Olfactory receptor genes have more variation than most gene families in the human genome. The only family with greater diversity is the major histocompatibility complex (MHC). Both families also exhibit high heterozygosity. Due to its association with disease, the MHC is well-studied. The explanation for the maintenance of MHC diversity is pathogen-driven selection–either through heterozygote advantage or frequency-dependent selection (see here for a review); a small number of papers (here’s two: 1 and 2) have also argued for divergent allele advantage. A diversity of pathogens will result in a diversity of MHC genes over time; as a species develops resistance to a disease, an evolutionary respones occurs in the disease-causing agent. The common analogy is the evolutionary arms race, also called the Red Queen Hypothesis.
If we apply that same model to olfaction in light of a few recent findings. there might be something worth pursuing. We know we can smell millions of odors. Such a diversity of odorants in the environment that vary from region to region may result in incredible diversity in the human olfactory receptor subgenome–especially if we look at it from the perspective of divergent allele advantage.