Environ. Sci. Technol. 2009, 43, 2118–2123
Enhancing the Ecological Significance of Sediment Contamination Guidelines through Integration with Community Analysis J U D I E . H E W I T T , * ,† MARTI J. ANDERSON,‡ CHRIS W. HICKEY,† SHANE KELLY,§ AND SIMON F. THRUSH† National Institute of Water and Atmospheric Research, New Zealand, Department of Statistics, University of Auckland, New Zealand, and Auckland Regional Council, New Zealand
Received August 6, 2008. Revised manuscript received December 15, 2008. Accepted December 19, 2008.
The underlying basis of sediment quality guidelines needs to be accepted both by the international scientific community and socially before they can be of use. Increasingly, this means that just saying that a certain number of species will be affected is not sufficient. Instead guidelines need to be related to changes in community composition and predicted changes in biodiversity and ecosystem function. This study derived guidelines for copper, zinc, and lead, from field-based SSDs, that predicted a 50% decrease in abundance of 5% of the taxa, well below present management guidelines. However, a multivariate model of effect demonstrated considerable changes in community composition occur at levels below these derived guidelines. Changes in the degree of rarity also occurred signaling potential changes to meta-community structure and resilience of the region. Furthermore, the most sensitive taxa indicated by the multivariate analysis were frequently of large size and those likely to affect oxygen, carbon, and nutrient exchanges between the water column and the seafloor, leading to ecological effects beyond the obvious change in composition. We suggest that guidelines should preferentially be field derived, backed where possible by experimental work. Community and functional responses should be calculated, from the same field studies, and explicitly mentioned whenever the guidelines are used to allow environmental costs to be more realistically determined.
Introduction Increasing population pressure and urbanization of the coastal zone has resulted in a variety of chronic impacts operating on coastal and estuarine ecosystems. In urbanized estuaries and harbors, stormwater runoff is a significant, often diffuse source of contaminants to the marine environment. Increasingly, environmental managers challenge scientists to predict trends in contaminant loading and define thresholds for ecologically significant effects. However, this is far from being an easy task. Laboratory-based dose-response experiments have predominantly been used by ecotoxicolo* Corresponding author e-mail:
[email protected]. † National Institute of Water and Atmospheric Research. ‡ University of Auckland. § Auckland Regional Council. 2118
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 6, 2009
gists to develop concentration guidelines for water and sediment quality, but the methods continue to change as ecotoxicologists seek to expand predictions to population dynamics and community structure (1-4). At the same time, community ecologists undertaking field assessments of contaminant impacts frequently identify significant ecological change in community structure and function at concentrations below the values derived through ecotoxicological assays. This study seeks a way to integrate ecotoxicology with a community ecology perspective (5-9) to produce ecologically meaningful guidelines. A number of sediment quality guidelines (SQGs) for contaminants have been developed, although no indication of the number or proportion of species affected nor the magnitude of effects are provided and several other problems exist (10). First, SQGs are usually not developed using the species present in the region for which protection is sought, with most data provided by a very limited number of test species that have been examined by standardized laboratory procedures. Second, the species on which the SQG is based may not necessarily coexist (i.e., they may not represent a natural community). Third, ecotoxicological assays conducted in laboratory conditions poorly reflect natural systems. Thus, the relationship between the SQGs and either community-level effects or ecosystem-function effects for a given contaminant is generally unknown. Identifying these relationships is especially important, as the effects of contaminant can vary depending on species density and diversity, the strength of interspecies interactions, and habitat type (11-14). This study attempts to link sediment quality guidelines to actual changes in community composition in the field, using comparisons between present low threshold effects level (TELs) for copper, lead and zinc, effect concentrations (FECs) developed from field-based species sensitivity distributions (f-SSDs) (15), and multivariate community analysis. The ecological relevance of the guidelines (both TELs and FECs) is primarily assessed in terms of changes in community composition. We then seek to further understand the relevance of the observed community changes by considering the response of rare species and organisms with different functional attributes, as these should influence ecosystem function.
