Spatiotemporal variability in the occurrence of juvenile Japanese jack mackerel Trachurus japonicus along coastal areas of the Kuroshio Current

To understand the population structure of the Japanese jack mackerel Trachurus japonicus in coastal areas adjacent to the Kuroshio Current (referred to as the “CAK”), we analyzed size composition and commercial landing data of juvenile fish in these areas for the period 2005– 2015. Trachurus japonicus does not undergo population-scale spawning migration, and thus, the connectivity between the spawning and ju-venile/adult habitat areas is important. Therefore, our primary aim was to assess the origin of juveniles landed in a number of subareas, including those spawned in local spawning grounds in January– May in the western part of the CAK (w- CAK), those spawned in May– July in the eastern part (e- CAK), and those spawned in February– March in the remote spawning ground in the southern East China Sea (s- ECS). Fishing periods starting in spring (spring onset) were commonly observed in the CAK, which involved relatively small size classes (50– 100 mm fork length [FL]). Back estimates based on the growth rate of T . japonicus suggested that the contributions from the s-ECS probably dominated most of the spring onsets in April– June because the smallest size class (50– 70 mm FL) occurred almost exclusively in April– May. In autumn, onset


| INTRODUC TI ON
Currents and tides transport and widely disperse the eggs, larvae, and juveniles of many marine organisms from the spawning grounds.
A proportion of them may reach nursery areas, from where recruitment to the adult habitats occurs. For migratory species, ontogenetic growth is followed by seasonal or spawning migration to spawning grounds. In fisheries science, this life history is called the "migration triangle" (Cadrin & Secor, 2009;Harden-Jones, 1968). Although this concept has been amended numerous times (e.g., the "combined contingent theory"; Secor, 1999), returning to the spawning ground, one of the main components of the triangle, is still widely used to explain the migration patterns of various marine fish species, such as the sardine (Sardinops melanostictus: Kuroda, 1991;S. sagax: van der Lingen et al., 2010), mackerel (Scomber scombrus: Uriarte & Lucio, 2001;S. japonicus: Yukami et al., 2018), and herring (Clupea harengus: Wheeler & Winters, 1984;C. pallasii: Tojo et al., 2007).
However, some fish species do not follow this migration pattern.
One species for which no population-scale spawning migration has been observed is the Japanese jack mackerel Trachurus japonicus, an ecologically and commercially important species in East Asian waters. Trachurus japonicus inhabits the coastal areas of China, Taiwan, Korea, and Japan. The pelagic eggs and hatched larvae of T. japonicus are transported passively by surface currents. When the larvae become juveniles exceeding approximately 40 mm in fork length (FL), they move from a surface-layer habitat to a benthic-layer habitat (Sassa et al., 2006(Sassa et al., , 2009. Because their eggs and larvae are mainly entrained in the Kuroshio Current (Kuroshio) and Tsushima Warm Current (TWC; Kasai et al., 2008), which flow eastward to northeastward, their distribution tends to shift downstream in these currents as they grow. No spawning migration from the downstream to upstream areas along the Kuroshio or TWC has been observed in T. japonicus at the population scale.
A complex population structure has also been suggested for the jack mackerel (genus Trachurus) in other regions. In the northeastern Atlantic and the Mediterranean Sea, multiple stocks of T. trachurus have been identified by integrating several approaches, including genetic markers, morphometry, parasite analyses, and life history traits (Abaunza et al., 2008;Comesaña et al., 2008). Although the genetic differentiation was low, the morphometrics and parasitebased methods suggested the separation of the Atlantic Ocean and the Mediterranean Sea populations, with a buffer zone in the western-most Mediterranean Sea. In the oceanic northeastern Atlantic Ocean, multiple population units of T. picturatus have also been detected with various approaches, including genetic markers, the otolith shape and microchemistry, and parasite associations (Moreira et al., 2020;Vasconcelos et al., 2018). Based on a phenotypic analysis of the otolith of T. picturatus off the Canary Islands, Tuset et al., (2019) suggested the contingent theory to explain its population structure. The main spawning the Chilean jack mackerel T. murphyi is in the open ocean off the coast of Chile, and its distribution, including its nursery and feeding grounds, includes wide areas extending to the coast of Ecuador to the north and to waters off New Zealand (Acros et al., 2001;Parada et al., 2017;Serra, 1991). Gerlotto et al., (2012) investigated the population structure of these four stocks (Chilean, Peruvian, central South Pacific, and western South Pacific) in a literature review. They noted that the main and permanent spawning ground off Chile suggests the metapopulation structure.
The spawning grounds of T. japonicus that contribute to its occurrence in these areas have been classified into the local spawning grounds in the coastal areas along the Kuroshio (collectively referred to as "CAK") and the remote spawning ground in the southern East China Sea (s-ECS) ( Figure 1). The local spawning grounds at the scale of bays and channels along the CAK, identified by sampling mature adults, have been documented since at least the 1950 s (Hashida et al., 2019;Hattori, 1964;Sakaji, 2001;Sakamoto et al., 1986;Sawada, 1974;Yakushiji, 2001;Yokota & Mita, 1958). Using maturation indices (mainly the gonadosomatic index) for adult T. japonicus, the spawning seasons in each of the areas within the CAK were estimated to be May-July in Sagami Bay (Kobata, 1972;Sawada, 1974), February-June in Kii Channel (Sakamoto et al., 1986), January-June in Sukumo Bay (Sakaji, 2001), and February-June (Yakushiji, 2001) and January-July (Hashida et al., 2019) in Bungo Channel (Figure 1).
Surveys of the ichthyoplankton in the ECS since the 2000 s have shown high densities of larvae in the southern ECS (s-ECS) from February to March, and it has been suggested that this is the primary spawning ground for the entire T. japonicus population (Sassa et al., 2006(Sassa et al., , 2009(Sassa et al., , 2016 Figure 1). However, the paths of the Kuroshio and TWC suggest that not all eggs spawned in the ECS are transported to the CAK, but that some are also transported to coastal areas by the TWC or retained within the ECS. Therefore, the relative contributions of these local and remote spawning areas have yet to be confirmed.
Until the large spawning ground was discovered in the s-ECS (Sassa et al., 2006), studies of the transport of T. japonicus to signals were associated with the landing of juveniles from the local spawning ground in an eastern subarea of the e-CAK. Despite the asymmetric transport and migration flows between the habitat areas of T. japonicus, its population levels may be sustained because the local and remote spawning grounds are used in different seasons.

