ArticlePDF Available

Abstract and Figures

1 Introduction 1.1 Background Sedimentation is a global issue where land-use change has resulted in excess sediment being delivered to and deposited on the beds of streams, rivers, estuaries and bays. Excess sediment directly affects the health of a waterway, decreasing its mauri or life-supporting capacity. Deposited fine sediment occurs naturally in the beds of rivers and streams. It usually enters a stream either because of terrestrial weathering processes, or bank erosion and in-stream fluvial processes. Sediment particles are transported and deposited in streams and receiving waters, such as lakes, estuaries and coastal bays, as the result of flowing water. Because sediment is naturally transported longitudinally through a river network, its state at any given point will be influenced by climate, geology, topography and current velocity. Human activities can impact on this natural sediment cycle by accelerating the delivery of sediment to streams and increasing the quantity of smaller particle sizes. The effect of excess in-stream sedimentation is recognised as a major impact of changing land use on river health. In particular, sediment alters the physical habitat by clogging interstitial spaces used as refugia by benthic invertebrates and fish, by altering food resources and by removing sites used for egg laying. As such, sediment can affect the diversity and composition of biotic communities. Excess sediment can also affect the aesthetic appeal of rivers and streams for human recreation. Although there is a general recognition of the significance of sedimentation in New Zealand, there are currently no widely accepted protocols for the measurement of deposited sediments, or guidelines to interpret the results in relation to ecological or recreational values. A number of regional councils have recognised the need to collect sediment information and have started to include some measure of deposited sediment in their monitoring programmes (Appendix 6.1). However, in the absence of established national guidance, different methodologies are currently being used. This lack of consistency could compromise the validity of any inter-region comparisons, or national state of the environment reporting. Furthermore, the absence of robust and tested methods may also compromise use of the data in any regulatory context (policy development, resource consents, prosecutions). The protocols and guidelines presented in this document were developed at the request of New Zealand Regional Councils to address a lack of national consistency. The aim of this document is to provide scientifically robust in-stream protocols and guidelines for the measurement of deposited sediment. The document also includes scientific justification and background information on the testing of these protocols.
Content may be subject to copyright.
Sediment Assessment Methods • Section 1. Introduction 1
Joanne Clapcott
Roger Young
Jon Harding
Christoph Matthaei
John Quinn
Russell Death
Sediment
Assessment
Methods
Protocols and guidelines for
assessing the effects of deposited
fine sediment on in-stream values
Joanne Clapcott
Roger Young
Jon Harding
Christoph Matthaei
John Quinn
Russell Death
Sediment
Assessment
Methods
Protocols and guidelines for assessing
the effects of deposited fine sediment
on in-stream values
Sediment Assessment Methods
Protocols and guidelines for assessing the eects of deposited ne sediment on in-stream values
Published by
Cawthron Institute
Private Bag 2
Nelson 7042
New Zealand
www.cawthron.org.nz
Citation
Clapcott, J.E., Young, R.G., Harding, J.S., Matthaei, C.D., Quinn, J.M. and Death, R.G. (2011) Sediment Assessment Methods: Protocols and
guidelines for assessing the eects of deposited ne sediment on in-stream values. Cawthron Institute, Nelson, New Zealand.
Copyright
© Cawthron Institute 2011. All rights reserved.
The information provided in this publication is for non-commercial reference purposes only. Whilst Cawthron Institute and the authors have
used all reasonable endeavours to ensure the information contained in this publication is accurate, no expressed or implied warranty is given
by Cawthron Institute (or the authors) as to the accuracy or completeness of the information. Neither Cawthron Institute nor the authors shall
be liable for any claim, loss, or damage suered or incurred in relation to, or as a result of, the use of the information contained within this
publication.
ISBN: 978-0-473-20105-0 (Print)
ISBN: 978-0-473-20106-7 (Online)
Cover images
Front cover - Wai-iti River (Kati Doehring); Okeover Stream (Jon Harding); Summer Warr trials the Shue method in the Porirua River (Juliet
Milne); Kati Doehring trials the in-stream visual assessment method in the Wai-iti River (Joanne Clapcott); Ashley River (Mary Beech).
Back cover – St Albans stream showing the eects of liquefaction following the Christchurch earthquake, February 2011 (Jon Harding); Shue
method in Heriot Burn (Justin Kitto); Ashley River (Mary Beech); St Albans stream (Jon Harding); Wairau River (Karen Shearer).
Graphic design
Downing Design, Nelson, New Zealand.
Acknowledgements
We are indebted to a large group of freshwater scientists and technicians who contributed to the development of protocols and guidelines
presented in this document. First and foremost, 12 regional councils eld tested protocols and provided valuable feedback and data, hence
we are very grateful to the following and their eld teams – Katrina Hansen (Northland Regional Council), Martin Neale (Auckland Regional
Council), Kevin Collier and Mark Hamer (Environment Waikato), Kimberley Hope and Chris Fowles (Taranaki Regional Council), Sandy Haidekker
(Hawkes Bay Regional Council), Kate McArthur and Carol Nicholson (Horizons Regional Council), Summer Warr and Alton Perrie (Greater
Wellingtin Regional Council), Fleur Tiernan (Marlborough District Council), Trevor James (Tasman District Council), Mary Beech (Environment
Canterbury), Justin Kitto and Rachel Ozanne (Otago Regional Council) and Kirsten Meijer (Environment Southland). Several of these people
also provided photos contained within. We also thank participants at workshops held at the Ministry for the Environment in Wellington 2009
and at the New Zealand Freshwater Sciences Society Conference in Whangarei in 2009 and in Christchurch in 2010. Adrian Meredith, Kate
McArthur, Graham-Sevicke Jones, Mary Beech, Juliet Milne, and Kirsten Meijer were members of the regional council steering committee
that developed the brief for the project. We thank Les Basher and Olivier Ausseil for their advice throughout. We thank Eric Goodwin for
contributing to the development of the boosted regression tree model and Kati Doehring for her assistance with spatial analysis. We thank
David Reid, Kati Doehring , Duncan Gray and postgraduate students of the School of Biological Sciences (University of Canterbury) for their
eldwork contribution. We thank postgraduate students at the University of Otago, NIWA laboratory sta and Environment Canterbury for
processing samples. We thank Annika Wagenho, Javad Ramezani and Logan Brown for providing additional data for guideline development
and we thank Summer Warr, Kate McArthur, Mary Beech and Oliver Ausseil for providing editorial feedback on earlier versions of this
document. The Sediment Assessment Methods were funded by the Envirolink Tools programme through the Ministry for Science and
Innovation. We thank the Ministry for the Environment for helping publish this document and we thank Cherie Johansson for proofreading
and arranging publication.
Sediment Assessment Methods • Section 1. Introduction 5
1. Introduction .............................................................................. 7
1.1 Background .........................................................................................................................................8
1.2 Scope ......................................................................................................................................................9
1.3 Dening deposited ne sediment ........................................................................................9
1.4 Sediment and in-stream values ...........................................................................................10
2. Sediment protocols ................................................................ 11
2.1 Guiding principles ........................................................................................................................12
2.1.1 Site selection ........................................................................................................................12
2.1.2 Sample collection .............................................................................................................12
2.2 Method selection and eld validation ............................................................................13
2.3 Recommended protocols .......................................................................................................14
SAM 1 - Bankside visual estimate of sediment ..........................................................15
SAM 2 – In-stream visual assessment of sediment .................................................17
SAM 3 – Wolman pebble count ..........................................................................................21
SAM 4 – Resuspendable sediment (Quorer method) ...........................................23
SAM 5 – Resuspendable sediment (Shue index) .................................................25
SAM 6 – Sediment depth ........................................................................................................27
3. Sediment guidelines .............................................................. 29
3.1 Guiding principles ........................................................................................................................30
3.1.1 Values-based assessment .............................................................................................30
3.1.2 Hard- versus soft-bottomed streams ....................................................................30
3.1.3 Accounting for temporal and spatial variability .............................................31
3.2 Determining sediment guideline values .......................................................................31
3.3 Recommended guidelines .....................................................................................................33
4. Supporting information ......................................................... 37
4.1 Review of sediment eects on biota and in-stream values ...............................38
4.1.1 Benthic invertebrates ......................................................................................................38
4.1.2 Fish..............................................................................................................................................44
4.1.3 Recreational and aesthetic values ..........................................................................46
4.2 Review of sediment assessment methods ...................................................................47
4.2.1 Percent cover of sediment ..........................................................................................48
4.2.2 Particle size distribution ................................................................................................50
4.2.3 Relative bed stability .......................................................................................................51
4.2.4 Embeddedness ...................................................................................................................52
4.2.5 Suspendible nes ..............................................................................................................52
4.2.6 Sediment depth .................................................................................................................54
4.2.7 Volume of nes in pools ................................................................................................55
Contents
Sediment Assessment Methods • Section 1. Introduction6
4.3 Protocol testing and validation ............................................................................................55
4.3.1 Do results vary for dierent habitats? ...................................................................56
4.3.2 Do results vary among dierent users? ...............................................................57
4.3.3 Do results vary in dierent land uses? ..................................................................59
4.3.4 How do results vary over time? ................................................................................60
4.3.5 How many replicates are required? .......................................................................62
4.3.6 How do results from dierent protocols compare? ....................................63
4.3.7 Are there cheaper, quicker methods? ..................................................................64
4.3.8 How well are sediment metrics related to in-stream biota? ..................66
4.3.9 Other useful things discovered along the way ..............................................67
4.4 Review of existing guidelines ...............................................................................................70
4.4.1 New Zealand ........................................................................................................................70
4.4.2 International .........................................................................................................................70
4.5 Guideline development ...........................................................................................................70
4.5.1 Sources of data ...................................................................................................................72
4.5.2 Correlation among sediment and biota .............................................................73
4.5.3 Predictive relationships between sediment and biota ..............................75
4.5.4 Boosted regression tree model to inform reference state .......................77
4.5.5 Data mining to inform reference state ................................................................82
4.5.6 Amenity values ...................................................................................................................84
4.5.7 Fish values ..............................................................................................................................88
5. References ............................................................................... 89
6. Appendices ............................................................................. 95
6.1 Survey of regional council objectives for sediment monitoring .............96
6.2 Survey of opinion on acceptable levels of sediment for
amenity values ................................................................................................................................98
6.3 Details of the boosted regression tree model used to predict variation
in ne sediment cover ............................................................................................................100
6.4 Volumetric Quorer method .........................................................................................104
Contents
Sediment Assessment Methods • Section 1. Introduction 7
Section 1
Introduction
Sediment Assessment Methods • Section 1. Introduction8
1 Introduction
1.1 Background
Sedimentation is a global issue where land-use change has resulted in excess sediment being delivered to and
deposited on the beds of streams, rivers, estuaries and bays. Excess sediment directly aects the health of a
waterway, decreasing its mauri or life-supporting capacity.
Deposited ne sediment occurs naturally in the beds of rivers and streams. It usually enters a stream
either because of terrestrial weathering processes, or bank erosion and in-stream uvial processes.
Sediment particles are transported and deposited in streams and receiving waters, such as lakes,
estuaries and coastal bays, as the result of owing water. Because sediment is naturally transported
longitudinally through a river network, its state at any given point will be inuenced by climate,
geology, topography and current velocity.
Human activities can impact on this natural sediment cycle by accelerating the delivery of sediment
to streams and increasing the quantity of smaller particle sizes. The eect of excess in-stream
sedimentation is recognised as a major impact of changing land use on river health. In particular,
sediment alters the physical habitat by clogging interstitial spaces used as refugia by benthic
invertebrates and sh, by altering food resources and by removing sites used for egg laying. As such,
sediment can aect the diversity and composition of biotic communities. Excess sediment can also
aect the aesthetic appeal of rivers and streams for human recreation.
Although there is a general recognition of the signicance of sedimentation in New Zealand, there are
currently no widely accepted protocols for the measurement of deposited sediments, or guidelines
to interpret the results in relation to ecological or recreational values. A number of regional councils
have recognised the need to collect sediment information and have started to include some measure
of deposited sediment in their monitoring programmes (Appendix 6.1). However, in the absence
of established national guidance, dierent methodologies are currently being used. This lack of
consistency could compromise the validity of any inter-region comparisons, or national state of the
environment reporting. Furthermore, the absence of robust and tested methods may also compromise
use of the data in any regulatory context (policy development, resource consents, prosecutions).
The protocols and guidelines presented in this document were developed at the request of New
Zealand Regional Councils to address a lack of national consistency. The aim of this document is to
provide scientically robust in-stream protocols and guidelines for the measurement of deposited
sediment. The document also includes scientic justication and background information on the
testing of these protocols.
This section contains an overview of the project which has been designed to develop protocols and
guidelines to assess the eects of sediment on in-stream values.
Deposited ne sediment is dened as inorganic particles deposited on the streambed that are less than
2 mm in size. ‘Sediment’ henceforth refers to deposited ne sediment, unless stated otherwise.
Sediment Assessment Methods • Section 1. Introduction 9
1.2 Scope
The key to establishing standardised protocols is identifying methodology that can be applied across a broad
range of conditions and yet be sensitive enough to distinguish change. Similarly, guidelines need to be
applicable to various river types present in New Zealand.
This document provides information on the development of protocols and guidelines for assessing
ne deposited sediment in wadeable rivers and streams in New Zealand. Recommended protocols
and related guidelines focus on providing a measure of sediment quantity that relates to specic
in-stream values.
In developing a series of protocols the following aspects are addressed:
Protocols cover both qualitative and quantitative measurements of deposited sediment.
Protocols are precise and directive enough to be undertaken by any reasonably experienced
freshwater scientist/technician. Level of skill, site selection, eld equipment, and eld and
laboratory procedures are described.
Protocols are scientically robust, repeatable, and relatively easy to use.
The key advantages and limitations of each protocol are outlined to help identify the protocol
best suited for the aim of the assessment.
Protocols do NOT address:
Suspended sediment (e.g., turbidity, clarity).
Sediment quality (e.g., associated contaminants, dissolved oxygen concentration,
decomposition potential).
Non-wadeable waterways.
Standing water bodies (have not been tested, but some protocols may be suitable for these
systems).
In developing a series of guidelines the following aspects are addressed:
Numerical guideline values are proposed for a range of waterway uses and values (e.g.,
biodiversity, sh habitat, and aesthetics). These have been based on current best estimates or
knowledge, which are provided, along with key limitations and needs for further research and
validation.
Numerical guideline values are a single value dening the threshold between an acceptable or
unacceptable state.
The applicability of numerical guidelines across a range of river types is described.
1.3 Dening deposited ne sediment
Sediment is the collective term for particles that are transported by natural processes (wind, water,
glaciers) and eventually deposited. In owing water, sediment can be dened by its composition,
locality and particle size. As such, sediment is organic or inorganic in nature and can be suspended
in the water column (causing turbidity) or deposited on the streambed. Using the Wentworth (1922)
classication system, sediment is characterised by particle size as mud and silt (<0.0625 mm) and sand
(0.0625-2 mm).
Sediment Assessment Methods • Section 1. Introduction10
During normal ow conditions, suspended sediment is dominated by particles less than 0.0625 mm
and can include colloids, clay, mud and silt. These smallest particles also form part of the deposited
sediment, and can be collectively referred to as suspendible sediments’. Larger particles deposited
on the streambed are collectively referred to as ‘bed load’. The movement of sediment is dependent
on channel morphology and ow. For example, higher water velocities are able to transport larger
particles.
In this document, deposited ne sediment refers to inorganic particles deposited on the streambed that are
less than 2 mm in size. ‘Sediment’ henceforth refers to deposited ne sediment, unless stated otherwise.
1.4 Sediment and in-stream values
Human activities, including urban development, agriculture and forestry, can accelerate the delivery
of sediment to streams or disrupt their natural downstream progression. When this occurs, resource
managers and stakeholders need to know to what degree this aects in-stream values and biota.
