14,000 ant species mapped in decade-long diversity project
Ants are small creatures that have a big impact on our world, but in something you might never have thought you needed to know about, very few have legal protection.
Ants perform several crucial ecological functions, including waste disposal, soil aeration, seed and nutrient dispersal, and both as predator and prey in complex food chains. In fact, ants and other similar invertebrates could be considered the true backbone of various ecospheres.
They must therefore be sought out and, in many cases, protected.
To improve our understanding of the global diversity of ants, researchers from the Graduate University of the Okinawa Institute of Science and Technology (OIST) combined machine learning methods with information from online repositories, museums and approximately 10,000 scientific publications for produce a global map of the diversity of ants up to about 20 km2.
A second map highlighting “species rarity” was constructed to display populations of highly localized ant species. They are particularly sensitive to environmental changes and are therefore of particular conservation concern.
Some regions, such as Okinawa in southern Japan, have been found to harbor many ant species in a single localized area – around 1000 times smaller than species in North America and Europe.
The researchers found that very few of these localized ant species were on land with any legal protection, such as conservation parks or reserves.
This new information, which took a decade to compile, could be key to conserving ant biodiversity.
Professor Evan Economo, who heads the Biodiversity and Biocomplexity Unit at OIST, says this project helps to “add ants and other terrestrial invertebrates in general to the discussion on biodiversity conservation. We need to know the locations of invertebrate centers of high diversity in order to know areas that can be the focus of future research and environmental protection.
More than 14,000 (of 15,000 currently known) ant species were included in the study, which scoured various sources to describe sampling locations and involved input from international researchers to identify and correct errors.
In many cases, the records were not specific enough to allow mapping, so computer estimates were made from the available data, which then underwent a range estimation process, during which researchers modeled by building shapes around data points or by modeling statistically.
Some regions of the world have been sampled much more than others, which affects estimates of ant diversity and distribution. To overcome this sampling bias, the researchers used machine learning, teaching an algorithm to predict how ant diversity would change if all regions of the world were sampled equally. The machine learning process resulted in a sort of “treasure map,” Economo says, as it identified areas where many unknown and unsampled species must have existed. We can be guided “to where we should explore next and look for new species with restricted ranges”.