Multiple Stressors in River Ecosystems: Status, Impacts and Prospects for the Future

Multiple Stressors in River Ecosystems: Status, Impacts and Prospects for the Future provides a comprehensive and current overview on the topic as written by leading river scientists who discuss the relevance of co-occurring stressors for river ecosystems. River ecosystems are subject to multiple stressors that threaten their ecological status and the ecosystem services they provide. This book updates the reader’s knowledge on the response and management of river ecosystems to multi-stress situations occurring under global change. Detailing the risk for biodiversity and functioning in a case-study approach, it provides insight into methodological issues, also including the socioeconomic implications.

• Presents a case study approach and geographic description on the relevance of multiple stressors on river ecosystems in different biomes
• Gives a uniquely integrated perspective on different stressors, including their interactions and joint effects, as opposed to the traditional one-by-one approach
• Compiles state-of-the-art methods and technologies in monitoring, modeling and analyzing river ecosystems under multiple stress conditions

• Language: English
• Publisher: Elsevier Science
• Release date: Aug 30, 2018
• ISBN: 9780128118009

Book preview

China

Preface

Sergi Sabater

Arturo Elosegi

Ralf Ludwig

Why do we need a book on multiple stressors in river ecosystems?

Well, first of all, no book has yet been written! This comes as a surprise, as it is quite clear that stressors rarely occur alone in the environment, particularly in river ecosystems, where they often act jointly to produce complex responses. Direct and indirect feedback between stressor effects result in interactions that range from synergistic to antagonistic and may produce ecological surprises. At the same time, ecosystems differ in their resistance and resilience to stressors, and they can show thresholds beyond which critical shifts occur between ecosystem states. Even more so—and this is extending the domain of typical considerations in the field—multiple stressors can affect the biodiversity and function of river ecosystems and consequently have an impact on the goods and services that societies derive from rivers. It is therefore time for a joint analysis and realistic appraisal of natural and human-driven stressors.

The creation of this book was triggered by acknowledging the existence of three main knowledge gaps: (i) adopting a multistressor perspective to assess current and future anthropogenic impacts on river ecosystems, in contrast to the current dominance of studies that usually focus on only one or several stressors at once; (ii) integrate the ecological studies of multiple stressors in rivers with socioeconomic studies so that the research findings can be placed into an interdisciplinary context appropriate for policy makers, water resource managers, and government agencies, as well as research scientists and environmental groups; and (iii) providing a contemporary compilation of case studies, applications, and research findings from around the world.

As such, this book takes complementary perspectives in order to reach an understanding of the types and consequences of multiple stressors in river systems. One series of chapters details the impacts of multiple stressors in regard to some of the main driving stressors and their implications. A second series of chapters embraces the geographical perspective and describes the relevance of multiple stressors in different ecoregions of the world, their effects on the river ecosystems, and their implications for biodiversity, structure, function, and human uses and perceptions. The final series of chapters account for the ways multiple stressor challenges can be approached, including managerial and social perspectives. Most of the chapters in the book use examples and provide updated references, as well as make use of pictures and graphs to illustrate the specific stressors and their effects on river ecosystems.

Finally, by summarizing the main findings and exposed evidence, the book aims to explore implications and provides recommendations for research, management, and policy. The rising number of stressors affecting river ecosystems, their increasing geographic extent, and their important consequences for both nature and society make it essential to adapt current river monitoring schemes and create a data repository to account for the evidence existing from many different sources. The worldwide examples reveal the necessity to consider scale dependencies in multiple stressor occurrences, as well as to undertake a more active scientific and social communication of risks associated thereto. All the evidence collected in the book, as well as the many uncertainties that can be envisaged, require prompt and concerted actions, not solely from the research side, but also from the societal side, where management and policy should be included. The challenges we face require moving to areas of the planet where many lessons can still be learned, and preserve as much as possible all other regions which have already suffered from long-lasting impairment.

We trust that the contents of this book, which has collected and combined the fragmented knowledge on multiple stressors in river ecosystems, can serve as a synthetic reference for researchers in all aquatic and environmental sciences. Given the breadth and interdisciplinarity of the presented materials, the target audience also includes graduate students and professionals in riverine ecology, as well as river and catchment managers.

