SNIFFER WFD119: Enhancement of the River Invertebrate Classification Tool (RICT).

Abstract

EXECUTIVE SUMMARY Project funders/partners: Environment Agency (EA), Northern Ireland Environment Agency (NIEA), Scotland & Northern Ireland Forum for Environmental Research (SNIFFER), Scottish Environment Protection Agency (SEPA) Background to research The Regulatory Agencies in the UK (the Environment Agency; Scottish Environment Protection Agency; and the Northern Ireland Environment Agency) now use the River Invertebrate Classification Tool (RICT) to classify the ecological quality of rivers for Water Framework Directive compliance monitoring. RICT incorporates RIVPACS IV predictive models and is a highly capable tool written in a modern software programming language. While RICT classifies waters for general degradation and organic pollution stress, producing assessments of status class and uncertainty, WFD compliance monitoring also requires the UK Agencies to assess the impacts of a wide range of pressures including hydromorphological and acidification stresses. Some of these pressures alter the predictor variables that current RIVPACS models use to derive predicted biotic indices. This project has sought to broaden the scope of RICT by developing one or more RIVPACS model(s) that do not use predictor variables that are affected by these stressors, but instead use alternative GIS based variables that are wholly independent of these pressures. This project has also included a review of the wide range of biotic indices now available in RICT, identifying published sources, examining index performance, and where necessary making recommendations on further needs for index testing and development. Objectives of research •To remove and derive alternative predictive variables that are not affected by stressors, with particular emphasis on hydrological/acidification metric predictors. •To construct one or more new RIVPACS model(s) using stressor independent variables. •Review WFD reporting indices notably AWIC(species), LIFE (species), PSI & WHPT. Key findings and recommendations : Predictor variables and intellectual property rights : An extensive suite of new variables have been derived by GIS for the RIVPACS reference sites that have been shown to act as stressor-independent predictor variables. These include measures of stream order, solid and drift geology, and a range of upstream catchment characteristics (e.g. catchment area, mean altitude of upstream catchment, and catchment aspect). It is recommended that decisions are reached on which of the newly derived model(s) are implemented in RICT so that IPR issues for the relevant datasets can be quickly resolved and the datasets licensed. It is also recommended that licensing is sought for a point and click system (where the dataset cannot be reverse engineered) that is capable of calculating any of the time-invariant RIVPACS environmental predictor variables used by any of the newly derived (and existing) RIVPACS models, and for any potential users. New stressor-independent RIVPACS models : Using the existing predictor variables, together with new ones derived for their properties of stressor-independence, initial step-wise forward selection discriminant models suggested a range of 36 possible models that merited further testing. Following further testing, the following models are recommended for assessing watercourses affected by flow/hydromorphological and/or acidity stress: • For flow/hydromorphological stressors that may have modified width, depth and/or substrate in GB, it is suggested that a new ‘RIVPACS IV – Hydromorphology Independent’ model (Model 24) is used (this does not use the predictor variables width, depth and substratum, but includes a suite of new stressor-independent variables). • For acidity related stressors in GB, it is suggested that a new ‘RIVPACS IV – Alkalinity Independent’ model (Model 35) is used (this does not use the predictor variable alkalinity, but includes new stressor-independent variables). • For flow/hydromorphological stressors and acidity related stressors in GB, it is suggested that a new ‘RIVPACS IV – Hydromorphology & Alkalinity Independent’ model (Model 13) is used (this does not use the predictor variables width, depth, substratum and alkalinity, but includes a suite of new stressor-independent variables). • Reduced availability of appropriate GIS tools at this time has meant that no new models have been developed for Northern Ireland. Discriminant functions and end group means have now been calculated to enable any of these models to be easily implemented in the RICT software. Biotic indices : The RIVPACS models in RICT can now produce expected values for a wide range of biotic indices addressing a variety of stressors. These indices will support the use of RICT as a primary tool for WFD classification and reporting of the quality of UK streams and rivers. There are however a number of outstanding issues with indices that need to be addressed: • There is a need to develop a biotic index for assessing metal pollution. • WFD EQR banding schemes are required for many of the indices to report what is considered an acceptable degree of stress (High-Good) and what is not (Moderate, Poor or Bad). • A comprehensive objective testing process needs to be undertaken on the indices in RICT using UK-wide, large-scale, independent test datasets to quantify their index-stressor relationships and their associated uncertainty, for example following the approach to acidity index testing in Murphy et al., (in review) or organic/general degradation indices in Banks & McFarland (2010). • Following objective testing, the UK Agencies should make efforts to address any index under-performance issues that have been identified, and where necessary new work should be commissioned to modify existing indices, or develop new ones where required so that indices for all stress types meet certain minimum performance criteria. • Testing needs to be done to examine index-stressor relationships with both observed index scores and RIVPACS observed/expected ratios. Work should also be done to compare the existing RIVPACS IV and the new stressor-independent models (developed in this project) as alternative sources of the expected index values for these tests. • Consideration should be given to assessing the extent to which chemical and biological monitoring points co-occur. Site-matched (rather than reach-matched) chemical and biological monitoring points would i) generate the substantial training datasets needed to refine or develop new indices and ii) generate the independent datasets for testing.