Research papersSensitivity of river eutrophication to multiple stressors illustrated using graphical summaries of physics-based river water quality model simulations
Graphical abstract
Introduction
Global water resources are affected by a complex mixture of stressors resulting from a range of drivers, including intensification of land use and climate change. Across the full spectrum of European river types, 90% of lowland rivers experience pressure from combinations of hydrological, morphological and water quality stressors (Schinegger et al., 2012). Understanding when and how these stressors interact to alter ecological status is essential for developing effective river basin management plans and influencing environmental policy (Hering et al., 2015, Jackson et al., 2016), yet it is challenging to untangle individual stressor contributions without very comprehensive observations (Dafforn et al., 2016).
River phytoplankton blooms represent a key threat to water quality and the structure and function of aquatic ecosystems (Smith and Schindler, 2009), pressurising anthropogenic water use particularly under projected future climate (Bussi et al., 2016). Essential to managing blooms effectively, is better understanding of their timing, magnitude and duration, and their physical and chemical controls. Whilst elevated chlorophyll concentrations are indicative of blooms it is the effects on dissolved oxygen (DO) which provide strong and holistic indicators of the wider health of river ecosystems. Consequently, preventing low DO concentrations, which are harmful across various trophic levels (e.g. for fish and benthic macro-invertebrates at 3–4 mg DO L−1 (Garvey et al., 2007)) is a primary objective of the EU Water Framework Directive. Oxygen concentrations are influenced by numerous factors including: (a) photosynthesis and associated autotrophic respiration, (b) water temperature, (c) aeration and mixing, (d) microbial decay of organic matter (Cox, 2003). As phytoplankton blooms develop they become increasingly susceptible to a population crash rendering rivers vulnerable to oxygen depletion following their decay. Oxygen concentration in turn influences biological responses with low levels causing biotic impairment (Comte et al., 2010). In such complex systems of abiotic-biotic feedback it is difficult to separate the influences of individual stressors (Rode et al., 2016). Therefore, in large basins, evaluating impacts on DO of changes in stressor combinations is somewhat intractable.
Considerable uncertainties in DO response are in part directly attributable to incomplete understanding of phytoplankton behaviour (Nguyen and Willems, 2016). It is commonly assumed in river quality models (Chapra, 1997) that a species or group of phytoplankton will take advantage of conditions whatever the temperature. Various observations support this assumption (e.g. in the Meuse (Descy et al., 2003) and Loire (Descy et al., 2012)). However, cool water diatom species proliferate and dominate the phytoplankton assemblage in many cases (Jeong et al., 2007). Moreover, in the River Thames (UK), the subject of the present study, phytoplankton biomass appears suppressed in mid-summer possibly due to short-lived nutrient limitation together with biological interactions (e.g. control by grazers at higher trophic levels) (Waylett et al., 2013, Bowes et al., 2016). Hence, an alternative hypothesis of biological response is that phytoplankton biomass is dominated by cool water diatom species which bloom in spring and early summer.
Combined effects of stressor interactions can be categorised, after Brown et al., (2013), as: (a) Additive; where there is no interaction and so the overall effect is the sum of their effects in isolation, (b) Synergistic; where the combined effect of two stresses is greater than the additive expectation, (c) Antagonistic; where the combined effect is less than the additive expectation, (d) Mitigating, where the effects of the two stressors cancel one another out, (e) Opposing; where the relationship switches from being additive to mitigating as one of the stressors increases.
Multi-stressor studies considering impacts on DO are scarce, although long-term lake monitoring studies of climate and land management impacts have been undertaken (e.g. North et al., 2013). In contrast, empirical statistical analyses from numerous observational studies have identified impacts of stressor combinations on aquatic biota. Schinegger et al. (2012) cite a wide breadth of studies identifying effects in habitats and organisms across a European-wide river survey. Some survey-based studies specifically categorise how stressors interact. Using empirical approaches linked to process-based modelling, Segurado et al. (2018) identified complex responses of fish and macrophyte biodiversity in Portuguese rivers to combinations of nutrient and flow stress, characterising opposing and antagonistic relationships. In multi-stressed Danish headwaters mitigation of single stressors is unlikely to have substantial benefits on benthic macroinvertebrate communities (Rasmussen et al., 2013).
