Improved Approaches to Sediment Quality Assessment · The project was designed to meet the needs of regulatory agencies for an improved sediment quality assessment framework to unambiguously

Environmental Trust Report 1

NSW Environmental Trust Research Project No 22003/RD/G0002

Improved Approaches to Sediment Quality Assessment

Final Report

CSIRO Land and Water Science Report 7/07

G.E. Batley, A.A. Chariton, S.L. Simpson, A.C. Roach and W.A. Maher

Prepared for NSW Environmental Trust

February 2007

Environmental Trust Report ii

Enquiries should be addressed to:

Dr Graeme Batley

Centre for Environmental Contaminants Research

CSIRO Land and Water

Private Mailag 7, Bangor NSW 2234

Phone 02 9710 6830

Email: [email protected]

Environmental Trust Final Report iii

Executive Summary In March 2004, a consortium comprising CSIRO’s Centre for Environmental Contaminants Research, the University of Canberra and the NSW Department of Environment and Conservation commenced a 2 ½ year project partly funded by The NSW Environmental Trust, on improved approaches to sediment quality assessment.

The project aimed to develop a more defensible way of establishing the risk of harm associated with contaminated sediments, by advancing the interim approach of the ANZECC/ARMCANZ sediment quality guidelines framework to include, in addition to chemistry and toxicity tests, the assessment of bioaccumulation and ecological effects. An expanded framework was to be proposed that allowed integration of all of the above lines of evidence in assessments of sediment quality, that will provide NSW EPA with a more sound basis for determining the need for sediment remediation in contaminated estuary and harbour sites.

The project investigated metal and PAH bioaccumulation, through transplant experiments at contaminated field sites and tests on laboratory spiked sediments, using bivalves and polychaete worms. The use of biomarkers of cellular integrity and function was developed in a parallel project. Current trends in measuring organism and benthic community health were reviewed, including the identification of feeding guilds and exposure pathways for organisms, and the use of newer indices for analysing sub lethal and community effects, such as benthic abundance and biomass curves, to provide more meaningful assessment of effects. Favoured approaches were successfully tested at field sites exhibiting a range of contaminants and degree of contamination The project has resulted in refining the use of ecological and bioaccumulation data as part of a new weight of evidence assessment framework for sediment quality assessment, and the approach is already being applied in sediment quality assessments in several states where there are major sediment dredging and/or remediation projects. Filling out the details of the newer lines of evidence and putting them into the WOE framework were the objectives of this study. This would substantially enhance the power of the existing assessments based largely on chemistry and ecotoxicology. A more robust framework is currently being developed for Agriculture Fisheries and Forestry Australia, independent of this study that will see the approach adopted by the Commonwealth as part of revised sediment quality guidelines. This is designed to improve the evaluation of assessment data from each of the lines of evidence, so that the findings are more transparent and less reliant on expert judgement. The full outcomes of this exercise are yet to be fully realised, however the pathways to adoption are set in place, and the benefits that will ensue are clear. This enhancing of the current ANZECC/ARMCANZ (2000) sediment quality guidelines framework, together with revising interim guidelines for certain common contaminants, will provide a defensible framework that DEC can use to enforce clean up of harbour and other sites.

Environmental Trust Final Report iv

Table of Contents Introduction............................................................................................................................ 2 Project Staff............................................................................................................................ 3 Project Description................................................................................................................ 4

Task 1. To develop passive and active biomonitoring for use in a sediment assessment framework........................................................................................................................................ 4 Task 2. To review current approaches for measuring estuarine benthic health ............................. 4 Task 3. Using the results of the review in Task 2, to undertake an assessment of benthic community health at key contaminated sites in NSW to evaluate the effectiveness of the approaches recommended by the review ....................................................................................... 5 Task 4. To field test our proposed improvements to the ANZECC/ARMCANZ sediment quality guidelines framework ...................................................................................................................... 5

Outcomes ............................................................................................................................... 6 Outputs ................................................................................................................................... 7

Task 1. Development of Passive and Active Biomonitoring............................................................... 7 Review of existing bioaccumulation data and biomarker experiments ........................................... 7 Review of Biomarker Responses .................................................................................................... 8 Experimental Studies ...................................................................................................................... 8

Task 2. Review of Current Approaches for Measuring Estuarine Benthic Health ........................... 11 Introduction.................................................................................................................................... 11 Univariate attributes and metrics................................................................................................... 11 Multivariate analysis of benthic communities ................................................................................ 12 Functional approaches .................................................................................................................. 14 Ecological stoichiometry:............................................................................................................... 14

Task 3. Assessment of Benthic Community Health at Key NSW Sites ........................................... 15 Introduction.................................................................................................................................... 15 Contaminant gradient within western Sydney Harbour ................................................................. 15 Non-metric dimensional scaling .................................................................................................... 17 PERMANOVA................................................................................................................................ 17 SIMPER analysis........................................................................................................................... 18 Canonical Correspondance Analysis (CCA) ................................................................................. 18 Lessons, future opportunities of benthic assemblage research.................................................... 19 Food web analysis......................................................................................................................... 20

Task 4. Field Testing Improvements to the ANZECC/ARMCANZ Sediment Quality Guidelines Framework......................................................................................................................................... 23

New WOE framework.................................................................................................................... 23 Application of WOE framework to key sites .................................................................................. 23

Issues, Changes and Opportunities .................................................................................. 26 Lessons from the program ............................................................................................................ 26

Acknowledgements ............................................................................................................. 26 References ........................................................................................................................... 27 Appendix 1. Publications ................................................................................................... 28

Papers in Preparation........................................................................................................................ 38 Appendix 2. Conference Presentations........................................................................... 39

Environmental Trust Final Report 2

Introduction In March 2004, a consortium comprising CSIRO’s Centre for Environmental Contaminants Research, the University of Canberra and the NSW Department of Environment and Conservation commenced a 2 ½ year project on improved approaches to sediment quality assessment, partly funded by The NSW Environmental Trust.

The project aimed to develop a more defensible way of establishing the risk of harm associated with contaminated sediments, by advancing the interim approach of the ANZECC/ARMCANZ sediment quality guidelines framework to include, in addition to chemistry and toxicity tests, the assessment of bioaccumulation and ecological effects. An expanded framework was to be proposed that allowed integration of all of the above lines of evidence in assessments of sediment quality, that will provide NSW EPA with a more sound basis for determining the need for sediment remediation in contaminated estuary and harbour sites. Specifically this project proposed to:

(i) Investigate metal and PAH bioaccumulation, through transplant experiments at contaminated field sites and tests on laboratory spiked sediments, using bivalves and polychaete worms. The use of biomarkers of cellular integrity and function will be developed in a parallel project, and these will be used to evaluate bioaccumulation thresholds.

(ii) Review current trends in measuring organism and benthic community health, including the identification of feeding guilds and exposure pathways for organisms, and the use of newer indices for analysing sublethal and community effects, such as benthic abundance and biomass curves, to provide more meaningful assessment of effects. Favoured approaches will be tested at field sites exhibiting a range of contaminants and degree of contamination. The project builds on the techniques for chemical and ecotoxicological assessment developed during an earlier highly successful Environmental Trust project that culminated in the production of a Handbook for Sediment Quality Assessment that to date has been downloaded from the CSIRO website over 34,000 times.

