Climate‐induced habitat suitability changes intensify fishing impacts on the life history of large yellow croaker (Larimichthys crocea)

Abstract Intense fishing pressure and climate change are major threats to the fish population and coastal fisheries. Larimichthys crocea (large yellow croaker) is a long‐lived fish, which performs seasonal migrations from its spawning and nursery grounds along the coast of the East China Sea (ECS) to overwintering grounds offshore. This study used length‐based analysis and habitat suitability index (HSI) model to evaluate the current life‐history parameters and overwintering habitat suitability of L. crocea, respectively. We compared recent (2019) and historical (1971–1982) life‐history parameters and overwintering HSI to analyze the fishing pressure and climate change effects on the overall population and overwintering phase of L. crocea. The length‐based analysis indicated serious overfishing of L. crocea, characterized by reduced catch, size truncation, constrained distribution, and advanced maturation causing a recruitment bottleneck. The overwintering HSI modeling results indicated that climate change has led to decreased sea surface temperature during L. crocea overwintering phase over the last half‐century, which in turn led to area decrease and an offshore‐oriented shifting of optimal overwintering habitat of L. crocea. The fishing‐caused size truncation may have constrained the migratory ability, and distribution of L. crocea subsequently led to the mismatch of the optimal overwintering habitat against climate change background, namely habitat bottleneck. Hence, while heavy fishing was the major cause of L. crocea collapse, climate‐induced overwintering habitat suitability may have intensified the fishery collapse of L. crocea population. It is important for management to consider both overfishing and climate change issues when developing stock enhancement activities and policy regulations, particularly for migratory long‐lived fish that share a similar life history to L. crocea. Combined with China's current restocking and stock enhancement initiatives, we propose recommendations for the future restocking of L. crocea in China.

tion, constrained distribution, and advanced maturation causing a recruitment bottleneck. The overwintering HSI modeling results indicated that climate change has led to decreased sea surface temperature during L. crocea overwintering phase over the last half-century, which in turn led to area decrease and an offshore-oriented shifting of optimal overwintering habitat of L. crocea. The fishing-caused size truncation may have constrained the migratory ability, and distribution of L. crocea subsequently led to the mismatch of the optimal overwintering habitat against climate change background, namely habitat bottleneck. Hence, while heavy fishing was the major cause of L. crocea collapse, climate-induced overwintering habitat suitability may have intensified the fishery collapse of L. crocea population. It is important for management to consider both overfishing and climate change issues when developing stock enhancement activities and policy regulations, particularly for migratory long-lived fish that share a similar life history to L. crocea. Combined with China's current restocking and stock enhancement initiatives, we propose recommendations for the future restocking of L. crocea in China.

| INTRODUC TI ON
Globally, heavily fishing activities and climate change are rapidly reducing the abundance of many marine organisms and increasing the likelihood of species extinction (Burgess et al., 2013;Cinner et al., 2012;Hoegh-Guldberg & Bruno, 2010;Payne et al., 2016).
For instance, intensive fishing and climate change have caused overfishing and declined catches in Canada, Iceland, and China (Du et al., 2014;Liang & Pauly, 2017;Pauly et al., 2001). Previous studies showed that fishing pressures and climate change can affect (i) the life-history strategy of individuals, via impacts on physiology, morphology, and behavior (Ba et al., 2016;Olafsdottir et al., 2016); (ii) the population dynamics, via changes to key population processes throughout an organism's life history and habitat suitability (Perry et al., 2005). Hence, bottlenecks of any life-history stage (e.g., spawning, hatching, larval survival, recruitment settlement, growth, and adult survival) and habitat suitability can cause overfishing of exploited species. In this context, recruitment bottleneck and habitat bottleneck are most well documented (Almany & Webster, 2006;Caddy, 2011). Correspondingly, the potential cause of overfishing is mismanagement because of a poor understanding of recruitment bottleneck and habitat bottleneck that constrain the productivity of the overall population.
Fishing alters the size structure by removing large fish exacerbated by size-selective gear. Heavy fishing can diminish the ability of fish to reproduce (recruitment overfishing) and/or constrain the overall recruitment ability before they can fully realize their growth potential (growth overfishing) (Diekert, 2012) via size truncation effect (STE) (Berkeley et al., 2004;Froese et al., 2008;Langangen et al., 2019;Ottersen et al., 2006). This effect states that population shifts with decreasing body sizes and advancing maturation characteristic of the life-history changes induced by fishing (Anderson et al., 2008;Bell et al., 2015;Berkeley et al., 2004). Hence, fishing for juveniles and mega-spawner can weaken the reproductive potential of fish stock, called "recruitment bottleneck" (Doherty et al., 2004).
Such bottlenecks are visible in long-term time series and are a common cause of the collapse in intensely fished stocks, for example, in Western cod, Pacific rockfish, and North Sea ground fish (Froese et al., 2008;Harvey et al., 2006;Poulsen et al., 2007).
Climate change-caused environmental conditions shift can have negative effects on the fish population (Graham et al., 2011;Johnson et al., 2011). In general, species' distribution patterns are relative to both life-history strategies (Anderson et al., 2013) and physiology tolerance on environmental variables, such as sea surface temperature (SST), chlorophyll-a concentration (Chl-a), sea surface salinity (SSS), and currents (Guan et al., 2013;Yu & Chen, 2018).
The environmental shift can selectively affect the habitat suitability of target species (Farrell et al., 2008). Lower habitat suitability of any life-history stage can lead to species-specific "habitat bottleneck" and later can have large consequences for loss of several fish's climatically suitable habitat, for example, Norwegian herring, Maine cod, and Mid-Atlantic Bight winter flounder (Bell et al., 2015;Pershing et al., 2015).

