e also acknowledge the Sociedade Botânica do Brasil (SBB) for financial support through the Scientia Amabilis 2024 grant. DATA AVAILABILITY STATEMENT The data supporting the findings of this study are available in the Supporting Information of this article (Appendix S6 ). Additionally, all data and code are openly accessible at https://github.com/ykilsztajn/fresh_dry_myrtaceae . REFERENCES Bewley , J. D. , K. J. Bradford , H. W. Hilhorst , and H. Nonogaki . 2012 . Development and maturation . In Seeds: Physiology of development, germination and dormancy , 3rd ed. , 27 – 83 . Springer , New York, New York, USA . Blonder , B. , V. Buzzard , I. Simova , L. Sloat , B. Boyle , R. Lipson , B. Agui
Open resource ↗ykilsztajn/fresh_dry_myrtaceae · lines:268-513Paper record
Herbarium‐based measurements are reliable predictors of fresh plant traits in Neotropical Myrtaceae
Applications in plant sciences · 15 Sept 2026 · 10.1002/aps3.70082
Abstract
Premise Herbarium specimens are increasingly used to extract morphological traits for ecological and evolutionary studies, yet the effects of tissue desiccation on trait measurements remain poorly understood. Here, we tested two hypotheses: (H1) whether higher tissue water content leads to greater measurement changes after herborization and (H2) whether fresh trait values can be reliably predicted from measurements of herborized specimens. Methods We evaluated the reliability of herbarium-based measurements by comparing fresh and dried traits of leaves, flowers, fleshy fruits, and seeds across 262 individuals representing 133 Neotropical Myrtaceae species. Phylogenetic least squares models and machine learning regressions were used to test H1 and H2. Results Leaves and flowers generally shrank after herborization, fruits size metrics tended to increase, and seeds were largely unaffected. Water content was significantly associated with the magnitude of herborization effects in flowers and some leaf and seed traits. Fresh trait values were accurately predicted from measurements of herborized specimens. Prediction errors were lowest for leaf traits, followed by fruits, flowers, and seeds. Discussion These results partially support H1 and support H2, indicating that herborized specimens can be reliably used for trait analyses when organ-specific responses are considered, providing a practical framework to account for potential desiccation bias in functional trait research.
Code and data availability
The paper's trait measurements (Appendix S6 species mean data) and analysis code are openly available in the authors' GitHub repository, explicitly stated in the Data Availability Statement.
Other versions of this study