Early life history affects fish size mainly by indirectly regulating the growth during each stage: a case study in a demersal fish

Introduction

Complex life cycles are common amongst organisms, but are immensely varied and can exhibit contrasting morphological, physiological, behavioural or ecological traits (Moran 1994). The genetic correlations may constrain evolution but will be modified by selection, thus leading to trade-offs in resource utilisation among different life history stages (Ebenman 1992). Disruption of genetic correlations enables life stages to respond independently to differing selection forces and to benefit from a change in resource use at some point in the life cycle (Johansson et al. 2010; Werner 1988). To maximise individual fitness, morphologically distinct phases are coupled with each other, or phenotypic induction early in ontogeny affects later phenotypes (Grol et al. 2014; Johansson et al. 2010). Fish are subject to high natural mortality, especially in the first year of life; therefore, clarifying the interrelationships of growth among each stage in the first year is key to understanding population fluctuations (Kristianen et al. 2000). Among these stages, the early life history is crucial to the fate of fish, and it is consequently worthy of study to determine how it affects growth in each subsequent stage (Catalán et al. 2020).

Pacific cod (Gadus macrocephalus) is a commercially important species in the North Pacific Ocean. In Japan, this species supports one of the largest fisheries extending southward from Hokkaido to Ibaragi Prefecture on the Pacific coast, and to the coast of Shimane Prefecture in the Sea of Japan (Hurst et al. 2010; Mishima 1984). In coastal waters of Japan, Pacific cod spawns on sandy mud bottoms at depths of 20–100 m from December to late March (Yoseda 1992). The larvae swim to the surface immediately after hatching and metamorphose to juveniles at 25 mm total length (DiMaria et al. 2010; Takatsu et al. 1995). The juveniles remain in the water column and feed on zooplankton until three or four months of age and settle to the bottom in June (Morioka and Kuwada 2002). Pacific cod migrate to the upper continental slope (200–400 m) during summer and become one of the most important species in this ecosystem (Narimatsu et al. 2015a). The effects of complex physiological and habitat changes on growth in the first year of life remains a major uncertainty. Otoliths, which are located in the inner ears of fish and are composed of aragonite (CaCO₃) precipitated on a protein matrix, provide a natural measure of various aspects in fish life histories, and potentially an environmental record (Campana 1999; Campana and Thorrold 2001). Periods of physiological or environmental stress (including metamorphosis, settlement and temperature) are recorded in the width and density of daily rings in otiliths (Campana 1990) and can provide data on the growth of each life history stage. The presence of accessory growth centre, defined as secondary growth centres around the central primordium and growth zone, has been confirmed in Pacific cod and its congener, Atlantic cod Gadus morhua (Campana and Neilson 1985; Narimatsu et al. 2007; Oeberst and Böttcher 1998). Rehberg-Haas et al. (2012) concluded that the first translucent ring under transmitted light in otoliths of age-0 Baltic cod (Gadus morhua) was deposited ~ 3 months post-hatch and formed in response to the process of settlement of juveniles from a pelagic to a demersal life. Similarly, a settlement check on the otolith of Pacific cod was confirmed on the basis of otolith microstructure and microchemistry analysis (Li et al. 2021; Wu et al. 2021). Pacific cod age determination based on otolith growth zone counts has been called into question because of difficulties in differentiating annual rings from checks, particularly in the first year (Andrews 2015; Kastelle et al. 2017; Mahé et al. 2012; Roberson et al. 2005). Therefore, we analysed the checks and growth rings in otoliths of Pacific cod to provide a reference for accurate age determination. In addition, we investigated the growth during each life history stage separated by checks and assessed whether and how phenotypic induction early in ontogeny affected later phenotypes.

