Artificial selection on tarsus length
In February and early March each year from 2002 to 2012 (11 years, Fig. 2) ca. 90% of the house sparrows on Leka (n =89-222 per year) and Vega (n =102-352 per year) were caught using mist-netting in or around farms and then held in separate aviaries for up to 12 days withad libitum water and food (sunflower seeds, cattle grain and bread) to obtain morphological measurements. During 2002-2005, adult sparrows with tarsi shorter (Leka, referred to as the highpopulation) or longer (Vega, referred to as the low population) than the island population mean ±0.3 SD, within each sex, were translocated to distant localities (Ranke et al., 2020) and thus removed from the breeding populations (see Kvalnes et al., 2017 for details of the selection procedure). Thus, both populations were subject to the same artificial selection pressure, but in opposite directions. On average, 56.4% (highpopulation) and 62.9% (lowpopulation) of the captured individuals were removed each year, whereas the selected birds were released back at their capture location. Thus, because ca. 10% of the individuals present at the time of selection were not captured, the artificially selected individuals constituted ca. 78% of the breeding populations (Kvalnes et al., 2017). In the first year (2002) of the selection experiment, there was no significant difference in average fledgling tarsus lengths between the highand low populations (n =167,βlow =0.22, lower and upper 95% confidence interval (CI)=[-0.25, 0.70] accounting for age at sampling and sex). In the years following the end of the selection experiment (from 2006) all birds were released back at the original capture location (Rønning et al., 2016).