I am an oceanographer, not an evolutionary biologist, but I am generally familiar with the very basic tools of tree construction. I'm trying to generate rooted trees as input for Unifrac, based on OTUs from v7 to v9 region (~440 bp) of 18S rDNA amplified from Antarctic seawater. There are 35 OTUs (excluding those <0.5% total abundance), mostly diatoms and haptophytes (other photosynthetic algae). No matter what sequences I try for an outgroup, invariably each candidate is clustered somewhere near an ingroup member. I even tried green algae (Chlamydomonas- not at all related to the red lineage) and two opisthokont sequences (S. cerevisiae and H. sapiens). Seems ridiculous that even an absurdly distant outgroup such as opisthokonts would cluster in the in-group. I'm on day 4 here and at my wits end. Am I missing something obvious?

Thanks!

Adam