If func.var in the call to sobol was NULL, this returns boxplots of sensitivity indices and total sensitivity indices. If there were functional variables, they are labeled with letters alphabetically. Thus, if I fit a model with 4 categorical/continuous inputs and 2 functional inputs, the functional inputs are labeled a and b. If func.var was not NULL, then posterior mean functional sensitivity indices are plotted, along with the functional partitioned variance. Variables and interactions that are excluded did not explain any variance.