Heritability Estimates of Growth-Related Traits in Oriental River Prawns, Macrobrachium nipponense

. Breeding programs in oriental river prawn populations would beneft from a knowledge of heritabilities for growth-related traits to assess their potential for genetic improvement. Tis study reports the results of heritability estimation and evaluation of epistatic interactions for growth-related traits in oriental river prawn ( Macrobrachium nipponense ). Te results showed that heritability estimates ( h 2 ) were low to medium in magnitude, ranging between 0.10 ( ± 0.02) and 0.22 ( ± 0.01) for females and between 0.08 ( ± 0.01) and 0.26 ( ± 0.02) for males. Te additive × additive epistasis was directional and synergistic for body weight (growth), body length, and two processing traits (abdominal meat weight and abdominal meat percentage). Te presence of biallelic epistasis provides additional new information for the formulation of selection programs. Elimination of some sources of nonadditive efects from phenotypes can lead to more accurate genetic evaluations, and more desirable genetic gains could be achieved through the integration of heritability, additive efects (breeding values), and epistasis in oriental river prawns.


Introduction
Te aquaculture industry of oriental river prawns is growing rapidly in China [1].Tis species is very popular in the consumer market of the middle and lower reaches of the Changjiang River.Its total yield has reached 228,765 tons in 2021 alone [2], with its culture primarily concentrated in the Jiangsu and Zhejiang provinces.Te sustainability of the aquaculture of oriental river prawn in China, nevertheless, is now confronting challenges.Genetic degradation of this species under culture has become a serious problem afecting yield and economic returns.Inbreeding, largely related to the biological features of prawns and seed production practices, are mainly responsible for this issue [3][4][5].Te seeds of oriental river prawns are produced in hatcheries through natural spawning.Brooder prawns may be wildcaught, but are more often selected from adult populations raised in grow-out ponds, and moreover, mature prawns often breed naturally in farmed ponds.Consequently, genetic degradation due to inbreeding often becomes established in cultured stocks.As a result, diminished growth/size at frst maturity occurs.Oriental river prawn juveniles produced from pond-reared parents have a smaller average size than those produced from lake-collected berried females [4,6,7].Although genetic improvement programs have been implemented through crossbreeding and selection in China to overcome the downsides incurred by inbreeding in oriental river prawns [1,8], reliable genetic evaluation for this species is yet to be conducted.
Genetic evaluation allows for monitoring changes in genetic variation and gains for desired traits to be improved, and its accuracy depends upon how much nonadditive variation is removed from phenotypic variability and how large the sample size is [9][10][11][12][13].Heritability estimates have been computed in commercial shrimp populations with diferent partitions of nonadditive components.Variation due to dominance may be confounded with additive genetic variation in estimates of heritability of Penaeus monodon [14,15], Penaeus vannamei [16][17][18], Macrobrachium rosenbergii [19][20][21], Procambarus clarkii [5], Exopalaemon carinicauda [22], and Fenneropenaeus chinensis [23].In the study of Wang et al. [24], the dominance efect and its interaction with the environment were eliminated, but epistasis was confounded with additive variation in red swamp crayfsh P. clarkii.
Epistasis, which describes the nonadditivity of efects between loci and acts as a source of genetic variation, should be a rule rather than an exception when estimating genetic parameters [9][10][11][12][25][26][27].Loci may interact in pairs or in groups of three or more.If a trait is afected by n loci, the number of additive efects is n, whereas the number of biallelic epistatic efects is n 2 .Tus, their cumulative efects may be large as n increases irrespective of higher-order epistatic interactions such as triallelic epistasis.Tis points to the potential for signifcant epistasis in the expression of quantitative traits.In this sense, epistasis should not be neglected.Without epistatic efects partitioned out, estimates of heritability may be overestimated or infated to varying degrees [10][11][12][13].For example, Wang et al. [28] reported that the additive-by-additive interaction had a signifcant efect on body weight and body length in red common carp (Cyprinus carpio), whereas Joshi et al. [29] found that the additive-by-additive epistasis played a dominant role in Oreochromis niloticus.In addition, epistatic interactions play an important role in explaining the expression of heterosis [8,30].For a more accurate genetic evaluation, it is, therefore, necessary to examine the infuence of epistasis.Although the presence of epistatic interactions has been suggested or implied in several aquatic animals, such as Oncorhynchus mykiss [31], Argopecten purpuratus [32], Oreochromis aureus [33], Ictalurus punctatus and Ictalurus furcatus [34], Crassostrea gigas [30,35], and Haliotis midae [36], studies on epistasis in aquaculture species are very few.In particular, no results about epistasis have been reported in crustaceans.
Predicated on an incomplete diallel cross design and the analysis model that included additive, dominance, epistasis, and their interactions with the environment, the heritabilities, additive efects (breeding values), and additive × additive epistasis of nine growth-related traits of oriental river prawns were obtained in the present study.Tese results would contribute to a more reliable assessment of the genetic potential of this species and to more efcient improvement programs.To the authors' knowledge, this is the frst report of heritabilities and epistatic interactions in oriental river prawn using an additive-dominanceepistasis model.

