Genotype by sex interaction at different phases during Nile tilapia culture period

Co(variance) components and genetic parameters were estimated for performance and morphometric traits in male and female Nile tilapia at different growth periods to verify the need for separate selection programs for the two genders. Data set contained information on 1,720 animals, collected in cage system. Two-trait analyses using Bayesian methodology were conducted and the records of males and females were considered distinct traits. Rates of additive genetic variance, phenotypic and heritability were higher for females in estimates for morphometric traits and higher for males in performance traits. Whereas common hatchery environment showed higher relative importance for males, the nursery caused greater variations in females. The reduction of the genetic correlation rates caused by growth increased the difference between genders and indicated the treatment of males and females as a distinct feature selection.


Introduction
Nile tilapia males revealed a higher growth rate when compared to females.The difference depends on several factors such as species capacity, ingestion, feed conversion, environmental factors and behavior (Toguyeni et al., 2002).Phenotypic differences between males and females in most water species are of great commercial interest.Since the differences are quantitative, several genes may be expressed in different manners between the genders, for instance, greater growth and late sexual maturity in female salmonids (Kause, Ritola Paananen, Mäntysaari & Eskelinen, 2003).Nguyen, Khaw, Ponzoni, Hamzah and Kamaruzzaman (2007) did not report genetic differences in Nile tilapias when they studied heredity for body weight and body shape in males and females, suggesting that there was no need of any differentiated selection between the genders.Non-different genetic correlations in studies by Rutten, Komen and Bovenhuis (2005a) show that body weight is controlled by the same genes in males and females.
Sexual dimorphism in the Nile tilapia is a fact, although the need of selection programs for males and females has never been reported (Lind et al., 2015;Lozano et al., 2014).This has been due to the fact that growth is involved with genderdetermining genes rather than with the reproduction onset (Toguyeni et al., 2002).Several studies are required to identify the differences between males and females due to increase in growth rates because of animals with high genetic capacity.If differences are detected, the use of different selection programs will be required to increase selection accuracy and genetic gain.
Current assay investigates the need for distinct selection programs for males and females by estimating co(variance) components, effects of a common environment, hereditability, genetic and phenotypic correlations for performance and morphometric traits of Nile tilapias in four different phases during the cultivation period.

Material and methods
Data obtained from the PeixeGen Research Group of the State University of Maringá, Maringá, Paraná State, Brazil, contained information on 1,720 animals from 58 families, coupled to pedigree information on 5,600 animals.Details on the formation process of the families have been described by Yoshida et al. (2013).Cultivation period ranged between June and October 2009 in cages at the Nile Tilapia Production Unit in the Rio do Corvo, Diamante do Norte, Paraná State, Brazil (22°39′21″ S; 52°51′36″ W).Two 6 m 3 (2 x 2 x 1.5 m) cages were used, with the same density and with specimens of all families in both cages.
During the evaluation period of production performance, the animals received a commercial diet composed of 2,800 kcal kg -1 digestible energy, 28% crude protein, 4.0% ether extract, 2.5% calcium, 0.9% phosphorus and 150 mg kg -1 vitamin C. Diet was provided three times a day following feed instructions of the distributor, taking age and biomass of fish and water temperature into account.
Four measurements were undertaken at intervals of approximately 37 days during the five-month culture period in the caged water tanks.Data on gender, live weight, standard length, width, height and daily weight gain (DWG) of each specimen, obtained by the ratio of live weight in each measurement and age, was registered.DWG1, DWG2, DWG3 and DWG4 refer respectively to daily weight gains till the first, second, third and fourth measurements.Table 1 shows the age of animals and phenotypic means of the analyzed traits in each biometry.
Estimate of the components of (co)variance and genetic parameters was calculated by two-traits analyses in which weights, weight gains and morphometric traits of males and females were distinct features, as follows: where, y i = observation vectors of the characteristics for males (1) and females (2); = vectors of environmental effects for males (1) and females (2), taking the caged water tank as fixed effect and age as co-variable; a, c, w and e are vectors of direct additive genetic effects, common environment effect of larva culture (due to the maintenance of the animals with their dams from spawning till the end of the reproduction season), effect of common nursery environment (related to the management in maintaining specimens of families in hapas distributed in different sites in nurseries) and of randomized errors respectively for males (1) and females (2); X, Z, C and W, which are the matrixes of incidents of identifiable environmental effects, direct genetic additives, common larva and nursery culture environment respectively for males (1) and females (2).If a, c, w and e have normal multivarious set distribution, then where A is the kin matrix between the animals; is the direct additive genetic variance; , and are variances of common environmental effects of larva culture, nursery and residues, respectively; identity matrix of the order h, with h equal to the number of hapas of the larva culture; identity matrix of the order c, with c equal to the number of hapas of nursery; identity matrix of the order n, with n equal to the number of observations.For two-traits analyses, , in which is the matrix of the characteristics´ genetic (co)variances; , in which is the matrix of the (co)variances relative to the effect of larva culture common environment; , in which is the matrix of (co)variances relative to the effect of the nursery´s common environment; in which is the matrix of residual (co)variances.Multiple Trait Gibbs Sampler for Animal Models (MTGSAM) was employed (Van Tassell & Van Vleck, 1995), which executes Bayesian estimates; normal a priori distribution may be taken into account for additive genetic effects, common environment of larva culture, nursery and residual.Inverted wishard distribution was taken into account for (co)variance components.
Initially 500,000 cycles were performed and increased till convergence was attained.Sampling interval comprised 10 cycles after the elimination of the first 50,000 cycles, totaling at least 45,000 samples.Hedielberger & Welch test (Cowles, Best & Vines, 1995) was employed to evaluate chain convergence, implemented in Convergence Diagnosis and Output Analysis (CODA), of R (version 2.12.0).

