Better together: uniting experiments and models to predict ecosystems under global change

Beijing – Accurately predicting how terrestrial ecosystems will respond to global change is essential for assessing and projecting the land carbon sink, and with it the pace of future climate change. It has become one of the central problems in global change science, yet neither of the field's two main tools can solve this endeavor on its own. Field experiments that deliberately manipulate global change drivers (e.g., CO₂ concentration) supply direct empirical evidence, but cover only a few years and a limited number of experimental sites. Process-based models that simulate how ecosystems work can project across the globe and centuries ahead, but rest on simplified, sometimes competing assumptions that leave their forecasts uncertain. The emerging solution is to couple the two, an approach known as data-model integration, in which experiments test and refine the assumptions inside models, while models in turn guide the design of more informative experiments.

A new invited review in Journal of Plant Ecology, led by PhD students He Lü and Xueqian Zhang with corresponding author Dr. Mingkai Jiang of Zhejiang University, China, maps the progress, challenges and future directions of this integration. The authors focus on the three major drivers central to both experimental and modeling work: elevated atmospheric CO₂, climate change in the form of warming and altered rainfall, and nutrient enrichment, principally nitrogen and phosphorus but also extending to other macro- and micro-nutrients. This review gives the community not just a status report, but a roadmap pointing to where future models and experiments are most needed.

The biggest payoff of data-model integration has come from CO₂. Drawing on long-running experiments that artificially expose forests to extra CO₂ (known as FACE experiments), the integration has improved how models represent photosynthesis, water use through stomatal behavior, the allocation of carbon within plant tissues, and the coupled cycling of carbon, nitrogen and phosphorus. Together, these advances have narrowed some of the largest uncertainties in projections of the terrestrial carbon sink. On the other two fronts, climate change and nutrient enrichment, field experiments have amassed a wealth of valuable evidence, but its integration with models lags well behind. Substantial gaps still remain in how models represent plant hydraulics (the movement of water from soil to leaf), soil microbial processes, and the belowground carbon costs plants pay to acquire nutrients. Most experiments cluster in the temperate Northern Hemisphere and often impose changes far more abrupt and extreme than the real world will bring, limiting how well models calibrated on them transfer to tropical, arid or Arctic ecosystems.

To close these gaps, the authors call for a coordinated shift in practice: prioritizing the quantification of mechanism-specific data; using multi-site experimental networks to benchmark and refine model processes systematically; and using models to design sharper targeted experiments. They further advocate open, ready-to-use data platforms and hybrid methods that couple process-based models with machine learning. Beyond these broad directions, the review offers more than twenty specific recommendations, each tied to a specific gap across the three drivers. Realizing the full potential of data-model integration, the authors conclude, requires its evolution from a technical exercise into a deeply integrated scientific paradigm, indispensable to the reliable, predictive understanding that future global change demands.

Original source: Lü H, Zhang XQ, Su J, Jiang MK (2026) Data-model integration in global change manipulative experiments: progresses, challenges, and future directions. Journal of Plant Ecology 19:rtag033. https://doi.org/10.1093/jpe/rtag033

About the Journal

Founded in 2008, the Journal of Plant Ecology (JPE) (https://academic.oup.com/jpe) is led by its Co-Editors-in-Chief, Professor Yuanhe Yang (Insititute of the Botany, CAS) and Professor Bernhard Schmid (University of Zürich). JPE is a journal of the Botanical Society of China and the Institute of Botany, Chinese Academy of Sciences, and is published by Oxford University Press. According to the latest Journal Citation Reports (JCR), JPE has an impact factor of 4.5 and is ranked in the Q1 quartile in the categories of Ecology, Forestry, and Plant Sciences.