550 likes | 588 Views
HCS 825 Presentation. By Robin Grom May 29, 2002. Trait Stability. An interesting concept…. Plant Breeding 101. There are many things to consider when aiming for a new phenotype…. A successful potato…. …breeding program:. Know how each trait responds to environmental conditions
E N D
HCS 825 Presentation By Robin Grom May 29, 2002
Trait Stability An interesting concept…
Plant Breeding 101 • There are many things to consider when aiming for a new phenotype….
…breeding program: • Know how each trait responds to environmental conditions • Be adaptable to these varying conditions • Understand false phenotypes
Outline • The environment • Importance to plant breeders • Genotype-by-environment interaction • Stability • Yield stability • Methods to calculate stability • Example analysis
Outline • The environment • Importance to plant breeders • Genotype-by-environment interaction • Stability • Yield stability • Methods to calculate stability • Example analysis
The environment • There are two classes of environmental factors that can affect a plant’s phenotype • Predictable • Unpredictable
Predictable factors… • Primarily under human control • Relatively unchanging • Row spacing • Planting date • Soil type
Unpredictable factors… • Not subject to human control • Randomly variable • Available moisture • Temperature extremes/averages • Light intensity • Relative humidity
Environmental factors can influence: • A plant’s rate of growth • The quality and/or quantity of its reproductive tissues • The plant’s overall commercial value
Outline • The environment • Importance to plant breeders • Genotype-by-environment interaction • Stability • Yield stability • Methods to calculate stability • Example analysis
The Environment to Plant Breeders F1 Progeny Parent 1 Parent 2
The Environment to Plant Breeders Environment A Environment B Genotype under consideration Environment C
Variations in phenotype due to the environment can make it difficult for breeders to analyze and choose the best genotypes to work with in the future…..
Outline • The environment • Importance to plant breeders • Genotype-by-environment interaction • Stability • Yield stability • Methods to calculate stability • Example analysis
G x E • “changes in the relative performance of genotypes across different environments” (Fehr, 1991) • “true” vs. “ghost” phenotypes
G x E • Important to breeders working with stability improvement • Allows breeders to anticipate and plan for environments that will show the most information about a phenotype
Outline • The environment • Importance to plant breeders • Genotype-by-environment interaction • Stability • Yield stability • Methods to calculate stability • Example analysis
Stability… Fehr, Principles of Plant Breeding: “the reliability of cultivar performance across locations and years…influenced by the genotype of individual plants and the genetic relationship among plants of the cultivar”
Stability…. Becker and Leon (1988): Stability is classified as either static or dynamic
Becker and Leon, cont. • Static stability: the performance of a genotype or cultivar remains unchanged regardless of environmental conditions • Dynamic stability: a measure of how the genotypic performance changes in a predictable manner across a range of environments
Outline • The environment • Importance to plant breeders • Genotype-by-environment interaction • Stability • Yield stability • Methods to calculate stability • Example analysis
Yield Stability • The ability of a genotype to sustain a marketable yield over a variety of environments • Subset of phenotypic stability
Static Yield Stability The quality and quantity of yield does not change over different environments
Dynamic Yield Stability • A measure of how the quality and/or quantity of yield changes over differing environments • Example with one genotype: • High yields in good environments • High yields in poor environments • Low yields in good environments
Yield Stability Key Points: • Cultivars with an average but consistent yield over a variety of conditions are generally more valuable than those with a higher overall yield that fluctuates dramatically with differing environments • “selections for high performance in an optimum environment generally have below-average stability, whereas selections under less-desirable conditions can result in the development of lines with above-average stability” (Berry, 1988)
Outline • The environment • Importance to plant breeders • Genotype-by-environment interaction • Stability • Yield stability • Methods to calculate stability • Example analysis
Methods for Calculating Stability A short history….
