“…These problems are also observed in CMIP5 models (e.g., Maloney et al ., ; Fuentes‐Franco et al ., ), which, as CMIP3 GCMs (Rauscher et al ., ), show increased future dryness in Mesoamerica by the end of the 21st century (Cook and Seager, ; Maloney et al ., ; Fuentes‐Franco et al ., ; ; Torres‐Alavez et al ., ). In the case of the North American monsoon (NAM) region, some GCMs have shown an improvement from CMIP3 to CMIP5; however, GCMs still tend to overestimate winter and spring precipitation and show a late retreat of the monsoon rainfall during autumn (e.g., Cavazos and Arriaga‐Ramirez, ; Geil et al ., ; Seth et al ., ; Torres‐Alavez et al ., ). A weak start of the monsoon and a late retreat has been projected for the end of the century (e.g., Cook and Seager, ; Maloney et al ., ; Seth et al ., ; Torres‐Alavez et al ., ), but with large uncertainties according to dynamically downscaled GCMs from CMIP5 (Bukovsky et al ., ).…”