In 2009, a survey of regional councils in New Zealand (Appendix 6.1) identied what in-stream values
were perceived as being aected by sediment. These in-stream values were ranked in declining
importance from invertebrate community composition and abundance, native sh spawning/habitat,
biodiversity, sports sh habitat (i.e., trout and other salmonids), to aesthetics and swimming.
Other values identied include mahinga kai (food-gathering places), interstitial space and groundwater
connectivity, phosphorus levels, river function and habitat integrity, and E. coli.
The primary values identied by regional councils (invertebrates, sh and amenity) strongly correlate
to qualities well recognised as being signicantly aected by excess in-stream sediment
(Owens et al. 2005). As such there is considerable literature on anecdotal and quantitative relationships
between deposited sediment and in-stream values (Section 4.1). This information was used to assist in
the selection of protocols and development of guidelines in this document.
Sediment Assessment Methods • Section 2. Sediment protocols 11
Section 2
Sediment Protocols
Sediment Assessment Methods • Section 2. Sediment protocols12
2 Sediment Protocols
2.1 Guiding principles
2.1.1 Site selection
The protocols provided in this document are measures of sediment quantity for a single site; developed
to provide a representative measure of sediment at a reach scale, but focus specically in runs (dened
below). There will be error associated with extrapolating information collected from a single habitat to
a larger spatial scale and practitioners are directed to other resources to determine the appropriateness
of such extrapolation (see Downes et al. 2002; Harding et al. 2009).
Although the sediment protocols presented in this document were developed for specic stream
habitats and in-stream values, this does not exclude their application to other in-stream habitats and
this is noted when applicable.
Rationale to inform site selection:
Include single run habitat – runs are intermediary between ries and pools and therefore
provide an average measure for a stream reach (see Section 4.3.1)
Assess the full run – by walking the length of the habitat for bankside visual assessments
(light refraction o surface water can impede assessments from a stationary point); by
systematically sampling in an upstream direction for in-stream visual and other protocols
Restrict to the wetted stream width – assessment of the wetted channel provides less error
(see Section 4.2.2); most methods are also restricted to the wetted channel
Avoid runs with aquatic plants – macrophytes entrap sediment and can have high seasonal
variability; macrophytes and periphyton also make visual assessments dicult (see Section
4.3.9)
Replicate across several runs – where time and resources allow a more accurate and robust
measure of reach-scale sediment can be obtained by sampling three run habitats.
2.1.2 Sample collection
The mobility of sediments means that at any point in a river their quantity will vary naturally over time.
Fine sediment movement is inuenced by channel slope, channel roughness and ow (discharge
and velocity). The relationship between these stream properties has been used to calculate bed load
movement and sediment load budgets (Gordon et al. 2004). Given that channel slope and roughness
are relatively stable, the stream property that will most aect short to medium term (months to years)
variability in sediment is ow. Therefore sampling to measure changes in sediment should take into
This section outlines the guiding principles to applying sediment protocols, such as where,
when and how.
Overviews, eld procedures and useful images for training purposes are provided for six
recommended protocols.
A summary of ndings from a literature review, protocol testing and data evaluation is provided to inform
method selection – for a full description of methods, references and data summaries the reader is referred
to Section 4.
Sediment Assessment Methods • Section 2. Sediment protocols 13
2.2 Method selection and
eld validation
1
An assessment of sediment quantity requires knowledge of areal cover, substrate size and interstitial
space (Cantilli et al. 2006). These requirements were used to review and assess potential sediment
protocols. Following a literature review (Section 4.2), protocols for six methods were developed and
tested (Table 2-1).
Table 2-1. Description of sediment methods and metrics trialled as part of the protocol testing and
validation stage.
Sediment
Component
Method Metric Description
Sediment
cover
Bankside
visual
% sediment cover A bankside semi-quantitative measure
of the relative cover of ne sediment in
comparison to other substrate classes
In-stream
visual
% sediment cover An in-stream (using an underwater viewer)
semi-quantitative measure of the relative
cover of ne sediment in comparison to
other substrate classes
Substrate
size
Wolman
pebble
count
% sediment (“W2”),
d16, d50, d86
A quantitative measure of the percent of
ne sediment calculated from at least 100
random substrate measurements
Interstitial
space
Quorer SIS (mg/m
2
),
SOS(mg/m
2
), %SIS
A quantitative measure of the amount of
suspendible inorganic sediment (SIS) and
suspendible organic sediment (SOS) on
the streambed
Shue
index
Shue index score A qualitative rank (1-5) measure of the
degree of suspendible ne sediment on
the streambed
Sediment
depth
Depth (mm) A quantitative measure of the depth of ne
sediment in runs
consideration the eects of discharge and velocity. Unless sampling is designed to specically assess
the eects of a discharge event, sampling should occur at a relatively stable point in the hydrograph.
Rationale to inform the timing of sampling:
Low to median discharge conditions – ne sediment is suspended during high ow; visual
assessments are dicult during high ow; it is unsafe to enter a waterway during high ow.
Low to median velocities – in-stream assessments are impeded by high velocities.
1
This section provides a summary of ndings from a literature review, protocol testing and data evaluation – a full description of methods,
references and data summaries is provided in Section 4.
Sediment Assessment Methods • Section 2. Sediment protocols14
Protocol testing and validation involved a national-scale eort by 12 regional councils, Cawthron
Institute (Cawthron), National Institute for Water and Atmospheric Research (NIWA), University of
Canterbury and University of Otago over a period of six months and covering 264 river sites.
Results of the protocol testing and validation showed a high degree of consistency in the output
provided by the dierent methods (Section 4.3.6). Sediment depth was the only metric not correlated
with other measures of sediment. Results indicated that the bankside visual estimate of % sediment had
the strongest and most consistent relationship with biological indicators of in-stream values (Section
4.3.8). The bankside visual estimate of % sediment was also strongly correlated with the more labour
intensive in-stream visual estimate of % sediment (Section 4.3.6). The bankside method is likely to be a
suitable measure for broad-scale state of the environment assessments. The bankside method provides
a single numerical value, whereas the in-stream visual method includes multiple visual observations
and therefore would be more suitable when a measure of error/variability is needed (Table 2-2).
Substrate size composition using a Wolman pebble count provides an assessment of % ne particles
as well as other useful substrate composition data, for example, d50 (i.e., the median particle size)
(Table 2-2). The Quorer method provides a quantitative measure of sediment in the surface and
subsurface layers and as such could also be used to indicate the embeddedness’ of particles and
interstitial space. The Quorer method has several alternative measures that can be applied to assess
suspendible sediment (i.e., SIS, SOS, suspendible benthic sediment volume (SBSV), Section 4.3.7). Whilst
the Shue method was only weakly correlated with Quorer results, it does provide a rapid assessment
of suspendible sediment in relation to amenity values (Section 4.3.6; see also Section 4.5.8).
Whilst the bankside visual estimate of % sediment was most consistently related to invertebrate
metrics, all protocols trialled showed a signicant correlation (p < 0.01) with the macroinvertebrate
metrics of stream biotic health, the Macroinvertebrate Community Index (MCI) and/or the number
of taxa belonging to the sensitive insect families Ephemeroptera, Plecoptera and Trichoptera (EPT)
(Section 4.3.8).
Table 2-2. Recommended sediment protocols based on protocol testing and validation.
2.3 Recommended protocols
Type of
assessment
Sediment component
Sediment cover Substrate
composition
Interstitial space
State of the
Environment
Bankside visual estimate
of % sediment
Wolman pebble count Quorer SIS
or Quorer SBSV
or Shue
Assessment of
eects
In-stream visual
estimate of % sediment
Wolman pebble count Quorer SIS
Sediment depth (mm)
Sediment Assessment Methods • Section 2. Sediment protocols 15
Habitat Rie Run Pool (Comments)
Habitat length (m)
% sediment
Ratio sand:ner
(silt, clay, mud)
Photo (Yes/No)
Rationale Rapid qualitative assessment of the surface area of the streambed
covered by sediment.
Equipment required • Field sheet • Camera
Application All streams
Type of assessment State of the environment (broad-scale survey)
Time to complete 5 minutes
Description of variables
Habitat length (m)
% sediment
ratio of sand:ner
Estimation of habitat length in metres.
A visual estimation from the stream bank of the proportion of the
habitat covered by sediment (<2mm).
Provides a rough indication of the relative components of sand
versus mud and silt.
Useful hints Complete at start of site survey/sampling.
Note that this measure is also part of the Stream Habitat
Assessment Protocols P2c (i.e., an estimate of all substrate
size classes).
Sediment Assessment Method 1 - Bankside visual estimate of %
sediment cover
Field procedure
Estimate habitat length (m) and the percentage of streambed within the wetted width
covered by sediment <2 mm in size (0-100%) from the stream bank, for each rie, run,
pool present.
Record percentages (%) in the table below.
Take a representative photograph.
Sediment Assessment Methods • Section 2. Sediment protocols16 Sediment Assessment Methods • Section 2. Sediment protocols16
Useful images
Run, rie and pool habitat locations (Image courtesy of Cathy Kilroy – from Biggs et al. 2002).
Notes:
The average value for each habitat present weighted by length is used to calculate %
sediment at the reach scale
If all habitats are not present record % sediment for a run habitat only.
The assessment of all substrate size classes can be obtained at the same time, but it is
not necessary for the determination of % sediment cover.
The table below can be used to assess all substrate size classes.
Habitat Rie Run Pool (Comments)
Habitat length (m)
% mud/silt (<0.06 mm)
% sand (0.06-2 mm)
% ne gravel (2-16 mm)
% coarse gravel (16-64 mm)
% cobbles (64-256 mm)
% boulders (>256 mm)
% bedrock (layer of solid rock)
Sediment Assessment Methods • Section 2. Sediment protocols 17
Sediment Assessment Method 2 – In-stream visual estimate of %
sediment cover
Rationale Semi-quantitative assessment of the surface area of the
streambed covered by sediment. At least 20 readings are made
within a single habitat
Equipment required
• Underwater viewer - e.g., bathyscope
(www.absolutemarine.co.nz) or bucket with a Perspex bottom
marked with four quadrats • Field sheet
Application Hard-bottomed streams
Type of assessment Assessment of eects
Time to complete 30 minutes
Description of variables
% sediment A visual estimate of the proportion of the habitat covered by
deposited sediment (<2 mm)
Useful hints Work upstream to avoid disturbing the streambed
being assessed.
Mark a four-square grid on the viewer to help with estimates –
determine the nearest 5% cover for each quadrat.
Calculate the average of all quadrats as a continuous variable
following data entry.
More than ve transects may be necessary for narrow streams, to
ensure 20 locations are sampled.
Field procedure
Locate ve random transects along the run.
View the streambed at four randomly determined locations across each transect,
starting at the downstream transect.
Estimate the ne sediment cover in each quadrat of the underwater viewer in
increments (1, 5, 10, 15, 20 …100%).
Record results in the table below.
Repeat for four more transects so that 20 locations are sampled in total.
Note: Estimation of cover in each quadrat is important during training but may not be necessary
for experienced viewers – instead one measurement per location could be recorded.
Sediment Assessment Methods • Section 2. Sediment protocols18 Sediment Assessment Methods • Section 2. Sediment protocols18
% sediment Transect 1 Transect 2 Transect 3 Transect 4 Transect 5
Location 1
Q1 Q2
Q3 Q4
Location 2
Location 3
Location 4
Useful images
Digital examples of percent cover of sediment on the streambed as seen through an
underwater viewer.
1% 5% 10% 15% 20% 25% 30% 40% 50%
An example of viewer locations (x) for the in-stream visual assessment of sediment.
Sediment Assessment Methods • Section 2. Sediment protocols 19
1%
1%
Real examples of percent cover of sediment on the streambed as seen through an
underwater viewer.
5%
5%
10%
10%
15%
15%
Sediment Assessment Methods • Section 2. Sediment protocols20 Sediment Assessment Methods • Section 2. Sediment protocols20
25%
30%
40%
50%
90%
100%
20%
20%
Sediment Assessment Methods • Section 2. Sediment protocols 21
Sediment Assessment Method 3 – Wolman pebble count
Rationale Semi-quantitative assessment of the particle size distribution,
including ne sediment, on the streambed. At least 100 particle
measurements are made within a single habitat.
Equipment required
• Gravelometer (www.envco.co.nz) or a ruler marked with
a modied Wentworth scale (e.g., 2, 8, 16, 32, 64, 128, 256,
>256mm, bedrock) • Field sheet
Application Hard-bottomed streams
Type of assessment State of the environment (broad-scale survey)
Assessment of eects
Time to complete 20 minutes
Description of variables
Particle size class The length of the particle B-axis in millimetres.
Useful hints Avoid bias in foot placement or in particle selection, i.e., be
rigorous about selecting the particle in the middle of the front of
the boot at regular paces across the stream.
Assess any particles picked up – this should include silt/clay
particles on top of larger particles.
This measure is similar to that of the Stream Habitat Assessment
Protocols P3c.
Field procedure
Sample beginning at the downstream end of a run and proceed across and upstream.
Select particles at the front of your foot.
Select at least 100 particles within the wetted width of a run.
Use a gravelometer or a ruled rod, to measure the B-axis size class. The B-axis would
prevent a particle from passing through a gravelometer/sieve.
Record particle size classes (on a modied Wentworth scale) as tally marks in the
table below.
Note: Measurement of particle size is important during training but may not be
necessary for experienced eld sta – instead the descriptive table may be a
useful guide.
Sediment Assessment Methods • Section 2. Sediment protocols22 Sediment Assessment Methods • Section 2. Sediment protocols22
Particle size class Count Description
Clay/silt
(<0.06 mm)
Not gritty between ngers and hard to pick
up but visible as particles
Sand
(>0.06-2 mm)
Gritty between ngers
Smaller than a match head
Small gravel
(>2-8 mm)
Match head to little nger nail size
Small-Med Gravel
(>8-16 mm)
Little nger nail to thumb nail size
Med-Large Gravel
(>16-32 mm)
Thumb nail to golf ball size (or circle when
thumb and index nger meet)
Large Gravel
(>32-64 mm)
Golf ball to tennis ball size (or st)
Small Cobble
(>64-128 mm)
Tennis ball to softball size (or circle when
thumb and index ngers of two hands meet)
Large Cobble
(>128-256 mm)
Softball to basketball size
Boulders
(>256 mm)
Basketball or greater
Bedrock Continuous layer of solid rock
Useful images
B-axis of a pebble
“B” intermediate
axis (mm)
A longest axis (mm)
Sediment Assessment Methods • Section 2. Sediment protocols 23
Sediment Assessment Method 4 – Resuspendible sediment
(Quorer method)
Rationale Quantitative measure of total suspendible solids deposited on
the streambed. Six samples are collected from a single habitat.
Samples are processed in the laboratory for Total Inorganic/
Organic Sediment by area (SIS and SOS, respectively) or
Suspendible Benthic Solids by Volume (SBSV).
Equipment required • Cylindrical tube (e.g., 45 cm length of 35 cm diameter plumbing
tube for gravel bed streams, or 60 cm length of 50 cm diameter
metal tube for cobble bed streams) • 7 x >120 ml screw topped
sample bottles • Stirrer • Ruler (e.g., broom handle marked with
1 cm graduations) • Field sheet
Application Hard-bottomed streams
Type of assessment State of the environment (broad-scale survey)
Assessment of eects
Time to complete 30 minutes
Description of variables
Sample
Average water depth
(m)
Average stirred depth
(m)
Sample number
The average of ve water depths inside the cylinder in metres.
The average of ve water depths inside the cylinder in metres to
the depth that the sediments were stirred. Measured after water
sample collection.
Useful hints A split garden hose placed around the top of the tube aids with
the insertion into coarse substrates.
Welded handles at hand-height assist with use of large diameter
corers used in cobble bed rivers.
