Transferability of a Bayesian Belief Network across diverse agricultural catchments using high-frequency hydrochemistry and land management data.

Biogeochemical catchment models are often developed for a single catchment and, as a result, often generalize poorly beyond this. Evaluating their transferability is an important step in improving their predictive power and application range. We assess the transferability of a recently developed Bayesian Belief Network (BBN) that simulated monthly stream phosphorus (P) concentrations in a poorly-drained grassland catchment through application to three further catchments with different hydrological regimes and agricultural land uses. In all catchments, flow and turbidity were measured sub-hourly from 2009 to 2016 and supplemented with 400-500 soil P test measurements. In addition to a previously parameterized BBN, five further model structures were implemented to incorporate in a stepwise way: in-stream P removal using expert elicitation, additional groundwater P stores and delivery, and the presence or absence of septic tank treatment, and, in one case, Sewage Treatment Works. Model performance was tested through comparison of predicted and observed total reactive P (TRP) concentrations and percentage bias (PBIAS). The original BBN accurately simulated the absolute values of observed flow and TRP concentrations in the poorly and moderately drained catchments (albeit with poor apparent percentage bias scores; 76 % ≤ PBIAS≤94 %) irrespective of the dominant land use, but performed less well in the groundwater-dominated catchments. However, including groundwater total dissolved P (TDP) and Sewage Treatment Works (STWs) inputs, and in-stream P uptake improved model performance (-5 % ≤ PBIAS≤18 %). A sensitivity analysis identified redundant variables further helping to streamline the model applications. An enhanced BBN model capable for wider application and generalisation resulted.

Cryosphere and land cover influence on stream water quality in Central Asia's glacierized catchments.

This work helps address recent calls for systematic water quality assessment in Central Asia and considers how nutrient and salinity sources, and transport, affect water quality along the continuum from the cryosphere to the lowland plains. Spatial and, for the first time, temporal variations in stream water pH, temperature, electrical conductivity, and nitrate and phosphate concentrations are presented for four catchments (485-13,500 km2), all with glaciers and major urban areas. The catchments studied were: Kaskelen (Kazakhstan), Ala-Archa (Kyrgyzstan), Chirchik (Uzbekistan) and the Kofarnihon (Tajikistan). Measurements were made in cryosphere, stream water, groundwater, reservoir and lake samples over a 22-month period at fortnightly intervals from 35 sites. The results highlight that glacier, permafrost and rock glacier outflows were primary and secondary nitrate sources (>1 mg N L-1) to the headwaters, and there were major increases in salinity and nitrate concentrations where rivers receive inputs from agriculture and settlements. Overall, the water quality complied with national and World Health Organization standards, however there were pollution hot-spots with shallow urban groundwaters contaminated with nitrate (>11 mg N L-1) and stream electrical conductivity above 800 µS cm-1 in some agricultural areas indicative of high salinity. Phosphate concentrations were generally low (<0.06 mg P L-1) throughout the catchments, though elevated (>0.2 mg P L-1) in urban areas due to effluent contamination. A melt water dilution effect along the main river channels was discernible, in the electrical conductivity and nitrate concentration seasonal dynamics, 100 s of km from the headwaters. Thus, the input of relatively clean water from the cryosphere is an important regulator of main channel water quality in the urban and farmed lowland plains adjacent to the Tien Shan and Pamir. Improved sewage treatment is needed in urban areas.

Spatial properties affecting the sensitivity of soil water dissolved organic carbon long-term median concentrations and trends.