Sourcing and Conditioning of Broodstock.
Te broodstock prawns used for reproduction came from the Siyang aquafarm of Jiangsu Zhengfeng Fisheries Science & Technology Co., Ltd.(Jiangsu Province, China), where three strains of oriental river prawn are maintained: Gaoyou Lake strain (GY), Taihu Lake strain (TH), and Hongze Lake strain (HZ).Te broodstock prawns demonstrated good vitality and were void of any exterior injury.Te area of the earthen ponds for broodstock acclimation was 0.1 hm 2 (water depth 1.2 m).Harmful organisms in the ponds were eliminated by applying quicklime (0.15 kg•m −2 at a water depth of 10 to 15 cm) 10 days prior to stocking of broodstock.Water infow to the ponds was fltered using a fne-mesh screen net to avert the entry of wild fsh and other harmful organisms into the ponds.With a sufcient number of broodstock sampled and measured (Table 1), the parents were stocked at a density of 5 indiv.m −2 .Formula pelleted feed produced by Qingyuan Seashell Biotechnology Co., Ltd., Qingyuan, China (crude protein 43%, fat 7%) was fed to the prawns per day at 3 to 4% body weight daily to provide necessary nutrients for gonad development and to prevent aggressive behaviors such as cannibalism.Males were conditioned separately from females.Water temperature ranged between 22 °C and 25 °C and dissolved oxygen was over 5 mg•L −1 .Tese technical steps ensured good conditioning of breeders and laid a solid foundation for spawning synchrony and highquality seeds.

Mating
Design.An incomplete diallel cross design was adopted in this study to realize the research objectives (Table 2).According to this mating design, the brooders of each combination were restocked (same density as above, and _ : \ � 1 : 3, standard body length >4 cm) in an earthen pond (0.1 hm 2 for each pond and water depth 1.2 m) for random mating.Each pond was planted with water weeds such as Alternanthera philoxeroides, Hydrilla verticillate, or Potamogeton crispus at the edges of the ponds to provide shelter for the breeders.It is necessary to make clear that due to the practical infeasibility to establish families (full-and half-sib) in oriental river prawn, the diallel cross design was adopted in the present study.

2.3.
Hatching and Larviculture for F 1 Generation.Mating in oriental river prawns is generally accomplished prior to spawning.Females begin to spawn within 24 hours after mating, and spawning occurs mostly at night [3].During hatching, the egg-bearing of females was checked every 3 days.Broodstock prawns were fed with the feed as indicated above.Te feeding amount was adjusted to 5 to 8% of their body weight.Water quality management was the same as described above.When the eyespots arose (fertilized eggs turned transparent and pale in color at this time), soybean milk was applied to enrich water and support natural diets (live zooplankton such as Artemia, rotifers, copepods, and cladocerans).Ample natural diets were used for the frst feeding of hatched larvae to ensure larval yield.Egg hatching took approximately 3 weeks (from fertilization to hatching out as zoea).When hatching ended (all zoeal larvae left their mother and fended for themselves), broodstock prawns were removed from the hatching ponds using specialized tools such as ground cages and larval rearing commenced.Larvae began to feed on Artemia nauplii on the second day after hatching out.Te density of Artemia nauplii was kept at a level equal to or larger than 10 4 indiv.L −1 .Te growth and feeding of larvae were 2 Aquaculture Research monitored each day.Dissolved oxygen was over 5 mg•L −1 , ammonia-N was less than 0.02 mg•L −1 , and the water temperature ranged from 24 °C to 26 °C during hatching and larval rearing.