The project was designed to meet the needs of regulatory agencies for an improved sediment quality assessment framework to unambiguously demonstrate risk of harm from estuary and harbour sites containing contaminated sediments, where sediment quality guidelines are exceeded, and remedial action is being debated. Chemistry and ecotoxicology alone are in many cases failing to demonstrate a system at risk. An extended framework including the assessment of bioaccumulation of contaminants and the effects on benthic ecology will give greater confidence in the risk predictions. While there are examples where such studies have been undertaken, their design and interpretation is frequently flawed. More scientifically-defensible methods of assessing the risk associated with contaminated sediments in NSW harbours and estuaries will provide guidance for industry and confidence for the NSW DEC and the community that risks are being appropriately managed.

While we are able to effectively manage discharges of contaminants to harbours and estuaries, improvements in ecosystem health also require effective management of the effects of sediment contaminants. Enhancing the current ANZECC/ARMCANZ (2000) sediment quality guidelines framework, together with revising interim guidelines for certain common contaminants, will provide a defensible framework that DEC can use to enforce clean up of harbour and other sites. This report details the findings of the above research program.


Project Staff The project involved the following staff: CSIRO Centre for Environmntal Contaminants Research Dr Graeme Batley, Project Manager Dr Stuart Simpson, Joint Leader, tasks 1-4 Dr Anthony Chariton, UC Postdoctoral Fellow David Spadaro, Experimental Scientist Vicky Burston, Honours Student (University of Wollongong) University of Canberra Professor Bill Maher, Joint Leader, tasks 1-3 Anne Taylor, PhD Student Department of Environment and Conservation Dr Tony Roach, Joint Leader, tasks 2-3


Project Description

Improvements to the way in which risk of harm is assessed in contaminated sediments, require revision of the interim ANZECC/ARMCANZ sediment quality guidelines assessment framework to accommodate all relevant lines of evidence, namely chemistry, toxicity, bioaccumulation and population and community ecological effects. This will involve the prescription of scientifically-validated protocols applicable to the range of NSW harbour and estuarine sediment types. The tasks required were as follows:

Task 1. To develop passive and active biomonitoring for use in a sediment assessment framework The recent Pellston workshop on "hazard identification approach for metals and inorganic metal substances" highlighted the need to incorporate bioaccumulation (of metals) into hazard assessments for waters and sediments. As a line of evidence, bioaccumulation provides evidence of exposure for individual species that may or may not be supported by chemistry, ecology or ecotoxicology. Species at a 'contaminated' field site will generally differ from the more sensitive species used in toxicity tests, but passive monitoring showing bioaccumulation will indicate exposure, although not necessarily toxic effects. Defining a threshold above which bioaccumulation can be considered excessive is a desirable goal that will be investigated in a PhD project linked to this study. Measurements of cellular integrity, cellular function and other biomarker responses will be related to bioaccumulation in laboratory and field studies. The findings of these studies will be incorporated in this task and the site assessments made in Task 4. Active bioaccumulation will be studied at field sites in transplant experiments. Transplant organisms will include species that are native to natural sediment sites (amphipods, bivalves and polychaete worms). Bioaccumulation experiments will consider both key metals and PAHs, and exposure pathways will be carefully considered during experimental design. Relationships between sediment contamination (total and bioavailable) and bioaccumulation (and selected biomarkers) will be developed. Ecological information on field sites (existing or new (Task 3)) will allow more complex relationships to be developed, e.g. describing the characteristics of the species accumulating contaminants and which species are absent - as an indicator of direct or indirect sensitivity to the contamination. The benefits and difficulties associated with the active and passive assessment approaches will be assessed and the recommeneded procedures will be proposed for use in the enhanced framework for sediment quality assessment.

Task 2. To review current approaches for measuring estuarine benthic health Over the past decade, considerable advances have been made in our understanding of benthic communities and in the development of statistical techniques for assessing benthic community health. Examples include the use of asymmetrical impact versus control analyses, abundance/biomass curves and similar indices, and hypothesis-based multivariate analyses based on permutation testing and canonical correspondence analysis. Even more recently, new multivariate analyses of community composition such as canonical analysis of principal coordinates and other forms of constrained ordination, and new indices of biotic integrity for estuarine benthic communities have been developed which may further improve our ability to detect effects from contamination. Considerable data on benthic communities in Sydney Harbour and other estuarine and marine sediments already exist in consultants' reports and internal reports of agencies such as Sydney Water, the Australian Museum and Sydney Olympic Park Authority. These reports will be reviewed and, where possible, the data will be reanalysed using some of the approaches reviewed to help identify areas with significant environmental effects from contamination which can potentially be used for assessments as


part of Task 3. Information will be obtained on issues of sample size and taxonomic resolution. The data will also be evaluated to determine if there are any consistent patterns of species composition, feeding types etc., which may lead to selection of possible indicator organisms in NSW sediments.

Task 3. Using the results of the review in Task 2, to undertake an assessment of benthic community health at key contaminated sites in NSW to evaluate the effectiveness of the approaches recommended by the review Benthic population effects and community assessments will be undertaken at locations that have been exposed to significant anthropogenic pressures (e.g. past or present industrial activity). These locations will be predominantly harbour areas adjacent to industrial sites and will include sites dominated by single contaminant groups (e.g. Five Dock Bay - predominantly metals; Hunter River - predominantly PAHs and hydrocarbons) and those with contaminant mixtures (e.g. Homebush Bay - metals and organics). Sediments at these sites will be characterized for chemical composition and physical properties. The results will be used to recommend more appropriate procedures for the assessment of effects on benthic communities.

Task 4. To field test our proposed improvements to the ANZECC/ARMCANZ sediment quality guidelines framework Full application of the framework will be undertaken at the above sites. The key findings will be summarised and recommended guidance on application of the framework will be documented. The reliability of risk predictions will be further enhanced by related CSIRO and University of Canberra research developing more robust guideline trigger values for the key contaminants being investigated here, since it is believed that in many cases the values are too conservative.


Outcomes This project has resulted in refining the use of ecological and bioaccumulation data as part of a new weight of evidence assessment framework for sediment quality assessment. The basic concept of WOE assessments was outlined as part of lectures by project staff to user groups in several capital cities in 2005/6 (See Appendix 2), as well as being documented in the Handbook of Sediment Quality Assessment (Simpson et al., 2005) produced as part of an earlier Trust grant. This approach is already being applied in sediment quality assessments in several states where there are major sediment dredging and/or remediation projects. Filling out the details of the newer lines of evidence and putting them into the WOE framework were the objectives of this study. This would substantially enhance the power of existing assessments based largely on chemistry and ecotoxicology. A more robust framework is currently being developed for Agriculture Fisheries and Forestry Australia, independent of this study, that will see the approach adopted by the Commonwealth as part of revised sediment quality guidelines. This is designed to improve the evaluation of assessment data from each of the lines of evidence, so that the findings are more transparent and less reliant on expert judgement. The full outcomes of this exercise are yet to be fully realised, however the pathways to adoption are set in place, and the benefits that will ensue are clear.


Outputs This section reports on the specific outputs of the four tasks undertaken for this project. It incorporates discussions of the methodology.

Task 1. Development of Passive and Active Biomonitoring This task had the following milestones: Year 1: Review of existing bioaccumulation data and biomarker techniques. Commence laboratory experiments investigating contaminant (Cd, Cu, Ni, Pb, Zn) bioaccumulation and biomarkers responses. Commence passive biomonitoring (field collection of organisms and analyses). Develop active biomonitoring (caged organism transplant) methods. Select organisms for PAH bioaccumulation studies and test methods.

