Study populations and data collection
The study was performed in four insular house sparrow populations off the coast of northern Norway (Fig. S1.1 in Appendix S1). The study periods differed between the populations with data from Hestmannøy (66°33’N, 12°50‘E) in the years 1994-2013, Træna (Husøy island, 66°30’N, 12°05‘E) in the years 2004-2013, and Leka (65°06’N, 11°38’E) and Vega (65°40’N, 11°55’E) both in the years 2002-2006. Hestmannøy and Træna were unmanipulated natural populations and are included in the primary analyses. The populations of Leka and Vega underwent artificial size selection (see Kvalnes et al., 2017; Pepke et al., 2021,submitted ) and were analyzed separately in a set of secondary analyses as replications of the primary analyses. All four islands are characterized by heathland, mountains, and sparse forest. The sparrows live closely associated with humans and within the study area they are found mainly on dairy farms (Hestmannøy, Vega and Leka), where they have access to food and shelter all year, or in gardens and residential areas (Træna), where they may be more exposed to weather conditions (Araya-Ajoy et al., 2019). Natural nests inside barns or artificial nest boxes were visited at least every 9th day during the breeding season (May-August) to sample fledglings (5-14 days old, with a median of 11 days). All individuals were ringed using a unique combination of a metal ring and three plastic color rings. Fledged juvenile sparrows and unmarked adults were captured using mist nets during the summer and autumn (September-October). These procedures ensured that approximately 90% of all adult birds were marked on all islands during the study period (Jensen, Steinsland, Ringsby, & Sæther, 2008; Kvalnes et al., 2017). For most fledglings, we measured tarsometatarsus (tarsus) length using digital slide calipers to nearest 0.01 mm and body mass to nearest 0.1 g with a Pesola spring balance (see details in Appendix S1). For 234 nestlings, no nestling morphological measurements were available. Following Schulte-Hostedde, Zinner, Millar, and Hickling (2005) nestling body condition was calculated as the residuals of a linear regression of mass on tarsus length (both log10-transformed). To avoid collinearity in models where both nestling age and tarsus length were included as covariates, we age-corrected tarsus length by using the residuals from a regression of tarsus length on age and age squared (to account for the diminishing increase in tarsus length with age). A blood sample (25 μL) was collected from all individuals, which was stored in 96% ethanol at room temperature in the field and subsequently at -20°C in the laboratory until DNA extraction.
Molecular sexing and pedigree construction
DNA extraction is described in Appendix S1. Sex of most fledglings (n =2641) was determined using amplification of the CHD-gene located on the avian sex chromosomes as described in Griffiths, Double, Orr, and Dawson (1998). 21 individuals were sexed exclusively based on their phenotype as adults and 84 nestlings could not be sexed. We used individual genotypes on 13 polymorphic microsatellite markers scored using the GeneMapper 4.0 software (Applied Biosystems) to assign parentage in CERVUS 3.0 (Kalinowski, Taper, & Marshall, 2007), as detailed in Rønning et al. (2016). Briefly, for each nestling, CERVUS calculates a LOD‐score (log‐likelihood ratio) for all putative parents, which is compared to the critical values generated by the simulated parentage analyses, resulting in a 95% parentage assignment confidence. Nestlings within the same clutch were assumed to have the same mother. Nestlings with missing (unassigned) parents were assigned dummy parents, assuming that nestlings within the same clutch were full siblings and thus had the same (unassigned) parents. The dummy parents were included in the pedigree as founders. We calculated individual inbreeding coefficients (F ) based on the microsatellite pedigree using the R package ‘pedigree’ (Coster, 2012). Pedigrees were ordered using the R package ‘MasterBayes’ (Hadfield, Richardson, & Burke, 2006) and pruned to only contain informative individuals. The pruned pedigrees included 4118 individuals (3093 maternities and 3130 paternities) in the natural populations, and 1057 individuals in artificially selected populations. Maximum pedigree depth was 13 generations, the number of equivalent complete generations (the sum of the proportion of known ancestors across all generations, Wellmann, 2021) was 1.510, and mean pairwise relatedness was 0.003.