In smaller streams, controlled experiments have characterised both synergistic (Wagenhoff et al., 2012) and antagonistic (Townsend et al., 2008) macroinvertebrate response to interactive nutrient and sediment stress. The range of stressors has sometimes been widened to consider climate-driven pressure whereby complex macroinvertebrate responses were found (Elbrecht et al., 2016, Piggott et al., 2015) which also depended on whether biota were assessed at population or community level. Due to the diversity of findings from controlled experiments, synthesising and scaling-up understanding to larger rivers is not straightforward. Nevertheless, bringing findings together from over 250 monitoring and controlled experiments, Jackson et al. (2016) categorised biotic response to paired stressors and identified antagonistic interaction to be most prevalent (41%), followed by synergistic (28%), additive (16%) and mitigating (15%). Of these, studies focusing on primary producers such as phytoplankton revealed synergistic relations to be more prevalent than the additive or antagonistic responses typically found for macroinvertebrates.
The inconsistencies apparent from observational studies imply considerable system complexity. Whilst basin-scale process-based river modelling studies are abundant, Hipsey et al. (2015) found that applications have primarily related degradation of aquatic habitat to water quality deterioration under specific scenarios, with limited exploitation of their potential to make more generic characterisation of responses arising from different combinations of stressors. Historically however, infrequent biological observations fail to match the high resolution possible with physical and chemical measurement (e.g. Rode et al., 2016) thereby limiting the value of ecosystem assessment (Dafforn et al., 2016). Advances in biomonitoring science and earth observation are now allowing much more comprehensive datasets of stressors and response variables (such as chlorophyll: Hunter et al., 2008) to be compiled at large scales, with consequent improvements in process-based ecological modelling. In turn the advent of high-performance computing has widened the scope for defining the nature of multi-stressor-response relationships; which is now possible even in the absence of driving data, by undertaking increasingly exhaustive sensitivity analysis of stressor combinations and their impacts.
The importance of abiotic-biotic feedbacks, physical factors and climatic variables, in addition to nutrient levels, in controlling phytoplankton and DO in large rivers has been widely demonstrated (e.g. Hardenbicker et al., 2014, Istvanovics et al., 2009, van Vliet and Zwolsman, 2008, Bowes et al., 2016, Minaudo et al., 2018). However, whilst detailed, these contrasting studies highlight that understanding arising of multi-stressor impacts cannot be comprehensive, instead being largely rudimentary typically only providing snapshots rather than response along stressor gradients of a system. From their work on the Thames, Bussi et al. (2016) highlight that modelling studies often only cover two climatic stressors, and recommend that sensitivity of model output should be assessed more systematically in terms of a greater number of stressor variables.
The aim of the present study was to make an ensemble of water quality model applications to assess the sensitivity of eutrophication in the Thames to a range of stressor variables and to develop a method for representing the outputs. Five stressor variables were chosen (water temperature, phosphate concentration, river flow, urbanisation and riparian shading) to which eutrophication in the Thames is known to be especially sensitive (Waylett et al., 2013, Bowes et al., 2016, Hutchins et al., 2018). The variables themselves reflect a range of pressure types including those related to climate, catchment management and population growth. Individually, they are not directly mappable to specific pressures in isolation (such as agricultural pollution or increased water demand). However, as a group of variables together they are a useful proxy for assessing the sensitivity of river water to a wide range of commonly occurring pressures acting in combination. Our analysis, although enabling us to provide some advice for managers, does not have a primary aim of identifying a detailed set of guidelines.
To achieve the aims of the study, main constituent objectives are:
- •
To establish a method of visualising eutrophication responses comprehensively and concisely.
- •
To execute a sensitivity analysis of the five stressors and to demonstrate their impact on three summary indicators of eutrophication derived from chlorophyll-a and DO concentrations.
- •
To detect which stressors predominate and identify different types of relationship (additive, synergistic, antagonistic, mitigating etc) acting on DO and chlorophyll-a indicators from stressor pairs.
