Maturity of Hairtail Varies with Latitude and Environment in the East China Sea

Abstract

Early maturity of economically important fish has become a global issue, which might affect the reproductive potential of fish stocks. We used the samples that were collected in a fisheries‐independent survey in the East China Sea annually around the end of August from 2002 to 2017 to investigate the maturation progress of female Hairtail Trichiurus japonicus in response to (1) spatial variation, (2) environmental variation, and (3) CPUE. The preliminary analysis indicated that the gonads of Hairtail in the southern area and the northern area began to develop at similar anal length (around 140–150 mm), but the development was faster in the southern area than in the northern area. The proportion of matured fish was significantly higher in the southern area than in the northern area in the group of fish with 170–210 mm anal length. Based on generalized additive models, we found that temperature and CPUE could affect the maturation schedules of female Hairtail in the East China Sea. The relationship between sea surface temperature and the probability of maturity was close to an s‐curve, while the stock density had an approximately negative effect on the degree of precocious puberty.

Length at maturity, as a vital life‐history parameter, has long been recognized as an important factor that influences egg sizes, spawning activity, and reproductive success in fishes (Gamboa‐Salazar et al. 2020; Saborido‐Rey and Junquera 1999). Notable reductions in the length at maturity of economically important fish have raised great concerns (Hutchings and Baum 2005). For these fish, not only maturity age and length but also the biological parameters and reproduction characteristics have changed a lot compared with the historical records in recent years (Cardoso and Haimovici 2014; Frey et al. 2015; Saborido‐Rey and Junquera 1999). The cause of changes in length at maturity has long been debated among scientists. Some have regarded it as the consequence of size‐selective fishing pressure (Allendorf and Hard 2009; Fenberg and Roy 2008; Heino et al. 2015; Nina et al. 2019), while others believed it might due to environment temperature, food availability, and habitat quality (Jonsson et al. 2013; Yoneda and Wright 2005). However, maturity is determined by a complex process that involves several confounding, subtle, and interacting factors. Without an understanding of the mechanisms behind the maturation process, estimates of the reproductive potential of spawning stock are potentially biased (Williams et al. 2016), which affects sustainable fishery management. Hence, it is essential to assess the temporal and spatial variability in the maturation process of economically important fishes and try to illuminate the cause of it.

For this study, we chose Hairtail Trichiurus japonicus as a sample fish, first, because the phenomenon of sexual precocity in Hairtail is very obvious and typical. In the 1960s, the minimum length at maturity was 200–210mm, while it decreased to 140–150mm in the late 1990s (Chen et al. 2013; Xu et al. 2003). Second, as one of the most economically important fish in China, Korea, and Japan, Hairtail is widespread in the subtropical West Pacific Ocean (Wang and Heino 2018) and bears high‐intensity fishing pressure (Wang et al. 2012). For instance, the Trichiurus fishery of the East China Sea (ECS) is the largest in China, on the order of 0.8 million metric tons worth RMB28 billion per year from 2002 to 2017 (Bureau of Fisheries of the Ministry of Agriculture 2003–2018). Third, Hairtail also plays an irreplaceable role in the ecosystem as one of the top predators. Its stock condition is closely linked to maintaining a sustainable fishery in China. However, despite the importance of Hairtail to science‐based management, knowledge on how local environmental conditions and commercial fishing affect life history traits like maturity and growth in this species is lacking. Unfortunately, although evidence on other fishes has shown that size‐selective fishing pressure can affect size at maturity in marine species at the population level (Fenberg and Roy 2008; Lappalainen et al. 2016; Sharpe and Hendry 2009), it is hard for us to consider the fishing pressure in the present scenario for two reasons. First, the capture production statistics that we can currently get are from the China Fishery Statistics Yearbook, which represents the catches of several Trichiurus species, although Hairtail is the dominant species. Second, China implemented a “zero increasing rate” in 1999 and fishing pressure was maintained at a stable level. Also, data from the 1960s is lacking, which might lead to biased estimates of the effect of fishing pressure on maturation rate. Considering that Hairtail displays sexual dimorphism in its maturation schedules (Wang and Heino 2018), in the present study we only considered environmental variability and CPUE and their mixed influences on the maturation schedule of female Hairtail.

