Further directions
Time-matched models have potential benefits beyond current estimations of biodiversity distributions. One such benefit is the ability to estimate how species are responding and will respond to climate change, based on fewer years of future climatic projections. Occurrence-environment matching approaches can capture extreme weather events hidden in long-term averages, events which could drastically affect population growth or even drive local extirpation depending on the species (Frederiksen et al. 2008, Bauerfeind and Fischer 2014, Marcelino et al. 2020). Studies have shown that even in cases where models that incorporate the frequency of extreme events have similar geographic predictions to standard average models in the current time, they could differ considerably in future geographic estimations (Moran-Ordonez et al. 2018). Given that the frequency of extreme events is becoming higher due to global warming, studies that incorporate extreme events intrinsically in model training (as in this study) or by using extreme-derived variables (e.g., Zimmermann et al. 2009, Feldmeier et al. 2018) should be considered in any conservation study that accounts for climate change.
Alongside investigating the differences between time-matched and standard approaches in climate change impacts on species ranges, further studies should explore the benefits of using time-matched remote sensing data. Derived remote-sensing variables such as tree forest cover (Hansen et al. 2013) or the Normalized Difference Vegetation Index may be used together with temperature and precipitation at the exact same temporal resolution to get almost real-time estimations of habitat suitability (Crego et al. 2022, Paz et al. 2022). For instance, based on the positive results using a ten-year resolution for estimating the C. mexicanus distribution, time-matching of traditional bioclimatic variables could be combined with the two decades of data captured by remote sensing instruments. When sufficient recent occurrence data exist, this represents a viable pathway that holds advantages over the common alternative of building a climate-only model and post-processing it by masking out regions no longer holding suitable habitat (e.g., based on remote sensing; Merow et al. 2022). This line of further research integrating disparate environmental data streams and temporal matching could enable us to obtain better model predictions and forecast how species respond to the changing environment.