We hope that this elaboration on multiple stressors in river ecosystems will enhance readers’ enthusiasm to delve more deeply into this fascinating and highly relevant field. Enjoy!

Chapter 1

Defining Multiple Stressor Implications

Sergi Sabater⁎,†; Arturo Elosegi‡; Ralf Ludwig§    ⁎ IEA (UdG), Girona, Spain

† Catalan Institute for Water Research (ICRA), Girona, Spain

‡ University of the Basque Country (UPV/EHU), Bilbao, Spain

§ Ludwig Maximilians Universitaet Muenchen (LMU), Munich, Germany

Abstract

Stressors rarely occur alone in the environment. Particularly in river ecosystems, many stressors often act jointly and produce complex responses. A realistic appraisal positions us in the necessary joint analysis of natural and human-driven stressors. In this chapter, we aim to define what a stressor is, how it affects the receptors, and the multiple ways in which stressors interact. We emphasize the existing literature analyses of the effects of multiple stressors, as well as the outcomes most commonly found. Multiple stressors can affect biodiversity and the functioning of river ecosystems, but also the goods and services that societies derive from rivers. Stressors differ in nature and need to be considered hierarchically, as they may differ in their associated energy as well as in their frequency of occurrence. Direct and indirect feedback between stressor effects result in interactions that range from synergistic to antagonistic and may produce ecological surprises. Ecosystems differ in their resistance and resilience to stressors, and they can show thresholds beyond which critical shifts occur between ecosystem states.

Keywords

Stressor; Stress; Receptor; Effect

Acknowledgments

This work has been supported by the European Community 7th Framework Programme Funding under Grant agreement no. 603629-ENV-2013-6.2.1-GLOBAQUA.

1.1 Global Environmental Change: A Source of Stress

The term Anthropocene was coined by Crutzen (2006) to name the current epoch in the Earth's history, in which a global environmental change caused by humans is affecting the sedimentary records on a planetary scale. This term responds to the modern manner of understanding humanity's position within the ecosystems, instead of considering them as strange and disruptive elements to nature. As a matter of fact the present state of the biosphere cannot be understood without human fingerprints; pristine ecosystems, if they still exist, are increasingly rare, and the actions of humans reach the most remote places on Earth. Global environmental change is general and affects both ecosystems and people.

As stated by Margalef (1997), humankind produces an acceleration in the use of power and the aggravation of many ecological problems to the global scale by using exosomatic energy derived from fossil sources. Effects include land use transformations and climate change. The melting of ice caps and the drowning of Pacific atolls as the sea level rises are among the vast evidence of shifting climate regimes throughout the planet. The transformation of land from forest to agricultural fields and to urban dwellings is also a relevant aspect of global environmental change. Satellite images have shown than in the last 30 years, over 170.000 km² have been converted from aquatic to terrestrial areas as wetlands were drained, whereas an additional 115.000 km² have followed the reverse path as land was drowned by reservoirs (Donchyts et al., 2016). Approximately 40% of the Earth's surface is currently occupied by croplands and pastures (Foley et al., 2005). The use of phosphorus and nitrogen fertilizers has increased worldwide two- and sevenfold, respectively, while the area of irrigated cropland has doubled during the last 40 years (Tilman et al., 2001). The steady rise in human population, the development of global economies, and the migration of people from rural to urban areas is increasing the demand for food and transportation, as well as the intensity of alterations of urban ecosystems (Grimm et al., 2008). These global demands exert strong pressure on water resources, with nearly 15% (40,000 km³ year− 1) of the world's total runoff is currently derived for hydropower, flood control, and water supply (Oki and Kanae, 2006). Thus global environmental change is deeply transforming biogeochemical cycles, such as that of carbon, nitrogen, or water.

