstem is likely to be the primary habitat for the upper and middle segment residents, as discussed above, the early hatched and early river-entering ayu would thus preferentially occupy the mainstem habitats and grow large, whereas later-hatched and later-river-entering ayu would avoid the mainstem because of their small size (i.e., inferior to interference competition) and enter the tributaries as growth sites.

Although there was no difference in river entry dates between the upper- and middle-segment residents (Fig. 4), fish sizes were significantly larger for the latter residents (Fig. 5). There are two possible reasons for this finding. First, primary production is higher in the middle mainstem (high order)³², and the microhabitats available to make territories are more abundant in the middle mainstem because of the wider river channel than in the upper mainstem. Another reason is that the upper segment is a famous fishing ground that attracts “tomozuri” anglers from all over Japan, so there is substantial fishing pressure for large ayu and they may be selectively removed¹⁷. These natural and anthropogenic factors likely influence the ecological traits of spawning populations.

This study also highlights the importance of river network continuity, including that of tributaries. River network continuity ensures the multiple strategies of habitat use and growth of ayu, which may support the spawning population size. The intake of weirs and dams impedes fish migration and dramatically alters the habitat available for migratory fish⁴². Therefore, rivers with a relatively consistent network continuity, such as the Nagara River, have high priority as conservation targets. In particular, the continuity of the mainstem habitat, which extends over 140 km and contributes substantially to the spawning population of the Nagara River, should be protected. In general, the installation and ongoing maintenance of fish passages and the removal (or partial removal) of existing dams should be considered to recover the habitats of migratory fish⁴³. This study provides fundamental information for the management of ayu resources. The contribution of adult spawners from the middle mainstem habitat to reproduction was considerable, both in terms of population proportion and body size, which affected egg numbers (larger fish had more eggs⁴⁴). In the Nagara River, “Ochi-ayu” fishing is practiced at several locations along the mainstem to catch ayu migrating downstream toward their spawning sites. Currently, there is a discussion on adjusting the “Ochi-ayu” fishing period to the delayed downstream migration of ayu due to climate warming¹⁰. However, adjusting the fishing period also poses the risk of overfishing. Based on our findings, limits on the number of fishing sites and days and/or catch should be considered in the mainstem which produces many adult spawners. Special attention should be paid to the highly efficient “yana” fishing method, which creates transient levee-like mounds to converge and catch ayu migrating downstream. Regulations are required to avoid catching all important adult spawners.

Methods

Study site

This study was conducted on the Nagara River, which has a channel length of 166 km and a basin area of 1985 km². The 42.0–53.0 km, 53.0–75.0 km, and > 75.0 km sections from the river mouth flow through the alluvial fan, valley bottom plain, and mountain area, respectively. The ayu can migrate to the check dam, which is located 143 km from the river mouth (Fig. 1a). The major ayu spawning sites are located in the main river channel of the alluvial fan. The Nagara River has five major tributaries (Yoshida, Kibijima, Itadori, Mugi, and Tsubo Rivers). The Tsubo and Itadori Rivers have no obstacles that block the upstream migration of ayu, whereas in the other three tributaries, the upstream migration is intercepted by check dams constructed in the middle or lower reaches (Fig. 1a).

In the Nagara River, the spring upstream migration of wild ayu occurs from February to June, according to an annual survey conducted by the Nagaragawa Estuary Barrage Operating & Maintenance Office of the Japan Water Agency (https://www.water.go.jp/chubu/nagara/15_sojou/kakosojou.html). Additionally, numerous hatchery-reared juvenile ayu (ca. 4,000,000 individuals) produced from adults caught at spawning sites the last year are released into the mainstem and tributaries from April to July by six fishery cooperative associations. Therefore, both wild and hatchery-reared ayu are widely distributed throughout the Nagara River system during early summer. However, the distribution of ayu is subsequently adjusted in response to flood disturbances⁴⁵, fluctuations in water temperature⁷, and possibly the foraging environment³².

Isoscape

An ⁸⁷Sr/⁸⁶Sr isoscape of the Nagara River system (Fig. 1a) was based on the Ida et al.²⁶. River water samples were collected at 115 points in the Nagara River system, which were strategically selected to cover the river system with higher spatial resolution in the middle and upper reaches (i.e., growing habitat for ayu)²⁶.

Fish samples

Adult ayu spawners (N = 133) were collected by one fisherman through traditional cast netting and angling methods at the Kagashima fishing site (47.4 km from the river mouth, Fig. 1a) from September 17 to December 17, 2022 (Fig. S1). The majority of samples (September 17–November 28) were acquired via the cast netting, locally called “sebari-ami” to capture ayu migrating toward downstream for spawning. In this method, fishermen preliminarily string ropes with stakes and place conspicuously colored (usually white) objects on the riverbed across the river channel. Then the fishermen with cast nets wait for schooling ayu migrating downstream during the daytime. Some ayu schools are startled by the ropes and colored objects and wander upstream, where the fishermen cast nets to catch the wandering ayu schools. Fish sampled only on December 17 were caught by “korogashizuri” angling, where fish are hooked. Angler hooked fish by pulling a line with many hooks attached along the riverbed. In the Nagara River, several major spawning sites are included in a reach of the alluvial fan (Fig. 1a). Adult spawners from different habitats are probably mixed in this reach because they migrate downstream with each flood¹⁰. The standard length (mm) of each sample was measured. The fish samples were stored in a freezer at − 20 °C until otolith extraction.

