Heritability estimations of ammonia tolerance and survival of ridgetail white prawn Exopalaemon carinicauda*

Ding LÜ , , Chengsong ZHANG , , Yang YU ,, Jianhai XIANG , Fuhua LI ,4 ,

1 Key Laboratory of Experimental Marine Biology, Institute of Oceanology, Chinese Academy of Sciences, Qingdao 266071,China

2 Laboratory for Marine Biology and Biotechnology, Pilot National Laboratory for Marine Science and Technology (Qingdao),Qingdao 266237, China

3 Center for Ocean Mega-Science, Chinese Academy of Sciences, Qingdao 266071, China

4 The Innovation of Seed Design, Chinese Academy of Sciences, Wuhan 430072, China

Abstract Ridgetail white prawn Exopalaemon carinicauda is an important economic marine species in China. To improve the germplasm quality of cultured E. carinicauda, for the first time, we estimated the heritability of ammonia tolerance and survival of E. carinicauda, covering 263 full-sib families in multiple generations. Analysis models including binary linear animal model (BLA), binary linear sire-dam model(BLSD), probit threshold animal model (PTA), and probit threshold sire-dam model (PTSD) were used.The estimated heritability of tolerance to ammonia in diff erent models ranged from 0.09 to 0.22, and that of survival rate ranged from 0.06 to 0.16. Concerning the tolerance to ammonia, both the PTSD and BLSD models showed comparatively better prediction accuracies (0.48 and 0.47, respectively) than PTA and BLA models (0.40 and 0.26, respectively), while to the survival rate, all four models presented relatively low predictive abilities (0.16-0.26). This study provided important information for the future breeding program of E. carinicauda.

Key word: heritability; Exopalaemon carinicauda; tolerance to ammonia; survival rate

1 INTRODUCTION

The ridgetail white prawn,Exopalaemoncarinicauda(Crustacean: Decapoda: Palaemonidae),is an economically important marine prawn,distributed naturally along China’s coast and the west coast of the Korean Peninsula (Li et al., 2003). With fast growth, strong reproduction ability, wide environmental adaptability, and delicious meat, it is one of the most important species in both aquaculture and fishery activities in China. However, farming on this species mainly relies on the capture of post-larvae or parent prawns from coastal areas, and this primitive method depletes wild resources (Bashirullah et al.,1989) and introduces the risk of pathogen dissemination, especially virulent viruses and bacteria(Lightner et al., 1998), which has led to serious economic loss in shrimp industry around the world(Flegel, 2009; Thitamadee et al., 2016). Besides,culturing wild prawns without artificial breeding greatly restricted the production effi ciency. To continue developing aquaculture industry ofE.carinicauda, genetic improvement on the main economic traits is particularly important.

Ammonia, especially nonionic ammonia, is the most common toxic environmental factor in aquaculture, due to direct discharge of cultivated animals and ammonization of residual feed and excrement (Armstrong et al., 1978; Yu and Hirayama,1986). It was reported that high concentrations of ammonia can lead to harmful eff ects on cultured prawns, such as reducing growth (Chen and Kou,1992; Chen and Lin, 1992a, b), weakening immune system (Liu and Chen, 2004; Wang and Chen, 2005)and increasing mortality (Chen et al., 1990; Lin et al.,1993). Therefore, tolerance of prawns to ammonia stress is an important trait, which enables them to be adaptive to relatively high concentration of ammonia thus to raise the production. Additionally, the output ofE.carinicaudadepends greatly on the survival rate during cultivation.

Selective breeding is a basic approach for genetic improvement in aquatic species, which off ers an opportunity of continuous genetic gain, and can be multiplied and expressed in a huge number of progenies in aquatic species (Gjedrem et al., 2012). In the past decade, numerous shrimp selective breeding programs in Asia and America were initiated and have made considerable achievements in a broad range of traits, including growth, survival, and resistance to a specific pathogen (Hetzel et al., 2000; Argue et al.,2002; Gitterle et al., 2007; Luan et al., 2012, 2014;Sui et al., 2016). However, studies and breeding programs ofE.carinicaudaare scarce.

To improve the germplasm quality of culturedE.carinicauda, a breeding project aimed at increasing the tolerance to ammonia and survival rate was conducted in 2012-2016. Although complete pedigree of the prawn of five generations and their phenotypic data were available, information on the genetic parameters remained limited. In this study, the heritability of ammonia tolerance and survival ofE.carinicaudawas estimated in four statistical models.The predictive performance of these models was compared to determine the most optimal fitting model.The results obtained in this study provide valuable information for subsequent selection breeding.