Results

Temporal variation in indicators

The otolith core was within a polygonal structure that pointed to the junction of the ventral (dorsal) side and the anti-sulcus (Fig. 1). The accessory growth centre (AGC) was apparent near this junction and 145 − 263 μm from the primordium; the average distance in each hatch year was 185 − 202 μm (Fig. 2A). In some otoliths, an obvious opaque zone connected to the polygon vertices was evident (Fig. 1B), but another (check 1; ① in Fig. 1), whose shape was similar to the entire contour of the otolith transverse section, was observed in all samples. It was located 314 − 584 μm from the core; the average distances for each of the three hatch years were 414, 465 and 460 μm (Fig. 2A). Based on ANOVA, there was no evidence for temporal variation in AGC size and Check 1 size for samples hatched in 2013 and 2014 in the Sea of Japan. Another very obvious and relatively wide dark band (check 2; ② in Fig. 1) was observed in all samples outside check 1 and 589–1274 μm from the core; the average distances in 2013, 2014 and 2020 were 750, 810 and 940 μm, respectively (Fig. 2A). The next opaque zone (③ in Fig. 1) was defined as the first annual ring. This was a judgment made in combination with the body length information for the fish, as the checks were similar to the annual ring. It was located at an average distance of 1346 μm from the core (Fig. 2A). Hatch year had a significant effect on Check 2 size (F_{1, 26} = 6.27, P < 0.05) and First annual ring size (F_{1, 26} = 10.58, P < 0.01). Individuals hatched in 2013 formed check 2 and the first annual ring at a shorter distance from the core than those hatched in 2014. The average distance between the AGC and check 1 (Check 1 growth) was 250 μm (Fig. 2B), and there was no difference between the samples collected in the Sea of Japan in each of the two hatch years. The mean distances between checks 1 and 2 (Check 2 growth) and between check 2 and the first annual ring (First annual ring growth) for individuals hatched in 2013 were 318 and 303 μm, and in 2014 were 362 and 334 μm (Fig. 2B). These distances did not show temporal variability between two years but were significantly smaller than the corresponding otolith indicator values for individuals collected on the Pacific coast of Hokkaido in 2020, where the Check 2 growth and First annual ring growth were 480 and 555 μm, respectively.

Figure  1.  Schematic diagram of the transverse plane for the otolith of Pacific cod (Gadus macrocephalus) captured in the Pacific ocean of Hokkaido (A SL = 302 mm) and the Sea of Japan (B SL = 189 mm) under the reflected light. Inset: zoomed core area with primordium (P) and accessory growth centre (AGC) under the transmitted light. Thick arrows indicate the polygonal structure surrounding the core. The accessory growth centre is formed when the radiation direction of daily increments changes. The number ①, ②, ③ indicate the check 1, check 2 and the first annual ring, respectively. Solid blue circle indicate the opaque zone on some otoliths that can be observed connected to the polygon vertices

DownLoad: Full-Size Img PowerPoint

Figure  2.  Box plots of otolith size indicators (A Accessory growth centre (AGC) size, Check 1 size, Check 2 size, First annual ring size) and growth indicators (B Check 1 growth, Check 2 growth and First annual ring growth) in three hatch years. Correlation graphs of Check 1 growth and Check 1 size, Check 2 growth and Check 2 size, First annual ring growth and First annual ring size (C). The box scatter plot for the distance from the core at each otolith indicator (D)

DownLoad: Full-Size Img PowerPoint

Interaction of indicators across life history stages

Path analysis provided no evidence that the direct and total effects between otolith indicators differed significantly among years (Supplementary Tables S1 and S2). Therefore, all data for the three hatch years (N = 75) were combined for subsequent analyses. AGC size had a significant impact on all otolith indicators expressed later (Fig. 3). It had significant direct and total effects on Check 1 growth (β = − 0.349) and Check 2 growth (β_DIR = 0.345, β_TOT = 0.326), and the total effect of AGC size on First annual ring growth (β_TOT = 0.21) was also significant, although no direct effect was detected. These results suggest that the larger the AGC size the shorter the distance between AGC and check 1, and the longer the distance between checks 1 and 2, and between check 2 and first annual ring. The growth between check 1 and check 2 was also significantly positively correlated with growth before the first annual ring (β = 0.387). The substantial variation for all three endogenous indicators was predicated by path analysis, including R² = 0.122 for Check 1 growth, R² = 0.109 for Check 2 growth and R² = 0.178 for First annual ring growth.

