Prey Abundance Regulating the Response of an Important Marine Eel Species Conger Myriaster Growth to Water Temperature

Whitespotted conger (Conger myriaster) is a commercially important species in East Asia but the sheries stock has drastically declined in recent years. Environmental changes are assumed to have profound impacts on the growth pattern of this species, yet the mechanisms that regulate growth remain poorly understood. Here we used otolith measurements to establish growth chronologies of 9-year period (2010–2018) for whitespotted conger in the Yellow Sea and evaluated the effects of environmental variables on the growth pattern. Linear mixed-effects models were used to explain growth variation with abiotic and biotic environmental variables, including seasonal water temperature, prey availability, and population density, and to assess age-dependent responses in growth and the interactions between abiotic and biotic factors. The results indicated that the growth of whitespotted conger positively correlated with spring sea bottom temperature (March-May) and prey per capita, and the inuence of prey availability became larger with increasing ages. The model detected signicant negative interactions between sea temperature and prey availability, indicating a higher degree of temperature-dependent growth when prey availability was low. Our ndings imply that the growth of whitespotted conger is less likely to be inuenced by food availability when the spring water temperature is high in the Yellow Sea, and a combination of cold spring and low food availability may result in substantial constrained growth. This study provided the rst evidence for the joint effects of abiotic and biotic factors on the growth variation of whitespotted conger, and the information may contribute to sustainable management policies. on conger This provided the rst evidence for the joint effects of abiotic and biotic factors on the growth variation of whitespotted conger.


Introduction
Whitespotted conger Conger myriaster (Brevoort 1856) is an eel species distributed in the coastal to deep waters around China, Korean Peninsula and Japan (Tokimura 2011). This population, which supports an important shery shared in East Asia, has been exploited mainly in the Yellow Sea and the East China Sea. As an economically important sh species, the population has been under intensive shing pressure and drastically declined in recent decades (Gorie and Tanda 2004;FAO 2016). Although there are increasing concerns about the population declines and many researchers and managers have been taking actions, restoring the population is challenging due to limited understanding of their complex life histories, which is typical for eel species. In addition, as coastal areas are characterized by dynamic and complicated environments driven by climate changes (Harley et al. 2006) and human activities (Selkoe et al. 2008), many biological processes of whitespotted conger may be largely impacted, making management of this species even more challenging.
Growth is one of the most important factors determining population dynamics, and modeling of body growth constitutes an essential part of sheries stock assessments (Lorenzen 2016). Currently, most sheries assessments models assume that the growth parameters are consistent for given species (Lorenzen, 2016), although many studies have shown that growth patterns actually change over time (e.g., Morrongiello et al. 2014), in response to environmental factors, e.g. temperature (e.g. Heather et al. 2018; Lee and Punt 2018) and food availability (Weatherley 1990). Ignoring the variation in growth may result in errors in stock assessment and misleading in sheries management, given the prevalent in uence of global climate changes (Rahikainen and Stephenson 2004).
Given the complex life history of this species, whitespotted conger is likely to be more vulnerable to the changes in environmental conditions. Growth of whitespotted conger has been investigated in several geographical areas (Katayama et  In this study, we used inter-annual measure of otolith increments as a proxy of somatic growth of whitespotted conger and adopted mixed-effects models to analyze individual growth trajectory regarding the in uence of temperature and population density on individual growth. We also evaluate prey availability effects by relating growth rates to the prey biomass data, as well as to assess age-dependent responses in growth and the interactions between prey availability and water temperature. This study aims to demonstrate the growth pattern of whitespotted conger in China Sea and evaluate the effects of the abiotic and biotic factors that regulate the growth of this species. The improved understanding of the growth patterns may provide critical information for responding to climate changes and contribute to management policies for sustainable sheries.  (Table 1). A total of 532 individuals were randomly selected from the catch, and the total length and body weight were measured and their sagittal otoliths were removed and cleaned for further analyses. Males were rarely captured in our surveys and all females and unsexed individuals were used in this study.

