Code, simulation outputs, figures, and supplementary information for the paper "One size does not fit all: ignoring spatial variation in occupancy and detectability underestimates the required number of sites" (Joshi, Zedrosser, Sollmann, Selva and Santoro). We used simulations to ask how many camera-trap sites are needed for reliable single-season occupancy estimation when occupancy and detection vary across space. We tested three scenarios: constant probabilities, simple spatial variation (one covariate per process), and complex variation (multiple or multilevel covariates). For each, we fitted the matching occupancy model in unmarked and found the smallest number of sites that kept the median RMSE at or below 0.10. The archive includes the R scripts (scenarios and figures), all simulation outputs, the final figures with captions, the supplementary information file, and a README explaining the structure and how to run everything. Analyses were run in R 4.5.2 with unmarked 1.5.0. All outputs are included, so the figures can be reproduced without re-running the simulations. Funding: Biodiversa+ (European Biodiversity Partnership), WildINTEL project (GA No. 101052342), co-funded by the European Commission, Agencia Estatal de Investigación (Spain, PCI2024-153489), the National Science Centre (Poland, UMO-2023/05/Y/NZ8/00104), the Research Council of Norway (NFR350962), and the German Research Foundation.