- •
To repeat the analysis using two contrasting versions of the model, so as to evaluate the impact of uncertainty reflecting incomplete understanding of biological processes controlling eutrophication.
Whilst ultimately of fundamental importance, the present study cannot identify which version of the two contrasting biological models (representing the two specific competing hypotheses of biological response described earlier) is best. This would require considerable additional biological monitoring and targeted experimental work. Instead, using the two model versions we illustrate how uncertainty in understanding of the biological system affects the multi-stressor assessment. Differing impacts on chlorophyll-a indicators arising from change in temperature stress are to be expected from the contrasting model formulations, but any dissimilarity in the nature of other stress-response interactions is far less clear. Therefore, we distinguish those circumstances of multiple stress where there is agreement as opposed to disagreement between models; thereby identifying where we can already be confident of making an appropriate management decision, and conversely where a better system understanding is necessary prior to prescribing interventions.
Section snippets
Eutrophication and multiple stressors in the river Thames
The River Thames (Fig. 1) with a catchment area of 9948 km2 (Marsh and Hannaford, 2008) extends upstream from the tidal limit for 257 km, and is the largest river wholly in England. Receiving mean annual rainfall of 720 mm (Marsh and Hannaford, 2008) the river basin is a source of water to 14 million people (Whitehead et al., 2015) and receives waste from almost 4 million inhabitants (Kinniburgh and Barnett, 2009). The upper Thames basin is predominantly underlain by Oolitic Limestone, with
Results and discussion
The section comprises the following narrative:
- 1.
Firstly scatterplots (Fig. 4a, Fig. 4b, Fig. 4c, Fig. 4d) are presented illustrating that relationships between the three indicators of eutrophication are not straightforward (Section 3.1).
- 2.
We then present a key (Fig. 5) which explains the contour plots (legends) and their interpretation regarding change in stress from present day conditions; and results for all 10 combinations of stressor pairs are presented for each indicator and each model (
Conclusions
We developed a sensitivity analysis framework to assess five types of stressor (water temperature, phosphate concentration, river flow, urbanisation and riparian shading) acting on river eutrophication as represented in a process-based water quality model. Using graphical illustration, we pinpointed findings in the context of model uncertainty (by running two models: one assuming a mixed population of phytoplankton (MP) the other assuming dominance by a cool water diatom Stephanodiscus
Declaration of Competing Interest
None.
Acknowledgements
Financial support primarily came from POLLCURB (a project funded under the NERC Changing Water Cycle programme, ref number: NE/K002317/1). Financial support was also received from MARS (Managing Aquatic ecosystems and water Resources under multiple Stress: grant agreement 603378 (EU-FP7 project)) and from NERC-CEH Water Resources Science area. Data for converting scalar multipliers into time series were provided by the NRFA and Mike Bowes of the CEH Thames Initiative. We would also like to
References (48)
- et al.
Changes in water quality of the River Frome (UK) from 1965 to 2009: is phosphorus mitigation finally working?
Sci. Total Environ.
(2011) - et al.
Nutrient and light limitation of periphyton in the River Thames: implications for catchment management
Sci. Total Environ.
(2012) - et al.
Identifying multiple stressor controls on phytoplankton dynamics in the River Thames (UK) using high-frequency water quality data
Sci. Total Environ.
(2016) - et al.
Impacts of climate change, land-use change and phosphorus reduction on phytoplankton in the River Thames (UK)
Sci. Total Environ.
(2016) - et al.
Mapping eutrophication risk from climate change: future phosphorus concentrations in English rivers
Sci. Total Environ.
(2018) A review of dissolved oxygen modelling techniques for lowland rivers
Sci. Total Environ.
(2003)- et al.
The interactive responses of water quality and hydrology to changes in multiple stressors, and implications for the long-term effective management of phosphorus
Sci. Total Environ.
(2013) - et al.
Managing aquatic ecosystems and water resources under multiple stress–an introduction to the MARS project
Sci. Total Environ.
(2015) What impact might mitigation of diffuse nitrate pollution have on river water quality in a rural catchment?
J. Environ. Manage.
(2012)- et al.