The objectives of the present study were to (1) identify annual and spatial patterns of variation in the maturation schedule of female Hairtail in the ECS during 2002–2017 and (2) understand the proximate and ultimate determinants of maturation in this species.

METHODS

Sample collection

Hairtail samples were collected annually in a fisheries‐independent survey in the ECS around the end of August during 2002–2017. The surveys were conducted by a pair of bottom trawlers (each with 183.25 kW of power), and the average trawling speed was about 2 knots. The size of the net mouth was 100 × 4 m, and the mesh size of the cod end was 2.5 cm. The survey covered a broad geographic area from 26°30′N to 35°00′N and 121°00′E to 127°00′E. The surveys had a systematic sampling design, with sampling stations evenly distributed at a 30′ interval between every pair of sampling sites and 1 h of towing duration for each sampling site. An SBE‐37 (SeaBird, America) was used to record temperature and salinity. Catch per unit effort (g/h) was standard based on the weight of individuals, using catch per hour of trawl in the fisheries‐independent survey.

The Hairtail were randomly sampled from random sites, with each Hairtail measured for basic biological information including anal length (AL), body weight, sex, and gonadal maturity. For the sampled sites, if the number of Hairtail was less than 30, we sampled all of them; if the number was more than 30, we sampled 30 fish out of the total. In total, we obtained effective samples of 6,016 Hairtail during 16 years (Table 1). All of the samples were grouped into two regions (Figure 1): Northern Area (NA; ≥30°N) and Southern Area (SA; <30°N). The two study regions in the present study were dominated by two different water masses with different characteristics. The SA was affected by high temperature and high salt currents in summer, while the NA was influenced by low salt and high‐nutrient salt currents (Figure 1). Although the strength and influence area of the currents vary slightly from year to year, the surface frontal line of the eastern side of the interaction between the two water systems is basically in the south of 30°N (Wen and Wang 1984).

1 FIGURE. Sampling locations of Hairtail. Collections from 2002 to 2017 in the ECS were grouped into two regions (green boxes): the Northern Area (NA; ≥30°N) and the Southern Area (SA; <30°N). The ocean currents are also shown in the sampling areas, which are (1) Kuroshio, (2) the Kuroshio branches current west of Kyushu, (3) Yellow Sea coastal current water, (4) the Chinese coastal current, (5) Changjiang River diluted water, (6) the Kuroshio branch current to the north of Taiwan, and (7) the Taiwan warm current.

1 TABLESample size and anal length (mm) range of immature and mature female Hairtail collected in the SA and NA around the end of August from 2002 to 2017; N = total sample size.

Year	N	Immature	Mature
SA < 30	NA ≥ 30	SA < 30	NA ≥ 30
n	Range	n	Range	n	Range	n	Range
2002	200	77	150–259	87	140–239	17	178–316	19	208–423
2003	290	105	137–255	107	128–237	36	184–289	42	194–425
2004	331	110	150–245	173	122–250	19	184–273	29	194–305
2005	216	73	160–250	98	142–258	30	193–286	15	206–284
2006	442	95	127–295	250	135–253	19	194–285	78	176–352
2007	182	59	129–237	51	133–210	23	152–285	49	176–298
2008	115	2	170–192	73	135–240	4	224–295	36	182–330
2009	641	132	105–232	308	102–244	66	163–306	135	160–371
2010	158	40	148–198	48	132–201	29	182–306	41	141–282
2011	209	51	135–212	33	105–210	92	163–305	33	185–280
2012	265	37	134–194	98	138–221	60	155–280	70	164–290
2013	460	125	155–237	199	130–237	87	170–298	49	162–311
2014	708	219	109–235	208	125–204	171	123–302	110	144–289
2015	336	100	132–244	118	132–211	76	156–410	42	164–245
2016	286	142	140–214	98	145–210	26	170–255	20	165–285
2017	1,177	217	142–240	722	120–245	107	140–329	131	157–302

Maturity determination

The gonadal development of the fish was determined based mainly on macroscopic examination (Chen 2004; Li 1982). The gonads of the fish were classified into six development classes, defined as follows: 1 = immature, 2 = developing, 3 = maturing, 4 = mature, 5 = ripe, and 6 = spent. For the data analysis, we reduced the six maturity classes to two functional classes: immature (stages 1–2) and mature (stages 3–6; Xu et al. 2003). The mature ratio differences of female Hairtail in different length intervals between SA and NA were tested by using Fisher’s exact test (P < 0.05).