Year 2: Completion of laboratory experiments investigating bioaccumulation-biomarker relationships. Complete passive biomonitoring work. Undertake active biomonitoring (caged organism transplant) experiments at contaminated field sites. Complete PAH bioaccumulation-bioavailability studies and develop relationships.

Year 3: Laboratory and field boaccumulation experiments completed and passive/active biomonitoring relationships developed.

All milestones were successfully completed, except for those in Year 3, which are part of a PhD project that is still being completed

Outputs in this task are discussed below:

Review of existing bioaccumulation data and biomarker experiments A comprehensive review of bioaccumulation data and of biomarker experiments was undertaken. In summary, the following points are worthy of mention:

1. Bioaccumulation of metals by invertebrates is complex and strongly affected by organism feeding behaviour (ingestion rates, selectivity of feeding) and the different sources of metals (overlying waters, pore waters, sediments and speciation.

2. Biota-to-sediment accumulation factors (BSAFs) commonly used as indices for

bioaccumulation for non-ionic organic chemicals were never intended for use with metals, although body concentrations of metals (body residues) may provide useful information on possible effects if strong and clear relationships exist between bioaccumulation and toxic effects. For metals that are sequestered into non-toxic forms or are regulated over the concentration range of interest, the use of body concentrations to predict effects is not appropriate. The usefulness of metal bioaccumulation data for sediment quality assessment purposes has yet to be fully evaluated.

3. If metal bioaccumulation data are to be used in the sediment quality assessment, care

should be taken to use a suitable depuration period (e.g. 24 h) for gut clearance before body concentrations of organisms are determined.

4. Bioaccumulation and biomagnification of non-ionic organic chemicals can be evaluated

through field measurements (body residues in field–collected organisms), laboratory measurements (bioaccumulation tests with laboratory animals), surrogate measures of bioaccumulation (biometric methods, gut fluid extraction) and modelling (biota-to-sediment


accumulation factors (BSAF) and theoretical bioaccumulation potential (TBP)). Each of these has been thoroughly reviewed.

Review of Biomarker Responses An extensive review of the biomarker literature was also undertaken to determine which techniques and which organism look the most promising for trace metal exposure and effects measurements in environmental assessment programs. Significant findings were:

1. The bivalve Tellina deltoidalis, and the Sydney cockle Anadara trapezia both common estuarine and marine organisms of south eastern Australia which have been used successfully in trace metal exposure studies, and are suitable for both laboratory studies and for deployment in cage trials.

2. A combination of biomarkers at several levels of biological organisation from early

sensitive molecular level responses, cell integrity measures, and the physiological response of the individual are being considered.

Molecular level responses include the direct measurement of increased levels of reactive oxygen species (ROIs): O2

- , H2O2, .OH; measurement of adaptive responses through increased activities of oxidant catalysing and reduction enzyme systems which inhibit the formation of ROIs by scavenging and reducing them to non-reactive molecules; cytoplasmic enzyme superoxide dismutase (SOD); antioxidant enzyme catalase (CAT); glutathione peroxidase (GPx) and reductase (GRx) enzyme systems; and reduced and oxidised glutathione (GSH and GSSG) ratio; lipid peroxidation via thiobarbituric acid reactive substances (TBARS) test for malonaldehyde (MDA, a byproduct of lipid peroxidation); methemoglobin (MetHb). Cell integrity biomarkers will be looked at to determine whether observed biochemical effects can be linked to higher order effects. These biomarkers will include decreased lysosomal activity and membrane stability; evidence of histopathological alterations; physiological effects measurements such as growth rate; and condition e.g. evidence of lesions.

Experimental Studies Bioaccumulation of metals: Studies are underway to investigate the most appropriate way of using metal accumulation data in a sediment quality risk assessment sense. The bioavailability of metals in sediments is dependent on (i) speciation (e.g. metal binding with sulfide, iron hydroxide, organic matter), (ii) sediment-water partitioning relationships (e.g. Kd = [Sediment-Cu, mg/kg]/[Water-Cu, mg/L), (ii) organism physiology (uptake rates from waters, assimilation efficiencies from particulates), and (iv) organism feeding behaviour. Studies are underway to extend our exposure-effects model to a wider range of benthic organisms, and to investigate the application of the model for deriving SQGs for metals in sediments. The collection and use of metal bioaccumulation data is an important component of this study. Bioaccumulation of PAHs: Although bioaccumulation tests are specified as a requirement for assessing the suitability of sediments for unconfined ocean disposal in Australia, these tests are seldom undertaken due to a lack of developed methods (Environment Australia, 2002). An honours student, Victoria Burston, from the University of Wollongong, undertook this component of the project. She investigated the application of surrogate (biomimetic) methods for assessing the bioavailability and bioaccumulation of hydrophobic organic contaminants in sediments. The focus was on PAHs with 28-d bioaccumulation of PAHs by a bivalve (Tellina deltoidalis) and a polychaete worm (Australoneris nephtys) being compared with the ‘bioavailable’ fraction estimated using 6-h XAD-2 desorption and gut-fluid mimic biomimetric methods. PAH-contaminated marine sediments were sourced from sites known to contain very high PAH concentrations.


Bioaccumulation bioassays usually take weeks to perform (e.g. 10-42 days), making them an expensive component of sediment quality assessments. Consequently, the application of surrogate methods that rapidly determine the bioavailable PAHs in sediments are becoming increasingly popular. Comparisons were made between the PAHs accumulated by the benthic bivalve, Tellina deltoidalis, and the fraction of bioavailable PAHs determined using a 6-h XAD-2 resin desorption method and a gut fluid mimic (GFM) extraction method. T. deltoidalis proved to be a useful organism for determining bioaccumulation of PAHs from sediments and there were significant positive relationships between PAH bioaccumulation by the bivalves and sediment PAH concentrations. These relationships were not improved by normalising the sediment PAH concentrations to the organic carbon concentration, as is prescribed in the ANZECC/ARMCANZ (2000) guidelines. The average percentage lipid content of the bivalves was 1.47±0.22% and BSAFs for total-PAHs ranged from 0.06 to 0.80 (kg OC/kg lipid), and were similar in range to those of other studies. Comparisons of the PAHs accumulated by T. deltoidalis and the XAD-2 and GFM extractable PAHs indicated that these methods required further development before they can be applied routinely as surrogate methods for assessing the bioavailability of PAHs in sediments. The XAD-2 and GFM methods both extracted varying amounts of PAHs from the sediments. Low concentrations of PAHs were extracted by the GFM method (0.2-3.6% of total-PAHs in sediments) and this may have been due to the high concentrations of organic carbon and/or oil. The GFM results were inadequate for generalising about the bioavailability of the PAHs in the sediments. The XAD-2 method extracted greater amounts of PAHs (3-34% of total-PAHs in sediments), however the total-PAH concentrations in the sediments provided a better, or equally good, prediction of PAH bioaccumulation by T. deltoidalis. It was anticipated that these surrogate methods could be used in conjunction with, or substitute for, more expensive organism bioaccumulation assays as part of weight-of-evidence assessments. The present study indicates that the XAD-2 and GFM methods require further refinement and future research should be directed towards lowering detection limits for both XAD-2 resin and GFM type extractions and obtaining comparative data for a greater range of sediment types, contaminant classes and concentrations, and organisms of different feeding guilds and with different gut chemistry. Such methods may also improve our understanding of the different mechanisms of PAH exposure, e.g. bioaccumulation from dissolved (filter feeding, absorption through membranes) or particulate (sediment ingestion) sources. A paper on this research published in Chemosphere is attached in Appendix 1. Passive and active biomonitoring: In order to assess the sensitivity of the selected organisms and to gain understanding of their uptake rates, a series of 25-day cadmium, lead and selenium bioaccumulation experiments have been performed on the bivalves Tellina deltoidalis and Gari modesta., while 45-day cadmium, lead and selenium bioaccumulation experiments have been performed on the bivalve, Anadara trapezia. Biomarker experiments based on trace metal exposure have been trialled on the three taxa using an antioxidant assay (measuring ratio changes between reduced and oxidised glutathione (GSH and GSSG)) and a lipid peroxidation assay (malondialdehyde). Experimental work has also commenced on testing a lysosomal stability assay (neutral red dye retention) using bivalves, as well as the use of fish liver cell lines as “reference” material for testing the various biomarker responses. Cadmium, lead and selenium exposure experiments using the bivalves, showed decreases in the ratios of GSH to GSSG and increased expression in malondialdehyde. Lysosomal stability of cells also decreased.


