lable at ScienceDirect

Environmental Pollution 196 (2015) 12e20

Contents lists avai

Environmental Pollution

journal homepage: www.elsevier .com/locate/envpol

What does impacted look like? High diversity and abundance ofepibiota in modified estuaries

Graeme F. Clark a, *, Brendan P. Kelaher b, Katherine A. Dafforn a, Melinda A. Coleman c,Nathan A. Knott d, Ezequiel M. Marzinelli a, Emma L. Johnston a

a School of Biological, Earth and Environmental Sciences, University of New South Wales, Sydney, NSW 2052, Australiab National Marine Science Centre & Centre for Coastal Biogeochemistry Research, School of Environment, Science and Engineering, Southern CrossUniversity, Coffs Harbour, NSW 2450, Australiac New South Wales Fisheries, Department of Primary Industries, Coffs Harbour, NSW 2450, Australiad NSW Department of Primary Industries, Jervis Bay Marine Park, Huskisson, NSW 2540, Australia

a r t i c l e i n f o

Article history:Received 15 April 2014Received in revised form8 September 2014Accepted 15 September 2014Available online

Keywords:BiodiversityPollutionIntroduced speciesMetalsNutrientsBioindicators

* Corresponding author.E-mail address: g.clark@unsw.edu.au (G.F. Clark).

http://dx.doi.org/10.1016/j.envpol.2014.09.0170269-7491/© 2014 Elsevier Ltd. All rights reserved.

a b s t r a c t

Ecosystems modified by human activities are generally predicted to be biologically impoverished.However, much pollution impact theory stems from laboratory or small-scale field studies, and fewstudies replicate at the level of estuary. Furthermore, assessments are often based on sedimentcontamination and infauna, and impacts to epibiota (sessile invertebrates and algae) are seldomconsidered. We surveyed epibiota in six estuaries in south-east Australia. Half the estuaries were rela-tively pristine, and half were subject to internationally high levels of contamination, urbanisation, andindustrialisation. Contrary to predictions, epibiota in modified estuaries had greater coverage and weresimilarly diverse as those in unmodified estuaries. Change in epibiota community structure was linearlycorrelated with sediment-bound copper, and the tubeworm Hydroides elegans showed a strong positivecorrelation with sediment metals. Stressors such as metal contamination can reduce biodiversity andproductivity, but others such as nutrient enrichment and resource provision may obscure signals ofimpact.

© 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Estuaries are a nexus of human settlement and marine envi-ronments, and bear the brunt of human activities occurring on bothland and water. Consequently, they are widely considered the mostimpacted of all marine habitats (Edgar et al., 2000; Kennish, 2002;Roy et al., 2001). At the same time there is great incentive to pre-serve their proper functioning, since estuaries provide keyecosystem services such as nutrient cycling, water filtration, and actas nursery grounds for economically-important species of fish(Morrisey, 1995). The task of minimising human impacts in theseecologically significant environments is one of themajor challengesfacing environmental managers today.

Many estuaries are subject to multiple stressors from industry,agriculture, urban and coastal development, and some are nowchronically contaminated (Scanes and Scanes, 1995). Heavy metals,fertilizers, sewage, and synthetic compounds are common forms of

pollution that enter the marine environment as a result of humanactivities. Some contaminants can bind to sediments and reside inestuaries for years or decades, periodically re-entering the watercolumn when sediments are disturbed (Roberts, 2012). Coastaldevelopment also affects physico-chemical water quality parame-ters (e.g. temperature, salinity, turbidity and dissolved oxygen) byaltering terrestrial run-off, hydrodynamics, dredging, or thedischarge of heated effluent. The physical underwater landscape ofmany urban estuaries is also extensively modified through theaddition of artificial structures (Dafforn et al., 2009), or by dis-placing sediment to accommodate shipping (Newell et al., 1998). Itis vital to identify and rank the strongest drivers of human impacts,and understand their effects in order to sustainably manage andconserve these ecosystems.

Estuaries encompass a diverse range of habitats, from soft sed-iments to rocky reefs, seagrass beds, mangroves and open watersystems (Morrisey, 1995). Past environmental management hasfocused on monitoring water quality alone (ANZECC, 2001), but hasbeen criticised for lacking ecological relevance (Scanes et al., 2007).Focus is now moving towards integrated ecosystem assessmentsthat include biotic components (Borja et al., 2008), and substantial

G.F. Clark et al. / Environmental Pollution 196 (2015) 12e20 13

research has been invested worldwide in developing estuarinehealth indicators. Some taxa have been identified as useful bio-indicators e species whose abundance or condition reflects envi-ronmental quality (Carignan and Villard, 2002). However, most ofthis research has focused on soft-sediment infaunal communities(Borja et al., 2008; Hewitt et al., 2009; Hyland et al., 2003; VanDolah et al., 1999), which are physically separated from water col-umn disturbances and may respond differently to other ecosystemcomponents.

