Environmental Pollution 112 (2001) 131±143
www.elsevier.com/locate/envpol
Using benthic recruitment to assess the signi®cance of contaminated sediments: the in¯uence of taxonomic resolution A.C. Roach a,*, A.R. Jones b, A. Murray b a
Water Studies Section, Environment Protection Authority, New South Wales, PO Box A290, Sydney South, NSW 2000, Australia b Division of Earth and Environmental Sciences, Australian Museum. 6 College Street, Sydney, NSW 2000, Australia Received 25 July 1999; accepted 18 March 2000
``Capsule'': Higher levels of taxonomic resolution can be used to describe variations in the structure of benthic communities in small-scale experiments. Abstract The use of small-scale experimental units as a means of evaluating the ecological eects of contaminated sediments was examined at the species, family, mixed and phylum levels of taxonomic resolution. Sediments were taken from various locations representing a range of contaminant loads. Containers with these sediments were placed in situ at a relatively uncontaminated location for 90 days. The containers were retrieved and the abundance of the macrofauna which recruited to the containers was estimated. The results showed that the composition of the benthic communities in the more highly contaminated sediments diered signi®cantly from those in less contaminated sediments. Analyses at the dierent taxonomic levels showed that all but the phylum level data showed some dierences in community structure among sediment types. The study showed that small-scale experiments are useful for examining the eects of contaminants and that higher levels of taxonomic resolution can be used to describe variations in the structure of benthic communities at this spatial scale. # 2001 Elsevier Science Ltd. All rights reserved. Keywords: Sediment quality; Taxonomic resolution; Recruitment; Benthos
1. Introduction Sediments are critically important as estuarine and marine habitats because of their large extent, the fact that they often support a diverse and dense assemblage of invertebrates (Gray, 1981; Wol, 1983; Jones et al., 1986) and because their productivity is important for many ®sheries (Mann, 1976; Graf, 1992). Unfortunately, they can concentrate contaminants such that there is a signi®cant impact on the health of aquatic ecosystems (Reynoldson, 1987; Chapman, 1989; Scott, 1989; Lamberson et al., 1992). Such sedimentary attributes have led to the widespread recommendation and use of benthic biota for pollution assessment and monitoring work (Bilyard, 1987; Coull and Chandler, 1992). Consequently, resource managers often need to consider sediments and their biota when undertaking environmental * Corresponding author. Tel.: +61-2-9795-5412; fax: +61-2-97955462. E-mail address:
[email protected] (A.C. Roach).
impact assessments, setting environmental quality objectives, monitoring the success of remediation, and assessing the risk of various activities to ecosystem and human health. Therefore, it is important to have techniques which measure the ecological eects of contaminated sediments. Methods for assessing the toxicity of sediments are numerous, but commonly include laboratory-based single species bioassays and correlative ®eld studies linking biological communities and the concentration of contaminants (Coull and Chandler, 1992). Some of these studies have shown signi®cant correlations between increasing mortality and reductions in the abundance of biota in the ®eld (e.g. Swartz et al., 1994; McGee et al., 1999). Despite this, laboratory bioassays have been widely criticised as being too simple or unrealistic in some circumstances since natural processes which in¯uence contaminant behaviour are often not taken into account (Burton, 1995; Campbell and Tessier, 1996). Further, the test organisms may not be representative of all the species found in the system of interest, though
0269-7491/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved. PII: S0269-7491(00)00124-X
132
A.C. Roach et al. / Environmental Pollution 112 (2001) 131±143
multispecies testing protocols go some way to addressing this problem. Another area of concern is that some of the endpoints used (e.g. growth and reproduction) are dicult to relate directly to eects on structure and function at a community level (Underwood and Peterson, 1988; Burton, 1995; Parkhurst, 1995; Clements, 1997; Power and McCarty, 1997) and therefore predictions about the ecological signi®cance (e.g. slight change in community structure and productivity or greatly reduced diversity and biomass) of a particular contaminant are dicult to make based on this information alone. Even observational studies on in situ biological communities have their drawbacks particularly when making inferences about the signi®cance of individual contaminants. Methodologies such as the apparent eects threshold (AET; AET Malek, 1992) and the sediment quality triad (SQT; Chapman, 1992) are based on correlative relationships between biological eects, community level changes and bioassays, and the concentration of contaminants. Such relationships can, however, be dicult to interpret because often there are many co-occurring variables, both contaminants and natural, making it impossible to distinguish the key variable(s) responsible for particular eects (Adams et al., 1992). Pre-existing spatial eects can also confound the interpretation of dierences along pollution gradients or between impact and control sites (Green et al., 1993). Data derived from these approaches are used to calculate sediment quality guidelines (SQGs) values (Long et al., 1995; MacDonald et al., 1996) though it is acknowledged that mixtures may have cumulative eects and that data based on these types of observations may make the SQG for an individual contaminant more protective than those derived approaches where only one contaminant was present (Long et al., 1995; Swartz and Di Toro, 1997). Alternative approaches which re¯ect community-level eects but remove confounding eects are needed to test the realism of existing sediment quality guidelines. Manipulative ®eld experiments oer an alternative approach which may ameliorate some of these concerns. In freshwater systems such approaches are widespread and generally considered reliable and sensitive indicators of environmental impacts (Costello and Thrush, 1991). In estuarine/marine systems, experimental approaches have included exposing pre-existing communities to a pollutant (Morrisey et al., 1996) or placing defaunated sediments which have varying pollutant levels into an open laboratory system (Tagatz et al., 1979) or the ®eld (Berge, 1990; Watzin et al., 1994; Watzin and Roscigno, 1997; Olsgard, 1999) and measuring the resultant variations in the communities which recruit to these sediments. Costello and Thrush (1991) reviewed these approaches for fresh and marine waters and found that whilst manipulative approaches showed
promise, extensive evaluation of their usefulness in marine systems was needed. Despite this there are still relatively few examples in the literature. A major problem associated with studying benthic communities is that sorting and identi®cation to species level is labour intensive and therefore expensive. Further, appropriate taxonomists may not be available. The use of higher taxonomic levels, has in some instances, been found to be an appropriate way to reduce these costs (i.e. person hours; Warwick, 1993) but its appropriateness for small-scale manipulative experiments is untested. This paper describes a study which examined the usefulness of small scale recruitment experiments as a tool for assessing the eects of contaminants on benthic communities and the importance of taxonomic resolution in this assessment. Speci®cally, it tested: (1) if there was a relationship between the level of sediment contamination and the patterns of recruitment of benthic communities measured using multivariate analyses and simple metrics; and (2) determine at which levels of taxonomic resolution dierences in patterns of recruitment were detected. 2. Methods 2.1. Field and laboratory Sediments with dierent concentrations of contaminants were taken from three locations at each of 10 sites (Fig. 1). These sites represented a contamination gradient from relatively unpolluted to severely polluted (Coade and Teutsch, 2000). At each location 10 l of sediment from the lower intertidal zone were placed in a pre-washed (i.e. distilled water, laboratory-grade detergent, distilled water then hexane) stainless steel bucket, taken to the laboratory and homogenised with a stainless steel stirrer. From each bucket approximately 780 ml of sediment was placed into ®ve PVC pots which were then frozen at ÿ18 C for up to 7 days to kill fauna in the sediments. Subsamples of sediment were also taken from each bucket for the analysis of grain size (50 ml), total organic carbon (TOC; 50 ml) and the concentrations of metals (50 ml) and a range of organic contaminants (100 ml) including oganochlorine pesticides (OCs), polychlorinated biphenyls (PCBs), polyaromatic hydrocarbons (PAHs), total petroleum hydrocarbons (TPHs). The pots were then placed at approximately 1.5 m below MLW at random within a 70 m2 area in Weeney Bay (Fig 1B; WB), an area with relatively low concentrations of contaminants in the sediments (Coade and Teutsch, 2000). The pots were recovered after 90 days and placed in a solution of 5% formalin in seawater. The sediment from each pot was sieved through a 500 mm mesh screen and the benthic fauna retained on
A.C. Roach et al. / Environmental Pollution 112 (2001) 131±143
133
Fig. 1. Map showing the locations of the estuaries and sites within each estuary from which the sediment samples were taken. Maps A±C show sampling sites as: AC, Alexandra Canal; CR, Cooks River; DC, Duck Creek; LB, Long Bay; LK, Lime Kiln Bay; MC, Mill Creek; PC, Prospect Creek; PH, Port Hacking; SP, Salt Pan Creek; WB, Weeney Bay.
