Multivariate analyses of the impact of offshore marine mining on the Benthic macrofauna off the west coast of Southern Africa

Abstract

Abstract by author:

The strategy for analysing multivariate data presented by Field et. al. (1982) was tested for its sensitivity in detecting the effects of offshore marine mining on macrobenthic communities. The technique has proven to be particularly sensitive and robust in elucidating changes in the structure of marine communities following organic pollution events. The primary aim of this study was to investigate its applicability in discerning community changes in an area exposed to physical disturbance of the seabed. Statistical testing, using analysis of similarities, reveals a highly significant difference between mined and unmined samples. Statistical testing also detects natural spatial heterogeneity across the 6 study areas. Aggregation of the data to higher taxonomic levels did not result in the loss of information, and in fact, improved the resolution of the community patterns. Multivariate analyses were therefore performed using the community data aggregated to genus-level. Hierarchical agglomerative clustering reveals two major groups of samples, the mined and the unmined samples. Within these two clusters, cluster analysis and non-metric multi-dimensional scaling distinguish between areas 1 and 2, areas 3 and 4, and areas 5 and 6. Area 2 provided a reference site where no mining is likely in the foreseeable future and area 1 was mined between the two sampling cruises, providing an indication of the changes induced by the mining process. Cluster analysis shows a shift in the grouping of samples from area 1. The unmined samples from area 1 grouped with the reference station (area 2) in the first survey. After the area was mined, the samples grouped with the mined samples from areas 3 and 4, indicating a shift in community structure. Multidimensional scaling ordination confirmed the groupings detected by cluster analysis and hence the groups of clusters can be accepted as real. Samples 3. 5 and 3. 7 from the Rockfish cruise and samples 4. 6 and 4. 9 from the Pentow Salvor cruise were conspicuous as outliers in both the cluster and ordination analyses. Geological results show that samples 3. 5 and 4. 9 also exhibit anomalously high proportions of gravel in the sediment, which possibly influences the community composition. Several species contributed to the overall dissimilarity between mined and unmined samples. The Amphipodal Ampelisca anomala and Hippomedon longimanus were reduced in abundance after mining. Furthermore, the Polychaeta Prionospio pinnata, Haploscoloplos kergttelensis and the Lumbrineris genus also appear to be sensitive to the effects of mining and were reduced in number. Organisms which showed an increase in relative abundance in mined areas included Macoma crawfordi, Nassarius vinctits, Tricolia capensis and Terebellides stroemii. There is a net increase in the proportion of Gastropoda and Bivalvia in the mined samples, which is possibly a reflection of the altered stratigraphy in mined areas. During the mining process, the fine sand component of an area is suspended in the water column and gradually disperses over a wide area. The net result is an increase in the relative proportion of the larger gravel and mud fractions. A meta-analysis of phylum-level community data was used to assess the severity of disturbance caused by marine mining vis-a-vis disturbance studies conducted on the N. E. Atlantic Shelf. The Namibian samples were consistently distinct from the samples collected off the N. E. Atlantic Shelf. It was hypothesized that this difference may reflect the anomalously oxygen-poor conditions on the continental shelf off the west coast of Southern Africa