Text mining of over 89,000 scientific papers reveals new evolutionary insights
New research led by ecologists at the University of Toronto shows that mutually beneficial partnerships between ants and plants that arise from the rapid diversification of both organisms do not in turn promote further diversification in ants as they do in plants.
In such co-operative relationships — known as mutualisms — plants provide ants with food or shelter, while ants protect plants against herbivores or disperse their seeds. Previous research has shown that plants with ant partners diversify faster, showing a bigger difference between extinction rates and the rise of new plant species than do other plants not involved in similar relationships.
“We wondered if plants similarly affect ant evolution and found that is not the case,” says Megan Frederickson, associate professor in the Department of Ecology & Evolutionary Biology (EEB) in the Faculty of Arts & Science at the University of Toronto, and senior author of a study published today in PLOS Computational Biology.
The researchers hypothesized that ants would first evolve mutualism with plants and then further diversify as plants do. But a computer-generated model based on data extracted from over 89,000 previous research studies suggests ants that are already rapidly diversifying are more likely to slow the process once they evolve a plant partnership.
“To our surprise, the intimate and often beneficial relationships that ants have with plants apparently did not help to generate the over 14,000 ant species on Earth today,” says Katrina Kaur, lead author of the study and a recent graduate of EEB’s Master of Science program. “Mutualism may put the brakes on the rise of new species or increase the threat of extinction because an ant’s fate becomes linked to its plant partner’s.”
Kaur, currently a PhD candidate at the University of British Columbia, and her colleagues collaborated with specialists at U of T’s SciNet high-performance supercomputer facility in order to address their query.
The researchers knew the data they needed was buried in thousands of scientific papers each discussing just one or a few ant species, so they wrote a computer program to “read” and extract data from the over 89,000 paper abstracts. They successfully assembled a large database of ant ecological interactions, mapped the data onto an ant evolutionary tree, then modelled how partnering with plants has affected ant diversification.
“To be able to comb through so much literature so quickly and with such accuracy was astounding,” said Frederickson. “The model could easily be adapted for use by researchers in other areas of biology as well as other disciplines altogether.”
The researchers plan to use their “text-mining” computer program to assemble an even larger database from thousands of additional papers in order to understand why ant-plant partnerships have different effects on ant versus plant evolution. A similar approach could also reveal insights about other species in mutualistic relationships, such as seed-dispersing birds or human-gut microbes.
Support for the research was provided by the Natural Sciences and Engineering Research Council of Canada.