the screen identi®ed. The polychaetes, crustaceans (except copepods) and molluscs were identi®ed to species level and the remaining organisms to the lowest practicable level. Metals were digested using nitric and hydrochloric acids according to the USEPA method 200.2 (Keith, 1992). This extraction will not recover the most strongly bound metals and may underestimate concentrations relative to guideline values (Long et al., 1995) particularly where the sediments are relatively uncontaminated (pers. obs). Following extraction, the concentrations of all metals except mercury were determined using inductively coupled plasma mass spectrometry using the USEPA method 6010 (Keith, 1992). Mercury was determined by cold vapour atomic absorption spectrometry using USEPA method 7471 (Keith, 1992). OCs, PCBs and PAHs were extracted by homogenisation with acetone/acetonitrile (1:1) from a mixture of approximately 5 g of sample and 10 g of anhydrous sodium sulphate. The extract was exchanged into hexane and cleaned according to the procedures outlined in the ¯orisil clean-up section, 983.21E of AOAC (1990). It was then analysed in accordance with USEPA method 8080 (Keith, 1992), except that capillary gas chromato-
graphy (GC) using electron capture detection was used. Two dierent polarity GC columns were used to identify the compounds, with gas chromatography±mass spectrometry (GC±MS) used to con®rm the compounds detected. Total organic carbon was analysed using an infrared combustion procedure (method 6B3 in Rayment and Higginson, 1992). The proportion of silt+ clay (=mud, <63 mm), sand (>63 mm, <2mm) and gravel (>2 mm) was determined by wet sieving. Estimates of the rates of recovery for metals were assessed using the standard reference material (SRM) BCSS-1 (Loring and Rantala, 1992). For all organic contaminants analyses the relative percent dierence for high and low concentrations and for surrogate spike recoveries for the high concentration were estimated. For metals the quality control results were in agreement with the certi®ed materials and for organics the results fell within an acceptable range of 15% (Tetra-Tech, 1986). 2.2. Data analysis Prior to analysis, data for the physico-chemical variables were transformed (log10+1) and examined for
134
A.C. Roach et al. / Environmental Pollution 112 (2001) 131±143
collinearity using the procedure of Clark and Ainsworth (1993). The number of variables was reduced where signi®cant collinearity occurred, de®ned as when r>0.90. Because the concentration of many metals is often related to the percentage of ®ne grained particles in the sample (Loring and Rantala, 1992), and the TOC aects the bioavailability of some organic contaminants (Zarba, 1992), the concentrations of metals were normalised with respect to the mean proportion of mud (particles <63 mm) found across all locations and the concentrations of organic contaminants were normalised with respect to the 1.2% TOC (Long et al., 1995). Locations with similar sediment quality were then grouped using medoid cluster analysis (Hintze, 1995). Principal components analysis was used to verify the results of the cluster analysis and to create composite variables (principal components) summarising the sediment quality data. A toxicity quotient based on the eects range median sediment quality guideline values (ERMQ; Long and Wilson, 1997) was also calculated for each site. Correlation was then used to test if there was a relationship between increasing levels of contamination represented by the principle components and ERMQ, and ecological characteristics such as community structure (i.e. centroids of each site on nMDS axis), number of species and number of individuals. For each group of sites formed by the cluster analysis the normalised sediment quality data were also compared with eects-based guideline values for individual contaminants (Long et al., 1995) and ERMQ guideline values (Long and MacDonald, 1998) to indicate whether the observed concentrations were likely to cause a biological impact. Although the metal values in the guidelines are based on bulk concentrations, the normalised concentrations were used here to maintain consistency throughout the data presentation and because the normalisation represents a concentration for sediment with a typical proportion of mud, making the value derived realistic in terms of these guidelines. Analysis of variance comparing variables among the contaminant groups was rejected as a means of testing if there were dierences in ecological characteristics among groups because the analysis would have been either unbalanced or required considerable data reduction to achieve a balanced design. Dierences in the structure of the benthic communities recruiting to the various sediment types were analysed using a multivariate approach. The raw data were transformed to their double square-root and ordinated using non-metric multidimensional scaling (nMDS) in the software package PRIMER (Clarke and Warwick, 1994). Dierences among and within sediment groups were tested for using a two factor nested analysis of similarities (ANOSIM; Factor 1=group, and Factor 2=location which was nested in group). The taxa important in discriminating among sediment
groups were identi®ed using the Kruskal±Wallis Hc statistic (KW-H) (Belbin, 1995). This was done in place of the similarity percentage (SIMPER) analysis (Clarke and Warwick, 1994) because the size (i.e. number of samples) of the data matrix was too large for the SIMPER program. Analyses were done at various levels of taxonomic resolution: at species, family, phylum, and to a level referred to as mixed. The mixed level represents a level of taxonomic grouping which can usually be done with minimal taxonomic training (i.e. polychaetes identi®ed to family level, crustaceans to subclass or order, molluscs to class, other taxa to class or phylum). The nMDS, ANOSIM and taxa breakdowns (KW-H) were repeated for each level. 3. Results 3.1. Identi®cation of contamination gradient Analyses of the sediment quality data showed that numerous variables were strongly inter-correlated. The data set was reduced to nine variables to minimise this collinearity. The nine variables were As, Cr, Cu (surrogate variable for Cd, r=0.95; Hg, r=0.91; Pb, r=0.92; Zn, r=0.98), Ag, chlordane, dieldrin (surrogate variable for Hexachlorobenzene [HCB], r=0.90), PCBs (surrogate variable for DDTs, r=0.91), PAHs and TPHs. The cluster analysis of these data showed that a ®ve group solution (hereafter referred to as Groups A±E) provided a good ®t and had few locations arbitrarily allocated to a group (Table 1). A plot of the ®rst two principal components showed that the allocation of most sites was reasonable (Fig. 2), although there was some variation within groups. The ®rst two axes explained 75% of the variance (Table 2). Comparison of the plot of the ®rst two components with the factor loadings and the mean contaminant concentrations for each of the ®ve cluster groups (Table 3), showed that locations in Group A had relatively higher concentrations of most contaminants except PCBs. Comparison with eects-based sediment quality guidelines (Long et al., 1995) showed that the mean normalised concentrations of lead, zinc, chlordane and dieldrin were all Table 1 Results from the medoid clustering analysis of the sediment quality data Cluster
Sites allocated to cluster
A B C D E
AC1, AC2, AC3, CR1, CR3, DC2 CR2, DC1, DC3, LB1, LB2, LB3, SP1 SP3, PC1, PC2, PC3 LK1, LK2, LK3, MC1, MC2, MC3, PH2, SP2 WB1, WB2, WB3, PH1, PH3
A.C. Roach et al. / Environmental Pollution 112 (2001) 131±143
Fig. 2. Plot of the ordination scores from the principal components. A±E refers to the groups formed in the cluster analysis.
