Fishing about and about fishing
COMPUTERS IN FISHERIES RESEARCH: FRIEND OR FOE?
--------------------------------------------------------------------------
To err is human; to really foul things up needs a computer…
--------------------------------------------------------------------------
Once upon a time, when I was young and computers had yet to appear on our scene, fishery scientists used to sail on board research and commercial fishing boats, sample and identify fish, examine their food, take water, plankton and benthos samples, and measure ambient temperatures. Then they were analyzing their samples in labs and tried to synthesize the lot into a meaningful pattern, somehow in a way similar to physicians diagnosing diseases. All this took the largest part of their working time.
Then came the computers and, sitting at them is now taking the largest part of fishery scientists’ working time, and the jury is still out on the question of whether fisheries knowledge and management are now better off. This is why the 2nd and updated edition of the book “Computers in Fisheries Research” (Megrey B.A. and E.Moksness (Eds.) 2009. (Springer Verlag, Heidelberg, German; www.springer.com), 421 p., drew my attention.
The first chapter introduces the technical advances and the formidable operational capabilities of contemporary computers. We shall remember that, apart from their general facilities, such as storing information, data processing, searching and communicating via Internet, (see Chapter 2), computers’ specific job in fishery science is to operate stock assessment models.
Otherwise, the book is rather about computer models (or algorithms), which after being fed with data are capable to process them and produce results representing some fishery-related reality. Computers themselves (hardware + their operating systems) are just the machines that do the technical job, the more advanced they are the faster they do it, but they don’t “foul the things up” – models do.
.
In the 3rd chapter, entitled “Guide to some Computerized Artificial Intelligence Methods”, S.B.Saila gets down to brass tacks. He points out some critical aspects of the data and information that, I think, should be remembered by every fisheries scientist as they start feeding with them their computer models. Fisheries data, however detailed, writes Saila “can be disorganized, ambiguous, incomplete, imprecise” and inadequate to meet the requirements of the models. He criticizes the commitment of fishery science to particular methodology, and quotes Professor Lofti Zadeh, the proponent of “fuzzy logic”, (see this Column of April, 2006: “FAO, tuna and fuzzy logic”), who pointed out that the more complex is a system the less is our ability to make precise and significant statements about its behaviour.
Saila reviews several recent methods of data processing, so far little applied in fishery science, designed for complex models fed with deficient data. Some of them are interesting, because without pretense of precise results, they offer a reasonable picture of the reality, or involve past information to reinforce recent data.
Marta Coll et al. discuss ecosystem modelling in Chapter 8. The models involved are complex and “have large inherent uncertainties” and “poor in predictive power and far from robust”. While “fisheries management is still mainly approached from the level of single-species management”, it has not yet been found how to combine fisheries management with ecosystem considerations. In somma, new and more sophisticated models are being developed, but they won’t be more useful as long as their variables are not fed with adequate data.
Until I’ve read Mark Maunder et al. Chapter 11 on computers in fisheries population dynamics, I had no idea that so many models, existing and under development, are available for stock assessments. But, I think that they all are still plagued with process errors, uncertainties and other maladies. If half of the monies and efforts spent on working out more and more sophisticated computer models would’ve been directed on improving and developing collection and laboratory processing of fishery and environmental data, sophisticated models might lead stock assessment closer to a real science.
But, it’s C.J.Walters, who in the last chapter (Computers and the Future of Fisheries) expresses both, well argued skepticism regarding the present status of the computerized basis for fishery management and hopeful optimism about its future development into something “that really deserves to be called science”.
This, in my opinion, requires that the models comprise all the major variables that affect stock size, which means that all that’s today covered by the term natural mortality is split into its main elements, such as predation, food scarcity, major hydrographic (temperature, salinity) shifts, and pollution (permanent or sporadic). All this requires sea-borne and laboratory research and collection, verification and quantification of the information and data needed for such a model. Where quantification is impossible, or can only be approximated, techniques that can handle ranges (from… to…) rather than precise figures must be involved. Obviously, truly scientific fish-stock assessment models cannot produce precise figures. Their products should not be presented in precise figures, but rather in the form of uncertainty estimates, as illustrated in Chapter 11. As Walters concludes: “…no complex computer modeling or data-gathering system can substitute for the fundamental scientific requirement of trying to challenge and reject our models through real field experience and experiments”.
Throughout the book one feels authors’ prevailing optimism about the future of the application of the existing and new models to population dynamics, but little attention to the supply of good data and information. Unfortunately, even the best and fastest computers and the most sophisticated models cannot improve lacking data. They can only process to bad or to worse what is fed into them. The saying “garbage in – garbage out” stands.
Within the frame of this page I was unable to give justice to all 13 chapters of the book. Whatever is the present contribution of computer models to fishery management; this book offers an excellent insight into the world of computers and modelling for fisheries and can be recommended not only to “computer freaks” among fishery scientists, but all, scientists, students and managers, who wish to broaden their knowledge associated with using computer models for stock assessment and management.