O1: Identifying the environmental correlates of Atlantic salmon run timing throughout the species’ range
Our results show that adult salmon migration timing or ‘run timing’ in North America have started occurring earlier over the past 29 years (1993-2021). Worryingly, the population that migrates the earliest is now at record low numbers.
We used climate data from the WorldClim database for each population and combined it with genetic findings from O2 to predict how vulnerable different salmon populations are to climate change (see O3).
O2: Characterising the genomic architecture of run timing throughout the species’ range
Although we had long-term records of when adult salmon returned, we did not have genetic samples for most of these fish. To overcome this, we used population-level return timing and linked it to population-level genetic differences. We did this for 11 North American salmon populations using a SNP array that scanned over 220,000 genetic markers.
We found many genes associated with migration timing. Some were already known from studies of European salmon, showing shared genetic traits across continents. Our findings also suggest that historical mixing between EU and North American populations at the end of the last Ice Age may have shaped these traits.
We then took a closer look at genetic differences using low-coverage whole genome sequencing on salmon from 7 North American populations. This more detailed analysis revealed a new genomic region strongly linked to migration timing. Interestingly, this region includes genes also associated with long-distance migration in birds, hinting at shared biological processes across species.
We also studied the genetics of “smolt migration” — the journey of young salmon from river to ocean. Thanks to our collaborators, we collected smolt samples from 10 rivers across 7 countries and recorded their migration dates. We genotyped the earliest and latest migrants using a 60K SNP chip to look for associations between genes and migration timing.
Given differences between North American and EU salmon, we analysed the continents separately. In EU populations, one genomic region stood out as being strongly linked to smolt migration and also linked to adult return timing. Early findings from North America suggest Ice Age population mixing may also have influenced smolt migration genetics.
O3: combining the outcomes of O1 and O2 in a modelling framework to predict the migration-mediated vulnerability of Atlantic salmon populations to future environmental change
Once we knew which genes influence migration timing and how this timing is shaped by the environment, we used machine-learning models to forecast how future climate conditions could threaten salmon. We calculated a "genomic offset", which is the gap between a population’s current genetic makeup and the genetic changes it would need to stay in sync with a changing climate.
In North America, we found no clear environmental link for early-migrating adults. But late-migrating and multi-peak populations showed strong genetic-environment links. Our models revealed that northern populations face the largest genomic mismatch, meaning they may struggle most to keep pace with climate change.
For smolts, we found no clear environmental link to their migration genes, so genomic offsets were not applied.
