# Adjusted R-squared for tree-based models

How can I use an evaluation metric like adjusted R-squared to evaluate tree-based models? It's not clear to me, since adjusted R-squared accounts for the number of predictors included in a given model, just how large k should be. Does this metric apply to tree-based models? If so, do I use, for the number of predictors, simply the number of columns in my input matrix? Or do I have to account for the number of splits/depth of the trees? Is there a better metric for evaluating the performance of continuous-regression tree-based models?