Statistical models

Generalized additive models (GAMs) were used to examine the contribution of the predictor variables to the probability of maturity in female Hairtail. The GAMs allowed for the incorporation of nonlinear terms through the application of smoothing functions, which had been typically used to estimate maturity (Winton et al. 2014; Wood 2007). In the present study, the proportion of maturity served as a response variable, and a binomial GAM was then fit, with weights equal to the size. Here, we focused only on the mature proportion of Hairtail that measured less than 200 mm AL because, first, it was the minimum anal length at maturity in the 1960s (Xu et al. 2003) and, second, Hairtail measuring less than 200 mm were mainly age‐1 fish (Yan et al. 2005). The initial explanatory variables were sea surface temperature (SST), sea bottom temperature (SBT), sea surface salinity (SSS), sea bottom salinity (SBS), geographic area, and CPUE. Before adding the variables in the models, we first determined the correlation among the predictor variables and the response variable by using Spearman’s rank correlation coefficient (Table 3). The variables that were significantly correlated with the mature proportion of female Hairtail (P < 0.05) were then tested for collinearity by using a Spearman’s rank correlation matrix（P < 0.05; Zuur et al 2010). The factors that were correlated with each other were excluded. The response variable was modeled with a binomial error distribution and logit link function. The best‐fit model was chosen based on the Akaike information criterion (AIC). All of the GAMs were conducted using the ‘mgcv’ package (v1.8‐33; https://www.rdocumentation.org/packages/mgcv), which is available in the statistical platform R, version 3.6.1.

RESULTS

Preliminary Analysis of Catch Composition

The AL of the Hairtail that were sampled in the present study ranged from 102 to 425 mm. The number of immature fish (n = 4,255) was greater than the number of mature fish (n = 1,761). Among them, there were considerable overlaps in AL of immature and mature individuals. The largest immature Hairtail (AL 295 mm, female) was collected in 2006 at 29°30′N, 124°30′E, which belonged to the Yushan fishing ground. The smallest mature Hairtail (AL 123 mm, female) was collected in 2014 at 26°30′N, 121°E, which belonged to the Mindong fishing ground.

A preliminary data analysis broadly showed geographic variation in maturation progress during summer (Figure 2). Almost all of the Hairtail that were less than AL 140 mm were immature. The proportions of matured fish in the AL 140–150‐mm size‐group were nearly the same between the SA and NA. The significant difference between the maturity ratios in the SA and NA was for fish in the AL 170–210‐mm size‐groups (P < 0.05). For instance, the maturity ratios that were obtained in the SA were more than twice those that were obtained in the NA for fish with AL 170–190 mm. The proportion of matured fish in the AL >220‐mm size‐groups showed no pronounced difference between the SA and NA (P > 0.05). The data indicated that the gonads of Hairtail in the SA and NA began to develop at similar anal length, but the development was faster in the SA than in the NA.

2 FIGURE. Number of immature and mature Hairtail with different length intervals sampled during 2002–2017. The total number of female Hairtail in the SA (<30°N) and NA (≥30°N) was 2,446 and 3,570, respectively. The columns with an asterisk represent a significant difference between the two areas (P < 0.05), which was tested by using Fisher’s exact test.

Environmental Variables

As shown in Figure 1, the SA is affected by the Kuroshio branches current to the north of Taiwan, which brings high‐temperature and high‐salinity water. The NA is mainly affected by Yellow Sea coastal current water, the Chinese coastal current, and Changjiang River diluted water, which make the water temperature and salinity in the NA relatively lower than they are in the SA (Table 2). During 2002–2017, the SST of both study regions showed an increasing tendency in fluctuation, and SST in the SA was almost always higher than that in the NA. The SBT of the two study regions fluctuated between 18°C and 21°C. Both the SSS and SBS in the SA were higher than in the NA during the study period.