Predictions for climate change are that air temperature and extreme weather events will increase, while spatial patterns in precipitation and runoff will also change (Milly et al., 2005) (Chapter 2). The Intergovernmental Panel on Climate Change (IPCC) predicts continued increases in greenhouse gases and a push of global temperatures by 2–4.5°C in the next 50 years (Stocker, 2014). Annual average river runoff might increase by 10%–40% at higher latitudes and decrease by 10%–30% over some dry regions. This indeed will affect river flow, with many wet systems becoming even wetter, whereas in drier areas permanent rivers may become intermittent as intermittent ones become ephemeral (Greve et al., 2014). Thus climate change will likely increase the frequency of floods and droughts (Chapter 6), which will in turn affect river geomorphology and habitat availability, and increase water temperature, as well as the concentration of sediments and nutrients (Sabater and Tockner, 2010; Chapter 16). River biodiversity will also suffer due to these changes, especially as a consequence of five major processes, namely the exploitation of water resources, pollution, the modification of flow regimes, the degradation of river habitats (Chapter 4), and the spread of invasive species (Dudgeon et al., 2006; Chapter 3), which could all be considered major stressors. These processes will have detrimental consequences for humans as well, who depend on goods and services provided by natural ecosystems (Kundzewicz et al., 2008; Chapters 17 to 19).

1.2 Stress, Stressor, Receptor, and Effect: Context and Evolution of the Terms

Before analyzing the effects of environmental change on ecosystems and on the associated human societies (socioecological systems), some key concepts should be formulated. Although the word stressor has been sometimes used as synonym of disturbance, the latter is a broader term with a long tradition in ecology (Pickett and White, 1985) and is used to define, when referring to river ecosystems, any relatively discrete event in time that is characterized by a frequency, intensity, and severity outside a predictable range, and that disrupts ecosystem, community, or population structure and changes resources or the physical environment (Resh et al., 1988). Thus a disturbance is any event, either natural or anthropogenic, that disrupts the population, community, or ecosystem structure; it is most often an event that by eliminating individuals creates void spaces.

The term stressor, on the other hand, focuses exclusively on anthropogenic disturbances (Segner et al., 2014; Crain et al., 2008; Piggott et al., 2016). A landslide, a large flood, or a tsunami are catastrophic disturbances, but they cannot be considered stressors under this conception. These, together with smaller disturbances such as sediment movement during a spate or a snag falling into a river, to name a few, are natural disturbances. The hydrological alteration caused by hydropeaking, or the presence of xenobiotics, are instead anthropogenic disturbances or stressors. Some stressors, under moderate amounts, may also act as a subsidy for particular groups of organisms (Box 1.1).

Box 1.1

Some Key Terms Used in This Book

Stressor: Any external abiotic or biotic factor derived from human intervention, which moves a receptor out of its normal operating range, causing either subsidy or stress.

Receptor: Any biological system impacted by a stressor. Although organisms are the entities most directly affected by stressors, the consequences can be detected at a large range of levels of complexity, from molecular entities to communities, and even ecosystems.

Stress: A reduction in the biological activity of a receptor as consequence of the presence of a stressor.

Subsidy: An increase in the biological activity of a receptor as a consequence of the moderate presence of a stressor.

Effect: Any change produced by a stressor on the receptor. It depends on the intensity, timing, and duration of the stressor, as well as on the sensitivity of the receptor to that particular stressor.

1.3 Types of Stressors According to Their Energy and Frequency

As already mentioned, natural disturbances are intrinsic to the structure and function of river ecosystems, and they shape their dynamic nature. River organisms tend to be adapted by natural selection to the disturbance regime historically existing in their habitats, which act as environmental filters (Angermeier and Winston, 1998). The high prevalence of natural disturbances makes it difficult to detect and assess the influence of anthropogenic stressors on river communities and ecosystems. In fact a large set of natural and anthropogenic disturbances, nested in spatial and temporal extents, interact to affect biological receptors (Fig. 1.1). For instance, on the continental scale, the distribution of fish in European lakes is driven by large-scale natural variations in climate and biogeography, whereas on the local scale it is driven by eutrophication, which is the result of both natural and anthropogenic factors (Brucet et al., 2013). Natural and man-made disturbances thus have a shared effect that cannot be easily disentangled (Feld et al., 2016), what forces any analysis to consider regionally the effects of a given set of stressors. This is the perspective taken in this book and outlined in the regional analysis undertaken later in the book.