Materials and Methods Sampling. An extensive field survey (84 sites) was performed on stormwater contamination (copper, lead, and zinc) and macrofaunal communities in unvegetated soft-sediment intertidal areas surrounding the city of Auckland, New Zealand (Supporting Information (SI) Figure S1). This area is highly urbanized and strong upward trends over time have become apparent in the concentrations of copper and zinc in estuarine sediments at many sites (16, 17). Copper, lead, and zinc are considered to be primary contaminants, with the concentrations of other contaminants low in comparison (16, 17). Sites were dispersed both spatially throughout the region and over a range in stormwater contamination (SI Figure S2). Care was taken to ensure that all sediment types were represented in the data set at sites with low, medium, or high contaminant levels (SI Figure S2). Sampling was spread over 3 years (2002, 2004, and 2005), with some sites sampled in more than one year in order to check whether temporal effects would confound results. However, all sites were sampled in October or November to minimize the potential effects of seasonal recruitment of 10.1021/es802175k CCC: $40.75
2009 American Chemical Society
Published on Web 02/06/2009
macrofauna. This resulted in a total of 95 sample stations. The area of each site was either 100 × 50 m or 100 × 100 m. In common with the majority of studies assessing the health of estuarine and coastal areas we focused on benthic macrofauna (18). Benthic macrofauna were sampled using a 13 cm diameter, 15 cm deep corer at 6-12 randomly selected positions at each site, with more replicates taken from the larger sites. As the sample sizes differed, the potential for bias in species richness and total average abundances causing overestimates of dissimilarity (19) was investigated but proved negligible. Core samples were sieved on a 0.5 mm mesh, preserved in 70% IPA and stained with 0.2% Rose Bengal. Macrofauna were identified and counted to the lowest practical level (SI Text S3). Replicates were averaged to produce a station mean for each taxon which was used in all subsequent analyses. Altogether, N ) 103 taxa were recorded. Sediment particle size was sampled using a 2 cm diameter × 2 cm deep corer, with six cores taken per site and pooled prior to analysis. Only the top 2 cm was sampled as the majority of the macrofauna present either live or feed no deeper than this, and this area is frequently mixed both by bioturbation and wave action. Particle size was determined, after digestion, in 9% hydrogen peroxide to disperse the sample and remove organics, using wet sieving. The g0.5 mm (coarse sand), 0.063-0.5 mm (medium to fine sand) and the 1 individuals, and Rm is the proportion of common taxa not exhibiting a negative relationship with m. Changes in Community Composition. There were two steps to this (see SI Figure S4). Initially, as there were good correlations among the three metals: rCu,Pb ) 0.85, rCu,Zn ) 0.88 and rPb,Zn ) 0.90 (p < 0.001 for all), a principal component analysis (PCA) was used to reduce these to a single variable (see also ref 22). The first PCA axis represented ∼94% of the variability and was subsequently used as the contaminant gradient in all the multivariate analyses (hereafter called the contaminant variable). After this, canonical analysis of principal coordinates (CAP (23, 24),) was conducted to relate the ecological communities to the contaminant variable. While this was done using a number of different distance measures (25), only the results based on the Bray-Curtis similarity of square root-transformed abundances are used here. The potential for differences in sediment particle size, generated by hydrodynamics, to confound the CAP was examined and dismissed (see SI Text S7). Comparisons between TELs, FECs, and Changes in Community Composition. After the gradient in community structure had been established, the TELs and FECs were converted into a contaminant variable value (see equation in SI Text S8). Note that for the FECs a value was obtained for both the upper and lower values of the 5% range. Comparisons of the change in community composition between the TELs and FECs were made visually. However, differences in community composition between stations with contaminant levels above and below the calculated FECs were assessed further in two ways. First, the degree of community change was calculated using the average Bray-Curtis dissimilarity between the two groups of