2 TABLEThe CPUE, average temperature, and average salinity of the sampling sites in SA and NA during 2002–2017.

Year	SST (°C)	SBT (°C)	SSS	SBS	CPUE (g/h)
SA	NA	SA	NA	SA	NA	SA	NA	SA	NA
2002	27.40	26.49	20.27	20.62	33.53	33.33	34.64	34.30	37,587	47,257
2003	28.19	27.01	19.65	18.48	33.58	32.02	34.58	33.52	45,537	45,930
2004	27.44	26.28	20.09	21.37	33.18	32.81	34.68	34.35	57,604	83,655
2005	27.67	26.59	19.01	21.12	33.58	31.31	34.80	33.92	31,607	55,071
2006	26.70	24.77	19.49	18.80	33.60	32.46	34.71	33.60	45,574	47,692
2007	28.93	27.33	20.08	19.80	33.20	31.56	34.50	33.59	93,114	57,681
2008	28.46	26.72	20.59	21.10	33.42	31.78	34.61	33.47	81,145	66,413
2009	29.13	26.72	21.22	19.46	33.00	31.63	33.89	32.95	128,694	66,885
2010	29.70	28.57	19.25	18.55	33.38	30.64	34.75	33.66	57,111	63,012
2011	28.24	26.27	19.31	20.35	33.27	31.96	34.67	33.47	35,794	55,230
2012	27.86	25.21	20.09	19.42	33.07	31.59	34.47	32.80	56,390	21,855
2013	26.46	26.29	22.19	20.96	31.84	31.13	33.68	32.57	51,892	69,440
2014	29.28	27.22	21.26	18.72	31.62	31.02	34.59	33.61	46,454	63,082
2015	27.73	26.49	19.69	17.94	32.76	30.61	34.82	32.86	45,325	51,774
2016	28.71	26.57	20.40	19.46	31.49	31.13	34.65	33.64	62,117	53,453
2017	29.35	28.68	20.01	19.71	30.65	30.47	34.68	33.44	33,834	102,542

During the study period (Table 2), the CPUE of Hairtail in the two regions fluctuated dramatically. In the SA, the CPUEs during 2007–2009 were higher than they were in the other years, especially in 2009, when the CPUE reached 128,694 g/h. In the NA, the highest CPUE occurred in 2017, followed by 2004, and the lowest CPUE was in 2012. In other years, the CPUEs of the two regions did not have a great difference between them.

Probability of Maturity

As shown in Table 3, year, SST, SBS, log₁₀ CPUE, and the geographic coordinates all had a significant correlation with the probability of maturity. To some extent, the SST and SBS variables may be acting as proxies for location (latitude, longitude), as the water mass mainly affected the temperature and salinity (Figure 1). So we treated the location variables as categorical variables. Considering that SBS had a significant correlation with SST and log₁₀ CPUE, respectively (Table 3), we did not put it in the models. Finally, the five best‐fitting models for describing the relationship between maturity probability and the predictor variables are shown in Table 4. The fifth GAM model gave the best fit to the data, with the lowest AIC and the highest percentage of deviance explained (the bold model in Table 4). The year had a high contribution to the percentage of deviance explained. The inclusion of SST and log₁₀ CPUE also improved the model fits. Compared with the fourth and fifth GAM models, we found that implementing an interaction between the smoother and a categorical variable could reduce the AIC. This indicated that the region affects the degree to which different variables (SST, log₁₀ CPUE, and year) influence the probability of maturity. So we considered these factors by regions (Figure 3). In the SA (Figure 3A), the representation of the relationship between SST and the probability of maturity was close to an s‐curve. In the NA (Figure 3B), although the probability fluctuated with increasing SST, it overall increased with increasing temperature. A negative effect was found with log₁₀ CPUE (Figure 3C, D), meaning that the probability of maturity was higher with low log₁₀ CPUE. During 2002–2017, the probability of maturity of female Hairtail that were less than 200 mm in both regions all first increased and then decreased (Figure 3E, F). The highest probability appeared around 2010.

3 FIGURE. Estimates from the GAMs for the effects of (A, B) sea surface temperature, (C, D) log10 CPUE, and (E, F) year on the probability of female Hairtail maturity in the SA and NA, respectively. The y‐axis denotes the effect of the predictor on the probability of maturity.

