We established our study sites in the cities of Basel, Lugano and Zurich, and their peri-urban forests. We characterised urban tree cover of each city using a rectangular grid with squares of 100x100 m. Within each square, we measured the area covered by urban trees using the European Union's Copernicus Land Monitoring Service information, Urban Atlas Street Tree Layer 2018 https://doi.org/10.2909/205691b3-7ae9-41dd-abf1-1fbf60d72c8c.

Then we assigned each square to four categories of urban tree cover that roughly represented the main types of urban uses:

1) low cover, industrial/commercial areas, 0-20% tree cover; 2) intermediate cover, residential areas, 20-40% tree cover; 3) high cover, urban parks and cemeteries, 40-60% tree cover; 4) peri-urban forests (natural/semi-natural forests), 60-100% tree cover.

Our study sites consisted of eight 1.2 km-long transects in each city, equally distributed across the four tree cover categories. We designed an experimental measurements of bird predation rate on herbivore insect larvae using artificial caterpillars. They mimicked a variety of common, cryptically-coloured herbivorous species, which are usually preyed by birds. We designed caterpillars with green water-resistant modelling clay (Staedtler Noris®), 4-6 cm long and approximately 0.5 cm wide. Caterpillars were threaded to a thin (Ø = 0.6 mm) zinc-plated iron wire and tethered to branches of trees or bushes at a height up to 2 m from the ground. We placed 2-4 caterpillar mimics every 200 m along each transect (five points per transect) or, when not possible due to absence of suitable branches, within 200 m from the point.

In total, we placed 270 caterpillar mimics at 115 points during the late breeding season (June-July 2023). Caterpillar mimics were checked after 10-14 days and replaced with new ones at new locations within 10 m from the previous ones. During each of the two visits, we counted the number of bird predation marks.