Fish biomass estimates along estuaries: A comparison of vertical acoustic sampling at fixed stations and purse seine catches

Abstract

Two techniques, the conventional purse seine method and a novel method using vertical hydroacoustic surveys at fixed positions, were applied simultaneously in the Gambia River estuary to sample a fish population. Environmental parameters were measured simultaneously. Four surveys were performed during different hydrological seasons to reflect the main sources of variation in the environmental conditions of the estuary. A total of 153 samples resulted from these surveys. To identify the environmental conditions in which the two methods yield similar pictures of the fish population, a regression tree was constructed in which the dependent variable was the difference between the standardised values of the catches from the purse seine and the estimated biomass from the acoustic samples. A highly significant correlation was found between the two approaches using a set of data (128 samples) that excluded the samples with specific environmental conditions (downstream samples with high transparency, and upstream ones with high transparency, high depth and high temperature). We conclude that, except under these specific environmental conditions, vertical hydroacoustic samples recorded at a fixed station can serve as a proxy of the catches from a purse seine and can allow the monitoring of fish biomass in estuaries without damaging the environment.

Introduction

The Gambia River estuary, one of the last aquatic ecosystems of West Africa not yet affected by strong environmental and human disturbances, plays a considerable role in the Gambian fisheries economy (Tobey et al., 2009). Information on fish resources (distributions and biomass) is needed to assess the adequacy of fisheries management, and, more broadly, ecosystem management (Mason and Brandt, 1999; Cury et al., 2008). Fish sampling techniques have biases, and their results are therefore difficult to interpret (Robinson et al., 1996; Kubečka et al., 2009). Thus, it is necessary to compare different approaches (McClatchie et al., 2000; Coll et al., 2007), using standardised protocols, to measure variation in the amount of biomass and changes in the spatial distributions of fish populations. Comparisons in space and time can be based on this information (Yule, 2000). Fish populations in the Gambia River estuary were sampled using two methods simultaneously: a conventional purse seine method (Albaret et al., 2004) and a novel method, using vertical hydroacoustic surveys at fixed positions (Guillard et al., 2004; Boswell et al., 2010).

If it is used blindly without searching for fish aggregations, purse seine sampling provides a good picture of the fish population in a given place (Albaret et al., 2004). However, the method is not as satisfactory for predominantly bottom-dwelling species. The principal information produced by this sampling method is the total weight of fish caught in the volume sampled by the seine and the species composition. In hydroacoustic sampling (Simmonds and MacLennan, 2005), the principal metric is based on the backscattering strength of a volume or a surface (the Sv or Sa values, respectively) (MacLennan et al., 2002), i.e., the integrated acoustic energy contained in a volume of water. This metric can serve as a proxy of the fish biomass (Boswell et al., 2007). The first approach is relatively well known and conventionally used in estuarine ecosystems but needs an efficient fishing team and is time-consuming and financially costly. Recent developments in acoustic technology and signal processing, in addition to decades of experience (Fernandes et al., 2002) at sea, in freshwater habitats and in estuaries have led to increased acceptance of acoustic surveys as a reliable technique for estimating fish abundance (Simmonds and MacLennan, 2005). More recently, acoustic surveys have been used in shallow waters (Lilja et al., 2003; Matveev, 2007; Boswell et al., 2010; Becker et al., 2011). However, few studies have been conducted in tropical estuarine environments (Guillard, 1998; Krumme and Saint-Paul, 2003). Nevertheless, the use of hydroacoustic surveys will increase, particularly in protection zones such as Marine Protected Areas, because no mortality results from the use of this method.

This paper analyses the fish population estimates produced by the two sampling methods conducted simultaneously and in spatial proximity. The similarities and differences between the two protocols were analysed and linked to major environmental features using the ‘regression tree’ method. Environmental patterns corresponding to a large difference between the two approaches were identified. A linear model is proposed to model the relationship between catches and hydroacoustic data in the Gambia estuary, and a sampling strategy to monitor such ecosystems is suggested.