3 TABLEThe Spearman’s rank correlation among the variables and the response variable. The upper triangular matrix shows the Spearman’s rank correlation coefficients and the lower triangular matrix shows the corresponding P‐values.

Spearman	Probability of maturity	Year	Longitude	Latitude	SST	SSS	SBT	SBS	log10 CPUE
Probability of maturity		0.259	−0.236	−0.306	0.253	−0.051	0.050	0.140	−0.129
Year	5.44 × 10–08		−0.016	−0.038	0.350	−0.399	0.073	−0.040	0.022
Longitude	7.28 × 10–07	0.102		0.123	0.106	0.212	0.184	0.136	0.017
Latitude	9.33 × 10–11	0.434	0.011		−0.574	−0.461	−0.044	−0.698	0.206
SST	1.06 × 10–07	7.61 × 10–14	0.029	6.12 × 10–39		0.068	0.155	0.405	−0.069
SSS	0.295	8.16 × 10–18	9.60 × 10–06	6.44 × 10–24	0.157		0.046	0.513	−0.160
SBT	0.287	0.132	1.25 × 10–04	0.359	0.001	0.344		−0.135	0.156
SBS	0.004	0.414	0.005	5.32 × 10–64	2.38 × 10–18	3.17 × 10–30	0.005		−0.219
log10 CPUE	0.008	0.653	0.727	1.66 × 10–05	0.152	0.001	0.001	4.53 × 10–06

4 TABLEGeneralized additive models (GAMs) fit. The five chosen models had a binomial distribution and logit link function. The models are ranked from worst to best fitting. The final model selected (in bold) was considered to be the simplest model, with the lowest AIC and the highest percentage of deviance explained. The region served as a categorical variable that divided the NA and SA; SST = sea surface temperature; CPUE = catch per unit effort; UBRE = unbiased risk estimator; s() indicates a smooth function.

GAM	Deviance explained (%)	UBRE	df	AIC	ΔAIC
1 region	10.20	1.318	2.0	1,469.7	186.4
2 region + s(Year, by = region)	29	0.884	12.9	1,283.2	46.3
3 region + s(Year, by = region) + s(SST, by = region)	35	0.775	22.3	1,236.9	25.0
4 region + s(Year, by = region) + s(SST) + s(log10CPUE)	37.20	0.719	22.7	1,211.9	16.5
5 region + s(Year, by = region) + s(SST, by = region) + s(log10 CPUE, by = region)	40.10	0.679	30.2	1,195.4	0.0

DISCUSSION

As the most important life history transition in animals, maturation is generally affected by a series complex determinants (Jorgensen et al. 2007; Tobin and Wright 2011; Wright et al. 2011). The results of the present study provide an evaluation of the effect of environmental factors and geographic location on the current degree of precocious puberty of Hairtail in the ECS. The variables SST and CPUE could affect the maturation schedules of female Hairtail.

The latitude variation in mature size was observed in the present study, and the same phenomenon has also been observed in many other fishes like Winter Flounder Pseudopleuronectes americanus (Winton et al. 2014), Walleye PollockGadus chalcogrammus (Williams et al. 2016), and Albacore Thunnus alalunga (Farley et al. 2014). The mature ratio difference between female Hairtail in the SA and NA was especially concentrated among fish measuring 170–210 mm AL (Figure 2), which suggests that the maturation process progressed faster in the SA than in the NA. The proportion of mature fish among smaller or larger fish did not show a significant geographical difference. Intuitively, it seemed that the gonads of Hairtail in the SA (<30°N) and NA (≥30°N) began to develop at a similar anal length. This is consistent with a postulated life history trade‐off to begin maturing when a size threshold is reached (Mcbride et al. 2013). Hairtail in the ECS belongs to a north–south migratory and multiple‐spawning species (Li 1982; Xu and Chen 2015). Its spawning period is long, and the spawning grounds are widely distributed in the ECS (Xu and Chen 2015). Every year from November to early March, Hairtail move from north to south for winter migration. From March to July, they move from south to north for reproductive migration (Xu and Chen 2015). The spawning period lasts from April to November (Li 1982). Since a genetic differentiation analysis based on the level of mitochondrial DNA showed that Hairtail from nearshore of the ECS constituted a panmictic population and can be treated as a unit in their management (Wu et al. 2019), the spatial variation in Hairtail length at maturity that was observed in this study was more likely driven by environmental factors.

