Variety Trial in Plant Breeding Lecture PDF

Variety Trial in Plant Breeding Efren E. Magulama, PhD Department of Crop Science College of Agriculture University of Southern Mindanao Importance Variety testing is an indispensable and essential step in the process of creating new improved varieties from breeding...

Variety Trial in Plant Breeding Efren E. Magulama, PhD Department of Crop Science College of Agriculture University of Southern Mindanao Importance Variety testing is an indispensable and essential step in the process of creating new improved varieties from breeding to adoption. The performance of the varieties can be compared and evaluated based on multi-trait data from multi-location variety tests in multiple years. Importance Innovation in crop variety is essential for increased grain production. In most countries, new and improved crop varieties have to undergo tests, registration, and approval before their benefits can be realized (Setimela et al., 2010; Yan, 2021). National and regional variety tests are often conducted to obtain agronomic information, such as agronomic characteristics, tolerance to disease, and quality. Why do we need field trials? In plant breeding programs, field trials are used to test genotypes for multiple traits and estimate (or predict) their genetic values. We use those measures of genetic value to: ❖Select crossing parents for population improvement; and ❖Identify candidate varieties for product development. Therefore, accurate estimates of genetic values are critical to drive genetic gain and ensure a high turnover of improved varieties. Field Trials Field trials, however, generate phenotypic values, which are a result of genetic and non-genetic effects. Efficient field trial designs and their analysis are key components of plant breeding programs to divide phenotypic values into components attributable to the genotype and to non-genetic effects. Hence, field trials lay the foundation for accurate and reliable estimates of genetic value. The phenotype as a function of genotype and environment Genetic value is determined by the entire set of expressed genes in an individual and their interactions (additive and non-additive genetic effects). Environmental effects cause the phenotypic value to deviate from the genetic value. The phenotype as a function of genotype and environment As stated by Falconer and Mackay (1995): “We may think of the genotype conferring a certain value on the individual and the environment the genotype is grown in causing a deviation of the observed phenotypic value from the underlying genetic value in one direction or the other. Hence, generally speaking, the effect of the environment is a source of error which reduced the precision of the estimated genetic value when based on the phenotypic value.” This relationship can be expressed in a simple model, where the phenotypic value (P) is described as a linear function of the genetic value (G) and the environment (E). The phenotype as a function of genotype and environment On the individual basis, the genetic value ca also be expressed as deviation from the population mean for a trait: The phenotype as a function of genotype and environment What is the “environment”? The environment is not a single factor but involves a plethora of biotic and abiotic factors that affect the phenotypic value: Physical and chemical soil attributes, including those induced by soil tillage and crop cultivation practice. Climatic factors, such as precipitation, temperature, and sunlight. Biological organisms, such as weeds, pests, and pathogens. Measurement error  The phenotype as a function of genotype and environment The total of these factors and their interactions result in an environment that is highly specific to the combination of location and season a genotype is grown in. We refer to the variation due to the effect of the location as spatial variation, and the variation due to the effect of the season as seasonal variation Figure 1. Spatial and seasonal variation of soil fertility in an experimental field at two different years. In the second year, a reduction of soil fertility can be observed at the left and bottom borders of the field. The phenotype as a function of genotype and environment Spatial and seasonal variation complicate the selection process in plant breeding programs for two reasons: The phenotypic performance of the same genotype can strongly vary when grown under different environmental conditions. This is an important phenomenon to consider when we derive genotypic values from phenotypic observations. A comparison of two or more genotypes grown in different environments may be unfair because phenotypic values might be significantly determined by differences in the environment rather than the underlying genetic