Urbanisation is rapidly altering ecosystems, leading to profound biodiversity loss. To mitigate these effects, we need a better understanding of how urbanisation impacts dispersal and reproduction. Two contrasting population demographic models have been proposed that predict that urbanisation either promotes (facilitation model) or constrains (fragmentation model) gene flow and genetic diversity. Which of these models prevails likely depends on the strength of selection on specific phenotypic traits that influence dispersal, survival, or reproduction. Here, we a priori examined the genomic impact of urbanisation on the Neotropical túngara frog (Engystomops pustulosus), a species known to adapt its reproductive traits to urban selective pressures. Using whole-genome resequencing for multiple urban and forest populations we examined genomic diversity, population connectivity and demographic history. Contrary to both the fragmentation and facilitation models, urban populations did not exhibit substantial changes in genomic diversity or differentiation compared to forest populations, and genomic variation was best explained by geographic distance rather than environmental factors. Adopting an a posteriori approach, we additionally found both urban and forest populations to have undergone population declines. The timing of these declines appears to coincide with extensive human activity around the Panama Canal during the last few centuries rather than recent urbanisation. Our study highlights the long-lasting legacy of past anthropogenic disturbances in the genome and the importance of considering the historical context in urban evolution studies as anthropogenic effects may be extensive and impact non-urban areas on both recent and older timescales. 

# Genomic footprints of (pre) colonialism: population declines in urban and forest túngara frogs coincident with historical human activity ## Description of the data and file structure The metadata file encompasses sampling location, corresponding population code (pop), and environmental data for forest (F) and urban (U) sites. The collection of light (in Lux), noise (in dB SPL, fast, max, A-weighted) and canopy cover data (percentage canopy cover estimated from pictures) was previously described and published in Halfwerk et al., 2019. The level of urbanisation (Urban_score) was calculated based on the type of landscape-cover for each sampling location using ‘Urbanisation Score’ software (Lipovits et al., 2015; Seress et al., 2014). This program accesses satellite images via GoogleMaps and applies a semi-automated approach to quantify the relative abundance of vegetation and impervious surfaces within a 1 km2 area around each sampling location. These values were then combined using principal component analysis (PCA) and an urbanisation score retained (PC1) for each location. ## Sharing/Access information Related data: Whole genome resequencing data used in this study is available from the European Nucleotide Archive (ENA) (PRJEB60348). ## Code/Software Main bash scripts for running software and R code used for analyses. Additional custom scripts are available from the corresponding authors upon request.

Full Methods description provided in manuscript: Moran et al., 2023 Genomic data: Whole genome resequencing data used in this study is available from the European Nucleotide Archive (ENA) (PRJEB60348). Environmental data: The collection of light (in Lux), noise (in dB SPL, fast, max, A-weighted) and canopy cover data (percentage canopy cover estimated from pictures) data was previously described and published in Halfwerk et al., 2019. The level of urbanisation (Urban_score) was calculated based on the type of landscape-cover for each sampling location using ‘Urbanisation Score’ software (Lipovits et al., 2015; Seress et al., 2014). This program accesses satellite images via GoogleMaps and applies a semi-automated approach to quantify the relative abundance of vegetation and impervious surfaces within a 1 km2 area around each sampling location. These values were then combined using principal component analysis (PCA) and an urbanisation score retained (PC1) for each location.