Proceedings of the 2nd International Conference on Environmental Interactions of Marine Renewable Energy Technologies (EIMR2014), 28 April – 02 May 2014, Stornoway, Isle of Lewis, Outer Hebrides, Scotland. www.eimr.org -1- EIMR2014-596 Modelling impact assessment in renewables development areas using the new R package, MRSea v0.1.1 LAS Scott-Hayward 1 , ML Mackenzie, CS Oedekoven Centre for Research into Ecological and Environmental Modelling, University of St Andrews, Scotland CG Walker Department of Engineering Science, University of Auckland, New Zealand. ABSTRACT For both developers and government licensing organisations it is important to have the ability to quantify spatially explicit change in the density and/or distribution of animals in and around marine renewables sites and, in particular, to identify if change occurs near renewables devices [1]. The publicly available MRSea package (Marine Renewable Strategic environmental assessment) [2] has recently been developed for analysing data collected for assessing potential impacts of renewable developments on marine wildlife, although the methods contained in this package have wide applicability. As a part of work commissioned by Marine Scotland, a number of candidate modelling methods were critically compared and the Complex REgion Spatial Smoother (CReSS) [3] with spatially adaptive knot placement using SALSA [4] was the recommended approach due to its success at locating spatially explicit impact-related change. The CReSS/SALSA approach was coupled with Generalised Estimating Equations (GEEs), which accommodate the spatial and temporal correlation that is generally inherent in baseline monitoring and impact assessment data. We present the capabilities of MRSea using an example data set from the package, which is based on offshore data collected from an existing renewables development. Specifically, we analyse a scenario where the animals have re-distributed across the study area between two time points, before and after construction of an offshore wind farm. We begin with correcting the observed counts from the survey data for imperfect detection, fit a spatial model with environmental covariates to the corrected counts, assess the fit of the model, run model diagnostics, make predictions and calculate uncertainty about these predictions. Most importantly for these applications, we identify spatially explicit significant differences in animal density before and after the construction. INTRODUCTION Previous assessment of the impact of renewable energy development has focused upon measuring differences in animal abundance prior to and following development. This approach suffered from the disadvantages of a) attributing any potential change to development as the causal agent, b) failing to acknowledge other forces that influence animal abundance and distribution and c) insensitivity to more subtle changes in animal populations, e.g. shifts in animal distribution to areas of habitat quality different than prior to renewable development. The statistical issues to be addressed in assessing animal population distribution and potential changes to those distributions are subtle and complex. If methods for addressing such questions were straightforward, then methods would be universally in use. However, such methods are at the leading edge of statistical development. This paper presents a new package MRSea, which was developed specifically to tackle the assessment of potential impacts of renewable developments on marine wildlife, although the methods are applicable to other studies as well. The functions of this package can be used to analyse segmented line transect data and nearshore vantage point data and include model fitting, diagnostic tools and non-parametric bootstrapping to estimate uncertainty. DATA The data were simulated based on off-shore survey data collected before an impact effect, for example the construction of a wind-farm, hereafter referred to as ‘the impact’. The impact effect was then imposed which reduced animal numbers in the impacted area and re-distributed these animals to the south east of the study region (Figure 1). Observed counts, with imperfect detection imposed, were lifted from this surface in the form of line-transects. This is the data set called dis.data.re within the MRSea package. 1 Corresponding author: lass@st-and.ac.uk