Heavy fishing activities and shifts in environmental conditions
can have combined effects on fishery collapse, especially for longlived species (Gascuel et al., 2014;Hsieh et al., 2009;Rose, 2004).
Specifically, some studies suggested that long-lived species are expected to have a slower demographic response to climate change (Berteaux et al., 2004;Wilson et al., 2010). Additionally, fishingcaused STE can exacerbate long-lived fish degradation via diminishing "bet-hedging" capacity, including the ability to migrate and avoid poor areas, having flexibility in spawning times and locations, and production of high-quality offspring that survive in a broader suite of environmental conditions, for adapting to rapid climate change (Bell et al., 2015). However, no example exists that demonstrates the STE and the climate-induced effects on long-lived migratory fish in the most heavily fishing (and minimally managed) marine ecosystem in the world: the East China Sea (ECS) (Szuwalski et al., 2017). To fill the knowledge gaps, we require a species that: first, under intensive fishing pressure; second, has specific habitat requirements; third, the habitat of which is affected by rapid climate-induced habitat suitability variation; and fourth, has been reliably assessed over a long period by field surveys.
In the following, we provide an appropriate example by discussing changes in the specific population dynamic of an overexploited, long-lived, migratory fish in the ECS, the large yellow croaker (Larimichthys crocea). The collapse of L. crocea represents an interesting example to explore both heavy fishing and climate change on the overall population: first, L. crocea ranked top among the four major marine economical fishes in China in the last century (Zhang et al., 2010) but suffered collapse since the 1980s. The latest International Union for Conservation of Nature (IUCN) Red List of Threatened Species labeled L. crocea as "critically endangered (CR)" (Liu et al., 2020). Second, L. crocea is a long-lived species with maximum age of 21 years in the 1960s . Accompanied by population collapse, the L. crocea population in the ECS is characterized by decreased maximum age and body size, and advanced maturation (Ye et al., 2012). Third, L. crocea is a migratory fish that conduct climatic migrations (e.g., movements driven by physiological tolerances of individuals to environmental factors such as temperature or salinity) and gametic migrations (e.g. movements that increase the reproductive success of individuals by promoting gonad K E Y W O R D S climate change, East China Sea, HSI model, Larimichthys crocea, length-based analysis, overfishing

T A X O N O M Y C L A S S I F I C A T I O N
Biogeography development, increasing sexual encounter rates, or increasing the survival of offspring) between offshore water and coastal water during autumn-winter and spring-summer, respectively ( Figure 1a).

| MATERIAL S AND ME THODS
In this study, we only considered L. crocea in the mid-southern ECS (120°E to 126°E, 25°N to 29°N) that have available data over longterm series. Also, we evaluated only overwintering distribution patterns and overwintering habitat suitability because they are strongly linked to the physiology fitness and survival rate during the juvenile and adult stages and corresponding climatic migration phase of L. crocea.

| Fishery data
To investigate how L. crocea population declines in the ECS in the last five decades, we analyzed two datasets on L. crocea in the ECS.