This method is not suitable for streambeds dominated by
large boulders.
Large cobbles can be removed from the corer prior to stirring.
Do not over-ll sample bottles because they expand when
frozen (samples should be frozen until analysis).
Field procedure
Collect a background water sample (i.e., control sample).
Insert an open-ended cylinder into the streambed in a run and measure water depth
at ve random locations within the cylinder. Record average water depth. Stir the upper
5-10 cm of sediment for 15 seconds.
Collect a sample of slurry (dirty water) and label.
Estimate average stirred depth (sediment + water).
Repeat Quorer method at ve more locations.
Freeze the six slurry samples and one background sample per site until
laboratory analysis.
Sediment Assessment Methods • Section 2. Sediment protocols24 Sediment Assessment Methods • Section 2. Sediment protocols24
Sample Average water depth (m) Average stirred depth (m)
Control na na
1
2
3
4
5
6
Notes
Suspendible inorganic sediment (SIS) and suspendible organic sediment (SOS) are
determined using the standard protocol for Total Suspended Solids (TSS method 2540D
in APHA 1998) and Volatile Suspended Solids (VSS method 2540E in APHA 1998).
o SIS (g/m
2
) = (TSS
(sample – control)
VSS
(sample – control)
) x average depth (m) in cylinder
o SOS (g/m
2
) = VSS
(sample – control)
x average depth (m) in cylinder
Stirred depth (m) is used to calculate SIS or SOS in g/m
3
.
Suspendible benthic sediment volume (SBSV) is determined using a settling assay
(See Appendix 6.4 for details).
The average value is calculated for each site.
Sediment Assessment Methods • Section 2. Sediment protocols 25
Sediment Assessment Method 5 - Resuspendible sediment
(Shue index)
Field procedure
Place a white tile on the streambed in a run, and measure/estimate water depth and
velocity at this point.
Stand 3 m upstream of the tile and disturb the streambed by moving feet vigorously for
ve seconds.
Allocate a score from 1-5 depending on the visibility and duration of the resulting
plume in relation to the white tile downstream.
Take a photo record of the plume where possible.
Repeat this process twice upstream.
Rationale Rapid qualitative assessment of the amount of total suspendible
solids deposited on the streambed. A score from 1-5 is assigned,
where 1 = little/no sediment and 5 = excessive sediment.
Equipment required • Camera • 10 cm x 10 cm white tile • Field sheet
Application All streams
Type of assessment State of the environment (broad-scale survey)
Assessment of eects (as support variable)
Time to complete 5 minutes
Description of variables
Water depth (m)
Water velocity
(fast/medium/slow)
Score
Photo
Depth of water in metres at tile location
Water velocity at tile location
A value of 1-5
Indication of whether a photo record was obtained
(preferably ‘Yes’)
Useful hints This method is best applied in an area where ow is between 0.2
0.6 m/sec and depth is between 20 and 50 cm.
Depth and velocity may be estimated and are mainly recorded to
ensure the method was applied in appropriate and comparable
conditions. Photos could be taken by a second team member on
the stream bank. Best completed at the end of sampling.
The average score is calculated for each site.
Sample Water Depth
(m)
Water velocity
(fast/medium/slow)
Score Photo
(yes/no)
1
2
3
Sediment Assessment Methods • Section 2. Sediment protocols26 Sediment Assessment Methods • Section 2. Sediment protocols26
Useful images
Resuspendible sediment index examples.
Score 1: No or small plume
Score 2: Plume briey reduces visibility at tile
Score 3: Plume partially obscures tile but quickly clears
Score 4: Plume partially to fully obscures tile but slowly clears
Score 5: Plume fully obscures tile and persists even after shuing ceases
Sediment Assessment Methods • Section 2. Sediment protocols 27
Sediment Assessment Method 6 –Sediment depth
Rationale Quantitative assessment of the depth of sediment in a run
habitat. At least 20 readings are made within a single habitat
Equipment required • Ruler or ruled rod • Field sheet
Application Hard-bottomed streams
Type of assessment Assessment of eects
Time to complete 30 minutes
Description of variables
Sediment depth (mm) A measure of the depth of sediment (mm).
Useful hints Determine the sampling grid rst to ensure an even cover of
edge and midstream locations.
Move upstream to avoid disturbing the streambed being
assessed.
Calculate the average depth for each site.
This method is usually only suitable when ne sediment is visible
from the stream bank.
Field procedure
Start downstream and randomly locate ve transects along the run.
Measure the sediment depth (mm) at four randomly determined locations across each
transect and record depth in the table below.
Depth (mm) Transect 1 Transect 2 Transect 3 Transect 4 Transect 5
Section 1
Section 2
Section 3
Section 4
Sediment Assessment Methods • Section 2. Sediment protocols28
Section 3
Sediment guidelines
30 Sediment Assessment Methods • Section 3. Sediment guidelines
3 Sediment Guidelines
This section outlines the guiding principles to applying sediment guidelines, such as their foundation on
values based assessments and their application in hard-bottomed streams during low ow.
Numerical guideline values are recommended for the protection of biodiversity, sh spawning habitat
and in-stream amenity values.
A summary of ndings from a literature review, a survey and data analysis to inform guideline
development is provided. Reference to scientic literature has been omitted for ease of reading – for more
detail readers are referred to Sections 4.1, 4.4 and 4.5.
3.1 Guiding principles
3.1.1 Values-based assessment
There is a common acceptance that excessive ne sediment deposited on stream and river beds can
adversely aect a number of environmental and community values, including, but not restricted to,
ecosystem health, amenity and recreational values. However, there is currently little guidance about
what constitutes acceptable and unacceptable levels of sediment in relation to the dierent in-stream
values.
The key question driving the development of these guidelines is:
What level of sedimentation corresponds to a signicant adverse eect on the dierent in-stream values?
The aim of these guidelines is to use the best current scientic information and knowledge available
nationally and internationally to answer this question in relation to three key in-stream values identied
by the regional councils (Section 1.4; see also Appendix 6.1):
macroinvertebrate communities health, as an indicator of overall aquatic ecosystem health
trout spawning
aesthetic, amenity and contact recreation values.
These guidelines are formulated as numerical thresholds, representing levels of in-stream
sedimentation beyond which specic in-stream values become impaired.
Rationale informing guideline development:
Guidelines relate to values – proposed guidelines provide a level of protection of the
identied primary in-stream values (invertebrates, sh, and aesthetics).
3.1.2 Hard- versus soft-bottomed streams
Whether a stream is naturally dominated by ne sediment is dependent on a number of factors
including stream size, slope, rainfall, catchment vegetation and geology. Streams naturally dominated
by sediment are usually very small streams with low slopes and low rainfall on sandy soils. Such soft-
bottomed’ streams currently account for approximately 20% of the length of rivers in New Zealand,
according to Freshwater Ecosystems of New Zealand (FENZ) classication (Leathwick et al. 2011).
Whereas, predictions from GIS models suggest less than 2% of all NZ streams would have greater
than 50% ne sediment cover in the absence of human land-use activities (Section 4.5.4). Together
these analyses indicate that the majority of streams in New Zealand are, or should be ‘hard-bottomed’,
31Sediment Assessment Methods • Section 3. Sediment guidelines
dominated by relatively coarse (gravel or larger) substrate.
During the protocol development stage sediment depth was trialled as a potential metric to assess
naturally soft-bottomed streams. However, sediment depth was poorly related to invertebrate biotic
metrics, possibly because many of the indicators used to assess stream condition in New Zealand are
developed primarily for application in hard-bottomed streams, for example %EPT. Sediment depth data
was also not as abundant as other protocol data.
Thus these guidelines focus on hard-bottomed streams and assume that an increase in sediment is
detrimental to fauna and ora naturally occurring in hard-bottomed streams. However, some protocols
reviewed in Section 4.2 may be applicable for assessing sediment accumulation in soft-bottom streams
(e.g., volume of sediment in pools). Guideline values are not provided for these untested methods.
Rationale informing guideline development:
Hard-bottomed streams – majority of waterways in New Zealand are or should be
dominated by coarse substrate.
3.1.3 Accounting for temporal and spatial variability
New Zealand is a geologically young and tectonically active country subject to strong erosive elements
(wind, rain) and land forming processes (tectonic uplift, volcanism and earthquakes). New Zealand
streams can be subject to high sediment loads on a continual or episodic nature and some river
systems have among the highest sediment bed loads recorded globally (Hicks et al. 2000). Furthermore,
human land-use activities can accelerate in-stream sediment delivery and alter downstream
transportation. For example, a storm event can lead to land slumping that delivers sediment to a
stream; the degree of slumping can be amplied due to vegetation clearance for agriculture, while
water abstraction can reduce the power of a stream to redistribute sediment.
The above example illustrates the need to consider temporal and spatial variability in sediment
distribution and accumulation when applying sediment guidelines. Therefore, it is important to
determine excess sediment in relation to what would occur naturally, i.e., in respect to a reference
condition. It is also important to be wary of undertaking sediment assessments at times of active
sediment movement, for example, immediately after periods of high ow.
Rationale informing guideline development:
Comparison to reference – New Zealand streams vary a lot over space and time.
While an upper limit may be applicable to protect certain values in all waterways, the degree of
departure from a reference state will provide a more sensitive assessment of sediment impact.
3.2 Determining sediment
guideline values
2
An evidence-based approach was used to develop guidelines for sediment quantity, based on reported
relationships and available data. This is sometimes referred to as ‘weight-of-evidence’ or consensus-
based’ approach and is widely used to dene guideline values and inform decision-making processes in
regards to sediment quality (e.g., MacDonald et al. 2000; Burton et al. 2002).
The methods used to determine sediment guidelines included:
2
This section provides a summary of ndings from a literature review, survey and data analysis to inform guideline development. Reference to
scientic literature has been omitted for ease of reading. For more detail readers should refer to sections 4.4 and 4.5.
32 Sediment Assessment Methods • Section 3. Sediment guidelines
a review of existing guidelines (Section 4.4)
a review of quantitative relationships between sediment and in-stream values
(Sections 4.1 and 4.5.7)
correlative analyses among sediment metrics (Section 4.5.2)
linear regression analyses among sediment metrics and biotic variables (Section 4.3.8)
data mining to inform reference state (Section 4.5.5)
boosted regression tree model to inform reference state (Section 4.5.4)
survey of amenity values (Section 4.5.6).
A review of existing sediment guidelines for waterways identied a wide range in sediment
criteria and standards because of a wide range in denitions of deposited ne sediment (i.e., anywhere
from <0.85 mm to <6.4 mm in size) and methods used to measure sediment (e.g., Wolman pebble
counts, embeddedness). Also, sediment guidelines have been developed to protect a range of values.
Generally, sediment guidelines include an absolute upper limit and a target deviation from reference.
In North America, upper limits range from less than 3% to less than 30% sediment with less than 5% to
less than 27% recommended deviation from reference.
Environment Canterbury Regional Council is the only New Zealand authority to currently include
sediment guidelines in regional planning, recommending between 10% and 40% absolute sediment
cover, depending on the management purposes dened in each water quality management unit .
A review of quantitative relationships between sediment and in-stream values in New Zealand
showed that sediment directly aects invertebrate community composition, EPT taxa richness
and abundance, specic taxa density and invertebrate drift. Anywhere between 10% and 10-fold
increases in sediment resulted in noticeable invertebrate responses, with changes amplied over time.
Few quantitative relationships have been observed between native sh and deposited sediment.
International literature suggests ideal sport sh habitat (i.e., salmonids) has less than 10% sediment, but
greater than 20% sediment will result in sh egg mortality.
Correlative analyses among sediment metrics showed that all visual estimates of sediment
cover were strongly correlated, i.e., in-stream visual, and bankside visual at a reach or run scale. This
suggests that guidelines developed for sediment cover can be assessed using any visual assessment
method. Quorer metrics were related to visual estimates of sediment cover at a run scale, but not at a
reach scale. All other sediment metrics were related to each other except % sediment calculated from
Wolman pebble counts and sediment depth. Results demonstrated the interdependence of sediment
components, i.e., cover, substrate size and suspendible sediment.
Linear regression analyses among sediment metrics and biotic variables showed few predictive
relationships and a wide range in biotic values at low values of sediment. A linear relationship between
bankside visual % sediment and MCI suggested a negative value of sediment at 120 MCI (i.e., MCI value
indicative of good health). A negative linear relationship between bankside visual % sediment and
%EPT richness suggested a value of 7% sediment at 50% EPT (i.e., EPT value potentially indicative of
good health). A negative linear relationship between SIS (log-transformed) and MCI suggested a value
of 22 g/m
2
at 120 MCI. A negative linear relationship between sediment depth (log-transformed) and
the total number of invertebrate taxa and EPT richness was also observed.
Data mining to inform reference state involved viewing the distribution of sediment data to
determine the 75th percentile for sites with greater than 80% native vegetation in their catchments,
and the 75th percentile for sites with greater than 120 MCI and greater than 50% EPT. These approaches
resulted in a similar value for each of the sediment metrics (Table 3-1).
33Sediment Assessment Methods • Section 3. Sediment guidelines
Table 3-1. Sediment reference values derived from two approaches to examine the distribution of
sediment data collated to develop sediment guidelines.
Summary of 75
th
percentile values
>80% native
vegetation >120 MCI >50% EPT
% sediment (bankside visual reach scale) 15 20 20
% sediment (bankside visual run scale) 20 20 20
% sediment (in-stream visual) 17 13 17
% sediment (Wolman pebble count) 17 8 20
SIS (g/m
2
) 405 429 953
SOS (g/m
2
) 28 43 69
%SIS 91 94 94
Shue index score 2 2 3
Sediment depth (mm) 9 4 63
SIS=Suspendible inorganic sediment, SOS – suspendible organic sediment, %SIS=percentage of
suspendible sediment that is inorganic
A boosted regression tree model to inform reference state was used to make national predictions
of sediment cover in the absence of land-use impacts. Percent sediment data from the New Zealand
Freshwater Fish Database (NZFFD) was used along with land-use and environmental descriptors for
each stream reach from FENZ. The model predicted a current national average of 29% sediment cover,
but when the inuence of land-use was factored out, the model predicted a national average of only
8% sediment cover. A range in predicted ‘reference sediment conditions was evident for dierent
stream types as classied by FENZ 20-level stream types, for example, 29.4% sediment for Group B (i.e.,
small warm coastal streams) down to 2.4% sediment for Group S (i.e., cold steep mountainous streams).
Finally, a survey of amenity values was conducted to inform the level of sediment acceptable
for swimming and recreation. Results suggested that amenity value changes from acceptable
to unacceptable between 12% and 27.5% sediment cover and swimming value decreases from
acceptable to unacceptable between a Shue index score of 2 and a Shue index score of 3.
Information from all of the above approaches was used to inform recommended guidelines. For
biodiversity values, weight was given to the results of data mining to inform absolute limits because
the results of the regression analyses were weak. The national sediment model provides guidance for
assessing deviation from predicted reference. For salmonid values, weight was given to relationships
reported in the literature. For amenity values, weight was given to results of the user survey.
3.3 Recommended Guidelines
The following are recommended guidelines for assessing the eects of deposited ne sediment on the
in-stream values of hard-bottomed streams.
