i
CONTAMINATED ESTUARINE SEDIMENTS: A STATISTICAL EXAMINATION
OF SAMPLING DESIGN
by
Megan R. Schiller
Thesis submitted to Plymouth University
in partial fulfilment of the requirements for the degree of
MRes Marine Biology
Plymouth University
Faculty of Science & Technology
in collaboration with
Marine Biological Association of the United Kingdom, Plymouth, UK
September 2014
Journal Format, Marine Pollution Bulletin
ii
Copyright Statement
This copy of the thesis has been supplied on condition that anyone who consults it is
understood to recognise that its copyright rests with the author and that no quotation
from the thesis and no information derived from it may be published without the
author’s prior written consent.
iii
CONTAMINATED ESTUARINE SEDIMENTS: A STATISTICAL EXAMINATION
OF SAMPLING DESIGN
Megan R. Schiller∗,a,b
, Irene Kaimia
and Nick D. Popeb
a
Plymouth University, Portland Square, Drake Circus, Plymouth PL4 8AA, UK
b
Marine Biological Association of the United Kingdom, The Laboratory, Citadel Hill,
Plymouth PL1 2PB, UK
Abstract
The statistical reliability of sediment sampling methodology was evaluated using metal
concentrations measured in two estuaries of southwest England. The elements of
sampling design reviewed included sieving and spatial variability. Sixty-four samples
were collected from each site and processed by standard techniques. Five metals (Cu,
Zn, Mn, Pb, Zn) were analyzed using spectrophotometry. A significant pretreatment
effect and significant global and local spatial correlation were observed in the majority
of metal data from both sites. Three sampling strategies were compared using Monte
Carlo resampling that differed in size of area resampled and sample size. Four sample
sizes were applied in each strategy, and mean concentration and RSD distributions
served as the basis of comparison. Random sampling from equal sub-regions totaling
the original sampling area was found to be the most precise method, regardless of
sample size, and should be applied in the design of future related studies involving
contaminated sediments.
Keywords:
metals, sediments, sieving, variance, southwest England
Introduction
Sediments are a critical feature for assessing the environmental health of coastal
habitats (Birch et al., 2001; Murray, 1996; Bubb and Lester, 1994). It is widely
understood that the by-products of manufacturing, agriculture and waste management
inevitably make their way to the ocean, but it may be the transition path of these
materials that yields the greatest consequences for both humans and animals. While
certain areas of acute toxicity have been minimized in recent decades, the effects on
the environment of chronic exposure to toxic by-products demand attention. Estuaries
are a common deposition zone for adverse materials from both upstream, freshwater
sources and marine-based events. The intricate physical, chemical and ecological
processes that characterize estuaries have captured the attention of researchers for
∗
Corresponding author. Tel.:+44(0)7907 626941. Email addresses: Megan.Schiller@postgrad.plymouth.ac.uk (M.R.
Schiller), Irene.Kaimi@plymouth.ac.uk (I. Kaimi), NDPO@mba.ac.uk (N.D. Pope).
iv
decades, particularly in light of how these processes have been impacted by
anthropogenic activity over time (Sun et al., 2012). Growing degradation from pollution
and development that accompany population growth, along with climatic change
leading to sea level rise and erosion, will likely increase the significance of estuarine
sediments as a mechanism for monitoring coastal zones.
The term ‘contaminant’ is applied to any unnatural or undesirable entity that exists in a
given system. In studies related to contamination of estuaries and related inter-tidal
areas, this definition includes heavy metals, organic matter, nutrients, bacteria, and oil
(Naser, 2013; Hawkins et al., 2002). Estuarine sediments are both a sink for
contaminants from coastal input and atmospheric deposition as well as a source to the
water column through resuspension and desorption processes (Ozseker et al., 2013;
Sun et al., 2012; Cukrov et al., 2011; Langston et al., 2010). Contaminants can occupy
estuaries by point or diffuse accumulation and increase in concentration over time.
The persistence and legacy of contaminants in sediments have thus been featured as
primary concerns for future research (Rainbow et al., 2011; Langston et al., 2010;
Martins et al., 2010). How sediment-bound contaminants affect organisms is still not
fully understood, and may evolve as new synthetic materials with potentially synergistic
(or antagonistic) effects are introduced to coastal environments. Environmental
agencies such as the US Environmental Protection Agency (EPA), UK Environment
Agency (EA) and UN Environment Programme (EP), along with international
conventions such as MEDPOL, HELCOM and OSPAR, have been active in the
awareness and prevention or monitoring of estuarine contamination. However,
estuaries will continue to build long-term reservoirs of contaminants with the potential
to leach into surrounding biotic habitat.
