R

Hf

AC

a

ARRAA

KEFECEM

C

h1

Ecological Indicators 69 (2016) 722–729

Contents lists available at ScienceDirect

Ecological Indicators

jo ur nal ho me page: www.elsev ier .com/ locate / ecol ind

eview

ow is ecosystem health defined and measured? A critical review ofreshwater and estuarine studies

llyson O’Brien ∗, Kallie Townsend, Robin Hale, David Sharley, Vincent Pettigroveentre for Aquatic Pollution Identification and Management (CAPIM), School of Biosciences, University of Melbourne, Victoria, 3010, Australia

r t i c l e i n f o

rticle history:eceived 15 January 2016eceived in revised form 22 April 2016ccepted 2 May 2016vailable online 2 June 2016

eywords:cosystem integrityreshwaterstuarineommunitycologyanagement

a b s t r a c t

Assessing ecosystem health is an ongoing priority for governments, scientists and managers worldwide.There are several decades of scientific literature discussing ecosystem health and approaches to assessit, with applications to aquatic and terrestrial environments incorporating economic, environmental andsocial processes. We conducted a systematic review of studies that assess ecosystem health to update ourcurrent understanding of how ecosystem health is being defined, and provide new ideas and directionson how it can be measured. We focused the review on studies that used the term ‘ecosystem health’ or theequivalent terms ‘ecosystem integrity’, ‘ecosystem quality’ and ‘ecosystem protection’, in lotic freshwa-ter and estuarine environments, and examined how many of these included explicit definitions of whatecosystem health means for their study system. We collected information about the temporal and geo-graphical distribution of studies, and the types of indicators (biological, physical or chemical) used in theassessments. We found few studies clearly defined ecosystem health and justified the choice of indica-tors. Given the broad use of the term it seems impractical to have an overarching definition of ecosystemhealth, but rather an approach that is able to define and measure health on a case by case basis. A combi-nation of biological, physical and chemical indicators was commonly used to assess ecosystem health inboth estuarine and freshwater studies, with a strong bias towards fish and macroinvertebrate communitymetrics (e.g. diversity, abundance and composition). We found only two studies that simultaneously con-sidered both freshwater and estuarine sections of the ecosystem, highlighting the significant knowledge

gap in our understanding of the transfer of flow, nutrients and biota between the different systems—allkey factors that influence ecosystem health. This review is the first to combine knowledge from bothfreshwater and estuarine ecosystem assessments and critically review how aquatic ecosystem health isdefined and measured since the late-1990s, providing the basis for setting achievable management goalsrelating to ecosystem health into the future.

© 2016 Elsevier Ltd. All rights reserved.

ontents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7232. Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7233. Results & discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 724

3.1. Defining ecosystem health . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7243.2. Temporal and spatial distribution of studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7253.3. Indicators used to assess ecosystem health . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 726

4. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Acknowledgement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

∗ Corresponding author.E-mail address: allyson@unimelb.edu.au (A. O’Brien).

ttp://dx.doi.org/10.1016/j.ecolind.2016.05.004470-160X/© 2016 Elsevier Ltd. All rights reserved.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 728 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 728

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 728

l Indic

1

tsdeeuEtm2mners

foadeatrCpeoc

atla(t(ttct2

iw(SK(1onotfHcet2at(

A. O’Brien et al. / Ecologica

. Introduction

‘Ecosystem health’ is a common term used in environmen-al science and management as a way to describe the state of aystem relative to a desired management target or reference con-ition (Rapport, 1989 Schaeffer et al., 1988). Other definitions ofcosystem health have emphasized the integration of ecological,conomic and human processes (Rapport et al., 1998) and meas-res of sustainability and system resilience (Costanza et al., 1992).cosystem health can be linked to management and political objec-ives (Fairweather, 1999) and used in the context of environmental

anagement legislation and policies (Karr, 1990; Negus et al.,009), as well as defining different societal, recreational or com-ercial benefits (Karr, 1999). Although ecosystem health is not a

ew term, but one that has been discussed and defined in the lit-rature since the late 1980s, achieving a state or condition thateflects a healthy ecosystem is an ongoing priority for government,cientists and managers worldwide (Burger et al., 2006).

Assessing ecosystem health can be approached in several dif-erent ways (Rapport et al., 1998). It can involve the identificationf certain characteristics that indicate a healthy ecosystem, such as

river system that is free from algal blooms, has a high biologicaliversity or a particular biotic score (Di Battista et al., 2016; Pontt al., 2006). Alternatively, the health of an ecosystem can be char-cterized by ecosystem services that humans depend on, such ashe provision of clean drinking water (Keeler et al., 2012), nutrientecycling and maintenance of biodiversity (US National Researchouncil, 2005). Ecosystem health assessments may also involveredictions about food web functioning and different trophic lev-ls required to sustain a healthy ecosystem (Thompson et al., 2012),r an ecosystem’s ability to maintain structure and function whenonfronted with external stress (Mageau et al., 1995).

In aquatic ecosystems, there is now more pressure than ever forppropriate monitoring and management strategies that will main-ain ecosystem health. With over 50% of the world’s populationiving within three kilometers of freshwater (Kummu et al., 2012)nd many of the largest cities around the globe situated on estuariesBoesch, 2000; Johnston et al., 2015b), these ecosystems are con-inually under pressure from an array of anthropogenic stressorsHalpern et al., 2008; Vorosmarty et al., 2010). Contamination, habi-at degradation and non-indigenous species are among the mainhreats that affect aquatic ecosystem health through changes inommunity composition and biological diversity, ecological func-ioning and provisioning of ecosystem services (Tolkkinen et al.,016).