Additional work examining the responses of these biomarkers to sub-cellular fractions of metals is due to commence after the purchase of an ultra-centrifuge. To date, the results of this sub-task clearly indicate that at environmentally realistic contamination concentrations changes in the expressions of biomarkers occur. This indicates that organisms are being stressed when exposed to metals, providing an early-warning for irreversible metabolic dysfunction. Like all diagnostic measures of health, individual measurements such as biomarkers need to be taken concurrently with other variables (e.g. body burdens, cellular integrity and growth), enabling a direct comparison to be made between the contaminant and the health of the organism, and ultimately the ecosystem. Passive biomonitoring is due to commence in the next 6 months. It should be noted that this sub-task is being performed by a University of Canberra PhD student (Anne Taylor) under the supervision of Prof. Bill Maher, and will not necessarily be completed within the time period of this grant. Active biomonitoring (caged organism transplant) experiments at contaminated field sites will be undertaken concurrently with the passive biomonitoring work.


Task 2. Review of Current Approaches for Measuring Estuarine Benthic Health This task had the following milestones:

Year 1: Commence literature review of new ecological statistical techniques. Collection of relevant assessment reports and review of data. Begin testing of new (favoured) techniques on existing data.

Year 2: Continuation of review and testing of statistical techniques for analysing data on biological communities. Apply and, where necessary, refine favoured techniques on existing data.

Year 3: New ecological statistical techniques selected and testing completed on past data.

All milestones were successfully completed, with the exception that the absence of adequate past data meant that new data needed to be acquired for testing ecological assessment approaches and this is covered in Task 3.

Outputs in this task are discussed below:

Introduction A comprehensive range of approaches for assessing estuarine benthic health were reviewed over the project’s duration.This included traditional structural approaches (e.g. abundance of taxa, metrics and indices), functional approaches (foodweb modelling, bioenergetics and ecological stoichiometry), as well as the range of statistical techniques for analysising biotic and environmental data, both individually and collectively. While several biological groups were investigated, including macrobenthic invertebrates, microphytobenthos, meiofauna fish and foodwebs, the review was weighted towards techniques which used benthic invertebrate assemblages (macrobenthos). Macrobenthos are the most frequently used group in field ecotoxicological studies as they reflect a wide range of direct and indirect responses to contaminants. Direct effects include changes in the composition and relative abundances of taxa due to differences in sensitivities among taxa and life-cycles. Indirect effects may arise from changes in fecundity, alterations to food-webs, and the cascading effects caused by the loss or increased prevalence of certain taxa. As responses vary within and among species, and are influenced by the temporal and spatial availability of the contaminant, bioassays using community end-points are designed to reflect the integrated responses of these complex interactions. Unlike lower levels of organisation, changes at a community level can be extrapolated to the ecosystem scale. It is their ecological relevance and holistic response to environmental contaminants which makes benthic communities an important and viable tool for risk assessment. The following provides a brief summary of the literature pertaining to these approaches:

Univariate attributes and metrics Univariate measurements are frequently used to summarise the structural attributes of a community, and generally include the total number of individuals; indices for richness, diversity, evenness; and the abundances of a priori selected taxa. An advantage of this approach is that the variables are amenable to several commonly used univariate statistical techniques (ANOVA, correlation and regression analyses). However, in order to provide robust, ecologically useful information, unvariate community attributes need to show a high level of response to contaminants and possess a low level of natural variability. Although changes in abundances have been observed across all broad taxonomic groups (e.g. polychaetes, decapods, bivalves, gastropods), some taxa appear to be more intolerant than others. For example, amphipods are frequently scarce or absent in contaminated environments. Conversely, some taxa appear to be relatively resilient to metals and can thrive under conditions of attenuated inter-specific competition and increased resources (e.g. food and shelter). For example, polychaetes from the Capitellidae family have regularly been observed in


high numbers and biomass in locations with enriched concentrations of metals, organics and nutrients. The simplest measurement of diversity is species (or taxa) richness. However, this measurement has three serious limitations: (i) it is improbable that all species will be counted, (ii) the boundary for the community being quantified is arbitrarily defined in space and time, and the number of individuals sampled, and (iii) the measurement negates the concept of heterogeneity, and the relative abundance of each species. Alternate measurements of diversity have been developed which individually measure or encompass the concepts of richness, sample size and evenness, producing an array of parametric and non-parametric indices (e.g. Shannon-Weiner Index, Simpson’s Index, Pielou’s Evenness). Numerous studies have shown a strong correlation between many of these variables and an increase in the concentration of contaminants, and consequently these measurements continue to form the basis of many benthic community studies. There is much criticism of univariate measurements regarding the loss of information which occurs when assemblages are reduced to a series of non-taxa specific variables. For example, two locations may have similar levels of diversity and evenness, even though the locations contain very different taxa, including a shift in the dominant taxa. There is also some concern about the suitability of applying these measurements to standard univariate statistical techniques (e.g. ANOVA), as their distributions often deviate from the assumptions of normality and the homogeneity of variances which underlie these techniques. Nevertheless, many univariate measurements are founded on frequently observed ecological observations and hypotheses, and commonly provide relevant information which may aid in discriminating between impacted and reference locations.

Table 1. General responses in univariate metrics to increased concentrations of metals and organics

Variable or Metric Response a Abundance (total) (–) Diversity (–) Evenness (–) Richness (–) Carnivores/omnivores (–) Deposit-feeders (+) Suspension-feeders (–) Amphipods (–) Biomass (–) k-selected taxa (–) r-selected taxa (+) No. of feeding guilds (–)

a (+) an increase in the metric in moderately contaminated locations; (–) a decrease in metric in contaminated locations. NB: responses to nutrients may differ from those listed.

Multivariate analysis of benthic communities In recent years, there has been a substantial increase in the use of multivariate techniques in benthic community studies. This has been primarily driven by an increase in the processing powers of personal computers, access to statistical packages founded on analysing and interpreting benthic community data (e.g. PRIMER - Plymouth Marine Laboratory), and the general acceptance of these techniques by the scientific community. In contrast to univariate approaches, multivariate techniques do not require the data to be reduced to a single variable, but rather comparisons are made between two or more sites by quantifying differences (and similarities) in their taxa and relative abundances. As a result, multivariate approaches can capture and reflect differences in whole assemblages.


