fitting nested structure model variogram using gstat

Dear All,
I have fitted basic model such as spherical, Gaussian, exponential, and
linear to empirical variogram using gstat library in R. The models were
fitted in omnidirectional and different directions to detect if the data
exhibit anisotropy.

mpvar<-variogram(yield~1,canmod.sp,cutoff=300,width = 7 , cressie=TRUE)
plot(empvar, xlab = "Lag distance (m)", ylab = "Semivariance", main =
"Omnidirectional empirical Variogram for Yield")


### directional variogram
empvar<-variogram(yield ~ 1,canmod.sp,cutoff=300,width =
7,alpha=c(0,45,90,135),cressie=TRUE)
plot(empvar,xlab = "Lag distance (m)", ylab = "Semivariance", main =
"Directional empirical Variogram

without defining the lag tolerance in my script, hope this would not have
negative impact on my result.

secondly, I intend to fit nested structure model by looking at the linear

combination of those basic model (spherical, gaussian, exponential, and
linear). Please how do I modify my R script in gstat to model such a
complex nested structure?

Thirdly, if it is a must to define my lag tolerance, will it be 1.5 times
whatever I use as my lag distance as specified by width=7?