As biologists gather bigger and bigger datasets involving thousands of molecular characters from entire genomes, they are increasingly able to reconstruct complicated evolutionary histories. More characters means more power, but when that’s not good enough we have to ask a question: is a tree really the best way to represent the relationships between these organisms?

A tree structure suggests that once two branches branch apart, they can never touch again. The species or clades they represent will have totally independent evolutionary histories from then on. But we know that species can in fact mix their histories again by hybridizing. When members of two different species hybridize, they combine the independent evolutionary histories of the parents into the new hybrid offspring. If those offspring can reproduce, it means that the two parent branches of the tree have joined to make a new branch. Now, there is a closed loop in the tree, called a reticulation, and our diagram looks more like a network than a tree. Reticulations can throw off our normal methods of reconstructing trees.

Biologists have known for a long time that hybridizations occur in nature, and in some groups they are actually quite common. But traditional methods of tree-building can’t account for this. New methods for detecting reticulations and building evolutionary networks are allowing biologists to incorporate hybridization events into our understanding of evolutionary history – and biologists are finding reticulations in evolutionary trees more and more often.

For instance, species of swordtail fish (Xiphophorus) in the rivers of Mexico and Guatemala have hybridized many times. Scientists were able to detect these ancient hybridization events using Bayesian methods that construct thousands of probable trees and look for sets of trees that have different shapes. The most likely overall tree shape is shown in black, but all of the species pairs linked by green arrows in the diagram formed clades in a significant number of the probable trees, meaning that they are probably linked to one another by past hybridization events.