Fish sampling data were obtained from 34 otter trawl surveys at 52 stations ( Figure 1 ) from 2006 to 2020. Apart from four survey cruises per year in January, April, July, and October in 2006 and 2007, surveys since 2008 were performed twice a year in January and July. Most survey cruises did not cover all 52 stations for a variety of reasons, but the samples were generally coherent in methodology. The surveys were undertaken by a commercial fishing vessel with a main engine of 600 HP. In every survey, each station was trawled once for 1 hour with a towing speed of 2.5~3.5 knots. The headrope length of the otter trawl net was 37.7 m and the cod-end mesh was 40 mm. Every trawl sample was sorted onboard and identified to the species level. Samples of each fish were counted, weighed, and recorded, and length and weight frequency data were also collected for major commercial fish. When fewer than 50 individuals were caught in a trap, all individuals were measured; otherwise, 50 individuals were randomly sampled for measurement. For each marine fish, total length (nearest cm) and body weight (nearest g) were measured. Based on the availability of data, we selected 13 species of major commercial fish (include 9 fish families, Table 1 ) with the highest frequencies of occurrence for analyses; according to data from the China Fishery Statistical Yearbook (2006-2020), these species account for about 45% of the total catch recorded in the Beibu Gulf.

To characterize ENSO events, data for the ONI in each month from 2006 to 2020 were obtained from products of the NOAA (National Oceanic and Atmospheric Administration) Climate Prediction Center (CPC; https://origin.cpc.ncep.noaa.gov/). The ONI is defined as 3-month running mean sea surface temperature (SST) departures in the Niño 3.4 region (5°N to 5°S, 120 to 170°W) and is a principal measure for monitoring, assessing, and predicting ENSO. Climate events are defined as five consecutive overlapping 3-month periods at or above a +0.5°C anomaly for warm (El Niño) events and at or below the −0.5°C anomaly for cold (La Niña) events. The threshold is further broken down into weak (with a 0.5–0.9°C SST anomaly), moderate (1.0–1.4°C), Strong (1.5–1.9°C), and very strong (≥2.0°C) events. The CPC considers that these anomalies must also be forecasted to persist for three consecutive months. In addition, the 1/4-resolution SST data were obtained from the OI SST V2 High Resolution Dataset provided by NOAA’s Physical Science Laboratory, Boulder, Colorado, USA (https://psl.noaa.gov, accessed on 1 October 2022). Monthly mean SST data were collected at all survey stations to calculate the monthly mean SST in the Beibu Gulf from 2006 to 2020.

Length frequency data of major commercial fishes from 2006 to 2020 was pooled and converted to monthly catches with the assumption that the samples were representative of the total catch of the month (Zhang et al., 2023). We used ELEFAN in the R package “TropFishR (1.6.3)” (Mildenberger et al., 2017) to estimate the natural mortality, asymptotic length, and growth coefficient of major commercial fishes, and the optimization routine for the fitting of von Bertalanffy growth function (VBGF) parameters is the “ELEFAN_GA” function uses the Genetic Algorithm package “GA” (Scrucca, 2013; Scrucca, 2017). Growth parameters are described by the VBGF (Von Bertalanffy, 1938):

where L_t represents mean length (cm) at age t, L_∞ represents asymptotic length (cm), k represents growth coefficient, and t ₀ is the theoretical age when L_t = 0.

To improve the precision of the growth parameters (L_∞ & k) estimation, the length frequency data of major commercial fishes was binned according to the maximum body length observed for the fish species (Wang K. et al., 2020), number indicating over how many length classes the moving average should be performed:

where L_max represents maximum body length (mm) of each fish being studied.

Natural mortality (M) was estimated by applied an updated version of the (Pauly, 1980) growth-based method (Then et al., 2015):

For optimum search and improvement in the accuracy of the output of ELEFAN_GA, we appropriately tuned key input settings (including Fix L_∞ (initial seed value of L _∞) and upper and lower limits of k) based on relevant literature (Taylor and Mildenberger, 2017; Mzingirwa et al., 2020; Wang K. et al., 2020) and used default TropFishR settings for all other parameters.

