I 've run a 4-gene 12,000-taxon super tree. Given I increased the bootstrap repliactes (200 to 1000), the support for the most internal nodes improved a lot (say 80 - 90% BS). However, the support for external nodes are remain low (some of them are 0-10% BS). Regarding to the knowledge that not all the taxa are fully sampled under the 4 genes, so there are great portion of missing data. Well, beside this, any other way to compensate for this shortage?