34 Sediment Assessment Methods • Section 3. Sediment guidelines
Sediment
measure
Sediment
value
Core method Supporting
data
Application
Sediment cover
(%)
< 20% OR within
10% cover of
reference
Bankside visual
estimate
Photo State of the
environment
reporting
< 20% OR within
10% cover of
reference
In-stream visual
estimate
Photo Assessment of
eects
Substrate size
(%)
< 20% OR within
10% cover of
reference
Wolman pebble
count
State of the
environment
reporting OR
Assessment of
eects
Suspendible
sediment
< 450 g/m
2
Quorer (SIS) State of the
environment
reporting OR
Assessment of
eects
In-stream value=Biodiversity*
[* includes native sh on the assumption that benthic invertebrates are their primary food source]
In-stream value=Salmonid spawning habitat
Sediment
measure
Sediment
value
Core method Supporting
data
Application
Sediment cover
(%)
< 20% OR within
10% cover of
reference
Bankside visual
estimate
Photo State of the
environment
reporting
< 20% OR within
10% cover of
reference
In-stream visual
estimate
Photo Assessment of
eects
Substrate size
(%)
< 20% Wolman pebble
count
State of the
environment
reporting OR
Assessment of
eects
35Sediment Assessment Methods • Section 3. Sediment guidelines
In-stream value=Amenity
Sediment
measure
Sediment
value
Core method Supporting
data
Application
Sediment cover
(%)
< 25% Bankside visual
estimate
Photo State of the
environment
reporting
< 25% In-stream visual
estimate
Photo Assessment of
eects
Suspendible
sediment
< 3 Shue index Photo State of the
environment
reporting
The following guidelines are recommended; that sediment should not exceed either:
1) 20% cover or 450 g/m
2
(SIS) to protect stream biodiversity and sh (native and trout) habitat.
2) 25% cover or Shue index score of 3 to protect stream amenity.
We recommend that these numerical guidelines provide upper limits on the amount of ne sediment
that will aect in-stream values, i.e., any amount of sediment greater than 20% cover will detrimentally
aect biodiversity and sh habitat. Note that there are likely to be lower limits at which in-stream value
levels will be negatively aected by sediment. The available data makes it dicult to locate those limits,
so for this reason it is recommended that a comparison of sediment values with a reference condition
is applied.
36 Sediment Assessment Methods • Section 3. Sediment guidelines
Section 4
Supporting
Information
38 Sediment Assessment Methods • Section 4. Supporting information
4 Supporting Information
This section contains detailed information used to support the development of protocols and guidelines
to assess the eects of sediment on in-stream values.
Included are literature reviews on sediment eects on biota and sediment assessment methods, and a
review of existing sediment guidelines.
Comprehensive details on protocol testing and validation, guideline development including the
prediction of reference values, and other useful things we learnt along the way are also included.
4.1 Review of sediment eects on
biota and in-stream values
4.1.1 Benthic invertebrates
The most commonly inferred causal pathway for invertebrate response to sediment is a change in
habitat. By denition, benthic invertebrates live on or in the streambed and hence any change to this
habitat will directly aect the invertebrate community. However, there is a wide range of responses of
benthic invertebrates to increased sediment including changes in invertebrate feeding and growth,
behaviour, community composition, diversity and abundance (Ryan 1991; Waters 1995; Wood &
Armitage 1997; Crowe & Hay 2004).
Invertebrate feeding can be directly aected by clogging of feeding apparatus (i.e., impeded lter-
feeding) and by loss of suitable habitat for attachment or feeding (Ryan 1991). Indirect eects on
invertebrate feeding may also occur, via changes in food source and nutritional content as well as the
adherence of toxicants to sediment (Ryder 1989; Collier 2002).
Sediment deposition can alter invertebrate behaviour. Interstitial spaces between substrata are used
by invertebrates to avoid predators and the scouring eects of high ow (Sedell et al. 1990). Increased
sediment deposition can lead to the short-term increase in invertebrate drift (Larsen & Omerod
2009; Molinos & Donohue 2009), and in the long term, invertebrate recolonisation through upstream
movement may also be disrupted by large-scale ne sediment accumulation (Luedtke & Brusven 1976).
Ultimately the clogging of both surface and subsurface habitats by sediment leads to changes in
invertebrate density and community composition (Waters 1995; Matthaei et al. 2006). As the level of
sediment increases, taxa that favour stony habitat such as EPT taxa (Ephemeroptera, Plecoptera and
Trichoptera) are replaced by burrowing taxa such as chironomids and worms (Wood & Armitage 1997;
Rabeni et al. 2005; Townsend et al. 2008).
In New Zealand, there is more information available on the quantitative relationships between
sediment and benthic invertebrates than for other in-stream values (Table 4-1).
39Sediment Assessment Methods • Section 4. Supporting information
Table 4-1 Quantitative relationships that have been documented between proportions of sediment and
invertebrate populations in New Zealand (adapted from Crowe & Hay 2004).
Taxon /
Community
descriptor
Experimental
method
Sediment
size
Quantitative
relationship
established
Source
Deleatidium Introduced
substrates
in relatively
sediment-free
stream
0.5-2 mm 12-17% increase
in interstitial ne
sediments resulted
in a 27-55% decrease
in abundance
Ryder
(1989)
Deleatidium,
hydrobiosid
caddisies
Sediment
additions into
stream section
0.125-1 mm Abundances
decreased as
amount of ne
sediment increased
Ryder
(1989)
Elmidae,
Oligochaeta,
Potamopyrgus
antipodarum
Sediment
additions into
stream section
0.125-1 mm No signicant
change in
abundance
Ryder
(1989)
Trichoptera,
Chironomidae
Introduced
substrates
in relatively
sediment-free
stream
0.5-2 mm Generally more
common on
substrates without
interstitial ne
sediments
Ryder
(1989)
Pycnocentrodes,
Austrosimulium, P.
antipodarum
Introduced
substrates
in relatively
sediment-free
stream
0.5-2 mm Abundance
not aected by
increased interstitial
ne sediments
Ryder
(1989)
Elmidae Introduced
substrates
in relatively
sediment-free
stream
0.5-2 mm Abundance
increased as amount
of interstitial ne
sediments increased
Ryder
(1989)
Total invertebrate
abundance
Introduced
substrates
in relatively
sediment-free
stream
0.5-2 mm 12-17% increase
in interstitial ne
sediments resulted
in a 16-40% decrease
in abundance
Ryder
(1989)
40 Sediment Assessment Methods • Section 4. Supporting information
Taxon /
Community
descriptor
Experimental
method
Sediment
size
Quantitative
relationship
established
Source
Total invertebrate
density, biomass,
taxa richness
Survey of 88 NZ
rivers
Silt <0.063
mm, sand
0.063-2
mm.
Decreased
invertebrate
density, biomass
and taxa richness
in rivers with high
proportions of
silt and sand in
surface sediments,
c.f. communities
in rivers with
coarser substrate
compositions
Quinn &
Hickey
(1990)
Total invertebrate
abundance, taxa
richness
Sediment
additions into
stream sections
<4 mm Following 21
days exposure,
total invertebrate
abundance and
taxa richness
had decreased
signicantly (c.f.
controls). Mean
total number
of individuals
decreased by 40-
55%, and mean taxa
richness decreased
by 15-30%
Dunning
(1998)
Helicopsyche,
Zephlebia
Sediment
additions into
stream sections
<4 mm Following 21
days exposure,
abundance of
Zephlebia and
Helicopsyche
had decreased
signicantly (c.f.
controls)
Dunning
(1998)
41Sediment Assessment Methods • Section 4. Supporting information
Taxon /
Community
descriptor
Experimental
method
Sediment
size
Quantitative
relationship
established
Source
Diptera Sediment
additions into
stream sections
<4 mm Following 21
days exposure,
abundances had
increased (c.f.
controls), but
not statistically
signicant
Dunning
(1998)
Potamopyrgus
antipodarum,
Elmidae
Sediment
additions into
stream sections
<4 mm Following 21
days exposure,
abundances had
not changed
signicantly (c.f.
controls) for either
Dunning
(1998)
% EPT taxa, QMCI,
MCI
Sediment
additions into
stream sections
<4 mm Following 21 days
exposure, %EPT
taxa and QMCI
had decreased
signicantly (c.f.
controls), whereas
MCI showed no
signicant change
Dunning
(1998)
Deleatidium drift Sediment
additions to
articial channels
containing
cobble substrate
and established
algae and
invertebrate
communities.
Deleatidium
added to
channels after
sediment
additions.
<2 mm 16% increase in
interstitial ne
sediment resulted in
a 80% mean increase
in numbers of
drifting Deleatidium
Suren &
Jowett
(2001)
42 Sediment Assessment Methods • Section 4. Supporting information
Taxon /
Community
descriptor
Experimental
method
Sediment
size
Quantitative
relationship
established
Source
Paracalliope
uviatilis, Oxyethira
albiceps, Hydrobiosis
sp. and chironomid
larvae drift
Sediment
additions to
articial channels
containing
cobble substrate
and established
algae and
invertebrate
communities
<2 mm 16% increase in
interstitial ne
sediment resulted in
a doubling of drift
rates
After 3 days,
abundances
of chironomid,
Oxyethira and
Hydrobiosis larvae
were signicantly
lower in sedimented
channels
Suren &
Jowett
(2001)
Chironomid
emergence, diurnal
drift patterns
Sediment
additions to
articial channels
containing
cobble substrate
and established
algae and
invertebrate
communities
<2 mm 16% increase in
interstitial ne
sediment had no
signicant eect
on chironomid
emergence or
diurnal drift patterns
Suren &
Jowett
(2001)
Ephemeroptera,
Trichoptera
Longitudinal
and temporal
sampling of
anthropogenic
point-source
inputs of ne
sediment to a
river
‘Sand’ A c. 10-fold increase
in percentage cover
by sand (c. 5%
cover at upstream
control vs. 50-54%
at downstream
sites), resulted in a
30-75% reduction in
Ephemeroptera and
a 70-80% reduction
in Trichoptera
Cottam
& James
(2003)
43Sediment Assessment Methods • Section 4. Supporting information
Taxon /
Community
descriptor
Experimental
method
Sediment
size
Quantitative
relationship
established
Source
Diptera, Oligochaeta Longitudinal
and temporal
sampling of
anthropogenic
point-source
inputs of ne
sediment to a
river
‘Sand’ A c. 10-fold increase
in percentage cover
by sand resulted
in a 0.5 to 2.4-fold
increase in Diptera
and a 1 to 8-fold
increase Oligochaeta
Cottam
& James
(2003)
Taxa richness, EPT
richness
Longitudinal
and temporal
sampling of
anthropogenic
point-source
inputs of ne
sediment to a
river
‘Sand’ A c. 10-fold increase
in percentage cover
by sand resulted in
a 40-50% reduction
in median taxa
richness, and a 25-
50% reduction in
median EPT richness
Cottam
& James
(2003)
Potamopyrgus
antipodarum &
Deleatidium
Laboratory
preference
trials using
cobbles subject
to diering
sediment and
algae treatments
<0.5 mm Both species
preferred a sediment
contaminated
version of their
respective food
source over
alternative alga
Suren
(2005)
Invertebrate density,
taxa richness, EPT
richness, specic
taxa density
Sediment
addition to
natural stream
channels
<2 mm Decrease in taxa
richness, EPT
richness and specic
taxa density. Eects
most signicant
in pasture streams
where pre-
treatment richness
and diversity were
highest.
10/20 taxa
unaected
Matthaei
et al. (2006)
44 Sediment Assessment Methods • Section 4. Supporting information
Taxon /
Community
descriptor
Experimental
method
Sediment
size
Quantitative
relationship
established
Source
Taxa richness, EPT
richness
Sediment
addition to
natural stream
channels
<2 mm
(mean=0.2
mm)
Increase from 35%
to 83% ne cover
correlated with
increased taxa and
EPT richness
Townsend
et al. (2008)
Invertebrate density,
EPT richness
Spatial survey of
32 streams
<1 mm With an increase
in ne sediment
cover there was
an increase in
invertebrate density,
and a decrease in
EPT taxa richness
Townsend
et al. (2008)
4.1.2 Fish
Sediment inuences sh directly through physical eects and indirectly through impacts on habitat
and food supply. Most physical eects are attributed to the gill damaging properties of suspended
sediment, which can limit sh growth and make sh susceptible to disease (Waters 1995). Suspended
sediment can also reduce the visual foraging eciency of sh including the avoidance of highly turbid
rivers by migratory species (Boubée et al. 1997; Rowe & Dean 1998). In comparison, deposited sediment
limits the amount of habitat available for spawning and can reduce the viability of egg survival (Wood
& Armitage 1997; Harvey et al. 2009). Salmonids are particularly susceptible to excess sediments that
suocate eggs in redds (Hay 2005).
Deposited ne sediment also reduces the amount of habitat and cover available to juvenile and adult
sh. Native sh species favour habitats with large interstices (e.g., gaps between cobbles) which are
important for refuge (Jowett & Boustead 2001; McEwan 2009). In terms of food availability, sediment
can alter the macroinvertebrate community in favour of less preferred food items for some sh species,
i.e., a reduction in drifting species. As such, sediment can aect the small-scale distribution of shes and
hence sh density and richness.
Information on the eects of deposited sediment on native New Zealand sh is limited to studies of
habitat and food preferences. For many species information is anecdotal at best. Few quantitative
relationships have been reported, although there are some established relationships between
suspended sediment and sh populations (Table 4-2).
45Sediment Assessment Methods • Section 4. Supporting information
Table 4-2. Quantitative and observational relationships that have been documented
between proportions of deposited sediment, suspended sediment and sh populations
in New Zealand.
Taxon /
Community
descriptor
Experimental
method
Fine
sediment
metric
Relationship
established
Source
Upland bullies
(adult)
Sediment
addition (12.4 kg/
m
2
) in articial
stream
96% <2
mm and
4% >2 mm
50% decline in
numbers after six
days
Jowett &
Boustead
(2001)
Redn bully,
Shortjaw kokopu
Survey of a
natural stream
0.5 mm
as part of
substrate
index
Presence
associated with
gravel and larger
substrates in day
and gravel and
smaller substrates
at night
McEwan
(2009)
Koaro Survey of a
natural stream
0.5 mm
as part of
substrate
index
Presence
associated with
larger substrates
day and night
McEwan
(2009)
Banded kokopu Spotlight survey
of a natural
stream
2 mm and
as part of
a substrate
index
Size-based
microhabitat
selection observed
with smaller sh
associated with
smaller substrate
sizes
Akbaripasand
et al. (2011)
Banded kokopu Laboratory
preference trials
17 NTU
25 NTU
50% avoidance
response
Boubée et al.
(1997)
Redn bully Laboratory
preference trials
1110 NTU No avoidance
behaviour
Boubée et al.
(1997)
Koaro Laboratory
preference trials
70 NTU 50% avoidance
response
Boubée et al.
(1997)
Inanga Laboratory
preference trials
420 NTU 50% avoidance
response
Boubée et al.
(1997)
Large eels Laboratory
preference trials
1110 NTU No avoidance
behaviour
Boubée et al.
(1997)
46 Sediment Assessment Methods • Section 4. Supporting information
Taxon /
Community
descriptor
Experimental
method
Fine
sediment
metric
Relationship
established
Source
Smelt Laboratory tank
experiment
640 NTU 59% reduction in
feeding rate
Rowe & Dean
(1998)
Banded kokopu Laboratory tank
experiment
20 NTU 45% reduction in
feeding rate
Rowe & Dean
(1998)
Redn Bully Laboratory tank
experiment
Between
40 and 640
NTU
50% reduction in
feeding rate
Rowe & Dean
(1998)
Banded kokopu Laboratory tank
experiment
120 mg/l
suspended
solids
60% avoidance
response
Rowe et al.
(2000)
Banded kokopu Suspended
sediment
addition to
natural stream
channel
25 NTU 40% moved
upstream when
NTU was below 25
NTU, after which
there was 0%
movement
Richardson
et al. (2001)
Longn eel, shortn
eel, Common bully,
Redn bully, Bluegill
bully, Torrentsh,
Inanga, Smelt, Koaro
Survey of natural
stream channel
Clarity Range in clarity
values explained
40% variation in
species richness
Richardson &
Jowett (2002)
Smelt Laboratory tank
experiment
1700 to
3000 NTU
50% mortality after
24 hours
Rowe et al.
(2004)
Inanga Laboratory tank
experiment
1750 to
2100 NTU
50% mortality after
24 hours
Rowe et al.