Based on the fundamental trade-offs between natural mortality (M) and parameters defining the maximum size and rate of growth in marine fish proposed by Gislason et al. (2010) and the empirical equation demonstrated by Charnov et al. (2013) (Equation (4), a life-history trait composite index (LTCI) was constructed to show the combinations of life-history traits, that is, life-history strategies. Given that for any combination of life-history traits, the points formed by the natural mortality (M) and growth state parameter ((L/L_∞ )^-1.5•k, a higher value indicates how fast the fish grows) of marine fishes on the coordinate axis are basically along the y = x line (Charnov et al., 2013), the LTCI was calculated as the distance between the point projected on the y = x line of each point and the zero point (origin). The value of LTCI was calculated as shown in Equation (5):

where L_∞ (cm) and k (year^-1) are the VBGF parameters and L (cm) is defined as the mean body length of the fish population. The value of LTCI will be positive if the projected point is on the right-hand side of the zero point, negative if it is on the left-hand side of the zero point, and 0 if it falls on the zero point ( Figure 2 ).

Cross-correlation function (CCF) analysis was used to explore the hysteretic relationship between climate event intensity and SST in time series, for removing spurious correlations from the affected time series (Baumgartner et al., 2018), this study performed prewhitened CCF analysis using the “prewhiten” function from the R package “TSA”. To investigate responses of fish populations to climate events, linear models were used to identify the relationships between the ONI and parameters of fish life-history traits. Spearman correlation coefficients were used to test the significance of these relationships (p < 0.05 was considered significant). Multivariate analysis of variance (MANOVA) and one-way analysis of variance (ANOVA) were employed to compare fish life-history trait parameters among climate periods. SPSS 27.0 (SPSS Inc.) was used to perform these statistical analyses.

Furthermore, the effects of climate events on the life history of the fish community were analyzed by using a hierarchical linear model and moderating effect estimation, and the R package “broom. mixed” and R package “bruceR” were used respectively. All R packages were operated in the R Environment (R Core Team, 2022).

In 2006−2020, 11 climate periods were identified according to the afore-mentioned criteria for the classification of climate events, of which four were normal climate periods, four were El Niño periods, and four were La Niña periods ( Table 2 ).

The hierarchical linear model results indicated that the relationship between the growth status parameter (ln((L/L_∞ )^-1.5•k)) and natural mortality (ln(M)) of fish community differ significantly by climate event type; the normal climate period regression slope (p _n = 0.772) was closest to 1.0, and it also was higher than the regression slopes of El Niño period (p _e=0.669) and La Niña period (p _l=0.593) ( Figure 3A ) The mean difference estimation also showed that climate events had an impact on LTCI in fish communities, and the mean diffidence of LTIC La Niña events also was the largest. In addition, regression analysis of the life history trait parameters of fish communities for each climate period was also shown the mean area of the 95% confidence ellipse (3.345 ± 0.382) was lowest in the normal period ( Figure 2 ), followed by the La Niña period (3.817 ± 0.613), and was highest in the El Niño period (4.819 ± 1.332). The mean major axis of the ellipse (a) was also lowest in the normal period (2.009 ± 0.158) and larger in the La Niña (2.303 ± 0.554) and El Niño periods (2.396 ± 0.081), while the mean minor axes of the ellipse (b) in the normal (0.534 ± 0.072) and La Niña periods (0.537 ± 0.041) were similar, with the highest value in the El Niño period (0.647 ± 0.203). Overall, the results of MANOVA showed that the growth state parameter, natural mortality, and LTCI of fish in different climatic event periods showed significant differences (p < 0.05).

The monthly mean SST in the Beibu Gulf ( Figure 4A ) presented significant and positive associations with ONI during 2006–2020 (r = 0.521). Considering the time with higher correlation values, there was a high correlation between the monthly mean SST in the Beibu Gulf and ONI without any lag ( Figure 4B ). In addition, although the effects of monthly mean SST on LTCI of fish communities were no significantly different in El Niño climate periods compared with normal and La Niña climate periods based on the results of moderate effect estimation (p = 0.07), the LTCI of fish communities strongly increased with monthly mean SST, while the LTCI of fish communities as a whole was lower than normal climate periods during La Niña climate periods ( Figure 5 ).

Except for T. lepturus and E. cardinalis, the natural mortality values and growth parameters of other fish showed a significant positive correlation ( Figure 6 ). Some fish exhibited relatively stable life histories between climate periods; for example, the life-history trait parameters of T. lepturus clustered in the lower-left area of the coordinate plane. In addition, one-way ANOVA showed that these fish life-history trait parameters did not differ significantly among climate periods (p > 0.05), whereas life-history trait composite index (LTCI) of P. macrocephalus and D. maruadsi differed significantly during the transition of climate periods (F = 38.986 and F = 14.921, respectively, all p < 0.05). Except for N. bathybius and P. anomala, all fishes did not display any considerable oscillation in LTCI with the transformation of climate periods ( Figure 7 ).

