Understanding spatial effects in species distribution models.

Area of Science:

• Ecology
• Environmental Science
• Statistical Modeling

Background:

• Species Distribution Models (SDMs) often incorporate spatial effects to enhance predictions and identify environmental drivers.
• Ecologists sometimes attempt to interpret the spatial patterns derived from these effects.
• However, spatial autocorrelation can arise from numerous unmeasured variables, hindering ecological interpretation of spatial effects.

Purpose of the Study:

• To demonstrate that spatial effects in statistical models can effectively smooth over the influence of multiple unaccounted drivers.
• To illustrate the challenges in ecologically interpreting spatial effects when unmeasured variables are present.

Main Methods:

• A simulation study was conducted to investigate the behavior of spatial effects.
• Model-based spatial models were fitted using geostatistics and 2D smoothing splines.
• The simulation focused on how spatial effects represent unmeasured covariates.

Main Results:

• Fitted spatial effects were shown to approximate the combined surface of unaccounted covariates.
• Both geostatistical and 2D smoothing spline methods produced similar results regarding the smoothing of unmeasured drivers.
• The study provides empirical evidence for the smoothing effect of spatial components.

Conclusions:

• Spatial effects in SDMs can obscure the individual contributions of unmeasured environmental drivers by integrating their influence.
• Direct ecological interpretation of spatial effects should be approached with caution due to potential confounding by unmeasured variables.
• Understanding the smoothing nature of spatial effects is crucial for accurate ecological inference in SDMs.