Figure  3.  Schematic diagram of the relationship among otolith microstructure indicators predicted by path analyses (N = 75). The numbers above each arrow indicate the bootstrap mean standardised effect between two indicators (direct effect β_DIR/total effect β_TOT). The direct effect and total effect are the same between two adjacent traits. Values marked with red star are statistically significant. Arrows with different colour (orange, blue, green and grey) represent the strength of correlation (r, see legend) between the two indicators

DownLoad: Full-Size Img PowerPoint

In addition, we explored the relationship between AGC size and the three size indicators not included in the path analysis (Check 1 size, Check 2 size and First annual ring size). There was no evidence that variation in the AGC size caused significant variation in Check 1 size (r = 0.058). A moderate number of correlations between AGC size and Check 2 size (r = 0.308) and between AGC size and First annual ring size (r = 0.31) were detected. More remarkably, the growth in each stage made a significant contribution to the body length at the end of the stage (Fig. 2C). For example, Check 1 size was strongly correlated with Check 1 growth (R² = 0.838), and the strong correlations between Check 2 size and Check 2 growth (R² = 0.806) and between First annual ring size and First annual ring growth were also observed (R² = 0.713). Greater variation was evident in the distance from the core of otolith indicators toward the later life history, which could be caused by variations in recent growth (Fig. 2D).

Discussion

Biomineralization processes in otoliths are influenced by environmental and physiological factors, and they can also reflect life history events that may interrupt somatic growth (Geffen et al. 2002). However, the various life history stages separated by checks are related, not isolated. In this study, no temporal or spatial difference were identified in the effects of early phenotypic induction on the later phenotype, which indicates that this relationship is stable amongst Pacific cod stocks. Path analysis showed that growth variation in the early life history stages (before the formation of the accessory growth centre) had a significant impact on growth variations in each subsequent stage, which made a significant contribution to the body length variation in Pacific cod at the end of each life stage (reflected by the distance from the primordium of the otolith).

Life history events reflected by checks in the otolith

In this study, we used otolith indicators from multiple Pacific cod individuals from different hatch years and environmental factors to determine if some opaque zones under reflected light similar to the checks were a consistent characteristic of cod. Two checks other than the AGC and the first annual ring were evident in all otoliths. However, it was unclear whether these features in the otolith were related to physiological and environmental changes the Pacific cod experienced during the first year of life. It has been reported that in the otolith the polygonal core area surrounded by the AGC contains 51 − 70 daily increments, and the otolith shape changes from circular to oblong (Narimatsu et al. 2007). In Gadiformes, walleye pollock (Gadus chalcogrammus) and European hake (Merluccius merluccius) the AGC is formed during larval metamorphosis, settlement to the sea floor or dietary shift (Arneri and Morales-Nin 2000; Morales-Nin and Aldebert 1997; Takatsu et al. 2001). However, during metamorphosis and settlement of Pacific cod (total length 25 and 45 mm, respectively), pectoral fin ossification is not completed and the diet does not change during AGC formation. Therefore, it has been speculated that pectoral fin ray ossification is related to formation of the AGC (Matarese 1989; Morioka and Kuwada 2002; Takatsu et al. 1995). The hypothesis that the AGC (which is the origin of the lobes on the otolith) is closely coupled with the ossification process has been proposed previously (Brown et al. 2001; Jearld Jr 1993; Lagardère and Troadec 1997).

Settlement is a crucial process for demersal fish, and variability in the transition of pelagic early stages to bottom habitat has been a cornerstone in the analysis of recruitment potential (Levin 1994). It is an adaptive strategy for pelagic larvae or juveniles to inhabit deep sea environments (Lin et al. 2012). Settlement checks in otoliths have been reported for fish including Atlantic cod (Gadus morhua), white hake (Urophycis tenuis) and other fish taxa among tropical reef fish and rockfish (Laidig 2010; Lang et al. 1996; Rehberg-Haas et al. 2012; Wilson and McCormick 1997). Settlement to a benthic habitat may be associated with changes in feeding because of growing energetic needs. Moreover, it has been demonstrated in western Baltic cod (WBC) that settlement is independent of season, but is a function of age and/or size of the fish (Neuman et al. 2001; Oeberst and Böttcher 1998; Rehberg-Haas et al. 2012). While metabolism remains relatively high during settlement, growth is often slowed or paused during settlement in environments having low temperatures and reduced food supply, and may be linked to reduced otolith opacity (Ciannelli et al. 2007; Høie et al. 2008). An opaque zone at a similar distance from the core as found in the present study has been reported in other studies of cod. For example, step changes found in the Sr: Ca ratio approximately 480 μm from the core and a corresponding check observed in the otolith of Pacific cod in the Yellow Sea suggest that settlement occurred approximately 80–120 days after hatching (Wu et al. 2021). The timing of prey switching inferred from stomach content analysis in juvenile WBC sampled from the Stollergrund in 2016 indicated that settlement took place in July at an age of 90–120 days (Plonus et al. 2021). Based on the narrow distance range (314 − 584 μm) that we identified for the location of the first opaque zone (check 1) in the otolith of Pacific cod, we strongly speculate that it is related to settling.

