Thermal Forcing Versus Chilling? Misspecification of Temperature Controls in Spring Phenology Models

Oct 28, 2024·
Xiaojie Gao (高孝杰)
Xiaojie Gao (高孝杰)
,
Andrew D. Richardson
,
Mark A. Friedl
,
Minkyu Moon
,
Josh M. Gray
· 2 min read
Abstract
We show that widely used modeling approaches that are calibrated using field-based observations misspecify the role of chilling under current climate conditions as a result of statistical artifacts inherent to the way that chilling is parameterized. Our results highlight the limitations of existing modeling approaches and observational data in quantifying how chilling affects the timing of spring leaf emergence.
Type
Publication
Global Ecology and Biogeography

This paper was not in the plan.

We started from using the state-of-the-art process-based models to investigate the chilling effect on spring phenology. The original idea was that if we compare models with and without the chilling factor, we would be able to understand if chilling is an important factor in controlling spring phenology and how these chilling parameters vary among time and space. However, when analyzing phenology and temperature datasets from multiple sources and spatiotemporal scales, we found it was challenging to robustly quantify chilling by these models. Sometimes, a chilling model would outperform other non-chilling models in fitting the data and in cross-validation, but if we remove just a few data points or add a small random noise to the data, the non-chilling models would become more signficant. This result let us think that maybe chilling is a subtle process and hard to identify. But, when we digged more into the problem, we found limitations of the model structures and the observational datasets. In other words, even if chilling is an important process in the controlled experiments, calibrating these process-based models with field-collected observational data cannot robustly identify the chilling effect.

This paper does not suggest that we should abandon process-based models, nor do we intend any disrespect to previous efforts. Instead, we aim to foster discussions and collaborations among modelers, remote sensing scientists, ecologists, plant biologists, and physiologists to better understand the role of chilling in spring leaf development by integrating controlled experiments with field observations. Specifically, address when plants start responding to temperatures, what range of temperatures influences spring phenology, and how and why these controls vary by species and/or climate conditions. This knowledge is key to understanding and modeling the chilling effect and predicting future phenological shifts driven by climate change, which could have significant ecological consequences.

As one of our reviewers said: “The field is mired in a bit of a mess and the only way out is likely through.” Discussions and collaborations are critical to solving the problem.