I'm not a phylogenetics expert when it comes to the algorithms used and such, but I believe what Marjanovic meant was that a morphological dataset like this one that tries to maximize taxonomic coverage will have a lot of highly incomplete taxa.
David M. is obviously the best-qualified person to clarify what he meant, but it didn't occur to me to read his comment on the difference between morphology and molecules as a statement about incompleteness. Sparse supermatrices with wildly varying levels of coverage are common in densely sampled molecular analyses as well; it's not a problem unique to fossil phylogenies.
Dromomeron gigas is just a partial femur, Ignotosaurus an incomplete ilium, Nyasasaurus a few vertebrae and partial humerus, etc.. It's common to have only 10% coded or less. So in these cases, it's very easy for the bootstrapping procedure to delete the few characters placing such taxa in their clades, or multiply those few characters that would otherwise be local autapomorphies and disagree with close relatives. And since the bootstrap percentage is telling you if the entire compliment of included taxa formed a clade in the bootstrap replicates, it's not useful if you have a lot of fragmentary taxa jumping around a basically stable backbone phylogeny. Ornithoscelida might have been recovered in every tree, but Nyasasaurus was in it 44% of the time for example. And because of this, Ornithoscelida gets a 66% bootstrap score, despite the fact every bootstrapped tree had theropods plus ornithischians forming a clade outside sauropodomorphs. Am I completely off base?
I don't know. I'd be in favor of testing this hypothesis empirically by re-running the bootstrap analysis on the original matrix and checking each pseudoreplicate for the presence of an Ornithoscelida-like clade (plus or minus Nyasasaurus and other highly incomplete taxa).