(2004)
Banded kokopu Laboratory tank
experiment
43000 NTU 10% mortality after
24 hours
Rowe et al.
(2009)
Redn Bully Laboratory tank
experiment
43000 NTU 15% mortality after
24 hours
Rowe et al.
(2009)
4.1.3 Recreational and aesthetic values
Excess ne sediment can detrimentally aect the amenity value of rivers and streams including
recreational use for swimming and other water sports, shing and general aesthetics. Poor water clarity
associated with suspended sediments, or bed sediments that are suspended on contact, usually results
in a negative experience for swimmers, as does the ‘feel’ of ne sediment under the toes.
47Sediment Assessment Methods • Section 4. Supporting information
Aesthetic value can be a very personal experience and dicult to measure. However, studies have
demonstrated how people prefer streams with good visual clarity for bathing (Smith et al. 1995) and
low turbidity for aesthetic value (Püger et al. 2010). This is reected in surface water quality guidelines
that set minimum clarity levels for recreational water use (e.g., >1.6 m black disc visibility, MfE 1994).
However, currently there are no quantitative relationships between deposited ne sediment and
recreational value.
4.2 Review of sediment
assessment methods
A wide range of methods have been applied in New Zealand and elsewhere to quantify sediment
in rivers and streams (Bunte & Abt 2001; Meredith et al. 2003; Sutherland et al. 2008). Although it is
recognized that there is not necessarily one universal method to assess ne sediment in streams, some
standardisation appeared necessary and was a key reason for initiating this project. Dierent properties
of sediment may relate more informatively to some in-stream values compared to others. A conceptual
model of the proximate stressors and causal pathways that lead to a response in benthic biota due
to increased deposited sediments can help identify the required focus of sediment metrics (Figure
4-1). This conceptual model suggests that at minimum sediment metrics should assess substrate size,
interstitial space and the coverage of ne sediment if all components of the issue are being evaluated.
However, the model does not take into account the interdependence of these components.
Figure 4-1. Conceptual model depicting the relationship between increased deposited sediment and
the eects on in-stream biota (adapted from Cantilli et al. 2006).
48 Sediment Assessment Methods • Section 4. Supporting information
A review of the most common protocols used globally was used to determine the most robust and
appropriate protocols for application in New Zealand (Table 4-3).
Table 4-3. Measures of in-stream sediment reviewed in this document (Section 4.2) Trialled protocols
and associated metrics tested in this study are in bold.
Sediment component Sediment measure Protocols and metrics
Substrate size Particle size distribution Wolman pebble count: %
sediment, d16, d50, d86, central
tendency
Volumetric sample sorting: %
sediment , d16, d50, d86, central
tendency
Relative bed stability Calculation using particle size
distribution, channel slope,
roughness, and ow
Surface cover Percent cover of sediment Bankside visual: % sediment
In-stream visual: % sediment
Interstitial space Suspendible sediment Quorer
#
: SIS, SOS, %SIS
Shue method: index score
Embeddedness In-stream visual: % embedded
Sediment depth Depth: mm
Volume of sediment in pools In-stream measure: V*
#
SIS=suspendible inorganic sediment, SOS=suspendible organic sediment
Below is a description of sediment methodologies that are currently used to assess cover, substrate
size (diversity and stability) and interstitial space or suspendible sediment. For each method, a boxed
summary of their likely relevance for quantifying deposited ne sediment in New Zealand is included.
4.2.1 Percent cover of sediment
Bankside estimates
The areal cover of sediment is a visual assessment of the relative surface area of the streambed covered
with deposited sediment. A visual assessment can be applied either at the habitat scale (e.g., run or
rie) or the reach scale (which might include multiple habitats). It may involve a single estimate to
provide a qualitative measure for the reach, or multiple assessments for individual habitats which are
combined to provide a value for the reach.
There is a long history of rapid bankside assessments of sediment cover in New Zealand. The New
Zealand Freshwater Fish Database (NZFFD) includes bankside assessment data from as early as 1909.
Similarly, almost all regional council SOE data are accompanied by a rapid habitat assessment, e.g., the
length of the habitat sampled, stream width, water depth and the relative proportion of substrate size
49Sediment Assessment Methods • Section 4. Supporting information
cover are all metrics readily recorded. Since the development of Stream Habitat Assessment Protocols
(Harding et al. 2009) a number of councils are routinely recording this information. Bankside visual
assessments result in a single measure for a habitat or reach.
Metric Percent cover of nes
Method Bankside visual assessment
Nature of data Semi-quantitative
Recommended measure SOE monitoring
In-stream estimates
An underwater viewer is used to systematically sample multiple patches of the streambed in an
in-stream visual assessment (Figure 4-2). Multiple estimates of ne sediment cover are averaged to
provide a measure for a given habitat or reach. Protocols for this technique have been developed
independently in New Zealand; in the Motueka River (Phillips & Basher 2005) and in Otago (Matthaei
et al. 2006). Matthaei et al. (2006) detected a signicant biological response to sedimentation assessed
using this technique. Best results would seem to be obtained when the resolution of measurements is
small enough to detect changes but large enough to minimise user bias (i.e., choosing measurements
which are multiples of 5%). As with other patch-scale assessments, replication needs to be sucient
to incorporate substrate variability (i.e., more samples are required in heterogeneous substrates).
These visual classication techniques generally require a high level of training to minimise user error
(Latulippe et al. 2001). Dividing the viewer into smaller elds (Figure 4-2) helps to decrease user bias
and are useful during user training, but do not necessarily improve the accuracy of visual assessment
(Bungton & Montgomery 1999a).
Figure 4-2. Photos of an underwater viewer used to assess ne sediment cover.
Metric Percent cover of nes
Method In-stream visual assessment with an underwater viewer
Nature of data Semi-quantitative
Recommended measure SOE monitoring
Eects-based assessments
50 Sediment Assessment Methods • Section 4. Supporting information
4.2.2 Particle size distribution
Pebble count
Surface particle size distribution, or central tendency, is assessed by a systematic grid method or a
pebble count method. Pebble counts are usually based on the Wolman technique (1954), where a
predetermined number of particles are measured in a reach or habitat. The B-axis (width) of each
particle is measured with a ruled rod or gravelometer (Figure 4-3). Particles are chosen from the front
of the boot or using a rod or stick placed on the streambed along designated transects. Alternatively,
particles are chosen by the random placement of a hoop. The B-axis size classes of between 60 (Harding
et al. 2009) and 400 (Bunte & Abt 2001) particles are measured depending on the goals of the study.
Figure 4-3. Photos of common tools used to measure particle grain size: a gravelometer (left) and a rod
graduated with Wentworth scale size classes (right).
Pebble counts are a simple and eective technique for assessing size distributions, but results can be
misrepresentative when the method is not conducted rigorously. For example, particle selection can
bias towards larger particles because these are easily seen and there is a tendency to avoid areas of
unstable footing (e.g., bedrock or large cobbles) (Bunte et al. 2009).
Due to operator bias, sampling error and the high replication required to detect a change in % nes
from pebble count data (Bevenger & King 1995; Bunte & Abt 2001), a pebble count is often not
recommended for the robust analysis of deposited sediment. However, with proper application, this
technique can be condently used to quantify percent sediment as well as other substrate attributes,
for example, mean particle size and particle size variability. Proper application requires user training,
appropriate equipment (e.g., gravelometer) and rigorous and careful application (i.e., the counting of
particles from representative habitats). This method is likely to be useful for sediment assessment as
well as general habitat assessment and the characterisation of sites (see Stream Habitat Assessment
Protocols for further information, Harding et al. 2009).
Metric Particle size distribution
Method Wolman pebble count
Nature of data Semi-quantitative
Recommended measure SOE monitoring
Useful for site establishment data and/or general habitat
assessment
51Sediment Assessment Methods • Section 4. Supporting information
Volumetric sampling
Collecting a grab sample of sediment is one way to overcome operator bias in calculating particle
size distributions. Systematic sampling of the streambed using a shovel or corer collects sediments
for quantitative sorting in the laboratory (Bunte & Abt 2001; Sutherland et al. 2010), however, this
method is very labour intensive. Furthermore, depending on the nature of substrate (ne or coarse) and
uniformity of the bed material, large quantities of substrate may need to be sampled. For example, sites
with heterogeneous substrate may require over 200 samples to determine d50 within + 10% (Mosley
& Tindale 1985). However, the sorting of sediments in the laboratory also allows for the determination
of ner grain sizes, i.e., the relative proportion of silt and clay (<0.063 mm), ne sand (0.063 – 125 mm),
medium sand (0.125 – 0.5 mm) and coarse sand (0.5 – 2 mm). Resulting data accurately reect the
particle size distribution of the sampled habitat.
Metric Particle size distribution
Method Volumetric sample sorting (systematic sample collection
and laboratory analysis)
Nature of data Quantitative
Recommended measure Eects-based assessments
Site specic values and/or research
Metric Relative bed stability
Method Calculation using particle size distribution, channel slope,
roughness, and ow
Nature of data Quantitative
Recommended measure Useful for site establishment data
Site specic values and/or research
4.2.3 Relative bed stability
Relative bed stability (RBS) is not a direct measure of sediment abundance; rather it is a measure of how
resistant a streambed is to substrate movement at a prescribed ow, usually bank-full ow. For example,
in a stream where the majority of sediment is ner than the substrate size moved during bank-full ows
then the RBS metric will indicate that the stream is relatively unstable. Generally the more ne sediment
that is present, the lower the RBS. There are several methods to calculate RBS, but they all involve an
assessment of median particle size, channel slope and bank-full channel dimensions (Jowett 1989;
Bungton & Montgomery 1999b; Gordon et al. 2004; Kaufman et al. 2009). In a recent study, K
aufmann et al. (2009) showed how a derivative of RBS decreased in relation to the level of human
disturbance in the catchment, although streams in soft sedimentary geologies appeared more
susceptible than others.
In New Zealand, studies suggest that RBS may not provide a good estimate of bed stability and
seldom correlates to bed load movement in river types other than homogenous gravel streams
(Death & Winterbourn 1994; Schwendel et al. 2009). Given the time involved in measuring data for RBS
calculation and the potentially system-limited application of this metric it is not recommended for
regular assessment of deposited ne sediment.
52 Sediment Assessment Methods • Section 4. Supporting information
4.2.4 Embeddedness
Embeddedness refers to the degree to which coarse particles are surrounded by ne particles and can
provide an indication of the availability or clogging of interstitial spaces. Common methods range from
subjective description of the proportion of streambed covered by ne sediment (an erroneous use of
the term embeddedness) (Platts et al. 1983) through to the measurement of the depth or width of a
particle surrounded by ne sediment (Burns & Edwards 1985). A comparison of methods suggested
that the United State Environmental Protection Agency (USEPA) method was most likely to provide
results that conformed with the expectation of embeddedness as an eect of altered sediment regimes
below a dam (Sennatt et al. 2006). The USEPA method involves estimating the fraction of the surface
area of at least 55 particles (>10 cm in diameter) which are surrounded by sediment (<2 mm).
For particles less than 2 mm, embeddedness is recorded as 100% (Peck et al. 2000).
4.2.5 Suspendible nes
Quorer method
A volumetric measure of sediment deposition on and within the stream bed can be gained by
re-suspending sediment in the water column and then collecting and weighing the suspendible
proportion of sediment deposited. The Quorer method was developed to measure the suspendible
surface and subsurface sediments in gravel-bed rivers (Lambert & Walling 1988; Quinn et al. 1997).
This method involves using an open-ended container or tube to isolate a patch of the streambed
(Figure 4-4). Surface and subsurface sediments are collected after stirring the streambed within the
corer. Samples are processed in the laboratory to provide relative measures of suspendible inorganic
sediment (SIS) and suspendible organic sediment (SOS), which need to be standardised to background
stream concentrations. Interstitial sediment is inferred from the amount of ne sediment recorded. The
percent of SIS in relation to total suspendible solids (i.e., SIS + SOS) provides a measure of sediment
composition and quality.
Collins and Walling (2007) showed that measures of suspendible nes can vary a lot within sites and
over time. The sensitivity of these measures was illustrated in Waikato hill-country streams where
suspendible sediment was shown to be signicantly greater in streams draining pasture and pine
catchments compared to streams draining native forest catchments (Quinn et al. 1997). Like all patch-
scale measures, the replication required to accurately characterise a site and to detect dierences over
space and time is dependent on substrate variability within each site.
Metric Embeddedness
Method In-stream visual assessment
Nature of data Qualitative
Recommended measure SOE assessments
Eects-based assessments
53Sediment Assessment Methods • Section 4. Supporting information
Figure 4-4. Photos of sample collection using the Quorer method.
Shue method
A rapid qualitative assessment of suspendible sediment may be gained by a subjective rating of
the sediment plume resulting from disturbing the streambed. Used by Environment Canterbury
and Tasman District Council, the ‘Shue’ index can provide useful eects-based assessments. The
method involves standing in the stream and disturbing the streambed by shuing for a set time
and subjectively ranking (1-5) the size and duration of the resulting sediment plume (Figure 4-5). This
method can be improved by taking photographs of the plume (for training and illustration purposes).
The Shue index also has the potential to directly assess the eects of sediment deposition on
the aesthetic or swimming value of streams, i.e., a river or stream that becomes highly turbid is less
attractive to many recreational users (see Section 4.5.6).
Metric Suspendible ne sediment
Method Quorer (in-stream corer to collect sediment and
laboratory analysis)
Nature of data Quantitative
Recommended measure SOE assessments
Eects-based assessments
54 Sediment Assessment Methods • Section 4. Supporting information
Figure 4-5. Suspendible nes index (Shue index).
Metric Suspendible ne sediment index
Method Shue method
Nature of data Qualitative
Recommended measure SOE assessments
Eects-based assessments
Metric Sediment depth
Method Ruler measurement of sediment depth
Nature of data Quantitative
Recommended measure SOE assessments
Eects-based assessments
4.2.6 Sediment depth
A quantitative measure of sediment depth can be made by inserting a ruler into the streambed.
This technique is most appropriate in naturally occurring soft-bottom streams or in specic habitats
where sediment accumulates, for example, the tail of pools. Replicate measurements can be averaged
to provide a mean depth for any given habitat (see also, Section 4.2.7 Volume of nes in pools)
Diculties in measuring ne sediment associated with macrophyte dominated streams can be
overcome by measuring sediment depth. Average sediment depth is multiplied by an areal estimate
of macrophyte cover to obtain volumetric estimates of sediment deposition (Heppell et al. 2009). In
their study, Heppell et al. (2009) demonstrated how sediment deposition was strongly correlated to the
seasonal growth and subsequent cover of macrophytes in lowland river reaches.
55Sediment Assessment Methods • Section 4. Supporting information
Metric Volume of nes in pools
Method In-stream measurement
Nature of data Quantitative
Recommended measure Eects-based assessments
Site specic values and/or research
4.2.7 Volume of nes in pools
The relative amount of sediment in pools (V*) has been shown to correlate to in-stream sediment
supply (Lisle & Hilton 1992; Hilton & Lisle 1993; Lisle & Hilton 1999). Fine sediment in pools can account
for 5-20% of ne sediment in the active channel (Lisle & Hilton 1999). However, nes in pools are usually
much ner and more readily suspended than deposited sediments in other parts of a river, and V*
can vary a lot over time (Lisle & Hilton 1999). Another aspect that limits the application of V* is that
sampling error can be high and requires between four and eight transects of four to 26 pools per reach
(Lisle & Hilton 1999). Lisle & Hilton (1999) recommend that the metric is most suitable for assessing
streams subject to high sediment loads (where V* is greater than 20%) and for long-term monitoring in
relation to a ow record, rather than broad-scale spatial monitoring.
4.3 Protocol testing and validation
Six protocols were trialled as part of the protocol testing and validation stage of the project (Table 4-4).
