Most long-read SV callers have parameters for the minimum read length that is considered when calling variants e.g 500bp or 1000bp. If you had a good N50 you might even consider raising this to 2000 or 5000bp, as I've seen done before.
But is there any strong disadvantage to lowering this number to 100 or 200bp, given these shorter reads may still be informative, particularly for small SVs and samples with low average read length?
I've not seen anything written about this (though I'm sure it exists), but I suspect the main disadvantage of using short reads is in creating informative breakpoint identification.
A short read might only have 1 of the 2 breakpoints for insertions, inversions, and translocations. I suspect most long read SV callers are used in conjunction with short read data which is relatively good at identifying small SVs of all classes. Hence, using short reads from nanopore/pacbio you are not providing anything new with the shorter reads, and cluttering the long read output with relatively low depth and uninformative individual breakpoints.
It may be worth running the low read length samples through and comparing with some of your better read-length samples. I suspect there will be a clear difference on the type, length, and quality of SV calls made.