values. The phenotype as a function of genotype and environment Phenotypes, however, are our primary source of information to derive genetic value. Therefore, to obtain an accurate predictor of the genetic value, we need to control and correct for spatial and seasonal variation as much as possible. We will now see how field trial designs and a proper statistical analysis enable us to achieve this. The principles of field experimentation In plant breeding trials, we routinely use a combination of experimental designs and statistical analysis to control and correct for environmental variation. In particular, Experimental designs help us to capture and control for environmental variation, and Statistical analysis helps us to correct for the factors of experimental variation captured by the experimental design. The principles of field experimentation While the design of a field trial always precedes its analysis, it is important that both measures are well coordinated. A good experimental design is ineffective without a suitable analysis. Likewise, a strong analysis will be of limited value if the basic principles of experimental design are ignored. As Fisher (1938) stated: “To consult the statistician after an experiment is finished is often merely to ask him to conduct a post-mortem examination. He can perhaps say what the experiment died off” Pillars of robust experimental designs The basic principles of a statistically valid experimental design comprise: 1.Replication 2. Randomization 3. Blocking (local control)  Principle of experimental design R.A. Fisher (1925; 1926) Principle of experimental design Replication is the repetition of an experimental treatment. In plant breeding, we mostly think of the repeated testing of selection candidates such as lines, hybrids, clones, or entire families when we refer to replication. Replication serves two purposes, and both of them are important prerequisites to obtain accurate treatment effects: Principle of experimental design It allows us to obtain an estimate of the experimental error (standard error of the mean of the treatments), i.e., the variability of a treatment due to non-systematic environmental effects. It simultaneously reduces the experimental error (standard error of the mean of the treatments). While a single phenotypic measurement, for example, can be strongly affected by random environmental conditions, the mean of multiple replicates will approach the true genetic value and therefore increase repeatability. Principle of experimental design Replication can be conducted within and across environments. Replication within an environment will increase the precision (repeatability) of the estimated genetic value only within that environment. Replication across environments will increase the precision (repeatability) of the estimated genetic value within the entire target population of environments. Principle of experimental design Randomization is the random allocation of treatments to experimental units, such as the random allocation of genotypes to plots, rows, and columns. In a randomized design, each experimental unit has an equal probability of receiving a particular treatment. Thereby, randomization ensures that treatment effects and environmental effects are independent of each other. In a field trial, randomization provides protection against unknown effects of spatial trends on genotypic performance. Principle of experimental design Therefore, it helps to avoid confounding between treatment effects and environmental effects. Randomization is also a prerequisite to obtaining a statistically valid and unbiased estimate of the experimental error. Systematic treatment allocation can result in an inflated or deflated estimate of the error and, therefore, in a considerable loss of the precision of treatment estimates. Thus, in combination with replication, randomization helps to minimize selection bias and ensures a fair comparison of treatments Principle of experimental design Blocking or local control as originally referred to by Fisher (1926) is the process of reducing the experimental error by dividing a heterogeneous experimental area into more homogeneous subunits. A reducing the experimental error by dividing a heterogeneous experimental area into more homogeneous subunits. A common example in plant breeding is the grouping of a field trial into blocks. Spatial trends in soil heterogeneity are often strong, which results in an increased experimental error. Experimental Designs Experimental designs which are commonly used in plant breeding and genetics includes 1. Augmented Design 2. Completely Randomized Design (CRD) 3. Randomized Block Design RBD) 4. Latin Square Design (LSD) 5. Split Plot Design (SPD) 6. Group Balanced Block Design (GBBD) 7. Lattice Design (LD) Randomized Block Design (RBD) The variability present in a population is of polygenic nature and these polygenic variations are three types viz. phenotypic, genotypic and environmental The statistical procedure which separate (or)splits the total variation into different component is called analysis of variance or ANOVA. ANOVA is useful in estimating the different components of variance. It provides basis to the last of significance and it is carried out only with replicated data obtained from standard statistical experimental results. ANOVA-RCBD F (calculated) is compared with F (table) value by looking at the F-table for replication (r-1) and error df values (r-1)(t-1). If the calculated F value is greater than F (table value) then it is significant. Genetic Variance Genetic variance It is the inherent variation which remains unaltered by the environment. It is variation due to genotypes. It is denoted by Vg and is calculated using the formula Coefficient of variation Coefficient of variation (1) Phenotypic co-efficient of variation: It is defined as the ratio of phenotypic standard deviation to the mean, expressed as % and can be calculated using the formula. Coefficient of variation (2) Genotypic coefficient of variation: GCV is the ratio of genotypic standard deviation to the mean expressed as% and is calculated using the formula (3) Environmental coefficient of variation: It is the ratio of environmental standard deviation to the mean expressed as % and in calculated using the formula Where, Vg, Vp and Ve are the genotypic, phenotypic and environmental variance and x is mean. Steps involved in ANOVA The different steps involved in preparation of ANOVA table can be illustrated by the following example. Question: Yield per plant (g) of an experiment with 8 chickpea genotypes evaluated in RCBD with three replications is given below in tabular form. Prepare the ANOVA table and workout components of variance, coefficient of variance and heritability (broad sense). Steps 1. Grand total=257 2. Grand mean= Grand total/no of observation=257/24=10.71 3. Correction factor (CF) = (G.T)²/n = (257)²2/24=2752.04 𝑇𝑆𝑆 = 𝛴𝑥 2 − 𝐶𝐹 4. TSS= Total sum of square of all observation-CF = 2875-2752.04 = 122.96 1 𝑇𝑅 𝑆𝑆 = ෍ Tr 2 −𝐶𝐹 5. TSS= sum of square of treatment /no of replication-CF = 8533/3-2752.04 = 98.96 𝑅 6. RSS= sum of square of replication/no treatment- CF = 22019/8- 2752.04= 0.34 7. ESS=TSS-RSS-TrSS = 122.96-98.96-0.34= 23.67 8. Vg = (MSTr- MSE)/r = (14.14- 1.69)/ 3= 4.15 9. Ve = MSE= 1.69 10. Vp = Vg + Ve = 4.15+1.69= 5.84 11. GCV= √Vg/G.M ×100=√ 4.15/10.71×100= 19.02% 12. PCV=√Vp/G.M×100=√5.84/10.71×100= 22.56% 13. ECV=√Ve/G.M×100=√1.69/10.71×100= 12.14% 14. Heritability (Broad sense) =VG/VP×100= 4.15/5.84×100=71.05% 15. Genetic Advance (GA) = K√VP×H= 2.06×√5.84×0.71= 3.54 16. G.A % of mean = GA/Grand Mean= 3.54/10.71×100= 33.03 Result Interpretation A. GCV And PCV are classified as suggested by Sivasubramonium & Madhavarmenon in 1973 as < 10% Low 10-20% Moderate > 30% High Interpretation of Coefficient of variance If, GCV> PCV, it means little influence of environment on the expression of character. Selection for improvement of such character would be rewarding. If, PCV> GCV, it indicates that apparent variation is not only due to genotype but also due to environment. Selection for improvement of such traits sometime may be misleading. If, ECV>PCV & GCV, it indicates that environment playing significant role in expression of such characters. Selection for improvement of such traits would be ineffective. Heritability Heritability: Jonnesen et. al. (1995) categorized heritability as 60%= Higher Other scale of broad sense heritability >40%=Low 40%-60%=Medium 60%-80%=High >80%=Very High Interpretation of heritability (H) If H is high: Means character less influenced by environmental effect, the selection for improvement of such character may not be useful because broad sense heritability includes both fixable (additive) and nonfixable (dominance and epistasis) genetic variance. If H is low: it reveals that character is highly influenced by environmental effect, and the genetic improvement through selection would be difficult due to masking effect of environment on the genotypic effect. Genetic Advance Genetic Advance: It is a measure of genetic gain under selection. It refers to the improvement of near genotype value of selection lines and families over the base population. Interpretation of Genetic Advance (GA) If GA is high: Means character is governed by additive gene action and selection will be rewarding for improvement of such character. If GA is low: It indicates that character is governed by non-additive gene action and Heterosis breeding may be useful Interpretation of Both Heritability and genetic advance: a) High heritability and High genetic advance means heritability mainly due to additive gene effect and hence selection may be effective. b) High heritability and low genetic advance is indicative of non- additive gene action and hence selection may not be rewarding. c) Low heritability with high genetic advance reveals that character governed by additive gene. Selection may be effective in this case. d) Low heritability and low genetic advance shows that character is highly influenced by environment and so selection would be ineffective.  