The capacity for environmental managers to monitor contaminated estuaries depends
on access to sufficient and reliable data. Too often sediment sampling is performed
without a conceptually-sound statistical framework, but instead done on an ad hoc
basis related to the characteristics of an individual site (Caeiro et al., 2003), the
location of point-source contamination (Naser, 2013; US EPA, 2001), or is otherwise
limited by budgetary constraints. The spatial variability of contaminants that may result
from physical or biological processes in estuaries is not necessarily incorporated into
the sampling design. Without appropriate consideration of spatial irregularity, or ‘field
variance’ (Birch et al., 2001), the value of data resulting from unfounded sampling
designs becomes limited. In large-scale studies and regional monitoring plans
involving multiple sites, this variance can be additive (Birch et al., 2001; Kelly et al.,
1994).
v
Heavy metals and organic matter are primary foci of sediment contamination surveys,
and their spatial and temporal trends in estuaries have been observed in previous
studies (Cukrov et al., 2011; Birch et al., 2001; Morrisey et al., 1994a). A range of
physical (tides, floods, river discharge), chemical (pH, redox state) and biological
(bioturbation) factors can contribute to these trends. Sediment topography and salinity
are subject to these factors, which in turn affect the distribution of chemical constituents
and organisms, and thus spatial variability (Sun et al., 2012; Chapman and Wang,
2001). Rodrigo et al. (2013) sampled an impacted estuary in Portugal and observed
spatial correlation between polycyclic aromatic hydrocarbon (PAH) concentrations in
sediments and their proximity to industrial, mining and urban discharge points. In
coastal sediments of the Black Sea, Turkey, Ozseker et al. (2013) uncovered both
spatial and temporal trends between urban discharge, Cu contamination and particle
size. The results of this study were attributed to the tendency for heavy metals,
particularly cationic metal ions such as Cu, to preferentially bind to fine particles. Work
performed by Pope and Langston (2011) in the Thames Estuary, UK, confirmed
relationships between metals that show co-varying distribution (Ag, Cd, Cu, Hg, Pb, Zn)
and both particle size and abundance of metal-binding substrates. Organic ligands
and iron and manganese oxy-hydroxide coatings are known to chemically attract
metals, partially determining the distribution of adsorbed contaminants (Jickells and
Rae, 1997). Included in the results of this study were significant correlations between
the proportion of bioavailable (1M HCl-extractable) metals and the salinity gradient, as
concentrations of the metals increased upstream. An extensive body of literature
illustrates the complexity and significance of observed physical and chemical patterns
in sediment contamination.
The efficacy of sampling designs in terms of precision of resulting contamination data
must be evaluated relative to spatial distribution. Techniques for the collection,
storage, categorization, and chemical and granulometric analyses of sediments
contaminated by metals have been defined and are widely used (US EPA, 2001). A
comprehensive list of techniques can be found in the EA Index of Methods of the
Examination of Waters and Associated Materials (1976-2011). Of particular interest,
due to their prevalence across studies, are sieving, particle size analysis, microwave
acid digestion, and atomic absorption spectrophotometry. Standard pretreatment
measures such as sieving may influence the analysis of samples, but this effect is
seldom quantified. Based on the focus of a given study, variations on these techniques
have been employed in the past (Benomar et al., 2012; Braungardt et al., 2011) but
without consideration of potential field variance. Using certain methods in line with
vi
previous related work has the potential to allow neglected statistical bias to continue
infiltrating data until appropriately identified. However, performing a study in a manner
that suits a specific objective without regard for baseline spatial or temporal factors
may weaken its applicability. The design of estuarine contamination studies is often
constricted by either of these scenarios, and would benefit from an evaluation of basic
methodology.
Southwest England has a considerable history of ore deposit mining that continues to
be evident in the upper catchments of several major estuaries (Rainbow et al., 2011;
EA, 2008; Turner, 2000; Burt, 1998). Two estuaries have been sampled using a
spatially and statistically relevant grid design, as determined by documented metal
contamination of the chosen sites (Dr. Nick D. Pope, personal communication).
Samples have been divided into partial (‘sieved’) and total (‘unsieved’) sediment
treatment batches by wet sieving for examination of pretreatment effect, and particle
size analysis conducted to characterize the size distribution of sediments. Heavy metal
concentrations served as the unit of analysis in quantifying the degree of spatial
randomness present in each site. The objective of this analysis was to address the
basic assumptions underlying typical sampling designs. Three approaches to sampling
have been simulated from collected data and compared with the intent to propose a
generalized sampling approach grounded in statistical reliability that can be used in
future surveys. Metal contamination data has been widely referenced by
environmental management (Turner, 2000), and future policy measures will need to
reflect accurate information to respond effectively to critical and long-term
contamination.

