Imagine that a cell contains the formula =A1+B2+C3, and another shows =Z9+B2+C3. A discrepancy analysis will quickly shows that the cells are similar their second and third terms, but dissimilar in the first.
But if A1 in turn refers to three other cells, and those refer to three more in turn, the discrepancy analysis will be dominated by the precedents of A1 (the cells identified by a black border below).
These cells take up much space on the report, and cause the cells that are matched, as shown by color and hyperlinks, to be displaced relative to each other, making the result hard to read. This reduced readability comes at no advantage, in the sense that it exposes no cells that are common to both calculations.
OAK can be instructed not include subtrees such as these that turn out to have no contents that cross-match with cells on the other side of the report.
If this option is enabled, the display of the subtree is suppressed.
The subtree is still examined by OAK, and would be shown if any matching cells were detected. All that is suppressed is the inclusion of this part of the precedent hierarchy in the final report.
Tip: If a discrepancy analysis seems to contain much less information than you expect, check that this option is turned off.