Hi Colin,
This is the way I do it:
I suggest that you base minimum corridor width for your target species on local expert knowledge and/or the literature. Once you have that distance, say, 2 km then do the following:
Run LM once with the default and then if the corridors are not wide enough, run again with larger truncation value. If they are too wide, look at the most narrow one and then query the values that are at the appropriate width (e.g. 1 km perpendicular from the least cost path). Note that value, say 150,000. Then run LM again using that cwd threshold.
Yes, to date, I always use the same cwd truncation threshold for linkage pafthways tool and pinchointmapper tool.
Good luck!
-John