Functional group classification using consensus clustering.

Area of Science:

• Ecology
• Biodiversity Science
• Computational Biology

Background:

• Functional diversity is key to understanding community structure and ecosystem function.
• Current metrics for functional diversity are complex and difficult to interpret, limiting their practical application.
• Categorizing species into functional groups offers a simpler approach but faces challenges in defining robust clusters due to trait variability and correlation.

Purpose of the Study:

• To develop a novel, robust, and interpretable method for classifying species into functional groups.
• To integrate trait uncertainty and correlation into the functional group classification process.
• To provide a scalable framework for quantifying functional biodiversity accessible to conservation organizations.

Main Methods:

• A multi-step consensus clustering approach was developed.
• The method incorporates trait uncertainty through resampling and trait correlation using Gaussian Mixture Models.
• Species were classified into functional groups based on a consensus matrix derived from clustered trait data.

Main Results:

• The method was applied to a global tree dataset (47,828 species, 18 traits), identifying 42 stable functional groups.
• The identified groups reflected ecological trade-offs and phylogenetic structure.
• Traditional diversity metrics were successfully applied to functional groups, yielding intuitive measures of functional richness and redundancy.

Conclusions:

• The proposed consensus clustering framework offers a scalable and interpretable solution for quantifying functional groups.
• This approach effectively handles trait uncertainty and correlation, enhancing the reliability of functional diversity assessments.
• The method facilitates the adoption of functional diversity metrics in practical conservation and restoration initiatives.