Global change manipulative experiments are developing rapidly in China

More global change manipulative experiments (GCMEs) in underrepresented regions such as semi-arid ecosystems, forests in the tropics and subtropics, and arctic tundra are needed to generate a robust projection of future terrestrial carbon-climate feedback.

Like Comment
Read the paper

    Over the past half century, scientists around the world have used global change manipulative experiments (GCMEs) to 1) better understand the mechanisms that control how terrestrial ecosystems respond to changing climate and atmospheric composition (e.g., climate warming, changing precipitation regimes, rising atmospheric CO2 concentrations, and elevated atmospheric inputs of nitrogen) and 2) develop and parameterize Ecosystem Models and Earth System Models simulating ecosystem responses to future scenarios.

    I studied terrestrial ecosystem responses to global change drivers using GCMEs as a Ph.D. candidate at University of Oklahoma and a post-doctoral associate at Oka Ridge National Lab and joined the Institute of Botany, Chinese Academy of Sciences in 2005. At that time there was only one reported GCME at ecosystem scale conducted in Tibetan Plateau, China (Klein et al. 2004). INTERFACE (An Integrated Network for Terrestrial Ecosystem Research on Feedbacks to the Atmosphere and ClimatE) created and released a global distribution map of GCMEs (Fig. 1). The map showed a clear message that compared with the United States and Europe, China has few GCMEs.

Figure 1. The map of GCMEs created in 2005. The global map was downloaded from the website ( of INTERFACE.

    I designed and built more than twenty GCMEs in a semi-arid temperate steppe in Duolun County, Inner Mongolia since 2005. Over the next five years, my lab published many peer-reviewed papers using the results of our GCMEs (e.g., Wan et al. 2009, Qiu 2014, Song et al. 2019, and Fig. 2). GCMEs are also rapidly increasing in China since 2005. I began to realize that a comprehensive review is urgently needed to summarize what has been done using GCMEs during the last few decades. Therefore, I asked my lab’s master and Ph.D. students to start collecting papers on GCMEs published over the past few decades. By April 2014, 683 papers were collected, but this didn’t yet cover all published literature, so I decided to let Jian Song, a master student in my lab at Henan University, to take the project over. With the help of the lab members, he extracted the basic experimental information (e.g., the locations, altitudes, ecosystem types, and climate conditions of experimental sites, global change drivers manipulated in experiments, the magnitudes etc.) from all the papers collected. Over the past five years (2010-2014), we could put the information together to show the state-of-the-art and future challenges of GCMEs. 

Figure 2. GCMEs in Duolun County, Inner Mongolia constructed by Dr. Shiqiang Wan’s lab. Left panel: A four-factor (CO2 enrichment, night-time warming, increased precipitation, and nitrogen addition) experiment constructed in May 2011 (Song et al. 2019). Right panel: A precipitation gradients plus nitrogen addition experiment constructed in May 2019. Photo credit: Shiqiang Wan, Jian Song, and Haidao Wang.

    In May 2014, I (the local host and co-sponsor), Jeffrey S. Dukes (Purdue University), and Aimée Classen (University of Vermont) organized an INTERFACE Workshop on ‘Using results from global change experiments to inform land model development and calibration.’ in Beijing (Dukes et al. 2014). In the workshop, I gave a talk on the summary of GCMEs using the information gathered by Jian from 975 peer-reviewed papers. After the workshop, I invited 28 experimentalists and modelers from around the world to participate in this project. Two more years passed (2015-2016), we cannot remember how many times we discussed the extracted data and revised the draft of this article. At that time, Jian finally collected 2230 papers and put the experimental information together. These results confirmed that GCMEs are developing rapidly in China since 2005 (Fig. 3). In addition to focusing on the GCMEs itself, one more year was spent to analyze the general patterns of the responses of terrestrial ecosystem carbon-cycle variables to single and multiple global change drivers and the climate sensitivity of carbon-cycle variables. With the great help of our co-authors, we finalized this article by the end of 2017. It was a very long story and meaningful journey for me and Jian. Despite the unprecedented difficulties, I believe Jian has not only learned a lot of knowledge and expertise skills during this journey, but more importantly, Jian has showed us the importance of patience and persistence. 

Figure 3. Global distribution of manipulative experiments. The global distribution of single factor and multifactor manipulative experiments (1973-2016, a). The number of papers published from GCMEs in China (1973-2018, b). W: warming, P: changing precipitation regimes, eCO2: elevated CO2, eN: enriched atmospheric N deposition, treatments in multifactor experiments are shown with multiple letters.

    Our paper, published in Nature Ecology & Evolution (, used meta-analysis to synthesize results on terrestrial carbon-cycling response to global change from 1,119 experiments as described in 2230 peer-reviewed papers published between 1973 and 2016, provided the most comprehensive dataset for global change impact research, and revealed the general patterns of carbon-cycling responses to global change drivers. Despite general trends, our paper underscores that the sensitivity of the terrestrial ecosystem carbon-cycling responses to multiple global change drivers is dependent on the background site climate and ecosystem conditions, and that powerful Earth System Models are best evaluated in the context of site-specific data. Towards this end, more GCMEs is needed in ‘hotspot’ regions to understand terrestrial carbon-climate feedbacks on a global scale. These underrepresented regions include semi-arid ecosystems, forests in the tropics and subtropics, and Arctic tundra. This article had a large team of 59 authors from 39 affiliations of 9 countries and presented a panoramic display for what we have done using GCMEs, what we have learned from GCMEs, and where GCMEs should go in the future. We believe that this paper will be one of the milestones on the development road of GCMEs to promote the rapid progress of GCMEs.  

By Shiqiang Wan (Professor of Ecology, Hebei University).



Dukes, J. S., Classen, A. T., Wan, S. & Langley, J. A. Using results from global change experiments to inform land model development and calibration. New Phytologist 204, 744-746 (2014).

Klein, J. A., Harte, J. & Zhao, X. 2004. Experimental warming causes large and rapid species loss, dampened by simulated grazing, on the Tibetan Plateau. Ecol. Lett. 7, 1170-1179.

Qiu, J. Land models put to climate test. Nature 510, 16-17.

Song, J. et al. Elevated CO2 does not stimulate carbon sink in a semi-arid grassland. Ecol. Lett. 22, 458-468 (2019).

Wan, S., Xia, J., Liu, W. & Niu, S. Photosynthetic overcompensation under nocturnal warming enhances grassland carbon sequestration. Ecology 90, 2700-2710 (2009).

Jian Song

Professor, Hebei University