Grow-Out.
Te larval development of oriental river prawns comprises 9 zoeal phases, which typically take about 3 weeks under the range of temperatures given above.When larvae metamorphose, they become postlarvae (PL, body length 0.9 to 1.1 cm), which are very analogous to adults in morphology and ecological habits.Ten, the PL from each mating combination was placed into 3 polyethylene hapas (10 m × 6 m × 1.5 m, 100 mesh cm −2 ) for 7 weeks, with a stocking density of 250 indiv.m −2 .Tere was a total of 27 hapas for 9 mating combinations.All hapas, which were set up in one earthen pond (2 hm 2 ), were 30 cm higher than the water surface and fastened to wood pillars to withstand the impact of winds and waves.One-quarter of each hapa area was planted with weeds for PL to inhabit.Te PL was fed daily from 8: 00 am to 9:00 am and from 4:00 pm to 5:00 pm.Prior to PL stocking, unwanted organisms (wild fsh, frog eggs, and leeches) were eradicated from the ponds using quicklime.Water management was the same as described above; in particular, water quality was checked each day to prevent eutrophication and anoxia.

Production of F 2 Generation.
Epistasis entails data from the F 2 generation, apart from those of F 0 and F 1 ofspring [10][11][12].Large numbers of F 1 progeny from each mating combination were sampled to become parents of F 2 males and females (grown separately; [10,11]).Ten, F 1 populations were sampled for data collection after water drainage.Te production of F 2 began in early May 2020 through matings between males and females coming from within the same mating combination (_ : \ � 1 : 3), and the growth of the F 2 populations ended in late Aug 2020.All technical measures used for the F 2 generation were roughly the same as those used for the F 1 generation.Note.GY, Gaoyou lake strain; TH, Taihu lake strain; HZ, Hongze lake strain.
Aquaculture Research well as their interactions with the environment.Te singletrait mixed model, in matrix notation [10,11], was as follows: where p is the vector of records for each individual trait (BW, BL, CHL, CTL, CTW, SAW, SAH, TMW, and TMP); X is the coefcient matrix relating individual trait records to fxed efects; b is the vector of fxed efects including the overall mean μ, pond efect, and year efect; U A is the coefcient matrix relating individual trait records to additive genetic efect; e A is the vector of additive genetic efects, and e A ∼(0, σ 2 A I); U D is the coefcient matrix relating individual trait records to dominance efects; e D is the vector of dominance efects, and e D ∼(0, σ 2 D I); U AA is the coefcient matrix relating individual trait records to additive × additive epistatic efects; e AA is the vector of additive × additive epistatic efects, and e AA ∼(0, σ 2 AA I); U AE is the coefcient matrix relating individual trait records to additive × environment interaction efect; e AE is the vector of additive × environment interaction efects, and e AE ∼(0, σ 2 AE I); U DE is the coefcient matrix relating individual trait records to dominance × environment interaction efects; e DE is the vector of dominance × environment interaction efects, and e DE ∼(0, σ 2 DE I); U AAE is the coefcient matrix relating individual trait records to additive × additive × environment interaction efects; e AAE is the vector of additive × additive × environment interaction effects, and e AAE ∼(0, σ 2 AAE I); H B is the coefcient matrix of hapa efects; e B is the vector of hapa efects, and e B ∼(0, σ 2 B I); e ε is the vector of residuals and e ε ∼(0, σ 2 ε I); and I is the identity matrix.Te above analysis model corresponded to the mating design used in the present study.Although this model did not account for additive genetic relationships among individuals, additive genetic relationships between individuals from 3 generations in this study were the same (no overlapping generations).Tese relationships were embedded into the mixed model by the diallel cross design [10,11].
Experimental data for the nine growth-related traits of oriental river prawn were collected for two years.Tus, years were regarded as two diferent environments to account for genotype-environment interactions.