stations. Values were calculated on raw data rather than transformed data, as changes in relative abundance of dominant taxa were considered to be important in capturing changes in communities associated with contaminants. A second measure of community change along the contaminant variable was determined using detrended canonical correspondence analysis by segments, where a 1-unit change along the axis is interpretable as a standard deviation of species turnover (26, 27). To investigate any community changes that may occur below TECs and FECs, k-means partitioning (28) of the metal concentration data was carried out and the optimal number of groups of stations occurring in the contaminant variable was identified using the Calinski-Harabasz criterion (29) (software courtesy of Pierre Legendre, University of Montreal). The degree of community change between these groups was then calculated using the average Bray-Curtis dissimilarities. Functional Responses. The ability of a species to affect other species and components of the ecosystem can vary markedly depending on its size, mobility, and its functional role. Therefore, the taxa that contributed to differences in community composition along the contaminant gradient, as identified by SIMPER (30), were examined for their potential to play significant roles in ecosystem change, using size, mobility, and feeding type information. The relationship between rare taxa (defined as those occurring at fewer than VOL. 43, NO. 6, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2119
FIGURE 1. Predicted EC50s derived for each metal for sensitive taxa plotted against the cumulative percentage (out of a total of N ) 103 taxa). five stations or with a mean abundance of 36%) for all species with EC50s below the derived FEC. However, the R2 values were generally more variable for lead and lower for zinc. Community Changes. A strong gradient in community composition was observed across the contaminant variable (Figure 2), with a canonical correlation of 0.89. Most of the sites occurred below the present TEL, with a strong degree of change occurring. The derived FECs occupied a range occurring approximately midway along the contaminant variable gradient. Communities found at stations above the FECs were, on average, 79% dissimilar to those below the FECs. On DCCA axis 1, the centroids of stations above and below the FECs were separated by 0.92 units, indicating >60% turnover in species composition. The full DCCA gradient covered 1.9 standard deviations. Over one-third of the sites in the CAP analysis occurred below the positions of the derived FECs (Figure 2). K-means 2120
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 6, 2009
FIGURE 2. Comparison between changes in community composition (derived by CAP) along the contaminant variable, the presently used TEL and the derived FECs. The derived FEC positions are given as a range (with FEC1 based on N and FEC2 based on Nm). The position of the five groups, determined by k-means partitioning (G1-G5) are superimposed. partitioning of metal concentrations indicated that an optimal number of five groups of stations occurred along the contaminant gradient (Figure 2). These groups differed from one another in community composition by more than 70%, on average. In particular, the communities found in group 1 (mean 3.8 ( 1.7, 4.2 ( 1.0, 25.9 ( 4.0 mg · kg-1 for copper, lead, and zinc) were, on average, 74% dissimilar from those in group 2 (mean 4.6 ( 1.5, 11.5 ( 3.2, 42.7 ( 12.5 mg · kg-1 for copper, lead, and zinc), and the average dissimilarity between groups 1 and 5 (mean 34.0 ( 7.7, 51.6 ( 18.0, 222.0 ( 45.0 mg · kg-1 for copper, lead, and zinc) was 90%. On DCCA axis 1, the centroids of these groups were separated by 0.32 units, indicating ∼50% turnover in species composition. Group 3 (mean 9.4 ( 2.2, 18.4 ( 5.5, 79.3 ( 18.2 mg · kg-1 for copper, lead, and zinc) which contained the FECs and group
TABLE 1. Information about Taxa Contributing Most to the Difference between K-Means Groups 1 and 2, Together with Their Average Abundance Per Corea taxa Austrovenus stutchburyi
group 1 group 2 Decreased 15.5 7.5
Corophidae
6.7
0.2
Aonides trifidaG
3.0
2.0
Notoacmea helmsiG
2.2
1.4
Colurostylis lemurumG
3.3
0.8
OrbinidaeG
2.0
0.4
Macomona lilianaG
2.7
1.4
Magelona ?dakiniG
2.6
0.0
Anthopleura aureoradiataG
1.3
0.7
Paphies australisG
0.9
0.2
Glycera spp.
0.9
0.3
Nucula hartivigana
Increased 15.1 23.1
Heteromastus filiformis
2.5
3.9
Polydorids
0.6
4.6
Aricidea sp.
