Takifugu rubripes used as broodstock were collected from three different cultured populations (China Dalian Tianzheng Industrial Company Limited cultured population, a population introduced from Japan in 2010 and a wild population caught in the Yellow Sea in 2009), and brought to China Yantai Tianyuan Aquatic Limited Corporation. The genetic background of the three populations can be found in the literature (Ma et al., 2016). In April 2012, individuals with mature gonads were selected from these three populations as parent fish. The selected parent fish reached sexual maturity at the same time after artificial cultivation and reproductive regulation. The broodstock of T. rubripes was induced by injection of luteinizing hormone releasing hormone A3 (LRH-A3) 250–300 g/kg + human chorionic gonadotrophin (hCG) 500 IU/kg (Ningbo hormone factory, Ningbo, China). The gonadal maturation of the broodstock was examined 24–48 h after induction. The abdomen of mature brood fish was obviously swollen. The female’s genital pore is filled with blood and slightly valgus, while milky semen flows out of the male fish when their abdomen is squeezed slightly. After gonadal maturation, the abdomen of female fish was squeezed to collect mature eggs in a dry white porcelain basin. The abdomen of the male was then squeezed to collect mature sperm. The sperm and eggs were then mixed. The sperm was activated by adding an appropriate amount of seawater, and the eggs were washed after standing for 5 min, until the fertilized eggs no longer adhered to the basin wall. The fertilized eggs were incubated in a 60 mesh sieve silk cone hatching net with a diameter of 80 cm. All the cone hatching nets were placed in a 40 m³ pond, with aeration and micro flow incubation. The incubation temperature was 18–19℃. Families were produced according to a nested mating design with two females nested within a male among 24 males and 48 females (8 males and 16 females in each population). First, eggs were obtained from two females and stored in two different beakers. Subsequently, the male fish were stripped and the milt was divided in two. Each portion was used to fertilize one of the two females. In total, 48 full-sib families (24 half-sib families) were obtained through artificial insemination.

The larvae were fed with enriched rotifers on the 4th day after hatching (DAH), and enriched rotifers and Artemia nauplii on the 15th DAH. Larvae were fed a small amount of surimi on the 28th DAH, and only surimi was fed on the 45th DAH. On the 60th DAH the families were transferred to a sea cage for cultivation. In the early stages of family cultivation, individuals were too small to carry out tagging identification and polyculture, so each family had to be cultivated in a separate pond. So the families were inevitably affected by environmental factors. In order to reduce the differences in environmental effects among different families, environmental condition standardization and quantity standardization methods were used during breeding. In order to eliminate the effect of stocking density on each family at different stages of development, four levels of standardized treatment were carried out in combination with the ecological habits and culture modes of T. rubripes. According to the specifications of the breeding pool and sea net cage in the seedling workshop of the experimental site, quantity standardization was as follows: In the first level of quantity standardization, the larvae were cultured in a 5 m³ round cultivation tank. On the 15th DAH, 15 000 larvae were retained for further cultivation, and the surplus larvae were transferred to production for cultivation. For the second level of quantity standardization, on the 30th DAH, 8 000 juveniles with fast growth and good vitality were selected and cultured in a 36 m³ cement pond (6 m×6 m×l m). In the third level quantity standardization, on the 60th DAH, 5 000 juveniles with fast growth and good vitality were selected and placed in a 125 m³ cage (5 m×5 m×5 m) for intermediate cultivation. The fourth level of quantity standardization involved 2 000 juveniles with fast growth and good vitality being selected and placed in a 125 m³ cage (5 m×5 m×5 m) for intermediate cultivation on the 120th DAH. Intermediate cultivation lasted for 60 days. When the fish were 6 months old, 50 individuals from each family were randomly selected and individually tagged with passive integrated transponders (for distinguishing families and individuals). Individuals tagged with passive integrated transponders were transferred to a 400 m³ concrete tank (20 m×20 m×l m) for polyculture, for another 6 months. Any dead individuals were removed from the tank once a day. The breeding conditions of each family were consistently maintained from the hatching of fertilized eggs to larval rearing and tagging polyculture. According to different developmental stages, the families were placed in the same workshop for unified breeding management. During the larval-culture period, water temperature, salinity, illumination intensity, pH, ammonia nitrogen, nitrite-nitrogen, and dissolved oxygen were maintained at 15℃ to 22℃, 25 to 32, 500 lx to 1000 lx, 7.8 to 8.3, ≤1 mg/L, ≤0.4 mg/L, and 5 mg/L to 10 mg/L, respectively. During the juvenile- and adolescent-culture period, the above seven indices were 22℃ to 24℃, 15 to 32, 500 lx to 1000 lx, 7.8 to 8.3, ≤1 mg/L, ≤0.4 mg/L, and ≥6 mg/L, respectively (Li et al., 2013). During the larval- and adolescent-culture period, the bait is minced trash fish and pellet feed, and the feeding amount is 0.7%–1.5% of the total weight of the fish. It is fed once or twice a day. The nutritional components of pellet feed include protein, lipids, vitamins, minerals and carbohydrates.