The most favoured approaches are ordination techniques which are designed to reduce the data into a ‘map’ which can subsequently be visually interpreted. Ordination techniques fall into two main catagories, unconstrained (indirect gradient) analysis and constrained (direct) gradient analysis. Unconstrained techniques do not use any a priori hypotheses (e.g. treatments) and simply reduce the dimensionality of data on the basis of some general criterion. These techniques are useful for describing broad patterns and are easily interpretable; however, spatial patterns provided by these techniques can sometimes cover-up natural differences in assemblages. Constrained ordinations use a priori hypotheses as the basis for producing the ordination plot, e.g. assemblages from several treatments. These approaches relate a matrix of response variables (e.g. benthic assemblages) with a matrix of predicator variables, e.g. environmental variables or treatments that identify factors or groups (as in the case of ANOVAs). Consequently, the effect of specific environmental variables on benthic assemblages can be identified, and in some cases, partitioned out. There are numerous multivariate techniques, and those which are most relevant to benthic community studies are described below. Non-metric multidimensional scaling (nMDS) is one of the most commonly used ordination techniques. This unconstrained technique provides an output which is presented as either a two or three dimensional ordination ‘map’ in which samples which are more similar to each other are positioned closer than those which are less similar. nMDS does not test for difference between samples, but rather provides a graphical representation of the data, which can be useful in gaining an understanding of how samples relate to each other, vary over time or respond to environmental variables. In addtition, nMDS has been shown to discriminate differences among samples in cases where univariate analyses have detected no differences. The dissadavantages of this approach include the loss of information and distortion in ordination maps which occurs from reducing the data sets to 2 or 3 dimensions, the lack of significance testing, and the post hoc layering of environmental variables on the ordination maps, e.g. bubble plots. nMDS and many other ordination techniques do not quantify the variance among or within the samples as traditionally measured by ANOVAs. Consequently, additional analysis is required in order to establish if predefined groups or treatments (i.e. location, site or time) contain significantly different assemblages. Analysis of Similarities (ANOSIM) is a function associated with the PRIMER software package which employs the same design layouts commonly used in ANOVAs (one-way, two-way or nested), and uses a permutation test to compute a test static (R) and a randomisation approach to generate levels of significance. Once significant differences among treatments are detected, pair-wise analyses can be performed to identify where the differences occurred. An alternative to ANOSIM is Permutational Multivariate Analysis of Variance (PERMANOVA) (Anderson, 2001). This technique is not limited to pre-defined distance measurements as in the case of ANOSIM (e.g. Bray-Curtis distance), and can handle complex designs and their interactions. Canonical Correspondance Analysis (CCA) is a non-linear constrained technique which is founded on identifying relationships between biotic communities and their environment. This technique develops synthetic environment gradients (ordination axes) which maximise the separation among species. A primary difference between CCA and linear techniques is founded on two assumptions which frequent benthic matrices and their complementary environmental data. Firstly, most species only occur in a proportionally small number of samples, resulting in a matrix with a high incidence of zeros (absence). Secondly, relationships between benthos and environmental variables are generally non-linear, with the occurrence of species often a unimodal function of the environmental variables. CCA also has several other qualities which make it ideal for benthos studies: (i) Variables which are not of primary interest (e.g. grainsize) can be partitioned out of the analysis, clarifying the roles of the primary environmental variables (e.g. contaminants), (ii) The significance of relationships between ordination axes and benthic communities can be tested using Monte


Carlo based permutations test, and (iii) Post-hoc analysis based on multiple regression analysis (forward selection) can be used to identify the key environmental variables, and their relative strength, which significantly influence the structure of benthic communities Canonical Analysis of Principal Coordinates (CAPS) is a relatively new constrained technique which is applicable to studies which use a priori defined groups or hypothese based on one or more quantitative explanatory variables. In contrast to other procedures which are restricted to a fixed type of distance measurement, e.g. CCA preserves a chi-square distance, CAPS is more flexible enabling the user to employ any type of dissimilarity or distance measurement. This makes the technique applicable to scenarios where the predefined measurement of distance may be ecologically irrelevant. In common with other canonical techniques such CCA, CAPS has the potential to elucidate dominant multivariate patterns which are founded on hypotheses rather than exploratory principles.

Functional approaches Food web research: Food web analysis is designed to assess the relative importance of trophic relationships within a defined ecosystem. In constrast to gut content examination, stable isotope ratio analyses of carbon and nitrogen provide diet information of animals integrated over time, the period of time being dependant upon the turnover time of the tissue being sampled. This technique is founded on the relationship between stable 13C/12C and 15N/14N isotope ratios for the sampled animals and their prey. Changes in ratios occur as the lighter isotopes are preferentially lost through metabolisms, resulting in consumers having a heavier composition than that of their prey. In the case of 15N/14N , an approximate 3‰ increase occurs with each successive trophic level. In the case of carbon, isotope values determine the origin of organic matter within the system. The use of stable isotopes is preferred over the gut content method in estuarine studies as most benthic invertebrates change their food preference at some stage in their life history. Furthermore, the size of the animals can make visual analysis difficult, producing highly variable results.

In the context of our research, food web analysis provides a potential way for examining how contaminant-induced structural changes in benthic assemblages influence trophic flow, nutrient cycling and the prey choice of fish. This information can be used to clarify the relative importance of benthic communities and shift in the dominance of specific taxa, as well as assist in identifying some of the primary drivers which contribute to ecosystem health.

Ecological stoichiometry: The concept of ecological stoichiometry was briefly examined as an approach for modelling the flow of energy of metabolic activity through ecosystems. Essentially, this approach is based on the law of conservation of mass and the law of definitive proportions and can potentially be used to model changes and the transfer of molecular building blocks (e.g. nitrogen and lipids) and energetics with undisturbed and disturbed ecosystems. Although this approach is intellectually interesting and does have scientific merit, ecological stoiciometry was not viewed as an approach which would be applicable to environmental managers as it required extensive modelling, field validation, and considerable research to be viable within NSW estuaries.


Task 3. Assessment of Benthic Community Health at Key NSW Sites This task had the following milestones:

Year 1: Collection of relevant assessment reports and review of data. Selection of study sites for testing new (favoured) methodologies. Commence sampling of new data.

Year 2: Application of favoured techniques (Task 2) to existing data on selected contaminated sites. Testing of techniques on newly collected ecological data and other data (e.g. chemical, bioaccumulation, biomarker response).

Year 3: Application of new methods to past and newly collected data (including chemical bioavailability and biomonitoring data) completed. Assessment of best approaches completed. All milestones were successfully completed, and research is continuing on the evaluation of food chain responses. Outputs in this task are discussed below:

Introduction Using the information obtained from the literature review (Task 2), a variety of techniques and approaches were applied to data collected from 15 locations (60 sites) across the Parramatta, Lane Cove, Hawkesbury and Georges Rivers. Initially, the statistical procedure were to applied to historic data, however, a paucity of concurrently collected biological and chemical data made this component defunct.