Heritability Estimates and Additive Genetic Prediction.
Variances for the mixed genetic model were estimated using restricted maximum likelihood (REML).Tis method has been applied to solving various types of complicated genetic models [10][11][12].Variances estimated by the mixed model for each trait were used to compute estimates of genetic variance components using the following formulae [10,11]: where  V A is the estimate of additive genetic variance;  V D is the estimate of dominance genetic variance;  V AA is the estimate of additive × additive genetic variance;  V AE is the estimate of the additive × environment interaction variance;  V DE is the estimate of dominance × environment interaction variance;  V AAE is the estimate of interaction variance between additive × additive and environment.Narrow-sense heritability (h 2 ) was estimated as the ratio of the additive genetic variance to the phenotypic variance (V p ), i.e., 2 /n − , where p i is the phenotypic record and n is the sample size.Te single-trait mixed model used here permits the estimation of variance components and heritabilities as well as the prediction of random efects [10][11][12]37].Te best linear unbiased prediction (BLUP) was used to predict the additive and biallelic epistatic efects of the traits under consideration.A repeated sampling technique (Jackknife) was used to compute the standard errors of varying estimates and predictions, and parameters were tested using t statistic for signifcance [10,11].All computations were conducted using QGAStation software (v 2.0, Zhejiang University, China).

Results
Survival rates ranged between 29.50% and 35.60% at the end of the growth period in generations F 1 and F 2 .Diferences in average survival among mating combinations and among hapas were not signifcant (P > 0.05).Between 10.50% and 12.10% of the females in the F 1 and F 2 generations were ovigerous.Tese females were excluded from the study because egg-bearing could retard their growth (i.e., heritabilities of growth-related traits in the ovigerous female population were not estimated in this study).In addition, no diseases occurred due to good management of the whole experimental process.Although there were lower survival rates in F 1 and F 2 generations, the progenies used for this study were sufcient.
3.1.Description of F 0 , F 1 , and F 2 Generations.Te phenotypic means and coefcients of variation of the 9 growth-related traits of oriental river prawns in F 0 , F 1 , and F 2 generations are given in Table 1.Tese statistics were obtained based on samplings from three generations.Body weight and morphological traits (BL, CHL, CTL, and CTW) of males were higher than those of females.Two processing traits, TMW and TMP, of males were higher than those of females.
Conversely, the SAW and SAH of females were higher than those of males.Second abdominal segment width and height are important for female fecundity.Tese two traits can form a ventral brood chamber in which eggs are carried.Te higher the SAW and SAH, the greater the egg-bearing capacity.Relative phenotypic variations (coefcient of variation, CV%) varied from 8.19% to 27.85%.

Variance Components.
Estimates of variance components of the 9 growth-related traits for females and males of oriental river prawn are shown in Tables 3 and 4. Variations of nearly all sources for the 9 growth-related traits were detected, but  V AAE of TMW failed to be detected in both sexes; the additive × environment interaction variance (  V AE ) of TMW was not detected in females, it was very small for males either.Values of estimated variance components varied with trait and sex.For example,  V A of BW was 0.11 ± 0.01 g 2 in females and 0.16 ± 0.02 g 2 in males; the additive × additive genetic variance (  V AA ) of BW was 0.05 ± 0.01 g 2 in females and 0.10 ± g 2 in males.Te additive-genetic variance (  V A ) of CHL was 5.05 ± 0.03 mm 2 in females and 7.01 ± 0.05 mm 2 in males; the additive × additive genetic variance (  V AA ) of CHL was 1.25 ± 0.02 mm 2 in females and 1.45 ± 0.02 mm 2 in males.Te additive-genetic variance and additive × additive genetic variance (  V A and  V AA ) of TMW were 0.01 ± 0.01 g 2 and 0.0056 ± 0.01 g 2 , respectively, in females, and were 0.02 ± 0.01 g 2 and 0.006 ± 0.01 g 2 , respectively, in males.In contrast to the estimates of additive and epistatic genetic variances for other traits, those of TMW were the smallest.