Assessing aquatic ecosystem health has been an ongoing themen the scientific literature for several decades, possibly originating

ith river health indicators such as the Index of Biotic IntegrityKarr, 1981, 1991) or even earlier with the development of theaprobic Index to assess water quality (Friedrich et al., 1992;olkwitz and Marsson, 1909) and biological surveys of whole river

Forbes and Richardson, 1913) and lake ecosystems (Thienemann,925). There are now many approaches for assessing the healthf freshwater and estuarine systems that use biological commu-ity measures either separately or in combination with chemicalr physical parameters. Multi-metric indices are useful in thathey can distill a lot of information and parameters onto a uni-orm scale that can be easily used to make management decisions.owever, they are not always transferrable between regions orountries, often depend on comparisons with appropriate refer-nce conditions and rarely provide information on what aspects ofhe ecosystem are responsible for changes in health (Borja et al.,

012; Pont et al., 2006). Measures of ecosystem function, suchs respiration, primary productivity, metabolism and decomposi-ion, are becoming more common in ecosystem health assessmentse.g. Bunn and Davies, 2000; Woodward et al., 2012). Combining

ators 69 (2016) 722–729 723

measures of ecosystem function with multi-metric indices andother indicators of ecosystem structure may provide a better rep-resentation of whole ecosystem health (Woodward et al., 2012).Further progress in the way ecosystem assessments are approachedmay continue to arise as molecular techniques, such as next-generation sequencing, become more cost-effective and new typesof biodiversity indicators are developed (Forbes and Richardson,1913; Zinger et al., 2014).

Considering the broad use of the term ‘ecosystem health’ andthe many different definitions and assessment approaches, it seemsunrealistic to have an overarching definition of ecosystem health(Costanza et al., 1992), but one defined relative to the systembeing studied (Dobbie and Negus, 2013). Palmer et al. (2005)highlight that endpoints to stream restoration are unlikely to beuniversally applied, and suggest that a ‘guiding image’ must bearticulated to describe the most dynamic, ecologically dynamicstate possible at a site, taking into account the likely influence ofirreversible anthropogenic changes, such as alterations to catch-ment hydrology and geomorphology, the installation of permanentinfrastructure (e.g. on floodplains or along coasts), and the spreadof invasive species that cannot be easily removed. In a similar vein,a range of information could be used to help determine what shouldreflect a ‘healthy ecosystem’, including historical information (e.g.photographs, maps, survey records), appropriate reference sites,analytical/process based models, stream classifications, or in someinstances common sense (Palmer et al., 2005).

Here, we present the results of a systematic review undertakento update our current understanding of how ecosystem health isdefined and measured in lotic freshwater and estuarine environ-ments. Our first aim was to identify studies seeking to measureecosystem health, and then to examine how many of these includedan explicit definition of what ecosystem health means for theirstudy system. Our second aim was to examine the spatial andtemporal distribution of studies that had the primary goal of mea-suring ecosystem health or equivalent (i.e. ecosystem integrity,quality or protection) in lotic ecosystems (i.e. freshwater, streams,rivers, waterway or estuaries), to examine if they are geograph-ically biased and whether these types of studies are becomingincreasingly common through time. Geographic biases can hinderconservation efforts, for example, if the location of research is mis-aligned with research needs or priorities (Lawler et al., 2006). Ourthird aim was to examine how ecosystem health is being assessed,in terms of the types of indicators that are being used (i.e. chemical,biological, physical), and for biological indicators, the level of orga-nization at which they are monitored (i.e. individual, population,community), and the taxonomic groups that are commonly stud-ied. We use the results of our study to critically evaluate ecosystemhealth assessments in lotic freshwater and estuarine environments,identify key knowledge gaps, and ultimately aid in the developmentof robust and appropriate approaches to assess ecosystem healthin the future.

2. Methods

The systematic review followed the method of Pickering andByrne (2014). We reviewed studies that assessed ecosystem healthin lotic freshwater and estuarine environments (rivers, streams andestuaries) which excluded marine, wetland, lake and lagoon sys-tems to keep the scope of the study manageable. Search termswere “ecosystem health” and similar phrases that are often usedinterchangeably; “ecosystem integrity”, “ecosystem quality” and

“ecosystem protection” plus freshwater or river* or stream* orwaterway or estuar*. These terms were used in a multi-databasesearch using Web of Science Core Collection, CABI: CAB Abstractsand Global Health, Current Contents Connect and SciELO Citation

7 l Indic

Ittbatwt‘f

eiwltc‘‘ItoahwbwticOslwitatreiiaoTah

t((emtphmio‘pcaoir

24 A. O’Brien et al. / Ecologica

ndex. There were no restrictions on the date range. We limitedhe search to peer-reviewed journal articles that could be accessedhrough the online database to minimize any bias created by authorackground knowledge and search patterns. Many bioassessmentnd biomonitoring studies have the purpose of assessing ecosys-em health (or quality, integrity or protection) and these studiesere captured in our literature search if they used these specific

erms. However, we did not include the terms ‘bioassessment’ orbiomonitoring’ in the literature search so that our search remainedocused on the central theme of our study.

It is possible studies that had the primary purpose of assessingcosystem health may have used the terms ‘river health’ or ‘streamntegrity’ or ‘estuary health’ instead of ‘ecosystem health’ and so

ould not have been selected by our search. To understand theikelihood of this affecting our search we did a comparison betweenhe number of records returned with what we think are otherommonly used phrases: ‘river health’ versus ‘ecosystem health’,waterway health’ versus ‘ecosystem health’, ‘estuary health’ versusecosystem health’, ‘stream integrity’ versus ‘ecosystem integrity’.n all cases, ‘ecosystem health’ or ‘ecosystem integrity’ returnedhe most number of records, and when we compared a sub-setf the records, around 20% were duplicated in both searches. Were confident that most studies that used different terms (i.e. riverealth, stream integrity or estuary health) were also identifiedith our search terms. If there were studies that were missed

y our search because they used different terms, we think theseere minor omissions and did not affect the main outcomes of

he review. A total of 2570 records were returned and importednto a reference management tool (Endnote version X6.0.1). Dupli-ates were deleted leaving 1300 records to be searched manually.ur focus was on lotic freshwater and estuarine environments and

o we excluded studies that were conducted in lakes, wetlands,agoons, mangroves and marine habitats as well any study that

as not a primary source of data (e.g. framework, management,ndex description, discussion papers and reviews). We recordedhe number of studies that specifically assessed ecosystem healthnd therefore used the term ‘ecosystem health’ or equivalenterms: ‘integrity’, ‘quality’ and ‘protection’. Studies that had indi-ect statements, such as ‘these results could have consequences forcosystem health’ were excluded. This information was typicallyn the final paragraph of the introduction in each paper and so,n most cases, to make this decision we accessed the paper. Welso recorded the number of studies that defined ecosystem healthr equivalent either within the introduction or method sections.he definition had to include specific predictions about how vari-bles, indicators or measures changed and/or related to ecosystemealth.

