Saturday, February 23, 2008

Trawling for Science

My trip to sea last week gave me a chance to see the methods used for the International Bottom Trawl Survey, a huge effort by all the countries that fish the North Sea to collect data about the fish populations that plays a significant part in ICES stock assessments and the resulting scientific recommendations about yearly quota allocations. Although a large amount of information about fish stocks also comes from sampling of commercial fish landings and the data reported by the fishing industry, this had generally been considered insufficient to assess the state of the fish population – first because scientists are also interested in the young fish that are not yet large enough to have become part of the commercial stock, and second because landings information provided by the fishing industry doesn’t always reflect the amount of fish actually taken out of the ocean, either because fish are caught but then discarded at sea (a legal practice in the EU) or are landed illegally. As a way of supplementing the data collected from the fish landed by the commercial fishing industry, most nations have developed fishery-independent surveys to monitor their fish stocks. Unlike in a country like Iceland (where I saw my first bottom trawl survey this summer) that independently controls its own fish stocks, in the North Sea, fish stocks are shared by a large number of countries and thus the responsibility for monitoring the fish stocks is also shared by a large number of countries.

There is a relatively long history of international cooperation to conduct surveys monitoring North Sea fish stocks, stretching back to 1967, when Scotland, England, Germany and the Netherlands began a joint spring survey of the population of juvenile herring to help predict future recruitment to the commercial herring fishery. They soon realized that they were collecting information on a number of other important species in addition to herring, including both cod and haddock, and by 1976 the survey was officially broadened from “The International Young Herring Surveys” to “The International Young Fish Surveys.” As the surveys gained prominence in official stock assessments and quota-based management, more nations began to participate in the surveys and also to develop their own national surveys generally conducted in the late summer or early fall to fill in the gap in annual data left by the international young fish survey. In 1990, ICES (the coordinating scientific body that helps integrate fisheries science throughout the North Atlantic, and in Europe in particular) combined these surveys into the North Sea International Bottom Trawl Survey (IBTS) conducted in February and August. Scotland participates in the North Sea IBTS, and in a similarly-coordinated bottom trawl survey on the west coast of Scotland in March and November, all conducted from the main Scottish research vessel, the FRV Scotia.

This is the Scotia in Lerwick the day I joined her for the final week of Scotland’s portion of the February IBTS in the North Sea.

Although standardization of trawling gear and methods was phased on only gradually beginning in 1977, today all of the nations participating in the IBTS use the same net and fishing methods, ideally producing standardized results. The North Sea is divided into “ICES Statistical Rectangles” 30 nautical miles on a side,* each of which is sampled twice by vessels from different countries, which is meant to eliminate the statistical effects of differences between the fishing power of different countries’ research vessels. Since the results of the survey play such a significant role in ICES’s assessment of the fish stocks and subsequent recommendations to the European Commission, the methods used in the survey have been heavily debated and often criticized. The standard IBTS method is to trawl for 30 minutes at 4 knots using a net called the GOV (Grande Overture Vertical) trawl designed by the Institut des Peches Maritimes and selected by ICES as the standard net for bottom trawl surveys in 1977. The net uses 120 millimeter meshes at the opening of the net but has very small (20 millimeter) meshes in the cod end where the catch collects so as to catch juvenile fish that have not yet entered the commercially-targeted fish stock.

The opening of the net showing the wide meshes and the ground gear:

The cod end of the net, with much smaller meshes:

To account for differences in the exact size of the net’s opening and thus the volume swept in each haul, acoustic sensors are attached to the net providing real-time data on the opening of the trawl doors and wings throughout the survey in addition to accurately recording the trawl speed and the exact time during which the net is actively trawling on the seabed. The crew and scientists on deck attach the bright orange Scanmar acoustic units just before setting the net at the beginning of each trawl and remove them at the end of the trawl as the net is hauled back in.

The catch from each haul is then emptied into a laboratory known as the “fish house” where the scientific crew sorts and measures the catch. Everything from the fish to the hermit crabs to the squid are sorted by species and counted and measured and the commercially important fish species – sprat, herring, mackerel, cod, haddock, whiting, Norway pout and saithe – are further sampled to determine weight, sexual maturity, and age, determined by removing an ear bone called the otolith which forms annual rings and can be read to determine age. It’s an incredible amount of data to collect, and all onto a series of multicolored forms (a yellow version for haddock and whiting, an orange one for cod, etc. allowing you to quickly find the one you’re looking for without rummaging through the stack) filled in for each trawl.

