The Jensen-Shannon divergence or distance matrix stores the $n*(n-1)/2$ pairwise distances/similarities between observations in an $n x p$ matrix where n correspond to the independent observational units and p represent the covariates measured on each individual.