<p>This study aims to quantify and regionalize future changes in daily minimum (T_min), maximum (T_max), and the diurnal temperature range (DTR = T_max − T_min) across South America for 2015–2100, addressing the current lack of dedicated sub-regional assessments of DTR projections using Coupled Model Intercomparison Project Phase 6 (CMIP6) models. Model skill was assessed against the Climatic Research Unit (CRU) Time-Series (TS) dataset, and the five best-performing members were used to build sub-regional ensembles for ten predefined domains to improve local signal detection. Trend analysis shows a continent-wide tendency toward DTR compression driven by stronger nighttime warming: median DTR trends are − 7.6 × 10⁻³ °C year⁻¹ (SSP2-4.5; Shared Socioeconomic Pathways) and − 1.28 × 10⁻² °C year⁻¹ (SSP5-8.5), with mean reductions between 2015 and 2025 and 2090–2100 of ≈ − 0.71&#xa0;°C (SSP2-4.5) and ≈ − 1.04&#xa0;°C (SSP5-8.5). The spatial pattern is heterogeneous: most domains exhibit the largest declines, while parts of the eastern coast show near-neutral DTR under the low-emission pathway and a clear tendency to increase under the high-emission pathway. Trend significance was evaluated using a modified Mann–Kendall test (α = 0.05) and Theil–Sen slope estimates; 5 of 10 sub-regions display statistically significant DTR trends under SSP2-4.5, increasing to 8 of 10 under SSP5-8.5, indicating stronger and more widespread signals under higher forcing. Crucially, regionalizing ensembles uncovers coastal signatures masked in continental means, offering a novel contribution to CMIP6-based assessments of DTR in South America. Given model limitations, targeted attribution studies and investments in high-resolution observational networks and regional climate modeling are recommended. Despite these uncertainties, the results provide valuable guidance for adaptation planning in health, agriculture, and energy sectors across South America.</p> Graphical Abstract <p></p> <p>This visual summary presents a study that quantifies projected changes in T_min, T_max, and the DTR across South America through 2100, using subregional ensembles of CMIP6 models composed of the five highest-skill members in each domain. It conveys the workflow and main results through integrated visual components: the top-left panel maps South America into 10 subregions; the central-top panel displays the multimodel CMIP6 catalogue; the top-right panel shows Taylor diagrams and historical series used to select the top five models per subregion; the central flow outlines spatial aggregation, skill evaluation, ensemble construction, trend estimation with the Theil–Sen estimator, and significance testing via a modified Mann–Kendall test; the bottom-right contains time-series panels of T_min and T_max under scenarios; and the bottom maps present projected DTR anomalies (SSP2-4.5, SSP5-8.5) while the bottom-left summarizes DTR by subregion, highlighting spatial heterogeneity. Projections indicate continent-wide DTR compression, with median trends of -7.6 × 10⁻³ °C year⁻¹ (SSP2-4.5) and − 1.28 × 10⁻² °C year⁻¹ (SSP5-8.5), and mean reductions from 2015 to 2025 to 2090–2100 of ≈ − 0.71&#xa0;°C and ≈ − 1.04&#xa0;°C. The decline reflects warming of both T_max and T_min, with a stronger rise in T_min. This asymmetric warming may be related to enhanced longwave radiation, higher humidity, cloud-cover changes, or land-use alterations. Most domains show robust, significant DTR declines, while parts of the eastern coast exhibit neutral or positive trends in some seasons and scenarios — patterns revealed only by the subregional ensemble approach. These shifts imply hotter nights, reduced thermal comfort, greater stress on health and energy systems, altered phenology, and uneven agricultural impacts between interior and coastal regions, underscoring the need for locally targeted adaptation and high-resolution attribution studies.</p>