0.2
2.9
Arthitica bifurca
0.3
2.0
Nereididae
0.6
1.1
Cossura consimilis
0.1
0.5
25-50 mm suspension feeding bivalve, mobile 5-10 mm burrowing amphipod, highly mobile 10-30 mm infaunal deposit feeding polychaete, mobile 5-8 mm grazing limpet, mobile 5-10 mm burrowing cumacean, highly mobile 30-50 mm subsurface deposit feeding polychaete, limited mobility 30-60 mm deposit feeding bivalve, adults’ feeding tentacles are highly mobile 10-30 mm infaunal deposit feeding polychaete, limited mobility 5-8 mm anemone, commonly living on Austrovenus shells 60-90 mm suspension feeding bivalve, adults with limited mobility 30-50 mm predatory polychaete, mobile 5-8 mm deposit feeding bivalve, mobile 10-30 mm infaunal deposit feeding polychaete, limited mobility 10-30 mm tube-dwelling polychaete, deposit or suspension feeder, sedentary 10-20 mm infaunal deposit feeding polychaete, limited mobility 2-5 mm deposit feeding bivalve, sedentary 30-60 mm predatory polychaete, highly mobile 5-10 mm infaunal deposit feeding polychaete, sedentary
a G ) Taxa with EC50 values below the derived FEC guideline. Sizes relate to adult longest dimension.
4 (mean 20.8 ( 4.0, 28.5 ( 5.8, 120.2 ( 29.8 mg · kg-1 for copper, lead, and zinc) which contained the presently used TELs and had an average dissimilarity of 82%. Functional Responses. A negative relationship between the number of rare taxa and the contaminant variable was observed (p < 0.0001, r2 ) 0.26). A number of taxa showed an ∼50% or greater decrease in relative abundance between k-means groups 1 and 2 (Table 1) indicating high sensitivity to the metals. These taxa represented a variety of functions (suspension feeders, deposit feeders, predators, and large and mobile organisms).
Taxa showing increases between groups 1 and 2 were generally smaller in size and were predominantly polychaetes.
Discusssion Use of Field Surveys. This study observed a strong gradient of change in community composition along a contaminant gradient of copper, lead, and zinc, with communities near the ends of the gradient exhibiting ∼90% dissimilarity. These large changes to community composition occurred well below the Australian and New Zealand ANZECC (2000) ISQGLow guidelines and even below the TELs presently utilized in the Auckland area. Guidelines for the individual metals derived from the f-SSDs were also lower than the present TELs. There is increasing evidence that SQGs developed from reviews of laboratory and field dose-response experiments are higher than those derived from field surveys (15, 32). There are a number of reasons why this may occur. First, field surveys incorporate the simultaneous effects of multiple stressors (15), which can result in some organisms showing increased responses to metal concentrations (20). Depthdependent effects for copper and zinc toxicity have been reported (32) as have additive effects of cadmium, copper, and zinc (33). Second, field surveys take into account regional differences and variability (34-36) and the presence of different species, allowing biological interactions, differential susceptibility of life stages, and life cycle variability to affect responses (37). Site history and genetics can also affect species responses (38, 39), and correlations between contaminants and interactions between these and other stressors are inherently location-specific. Deriving guidelines from field surveys has some disadvantages. Determining a relationship between species abundances and a single contaminant requires the use of covariables to reduce variability, and even so, unexplained variability may remain high. As with any field study, predictions are limited to effects that occur within the gradient of contaminant already existing; effects can not be extrapolated to markedly higher contamination scenarios, or to changes in the ratios of different contaminants, except perhaps through the application of general ecological theory. Results are also not directly applicable to locations outside the study area, suggesting that the cost of location-specific studies needs to be examined against the increased confidence in results. Furthermore, while ecological confidence in the consequences of guidelines is increased, confidence in causality has been decreased (although there are a number of methods for increasing confidence (40) which this study employed: gradients, sampling over time, and analysis of other potential stressors). Use of Multivariate Models. Despite our FECs being field derived and location-specific, communities in two groups occurring below these thresholds were greater than 60% dissimilar. Multivariate models have frequently been used in the ecological literature to demonstrate impacts and define health (e.g., refs 4, 32, 41, but see also ref 42). The increased sensitivity displayed by the multivariate model is probably due to two major factors. First, multivariate models can include all taxa, even those too patchily distributed (both at the station and study area scale) to be included in the guideline development method (32). In our study, these rare, restricted-range, taxa occurred more frequently in less polluted stations. Second, multivariate models, unlike SSDs, explicitly include the effect of taxa increasing with increased metal concentrations. While species directly benefiting from increased heavy metal concentrations may be few, species may exhibit increases in abundances due to releases from competition and other interspecific interactions. Thus, preserving all of the information on the abundance of each taxon is generally expected to increase sensitivity of observed responses. Importantly, while community data is a summary VOL. 43, NO. 6, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2121