Juvenile cod settle in June in the demersal zone in a depth range of 40–120 m (Kitagawa 2002; Narimatsu et al. 2015b). According to the 'shallow water refuge' paradigm, nearshore nursery areas supply ample food where Pacific cod can avoid predators and competition from larger conspecifics (Werner and Gilliam 1984). During summer they migrate to the upper continental slope (200–400 m) and become one of the most important species in that ecosystem (Kitagawa 2002; Narimatsu et al. 2015a, b). Predominantly remaining associated with the bottom in the late demersal stage is a common expansion phase for many coastal species when their home range increases (Hüssy et al. 2003; Pittman and McAlpine 2003). Depth changes are typically associated with inshore − offshore shifts; these result in a kind of trophic relay whereby production is transported offshore (Pittman and McAlpine 2003). A water temperature increase and a decrease in prey abundance in shallow waters are two important factors that promote the migration of cod to deeper waters (Takatsu et al. 2001). The combined effects of a stratified environment and vertical migration may cause the formation of another translucent zone (TZ) under transmitted light on the otoliths in summer, resulting in a 'summer ring' rather than a 'winter ring' (McQueen et al. 2019). This is consistent with the results of research on Baltic Sea cod, and a positive relationship between high temperatures and TZ formation in Atlantic cod has been demonstrated in laboratory experiments (Krumme et al. 2020; Neat et al. 2008; Plonus et al. 2021). This relationship corresponds to the period during which ambient temperature peaks and exceeds the optimal temperature for growth (Björnsson et al. 2001; Pilling et al. 2007). The temperature-dependent precipitation of aragonite may explain the decreasing cod otolith opacity with increasing temperature (Fablet et al. 2011). However, increased metabolic rates and energy requirements in fish resulting from high temperatures may not be met by food availability (Hüssy and Mosegaard 2004; Krumme et al. 2020). Analysis of the stomach fullness index (SFI) has shown that the peak summer months are a period of reduced feeding opportunities and that the index increased once TZ formation started (Funk 2017; Plonus et al. 2021). This potentially physiologically stressful period leads to the formation of the TZ. Based on our results, we speculate that check 2 is a sign that the cod have responded to unfavourably high water temperatures in shallow waters and have moved to deeper, cooler and low oxygen saturation water.

We observed that the hatch check and the opaque zone around the polygon core (Fig. 1B) was present in some but not all individuals, so these were therefore not analysed in this study. However, their representational significance and the potential growth information they reflect are worthy of further investigation. We have to refer to body length information for age identification because of the presence of the checks on the otolith. Although the results of this study are consistent with many previous studies on Baltic cod, additional methods including stable isotope analysis should be applied cautiously to the interpretation of checks on the otoliths, as the research in this field is poorly developed. Next, we plan to collect young-of-the-year Pacific cod from multiple waters, preferably containing both pre-settlement and post-settlement individuals. The daily increment will be used to accurately determine the time of formation of each check and to make inter-regional comparisons.