A Spearman correlation analysis indicated that the LTCI values for D. maruadsi, P. macrocephalus, and U. sulphureus were significantly correlated with ONI in 2006–2020 (r = 0.697, 0.762, and 0.905, respectively, all p < 0.05). The most consistent changes in the LTCI of the three fish species occurred in 2006–2012, when transitions between warm and cold events was most frequent ( Figure 8 ). Furthermore, in the latter six climate periods, there was a negative correlation between the LTCI of T. japonicus and D. maruadsi. While the relationships were not statistically significant, the LTCI of fishes belonging to the same genus showed the opposite trends in some climate periods, including U. sulphureus and U. japonicus, S. tumbil and S. undosquamis, and N. virgatus and N. bathybius.

Despite a large and growing body of evidence for the role of anthropogenic activities in the population dynamics of marine species (Pinsky and Palumbi, 2014; Sherman et al., 2017; García-March et al., 2020; Strøm et al., 2021), including dramatic fluctuations of pelagic and demersal fishes in the South China Sea over space and time scales (Masrikat, 2012; Yang et al., 2021; Zhang et al., 2022), our results suggest that the combination of life-history strategies for the Beibu Gulf fish community were most strongly driven by climate events. Climate events significantly altered the slope of the regression of life-history trait parameters for fish communities, with slopes closest to 1.0 in the normal period ( Figure 3 ). However, our results also revealed that both climate events would lead to a decrease in the mean correlation coefficient, especially La Niña periods, which can be explained by the fact that the combinations of life-history traits for most fish species fall in the lower-left quadrant (slower growth and lower M). Although the mean correlation coefficient of the El Niño period was between those for the normal and La Niña periods, the larger mean area of the confidence ellipse for the combinations of life-history traits in the El Niño period differed from those of other periods, and the former period tends to have a higher diversity of fish life-history strategies. While other studies have found that this climate variability may influence the biomass of certain fish species substantially via environmental variables (Yatsu et al., 2008; Yen and Lu, 2016), our results highlight the regular dynamics of life-history strategy selection in fish communities.

An important and relatively underappreciated mechanism underlying this dynamic may be related to ocean-atmosphere coupling. Temperature in all regions has pervasive effects on biological processes. Even though the mean SST in Beibu Gulf is on the rise under the background of global warming, this study found a positive correlation between the monthly mean SST in Beibu Gulf and ONI without any lag. This response to water temperature anomaly in the equatorial Pacific is bound to affect fish growth and development (Gillanders et al., 2012). In general, species characterized by fast growth and short generation times (high M values) would respond quickly to environmental changes (King and Mcfarlane, 2006), and the increase in seawater temperature caused by warm events may accelerate their growth and maturation. From the results of this study analysis, however, the relationship between LTCI of fish community and SST in the Beibu Gulf has a certain correlation with the type of climate events. On the one hand, since Beibu Gulf was a subtropical bay and its fishes were mainly warm-water species (Liang, 2022), temperate fishes tend to have smaller heating tolerances than tropical fishes (Payne et al., 2021). Therefore, an increase in environmental temperature would reduce the physiological function of fish and then affect the fish’s life history (Dembski et al., 2006; Holt and Jorgensen, 2015), especially during the El Niño climate periods when the SST is significantly warmer, and this study observed a significant increase in LTCI in fish with increasing SST. On the other hand, fishes tend to choose more “conservative” life-history strategy (slower growth and lower M) selection because of the lack of suitable water temperature in the environment (Soyano and Mushirobira, 2018). This was also evident in the event of cold events, where significantly lower seawater temperatures undoubtedly result in a lower overall LTCI of fish communities compared to normal climate periods. Life-history traits of signal species evolve based on environmental conditions to maximize sustainable reproduction; however, changes in the life-history strategy of fast-growing species, as the most important component of the ecosystem (Waples and Audzijonyte, 2016), will inevitably affect their ecological functions. In addition, during El Niño periods, the South China Sea generally exhibits weakened circulatory gyres and warming of the SST, consistent with the weakening of East Asian winter monsoon (Qiu et al., 2015). Previous studies have suggested that warmer temperatures increase metabolic rates and therefore increase the energy demand of fishes, while simultaneously decreasing the quality and quantity of available prey (Overland et al., 2010). As a result, combinations of life-history traits of fish within communities have become more diverse, even if some combinations have deviated significantly from evolutionarily stable strategies. This mechanism may explain our finding that life-history strategies for the fish community in Beibu Gulf were sensitive to climate events. While marine species fluctuate strongly, even in the presence of fishing (Adams, 1980; Hu et al., 2015), the strong interaction we found between climate events and life-history traits suggested that the composition and structure of the fish community in the subtropical semi-enclosed bay were also primarily caused by climate teleconnection.

