In the CLR transform, each entry is divided by the geometric mean of its sample and then the logarithm is taken.
clr( abundances, minocc = 0, pseudocount = 1, scale.factor = 1, omit.zeros = FALSE )
a matrix with taxa as rows and samples as columns
prevalence filter; ignored if 0 or below
zero replacement value; should be above 0
scaling factor; abundances will be multiplied with this factor
if TRUE, ignore pseudocount and compute clr transform only on non-zero elements
When omit.zeros is set to true, the default behaviour of clr in the compositions package when applied column-wise is reproduced.