of all contaminants and disturbances present, the use of canonical ordination assures that the model is unaffected by unknown contaminants, unless their presence is highly correlated with the contaminants. If their presence is highly correlated, then the results integrate their effects as well. Recently, a method to develop SQGs directly from a multivariate model has been suggested (32), based on an adaptation of the doubling of standard errors as a “rule of thumb” for differences used in traditional statistical tests. This translates into a guideline of 4 times the background level of contaminants. However there are a number of problems with this guideline. Most importantly, the presence of sliding baselines around the world (43) mean that it is unlikely that we will know what background levels truly are. Second, the average concentration eliciting effects on metal was 3.6 times the background, and in silt guidelines were underprotective (32). Finally, such guidelines do not explore the ecological significance of change. Community effects on rare species and functional types, such as we observed, directly address present theories relating community change to stress and along gradients in disturbance. Rare species collectively dominate community structure and contribute substantively to traditional measures of structural biodiversity (e.g., 44). They are also important in maintaining the stability and resilience of ecosystem functioning, especially in changing environments (45-47). Rarity is often associated with habitat specificity, low environmental tolerance, or restricted dispersal ability (48, 49), characteristics that make such taxa vulnerable to human activities and habitat degradation (50-52). We observed decreased numbers of rare taxa with increased contamination suggesting that contamination is likely to affect resilience. When the contamination is diffuse and chronic, as with stormwater inputs, increased fragmentation of species and communities is likely, altering the meta community structure of the area and leading to ecological effects beyond the obvious change in composition (e.g., 53, 54). Further, ecological significance of the changes in community composition can be derived from investigating functional roles of sensitive organisms. We observed a decline in the number of large organisms with increasing contaminant levels. Large animals living in marine sediments can make a disproportionately large contribution to ecosystem functions such as deep-bioturbation of sediments, modification of benthic boundary layer flows, or through providing large packages of labile carbon (food) for fish and other larger consumers, see refs 55, 56, and references therein. Moreover, many of the affected taxa were highly mobile, increasing the potential effect on bioturbation rates and thus to oxygen and nutrient exchanges between the water column and the seafloor (57-59). In particular, Macomona and Austrovenus have both been recorded as influencing community composition and sediment-water fluxes (55, 60, 61). Numbers of suspension-feeding taxa also decreased, increasing the potential effect on benthic-pelagic coupling (e.g., ref 62, 63). Such changes are likely also to affect sediment burial and resuspension. Changes in these factors need to be considered when modeling the potential for contaminated sediments to be sequestrated and the likely effects on overall ecosystem health. A Way Forward. The degree to which contaminant guidelines are tested in law at present requires a strong ability to link cause and effect. However, we suggest that guidelines developed from individual species dose responses in experimental situations need to tested against f-SSDs and multivariate models in a range of locations and that such information is included in court cases. We feel that guidelines should be primarily location-specific, developed from f-SSDs, and associated with multivariate models of the degree of community change and an analysis of the species being lost 2122
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 6, 2009
and their associated functions. While this precludes a reliance on a single figure and requires a more risk-based analysis, the approach does allow a direct answer to the “so what” question and a realistic assessment of the environmental costs associated with accepting different guideline levels.
Acknowledgments We thank the Auckland Regional Council for providing the data and the funding for the multivariate model development. The New Zealand Foundation for Science and Technology (C01X0504) provided funding for development of the fSSD’s.
Supporting Information Available More information on the statistical methods, location of sampling sites and distribution of metal concentrations and sediment characteristics. This material is available free of charge via the Internet at http://pubs.acs.org.