Five growth traits, BW, BL, TL, CM, and TKL (Fig. 1), were evaluated at two growth stages. BL, TL and TKL of each T. rubripes were directly measured using Vernier calipers. With the help of the thin line, CM was indirectly measured using Vernier calipers. BW was obtained by weighing using a digital balance. Measurements were taken at 6 and 12 months. A total of 50 individuals were randomly sampled and measured in all 48 full-sib families at 6 months, and all individuals retained in the 48 families (some individuals died or escaped) were measured at 12 months. Prior to measurements, all individuals were anaesthetized with 100 μL/L MS222 (tricaine methane sulfonate) (Maya Reagent, Jiaxing, China).

The genetic parameters for BW, BL, TL, CM, and TKL were estimated using an animal model. The model can be written as

where μ is the mean value of population; y_ij is the observation of each trait (BW, BL, TL, CM, and TKL); a_i is the random additive effect of individual i and is distributed ~N (0, A$ {\sigma }_{{\rm{a}}}^{2} $), where A is the numerator relationship matrix; f_j is the full-sib family j as a random effect and e_ij is the random residual error of individual i. Heritability and genetic correlations were estimated using univariate animal model and bivariate animal model, respectively.

The narrow sense heritability was calculated as the proportion of additive genetic variance to total variance as follows:

where ${\sigma }_{{\rm{a}}}^{2}$, ${\sigma }_{{\rm{f}}}^{2}$ and ${\sigma }_{{\rm{e}}}^{2}$ are additive variance components, the full-sib family variance component, and the residual variance component, respectively.

The phenotypic correlation and genetic correlation between the five growth traits were calculated based on phenotype value and breeding value.

Phenotypic correlation (r_p) can be written:

Genetic correlation (r_g) can be written:

where, p_x and p_y are the measurement results of x- and y- growth traits and cov(p_x, p_y) is covariance of p_x and p_y; a_x, a_y are additive genetic effects of x- and y- growth traits, and cov(a_x, a_y) is covariance of a_x and a_y.

Heritability could be classified as low (0.05–0.15), medium (0.20–0.40), high (0.45–0.60) and very high (>0.65) levels. Correlations could be classified as low (0–0.40), medium (0.45–0.55) and high (0.60–1), independent of the sign (Cardellino and Rovira, 1987). When p<0.05 or p<0.01, the correlation reached significant or highly significant level.

Variance components, heritabilities and genetic correlations for BW, BL, TL, CM, and TKL were estimated using derivative-free restricted maximum likelihood estimation as implemented in the MTDFREML package (Boldman et al., 1995). The phenotypic correlations among BW, BL, TL, CM, and TKL were estimated as Pearson’s correlations using the SPSS 13.0 software package (Norušis, 2005). The t test was used to test the significance level of each correlation coefficient from zero. Before analysis, all data were checked for outliers using box plots.

Five growth traits at two growth stages were included in these analyses: 6-month body weight (BW6), 6-month body length (BL6), 6-month total length (TL6), 6-month chest measurement (CM6), 6-month trunk length (TKL6), 12-month body weight (BW12), 12-month body length (BL12), 12-month total length (TL12), 12-month chest measurement (CM12) and 12-month trunk length (TKL12). The growth trait distributions are shown in Table 1. The coefficient of variation (CV) of the four growth traits was generally high at over 10%, with up to 19.238% and 17.649% for BW6 and BW12, respectively. The genetic diversity of BW was the highest. CV decreased gradually with the growth of juveniles. On the whole, the growth tendency of the 48 full-sib families remained the same at the 6-month and the 12-month stages (Figs 2–6).

Estimates of the variance component and the heritability of five growth traits are presented in Table 2. It was found that heritability was 0.37 for BW6, 0.19 for BL6, 0.35 for TL6, 0.29 for CM6, 0.26 for TKL6, 0.36 for BW12, 0.26 for BL12, 0.25 for TL12, 0.31 for CM12 and 0.15 for TKL12. The heritabilities of the other traits were moderate, except for BL6 and TKL12.

Phenotypic and genetic correlations for all growth traits of 6- and 12- month old T. rubripes are shown in Tables 3 and 4. The ranges of phenotypic and genetic correlations estimated for the traits at 6 months were 0.010–0.727 and 0.025–0.725, respectively. The phenotypic correlation between BW6 and BL6 (0.727), and the genetic correlation between BW6 and BL6 (0.725) were the strongest. The ranges of phenotypic and genetic correlations estimated for traits at 12 months were 0.001–0.697 and −0.002–0.706, respectively. The phenotypic correlation between BW12 and TKL12 (0.697) as well as the genetic correlation between BW12 and TKL12 (0.706) were the strongest.