Study populations and field methods
The study comprises data from eight subpopulations of an insular house sparrow metapopulation located in the Helgeland archipelago in northern Norway (66°30′N, 12°30′E; Figure 1). These 8 islands are part of a larger metapopulation study system covering ca. 1600 km2, where extensive field work during the breeding seasons (May-August) and autumns (September-October) since 1993 has ensured that most house sparrows have been individually marked and followed through their lives from hatching to death (Baalsrud et al., 2014; Ranke et al., 2021; Saatoglu et al., 2021). From each ringed individual we collected a small blood sample (~25 μl) to obtain DNA for genotyping purposes, genetic parentage determination and pedigree construction (Niskanen et al., 2020). The subpopulations on islands in this metapopulation are almost solely interconnected by natal dispersal (breeding dispersal < 0.3%; Altwegg, Ringsby, & Sæther, 2000; Jensen et al., 2013; Tufto, Ringsby, Dhondt, Adriaensen, & Matthysen, 2005) with an overall dispersal rate of 22.2% (Saatoglu et al., 2021). However, dispersal rates also differ among islands and years, and depend on environmental conditions related to the habitat type of the subpopulations (Ranke et al., 2021; Saatoglu et al., 2021). On five islands (Aldra, Gjerøy, Hestmannøy, Indre Kvarøy and Nesøy), which are also denoted as “farm” islands, house sparrows nest in a colony-like manner on and near dairy farms, and thus have higher local densities but are also more sheltered during harsh weather conditions (Araya-Ajoy et al., 2019). Conversely, in the subpopulations located farther from the mainland (Myken, Træna and Selvær), denoted “non-farm” islands, the house sparrows live and breed in nest boxes and other suitable cavities on the houses in small villages. Non-farm islands have less sheltering opportunities and probably less stable food availability during winter-time (Araya-Ajoy et al., 2019). We have previously shown that the dispersal rate is higher among the non-farm islands (24.3%) than among the farm islands (9.6%), and that the dispersal rate is higher from non-farm to farm islands (7.2%) than from farm to non-farm islands (2.1%; see Saatoglu et al., 2021).