Scholarlink

<p>Establishing the relationships between organismal phenotypes, their ecology, and the environment through time is a central goal in evolutionary biology. A robust understanding of how functional traits vary across ecological niches in extant lineages can help contextualise fossil morphologies and palaeoecological interpretations. This is evident in the well-established association between foraging ecology and trait variation in waterfowl. Here we apply supervised machine learning (random forest) and linear discriminate analysis trained on a large, extant geometric morphometric dataset to predict the foraging ecologies of nine extinct waterfowl species. We find that both approaches reliably predict palaeoecology for extinct species with verified foraging ecologies and provide sensible predictions for lesser-known species. Interestingly, we found that the Hawaiian moa-nalo and the New Zealand <i>Cnemiornis calcitrans</i> likely occupied ecological niches no longer represented in modern waterfowl. These species are even more morphologically and ecologically distinct than previous descriptions suggested. Our study further underscores the utility of predictive modeling for supporting palaeontological research.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Geometric morphometrics enables accurate predictions of palaeoecology and reveals unique adaptations to an expanded niche space in extinct waterfowl

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月4日

查看全文
  • Ray Mariano Chatterji,
  • Shante Williams,
  • Janet C. Buckner

摘要

Establishing the relationships between organismal phenotypes, their ecology, and the environment through time is a central goal in evolutionary biology. A robust understanding of how functional traits vary across ecological niches in extant lineages can help contextualise fossil morphologies and palaeoecological interpretations. This is evident in the well-established association between foraging ecology and trait variation in waterfowl. Here we apply supervised machine learning (random forest) and linear discriminate analysis trained on a large, extant geometric morphometric dataset to predict the foraging ecologies of nine extinct waterfowl species. We find that both approaches reliably predict palaeoecology for extinct species with verified foraging ecologies and provide sensible predictions for lesser-known species. Interestingly, we found that the Hawaiian moa-nalo and the New Zealand Cnemiornis calcitrans likely occupied ecological niches no longer represented in modern waterfowl. These species are even more morphologically and ecologically distinct than previous descriptions suggested. Our study further underscores the utility of predictive modeling for supporting palaeontological research.