BREMERTON - Oxygen levels in Hood Canal have been plummeting this year, raising concerns about the potential for a massive fish kill.

Jan Newton, a University of Washington oceanographer, says it's hard to predict how high that potential is.

In 2004, oxygen levels in the narrow channel between the Olympic and Kitsap peninsulas reached a record low, but no wide swaths of fish suffocated. In 2002 and 2003, there was more oxygen in the water but fish died by the thousands.

"That's the dynamic we're trying to understand," said Newton, who is leading an intensive, multimillion-dollar research effort into the cause of the low-oxygen problem.

Other researchers also are keeping watch on Hood Canal, and few are finding encouraging signs.

Greg Bargmann of the Washington Department of Fish and Wildlife noted that rockfish populations at the southern end of the canal were struggling before the 2003 fish kill. And recent surveys have shown that rockfish have failed to rebound and may be declining further, he said.

Growth of eelgrass, which is important to juvenile salmon and other sea creatures, has declined for four consecutive years, said Pete Dowty of the Washington Department of Natural Resources. Of five regions his group has studied throughout Puget Sound, only Hood Canal is showing that kind of decline, he said.

Some inactive geoduck beds in Hood Canal also have shown a troubling decline in density, according to state shellfish biologists.

While many species are suffering from low-oxygen problems, fish get the most attention. That may be due to their economic value or the fact that dead fish often float to the surface.

Fortunately, fish can swim away from some problems, but that complicates the effort to predict fish kills.

If oxygen levels are exceedingly low in deep water, for example, fish can escape by swimming closer to the surface. In 2002 and 2003, something happened that brought the low-oxygen water right up to the surface, causing the fish to die.

One theory is that winds out of the south trigger fish kills by pushing surface waters to the north. At the dead end in southern Hood Canal, displacement of surface water may bring dangerous low-oxygen water up from the depths.

"It looks like 06 will be a bad year for oxygen," Newton said, "and one of the things we will be watching is the local winds."

Over the past two years, researchers have installed four buoys loaded with instruments to monitor water conditions at all depths, 24 hours a day. That information is added to results of samples taken by volunteers and professionals. The buoys also monitor the growth and movement of plankton, and some of the findings are surprising, Newton said.

In the presence of sunlight, plankton feed on available nitrogen, which comes from many natural and man-made sources. When the plankton die, they sink and decompose, using up oxygen other sea creatures need.

Newton said researchers always understood that the shape of Hood Canal - long, deep and enclosed at one end - constrains mixing of the water layers. As a result, plankton blooms start earlier and last longer than in the main part of Puget Sound.

"What's new," Newton said, "is how early they might start and how late they might go."

In most areas of Puget Sound, the plankton season generally lasts from May to October. In Hood Canal, plankton blooms may stretch from February into the fall, then come back strong as late as December.