It is increasingly clear that increases in dissolved organic carbon in upland waters in recent decades have often been dominated by acid deposition, but reasons for substantial variation in rates of change remain unclear. This paper focuses on the extent to which spatial properties, such as variation in soil properties, atmospheric deposition and climate, affect the sensitivity of DOC concentrations in soil water. The purpose is to i) examine evidence for differences in site average concentrations and trends in soil water DOC between sites with contrasting ecosystem properties, i.e. vegetation cover and soil type, and ii) identify the wider combination of site characteristics that best explain variation in these DOC metrics between sites. We collated soil water and deposition chemistry, soil chemistry and meteorological data from 15 long-term UK monitoring sites (1992-2010) covering a range of soils, vegetation, climate and acid deposition levels. Mineral soils under forests showed the greatest range of long-term mean DOC concentrations and trends. Regression analysis indicated that acid and sea-salt deposition, and soil sensitivity to acidification were the factors most strongly associated with spatial variation in mean DOC concentrations. Spatial variation in DOC trends were best explained by Al saturation and water flux. Overall, the sensitivity of DOC release from soil to changes in pollutant deposition could be related to the type of vegetation cover and soils chemistry properties, such as Al saturation, divalent base cation content and hydrological regime. The identification of the ecosystem properties that appear most influential in modifying DOC production and responses to long-term drivers, helps elucidate potential mechanistic explanations for differences in DOC dynamics across seemingly similar ecosystems, and points to the importance of DOC mobility in regulating its dynamics.

A Bayesian network to simulate macroinvertebrate responses to multiple stressors in lowland streams.

Aquatic ecosystems are affected by multiple environmental stressors across spatial and temporal scales. Yet the nature of stressor interactions and stressor-response relationships is still poorly understood. This hampers the selection of appropriate restoration measures. Hence, there is a need to understand how ecosystems respond to multiple stressors and to unravel the combined effects of the individual stressors on the ecological status of waterbodies. Models may be used to relate responses of ecosystems to environmental changes as well as to restoration measures and thus provide valuable tools for water management. Therefore, we aimed to develop and test a Bayesian Network (BN) for simulating the responses of stream macroinvertebrates to multiple stressors. Although the predictive performance may be further improved, the developed model was shown to be suitable for scenario analyses. For the selected lowland streams, an increase in macroinvertebrate-based ecological quality (EQR) was predicted for scenarios where the streams were relieved from single and multiple stressors. Especially a combination of measures increasing flow velocity and enhancing the cover of coarse particulate organic matter showed a significant increase in EQR compared to current conditions. The use of BNs was shown to be a promising avenue for scenario analyses in stream restoration management. BNs have the capacity for clear visual communication of model dependencies and the uncertainty associated with input data and results and allow the combination of multiple types of knowledge about stressor-effect relations. Still, to make predictions more robust, a deeper understanding of stressor interactions is required to parametrize model relations. Also, sufficient training data should be available for the water type of interest. Yet, the application of BNs may now already help to unravel the contribution of individual stressors to the combined effect on the ecological quality of water bodies, which in turn may aid the selection of appropriate restoration measures that lead to the desired improvements in macroinvertebrate-based ecological quality.

Responses of Aquatic Plants to Eutrophication in Rivers: A Revised Conceptual Model.

Compared to research on eutrophication in lakes, there has been significantly less work carried out on rivers despite the importance of the topic. However, over the last decade, there has been a surge of interest in the response of aquatic plants to eutrophication in rivers. This is an area of applied research and the work has been driven by the widespread nature of the impacts and the significant opportunities for system remediation. A conceptual model has been put forward to describe how aquatic plants respond to eutrophication. Since the model was created, there have been substantial increases in our understanding of a number of the underlying processes. For example, we now know the threshold nutrient concentrations at which nutrients no longer limit algal growth. We also now know that the physical habitat template of rivers is a primary selector of aquatic plant communities. As such, nutrient enrichment impacts on aquatic plant communities are strongly influenced, both directly and indirectly, by physical habitat. A new conceptual model is proposed that incorporates these findings. The application of the model to management, system remediation, target setting, and our understanding of multi-stressor systems is discussed. We also look to the future and the potential for new numerical models to guide management.

Modelling the bioaccumulation of persistent organic pollutants in agricultural food chains for regulatory exposure assessment.