K E Y W O R D S
commercial landing data, jack mackerel, population structure, spawning ground, transport Japanese waters mainly focused on the eastern ECS near Japan (Kim & Sugimoto, 2002). After the discovery of the spawning ground in the southwest ECS by Sassa et al., (2006), interest shifted to how eggs and larvae are transported from this spawning ground to the fishing grounds in Japanese waters. Kasai et al., (2008) conducted particle-tracking experiments that incorporated fish mortality to investigate the transport of T. japonicus larvae from the ECS to the Sea of Japan and the Pacific side of Japan. Their numerical experiments showed that larvae can reach the Sea of Japan or the Pacific side from the ECS and that advection and diffusion transports ~80% of them to the Pacific side (although survival was not considered).
The studies cited above clarified the transport patterns of T. japonicus larvae from the s-ECS, but more detailed studies of the distribution of larvae and juveniles after they enter the Pacific side of Japan (i.e., within the CAK) are required to understand the roles of the various spawning grounds. There is some evidence that T. japonicus originating in the s-ECS reaches specific areas of the CAK. Xie and Watanabe (2007) used a sagittal otolith analysis to assess the F I G U R E 1 Hydrography and habitat areas for T. japonicus. Synoptic paths of the Kuroshio Current and Tsushima Warm Current (TWC) and the spawning ground (SG) in the southern East China Sea (s-ECS) are shown in (a). The coastal area of the Kuroshio (CAK) in Japan, demarcated by the rectangle and the dotted line in (a), is enlarged in (b-d), which show (b) the entire CAK, (c) the western part of the CAK (w-CAK), and (d) the eastern part (e-CAK). In (b), the long-term mean latitude of the Kuroshio Current is shown at intervals of 1° longitude with black dots and a line. Locations of temperaturemonitoring buoys are shown as triangles.
Star in (c) shows the location of the port for purse seine landings, and the circles in (c) and (d)   In the cited above, Xie and Watanabe (2007)  Shizuoka, SO; and Kanagawa, KN). Purse seine netting was used in one subarea (EH) (Figure 1). The daily landing data from 44 fisheries units in these eight subareas were used in this study (Table 1). As mentioned above, the spawning periods in each area partly overlap, but the spawning periods occur later from west to east. In this study, the term "main spawning season" is used to refer to the period in which most adults are mature (after the first maturation).
We assumed in this study that the main spawning season is January-May in the western part of the CAK (corresponding to the waters off the KG-WK subareas; Table 1; referred to as the western CAK or the w-CAK; Figure 1c) and May-July in the eastern part (corresponding to waters off the ME-KN subareas; Table 1; referred to as eastern the CAK or the e-CAK; Figure 1d). These assumptions on the main spawning seasons, based on the surveys of the gonadosomatic index in the previous studies (see Introduction), are consistent with the occurrences of T. japonicus eggs and larvae (Sakaji et al., 2008(Sakaji et al., , 2009Takasuka et al., 2006Takasuka et al., , 2007Takasuka et al., , 2010Takasuka et al., , 2011Takasuka et al., , 2014Takasuka et al., , 2015Takasuka et al., , 2016Takasuka, Nashida, Udagawa, & Sakaji, 2012, 2013; these reports, covering the period 2006-2016, are hereafter referred to as "FRA_06-16"; egg distribution data have been available since 2013).

