A new (old) approach to genotype-based phylogenomic inference within species, with an example from the saguaro cactus (Carnegiea gigantea)

AO_SCPLOWBSTRACTC_SCPLOWGenome sequence data are routinely being used to infer phylogenetic history within and between closely related diploid species, but few tree inference methods are specifically tailored to diploid genotype data. Here we re-examine the method of "polymorphism parsimony" (Inger 1967; Farris 1978; Felsenstein 1979), originally introduced to study morphological characters and chromosome inversion polymorphisms, to evaluate its utility for unphased diploid genotype data in large scale phylogenomic data sets. We show that it is equivalent to inferring species trees by minimizing deep coalescences--assuming an infinite sites model. Two potential advantages of this approach are scalability and estimation of a rooted tree. As with some other single nucleotide polymorphism (SNP) based methods, it requires thinning of data sets to statistically independent sites, and we describe a genotype-based test for phylogenetic independence. To evaluate this approach in genome scale data, we construct intraspecific phylogenies for 10 populations of the saguaro cactus using 200 Gbp of resequencing data, and then use these methods to test whether the population with highest genetic diversity corresponds to the root of the genotype trees. Results were highly congruent with the (unrooted) trees obtained using SVDquartets, a scalable alternative method of phylogenomic inference.