2 MATERIAL AND METHOD

2.1 Material

The breeding program was performed at Nantong Juhao Company, Jiansu Province, China. The base generation (G0), including 57 full-sib families, was bred by collecting females carrying eggs from Laoba port near Hai’an County, Jiangsu Province in 2012.When the off spring were cultured to the stage of postlarvae, 500 individuals from each family were selected randomly and reared in 200-L white plastic tank separately based on a family-culturing unit. To ensure that the rearing conditions were kept as consistently as possible, all the plastic tanks were put in the same workshop. Eighty percent of the water was changed every 2 d, using disinfected seawaterfrom the same reservoir. Water temperature was varied with air temperature (22-32 ℃), at the salinity of 20±2, pH of 8.2-8.5. The light source was natural light. After 50 d, the survival rate of each family was calculated. Meanwhile, another 40 individuals from each family were selected randomly to test their tolerance to ammonia exposure. The testing ammonia concentrations were determined by preliminary experiments to ensure that about half of the prawns died within 96 h after ammonia exposure. The total ammonia nitrogen concentrations used for formal testing were 50 mg/L in G0 and G1; 55 mg/L in G2-G4. After the tolerance to ammonia nitrogen was tested for each family, the top 20 families for each trait were selected to produce G1 by mating among families in 2013. The subsequent generations (G2-G4) were established using similar approaches as G1.A total of 263 full-sib families in 5 generations were used for data analysis. In total, 10 220 prawns were involved in the ammonia tolerance test and 59 250 prawns in the survival test (Table 1). Phenotypes of alive and dead individuals after tolerance to ammonia nitrogen test and survival test were recorded as 1 and 0, respectively.

Table 1 Numbers of full-sib families and prawns tested

2.2 Genetic analysis

In this study, genetic parameters for two traits were analyzed using both linear mixed model and threshold generalized linear mixed model. Additionally, each type of model contained both animal model and siredam model. Four models were applied as follows:

2.3 Model comparison

To compare predictive abilities of four models in this study, we used phenotype dataset containing the first four generations (2012-2015) to calculate family estimated breeding value (EBVs) in 2016 from each model. Then, Spearman correlation coeffi cients between family EBVs and actual family average phenotype (survival rate) in 2016 were calculated.Since we paid more attention to the ranking diff erence of two sets of data, the Spearman correlation coeffi cient was adopted. The EBVs for families were estimated using ASReml and Spearman correlation coeffi cients were calculated using R (R Core Team,2018).

3 RESULT

3.1 Descriptive statistics for phenotype data

The coeffi cient of variation of the survival rate in G0 was 27.21%, being higher than those of subsequent generations, ranging from 5.40% to 10.67%, and the data of cultured survival rate in 2013 were missing(Fig.1).

The coeffi cients of variation in family survival rate after 96-h acute ammonia stress test are relatively larger, ranging from 20.50% to 59.08% (Fig.1). And the overall survival rate after tolerance to ammonia test in 2012-2016 was 47.72%, 49.35%, 50.76%,48.71%, and 44.22%, respectively.

3.2 Genetic analysis

Variance components, heritability for ammonia tolerance and survival calculated using diff erent models are shown in Table 2. The heritability estimatesof tolerance to ammonia ranged from 0.09±0.01 to 0.22±0.03, and those of survival ranged from 0.06±0.01 to 0.16±0.02. Heritability for tolerance to ammonia was overall larger than that for survival. For both traits, estimates from sire-dam models (models 2 and 4) were close and they were higher than those from the other models. Estimated heritability using diff erent models was highly significantly diff erent from zero (P＜0.01) through Z-score.

Fig.1 Distribution of family survival rate after being raised for 50 d under normal conditions (a) and after 96-h acute ammonia stress (b) in each generation

Table 2 Variance components, heritability of tolerance to ammonia and survival in four statistical models

Table 3 Genetic correlation ( γ g) and phenotypic correlation( γ p) between tolerance to ammonia and survival in models 1 and 3

Genetic and phenotypic correlations between ammonia tolerance and survival calculated using animal models are shown in Table 3. Genetic correlations were 0.20±0.10 and 0.20±0.04 based on model 1 and model 3, respectively. Phenotypic correlations were 0.29±0.13 and 0.09±0.02,respectively. All of them were low but significantly diff erent from zero (P＜0.05).

The family EBVs of two traits were also obtained from diff erent models and their relationships are presented in Table 4. The Pearson correlation coeffi cients between EBVs calculated using diff erent models were all greater than 0.9.

