The global impacts of biodiversity loss and climate change are interlinked but the feedbacks between them are rarely assessed. Areas with greater tree diversity tend to be more productive, providing a greater carbon sink, and biodiversity loss could reduce these natural C sinks. Here, we quantify how tree and shrub species richness could affect biomass production at biome, national and regional scales. We find that greenhouse gas mitigation could help maintain tree diversity and thereby avoid a 9-39% reduction in terrestrial primary productivity across differ biomes, which cold otherwise occur over the next 50 years. Countries that will incur the greatest economic damages from climate change stand to benefit the most from conservation of tree diversity and primary productivity, which contributes to climate change mitigation. Our results emphasize an opportunity for a triple win for climate, biodiversity and society, and highlight how these co-benefits must be focused by reforestation programs.

Supplementary Data S1: A CSV file for the list of target species.

Supplementary Data S2: A CSV file containing the estimations of the effect sizes (lnRR) to alleviate the loss of woody plant species [-log(ΔSR_mitigation/ΔSR_baseline)] for terrestrial biomes, based on five Shared Socioeconomic Pathways (SSPs). Results are provided for ensembled means across three General Circulation Models (GCMs) and each of them. The upper and lower 95% confidence intervals (UpperCI and LowerCI) are provided. The number of coarse grids analyzed for each biome (NoCoarseGrids) is also included. BiomeIDs are consistent with those used in the paper. Abbreviations for GCMs: Ensembled, ensembled result; MIROC, MIROC-ESM-CHEM; HadGEM2, HadGEM2-ES; GFDL, GFDL-CM3.

Supplementary Data S3: A CSV file containing the estimations of the effect sizes (lnRR) of tree diversity-dependent productivity conservation [-log(ΔP_mitigation/ΔP_baseline)] for terrestrial biomes, based on five SSPs. Results are provided for ensembled means across three GCMs and each of them. The upper and lower 95% confidence intervals (UpperCI and LowerCI) are provided. The number of fine grids analyzed for each biome (NoFineGrids) is also included. BiomeIDs are consistent with those used in the paper. Abbreviations for GCMs: Ensembled, ensembled result; MIROC, MIROC-ESM-CHEM; HadGEM2, HadGEM2-ES; GFDL, GFDL-CM3.

Supplementary Data S4: Raster files containing the estimations of tree diversity-dependent productivity conservation [-log(ΔP_mitigation/ΔP_baseline)] at a spatial resolution of 30 arcminutes (maps in geographic coordinates of WGS84; Extended Data Figure S7), which were calculated using five SSPs and three GCMs. Value of -999 is for infinite values (denominator or numerator is zero and thus log-ratio cannot be calculated in these grids). Abbreviations for GCMs: MIROC, MIROC-ESM-CHEM; HadGEM2, HadGEM2-ES; GFDL, GFDL-CM3.

Supplementary Data S5: A CSV file containing the estimations of the effect sizes (lnRR) of tree diversity-dependent productivity conservation [-log(ΔP_mitigation/ΔP_baseline)] of the IPBES subregions, based on five SSPs. Results are provided for ensembled means across three GCMs and each of them. The number of fine grids analyzed for each subregion (NoFineGrids) is also included. Abbreviations for GCMs: Ensembled, ensembled result; MIROC, MIROC-ESM-CHEM; HadGEM2, HadGEM2-ES; GFDL, GFDL-CM3.

Supplementary Data S6: A CSV file containing the estimations of the effect sizes (lnRR) of tree diversity-dependent productivity conservation [-log(ΔP_mitigation/ΔP_baseline)] of countries (ADM0_A3_IS), based on five SSPs. Results are provided for ensembled means across three GCMs and each of them. The number of fine grids analyzed for each country (NoFineGrids) is also included. Abbreviations for GCMs: Ensembled, ensembled result; MIROC, MIROC-ESM-CHEM; HadGEM2, HadGEM2-ES; GFDL, GFDL-CM3