Literature Cited (1) Chapman, P. M.; Mann, G. S. Sediment quality value (SQVs) and ecological risk assessment (ERA). Mar. Pollut. Bull. 1999, 38, 339–343. (2) Forbes, V. E.; Calow, P.; Sibly, R. M. Are current species extrapolation models a good basis for ecological risk assessment. Environ. Toxicol. Chem. 2001, 20, 442–447. (3) Hickey, C. W.; Golding, L. A. Response of macroinvertebrates to copper and zinc in a stream mesocosm. Environ. Toxicol. Chem. 2002, 21, 1854–1863. (4) Hewitt, J. E.; Anderson, M. J.; Thrush, S. F. Assessing and monitoring ecological community health in marine systems. Ecol. App. 2005, 15, 942–953. (5) Chapman, P. M. Current approaches to developing sediment quality criteria. Environ. Toxicol. Chem. 1989, 8, 589–599. (6) Calow, P.; Forbes, V. E. Does ecotoxicology inform ecological risk assessment. Environ. Sci. Technol. 2003, 37, 146A–151A. (7) Long, E. R.; Morgan, L. G. the Potential for Biological Effects of Sediment-Sorbed Contaminants Tested in the National Status and Trends Program; NOAA: Seattle, WA, 1990. (8) MacDonald, D. D.; Carr, R. S.; Calder, F. D.; Long, E. R.; Ingersoll, C. G. Development and evaluation of sediment quality guidelines for Florida coastal waters. Ecotoxicology 1996, 5, 253–278. (9) Long, E. R.; MacDonald, D. D.; Smith, S. L.; Calder, F. D. Incidence of adverse biological effects within ranges of chemical concentrations in marine and estuarine sediments. Environ. Manage. 1995, 19, 81–97. (10) Underwood, A. J.; Peterson, C. H. Towards an ecological framework for investigating pollution. Mar. Ecol.: Prog. Ser. 1988, 46, 227–234. (11) Adams, S. M. Assessing cause and effect of multiple stressors on marine systems. Mar. Pollut. Bull. 2005, 51, 649–657. (12) Peterson, C. H.; Black, R. Density-dependent mortality caused by physical stress interacting with biotic history. Am. Nat. 1988, 131, 257–270. (13) Munns, J. W. R. Assessing risks to wildlife populations from multiple stressors: overview of the problem and research needs. Ecol. Soc. 2006, 11, 23. (14) Coco, G.; Thrush, S. F.; Green, M. O.; Hewitt, J. E. The role of feedbacks between bivalve (Atrina zelandica) density, flow and suspended sediment concentration on patch stable states. Ecology 2006, 87, 2862–2870. (15) Leung, K. M. Y.; Bjørgesæter, A.; Gray, J. S.; Li, W. K.; Lui, G. C. S.; Wang, Y.; Lam, P. K. S. Deriving sediment quality guidelines from field-based species sensitivity distributions. Environ. Sci. Technol. 2005, 39, 5148–5156. (16) Reed, J.; Webster, K. Marine Sediment Monitoring Programmes2003 Results, TP 246; Auckland Regional Council: Auckland, 2004. (17) McHugh, M.; Reed, J. Marine Sediment Monitoring Programme: 2005 Results, TP 316; Auckland Regional Council: Auckland 2006. (18) Borja, A.; Dauer, D. Assessing the environmental quality status in estuarine and coastal systems: comparing methodologies and indices. Ecol. Ind. 2008, 8, 331–337. (19) Chao, A.; Chazdon, R. L.; Colwell, R. K.; Shen, T. A. statistical approach for assessing similarity of species composition with incidence and abundance data. Ecol. Lett. 2005, 8, 148–159.