Results and discussion
All samples converged in the two-traits analyses by Bayesian methodology.Results showed increasing rates of additive and phenotypic genetic variance for the two genders.The males´ genetic variances were greater when compared to the females´ for weight and DWG, whereas the opposite was detected for the morphometric traits.Ratio of genetic and phenotypic variance rates between the genders decreased for weight and DWG over time.In the case of morphometric traits, variance ratios were higher than 1.40, in spite of oscillations mainly in intermediate biometry, with a reduction in rates throughout the measurements (Tables 2 and 3).
Hereditability estimates for weight and DWG ranged between medium and high magnitude (Table 4).Morphometric traits in females revealed high greatness rates which remained constant during the evaluation period, whereas hereditability in males ranged between medium and low magnitude (Table 5).a hatchery) till identification.Therefore, restriction of physical space and different environmental conditions from where the specimens are selected may have contributed for not observing sexual dimorphism in the first biometry.However, transference to the commercial culture system in caged water tanks where the environment is proper for the breeding of animals, with adequate density, better water quality and less temperature variation, may have enhanced males´ performance.It must have provided the best conditions for their genetic capacity and evidenced the differences in the growth of the animals.
According to Oliveira et al. (2013), studies on growth curves of Nile tilapia which were genetically improved within Brazilian conditions of culture, showed that sexual dimorphism may be observed by body weight as from 165 days of life.Since in current assay animals aged between 170 and 285 days were used, the evaluation of animal performance during measurements evidenced differences between the genders.
Only Rutten et al. (2005a) had previously reported experiments on the sexual dimorphism of the Nile tilapia throughout the culture period.Although genetic correlation rates for body weight between males and females close to 1 and the lowest greatness in the last measurements were reported, the authors failed to observe sexual dimorphism.
Further, Nguyen et al. (2007) estimated genetic parameters for males and females and the possibility of applying different selection strategies for the two genders but results disagreed with those in current research.The above-mentioned authors reported that, due to high positive correlations (0.91-0.96) for performance and morphometric traits in males and females, there was no need to deal with the genders as distinct in genetic improvement programs.Their results may be because the authors used information derived from only one biometry and sexual dimorphism may not have manifested itself sufficiently to require differentiated selection programs.Moreover, results may be divergent since estimates of genetic parameters are intrinsic to the population and to the environment in which the animals were assessed (Santos et al., 2011).
In their studies on GIFT strain, Bentsen et al. (2012) registered the genetic correlations for weight between males and females in hatcheries (0.78) and in caged water tanks (0.88), demonstrating a medium interaction between genotype and gender.The authors suggested that growth rate was more influenced in hatcheries due to early sexual maturity and reproduction when compared to animals in caged tanks.The above shows that the verification of the genotype and gender interaction for tilapias in caged tanks had not been performed prior to current assay.

Conclusion
Results demonstrate that the two common environments reveal important differences for males and females.Hereditability rates of genetic and phenotypic correlation reveal that the traits under analysis respond distinctly to selection for the two genders.Specific detection programs for males and females are required as the age of animals increases.

Table 1 .
Age (days)and phenotypic means in performance, morphometry and quantity (n) of males and females and their respective standard deviation (±SD) in different measurements (MEA) of the Nile tilapia.

Table 2 .
Estimates of additive genetic ( ), phenotypic ,residual ( ) variance rates for males and females of their respective variances for performance in different measurements (MEA).

Table 3 .
Estimates of additive genetic ( ), phenotypic , residual ( ) variance rates for males and females of their respective variances for morphometric traits in different measurements (MEA).