Information recorded from each study included the study sys-em (freshwater or estuary), the exact term used in the studyecosystem health, integrity, quality or protection), type of studyexperiment, survey or modelling) and measures used to assesscosystem health (chemical, physical or biological). Chemicaleasures used were further recorded as physico-chemistry (e.g.

urbidity, dissolved oxygen, pH, electrical conductivity, and tem-erature), nutrients, contaminants, (e.g. metals, pesticides andydrocarbons) or chlorophyll, with an ‘other’ category included foretrics that were less common (e.g. turbidity, E.coli and other faecal

ndicators). Physical measures were recorded as channel morphol-gy/dimension, flow, substrate, land use or vegetation, with an

other’ category to record the less common metrics (e.g. redoxotential, altitude, annual runoff and sediment deposition). A ‘Pro-essing’ category included indicators that could not be categorized

s physical or chemical, and were indicative of ecosystem functionr ecological processes, such as measures of primary productiv-ty, respiration, biochemical oxygen demand and decompositionates.

ators 69 (2016) 722–729

Biological measures were categorized according to levels ofbiological organization: individual, population or community.Individual measures included morphology, pathology, behavior,bioaccumulation/bio-concentration, survival, histological, molec-ular, larval settlement, growth and reproduction. Populationmeasures included abundance, reproductive output/fecundity,population growth and biomass. The majority of studies usedcommunity measures and so these were further categorized asmeasures of tolerance (e.g. Hilsenhoff, SIGNAL, Index of Biologi-cal Integrity), diversity (e.g. Shannon-Wiener, Simpson, Evenness),abundance (total number of taxa, number of individuals, numberof individuals in each taxonomic group), composition (% of eachtaxonomic group), biomass and traits (e.g. guild, functional group,native/introduced) (Aazami et al., 2015), with an ‘other’ categorythat included metrics that were less common (e.g. fish parasiteload, disease, catch per unit effort, stable isotope analysis). Thetaxonomic groups for each biological measure were recorded asbacteria, biofilms, zooplankton, algae (including macroalgae andphytoplankton), plants, macroinvertebrates or fish.

3. Results & discussion

3.1. Defining ecosystem health

We found less than 15% of studies that used the term ‘ecosystemhealth’ or ‘ecosystem integrity’ or ‘ecosystem quality’ or ‘ecosystemprotection’ had the primary aim of assessing it in lotic freshwa-ter and estuarine environments (n = 191 studies). Thirty-three ofthese studies were in estuaries, 156 were freshwater and two stud-ies in both the freshwater and estuarine sections of the ecosystem.‘Ecosystem health’ was the most commonly used term (127 studies)with fewer studies using the equivalent ‘ecosystem integrity’ (42studies), ‘ecosystem quality’ (18 studies) and ‘ecosystem protec-tion’ (4 studies). To reflect these results we use the term ‘ecosystemhealth’ throughout this manuscript to encompass all of these sim-ilar search terms.

The majority of studies used ecosystem health as a generaldescriptive term, often helping to put the study in a broader con-text with potential management applications but without explicitlyassessing ecosystem health (n = 1109). These studies typically usedsingle parameters, such as water quality, flow or single species(e.g. Hasenbein et al., 2013) as an indicator of ecosystem healthand used indirect statements such as ‘the present investigationsuggests that contamination of aquatic ecosystems with oil couldpotentially affect ecosystem health. . .’ based on toxicity in a singlespecies (Ramadass et al., 2014), ‘detection of the protozoa in fishmay also be a good sentinel for environmental contamination orecosystem health’ (Reid et al., 2010) or ‘our recommendations willensure ecosystem integrity’ (Aloo et al., 2013). Studies measuringthe presence of contaminants were also common and often referredto those contaminants having potential consequences for ecosys-tem health (e.g. Aboul-Kassim and Williamson, 2003; Marentetteet al., 2010). Only studies that specifically stated that they assessedecosystem health (n = 191) were considered further in this review.

Defining ecosystem health by having clear study objectives isimportant to determine if management targets have been achievedor whether actions should be adapted (Kolkwitz and Marsson,1909; Thienemann, 1925). This also requires an understandingof uncertainty or threshold limits that will trigger managementintervention if targets are not met (Friedrich et al., 1992). Only28 studies had clearly defined objectives and specific predictions

about parameters used to measure ecosystem health. Seven ofthese studies were in estuaries and 21 were in freshwater envi-ronments. These studies included clear predictions about howindicators used in the study were expected to change if the

A. O’Brien et al. / Ecological Indicators 69 (2016) 722–729 725

F h (or ‘f

emgan

ecFneiIadtat2sas(s

ig. 1. Geographic and temporal distribution of studies that assess ecosystem healtreshwater systems.

cosystem was healthy, such as, low occurrence of larval defor-ities (Diggins and Stewart, 1998), ecosystem health index score

reater than 85 (Davies et al., 2010), presence of planktonic largend juvenile stages of estuarine fish species (Ramos et al., 2012) oritrogen concentrations below 2 mg N/L (Wiegand et al., 2010).

There is no overall agreement on what it means to have a healthycosystem (Palmer and Febria, 2012), yet it is still necessary to havelear definitions of ecosystem health on a study-by-study basis.or example, characteristics like low concentrations of contami-ants and reduced biotic richness can indicate a severely degradedcosystem that is adjacent to protected forests, but may converselyndicate a reasonably healthy ecosystem in an urban environment.n our review, we found an absence of clear study objectives and

lack of justification about basic study design, including samplingesign, replication, choice of parameters and expected variabilityhat either would or would not be acceptable depending on man-gement targets. There is generally a bias in the literature towardshe development of ecosystem health indicators (e.g. Chiu et al.,011) and less discussion about the selection of reference sites (butee Borja et al., 2012), defining levels of uncertainty and accept-

ble limits (Dobbie and Clifford, 2015), meaningful effect sizes andampling protocols that will detect ecologically relevant changesDobbie and Negus, 2013). It is possible that decisions about whattudy design and parameters to use may be based on uncited

integrity’, ‘quality’ or ‘protection’, all used in the literature search) in estuarine and

government policies, previously published literature or reflect theauthor’s personal interests (Lindenmayer and Likens, 2010) ratherthan statistical procedures and ecological importance. Irrespectiveof the rationale, our review demonstrates the need for more care-ful planning and justification of approaches for ecosystem healthassessments in both freshwater and estuarine environments.

