Ecosystem effects of bottom trawling in the Benguela current system: experimental and retrospective data analyses

Abstract

Bottom-trawling is one of the most widespread sources of physical disturbance to the continental shelf substrates throughout the world. Previous studies have shown that degradation and ecosystem changes have occurred in intensively fished areas. To date it has been difficult to attribute ecosystem changes to bottom-trawling intensities at a spatial scale that is representative of commercial fishing effort. This thesis investigates ecosystem effects of bottom-trawling on demersal fish assemblages (1990 – 2006) off Namibia and their relationship to environmental variables. The study quantifies the bottom-trawl intensity using commercial data (2000 – 2006) from the Namibian hake-directed bottom-trawl fishery for the first time. The thesis further investigates short-term effects of bottom-trawling on demersal fish assemblages in heavily and lightly trawled areas experimentally. Finally, it assesses the changes in diversity, density, and structural composition of demersal fish assemblages in the northern and southern Benguela ecosystems, using the hake biomass survey data for 2007. The major research questions are answered using various approaches. Changes in the structure and composition of the demersal fish assemblages are assessed using a variety of ecosystem indicators known to capture such changes, which may be induced by bottomtrawling. They include size-based indicators such as mean length, slope and height of the size spectra and proportions of abundances by size classes; and species-based indicators such as dominance curves, dendrograms, multidimensional scaling and various diversity indices. Indicators were inferred using directions of change of the indicator in question, in relation to the expected response of the indicator to varied bottom-trawl effects and environmental variability. Bottom-trawl intensity is assessed by mapping the average number of hours trawled in a 1’ latitude x 1’ longitude grid square and as the average number of trawls per grid square over a seven-year period, using ArcGIS software. Effort and catch per unit effort are estimated and modelled using general linear models first, and then generalized additive models or big additive models, with longitude, latitude, depth, year, seasons, vessel type, and regions as explanatory factors. Bottom-trawling intensity varied spatially off Namibia, with the average number of trawls per grid square patchily distributed. Southern Namibia was the most heavily trawled with high fishing effort (54.18 %), followed by central Namibia (25.06 %), with northern region the least trawled (20.77 %). Most of the effort was concentrated in water depths ranging 200 to 399 meters, and freezer vessels expended more effort (71.56 %) than the wetfish vessels (28.44 %). Effort varied significantly (p < 0.001) both annually and seasonally with the highest effort recorded in autumn and lowest in summer. Temporal and seasonal variations affected the catch per unit effort, and wetfish vessels accounted for higher catch per unit effort (64.11 %) than the freezer vessels (35.89 %). Significant (p < 0.05) changes in the mean catch rates, mean catch length, heights and slopes of the size spectra are observed over the years under review, and environmental variables (Benguela Niños, low oxygen waters, sea surface temperature, Angola-Benguela Front position, upwelling favorable wind anomalies, upwelling index), and M. capensis recruitment index shows minimal effects on the mean catch rate, mean catch length and heights and slopes of the size spectra, thus the other effects may be partly due to direct and indirect effects of fishing. Cluster analysis and ordination by non-metric multidimensional scaling, show changes in the demersal fish community off Namibia and identified three main assemblages temporally, in 1990–1993, 1994–2000, and 2001–2006. Dominance curves suggest that the three assemblages are all moderately disturbed. Permutational multivariate analysis of variance and a distance-based linear model suggest multi-factor combined effects of regions, depth, bottom oxygen, temperature and vessels play important roles in the observed changes in demersal fish community structure. Effects of hypoxia and changing environmental conditions (bottom oxygen and bottom temperature) are more pronounced on the northern and central continental shelf and play an important role in the distribution of demersal fish species. Diversity increases with increasing latitude (towards the south) and depth, and is minimally affected by the current levels of fishing. The generalised additive model suggests that total abundance and latitude are important predictors of diversity. Demersal fish assemblages are compared between lightly and heavily trawled areas experimentally, for the first time in Benguela waters. Heavily trawled areas are characterized by low fish diversity, high abundance of small individuals and dominance by a few species. Lightly trawled areas are less dominated, exhibited high evenness and high diversity. This empirical study unravelled the immediate effects of heavy trawling activities as a partial cause of changes in the assemblage structure. Cluster analysis and multi-dimensional scaling identify three demersal fish assemblages, namely the continental shelf and slope of northern Benguela, and southern Benguela at 19 % Bray- Curtis similarity, and their W–statistics suggest that the northern Benguela assemblages are moderately disturbed. Fish density and diversity increase with increasing latitude and depth. The separation of the two countries’ demersal assemblages, as identified by cluster analysis and multidimensional scaling, coincides with the boundary between the northern and southern Benguela systems. The changes in composition of demersal fish assemblages observed in this analysis are a reflection of both direct and indirect effects of bottom-trawling, complicated by varying environmental conditions, such as low oxygen water important predictors of diversity.