The most likely case here is that your read R has been pushed higher in the taxonomy, because it is the only or one of very few reads with significant matches to taxon T.
There are two LCA parameters guiding this behaviour.
Min Support Percent sets the threshold of reads in percent of all reads, that have to be theoretically assigned to a taxon, before this taxon is shown in the taxonomy and the reads are actually assigned to it.
Min Support does the same, but for an absolute number of reads.
In this example, this means that one read being assigned to a taxon is enough for a taxon to be shown and the read to be assigned, but you also need at least 0.1 % of all reads to be assigned to that taxon. Otherwise the reads will be assigned to the first taxon higher up in the taxonomy which has enough reads assigned to get over the threshold.
If your dataset has 10 000 reads, with a Min Support Percent value of 0.1 your would need at least 10 reads assigned to a taxon because Min Support Percent overrides Min Support.
If you want to set the threshold by an absolute number of reads, set Min Support Percent to 0.0 and if you do not want to use these thresholds at all, set Min Support Percent to 0.0 and Min Support to 1.
If you use Min Support Percent, you should set it in a reasonable manner in relation to the number of reads in your dataset (very large datasets should use lower values than 0.1).