Fig. 1.1 Contrasting stressors in river ecosystems. (A) Cloth washing, a low intensity stress with low persistence. (B) Dredging, a very intense stressor with likely short-term effects under natural hydrological dynamics. (C) Severe chemical pollution may affect a long river reach and likely will produce lasting legacy effects even after the pollution source is removed. (D) Severely incised stream in an agricultural area. (E) A small weir that affects a short reach, but local effects are long lasting. (F) Channelized river also affected by pollution and water abstraction. (G) Urban stream, affected by chemical and physical stressors, most of these long lasting or irreversible. (H) A new river being dug in an artificial landscape after an open mine closure in Germany. Extremely acidic conditions and legacy metal pollution in the entire catchment will likely result in long-lasting legacy stress.

An energy-based perspective can help to further understand the types and effects of stressors. Local stressors differ from larger-scale or global stressors in their associated energy (or intensity), as well as in their temporal occurrence (frequency; Fig. 1.2). Stressors either occur at spatial and temporal scales to which river systems cannot cope with (e.g., channelization and damming; Nilsson et al., 2005), or are novel to organisms (e.g., xenobiotic substances such as pesticides and pharmaceutical products; Ricart et al., 2010). Some simply may move a species outside of its normal physiological state (i.e., its homeostasis). But when the intensity level falls beyond the range to which organisms are adapted, local extinctions may occur, thus affecting biodiversity and potentially ecosystem functioning. Alternatively, if the homeostatic mechanisms of organisms are enough to withstand the level of stress, subtle changes can occur at the physiological level. These, in turn, can ultimately affect communities and ecosystems through ecological interactions. This is the case when tolerant species gain a competitive advantage and outcompete the less tolerant others.

Fig. 1.2 Distribution of stressors according to their associated energy, measured in terms of intensity and frequency. The more intense the stressor, the less reversible is its effect on the biological receptor.

Stressors therefore encompass a wide range of possible actions and effects because they have different associated energies, frequencies, and spatial and temporal scales (Fig. 1.2). The effects may be noticeable only at the level of individuals; however, they may be perceived at the community level or ultimately at the whole ecosystem level, depending on their intensity (Segner et al., 2014). However, in real-world situations, it is difficult to disentangle the effects from different stressors. For instance, changes in land uses result in cocktails of stressors (Feld et al., 2016), as when agriculture affects aquatic biodiversity through diffuse pollution, siltation, and in-stream habitat degradation. Similarly, urbanization deeply affects water fluxes; it is associated to point source sewage inputs, and tends to create barriers to the movement of fauna and flora.

We can, therefore, envisage a hierarchy of stressors (Stevenson and Sabater, 2010) in terms of their intensity, frequency, and spatial scale. This hierarchy determines that some stressors produce localized effects on shorter time scales, while those on larger scales are likely to occur less frequently. The former may be transient or reversible while the latter produce irreversible effects (Fig. 1.2). The relationship between intensity and frequency of perturbations is, generally, inversely related (Margalef, 1997; Vitousek, 1994); the less energetic stressors tend to occur much more frequently than those associated with higher energies. Life is able to persist when it can adapt to situations where stressor energy and frequency of occurrence allows recovery.

1.4 Multiple Stressors in the Literature

Ecologists study the interactions among organisms and their environment and have long developed predictive models of the response of ecosystems to environmental change, some of which are very pertinent to the present book (Chapter 14). In the subsidy-stress perturbation theory, Odum et al. (1979) distinguished two types of environmental variables that depend on the general shape of the relationships describing their effects on biological activity. For some variables, such as temperature or nutrient concentration (the so-called concentration variables), there is a hump-shaped relationship between their concentration or level and biological activity. Moderate increases from base levels promote biological activity above the normal operating range, producing what is called a subsidy effect. Nevertheless, beyond a threshold, these same variables cause the reduction of biological activity, which first returns to a normal range and later may even fall below normal, in what is known as the stress effect. For other variables such as toxic substances (termed as pollutant variables) the shape of the relationship is much simpler, as the stress (i.e., the level of depression in biological activity) is roughly dependent on their concentration.