Otolith measurements
The otoliths were embedded in epoxy resin (Epo x; Struers, Copenhagen, Denmark), then sectioned along the dorsal-ventral axis across the core with a diamond circular saw (Isomet low speed, Buehler). The sections (2 mm) were mounted on glass slides and polished with 800-1200 grit grinding paper (Minimet 1000 grinder-polisher, Buehler) until a clear view of the primordium was evident, and then pictured by optimal microscope (Olympus bx51). Sections with unclear annulus were discarded for subsequent age determination and measurement. We estimated age by counting annuli on the otolith images, where individual ages were represented as numbers of complete annuli. Ages were read twice from the otolith images and re-read after half a month, and the results that differed in two reading were then reexamination by another well-experienced reader. The birth-date of was assumed to be 1, December according to literature (Lee and Byun 1996). Using the 'list year' technique by subtracting the age from the collection year, we aligned each annual increment into the appropriate calendar year of the formation (Yamaguchi 1991). Then, annuli were measured from the nucleus to the edge along the axis of maximum growth, which was perpendicular to growth-increment boundaries.
Increment width series were constructed using ImageJ image analysis software (Rasband 1997(Rasband -2012. The rst increments (birth year) were removed because the increment width mainly depended on the birth date rather than the corresponding annual environmental variables. Marginal increments were also removed from analysis because they did not comprise a complete year of growth. In total, 264 otolith samples were used to establish the growth chronologies of whitespotted conger ( Supplementary Fig. S1).
Some samples were excluded because of the low annulus clarity (N = 167) or small age (N = 101). Additional analyses were conducted to ensure the increment of otolith was proportional to somatic growth ( Supplementary Fig. S2), which is the precondition for using otolith increments to construct the biological chronology.

In uencing factors of growth
Firstly, the variability in growth patterns were examined for C. myriaster. The tested factors included the age of sh when a given increment in otolith was formed (age), the calendar year of the increment formed (Year), and the identity of shes (Fish ID). The rst factor account for expected age-related trends, the second account for the effects of annual variation, and the last Fish ID allowed individual to have diverged growth from average and handled the correlations among increment measurements of the same individuals (Morrongiello and Thresher 2015).
We then considered both biotic and abiotic factors that in uence the growth rate of C. myriaster. Environmental factors included three categories of variables, seasonal bottom temperature, prey availability, and population density ( Supplementary Fig. S3

Mixed-effect modeling
Hierarchical mixed effect models have been developed to determine growth variation following Morrongiello and Thresher (2015). We used the data of 2-7 ages to generate a general population chronology, i.e., an estimate of the mean inter-annual growth rates, which was developed by extracting the conditional effects from models (Morrongiello and Thresher 2015). Our model used otolith increments as response variables, and used the factors mentioned above as explanatory variables. Growth increments were log-transformed to meet model assumptions.
The models were developed by analyzing xed effect (Age) with a series of random effect structures (combinations of random slope and/or intercepts of Fish ID, Age, Year). Models were compared using Akaike's information criterion (AIC; Bozdogan, 1987), and the model that results in the smallest AIC was chosen as the optimal model. To facilitate model convergence and interpretation of interaction terms, all predictor variables were mean centered (Morrongiello et al. 2014).
The effects of seasonal sea bottom temperature, PPC and population density were added to the model separately. Signi cantly correlated variables (Pearson's product-moment correlation P < 0.05) were not be included in the model simultaneously. To assess the potential synergistic effects of different factors, interaction terms were tted among the combination of environmental variables. Speci cally, we examined the age-speci c and density-dependent response of conger growth to water temperature, which were derived from the interaction effects of the mixed effects models.
The variance of the model was assessed on marginal R 2 describing the variance of xed effects (Morrongiello and Thresher 2015). We obtained the predictions and con dence intervals using the arm and Effects packages (Fox 2003; Gelman and Hill 2006). All the data analyses were implemented in R 3.3.2.