| Commercial landing data
The daily landing data by weights at each fishery unit were first averaged on a weekly basis to reduce the bias caused by occasional no-landing records ( Figure 2a). It was often impossible to distinguish between whether a no-landing event occurred despite fishing effort or there was no effort because fishing was suspended for an unspecified reason (e.g., adverse weather conditions). Records of zero catch after the calculation of the weekly means were therefore regarded as missing values.
The weekly landing data at each fishing unit were log-transformed and standardized. The standardized data were then averaged within subareas for each week of the year, to obtain C ( Figure 2a). The conversion processes for the weekly landing data for the subareas with single fisheries units altered the amplitude of the data, but not the underlying pattern. Similarly, the weekly long-term mean (C) was calculated for each subarea by averaging C over subareas and years. The confidence intervals were calculated with the bootstrap method.

| Length compositions and equivalent numbers of individuals
The landings of these units are dominated by 0-and 1-year-old fish . The smallest length in the market size categories is about 50-150 mm FL (Table 1; hereinafter called "juveniles"). Subsamples were taken for scientific analysis from these commercially based landings almost monthly, and their length compositions were recorded.
The FL composition data were used to assess the size of T. japonicus corresponding to the landing data ( Figure 2b). The mid-values of the bin at the smallest frequency mode (e.g., 55, 65, and 75 mm FL), L, were then regarded as the FL of the juveniles in that week ( Figure 2b).
Although the approximate FL range for the commercial landing data was assumed based on interviews with local fishermen (Table 1) where W and L are the weight index and FL, respectively ( Figure 2c).
The linear coefficient of 1.033 × 10 −5 for the length-weight relationship is not included in Equation (1), because L in Equation (1)  An index of the catch in numbers, called the "equivalent numbers of individuals," was then estimated by dividing the landing of the smallest market categories by W. For the weekly data, this was denoted as "N E, " and the weekly long-term mean, calculated from the weekly long-term mean landing and the weight index, was denoted as "N E " (Figure 2d).

| Onset of the fishing period: A weight-based index to detect increases in numbers
Although the average seasonal fluctuation pattern can be examined with N E , calculated from the long-term mean data (Figure 2d), the patterns in single years cannot be similarly examined with N E . This is because of the sparseness of L (0-3 times per month), which was used to calculate N E . However, the weekly mean catch in weight C is dense, although its seasonal peak is probably influenced by the increase in the weight of individuals. Therefore, considering the relationship between N E and C, we introduced another index, the onset of the fishing period from C, which is less affected by the weight increase ( Figure 2e).
The definition of "onset" is based on the relationship between N E and C (Figure 2e). We defined the fishing periods as those in which the standardized weekly year-to-year landing data continuously exceeded the 25th percentile of the log-normal distribution (0.5094) for periods of ≥5 weeks. The onset and termination weeks were then specified for each fishing period. The criterion of the 25th percentile was selected because it properly detected major onsets of the weekly long-term mean equivalent number of individuals, and the 5-week criterion was selected to detect monthly scale signals.
The detections of onsets based on N E and C are actually consistent over the percentile range of 20-35%. Nevertheless, the sensitivity test showed that adopting a different percentile criterion within this range does not change the seasonal onset pattern when applied to the weekly data ( Figure S1).