Sessile invertebrates and algae, or epibiota, grow on hardsubstrates in estuaries and can be useful bioindicators. Mostsessile invertebrates are filter feeders and are exposed to con-taminants ingested as food, or absorbed in dissolved form directlyfrom the water column. These pathways of exposure are alsorelevant to other ecosystem components, such as fish andplankton. Marine invertebrates require trace amounts of somemetals (e.g. copper, lead and zinc) for metabolic function, butelevated concentrations interfere with metabolic processes andcan cause sub-lethal impacts or mortality. Particulate matter in thewater column can also physically disturb filter-feeders by cloggingfeeding apparatus or burying individuals (Burton and Johnston,2010). Sediment resuspension events from dredging or vesseltraffic have been shown to harm sessile invertebrates through theremobilisation of contaminants and/or increased suspendedsediment (Knott et al., 2009). To assess the response of epibiota tocontamination and to understand the utility of these fauna asbioindicators, we need to quantify their distributions relative tocontamination at large-scales in the field.

The abundance and diversity of introduced species is anotherimportant measure of ecological integrity. Epibiota represent alarge proportion of marine non-native species (Ruiz et al., 1997),primarily due to their affinity to be transported by shipping. Non-native species are frequently transported as adults on the hulls ofvessels, or as larvae in ballast water. Some species can acquiretolerance to heavy-metals (Piola and Johnston, 2006), which fa-cilitates their transport on vessels coated in anti-fouling paint andprovides them a competitive advantage against less-tolerant na-tives in polluted environments (Piola and Johnston, 2007). Moni-toring epibiota in urbanised estuaries is an essential task forunderstanding the spread and impacts of marine invaders.

Here we assess the distribution of epibiota relative to humanmodification at large scales, across multiple estuaries. We use theterm ‘modification’ to acknowledge the multiplicity of ways thathumans impact estuaries. Many anthropogenic stressors arespatially correlated so their effects cannot be distinguishedwithout experiments, but together they represent the collectiveimprint of human activities. Broad-scale surveys point to likelyassociations between ecological and environmental patterns, andindicate whether there is real-world support for predicted trends.We conducted a survey of sessile invertebrates at 84 sites acrosssix estuaries in south-eastern Australia, and tested for differencesbetween communities in replicate modified and relatively un-modified estuaries. At each site we collected environmental data,including water quality measurements (e.g. temperature, salinityand turbidity), and metal and organic contamination. The modi-fied estuaries included Port Jackson and Port Kembla, which havesome of the highest concentrations of benthic sediment metals inthe world (Birch, 2000; He and Morrison, 2001). For this reasonwe focused on metal contamination as a proxy for modificationwhen testing for potential bioindicators, but emphasize that otherstressors covary with metal contamination. We were also inter-ested in how introduced species were distributed relative to hu-man modification, so conducted additional analyses to examinethis.

2. Materials and methods

2.1. Study design

We surveyed the distribution of sessile invertebrates in six estuaries in NewSouthWales (NSW), Australia (Fig. 1). Estuaries were classified as either “modified” or “un-modified” by human activities. These labels are used in relative terms e unmodifiedestuaries are subject to less disturbance but still accommodate some human activities.Port Kembla, Port Jackson and Botany Bay are heavilymodified estuarieswith 80e100years of industrialisation andurbanisation (Birch,1996; He andMorrison, 2001; Irvineand Birch,1998). Port Hacking, TheClyde and Jervis Bayare estuaries that are relativelyless modified by urbanisation and have no history of major industry. In addition,trawling and commercial fishing are restricted in the Marine Protected Areas of TheClydeand Jervis Bay (ANZECC,1999). Eachestuarywasdivided into “inner” and “outer”zones, and in each zonewe selected 7 sites. Zoneswere determined by natural breaksin the geomorphology and water quality of estuaries, where the prevailing influencechanged from predominantly oceanic to estuarine.

2.2. Sampling units

At each site we deployed a 40 � 40 cm grey PVC panel, held to the seabed witheither a sand-screw (Wombat®) or 30 kg weight, depending on site suitability. Asettlement plate (11 �11 cm roughened black Perspex) was attached to each side ofeach panel, such that two settlement plateswere deployed per site. Settlement plateswere the sampling unit on which sessile invertebrate communities were censused.Panels were deployed for 3 months between November 2009 and February 2010.Upon collection, settlement plates containing assemblages were removed frompanels and preserved in 7% formaldehyde buffered in seawater. Communities werecensused by placing a 10� 10 cm grid over each settlement plate, and recording taxathat occurred under 100 evenly spaced points. All organisms were identified to thelowest possible taxonomic level. The origin of each taxa (native, cryptogenic, intro-duced, or unidentified) was determined with reference to the literature.

2.3. Environmental variables

At each site wemeasured a suite of environmental variables including metal andPAH concentrations, andwater quality variables. Temperature, salinity, turbidity, pH,oxygen and chlorophyll fluorescence were measured at the same depth as the set-tlement panels using a YSI-Sonde 6600-v2 (Yellow Springs, USA). This was donetwice, as panels were deployed and retrieved.

A detailed description of methodology used to quantify contamination is givenin Dafforn et al. (2012). Briefly, at each site we measured metal contamination in (i)benthic surface sediments collectedwith a Van Veen grab, (ii) sediments collected insediment traps attached to each panel, and (iii) tissue of oysters (Saccostrea glom-erata) deployed in mesh bags attached to each panel. Both sediments traps andoysters were deployed for the full 3 months of panel deployment.

Sediments and oyster tissue were acid digested and analysed for metal con-centrations (Cu, Zn, Pb, As) using ICP-AES (Perkin Elmer, Optima7300DV, USA).Benthic sediments and oyster tissues were analysed for the following PAHs usingMethod 8260 (USEPA,1996): naphthalene, acenaphthylene, acenaphthene, fluorene,phenanthrene, anthracene, fluoranthene, pyrene, benzo(a)anthracene, chrysene,benzo(a)pyrene, benzo(b & k)fluoranthene, indeno(1,2,3-cd)pyrene, dibenzo(a,h)anthracene and benzo(g,h,i)perylene. There was insufficient sediment in the sedi-ment traps to analyse PAH concentrations. Individual sediment contaminants werehighly correlated with one another, so we derived a single measure of toxicity bycalculating a mean sediment quality guideline quotient (mSQGQ). This quotient wasobtained by scaling contaminant concentrations against their guideline values andhigh sediment quality guidelines (Simpson et al., 2013), then summing scaledconcentrations at each site (Long et al., 2006). These ANZECC/ARMCANZ sedimentquality guideline values are comparable to standards by other regulatory agencies(CCME, 2002; EUWFD, 2010).

2.4. Statistical analysis

Community-level differences between samples from modified and unmodifiedestuaries were tested with PERMANOVA (Anderson, 2001). Data were square-roottransformed and samples related by BrayeCurtis similarity. Fixed factors (levels inparentheses) were Modification (modified, unmodified) and Zone (inner, outer). Es-tuary (n¼6)andSite (n¼84,nested inEstuary)wererandomfactors.Multivariatedatawere visualized with an nMDS ordination, using the metaMDS function in the Rpackage vegan (Oksanen et al., 2012). We related environmental variables to nMDSaxes with vector plots, and plotted correlations between nMDS axes and species dis-tributions. Metal concentrations in benthic sediment, sediment trap, and oystermetalwere highly correlated, so to avoid co-linearity we used benthic metal concentrationsto representmetal contamination. Correlations between a key environmental variable(Cu in benthic sediment) and nMDS axes were illustrated with a surface plot.

Univariate tests for differences in response variables according to Modification,Zone, and their interaction were tested with generalized linear mixed models(Bolker et al., 2009) using the lme4 package (Bates et al., 2011) in R v.3.0.1. Datawereaveraged for each Site prior to analysis. Modification and Zone were fixed factors,and Estuary was a random factor. We assumed Poisson distributions and estimated

Fig. 1. Location of study sites in New South Wales, Australia.

Table 1Permutational multivariate analysis of variance (PERMANOVA) testing differencesbetween disturbance categories (Modified vs. Unmodified) and Zones (Inner vs.Outer) of estuaries. Significant P-values (<0.05) involving fixed factors are in bold.

G.F. Clark et al. / Environmental Pollution 196 (2015) 12e2014

parameters with Laplace approximations (Breslowand Clayton,1993). P-values wereobtained with Z-tests (Zuur et al., 2009). To find which species were most correlatedwith contaminants and may therefore be useful bioindicators, we used boostedregression trees (Elith et al., 2008) in the R package dismo and conditional randomforests (Strobl et al., 2008) in the party package. Both of these methods handlecomplex interactions and correlated predictor variables, but since they differ inapproach we used both for comparison.

Source df SS MS Pseudo-F P(perm)

Modification 1 32,543 32,543 2.9671 0.0315Zone 1 18,577 18,577 3.2630 0.0108Modification x Zone 1 8107 8107 1.4240 0.2033Estuary (Mod) 5 55,915 11,183 5.1627 0.0001Zone x Est (Mod) 5 28,950 5790 2.6730 0.0001Site (ZoxEst (Mod)) 63 1,39,700 2218 4.0216 0.0001Residual 73 40,252 551Total 149 3,46,330

3. Results

3.1. Community patterns

The structure of epibiotic communities differed betweenmodified and unmodified estuaries (Table 1). Sites in an nMDS plotformed groups according to estuarine modification (Fig. 2a),

G.F. Clark et al. / Environmental Pollution 196 (2015) 12e20 15

although there was considerable variation within some estuaries(e.g. Port Jackson). Stress in the 2-dimensional nMDS was relativelyhigh (0.249), but similar patterns occurred in a 3-dimensionalnMDS with lower stress (0.14), so 2-d patterns are interpretable(Clarke and Gorley, 2006).

Sites in modified estuaries were generally more turbid, andoysters in modified estuaries accumulated more copper, manga-nese and PAHs (polycyclic aromatic hydrocarbons) (Fig. 2b). Faunalassemblages within these estuaries were characterized by higherabundance of several introduced species, including the solitaryascidians Ciona intestinalis and Styela plicata, and the bryozoanBugula neritina (Fig. 2c). Other bryozoans, coralline algae, and scarsof deceased invertebrates (e.g. barnacles) characterized unmodifiedestuaries. Change in community structure from unmodified tomodified estuaries was linearly correlated with copper in sus-pended sediments (Fig. 2d). Copper in suspended sedimentgenerally exceeded sediment quality guideline values (65 mg/kgdry wt) in modified estuaries.

3.2. Substrate cover and species richness

Space occupation e a proxy for resource use e was greater inmodified than unmodified estuaries (Fig. 3a). Species richness was

Fig. 2. Community patterns. (A) Multi-dimensional scaling plot (nMDS) based on BrayeCubetween samples. Acronyms for estuaries are: PH ¼ Port Hacking, JB ¼ Jervis Bay, CL ¼ Thebetween environmental covariates and nMDS axes. (C) Correlations between species and nMdisplayed and other species appear as grey crosses. (D) Contour plot of copper (mg/kg dry

higher in the outer zones of unmodified than modified estuaries,but did not differ with modification in inner zones or whole estu-aries (Fig. 3b). When grouped by origin (native, introduced orcryptogenic), most variation in distribution occurred in substratecover rather than diversity. Introduced species occupied morespace in the inner zones of modified than unmodified estuaries,while native species dominated communities in the outer zones ofmodified estuaries (Fig. 3a). Introduced species constitutedapproximately 75% of biotic cover in the inner zone of Port Jacksone a contaminated harbour with frequent shipping. In contrast,there were few introduced species in the outer zone of the Clyde e

a relatively pristine and well-flushed estuary. The only significantdifference in species richness concerned cryptogenic species, whichwere more diverse in unmodified estuaries (Fig. 3b). These speciescontributed to species diversity (Fig. 3b) but occupied relativelylittle space (Fig. 3a). There was a negative correlation between thecover of introduced species and barnacles (Poisson GLMM,b ¼ �0.024, P < 0.001).

3.3. Major taxonomic groups

The cover of some major taxonomic groups differed withmodification when inner and outer zones were considered

rtis similarity. Axes are scaled such that 1 unit equates to a doubling in dissimilarityClyde, BB ¼ Botany Bay, PJ ¼ Port Jackson, PK ¼ Port Kembla. (B) Vector correlationsDS axes. Where species positions overlap, the name of the more abundant species is

wt) found in benthic sediments, relative to nMDS axes.

Fig. 3. Introduced, cryptogenic and native species. (A) Mean percentage cover (alive and dead) across whole estuaries, and in inner and outer zones. Estuary abbreviations followthose in Fig. 2. (B) Mean species richness per plate in modified and unmodified estuaries, considering whole estuaries and inner and outer zones separately. Error bars are standarderror. P-values refer to factors Modification (M), Zones (Z), and their interaction (M � Z). P-values interpreted as significant (a ¼ 0.05) are in bold.

G.F. Clark et al. / Environmental Pollution 196 (2015) 12e2016

separately (Fig. 4). The cover of bryozoans and sponges varied inopposite directions in inner and outer zones, relative to level ofmodification (Fig. 4b). Barnacles, bryozoans and polychaetes weremost abundant in inner zones (Fig. 4). Barnacles were dominant inboth modified and unmodified inner zones, while bryozoans,polychaetes and sponges were more abundant in modified innerzones. The strongest differences between inner zones were in thecover of sponges and polychaetes, which were three to four timesmore abundant in modified than unmodified inner zones (Fig. 4b).Several taxa showed strong differences in cover in the outer zonesof modified vs. unmodified estuaries (Fig. 4). Modified outer zoneswere dominated by barnacles, while unmodified outer zones held amore even mix of taxa with higher cover of bryozoans, algae andsponges (Fig. 4b).

3.4. Potential bioindicators

Boosted regression trees and conditional random forests foundsimilar results and identified species most correlated with sedi-ment toxicity (mSQGQ). The tubeworm Hydroides elegans was by

far the best predictor of toxicity (Fig. 5a,b), to which it was posi-tively correlated (Fig. 5c). The strongest negative association wasfound for live barnacles Balanus variegatus. The introduced speciesSchizoporella errata (bryozoan) and Diplosoma listerianum (colonialascidian) had weak positive correlations with mSQGQ (Fig. 5c).

4. Discussion

Patterns we observed in epibiota challenge conventional beliefsabout the nature of human impacts in marine ecosystems. Half theestuaries we sampled were highly modified, with extensive ur-banisation, industrialisation, and internationally high levels ofmetal contamination (Birch, 2000; He and Morrison, 2001). Giventhese conditions, we predicted that modified estuaries would belower in biodiversity and contain impoverished assemblages(Johnston and Roberts, 2009). On the contrary, communities inmodified estuaries had greater species cover and were comparablein diversity to those in unmodified estuaries. Similar patterns werefound in fish (McKinley et al., 2011) and infaunal polychaetes inthese same estuaries (Dafforn et al., 2013), implying that this trend

Fig. 4. Major taxonomic groups. (A) Percentage cover of major taxa (alive only) across whole estuaries, and in inner and outer zones. Abbreviations of estuaries follow those in Fig. 2.(B) Mean percentage cover of major taxa (alive only) in modified and unmodified estuaries, considering whole estuaries and inner and outer zones separately. Error bars arestandard error. P-values refer to differences between levels of modification (M), zones (Z), and their interaction (M � Z). P-values interpreted as significant (a ¼ 0.05) are in bold.

G.F. Clark et al. / Environmental Pollution 196 (2015) 12e20 17

is common across trophic levels and faunal groups. Tests for humanimpacts on estuarine epibiota are rarely conducted at spatial scalesas large as in this study, and data here suggest that for some faunathere may be no simple change in aggregate ecological metricsrelative to human modification. A likely explanation for thesepatterns is the multiplicity of stressors that impact modified estu-aries, which in combination can have effects that are not easilypredictable.

Multiple natural and anthropogenic stressors shape speciesdistributions, and the ecological effects of these stressors vary innature and extent. Positive effects of some stressors may counteractnegative effects of others, potentially masking signals of humanimpact (Crain et al., 2008). For example, metal contamination isknown to reduce native species diversity (Piola and Johnston,2007), but other stressors may increase diversity (Connell, 1978)and abundance (Gray, 1979; Pearson and Rosenberg, 1978) at localscales. Nutrients promote primary production (Falkowski et al.,1998) and therefore potential food for filter-feeding invertebrates.Growth and biomass of bryozoans, for example, can increase inresponse to nutrient enrichment (Hartikainen et al., 2009). Artifi-cial structures provide additional habitat for epibiota (Dafforn et al.,2009), and propagule pressure is higher in areas of reduced flush-ing and more frequent shipping (Floerl and Inglis, 2003; Johnston

et al., 2008; Ruiz et al., 2000). Furthermore, stressors can interactwith one another (Crain et al., 2008) or change the variability ofspecies abundance rather than the mean (Warwick and Clarke,1993). Therefore, small-scale aggregate measures of species di-versity and abundance may not adequately capture complex effectsof human disturbance, and inference based on these measuresalone risks being overly reductionist.

In lieu of using preconceived changes in diversity and abun-dance to signal impact, characteristics shared by communities inmodified areas are potential indicators (Clarke andWarwick, 2001;Hewitt et al., 2005) andmechanistic links to known stressors add toweight of evidence. Compared with unmodified estuaries, com-munities in modified estuaries used more of the primary limitingresource (substrate), and had more introduced species in their in-ner zones. This is consistent with the predicted ecological conse-quences of altered hydrology (Floerl and Inglis, 2003), and theincreased arrival of non-native species via shipping (Ruiz et al.,2000). Using converging community characteristics as a signal ofimpact does not make assumptions about the direction or nature ofimpacts (Walters and Coen, 2006), which may vary with theidentity and combination of stressors. It does, however, assumethat communities will respond similarly, which may not hold ifthere are major differences in initial species composition.

Fig. 5. Potential bioindicators. (A) Ranked variable importance of taxa to sediment toxicity (mSQGQ) according to boosted regression trees, and (B) conditional random forests. (C)Scatterplots showing the relationship between sediment toxicity and potential bioindicators. Fitted lines are Poisson GLM. Plotted taxa were those in the 15 most importantpredictors according to both boosted regression trees and conditional random forests.

G.F. Clark et al. / Environmental Pollution 196 (2015) 12e2018

Arguments can be made for and against whether increasedalpha diversity through species invasion should be viewed asdesirable (Sax and Gaines, 2003). It is generally agreed that whilesome non-native species can improve ecosystem function, in-vasions lead to biotic homogenization and a decline in beta di-versity at larger spatial scales (McKinney and Lockwood, 1999;Olden et al., 2004). Patterns in our study suggest that these con-cepts could be generalized to include the variety of other humanactivities that potentially increase alpha diversity, productivity, andperhaps ecosystem function, such as the provision of artificialhabitat (Dafforn et al., 2009) and nutrient enrichment (Falkowskiet al., 1998). We found convergence of community traits

associated with human disturbance, analogous to biotic homoge-nization through invasion, but at the community rather than spe-cies level. Ecosystem health may be best judged by comparingequivalent ecosystems at large scales and determining the diversityof communities, rather than species.

One hypothesis to explain the maintenance of species rich-ness in modified estuaries is that native species loss iscompensated by introduced species gain, as occurs during biotichomogenization (McKinney and Lockwood, 1999). Introducedspecies are generally more abundant and diverse in areas of highhuman activity, due to higher propagule pressure (Clark andJohnston, 2009), pollution tolerance (Piola and Johnston, 2007)

G.F. Clark et al. / Environmental Pollution 196 (2015) 12e20 19

and larval retention (Floerl and Inglis, 2003). While intuitive, thereplacement hypothesis is not well supported by our data. In theinner zones, where contaminants were most concentrated, nativediversity was higher in modified than unmodified estuaries. Asimilar number of introduced species were present in each es-tuary, but they were more abundant in modified estuaries. Itappears that introduced species can establish in pristine estu-aries but remain at relatively low abundance, possibly due toexisting environmental constraints or stronger ecological in-teractions such as predation.

Another hypothesis is that a finite sampling period (3 months inthis study) creates bias in favour of fast-recruiting taxa, preferen-tially sampling weedy species that may be more common inmodified estuaries. Rarer, long-lived species may still inhabit pris-tine estuaries, but require more sampling to be detected. Again, ourdata provided little evidence in support of this hypothesis. If it weretrue, we would expect steeper rarefaction curves in unmodifiedthan modified estuaries, indicating a higher proportion of unsam-pled species. Rarefaction curves were, however, similar in eachestuary, suggesting that sampling efficiency was comparable be-tween levels of modification. Nonetheless, sampling for longerperiods and at different times of year would provide additionalinformation to assess impacts on epibiota at different stages ofdevelopment.

We found a strong potential bioindicator in the polychaetetubeworm H. elegans, which was positively correlated with sedi-ment toxicity. Adults of this species can exhibit some tolerance toheavy metals and are commonly found on the hulls of vesselscoated in antifouling paint (Floerl et al., 2004). Multiple field ex-periments that artificially increased water-borne copper concen-trations produced recruitment assemblages with higher densitiesof H. elegans (Johnston and Keough, 2002, 2003), suggesting thateither low concentrations of contaminants induce settlement (i.e.hormesis), or that tolerant species indirectly benefit from reducedcompetition with more sensitive species (Johnston and Keough,2003). The consensus between small-scale experiments and ourlarge-scale survey provides a mechanistic link between contami-nation and the distribution of this species, and demonstrates thatprocesses detected in experimental studies can translate into large-scale ecological patterns.

The distribution of some other taxa relative to modification wasless intuitive, and contrary to expectations based on the literature.Sponges are often considered bioindicators of pristine environ-ments due to their sensitivity to heavymetals (Cebrian et al., 2007),but in this study were more abundant and diverse in modifiedestuaries with higher contamination. The most likely explanation isthat benefits of nutrients, low light, and extensive artificial habitatsoutweigh harm caused by metal contamination. We did not mea-sure nutrients directly, but modified estuaries had higher turbidityand chlorophyll, indicative of nutrient enrichment. Turbidity alsoreduces light attenuation through the water column, creating lowlight conditions favoured by non-coral invertebrates (Miller andEtter, 2008). There is limited understanding of the impacts of nu-trients on epibiota, and experimental studies are needed to fill thisknowledge gap.

In conclusion, this study demonstrates that human activitieshave measurable impacts on estuarine epibiota, but the nature ofimpacts may not be readily predictable due to the number andvaried effects of multiple stressors. Impacts can be observed as theconverging characteristics of stressed communities and might onlybe detectable through large-scale comparisons of communitiesexposed to different amounts of stress. Conservation efforts aimedat maintaining alpha diversity may be less successful at preservingecological integrity than efforts aimed at maintaining diversity ofboth species and communities at larger scales.

Acknowledgements

We thank members of the Subtidal Ecology and Ecotoxicologylaboratory and volunteers for their assistance. We also thankBluescope Steel and the NSW Marine Parks Authority for theirsupport. This research was primarily funded by ARC Linkage GrantLP0990640.

References

Anderson, M.J., 2001. A new method of non-parametric multivariate analysis ofvariance. Austral Ecol. 26, 32e46.

ANZECC, 1999. Strategic plan of action for the national representative system ofmarine protected areas: a guide for action by Australian governments. In:Australian and New Zealand Environment and Conservation Council Task Forceon Marine Protected Areas. E.A, Canberra.

ANZECC, 2001. Australian and New Zealand Guidelines for Fresh and Marine WaterQuality: Aquatic Ecosystems e Rationale and Background Information.Australian and New Zealand Environment and Conservation Council.

Bates, D., Maechler, M., Bolker, B., 2011. lme4: Linear Mixed-effects Models Using S4Classes. R package version 0.999375-40.

Birch, G.F., 1996. Sediment-bound metallic contaminants in Sydney's estuaries andadjacent offshore, Australia. Estuar. Coast. Shelf Sci. 42, 31e44.

Birch, G.F., 2000. Marine pollution in Australia, with special emphasis on centralNew South Wales estuaries and adjacent continental margin. Int. J. Environ.Pollut. 13, 573e607.

Bolker, B.M., Brooks, M.E., Clark, C.J., Geange, S.W., Poulsen, J.R., Stevens, M.H.H.,White, J.-S.S., 2009. Generalized linear mixed models: a practical guide forecology and evolution. Trends Ecol. Evol. 24, 127e135.

Borja, A., Bricker, S., Dauer, D.M., Demetriades, N.T., Ferreira, J.G., Forbes, A.T.,JHutchings, a., Jia, X., Kenchington, R., Marques, J.C., Zhu, C., 2008. Overview ofintegrative tools and methods in assessing ecological integrity in estuarine andcoastal systems worldwide. Mar. Pollut. Bull. 56, 1519e1537.

Breslow, N.E., Clayton, D.G., 1993. Approximate inference in generalized linearmixed models. J. Am. Stat. Assoc. 88, 9e25.

Burton, G.A., Johnston, E.L., 2010. Assessing contaminated sediments in the contextof multiple stressors. Environ. Toxicol. Chem. 29, 2625e2643.

Carignan, V., Villard, M.C., 2002. Selecting indicator species to monitor ecologicalintegrity: a review. Environ. Monit. Assess. 78, 45e61.

CCME, 2002. Canadian Sediment Quality Guidelines for the Protection of AquaticLife e Summary Tables Update 2002.

Cebrian, E., Uriz, M., Turon, X., 2007. Sponges as biomonitors of heavy metals inspatial and temporal surveys in northwestern Mediterranean: multispeciescomparison. Environ. Toxicol. Chem. 26, 2430e2439.

Clark, G.F., Johnston, E.L., 2009. Propagule pressure and disturbance interact toovercome biotic resistance of marine invertebrate communities. Oikos 118,1679e1686.

Clarke, K.R., Gorley, R.N., 2006. PRIMER V6: User Manual/Tutorial.Clarke, K.R., Warwick, R.M., 2001. Changes in Marine Communities: an Approach to

Statistical Analysis and Interpretation, second ed. PRIMER-E, Plymouth.Connell, J.H., 1978. Diversity in tropical rain forests and coral reefs. Science 199,

1302e1310.Crain, C.M., Kroeker, K., Halpern, B.S., 2008. Interactive and cumulative effects of

multiple human stressors in marine systems. Ecol. Lett. 11, 1304e1315.Dafforn, K.A., Glasby, T.M., Johnston, E.L., 2009. Links between estuarine condition

and spatial distributions of marine invaders. Divers. Distributions 15, 807e821.Dafforn, K.A., Kelaher, B.P., Simpson, S.L., Coleman, M.A., Hutchings, P.A., et al., 2013.

Polychaete richness and abundance enhanced in anthropogenically modifiedestuaries despite high concentrations of toxic contaminants. PLoS ONE 8 (9),e77018. http://dx.doi.org/10.1371/journal.pone.0077018.

Dafforn, K.A., Simpson, S.L., Kelaher, B.P., Clark, G.F., Komyakova, V., Wong, C.K.C.,Johnston, E.L., 2012. The challenge of choosing environmental indicators ofanthropogenic impacts in estuaries. Environ. Pollut. 163.

Edgar, G.J., Barrett, N.S., Graddon, D.J., Last, P.R., 2000. The conservation significanceof estuaries: a classification of Tasmanian estuaries using ecological, physicaland demographic attributes as a case study. Biol. Conserv. 92, 383e397.

Elith, J., Leathwick, J.R., Hastie, T., 2008. A working guide to boosted regressiontrees. J. Animal Ecol. 77, 802e813.

EUWFD, 2010. Technical Guidance for Deriving Environmental Quality Standards.Draft Version 5.0. Implementation Strategy for the Water Framework Directive(2000/60/EC).

Falkowski, P.G., Barber, R.T., Smetacek, V., 1998. Biogeochemical controls andfeedbacks on ocean primary production. Science 281, 200e206.

Floerl, O., Inglis, G.J., 2003. Boat harbour design can exacerbate hull fouling. AustralEcol. 28, 116e127.

Floerl, O., Pool, T.K., Inglis, G.J., 2004. Positive interactions between nonidigenousspecies facilitate transport by human vectors. Ecol. Appl. 14, 1724e1736.

Gray, J.S., 1979. Pollution induced changes in populations. Philosophical Trans. R.Soc. Ser. B 286, 546e561.

Hartikainen, H., Johnes, P., Moncrieff, C., Okamura, B., 2009. Bryozoan populationsreflect nutrient enrichment and productivity gradients in rivers. Freshw. Biol.54, 2320e2334.

G.F. Clark et al. / Environmental Pollution 196 (2015) 12e2020

He, Z., Morrison, R.J., 2001. Changes in the marine environment of Port KemblaHarbour, NSW, Australia, 1975-1995: a review. Mar. Pollut. Bull. 42, 193e201.

Hewitt, J.E., Anderson, M.J., Hickey, C., Kelly, S., Thrush, S.F., 2009. Enhancing theecological significance of contamination guidelines through integration withcommunity analysis. Environ. Sci. Technol. 43, 2118e2123.

Hewitt, J.E., Anderson, M.J., Thrush, S.F., 2005. Assessing and monitoring ecologicalcommunity health in marine systems. Ecol. Appl. 15, 942e953.

Hyland, J., Balthis, W., Engle, V., Long, E.R., Paul, J., Summers, J., Van Dolah, R., 2003.Incidence of stress in benthic communities along the US Atlantic and Gulf ofMexico coasts within different ranges of sediment contamination from chem-ical mixtures. Environ. Monit. Assess. 81, 149e161.

Irvine, I., Birch, G.F., 1998. Distribution of heavy metals in surficial sediments of portJackson, Sydney, New South Wales. Aust. J. Earth Sci. 45, 247e304.

Johnston, E.L., Keough, M.J., 2002. Direct and indirect effects of repeated pollutionevents on marine hard-substrate assemblages. Ecol. Appl. 12, 1212e1228.

Johnston, E.L., Keough, M.J., 2003. Competition modifies the response of organismsto toxic disturbance. Mar. Ecol. Prog. Ser. 251, 15e26.

Johnston, E.L., Piola, R.F., Clark, G.F., 2008. The role of propaugle pressure in invasionsuccess. In: Rilov, G., Crooks, J.A. (Eds.), Marine Bio invasions: Ecology, Con-servation, and Management Perspectives. Springer.

Johnston, E.L., Roberts, D.A., 2009. Contaminants reduce the richness and evennessof marine communities: a review and meta-analysis. Environ. Pollut. 157,1745e1752.

Kennish, M.J., 2002. Environmental threats and environmental future of estuaries.Environ. Conserv. 29, 78e107.

Knott, N.A., Aulbury, J.A., Brown, T.H., Johnston, E.L., 2009. Contemporary ecologicalthreats from historical pollution sources: impacts of large-scale resuspension ofcontaminated sediments on sessile invertebrate recruitment. J. Appl. Ecol. 46,770e781.