Variability in growth
Different combinations of xed and random effects made up the intrinsic models for chronologies. The best intrinsic models included the sh ID random intercept and the year random intercept ( Table 2). The selected model with both random and xed effects explained 29.3% of the variance in otolith growth according to the marginal R 2 .
"Age" as a xed effect explained a large portion of the variance of growth, and a relatively high proportion of otolith growth was explained by Year and Fish ID, and a considerable level of unexplained variance persisted as residuals ( Table 4). The growth rates of otolith decreased gradually with increasing age (Fig.  2a). Best linear unbiased predictors for the Year random effect showed considerable inter-annual variation in whitespotted conger growth over the last 9 years (Fig. 2b). Marked below-average growth rates were observed in 2010-2011, whereas above-average growth was detected in 2016-2018 (Fig. 2b).

Biotic and abiotic effects
We further examined the effects of conger density, prey per capita, and seasonal bottom temperature. AIC scores showed that spring (March-May) bottom temperature was the most critical factors for growth modelling among the seasonal averages (Table 3). Model AIC was lowest when including sh age, spring bottom temperature (Spring BT), PPC, and interaction between age and PPC and between Spring BT and PPC (Table 3). Including these xed effects increased model marginal R 2 by 0.264. However, density was not included into the optimal model. The spring bottom temperature showed a signi cant positive in uence on otolith growth, with an average increase rate of 21.63% per 1 °C (Table 4, Fig. 3a). The prey per capita also had a signi cant positive in uence on otolith growth (Table 4, Fig. 3b). High prey per capita resulted in an increase of 39.38% in mean growth of otolith (Table 4, Fig. 3b)

Interaction effects
The interaction between Age and PPC showed a signi cant positive relationship with species growth ( Table 4, Fig. 4a). Based on the optimal model, high PPC resulted in a 14.47% increase in otolith growth of individual with 2 years and 69.81% in 7 years sh otolith growth (Supplementary Table. S1). Furthermore, we found a signi cant negative relationship of species growth with Spring BT and PPC interaction ( Table 4, Fig. 4 b), indicating a higher degree of temperature-dependent growth when prey availability was low. The coe cients of interaction implied that increase of 1°C in spring BT led to 41.22%, 27.59%, 16.11%, and 5.89% growth increase when PPC was at the level of 1, 1.5, 2 and 2.5, respectively (Fig. 4).