3.2. Temporal and spatial distribution of studies

The number of studies with the primary goal of assessing aquaticecosystem health has increased over time (Fig. 1). We found onlyfreshwater studies (n = 21) prior to 2004, with 136 freshwaterstudies, 33 estuarine and two in both environments since. The dom-inance of freshwater studies prior to 2004 most likely reflects theinfluence of some key papers describing the concept of ecosystemhealth (Downes et al., 2002; Karr, 1999; Rapport, 1989) and theearly development of the Index of Biotic Integrity for freshwaterfishes (Karr, 1981). The number of ecosystem health publicationshas steadily increased since the late 1990s and now includes estu-arine and freshwater, as well as marine environments (e.g. Barras

et al., 2014; Blomberg and Montagna, 2014). Although there arejournals dedicated to aquatic ecosystem health (Journal of AquaticEcosystem Health, Ecosystem Health, and Aquatic Ecosystem Health& Management) we found studies published across a wide range

726 A. O’Brien et al. / Ecological Indicators 69 (2016) 722–729

F indici comm

ot0(y

tAUibieCwPtfbiihAmttcaw

toUbmeb

ig. 2. Proportion of freshwater and estuarine studies that used (a) one, two or threendicators in the study, and (d) biological indicators at the individual, population or

f ecological, conservation and management journals (Supplemen-ary Table S1 in the online version at DOI: 10.1016/j.ecolind.2016.5.004). Ecological Indicators had the most number of publicationsn = 21) with over half of these studies published in the last threeears (n = 11).

Based on geographic biases found by other reviews on conserva-ion (Lawler et al., 2006) and biodiversity research (Trimble and vanarde, 2012), we expected there may be more publications from theS and Europe with fewer studies from Asia, Africa and South Amer-

ca. Instead, we found no strong bias in the distribution of studiesetween countries (Fig. 1). There were 35 countries represented

n this review and they all had around 80% freshwater and 20%stuarine studies. The exceptions to this were Korea, Canada andhina, which all had significant biases towards freshwater studies,hile Portugal and the UK had mostly estuarine studies (Fig. 1).

ublications from China and Korea were particularly prevalent inhe last three years (68% of studies from China and 62% of studiesrom Korea were published in 2013–2015). The geographic distri-ution of ecosystem health assessments should ideally be focused

n regions where this type of knowledge and research is mostmportant, such as areas where impacts from human pressures areigh and ecosystem health is most threatened (Trimble and vanarde, 2012). The increase in publications from China and Koreaost likely reflects the general increase in scientific output from

hese countries (Veugelers, 2010), but may also be an indicationhat new methods and approaches are being developed in theseountries. Overall, these results indicate a positive shift towardsddressing knowledge gaps and progressing our understanding ofhat ecosystem health means for different ecosystems.

Our review identified only two studies that assessed ecosys-em health across both the freshwater and estuarine sectionsf the ecosystem (Tolan and Nelson, 2009; Zhao et al., 2014).nderstanding the transfer of water, nutrients, contaminants and

iota between the two systems is important for the develop-ent of appropriate indicators and approaches to monitoring

cosystem health. Cross-system subsidies of energy, nutrient andiota is reasonably well-understood for terrestrial and freshwater

ators to measure ecosystem health, (b) physical indicators in the study, (c) chemicalunity level of organization.

(Giling et al., 2015) as well as terrestrial and marine environments(Conway-Cranos et al., 2015; Treasure et al., 2015) but, as ourreview highlights, there are still substantial knowledge gaps thatexist for freshwater-estuarine ecotones.

3.3. Indicators used to assess ecosystem health

In any monitoring and assessment program, logistical con-straints (e.g. time, cost) are likely to mean that not all possiblevariables can be monitored, and thus a subset of indicators mustbe selected that relate to the goals of the project (Cairns et al.,1993; Dale and Beyeler, 2001). In our review, most studies took theapproach of using a combination of biological, physical and chem-ical indicators to assess ecosystem health, with over 80% of studiesusing two or three types of indicators (Fig. 2a). There were few stud-ies that used only one type of indicator, although this was higherfor estuarine studies (39%) compared to freshwater studies (15%)(Fig. 2a).

The use of multiple indicator types in a study may provide abetter assessment, but this needs to be accompanied by clear pre-dictions about how they represent ecosystem health. Failing tojustify the selection of indicators is one of the main reasons moni-toring programs fail (Lindenmayer and Likens, 2010). Results fromthis review have already shown that there are generally not a pri-ori definitions of ecosystem health (Palmer et al., 2005) and inmany cases there is no clear understanding of indicator responsesto changes in ecosystem condition. Therefore, it is difficult to deter-mine which combination of indicators is likely to reliably evaluatethe state of the ecosystem and success of the program.

