ACOUSTIC REMOTE-SENSING OF BENTHIC ACTIVITY - A STATISTICAL APPROACH

Acoustic backscatter at 40 kHz was recorded from a circular area of si lt 75 m in radius every 0.1 d at a water depth of 91 m off the coast o f northern California for 49 d in winter 1988-1989. Data were resolved into pixels that extended 7.5 m radially and 5 degrees azimuthally. W e analyzed the relatively small change in backscatter intensity and ph ase from a given pixel. In a time-series analysis, this change in back scatter was decomposed into three components putatively due to large, rare, transient nekton and epibenthos; more abundant benthopelagic org anisms that caused volume reverberation; and more sedentary benthos. B enthic change in backscatter was presumed due to movement of animals a nd other biogenic modifications of sound speed in the sediment-water m edium. Patterns of benthic activity and volume backscatter near the se abed changed abruptly twice, with the middle interval showing highest levels. During this middle interval, winter storms disturbed surficial sediments, offshore transport occurred near the seabed, and the benth ic component was dominated by activity perfectly in phase with and lin early proportional to median, local, downwelling irradiance. We suspec t that this circadian pattern was due to the burrowing urchin Brisaste r latifrons that is known from previous studies at this site to emerge on the sediment surface at night when food resources are available th ere and to reburrow during the lighted part of the day. Extensive corr elative physical data collected at the site as part of interdisciplina ry studies (CODE and STRESS) allowed us to exclude other physical forc ings as likely explanations for diurnal change in backscatter. In a sp atial analysis, the transient component was removed explicitly pixel b y pixel, and volume reverberation was removed implicitly to leave a be nthic remainder that could be analyzed for spatial variation. Benthic activity showed distinct patchiness with coherence scales of roughly 5 -10 m and negative spatial autocorrelation in activity levels near the largest spatial lag available (similar to 140 m). Patches did not mov e appreciably during the study.