Comparison of habitat models for scarcely detected species

详细信息    查看全文

• 作者：Auriane Virgili^a ; ^{auriane.virgili@univ-lr.fr} ; Mé ; lanie Racine^b ; Matthieu Authier^b ; Pascal Monestiez^a ; ^c ; Vincent Ridoux^a ; ^b
• 关键词：Habitat modelling ; Scarce species ; GAM ; GLM ; MaxEnt ; Poisson ; Negative binomial ; Tweedie ; Zero-inflated Poisson
• 刊名：Ecological Modelling
• 出版年：2017
• 出版时间：24 February 2017
• 年：2017
• 卷：346
• 期：Complete
• 页码：88-98
• 全文大小：1528 K
• 卷排序：346

文摘

Importance of selecting appropriate models for rare species habitat predictions. Presence-absence models made better predictions than the presence-only model. GAMs with a Negative Binomial and Tweedie distributions better predict the habitats. Zero-inflated Poisson distributions exhibited less convincing results. 92% zeros in the dataset does not necessarily mean zero-inflation.