Effect of phenotypic induction early on later phenotypes

Pacific cod undergo a series of stages during their early life history, including from the egg to yolk-sac larva, and larva to the pelagic juvenile. Fish have long been recognised as being very biologically sensitive during these transitional phases, and may be subject to massive mortality (Cushing and Horwood 1994; Marchetti 1965). There is increasing evidence that environmental conditions early in ontogeny has a significant influence on the subsequent development of fish species (Pechenik et al. 1998; Relyea and Hoverman 2003). The earlier the environmental disturbance, the stronger its likely long-term effects; this is termed the 'silver spoon effects' (Jonsson and Jonsson 2014; Lindström 1999). Larvae are subject to variation in the quantity and quality of food items available (McCormick and Molony 1992). The previous study demonstrates convincingly that even brief periods of poor diet have persistent and lingering influences on subsequent post-metamorphic growth rates, energy stores and size at settlement (Wacker and von Elert 2002). The negative impact of AGC size on check 1, now considered the settlement check (Fig. 3), is consistent with the conclusion of Jonsson and Jonsson (1993) that it is an increased early growth rate that leads to earlier age-at-life-stage transitions. This provides further proof that settlement is a function of body length. The faster the growth of the individual at an early stage, the sooner the appropriate body length at the time of settlement will be reached. The earlier the settlement the better the survival of age-0 Atlantic cod (Geissinger et al. 2021). As high temperature mediates the movement of Pacific cod to deeper waters after settlement, earlier settlers spend more time in an environment that is favourable for growth that results in a greater amount of absolute growth than later settlers. This is consistent with our results showing that AGC size was positively correlated with Check 2 growth (Fig. 3). Individuals enter deeper water with greater body length variability (Fig. 2D). The size advantage may be translated into a competitive advantage, especially in the bottom environment where nutrition is scarce (Bertness 1989; McCormick and Molony 1992). Large individuals potentially swim faster, have more energy reserves and achieve greater absolute growth at age one. The likelihood of broad dispersal within the life cycle creates the potential that local production of larvae is decoupled from subsequent recruitment of juveniles (Phillips 2002), as occurs for example in the intertidal Cocos frillgoby Bathygobius cocosensis (Thia et al. 2018). However, the influence of early growth variation on each subsequent stage revealed in this study shows that the conditions Pacific cod encounter during their early life history may leave lasting effects on growth rate throughout the life of the fish. Differences in growth may be dependent on the environmental conditions in the habitat during later life stages (Moran and Emlet 2001). When environmental conditions in the benthic habitat are favourable, early life conditions may have a minor effect on subsequent growth. Conversely, the difference in impact on high-quality and low-quality individuals will be obvious. Low temperature, low oxygen content and lack of food in the bottom layer may cause the body length variation of Pacific cod to increase in later life stages.

We found no or moderate evidence that AGC size had a significant effect on the diameter of otoliths at the end of each subsequent stage (Check 1 size, Check 2 size and First annual ring size), similar to previous studies that the growth in the early stages leads to differences in individual size at settlement (McCormick and Molony 1992; Wacker and von Elert 2002). However, our result did show a slight variation in AGC size and gradually increasing differences in the three following otolith indicators (Fig. 2D). Delaying any of the transitions may increase mortality or reduce the ability of juveniles to compete successfully for space or food (Pechenik et al. 1993; Wendt 1996). A range in settlement times during this period results in multiple size classes for age-0 Pacific cod entering winter, creating a size-structured population from fall to spring. Differences among individuals in growth compensation may be related to whether energy reserves have been replenished at the end of the early life history. If no or reduced replenishment of internal energy stores takes place, the growth capacity of individuals will vary over time (Wacker and von Elert 2002). Analysis indicated that the differences in fish body length at the end of each stage arise strongly from differences in growth during this stage, which is controlled by the growth at the early stage. Therefore, the effect of early life history stages on subsequent body length mainly by indirectly mediate the growth during each life stage.