Contaminant gradient within western Sydney Harbour Principal component analysis (PCA) of the environmental data collected from the Parramatta, Lane Cove, Georges and Hawkesbury Rrivers found the locations within the Parramatta River to be quite distinct from both the external (Georges and Hawkesbury Rivers) and internal (Lane Cover River) reference locations (Figure 1). Elevated concentrations of metals and metalloids (Cd, Cu, Pb, Zn, As, Cr, Ni and Ag) were the predominant variables which discriminated the Parramatta River locations from the reference locations. Trends in organic contaminants were less pronounced than metals. A notable exception was Iron Cove (IC), which was clearly separated by the ordination due to its relatively high concentrations of organics (chlordane, heptachlor, dieldrin, PAH, TPH and PCBs). This location also had higher concentrations of organic carbon (OC) than the other sampled locations. An additional feature of note was that acid volatile sulfide concentrations increased with metal concentrations. This suggests that the AVS-SEM model may be unsuitable for predicting toxicity within Sydney’s estuarine environments. PCA was also used to identify the influence of various metal extraction techniques (total particulate metals (TPM) by 2:1 hot conc HCl:HNO3 digestion, dilute acid-soluble metals (ASM) by cold 1-M HCl digestion of total sediment or the fine sediment fraction (<63 µm)), as well as metals in pore waters, AVS, depth-stratified ASM samples, and various other environmental variables on ordination patterns. The ‘broken-stick model’ was used as a stopping mechanism to indicate which components contained interpretable information and to compensate for changes in variance which inherently occur from altering the number of variables used in each analysis. Briefly, for each component, if the observed eigenvalue exceeds the calculated broken-stick value then information is considered meaningful, and is retained. Conversely, if the eigenvalue is less than the calculated broken-stick value, then information from subsequent components are not interpreted in the analysis. It is emphasised that this analysis is purely based on describing the gradient, i.e. a summary of the spatial patterns of the environmental variables, and does not reflect the biological availability or toxicity of any contaminants.


-1.0 1.0

-1.0

1.0

Cr

Ni Cu

Zn

As

Ag

CdPb

NH3

fines

OC

AVS

TPHPCBs

Chlordane

DieldrinHeptochlor

PAH

DDT

BP BP

BP

BP

BBBB BB

BB

COB COBCOB

COB

DC

DC

DC

DC

FD

FD

FD

FD

FOM

FOM

FOM

FOM

HC

HC

HC

HC

CBCB

CB

CB

IC

IC

ICIC

KB

KB

KBKB

MB

MB

MB

MB

MJB

MJB

MJB

MJB

PBPB PB

PB

TB

TB

TB

TB

WB

WB

WBWB

Figure 1. PCA of the 60 sites (15 locations) sampled

The first two components account for 54 % of the variance. Vectors are only shown for the strongest environmental variables.

The results from these analyses suggest that all types of metal analysis explained similar percentages of variance (63-65%) over the first 4 principal components (PCs). However, only the analysis for the TPM of the full sediment and the ASM of the <63 µm sediments had the interpretable information restricted to the first 4 PCs. For the first two PCs, ASM extractions using the fine-fractioned sediments explained marginally more variance than the nitric acid digested sediments. The analyses also found that there were no benefits in the addition of stratified metal measurements, indicating that future sampling programs were able to focus on the surficial layers. Additional analyses were performed to identify if the gradient was best described using particulate metals (<63 µm ASM), pore water metals or combination of the two. The results clearly showed that <63 µm ASM alone explained a greater amount of variance than either the pore waters or combination of particulate and porewater metals.


Non-metric dimensional scaling The ordination map of the benthic data (only shown for 2005) illustrates an underpinning trend where assemblages collected from locations within the Parramatta River (blue) are separated from the external reference locations (green), and to a lesser degree, the internal reference locations (red) (Figure 2). Three sites from Duck Creek (DC2-4) and one from Field of Mars (FOM1) have assemblages which are markedly different from all other assemblages, with the latter having several impoverished communities even though there was no indication of extensive contamination.

Figure 2. nMDS plot for the 2005 benthic community data

PERMANOVA PERMANOVA detected a significant difference in benthic communities among locations. Post-hoc analysis found that most locations had dissimilar assemblages, with the exceptions listed below in Table 2. One observation that appears not to fit the hypothesised trend is the similarity between Field of Mars (an internal reference location) and Hen and Chicken Bay, a highly contaminated location.

Table 2. Locations with similar (no significantly different) assemblages as detected using PERMANOV pair-wise comparions.

Location Locations with Similar Assemblages

Brays Bay Hen and Chicken Bay, Iron Cove

Hen and Chicken Bay Iron Cove, Five Dock, Field of Mars

Iron Cove Morrison

Kyle Bay Woodforde


SIMPER analysis Table 3 shows the taxa which contributed to 90% of the total abundance of each location. Spoindae were the most abundant taxa in all but two locations, Duck Creek and Cogra Bay. In both Brays Bay and Hen and Chicken Bay spionids made up over 90% of the total abundance. Other notable trends include a greater presence of bivalves from Hawkesbury River locations, and the relatively high abundance of melitid amphipods in two contaminanted locations, Iron Cove and Morrisons Bay.

Table 3. SIMPER analysis indicating the taxa which contributed to 90 % of the total abundance for each location

Canonical Correspondance Analysis (CCA) CCA analysis provided more defined separations in benthic assemblages than the nMDS. For example, sites from the two Hawkesbury locations (CB and PB) were clearly separated from the Georges River reference locations, a pattern which was not identified in the nMDS.

Overall the CCA demonstrated that elevated concentrations of particulate trace metals were influencing benthic community structure within the harbour, however, marked differences were only occurring at locations with significantly elevated concentrations, e.g. Iron Cove. The analysis also found no strong relationships between grain size, organic carbon, salinity and benthic community structure, although, the sampling strategy was stratified to minimise the variability in these variables. CCA was also able to identify a contaminant and phase specific reponse in Duck Creek, where elevated concentrations of pore water chromium appear to be significantly modifying the benthos.

River Location Taxa Contributing to 90% of the Total Abundance

Parramatta Brays Bay Spionidae

Duck Creek Sabellidae, Galeommatidae

Five Dock Spionidae, Capitellidae, Sabellidae, Oligochaeta

Hen and Chicken Bay

Spionidae

Iron Cove Spionidae, Opheliidae, Melitidae

Morrisons Bay Spionidae, Opheliidae, Melitidae, Photidae

Majors Bay Spionidae, Opheliidae, Capitellidae, Sabellidae

Lane Cove Woodforde Bay Spionidae, Capitellidae, Sabellidae, Cirratulidae, Lumbrineridae

Tambourine Bay Spionidae, Opheliidae, Capitellidae

Boronia Park Spionidae, Capitellidae, Pilargidae, Opheliidae

Field of Mars Spionidae, Sabellidae, Nereididae, Capitellidae

Georges Coronation Bay Spionidae, Capitellidae, Bivalve(unid), Magelonidae, Lumbrineridae

Kyle Bay Spionidae, Nephtyidae, Sabellidae, Capitellidae

Hawkesbury Porto Bay Spionidae, Trichobranchidae, Lumbrineridae, Capitellidae Psammobiidae

Cogra Bay Diplodontidae, Spionidae, Magelonidae, Capitellidae, Psammobiidae, Cirratulidae, Lumbrineridae


-0.6 1.0

-0.6

1.0

Cr

Ni

Cu

As

CdPb

PW-CrPW-Ni

PW-Cu

PW-NH3

AVS

BPBP

BP BB

BBBB

BB

COB

COB

COB

COB

DC

DC

DC

DC

FD

FD

FDFDFOM

FOM FOM

FOM

HCHC

HCHC

CBCB CBCB

IC IC

IC

IC

KB

KBKB

KB

MB

MB

MB

MB

MJBMJB

PB

PB

PBPB

TB TB

TB

TB

WBWB

WBWB

Figure 2. CCA from the data collected in 2005 from Sydney and adjacent estuaries

Fine-fractioned 1M HCl digested sediments were used for the metal analysis. Variables measured from pore waters contain the prefix PW.