Multi-environment trials and genotype-by environment (GxE) interaction Understanding GxE interactions is important for breeders to identify varieties that show high performance on average and also exhibit high performance stability across the entire trials. Multi-environment trials and genotype-by environment (GxE) interaction Multi-environment trials and genotype-by environment (GxE) interaction Patterns of genotype-by-environment (GxE) interaction Genotype-by-environment interaction occurs in various patterns which can be classified into: Non-cross over GxE interaction (scaling). Cross over GxE interaction. A combination of both. Yield Trials Yellow QPM hybrids performance trials 2011-2012 4 locations, 2 seasons 10 genotypes Malaybalay 7 yellow QPM Maramag 3 check varieties Kabacan Banga RCBD 10 entries 4 reps 15 m² plot size Data source: Magulama et al (2012) EE Magulama Methods of Stability Analysis 43 Table1. Grain yields (t/ha) of 10 maize genotypes in eight growing environments. ENVIRONMENTS¹ GEN GENOTYPES 1 2 3 4 5 6 7 8 MEAN G1 7.76 6.66 5.23 6.93 7.87 4.17 1.90 4.01 5.57 G2 7.62 5.62 3.57 6.66 7.02 4.26 1.25 2.64 4.83 G3 8.80 5.43 4.59 5.81 6.46 4.31 2.29 3.63 5.16 G4 8.96 5.73 5.71 7.82 6.85 4.36 2.37 3.35 5.64 G5 6.93 5.83 3.91 6.78 7.82 5.03 5.04 4.93 5.78 G6 10.34 7.33 6.27 6.78 7.21 4.80 3.00 4.80 6.32 G7 9.34 6.52 4.75 8.18 6.45 3.72 2.20 2.68 5.48 G8 10.19 7.39 4.43 5.99 6.25 5.27 5.46 4.37 6.17 G9 9.10 7.58 5.14 7.03 5.30 3.83 4.20 3.18 5.67 G10 8.12 5.80 4.52 7.27 4.62 4.60 1.92 2.63 4.94 ENV MEANS 8.72 6.39 4.81 6.93 6.58 4.43 2.96 3.62 5.56 1/Average of 4 reps/environment ENV 1-4= Dry Season ENV 5-8=Wet Season Data source: Magulama et al (2012) EE Magulama Data source: Methods of Stability Analysis Magulama et al (2012) 44  Table 2. Pooled analysis of variance for grain yield. F-table Sources of Variation df SS MS F-count 5% 1% P-value ENV (E) 7 1,036.48 148.07 49.60** 2.05 2.72 0.0000 REP*ENV 24 71.65 2.99 2.72** 1.57 1.88 0.0001 GENOTYPES (G) 9 67.09 7.45 2.44** 1.92 2.49 0.0000 GxE 63 192.52 3.06 2.79** 1.37 1.56 0.0000 Pooled Error 216 236.82 1.10 Total 319 1,604.56 cv = 18.85% Data source: Magulama et al (2012) **- significant **- at 1% levelat 1% level significant Data source: Magulama et al (2012) EE Magulama Methods of Stability Analysis 45  Table 4. Pooled ANOVA and AMMI analysis of 10 genotypes tested in 8 environments. Explained SS Sources of Variation df SS MS F-test P-value (%) ENV (E) 7 1,036.48 148.07 49.60** 0.0000 64.60 REP*ENV 24 71.65 2.99 2.72** 0.0001 4.47 GENOTYPES (G) 9 67.09 7.45 2.44** 0.0000 4.18 G*E 63 192.52 3.06 2.79** 0.0000 12.00 Poolled Error 216 236.82 1.10 Corrected Total 319 1,604.56 Genotypes 9 16.766 1.863 5.18 Environments 7 259.079 37.011 79.97 Genotype x Environment 63 48.138 0.764 14.86 AMMI Component1 15 18.961 1.264 2.078* 0.028 39.39 AMMI Component2 13 15.936 1.226 3.233** 0.003 33.10 AMMI Component3 11 6.398 0.582 2.034 0.071 13.29 AMMI Component4 9 2.960 0.329 1.264 0.330 6.15 GxE Residual 15 3.902 0.260 Total 79 323.983at 5% level, **- significant at 1% level * - significant CV =18.85% Data source: Magulama et al (2012) EE Magulama Methods of Stability Analysis 46  Table 5. AMMI mean grain yield (t/ha), AMMI stability values and rank orders of 10 yellow QPM genotypes tested across 8 environments GEN VARIETIES MEANS RANK PCA1 PCA2 ASV RANK 1 CML 161 x CML 165/CML 172 5.5664 6 0.4107 0.7613 0.9047 7 2 CML 161 x CML 165/CML 192 4.8303 10 0.4052 0.6355 0.7977 6 3 CML 501 x CLRY007/CML 172 5.1638 8 -0.0272 -0.0974 0.1026 1 4 CML 501 x CLRY007/SMYL 114 5.6425 5 0.6297 0.0868 0.7544 5 5 CML 161 x CML 165 5.7834 3 -1.1879 1.1150 1.8004 10 6 CML 501 x CLQ 2450 6.3156 1 0.2119 -0.3051 0.3958 2 7 CML 501 x CLQRYL 007 5.4786 7 0.7137 -0.3181 0.9070 8 8 P30B80 6.1685 2 -0.0119 -0.7102 0.7104 4 9 IPB Hy4y6 5.6687 4 -0.3333 -0.9272 1.0085 9 10 USM var 5 4.9354 9 0.3657 -0.4345 0.6150 3 ASV –AMMI Stability Value (Purchase,1997) Data source: Magulama et al (2012) EE Magulama Methods of Stability Analysis 47  Phases of variety trials National Coordinated Tests RCBD Multilocation trials RCBD Advanced trials Wet Season RCBD Dry Season Preliminary trials Luzon Lattice design Visayas Observation Mindanao nursery Augmented ❑ Regional Variety design ❑ National Variety