Materials and methods

Fish collection and otolith preparation

Pacific cod were collected using bottom trawls at 200–350 m depth in the Sea of Japan in May 2015 and off the Pacific coast of Hokkaido in June 2021 (Fig. 4). All specimens were conducted in accordance with the guidelines and approval of the respective Animal Research and Ethics Committees of Ocean University of China (Permit Number: 20141201. http://www.gov.cn/gongbao/content/2011/content_1860757.htm). Fish lengths (total and standard lengths, to the nearest 1 mm) and weights (total weight, to the nearest 1 g) were measured on board. The left and right sagittal otoliths were excised, cleaned, sonicated, and dried under clean conditions, then stored in plastic tubes. As this research focused on the life history of first year Pacific cod, individuals with body lengths less than 400 mm (Table 1) were selected, referring to the Resource Assessment Report of the Fisheries Agency of Japan (http://abchan.fra.go.jp/index1.html) for subsequent analysis. One otolith selected at random from each fish was placed in a mould and embedded in epoxy resin (#105; Struers AS, Tokyo, Japan). A rectangular parallelepiped block was cut along the transverse section using a diamond circular saw (Leica sp1600; Leica Microsystems GmbH, Wetzlar, Germany), with the otolith cutting position determined by the obvious protrusions on the distal plane and the collum on the proximal plane (Supplementary Fig. S1). Microscopic examination of the block showed that irrespective of the side, the plane area containing the otolith core was the largest. Otolith sections were fixed to glass slides using thermoplastic glue, with the side furthest from the core facing upwards to avoid over-grinding, and were ground using a series of waterproof sandpaper (#600–4000). When the accessory growth centre (AGC) became evident, the otolith was removed from the slide, turned over, remounted, and then subsequently ground until the primordium became clearly visible. To avoid missing checks and misinterpreting the age, we ensured the otolith was polished to the primordium.

Figure  4.  Sampling locations for Pacific cod in Pacific Ocean off Hokkaido (A) and the Sea of Japan (B) marked with red dots. Bathymetric contours within the sampling depth range (0–350 m) was displayed in the figure

DownLoad: Full-Size Img PowerPoint

Table  1.  Fish information for otolith microstructure analysis

Region	Capture season	Age	Hatch year	N	Standard length (mm)
Sea of Japan	May 2015	1+	2014	14	210 ± 32
		2+	2013	14	277 ± 26
Pacific Ocean off Hokkaido	June 2021	1+	2020	47	340 ± 40
N = number of specimens per hatch year. Standard length = mean ± standard error

 | Show Table

DownLoad: CSV

Photographs of entire otoliths were taken at 25× magnification under reflected light, and the core area was photographed at 100× magnification under transmitted light using a Leica DM2500 microscope (Leica, Germany) and a Radeon R5 430 imaging system. The opaque zones visible under reflected light identified the annual rings of Pacific cod (Hattori et al. 1992). Coupled with the body length data and the microstructure characteristics of otoliths, 75 age one and two individuals were selected for subsequent analysis.

Otolith microstructure in the first year

The axis running straight up the middle of the otolith is preferred for age determination (Mahé et al. 2012). This was available for all samples in our study; therefore, the anti-sulcus served as the trajectory for counting and measuring distance (Fig. 1). A linear regression analysis between the standard length (SL) of the fish (N = 75) and the distance from the primordium to the edge of the anti-sulcus of the otolith was performed to validate that otolith length was a proxy for fish length (Supplementary Fig. S2; R² = 0.6).

On the otoliths of all samples, two distinct checks (check 1—① and check 2—②) were observed inside the first annual ring (Fig. 1; Supplementary Fig. S3); these were evident as a narrower band closer to the core and a relatively wider dark zone further from the primordium. The change in the direction of radiation for daily increments formed an AGC around the core. These otolith microstructures acted as a proxy for a transition in the life history of the fish, and four nodes (AGC, check 1, check 2 and the first annual ring) were used for stage analysis based on these otolith traits. We defined four size indicators (AGC size, Check 1 size, Check 2 size and First annual ring size) to characterise the body length of fish when these features formed on the otoliths. The distance from the primordium to each node reflected these four indicators and revealed the overall impact of growth up to each stage (Table 2). They have been used to indicate habitat suitability for this life history stage and predator-mediated selection on juvenile fish (Holmes and McCormick 2006; Vasconcelos et al. 2013). We also measured the distance between each two adjacent nodes on the otolith and constructed four growth indicators. The distance from the primordium to the AGC simultaneously reflects the body length and growth at this stage. To avoid repetition, we only used this distance as a growth indicator for subsequent analysis to reveal the individual development from hatching (or maternal investment) to metamorphosis (Narimatsu et al. 2007). Although some studies have suggested that the settlement of Pacific cod and its subsequent movement to deeper water layers may cause the formation of checks on the otoliths (Li et al. 2021; Takatsu et al. 2001; Wu et al. 2021), the specific reasons for the formation of check 1 and check 2 remain uncertain. Therefore, we did not define what life history events correspond to them here (see "Discussion").