The types of physical and chemical indicators used in ecosystemhealth assessments were mostly flow and channel morphol-ogy/dimension (in freshwater studies only) (Fig. 2b) and nutrientsand physico-chemistry (in freshwater and estuarine studies)

(Fig. 2c). Biota, more often than not, are used as indicators toreflect the condition of aquatic ecosystems (Vilmi et al., 2016).This was reflected in our results, with 82% of studies includ-ing biological indicators at the community level of biological

A. O’Brien et al. / Ecological Indicators 69 (2016) 722–729 727

F mmun

oab(yt2ecocHda(uw(afiebpo1

twepmitTfofi

ig. 3. Proportion of freshwater and estuarine studies that used (a) the different co

rganization (Fig. 2d). Biological diversity, abundance, tolerance,nd composition were all common community metrics used acrossoth freshwater and estuarine systems to assess ecosystem healthFig. 3a). Community metrics has been successfully used for manyears either as single or multi-metric indices to represent ecosys-em health and inform management decisions (e.g. Chiu et al.,011). Community metrics are arguably the most ecologically rel-vant as they are more complete assessments of communities,ompared to individual or population metrics, and can be extrap-lated to the state of an ecosystem through changes in speciesomposition or food web structure (Attrill and Depledge, 1997).owever, detecting change at the community-level can often beifficult against highly variable background conditions (Gessnernd Chauvet, 2002). This is when individual or population metricse.g. individual toxicity tests or population growth) may be moreseful indicator of ecosystem health and possibly useful as ‘earlyarning’ management tools to prevent ecosystem degradation

Corsi et al., 2003). However, there are so few ecosystem healthssessments done at the individual and population level and evenewer across multiple levels of biological organization, so makingnformed decisions about the most relevant indicators to use incosystem health assessments is difficult. Understanding the linksetween different levels of biological organization rather than sim-ly extrapolating from one level to another continues to be an areaf research that needs further development (Attrill and Depledge,997; O’Brien and Keough, 2014).

Taxonomic groups used to assess ecosystem health were biasedowards fish and macroinvertebrates for both estuarine and fresh-ater studies (Fig. 3b). Fish are likely to be responsive to many

nvironmental changes. For example, abundance, diversity, andopulation demographic rates often decrease after changes in flowagnitude (Poff and Zimmerman, 2010). Macroinvertebrates live

n a wide variety of habitats, respond to disturbance, and are likelyo be sensitive to water quality and pollution (Sharley et al., 2008;ownsend et al., 2009). In addition, both of these taxonomic groups

orm the basis for biotic indices used in biomonitoring (e.g. Indexf Biotic Integrity for fish in the USA, SIGNAL index and AUSRIVASor macroinvertebrates in Australia, and biotic indices in estuar-es (Pinto et al., 2009)). However, conceptual links between these

ity metric and (b) taxonomic groups used in the ecosystem health assessments.

taxa and ecosystem health are often assumed rather than demon-strated, creating potential uncertainty about whether they are themost suitable indicators to be used. Our results highlight that thereare many parts of the biological community that are not beingmeasured, particularly microbiota (e.g. diatoms, microalgae, bacte-ria, fungi). Microbiota could have different sensitives to humandisturbance and could be equally as good or better as indicatorsof ecosystem health (Chen et al., 2016). It may be the case thatmultiple taxonomic groups are routinely monitored to providethe most accurate assessment of ecosystem health (Vilmi et al.,2016). Rapidly developing new technologies and environmentalDNA monitoring tools (Rees et al., 2014) will enable more flexi-ble and cost-effective ways to sample biological communities, sowe expect that microbiota and other cryptic taxa that previouslyrequired a high level of taxonomic expertise and sampling effort canbe included in more comprehensive ecosystem health assessmentsin the future.

Thirty-five studies (two estuarine and 33 freshwater) includedmeasures of ecosystem processes (e.g. productivity, respiration,photosynthesis, litter breakdown). Although the effects of anthro-pogenic stressors on ecosystem functioning has been demonstrated(Johnston et al., 2015a) and the importance of including meas-ures of ecosystem function in monitoring programs documented(Palmer and Febria, 2012), our results show there is still an empha-sis on measuring traditional community structure and compositionmetrics in particular taxonomic groups (i.e. fish and macroinverte-brates). Furthermore, physical and chemical stressors contributingto ecosystem decline are often overlooked in ecosystem healthassessments. For example, we found that few studies measuredcontaminants (excluding nutrients) in their assessments of ecosys-tem health (3% were from estuarine and 13% from freshwatersystems). Contaminants are likely to be one of the key stressorsaffecting ecosystem health but they are not monitored in manystudies. This may be because contaminants are not specificallydefined as a threat and therefore not monitored, or they may be

omitted from the program due to lack of expertise or high costof sample analysis. However, contamination is recognized as oneof the major threats to biodiversity worldwide, along with cli-mate change, habitat loss and exotic species (Lawler et al., 2006).

7 l Indic

Tstc

4

bwstcphhntomago

dtwmaodfae

e(siaitt(aglsuiitteatome

A

tL

28 A. O’Brien et al. / Ecologica

herefore, the risk of contaminants affecting ecosystem healthhould be routinely considered and, if possible, included in ecosys-em health assessments, with advances in portable analytical toolsritical for this to occur (Gałuszka et al., 2015; Vrana et al., 2005).

. Conclusions

Given many ecosystems globally have been irreversibly changedy humans, understanding the limits for ecosystem improvementithin a given system is critical when developing targets for con-

ervation and restoration. It is also becoming increasing importanthat targets are in line with social and cultural expectations of theommunity. Without clear definitions of ecosystem health it is notossible to set targets and assess whether management actionsave been effective. This review has revealed that many ecosystemealth assessments in freshwater and estuarine environments doot define what a healthy ecosystem is or clearly state their objec-ives in terms of measuring ecosystem health, or justify their choicef indicator/s. Future ecosystem health assessments will need toore carefully set study objectives, and justify choice of indicators

nd sampling design. This will lead to better outcomes that are tar-eted towards not only environmental goals, which was the focusf this review, but also social, culture and economic values.

This study revealed a strong bias toward biological communityata for fish and invertebrates, with few studies focusing on otheraxonomic groups such as algae, bacteria and other microbiota,hich we feel will become increasingly important and easier toeasure with advances in molecular technologies. Given estuaries

nd freshwater systems are intrinsically linked through transfersf water, nutrients, contaminants and biota, this review clearlyemonstrates more studies are required that consider both thereshwater and estuarine components simultaneously. This willllow environmental managers to better implement whole riverinecosystem management strategies.