The intermediate disturbance hypothesis (Grime, 1973; Connell, 1978) proposed a hump-shaped relationship between the frequency and intensity of the disturbance and diversity of biological communities. According to this hypothesis, competitive exclusion is a strong force that reduces the diversity of biological communities occurring in environments with a low level of disturbance. As the intensity or the frequency of disturbance increases, the environmental variability reduces competitive exclusion and promotes coexistence, thus resulting in more diverse communities. Further increases in the intensity or frequency of disturbance, though, again reduce the biological diversity, as few species are able to withstand such a strong disturbance regime.

On the other hand, much has been discussed on the characteristics of disturbances (both natural and anthropogenic) according to their duration and temporal evolution. Pulse disturbances are short, discrete disturbances that even if they are powerful in their actions (e.g., a flood) will allow the community to recover when predisturbance conditions return (Smith et al., 2009). Press disturbances, on the other hand, are maintained through time. In this case, predisturbance conditions are not recovered; and although the community can adapt to the new situation, usually the original community is deeply altered. Finally, ramp disturbances increase in intensity with time (e.g., agricultural intensification in a basin; Lake et al., 2007). Evidently, the characteristics of stressors (e.g., pulse, press, or ramp) are essential in predicting their potential ecological consequences.

Although there is a large body of literature focusing on the occurrence and effects of multiple stressors in river ecosystems, it is strongly biased toward certain stressors, groups of organisms, or regions. Noges et al. (2016) recently reviewed the literature and found over 75 papers on the effects of multiple stressors in European freshwaters. The best described effects are those corresponding to nutrient concentration and hydrological alterations, either alone or combined (half of the papers), or the two combined with other stressors (including morphological stressors, thermal influences, chemical pollution, and biological stressors). These stressors are frequently associated with restoration actions performed in European rivers (Verdonschot et al., 2013). Most of the papers in the literature dealt with the effects of single stressors or, at most, with a paired combination of stressors. Analyzing more than two stressors is rare in the scientific literature and is practically constrained to laboratory studies (Brennan and Collins, 2015).

The receptors most commonly analyzed in the literature are organisms such as bacteria, algae, invertebrates, and fish. The plethora of papers associated with the monitoring of water bodies following the EU Water Framework Directive (WFD) focus mostly on invertebrates, followed by fish, macrophytes, phytobenthos, and zooplankton (Noges et al., 2016). Structural elements are also dominant outside from Europe. We performed a bibliometric analysis in the ISI Web of Science that yielded 509 papers in the last 5 years (2012–2016). The search was limited to papers showing results from surveys or experimental research and intentionally excluded meta-analyses, reviews, and methodological papers. When some of the papers used more than one descriptor (e.g., algae and invertebrates), counts were performed separately. The results showed that the variables most often analyzed are structural, especially community composition and biomass, although some functional variables such as growth and mortality are also measured (Table 1.1). However, a rising number of papers in the more recent years also incorporate physiological, behavioral (e.g., movement and feeding habits), and integrative measurements (e.g., gross primary production ad organic matter decomposition), as well as some specific genetic biomarkers.

Table 1.1

Growth includes survival and mortality estimates. Integrative metabolism includes also organic matter decomposition. Physiological measurements are inclusive of photosynthesis, enzymatic activities, and fatty acid estimates.

Finally the effects that stressors cause across ecosystems are largely neglected, although it is well known that interactions are common. For instance, Paetzold et al. (2011) observed that chemical pollution in a stream affected the aquatic invertebrates, reducing the density of larvae and the number of emerging adults, which ultimately affected terrestrial ecosystems (i.e., riparian spiders) as well. These unusual studies are much needed to understand the complexity of the effects of multiple stressors in river ecosystems and all of their compartments (Chapter 5).

1.5 Studying the Effects of Multiple Stressors

The high number of and the complex interactions among stressors make it difficult to predict the responses of river ecosystems to global environmental change beyond general expectations, such as a decline in biodiversity. A major issue is therefore how to identify causality in multistress situations (Chapter 15), a question that can be only solved through the careful design of the studies and appropriate definitions of the spatial and temporal scales of observation. We outline here some of the most common approaches followed by researchers, as well as the requisites to identify causality.

1.5.1 Experimental and Observational Approaches

To evaluate and forecast the patterns of response of the ecosystems to multiple stressors, observational studies must be combined with manipulative experiments, both in the field and in the laboratory. Studies should be designed to encompass the potential effects of the stressors, as well as the different responses of the receptors (Section 1.2).

Surveys or monitoring schemes are designed to characterize environmental and biological patterns in a set of sites, and they may be useful in describing the distribution patterns of organisms according to driving pressures, but they are rarely sufficient to identify causality. On the other hand, manipulative experiments in the laboratory (e.g., using micro or mesocosms), as well as in the field (e.g., experimental induction of stressors and translocations), can prove causative relationships if they are well designed. Laboratory experiments, in particular, can shed light on the relevance of the factors tested and on the relationship between stress intensity and response, as well as on the mechanisms involved (Hall et al., 1999). Nevertheless laboratory experiments are necessarily simplifications of the reality, as they replace a multiplicity of factors by a few interactions.

For instance, Corcoll et al. (2015) used mesocosms (artificial streams) to analyze the interaction between two common stressors on biofilm communities: pharmaceutical compounds and flow intermittency. Moderate concentrations of pharmaceuticals affected algal biomass and bacterial taxa richness, whereas the effects of flow intermittency were much stronger. When both stressors were combined, algae were more affected than bacteria. The experiment determined the relative contribution and the interaction strength of the two stressors on stream biofilms and showed some of the mechanisms involved in a way that would have been impossible in the field, although at the cost of gross simplification.

As an example of a field-based observational study, Ponsatí et al. (2016) measured concentrations of DOC, DIN, and organic microcontaminants (e.g., pharmaceutical compounds, industrial organic compounds, and herbicides) in a series of sites across a gradient of pollution. A multivariate analysis (variance partitioning) showed that, although organic microcontaminants were significantly related to the responses in biofilm, their interaction with nutrients had a higher share of the biofilm variance. This study also showed the interaction between chemical pollution and hydrological patterns. During high water flow, biofilms were thinner, more active, and potentially more reactive to chemical influences, while the reverse occurred during low flow conditions. The design was useful to highlight the different influences of environmental and pollution variables, but they could only indicate potential causalities.

Manipulative experiments in the field provide a more robust cause-driven analysis than observational studies, albeit in situations less controlled than in the laboratory. For instance, Sabater et al. (2011) fertilized over 2 years a forested Mediterranean stream to measure the response of different biological compartments to moderate nutrient addition. The hypothesis was that nutrient enrichment would enhance the biomass of primary producers, thus triggering a bottom-up effect. The experiment showed that biofilms responded with a shift towards autotrophy that was maintained for as long as the experimental nutrient enrichment; however, the response was affected by the natural variability of the system that complicated the interpretation of the effects.

Manipulative experiments such as the one described above are costly in time and resources and can even raise ethical questions about the convenience of stressing a river on purpose. Researchers can also take advantage of the opportunities offered by large-scale environmental changes, such as the logging of a forested catchment, the conversion of an area from forest to agriculture, or the building of a reservoir within a river. Restoration projects can also be used to gain information on the effects of the stressors that will be removed (e.g., the elimination of a small dam in a river). Many of these studies are hampered by the lack of data before restoration. In any case, results must be interpreted with caution because legacy effects of the impact may remain for a long period of time (Baumgartner and Robinson, 2015) and also because the trajectories of degradation and restoration can widely differ (Lake et al., 2007). These studies require a careful experimental design, as we will see in the next section.

These examples illustrate that the scales of occurrence and the experimental designs are key to understanding the cause-effect relationship and also that several approximations can be combined from lower to larger-scales perspectives.

1.5.2 Conditions for Adequate Design of Multiple-Stressor Studies

Ecosystems are inherently complex entities whose response to environmental variables is difficult to forecast; this is a possible consequence of nonlinear responses as well as of complex interspecific interactions (Wootton et al., 1996). Their response to multistressor situations is even less tractable and demands the most rigorous experimental design, which must include definitions of clear and accurate hypotheses, selections of the most meaningful variables, and a knowledge of the spatial and temporal scales at which these variables may respond (Downes, 2010).

