Lecture 4: Basic Designs for Estimation of Genetic Parameters

Heritability Narrow vs. board sense Narrow sense: h 2 = V A /V P Board sense: H 2 = V G /V P Slope of midparent-offspring regression (sexual reproduction) Slope of a parent - cloned offspring regression (asexual reproduction) When one refers to heritability, the default is narrow-sense, h 2 h 2 is the measure of (easily) usable genetic variation under sexual reproduction

Why h 2 instead of h? Blame Sewall Wright, who used h to denote the correlation between phenotype and breeding value. Hence, h 2 is the total fraction of phenotypic variance due to breeding values Heritabilities are functions of populations Heritability measures the standing genetic variation of a population, A zero heritability DOES NOT imply that the trait is not genetically determined Heritability values only make sense in the content of the population for which it was measured.

Heritabilities are functions of the distribution of environmental values (i.e., the universe of E values) Decreasing V P increases h 2. Heritability values measured in one environment (or distribution of environments) may not be valid under another Measures of heritability for lab-reared individuals may be very different from heritability in nature

Heritability and the prediction of breeding values If P denotes an individual's phenotype, then best linear predictor of their breeding value A is The residual variance is also a function of h 2 : The larger the heritability, the tighter the distribution of true breeding values around the value h 2 (P -  P ) predicted by an individual’s phenotype.

Heritability and population divergence Heritability is a completely unreliable predictor of long-term response Measuring heritability values in two populations that show a difference in their means provides no information on whether the underlying difference is genetic

Sample heritabilities Peoplehshs Height0.65 Serum IG0.45 Pigs Back-fat0.70 Weight gain0.30 Litter size0.05 Fruit Flies Abdominal Bristles0.50 Body size0.40 Ovary size0.3 Egg production0.20 Traits more closely associated with fitness tend to have lower heritabilities

Estimation: One-way ANOVA Simple (balanced) full-sib design: N full-sib families, each with n offspring One-way ANOVA model: z ij =  + f i + w ij Trait value in sib j from family i Common MeanEffect for family i = deviation of mean of i from the common mean Deviation of sib j from the family mean  2 f = between-family variance = variance among family means  2 w = within-family variance  2 P = Total phenotypic variance =  2 f +  2 w

Covariance between members of the same group equals the variance among (between) groups The variance among family effects equals the covariance between full sibs The within-family variance  2 w =  2 P -  2 f, -

One-way Anova: N families with n sibs, T = Nn FactorDegrees of freedom, df Sums of Squares (SS) Mean sum of squares (MS) E[ MS ] Among-familyN-1SS f =SS f /(N-1)  2 w + n  2 f Within-familyT-NSS W =SS w /(T-N) 2w2w N X i =1 n X j =1 (z ij °z i ) 2 n N i =1 (z i °z) 2 X

Estimating the variance components: 2Var(f) is an upper bound for the additive variance Since

Assigning standard errors ( = square root of Var) Nice fact: Under normality, the (large-sample) variance for a mean-square is given by æ 2 (MS x ) ' 2 ( MS x ) 2 df x +2 ( ( ) )

Estimating heritability Hence, h 2 < 2 t FS An approximate large-sample standard error for h 2 is given by - - -

Worked example FactorDfSSMSEMS Among-familes9SS f =  2 w + 5  2 f Within-families40SS w = 2w2w 10 full-sib families, each with 5 offspring are measured V A < 10 h 2 < 2 (5/25) = 0.4 Var(w) = MS w = 20 Var(z) = Var(f) + Var(w) = 25

Full sib-half sib design: Nested ANOVA Full-sibs Half-sibs

Estimation: Nested ANOVA Balanced full-sib / half-sib design: N males (sires) are crossed to M dams each of which has n offspring Nested ANOVA model: z ijk =  + s i + d ij + w ijk Value of the kth offspring from the kth dam for sire i Overall mean Effect of sire i = deviation of mean of i’s family from overall mean Effect of dam j of sire i = deviation of mean of dam j from sire and overall mean Within-family deviation of kth offspring from the mean of the ij-th family  2 s = between-sire variance = variance in sire family means  2 d = variance among dams within sires = variance of dam means for the same sire  2 w = within-family variance  2 T =  2 s +  2 d +  2 w

Nested Anova: N sires crossed to M dams, each with n sibs, T = NMn FactorDfSSMSEMS SiresN-1SS s = SS s /(N-1) Dams(Sires)N(M-1)SS s = SS d /(N[M-1]) Sibs(Dams)T-NMSS w = SS w /(T-NM) Mn N X i =1 M i X j =1 (z i °z) 2 n N X i =1 M X j =1 (z ij °z i ) 2 N X i =1 M X j =1 n X k =1 (z ijk °z ij ) 2

Estimation of sire, dam, and family variances: Translating these into the desired variance components Var(Total) = Var(between FS families) + Var(Within FS) Var(Sires) = Cov(Paternal half-sibs)  2 w =  2 z - Cov(FS)

Summarizing, Expressing these in terms of the genetic and environmental variances,

4t PHS = h 2 h 2 < 2t FS Intraclass correlations and estimating heritability Note that 4t PHS = 2t FS implies no dominance or shared family environmental effects

Worked Example: N=10 sires, M = 3 dams, n = 10 sibs/dam FactorDfSSMSEMS Sires94, Dams(Sires)203, Within Dams2705,40020

Parent-offspring regression Shared environmental values To avoid this term, typically regressions are male-offspring, as female-offspring more likely to share environmental values Single parent - offspring regression The expected slope of this regression is: Residual error variance (spread around expected values) 

The expected slope of this regression is h 2 Hence, even when heritability is one, there is considerable spread of the true offspring values about their midparent values, with Var = V A /2, the segregation variance   Midparent - offspring regression  Residual error variance (spread around expected values) Key: Var(MP) = (1/2)Var(P)

Standard errors Squared regression slope Total number of offspring Midparent-offspring regression, N sets of parents, each with n offspring Midparent-offspring variance half that of single parent-offspring variance Single parent-offspring regression, N parents, each with n offspring Sib correlation

Estimating Heritability in Natural Populations Often, sibs are reared in a laboratory environment, making parent-offspring regressions and sib ANOVA problematic for estimating heritability Why is this a lower bound? Covariance between breeding value in nature and BV in lab A lower bound can be placed of heritability using parents from nature and their lab-reared offspring,   where is the additive genetic covariance between environments and hence  2 < 1

Defining H 2 for Plant Populations Plant breeders often do not measure individual plants (especially with pure lines), but instead measure a plot or a block of individuals. This can result in inconsistent measures of H 2 even for otherwise identical populations. Genotype iEnvironment j Interaction between Genotype I and environment j Effect of plot k for Genotype I in environment j deviations of individual Plants within this plot If we set our unit of measurement as the average over all plots, the phenotypic variance becomes Number of EnvironmentsNumber of plots/environmentNumber of individuals/plot Hence, V P, and hence H 2, depends on our choice of e, r, and n