Defining realistic goals is an important early step for everycosystem health assessment, as it is in for any monitoring programDownes et al., 2002). The definition should (1) explicitly define thepatial extent of the ecosystem, which in the case of the studies usedn this review, be the river or stream reach or extent of the estuary;nd (2) reflect the diversity of current and historical conditions,ncluding, geology, land use, climate, vegetation, and species dis-ributions. Conceptual models would be useful tools at this stageo help define the inputs and outputs to the specific ecosystemLindenmayer and Likens, 2010), and to evaluate the likely scalest which these operate. This information should then be used touide the selection of appropriate indicators, noting that there isikely to be no single indicator that can unambiguously demon-trate a healthy ecosystem (Boulton, 1999). We suggest indicatorssed in ecosystem health assessments include at least one biolog-

cal indicator species or group plus other chemical and/or physicalndicators chosen to reflect the key ecological linkages of the sys-em (identified by conceptual model), the spatial scales over whichhey operate, and the objectives of the assessment, such as, to detectarly warning signs of degradation, or assess and diagnose compli-nce (Cairns et al., 1993). Defining ecosystem health and using thiso guide the selection of indicators, will hopefully aid in the rigor-us collection and effective communication of information to policyakers and the general public and help guide the management of

cosystems (Boulton, 1999).

cknowledgement

We acknowledge funding support from Melbourne Water andhe Australian Research Council under Projects LP140100343 andP140100565.

ators 69 (2016) 722–729

References

Aazami, J., Sari, A.E., Abdoli, A., Sohrabi, H., Brink, P.J.v.d., 2015. Assessment ofecological quality of the Tajan River in Iran using a MultimetricMacroinvertebrate Index and species traits. Environ. Manag. 56.

Aboul-Kassim, T.A.T., Williamson, K.J., 2003. Forensic analysis and genetic sourcepartitioning model for Portland harbor contaminated sediments. J. Environ. Inf.1, 58–75.

Aloo, P.A., Oyugi, D.O., Morara, G.N., Owuor, M.A., 2013. Recent changes in fishcommunities of the equatorial Lake Naivasha, Kenya. Int. J. Fish. Aquac. 5.

Attrill, M.J., Depledge, M.H., 1997. Community and population indicators ofecosystem health: targeting links between levels of biological organisation.Aquat. Toxicol. 38, 183–197.

Barras, C., Jorissen, F.J., Labrune, C., Andral, B., Boissery, P., 2014. Live benthicforaminiferal faunas from the French Mediterranean Coast: towards a newbiotic index of environmental quality. Ecol. Indic. 36, 719–743.

Blomberg, B.N., Montagna, P.A., 2014. Meta-analysis of Ecopath models revealssecondary productivity patterns across the Gulf of Mexico. Ocean Coast.Manag. 100, 32–40.

Boesch, D.F., 2000. Measuring the health of the Chesapeake Bay: towardintegration and prediction. Environ. Res. 82, 134–142.

Borja, A., Dauer, D.M., Gremare, A., 2012. The importance of setting targets andreference conditions in assessing marine ecosystem quality. Ecol. Indic. 12,1–7.

Boulton, A.J., 1999. An overview of river health assessment: philosophies, practice,problems and prognosis. Freshw. Biol. 41, 469–479.

Bunn, S.E., Davies, P.M., 2000. Biological processes in running waters and theirimplications for the assessment of ecological integrity. Hydrobiologia 422,61–70.

Burger, J., Tsipoura, N., Gochfeld, M., Greenberg, M.R., 2006. Ecologicalconsiderations for evaluating current risk and designing long-termstewardship on Department of Energy Lands. In: Leschine, T.M. (Ed.),Long-Term Management of Contaminated Sites. , pp. 139–162.

Cairns, J., McCormick, P.V., Niederlehner, B.R., 1993. A proposed framework fordeveloping indicators of ecosystem health. Hydrobiologia 263, 1–44.

Chen, X., Zhou, W., Pickett, S.T.A., Li, W., Han, L., Ren, Y., 2016. Diatoms are betterindicators of urban stream conditions: a case study in Beijing, China. Ecol.Indic. 60, 265–274.

Chiu, G.S., Guttorp, P., Westveld, A.H., Khan, S.A., Jun, L., 2011. Latent health factorindex: a statistical modeling approach for ecological health assessment.Environmetrics 22, 243–255.

Conway-Cranos, L., Kiffney, P., Banas, N., Plummer, M., Naman, S., MacCready, P.,Bucci, J., Ruckelshaus, M., 2015. Stable isotopes and oceanographic modelingreveal spatial and trophic connectivity among terrestrial, estuarine, andmarine environments. Mar. Ecol. Prog. Ser. 533, 15–28.

Corsi, I., Mariottini, M., Sensini, C., Lancini, L., Focardi, S., 2003. Fish as bioindicatorsof brackish ecosystem health: integrating biomarker responses and targetpollutant concentrations. Oceanol. Acta 26, 129–138.

Costanza, R., Norton, B., Haskell, B., 1992. Ecosystem Health: New Goals forEnvironmental Management. Island Press.

Dale, V.H., Beyeler, S.C., 2001. Challenges in the development and use of ecologicalindicators. Ecol. Indic. 1, 3–10.

Davies, P.E., Harris, J.H., Hillman, T.J., Walker, K.F., 2010. The sustainable riversaudit: assessing river ecosystem health in the Murray-Darling Basin, Australia.Mar. Freshw. Res. 61, 764–777.

Di Battista, T., Fortuna, F., Maturo, F., 2016. Environmental monitoring throughfunctional biodiversity tools. Ecol. Indic. 60, 237–247.

Diggins, T.P., Stewart, K.M., 1998. Chironomid deformities, benthic communitycomposition, and trace elements in the Buffalo River (New York) Area ofConcern. J. N. Am. Benthol. Soc. 17, 311–323.

Dobbie, M.J., Clifford, D., 2015. Quantifying uncertainty in environmental indices:an application to an estuarine health index. Mar. Freshw. Res. 66, 95–105.

Dobbie, M.J., Negus, P., 2013. Addressing statistical and operational challenges indesigning large-scale stream condition surveys. Environ. Monit. Assess. 185,7231–7243.

Downes, B.J., Barmuta, L.A., Fairweather, P.G., Faith, D.P., Keough, M.J., Lake, P.S.,Mapstone, B.D., Quinn, G.P., 2002. Monitoring Ecological Impacts: Conceptsand Practice in Flowing Water. Cambridge University Press, New York, USA.