In monitoring or survey schemes where multiple stressors occur, reference or least disturbed sites are frequently included (Tornés et al., 2007). In these situations, causality can be explored by means of direct ordination analyses such as redundancy analysis (RDA) or canonical correspondence analysis (CCA), although these approaches are mostly correlational and thus provide weak evidence for causality.

The experimental approach, in which an environmental variable is manipulated on purpose, gives a much stronger evidence of causality, especially if there are control sites unaffected by the manipulation. The multiple variants of the BACI (before-after/control-impact) design stand out as the most robust. The BACI design basically includes one or several control reaches as similar as possible to the reach or reaches that are going to be impacted (I); all reaches are studied before (B) and after (A) the impact, and the effect of the impact is measured from the changes in the differences between C and I reaches from the B to the A period (Downes, 2010). This is one of the most realistic designs, but it requires a precise knowledge of the sites and moments when impacts will occur and a good planning of the research. Also, care must be given to the duration of the experiment, as the response of an ecosystem is not necessarily immediate as shown, for instance, by the delay often found to detect the effects of restoration projects (Acuña et al., 2013). It is also debatable whether control and impact reaches should be located in the same river or placed in different tributaries with analogous characteristics (Stewart Oaten and Bence, 2001).

The BACI design has been applied to reach-scale manipulations to assess the effect of several types of stressors, including drought by water abstraction (Arroita et al., 2017), wood removal (Warren and Kraft, 2003), increased nutrient concentration (Artigas et al., 2013), or channelization (Roberts et al., 2016). Nevertheless, the complexity of reach-scale manipulations makes it extremely difficult to perform experiments with combinations of different stressors and limits fully factorial experiments with different levels of each stressor to less realistic laboratory situations.

Moving from small-scale experimental results to patterns observable at larger scales is challenging and requires shifts in uncertainty and realism. Upscaling can be approached by quantitative techniques to measure the relevance of stressors and forecast their effects. When insufficient experimental data result in a weak ability to infer causal relationships, one possibility is to use causal criteria analysis (Norris et al., 2011), a weight-of-evidence approach to evaluate the strength of evidence for one or a number of putative cause-effect relationships. Meta-analysis, on the other hand, uses the results assembled from a number of studies to determine an overall effect-size associated with some treatment (e.g., Ferreira et al., 2015). The Bayesian Belief Networks (BBNs) also offer a framework to describe the chain of causal relationships in multiple stressor situations and are used to quantify the relative influence of individual linkages (Borsuk et al., 2004). In BBNs, cause and effect relationships are combined in an influence diagram to provide a visual representation, though pathways of influence can be also quantified. For instance, Mantyka-Pringle et al. (2014) analyzed by means of BBNs the independent and combined effects of climate and land use changes on freshwater macroinvertebrates and fish, and they identified impacts on small spatial scales. Similar outputs are also provided by the Artificial Neural Networks (ANN), which also use a graphical output to outline relationships between variables or locations. ANNs have been recently applied to connect the results from mesocosm studies to landscape scales (Magierowski et al., 2015).

1.5.3 Potential Interactions Among Multiple Stressors

The ecological effects of stressors can be very complex. Some stressors, such as nutrients or essential elements, produce nonlinear responses, which promote biological activity at low to medium concentrations but can become toxic above a certain threshold (Giorgi, 1995). Further, stressors may interact, and the impact may change in the presence of additional stressors. As an example, anomalous sediment stability promotes algal biomass accrual in regulated rivers, but low water temperatures may counteract this effect (Segura et al., 2011). The sensitivity to stressors differs widely among groups of organisms, and interactions between species occur in the form of competition or cotolerance (Vinebrooke et al., 2004). In other cases the effects of stressors can be attenuated if there is a high biological complexity. Several species within a community perform the same function, and a given species may perform more than one function at the same time (Hector and Bagchi, 2007). This functional redundancy makes biological communities highly adaptable to occurring stressors.

