This document summarizes a study on exploiting the "Doddington Zoo" effect in biometric fusion. The Doddington Zoo refers to how some biometric subjects ("lambs" and "wolves") have higher false acceptance rates than others ("sheep"). The study shows that selectively fusing biometric data based on which subjects are lambs/wolves versus sheep can increase throughput with minimal cost. User-specific fusion of iris and fingerprint data from two datasets achieved equal error rates of 2% and 1.1%, lower than non-selective multibiometric fusion.