A fish tale that slams science

How cod ecology reveals the limitations of science and mathematical modeling.

Dear readers, I recently came across a fascinating story about how scientists had failed to predict cod populations off the coast of Canada, with disastrous results. It was interesting to me because the ecologists’ prediction problem is also found in psychology and mental health research. Many of the ‘systems’ scientists care about, from minds to ecosystems, are so complex that you can’t capture them with deterministic models. How do we deal with this ‘modeling gap’? Perhaps we can learn something from a cautionary fish tale. I hope you enjoy it! With thanks to my friend Dr. Daan Mes, an expert in marine biology, for his helpful comments on a draft of the piece.

~ Jeroen

---

Now that I live in Washington DC, I take every opportunity I get to sample the seafood sold at a floating market down by the wharf. It’s the oldest open-air fish market in continuous operation in the United States, dating back to 1805. But while the market is a local feature, the fish itself is not. This is partly due to globalization: tilapia is farm-grown in countries like China and Indonesia, frozen, shipped, and sold defrosted at the wharf. But even the cod, a fish historically abundant in the Atlantic Ocean, is rarely caught nearby. As it turns out, the reason for this is not economical, it’s ecological. And it provides a valuable view into the limitations of science and mathematical modeling.

Patrons buy seafood at the DC Wharf.

The story starts about five hundred years ago, when Spanish and Portuguese fishermen sailed West to find huge populations of Atlantic cod swimming around what we now call the Grand Banks of Newfoundland. The fishing was good and the fishermen hauled home barrel upon barrel of Basque-style salted bacalao, roughly 150 metric tons per year. This became a staple protein in Europe for centuries and for a while, all was well.

Enter modernity. By the middle of the twentieth century, engineers had developed such powerful bottom-scraping nets and fish-finding sonar systems that cod landings shot up to over 1,000,000 tons in 1968 (see figure). European freezer trawlers took home a big share of this pie, which upset local fishermen who had started seeing sub-par yields. The Canadian government therefore claimed an exclusive economic zone not 3 but 200 miles off the coast, which contained about 90% of the Grand Banks area. The locals then deployed massive trawlers of their own. This led to a decrease in overall cod landing but a recovery in Canadian cod production, and for a while, all was well.

Soon, however, the yearly yield of cod stagnated again. This is when fishermen and policymakers got worried. Could it be that there was something wrong with the cod population itself? Clever ecological modelers came up with a way of calculating a ‘maximum sustainable yield’ (MSY), set at 16% of the total population, which should theoretically leave enough fish to repopulate each year. For a while, all was well, as this mathematical might settled everyone’s nerves. But fishing floundered further and the Grand Banks cod population collapsed almost entirely in 1992. The government quickly called a moratorium on cod fishing, which was renewed indefinitely in 1993. It marked the end of what was once the greatest fishery on Earth.

Atlantic northwest cod capture in millions of tons, graphed by Wikipedia user Epipelagic. While the source data link for this figure is broken, the picture is corroborated by alternate sources e.g. here and here.

The collapse of the Grand Banks cod industry was a tragedy for Canada and in particular the provinces of Newfoundland and Labrador. Centuries-old fishing towns crumbled as young people moved away to the cities, mirroring declines in the American ‘Rust Belt’ and other once-famed industrial zones. When asking who was responsible, it is easy to point to greedy fishermen or zealous politicians as the cause of the Grand Banks cod’s demise. But as reported by science journalist Deborah Mackenzie, and retold well by Duke ecologists Orrin Pilkey and Linda Pilkey-Jarvis in their sobering book Useless arithmetic, some of the blame should land on scientists, too.

This is because the magic MSY is in fact a hopeless knot of faulty assumptions. It is obtained by computing how many baby cod will be born from a given population. If you stop fishing at the point where the projected number of babies equals the number of fish caught, you get a population that replenishes itself back to 100% before the next season starts. Makes sense, right?

Well, not so fast. First, as Mackenzie quips in a 1995 New Scientist article, ‘the central problem is that fish live in the sea’. That is to say: you cannot count them all. Instead, scientists take samples of fish, assess their age, and infer what the total population must look like to have produced such a sample. This is where practical and political concerns come in. While the Canadian government attempted to sample the cod population in the 1980s, their ships caught so much less than professional fishermen that the ecological modelers deferred to the pros’ samples to infer population size. In doing so, the modelers ignored a selection bias: the pros used better tech and only fished in the highest-yielding spots, so these numbers cannot be extrapolated to the entire region.

A second assumption has to do with the way cod make babies. In humans, the number of kids in a population depends heavily on the number of parents, because one pair of parents usually has just one kid at a time. In cod, on the other hand, a single fish can produce eight million eggs at a time. This means that the number of cod babies who make it to adulthood depends much less on the existing population size and much more on environmental factors like food and predation. This was encapsulated in the model that produced the MSY by assuming a fixed ‘recruitment’ number each year, independent of population size. The issue is that in severely depleted populations, recruitment DOES depend on the number of breeding adults, rendering the model overly optimistic.

A third problem is that the fishing industry has far-reaching and often unforeseeable effects on the ecosystem as a whole. New trawl nets invented in the 1980s, write the Pilkeys, severely damage the ocean floor. ‘Juvenile cod survive best in areas that have rough bottoms, hiding from predators behind and within the many nooks and crannies afforded by such a seafloor. […] Since an area equal to all of the world’s continental shelves is trawled every two years, the habitats provided by an uneven seafloor disappear into geologic history.’ These downstream effects of human activity are rarely accounted for in even the best ecological models.

Due to all these issues and then some, the catch allowed by Canadian authorities proved much too high—and using MSY to manage the cod population proved to be a fundamentally flawed approach. Retrospective analyses showed that not 16%, but 60% of the total cod population was pulled out of the water in the final years of the Grand Banks cod industry, decimating recruitment and necessitating the fishing ban. A faulty mathematical model is thus why my mid-Atlantic fish market sells Pacific and Alaskan cod.

There is reason for optimism, though. Just a month ago, Canadian authorities announced an end to the cod fishing moratorium, 32 years after it began. By finishing fishing just in time, the cod populations were given a last chance to recover. Fishing will commence carefully this year and the yield is limited to 18,000 tons—much, much lower than the ‘MSY’ of 235,000 tons set by the Canadian fisheries minister in 1989. Hopefully, the authorities have updated their assumptions and now leave enough room to accommodate residual model error.

Now that the moratorium is lifted, will Newfoundland reinstate the codfish postage stamp?

Ecology is not my field of study—though I hope I did it justice here—but I see this story as a broader cautionary tale. As scientists attempt to understand larger and more complex systems, we end up dealing with phenomena we can never fully understand, including ecosystems but also economies and mental health. It is impossible to precisely predict outcomes of human interventions in such systems, even with tons of experience. That said, we are eager to intervene in anything we care about, so what are we to do? Some scholars argue for taking a more holistic approach, avoiding the temptation (and failure) to reduce a complex system to its constituent parts. Furthermore, it appears that complex systems can have some stable or predictable features, such as alternative equilibria and the cascades of events that disturb them. A focal point of social science going forward will be how we can leverage these approaches to optimize conditions of life for people—and fish—around the world.

This post appeared first on 3 Quarks Daily.