Reversal of response to artificial selection on body size in a wild passerine

In quantitative genetics one often assumes that stabilizing selection maintain the intermediate phenotype which typically has a higher mean fitness in the average environment compared to larger or smaller phenotypes. We investigated the presence of such a phenotype in an eleven-year artificial selection experiment for long and short tarsus length (a proxy for body size) in wild house sparrow populations. The study system included 3 island populations in northern Norway where large and small individuals were artificially selected at the islands Leka and Vega, respectively, whereas the population at Hestmannøy was used as control. The artificial selection regimes at Leka and Vega were implemented towards the breeding population during 2002 – 2005, and then closely monitored during 2006 - 2012. The experimental selection resulted in strong selection in the imposed directions. However, because selected parents produced fewer recruits the artificial selection was counteracted by weak natural selection against individuals carrying extreme trait values. Significant responses to the artificial selection regimes were observed both at the phenotypic and genetic level (2002 – 2005), followed by a significant return towards pre-experimental means (2006 - 2012). This study demonstrates the evolvability of phenotypes and that selection may favour an intermediate phenotype in wild populations.