I do not think the vegetation suppression tool considers a value of 0 (or whatever masked value the water areas may have) to be a masked value in this case. It treats those pixels like any other pixels and end up with a value after vegetation suppression. Did you create a mask for the water? If so, you could apply this mask to the vegetation suppressed image. If not, do the water values have a single value? If so, you could create an ROI and use this to create a mask that could be applied to the data to convert those values back to 0.