Fisheries Have a Data Infrastructure Problem
Dilapidated data infrastructure, both in the United States and globally, stands as one of the biggest obstacles to research on fishery management and its impacts. Upgrading, standardizing, and integrating those systems must become a priority in every country’s fishery management policies.
Fisheries management around the world relies heavily on data, whether to estimate fish populations, track changes in their range due to global warming, or assess the impacts of a policy on stocks and fishing communities. Without good data, even well-intentioned policies can inflict unnecessary economic damage, fail to preserve fish stocks, or both.
The good news is that much of the data that government agencies and researchers need to craft sustainable fishing policies is already being collected. Furthermore, new and better data sources are coming online, thanks to high-tech equipment on boats, sensors such as Satellite-AIS, and online seafood auctions. We’re living in a golden age of data, and fisheries should be no exception.
There is a problem, however. The systems in place for storing, tracking, and querying the data are cumbersome and outdated. Dilapidated data infrastructure, both in the United States and globally, stands as one of the biggest obstacles to research on fishery management and its impacts. Upgrading, standardizing, and integrating those systems must become a priority in every country’s fishery management policies.
For example, population assessments are done by stock—that is, regional subsets of fish species—but most management happens at the level of the fishery, which can include multiple species and stocks. By contrast, most fish prices are determined at the species or product level, and foreign trade data are broken down by product types and species groupings that change over time. In order to understand the bigger picture of how management affects stock status, or how the market affects fishing behavior, we must piece the data together across these multiple domains. To make matters worse, each data source utilizes different naming conventions and has coverage over varying geographic regions and time scales. Linking these datasets requires an extremely nuanced understanding of regional management idiosyncrasies, reporting caveats, and where to find documentation and keyfiles. A standardized approach would greatly facilitate the process of providing accurate and timely insights into how our fishery policies are working. At a minimum, there should be an identification number for each stock, species, and fishery, and yearly keyfiles that match the stocks to species and fisheries they belong to.
Of course, data can only be cleaned and merged when they are accessible in the first place, and much of the information critical for evaluating policy measures in terms of both their biological and socioeconomic outcomes simply does not exist in a readily usable form. In the United States, for instance, changes to policy are often embedded within PDFs scattered throughout regional offices’ websites and can only be searched systematically through the Federal Register. This includes even the most basic metrics, such as trip/day catch limits and yearly total allowable catches by stock. Even if researchers write a web-scraping program to collect relevant files, they must then meticulously comb through each document manually, piecing together the chronology and context for each number. Errors and omissions are inevitable. For variables that are not catalogued in the Federal Register, the process is even more onerous, involving collection and review of periodic reports, memos, and meeting minutes. In still other cases, such as port-, sub-annual-, and other finer-scale data, special requests may be needed, and access is often limited to use for specific funded projects or partnerships.
These might seem like isolated issues, but the upshot is that they dramatically constrain the research questions we can answer. The authors of this post study the efficacy of fishing policies and how they impact fish populations and fishing behavior. Each of us estimates that we spend 50 to 80 percent of our research time wrangling data instead of analyzing it. For those without backgrounds in data management or the resources to hire specialized assistants, these obstacles can be insurmountable, discouraging quantitative research in this field altogether.
It does not have to be this way. An example of the difference made by better data management comes from Norway, whose Directorate of Fisheries and regional sales organizations maintain clean and consolidated datasets of landings, vessels, owners, and seafood buyers, all broken down by standardized species, product, and gear codes. This part of why so much of the empirical literature on fisheries management centers on data-rich countries, and why they are better positioned to track seafood sourcing, climate impacts on fisheries, community sustainability, and other important current issues.
Every country that has a national fishing policy should add to it both a mandate and funding for data management and accessibility. This includes the United States, where the Magnuson-Stevens Fishery Conservation and Management Act is overdue for reauthorization. For countries that want to manage their fisheries effectively, this is an investment that would pay dividends far beyond the cost of buying new computers, software, and expertise. It could make a lasting difference in the global effort to secure one of our most important natural resources for future generations, along with the food and jobs it supplies and the communities and cultures that rely on it.