When applied to structured data, conventional random cross‐validation techniques can lead to underestimation of prediction error, and may result in inappropriate model selection. We present the r package block CV , a new toolbox for cross‐validation of species distribution modelling. Although it…