signi®cantly greater than the eects range median (ERM) and a further eight contaminants exceeded their eects range low concentrations (ERL). This group also had the highest mean ERMQ (Table 3). Locations in Group B were the next most contaminated, exceeding ERM concentrations for three contaminants and exceeding the ERL for four contaminants. This group had the next highest mean ERMQ. Group C locations showed some contamination but they had dierent characteristics compared to Groups A and B. The PCA separated the Group C locations along the second axis which was largely associated with higher concentrations of PCBs. Mean concentrations of PCBs, DDT and chlordane were higher than the corresponding ERM concentrations and the concentrations of lead and zinc exceeded the ERL concentrations. The ERMQ for this group was only slightly lower than that for group B locations but the overall quotient value was determined mostly by the concentrations of PCBs and DDTs, whereas metals contributed most to the quotient for Group B. Locations in Groups D and E were least contaminated but not pristine. Group D locations had concentrations of PCBs which exceeded the ERL. Group E locations had concentrations levels of As and
135
Hg which exceeded the ERL. Despite dierences in the composition of contaminants the ERMQ of these groups were almost identical. TPH does not have a guideline value but the mean concentrations were highest in Group A, elevated and similar in Groups B and C, and below the limits of detection in Groups D and E. Comparisons of the bulk metal concentrations with the guidelines were also made. These comparisons, which are not shown, found that the relative dierences in levels of contamination among groups were similar to those observed using the normalised metal concentrations. The main dierences were that locations in Group C appeared slightly more contaminated than the normalised data, whereas those in Groups B and E appeared slightly less contaminated and those in Groups A and D were unchanged. 3.2. Ecological response to dierent levels of contamination A total of 135 dierent taxa were identi®ed and used in the species analysis. The data were aggregated to 63 taxa at the family level, 39 taxa at the mixed level, and nine phyla. The species ordination into two dimensions provided an unacceptable ®t with stress=0.32 (Clarke, 1993). The three-dimensional solution provided a more acceptable ®t (stress=0.24) and, for the purposes of comparison among taxonomic levels, the three dimensional solution was used. For simplicity only the two-dimensional ordination plots which gave the best separation of locations are shown. When the ordination scores were plotted with respect to their sediment quality group (Fig. 3a±d), dierences in the community structure among some of the sediment types could be seen. Separation of the Groups A and B from Groups C, D and E was apparent in the plots at the species, family and mixed levels. These dierences, however, degenerated markedly at the phylum level. ANOSIM con®rmed
Table 2 Results from the principal components analysis of the sediment quality data No.
1 2 3 4 5 6 7 8 9
Eigenvalue
5.182 1.581 0.562 0.500 0.424 0.308 0.284 0.110 0.049 a
Variance (%)
57.57 17.57 6.24 5.56 4.71 3.43 3.16 1.22 0.55
Variance
57.57 75.14 81.38 86.94 91.65 95.08 98.23 99.45 100.00
Variablesa
As Cr Cu Ag Chlordane Dieldrin PCB PAH TPH
Factor Loadings PCA1
PCA2
PCA3
ÿ0.732 ÿ0.910 ÿ0.948 ÿ0.805 ÿ0.766 ÿ0.796 0.058 ÿ0.798 ÿ0.641
ÿ0.488 ÿ0.089 0.076 ÿ0.193 0.419 ÿ0.012 0.876 ÿ0.134 0.574
-0.287 ÿ0.136 ÿ0.1727 ÿ0.246 ÿ0.003 0.320 ÿ0.307 0.346 0.346
PCB, polychlorinated biphenyl; PAH, polyaromatic hydrocarbon; TPH, total petroleum hydrocarbons.