In the present study, the Hairtail in the northern latitudes were mature at larger lengths than were those occupying the southern latitudes. This might be due to the warmwater environment in the SA, which consists of seawater with high temperature and salinity, whereas the water system in NA is coastal water with low salinity and high‐nutrient salt (Beardsley et al. 1985; Chern and Wang 1990; Ichikawa and Chaen 2000). The water temperature difference is a known driver of geographical variation in fish maturation rates (Farley et al. 2014; Lourenco et al. 2015; Wang et al. 2008), as ectotherms tend to grow faster and mature earlier at warmer temperatures than at cooler temperatures (Angilletta et al. 2004; Brunel and Dickey‐Collas 2010; Matta et al. 2016). Our results also suggested that SST significantly affected the maturity schedules in female Hairtail (Table 4). The probability of maturity increased with increasing SST until approximately 28°C (Figure 3). The temperature effect on the size at maturity was the so‐called “temperature–size rule” (Atkinson 1994; Atkinson and Sibly 1997). Previous studies have suggested that it was the adaptive changes in energy allocation to reproduction and other competing needs (Angilletta et al. 2004; Berrigan and Charnov 1994; Irie et al. 2013), as the temperature could modify metabolic rates and physiological activities including oxygen consumption, feeding, and enzyme reaction rates (Garcialopez et al. 2006; Hewitt and Duncan 2001). The potential warming of ocean waters and the effects of this warming on the life history traits of various marine organisms have raised great concerns (Baudron et al. 2014; Sheridan and Bickford 2011). Evidence has shown that fish populations around the world have already been strongly affected by warming (Christopher et al. 2019). Our findings provide clear evidence for warming effects on the onset of puberty in subtropical fish and highlight the importance of accounting for the effects of climate change in fisheries assessment and management.

Another notable finding in our study was the effect of CPUE on the probability of maturation (Table 4). In the present study, a negative effect was found with increasing stock density—high stock density was related to a low probability of mature female Hairtail. This concurs with the point that sizes at maturity are highly plastic parameters that change under external pressure, particularly with a decrease in the population abundance (Adams 1980; Wootton 1990). Gwladys et al. (2009) thought that the density effects influenced maturation through growth, as the juvenile growth rate is higher when the stock density is low in Norway Pout stock. This density dependence might be linked to local aggregation, food availability, or perhaps due to density‐ and size‐dependent juvenile mortality (Engelhard and Heino 2004; Gwladys et al. 2009). Several previous studies also observed the same trend. Li et al. (2011) found a correlation between stock density and the declines in length at reproductive maturity in small Yellow Croaker Larimichthys polyactis. Engelhard and Heino (2004) observed that during a collapse of the population, size at maturity increased slightly in Norwegian Atlantic Herring Clupea harengus. Stahl and Kruse (2008) reported some evidence for a density‐dependent relationship between length at 50% maturity and stock biomass among Walleye Pollock in the eastern Bering Sea. Roff (2002) suggested that slow growing, late maturing, and high rates of early survival occurred when the population size of a species reached a stable equilibrium.

In the present study, we also noticed that in both the SA and NA the probability of maturity of Hairtail that measured less than 200 mm has decreased in recent years (Figure 3). It might be good news that indicates that the actions that were directed at fish management worked, such as the midsummer moratorium and the ECS Hairtail national aquatic germplasm resources reserve. It was also worth mentioning that the explanatory power of the full model (Table 4) was 40.1%. This implies that the variables that were used (temperature, CPUE, year, etc.) represented only some aspects of a much more complex process that determines maturity in Hairtail, which suggests that there are confounding, subtle, and interacting factors that affect the maturation schedules in this species.

ACKNOWLEDGMENTS

This study was sponsored by the Shanghai Sailing Program (18YF1429800), Central Public‐Interest Scientific Institution Basal Research Fund, East China Sea Fisheries Research Institute, Chinese Academy of Fishery Sciences (L32201921860). There is no conflict of interest declared in this article.