New models for estimating bioaccumulation of persistent organic pollutants in the agricultural food chain were developed using recent improvements to plant uptake and cattle transfer models. One model named AgriSim was based on K OW regressions of bioaccumulation in plants and cattle, while the other was a steady-state mechanistic model, AgriCom. The two developed models and European Union System for the Evaluation of Substances (EUSES), as a benchmark, were applied to four reported food chain (soil/air-grass-cow-milk) scenarios to evaluate the performance of each model simulation against the observed data. The four scenarios considered were as follows: (1) polluted soil and air, (2) polluted soil, (3) highly polluted soil surface and polluted subsurface and (4) polluted soil and air at different mountain elevations. AgriCom reproduced observed milk bioaccumulation well for all four scenarios, as did AgriSim for scenarios 1 and 2, but EUSES only did this for scenario 1. The main causes of the deviation for EUSES and AgriSim were the lack of the soil-air-plant pathway and the ambient air-plant pathway, respectively. Based on the results, it is recommended that soil-air-plant and ambient air-plant pathway should be calculated separately and the K OW regression of transfer factor to milk used in EUSES be avoided. AgriCom satisfied the recommendations that led to the low residual errors between the simulated and the observed bioaccumulation in agricultural food chain for the four scenarios considered. It is therefore recommended that this model should be incorporated into regulatory exposure assessment tools. The model uncertainty of the three models should be noted since the simulated concentration in milk from 5th to 95th percentile of the uncertainty analysis often varied over two orders of magnitude. Using a measured value of soil organic carbon content was effective to reduce this uncertainty by one order of magnitude.

False positive and false negative errors in the design and implementation of agri-environmental policies: A case study on water quality and agricultural nutrients.

When designing and implementing agri-environmental policies to reduce nutrient loss, action programmes may falsely address areas where the nutrient issue from agricultural activity is not currently important and is not likely to become so in the future (a false positive), or may fail to address areas where the agricultural nutrient issue is currently important or may likely become so in the future (a false negative). Based on a case study of the Louros watershed in Greece, this work identifies database and modelling sources of false positives and negatives and proposes a decision making process aimed at minimizing the possibility of committing such errors. The baseline is well simulated and shows that the Louro's watershed falls behind a Good Environmental Status, at least marginally. Simulated mitigation measures show that the river's status can be upgraded to "Good", at least as concerns nitrates and ammonium. Simulated climate change does not seem to exert an important positive or negative effect. Land use changes forecasting considerably less cultivated area have a significant effect on Total Phosphorous but not on nitrates or ammonium concentrations. The non-linearity between nutrient disposition (inputs) and nutrient concentration in downstream water bodies (output) and the many factors that affect the nutrient disposition-transportation-concentration chain, highlights the importance of simulating the effects of mitigation actions and of future climate and land use changes before adopting and establishing agri-environmental measures.

Sensors in the Stream: The High-Frequency Wave of the Present.

New scientific understanding is catalyzed by novel technologies that enhance measurement precision, resolution or type, and that provide new tools to test and develop theory. Over the last 50 years, technology has transformed the hydrologic sciences by enabling direct measurements of watershed fluxes (evapotranspiration, streamflow) at time scales and spatial extents aligned with variation in physical drivers. High frequency water quality measurements, increasingly obtained by in situ water quality sensors, are extending that transformation. Widely available sensors for some physical (temperature) and chemical (conductivity, dissolved oxygen) attributes have become integral to aquatic science, and emerging sensors for nutrients, dissolved CO2, turbidity, algal pigments, and dissolved organic matter are now enabling observations of watersheds and streams at time scales commensurate with their fundamental hydrological, energetic, elemental, and biological drivers. Here we synthesize insights from emerging technologies across a suite of applications, and envision future advances, enabled by sensors, in our ability to understand, predict, and restore watershed and stream systems.

Assessment and improvement of biotransfer models to cow's milk and beef used in exposure assessment tools for organic pollutants.