Estimates of Heritability.
Estimates of heritability for the 9 growth-related traits for females and males of oriental river prawn are given in Tables 5 and 6, respectively.Ratios of the other 6 variance components to phenotypic variances are also presented in the two Tables.All heritability estimates were obtained using the data collected at the time when the selection program was generally pursued in China [4,5].Tere exists sexual dimorphism in the oriental river prawn.Considering that selection pressure exerted on males is greater than that on females in this species during actual selective breeding, heritability estimation was undertaken within each sex of the oriental river prawn in this study.Heritability estimates for the 9 growth-related traits were statistically signifcant at a 5% level.Heritabilities were low to moderate in value and varied from 0.10 (±0.02) for CTW to 0.22 (±0.01) for BW in females and from 0.08 (±0.01) for SAW to 0.26 (±0.01) for BW in males.Heritabilities of BW, CTL, and two processing traits (TMW and TMP) stood within moderate ranges.Te percentages of phenotypic variation explained by the  V AA of BW, BL, CHL, CTL, SAW, SAH, TMW, and TMP were signifcant in females (P < 0.05), while the percentages of variation attributable to the  V AA of CTW, SAW, and SAH statistically did not difer from zero in males (P > 0.05).

Prediction of Genetic Efects.
Predicted values of additive and epistatic (additive × additive) genetic efects of the 9 growth-related traits for females and males from 3 strains of oriental river prawn are provided in Tables 7 and 8, respectively.Te additive genetic efects of BW, BL, CHL, CTL, CTW, TMW, and TMP were signifcant (P < 0.05).Te epistatic genetic efects of BW, BL, CHL, TMW, and TMP were signifcant (P < 0.05), while those of CTL and CTW were nonsignifcant (P > 0.05) in females and males.Te additive and epistatic genetic efects for SAW and SAH were not detected in each sex of the 3 strains.For BW, the additive genetic efects in the HZ strain were greater than those in GY and TH strains in both sexes; the epistatic genetic efects in each sex of the 3 strains were all positive and favorable to growth.Additive and epistatic genetic efects for BL, CHL, TMW, and TMP were similar to those for BW.

Discussion
For more accurate genetic evaluation, the additive genetic variations of the 9 growth-related traits in oriental river prawn were estimated using an incomplete diallel cross design and a complex linear mixed model.Additive genetic variation is tightly enveloped by nonadditive variations, thus unveiling additive genetic variation is not an easy task [9,12].What matters the most is to estimate the additive genetic variation as accurately as possible [10,11,38].Te mere separation of dominance and dominance × environment interaction from phenotype may not sufce [24].Empirical evidence confrmed the presence of epistasis in red common carp, C. carpio [28], and in Nile tilapia, O. niloticus [29].Unlike the studies of Xu et al. [39] and Li et al. [40] in whiteleg shrimp, P. vannamei, of Kitcharoen et al. [37] and Luan et al. [41] in giant river prawn M. rosenbergii and analogous studies in other crustaceans, the variances due to additive × additive, additive × environment, and additive × additive × environment interactions were separated from the additive variance in the present study.Tus, relatively "cleaner" estimates of the additive genetic variations were obtained (Tables 3 and 4).For instance, the CHL variances attributable to dominance, additive × additive, dominance × environment, additive × environment, and additive × additive × environment were 5.02, 1.06, 4.13, 2.32, and 2.11 mm 2 in females, respectively, and 6.04, 1.51, 6.85, 2.78, and 2.46 mm 2 in males, respectively.Tese results clearly show the need for separating nonadditive sources of variation out of phenotypic variances.Te heritability estimates of growth (body weight), body length, and two processing traits fell close to the lower end of the range (Tables 5 and 6).Only one report of heritability estimates of body weight and body length in 97-day-old oriental river prawn has been published to date [42].Estimates obtained in Aquaculture Research this study are comparable to theirs.Due to the inability to analyze the maternal efect with incomplete diallel (cross design), it, if any, was confounded with the additive efects in the present study.Joshi et al. [29,43] reported the presence of maternal efect on body weight at harvest, body depth, body length, and fllet weight in Nile tilapia, O. niloticus.Separation of various nonadditive sources of variation including the common environmental efect as well as larger sample sizes guarantees relatively more accurate genetic evaluation.Te data from 3 generations of oriental river prawns were used for the separation of additive epistasis (Table 1).If the generation gap is larger, the experimental design used in this study may be unfeasible because this would take too long.
Tere are many possible types of epistasis.For biallelic epistasis, for example, there are 3 forms, i.e., additive × additive, additive × dominance, and dominance × dominance [26].Te last two epistatic interactions cannot be separated from phenotype based on the analysis model and mating design used in the present study, thus they were confounded with the additive efects.Tey are, however, generally ignored in genetic analyses of animals and plants due to their relatively weaker efects [10-12, 26, 27].Since additive × additive epistasis is a type of genetic efect that can be modifed through selection, it needs to  Table 4: Estimates of variance components (±SE) for nine traits in male Macrobrachium nipponense using the REML approach.