| Coastal temperature, Kuroshio axis, and population level
The coastal water temperature and the stream of the Kuroshio are

| Seasonal landing fluctuations
The patterns of the seasonal fluctuations in landing in the eight subareas along CAK were estimated from the weekly long-term means of the standardized landing data (Figure 5) The deepest layer was selected if thermistors were moored in multiple layers.
was one peak in the standardized landing data in July-September in the KN, SZ, ME, and KG subareas, and in June in the EH subarea, in the middle of the short fishing season that exclusively targeted very small FL (≤100 mm). Although less clear than in these subareas, increases were observed in August-September in the WK subarea, in September-October in the KO subarea, and in June-August in the MZ subarea.
Because the seasonal fluctuations were gradual, the onsets defined by the 25th percentile were detected well before the peak, except in the EH subarea, whereas second onsets were observed during autumn in the ME and KG subareas. The first onsets appeared from mid-March to May in the KN, SZ, ME, WK, EH, and KG subareas, and during February in the KO and MZ subareas.
Fluctuations in the long-term mean weekly equivalent number of individuals (N E ) showed a markedly different pattern ( Figure 6) from that of the weight-based landing data ( Figure 5). Although N E was generally low from late autumn to early winter, marked increases were observed from late winter to spring. Peaks in N E on the seasonal scale were then observed from late March to May in the four subareas, whereas in MZ, a second peak appeared at the end of May.
The timing of the onset was nearly concurrent with that of the landings. This indicates that it is possible to detect the mass emergence of small juveniles through onset signals.

F I G U R E 3
Weekly long-term mean fork lengths for the minimum market size classes in the four subareas for which the available data represented the seasonal variability. Asterisks indicate periods in which 1 year was used to calculate the long-term means 12 24 36 48 Week

| Year-to-year variations in weekly landings
The weekly landing time series for individual years indicated intermittent fishing periods, the timing, duration, and magnitude of which differed among years (Figures 7-9). As shown in Figures 5 and 6, the peaks in the landing time series lagged behind those for N E , but the onsets of the fishing periods almost coincided with these peaks. Therefore, we focused mainly on the onsets of the fishing periods.
Apart from the EH subarea, where the fishing season was concentrated in the April-June period, weekly landings occurred almost throughout the year and most frequently in April-October. There

| Environmental conditions and relationship with population level
The weekly equivalent numbers of individuals (N E ) were compared within the corresponding weeks ( Figure 11). Although the temperature ranges differed among subareas and seasons, the range in which substantial numbers of individuals were found was 15-25°C (Figure 11a,c,e,g). High N E values were mainly detected during January-March, but the temperature range in this period differed among subareas. The linear correlation between temperature and N E was not significant, except for the MZ subarea during October-December (r = −0.64, p < 0.05). The effects of abrupt intrusions of the Kuroshio into coastal areas, which typically cause rapid temperature rises over 1-2 days (e.g., Akiyama & Saitoh, 1993), were not evident in the data.
The latitude of the Kuroshio axis generally indicated the distance of the mainstream from the coast, and this distance is smaller at higher latitudes (Figure 11b, d, f, h). Although there was substantial F I G U R E 6 Long-term mean equivalent numbers of individuals (N E ) for the four subareas, calculated from the long-term mean values for landings and fork length. Onsets of N E (see text for details) are indicated by black triangles  Assuming that the cohorts causing onsets and peaks were the same, we can infer the spawning season corresponding to each onset from the length obtained for the season of the peak in N E . In this context, the growth rate of T. japonicus recorded by Ochiai et al., (1983) (Ochiai et al., 1983), and based on this, the growth periods for the other sizes can be estimated. Although there is great variability in growth rates, especially in the larval stage, those individuals with a very low growth rate are less likely to survive (Takahashi et al., Although we did not have the size composition data at weekly intervals (unlike the landing data), increases in size from 50 mm FL in the four available subareas in CAK typically commenced no earlier than March and predominantly ended in June (Figures 3 and 4).

