[R] MCMCglmm Intersex covariance set to 0

Rapkin, James jr297 at exeter.ac.uk
Tue May 12 14:02:41 CEST 2015

Dear all

I'm trying to examine the effect that the additive genetic covariance (B) between the sexes has on the predicted response to selection on life-history traits in the sexes. 

Using the method of Argrawal and Stinchcombe (2009) this is done by calculating the response to selection following the method of Lande (1980):

(Delta_Z_Male		  = 1/2 		( Gm	B	 (beta m
 Delta_Z_Female)			    B	Gf)	    beta f)	

when B was estimated from my breeding design versus when it was set to zero to reflect the case where genetic covariance between the sexes does not constrain the evolution of shared life-history traits. 

My experiment was a half-sib quantitative breeding design which looked at nutrient intake, specifically protein and carbohydrate, using artificial diets of known nutritional content, in males and females. 

My life history traits are lifespan and reproductive effort. 

I have used an MCMCglmm model to calculate the response to selection when B was predicted from my breeding design using the following as an example:

prior1.1 <- list(G = list(G1 = list(V = diag(4), n = 1.002)), R = list(V = diag(4), n = 1.002))

model1.1 <- MCMCglmm(cbind(Total_P_Eaten_Male_Daily, Total_C_Eaten_Male_Daily, Total_P_Eaten_Female_Daily, Total_C_Eaten_Female_Daily) ~ trait -1, random = ~us(trait):animal, 
                     rcov = ~us(trait):units, family = c("gaussian", "gaussian", "gaussian", "gaussian"), 
                     pedigree = Ped, data = Data, nitt=15000, thin = 50, burnin=100,
                     prior = prior1.1, verbose = T)

modelBM1 <- MCMCglmm(Z_MLS ~ Z_MP + Z_MC-1, data = Male.Data, nitt=15000, thin=50, burnin=100)
modelBF1 <- MCMCglmm(Z_FLS ~ Z_FP + Z_FC-1, data = Female.Data, nitt=15000, thin=50, burnin=100)

# Calculate delta_z
for(i in 1:298){
  MP_MP <- model1.1$VCV[i, 1]
  MP_MC <- model1.1$VCV[i, 2]
  MP_FP <- model1.1$VCV[i, 3]
  MP_FC <- model1.1$VCV[i, 4]
  MP_LS <- modelBM1$Sol[i, 1]
  MC_LS <- modelBM1$Sol[i, 2]
  FP_LS <- modelBF1$Sol[i, 1]
  FC_LS <- modelBF1$Sol[i, 2]
MP_LS_delta_z[i]<- ((MP_MP*MP_LS)+(MP_MC*MC_LS)+(MP_FP*FP_LS)+(MP_FC*FC_LS))
#Summary of deltaz including confidence

This has allowed me to calculate the confidence interval of my predicted responses and I would like to be able to do this for when my between sex genetic covariance is set to 0, but I am having trouble doing this. 

Is there a way to set my B to 0? I know how to set all my off diagonal components to 0 but this isn't exactly what I want to do.

Here is an example of my G matrix:

					G1 	G2	G3	G4
				G1	G1,1	G1,2	G1,3	G1,4
				G2	G1,2	G2,2	G2,3	G2,4
				G3	G1,3	G2,3	G3,3	G3,4
				G4	G1,4	G2,4	G3,4	G4,4

I have read the help (?MCMCglmm) and searched through the R help pages but if the answer is there I'm not able to find/understand it. 

Any help would be greatly appreciated. 

Thank you. 

James Rapkin

More information about the R-help mailing list