| Life-history parameters of L. crocea in the ECS
To understand the life-history parameters of L. crocea in the study area, we analyzed length frequency data with the electronic length F I G U R E 1 (a) Life-history migration patterns of Larimichthys crocea in the ECS, e.g., L. crocea spawns inshore and "overwinters" offshore; and (b) the fishing area (gray area) during winter 1971-1982 and the survey stations (crosses) in the winter of 2018 for L. crocea in the mid-southern East China Sea. frequency analysis (ELEFAN) approach using the Tropfishr package . Size frequency of the commercial fishing catch during 1970-1982 is not available but we gathered the life-history information from published literature (Liu & de Mitcheson, 2008;Xu et al., 1984aXu et al., , 1984bYe et al., 2012;Yu & Lin, 1980) coinciding in time and space with the available data. in Tropfishr package.
We fit the von Bertalanffy growth function (VBGF) through the length frequency and life-history data to estimate life-history parameters (e.g., maximum length, weight at length, length at maturity (L mat ), and VGBF parameters including von Bertalanffy growth constant (K)) and asymptotic length (L inf ), age at zero length (t 0 ) (Brey & Pauly, 1986;Pauly & David, 1981;Sparre & Venema, 1998), and estimated mortality parameters (e.g., total mortality (Z), natural mortality (M), and fishing mortality (F)) (see details in Supporting Information). Specifically, the VBGF and mortality parameters were estimated following four ap-

| L. crocea overwintering distribution patterns and overwintering habitat suitability in the ECS
L. crocea is overall a habitat specialist during overwintering phase.
Previous studies have shown that L. crocea has strong depth, temperature, and salinity preferences, while pH, dissolved oxygen, light intensity, sound, water velocity, and other factors may affect its distribution pattern, survival, and growth at different life stages (Liu, 2013;Wang et al., 2016Wang et al., , 2017. Unfortunately, detailed data on seasonal environmental data were limited between 1971 and 1982. Hence, only depth, SST, and SSS data were available as model inputs to determine suitable habitats for L. crocea. We used the HSI model to predict the overwintering habitat suitability of L. crocea in the ECS, which is a type of species distribution model (SDM) used for evaluating organisms-habitat relationships based on limited data or expert knowledge. In our HSI model, we used standardized abundance, HSI, as the response variable, and three environmental variables with the strongest correlation and the best data availability, depth, SST, and SSS as predictors. Firstly, we constructed both fitting-based (Hua et al., 2020;Lee et al., 2019;Yu et al., 2020) and regression-based (Chang et al., 2018;Jin et al., 2020) suitability index (SI) models to describe the relationship between each environmental variable and L. crocea abundance (Supporting Information). Then, we combined two types of SI models into HSI models, respectively. For each type of SI model, we used two empirical HSI models: the arithmetic mean model and the geometric mean model ( Figure S1), under different environmental variable combinations (Lee et al., 2019).
For model validation and selection, we used catch data (abundance) from 1971 to 1980 and corresponding environmental data as training dataset, catch data from 1981 to 1982, and corresponding environmental data as testing dataset. We assumed a positive linear relationship between predicted HSI values and L. crocea abundance and evaluated the goodness of fit of the above relationship for each HSI model based on R 2 and the Akaike information criterion value adjusted for small sample size (AIC c ) (Chang et al., 2013). A fitting-based arithmetic mean model with two variables (e.g., depth and SST) yielded the maximum R 2 and the minimum AIC c value (Supporting Information), thus was selected as the final HSI model.
Correspondingly, fitting-based SI model was selected as the best SI model. Finally, we retrained the SI model (see parameters and statistical test results in Supporting Information) and the HSI model using catch data (abundance) from 1971 to 1982 and corresponding environmental data.

| Catch, hatchery release, and life-history parameters of L. crocea in the mid-southern ECS
Based on the reported landing information, the overall production of L. crocea in the ECS has been continuously declining since the 1970s. The variation in landing data indicated that the recent annual catch is now less than 4400 tons, which has declined by >90%, The life-history parameter results show that since the 1980s, using length-based data, we fit a series of ELEFAN models and most of the workflows were within feasible ranges for data-limited measurement (Hordyk et al., 2015). The best model (ELEFAN S.A. with bin = 10, MA = 11, see details in Table S3) Table 1). Still, the fishing mortality (F) and exploitation rate (E = F/Z) were predicted to be 1.57 and 0.84, respectively, which are continuously increasing compared with the 1980s ( Table 1).

