Catchment Risk Assessment of Steroid Oestrogens from Sewage Treatment Works

This project has developed a regional catchment-based risk assessment for steroid oestrogens in England and Wales. Using the Low Flows 2000 water quality (LF2000-WQX) model, which can predict river concentrations of contaminants discharged through sewage treatment plants (STPs), the project has focussed on predicting concentrations of the three most potent steroid oestrogens in rivers and the associated risk of endocrine disruption in fish. The model could equally well be applied to any other chemical of concern found in treated sewage effluent, where loading can be estimated on a per capita basis. The model, believed to be the first of its kind, covers the inland waters of England and Wales in unprecedented detail. It includes 357 catchments covering 122,000km2 and incorporates more than 2,000 STPs serving over 29 million people (STPs discharging to estuaries and coastal waters and those serving very small communities are excluded). The model predicts the concentrations of three steroid oestrogens – oestradiol (E2), oestrone (E1) and ethinyloestradiol (EE2) – and the associated risk of endocrine disruption for 10,313 individual river reaches (21,452 km). The scale of this assessment underlines the usefulness of computer-based risk assessment methods. Developing this regional model was only possible due to the remarkable cooperation of different groups within the Environment Agency and the UK water industry in establishing the underlying database. The model calculates how much of each of the three oestrogens would be discharged into the receiving waters from data on the flow of the STPs and the populations they serve. The in-stream concentrations are then calculated based on dilutions down through the river network, including the effects of the natural attenuation rate. During the development of the model, refinements were added to allow one of the oestrogens, oestradiol, to be converted in-stream into its first degradation product, oestrone, which is another oestrogen. This simulates what has been observed in the field and allows a more accurate prediction of overall oestrogenicity. In addition, an approach has been developed that allows users to identify and calculate what additional levels of improvement are required for the most polluting STPs in order for there to be no predicted risk of endocrine disruption in their catchment. Three specific tasks were required to generate the factors underpinning the model. First, the most recent literature and data on STP oestrogen removal efficiencies were reviewed. Primary treatment plants, activated sludge plants (with and without tertiary treatment) and biological filter plants (with and without tertiary treatment) were all considered. The latest UK data indicated that the removal efficiency for E1 in biological filter plants without tertiary treatment was significantly different to that previously determined, being reduced by around 30 per cent. For all other types of STP, the new data indicated that a slightly improved removal efficiency of 69 per cent should be used. In the cases of E2 and EE2, only a slight modification was necessary, increasing the removal efficiency to 83 per cent for all treatment types. Second, recent scientific studies measuring the effects of steroid oestrogens were reviewed. This allowed PNECs (predicted no effect concentrations) of 0.1ngL-1, 1ngL-1 and 3ngL-1 to be established for EE2, E2 and E1 respectively. A method for calculating the E2 equivalent concentrations was also developed. This divides each steroid by its respective PNEC to produce a measure of relative potency and these values are then summed, as the effects of steroids have been shown to be additive. Thus the [E2 equivalent] = [EE2]/0.1 + [E2]/1 + [E1]/3 (with the square brackets denoting concentrations). Finally, the risk class boundaries were also reviewed and it was established that the currently proposed total steroid oestrogen PNEC (1ngL-1 E2 equivalent) remained valid for distinguishing ‘no risk’ sites from ‘at risk’ sites. The review also determined that the boundary between ‘at risk’ and ‘high risk’ sites should be set at 10ngL-1 E2 equivalent. This Science Report – Catchment Risk Assessment of Steroid Oestrogens from Sewage Treatment Work v was estimated to be equivalent to the lowest measured population effect end-point for E2 in published literature. Overall, the majority of the reaches in England and Wales (61 per cent using mean concentrations and expressed as a percentage of the total river length modelled) are predicted not to be ‘at risk’ from endocrine disruptive effects in fish (< 1ngL-1 E2 equivalent). However, a significant proportion remains ‘at risk’ (>1 ngL-1 E2 equivalent; 39 per cent of length of the modelled reaches under mean conditions). These risk proportions are not evenly distributed throughout England and Wales. The lowest proportions predicted to be ‘at risk’ are in Wales and the South West (5 per cent and 16 per cent respectively). In the Southern, North East, and North West regions, 34 per cent, 38 per cent, and 34 per cent of the reach lengths are predicted to be ‘at risk’ respectively. The highest proportions of reaches predicted to be ‘at risk’ are in the Thames, Midlands and Anglian regions, with 67 per cent, 55 per cent, and 50 per cent respectively. Key factors influencing the proportion of river reaches classified as being ‘at risk’ are the location of densely populated areas and the available dilution (which is a function of rainfall, evaporation and upstream water use). The proportion of lengths predicted to be ‘at risk’ seems rather high, but the high proportion of intersex individuals reported in wild roach in two national surveys (Environment Agency 1995 and 2003) suggests the predicted risk is reasonable, at least for this species. A very small proportion of reaches, around 1–2 per cent, were predicted to be at ‘high risk’ (>10ngL-1 E2 equivalent). However, many of these ‘high risk’ reaches were short stretches of headwaters or ditches composed almost entirely of sewage effluent. For this reason, consideration will need to be given to the most appropriate use of this model in determining which options for improving the removal of oestrogens from the environment will provide the greatest benefit for fish populations and their habitats. A more detailed risk assessment was carried out using these same methods for 12 sites defined as Special Areas of Conservation (SACs). A simpler ‘no risk’ (<1ngL-1 E2 equivalent) or ‘at risk’ (E2 equivalent >1ngL-1) assessment was used, which incorporated a lower predicted no effect concentration for EE2 of 0.06ngL-1. Four of these sites were predicted to have at least one reach ‘at risk’ under mean concentrations, rising to seven sites under 90th percentile concentrations (Chapter 8). This risk assessment was based on readily available data sets and due diligence has been taken in the quality control of these data. However, there are limitations associated with the data and certain outstanding issues that will need to be addressed in the longer term, which both contribute uncertainty to the model. For instance, a correct association needs to be made between each STP and its receiving water course and it would be advantageous to use measured dry weather flows rather than estimated values. Further improvements would include having more detailed estimates of STP steroid removal efficiencies, or even measured values for individual STPs, and refining the PNEC, which may alter the risk category thresholds and the calculation of E2 equivalent concentrations. Also, it is recommended that the predicted environmental concentrations produced by the model should be compared with measured data in water bodies. Furthermore, the predicted risk for fish should be compared to observed effect data, so that the risk assessment can be refined accordingly. This model gives a detailed and comprehensive picture of the likely levels of exposure of freshwater fish populations to steroid oestrogens. It should therefore help in the development of a rational and cost effective strategy to reduce the risk of population decline, by targeting areas where steroid oestrogen reduction would prove of greatest benefit to fish stocks and to the wider environment.