Usual disclaimer that qiime1 is no longer being updated, and qiime2 will be preferable for new options/support in the future.
You might want to go back to the rep_set.fna sequence, filter the mitochondria and chloroplast there (http://qiime.org/scripts/filter_fasta.html
and pass your filtered OTU table as -b input), and then on the resulting filtered fasta file, do alignment, alignment filtering, and tree building. This would be so any alteration to filtering (the base positions that were highly gapped, for example) would reflect an alignment without the plastic/mitochondria sequences. However, it might not change the tree-apart from getting rid of the unwanted tips.
But if you wanted to filter at the tree level itself, you have to pass a tab-delimited text file with the OTU IDs as the first column, rather than the taxonomy strings.
You could convert your (already filtered) biom OTU table to tab delimited form, with this example command: