Hp

RA

a

ARR1A

KEEAFFrD

1

isr2hetaatfisd

1h

Ecological Indicators 39 (2014) 179–188

Contents lists available at ScienceDirect

Ecological Indicators

j o ur na l ho me page: www.elsev ier .com/ locate /eco l ind

ow do low magnitudes of hydrologic alteration impact riverine fishopulations and assemblage characteristics?

obert J. Rolls ∗, Angela H. Arthingtonustralian Rivers Institute, Griffith University, Nathan, Queensland 4111, Australia

r t i c l e i n f o

rticle history:eceived 15 August 2013eceived in revised form0 December 2013ccepted 14 December 2013

eywords:nvironmental flowsLOHA (Ecological Limits of Hydrologiclteration)low classificationlow alteration–ecological responseelationshipsam management

a b s t r a c t

Water managers need quantitative information on the effects of hydrologic alteration on aquatic biotato guide ecologically sensitive water management strategies such as water releases from dams. A keygap in the global research literature is determining whether low levels of hydrologic alteration havesignificant effects on fish populations and assemblage characteristics. This study quantified patterns offish response to flow regime alteration in a sub-tropical region where many rivers have regulated flowregimes but 57% of ecologically relevant flow metrics have changed by <20%. We tested for flow regulationeffects on 17 (univariate and multivariate) response variables representing fish population abundanceand assemblage characteristics using a field design based on the environmental flow assessment frame-work known as ELOHA (Ecological Limits of Hydrologic Alteration). Ecological response variables that arereadily quantified and sensitive to variation and alteration in flow regimes are critical to the applicationof environmental flow frameworks such as ELOHA. In this study only three of 17 response variables rep-resenting fish population abundance and assemblage attributes showed significant differences betweenregulated and unregulated reaches (densities of both Pseudomugil signifier and Melanotaenia duboulayi,and fish assemblage composition). Effects associated with flow regulation were most evident where his-

torically intermittent flow regimes have become more perennial as a consequence of managed waterreleases from dams. Our study provides positive evidence that dams and regulated flow regimes can bemanaged with sensitivity such that there are few significant changes in populations of most fish species,and little change in fish assemblage characteristics. However, it must be cautioned that the magnitude offlow regime alteration may interact with the duration of exposure (i.e. years to decades) such that otherecological impacts emerge over time as species and assemblages adjust to altered flow regimes.

. Introduction

Human impacts on river flow regimes alter the magnitude, tim-ng, frequency, duration and variability of flow events, with thepecific changes depending on the nature and magnitude of flowegime manipulation (Poff et al., 1997; Magilligan and Nislow,005). Such changes to the natural flow regime contribute to alteredabitat structure, impacts on life history processes, growth andcosystem production, and the facilitation of non-native biota;hese impacts contribute to changes in aquatic biodiversity (Bunnnd Arthington, 2002; Naiman et al., 2008; Strayer, 2010). Differentquatic biota, such as fish, macroinvertebrates and riparian vegeta-ion, respond in different ways to flow regime alteration, however

sh have been identified as the only taxonomic group that con-istently responds negatively (in terms of population abundance,emographics and species composition) to flow magnitude change

∗ Corresponding author. Tel.: +61 428 332 787; fax: +61 7 3735 3819.E-mail address: r.rolls@griffith.edu.au (R.J. Rolls).

470-160X/$ – see front matter © 2013 Elsevier Ltd. All rights reserved.ttp://dx.doi.org/10.1016/j.ecolind.2013.12.017

© 2013 Elsevier Ltd. All rights reserved.

(Poff and Zimmerman, 2010). Identifying the characteristics of therelationships between the flow regime and ecological patterns andprocesses is necessary to determine levels or thresholds where flowalteration is associated with ecological change (sensu Underwoodet al., 2000; Downes, 2010). If clear relationships between typesand degrees of flow alteration and ecological response can be found,then this information can make a significant contribution to defin-ing environmental flow rules such as water releases from damsand/or appropriate levels of water extraction.

Knowledge of the environmental flow requirements of aquaticspecies and biological assemblages is derived from a rangeof approaches that relate flow regime alterations to ecologicalresponses (Tharme, 2003; Acreman and Dunbar, 2004). The Ecolog-ical Limits of Hydrologic Alteration (ELOHA) framework is based ongrouping (i.e. classifying) rivers at regional scales based on varia-tion in flow regimes, and then testing for differences in ecological

responses to flow regime alterations within those classes to pro-duce flow alteration–ecological response relationships for eachflow class (Poff et al., 2010). In theory, flow regime classes iden-tified by classification may be regarded as management units that

1 ogical

sifltmva

aaflriatHabcsasl(wfpss

iflaippbflbeUluoTfir

latr2rcrtirtcaclp

80 R.J. Rolls, A.H. Arthington / Ecol

hare ecological attributes and therefore could be managed in sim-lar ways with regard to the design and allocation of environmentalows (Arthington et al., 2006; Kennard et al., 2010). However, forhe ELOHA approach to be useful in the development of environ-

ental flow rules/regimes, it is necessary to identify ecologicalariables that are both readily quantified and reflect population andssemblage level responses to different types of flow alteration.

Ecological evidence indicates that the effects of flow regimelteration on freshwater fish occur as a sequence of processes thatre dependent on the magnitude and/or duration of exposure toow alterations (Table 1). For example, lower population size inegulated versus unregulated (or less regulated) rivers (stage 1n the sequence) has been attributed to the effect of flow regimelteration on population recruitment by modifying environmen-al conditions such as prey and habitat (e.g. Freeman et al., 2001;umphries et al., 2002). Increasing flow regime change expresseds monthly and annual flow deviations from natural conditions haseen associated with increasing abundance of non-native commonarp, Cyprinus carpio (stage 2), and declining richness of native fishpecies in the Murray-Darling Basin, Australia, thereby altering fishssemblage composition (stage 3) (Gehrke et al., 1995). Reducedtream flow variability as a consequence of regulation has beeninked with species extinction resulting in lower species diversitystage 4, Table 1), and this is particularly evident for fish speciesith preferences for specific hydraulic habitats such as shallow,

ast-velocity riffles (e.g. Meador and Carlisle, 2012). These examplesrovide evidence of patterns in fish populations and assemblagetructure that reflect the effects of flow regime alteration on aequence of ecological processes.

Many environmental flow methods are based on thedea/assumption that the magnitude of ecological response toow regulation is proportional to the magnitude of hydrologiclteration from pre-impact conditions. Evidence from studiesn rivers with high levels of flow regime change support thisrediction. For example, measures of fish abundance, demographicarameters and diversity decrease by at least 50% in response tooth decreased (−50% to −100%) and increased (+50% to 100%)ow magnitude (Poff and Zimmerman, 2010). Likewise it mighte expected that if hydrologic alterations are kept to low lev-ls there will be very small or no detectable ecological impact.nfortunately, the lack of published ecological response data for

ow to moderate levels of hydrologic alteration leaves a gap innderstanding of potential subtle impacts of flow regime changen fish populations and assemblages (Poff and Zimmerman, 2010).his knowledge gap is important from a management perspective,or example, in defining thresholds of ecological response to assistn establishing limits to water abstraction, or setting particularelease volumes from dams.

This study was designed to quantify the effects of relativelyow levels of hydrologic alteration on fish population size andssemblage characteristics within an ELOHA framework. Firstly, weested for differences in fish response variables between currentlyegulated and unregulated (i.e. sampling control, sensu Downes,010) reaches in rivers of different natural (pre-development) flowegime type (i.e. reference flow classes). By testing for ecologi-al differences between regulated versus unregulated referenceeaches, we can test the hypothesis that flow regulation initiateshe impact sequence described above (Table 1), and determinef such ecological changes differ across a range of reference flowegime classes. Secondly, we applied a modified survey design thatested for differences in fish response variables across flow regimelasses that incorporate different types and magnitudes of flow

lteration based on recent gauged flow records (i.e. current flowlasses), thereby testing the hypothesis that fish respond to hydro-ogic differences and regime shifts brought about by water releaseatterns from dams. By testing for differences using two different

Indicators 39 (2014) 179–188

experimental designs we can strengthen our inference of relation-ships between the magnitude and type of flow regime alterationand associated ecological changes in rivers of contrasting hydro-logical character.

2. Methods

2.1. Study area and flow regimes

This study was conducted between the South Coast andMary River catchments of South East Queensland (SEQ), Australia(see Arthington et al., 2012). Briefly, the climate is subtropicaland influenced by both tropical and temperate weather pat-terns. Rainfall predominantly occurs during the summer period ofDecember–March, and has a distinct increasing gradient from westto east (inland to the coast) of 800–1400 mm average per annum.SEQ is inhabited by ∼2.8 million people, predominantly in the GoldCoast, Brisbane and the Sunshine Coast regions.

Rivers in the study region generally have late summer–earlyautumn high discharge regimes, with periods of low dischargeoccurring from August to November (Pusey et al., 2004). However,temperate weather systems that produce winter rain in south-ern Australia may also produce significant rainfall in the studyarea from autumn to mid-winter. As the occurrence and inten-sity of summer and autumn–winter rainfall is irregular, dischargeregimes of rivers and streams in the region vary considerably. Theflow regimes of many rivers have been altered by the construc-tion and operation of dams and weirs, unsupplemented extraction(extraction of water from natural river flows, as opposed to supple-mented extraction, where explicit releases are made from storagesfor extraction downstream), inter- and intra-basin water transfers,and land use changes. Twenty-four dams >15 m wall height nowexist in SEQ, with total storage capacity of 38% of the total naturalmean annual runoff for the study region.

Two regional hydrological classifications have been undertakento explore patterns of flow regime variability and the effects of flowregulation in the study area (Arthington et al., 2012; Mackay et al.,in press). Each classification was based on 35 minimally redundanthydrologic metrics describing the major facets of the natural flowregime (magnitude, frequency, timing and duration of dischargeevents, and discharge variability/predictability; sensu Poff et al.,1997). The metrics analysed are known to be ecologically relevant,sensitive to hydrologic alterations due to human activities, andpotentially amenable to management through ecologically sensi-tive dam operations and water abstraction rules. The purpose ofundertaking a classification based on modelled pre-developmentflow data (i.e. the reference classification) was to represent vari-ation in flow regimes under near natural conditions (i.e. withoutthe effect of human alterations to hydrology caused by stream flowmanipulation, extraction and land use change). The aim of under-taking a separate classification based on gauged flow records (i.e.current conditions) was to quantify differences in flow regimes thatincorporate the influence of human alterations. Classification wasundertaken using model-based hierarchical agglomerative clus-tering based on Gaussian finite mixture models, as implementedin the Mclust package for R (Fraley and Raftery, 2008; R CoreDevelopment Team, 2010). Cluster analysis distinguished six refer-ence classes (based on modelled pre-development flow data) andfive current classes based on stream gauge data (Table 2). The sixthreference flow class was removed from further analysis because itcontained very few replicate river reaches. Specific details of the

methods and results of the flow classifications are presented inArthington et al. (2012), Mackay et al. (in press) and Supplemen-tary material S1. Briefly, six flow metrics discriminated betweenreference flow classes (mean annual 1-day minimum, mean annual

R.J. Rolls, A.H. Arthington / Ecological Indicators 39 (2014) 179–188 181

Table 1Predicted sequence of changes in fish populations and assemblages as a consequence of flow regime alterations. Definitions of all terms are provided in text. Directions (i.e.increases or decreases) in potential response variables are included.

Key consequence of flowregime alteration

Stage in sequence Potential ecological response Example references

Native species: alteration infish population size

1 Population abundance (±) Freeman et al. (2001) and Humphrieset al. (2002)

Non-native species:establishment andproliferation of non-nativespecies

2 Non-native species richness (+)Total abundance of non-native species(+)Proportion of species that arenon-native (+)Proportion of individuals that arenon-native (+)

Gehrke et al. (1995), Bunn andArthington (2002) and Poff andZimmerman (2010)

Altered assemblage abundanceand composition

3 Total density of all species inassemblage (±)Assemblage composition (±)

Gehrke et al. (1995), Bunn andArthington (2002) and Poff andZimmerman (2010)

Loss of sensitive species 4 Total species richness (±)e speces den

Bunn and Arthington (2002), Poff and

1aycdcoamaytm(

totfl(cis

THoch

leading to reduced diversity NativSpeci

-day maximum, mean daily flow in March and September, meannnual maximum discharge and mean number of zero flow days perear). The principal gradient separating reference classes was dis-harge magnitude, with classes 5 and 6 having substantially higherischarge magnitude per unit of catchment area than referencelasses 1–4. This hydrologic gradient reflects geographic patternsf rainfall in south-east Queensland (Bridges et al., 1990; Youngnd Dillewaard, 1999). For the current flow classification, six flowetrics discriminated between flow classes (mean annual 1-day

nd 30-day minimum flows, mean number of zero flow days perear, 10-year average return interval flood, high flow spell dura-ion based on 25th percentile exceedance flow, and constancy of

ean daily discharge based on Colwell’s index of flow constancyColwell, 1974).

In summary, each classification described a gradient of perennialo varying forms of intermittent flow regime (Table 2). Comparisonsf change (modelled versus gauged data) in the 35 flow metricshat form the basis of the flow classifications revealed that 57% ofow metrics have changed by less than 20% due to flow regulationArthington et al., 2012; Mackay et al., in press). This analysis indi-

ated that the magnitude of flow regime change (i.e. the changen multiple individual flow metrics) is relatively minor across thetudy area.

able 2ydrologic characteristics distinguishing flow regime classes identified in the reference (f South East Queensland (Arthington et al., 2012; Mackay et al., in press). Numerals indurrent flow classes. Numbers in brackets indicate the number of river reaches where flydrological metrics that distinguish each flow class.

Class number Reference classification Flow class description C

1 5 (2) Intermittent stable: low min,low max flow, low proportionof zero flow days

3

2 7 (3) Intermittent unpredictable: lowmin flow, low max flow,moderate proportion of zeroflow days

3

3 1 (0) Perennial: moderate min flow,low max flow, very lowproportion of zero flow days

7

4 2 (0) Highly intermittent: low minflow, low max flow, highproportion of zero flow days

2

5 5 (1) Intermittent: moderate minflow, moderate max flow,moderate proportion of zeroflow days

5

ies richness (−)sity (−)

Zimmerman (2010) and Meador andCarlisle (2012)

2.2. Sampling design and methods

Forty study sites were selected from 20 rivers (i.e. two ran-dom replicate sites in each river reach) that reflected the majorregional gradients of flow variability and flow regime alterationin SEQ (Table 2). Sites were selected close to stream flow gaugesand were predominantly located in upland catchments that wereminimally disturbed by land use and management practices closeto stream channels. However, influences of broader catchmentland-use (forestry, grazing, agriculture, residential and industrialuse) could not be eliminated entirely by site selection becausethis was focused on flow characteristics and site accessibility.Land-use patterns were therefore assessed and their influence onfish assemblages analysed using distance-based multivariate linearregression (Arthington et al., in press). The majority of catchmentarea (84%) upstream of study sites was in relatively natural condi-tion, such that land-use characteristics accounted for only a smallproportion of variation in species presence–absence and relativeabundance patterns of fish assemblages (Arthington et al., in press).

All sites were sampled three times at approximately four-month

intervals between July 2009 and May 2010. Fish were sampledusing multiple-pass backpack electrofishing and supplementaryseine netting at each site. Multiple pass electrofishing has been

modelled flow) classification and the current (gauged flow) classification for riversicate the number of separate river reaches sampled in each of the reference andow is regulated by a dam or large weir. See text and Supplementary Figure 1 for

urrent classification Flow class description

(2) Perennial: high min flow, low flood magnitude, shortduration of high flow pulses, no zero flow days, highconstancy of flow

(1) Highly intermittent-unpredictable summer: low minflow, low flood magnitude. Long duration of high flowpulses, high proportion of zero flow days, moderateconstancy of flow

(2) Rarely intermittent-unpredictable: moderate min flow,moderate flood magnitude, moderate duration of highflow pulses, low proportion of zero flow days, lowconstancy of flow

(1) Unpredictable: low min flow, high magnitude of largefloods, moderate duration of high flow pulses, highproportion of zero flow days, low constancy of flow

(0) Intermittent-unpredictable: moderate min flow, highmagnitude of large floods, short duration of high flowpulses, low proportion of zero flow days, lowconstancy of flow

1 ogical

sasnpthAwutwarspa4la

2

“tqs1aNoncner(aabib2satfl

2

dlsartiirtrrtu

82 R.J. Rolls, A.H. Arthington / Ecol

hown to improve the ability to detect real differences in fishssemblages other than variations due to inadequate or incon-istent sampling (e.g. Kennard et al., 2006). At each site, blockets were used to separate each habitat unit present (riffle, run,ool) and one person operated a Smith-Root LR-24 backpack elec-rofisher while another collected stunned fish with a dip net. Eachabitat was repeatedly sampled until no more fish were collected.dditional seine netting (10 m long × 1.5 m high × 5 mm mesh size)as undertaken where practical to ensure all species and individ-als were collected. Fish were identified to species and returnedo the site after sampling (apart from all non-native species thatere euthanased as required by sampling permits). Due to vari-

tions in geomorphology and river flow, not all sites comprisediffle, run and pool habitat units. During some sampling periods,ome sites were completely dry and therefore not sampled (5 sam-les). The density of each fish species or groups of species sampledt each site (based on available habitat units) was standardised to50 m2 of wetted site surface area to allow for meaningful popu-

ation and assemblage comparisons among flow regime conditionsnd sampling times.

.3. Fish response variables

From the resulting dataset, we calculated metrics (hereafterresponse variables”) of fish population and assemblage structurehat reflect the documented and predicted sequence of conse-uences of flow regime alteration (Table 1). We used the rawtandardised abundances (densities of individuals per 450 m2) of0 species that occurred in >25% of all samples to assess alter-tion of fish population sizes (abundances) (i.e. stage 1 of Table 1).on-native species richness (number of species), total densityf non-native species, the proportion of species that are non-ative and the proportion of individuals that are non-native werealculated to represent the establishment and proliferation of non-ative species (stage 2, Table 1). Total density of all species inach assemblage and assemblage composition were used as theesponse variable for assemblage level responses. The compositioni.e. presence and density) of the entire species assemblage (nativend non-native species) was used to test how patterns in speciesssemblages varied across the two separate flow classifications andetween regulated and unregulated reaches (stage 3, Table 1). Sim-

larity of assemblage composition between samples was calculatedased on the modified Gower (log base 2) measure (Anderson et al.,006). Finally, total species richness, native species richness andpecies density (the ratio of the number of all species: number ofll individuals) indicate patterns of species diversity and poten-ially, losses in species that may be unable to persist in particularow conditions.

.4. Data analysis

Patterns in each fish response variable were tested to compareifferences among flow classes, between regulated and unregu-

ated reaches and over time using a three-factor mixed modeltatistical design (Supplementary material S2). Flow classes (1–5)re fixed, as we were specifically interested in comparing the fishesponse variables across flow regimes of SEQ. The factor “regula-ion” (i.e. comparing currently unregulated and regulated reaches)s fixed and nested within each flow class. Nested factors are typ-cally considered random, however in the context of this studyegulated and unregulated reaches were deliberately chosen. Fur-hermore, the characteristics of regulated and unregulated flow

egimes are unique to each flow class and therefore not compa-able across classes. Sampling time (n = 3) was treated as randomo determine if patterns of fish response to flow variability and reg-lation varied across the three sampling times or were consistent

Indicators 39 (2014) 179–188

through time. Replicate sample sites from each river (n = 2) weretreated as random and formed the residual error term in the model(Supplementary material S2).

Statistical tests were carried out twice for each response vari-able to test the two main hypotheses of our study. Firstly, wetested patterns using the statistical design based on the referenceflow classification. The purpose of this analysis was to test if therewere any ecological differences between currently regulated andunregulated reaches that, under modelled reference conditions,had similar flow regimes. If there are differences between reachesthat are regulated and unregulated now, but were hydrologicallysimilar in the past, we can infer that these differences are due tothe effects of flow regime change (cf. Catford et al., 2011). Testingfor patterns across reference flow classes allowed us to determineif and how differences between unregulated and regulated reachesvaried according to flow regime class.

Secondly, we tested patterns in fish response variables usingan identical statistical design based on the current flow regimeclassification. For this test we expected that there would be signifi-cant differences in response variables across flow regime classesthat incorporate different types and magnitudes of flow alter-ation based on recent gauged flow records. We also expected thatthere would be no differences between regulated and unregulatedreaches within current flow classes because these reaches haveexperienced similar flow regimes based on flow class membership.If there are significant ecological differences between regulated andunregulated sites within a current flow class, then the ecologicalsystem at some stream reaches has not fully adjusted to the prevail-ing hydrological conditions, which classification has revealed arerelatively similar for both regulated and unregulated sites withineach current flow class. Alternatively, flow patterns and flow regu-lation may not be the only factors influencing fish population andassemblage patterns in rivers of our study area.

Events that occur seasonally (e.g. spawning) are likely to resultin temporal differences in fish response variables throughout theyear. Our experimental design tested for temporal variability of fishresponses by estimating the interaction between the factors “time”and “flow class”, “regulation” or “regulation (flow class)”. All testswere done using permutational analysis of variance (PERMANOVA;Anderson, 2001), which is useful for mixed model analyses and iscapable of both handling unbalanced designs and testing patternsin univariate or multivariate data. Differences in each univariateresponse variable (metric) were compared by Euclidean distanceand the resultant resemblance matrix tested using the statisti-cal design (type III sums of squares) run with 9999 permutations.This approach is analogous to traditional ANOVA, however has theadvantage of not requiring the assumption of normal distribution.Post hoc pairwise tests were carried out where differences betweenfactors of interest (flow class, regulation (flow class)) were signifi-cant. PERMANOVA also calculates the components of variation foreach test in the analysis and is useful for interpreting effect sizeof significant terms in the model. PERMANOVA was undertakenusing the PERMANOVA+ add-on package in PRIMER v6 (Clarke andGorley, 2006; Anderson et al., 2008).

3. Results

3.1. Fish assemblage characteristics

Thirty-five fish species were sampled across the study regionbetween 2009 and 2010. Six native species (Anguilla reinhardtii,

Hypeleotris galii, H. klunzingeri, Melanotaenia duboulayi, Retropinnasemoni and Tandanus tandanus) occurred in >50% of all samples.The 10 species that were collected in >25% of samples (A. reinhardtii,Gobiomorphus australis, H. galii, H. klunzingeri, H. sp. 1, M. duboulayi,

ogical

PGndo

3

AfcfidwsPPd2a1dflci(fr

dogncHcbUwbc

3

adehfll(tflrtt(

3

ssu

R.J. Rolls, A.H. Arthington / Ecol

seudomugil signifier, R. semoni, T. tandanus and the non-nativeambusia holbrooki) accounted for 85% of all fish sampled. Five non-ative species were sampled, and comprised <4% of all individualsetected. G. holbrooki was the most widespread non-native species,ccurring in 30% of samples (Supplementary material S3).

.2. Population patterns

When tested across the reference flow classes, densities of. reinhardtii, H. galii, H. sp. 1, P. signifier and G. australis dif-

ered significantly between classes and these differences wereonsistent among sampling times (Table 3). Pseudomugil signi-er and M. duboulayi were the only species with densities thatiffered significantly between unregulated and regulated reachesithin reference flow classes on all three sampling occasions (non-

ignificant regulation (flow class) × sampling time interaction).ost hoc pairwise tests revealed that differences in abundances of. signifer between unregulated and regulated reaches were onlyetected in reference flow class 1 (class 1, T = 28.40, P = 0.003, class, T = 1.05, P = 0.42, class 5, T = 2.32, P = 0.15), with a standardisedverage density of 9.7 individuals in unregulated reaches versus.4 in regulated reaches (Table 3). Abundances of M. duboulayiiffered between unregulated and regulated reaches in referenceow class 2 (class 1, T = 1.49, P = 0.292, class 2, T = 10.57, P = 0.005,lass 5, T = 1.86, P = 0.21), with a standardised average density of 30ndividuals in unregulated reaches versus 6.4 in regulated reachesTable 3). None of the remaining eight species tested (Table 3) dif-ered significantly in density between unregulated and regulatedeaches.

Densities of A. reinhardtii, H. sp. 1, P. signifier, and G. australisiffered significantly across current flow classes on all samplingccasions (Table 4). Densities of seven species (A. reinhardtii, H.alii, H. sp. 1, M. duboulayi, P. signifier and G. australis) differed sig-ificantly between unregulated and regulated sites within specificlasses (Table 4). For five of these species (A. reinhardtii, H. galii,. sp. 1, P. signifier, G. australis) these differences were temporallyonsistent, whereas M. duboulayi showed inconsistent differencesetween unregulated and regulated sites over sampling times.nder current flow conditions, R. semoni, T. tandanus, H. klunzingeriere the only species that did not differ significantly in abundance

etween regulated and unregulated reaches within current flowlasses.

.3. Non-native species responses

Non-native species richness and the proportion of both speciesnd individuals that are non-native were very low throughout theata set, and hence were low across all flow classes in both the ref-rence and current flow classifications. Despite this, densities of G.olbrooki differed significantly across both reference and currentow classes and were not significantly different between unregu-

ated and regulated sites within both reference and current classesTables 3 and 4). None of the remaining three variables representinghe establishment of non-native species differed across referenceow classes, or between unregulated and regulated reaches withineference flow classes (Table 3). In the current flow classificationhere was a significant difference in the proportion of individualshat were non-native between unregulated and regulated reachesTable 4).

.4. Assemblage abundance and composition

Assemblage abundance (total density of all individuals inamples) differed significantly among reference flow classes incon-istently through time (Table 3) but did not differ betweennregulated and regulated reaches within any reference flow class.

Indicators 39 (2014) 179–188 183

Assemblage composition differed significantly among referenceflow classes (Table 5; 22.9% variance explained), and there weresignificant differences in species composition between unregulatedand regulated reaches (8.2% variance explained) within referenceflow classes 1 and 5, but not 2 (post hoc pairwise tests: class1, T = 2.23, P = 0.009, class 2, T = 1.25, P = 0.27, class 5, T = 2.50,P = 0.002).

When tested across current flow classes, fish assemblage abun-dance differed significantly across classes, varied through timeand did not differ between unregulated and regulated reacheswithin flow classes (Table 4). Assemblage composition differed sig-nificantly across all current flow regime classes (Table 5; 15.4%variation explained, all post hoc comparisons P < 0.001), and alsobetween unregulated and regulated reaches within current flowclasses (20.9% variation). Post hoc pairwise tests identified thatthese differences between unregulated and regulated reaches weresignificant in current flow classes 1, 3 and 4, but not 2 (class 1, T = 3.2,P = 0.006, class 2, T = 1.64, P = 0.07, class 3, T = 2.61, P = 0.002, class 4,T = 2.25, P = 0.034).

3.5. Species diversity

None of the three variables representing fish species diver-sity (total species richness, native species richness and speciesdensity) differed significantly among reference flow classes, orbetween unregulated and regulated reaches within any referenceclass (Table 3). When tested against the current flow classifica-tion, both total species richness and native species richness differedamong flow classes on all sampling occasions (Table 4). Total andnative species richness also differed significantly between unregu-lated and regulated reaches within flow class 3 (T = 8.1, P = 0.009),but not within class 1 (T = 0.11, P = 0.91), class 2 (T = 0.77, P = 0.52)or class 4 (T = 3.84, P = 0.068).

4. Discussion

Anthropogenic flow regime change is predicted to result in asequence (Table 1) of effects on riverine fish at levels of organisa-tion varying from individuals to species assemblages (e.g. Freemanet al., 2001; Gehrke and Harris, 2001; Bunn and Arthington, 2002;Poff and Zimmerman, 2010). The first aim of this study was totest if fish responses variables representing this sequence differedbetween regulated and unregulated reaches in streams and riversof contrasting natural (pre-development) flow variability. Threeof 17 response variables representing population and assemblagelevels of the predicted sequence showed significant differencesbetween regulated and unregulated reaches (densities of P. signi-fier and M. duboulayi, and fish assemblage composition). Secondly,we predicted that if the flow regime is a key determinant of fishpopulation size and assemblage structure that there would be dif-ferences in response variables across current flow regime classesthat incorporate different magnitudes and types of flow regula-tion. We detected significant differences in the population size offive of 10 fish species (A. reinhardtii, H. sp. 1, P. signifier, G. australis,G. holbrooki) across current flow classes, and in total and nativespecies richness, assemblage abundance (total abundance of all fishspecies) and assemblage composition across current flow regimeclasses. These differences in univariate and multivariate responsevariables were apparent in reaches that have experienced shiftsfrom naturally intermittent and unpredictable flow regimes to therarely intermittent current flow class (3) as a consequence of man-

aged flow releases from dams. However, contrary to predictions,there were also significant differences in fish response variablesbetween regulated and unregulated reaches of similar hydrologi-cal character within some current flow classes. When these findings

184

R.J.

Rolls,

A.H

. A

rthington /

Ecological Indicators

39 (2014)

179–188

Table 3Univariate PERMANOVA comparisons of fish population and assemblage response variables between (a) reference flow classes and (b) unregulated (UR) and regulated (R) sites within each reference flow class, in the predictedorder of sequence of effects (Table 1). Significance values and the component of variation explained are indicated by P and VC, respectively. Underlined P values indicate a non-significant interaction between each main effect andsampling time (indicating temporal consistency). To guide interpretation, values under each flow class, and unregulated and regulated groups within each flow class indicate the mean value for each response variable in eachsample group.

P VC (a) Differences among flow classes P VC (b) Differences between unregulated and regulated sites within each flow class

Flow class Class 1 Class 2 Class 3 Class 4 Class 5

1 2 3 4 5 UR R UR R UR R* UR R* UR R

1. Native species population sizeAnguillareinhardtii

0.005 28.1 9.6 5.9 7.7 6.5 27.3 0.083 4.3 6.8 8.5 5.2 17.0 3.1 – 6.0 – 24.6 23.4

Melanotaeniaduboulayi

0.024 14.4 8.5 39.2 1.6 30.1 4.2 0.002 5.1 18.5 70.7 30.0 6.4 3.0 – 21.4 – 2.4 2.1

Retropinnasemoni

0.123 8.6 37.4 38.8 0.9 11.1 16.4 0.375 0.7 36.7 37.0 30.4 36.0 1.1 – 29.7 – 13.4 42.4

Tandanustandanus

0.103 5.5 5.3 4.4 0.5 9.2 1.8 0.304 2.8 5.2 1.5 5.8 5.3 1.3 – 2.5 – 1.2 13.2

Hypseleotrisklunzingeri

0.016 7.6 7.1 21.7 10 47.5 4.7 0.517 0 45.7 5.3 15.4 17.7 8.1 – 1.8 – 3.8 2.0

Hypseleotris galii 0.009 11.6 4.7 25.1 0.4 12.1 6 0.966 0 6.0 41.7 21.8 2.8 1.3 – 4.7 – 7.1 13.1Pseudomugilsignifer

0.001 12.5 24.9 10.4 2.6 2.6 0.6 0.007 13.7 9.7 1.4 13.6 6.2 5.0 – 63.2 – 0.8 0.0

Hypseleotris sp. 1 0.013 24.9 4.4 2.3 0 21.3 0 0.146 3.3 9.5 1.8 1.6 8.5 0.6 – 3.3 – 0.0 0.0Gobiomorphusaustralis

0.006 40.9 0.1 0.2 0 0 43.9 0.772 0 0.2 0.2 0.2 18.5 0.0 – 0.0 – 41.0 0.0

2. Non-native speciesGambusiaholbrooki

0.004 27.9 0.4 2.4 1.3 41 2.8 0.156 0.1 20.1 4.4 1.0 2.4 1.3 – 0.9 – 2.8 0.1

Non-nativespecies richness

0.162 9.1 0.1 0.5 0.5 0.9 0.5 0.151 2.9 0.7 0.8 0.2 0.4 0.5 – 0.1 – 0.5 0.2

Proportion ofspecies that arenon-native

0.216 6.2 0.02 0.07 0.06 0.08 0.07 0.260 3.5 0.1 0.1 0.0 0.0 0.1 – 0.0 – 0.1 0.0

Proportion ofindividuals thatare non-native

0.144 14.8 0.01 0.02 0.05 0.13 0.02 0.185 0.8 0.1 0.0 0.0 0.0 0.0 – 0.0 – 0.0 0.0

3. Assemblageabundance

0.011 15.5 111.2 162.5 39.7 250.2 142.7 0.570 0 180.8 178.0 143.4 146.3 39.4 – 153.9 – 134.6 96.8

4. Species diversitySpecies richness 0.085 5.4 8.1 7.5 7 10.4 6.9 0.081 7.8 8.9 6.6 7.4 9.7 7.5 – 8.9 – 6.4 5.2Native speciesrichness

0.111 5.8 7.9 7 6.5 9.6 6.4 0.108 8.2 8.2 5.8 7.2 9.3 7.0 – 8.8 – 5.8 5.0

Species density 0.067 11.9 13.6 24.9 5.2 24.7 20.6 0.441 0.2 19.6 34.3 20.6 15.0 4.9 – 16.5 – 21.1 18.1

Asterisk (*) indicates no sites within the specific flow class or with flow alteration were available for assessment (see Table 2).

R.J.

Rolls,

A.H

. A

rthington /

Ecological Indicators

39 (2014)

179–188

185

Table 4Univariate PERMANOVA comparisons of predicted fish population and assemblage response variables between (a) current flow classes and (b) unregulated (UR) and regulated (R) sites within each current flow class.

P VC (a) Differences among flow classes P VC (b) Differences between unregulated and regulated sites within each flow class

Flow class Class 1 Class 2 Class 3 Class 4 Class 5

1 2 3 4 5 UR R UR R UR R UR R UR R*

1. Native species population sizeAnguillareinhardtii

0.002 14.3 9.3 4.9 10.5 4.9 24.3 0.002 16.2 10.8 8.5 6.5 2.5 4.9 23.5 8.4 1.4 24.3 –

Melanotaeniaduboulayi

0.124 5.5 48.6 21.7 19 17.1 2.3 0.043 26.5 4.5 70.7 30.8 8.0 24.0 7.4 26.7 7.5 2.3 –

Retropinnasemoni

0.093 8.4 47.6 18 32.5 22.4 19.2 0.140 12.9 69.0 37.0 11.1 28.4 27.7 43.7 42.8 2.0 19.2 –

Tandanustandanus

0.129 5 1.2 8 4.9 2.3 3.6 0.434 0.2 0.5 1.5 9.2 6.1 5.2 4.1 3.1 1.5 3.6 –

Hypseleotrisklunzingeri

0.058 14.2 3.7 61.7 14.7 2.4 3.4 0.986 0 0.5 5.3 47.5 83.0 13.6 17.1 0.3 4.5 3.5 –

Hypseleotris galii 0.442 0.24 27.8 9.1 12.6 4 8.3 0.021 16.8 0.0 41.7 12.1 4.7 17.2 1.9 0.9 7.2 8.3 –Pseudomugilsignifer

0.002 26.5 3.3 9 10.3 49.6 0.6 0.001 51.5 7.2 1.4 2.6 18.6 12.4 5.4 89.2 10.0 0.6 –

Hypseleotris sp. 1 0.016 14.5 1.2 15.2 3.3 2.8 0 0.025 17.9 0.0 1.8 21.3 5.9 1.3 8.2 3.5 2.1 0.0 –Gobiomorphusaustralis

0.012 15.7 0.3 0.04 8.4 0 32.8 0.003 12.1 0.5 0.2 0.0 0.1 0.1 27.7 0.0 0.0 32.8 –

2. Non-native speciesGambusiaholbrooki

0.014 10.2 3 25.9 1.2 0.7 2.3 0.066 21.1 0.2 4.4 41.0 3.2 1.1 1.3 0.0 1.4 2.3 –

Non-nativespecies richness

0.156 10.4 0.8 0.8 0.3 0.1 0.5 0.888 0 0.8 0.8 0.9 0.7 0.3 0.3 0.0 0.2 0.5 –

Proportion ofspecies that arenon-native

0.059 13.6 0.12 0.08 0.03 0.01 0.07 0.956 0 0.1 0.1 0.1 0.1 0.0 0.0 0.0 0.0 0.1 –

Proportion ofindividuals thatare non-native

0.110 6.6 0.03 0.08 0.01 0.01 0.02 0.046 21.3 0.0 0.0 0.1 0.0 0.0 0.0 0.0 0.0 0.0 –

3. Assemblageabundance

0.026 6.4 151.2 216 134.4 121.8 127 0.165 11.5 97.7 178.0 250.2 164.7 125.1 156.0 203.3 40.3 127.0 –

4. Species diversitySpecies richness 0.006 25.4 6.7 10.5 8.1 8.2 6.1 0.010 21.3 6.8 6.6 10.4 10.7 7.5 9.7 10.2 6.3 6.1 –Native speciesrichness

0.003 29.7 5.9 9.7 7.9 8.2 5.7 0.019 23.5 6.0 5.8 9.6 10.0 7.2 9.4 10.2 6.2 5.7 –

Species density 0.369 0.9 27.6 20.8 17.4 13.3 20.5 0.107 16.1 14.1 34.3 24.7 15.0 17.8 16.4 20.4 6.4 20.5 –

Asterisk (*) indicates no sites within the specific flow class or with flow alteration were available for assessment (see Table 2).

186 R.J. Rolls, A.H. Arthington / Ecological Indicators 39 (2014) 179–188

Table 5Multivariate PERMANOVA comparisons testing differences in fish assemblage composition among flow classes, sample times and between regulated and unregulated siteswithin each class of the reference and current flow classification. Significant P values are indicated in bold and variance components (VC) are included.

Source of variation df MS Pseudo-F P VC

Reference classificationFlow class 4 28.0 7.48 0.001 22.9Sample time 2 8.6 2.22 0.004 4.1Regulation (flow class) 3 9.7 3.85 0.003 8.2Flow class × time 8 3.7 0.96 0.584 0.0Regulation (FC) × time 6 2.5 0.65 0.983 0.0Residual 91 3.9 64.8

Current classificationFlow class 4 22.3 8.04 0.001 15.4Sample time 2 6.9 1.73 0.024 1.6Regulation (flow class) 4 15.3 5.27 0.001 20.9

alaa

4a

acfl2ehccrcmcrefitaattdai

wttttlctkarirI(

Flow class × time 8 2.8

Regulation (FC) × time 8 2.9Residual 88 4.0

re considered together they suggest that the effects of flow regu-ation in SEQ are apparent largely at the first step (i.e. populationbundance level) in the ecological impact sequence with somessociated changes in fish assemblage composition (Table 1).

.1. Patterns in population size and assemblage structuressociated with flow regimes

Changes (predominantly declines) in the abundances of nativequatic species, reduced total species richness and shifts in theomposition of assemblages are well-documented consequences ofow regulation (Bunn and Arthington, 2002; Poff and Zimmerman,010). In this study, the effects of flow regime alteration were mostvident in streams where historically intermittent flow regimesave become more perennial. For example, we detected signifi-ant differences in M. duboulayi and P. signifer populations betweenurrently unregulated and regulated reaches that had similar flowegimes under pre-development conditions. There were signifi-antly lower densities of both species in regulated reaches withore perennial flows compared to reference reaches. This finding

oncurs with previous evidence of lower abundances of P. signifer inegulated rivers when compared with unregulated rivers in north-rn New South Wales, Australia (Gehrke and Harris, 2001). Onlyve non-native species were recorded in our study area howeverhey represented <4% of the total number of fish sampled. Thebundance of G. holbrooki, the dominant non-native species, variedcross current flow classes and was highest in the highly intermit-ent unpredictable summer flow class (2). This is consistent withhe responses of other non-native species to regulated flow con-itions (e.g. Gehrke and Harris, 2001; Bunn and Arthington, 2002)nd the ability of G. holbrooki to dominate fish populations underntermittent flow conditions (Ho et al., 2013).

A key premise of the ELOHA framework is that river reachesith similar flow regimes will have similar ecological characteris-

ics (e.g. patterns in structure and function) with the corollary beinghat rivers of different flow character will allow different specieso dominate (Poff et al., 2010). One implication of this premise ishat if two rivers have similar flow regimes, even if one is regu-ated and the other is not, then they will have similar ecologicalharacteristics but only if flow is the major factor shaping ecosys-em structure and function. Three species (R. semoni, T. tandanus, H.lunzingeri) were the only species that did not differ significantly inbundance between regulated and unregulated reaches within cur-ent flow classes. Patterns such as these suggest that these species

n particular have adjusted to the characteristics of current flowegimes, including any differences attributable to flow regulation.n contrast, although densities of five of 10 common fish speciesA. reinhardtii, H. sp. 1, P. signifier, G. australis, G. holbrooki), total

0.69 0.980 0.00.72 0.972 0.0

62.1

and native species richness, assemblage abundance (total abun-dance of all fish species) and assemblage composition differedacross current flow regime classes, these response variables alsodiffered between regulated and unregulated reaches within someflow classes. Such patterns suggest that these particular speciesand the assemblages to which they contribute have not adjusted tothe characteristics of the prevailing regulated flow regimes, or thereaches we compared differed in characteristics other than flow.The latter interpretation seems less likely given the distribution ofstudy reaches and their catchment and in-stream characteristics.Separate analysis has shown that land-use patterns in catchmentsof the study area (and by inference, any associated impacts on studyreaches) account for only a small proportion of variation in fishpresence–absence and relative abundance patterns; reach habitatstructure was also a minor explanatory variable in multivariatemodels (Arthington et al., in press). Most variation in fish assem-blage composition was found to be a function of regional climateand patterns of rainfall and runoff, such that spatial patterns of mostspecies (24 of 35) were strongly associated with gradients in min-imum flows, number of zero flow days and number of high flowpulses, the magnitude of the 1-year flood ARI and the constancyand predictability of flows (Arthington et al., in press). The over-whelming importance of flow regime characteristics in structuringthe fish assemblages of the study area supports our emphasis onflow regime changes in regulated reaches as the probable impactfactor, rather than stresses associated with land-use in surroundingupland catchments or differences in reach habitat structure.

4.2. Potential processes underlying patterns of association withflow regimes

By identifying how patterns of difference in response variablesreflecting native fish population size, non-native species prolifer-ation and assemblage composition are associated with differencesin flow regimes, we can begin to infer the underlying flow-relatedecological processes (sensu Underwood et al., 2000; Downes, 2010).Such evidence is needed to support and strengthen appropriatemanagement actions advocated by environmental flow frame-works, such as ELOHA.

Few differences in fish response variables between regulatedand unregulated reaches that are comparable in relation to theirhistorical flow regimes may be attributed to the relatively low mag-nitude of flow regime alteration in our study region, where 57%of ecologically relevant flow metrics have changed by <20%. This

interpretation is consistent with previous studies which show thatfish abundance, demographic parameters and diversity decreaseby at least 50% in response to both decreased (−50% to −100%) andincreased (+50% to 100%) flow magnitude (Poff and Zimmerman,

ogical

2cc12atbKsovepstcifr

ilmtetstaTtrr

rpfibtimuafahaawsrsTiflc

4

afless

R.J. Rolls, A.H. Arthington / Ecol

010). Clear ecological effects of flow alteration have been asso-iated with other marked changes in the flow regime, such asomplete reversal of flow seasonality (e.g. Weisberg and Burton,993; Gehrke et al., 1995; Carlisle et al., 2010; Poff and Zimmerman,010). In our study area, relatively low levels of hydrologic alter-tion appear to have generated modified flow regimes that areolerable for species that have natural distributions across a muchroader range of flow regime types in Australia (Pusey et al., 2004;ennard et al., 2010). Species unaffected by flow regulation pre-umably have the traits or characteristics enabling them to persistver a spectrum of hydrological conditions beyond the ranges ofariability experienced in the rivers of south-east Queensland. Thexception appears to be that a shift from intermittency to a moreerennial type flow regime has implications for pelagic nativepecies such as P. signifier and M. duboulayi. These small speciesend to form large shoals in run and pool type habitats or takeover in vegetated pool margins (Pusey et al., 2004). Elevated veloc-ties and more constant flow levels associated with water releasesrom dams may disrupt habitat conditions and/or interfere witheproductive behaviours and population recruitment.

Stream ecosystems are structured by, and therefore reflect, thenfluence of flow patterns over temporal scales ranging from mil-iseconds to millennia (Biggs et al., 2005). The response variables

easured during this study reflect the outcomes of adjustmentso environmental conditions over periods of years to centuries. Forxample, changes in species richness in response to environmen-al change such as flow regulation can only occur after individualpecies have colonised and established, or declined in abundanceo the point of local extinction. Such processes may not be immedi-te upon flow regime change nor likely even after several decades.herefore temporal aspects of changes in river flow regimes (e.g.he length of time that a river has experienced a regulated flowegime) are also critical to interpretation of the effects of flowegime alterations (Murchie et al., 2008).

In this study, streams with regulated flow regimes have expe-ienced flow regulation for 21–46 years prior to sampling. It isossible that limited evidence or absence of significant effects onsh species and assemblages associated with flow regulation maye attributed in part to the relatively short duration of exposureo flow regime change, coupled with the relatively minor changesn magnitude (i.e. <20%) of more than half the ecologically relevant

etrics distinguishing flow regime classes and patterns of flow reg-lation. Other studies support this observation. For example, fishssemblage composition showed very limited change immediatelyollowing impoundment of the White River, Arkansas, howeverfter 30 years of regulation many flowing-water specialist speciesad disappeared and non-native salmonids had established (Quinnnd Kwak, 2003). We suggest that the effects of flow regulationre dependent on the longevity of local biota, such that speciesith short life spans (e.g. P. signifier, M. duboulayi, R. semoni,

pecies of Hypseleotris and G. holbrooki) would be expected to showesponses much earlier than long-lived species (e.g. eels, Anguillapp.) exposed over the same timeframe (Mims and Olden, 2013).his hypothesis could be tested simply by ongoing long-term mon-toring of the ecology of streams subjected to various periods ofow regulation to determine if, and at what point in time, furtherlear ecological changes become evident.

.3. Conclusions

This study provides much needed evidence of fish populationnd assemblage level impacts associated with low magnitudes of

ow regime alteration by dams. Our findings indicate that theffects of flow regulation in SEQ are apparent largely at the firsttep (i.e. population abundance levels) in the ecological impactequence with some attendant changes in native fish assemblage

Indicators 39 (2014) 179–188 187

composition and the proliferation of one common non-nativespecies, G. holbrooki. These results are consistent with other studiesthat have documented significant impacts on native fish popula-tion abundance where naturally intermittent streams and rivershave been converted to systems with more perennial flows as aconsequence of managed flow releases from dams (e.g. Gehrke andHarris, 2001; Reich et al., 2010; Mims and Olden, 2013), and withstudies on the population responses of G. holbrooki to seasonalflooding in unregulated rivers (Ho et al., 2013). This study providespositive evidence that dams and flow regimes can be managed withsensitivity such that there are no significant changes in populationsof most fish species, and little change in assemblage characteristics.Low levels of hydrologic alteration based on percentage changes(e.g. 10–20%) have been proposed as a presumptive flow standardwhen regional scale environmental flow studies are not feasible dueto cost and/or time constraints (e.g. Richter et al., 2012). However,standards of this type must be supported by ecological evidence forrivers of different hydrological and ecological character (Arthingtonet al., 2006; Poff et al., 2010). It must be cautioned that interactionsof the magnitude of flow alterations and durations of exposure (i.e.years to decades) have the potential to induce ecological harm, andsuch interactions require further investigation.

Acknowledgements

This project was funded by the Australian National Water Com-mission under the Raising National Water Standards programme,with administrative support and guidance from the InternationalWater Centre and Griffith University, Brisbane. The support andcontributions of land owners who allowed access to their prop-erties around SEQ are gratefully acknowledged. We thank SofieBernays, Tim Howell, David Sternberg and Ben Stewart-Koster fortheir efforts with field sampling. This research was undertakenunder Queensland Fisheries Permit PRM00157K and Griffith Uni-versity Animal Ethics Approval No. ENV/21/08/AEC. We appreciatethe constructive advice provided by four referees to improve thecontent and clarity of the manuscript.

Appendix A. Supplementary data

Supplementary data associated with this article can befound, in the online version, at http://dx.doi.org/10.1016/j.ecolind.2013.12.017.

References

Acreman, M., Dunbar, M.J., 2004. Defining environmental river flow requirements –a review. Hydrol. Earth Syst. Sci. 8, 861–876.

Anderson, M.J., 2001. A new method for non-parametric multivariate analysis ofvariance. Austral Ecol. 26, 32–46.

Anderson, M.J., Ellingsen, K.E., McArdle, B.H., 2006. Multivariate dispersion as ameasure of beta diversity. Ecol. Lett. 9, 683–693.

Anderson, M.J., Gorley, R.N., Clarke, K.R., 2008. PERMANOVA+ for PRIMER: Guide toSoftware and Statistical Methods. PRIMER-E, Plymouth, United Kingdom.

Arthington, A.H., Bunn, S.E., Poff, N.L., Naiman, R.J., 2006. The challenge of providingenvironmental flow rules to sustain river ecosystems. Ecol. Appl. 16, 1311–1318.

Arthington, A.H., Mackay, S.J., James, C.S., Rolls, R.J., Sternberg, D., Barnes,A., Capon, S.J., 2012. Ecological limits of hydrologic alteration: a testof the ELOHA framework in south-east Queensland, Waterlines ReportSeries 75. National Water Commission, Canberra, Australia, Available from:http://www.nwc.gov.au/publications/waterlines/75 (accessed 26.10.13).

Arthington, A.H., Rolls, R.J., Sternberg, D., Mackay, S.J., James, C.S., 2014. Fish assem-blages in sub-tropical rivers: low flow hydrology dominates hydro-ecologicalrelationships. Hydrol. Sci. J. (in press) http://www.tandfonline.com/doi/abs/10.1080/02626667.2013.844345#.Ur61utIW2E4

Biggs, B.J.F., Nikora, V.L., Snelder, T.H., 2005. Linking scales of flow variability to lotic

ecosystem structure and function. River Res. Appl. 21, 283–298.