Here’s the fish house empty of people but prepared for a catch to come in.

Scientists stand on each side of each of the fish bins fed by the conveyor belt, first sorting the fish by species into the orange baskets on the floor and for less numerous species into the smaller bins stacked at the end of the table. As each crew finishes sorting their portion of the catch, the single-species baskets are then combined, weighed (information for a master white sheet), and given to smaller teams to count and sample further (filling in the multicolored sheets) in what was described to me as an organized chaos. It seemed particularly chaotic while I was there since I was the third visitor in the fish house in addition to the usual six scientific staff, once I got a sense of what was going on I could see that everyone had a sense of what needed to be done and took on whatever task was needed to keep the system going smoothly.

Although there were already plenty of hands working in the fish house and I was by far the least knowledgeable or experienced in fisheries biology, I was able to help sort the fish by species and got to know a few new species I hadn’t seen before including mackerel (which I hadn’t actually seen up close before but are a really gorgeous blue with black stripes), Norway pout (tricky to tell from a similar species called poor cod), a number of flatfishes including saithe (easy to tell from the other flatfishes by its bright orange spots, which I am embarrassed to admit I had thought were artificial when I first saw them in an aquarium), and my first live monkfish. There were also lots and lots of haddock and whiting, which I hadn’t seen much of during the shrimp survey in Iceland, and relatively few cod – though the cod we did catch were often very big, one nearly a meter long and over nine kilos.

Here’s Keith weighing one of the particularly large cod:

When I wasn’t sorting fish, I mostly helped by recording data for the measurements of the demersal species – cod, haddock and whiting. In addition to measuring all the fish, one fish from each length (by centimeter) from each trawl was further sampled by weighing the fish, opening its belly to determine sexual maturity, and removing the otolith, providing the information to assess the age distribution and contributions to the spawning stock within the overall fish stock.

Here’s Craig cutting open one of many many haddock he sampled (so fast he could gut a fish and determine its sexual stage faster than I could write it down):

Ken looks at a cod’s guts to determine its sexual maturity:

And Catherine records cod information on the otolith envelopes, the same job I did most of the time:

In some ways the process of writing everything on paper forms before entering it into the official database seemed old-fashioned, particularly after having seen the process in Iceland, where everything is entered directly into an electronic database in the scales that is then uploaded to their database. But the IBTS process is much more people-intensive (with six or more people working at once rather than three) and even more of the analysis is conducted aboard the ship, since one of the scientists also read the ages of all the demersal fish species during the trip, so it might be too complicated to put it all in one centralized database from the start rather than having a paper backup. Plus, if it ain’t broke, don’t fix it – and this system seems to work pretty well. Any sampling process that gathers so much data in such a short time is liable to have some gaps, and I did hear from one of the other visitors on the ship who works with IBTS data to model haddock populations that she has indeed come across some inexplicable problems with the data either from information that was improperly entered in the system or recorded inaccurately, but it seems that it works as well as I imagine anything could in this type of high-intensity sampling environment.

At the end of the survey, the data from all of the ships collecting data for the IBTS is worked up and the results are compiled at an ICES working group meeting to produce a picture of fish distributions in the North Sea, something like this one for cod.

The difficulty, of course, is in figuring out what exactly a figure like this means about the actual fish in the sea - translating the statistics generated from the IBTS into an actual understanding of fish populations. This is a tricky process that requires making assumptions not only about the data generated by the IBTS but also about commercial fish catches and natural mortality of fish in the environment. Although a lot of criticisms are leveled at the science, including one accusation from a fisherman I met in Denmark who was convinced that the scientists at ICES were in collusion to only produce results that said what the politicians in Brussels want them to say, I think the real issue lies in the assumption of what can be concluded from the data, the other assumptions that scientists have to make even without accurate data, and how the scientific assessments are then used to make fisheries management policies.