Protocols were chosen following a literature review and an expert assessment (by the authors) of
their potential applicability in wadeable rivers and streams in New Zealand. Protocols included
bankside visual estimate (% sediment), an in-stream visual estimate (% sediment), Wolman pebble
count (% sediment), Quorer suspendible sediment (g/m
2
), Shue method (index score), and sediment
depth (m).
Draft protocols for applying each of these metrics were trialled at a total of 174 sites by councils during
their sampling programmes in 2009/2010 (Table 4-4). Not all protocols were applied at all sites. Councils
were provided with a document outlining project goals, eld protocol applications and
eld sheets.
Specic aspects of the protocols (e.g., user variability, interhabitat variability) were further tested at an
additional 63 sites in 2010 (Table 4-4). Data from previous studies, where the selected protocols had
been used, were also collated to be used in analyses; n = 90 sites.
56 Sediment Assessment Methods • Section 4. Supporting information
4.3.1 Do results vary for dierent habitats?
Variability in sediment levels between habitats was tested by applying draft protocols in runs, ries and
pools at twelve streams in the Tasman region. Streams were all located on ‘Moutere gravels sedimentary
geology where stream form (i.e., substrate composition and ow) is relatively similar among habitats
compared to other geologies. This means any observed dierence in sediment metrics among habitats
is likely to be greater in other geological settings. Stream catchments ranged in native vegetation cover
from 2% to 99%; any signicant habitat eects should reect a consistent response across streams
subject to varying land uses.
Run and rie habitats had signicantly less sediment than pool habitats according to the in-stream
visual protocol, Quorer and sediment depth metrics (Figure 4-6). This result was repeated for sediment
cover based on the bankside visual, Wolman pebble count and for the Shue index score, however,
the values were not statistically signicant among habitats. Because runs are usually intermediary in
ow and form to rie and pools, results from this survey suggest that the application of protocols in
run habitats should provide a representative assessment of ne sediment at a reach scale.
Furthermore, the close similarity between data collected from runs and ries also implies that it
is reasonable to compare sediment data collected from runs with biotic data collected from ries
(Section 4.3.8).
Table 4-4. Number of sites where sediment protocols were trialled by regional councils in 2009/2010
and data from additional and previous studies (totals in parentheses).
Protocol
Region
Bankside
visual
In-stream
visual
Wolman
pebble
Quorer
Shue
index
Sediment
depth
Northland 10 6 7 7 10 10
Auckland 32 32
Waikato 10 10 10 10 10
Horizons 37 37 40
Hawkes Bay 8 8 8
Taranaki 6 6 6 6 4 6
Wellington 10 10 10 10
Marlborough 4 5 5 5 5 4
Tasman 4 4 4 4 4 3
Canterbury 29 29 29 29 29 29
Otago 16 16 16 8 16
Southland 3
Additional
studies 64 106 111 162 64 106
Total 166 (230) 124 (230) 127 (228) 119 (281) 88 (152) 52 (158)
57Sediment Assessment Methods • Section 4. Supporting information
Pool Riffle Run
HABITAT
0
10
20
30
40
50
60
70
80
90
RAPID
Pool Riffle Run
HABITAT
0
20
40
60
80
100
VISUAL
Pool Riffle Run
HABITAT
0
10
20
30
W2
Bankside visual (% sediment)
Habitat: F
(2, 33)
=1.51, p=0.231
In-stream visual (% sediment)
Habitat: F
(2, 684)
=28.53, p < 0.001
Post hoc: Pool > Run ≥ Rie
Wolman (% sediment)
Habitat: F
(2, 33)
=1.56, p=0.225
Pool Riffle Run
HABITAT
0.5
1.0
1.5
2.0
2.5
3.0
3.5
SCORE
Pool Riffle Run
HABITAT
0.00
0.02
0.04
0.06
0.08
0.10
0.12
DEPTH (m)
Quorer SIS (g/m
2
)
Habitat: F
(2, 144)
=7.42, p=0.001
Post hoc: Pool > Run ≥ Rie
Shue index score
Habitat: F
(2, 72)
=1.75, p=0.181
Sediment depth (m)
Habitat: F
(2, 684)
=8.99, p<0.001
Post hoc: Pool > Rie=Run
Figure 4-6. Variation in sediment metrics among run, ries and pools illustrated with box plots of
the mean, upper and lower quartiles and outliers, and results of analysis of variance of metrics
among habitats.
4.3.2 Do results vary among dierent users?
In order to test the amount of variation which might occur between observers in the eld, ten
freshwater researchers were asked to make bankside visual estimates of % sediment in the same
reach. Findings indicated that having only one to two sta making observations can lead to poor
accuracy in bankside visual estimates. However, in this trial observers were not given any training or
allowed to discuss their estimates with each other. Therefore, to maximize consistency of results, it is
recommended that observers are trained (e.g., by showing a series of photographs with the level of
sediment cover shown or doing training assessments at sites covering a range of measured % cover
values) and, if possible, either using a consistent single observer or two observers in consultation.
Observed trials were conducted in a total of three reaches where the percentages of ne sediment
cover (based on Wolman pebble counts) were 10%, 40% and 80%, respectively.
In general, low sediment sites were correctly assessed as having low levels of bed sediment, moderate
sediment sites were assessed as moderate and high sediment sites assessed as having high sediment
using bankside visual estimates (Figure 4-7).
58 Sediment Assessment Methods • Section 4. Supporting information
Bed sediment cover (%)
0 20 40 60 80 100
Bankside visual estimated (%)
0
20
40
60
80
100
Figure 4-7. Median values and 25th and 75th quartiles and range of bankside visual estimates of
sediment cover by 10 observers at three reaches with varying sediment levels measured using the
Wolman pebble count (10, 40, and 80%).
In-stream visual assessments using an underwater viewer were also made with 10 observers in the
three reaches with diering levels of bed sediment (Figure 4-8). These results indicate that observers
were able to accurately determine sites with low sediment levels; however the variation in observations
increased markedly as sediment levels increased. In the high sediment reach, some observers had
diculty agreeing that % sediment cover was greater than 50%, despite being supplied with diagrams
to help estimates. If in-stream visual estimates are used, then training of eld sta is essential to reduce
variability and improve accuracy. If possible, taking photos will allow quality control of observations.
User-variability of the Shue method or the Wolman pebble count method was not tested.
Reach sediment levels
Low Moderate High
In-stream estimates (% fines)
0
20
40
60
80
100
Figure 4-8. Mean values for in-stream visual estimates of sediment cover in three stream reaches with
varying sediment levels: low = 10%, moderate = 40%, high = 80% (n = 10, ± SE) as assessed using
Wolman pebble counts.
59Sediment Assessment Methods • Section 4. Supporting information
4.3.3 Do results vary in dierent land uses?
Whether metric values dier between dierent land uses may reect their sensitivity to anthropogenic
pressure and/or natural environmental variation. For instance, small Waikato hill-country streams
draining native forest had 2-3-fold lower SIS Quorer values than those draining pasture and pine forests
(Quinn et al. 1997). SIS also tended to be higher in small Coromandel streams draining clear-cut pine
plantations than non-harvested pine and native forest catchments and logged sites with continuous
riparian buers (Quinn et al. 2004).
To examine broad-scale spatial variability draft protocols were applied at 50 sites in the Canterbury
region; predominantly rst to third order streams on low gradient alluvial oodplains. Sites were
grouped into ve land-use and waterway categories: agricultural, urban, forest, spring-fed or mountain
streams with 10 streams in each category. This was done in order to test potential variability in the
protocols across both a gradient of sediment stress but also variable physical habitats.
Using the bankside visual protocols, some land uses and stream types had markedly higher levels of
sediment than others (Figure 4-9). The agricultural and urban streams averaged 35-40% sediment.
The forested streams of Banks Peninsula also had relatively high levels of sediment presumably due to
ultrane windblown loess soils that dominate these catchments, whereas spring-fed and mountain
streams averaged about 20% sediment. The variation in % sediment (error bars) within each grouping
of streams was similar among land uses, indicating that the protocols worked equally well under
diering stream conditions.
Using Wolman pebble counts of % sediment, agricultural, urban and forested streams also showed
relatively high % sediment, whereas spring-fed and mountain streams had lower % sediment consistent
with the bankside visual estimates.
0
10
20
30
40
50
60
Agricultural Urban Forest Spring Mountain
Land use & stream types
Bankside visual (% sediment)
0
10
20
30
40
50
60
70
Agricultural Urban Forest Spring Mountain
Land use & stream types
Wolman count (% sediment)
Figure 4-9. Mean % sediment for bankside visual and Wolman pebble counts in diering land uses and
stream types in the Canterbury Plains (± 1SE, n = 10).
Comparisons of data from the Quorer and Shue methods showed slightly dierent results (Figure
4-10). The Quorer (using SIS values) showed urban streams had higher quantities of suspendible
inorganic matter than the other four types of streams. The Quorer method diers from the bankside
visual and Wolman method in that it measures the quantity of ne sediment on and within the upper
layer of the streambed, rather than cover. The Shue method indicated that agricultural, urban and
forested streams had higher amounts of bed sediment than spring-fed and mountain streams. The
Shue method does not distinguish between organic and inorganic components of suspendible
sediment, which may contribute to the dierent trend among land uses. The Shue results ranked
stream types in a similar order as the bankside visual assessment – this suggests that protocols that do
not distinguish between organic and inorganic components of sediment may result in similar spatial
trends.
60 Sediment Assessment Methods • Section 4. Supporting information
0
1000
2000
3000
4000
5000
Agricultural Urban Forest Spring Mountain
Land use & stream types
Quorer SIS (g/m2)
0
1
2
3
4
5
Agricultural Urban Forest Spring Mountain
Land use & stream types
Shuffle index
Figure 4-10. Mean values for Quorer SIS (g/m
2
) and Shue index scores in diering land uses and
stream types in the Canterbury Plains (±1SE, n = 10).
4.3.4 How do results vary over time?
The draft protocols were tested in one season (summer 2009/2010). However, historic datasets show
that in-stream visual methods, Quorer SIS and Wolman pebble counts can be used to detect
signicant trends in sediment over time. Temporal trends in sediment may be caused by seasonal
ow inuences on the distribution of sediment, or pulses in sediment as a result of land use or natural
disturbance events.
Datasets exist for Quorer and Wolman pebble counts from 11 years in the Whatawhata streams where
SIS and % sediment, averaged across all sites, were 28% and 29% higher, respectively, in early autumn
(March) than in spring (September) (Quinn et al. 2009). Both SIS and % sediment increased at a native
forest stream after a tree fell into the reach and accumulated sediment upstream, demonstrating that
these methods are able to detect changes due to natural inuences even at relatively small scales.
Percent sediment also showed a signicant decrease in a small pasture stream following native forest
riparian planting (Quinn et al. 2009).
Quorer derived SIS showed signicant temporal patterns along the Tongariro River, a gravel-cobble bed
river, in a one-year study of hydro-electricity generation related impacts on stream habitat and biota
(Quinn & Vickers 1992) (Figure 4-11). SIS varied markedly with season and tended to be higher in winter
and lower in spring, however this trend was strongly inuenced by larger scale processes inuencing
the reach. This study indicates that SIS values at any site might uctuate by 100% or more within a year.
In general, results from all methods detected dierences in sediment values associated with dierent
land uses. The similarity in variability among land uses suggests that the methods work similarly well in
a range of stream settings.
61Sediment Assessment Methods • Section 4. Supporting information
Figure 4-11. Seasonal patterns of Quorer suspendible inorganic sediment (g/m
2
) in run habitat at sites
down the Tongariro River in relation to the inuences of sand input (from Rangipo Desert) and dams.
Letters (above the bars) and lines (in seasons legend) indicate signicant dierences (p < 0.05) using
two-way ANOVA with Bonferroni post-hoc multiple comparisons (adapted from Quinn & Vickers 1992).
Temporal variability has been observed over several years at sites throughout the Motueka River
catchment using a method comparable to the in-stream visual assessment (L. Basher, pers. com.).
A large ood in March 2005 mobilised sediment along the river resulting in greater than 20% of the
bed having more than 50% sediment cover (Figure 4-12). Within six months, much of this sediment had
been ushed from the site.
Figure 4-12. Temporal changes in % sediment cover in a run site on the Motueka River.
A signicant temporal trend was observed over a 5-yr period (p = 0.003). Data courtesy of Les Basher,
Landcare Research.
62 Sediment Assessment Methods • Section 4. Supporting information
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0 5 10 15 20 25 30
300 g/m
2
change
400 g/m
2
change
500 g/m
2
change
Sample size
SIS g/m
2
(median=356, SD=210)
Power
Figure 4-13. The eect of sample size and the desired eect size on statistical power of comparisons of
suspendible inorganic sediment (SIS) measurements.
A similar analysis can be conducted for the other protocols which involve replicate measures at a site.
With 20 replicate measurements the proposed in-stream visual protocol should enable satisfactory
statistical power to detect a change of 15% ne sediment cover at a site (Table 4-5). Similarly, the
proposed Shue method involving three replicate measurements should enable satisfactory power to
detect a change of one unit in the Shue index score.
4.3.5 How many replicates are required?
The number of replicates required for a sampling programme depends on the expected variability
among replicate measurements, the size of the eect, and the statistical power required to detect
an eect of this size. Statistical power is maximised when there are a large number of replicate
measurements, low variability among replicate measurements, and there is a large eect. Typically a
statistical power of 0.8 is considered satisfactory – i.e., there is an 80% chance of detecting an eect of a
certain size given the number of replicates and the variability among those replicates.
By examining the observed variability among replicate measurements for each of the protocols it is
possible to give some guidance on the number of replicate samples required to detect an eect of a
particular size. For example, using the average variability observed within sites for the protocol trials
only three to four replicate measurements of suspendible sediment are required to have satisfactory
statistical power to detect a 500 g/m
2
change in suspendible sediment. However, six replicate
measurements are required to have the same statistical power to detect a change of 400 g/m
2
(Figure
4-13). The proposed protocol for suspendible sediment assessment which involves six replicate
measurements (Section 2.3) will enable satisfactory power to detect a change in SIS of 400 g/m
2
.
63Sediment Assessment Methods • Section 4. Supporting information
Table 4-5. Summary of the number of replicates required for sediment protocols to condently (power
= 0.8) detect a range of eect sizes.
Protocol Size of eect to be
detected
Number of replicates
required to have
satisfactory statistical
power (0.8)
Suspendible sediment 300 g/m
2
400 g/m
2
500 g/m
2
9
6
4
In-stream visual 10% change in cover
15% change in cover
25% change in cover
36
18
10
Shue 0.3 units
0.5 units
1 unit
26
10
3
Sediment depth 10 mm
20 mm
30 mm
50+
16
8
4.3.6 How do results from dierent protocols compare?
Data from regional council trials were used to examine the correlations between data from dierent
protocols, including bankside visual (Bankside), in-stream visual (In-stream), % sediment from Wolman
pebble counts (Wolman), suspendible inorganic sediment (SIS), suspendible organic sediment (SOS),
Shue index score (Shue) and sediment depth (Depth). Data were transformed to improve normality
where necessary. The only protocol which did not compare well was sediment depth. Pearson
correlation coecients showed strong relationships between visual estimates of ne sediment cover,
for example, bankside and in-stream visual r = 0.89, p < 0.01 (Figure 4-14). The Quorer metrics were
also highly correlated, for example, log-SIS and log-SOS, r = 0.85, p < 0.001. The Shue index was
signicantly correlated with all other sediment measures, except sediment depth (Figure 4-14).
64 Sediment Assessment Methods • Section 4. Supporting information
RUN
VISUALAVE
W2
LOGSIS
LOGSOS
SCORE
RUN
LOGDEPTH
VISUALAVE W2 LOGSIS LOGSOS SCORE LOGDEPTH
Bankside Instream Wolman LogSIS LogSOS Shuffle LogDepth
LogDepth Shuffle LogSOS LogSIS Wolman Instream Bankside
0.89 (103)
0.709 (107) 0.677 (82)
0.442 (93) 0.423 (94) 0.449 (69)
0.339 (93) 0.418 (93) 0.352 (69) 0.847 (108)
0.563 (72) 0.628 (72) 0.485 (84) 0.51 (67) 0.439 (67)
0.461 (39) 0.123 (47) 0.296 (48) 0.175 (40) 0.119 (40) 0.294 (48)
RUN
VISUALAVE
W2
LOGSIS
LOGSOS
SCORE
RUN
LOGDEPTH
VISUALAVE W2 LOGSIS LOGSOS SCORE LOGDEPTH
Bankside Instream Wolman LogSIS LogSOS Shuffle LogDepth
LogDepth Shuffle LogSOS LogSIS Wolman Instream Bankside
0.89 (103)
0.709 (107) 0.677 (82)
0.442 (93) 0.423 (94) 0.449 (69)
0.339 (93) 0.418 (93) 0.352 (69) 0.847 (108)
0.563 (72) 0.628 (72) 0.485 (84) 0.51 (67) 0.439 (67)
0.461 (39) 0.123 (47) 0.296 (48) 0.175 (40) 0.119 (40) 0.294 (48)
Figure 4-14. Correlations among sediment protocols. Values are the Pearson correlation coecients (r)
with the number of sites in parentheses, p < 0.01. Italicised values are where p < 0.05 and red values are
where p > 0.1.
4.3.7 Are there cheaper, quicker methods?
The Quorer method provides a quantitative assessment of surface and interstitial sediment. The sample
collection in the eld is relatively quick – however, the method has a nancial cost associated with
laboratory analysis for total and volatile suspended solids. Therefore alternatives were investigated to
reduce this processing cost. These included measuring turbidity in the Quorer method sample in the
laboratory, and measuring the volume of suspendible sediment in the Quorer method sample using a
settling assay.
Turbidity of a Quorer method sample was measured in the laboratory prior to settling the sample
to calculate suspendible benthic sediment volume (SBSV). Turbidity (NTU values) was signicantly
related to Quorer SIS (g/m
2
) values and a stronger positive relationship was noted with Quorer SBSV
values (Figure 4-15). Comparisons of turbidity of the Quorer method sample with other sediment data
collected from the same 50 sites showed no relationships. It appears that more research is required to
determine the conditions in which turbidity within the Quorer method sample might provide a useful
measure of suspendible sediment, if at all.
65Sediment Assessment Methods • Section 4. Supporting information
Turbidity (NTU)
SIS (gm
2
)
r2 = 0.24
P < 0.001
10,000
1,000
100
1
10 100 1000 10,000
Turbidity (NTU)
SBSV
0.01
0.1
1
10
100
10 100 1,000 10,000
r2=0.76
P<0.001
Figure 4-15. Comparison of metrics generated from Quorer method samples collected from 50 sites
on the Canterbury Plains, including turbidity (NTU), suspendible inorganic sediment (g/m
2
) and
suspendible benthic sediment volume (ml/m
2
). Note the log-scale of both axes.
Ninety-three Quorer method samples were processed to measure suspendible inorganic sediment
(SIS) and suspendible benthic sediment volume (SBSV) using a settling assay. This assay involved letting
the samples settle in a cylinder and measuring the volume of settled sediment. The values from these
two methods were generally closely correlated for any given sample (Figure 4-16). A linear regression
showed a signicant positive relationship (r
2
= 0.43, F (1, 48) = 35.79, p < 0.001) and results suggested
that SBSV could provide a surrogate for SIS. An SIS value of 400 g/m
2
is in the region of a SBSV value
of 3000 ml/m
2
. However, the same method should be used for all samples which are going to be
compared; this trial indicated that not all data points fall directly along the trend line and converting
from one measure to the other will introduce some error.
Figure 4-16. Comparison of measurements of amounts of suspendible inorganic sediment (SIS) based
on dry weights with areal suspendible benthic sediment volume from settled volumes for 93 samples
(ve outliers, with SIS >2000 g/m
2
, are not shown).
SIS (g/m
2
)
SBSV (mL/m
2
)
0
2000
4000
6000
8000
10000
12000
14000
16000
0 200 400 600 800 1000 1200 1400 1600
66 Sediment Assessment Methods • Section 4. Supporting information
4.3.8 How well are sediment metrics related to in-stream biota?
Data from regional council trials was used to calculate all possible sediment metrics (Table 2-1). For the
bankside visual assessment, % sediment cover data was available at both a reach and run-scale, so both
metrics were examined. Sediment metrics were then compared to eight macroinvertebrate variables as
representative measures of in-stream biota (Table 4-6). A series of correlations, regressions and analysis
of covariances (ANCOVAs) were used to examine the relationship between sediment metrics and
biological variables.
Table 4-6. Description of macroinvertebrate variables.
Biotic variable Description
No. of taxa Number of taxa
No. of individuals Number of individuals
EPT abundance Number of individuals belonging to the sensitive Ephemeroptera,
Plecoptera or Trichoptera taxa
%EPT richness Percentage of taxa belonging to the sensitive Ephemeroptera,
Plecoptera or Trichoptera taxa
%EPT abundance Percentage of individuals belonging to the sensitive Ephemeroptera,
Plecoptera or Trichoptera taxa
MCI Macroinvertebrate Community Index (calculated from presence/
absence data)
QMCI Quantitative Macroinvertebrate Community Index (calculated from
abundance data)
SQMCI Semi-quantitative Macroinvertebrate Community Index (calculated
from rank abundance data)
First Pearson and Spearmans rank correlations were used to identify potential relationships. Of the 12
sediment variables suitable for this analysis
3
, the three visual assessments (Bankside reach, Bankside run
and In-stream visual) performed best because they were signicantly related to the largest number of
invertebrate response variables. These metrics were signicantly correlated with seven (In-stream) or six
(Bankside reach, Bankside run) of the eight invertebrate variables. Percent sediment from the Wolman
pebble count and Shue index score were related to three invertebrate variables each, and %SIS
(Quorer) to two variables.
Non-linear relationships in addition to the linear ones were identied using scatter plots and non-linear
regressions. In some cases, log-transformations were used to help full the assumptions of the analysis.
This analysis was computed for the seven better-performing sediment variables identied in the
correlation analyses (the three visual % sediment metrics, log-SIS, log-SOS, % sediment from Wolman
pebble counts and median particle size from Wolman pebble counts [d50]). Seven invertebrate
measures (No. of taxa, No. of individuals, EPT abundance, %EPT richness, %EPT abundance, MCI, QMCI)
had sample sizes that were large enough across all seven sediment predictors to allow running this
3
The database for sediment depth was too small to be included. A separate analysis showed a negative correlative relationship with No. of taxa,
but no other invertebrate variables were signicantly related to sediment depth.
67Sediment Assessment Methods • Section 4. Supporting information
analysis. Once again, the three visual % sediment metrics performed best, and ‘Bankside reach was
the top performer overall. It was signicantly related to all key invertebrate metrics (r
2
values ranged
from 0.13-0.32); all ve invertebrate metrics showing either linear or quadratic declines with increasing
sediment. Log-SOS (Quorer) had the strongest relationship with MCI and QMCI (r
2
= 0.24 in both
cases) but was relevant for fewer invertebrate metrics than the % sediment measures. The relationship
between Bankside reach and MCI had similar strength (r
2
= 0.16). Percent sediment in Wolman counts
and Shue index were both less relevant (in terms of the number of invertebrate metrics aected
and also the r
2
values of these relationships). The generally fairly low r
2
-values for these relationships
indicate that several other factors inuenced the investigated invertebrate response variables besides
the amount of deposited ne sediment at the study sites. The potential roles of two of these additional
factors were examined in our next analysis.
Finally, the inuence of region and stream size in the relationships between six sediment metrics (three
% sediment metrics, log-SIS, log SOS and % sediment in Wolman pebble counts) and MCI and %EPT
richness was investigated. Adding region as a predictor had a signicant eect for all six sediment
variables and increased the r
2
-values of the linear models to a precision (26-54% of the variation in
the data explained) that is fairly high for ecological survey data. In every single case, the eect size for
region (range 0.17-0.38; eect sizes can theoretically range from 0.0 to 1.0) was greater than the eect
size of the sediment predictor in question (range 0.05-0.18). Nevertheless, all three visual % sediment
metrics (and also Wolman % sediment, but neither SIS nor SOS) were still signicantly and negatively
correlated with MCI and %EPT. Based on their eect sizes, In-stream visual and Bankside reach were the
best sediment predictors for MCI, and Wolman % sediment and In-stream were the best predictors for
%EPT richness. In conclusion, regional variation and/or variation between dierent operators played
an important role in this study as expected, but this variation had relative little eect on the main
conclusions drawn from the previous analyses.
In practise, dierences between geographical regions should not be a major problem for determining
sediment-invertebrate relationships during future biomonitoring in New Zealand because regional
councils usually collect all their data within a single geographical region. However, because the
factor region also included potential dierences between dierent operators, training all operators
using standardised criteria to minimise between-operator variation in all assessments of sediment is
recommended.
Adding stream size (using stream width and depth data) as a covariate had no signicant eect: p =
0.11 for Wolman % sediment and MCI, and p > 0.29 in all other cases. The results indicated that the
observed relationships between the six sediment predictors and MCI or %EPT were independent of
stream size.
4.3.9 Other useful things discovered along the way
Bankside visual estimate
Can I stand in one place to make an assessment (e.g., on a bridge)?
The bankside visual assessment should take into account the full sample reach. Usually, it is necessary
to walk along the river bank (sometimes both sides in larger rivers) to estimate sediment cover for the
full sample habitat. Bankside reach-scale and run-scale assessments of % cover are highly correlated,
but analyses show a stronger relationship between run-scale assessments and in-stream values.
In-stream visual estimate
Do I need to record a measurement for every quadrant on the viewer?
Quadrants are very useful for training purposes and quality control among users. Once a user ‘has their
eye in it is not necessary to record data for every quadrant.
68 Sediment Assessment Methods • Section 4. Supporting information
Limitations with the bathyscope
The scope can be dicult to use in shaded areas where the stream bed is hard to see. It is dicult to
use in faster water and if you have shallow, fast water the scope can cause turbulence which will entrain
sediment, altering your readings.
Wolman pebble count
To work well this needs the observer to take care in randomly selecting particles and ensuring that they
record ne sediment among larger particles.
Quorer method
How long do I stir the sediment before collecting a sample?
A small experiment was conducted where samples were collected after 15, 30, 45 and 60 seconds of
stirring. Results (Figure 4-17) suggested that 15 seconds was ample to provide an accurate measure of
SIS.
Time (sec)
Turbidity (NTU)
0
500
1000
1500
2000
2500
15 30 45 60
Figure 4-17. Relationship between turbidity and time of stirring in a Quorer sampler. N = 18.
Are there any physical limitations to applying the Quorer method?
The Quorer method is limited to locations where a corer can be deployed to form a tight seal on the
streambed (inuenced by streambed roughness, substrate size and current velocity) without the stream
water over-topping into the cylinder (inuenced by depth). In general, the method is limited to depths
and velocities below approximately 0.5 m and 0.5 m/s, respectively, and substrate sizes up to gravel/
cobble (not boulders). These limitations can be met to some extent by increasing the Quorer diameter
(to deal with cobble/gravel beds) and depth. Large Quorers have been used with success by NIWA
and Tasman District Council scientists (Figure 4-18). A metal Quorer method sampler with handles can
be useful.
69Sediment Assessment Methods • Section 4. Supporting information
Figure 4-18. Large corers successfully applied to extend the physical limitations of the Quorer method.
Shue method
Both water depth and ow appear to bias this protocol when assigning scores based on views of
a white tile. Possible renements include using a pilot measure to inform the assessor as to which
way the ow is going and where to place the tile, attaching the white tile to a pole and ensuring
readings are made at 200 mm depth. The categorical nature of the index could be rened in future
to a continuous variable which takes into account both ow and depth; by measuring the depth
of the plume in relation to a white pole and the time it takes for the sediment plume to reach and
subsequently clear the tile. However, it would be dicult to take into account the horizontal dispersion
of the plume in the water column. Another possible approach involves reading the eect of standard
bed disturbance in horizontal clarity with a water sample from the Quorer method measured in a mini
Stream Health Monitoring and Assessment Kit (SHMAK) clarity tube. There are currently no data to
validate this approach, but see Section 4.3.7 for discussion on surrogate measures.
Sediment depth
Limited data and weak relationships observed during protocols development suggested sediment
depth is not a very sensitive indicator of sediment eects on biota (however, guideline development
analyses indicated a relationship with taxa richness – see Section 4.5.3). Sediment depth might be a
valuable measure for eects-based assessments. Measuring sediment depth in pools may also provide
a more sensitive measure, but this was not tested in these trials.
70 Sediment Assessment Methods • Section 4. Supporting information
4.4 Review of existing guidelines
4.4.1 New Zealand
Environment Canterbury is the only Regional Council to currently provide numerical guidelines and
include numerical objectives for the areal coverage of ne sediments within a spatial framework for
the region (Environment Canterbury 2011, see Hayward et al. 2009 for rationale). Objectives range from
10% to 40% cover depending on the surface water management unit of interest. These objectives have
been calculated from data collected at 144 sites measured since 1999.
There are currently no national standards or guidelines to assess the eects of sediment on in-stream
values in New Zealand.
4.4.2 International
Many river-type specic guidelines have been developed for areas of the United States and Canada
(Table 4-7). To summarise, the most common sediment criteria in northern America are for the percent
of sediment calculated from pebble counts or by mass and substrate embeddedness assessed using
the USEPA qualitative method. Unfortunately recommended values for each state and province are
dicult to interpret because of inconsistencies in the denition of ne sediment, i.e., anything from 0.64
mm to 6.4 mm.
4.5 Guideline development
An essential criterion for any guideline is that it must relate to a demonstrable eect that can be
quantied (Jones et al. 2011). Thus manipulative experiments (laboratory and eld-scale experiments)
are often used to identify chronic and acute concentrations of contaminants (Table 4-8). There are
examples of manipulative experiments in New Zealand that have identied biotic responses to
sediment additions, although none specically tested sediment thresholds (e.g., Ryder 1989; Dunning
1998; Suren & Jowett 2001; Matthaei et al. 2006; but see also Wagenho 2011). Decreases in mayy,
stoney and caddisy richness have been associated with increases in sediment cover (Table 4-1).
Notably, Townsend et al. (2008) observed a decrease in EPT richness associated with higher sediment
levels in a survey of 32 streams, whereas the response of EPT to experimental sediment and nutrient
addition in nine agricultural streams was more complex.
More commonly, correlations with eld survey data are used to develop sediment guidelines
(Sutherland et al. 2008). While surveys do not provide proof of cause and eect and can be
confounded by multiple stressor eects, they do identify sediment levels associated with changes
in in-stream values.