Predicting the response of multiple stressors cannot be dissociated from environmental stochasticity, or the degree of unpredictability that accompanies the variability of environmental conditions. It is therefore necessary to understand the mechanisms associated with the interactions among stressors, and it is useful to predict the combined impact of several stressors based on the similarity of their independent impacts. Folt et al. (1999) described three models for interactions among stressors, named as additive, multiplicative or comparative or nonadditive (i.e., when the effects of the combined stressors is not the sum of their effects). According to his model, synergism and antagonism could derive both from additive and multiplicative models. The nonadditive model accounted for situations where a single stressor takes precedence over the other(s) in their combined effect. Folt et al. (1999) paralleled the nonadditive model to the predictions of the Liebig's law of the minimum, in which the effects of nutrients on an organism's growth depend on the limiting nutrient. As such, when the worst stressor is present the other stressors have no additional impact. On the other hand the additive model predicted that the effects of multiple stressors could be greater (synergism) or lower (antagonism) than the sum of the effects elicited by individual stressors. Folt et al. (1999) considered the additive model as particularly suitable to describe stressors effects on physiological processes. The multiplicative model was a probabilistic complication of the former, which is also able to produce synergism or antagonism.

Crain et al. (2008) and Piggott et al. (2016) further shaped these concepts and stressed that the direction of the interaction is as important as the interaction, per se. The stressor A (Fig. 1.2) may decrease or increase the response with respect to the control. The same for the stressor B. Therefore the combined effect of stressors A and B may occur in the same or in opposite directions (e.g., positive or negative) than those predicted based on their independent effects (Fig. 1.3).

Fig. 1.3 Types of interactions between stressors. The figure shows the difference between a control (CT), a stressor A, a stressor B, or the interation of the two stressors (A + B). The interaction types (additive, synergistic, and antagonistic) depend on A + B response and may be double negative (Panel A), opposing (Panel B), or double positive (Panel C). From Crain, C. M., Kroeker, K., Halpern, B. S., 2008. Interactive and cumulative effects of multiple human stressors in marine systems. Ecol. Lett. 11, 1304–1315. With Permission of Wiley and Sons.

There are few empirical evidences of these predictions. Among these, in an analysis of long-term changes in boreal lakes, Christensen et al. (2006) showed that the interactions between warming, drought, and acidification were driving changes in plankton communities. Whereas the interaction among these stressors was synergistic for consumers, it was antagonistic for producers, making it especially difficult to predict the consequences of global change. A meta-analysis of factorial experiments on the impacts of multiple stressors in freshwater, marine, and terrestrial communities (Darling and Côté, 2008) determined that animal mortality did not often derive synergistically from the combined action of two stressors, and only one-third of experiments displayed truly synergistic effects.

1.6 Ecosystem Responses to Multiple Stressors

The ecological effects of anthropogenic stressors depend, among others, on the intensity and duration of the stress. Short, low-magnitude effects, such as a short event of moderate nutrient pollution, can produce little changes in the communities because most species are adapted to withstand these changes. Furthermore, the changes tend to be transient, and the community can recover quickly after cessation of the disturbance. However, persistent disturbances can exert stronger effects, leading to structural simplification of the community. High-intensity disturbances can cause strong mortality, but the recovery of the community depends on factors such as the availability of refugia.

The sensitivity of organisms to a given stressor is largely determined by their physiological or ecological properties, of which traits can be used as a proxy. Physiological traits include characteristics such as the ability for detoxification or the capacity for osmotic regulation, whereas ecological traits include characteristics such as feeding types, reproductive strategies, or life histories. The presence or lack of these traits defines the potential resistance of an organism to a given stressor. Some traits, such as living in the water column or onto the water surface, make a difference in the exposure risk to a pollutant in water, for instance. Some other traits, such as the respiration mode (e.g., aerial, through gills, through the tegument, etc.) determine the resistance of organisms to anoxia (Dolédec and Statzner, 2010).

Other factors also modulate the potential response to a stressor. Habitat heterogeneity may increase resistance and resilience to stressors by facilitating refugia to organisms. In large floodplains of high-gradient rivers, for instance, the harsh natural conditions imposed by high-energy flows and sediment mobility may be further impaired by human-driven stressors (e.g., river fragmentation, hydraulic stress, or introduced species; Tockner et al., 2010). In these cases, some periods of low flow become windows of opportunity for the recovery of biological communities, facilitating their resistance to some of these anthropogenic stressors. The windows of opportunity may be environmental opportunities within the physical template of the system, when organisms may recover more easily than during periods of harsher