A UAF mathematician builds software to untangle interbreeding species' family trees

Evolutionary family trees are usually drawn as clean branches: one species splits into two, those split again, and so on back to a common ancestor. Nature does not always cooperate. Species sometimes interbreed — a process called hybridization — mixing the genes of two separate lineages into something new, which tangles the tidy branching picture into something closer to a web. A University of Alaska Fairbanks researcher is building the statistical tools to make sense of it.

The standard math used to reconstruct evolutionary history assumes a tree — branches that split apart and never rejoin. When lineages merge instead of just dividing, a tree can't capture what happened. UAF mathematics professor Elizabeth Allman is developing software that infers networks instead of trees, reading the signal of past hybridization out of full genome sequences or the patterns across thousands of individual genes. Once built, the software will be free to use and tested against both simulated data and real-world genetic data.

The methods have uses well beyond plants and animals. The same math that untangles interbreeding species also helps trace how viruses are related to one another — including the strains of the virus that causes COVID-19 — which in turn helps researchers understand how a disease spreads and why some versions are more dangerous than others.

The work is funded by a $289,888 National Science Foundation grant, running from 2026 through 2029, under the agency's mathematical biology program. Allman, whose research focuses on whether the methods scientists use to build these trees and networks are mathematically sound, will also use the grant to run a summer school and develop teaching materials.