Calculating Stability… Finlay and Wilkinson, 1963 • Used a linear regression model • Averaged each individual’s mean yield • Formed a regression of these against the mean of the entire population for that environment • Resulting regression coefficient used as a measure of the phenotypic stability of the varieties in question
Calculating Stability… • Eberhart and Russell • Calculated “an environmental index for each of the environments used for a series of trials as the mean of all varieties at one environment minus the grand mean of all environments of the trials. They then pointed out that both the regression coefficient and the deviation from regression of a variety on the environmental indices should be considered as parameters for measuring the stability of a variety” (Tai, 1971)
Calculating Stability… • Tai, 1971 • Tai partitions G x E interactions into two parts: • Linear response to environmental effects, measured by (α) • Deviation from the linear response, measured by (λ) • Perfectly stable variety: α,λ = -1,1 • Average stability: α,λ = 0,1
Outline • The environment • Importance to plant breeders • Genotype-by-environment interaction • Stability • Yield stability • Methods to calculate stability • Example analysis
General Information • Used historic potato yield trial data from plants grown in Ohio to determine the yield stability of each line • Project spanned approximately 20 years • Certain lines were grown in more than 1 study per year • Varieties changed as lines were added or discarded
Stability Analysis • Line Regression Model • Used method similar to Finlay and Wilkinson • SAS code was developed for the regression analysis of individual variety means vs. the trial mean • (The SAS program is available on most campus computers, or through the SAS Institute website for purchase (www.sas.com)
Data Alteration • Many lines over various years were provided for study • To make the regression model reliable, only varieties that were present for over 3 years were used • Late 1980’s: 5 varieties over 5 years • Late 1990’s: 9 varieties over 5 years • Two varieties were present in each group • Varieties with yield data for more than one study per year were averaged within that year
Dr. Baker’s Basic Model for Joint Regression Analysis • The basic statistical model for joint regression analysis is: X[ijk] = u + e[i] +r[ik] + g[j] + b[j]e[i] + d[ij] + w[ijk] • where u is the grand mean of all experiments, • e[i] is the effect of environment (experiment) i, • r[ik] is the effect of replication k in experiment i, • g[j] is the effect of genotype j, • b[j] is a linear regression coefficient for genotype j, • d[ij] is the deviation from regression for genotype j evaluated in environment i, and • w[ijk] is the random effect associated with plot ijk. (http://duke.usask.ca/~rbaker/gxe4.html)
While I tried…. using Dr. Baker’s method of joint regression analysis, the lack of replicates in the Ohio trial prevented me from using this approach. Instead, SAS code was developed for the regression analysis of individual variety means vs. the trial mean.
SAS code data one; infile'a:latenines2.csv' delimiter = ','firstobs = 4; input year gen yld; procsortdata = one; by gen; procsummarynway; class year; var yld; outputout = two mean = myld; procprint; procsortdata = one; by year; data three; merge one two; by year; procprint; procsortdata = three; by gen;
SAS code, cont. procreg; model yld = myld; by gen; data three; set three; residual = yld-myld; procreg; model residual = myld; by gen; run; (Code provided by Dr. Francis)
Source Root MSE DF 70.18981 Sum of Squares R-Square Mean Square F-value 0.4242 Pr > F DependentMean 250.400000 Adj. R-Square 0.2323 Model 1 10889 10889 2.21 0.2338 Coeff. Var. 28.03107 Error 3 14780 4926.608 Corrected Total 4 25669 Sample Output (Superior) Analysis of Variance
Variable DF Parameter Estimate Standard Error t-Value Pr > |t| Intercept 1 73.71043 122.92122 0.60 0.5910 Myld 1 0.68567 0.46120 1.49 0.2338 Output, cont.
Stability Parameters for Comparing Varieties • Slope (or regression) = 1; Var[dev] = 0 means variety is stable (from Eberhart and Russell) • Remember, cultivars with an average but consistent yield over a variety of conditions are generally more valuable than those with a higher overall yield that fluctuates dramatically with differing environments
Results of Individual 1980’s Analyses Using Eberhart and Russell’s parameters, Norchip appears to be the best variety in this group with m=.981 and Var[dev]=165.1