Feature - Plenty more fish in the sea?
Once thought to be an endless source of bounty, the oceans' supply of fish has plummeted dramatically. The biggest cause of this deterioration - bigger than pollution - is overfishing.
Rampant over-fishing around the globe has caused some of our favorite species to hover near collapse. There has been a 90 percent drop in the number of tuna, salmon, cod, swordfish, rays, pollack and haddock since the 1950s, wrote Jeremy Jackson, director of the Scripps Center for Marine Biodiversity and Conservation at the University of California, San Diego, in a paper published last year in the Proceedings of the National Academy of Sciences.
The effects of over-fishing are being exacerbated by global climate change, which changes the distribution of species and the location of biodiversity hotspots. Under these conditions, good fisheries management is crucial.
But how can stocks be controlled if we don't know their location and numbers? A new grid-based tool, AquaMaps, looks like it could provide the answer. "In fisheries, we have problems when we try to do spatial analysis with data," says Anton Ellenbroek of the UN's Food and Agriculture Organization, because "there are no clear boundaries in the ocean."
The nature of the data compounds the problem. Large-scale, marine data sets frequently come with built-in biases as they tend to concentrate in certain areas, such as along the continental shelf. Furthermore, some data is over a century old (a time period over which many species will have been taxonomically reshuffled).
AquaMaps addresses the difficulties of modeling the global distribution of marine species.
Since 2004, a collaboration of biologists and modelers from academic and international non-governmental organizations has been developing AquaMaps. The maps give the likelihood of finding a certain species in an area. Relatively speaking, their resolution is high: each pixel represents an area of just 55 by 55 kilometers (about 34 x 34 miles). The novelty and inherent strength of AquaMaps lies in its combination of survey data (where a species has been found) with environmental data and habitat profiles (where a species should be found).
Generating a map is a computationally intense task; to draw one multi-species map requires 125 million computations. To help manage and process their data, the AquaMaps consortium has recently teamed up with D4Science, who will provide a virtual research environment that will use the computing infrastructure of the Enabling Grids for EsciencE (EGEE) project.
From days to seconds
The first phase of this work, which was presented at the EGEE User Forum in Catania this March, was the creation of a broad basis of maps, charting the distribution of marine species, families, orders, classes and phyla. The improvement is significant: Where a query used to take three days to be returned, a map is now available within seconds. "This will make our reporting much easier," says Ellenbroek.
"The crux of the issue for AquaMaps," says Pedro Andrade, a software engineer working with D4Science and EGEE, "is that their data requirements limit their capabilities. Currently they can search distribution maps for about 3,500 species. With enough support, this could go up to 50,000. We can give them a distributed-computing environment which could drastically reduce the time needed for searching and processing."
AquaMaps also needs to manage constantly evolving data that is located around the world.
Northern bluefin tuna might be found to prefer a slightly different surface temperature, or a new
The grid is the perfect solution, Pedro Andrade said: "A grid-computing environment, which easily allows for data stored at one outpost to be accessed and used within an entire organization, fits its (AquaMaps') information requirements very well."
The next phase of AquaMaps will see users creating their own maps according to their desired
-Danielle Venton, EGEE. This story also appeared in the June issue of the OMII-UK newsletter