Sample_MRSchiller

  • 1.
    i CONTAMINATED ESTUARINE SEDIMENTS:A STATISTICAL EXAMINATION OF SAMPLING DESIGN by Megan R. Schiller Thesis submitted to Plymouth University in partial fulfilment of the requirements for the degree of MRes Marine Biology Plymouth University Faculty of Science & Technology in collaboration with Marine Biological Association of the United Kingdom, Plymouth, UK September 2014 Journal Format, Marine Pollution Bulletin
  • 2.
    ii Copyright Statement This copyof the thesis has been supplied on condition that anyone who consults it is understood to recognise that its copyright rests with the author and that no quotation from the thesis and no information derived from it may be published without the author’s prior written consent.
  • 3.
    iii CONTAMINATED ESTUARINE SEDIMENTS:A STATISTICAL EXAMINATION OF SAMPLING DESIGN Megan R. Schiller∗,a,b , Irene Kaimia and Nick D. Popeb a Plymouth University, Portland Square, Drake Circus, Plymouth PL4 8AA, UK b Marine Biological Association of the United Kingdom, The Laboratory, Citadel Hill, Plymouth PL1 2PB, UK Abstract The statistical reliability of sediment sampling methodology was evaluated using metal concentrations measured in two estuaries of southwest England. The elements of sampling design reviewed included sieving and spatial variability. Sixty-four samples were collected from each site and processed by standard techniques. Five metals (Cu, Zn, Mn, Pb, Zn) were analyzed using spectrophotometry. A significant pretreatment effect and significant global and local spatial correlation were observed in the majority of metal data from both sites. Three sampling strategies were compared using Monte Carlo resampling that differed in size of area resampled and sample size. Four sample sizes were applied in each strategy, and mean concentration and RSD distributions served as the basis of comparison. Random sampling from equal sub-regions totaling the original sampling area was found to be the most precise method, regardless of sample size, and should be applied in the design of future related studies involving contaminated sediments. Keywords: metals, sediments, sieving, variance, southwest England Introduction Sediments are a critical feature for assessing the environmental health of coastal habitats (Birch et al., 2001; Murray, 1996; Bubb and Lester, 1994). It is widely understood that the by-products of manufacturing, agriculture and waste management inevitably make their way to the ocean, but it may be the transition path of these materials that yields the greatest consequences for both humans and animals. While certain areas of acute toxicity have been minimized in recent decades, the effects on the environment of chronic exposure to toxic by-products demand attention. Estuaries are a common deposition zone for adverse materials from both upstream, freshwater sources and marine-based events. The intricate physical, chemical and ecological processes that characterize estuaries have captured the attention of researchers for ∗ Corresponding author. Tel.:+44(0)7907 626941. Email addresses: Megan.Schiller@postgrad.plymouth.ac.uk (M.R. Schiller), Irene.Kaimi@plymouth.ac.uk (I. Kaimi), NDPO@mba.ac.uk (N.D. Pope).
  • 4.
    iv decades, particularly inlight of how these processes have been impacted by anthropogenic activity over time (Sun et al., 2012). Growing degradation from pollution and development that accompany population growth, along with climatic change leading to sea level rise and erosion, will likely increase the significance of estuarine sediments as a mechanism for monitoring coastal zones. The term ‘contaminant’ is applied to any unnatural or undesirable entity that exists in a given system. In studies related to contamination of estuaries and related inter-tidal areas, this definition includes heavy metals, organic matter, nutrients, bacteria, and oil (Naser, 2013; Hawkins et al., 2002). Estuarine sediments are both a sink for contaminants from coastal input and atmospheric deposition as well as a source to the water column through resuspension and desorption processes (Ozseker et al., 2013; Sun et al., 2012; Cukrov et al., 2011; Langston et al., 2010). Contaminants can occupy estuaries by point or diffuse accumulation and increase in concentration over time. The persistence and legacy of contaminants in sediments have thus been featured as primary concerns for future research (Rainbow et al., 2011; Langston et al., 2010; Martins et al., 2010). How sediment-bound contaminants affect organisms is still not fully understood, and may evolve as new synthetic materials with potentially synergistic (or antagonistic) effects are introduced to coastal environments. Environmental agencies such as the US Environmental Protection Agency (EPA), UK Environment Agency (EA) and UN Environment Programme (EP), along with international conventions such as MEDPOL, HELCOM and OSPAR, have been active in the awareness and prevention or monitoring of estuarine contamination. However, estuaries will continue to build long-term reservoirs of contaminants with the potential to leach into surrounding biotic habitat. The capacity for environmental managers to monitor contaminated estuaries depends on access to sufficient and reliable data. Too often sediment sampling is performed without a conceptually-sound statistical framework, but instead done on an ad hoc basis related to the characteristics of an individual site (Caeiro et al., 2003), the location of point-source contamination (Naser, 2013; US EPA, 2001), or is otherwise limited by budgetary constraints. The spatial variability of contaminants that may result from physical or biological processes in estuaries is not necessarily incorporated into the sampling design. Without appropriate consideration of spatial irregularity, or ‘field variance’ (Birch et al., 2001), the value of data resulting from unfounded sampling designs becomes limited. In large-scale studies and regional monitoring plans involving multiple sites, this variance can be additive (Birch et al., 2001; Kelly et al., 1994).