Which offers more scope to suppress river phytoplankton blooms: reducing nutrient pollution or riparian shading?
Sci. Total Environ.
(2010)
Combined impacts of future land-use and climate stressors on water resources and quality in groundwater and surface waterbodies of the upper Thames river basin, UK
Sci. Total Environ.
Delayed influence of dam storage and discharge on the determination of seasonal proliferations of Microcystis aeruginosa and Stephanodiscus hantzschii in a regulated river system of the lower Nakdong River (South Korea)
Water Res.
River flow forecasting through conceptual models: part 1 – a discussion of principles
J. Hydrol.
A catchment scale evaluation of multiple stressor effects in headwater streams
Sci. Total Environ.
Understanding multiple stressors in a Mediterranean basin: Combined effects of land use, water scarcity and nutrient enrichment
Sci. Total Environ.
Eutrophication science: where do we go from here?
Trends Ecol. Evol.
Impact of summer droughts on the water quality of the Meuse river
J. Hydrol.
Physico-chemical factors alone cannot simulate phytoplankton behaviour in a lowland river
J. Hydrol.
Dynamic modelling of multiple phytoplankton groups in rivers with an application to the Thames river system in the UK
Environ. Modell. Software
Managing for interactions between local and global stressors of ecosystems
PLoS One
Surface Water-quality Modeling
Assessment of stream biological responses under multiple-stress conditions
Environ. Sci. Pollut. Res.
Big data opportunities and challenges for assessing multiple stressors across scales in aquatic ecosystems
Mar. Freshw. Res.
Modelling the impact of benthic filter feeders on the composition and biomass of river plankton
Freshw. Biol.
Cited by (10)
Spatiotemporal variations and controlling mechanism of low dissolved oxygen in a highly urbanized complex river system
2024, Journal of Hydrology: Regional StudiesDistinct indicators of land use and hydrology characterize different aspects of riverine phytoplankton communities
2022, Science of the Total EnvironmentCitation Excerpt :Higher phytoplankton abundance may be due to increases in urban land use and decreases in forest habitat (Kakouei et al., 2021). On one hand, river in urban area needs special concern on phosphorus control (Hutchins and Hitt, 2019). Additionally, urban area percentage was detected indirectly impact on phytoplankton functional groups through the influence of phosphate and total phosphorus concentration in the river Treene (Qu et al., 2019).
Multiple stressors influencing the general eutrophication status of transitional waters of the Brazilian tropical coast: An approach utilizing the pressure, state, and response (PSR) framework
2022, Journal of Sea ResearchCitation Excerpt :The effects of multiple stressors on coastal ecosystems are poorly understood (Deschutter et al., 2017), especially in the assessment of trophic status. Modeling approaches are limited to subsystems (reservoirs and lakes) (Hartnett et al., 2012, Hutchins and Hitt, 2019; Rankinen et al., 2019, Andersen et al., 2020). The current global trend is the implementation of legislative measures that assess the ecological integrity of marine systems (estuarine, coastal, and offshore) from this perspective (Bricker et al., 2003; Ferreira et al., 2006; Borja et al., 2008).
Quantile regression illuminates the successes and shortcomings of long-term eutrophication remediation efforts in an urban river system
2021, Water ResearchCitation Excerpt :Given their critical role in the hydrological and ecological connectivity of surface waters, analyses of eutrophication responses by rivers are still desperately needed (Dodds and Smith, 2016). Relative to lakes, river ecosystems are spatially and temporally dynamic, with complex interacting factors driving water quality changes (Basu and Pick, 1995; Bowes et al., 2016; Hutchins and Hitt, 2019). Water temperature, water residence time (discharge), transparency (light), and pH have all been found to be correlated with temperate river phytoplankton biomass in previous studies, in addition to nutrients (Basu and Pick, 1996; Balbi, 2000; Bukaveckas et al., 2011; Bowes et al., 2016).
Citizen scientists filling knowledge gaps of phosphate pollution dynamics in rural areas
2024, Environmental Monitoring and AssessmentA Simulation Experiment on Quality Dynamics of Reclaimed Water under Different Flow Exchanges
2022, International Journal of Environmental Research and Public Health