| Overwintering habitat suitability of L. crocea in the ECS
We used both fitting-based and regression-based methods to construct SI models of each environmental variable and employed both the arithmetic mean model and geometric mean model under different environmental variable combinations to calculate HSI values. A fittingbased arithmetic mean model with two variables (e.g., depth and SST) yielded the maximum R 2 and the minimum AIC c value (Supporting Information), thus was selected as the final HSI model. The statistical analysis of fitting-based SI models (Supporting Information) shows they were all significant (p < .05). As shown by the SI curves ( Figure S1), the optimal range for depth, SST, and SSS during winter in our study area was 36-72 m, 18.2-20.5°C, and 33.89-34.27, respectively. The recent five decades' cooling trend in winter is remarkable, with the reduced average SST (−0.028°C/year, R 2 = .31, p < .05) between 1982 and 2019 in the mid-southern ECS in winter ( Figure 4). The cooling trend in our study area may be influenced by the Kurushio extension; specifically, in the latest IPCC report (2019), the Kurushio extension exhibits long-term cooling, which is consistent with our result. Also, another study revealed the cooling trend along China's and Japan's coast (−0.69 ± 0.44°C/ decade), opposing the overarching global warming trend, especially in the winter season due to the extreme cold events (Bindoff et al., 2019;Liao et al., 2015). Consisted with the cooling trend of SST during the overwintering phase of L crocea, the results of HSI models show the mean habitat overwintering suitability of the 1970s (1971-1980), 1980s (1981-1990), 1990s (1991-2000), 2000s (2001-2010), and 2010s (2011-2019) shifting in our study area. Figure 5a shows that there was no significant change (p > .05) in the average and optimal habitat area from the 1970s to 1990s.
However, the percentage of optimal habitat decreased significantly (p < .05) from 13%, 12%, and 13% in the 1970s, 1980s, and 1990s to 4% and 5% in the 2000s and 2010s. Figure 5b shows that the spatial distribution of habitat suitability also changed: the optimal area has moved toward a southeast direction, with suitable habitats becoming offshore oriented. Unfortunately, regarding data availability, the HSI models conducted in our study may be biased because we used catch data during 1971-1982 as a measure of abundance (e.g., highly dependent on the effort). Hence, abundance data obtained from the scientific cruise is more convincing than using catch data as abundance data and should be encouraged in future studies.

| Fishing-induced life-history variation
Overall, this study provides evidence of serious fishing-induced life-history variation in L. crocea population and represents a glimpse of fishery collapse. The observed life-history parameters show that the body size of L. crocea has on average, decreased during the last five decades ( Table 1). Previous studies in the 1970s and 1980s revealed that the main catch of L. crocea consisted of 2or 3-year-old (400-800 g) individuals (Yu & Lin, 1980), while 95% of catches in this study were individuals aged 0 or 1 year. Still, we indeed find that the maturation proportion of L. crocea is dramatically declined during the last five decades (Figure 2b). It is likely that the decrease in average body size, size truncation, and matu-  Fishing-induced life-history variation may also constrain species distribution because the migratory ability of a species is strongly dependent on dispersal characteristics, such as morphological traits (Hsieh et al., 2009). To substantiate our finding on the potential negative effect of life-history variation on the overall population, we compared the historical and recent distribution of L. crocea. The result showed that ~70% of potential catch areas have disappeared ( Figure S2), with the highest disappearance rate in offshore areas (122°E-125°E), which follows the life-history variation in L. crocea during the last five decades. Consequently, constriction of geographic distribution, associated with a decline in body size, may reduce the ability to respond to climatic stress, by limiting movement (Reusch et al., 2005).