Long, E.R., Ingersoll, C.G., Macdonald, D.D., 2006. Calculation and uses of meansediment quality guideline quotients: a critical review. Environ. Sci. Technol. 40,1726e1736.

McKinley, A., Taylor, M.D., Miskiewiecz, A., Johnston, E.L., 2011. High levels ofsediment contamination have little influence on estuarine beach fish commu-nities. PLoS ONE 6, e26353.

McKinney, M.L., Lockwood, J.L., 1999. Biotic homogenization: a few winnersreplacing many losers in the next mass extinction. Trends Ecol. Evol. 14,450e453.

Miller, R.J., Etter, R.J., 2008. Shading facilitates sessile invertebrate dominance in therocky subtidal Gulf of Maine. Ecology 89, 452e462.

Morrisey, D.J., 1995. Estuaries. In: Underwood, A.J., Chapman, M.G. (Eds.), CoastalMarine Ecology. University of New South Wales Press Ltd, Sydney, pp. 152e170.

Newell, R.C., Seiderer, L.J., Hitchcock, D.R., 1998. The impact of dredging works incoastal waters: a review of the sensitivity to disturbance and subsequent re-covery of biological resources on the sea bed. Oceanogr. Mar. Biol. Annu. Rev. 36,127e178.

Oksanen, J., Blanchet, F.G., Kindt, R., Legendre, P., Minchin, P.R., O'Hara, R.B.,Simpson, G.L., Solymos, P., Stevens, H.H., Wagner, H., 2012. Vegan: CommunityEcology Package.

Olden, J.D., Leroy Poff, N., Douglas, M.R., Douglas, M.E., Fausch, K., 2004. Ecologicaland evolutionary consequences of biotic homogenization. Trends Ecol. Evol. 19,18e24.

Pearson, T.H., Rosenberg, R., 1978. Macrobenthic succession in relation to organicenrichment and pollution of the marine environment. Oceanogr. Mar. Biol.Annu. Rev. 16, 229e311.

Piola, R.F., Johnston, E.L., 2006. Differential resistance to extended copper exposurein four introduced bryozoans. Mar. Ecol. Prog. Ser. 311, 103e114.

Piola, R.F., Johnston, E.L., 2007. Pollution reduces native diversity and increasesinvader dominance in marine hard-substrate communities. Divers. Distribu-tions 14, 329e342.

Roberts, D.A., 2012. Causes and ecological effects of resuspended contaminatedsediments (RCS) in marine environments. Environ. Int. 40, 230e243.

Roy, P.S., Williams, R.J., Jones, A.R., Yassini, I., Gibb, P.J., Coates, B., West, R.J.,Scanes, P.R., Hudson, J.P., Nicholi, S., 2001. Structure and function of south-eastAustralian estuaries. Estuar. Coast. Shelf Sci. 53, 351e384.

Ruiz, G.M., Carlton, J.T., Grosholz, E.D., Hines, A.H., 1997. Global invasions of marineand estuarine habitats by non-indigeneous species: mechanisms, extent andconsequences. Am. Zool. 37, 621e632.

Ruiz, G.M., Fofonoff, P.W., Carlton, J.T., Wonham, M.J., Hines, A.H., 2000. Invasion ofcoastal marine communities in North America: apparent patterns, processes,and biases. Annu. Rev. Ecol. Syst. 31, 481e531.

Sax, D.F., Gaines, S.D., 2003. Species diversity: from global decreases to local in-creases. Trends Ecol. Evol. 18, 561e566.

Scanes, P., Coade, G., Doherty, M., Hill, R., 2007. Evaluation of the utility of waterquality based indicators of estuarine lagoon condition in NSW, Australia. Estuar.Coast. Shelf Sci. 74, 306e319.

Scanes, P., Scanes, K., 1995. Environmental problems due to disposal of wastes. In:Underwood, A.J., Chapman, M. (Eds.), Coastal Marine Ecology of TemperateAustralia. UNSW Press, Sydney, pp. 297e310.

Simpson, S.L., Batley, G.E., Chariton, A.A., 2013. Revision of the ANZECC/ARMCANZSediment Quality Guidelines. CSIRO Land and Water Science Report 08/07.

Strobl, C., Boulesteix, A., Kneib, T., Augustin, T., Zeileis, A., 2008. Conditional variableimportance for random forests. BMC Bioinformatics 9, 307.

USEPA, 1996. Method 8260B Volatile Organic Compounds by Gas Chromatography/Mass Spectrometry (GC/MS). US Environmental Protection Agency,Washington.

Van Dolah, R., Hyland, J., Holland, A., Rosen, J., Snoots, T., 1999. A benthic index ofbiological integrity for assessing habitat quality in estuaries of the southeasternUSA. Mar. Environ. Res. 48, 263e269.

Walters, K., Coen, L.D., 2006. A comparison of statistical approaches to analyzingcommunity convergence between natural and constructed oyster reefs. J. Exp.Mar. Biol. Ecol. 330, 81e95.

Warwick, R.M., Clarke, K.R., 1993. Increased variability as a symptom of stress inmarine communities. J. Exp. Mar. Biol. Ecol. 172, 215e226.

Zuur, A.F., Ieno, E.N., Walker, N.J., Saveliev, A.A., Smith, G.M., 2009. Mixed EffectsModels and Extensions in Ecology with R. Springer, New York.