| Autumn onsets
Unlike the spring onsets, those in autumn were not necessarily composed of size classes ≤100 mm FL (Figures 3, 4 and 10

| Transport and population structure
The temperature at 10-30 m recorded by monitoring buoys near the set nets showed that juvenile T. japonicus were mainly landed at the  (Figure 11). This is consistent with the previous finding of bottom water temperature of >15°C in the East China and Yellow Seas during all seasons (Sassa et al., 2009), and 20-50 m layer (where acoustic signals were detected) temperature of 19-21°C in the Sea of Japan during June-July (Nakamura & Hamano, 2009 Kasai et al., 1993), as had been partly confirmed for the EH subarea by Kim et al., (2007) and Hashida et al., (2017). Local-scale migration of schools could also moderate the apparent relationship between temperature or the stream axis and landings. The large interannual variability in the seasonal fluctuations (Figures 7-9) also suggests that the stochasticity in landings was influenced by the local-scale movement of schools. The relationships were unclear partly because the amount of weekly means used in the present study were not sufficient to detect correlations. A detailed analysis of the relationships involved will be the subject of future research.
Because the population of T. japonicus is probably highest in the ECS, allochthonous larval supplies from the s-ECS may be important for subpopulations in the CAK, as well as for the autochthonous reproduction within the local areas. If the recruitment estimated with the VPA accurately reflects the population in the entire CAK, then the total number of individuals within the CAK increases as the population increases. Although we did not compare the VPA-based recruitment with the total number of individuals, but with the sum of N E in each subarea in April-June, a positive relationship was expected for any subarea in which the connectivity to a major subpopulation (e.g., those transported from s-ECS) was sufficient. However, the relationship was not statistically significant, possibly because of the highly stochastic variability in landings. More detailed data are required to more fully explore this statistical relationship.
In this study, the pattern of occurrence of juvenile T. japonicus in the CAK was investigated, mainly using size composition and landing data for the smallest marketable size range. Our results suggest that the remote s-ECS spawning ground was the major influence on landings from spring to summer over most of CAK, although contributions from local or neighboring spawning grounds were also likely in some cases. The importance of neighboring spawning grounds was most clearly evident in the KN subarea, where onsets of autumn landings were probably initiated by small individuals.
The larvae and juveniles of T. japonicus transported to the Sea of Japan from the s-ECS by the TWC are also thought not to return to the s-ECS, and it has been suggested that the populations in the s-ECS and the Sea of Japan form a metapopulation (Sassa et al., 2016).
Although T. japonicus share its metapopulation structure with other Trachurus species (e.g., T. picturatus: Moreira et al., 2020;T. murphyi: Gerlotto et al., 2012), one way in which it differs from those inhabiting in the northeastern Atlantic and the southeastern Pacific Oceans is in its apparent unidirectional dispersal, which may be caused by the strong streams of the Kuroshio and the TWC. Roles of disturbances diverged from the mainstream in retention and dispersal of larvae and juveniles may be important for T. japonicus in forming the population structure.
Given the source-sink network caused by the unidirectional transport of the Kuroshio, sustainability of the adult stock in the s-ECS is necessary. Numerical experiments by Kasai et al., (2008) suggested that 10-15% (their Figure 10a)  Migratory and resident forms of T. japonicus with slightly differing morphologies are known to occur in various coastal areas of Japan, including the CAK (e.gAzeta & Ochiai, 1962;Kanaji et al., 2010). Both forms occur in the same fishing grounds, but although the migratory form originates in the remote s-ECS, the differences between them are phenotypic and not genetically determined. The effects of their differing life histories may be critical in sustaining the population of T. japonicus, despite the asymmetric transportation and migration flows between distant habitat areas.

ACK N OWLED G M ENTS
This study was supported by Grant-in-Aid for JSPS Research Fellows

CO N FLI C T O F I NTE R E S T
The authors have no conflict of interest to declare.

AUTH O R S CO NTR I B UTI O N S
KI and SI conceived and designed the study; TT, HH, DH, TO, TT, MK, YS, ST, and RF provided the commercial landings data; CW provided the length composition data; TK provided the coastal water temperature data; KI compiled and analyzed the data and drafted the manuscript; CW, TK, and SI contributed to revise the manuscript; All authors participated in discussion and approved the final version of the manuscript.

DATA AVA I L A B I L I T Y
The data that support the findings of this study are available from the corresponding author upon reasonable request.