3.3 Comparisons among diff erent models

Spearman correlation coeffi cients between family EBVs based on data of 2012-2015 and true family average phenotype (survival rate) in 2016 are presented in Table 5. For tolerance to ammonia,Spearman correlation coeffi cients under sire-dammodels (models 2 and 4) were larger than those obtained from animal models (models 1 and 3), which reached a moderate correlation level. However, for survival, all models showed a low predictive ability(Table 5).

Table 4 Pearson correlation coeffi cients among EBVs from diff erent models

4 DISCUSSION

4.1 Genetic parameter

Understanding the additive genetic variance existing in the target population was the foundation ofinitiating a selection-breeding program. One of the main objectives of this study was to estimate the heritability of tolerance to ammonia and survival inE.carinicauda. The results showed that heritability for tolerance to ammonia or survival based on four models was all highly significantly diff erent from zero (P＜0.01), which indicated that improvement on these two traits inE.carinicaudawas theoretically feasible. Furthermore, heritability for tolerance to ammonia estimated in this study belonged to the medium, based on the following categorization: low(0.05-0.15), medium (0.20-0.40), high (0.45-0.60),and very high (＞0.65) (Cardellino and Rovira, 1987),and that for survival was lower than tolerance to ammonia. Until present, knowledge about the heritability of ammonia tolerance or survival in crustaceans is still limited. The heritability for tolerance to ammonia inLitopenaeusvannameijuveniles was estimated to be 0.148-0.264 (Li et al.,2016; Lu et al., 2017), and it was reported as 0.13±0.06 inMarsupenaeusjaponicus(Jiang et al., 2017). For the survival trait, the heritability was estimated to be 0.04-0.10 inL.vannamei(Gitterle et al., 2005, 2007)and 0.02-0.14 inMacrobrachiumrosenbergii(Luan et al., 2014; Vu et al., 2017). Estimates in the present and previous studies were similar for both traits. What called for special attention was that we did our best to maintain the environmental conditions consistent among the families in the trait-testing process, but subtle diff erences were inevitable. Besides, theabsence of common environmental eff ects may lead to the bias of heritability estimates (Gallardo et al.,2010).

Table 5 Spearman correlation coeffi cients between family EBVs within G4 based on dataset consisting of G0–G3 and actual family average phenotype(survival rate) within G4 in diff erent models

Analysis of the genetic correlation and phenotypic correlation between ammonia tolerance and survival showed a low but significant positive genetic correlation (0.20±0.10 and 0.20±0.04, respectively).This result suggested that selection on one trait would not cause a negative response to another one.

4.2 Model comparison

In previous quantitative genetics studies, the predictive abilities of statistic models were evaluated through model cross-validation (Gitterle et al., 2006;Saatchi et al., 2011; Rezende et al., 2012; Lever et al.,2016). In this study, we used a correlation between the actual family average phenotype in G4 and their family EBVs calculated from each model as the criteria for model predictive abilities to make the statistical model closer to practical application.Although the family EBVs calculated from diff erent statistical models showed high correlations (Table 4),the predictive abilities were varied considerably among models (Table 5). For the tolerance to ammonia, the predictive abilities of threshold siredam model (model 4) were slightly higher than the linear sire-dam model (model 2), and clearly higher than animal models (models 1 and 3), suggesting that sire-dam models (both linear and threshold) are the optimal models. Previous studies indicated that animal threshold model was not suitable, because it tended to give a biased estimation of genetic parameters (Hoeschele and Tier, 1995; Luo et al.,2001; Stock et al., 2007). Therefore, threshold siredam model should be a preferred model for genetic evaluation on threshold traits in future selection breeding.

The predictive abilities of diff erent models for survival traits were all unsatisfactory. There was no or weak correlation between predicted family EBVs and actual phenotype values because survival is complicated and influenced by many factors. During aquaculture, the survival trait contains various traits,such as resistance to diff erent pathogens, tolerance to diff erent physical and chemical stresses and other lethal factors. Considering that the low predictive accuracies were obtained in survival trait, it is not an ideal breeding target. In fact, selective breeding aimed at increasing resistance to specific diseases, which is measured as survival under challenge testing with a specific pathogen, seemed to be a more targeted approach to improve the survival rate of aquaculture(Ødegård et al., 2011). Besides, as the accumulation of data and pedigree depth, selection based on EBVs instead of phenotypes is becoming increasingly urgent and eff ective to accelerate the progress of selective breeding.

5 CONCLUSION

Significant additive genetic variations for tolerance to ammonia and survival existed inE.carinicauda.Sire-dam models presented relatively high prediction accuracies for tolerance to ammonia of the prawn,which should be a preferred statistical model for genetic evaluation.

6 DATA AVAILABILITY STATEMENT

All data generated and/or analyzed during this study are included in the manuscript.