Methods
In this study, I used the Mexican small-eared shrew, Cryptotis mexicanus , to compare the model performance of the time-matched approach against the standard 30-year approach (STA). The workflow consisted of five steps: (i) collection of occurrence and background data, (ii) generation of bioclimatic variables, (iii) comparison of environmental values, (iv) training and selection of models, and (v) comparison between models (Fig. 2). Cryptotis mexicanus is a cloud forest specialist that inhabits the Sierra Madre Oriental and the mountainous areas of Oaxaca State in Mexico. It is a good model species for exploring predictions of environmental suitability using several temporal resolutions for three reasons. First, its geographical distribution seems largely constrained by moisture and temperature, which allows for informative ENMs based on a few bioclimatic variables (Guevara et al. 2018, Guevara 2020). Second, shrews are considered one of the mammalian groups with the lowest individual dispersal capabilities (Schloss et al. 2012, Urban et al. 2013), which enables me to test the reliability of time-matched ENMs in a species with low mobility. Third, it has a fairly short, estimated generation time of 330 days, given its body size (Pacifici et al. 2013), which allows me to explore the existence of different time lag relationships between population dynamics and the temporal resolutions of the environmental data (one, five, and ten years prior to the records).