(20) Thrush, S. F.; Hewitt, J. E.; Hickey, C. W.; Kelly, S. Multiple stressor effects identified from species abundance distributions: Interactions between urban contaminants and species habitat relationships. J. Expt. Mar. Biol. Ecol. 2008, 366, 160–168. (21) Grist, E. P. M.; Leung, K. M. Y.; Wheeler, J. R.; Crane, M. Better bootstrap estimation of hazardous concentration thresholds for aquatic assemblages. Environ. Toxicol. Chem./SETAC 2002, 21, 1515–1524. (22) Smith, E. P.; Robinson, T.; Jayfield, L.; Norton, S. B. Predicting sediment toxicity using logistic regression: a concentrationaddition approach. Environ. Toxicol. Chem. 2003, 22, 565–575. (23) Anderson, M. J.; Robinson, J. Generalized discriminant analysis based on distances. Aust. NZ J. Stat 2003, 45, 301–318. (24) Anderson, M. J.; Willis, T. J. Canonical analysis of principal coordinates: a useful method of constrained ordination for ecology. Ecology 2003, 84, 511–525. (25) Anderson, M. J.; Hewitt, J. E.; Ford, R. B.; Thrush, S. F. Regional Models of Benthic Ecosystem Health: Predicting Pollution Gradients from Biological Data, TP317; Auckland Regional Council: Auckland, 2006. (26) Hill, M. O.; Gauch, H. G. Detrended correspondence analysis, an improved ordination technique. Vegetatio 1980, 42, 47–58. (27) ter Braak, C. J. F.; Smilauer, P. CANOCO Release 4 Reference Manual and User’S Guide to Canoco for Windows-Software for Canonocal Community Ordination; Microcomputer Power: New York, 1998. (28) MacQueen, J. Some methods for classification and analysis of multivariate observations. In Proceedings of the 5th Berkeley Symposium on Mathematical Statistics and Probability; Le Cam, L. M., Neyman, J., Eds.; University of California Press: Berkeley, 1967; Vol. 1. (29) Calinski, T.; Harabasz, J. A dendrite method for cluster analysis. Commun. Stat. 1974, 3, 1–27. (30) Clarke, R. T.; Gorley, R. N. Primer v6; PrimerE: Plymouth, 2006. (31) ANZECC; ARMCANZ. Australian and New Zealand Guidelines for Fresh and Marine Water Quality; Australian and New Zealand Environment and Conservation Council and Agriculture and Resource Management Council of Australia and New Zealand: Canberra, 2000; Vol. 4. (32) Bjørgesæter, A.; Gray, J. S. Setting sediment quality guidelines: A simple yet effective method. Mar. Pollut. Bull. 2008, 57, 221– 235. (33) Kraak, M. H. S.; Lavy, D.; Schoon, H.; Toussaint, M.; Peeters, W.; Van Straalen, N. Ecotoxicity of mixtures of metals to the zebra mussel Dreissena polymorpha. Environ. Toxicol. Chem. 1994, 13, 109–114. (34) Wenning, R. J.; Ingersoll, C. G. Summary of the SETAC Pellston Workshop on Use of Sediment Quality Guidelines and Related Tools for the Assessment of Contaminated Sediments, 17-22 August; SETAC: Fairmont, 2002. (35) Long, E. R.; Field, L. J.; MacDonald, D. D. Predicting toxicity in marine sediments with numerical sediment guidelines. Environ. Toxicol. Chem./SETAC 1998, 17, 714–727. (36) Posthuma, L.; Suter, G. W., II; Traas, T. P. Species Sensitivity Distributions in Ecotoxicology; Lewis Publishers: Boca Raton, FL, 2002. (37) Calow, P. Handbook of Environmental Risk Assessment and Management; Blackwell: London, 1998. (38) Weisberg, S. B.; Ranasinghe, J. A.; Dauer, D. M.; Schaffner, L. C.; Diaz, R. J.; Frithsen, J. B. An estuarine benthic indiex of biotic integrity (B-IBI) for Chesapeake Bay. Estuaries 1997, 20, 149– 158. (39) Grosberg, P.; Cumming, C. W. Genetic structure in the sea: from populations to communities. In Marine Community Ecology; Bertness, M. D., Gaines, S. D., Hay, M., Eds.; Sinauer Associates Inc.: Sunderland, 2001. (40) Peters, R. H., A Critique of Ecology; Cambridge University Press: Cambridge, 1991. (41) Olsgard, F.; Gray, J. S. A comprehensive analysis of the effects of offshore oil and gas exploration and production on the benthic communities of the Norwegian continental shelf. Mar. Ecol.: Prog. Ser. 1995, 122, 277–306.