Discuss
Our study used otolith increments to reconstruct the growth trajectory of whitespotted conger and illustrated a considerable inter-annual variation in their growth rates in recent years. After accounting for the intrinsic variations of growth rates resulting from age and the individual levels, we identi ed signi cant effects of environmental factors that drive annual growth, which enabled important implication for stock assessments and sustainable management for sheries.
The study identi ed a remarkable pattern of increasing growth rate of whitespotted conger over the last nine years. Speci cally, marked below-average growth rates were observed in 2010-2011, and aboveaverage growth was detected in 2016-2018. The variation may be attributable to the large-scale climate events. During 2010, the tropical Paci c had a persistent La Nin ã event, with a second-year sea surface cooling that occurred in the autumn of 2011 (Feng et al. 2015). The cool ocean conditions may slow down metabolism and temporarily decrease primary production, resulting in lower food availability, which, in turn, eventually might result in decreasing sh growth. Instead, extreme ocean warming events occurred during boreal summers in 2016 to 2018 in the South Yellow Sea and the East China Sea (Gao et al. 2020). Warm ocean conditions caused increases in metabolic rates, extending the whitespotted conger growing season. Therefore, the remarkable inter-annual variation of whitespotted conger growth could be a consequence of climate events which induced changes in aquatic temperature and food availability, though other factors such as eutrophication and shing pressure may also have played a role.
Our study showed that growth of whitespotted conger was sensitive to changes in spring temperature rather than other seasons. The spring bottom temperature showed a signi cant positive in uence on otolith growth, with an average increase rate of 21.63% per 1°C. The distinguishing effects of spring temperature may correspond to the fast growth period of conger eel (Gorie and Tanda In the last nine years, the spring and summer mean-bottom temperature in study area was ranged of 10.81-12.60 ℃ and 21.11-22.75 ℃, respectively. Therefore, the increase of temperature in the coastal water of the Yellow Sea still fell into the broad suitable temperature range of whitespotted conger and bene t its growth rates in recent years. By the end of this century, sea surface temperature in the Bohai, the Yellow Sea and the East China Sea will rise by 2°C and 4°C under the high and medium concentration emission scenarios, respectively (Tan et al. 2018). Combined climate predictions, temperature-dependent growth could lead to a signi cant implication for the management of whitespotted conger.
The positive correlation between growth and prey availability is in accord with the common sense that high food availability would increase sh growth. In addition, the whitespotted conger growth was in uenced by the prey per capita depending on sh age, and the older were more sensitive to the increase of prey availability. The ontogenetic shift in diet were observed in this species (Mu et al. 2019), and different food resources were consumed by small and large conger eels (Mu et al. 2019), which may lead to differential growth suppression among age classes (Walters and Post, 1993). Additionally, we found prey availability in uenced the temperature effect on otolith growth of whitespotted conger. When prey resource was relatively poor, the growth rates were signi cantly improved with the increased of temperature, and when the prey was abundant, the growth rate was at a higher level and was less affected by the water temperature. The results implied a combined effect on growth limitation by low water temperature and prey per capita. This response could be related to the predation success of whitespotted conger. Feeding rate is a function of the encounter rate between predator and prey, the predator's attack frequency, and the capture success (Elliott and Leggett 1997). Swimming speeds and activity rates are two most important factors determine the encounter rates (Gerritsen and Strickler 1977), and generally increase with temperature range from 10 to 26°C for whitespotted conger (Hori et al. 2019).
The attack rate of predators has also been shown to increase with temperature (Elliott and Leggett 1996). Therefore, the in uences of high temperature on conger's behaviors are likely to bene t predation success when prey resource is poor, and conversely, the capture ability of sh predators would be reduced at low temperatures (Persson 1986;Bergman 1987). Consequently, the increasing water temperature in the future enables whitespotted conger to remain high predation pressure on their preys when the forage population fall to a low level, which may result in substantial changes in community structure.
Many previous researches support an inverse relationship between growth of eel and its density (Machut et al. 2007;Boulenger et al. 2016), e.g., a negative linear relationship was found of density-dependent growth for European eel in a small river of western France (Boulenger et al. 2016). Density-dependent regulation of growth is often related to intraspeci c competition (Lorenzen and Enberg 2002), and occurrence of different forms of density-dependent growth depending on the nature of the life history and individual behaviors (Post et al.1999). Although the density effect was not selected into the optimal model, the prey-per-capita, which calculated by total catch of three forage species divided by the catch of whitespotted conger, takes into account the effect of intraspeci c competition. In addition, the interspeci c competition may also in uence the results, as there are more diverse species completing for the same preys in the marine environment, which was not considered in our study.

Conclusions
This study examined the otolith growth of whitespotted conger with explicit consideration of biotic and abiotic variables in mixed effects models to reveal the critical driving factors. We identi ed that spring temperature and prey availability had signi cant in uence on conger growth. This study provided the rst evidence for the joint effects of abiotic and biotic factors on the growth variation of whitespotted conger. The improved knowledge is essential for effective management of these ecologically important species, and is key to understanding the broad implications of global climate change.

Declarations Acknowledgment
We are grateful to the members of Ecosystem Assessment and Evaluation Laboratory from College of Fisheries in Ocean University of China for the sample collections.  Locations of Conger myriaster specimen collections for otolith biochronology analysis Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors.

Figure 2
Intrinsic effects for Conger myraister with 95% Cis in the Yellow Sea The effect of spring bottom temperature, density, and prey availability on otolith growth rates of Conger myriaster.