Boxplot of annual Rh (g C m<sup>−2</sup>)

<p><strong>Figure 2.</strong> Boxplot of annual Rh (g C m<sup>−2</sup>). The box and whisker bars indicate the full range (thin line) and 25th–75th percentiles (upper and bottom edges of the bar) and median (line in the bar) of Rh for the specific year and location of the measurements. The alphabets on the <em>x</em>-axis denote observations (SRDB-V2, blue), the ESMs with the carbon cycle only (CanESM2, GFDL-ESM2M, HadGEM2-ES, INMCM4, MIROC-ESM, and MPI-ESM-LR, red) and with the carbon–nitrogen cycle (CCSM4 and NorESM1-M, purple). The number of records in the panels is 56, 219, 40, 48, and 37, respectively.</p> <p><strong>Abstract</strong></p> <p>Soil microbial respiration (Rh) is a large but uncertain component of the terrestrial carbon cycle. Carbon–climate feedbacks associated with changes to Rh are likely, but Rh parameterization in Earth System Models (ESMs) has not been rigorously evaluated largely due to a lack of appropriate measurements. Here we assess, for the first time, Rh estimates from eight ESMs and their environmental drivers across several biomes against a comprehensive soil respiration database (SRDB-V2). Climatic, vegetation, and edaphic factors exert strong controls on annual Rh in ESMs, but these simple controls are not as apparent in the observations. This raises questions regarding the robustness of ESM projections of Rh in response to future climate change. Since there are many more soil respiration (Rs) observations than Rh data, two 'reality checks' for ESMs are also created using the Rs data. Guidance is also provided on the Rh improvement in ESMs.</p>