Lessons, future opportunities of benthic assemblage research The key finding from this research which focussed on assessing the health of estuarine sites near Sydney, NSW, using benthic assemblages were:

• Complementary benthic and environmental data sets should always be obtained to maximise the choice of statistical techniques (both constrained and unconstrained) and to identify major variables which may be contributing to changes in benthic assemblages.

• No single statistical approach is sufficient for assessing benthic assemblages.

• In those locations studied, trace metals (both in the sediment and pore water) are the major group of contaminants influencing benthic community structure.

• The findings from the research indicate that there is either a relatively high threshold response in benthic communities to contaminants, or a lack of sensitivity in the applied techniques for determining changes in moderatetly contaminated environments.

• Reference locations from the Georges River were very similar to internal reference sites such as Tambourine Bay and Woodford Bay, enabling spatially distinct locations to be incorporated into future studies. Field of Mars (Lane Cove River) appears to be unsuitable as a reference location even though concentrations of measured contaminants were low. Benthic assemblages sampled from the Hawkesbury River reference locations were significanty different to the Georges River locations and internal references, indicating that these locations may not be suitable for comparions within the Harbour.

• The collection of benthic data is timely, costly and requires expert taxonomic skills, the latter being limited within the region. As a result, more rapid and cost-effective techniques need to be developed. A Trust project on the use of genomics for this purpose is about to commence.

• Efforts in this study to reduce the number of replicates through composite sampling were unsuccessful.


• Techniques for comparing the environmental condition (not biological) of a location based on its contaminant loadings and physical condition would permit greater application of techniques founded on a priori groupings (e.g. CAPS), and provide a template for the potential development of an Index of Biotic Integrity.

• The amount of data required to develop an Index of Biotic Integrity is currently unavailable, with such an approach requiring a significant amount a temporal and spatial sampling.

• The analytical detection limits of organics are often greater than their ISQG-low value. Consquently protocols for classifying a site as a ‘reference site’ are difficult.

• Environmental assessment needs to incorporate a greater diversity of taxa than the current reliance of benthos, as this process is not a panacea, but rather a line of evidence.

Food web analysis During 2006, materials were obtained to study the food webs of four locations, two reference sites (Boronia Park and Tambourine Bay) and two contaminated sites (Hen and Chicken Bay and Five Dock Bay). The primary aims and objectives of this sub-task were:

(i) To identify whether fish in contaminated locations alter their diet to

accommodate changes in benthic communities, or consume their same diet and therefore alter the amount of energy required to consume their prey of choice.

(ii) To examine differences in the trophic organisation and food sources between

contaminated and reference location food webs.

Replicate samples (5) from each location were collected for seven species of fish, benthos (5 species), meiofauna, microphytobenthos, seston, mangrove leaves, waters, macroalgae and sediment. Extensive delays in the analysis of these materials have occurred through equipment failure and the loss of operating staff.

Figure 3 illustrates the mean δ13C and δ15N from fish collected from Boronia Park and Hen and Chicken Bay. Carbon signatures (δ13C) for all seven fish species significantly differed between Boronia Park and Hen and Chicken Bay. With the exception of tailor, fish from Hen and Chicken Bay were more enriched in carbon isotopes. The fish sampled from the two locations encompassed two trophic levels (δ15N > 3‰ per trophic level), with tailor having being the most enriched species and mullet the least. No significant difference in δ15N occurred in toadies, silver biddy, gobies and tailor sampled from the two locations. δ15N was more enriched in mullet and whiting collected from Boronia Park, conversely, enrichment was greater in bream sampled from Hen and Chicken Bay. The significance of these findings, as well as those from Tambourine Bay and Five Dock Bay, is currently being assessed using the data collected from a suite of potential carbon sources (waters, sediments, primary producers and consumers) and gut-content analyses.


Table 4. Mean δ15N of fish sampled from Boronia Park and Hen and Chicken Bay (n=5)

δ15N Location Species mean 1 SE

Boronia Park Mullet 10.74 0.27 Boronia Park Goby 11.90 0.18 Boronia Park Bream 12.36 0.20 Boronia Park Sand Whiting 13.28 0.41 Boronia Park Toady 13.64 0.12 Boronia Park Silver Biddy 14.64 1.83 Boronia Park Tailor 15.20 0.20 Hen and Chicken Mullet 9.42 0.34 Hen and Chicken Sand Whiting 11.58 0.29 Hen and Chicken Goby 12.04 0.25 Hen and Chicken Silver Biddy 13.24 0.43 Hen and Chicken Toady 13.58 0.62 Hen and Chicken Bream 13.80 0.08 Hen and Chicken Tailor 15.95 0.51


0

2

4

6

8

10

12

14

16

18

-24 -22 -20 -18 -16 -14 -12

Delta13 C

Del

ta 1

5 N

BP Silver BiddyHC Silver BiddyBP BreamHC BreamBP GobyHC GobyBP MulletHC MulletBP Tailor HC Tailor BP ToadyHC ToadyBP Sand WhitingHC Sand Whiting

Figure 3. Dual isotope plot to indicate difference in organic carbon sources and trophic positions of seven species of fish collected from Boronia Park and Hen and Chicken Bay (Error bars represent one standard error)


Task 4. Field Testing Improvements to the ANZECC/ARMCANZ Sediment Quality Guidelines Framework This task had the following milestones:

Year 2: Dicussion of new framework advanced

Year 3: Revised/extended ANZECC/ARMCANZ sediment quality guideline framework prepared and tested at key field sites. All milestones were completed. The application of the WOE framework including the new lines of evidence was undertaken, however use of these data in the WOE framework currently being developed for AFFA has yet to be undertaken.

New WOE framework The new WOE framework involves scoring on a 1 to 3 basis for each of the lines of evidence, with a final assessment as shown in Table 5. This framework is described in more detail elsewhere (Simpson and Batley, 2007).

Application of WOE framework to key sites The ecological data that were obtained for Sydney Harbour sites (Iron Cove, Five Dock Bay, Hen and Chicken Bay and Boronia Park) as part of Task 3, were examined as a line of evidence using the proposed weight-of-evidence framework. The other lines of evidence included TPM, ASM, AVS-SEM, organic contaminants (normalised to 1% OC), and acute amphipod toxicity testing. No bioaccumulation data were obtained. The test species in the amphipod test were 7-14-day old Melita plumulosa. Sediments were considered toxic in cases where survival was less than 80 %. Benthic communities from the impacted locations were compared to the internal reference locations (Boronia Park and Tambourine), and PERMANOVA was used to identify differences among benthic assemblages. Table 6 shows the data obtained. At Iron Cove, Hen and Chicken Bay and Five Dock Bay, TPM concentrations of copper or zinc exceeded the SQG trigger values (TVs) by greater than five times. ASM concentrations did not exceed the TVs indicating that the metals were mineralised.There was an excess of AVS over SEM in three of the sites, however very high AVS concentrations often indicate poor benthic community health due to few organisms bioturbating the sediments. The ISQG-low values for organics were exceeded by detection limits, so results were inconclusive. Results were analysed on the assumption that the exceedences were significant. Only the Hen and Chicken Bay sediments were toxic to the juvenile amphipod, and dissolved copper was measured in toxicity tests indicating oxidation of AVS in the surface sediments. PERMANOVA seemed to find everything significant, except locations which are have very low diversity and variability (i.e. significantly impacted). In this example Boronia Park was seen as different to the other local reference, Tambourine Bay. Using the WOE framework, one site was ranked as showing significant adverse effects, two with possible adverse effects and one with no adverse effects (Table 7). Incomplete though this particular set was, some confidence can be had that the approach can lead to reasonable assessment conclusions.


