Saturday , April 17 2021

Using artificial intelligence, you can predict which plants will disappear



The International Nature Conservation Association (IUCN, its acronym in English) work continuously by nominating Red list which analyzes species and categorizes them, from "minor concern" to "extinct", through intermediate categories.

These classifications are used to define conservation actions, therefore they are "tremendous useful" but "to include a species, they must be individually assessed, which requires a predefined protocol, the means available and the presence of specialists who carry out the assessment, making it a slow process , "he said. Telam Anahi Espíndola, Co-author of the study in Argentina.

"The basic method we used was" random forest, "which is known for its ability to classify and predict data," he said, explaining that "in this case, we try to predict the probability that a species is or is not being compromised by using data on its range of prevalence, for the desired climatic conditions and some morphological characteristics ".

Espindol, a professor of entomology at the University of Maryland in the United States, explained that "this method allows us to use all species that have already been assessed by the IUCN to train and create our" random forest "classification using species characteristics as predictive variables."

"As soon as a sufficiently precise classification model has been obtained, we can use the same model for the species we know for the modeling parameters used (range, preference climatic conditions and morphology), but for which we do not know the risk of extinction."

In this respect, the co-author of the article is published in a specialized journal PNAS He stated that he would use and classify this data "to calculate the probability that these species that have not yet been evaluated on the IUCN Red List are at risk."

These systems are "extremely useful" in the sense that it is "relatively precise and can be analyzed without access to important computing resources," it also has an "advantage" that it is based only on "public data" (open access), that is, everyone can perform these analyzes and use their own results.

"In addition, this method can be adapted to any geographic or taxonomic role, as it can also be used at national, regional or local level, and allows identification of species that IUCN should prioritize," he said. Espíndola and described this instrument as "helpful and complementary to these assessments".

The specialist pointed it out Of the 150,000 analyzed species, "about 10% (15,000) are highly likely to belong to conservation categories that are not" minor concerns ".

"From a global perspective, we identify regions that are more likely to endanger species, such as some northern Andean regions of South America, or Brazilian Atlantic forests." These regions are characterized by high endemism and the presence of many rare species, "he concluded.


Source link