Bioclimatic variables
Climate data were obtained from the Terraclimate dataset, which provides
monthly temperature and precipitation data at a resolution of 5 arcmin
(Abatzoglou et al. 2018). The ClimateR R package
(https://github.com/mikejohnson51/climateR)
was used to download the climate data in R. Four bioclimatic variables
– maximum temperature of the warmest month (bio05), minimum temperature
of the coldest month (bio06), precipitation of the wettest month
(bio13), and precipitation of the driest month (bio14) – were selected
for model training. These variables were chosen based on previous
studies of the species, which have established their usefulness when
capturing the averages of the extreme temperature and precipitation
conditions (Guevara et al. 2018).
To analyse the impact of time-matched approaches in comparison to models
without time-matching, I created bioclimatic variables of different
temporal resolutions. First, a single set of the standard 30-year
bioclimatic average variables (STA; Fig. 1a) was created using monthly
data from January 1971 to December 2000, based on a modified version of
the biovars() function of the dismo R package (Hijmans et al. 2021) that
use terra R package (Hijmans 2022) for faster processing. Multiple
bioclimatic variables were then generated using a one-forward month
moving window to explore predictions using several temporal resolutions
of occurrence-environment matching. For instance, the one-year
occurrence-environment matching (T01) used bioclimatic values generated
within the 12 months preceding the occurrence observation, including the
month of observation. Similarly, the five-year temporal resolution (T05)
used 60 months, and the ten-year temporal resolution (T10; Fig. 1b) used
120 months. Environmental values for both occurrence and background
points were obtained by aligning the collection date with the last month
of each window period. For example, if an occurrence record was observed
in January 1971, the bioclimatic values used for ten-year matching
corresponded to the average of 120 months between February 1961 and
January 1971.