Using acoustics to investigate changes in efficiency of a sandeel dredge

Abstract

Acoustics biomass measurements and a sandeel dredge tows were used to sample sandeels (Ammodytes marinus), during six surveys conducted on the Dogger Bank, North Sea between 2000 and 2002. Presented here, is an approach that uses the acoustically determined biomass to estimate the apparent efficiency of the dredge. Simulations were used to quantify the main sources of error in the estimate of acoustic biomass. The estimated dredge efficiencies varied between surveys but were comparable in 2000 and 2001. The dredge appeared to have performed poorly in June 2002. In principle, the use of ‘acoustically derived’ estimates of dredge efficiency will make it possible to calculate the biomass of sandeels at times when they are not readily available to acoustics (e.g. in April and November). However, before this can be achieved with confidence it is necessary to determine what proportion of sandeels may remain buried in the sediment during daylight hours and better define the target strength of sandeels at 120 kHz.

Introduction

With the exception of spawning in December and January (Gauld and Hutcheon, 1990), observations on sandeels (Ammodytes marinus) in the North Sea suggest they rarely emerge from being buried in the sea-bed during the September to March over-wintering period (Cameron, 1958, Macer, 1966, Wright and Bailey, 1993). Emergence in the spring corresponds closely with the increasing abundance of zooplankton and is triggered by temperature, photoperiod and food availability (Winslade, 1971). During peak occurrence in May–July, sandeels exhibit strong diel behaviour. At night, they bury in sandy sediments, whilst during daylight hours they form shoals in the water column and forage (Winslade, 1971, Winslade, 1974, Robards et al., 1999) over the shallow (<150 m), turbulent, sandy areas they inhabit (Macer, 1966, Reay, 1970, Wright and Bailey, 1993, Wright et al., 2000, Freeman et al., 2004). Preliminary feeding experiments (Cude, 2002 unpublished data) show that sandeels feeding on the Dogger bank, take 9–13 h to digest 50 wt.% of their stomach contents, thus lending the support to the belief that sandeels emerge from the sediment each day to feed in the water column. Such dynamic behaviour means that sandeels vary in their availability to different sampling gears, and consequently, more than one approach is required to effectively assess their abundance and distribution at different times of the year, and even at different times of day. Two approaches that can be put to use are fishing and acoustics. During April, May, June and November, we surveyed sandeels on the Dogger Bank, North Sea, using a sandeel dredge (modified scallop dredge) at night and acoustics during daytime. The purpose of this short communication is to present preliminary results of an approach that uses the acoustic data to provide insight to the efficiency of the sandeel dredge.

Continuous acoustic surveys are characterised by their ability to provide rapid coverage of survey areas with undisturbed observations (e.g. Fernandes et al., 2002). Acoustics can be used to estimate absolute abundance, provided that the acoustic target can be positively identified and a species-specific value for target strength is available for a given acoustic operating frequency (MacLennan and Simmonds, 1992, MacLennan et al., 2002). However, the uncertainty associated with absolute abundance estimates is often large, since many factors contribute to the variability in acoustic assessments (MacLennan and Simmonds, 1992, Simmonds et al., 1992, Arnold, 1996, ICES, 1997, Fréon and Misund, 1999, Rose and Gauthier, 2000). Because of such large uncertainties, acoustic abundance estimates are in practice more appropriately considered as relative abundances that may, in some circumstances, come close to true values. With particular regard to sandeels, acoustics can only provide a usefully close estimate of abundance if all the stock in a survey area is present in the water column during the survey period. In contrast, dredge sampling offers the opportunity to confirm the identity of the species (and the lengths and weights of individuals) and can be used to catch sandeels at times when they are buried in the sediment. However, since the practicalities of available survey time generally dictate that only a limited number of dredges can be taken, overall coverage of a survey area is poorer than that provided by acoustics. The dredge catch rate is a relative index of abundance. Only if we know what the dredge catch represents as a proportion of the fish available to be caught, can we derive an estimate of actual abundance. This proportion is known as the dredge efficiency and is a function of fish behaviour and selectivity of the sampling gear.

On the basis of the knowledge of sandeel behaviour documented above, we assume herein that the overwhelming majority of sandeels in our survey area are present in the water column during the daytime in May and June, and hence are completely available to acoustic gear. Using standard methods, we calculate estimates of acoustic biomass and use this to derive estimates of apparent dredge efficiency for our sandeel dredge. Note that we choose to convert sandeel numbers to biomass since this is the unit of assessment upon which fishing quotas (tonnage) are based. In addition, we performed Monte Carlo style simulations of the acoustic biomass calculations to explicitly consider the impacts of major sources of variability on our estimates of apparent dredge efficiency. Estimates of the ‘acoustically derived’ apparent dredge efficiency are tentatively used to provide biomass estimates from dredge surveys performed at times when sandeels are not fully available to acoustic sampling. Key requirements for improving the method are discussed.