One of the major criticisms leveled at bottom trawl surveys, both the surveys done in Iceland and the IBTS survey, is that the gear used is out of date and very different from the gear used by commercial fishermen. This is, of course, entirely true – no fisherman could stay competitive using the same net design for thirty years. This means that what the IBTS finds is often very different from what the fishermen are seeing in their own catches, and not just because the IBTS uses smaller meshes but also because of the design of the GOV trawl. Studies of the fish escaping from the GOV trawl indicate that up to 75% of the cod swept by the net escape through the ground gear. At a first glance, this seems an obvious problem – the scientific survey is not measuring everything that is actually out there to be caught. However, this does not necessarily lead to flawed results, because the IBTS does not claim to accurately measure everything in the ocean, but rather to assess trends over time in the fluctuations of fish stocks. So, if the IBTS finds more fish this year than last year, they can conclude that the fish stocks are improving – an assessment that requires a time series using the same methods over a large time span to allow for accurate comparisons between past and present fish stocks.

It is possible, however, (albeit with considerable expense) to trial a new trawl net along with the old GOV trawl to determine how two sets of methods fit together and allow a continuation of the time series even with an update in the methods. This has been considered as a possibility within the scientific community partly as a way of more closely aligning the IBTS methods with current commercial fishing methods but also, more importantly, because the GOV net is not easy to work with because it has a tendency to tear very easily. Our first day out of Lerwick, we went to a particularly rough ground where the lead scientist says the trawl had torn nearly every year and had to be particularly careful to avoid a tear in the net – both getting advice from skippers in Lerwick and running an acoustic scan over the ground before actually trawling. A few days after successfully avoiding the expected tear, though, the net tore on the second trawl of the morning right as the net was shot, meaning that we couldn’t use that haul and then it took a few hours for the crew to repair the tear, all the way from the wing to the cod end.

Repairing the net by hand, basically tying in new meshes to replace those that were ripped:

In addition to the inconvenience of having to spend a few hours repairing the net rather than working, the scientist overseeing all the operations on deck also pointed out that this increases the variability between different countries performing the survey because the repaired net will never be exactly to specifications. And since the GOV tears so frequently, it means most of the IBTS is conducted by trawls not quite to official standards.

Even more important than a net not being quite to specifications, I think an even more important question is whether it is actually valid to assume that trawling the same area each year with the same gear will catch the same proportion of the total fish population in the area. A number of fishermen have observed that the fish’s behavior has changed over time, probably the result of both environmental changes (especially changes in ocean temperature) and changes in behavior as the overall fish population fluctuates and thus the density of fish in the area changes. A cooperative study between the Danish fisheries research institute in Hirtshals and local fishermen looked at the populations of cod on rocky ground where the IBTS can’t survey and found much higher fish populations than on the smoother ground where trawling is possible, indicating that a trawl survey is inherently biased by the type of area it is able to include in the survey. If, as fishermen have observed and this study’s preliminary results suggest, cod prefer the rocky ground, a decrease in population might seem much more drastic in the survey results because the largest signs of a decline in population would probably be found on the least favorable grounds – the ones surveyed – even if the population is maintained on the rougher ground.

Of course, there isn’t enough data available to tell whether this is truly a problem or not, but it does indicate what I see as a larger issue with any attempts to determine fish populations: we just don’t know enough to say for sure what’s actually going on in the sea. And it’s not that scientists don’t realize this is an issue – most fisheries scientists I’ve met have openly told me that what they do is like taking a “stab in the dark,” because no matter how much information you collect and how precise your methods, it is impossible to completely describe a complex ecological system. From seeing the IBTS for myself, I am certainly impressed by the amount of data collected and the amount of work by nations all around the North Sea that goes into conducting the survey – and that just one piece of the scientific data that goes into stock estimates along with samplings of commercial landings, observer programs aboard commercial fishing vessels, and a slew of other studies designed to better confine the assumptions that go into making the final stock estimates. No matter how good the science, however, I’ve been thinking a lot about the limitations of predictive scientific models and thus the limitations of a management system relying on predictive scientific stock assessments. It’s too big an issue for me to tackle right now, but my thoughts coming out of spending a week observing the IBTS are both that there is an immense amount of data being collected on the North Sea fish stocks – it is likely the best-studied piece of ocean anywhere in the world – but I still don’t think it’s enough to tell us what exactly is going on for the fish in the North Sea.

*For those of you confused as to why these are rectangles rather than squares, I am assuming that this is taking into account the fact that a nautical mile is not a fixed length but changes at different latitudes.

No comments: