The Mossing Link: Building a Trait-Based Database as a Foundation for Bryophyte Integration in Ecosystem Models

Abstract

Bryophytes—mosses, liverworts, and hornworts—play an important role in carbon storage, nutrient cycling, and hydrological processes across terrestrial ecosystems, yet they remain largely absent from the land surface components of Earth System Models (ESMs). Their exclusion limits the accuracy of terrestrial biosphere simulations, particularly in high-latitude and moisture-limited regions where bryophytes can dominate primary production and mediate critical biogeochemical fluxes. The Bryophyte Functional Integration for Terrestrial Ecosystem models (BryoFITE) project has been developed to address this omission. The objective of this study is to establish a systematic, reproducible workflow for synthesizing bryophyte trait data relevant to terrestrial ecosystem models. This workflow integrates automation with expert curation and relies on tools already widely used in ecological research. We identified relevant literature by constructing a comprehensive matrix of subject-parameter search terms. We then used these search terms to execute a total of 720 targeted searches across Google Scholar and Scopus, retrieving over 84,000 records. We applied a combination of fuzzy string matching and supervised machine learning to deduplicate and classify the resulting literature, yielding a refined group of ∼35,000 unique entries. A Random Forest model, trained on a manually flagged subset of records, excluded over 50% of records as unlikely to contain relevant data based on manual and model-based classification. The remaining records were ranked by likelihood of containing usable data, enabling us to prioritize data entry toward the literature with the highest probable value for this research. We then developed a modular, cloud-based database using Google Sheets with custom JavaScript automation for collaborative, flexible data entry. The BryoFITE database captures ecophysiological parameters, including photosynthetic capacity, desiccation tolerance, and nutrient-use traits, critical for modeling bryophyte function. Our preliminary results demonstrate the utility of this workflow for synthesizing heterogenous trait data and highlight its potential to accelerate the development of bryophyte-specific plant functional types for integration into land models, to advance our capacity to simulate plant–climate interactions in ESMs, and to use probability-based workflows for integrating diverse data for model parameterization, evaluation, and development.