The aim of this study was to assess and improve the accuracy of biotransfer models for the organic pollutants (PCBs, PCDD/Fs, PBDEs, PFCAs, and pesticides) into cow's milk and beef used in human exposure assessment. Metabolic rate in cattle is known as a key parameter for this biotransfer, however few experimental data and no simulation methods are currently available. In this research, metabolic rate was estimated using existing QSAR biodegradation models of microorganisms (BioWIN) and fish (EPI-HL and IFS-HL). This simulated metabolic rate was then incorporated into the mechanistic cattle biotransfer models (RAIDAR, ACC-HUMAN, OMEGA, and CKow). The goodness of fit tests showed that RAIDAR, ACC-HUMAN, OMEGA model performances were significantly improved using either of the QSARs when comparing the new model outputs to observed data. The CKow model is the only one that separates the processes in the gut and liver. This model showed the lowest residual error of all the models tested when the BioWIN model was used to represent the ruminant metabolic process in the gut and the two fish QSARs were used to represent the metabolic process in the liver. Our testing included EUSES and CalTOX which are KOW-regression models that are widely used in regulatory assessment. New regressions based on the simulated rate of the two metabolic processes are also proposed as an alternative to KOW-regression models for a screening risk assessment. The modified CKow model is more physiologically realistic, but has equivalent usability to existing KOW-regression models for estimating cattle biotransfer of organic pollutants.

Assessment of plant uptake models used in exposure assessment tools for soils contaminated with organic pollutants.

The aim of this study was to evaluate and improve the accuracy of plant uptake models for neutral hydrophobic organic pollutants (1 < logK(OW) < 9, -8 < logK(AW) < 0) used in regulatory exposure assessment tools, using uncertainty and sensitivity analyses. The models considered were RAIDAR, EUSES, CSOIL, CLEA, and CalTOX. In this research, CSOIL demonstrated the best performance of all five exposure assessment tools for root uptake from polluted soil in comparison with observed data, but no model predicted shoot uptake well. Recalibration of the transpiration and volatilisation parameters improved the performance of CSOIL and CLEA. The dominant pathway for shoot uptake simulated differed according to the properties of the chemical under consideration; those with a higher air-water partition coefficient were transported into shoots via the soil-air-plant pathway, while chemicals with a lower octanol-water partition coefficient and air-water partition coefficient were transported via the root. The soil organic carbon content was a particularly sensitive parameter in each model and using a site specific value improved model performance.

An analysis of long-term trends, seasonality and short-term dynamics in water quality data from Plynlimon, Wales.

This paper examines two hydrochemical time-series derived from stream samples taken in the Upper Hafren catchment, Plynlimon, Wales. One time-series comprises data collected at 7-hour intervals over 22 months (Neal et al., 2012-this issue), while the other is based on weekly sampling over 20 years. A subset of determinands: aluminium, calcium, chloride, conductivity, dissolved organic carbon, iron, nitrate, pH, silicon and sulphate are examined within a framework of non-stationary time-series analysis to identify determinand trends, seasonality and short-term dynamics. The results demonstrate that both long-term and high-frequency monitoring provide valuable and unique insights into the hydrochemistry of a catchment. The long-term data allowed analysis of long-term trends, demonstrating continued increases in DOC concentrations accompanied by declining SO(4) concentrations within the stream, and provided new insights into the changing amplitude and phase of the seasonality of the determinands such as DOC and Al. Additionally, these data proved invaluable for placing the short-term variability demonstrated within the high-frequency data within context. The 7-hour data highlighted complex diurnal cycles for NO(3), Ca and Fe with cycles displaying changes in phase and amplitude on a seasonal basis. The high-frequency data also demonstrated the need to consider the impact that the time of sample collection can have on the summary statistics of the data and also that sampling during the hours of darkness provides additional hydrochemical information for determinands which exhibit pronounced diurnal variability. Moving forward, this research demonstrates the need for both long-term and high-frequency monitoring to facilitate a full and accurate understanding of catchment hydrochemical dynamics.

Sewage effluent clean-up reduces phosphorus but not phytoplankton in lowland chalk stream (River Kennet, UK) impacted by water mixing from adjacent canal.