The life-history traits of most fish species were optimized across all periods, i.e., variation in mortality coincided with variation in growth parameters; however, the variation in life-history traits of T. lepturus and E. cardinalis differed significantly from those of other species. Ribbonfish (T. lepturus) is a benthonic-pelagic species found worldwide in tropical and subtropical regions, mainly between 45°S and 60°N (Taghavi Motlagh et al., 2021). While T. lepturus is relatively short-lived and fast-growing (Clain et al., 2023) compared to larger fish in the marine ecosystem, they had the lowest mean mortality rate among the fish analyzed in this study and the distribution of M values was very similar at all periods, suggesting that their life-history traits could mainly be described as “longevity-slow.” Although we were unable to find significant differences between growth parameters among climatic periods, relatively sharp variation in the growth of ribbonfish was clearly influenced by climatic events. Past research has suggested that T. lepturus possesses flexible reproduction strategies; in environments with marked temperature cycles or weak fishing pressure, populations are more likely to reach maturity after longer durations (Al-Nahdi et al., 2009). However, in warmer and less productive environments, lower batch fecundity may be compensated for, in part, by a prolonged spawning period due to a smaller size (Martins and Haimovici, 2000). The increase in growth parameters in the case of relatively fixed mortality will promote a shift in the combination of life-history traits to the fourth quadrant. Therefore, T. lepturus would expend a great deal of energy to mature and begin reproduction early, retaining a small size and not capitalizing on the benefits of increased fecundity that come with a large size and long-life cycle (Waples and Audzijonyte, 2016). However, the reproduction strategy of a multiple batch spawner with a prolonged spawning period seems to allow for such a life-history strategy to exist.

We also observed that the life-history strategy of E. cardinalis was relatively constant in most climate periods. E. cardinalis is a small, warm-water, near-demersal fish with a fast rate of sexual maturity and a high reproductive rate (Zhang et al., 2020). The species exhibits significant periodic fluctuations in stock density (Sun and Lin, 2004) and can withstand a high fishing intensity (Chen, 2005). While E. cardinalis is one of the most economically important species in the northern South China Sea, the substantial increase in catch due to overfishing is a major threat (Zhang et al., 2020; Xu et al., 2021), and the species was even listed as endangered in a recent International Union for Conservation of Nature (IUCN) red list (Xu et al., 2022). Having a relatively unique niche in the ecosystem may be the main reason for the relatively stable life-history strategy selections of these fishes (Olden et al., 2006), with less interspecific competitive pressures making them more resistant to disturbances (Lewis, 2022). We also suggested that the relatively fixed combination of life-history traits may be a response of the population to overfishing; this stable life-history strategy would reduce the negative impacts of heavy exploitation on survival, and the effects of climate events may not be dominant.

Of note, P. anomala and N. bathybius showed substantial differences in LTCI among climate periods. Given that fishing may lead to populations with an elevated sensitivity to climate variability, interaction effects between fishing and climate cannot be ignored. Several recent studies have shown that the fishery resources in the Beibu Gulf have been overfished (Su L. et al., 2021; Zhang et al., 2021; Su et al., 2022), and loopholes in local management measures and the increased fishing pressure from Vietnam have resulted in the annual decreases of fishery resources in the region (Huang and Huang, 2013; Tian et al., 2022). This not only results in the replacement of dominant species with less valuable, small-sized, low trophic level species (Zhang et al., 2020) but also increases the magnitude of population fluctuations driven by environmental variability (Hsieh et al., 2006; Planque et al., 2010; Hidalgo et al., 2011; Rouyer et al., 2011; Shelton and Mangel, 2011). Therefore, synergistic interactions between climate anomalies and fishing are realistic (Waples and Audzijonyte, 2016). Fish whose life-history strategies are significantly affected by climate events are more frequent in overexploited areas.