Table 5. Weight-of-Evidence Decision Matrix (not all possible LOE or cases included)

Line of Evidence a

Case Chemistry (metals-

organics)

Toxicity Bioaccumulation / Biomagnification

Ecology Weight-of-evidence (WOE) Score Overall Assessment

W1 3 3 2 or 3 3 3 Significant adverse effects from sediment contamination

W2 3 3 2 or 3 2 3 Significant adverse effects from sediment contamination

W3 2 or 3 3 2 2 3 Significant adverse effects from sediment contamination

W4 2 or 3 2 1 or 2 2 2 Possible adverse effects from sediment contamination

W5 2 2 or 3 1or 2 2 2 Possible adverse effects from sediment contamination

W6 2 2 1 or 2 2 or 3 2 Possible adverse effects from sediment contamination

W7 2 or 3 2 or 3 2 or 3 1 2 Toxic chemical stressing system but resistance has developed at community level

W8 1 2 or 3 1 2 or 3 2 Unmeasured toxic chemicals causing effects on communities

W9 1 2 or 3 1 1 2 Unmeasured physical or chemical causes of toxicity

W10 2 or 3 1 1 2 or 3 2 Chemicals are not bioavailable or community change not due to chemicals

W11 1 1 1 2 or 3 1 Changes not due to measured contaminants

W12 1 or 2 1 1 or 2 1 1 No adverse effects

W13 1 1 1 1 1 No adverse effects

W14 2 or 3 1 1 1 1 Contaminants unavailable a Values listed in each line of evidence category are the highest scoring assessment in that category, e.g. under chemistry, metals may score 2, organics 3, so 3 is recorded. The greater the number of 3s recorded in a category, the greater is the weight that line of evidence category assumes. The same applies then to the overall assessment which combines all LOEs.


Table 6. Scoring for Lines of evidence measured for selected Sydney Harbour sediments Location Total

metals ASM AVS-SEM Organics Amphipod

toxicity Benthic ecology Iron Cove

3 1 2 2 1 Sig. different to reference

Hen and Chicken

3 1 2 2 3 Sig. different to reference

Five Dock

3 1

1 2 1 Sig. different to reference

Boronia Park

1 1 1 1 1 Sig. different to Tambourine Bay

Table 7. WOE scoring for Sydney Harbour sediments Chemistry Ecotoxicity Bioaccumulation Ecology Overall

WOE Score Iron Cove 2 1 - 3 2 Hen and Chicken

3 3 - 3 3

Five Dock 1 1 - 3 2 Boronia Park

1 1 - 2 1


Issues, Changes and Opportunities

Lessons from the program A major impediment to the planned exercise was the difficulty in obtaining historical benthic data for Sydney Harbour sites. It was our belief that considerable data already existed from earlier studies by Sydney Water, the Ecology Lab, Sydney University and other agencies. In fact, the number of available datasets was extremely limited, and in many cases the datasest were incomplete. Some results had been collected by overseas consultants and not retained in Australia and not able to be tracked down through these consultants, while in other cases there was a reluctance to make the data available. These difficulties meant that a new and complete dataset needed to be obtained as part of this study. This added considerably to the workload and delayed the project. The advantage, however, was that the datasets were complete and allowed a thorough scientific assessment which would not have been possible using the historical data. In hindsight, we were probably too optimistic in our expectation that there was a lot of information on Sydney Harbour available, and that we would have no difficulty in obtaining this. The value of grey literature that is incomplete or has not been adequately peer reviewed is therefore questionable. Other issues Largely as a consequence of the above, the project took longer than anticipated. This meant that CSIRO (and to a lesser extent University of Canberra and DEC) needed to provide additional financial contributions for analyses of collected samples and in kind support to the project. The PhD project, being undertaken at University of Canberra, that provided input into Task 1, has a further year to run. This means that some of the data that were being provided on bioaccumulation were not available at the time of writing this report. They will nevertheless be incorporated in later input to assessment protocol revision. The revised format required for the Trust Final Report does not allow for extensive scientific reporting. Greater details of literature reviews, experimental methodology and discussions of findings will lbe the subject of publications currently in preparation. These will appropriately acknowledge the Trust.

Acknowledgements A number of people made very significant contributions to this project. Dr Dianne Jolley (University of Wollongong) co-supervisor of the honours student Ms Vicky Burston). Kim Chau and Dr Gary Low of the Analytical Division of NSW DEC assisted with the analyses of hydrocarbons in sediment extracts and biological tissues (for Vicky Burston). Sheryl Tang, Richard Gardiener, Steve Jacobs, Max Carpenter, Chris Rush and Joanne Ling who assisted in biological survey work and biota identification. David Spadaro and Ian Hamilton undertook most of the physico-chemical characterisation, chemical analyses, toxicity tests and assistance with biological survey work.



References Simpson, S.L., Batley, G.E., Chariton, A.A., Stauber, J.L., King, C.K., Chapman, J.C., Hyne, R.V.,

Gale, S.A., Roach, A.C., and Maher, W.A. (2005). Handbook for Sediment Quality Assessment. CSIRO, Bangor NSW, 117 pp.

Simpson, S.L.and Batley, G.E. (2007). Revision of the ANZECC/ARMCANZ sediment quality

guidelines. CSIRO Land and Water Report, in press.


Appendix 1. Publications











Papers in Preparation

1. Chariton, A.A., Roach, A.C., Simpson, S.L and Batley, G.E. The influence of metal

extraction techniques and porewaters constituents on interpretation of principal components.

2. Chariton, A.A., Roach, A.C., Simpson, S.L and Batley, G.E. A comparison of

statistical techniques for detecting the ecological effects of contaminated estuarine sediments.

3. Chariton, A.A. Are the diets of estuarine fish altered by contaminated induced

changes in benthic communities?

4. Chariton, A.A. Contaminant induced changes in the carbon and nitrogen signatures from estuarine food webs.


Appendix 2. Conference Presentations

Environmental Trust / CSIRO Workshop on Sediment Quality Assessment, Australian Museum, Sydney, June 2, 2005

Frameworks for sediment quality assessment Graeme Batley

Weight of evidence assessments Stuart Simpson

Sediment sampling Stuart Simpson

Sediment chemistry and bioavailability Stuart Simpson

Sediment TIEs Stuart Simpson

In situ assessments Anthony Chariton

Australian Water Association Ecotoxicology Workshop, Canberra, June 24, 2005

Sediment chemistry and bioavailability Stuart Simpson

Sediment TIEs Stuart Simpson

INTERACT Conference, Perth, September 25, 2006

Measuring relationships between contamination and benthic community structure: assessment of approaches

Tony Roach

SETAC North America Annual Meeting, Montreal, Canada, November 7, 2006

Relationships between sediment contamination and benthic community structure: an assessment of approaches

Anthony Chariton

Documents

Improved Approaches to Sediment Quality Assessment · The project was designed to meet the needs of regulatory agencies for an improved sediment quality assessment framework to unambiguously