Ring count and isotope analysis of otoliths

Sagitta otoliths were extracted from the skulls, washed with ultrapure water, and air-dried in a desiccator with clean circulation (DG851; Yamato Scientific Co., Ltd., Tokyo, Japan). The otoliths were then embedded in epoxy resin (SpeciFix Resin; Struers, Ballerup, Denmark) and polished to the core on polyester sheets coated with diamond powder of decreasing particle size (2000–8000 grit). Each polished sample was fixed onto a glass slide and washed with ultrapure water. All otolith samples were measured for length, and the number of rings (i.e., age in days) in each sample was counted from the nucleus to the posterior edge by the same one person using a marker projector attached to an optical microscope. The daily ring formation of ayu otoliths was validated by the Tsukamoto & Kajihara²².

To estimate the habitat-use history of ayu individuals based on the correspondence between ⁸⁷Sr/⁸⁶Sr values of the isoscape and otolith, the ⁸⁷Sr/⁸⁶Sr of sagitta otoliths was measured along a core-to-edge otolith transect using Neptune Plus MC–ICP–MS with a laser ablation system (LSX-213 G2 + equipped with a HelEX 2-volume cell, CETAC Technologies, USA). The ablated material was transported to the ICP via a He gas carrier stream where it was combined with N2 gas. The laser operating conditions were set to a repetition rate of 8–15 Hz, traverse speed of 3 μm/s, and spot size of 30 μm. The signal intensities (V) of ⁸⁸Sr, ⁸⁷(Sr + Rb), ⁸⁶(Sr + Kr), ⁸⁵Rb, ⁸⁴Sr(Sr + Kr), ⁸³Kr, and ⁸²(Kr + Se) were measured using Faraday cups with an integration time of 1.0 s/cycle. Prior to each ablation sequence, the background intensities were measured for 60 cycles along the transect, with the median value applied as a blank correction during sample measurement. The ⁸⁷Sr/⁸⁶Sr ratios underwent mass bias correction through normalization to ⁸⁶Sr/⁸⁸Sr = 0.1194, following the removal of Kr and Rb isobaric interference⁴⁶. To ensure compatibility with the water sample data, mass-bias-corrected Sr isotope ratios were normalized to ⁸⁷Sr/⁸⁶Sr = 0.70918 using the coral aragonite standard DH05Ishi-1^25,47. The analytical performance of our in-situ Sr isotope measurements was validated using the homogeneous otolith reference material PO14SMN-01. Throughout the analytical sessions, the measured mean ⁸⁷Sr/⁸⁶Sr values (± SD) were 0.70922 ± 0.000004 (N = 78) for DH05Ishi-1 and 0.70921 ± 0.000006 (N = 78). These results showed a close agreement with the established global seawater ⁸⁷Sr/⁸⁶Sr value of 0.70918⁴⁸, confirming the measurement accuracy.

Life history traits

Grouping for habitat use was visually and comprehensively determined based on the ⁸⁷Sr/⁸⁶Sr profile along a core-to-edge otolith transect, its numerical value, and isoscape. The seawater ⁸⁷Sr/⁸⁶Sr ratio (0.70918)⁴⁸ observed in the front half of the otolith transect was considered a sign of wild ayu. Isoscape of ⁸⁷Sr/⁸⁶Sr ratios (Fig. 1a) and its distribution in each of the different river segments (Fig. 1b) provided criteria for different habitat use. Habitat use patterns (Groups 1–5) were identified from the range and fluctuation of ⁸⁷Sr/⁸⁶Sr ratios, and the spatial overlap of these patterns was considered to identify residence types (Type I–III) (see Results in detail).

The hatch date, river entry date, and ocean residence time (days) were estimated for individual wild ayu (N = 82) which had countable, clear otolith rings. The hatching date was determined by subtracting age in days (ring count) from the capture date. Although the saturation time of Sr concentration in the otolith is approximately 20 d after exposure to ambient water, changes in concentration values begin immediately⁴⁹. Therefore, the river entry date and ocean residence time were determined based on the relative location (proportion 0–1) of the change point where the ⁸⁷Sr/⁸⁶Sr value deviated from that of seawater (0.70918)⁴⁸ along the core-to-edge otolith transect. Ocean residence time was calculated by multiplying the age in days (ring count) by the relative location of the change point. The river entry date was estimated by adding the ocean residence time to the hatching date.

Statistical analysis

Linear regression analyses were performed to examine the relationships between the hatch date and three ecological traits (river entry date, ocean residence time, and standard length) of wild ayu (N = 82). The response variable was each ecological trait, and the explanatory variable was hatch date. Two-sample Kolmogorov-Smirnov tests were performed to examine the differences in river entry dates among the three habitat-use types (upper segment residents, middle segment residents, and middle tributary residents). In addition, the Kruskal-Wallis test and multiple comparisons with Dunn’s test were performed to examine the standard lengths among the three habitat-use types. All statistical analyses were performed using the R ver. 4.1.0⁵⁰ with a significance level of 0.05.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

S. N., T. O., R. F., and M. H. designed the study. S. N. and R. F. pretreated the otoliths, and S. N., T. O., and T. I. analyzed the otoliths. S. N. and T. O. conducted the data analyses, wrote the manuscript, and prepared the figures. All the authors reviewed the manuscript.

Data availability

The data that supports the findings of this study is not publicly available but may be available from the corresponding author upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.