Bridges, E.M., Ross, D.J., Thompson, C.H., 1990. Soils of the Mary River alluvia nearGympie, Queensland, CSIRO Division of Soils, Divisional Report No. 109.

Bunn, S.E., Arthington, A.H., 2002. Basic principles and ecological consequences ofaltered flow regimes for aquatic biodiversity. Environ. Manage. 30, 492–507.

1 ogical

C

C

C

C

D

FF

G

G

H

H

K

K

M

M

M

M

88 R.J. Rolls, A.H. Arthington / Ecol

arlisle, D.M., Wolock, D.M., Meador, M.R., 2010. Alteration of streamflow magni-tudes and potential ecological consequences: a multiregional assessment. Front.Ecol. Environ. 9, 264–270.

atford, J.A., Downes, B.J., Gippel, C.J., Vesk, P.A., 2011. Flow regulation reduces nativeplant cover and facilitates exotic invasion in riparian wetlands. J. Appl. Ecol. 48,432–442.

larke, K.R., Gorley, R.N., 2006. Primer v6: User Manual/Tutorial. Primer-E, Plymouth,United Kingdom.

olwell, R.K., 1974. Predictability, constancy, and contingency of periodicphenomena. Ecology 55, 1148–1153.

ownes, B.J., 2010. Back to the future: little-used tools and principles of scien-tific inference can help disentangle effects of multiple stressors on freshwaterecosystems. Freshw. Biol. 55, 60–79.

raley, C., Raftery, A., 2008. Package ‘Mclust’, version 3.1–3.reeman, M.C., Bowen, Z.H., Bovee, K.D., Irwin, E.R., 2001. Flow and habitat effects

on juvenile fish abundance in natural and altered flow regimes. Ecol. Appl. 11,179–190.

ehrke, P.C., Brown, P., Schiller, C.B., Moffatt, D.B., Bruce, A.M., 1995. River regula-tion and fish communities in the Murray-Darling river system, Australia. Regul.Rivers: Res. Manage. 11, 363–375.

ehrke, P.C., Harris, J.H., 2001. Regional-scale effects of flow regulation on lowlandriverine fish communities in New South Wales, Australia. Regul. Rivers: Res.Manage. 17, 369–391.

o, S., Bond, N.R., Thompson, R.M., 2013. Does seasonal flooding give a native speciesan edge over a global invader? Freshw. Biol. 58, 159–170.

umphries, P., Serafini, L.G., King, A.J., 2002. River regulation and fish larvae: varia-tion through space and time. Freshw. Biol. 47, 1307–1331.

ennard, M.J., Pusey, B.J., Harch, B.D., Dore, E., Arthington, A.H., 2006. Estimatinglocal stream fish assemblage attributes: sampling effort and efficiency at twospatial scales. Mar. Freshw. Res. 57, 635–653.

ennard, M.J., Pusey, B.J., Olden, J.D., Mackay, S.J., Stein, J.L., Marsh, N., 2010. Clas-sification of natural flow regimes in Australia to support environmental flowmanagement. Freshw. Biol. 55, 171–193.

ackay, S.J., Arthington, A.H., James, C.S., 2014. Classification and comparison ofnatural and altered flow regimes to support an Australian trial of the EcologicalLimits of Hydrologic Alteration (ELOHA) framework. Ecohydrology (in press).

agilligan, F.J., Nislow, K.H., 2005. Changes in hydrologic regime by dams. Geomor-

phology 71, 61–78.

eador, M.R., Carlisle, D.M., 2012. Relations between altered streamflow variabilityand fish assemblages in eastern USA streams. River Res. Appl. 28, 1359–1368.

ims, M.C., Olden, J.D., 2013. Fish assemblages respond to altered flow regimes viaecological filtering of life history strategies. Freshw. Biol. 58, 50–62.

Indicators 39 (2014) 179–188

Murchie, K.J., Hair, K.P.E., Pullen, C.E., Redpath, T.D., Stephens, H.R., Cooke, S.J., 2008.Fish response to modified flow regimes in regulated rivers: research methods,effects and opportunities. River Res. Appl. 24, 197–217.

Naiman, R.J., Latterell, J.J., Pettit, N.E., Olden, J.D., 2008. Flow variability and thebiophysical vitality of river systems. C. R. Geosci. 340, 629–643.

Poff, N.L., Allan, J.D., Bain, M.B., Karr, J.R., Prestegaard, K.L., Richter, B.D., Sparks, R.E.,Stromberg, J.C., 1997. The natural flow regime. BioScience 47, 769–784.

Poff, N.L., Richter, B.D., Arthington, A.H., Bunn, S.E., Naiman, R.J., Kendy, E., Acreman,M., Apse, C., Bledsoe, B.P., Freeman, M.C., Henriksen, J., Jacobson, R.B., Kennen,J.G., Merritt, D.M., O’Keefe, J.H., Olden, J.D., Rogers, K., Tharme, R.E., Warner,A., 2010. The ecological limits of hydrologic alteration (ELOHA): a new frame-work for developing regional environmental flow standards. Freshw. Biol. 55,147–170.

Poff, N.L., Zimmerman, J.K.H., 2010. Ecological responses to altered flow regimes:a literature review to inform the science and management of environmentalflows. Freshw. Biol. 55, 194–205.

Pusey, B., Kennard, M., Arthington, A., 2004. Freshwater Fishes of North-EasternAustralia. CSIRO Publishing, Collingwood, Australia.

Quinn, J.W., Kwak, T.J., 2003. Fish assemblage changes in an Ozark River afterimpoundment: a long-term perspective. Trans. Am. Fish. Soc. 132, 110–119.

R Core Development Team, 2010. R: A Language and Environment for StatisticalComputing. R Foundation for Statistical Computing, Vienna, Austria.

Reich, P., McMaster, D., Bond, N., Metzeling, L., Lake, P.S., 2010. Examining theecological consequences of restoring flow intermittency to artificially peren-nial lowland streams: patterns and predictions from the Broken–Boosey creeksystem in northern Victoria, Australia. River Res. Appl. 26, 529–545.

Richter, B.D., Davis, M.M., Apse, C., Konrad, C., 2012. A presumptive standard forenvironmental flow protection. River Res. Appl. 28, 1312–1321.

Strayer, D.L., 2010. Alien species in fresh waters: ecological effects, interactions withother stressors, and prospects for the future. Freshw. Biol. 55, 152–174.

Tharme, R.E., 2003. A global perspective on environmental flow assessment:emerging trends in the development and application of environmental flowmethodologies for rivers. River Res. Appl. 19, 397–441.

Underwood, A.J., Chapman, M.G., Connell, S.D., 2000. Observations in ecology: youcan’t make progress on processes without understanding the patterns. J. Exp.Mar. Biol. Ecol. 250, 97–115.

Weisberg, S.B., Burton, W.H., 1993. Enhancement of fish feeding and growth after an

increase in minimum flow below the Conowingo Dam. N. Am. J. Fish. Manage.13, 103–109.

Young, P.A.R., Dillewaard, H.A., 1999. Southeast Queensland. In: Sattler, P.S.,Williams, R.D. (Eds.), The Conservation Status of Queensland’s bioregionalecosystems. Environmental Protection Agency, Brisbane.