Global analysis of carbon concentration data in roots

Due to the complexity and cost of quantifying carbon in roots (CR), more accessible alternatives are being sought to obtain this data, since the carbon cycle is an essential factor in assessing climate change. The aim of this study was to assess the feasibility of using Global Roots Traits (GRooT) data for environmental modeling in different global ecoregions. GRooT is a database with 144,222 records of 38 functional traits compiled from various studies around the world. Filtering and removal of inconsistent data was carried out using the R software and its “dplyr”, “sf” and “rnaturalearth” libraries. In addition, the displaced geographical coordinates were corrected using Q-GIS software. Initially, 6,124 CR samples were found and, after filtering the data, a total of 4,454 samples with the necessary parameters for modeling were obtained. Of these samples, 2,727 are in Europe, 936 in America, 759 in Asia, 24 in Africa, 8 in Oceania and 0 in Antarctica. The samples are distributed in 28 countries, most of which are located in Germany (1,612), Poland (906) and China (720). The countries with the fewest observations (3) were Estonia, Ethiopia, Spain, Thailand and Yemen. In the ecoregions, the number of samples exceeded 200 in the Temperate Forests, Wet Broadleaved Tropical and Subtropical Forests (HTF) and Temperate Coniferous Forests, while the Boreal/Taiga Forests (BTF), Mountain Grasslands and Shrublands (MAP), Flooded Grasslands and Savannas, Dry Broadleaved HTF, and Lakes were below 20 observations. The rest of the ecoregions had a number of samples within this range. The average CR concentration was above 480 mg.g^-1. in the FBT, Dry Broadleaf FTS and PAM ecoregions. In contrast, the Temperate Grasslands, Savannas and Scrublands, Temperate Forests, and Mediterranean Forests, Woodlands and Scrublands had concentrations below 422 mg.g^-1. The other ecoregions had average concentrations between this range. There was an uneven distribution in the number of samples across the Earth's surface, with some ecoregions having a high number of samples, while others contained relatively smaller amounts. The number of samples from Germany, for example, represented 48% of the data from the Temperate Forests, which showed a concentration of almost half of the data from this ecoregion in a single country. The GRooT has limitations for global modeling due to the unbalanced distribution of samples between ecoregions. This imbalance can compromise model performance and highlights the need to expand data sampling in certain regions, such as the Amazon rainforest, southern Africa and Antarctica.