(42) Flaten, G. R.; Bolen, H.; Grung, B.; Kvalheim, O. M. Quantifying disturbances in benthic communities: comparison of the community disturbance index (CDI) to other multivariate methods. Ecol. Ind. 2007, 7, 254–276. (43) Dayton, P. K.; Tegner, M. J.; Edwards, P. B.; Riser, K. L. Sliding baselines, ghosts, and reduced expectations in kelp forest communities. Ecol. App. 1998, 8 (2), 309–322. (44) Harper, J. L.; Hawksworth, D. L. Biodiversity: measurement and estimation. Preface. Philos. Trans. R. Soc. 1994, 345, 5–12. (45) Loreau, M.; Naeem, S.; Inchausti, P.; Bengtsson, J.; Grime, J. P.; Hector, A.; Hooper, A.; Huston, D. U.; Raffaelli, D.; Schmid, B.; Tilman, D.; Wardle, D. A. Biodiversity and ecosystem functioning: current knowledge and future challenges. Science 2001, 294, 804–808. (46) Folke, C.; Holling, C. S.; Perrings, C. Biological diversity, ecosystems and the human scale. Ecol. App. 1996, 6, 1018– 1024. (47) Naeem, S.; Li, S. Biodiversity enhances ecosystem reliability. Nature 1997, 390, 507–509. (48) Gaston, K. J. Rarity; Chapman and Hall: London, 1994. (49) Gaston, K. J.; Blackburn, T. M.; Lawton, J. H. Interspecific abundance-range-size relationships: an appraisal of mechanisms. J. Anim. Ecol. 1997, 66, 579–601. (50) Thomas, C. D.; Mallorie, H. C. Rarity, species richness and conservation: Butterflies of the Atlas Mountains in Morocco. Biol. Conserv. 1985, 33, 95–117. (51) Gray, J. S. Marine biodiversity: patterns, threats and conservation needs. Biodiv. Conserv. 1997, 6, 153–175. (52) Jackson, J. B. C. What is natural in the coastal oceans. Proc. Natl. Acad. Sci. U. S. A. 2001, 98, 5411–5418. (53) Hanski, I. the Shrinking World: Ecological Consequences of Habitat Loss; International Ecology Institute: Oldendorf/Luhe, 2005. (54) Holyoak, M.; Leibold, M. A.; Mouquet, N.; Holt, R. D.; Hoopes, M. F. Metacommunities: A framework for large-scale community ecology. In Metacommunities: Spatial Dynamics and Ecological Communities; Holyoak, M., Leibold, M. A., Holt, R. D., Eds.; Chicago University Press: Chicago, 2005. (55) Thrush, S. F.; Hewitt, J. E.; Gibb, M.; Lundquist, C.; Norkko, A. Functional role of large organisms in intertidal communities: Community effects and ecosystem function. Ecosystems 2006, 9, 1029–1040. (56) Thrush, S. F.; Dayton, P. K. Disturbance to marine benthic habitats by trawling and dredging - Implications for marine biodiversity. Annu. Rev. Ecol. Syst. 2002, 33, 449–473. (57) Kristensen, E. Exchange o f. oxygen and inorganic nitrogen in a bioturbated estuarine sediment-water system. J. Coastal Res. 1985, 2, 14–23. (58) Lohrer, A. M.; Thrush, S. F.; Gibbs, M. M. Bioturbators enhance ecosystem performance via complex biogeochemical interactions. Nature 2004, 431, 1092–1095. (59) Bird, F. L.; Ford, P. W.; Hancock, G. J. Effects of burrowing macrobenthos on the flux of dissolved substances across the water-sediment interface. Mar. Freshwater Res. 1999, 50, 523– 532. (60) Thrush, S. F.; Hewitt, J. E.; Pridmore, R. D.; Cummings, V. J. Adult/juvenile interactions of infaunal bivalves: contrasting outcomes in different habitats. Mar. Ecol.: Prog. Ser. 1996, 132 (1-3), 83–92. (61) Whitlatch, R. B.; Hines, A. H.; Thrush, S. F.; Hewitt, J. E.; Cummings, V. J. Benthic faunal responses to variations in patch density and patch size of a suspension-feeding bivalve inhabiting a New Zealand intertidal sandflat. J. Exp. Mar. Biol. Ecol. 1997, 216, 171–190. (62) Newell, R. I. E. Ecosystem influences of natural and cultivated populations of suspension-feeding bivalve molluscs: A review. J. Shellfish Res. 2004, 23, 51–61. (63) Hewitt, J. E.; Thrush, S. F.; Gibbs, M.; Lohrer, A.; Norkko, A. Indirect effects of Atrina zelandica on water column nitrogen and oxygen fluxes: the role of benthic macrofauna and microphytes. J. Exp. Mar. Biol. Ecol. 2006, 330, 261–273.
ES802175K
VOL. 43, NO. 6, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2123
