Same as Figure 1 with objects gathered by dynamical class. Underlying contours are a density map of all objects obtained by bootstrap-sampling each object 100 times assuming its albedo (color) follows a lognormal (normal) distribution set by the observing uncertainties.