Implementation of a new phylonumerics approach to construct a mass tree representing over 6000 H1N1 human influenza strains has enabled ancestral and compensatory descendant mutations to be identified in N1 neuraminidase that promote antiviral resistance and restore viral fitness. Adjacent to the H275Y resistance mutation site, mutations S299A and S247N, respectively, lead the evolution of oseltamivir-resistant strains and restore viral fitness to those strains thereafter. Importantly the mass tree phylonumerics approach can identify such mutations globally, without any positional bias, so that functionally linked or compensatory mutations remote in the sequence or structure of the protein can be identified and interrogated. This is achieved using mass map datasets commonly employed for protein identification in proteomics applications, thus avoiding the need for either gene or protein sequences that are central to other phylogenetic methods.