| Climate-induced overwintering habitat degradation may intensify the effect of overfishing
The best HSI model (next best model: ∆AIC c = 2, for other models, Table S2) to explain catch patterns under unexploited status over time included SST and depth. The result of SI suggested that the optimal overwintering temperature range (18.2-20.5°C) and depth (36-72 m) of L. crocea mirror previous lab-based observation of optimal growth temperature (17-24°C) and empirical observation of optimal overwintering depth (50-60 m) (Liu, 2013;Xu & Chen, 2011). It is worth noting that SST in the mid-southern ECS has decreased by an average of 1°C between the 1980s and 2010s with an annual decrease in SST rate -0.028°C/year (Figure 4). HSI variation results in the last five decades suggest that the cooling trend of SST in the ECS has significantly reduced the proportion of optimal and average overwintering habitats for L. crocea ( Figure 5b).
Consequently, in response to the SST decrease in winter, migratory species, like L. crocea, are expected to respond in two ways as follows. Generally, marine organisms respond to climate change through shifts in distribution (Guisan & Thuiller, 2005). For instance, in the North Sea, both exploited and unexploited fish species have shifted to higher latitudes and deeper water between 1977 and 2001 in response to rising sea temperatures (Perry et al., 2005); in the Eastern Tropical Pacific, demersal species were projected to move into shallow water by the mid-21st century in response to high greenhouse gas emissions (Representative Concentration Pathways, RCP8.5) and strong migration (RCP2.6) scenarios (Clarke et al., 2020). Alternatively, it is also commonly observed that species stay in poor habitats against climate change but suffered climate-induced life-history variation. Particularly, it occurs when marine fishes are living in temperatures outside their physiology optima: this results in reduced aerobic scope, which negatively affects their growth and reproduction (Pearson & Dawson, 2003;Pörtner & Knust, 2007;Toresen et al., 2019).
Hence, species like L. crocea must either migrate to remain within a suitable habitat or suffer the consequences (Bell et al., 2015).
Interestingly, L. crocea overwintering distribution pattern did not shift alongside a decrease in winter SST, which is a good indicator that temperature per se did not explain the overall shift of L.
crocea distribution. Such absence of a clear systematic impact of temperature may be due to the life-history parameters degradation, which could constrain the hedging capacity against climate change (Thorson et al., 2017). For example, STE caused a change in the length age structure is the main diver of interannual shifts in summer flounder distribution, while temperature had little influence on the change in distribution (Bell et al., 2015).
More broadly, the "match/mismatch hypotheses" may explain the combined effects of heavy fishing and climate change on the decrease in the overall population (Cushing, 1990;Edwards & Richardson, 2004). Based on our study, we suggest that fishinginduced life-history variation leads to the "mismatch" of L. crocea optimal overwintering habitat. Specifically, our study demonstrated that L. crocea has both life-history variation and size truncation compared with the 1980s, with significantly smaller body size and advanced maturation (Figure 2b). This truncation in overall size structure can significantly affect swimming ability, such as reducing the sustained swimming time and average swimming speed, namely size dependent, consequently reducing the distribution range of L. crocea (Jorgensen et al., 2008;Opdal & Jørgensen, 2015. Given the climate-induced changes in overwintering habitat suitability that occurs in the mid-southern ECS, the fishing-induced life-history population variation that constrains dispersal capability could pose a significant "mismatch" of optimal overwintering habitat to L. crocea-like migratory species (Figure 6). Such applicability of "mismatch hypotheses" to the specific long-lived migratory fish exposed to fishing and climate change had rarely been demonstrated.

| Management implication
In China, approximately 50% of China's fish stocks have been overexploited or collapsed (Cao et al., 2017). Our approach can provide insight in anticipation of stock enhancement and management that may facilitate conservation and re-stocking. While our results highlight that, as well as the previously described "lack of timely, effective or sufficient management, combined with heavy fishing pressure, particularly at spawning and overwintering grounds were major factors responsible for croaker stock declines" (Liu & de Mitcheson, 2008), climate change-induced overwintering habitat is another potential reason for the stock depletion. This is highly worrying because long-lived migratory fish like L. crocea decline even faster were both heavy fishing and climate-induced habitat suitability synergies (Färber et al., 2018).
The severe situation has led to an urgent need to reevaluate fishery management and calls for a species-specific F I G U R E 5 Decadal variations in (a) spatial distribution of predicted habitat suitability and (b) area percentage of optimal, average, and poor habitat since the 1970s. The areas with habitat suitability index (HSI) value >0.7, 0.7 > HSI value >0.3, and HSI value <0.3 were regarded as optimal, average, and poor habitat, respectively.
or life-history-based approach to stock enhancement (Dubik et al., 2019;Lotze et al., 2011;Pinsky et al., 2018;Young et al., 2006).  (Wilson et al., 2008). Also, fishery managers often deploy hatchery releases to address the recruitment bottleneck of species' restocking (Kitada, 2018;Myers et al., 2004;. Because L. crocea's suitable overwintering habitats have shifted toward offshore areas, to tackle both recruitment and habitat bottleneck, we recommend that stakeholders choose larger juveniles, even megaspawner for hatchery release to keep pace with the shifting of suitable habitats caused by climate change.

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

DATA AVA I L A B I L I T Y S TAT E M E N T
The data generated for this study are available at Dryad Data Repository: https://doi.org/10.5061/dryad.08kpr r538.