Information is provided on phosphorus in the River Kennet and the adjacent Kennet and Avon Canal in southern England to assess their interactions and the changes following phosphorus reductions in sewage treatment work (STW) effluent inputs. A step reduction in soluble reactive phosphorus (SRP) concentration within the effluent (5 to 13 fold) was observed from several STWs discharging to the river in the mid-2000s. This translated to over halving of SRP concentrations within the lower Kennet. Lower Kennet SRP concentrations change from being highest under base-flow to highest under storm-flow conditions. This represented a major shift from direct effluent inputs to a within-catchment source dominated system characteristic of the upper part to the catchment. Average SRP concentrations in the lower Kennet reduced over time towards the target for good water quality. Critically, there was no corresponding reduction in chlorophyll-a concentration, the waters remaining eutrophic when set against standards for lakes. Following the up gradient input of the main water and SRP source (Wilton Water), SRP concentrations in the canal reduced down gradient to below detection limits at times near its junction with the Kennet downstream. However, chlorophyll concentrations in the canal were in an order of magnitude higher than in the river. This probably resulted from long water residence times and higher temperatures promoting progressive algal and suspended sediment generations that consumed SRP. The canal acted as a point source for sediment, algae and total phosphorus to the river especially during the summer months when boat traffic disturbed the canal's bottom sediments and the locks were being regularly opened. The short-term dynamics of this transfer was complex. For the canal and the supply source at Wilton Water, conditions remained hypertrophic when set against standards for lakes even when SRP concentrations were extremely low.

An assessment of the fine sediment dynamics in an upland river system: INCA-Sed modifications and implications for fisheries.

There is a need for better links between hydrology and ecology, specifically between landscapes and riverscapes to understand how processes and factors controlling the transport and storage of environmental pollution have affected or will affect the freshwater biota. Here we show how the INCA modelling framework, specifically INCA-Sed (the Integrated Catchments model for Sediments) can be used to link sediment delivery from the landscape to sediment changes in-stream. INCA-Sed is a dynamic, process-based, daily time step model. The first complete description of the equations used in the INCA-Sed software (version 1.9.11) is presented. This is followed by an application of INCA-Sed made to the River Lugg (1077 km(2)) in Wales. Excess suspended sediment can negatively affect salmonid health. The Lugg has a large and potentially threatened population of both Atlantic salmon (Salmo salar) and Brown Trout (Salmo trutta). With the exception of the extreme sediment transport processes, the model satisfactorily simulated both the hydrology and the sediment dynamics in the catchment. Model results indicate that diffuse soil loss is the most important sediment generation process in the catchment. In the River Lugg, the mean annual Guideline Standard for suspended sediment concentration, proposed by UKTAG, of 25 mg l(-1) is only slightly exceeded during the simulation period (1995-2000), indicating only minimal effect on the Atlantic salmon population. However, the daily time step simulation of INCA-Sed also allows the investigation of the critical spawning period. It shows that the sediment may have a significant negative effect on the fish population in years with high sediment runoff. It is proposed that the fine settled particles probably do not affect the salmonid egg incubation process, though suspended particles may damage the gills of fish and make the area unfavourable for spawning if the conditions do not improve.

Modeling the stream water nitrate dynamics in a 60,000-km2 European catchment, the Garonne, southwest France.

The spatial and temporal dynamics in the stream water NO(3)-N concentrations in a major European river-system, the Garonne (62,700 km(2)), are described and related to variations in climate, land management, and effluent point-sources using multivariate statistics. Building on this, the Hydrologiska Byråns Vattenbalansavdelning (HBV) rainfall-runoff model and the Integrated Catchment Model of Nitrogen (INCA-N) are applied to simulate the observed flow and N dynamics. This is done to help us to understand which factors and processes control the flow and N dynamics in different climate zones and to assess the relative inputs from diffuse and point sources across the catchment. This is the first application of the linked HBV and INCA-N models to a major European river system commensurate with the largest basins to be managed under the Water Framework Directive. The simulations suggest that in the lowlands, seasonal patterns in the stream water NO(3)-N concentrations emerge and are dominated by diffuse agricultural inputs, with an estimated 75% of the river load in the lowlands derived from arable farming. The results confirm earlier European catchment studies. Namely, current semi-distributed catchment-scale dynamic models, which integrate variations in land cover, climate, and a simple representation of the terrestrial and in-stream N cycle, are able to simulate seasonal NO(3)-N patterns at large spatial (>300 km(2)) and temporal (> or = monthly) scales using available national datasets.

