The idea that there’s safety in numbers might be well known, but it’s one that is difficult to test in many species.
A recent University of Washington study based on decades of historical data found Pacific salmon in larger groups are less likely to be eaten by predators.
The study, published June 29 in the journal Science Advances, found unique support for the “safety in numbers” hypothesis by looking at data collected from 1956 to 1991 for four species of Pacific salmon — sockeye, chum, coho and pink.
“With salmon, most people think of them spawning in freshwater streams, but there’s also this huge amount of time they spend in the ocean feeding and growing,” said lead author Anne Polyakov, a doctoral student in UW’s Quantitative Ecology and Resource Management Program and the School of Aquatic and Fishery Sciences.
The historic salmon catch data across the North Pacific Ocean was collected by UW’s Fisheries Research Institute in partnership with the International North Pacific Fisheries Commission.
“One of the reasons why this study is so unique is that we essentially can’t observe these fish at all in their natural ocean environment, and yet we’re able to pull out these really strong results on how grouping affects predation risk and foraging success for individual fish using this incredibly valuable data set,” Polyakov said in a UW news release about the study.
The study’s authors analyzed catch data from purse seine nets, which capture all of the fish in a small area. Researchers estimated the size of schools based on how many fish were caught in these nets.
The data also included records of what was in salmon’s stomachs, giving researchers a way to gauge feeding success, according to UW.
The study found that sockeye and chum in larger groups had less food in their stomachs, suggesting fish sacrificed a meal for the safety that comes with being in a larger group.
The authors said they hope their paper inspires group size distributions and the benefits and costs of grouping to be considered in fisheries management models.
--------