Table  2.  Otolith microstructure indicators along the life-history of the first year for Pacific cod

Indicator (μm)	Description	Code	CV
Accessory growth centre size	The distance between the primordium and the accessory growth centre	P-AGC	0.129
Check 1 growth	The distance between the accessory growth centre and the check 1	AGC-①	0.248
Check 1 size	The distance between the primordium and the check 1	P-①	0.131
Check 2 growth	The distance between the check 1 and the check 2	①–②	0.271
Check 2 size	The distance between the primordium and the check 2	P-②	0.153
First annual ring growth	The distance between the check 2 and the first annual ring	②–③	0.316
First annual ring size	The distance between the primordium and the first annual ring	P-③	0.173
The code refers to the description in Fig.. The coefficient of variation (CV) for each indicator was summarized (N = 75)

 | Show Table

DownLoad: CSV

Temporal variation in indicators

As the collection time for the Hokkaido and Sea of Japan samples did not overlap, we used only the Sea of Japan samples for inter-year comparison of otolith indicators. The corresponding incubation years for 1- and 2-year-old individuals were 2014 and 2013, respectively. The effect of year on the otolith indicator was analysed using F values from a one-way ANOVA.

Interaction of indicators across life history stages

Path analysis is an extension of the regression model used to analyse the causal relationships among more complex variables (Lleras 2005). A variable can be either a cause variable or an outcome variable. A variable can directly affect another variable or indirectly affect it by affecting a third variable. Path analysis is used to sort out the direct and indirect influences involved (Lleras 2005). We used path analysis to explore how phenotypic induction early in the ontogeny of Pacific cod affects later phenotypes. In this study, for example, 'AGC' may either directly affect 'Check 2 growth' or indirectly affect 'Check 2 growth' by affecting 'Check 1 growth', so the impact of 'AGC' on 'Check 2 growth' consists of two parts, namely 'β_dir' and 'β_ind', which together form 'β_tot'. The same is true for the later life history traits. Correlations among the seven otolith microstructure indicators were tested (Supplementary Fig. S4 and Table S3), and three of the seven that showed auto-correlation with other factors were eliminated. The remaining four growth indicators were included as variables in the path analysis (Supplementary Table S4). How the growth of each stage affects the body length of the fish at the end of the stage was determined by: (1) the correlation between growth and matching size indicators; and (2) the relationship between size and life stage.

Path analysis was undertaken using the R package plspm. The four growth indicators were imported and defined as exogenous (Accessory growth centre size) and endogenous (Check 1 growth, Check 2 growth and First annual ring growth) type. Each indicator was assigned to its respective 'block' (latent factor) and standardised (mean = 0, variance = 1) using the scaled argument in plspm (Thia et al. 2018). Permutation and combination produced a total of six paths. The 95% confidence interval (2.5 and 97.5 percentiles) and the indicator effect of 1000 bootstrap replicates were generated using the boot.val argument in plspm. Indicators that covaried non-significantly were evaluated as to whether the bootstrap confidence interval overlapped with zero. The outer model of the four indicators (including the total explained variation, R²) and the inner model in pairs were exported. The direct and indirect effects on each path and the confidence interval were used for visual representation in figures.

Four models were constructed, based on the three hatch years individually and the pooled data. Differences in the bootstrap means of β_DIR or β_TOT between pairs of hatch years for each pathway were tested using Welch's t test (Thia et al. 2018; Welch 1947, 1951). Taking into account the number of samples in each hatch year (especially for 2013 and 2014), we combined all samples if the difference of β_DIR or β_TOT between pairs of hatch years was not significant for subsequent analysis.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1007/s42995-022-00145-y.

Acknowledgements

This work was financially supported by China Scholarship Council (202006330094).

Author contributions

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

Data availability

All data generated or analysed during this study are included in this published article (and its supplementary information files).

Declarations

Conflict of interest

Yongjun Tian is the Editorial Board Member of the journal, but 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 accordance with the guidelines and approval of the respective Animal Research and Ethics Committees of Ocean University of China (Permit Number: 20141201), and does not contain any studies with human participants.