6
Aquaculture Research be analyzed together with additive and dominance efects [12].Terefore, only additive × additive epistasis (biallelic) was examined in this study using the experimental data of 3 generations of oriental river prawns (Table 1).Results given in this study showed that the existence of additive × additive epistasis was dependent on traits and was directional (Tables 7 and 8).For instance, the additive × additive epistasis, although not signifcant, was negative for cephalothorax length and width.In contrast, the additive × additive epistasis was positive for body weight, body length, and chela length in both sexes of oriental river prawn, indicating that the loci underlying these traits interacted epistatically in a synergistic manner.Moreover, the Table 6: Proportions and standard errors of phenotypic variation accounted for by variance components of the 9 growth-related traits in male Macrobrachium nipponense.additive × additive epistasis for traits with higher heritabilities was relatively larger.For instance, the heritability of body weight for females was 0.22 ± 0.01, and the additive × additive epistasis ratio for this trait was 0.10 ± 0.04, 0.15 ± 0.02, and 0.18 ± 0.03 in the Gaoyou strain, Taihu strain, and Hongze strain of oriental river prawn, respectively.Tis is because higher levels of additive variation occurred [12,13].Additive × additive epistasis was detected in other aquatic animals.For instance, Wang et al. [28] concluded that synergistic additive × additive epistasis was present for the body weight and body length of the red common carp, C. carpio.Palti et al. [33] reported strong synergistic epistasis between two microsatellite marker loci (UNH159 and UNH216), where an additional deleterious mutation could greatly decrease the ftness of blue tilapia O. aureus.Results presented in this study are analogous to those of these two studies, but they are inconsistent with those of Pante et al. [31], who implied no presence of additive × additive interaction for growth (body weight at harvest) of rainbow trout, O. mykiss.
Based upon the linkage disequilibrium between many candidate loci under molecular-driven selective breeding, Dale-Kuys et al. [36] suggested the occurrence of possible epistatic interactions in the genome of South African abalone (H.midae) but did not demonstrate the directionality and magnitude of the epistatic interactions.In this study, the additive × additive epistasis was also synergistic for two processing traits (tail meat weight and tail meat percentage) in both sexes of oriental river prawns (Tables 7  and 8).Tis result is similar to that of Argue et al. [34], who found that greater epistatic interactions could signifcantly increase the dress-out percentage and fllet percentage in channel catfsh I. punctatus.Results of this study indicate that epistasis is an important genetic basis for economic traits such as body weight (growth) and processing traits in oriental river prawns.
For Mendelian traits such as coloration, similar results regarding the existence of epistasis have also been reported in a couple of aquatic species.Winkler et al. [32] suggested the existence of a dominant epistasis that governs the general pattern of coloration in scallop A. purpuratus.Ge et al. [35] found an epistatic efect of shell background color on shell foreground pigmentation in Pacifc oyster, C. gigas.
Because the variation ascribable to additive × additive epistasis is heritable, it, in conjunction with additive variation, should be taken into account when formulating an improvement program.Predicated on the results of the present study (Tables 5 and 8), heritabilities, additive × additive epistasis, and additive genetic efects can provide detailed information for the traits to be improved, which may result in additional genetic gains when implementing selection programs.According to the directionality of additive × additive epistasis, the traits that have higher heritability, additive efects, and synergistic epistatic interactions can be considered candidates for improvement.In this study, for example, the breeding value (additive efect) and additive × additive epistasis for body weight in the Hongze Lake strain of oriental river prawn were relatively higher, 0.17 ± 0.03 g and 0.18 ± 0.03 g, respectively, for females and 0.15 ± 0.02 g, and 0.12 ± 0.03 g, respectively, for males.Hence, if this strain were selected as the parents, preferable genetic gains would be garnered.For the body weight of oriental river prawn in the Gaoyou Lake and Taihu Lake strains, the higher synergistic additive × additive epistasis would also result in better genetic gains if taking the two strains as parents for selective breeding, albeit with relatively lower additive efects.Although signifcant synergistic additive × additive epistasis and additive efects occurred with two processing traits, it was very difcult for them to be directly selected or improved.Te epistatic interactions, additive efects, and heritabilities of other traits, such as cephalothorax length, cephalothorax width, and second abdominal segment width and height, were absent or had very low values, thus they would be of little beneft to the selection program.
Te choice of mating design is very important when estimating genetic parameters for aquaculture species [44].Te nested design has been used by many researchers such as Pérez-Rostro and Ibarra [16], Li et al. [5], and Joshi et al. [29].A diallel cross design, however, was used in this study because the establishment of families and identifcation of individuals in oriental river prawn was impracticable due to the biological characteristics of this species.Although genetic markers can be used to construct families, the cost would be prohibitively high for this study.Tus, the individual animal model used by some researchers such as Joshi et al. [43] and Cardoso et al. [45] was not appropriate for the data analysis in this study.In many aquaculture species, individual animal identifcation is a challenge, which was the case in the current study.Tis is because the oriental river prawn is morphologically very small, and ecdysis takes place about 20 times throughout its life [46].Tis would explain why strategic mating designs could be used to try to devise selection programs in oriental river prawns.Te diallel design has been widely applied to genetic assessment in many aquaculture species, such as red common carp, C. carpio [28]; giant river prawn, M. rosenbergii [19,47,48]; Pacifc oyster, C. gigas [49]; red swamp crayfsh, P. clarkii [24,50]; and pearl oyster, Pinctada fucata [46].Based upon the single-trait mixed genetic analysis model utilized in this study, the diallel design can be used for genetic evaluation of oriental river prawn using the specifc approach of Zhu [10,11].Since the genetic linkage map for the oriental river prawn has yet to come out, genomic information was not utilized for the genetic assessment of the oriental river prawn in the present study.

