Multi-scale fundamental phenology mechanism in response to global climate change

Abstract

Plant phenology, which involves the timing of life cycle events in plants, is a sensitive bio-indicator of global climate change, and is integral to numerous crucial ecological processes and functions of ecosystems. Understanding the factors driving phenological changes is essential for predicting ecological responses to climate change. This thesis focuses on three fundamental questions: 1) Since factors affecting spring phenology arise from conventional models mainly considering environmental variables such as temperature and photoperiod, is it possible that other variables, including solar radiation, or vapor pressure deficit, also have a substantial impact as well? 2) What is the key environmental variable(s) responsible for existing model residuals over the Northern Hemisphere (>30°N), and does the factor(s) remain consistent throughout the region, or do they differ across different PFTs? 3) Is it possible to enhance the accuracy of spring phenology modeling? Could the integration of underappreciated factors or the employment of the Farquhar-Medlyn photosynthesis model lead to a more precise simulation of the spring phenology process? To address question 1, we utilized PhenoCam observations, satellite derived MCD12Q2 phenology product as well as environmental variables from Daymet/GLDAS dataset to evaluate prognostic models in the Northern and Eastern United States ecosystems. We evaluated ecosystem-scale phenology models to determine if they underestimated environmental factors. The results indicated that: 1) All models showed good capability in characterizing spring phenology, with the OPT model performing best (RMSE = 8.04 ± 5.05 days), followed by SEQ (RMSE = 10.57 ± 7.77 days) and GDD (RMSE = 10.84 ± 8.42 days). 2) All models had high residuals, which correlated closely with solar radiation (r = 0.45-0.75). 3) Revised models that included solar radiation performed significantly better, reducing RMSE by 22.08%. These findings highlight the critical but underappreciated role of solar radiation in constraining spring phenology in temperate ecosystems and suggest improvements for phenology modeling and related ecological processes. To address question 2, we used leaf unfolding data (LUD) data derived from the AVHRR and environmental variables from GLDAS, TerraClimate, and CRU_TS v4.03. This thesis analyzed phenology models on a hemispheric scale, focusing on the Northern Hemisphere. We examined the associations of model residuals with solar radiation (SR), precipitation (P), soil moisture (SM), and vapor pressure deficit (VPD). Our findings showed that SR and water stress significantly influenced spring phenology variability, with water stress being dominant in grasslands and SR in other ecosystems. Based on these results, we revised the OPT model to include SR (OPT-S) and VPD (OPT-VPD). The results indicated that all three models performed well, but OPT-S had the highest accuracy (median RMSE = 4.05 days), followed by OPT-VPD (median RMSE = 6.26 days), and the original OPT model (median RMSE = 7.08 days). The improved performance of the revised models likely stems from their better approximation of early-season plant photosynthesis potential by considering SR and water conditions, compared to photoperiod alone. Overall, this study enhances our understanding of key environmental drivers of spring phenology in the Northern Hemisphere, contributing to more accurate ecological forecasts in the context of global environmental change. To address question 3, we utilized phenology records PEP725 and environmental variables from ERA5 Land Hourly dataset and the Mauna Loa Observatory CO2 dataset to test whether incorporating the Farquhar-Medlyn photosynthesis model lead to a more precise simulation of the spring phenology process across the European region. The results demonstrate that incorporating the Farquhar Medlyn model into the existing process-based OPT model significantly improves accuracy, followed by CFT and GDD. Furthermore, results also showed that different environmental gradients impact the leaf unfolding date differently across species. However, the variability associated with carbon gain, water stress, and light intensity plays a crucial role in explaining LUD across various sites. Interestingly, earlier models have underestimated the influence of CO2 and water stress during the photosynthesis period. These findings strongly support the plant principle that plants optimize strategies to maximize resource gain while minimizing water loss. Overall, this thesis enhances our comprehension of plant spring phenology. The findings advance our understanding of plant phenology and its driving factors across both spatial and temporal dimensions, emphasizing the fundamental biophysical mechanisms that underlie the diverse spring phenology strategies observed worldwide. The insights gained from these investigations can guide future plant phenology research and aid in the development of more accurate and comprehensive models for predicting plant ecosystem responses to climate change.