Fairweather, P.G., 1999. State of environment indicators of ‘river health’: exploringthe metaphor. Freshw. Biol. 41, 211–220.

Forbes, S.A., Richardson, R.E., 1913. Studies on the biology of the Upper IllinoisRiver III. Bull. Ill. State Libr. Nat. Hist. 9, 481–574.

Friedrich, G., Chapman, D., Beim, A., 1992. The use of biological material. In:Chapman, D. (Ed.), Water Quality Assessments—A Guide to Use of Biota,Sediments and Water in Environmental Monitoring. Chapman and Hill, p. 651.

Gałuszka, A., Migaszewski, Z.M., Namiesnik, J., 2015. Moving your laboratories tothe field—Advantages and limitations of the use of field portable instrumentsin environmental sample analysis. Environ. Res. 140, 593–603.

Gessner, M.O., Chauvet, E., 2002. A case for using litter breakdown to assessfunctional stream integrity. Ecol. Appl. 12, 498–510.

Giling, D.P., Mac Nally, R., Thompson, R.M., 2015. How might cross-System

subsidies in riverine networks be affected by altered flow variability?Ecosystems 18, 1151–1164.

Halpern, B.S., Walbridge, S., Selkoe, K.A., Kappel, C.V., Micheli, F., D’Agrosa, C.,Bruno, J.F., Casey, K.S., Ebert, C., Fox, H.E., Fujita, R., Heinemann, D., Lenihan,H.S., Madin, E.M.P., Perry, M.T., Selig, E.R., Spalding, M., Steneck, R., Watson, R.,

l Indic

H

J

J

K

K

K

KK

K

K

L

L

M

M

N

O

P

P

P

P

P

R

A. O’Brien et al. / Ecologica

2008. A global map of human impact on marine ecosystems. Science 319,948–952.

asenbein, M., Komoroske, L.M., Connon, R.E., Geist, J., Fangue, N.A., 2013.Turbidity and salinity affect feeding performance and physiological stress inthe endangered delta smelt. Integr. Comp. Biol. 53, 620–634.

ohnston, E.L., Mayer-Pinto, M., Crowe, T.P., 2015a. Chemical contaminant effectson marine ecosystem functioning. J. Appl. Ecol. 52, 140–149.

ohnston, E.L., Mayer-Pinto, M., Hutchings, P.A., Marzinelli, E.M., Ahyong, S.T.,Birch, G., Booth, D.J., Creese, R.G., Doblin, M.A., Figueira, W., Gribben, P.E.,Pritchard, T., Roughan, M., Steinberg, P.D., Hedge, L.H., 2015b. Sydney Harbour:what we do and do not know about a highly diverse estuary. Mar. Freshw. Res.66, 1073–1087.

arr, J.R., 1981. Assessment of biotic integrity using fish communities. Fisheries 6,21–27.

arr, J.R., 1990. Biological integrity and the goal of environmental legislation:lessons for conservation biology. Conserv. Biol. 4, 244–251.

arr, J.R., 1991. Biological integrity: a long-neglected aspect of water resourcemanagement. Ecol. Appl. 1, 66–84.

arr, J.R., 1999. Defining and measuring river health. Freshw. Biol. 41, 221–234.eeler, B.L., Polasky, S., Brauman, K.A., Johnson, K.A., Finlay, J.C., O’Neill, A., Kovacs,

K., Dalzell, B., 2012. Linking water quality and well-being for improvedassessment and valuation of ecosystem services. Proc. Natl. Acad. Sci. 109,18619–18624.

olkwitz, R., Marsson, M., 1909. Ökologie der tierischen Saprobien Beiträge zurLehre von der biologischen Gewässerbeurteilung. 2, 152.

ummu, M., de Moel, H., Porkka, M., Siebert, S., Varis, O., Ward, P.J., 2012. Lost food,wasted resources: global food supply chain losses and their impacts onfreshwater, cropland, and fertiliser use. Sci. Total Environ. 438, 477–789.

awler, J.J., Aukema, J.E., Grant, J.B., Halpern, B.S., Kareiva, P., Nelson, C.R., Ohleth,K., Olden, J.D., Schlaepfer, M.A., Silliman, B.R., Zaradic, P., 2006. Conservationscience: a 20-year report card. Front. Ecol. Environ. 4, 473–480.

indenmayer, D., Likens, G.E., 2010. Effective ecological monitoring Collingwood,Vic.: CSIRO Publishing; London: Earthscan, c2010.

ageau, M.T., Costanza, R., Ulanowicz, R.E., 1995. The development and initialtesting of a quantitative assessment of ecosystem health. Ecosyst. Health 1,201–213.

arentette, J.R., Gooderham, K.L., McMaster, M.E., Ng, T., Parrott, J.L., Wilson, J.Y.,Wood, C.M., Balshine, S., 2010. Signatures of contamination in invasive roundgobies (Neogobius melanostomus): A double strike for ecosystem health?Ecotoxicol. Environ. Saf. 73, 1755–1764.

egus, P., Moller, G., Blessing, J., Davis, L., Marshall, J., Dobbie, M.J., 2009. Streamand Estuary Assessment Program—an Assessment Framework for RiverineEcosystems. Queensland Department of Environment and ResourceManagement.

’Brien, A.L., Keough, M.J., 2014. Ecological responses to contamination: ameta-analysis of experimental marine studies. Environ. Pollut. 195,185–191.

almer, M.A., Febria, C.M., 2012. The heartbeat of ecosystems. Science 336,1393–1394.

almer, M.A., Bernhardt, E.S., Allan, J.D., Lake, P.S., Alexander, G., Brooks, S., Carr, J.,Clayton, S., Dahm, C.N., Shah, J.F., Galat, D.L., Loss, S.G., Goodwin, P., Hart, D.D.,Hassett, B., Jenkinson, R., Kondolf, G.M., Lave, R., Meyer, J.L., O’Donnell, T.K.,Pagano, L., Sudduth, E., 2005. Standards for ecologically successful riverrestoration. J. Appl. Ecol. 42, 208–217.

into, R., Patricio, J., Baeta, A., Fath, B.D., Neto, J.M., Marques, J.C., 2009. Review andevaluation of estuarine biotic indices to assess benthic condition. Ecol. Indic. 9,1–25.

off, N.L., Zimmerman, J.K.H., 2010. Ecological responses to altered flow regimes: aliterature review to inform the science and management of environmentalflows. Freshw. Biol.

ont, D., Hugueny, B., Beier, U., Goffaux, D., Melcher, A., Noble, R., Rogers, C., Roset,

N., Schmutz, S., 2006. Assessing river biotic condition at a continental scale: aEuropean approach using functional metrics and fish assemblages. J. Appl. Ecol.43, 70–80.

amadass, K., Megharaj, M., Venkateswarlu, K., Naidu, R., 2014. Toxicity andoxidative stress induced by used and unused motor oil on freshwater

ators 69 (2016) 722–729 729

microalga, Pseudokirchneriella subcapitata. Environ. Sci. Pollut. Res. 22,8890–8901.

