testing collinearity in MLM Given this linear mixed model 
dat <- data.frame(blocks = rep(1:15, each = 2), 
                  soil = runif(30, 0, 10), 
                  nitro = runif(30, 0, 10), 
                  temp = rnorm(30, 10, 3)) 

mod <- lmer(soil ~ nitro * temp + (1 | blocks), data = dat)

How can I check if there is multi-collinearity between the predictors? 
 A: You have created the variables by sampling from a normal distribution where each variable is sampled independently of the others. Therefore there should be no multicollinearity between the variables, however there will be high correlation between the variables and the interaction term:
> set.seed(22)
> dat <- data.frame(blocks = rep(1:15, each = 2), 
                   soil = runif(30, 0, 10), 
                   nitro = runif(30, 0, 10), 
                   temp = rnorm(30, 10, 3)) 

> mm <- model.matrix(soil ~ nitro * temp, dat)

> cor(mm[, -1])

                nitro       temp nitro:temp
nitro       1.0000000 -0.1860283  0.8950713
temp       -0.1860283  1.0000000  0.1661519
nitro:temp  0.8950713  0.1661519  1.0000000

As per the answer by Dimitris, you can assess the impact of this through the variance inflation factors.
A: To assess the level of multi-colinearity between you predictors you can calculate the variance inflation factors. The function vif() from the car package should work for linear mixed models fitted by lmer().
