Demographic consequences of rapid environmental change and extreme climatic events (ECEs) can cascade across trophic levels with evolutionary implications that have rarely been explored. Here, we show how an ECE in high Arctic Svalbard triggered a trophic chain reaction, directly or indirectly affecting the demography of both overwintering and migratory vertebrates, ultimately inducing a shift in density-dependent phenotypic selection in migratory geese. A record-breaking rain-on-snow event and ice-locked pastures led to reindeer mass starvation and a population crash, followed by a period of low mortality and population recovery. This caused lagged, long-lasting reductions in reindeer carrion numbers and resultant low abundances of Arctic foxes, a scavenger on reindeer and predator of migratory birds. The associated decrease in Arctic fox predation of goose offspring allowed for a rapid increase in barnacle goose densities. As expected according to r- and K-selection theory, the goose body condition (affecting reproduction and post-fledging survival) maximising Malthusian fitness increased with this shift in population density. Thus, the winter ECE acting on reindeer and their scavenger, the Arctic fox, indirectly selected for higher body condition in migratory geese. This high Arctic study provides rare empirical evidence of links between ECEs, community dynamics and evolution, with implications for our understanding of indirect eco-evolutionary impacts of global change.

First dataset 

The first data set includes the data to run the correlative analysis, presented in Figure 2 in the article. This was based on time series data on Svalbard reindeer abundance, a proxy of Arctic fox abundance and abundance data of barnacle geese on high Arctic Svalbard. Annual estimates of barnacle goose population size at Ny-Ålesund were obtained from the results of an integrated population model. All data are accessible in published articles or online. The percentage of occupied dens was used as a proxy of abundance of Arctic foxes, based on annual records of known den sites around Ny‐Ålesund with pup production during summer (Fuglei et al. 2003) and were obtained from MOSJ (2021). Data of reindeer abundances (population counts) were available from Supplementary Information Figure S1 in Hansen et al. (2019). Data of goose abundance (total population size estimated with an integrated population model) were obtained from (Layton-Matthews et al. 2019). Please see respective publications for details on the data collection methods. 

Second dataset

The second dataset includes individual-level data to analyse demography-trait relationships (Figure 3) and to model eco-evolutionary dynamics (Figure 4). The data includes a unique bird identification number, observation year, body mass (g), tarsus length (cm), whether that individual survived the following breeding season (I) and the number of recruits of both sexes contributed by that individual to the following breeding season (B). These data were available over the study period 1991-1999. Females were caught during the breeding season, more specifically during the moulting period when parents raise their offspring, and ringed with colour and metal identification rings. Recapture data were based on daily observations of ringed individuals around Ny-Ålesund. We assessed reproduction based on observations of adults with fledged offspring at the beginning of August. The number of recruits the following year was based on parentage data where families were monitored after capture (and ring marking) to attribute parent–offspring relationships (ref 64 in article). We encourage persons wishing to use these data in further analyses to contact the data owner and curator: Maarten J.J.E. Loonen (email: m.j.j.e.loonen@rug.nl).