Among the commercially valuable fish species included in this study, only the LTCI values for D. maruadsi, P. macrocephalus, and U. sulphureus were significantly correlated with the fluctuations in ONI, which represents climate period transformations. Given the delayed response trait of the fish population to environmental changes, it is reasonable to estimate fish life-history traits separately by the types of climate periods over a longer time scale in this study. The life-history traits of these fishes are characterized by higher M and faster growth during the normal or El Niño periods. The LTCI for D. maruadsi generally decreased after the shift from El Niño to La Niña or normal conditions. Climate events will affect a range of abiotic factors that are tightly linked to the production and distribution of fish populations. Shallow areas will exhibit larger increases in sea temperatures than those in deeper waters. As warm-water fishes, D. maruadsi, P. macrocephalus, and U. sulphureus will grow faster with increased reproduction due to seawater warming, which expands aerobic scope initially (Holt and Jorgensen, 2015), and the increase in metabolic requirements intensify foraging and reduce survival. However, under higher or lower mortality, evolution toward a faster or slower growth, respectively, optimizes their combination of life-history traits to ensure the stability of population dynamics. High natural mortality is essential if fish growth is fast, as the stability of the ecosystem only limits population size within the carrying capacity. Furthermore, La Niña events also strongly affect the surface and subsurface patterns of the South China Sea through the anomalous cyclonic circulation in the low-level atmosphere (Kuo and Tseng, 2020). Like the west coast of the Atlantic (Mcphaden et al., 2006), tropical cyclones tend to be reduced in number and intensity during El Niño periods but stronger and more numerous during La Niña periods (Huang and Xu, 2010; Hsu et al., 2013; Chen and Duan, 2018). The frequent occurrence of weather phenomena reduces anthropogenic disturbances and the lower SST during La Niña periods (Qiu et al., 2015) results in a shift to increased growth over rapid reproduction.

We also found that changes in life-history strategies can differ among species within the same genus in response to climate events. In particular, for U. sulphureus and U. japonicus, S. undosquamis and S. tumbil, and N. virgatus and N. bathybius, the LTCI values of two species in the same genus were generally similar in normal periods, with divergence in opposite directions in response to climate events. These species of the same genus not only share very similar biological parameters but also have similar population characteristics, such as diet, habitat preference, and spawning period, suggesting that their ecological niche is basically the same (Lewis, 2022). However, the oceanographic features that change following ENSO events include surface seawater temperatures, the vertical and thermal structure of the ocean (especially in coastal regions) (Wu et al., 2014), and the coastal and upwelling currents (Chu et al., 2017; Dang et al., 2021). These effects will undoubtedly have a significant negative impact on the growth and fecundity of fish and consequently intensify competition for survival between species with similar ecological niches. Thus, species of the same genus, through the complementarity of life-history strategies, avoid intense competition for ecological niches during climate anomalies because the carrying capacity of the ecosystem can be increased by reducing the similarity in life stages.

Finally, T. japonicus and D. maruadsi are generally considered to belong to the same feeding group at the ecosystem level and are both dominant species in offshore areas of China (Li et al., 2019; Su et al., 2022); accordingly, they have very similar ecological niches in long-term cohabitation. While the correlation was not statistically significant, we observed similar trends in LTCI for the two species over the first six climate periods in this study. Combined with the observation that there was a negative correlation between the LTCI value of T. japonicus and D. maruadsi in the latter six climate periods, we can conclude that the regularly alternating climate periods may explain the consistency of direction of change in life history strategies. The continued decline of fishery resources and increasing fishing pressure in the Beibu Gulf may also be responsible for the complementarity of life-history strategies between D. maruadsi and T. japonicus in the latter six climate periods. Overall, given the different relationships between life-history strategies and study periods, we emphasize the crucial importance of analyses of the effect of climate events on fish life history; overfishing weakens the resilience of fish populations to climate events, while interactions within communities also affect the response of fish life-history strategies to climate events.

There are some limitations in this study. The ELEFAN assumes that the body size composition of the survey data is representative of the exploited population major commercial fishes, and the study assumes that these survey data can be adequately analyzed in the corresponding climate event periods. However, the time span of each period of climate event is not uniform, and the composition of seasons varies from period to period. The energy storage of fish to adapt to seasonal environments is also reflected in changes in their life history (Giacomini and Shuter, 2013), which may affect our judgment of the influence of climate events on fish life-history trait parameters. For instance, although the effects of different types of climate events on the life history trait parameters of fish populations can be observed at the community level, these changes were relatively difficult to notice at the individual species level. Furthermore, the ONI has become the de-facto standard that NOAA uses for classifying warm and cool events in the eastern tropical Pacific. Although we have demonstrated a correlation between temperature changes in the Beibu Gulf and anomalies in the distant eastern equatorial Pacific, considering that ENSO’s teleconnections to other regions have impacts beyond just affecting water temperature [e.g., rainfall, seawater salinity (Zhu et al., 2014; Ramos et al., 2019)], so we did not directly use the local sea surface temperature anomalies as a direct factor in this study. In conclusion, our research results have demonstrated a certain correlation between the changes in life-history traits of major commercial fishes in the Beibu Gulf and ENSO events, and identifying specific influencing environmental factors will be the focus of our further research. Therefore, in future studies on the relationship between climate change and fisheries in the Beibu Gulf, more factors will be taken into account when assessing the effects of climate events on the life-history trait parameters of fish, such as the temporal continuity of survey data.