136
A.C. Roach et al. / Environmental Pollution 112 (2001) 131±143
Table 3 Summary of contaminant concentrations for groups derived from cluster analysesa A Asb (mg kgÿ1 dw)
B
C
D
Pbb (mg kgÿ1 dw)
Mean Standard error Mean Standard error Mean Standard error Mean Standard error Mean Standard error Mean Standard error Mean
22.6 3.1 8.96 2.91 123.9 12.6 411.4 80.4 0.75 0.19 3.125 0.730 1085.3
14.7 2.5 2.03 0.38 86.1 22.9 166.6 38.1 0.34 0.12 2.780 1.240 431.9
5.3 0.8 1.09 0.15 22.4 1.3 49.8 3.5 0.12 0.02 0.315 0.070 116.8
Znb (mg kgÿ1 dw)
Standard error Mean
214.6 2213.5
102.2 717.1
4.9 273.7
Cdb (mg kgÿ1 dw) Crb (mg kgÿ1 dw) Cub (mg kgÿ1 dw) Hgb (mg kgÿ1 dw) Agb (mg kgÿ1 dw)
ERLc
E
7.0 1.2 0.57 0.21 20.5 3.0 39.1 9.9 0.08 0.02 0.198 0.050 67.3 17.2 158.5
16.4 2.0 0.64 0.24 35.1 5.7 20.5 2.1 0.29 0.04 0.194 0.060 54.2 5.3 111.4
Chlordane (mg kgÿ1 dw)
Standard error Mean Standard error Mean
388.9 0.003 0.002 0.023
135.6 0.000 0.000 0.015
8.3 0.000 0.000 0.018
37.3 0.000 0.000 0.004
21.5 0.000 0.000 0.000
Dieldrin (mg kgÿ1 dw)
Standard error Mean
0.003 0.011
0.003 0.001
0.004 0.000
0.002 0.000
Total DDT (mg kgÿ1 dw)
Standard error Mean
0.002 0.079
0.001 0.015
0.000 0.313
Total PCBs (mg kgÿ1 dw)
Standard error Mean
0.034 0.128
0.005 0.083
Standard error Mean Standard error Mean Standard error Mean Standard error
0.036 4.16 1.74 10245 6547 2.65 0.3
0.026 1.47 0.43 1840 712 1.13 0.2
HCB (mg kgÿ1 dw)
Total PAHs (mg kgÿ1 dw) TPHs ERMQ
8.2 1.2
ERMc 70 9.6
81
370
34
270
0.15
0.71
1.0
3.7
47
218
150
410
na
na
0.0005dl
0.006
0.000 0.000
0.00002dl
0.008
0.000 0.009
0.000 0.000
0.002
0.046
0.131 0.572
0.005 0.121
0.000 0.022
0.023
0.180
0.242 0.00 0.00 1758 750 1.07 0.2
0.023 0.30 0.13 10d 0 0.27 0.03
0.008 0.41 0.38 10d 0 0.28 0.02
4.79 na
na
a
Letters A±E refer to the site groups from the cluster analysis. Single underline indicates signi®cantly (P<0.05) greater than the ERL. Double underline indicates that the Mean was signi®cantly greater than the ERM. Metals normalised to average%mud and organochlorine compounds normalised to 1% total organic carbon. HCB, hexachlorobenzene; Chlordane, sum of alpha-chlordane, gamma-chlordane and oxychlordane; Total PCBs sum of the arochlors 1248, 1254 and 1260; Total DDTs, sum of the isomers p,p0 -DDE, p,p0 -DDD and p,p0 -DDT. Total PAHs, sum of naphthalene, acenapthylene, acenaphthene, ¯uorene, phenanthrene, anthracene, ¯uoranthene, pyrene, benzo(a)anthracene, chrysene, benzo(b)¯uoranthracene, benzo(k)¯uoranthracene, benzo(a)pyrene, ind(123-cd)pyrene, dibenz(ah)anthracenebenzo, benzo(ghi)perylene; TPHs, total petroleum hydrocarbons; ERMQ, eects range median quality guideline values. b Data normalised wrt to 40% mud. c Values from Long et al., (1995). Except for chlordane which was from MacDonald et al., (1996). d Indicates that this values was below the detection limit for this study.
the signi®cant dierences among groups (Table 4), although the results varied with taxonomic level. At the species level, the most contaminated groups (A and B), were signi®cantly dierent from each other and all other groups, but there were no signi®cant dierences among Groups C, D and E. Dierences between Group A and all other groups were also detected at the family and mixed levels. At the phylum level signi®cant dierences were only detected for comparisons between Groups A and B, and A and D. Dierences between Groups B and C were not detected at the family and mixed levels, but dierences between Groups
B and D, and Groups B and E remained signi®cant. At the phylum level there were no signi®cant dierences between Group B and Groups C, D and E. Correlation of the biological nMDS axes with the ®rst principal component and the ERMQ showed a signi®cant correlation (P<0.04 or less) between one or more axes from the species, family and mixed level analyses, con®rming the relationship between community structure and overall level of contamination. Likewise, a signi®cant correlation (P<0.001) was found between TOC and at least one of these nMDS axes, but no correlation was found between any
A.C. Roach et al. / Environmental Pollution 112 (2001) 131±143
137
Fig. 3. Plot of ordination scores from nMDS analyses at each taxonomic level.
nMDS axis and percentage mud. At the phylum level, signi®cant correlation with the ®rst principal component was found for only one nMDS axis and no correlation was found with TOC. There was no signi®cant correlation between any of the nMDS axes and percentage mud. There was a signi®cant correlation between the ®rst principal component, the ERMQ and some of the general univariate population measures, including the number of species (r=0.30, P<0.1; r=ÿ0.34, P<0.06, respectively), the number of individuals (r=0.41, P<0.02; r=ÿ0.37, P<0.04) and the number of polychaetes (r=0.45, P<0.1; r=ÿ0.44, P<0.02). There was no signi®cant relationship between any general population measure and the proportion of mud, but there were signi®cant relationships between TOC and the number of species (r=ÿ0.36, P<0.05), number of individuals (r=ÿ0.42, P<0.02), number of polychaetes (r=ÿ0.46, P<0.01). Generally, however, even the signi®cant r values were small indicating that the relationships between these variables did not exhibit a particularly tight linear relationship. Taxonomic aggregation had a varied eect on the KW-H values for dierent taxa (Table 5). The KW-H for
the lumbrinerids and the capitellids was relatively low despite Lumbrineris latreilli and Mediomastus australiensis having relatively high values. This indicates that other species cancelled out their importance in the family level analysis. The KW-H for the harpacticoid copepods decreased overall when aggregated with other copepod orders at the mixed level but increased in particular between group comparisons (i.e. Group A versus B, and A versus D). The merging of the spionids had an averaging eect with the overall value at the family level being lower than that for Spio paci®ca but higher than Prionospio aucklandica. The overall KW-H for the Dorvilleidae and the Cirratulidae increased when compared with comparisons for species of these families. These breakdowns show that the polychaetes and the crustaceans were the taxa which contributed most to the dierences among groups at all levels. Other taxa such as the oligochaetes, nemerteans and molluscs had middle range KW-H values at the species, family and/or mixed level aggregations. There was no evidence of a speci®c indicator(s) of pollution as no particular species or aggregated taxon showed consistently high KW-H values overall or for speci®c comparisons among more or less contaminated groups.
138
A.C. Roach et al. / Environmental Pollution 112 (2001) 131±143
Table 4 Results of the nested ANOSIM analyses and the pairwise analyses comparing communities among sediment quality groups for each taxonomic level. Perm., the number of permutations used in the analysis. Signi®cance attributed when level 5% or less. Pair-wise comparisons are not adjusted for multiple comparisons R
Perm.