Conclusion
Te existence of ample additive genetic variation is a prerequisite for desirable genetic gains for traits to be improved in selective breeding.Tus, it is necessary to make an accurate genetic assessment of the existing additive genetic variation in a population prior to developing a selection program.In the present study, due to sufciently large sample sizes, the additive × additive epistasis, other nongenetic efects, and their interactions with the environment were accounted for when estimating additive genetic variance and breeding values.Tere was greater relative additive genetic variation for body weight (growth) and processing traits to be utilized for the selective breeding of oriental river prawns.Te level of epistasis varied according to the trait of 8 Aquaculture Research interest.Te genetic determinism of these traits depends quantitatively on not only signifcant additive genetic efects but also synergistic epistatic interactions.Te epistatic interactions provided new information on the genetic determinism of growth-related traits in oriental river prawns.Desirable genetic gains for this commercial species would be achieved for traits such as growth with superior additive genetic efects and synergistic epistatic interactions.

Table 1 :
Phenotypic means and coefcients of variation (CV) of nine growth-related traits of three oriental river prawn strains for 3 generations.

Table 2 :
Mating design of an incomplete diallel cross between three strains of Macrobrachium nipponense.

Table 3 :
Estimates of variance components (±SE) for nine traits in female Macrobrachium nipponense using the REML approach.

Table 5 :
Proportions and standard errors of phenotypic variation accounted for by variance components for the 9 growth-related traits in female Macrobrachium nipponense.

Table 7 :
Predicted values of additive and epistatic (additive × additive) genetic efects for nine traits in three strains of female Macrobrachium nipponense using the BLUP approach.

Table 8 :
Predicted values of additive and epistatic (additive × additive) genetic efects for nine traits in three strains of male Macrobrachium nipponense using the BLUP approach.