Look at a cactus and the desert adaptation seems obvious. Thick and swollen with water, that stem is what carries the plant through months with no rain. On top of that, it costs the plant carbon and some photosynthesis.

On a new reconstruction of the family tree, something far less visible came first. The red pigment that colors beets and cactus fruit evolved earlier, and it may be what made the costly stem worth having. That puts the trait everyone points to second in line.

Tom Carruthers ran the work as a postdoctoral researcher in Ecology and Evolutionary Biology at the University of Michigan, and he’s now at University College Dublin. The team ran no experiments. They built an evolutionary tree of more than 4,400 species, scored each one for pigment, water storage and drought, and worked out the order the traits arrived in.

Almost every flowering plant on Earth builds its reds and purples out of the same family of compounds, and one group doesn’t. Across roughly 12,000 species that include beets, bougainvillea, amaranth, and cacti – those compounds are gone, and a different set of red and yellow pigments does the job. Botanists call them betalains. Nothing else in the plant kingdom makes them. Plants under stress make more: salt, drought and strong ultraviolet light all push production up, and lab work has shown the pigment protects cells under salt stress.

The same group is also full of desert specialists, and researchers had long noticed the overlap. Nobody had worked out which came first. Dating a plant lineage takes fossils. Gregory Stull, of the Smithsonian’s National Museum of Natural History, assembled the fossil record for the group, and the authors picked 14 species out of it. Those 14 include pollen grains from Cretaceous sediments, a set of seeds, three fruits and one flower cluster. Each fossil was scored for 50 physical traits alongside the living species in the genetic dataset. That let them sit in the tree as tips rather than as fixed calibration points. The authors accept more uncertainty that way, and argue that showing it is the honest version. The group itself dates to the Early Cretaceous, roughly 100 to 110 million years ago.

Betalains turned up four separate times in the tree, all of them during the Cretaceous. Those four lineages had nothing in common but this group’s ancestry. Of the 3,941 species the team could classify by climate, 1,308 counted as drought tolerant, meaning more than three quarters of collected specimens came from arid areas. Among those drought-tolerant species, 75% carry betalains, against 55% of the full dataset. Restrict the definition to hot arid places and the figure climbs to 85 percent.