Predicting Quaternary hominin climatic niches using faunal assemblages with random forest algorithms - Archive ouverte HAL Accéder directement au contenu
Poster De Conférence Année : 2023

Predicting Quaternary hominin climatic niches using faunal assemblages with random forest algorithms

Résumé

The Quaternary period (2.58 MYA - today) has been characterized by a series of abrupt climatic fluctuations and a global trend of cooling and aridification. In Africa, these changes had substantial impact on landscapes and led to shifts in faunal communities, which resulted in the gradual establishment of present-day faunas. This period also coincides with major events in the hominin evolution, including the appearance and diversification of the genera Homo and Paranthropus , the development of new lithic technologies, and the adoption of new food and resources acquisition strategies. Therefore, understanding the intricate relationship between the human evolutionary history and the environmental and climatic shifts of the Quaternary period is a fundamental focus within the field of human origins research. For palaeoenvironmental reconstructions based on terrestrial vertebrate faunas, two clades of mammals have historically received great attention: rodents and bovids. Both provide useful ecological information, yet they are rarely used simultaneously for palaeoenvironmental reconstruction as they have specific modes of accumulation and preservation of their remains. In this contribution, we developed a novel approach for retrodicting past climate conditions of several African hominin-bearing fossil sites from the Quaternary based on the composition of the rodent and bovid palaeocommunities, using random forest (RF) algorithms for robust predictive modeling.
Fichier non déposé

Dates et versions

hal-04346386 , version 1 (15-12-2023)

Identifiants

  • HAL Id : hal-04346386 , version 1

Citer

Pierre Linchamps, Emmanuelle Stoetzel, Raphaël Hanon, Raphael Cornette, Pierre Latouche. Predicting Quaternary hominin climatic niches using faunal assemblages with random forest algorithms. 1st Symposium in Artificial Intelligence and Applied Mathematics for History and Archaeology (IAMAHA), Nov 2023, Nice, France. ⟨hal-04346386⟩
62 Consultations
0 Téléchargements

Partager

Gmail Facebook X LinkedIn More