Sig. level
Species Clusters: C Locations (C) AvB AvC AvD AvE BvC BvD BvE CvD CvE DvE
0.485 0.187 0.421 0.984 0.935 1.000 0.423 0.537 0.497 ÿ0.147 ÿ0.031 0.174
10000 10000 1716 210 3003 462 330 6435 792 495 126 1287
0.0* 0.0* 0.2* 0.5* 0.0* 0.2* 3.3* 0.1* 1.5* 81.4 57.1 9.0
Mixed Clusters: C Locations (C) AvB AvC AvD AvE BvC BvD BvE CvD CvE DvE
0.428 0.138 0.344 0.952 0.915 1.000 0.138 0.370 0.473 ÿ0.007 0.094 0.075
10000 10000 1716 210 3003 462 330 6435 792 495 126 1287
0.0* 0.0* 1.4* 0.5* 0.0* 0.2* 16.7 0.1* 0.8* 48.1 24.6 27.7
Family Clusters: C Locations (C) AvB AvC AvD AvE BvC BvD Be CvD CvE DvE
0.273 0.161 0.254 0.595 0.724 0.901 ÿ0.024 0.171 0.222 ÿ0.134 ÿ0.169 0.124
10000 10000 1716 210 3003 462 330 6435 792 495 126 1287
0.0* 0.0* 2.7* 1.0* 0.0* 0.2* 50.9 1.5* 4.4* 76.8 92.1 16.1
Phylum Clusters: C Locations (C) AvB AvC AvD AvE BvC BvD Be CvD CvE DvE
0.134 0.155 0.290 0.000 0.425 0.072 ÿ0.146 0.090 0.119 ÿ0.04 ÿ0.006 0.097
10000 10000 1716 210 3003 462 330 6435 792 495 126 1287
1.6* 0.0* 1.1* 38.6 0.4* 20.6 81.8 10.8 15.2 55.6 49.2 19.7
4. Discussion 4.1. Usefulness of ®eld experiment approach This study showed that the patterns of recruitment of estuarine macroinvertebrates can be in¯uenced by sediment quality at small spatial scales. Eects were manifest as alterations in the multivariate composition of the community and as reductions in gross univariate community characteristics such as total number of individuals and number of taxa with increasing loads of pollutants. Such ecological responses also occur following contamination over much larger spatial scales (Rygg, 1985) indicating that small scale ®eld experiments can produce responses which are realistic in terms of the natural communities (Clements, 1997). Contaminated sediments may aect the structure of benthic communities via various mechanisms. For example, recruitment may be reduced either through direct toxic eects to settling larvae or the alteration of cues important for their settlement (Menzie, 1984; Watzin and Roscigno, 1997). Similarly contaminants can aect adult invertebrates causing direct toxicity, inducing migration and altering biological interactions (Reynoldson, 1987; Scott, 1989). Costello and Thrush (1991) suggested that ®eld experiments, particularly colonisation experiments, should be a useful measure of environmental stress because they essentially provide a bioassay of multiple taxa which are natural components of benthic systems. These organisms are exposed at their sensitive early life stage and in situ experiments control for natural processes which may also greatly in¯uence spatial pattern. The few examples cited earlier, and this study, highlight the potential of ®eld experiments to detect eects of contaminated sediments generally. The sensitivity to contamination probably varies among phyla. For example, Warwick and Clarke (1994) found that generally annelids were the least sensitive to pollution, followed by molluscs, crustaceans and echinoderms in order of increasing sensitivity. This study did not ®nd one particular species or taxon speci®cally associated with clean or contaminated sediments. Polychaetes and crustaceans were the numerically dominant taxa and both appeared equally responsible for dierences in community structure among the various sediment groups, though only polychaete abundance was correlated with the level of contamination. These results suggest that taxa which are relatively less sensitive to contamination are capable of showing patterns of eects. In general, the likelihood of toxicity increases with increasing exceedance of sediment quality guidelines (Carr et al., 1996; Long et al. 1998a, b; Hyland et al., 1999). Yet these studies also showed that not all sediments which exceed the guidelines will be toxic and, conversely, that sediments below the guideline con-
A.C. Roach et al. / Environmental Pollution 112 (2001) 131±143
centrations can show some level of toxicity. In particular, Hyland et al. (1999) who studied in situ benthic communities observed degradation at levels of contamination well below suggested ERMQ guideline levels. The sediments included in this study did not have ERMQ values as low as those in the study by Hyland et al. (1999) and as such cannot be used to test if recruitment is likely to be as sensitive as in situ communities. This study did ®nd, however, that sediments which exceeded the guidelines values did not always have an impact on the recruitment of benthic communities. Samples allocated into Groups A, B and C had concentrations of contaminants and ERMQ values which are frequently associated with biological eects. Those samples allocated to Groups D and E were relatively unpolluted, the concentrations and ERMQ values being only occasionally associated with biological eects (Long and MacDonald, 1998; Long et al., 1998a). Thus it is reasonable to expect community dierences between the Groups A, B and C, and Groups D and E. Instead, dierences occurred among the two most polluted Groups A and B, and the two least polluted groups, D and E but Group C was not separated from Groups D and E. In terms of managing the environment, guidelines which sometimes give false positives are less problematic than those which underestimate eects because they cause managers to be overly protective (Fairweather, 1991). False positives are potentially a problem, however, when guidelines alone are used to make decisions to remediate because it is assumed that the eects predicted are actually occurring. The ®nancial costs associated with such decisions are unnecessary. This issue reinforces the point made by many authors (MacDonald et al., 1996; Swartz and DiToro, 1997; Long et al., 1998a, b) that sedimentbiota interactions are complex and that predictions of toxicity/bioavailability based on sediment quality guidelines alone are not always reliable and that eectsbased studies are necessary for a proper evaluation of environmental risks when contaminant levels exceed guideline values. There was strong inter-correlation among contaminants, making it dicult to ascribe eects to particular contaminants. This is a common problem in observational studies on benthic communities (Scott, 1989). The identi®cation of causal factors was not a speci®c aim of this study but the results of this study show that community responses in small scale experiments can be related to the overall level of contamination. Therefore the use of these types of experimental approaches which incorporate sediment-spiking techniques (Murdoch et al., 1997) and homogenous sediments, either natural or arti®cial (Watzin et al., 1994; Gonzalez, 1996) thereby controlling the correlation among contaminants and between contaminant concentrations and sediment characteristics (i.e. grain size and TOC)
139
