SAMPLING ERRORS IN ANCHOVY EGG ABUNDANCE ESTIMATES USING THE PAIROVETNET

A replicate haul experiment was carried out using the vertically towed PAIROVET net in an area of high abundance of anchovy eggs in the Bay of Biscay. The main aim of the experiment was to compare the performan ce of ships of different size working in the same area. In addition, w e tested the performance of the PAIROVET net and the deployment method ology used to sample anchovy eggs for biomass assessment purposes. We concluded that under the experimental conditions experienced in this c ruise, different ships yield similar performances of the vertically to wed PAIROVET net if the methodology is kept similar. The ANOVA suggest s that the systematic central sampling (SCS) scheme currently in use f or the Bay of Biscay anchovy biomass estimations (one station out at 3 miles) adequately represents the surrounding area along the transect line. The among-station component of variance within the unit sampling area of the current SCS turned out to be comparable and, in some case s, smaller than the within-station variance. Consequently, the spatial design of the SCS adopted during routine surveys appears to be a reli able strategy for sampling anchovy eggs in areas of high egg abundance , as in this experiment. In addition, it ensures fairly independent da ta units. In the area of the experiment, the most disperse egg stages (like 2- and 3-day-old eggs) show a larger within station component of variance than the spatial component of variance within the rectangles of the current SCS. This suggests that for these stages, the sampling can still be improved cheaply by increasing the precision of egg abun dance estimates per station with larger sample sizes, without changing the spatial sampling scheme or the number of stations. However, the b enefits of this strategy will hardly affect the precision of the 1-day -old egg estimates because of their patchy distribution.