XPath to identify a parent based on two related children - xpath

Question

Using the following Example, I am trying to create an XPath that will identify the id of Apples where their branchID matches the ID of a branch, but the treeID doesn't match the branch's treeID.
for instance: //growth[#type="apple"][branchID=//branch/#id]/#id - results would be
Granny
Empire
Gala
And //growth[#type="apple"][treeID!=//branch/treeID]/#id - results would be
Granny
Empire
Gala
But I want a query that will only return:
Granny
<xml>
<growth type="apple" id="Granny">
<branchID>ABCD</branchID>
<treeID>456</treeID>
</growth>
<growth type="apple" id="Empire">
<branchID>ABCD</branchID>
<treeID>123</treeID>
</growth>
<growth type="apple" id="Gala">
<branchID>EFGH</branchID>
<treeID>456</treeID>
</growth>
<growth type="flower" id="white">
<branchID>EFGH</branchID>
<treeID>123</treeID>
</growth>
<growth type="flower" id="yellow">
<branchID>ABCD</branchID>
<treeID>456</treeID>
</growth>
<branch id="ABCD">
<treeID>123</treeID>
</branch>
<branch id="EFGH">
<treeID>456</treeID>
</branch>
<tree id="123" />
<tree id="456" />
<tree id="789" />
</xml>

Answer 1

So the required item(s) (where the branchID/treeID pair does not exist in a branch element) will have different index values for the branchID in the list of all //branch/#id values and the treeID in the list of all //branch/treeID values...
//growth[#type="apple"][index-of(//branch/#id,branchID)!=index-of(//branch/treeID,treeID)]/#id
Granny
Nice question to give the little grey cells some exercise on a Friday lunchtime :o)