Studies of environmental impact assessment (EIA) base their final conclusions and/or decisions, typically, on available information that has been extracted from existing environmental studies. To solve this, conclusions and desicisons taken after an EIA can be re-stated in the form of hypothyeses, and the appropriate monitoring programs are put in place to test them. These monitoring programs should be designed to detect the potential impacts that a given development might cause. An important aspect to be developed to reliably detect an anthropogenic impact is the development of an effective and precise analytical framework able to differentiate between human induced changes and the natural spatio-temporal variability shown by most populations. Among all the possibilities, "Beyond-BACI designs" allow addressing unequivocally these issues, but until the beginning of the millennium their use was limited to the analysis of single variables (e.g. abundance of a single species or any measure of biodiversity). In this paper we describe an asymmetrical sampling design that will allow the use of an analytical procedure able to unambiguously detect if there is any effect of a FSO on the natural conditions of the surrounding marine areas and assemblages living in it, more specifically, pelagic and soft-bottom benthic assemblages. The FSO will be located in the Gulf of Paria, Venezuela.

The proposed sampling design consists of one potentially impacted location (1 km radius area around the exact location of the FSO) and three reference locations (same area as the impacted one) sampled at two spatial scales: three sites within each location (separated by 2 −3 km) and replicate units within each site (separated by 100´s of meters). This spatial layout of sampling sites will be sampled three times before and three after the placement of the FSO.

In this study, the comparison of the different variance components for the assessment of potential impacts will be done using a recently developed multivariate method: dissimilarity-based permutational MANOVA, for the analysis of multi-species data matrices in response to complex sampling designs such as the one described above.

This type of experimental design and associated analyses are complicated and require a considerable increment of replication effort beyond to what we are accustomed to, involving costs above what has traditionally been the case. This expenditure, however, is warranted because, among other reasons that will be discussed in the paper, it does protect against unfounded claims of impacts, and realistically determines the level of a predicted impact so that corrective or mitigating options can be developed and implemented.

You can access this article if you purchase or spend a download.