have some potential as a tool to evaluate the ecological signi®cance of particular contaminants. Some caution must be urged at this point, however, because the more arti®cial a substrate or the more it is processed (e.g. physically disturbed, dried, or frozen) prior to use the greater the potential for artefacts. Important factors which can contribute to biological (e.g. larval recruitment cues, post recruitment growth and survival) and chemical responses (e.g. speciation, transformation and bioavailability) such as the concentrations of dissolved organic carbon, unionized ammonia and amorphous sul®des, microbial activity and redox potential may take some time to reach, or may not reach, typical levels when arti®cal sediments are placed in the ®eld. The organisms themselves will probably also modify the chemistry of their environment through bioturbation/bioirrigation and a suite of other interactions. These factors are, however, also problematic to some degree in laboratory studies (Campbell and Tessier, 1996). It is important, therefore, that any other potentially signi®cant changes in chemistry are also taken into account, as much as possible, when designing studies and interpreting the results of studies which aim to establish relationships between concentrations of particular contaminants and biological endpoints, whether they be laboratory or in situ repsonses. 4.2. Taxonomic resolution The interpretation of the present results varied with taxonomic level. In general there were smaller dierences among sites, locations and groups with decreasing taxonomic resolution. Importantly, tests of speci®c hypotheses using analysis of similarities also varied. The species-level analysis dierentiated Group C locations from those of Group B but the family and mixed levels did not, although signi®cant dierences between the most (A and B) and least polluted (D and E) groups were preserved. The analysis at phylum level was markedly dierent with far fewer signi®cant dierences among groups being found and little dierentiation between the most and least contaminated sites. This suggests that aggregation to either family or mixed levels will identify major eects, albeit with some loss of resolution, but that phylum level analysis is unreliable. In their analysis of taxonomic resolution, James et al., (1995) found that the number of species per family, and the relative dominance of particular species in a family aected whether aggregated data retained the same relative dierences among sites as species-level data. Similar results were found in this study but the overall trend was variable. Some species contributed more to dierences among groups than when they were combined with other species of the same family. In direct contrast, particular families contributed more signi®cantly to
140
A.C. Roach et al. / Environmental Pollution 112 (2001) 131±143
Table 5 W-H breakdowns among groups. Hab..Hbe are the KW-H statistics for comparisons between speci®c groups. KW-H is the statistic across all groups Hab
Hac
Had
Hae
Hbc
Hbd
Hbe
KW-H
Species level Lumbrineris latreilli Lysilla paci®ca Callianassa arenosa Hesionid sp.1 Mediomastus australiensis Harpacticoida Spio paci®ca Limnoporeia yarrague Aricidea fauveli Prionospio aucklandica Loxoconcha sp. Sphaerosyllis sp.4 Oligochaeta Sphaerosyllis semiverrucosa Rhinothelepus lobatus Nemertea Scyphoproctus towraensis Caulleriella tricapillata Tharyx sp.2 Exogone sp.1
1.27 1.49 4.77 0.37 0.06 1.60 3.97 1.23 0.04 3.68 3.34 4.89 4.27 2.42 1.66 3.86 2.96 1.97 0.01 3.68
0.08 1.57 2.10 2.25 1.74 3.01 2.31 3.48 5.05 2.39 1.47 1.87 3.01 4.94 ± 0.90 0.09 5.43 0.01 1.39
4.54 3.64 0.18 6.68 6.46 2.82 0.07 0.28 0.96 4.37 1.36 3.95 4.39 4.29 1.13 0.45 3.85 2.83 1.27 0.31
1.41 0.08 1.86 0.02 0.87 7.47 2.53 0.30 0.39 5.40 4.74 4.49 1.10 2.80 0.02 2.46 3.49 0.93 0.90 1.18
1.87 5.57 0.25 1.20 1.38 0.54 0.16 0.75 5.16 0.15 0.57 0.61 0.09 0.79 1.40 0.91 1.23 0.94 0.00 0.30
8.78 0.47 8.38 4.67 6.44 0.45 6.21 3.08 0.78 0.04 1.20 0.03 0.00 0.09 5.19 2.51 0.08 0.01 1.64 2.36
4.38 2.55 0.71 0.22 0.58 4.26 0.07 0.33 0.24 0.51 0.28 0.04 1.77 0.07 1.95 0.11 0.02 0.13 0.76 0.73
12.67 12.53 10.97 10.73 9.92 9.57 9.06 7.71 7.42 6.88 6.40 6.17 6.07 5.73 5.66 5.52 5.48 5.46 4.89 4.68
Family level Callianassidae Terebellidae Harpacticoida Hesionidae Phoxocephalidae Glyceridae Spionidae Paraonidae Dorvilleidae Leptonidae Oligochaeta Lumbrineridae Cirratulidae Sabellidae Nemertea Nephtyidae Syllidae Nebaliidae Ampharetidae Podocopina
4.78 0.67 1.60 0.41 1.87 0.96 6.92 0.03 6.42 0.79 4.27 0.29 1.78 0.83 3.86 3.69 3.20 1.87 2.11 2.73
2.10 1.00 3.01 2.47 5.02 1.23 2.64 5.05 1.09 ± 3.01 0.12 4.26 1.40 0.90 0.06 3.67 0.06 ± 1.96
0.18 3.52 2.81 6.07 0.03 2.21 4.78 0.96 1.32 ± 4.39 3.16 3.81 2.47 0.45 2.18 1.98 0.33 0.76 0.67
1.86 0.01 7.47 0.03 0.30 0.11 3.16 0.40 2.06 2.25 1.10 2.75 0.51 1.07 2.45 0.49 3.70 0.00 0.00 1.64
0.25 3.11 0.54 1.31 1.00 0.11 1.20 5.16 1.11 0.53 0.08 0.02 0.55 2.95 0.91 2.61 0.00 0.99 1.83 0.00
8.38 1.31 0.45 3.69 2.94 6.41 0.28 0.78 3.13 1.18 0.00 2.05 0.29 1.92 2.51 0.47 0.01 1.15 0.65 0.67
0.71 1.04 4.26 0.23 0.69 1.59 0.63 0.24 0.74 0.72 1.77 1.69 0.26 0.04 0.11 0.43 0.04 1.86 2.03 0.00
10.97 9.65 9.57 9.52 9.31 8.41 8.00 7.42 7.15 6.29 6.07 5.80 5.75 5.64 5.52 4.82 4.72 3.62 3.52 3.44
Mixed level Decapoda Terebellidae Hesionidae Glyceridae Spionidae Paraonidae Bivalvia Dorvilleidae Copepoda Oligochaeta Lumbrineridae Cirratulidae Sabellidae Nemertea Amphipoda Nephtyidae Syllidae
4.87 0.67 0.41 0.96 6.92 0.03 3.91 6.42 5.22 4.27 0.29 1.78 0.83 3.85 0.77 3.69 3.20
2.10 1.00 2.47 1.23 2.64 5.05 2.14 1.08 3.01 3.01 0.12 4.27 1.40 0.90 2.17 0.06 3.67
0.15 3.52 6.07 2.22 4.79 0.96 0.07 1.31 4.39 4.39 3.16 3.81 0.16 0.45 0.02 2.19 1.98
2.90 0.01 0.03 0.11 3.16 0.39 2.60 2.06 7.47 1.10 2.75 0.51 1.07 2.46 0.72 0.49 3.70
0.33 3.11 1.31 0.11 1.00 5.16 0.27 1.11 0.22 0.09 0.02 0.55 2.95 0.91 0.79 2.61 0.00
7.91 1.31 3.69 6.41 0.28 0.78 4.32 3.13 0.05 0.00 2.05 0.29 1.92 2.51 1.32 0.47 0.01
0.31 1.04 0.23 1.59 0.60 0.24 2.03 0.74 0.44 1.77 1.69 0.26 0.04 0.11 0.00 0.44 0.04
11.16 9.65 9.52 8.41 8.01 7.42 7.24 7.15 7.05 6.07 5.80 5.75 5.64 5.52 4.85 4.82 4.73
A.C. Roach et al. / Environmental Pollution 112 (2001) 131±143
141
Table 5 (continued) Hab
Hac
Had
Hae
Hbc
Hbd
Hbe
KW-H
Ostracoda Leptostraca Ampharetidae
2.72 1.87 2.11
2.77 0.06 ±
0.67 0.33 0.76
1.82 0.00 0.00
0.06 0.99 1.83
0.72 1.15 0.65
0.00 1.86 2.03
4.20 3.62 3.52
Phylum level Annelida Crustacea Nemertea Mollusca
5.93 4.13 3.85 0.70
3.69 4.29 0.90 0.60
5.81 2.46 0.45 0.06
4.07 5.68 2.46 3.16
0.07 0.01 0.91 0.00
0.06 0.90 2.51 1.27
0.02 0.07 0.11 0.84
8.02 7.70 5.51 5.45
