3Loss of habitat connectivity is a well-known threat to biodiversity and metapopulation 4 persistence and anticipated to strongly affect evolutionary dynamics. While selection is 5 typically expected to act on dispersal-related traits that are expressed at the metapopulation-6 level, demographic changes in complex spatial networks may cause life histories to diverge 7 between local patches as well. However, the exact consequences of reduced connectedness on 8 life-history traits have never been quantified in detail. 9 We used experimental metapopulations of two-spotted spider mites to investigate how 10 connectedness loss affects trait evolution. Mites were introduced in spatial networks with 11 different levels of connectedness among local patches. In order to separate processes of local 12 and metapopulation-level selection, we implemented a randomization treatment by 13 redistributing mites across all local patches of certain metapopulations. 14 A decreased metapopulation connectedness selected for increased starvation resistance and 15 delayed dispersal timing. Reshuffling did not alter any of these evolutionary processes, 16 indicating that metapopulation-level selection was prominent. Intrinsic growth rates were, 17 however, systematically lower in individuals that originated from the most connected local 18 patch. Because this pattern was maintained under reshuffling, fast epigenetic evolution must 19 have been acting as well. 20 We thus demonstrate selection on multiple traits following connectedness loss related to 21 population growth rate and dispersal, but not dispersal rate per se. We attribute these 22 evolutionary dynamics to feedbacks between costs of dispersal and spatial demographic 23 variation. With longer dispersal, starvation resistance improved, possibly to face longer periods 24 2 in the matrix without resources; at the same time, lower dispersal costs translated into a quicker 25 decision to disperse as the net loss/gain ratio was more favourable.
26Our results demonstrate the need to incorporate multivariate trait evolution at multiple spatial 27 scales to understand eco-evolutionary dynamics in metapopulations. 28 65 2012). It can be assumed that connectedness loss mostly influences the costs of the transfer 66 phase, thereby imposing selection by lowering departure rates (Bonte et al., 2003). Evolutionary 67 adaptations to reduce costs of the other dispersal phases may, however, be prominent as well 68 and result in dispersal syndromes (Clobert et al., 2012). Evolutionary changes in (pre-) 69 departure behaviours (Derr, Aldent and Dinglet, 1981; Bell et al., 2005; Cañete et al., 2007), 70 dispersal timing (Lakovic, Poethke and Hovestadt, 2015), movement (Korb and Linsenmair, 71 2002; Baguette and Van Dyck, 2007) and post-settlement strategies (Vessby and Wiktelius, 72 2003; Hansson, Bensch and Hasselquist, 2004) in response to altered habitat connectedness are 73 becoming well-documented. Because interactions and/or trade-offs between two or more of 74 these traits are also...