What is it about?

Birds have adapted a stunning variety of specialized traits and lifestyles to thrive across environments – from eagles soaring on mountain winds to puffins diving into seas. But can measurements of wings, legs and beaks predict how a bird makes its living? This research explores whether bird lifestyles linked to habitats and movement have a measurable morphological basis. Using a dataset of nearly 11,000 species spanning almost the entire diversity of modern birds, machine learning models were developed to classify species as “aerial”, “terrestrial”, “aquatic” and more based solely on traits like wing shape, tail length, and foot size. The best model performed with 84% accuracy, showing a strong capability to infer lifestyle from physical form. Aerial species with adaptations suiting airborne living were especially distinguishable. The research confronts existing categorical schemes used to pigeon-hole birds and suggests better criteria reflecting morphological patterns. By revealing a tight coupling of lifestyle and anatomy, these computable techniques can automate the identification of poorly known species’ habits and niches when specimens arise. They also allow projecting potential vulnerabilities of groups sharing traits as environments shift. Understanding the morphological constraints and tradeoffs birds navigate provides insight into the evolutionary playbook enabling their worldwide success.

Featured Image

Why is it important?

Linking morphology to lifestyle and performance allows predicting how organisms exploit environments – key to ecology and conservation. For diverse, data-sparse clades like birds, tools revealing these connections can uncover hidden ecological patterns and roles. By demonstrating lifestyle habits have measurable anatomical roots, this work enables the estimation of niche occupancy from readily observable traits. The technique provides means to rapidly assess unstudied species' likely behaviors and habitat usage when specimens emerge, and model vulnerabilities of suites of birds sharing morphological features as conditions change. Beyond prediction, better quantifying adaption facilitates understanding evolutionary pathways and constraints generating Earth’s spectacular biodiversity. Connecting form, function, and fitness guides strategies for biodiversity maintenance. Determining the morphological variability supporting avian lifestyles across geographies could indicate the potential for groups to persist if environments alter. Overall this research opens multiple avenues for harnessing the information-rich tapestry of relationships between bird form and function to address scientific and conservation challenges.

Perspectives

As an ecologist, I've always been captivated by birds and the spectacular ways evolution has shaped their morphologies for aerial, aquatic, and terrestrial living. But subjective classification schemes used to categorize species have always bothered me – where are the quantifiable links to anatomical adaptions? By leveraging new large datasets and machine learning, my team and I uncovered clear signals in body plans correlating with habits and habitats. But this is only a start – better phenotyping birds will unravel more precise structures within these lifestyle groups. We envision these techniques applied to elucidate poorly known tropical bird species' roles when only museum samples exist. I'm excited by the possibilities these approaches open up – automating the identification of cryptic species, calculating the resilience of suites of birds to environmental change, and projecting modifications if warming proceeds. More broadly, quantifying the relationship between form and function will provide fundamental insights into the drivers and constraints guiding the evolution of birds' dazzling diversity.

Luis Javier Madrigal Roca

Read the Original

This page is a summary of: Assessing the predictive value of morphological traits on primary lifestyle of birds through the extreme gradient boosting algorithm, PLoS ONE, January 2024, PLOS,
DOI: 10.1371/journal.pone.0295182.
You can read the full text:

Read
Open access logo

Contributors

The following have contributed to this page