dierences among groups than any particular species did in the analysis of the species level data. In general the polychaetes and crustaceans dominated the communities and contributed most to the dierences among groups. This strongly in¯uenced the eect of aggregation from family to mixed level. In the mixed level analysis, the polychaetes were represented at the family level, crustaceans were mainly aggregated at the order level and had relatively few families per order, whereas the other taxa which had little eect at the species and family level were aggregated to class or higher. Whilst the number of taxa decreased from 63 to 39 in the mixed level analysis, there was little change in the relative abundance of the major taxa which contributed to differences in the family-level analysis. In the phylum-level analysis, data were proportionally aggregated to a much greater degree with only nine phyla being used. The resulting analysis indicated that aggregating to this level merged the taxa which were both important and redundant in detecting dierences among groups. This result agrees with the suggestion by James et al., (1995) that the eect of aggregation may depend upon the assemblage examined. It appears that family-level taxonomy, but not phylum level, is sucient to detect some types of impact or to describe the spatial and temporal patterns of marine benthic communities (Herman and Heip, 1988; Warwick, 1988; Ferraro and Cole, 1992; James et al., 1995). Our results are relatively consistent with this generalisation but they emphasize that whilst a signi®cant result at a higher taxonomic level is probably reliable a non-signi®cant result may not be reliable. In this situation, identi®cation to lower levels may be warranted especially if dierences among particular treatments are of importance to the hypothesis being tested or if concurrent environmental data suggests that dierences may occur. 5. Conclusions The major implications of this study are that contaminated sediments can aect the recruitment of
benthic invertebrates and that small-scale manipulative ®eld experiments can provide a measure of these impacts. Such approaches have promise as tools for the development and veri®cation of sediment quality criteria for the protection of benthic communities especially when combined with other experimental techniques such as using arti®cial sediment and spiking. These approaches could control for factors which may normally confound observational studies, although development is required to allow us to understand the signi®cance of potential artefacts. In addition, this study supports the growing body of evidence that higher taxonomic levels are sucient to detect major trends in the data. There was, however, some loss of detail when using data aggregated to the family or mixed level. Data aggregated to the phylum level performed very poorly and is not recommended. Acknowledgements We would like to thank Matthew Dasey, Leigh Gray, Scott Herring, Jamie Potts, Murray Root and John Runcie for their assistance in setting up and running this experiment and also to Rick Johnson, Andrew Parker (who also identi®ed the ostracod fauna), Rachael Evans and Robin Marsh, for sorting the samples. Dr. Don Morrisey (National Institute of Water and Atmospheric Research, New Zealand) and Dr. Micheal Warne (Centre for Environmental Toxicology) and an anonymous reviewer are thanked for their comments on the manuscript. Dr. Peter Scanes, Dr. Klaus Koop and Dr. David Leece are thanked for their comments and for facilitating this study. References Adams, W., Kimerle, R., Barnett, J., 1992. Sediment quality and aquatic life assessment. Environmental Science and Technology 26, 1865±1875. AOAC, 1990. Association of Ocial Analytical Chemists, 15th Edition. Ocial Methods of Analysis Belbin, L., 1995. PATN Pattern Analysis Package. CSIRO Division of Wildlife and Ecology, Canberra.
142
A.C. Roach et al. / Environmental Pollution 112 (2001) 131±143
Berge, J., 1990. Macrofauna recolonization of subtidal sediments. Experimental studies on defaunated sediment contaminated with crude oil in two Norwegian fjords with unequal eutrophication status. 1. Community responses. Marine Ecology Progress Series 66, 103±115. Bilyard, G.R., 1987. The value of benthic infauna in marine polution monitoring studies. Marine Pollution Bulletin 18, 581±585. Burton, G., 1995. Critical issues in sediment bioassays and toxicity testing. Journal of Aquatic Ecosystem Health 4, 151±156. Campbell, P., Tessier, A., 1996. Ecotoxicology of metals in the aquatic environment: geochemical aspects. In: Newman, M., Jagoe, C. (Eds.), Ecotoxicology: A Hierarchical Treatment. CRC Press, Boca Raton, pp. 11±58. Carr, R., Chapman, B., Presley, Biedenbach, Robertson, L., Boothe, P., Kilada, R., Wade, T., Montagna, P., 1996. Sediment porewater toxicity assessment studies in the vicinity of oshore oil and gas production platforms in the Gulf of Mexico. Canadian Journal of Fisheries and Aquatic Sciences 53, 2618±2628. Chapman, P.M., 1989. Current approaches to developing sediment quality criteria. Environmental Toxicology and Chemistry 8, 589± 599. Chapman, P.M., 1992. Sediment quality triad approach. In: Sediment Classi®cation Methods Compendium, Chap. 10. US Environmental Protection Agency Clarke, K.R., 1993. Non-parametric multivariate analyses of changes in community structure. Australian Journal of Ecology 18, 117±143. Clarke, K.R., Ainsworth, M., 1993. A method of linking multivariate community structure to environmental variables. Marine Ecology Progress Series 92, 205±219. Clarke, K.R., Warwick, R.M., 1994. Change in Marine Communities: An Approach to Statistical Analysis and Interpretation. Natural Environment Research Council, UK. Clements, W., 1997. Ecological signi®cance of endpoints used to assess sediment quality. In: Ingersoll, C., Dillon, T., Biddinger, G. (Eds.), Ecological Risk Assessment of Contaminated Sediments. SETAC Press Florida, pp. 123±134. Coade, G., Teutsch, M., 2000. Organochlorine contaminants in the sediments of the Georges River and Sydney Harbour. Environment Protection Authority, NSW. In press Costello, M.J., Thrush, S.F., 1991. Colonization of arti®cial substrata as a multi-species bioassay of marine environmental quality. In: Bioindicators and Environmental Management. Academic Press, pp 401±418 Coull, B.C., Chandler, G.T., 1992. Pollution and meiofauna: ®eld, laboratory and mesocosm studies. Oceanography and Marine Biology Annual Review 30, 191±272. Fairweather, P.G., 1991. Statistical power and design requirements for environmental monitoring. Australian Journal of Marine and Freshwater Research 42, 555±568. Ferraro, S., Cole, F., 1992. Taxonomic level sucient for assessing a moderate impact on macrobenthic communities in Puget Sound, Washington, USA. Canadian Journal of Fisheries and Aquatic Sciences 49, 1184±1188. Gonzalez, A., 1996. A laboratory-formulated sediment incorporating synthetic acid-volatile sul®de. Environmental Toxicology and Chemistry 15, 2209±2220. Graf, G., 1992. Benthic-pelagic coupling: a benthic view. Oceanography and Marine Biology Annual Review 30, 149±190. Gray, J.S, 1981. The Ecology of Marine Sediments. Cambridge University Press, Cambridge. Green, R.H., Boyd, J., MacDonald, J., 1993. Relating sets of variables in environmental studies: the sediment quality triad as a paradigm. Environmetrics 4, 439±457. Herman, P., Heip, C., 1988. On the use of meiofauna in ecological monitoring: who needs taxonomy? Marine Pollution Bulletin 19, 665±668.