Ramos, S., Amorim, E., Elliott, M., Cabral, H., Bordalo, A.A., 2012. Early life stages offishes as indicators of estuarine ecosystem health. Ecol. Indic. 19, 172–183.

Rapport, D.J., Costanza, R., McMichael, A.J., 1998. Assessing ecosystem health.Trends Ecol. Evol. 13, 397–402.

Rapport, D.J., 1989. What consitutes ecosystem health? Perspect. Biol. Med. 33,120–132.

Rees, H.C., Maddison, B.C., Middleditch, D.J., Patmore, J.R.M., Gough, K.C., 2014.Review: The detection of aquatic animal species using environmental DNA—areview of eDNA as a survey tool in ecology. J. Appl. Ecol. 51, 1450–1459.

Reid, A., Lymbery, A., Ng, J., Tweedle, S., Ryan, U., 2010. Identification of novel andzoonotic Cryptosporidium species in marine fish. Vet. Parasitol. 168, 190–195.

Schaeffer, D.J., Henricks, E.E., Kerster, H.W., 1988. Ecosystem health: Measuringecosystem health. Environ. Manage. 12, 445–455.

Sharley, D.J., Hoffmann, A.A., Pettigrove, V., 2008. Effects of sediment quality onmacroinvertebrates in the Sunraysia region of the Murray–Darling RiversAustralia. Environ. Pollut. 156, 689–698.

Thienemann, A., 1925. Die binnengewässer mitteleuropas Binnengewässer. 1,1-255.

Thompson, R.M., Brose, U., Dunne, J.A., Hall, R.O.J., Hladyz, S., Kitching, R.L.,Martinez, N.E., Rantala, H., Romanuk, T.N., Stouffer, D.B., Tylianakis, J.M., 2012.Food web: reconciling the structure and function of biodiversity. Trends Ecol.Evol. 27, 689–697.

Tolan, J.M., Nelson, J.M., 2009. Relationships among nekton assemblage structureand abiotic conditions in three Texas tidal streams. Environ. Monit. Assess.159, 15–34.

Tolkkinen, M.J., Mykra, H., Virtanen, R., Tolkkinen, M., Kauppila, T., Paasivirta, L.,Muotka, T., 2016. Land use impacts on stream community composition andconcordance along a natural stress gradient. Ecol. Indic. 62, 14–21.

Townsend, K.R., Pettigrove, V.J., Carew, M.E., Hoffmann, A.A., 2009. The effects ofsediment quality on benthic macroinvertebrates in the River Murray Australia.Mar. Freshw. Res. 60, 70–82.

Treasure, A.M., Ruzicka, J.J., Moloney, C.L., Gurney, L.J., Ansorge, I.J., 2015. Land-seainteractions and consequences for sub-Antarctic marine food webs.Ecosystems 18, 752–768.

Trimble, M.J., van Aarde, R.J., 2012. Geographical and taxonomic biases in researchon biodiversity in human-modified landscapes. Ecosphere 3.

U.S. National Research Council, 2005. Valuing Ecosystem Services: Toward BetterEnvironmental Decision Making. The National Academies Press, Washington D.C.

Veugelers, R., 2010. Towards a multipolar science world: trends and impact.Scientometrics 82, 439–456.

Vilmi, A., Karjalainen, S.M., Nokela, T., Tolonen, K., Heino, J., 2016. Unravelling thedrivers of aquatic communities using disparate organismal groups anddifferent taxonomic levels. Ecol. Indic., 108.

Vorosmarty, C.J., McIntyre, P.B., Gessner, M.O., Dudgeon, D., Prusevich, A., Green, P.,Glidden, S., Bunn, S.E., Sullivan, C.A., Liermann, C.R., Davies, P.M., 2010. Globalthreats to human water security and river biodiversity. Nature 467, 555–561.

Vrana, B., Allan, I.J., Greenwood, R., Mills, G.A., Dominiak, E., Svensson, K., Knutsson,J., Morrison, G., 2005. Passive sampling techniques for monitoring pollutants inwater. TrAC Trends Anal. Chem. 24, 845–868.

Wiegand, J., Raffaelli, D., Smart, J.C.R., White, P.C.L., 2010. Assessment of temporaltrends in ecosystem health using an holistic indicator. J. Environ. Manag. 91,1446–1455.

Woodward, G., Gessner, M.O., Giller, P.S., Gulis, V., Hladyz, S., Lecerf, A., Malmqvist,B., McKie, B.G., Tiegs, S.D., Cariss, H., Dobson, M., Elosegi, A., Ferreira, V., Graca,M.A.S., Fleituch, T., Lacoursiere, J.O., Nistorescu, M., Pozo, J., Risnoveanu, G.,Schindler, M., Vadineanu, A., Vought, L.B.M., Chauvet, E., 2012.Continental-scale effects of nutrient pollution on stream ecosystemfunctioning. Science 336, 1438–1440.

Zhao, Y.W., Xu, M.J., Yu, L., Yin, X.A., Zhang, Y., 2014. Identifying sensitive indices inthe response of aquatic biota to landscape pattern changes: a case study of theTaizi River Basin in North China. River Res. Appl. 30, 1013–1023.

Zinger, L., Boetius, A., Ramette, A., 2014. Bacterial taxa-area and distance-decayrelationships in marine environments. Mol. Ecol. 23, 954–964.