Connect with us

Science

World's loneliest tree species can't reproduce without a mate. So AI is looking for one hidden in the forests of South Africa.

Published

on

/ 8728 Views

The world's loneliest tree may soon find a mate, thanks to artificial intelligence (AI).

Only a single, male specimen of the Wood's cycad (Encephalartos woodii) has ever been discovered in the wild. In 1895, botanist John Medley Wood found the solitary plant in what is now the Ngoye Forest Reserve in KwaZulu-Natal, South Africa.

Cycads, a primitive type of seed plant, first emerged around 300 million years ago, before dinosaurs roamed Earth. In cycads, male and female reproductive structures — called cones — are produced by separate plants.

However, E. woodii also reproduces via offshoots, meaning that the parent plant sends out shoots that turn into other adult plants. In the 20th century, botanists removed some of the offshoots and transplanted some of the original trunks from the lonely male in South Africa. These have given rise to about 500 separate plants that live in botanical gardens around the world.

Related: 100-foot 'walking tree' in New Zealand looks like an Ent from Lord of the Rings — and is the lone survivor of a lost forest

However, sexual reproduction is necessary for the long term viability of the species. To help the male offshoots find mates, scientists have launched a series of drone flights over the cycad's remote and inaccessible native forests.

The scientists then use AI algorithms to sift through the visual imagery collected by the drone. "Our project's approach to using AI focuses on the visual identification of cycads, which resemble palm trees when viewed from above," said Laura Cinti, co-founder of C-LAB, an art and Science collective, and a research fellow at the University of Southampton in England. "Initially, we adopted detection models that are routinely used in the palm oil industry for counting palm trees, but to optimize it for our specific viewpoint and unique shape of cycads, we trained our own image recognition algorithms," Cinti told Live Science.

Trending