Hintze, J., 1995. NCSS 6.0 Users Guide. Number Cruncher Statistical Systems, Kaysville. Hyland, J.L., Van Dolah, R.F., Snoots, T.R., 1999. Predicting stress in benthic communities of southeastern U.S. estuaries in relation to chemical contamination of sediments. Environmental Toxicology and Chemistry 18, 2557±2564. James, R., Lincoln Smith, M., Fairweather, P., 1995. Sieve mesh-size and taxonomic resolution needed to describe natural spatial variation of marine macrofauna. Marine Ecology Progress Series 118, 187±198. Jones, A.R., Watson-Russell, C.J., Murray, A., 1986. Spatial patterns in the macrobenthic communities of the Hawkesbury estuary, New South Wales. Australian Journal of Marine and Freshwater Research 37, 521±543. Keith, L., 1992. Compilation of E.P.A.'s Sampling and Analysis Methods. Lewis Publishers, Chelsea. Lamberson, J.O., DeWitt, T.H., Swartz, R.C., 1992. Assessment of sediment toxicity to marine benthos. In: Burton, Jr., G.A. (Ed.), Sediment Toxicity Assessment. Lewis Publishers, Boca Raton, pp. 183±212. Long, E.R., MacDonald, D.D., 1998. Recommended uses of emperically derived sediment quality guidelines for marine and estuarine ecosystems. Human and Ecological Risk Assessment 4, 1019±1039. Long, E.R., MacDonald, D.D., Smith, S., Calder, F., 1995. Incidence of adverse biological eects within ranges of chemical concentrations in marine and estuarine sediments. Environmental Management 19, 81±97. Long, E.R., Field, L.J., MacDonald, D.D., 1998a. Predicting toxicity in marine sediments with numerical sediment quality guidelines. Environmental Toxicology and Chemistry 17, 714±727. Long, E.R., MacDonald, D.D., Cubbage, J.C., Ingersoll, C.G., 1998b. Predicting the toxicity of sediment-associated trace metals with simultaneously extracted trace metal: acid-volatile sul®de concentrations and dry weight-normalized concentrations: a critical comparison. Environmental Toxicology and Chemistry 17, 972±974. Loring, D., Rantala, R., 1992. Manual for the geochemical analyses of marine sediments and suspended particulate matter. Earth-Science Reviews 32, 235±283. MacDonald, D., Carr, R., Calder, F., Long, E., Ingersoll, C., 1996. Development and evaluation of sediment quality guidelines for Florida coastal waters. Ecotoxicology 5, 253±278. Malek, J., 1992. Apparent eects threshold approach. In: Sediment Classi®cation Methods Compendium, Chap. 11. US Environmental Protection Agency Mann, K.H., 1976. Production on the bottom of the sea. In: Cushing, D.H., Walsh, J.J. (Eds.), The Ecology of the Seas. Saunders, Philadelphia, pp. 225±250. Menzie, C., 1984. Diminishment of recruitment: a hypothesis concerning impacts on benthic communities. Marine Pollution Bulletin 15, 127±128. Morrisey, D.J., Underwood, A.J., Howitt, L., 1996. Eects of copper on the faunas of marine soft-sediments: an experimental ®eld study. Marine Biology 125, 199±213. Murdoch, M., Chapman, P., Norman, D., Quintino, V., 1997. Spiking sediment with organochlorine for toxicity testing. Environmental Toxicology and Chemistry 16, 1504±1509. Olsgard, F., 1999. Eects of copper contaminaton on recolonisation of subtidal marine soft sediments Ð an experimental ®eld study. Marine Pollution Bulletin 38, 448±462. Parkhurst, B.R., 1995. Are single species toxicity test results valid indicators of eects to aquatic communities. In: Cairns, J., Niederlehner, B.R. (Eds.), Ecological Toxicity Testing: Scale, Complexity and Relevance. Lewis, pp. 105±121. Power, M., McCarty, L., 1997. Fallacies in ecological risk assessment practices. Environmental Science and Technology 31, 370±375. Rayment, G.E., Higginson, R., 1992. Australian Laboratory Handbook of Soil and Water Chemical Methods. Inkata Press, Sydney.
A.C. Roach et al. / Environmental Pollution 112 (2001) 131±143 Rygg, B., 1985. Eect of sediment copper on benthic fauna. Marine Ecology Progress Series 25, 83±89. Reynoldson, T., 1987. Interactions between sediment contaminants and benthic organisms. Hydrobiologia 149, 53±66. Scott, K., 1989. Eects of contaminated sediments on marine benthic biota and communities. In: Contaminated Marine Sediments Ð Assessment and Remediation. National Research Council. National Academy Press, Washington, pp. 132±154 Swartz, R.C., Di Toro, D.M., 1997. Sediments as complex mixtures: an overview of methods to assess ecotoxicological signi®cance. In: Ingersoll, C.G., Dillon, T., Biddinger, G.R. (Eds.), Ecological Risk Assessment of Contaminated Sediments. SETAC Press, Pensacola, pp. 255±270. Tagatz, M., Ivey, J., Lehman, H., Oglesby, J., 1979. Eects of Sevin on the development of experimental estuarine communities. Journal of Toxicology and Environmental Health 5, 643±651. Tetra-Tech, 1986. Recommended Protocols for Measuring Organic Compounds in Puget Sound Sediment and Tissue Samples. Tetra Tech, Bellevue, Washington. Underwood, A.J., Peterson, C.H., 1988. Towards an ecological framework for investigating pollution. Marine Ecology Progress Series 46, 227±234.
143
Warwick, R.M., 1988. The level of taxonomic discrimination required to detect pollution eects on marine benthic communities. Marine Pollution Bulletin 19, 259±268. Warwick, R.M., 1993. Environmental impact studies on marine communities: pragmatical considerations. Australian Journal of Ecology 18, 63±90. Warwick, R.M., Clarke, K.R., 1994. Relearning the ABC: Taxonomic changes and abundance/biomass relationships in disturbed benthic communities. Marine Biology 118, 739±744. Watzin, M.C., Roscigno, P.R., 1997. The eects of zinc contamination on the recruitment and early survival of benthic invertebrates in an estuary. Marine Pollution Bulletin 34, 443±455. Watzin, M.C., Roscigno, P.R., Burke, W.D., 1994. Community-level ®eld method for testing the toxicity of contaminated sediments in estuaries. Environmental Toxicology and Chemistry 13, 1187± 1193. Wol, W.J., 1983. Estuarine Benthos. In: Ketchum, B.H. (Ed.), Ecosystems of the World. Chapter 6. Elsevier, Amsterdam, pp. 151±182. Zarba, C., 1992. Equilibrium partitioning approach. In: Sediment Classi®cation Methods Compendium, Chap. 6. US Environmental Protection Agency