Chlorophyll-a in the rivers of eastern England.

Chlorophyll-a concentration variations are described for two major river basins in England, the Humber and the Thames and related to catchment characteristics and nutrient concentrations across a range of rural, agricultural and urban/industrial settings. For all the rivers there are strong seasonal variations, with concentrations peaking in the spring and summer time when biological activity is at its highest. However, there are large variations in the magnitude of the seasonal effects across the rivers. For the spring-summer low-flow periods, average concentrations of chlorophyll-a correlate with soluble reactive phosphorus (SRP). Chlorophyll-a is also correlated with particulate nitrogen (PN), organic carbon (POC) and suspended sediments. However, the strongest relationships are with catchment area and flow, where two straight line relationships are observed. The results indicate the importance of residence times for determining planktonic growth within the rivers. This is also indicated by the lack of chlorophyll-a response to lowering of SRP concentrations in several of the rivers in the area due to phosphorus stripping of effluents at major sewage treatment works. A key control on chlorophyll-a concentration may be the input of canal and reservoir waters during the growing period: this too relates to issues of residence times. However, there may well be a complex series of factors influencing residence time across the catchments due to features such as inhomogeneous flow within the catchments, a fractal distribution of stream channels that leads to a distribution of residence times and differences in planktonic inoculation sources. Industrial pollution on the Aire and Calder seems to have affected the relationship of chlorophyll-a with PN and POC. The results are discussed in relation to the Water Framework Directive.

Phosphorus sources, speciation and dynamics in the lowland eutrophic River Kennet, UK.

This paper examines the behaviour of phosphorus (P) in a lowland chalk (Cretaceous-age) stream, the upper River Kennet in southern England, which has been subject to P remediation by tertiary treatment at the major sewage treatment works in the area. The effects of treatment are examined in relation to boron, a conservative tracer of sewage effluent and in terms of the relative contributions of soluble reactive phosphorus (SRP) loads from point and diffuse sources, and in-stream SRP loads. These results indicate a baseline reduction in in-stream SRP concentrations immediately following P-treatment of approximately 72%. Subsequent high flows result in a greater contribution of diffuse inputs and increases in SRP levels relative to the initial post-treatment period. The dynamics of SRP and particulate phosphorus (PP) are examined under a wide range of river flow conditions. Given the flashy nature of near-surface runoff in the River Kennet, sub-weekly (daily automated) sampling was used to examine the dynamics in SRP and PP concentrations in response to storm events. Simple empirical models linking weekly SRP concentrations with flow were developed. The empirical models were successfully applied to the daily data, to partition TP measurements and provide an estimate of daily SRP and PP concentrations. Mass balance studies were used to examine net gains and losses along the experimental river reach and indicate large net losses (up to 60%) during the extreme low flows and high SRP concentrations prior to P-treatment, which may be linked to extensive epiphytic growth. Phosphorus dynamics and response to P-treatment are discussed in relation to hydrological controls in permeable chalk catchments and wider implications for eutrophication management are examined.

Water quality functioning of lowland permeable catchments: inferences from an intensive study of the RIVER KENNEt and upper River Thames.

This paper brings together information on the water quality functioning of the River Kennet and other parts of the upper River Thames in the south east of England. The Kennet represents a groundwater fed riverine environment impacted by agricultural and sewage sources of nutrient pollution. Descriptions of the general water quality of the area, nutrient sources, sinks and within river processes are provided together with biological responses to driving issues of agriculture, sewage treatment and climatic change. Models are developed and applied to assess the key processes involved for a highly dynamic system and to provide initial estimates of the likely responses to environmental change. Furthermore, the economic aspects of pollution control are reviewed, together with legislation issues, which are presented within the context of a landmark case known as the 'Axford Inquiry', the implications of which extend to regional and national dimensions. The paper concludes with a discussion on the present state of knowledge, key issues and future research on the science and management of groundwater fed nutrient impacted riverine systems.