Unified theory resolves phenological paradoxes in biocollection data by modeling phenophase duration during Bayesian...

Area of Science:

• Ecology
• Botany
• Data Science

Background:

• Phenology research increasingly uses digitized biocollection data.
• Existing methods lack a general theory connecting specimen collection dates to underlying phenological processes.

Purpose of the Study:

• To derive a unified mathematical theory linking biocollection data to phenological events.
• To develop a method for accurate estimation of phenological timing, duration, and environmental sensitivity.

Main Methods:

• Developed a unified mathematical theory connecting biocollection data to phenophase onset, duration, cessation, and peak timing.
• Embedded the theory within a Bayesian Gaussian process framework.
• Utilized 5363 herbarium records from 13 spring ephemeral species.

Main Results:

• The theory demonstrates that both phenophase duration and onset timing influence specimen collection dates.
• Failure to model these latent events leads to inaccurate phenological timing estimates.
• The Bayesian Gaussian process framework successfully disentangles the effects of onset and duration, providing explicit models of phenophase dynamics as functions of environmental factors.
• Variation in phenophase duration was found to confound estimates of phenological sensitivity in all examined species.

Conclusions:

• Accurate phenological inferences require explicit modeling of phenophase duration and onset timing.
• The developed theory and methodology improve the estimation of phenological sensitivity to environmental change.
• The R package phenoCollectR facilitates theory-driven phenological research using biocollection data.