I have a polygon feature class where each polygon has one of two attributes, "Admin" or "Active".  These polygons are related spatially in a daisy-chain manner. That is, any Admin polygon overlapping an Active polygon will be grouped with that Active polygon into a larger "Active" site (ultimately dissolved into individual polygons), and any Admin polygon that then overlaps that newly created larger Active area is then also included in that Active site, until there are no more Admin polygonsoverlapping the large Active area. 

 

I am using Neighbor Finder to select the polygon that are overlapping, with the Active polygons as the Candidates and the Admin polygons as the Base.  The trick is that once I get the first set of Admin polygons selected, I need to then use those to select more Admin polygons that overlap the Admin polygons selected in the first step. I have done this manually in ArcMap and it should take 5 steps to get all the polygons that are related in this daisy-chain effect.  I thought I would be able to take the Matched and Unmatched Candidate features from the first step (bc those are all going to be Active), and use them as the candidate input for a second Neighbor Finder, and the unmatched base (Admin polygons) as the base input for the second Neighbor Finder (and repeat this through a total of 5 Neighbor Finders).

 

However, the polygons selected (including original Active polygons) in the previous steps are dropped in the subsequent steps.  Is there a trick to doing this without losing the polygons selected in each previous step?