A comprehensive monitoring programme was carried out in the Aire, Calder, Went and Rother catchments in the UK. A total of 804 effluent samples from 36 sewage treatment works (STWs) and 1100 water samples from 54 river sampling sites were analysed. Concentrations of linear alkylbenzene sulfonate (LAS), boron and other water quality determinands in STW effluent and river waters over a 2-year period (August 1996–August 1998) are reported. The data illustrate the temporal and spatial variations in concentrations of LAS and boron in river waters and effluents. Concentrations of LAS in effluents reflect the biological treatment employed and the influence of tertiary treatment is clearly demonstrated.

As a result of the introduction of tighter discharge limits and effluent treatment processes at source, the concentration of alkylphenol ethoxylates and nonylphenol present in the final effluent discharge from a sewage treatment works that treats trade effluent from the textiles industry was reduced. The estrogenic effects of the final effluent discharge to the Aire River were compared over a four‐year period during which various treatment measures were introduced. Male rainbow trout exposed to the effluent on four occasions in consecutive years (1994–1997) showed a reduction in the level of induced vitellogenesis between 1994 and 1997. A marked decrease in gonadosomatic index (GSI) and increase in heptaosomatic index (HSI) was measured in fish exposed to the effluent in 1994. In successive years, these differences diminished, and in the case of the GSI no measurable difference was observed between fish exposed to the final effluent or those in the control group in 1997. However, an increase in HSI was still measurable in 1997 in fish exposed to the final effluent and at sites farther downstream. The reduction in the effects of the effluent paralleled the reduction in the concentration of nonylphenol as well as its mono‐ and diethoxylates, which have been demonstrated to produce estrogenic effects in trout exposed to these compounds in the laboratory. This study demonstrates that the setting of more restricted discharge limits for known estrogenic chemicals of industrial origin can lead to significant reductions in the estrogenic activity of the watercourses into which the effluents are discharged.

Chemical analyses were combined with a biological assay to investigate the main estrogenic chemicals as they passed through a sewage treatment works (STW) and entered a river. The STW studied was unusual in that it received wastewater from the textile trade. This wastewater was shown to contain high concentrations of alkylphenol polyethoxylates and their degradation products, such as nonylphenol. High-performance liquid chromatography fractionation, combined with biological assay, showed that the majority of the estrogenic activity was contributed by the alkylphenolic chemicals and the natural estrogens 17 beta-estradiol and estrone. Despite removal of a high proportion of (lie alkylphenolic chemicals by the various treatment processes within the STW, concentrations in the final effluent were still high compared to most other STW effluents in the United Kingdom. The effluent was very estrogenic to caged fish, as was the river water 2 and 5 km downstream of the STW, even though less so. Using various approaches, attempts were made to determine which group of chemicals contributed most to the estrogenic activity of the effluent. The analysis suggested that, in this unusual situation, the alkylphenolic chemicals may contribute the majority of the estrogenic activity of the effluent. However, this conclusion was based on a number of uncertainties that are presently unresolved and hence can be considered only tentative.

The concentrations of boron in the Aire and Calder which have been measured in order to validate the GREAT-ER model have been used to derive a predicted environmental concentration (PEC) for boron in the Aire catchment in Yorkshire. The data have been obtained from 33 sites, with monthly sampling over a 2-year period. The 90th percentile of the means of measured concentrations (242 microg l(-1)) is higher than the 90th percentile predicted by the GREAT-ER model (186 microg l(-1)), if river stretches are normalised by their flow volumes, and unpolluted headwaters are included in the catchment averaging procedure. However, if the river stretches are normalised by the length of the respective stretch, and unpolluted headwaters are included, a lower regional PEC of 28 microg l(-1) is obtained. If normalisation by stretch length is appropriate for regional risk assessment, then the 90th percentile of the means of the measured data will overpredict the regional boron concentration in a catchment such as the Aire, with a low natural level of boron in the headwaters. Monitoring data for boron in the effluents from sewage treatment works in the Aire catchment, and for boron in the Went and Rother catchments, with a higher background boron level, are also given.

An in-stream removal experiment has been carried out in Red Beck, a small stream which receives effluent from Shibden Head Sewage Treatment Works. This trickling filter works serves a purely domestic population of 9408 but is scheduled to be closed, and the flows diverted to another works, as part of Yorkshire Water's continuing capital investment programme. An anionic detergent, linear alkylbenzene sulfonate (LAS), boron, and standard water quality parameters have been measured at seven sites downstream of the effluent discharge point. Time of travel has been measured by detection of a fluorescent dye added to the effluent sampling chamber, and the increase in flow as the river proceeds through the catchment has been determined from current flow measurements, and from boron dilution data. Assuming a first order removal mechanism, the overall half-life for LAS removal is just over 2 h (2 h 14 min). Faster removal takes place in the upper portion of the stream, and removal over the last five sampling points is somewhat slower, with a half-life of 2 h 40 min. This removal may comprise both primary biodegradation and the deposition of suspended matter to which the surfactant has been adsorbed. There was no significant difference in the removal half-lives of the individual alkyl chain length homologues.

The results of a linear alkylbenzene sulphonate (LAS) monitoring exercise carried out during August and September, at six different trickling filter sewage treatment plants in the Aire and Calder catchments in the UK are described. The average LAS concentration in flow proportional, composite, crude sewage was 3.25 mg l−1 and ranged from 1.10 to 5.58 mg l−1. Based on consumer usage data and boron measurements these results indicate that up to 60% of the LAS was removed in the sewers prior to reaching the sewage treatment plants. The concentration of LAS in flow proportional, composite final effluents from well-operated plants ranged from 4 to 460 μg l−1, with an average value of 240 μg l−1. Removal through the plants ranged from 70–99% with mean values of 90.7, 93.4, 93.0, 74.6, 97.3 and 92.4% for the six plants. A simple model is described which can be used to accurately predict the diel variation in LAS concentrations in the final effluent when measured influent concentrations, flow and settling tank volumes are known.