  • 5.
    v Heavy metals andorganic matter are primary foci of sediment contamination surveys, and their spatial and temporal trends in estuaries have been observed in previous studies (Cukrov et al., 2011; Birch et al., 2001; Morrisey et al., 1994a). A range of physical (tides, floods, river discharge), chemical (pH, redox state) and biological (bioturbation) factors can contribute to these trends. Sediment topography and salinity are subject to these factors, which in turn affect the distribution of chemical constituents and organisms, and thus spatial variability (Sun et al., 2012; Chapman and Wang, 2001). Rodrigo et al. (2013) sampled an impacted estuary in Portugal and observed spatial correlation between polycyclic aromatic hydrocarbon (PAH) concentrations in sediments and their proximity to industrial, mining and urban discharge points. In coastal sediments of the Black Sea, Turkey, Ozseker et al. (2013) uncovered both spatial and temporal trends between urban discharge, Cu contamination and particle size. The results of this study were attributed to the tendency for heavy metals, particularly cationic metal ions such as Cu, to preferentially bind to fine particles. Work performed by Pope and Langston (2011) in the Thames Estuary, UK, confirmed relationships between metals that show co-varying distribution (Ag, Cd, Cu, Hg, Pb, Zn) and both particle size and abundance of metal-binding substrates. Organic ligands and iron and manganese oxy-hydroxide coatings are known to chemically attract metals, partially determining the distribution of adsorbed contaminants (Jickells and Rae, 1997). Included in the results of this study were significant correlations between the proportion of bioavailable (1M HCl-extractable) metals and the salinity gradient, as concentrations of the metals increased upstream. An extensive body of literature illustrates the complexity and significance of observed physical and chemical patterns in sediment contamination. The efficacy of sampling designs in terms of precision of resulting contamination data must be evaluated relative to spatial distribution. Techniques for the collection, storage, categorization, and chemical and granulometric analyses of sediments contaminated by metals have been defined and are widely used (US EPA, 2001). A comprehensive list of techniques can be found in the EA Index of Methods of the Examination of Waters and Associated Materials (1976-2011). Of particular interest, due to their prevalence across studies, are sieving, particle size analysis, microwave acid digestion, and atomic absorption spectrophotometry. Standard pretreatment measures such as sieving may influence the analysis of samples, but this effect is seldom quantified. Based on the focus of a given study, variations on these techniques have been employed in the past (Benomar et al., 2012; Braungardt et al., 2011) but without consideration of potential field variance. Using certain methods in line with
  • 6.
    vi previous related workhas the potential to allow neglected statistical bias to continue infiltrating data until appropriately identified. However, performing a study in a manner that suits a specific objective without regard for baseline spatial or temporal factors may weaken its applicability. The design of estuarine contamination studies is often constricted by either of these scenarios, and would benefit from an evaluation of basic methodology. Southwest England has a considerable history of ore deposit mining that continues to be evident in the upper catchments of several major estuaries (Rainbow et al., 2011; EA, 2008; Turner, 2000; Burt, 1998). Two estuaries have been sampled using a spatially and statistically relevant grid design, as determined by documented metal contamination of the chosen sites (Dr. Nick D. Pope, personal communication). Samples have been divided into partial (‘sieved’) and total (‘unsieved’) sediment treatment batches by wet sieving for examination of pretreatment effect, and particle size analysis conducted to characterize the size distribution of sediments. Heavy metal concentrations served as the unit of analysis in quantifying the degree of spatial randomness present in each site. The objective of this analysis was to address the basic assumptions underlying typical sampling designs. Three approaches to sampling have been simulated from collected data and compared with the intent to propose a generalized sampling approach grounded in statistical reliability that can be used in future surveys. Metal contamination data has been widely referenced by environmental management (Turner, 2000), and future policy measures will need to reflect accurate information to respond effectively to critical and long-term contamination.