71Sediment Assessment Methods • Section 4. Supporting information
Location Performance criteria Standard (target)
Alaska % ne sediment
(0.1 mm – 4.0 mm by mass)
≤ 5% above reference or
≤ 30% absolute
Arizona % sediment in ries (Wolman) ≤ 35%
British
Columbia
% ne sediment in redds (by mass)
Geometric mean diameter
≤ 10% (<2 mm) or
≤ 25% (<6.35 mm)
≥ 12 mm
California Geometric mean diameter
% embeddedness in ries
% ne sediment in redds (by wet
mass)
> 69 mm
≤ 25%
≤ 14% (<0.85 mm) or
≤ 30% (<6.4 mm)
Colorado % sediment (Wolman)
% embeddedness
90-100% of expected condition=fully
supporting
73-89% of expected
condition=partially supporting
90-100% of expected condition=fully
supporting
73-89% of expected
condition=partially supporting
Hawaii Fine sediment depth in
hard-bottom streams
≤ 5 mm
Idaho % ne sediment in ries
(by mass)
Rie stability index (RSI)
≤ 10% (<0.85 mm)
≤ 70 RSI
Montana % ne sediment in ries (by mass) ≤ 30% (<6.35 mm)
New
Brunswick
% sediment (Wolman + visual
estimate)
Median particle size
% sediment in ries (by mass)
≤ 7.2% (<2 mm)
≤ 9.3% (<6.35 mm)
> 56.9 mm
≤ 3% (<2 mm)
New Mexico % embeddedness ≤ 33%=fully supporting
> 33% is compared to reference
% sediment in ries (Wolman) < 20%=fully supporting
> 20% compared to reference
Oregon % ne sediment in ries (by mass) < 20%
Prince
Edward
Island
% sediment (Wolman + visual
estimate)
Median particle size
Relative bed stability (RBS)
≤ 12.9% (<2 mm)
≤ 12.7% (<6.35 mm)
> 47.4 mm
≤ 3.8 RBS
Table 4-7. Sediment criteria and standards for the United States and Canada (from Sutherland et al.
2008 and Culp et al. 2009)
72 Sediment Assessment Methods • Section 4. Supporting information
Table 4-8. Summary of approaches used to dene sediment criteria summarised from Jones et al. (2011).
Guideline approach Advantages Disadvantages
Laboratory assessments Controlled
Denition of mortality limits
Individual response observable
Limited to target organisms/
populations
Limited treatment options
Dicult to scale to stream
Field-scale experimental
manipulations
(experimental channels
and simulated events)
Similar to natural conditions
Population and community
level response observable
Limited treatment options
High logistical requirements
Case studies of pollution
events
Population through to
ecosystem response observable
Logistically dicult
Requires opportunistic
sampling often results in lack of
‘before data
Dicult to discern sediment
eect versus other/background
eect
Correlation with eld
survey data
Natural conditions and results
relevant at management scale
Eects in the presence of
multiple pressures observable
Population through to
ecosystem response observable
Does not provide proof of
cause and eect
Separating eects of co-
variables (natural variability) is
dicult
4.5.1 Sources of data
An historic dataset was combined with data collected from our research to provide sediment
information for 454 sites. Sites ranged from rst to seventh order streams and had a wide spatial
coverage – from Northland to Southland and all regions in between, except the West Coast. Classifying
sites by stream type showed that ve of the 20 FENZ 20-level groups were represented by the data
(Groups A, C, D, G, H; for a description of stream types see Leathwick et al. 2011), but these ve groups
account for 83% of the national river network (Leathwick et al. 2011).
73Sediment Assessment Methods • Section 4. Supporting information
4.5.2 Correlation among sediment and biota
Initially, sediment measures were compared with each other. Variables were transformed where
necessary to improve the normality of data distributions.
Correlation analyses identied similarity among all sediment measures (Figure 4-19). All three visual
assessments of % cover (Bankside reach, Bankside run, In-stream visual) were strongly related (r >
0.85). Quorer measures (log-SIS, log-SOS) were signicantly related to each other (r = 0.86, p < 0.01)
and to bankside and in-stream visual assessments of % sediment cover at the run-scale (r = 0.42, p <
0.01), but not the reach-scale measure. The Shue index and Wolman measure of % sediment were
signicantly related to all metric measures (r > 0.40), except log-sediment depth. Sediment depth was
only signicantly related to ve other metrics and the strongest of these relationships was with log-SOS
(r = 0.48).
% fines
Bankside
Reach visual
% fines
Bankside
Run visual
% fines
In-stream
visual
% fines
Wolman
LogSIS LogSOS Shuffle index
score
LogDepth
% fines
Bankside
Run visual
% fines
In-stream
visual
% fines
Wolman
LogSIS
LogSOS
Shuffle index
score
LogDepth
0.86 (125)
0.92 (84)
0.80 (90)
0.28 (86)
0.33 (85)
0.67 (52)
0.45 (32)
0.85 (167)
0.70 (171)
0.44 (186)
0.44 (186)
0.49 (134)
0.37 (98)
0.68 (146)
0.47 (206)
0.42 (206)
0.45 (134)
0.35 (153)
0.43 (232)
0.44 (233)
0.40 (146)
0.25 (107)
0.86 (355)
0.43 (125)
0.46 (142)
0.44 (126)
0.48 (143) 0.39 (105)
% fines
Bankside
Reach visual
% fines
Bankside
Run visual
% fines
visual
% fines
LogSIS LogSOS
0.80 (90)
0.28 (86)
0.33 (85)
0.67 (52)
0.45 (32)
0.85 (167)
0.70 (171)
0.44 (186)
0.44 (186)
0.49 (134)
0.37 (98)
0.68 (146)
0.47 (206)
0.42 (206)
0.45 (134)
0.35 (153)
0.43 (232)
0.44 (233)
0.40 (146)
0.25 (107)
0.86 (355)
0.43 (125)
0.46 (142)
0.44 (126)
0.48 (143) 0.39 (105)
% fines
Bankside
Reach visual
% fines
Bankside
Run visual
% fines
In-stream
visual
% fines
Wolman
LogSIS LogSOS Shuffle index
score
LogDepth
% fines
Bankside
Run visual
% fines
In-stream
visual
% fines
Wolman
LogSIS
LogSOS
Shuffle index
score
LogDepth
0.86 (125)
0.92 (84)
0.80 (90)
0.28 (86)
0.33 (85)
0.67 (52)
0.45 (32)
0.85 (167)
0.70 (171)
0.44 (186)
0.44 (186)
0.49 (134)
0.37 (98)
0.68 (146)
0.47 (206)
0.42 (206)
0.45 (134)
0.35 (153)
0.43 (232)
0.44 (233)
0.40 (146)
0.25 (107)
0.86 (355)
0.43 (125)
0.46 (142)
0.44 (126)
0.48 (143) 0.39 (105)
% fines
Bankside
Reach visual
% fines
Bankside
Run visual
% fines
visual
% fines
LogSIS LogSOS
0.80 (90)
0.28 (86)
0.33 (85)
0.67 (52)
0.45 (32)
0.85 (167)
0.70 (171)
0.44 (186)
0.44 (186)
0.49 (134)
0.37 (98)
0.68 (146)
0.47 (206)
0.42 (206)
0.45 (134)
0.35 (153)
0.43 (232)
0.44 (233)
0.40 (146)
0.25 (107)
0.86 (355)
0.43 (125)
0.46 (142)
0.44 (126)
0.48 (143) 0.39 (105)
Figure 4-19. Correlations among metrics from data collated for guideline development. Values are the
Pearson correlation coecient (r) with number of sites in parentheses, p < 0.01. Red values are where p
> 0.1. Note log
10
transformations of SIS, SOS and sediment depth data.
74 Sediment Assessment Methods • Section 4. Supporting information
Sediment data was then compared in relation to biotic metrics representing in-stream values: MCI,
%EPT richness, taxonomic richness, EPT taxa richness, % trout, % native sh, koura abundance, and eel
abundance. There were no signicant relationships between the Shue index score and any of the
values metrics. There were also no signicant correlations between any of the sediment measures and
sh metrics; however, both MCI and %EPT richness were related to sediment metrics. Log-sediment
depth had a signicant relationship with the number of invertebrate taxa and the number of EPT taxa
(Figure 4-20).
-0.45 (124)
-0.37 (141)
% fines Bankside Reach visual
MCI
% fines Bankside Run visual
MCI
% fines In-stream visual
MCI
LogSIS
MCI
-0.38 (168)
-0.40 (183)
% fines Bankside reach visual
% EPT richness
% fines Bankside Run visual
% EPT richness
% fines Instream visual
% EPT richness
% fines Wolman
% EPT richness
-0.56 (126)
-0.43 (143)
-0.47 (113)
-0.42 (120)
LogSOS
MCI
LogDepth
No. taxa
-0.48 (154)
-0.38 (93)
LogDepth
No. EPT taxa
-0.52 (77)
-0.45 (124)
-0.37 (141)
% fines Bankside Reach visual
% fines Bankside Run visual
LogSIS
MCI
-0.38 (168)
-0.40 (183)
% fines Bankside reach visual
% EPT richness
% fines Wolman
% EPT richness
-0.56 (126)
-0.43 (143)
-0.47 (113)
-0.42 (120)
LogSOS
MCI
LogDepth
No. taxa
-0.48 (154)
-0.38 (93)
LogDepth
taxa
-0.52 (77)
-0.45 (124)
-0.37 (141)
% fines Bankside Reach visual
MCI
% fines Bankside Run visual
MCI
% fines In-stream visual
MCI
LogSIS
MCI
-0.38 (168)
-0.40 (183)
% fines Bankside reach visual
% EPT richness
% fines Bankside Run visual
% EPT richness
% fines Instream visual
% EPT richness
% fines Wolman
% EPT richness
-0.56 (126)
-0.43 (143)
-0.47 (113)
-0.42 (120)
LogSOS
MCI
LogDepth
No. taxa
-0.48 (154)
-0.38 (93)
LogDepth
No. EPT taxa
-0.52 (77)
-0.45 (124)
-0.37 (141)
% fines Bankside Reach visual
% fines Bankside Run visual
LogSIS
MCI
-0.38 (168)
-0.40 (183)
% fines Bankside reach visual
% EPT richness
% fines Wolman
% EPT richness
-0.56 (126)
-0.43 (143)
-0.47 (113)
-0.42 (120)
LogSOS
MCI
LogDepth
No. taxa
-0.48 (154)
-0.38 (93)
LogDepth
taxa
-0.52 (77)
Figure 4-20. Correlations among sediment measures and metrics of in-stream values. Values are the
Pearson correlation coecient (r) with number of sites in parentheses. Only signicant relationships
(p < 0.01) are shown.
75Sediment Assessment Methods • Section 4. Supporting information
4.5.3 Predictive relationships between sediment and biota
Based on the results of correlation analyses, the strength of relationships between biotic indices and
sediment metrics was investigated using linear regression of the combined data set.
1. Percent sediment and MCI and %EPT richness
The strongest relationship observed between MCI and % sediment was for the reach scale bankside
measure: r
2
= 0.20, p < 0.001 (Figure 4-21). Using the linear relationship (y=113.19-0.29x) to predict
the sediment value at 120 MCI (i.e., the value separating clean waters from possible pollution) leads
to a theoretical value of -23% sediment cover, i.e., an absence of sediment. Also, it is apparent from
Figure 4-21 that there is a large spread of MCI values at 0% sediment cover, anywhere between 90 and
142 MCI, probably reecting the eects of factors other than % sediment on MCI. Clearly a regression
approach is not a meaningful or sensitive way to assign guideline values.
The bankside reach scale estimate of % sediment had the strongest relationship with %EPT richness: r
2
= 0.32, p < 0.001 (Figure 4-21). Using the linear relationship (y=52.49-0.33x) to predict sediment cover at
a %EPT value indicative of clean water (50% EPT) results in sediment cover value of 7%. However, there
was a wide range of %EPT values at less than 7% sediment; 7% EPT – 90% EPT.
0 20 40 60 80 100
Bankside reach (% sediment)
0
60
120
180
MCI
0 20 40 60 80 100
Bankside reach (% sediment)
0
60
120
180
MCI
0 20 40 60 80 100
Bankside reach (% sediment)
0
20
40
60
80
100
%EPT richness
0 20 40 60 80 100
Bankside reach (% sediment)
0
20
40
60
80
100
%EPT richness
Figure 4-21. Linear relationship (with 95% prediction condence intervals) between a reach scale
bankside estimate of % ne sediment cover and the MCI metric (n = 124), and %EPT richness (n = 126).
The blue line indicates the % ne sediment value where %EPT exceeds 50%.
2. Suspendible sediment and MCI
The relationship between measures of organic and inorganic suspendible sediment and MCI was
examined. There were relatively weak yet signicant linear relationships with MCI for log-transformed
SOS: r
2
= 0.26, p < 0.001 and log-transformed SIS: r
2
= 0.18, p < 0.001 (Figure 4-22). Using the linear
relationship (y=125.90-13.93x) to predict the log-SOS value at 120 MCI (i.e., clean water) results in a
value of 0.42, equivalent to a back-transformed value of 2.65 g/m
2
. Similarly, for log-SIS the predicted
value at 120 MCI based on the linear relationship (y=135.62-11.67x) was 1.34, equivalent to a back-
transformed value of 21.8 g/m
2
.
76 Sediment Assessment Methods • Section 4. Supporting information
0 1 2 3 4
LogSOS
0
60
120
180
MCI
0 1 2 3 4
LogSOS
0
60
120
180
MCI
1 2 3 4 5
LogSIS
0
60
120
180
MCI
1 2 3 4 5
LogSIS
0
60
120
180
MCI
Figure 4-22. Linear relationships (with 95% prediction condence intervals) between MCI and
suspended organic sediment (n = 154), and suspended inorganic sediment (n = 183). Note that
suspended sediment metrics are log-transformed. The blue line indicates the suspended sediment
value where MCI exceeds 120.
3. Sediment depth and taxa and EPT taxa richness
Notably the only sediment metric signicantly related to total numbers of invertebrate taxa and total
numbers of EPT taxa was sediment depth (Figure 4-23), although the relationships were weak: No.
taxa r
2
= 0.20, p < 0.001; No. EPT taxa r
2
= 0.27, p < 0.001 [Linear relationships: No. taxa (y = 19.339-
2.448x). No EPT taxa (y = 8.962-2.074x)]. There are no guideline values for each of these invertebrate
metrics; however, these relationships suggest sediment depth may be an indicator of the eects of ne
sediment accumulation on invertebrate diversity.
-2 -1 0 1 2 3
LogDepth
0
10
20
30
40
No. taxa
-2 -1 0 1 2 3
LogDepth
0
10
20
30
40
No. taxa
-2 -1 0 1 2 3
LogDepth
0
5
10
15
20
No. EPT taxa
-2 -1 0 1 2 3
LogDepth
0
5
10
15
20
No. EPT taxa
Figure 4-23. Linear relationships (with 95% prediction condence intervals) between the number of
taxa and the number of invertebrate taxa and log-transformed sediment depth (mm).
77Sediment Assessment Methods • Section 4. Supporting information
4.5.4 Boosted regression tree model to inform reference state
An output of a parallel research project (MSI contract C01X1005) was the development of a regression
model that determines the relationship between ne sediment cover and environmental predictors.
This model was then used to predict the relative proportion of ne sediment cover in every stream
reach in New Zealand (see Appendix 6.3 for model details). Sediment data used in model development
was sourced from the Freshwater Fisheries database. The % ne sediment cover was calculated from
a bankside estimate of the relative proportion of substrate size classes. Environmental variables were
sourced from the FENZ database and included measures of land use, climate, geology, morphology
and topography as described in Leathwick et al. (2011). The resulting boosted regression tree model
had a cross-validation error of 0.67 and explained 45% of the variance in % ne sediment cover (n =
10,026), which indicate good model performance. The tted functions of explanatory variables were
as expected with ne sediment cover increasing in response to decreasing upstream average slope
and segment slope, decreasing native vegetation cover in the catchment, segment ow and ow
stability. Sediment cover increased in response to increasing mean air summer and temperature
annual variability and land-use intensication. Fitted functions from the model were used to predict
ne sediment cover in each NZREACH stream segment in New Zealand (Figure 4-24). Predicted ne
sediment cover values ranged from 0% to 100% and the national average value was 29.4%.
78 Sediment Assessment Methods • Section 4. Supporting information
Figure 4-24. Predicted contemporary sediment cover for stream segments in the New Zealand
river network. For more information on the source model see Appendix 6.3.
79Sediment Assessment Methods • Section 4. Supporting information
Figure 4-25. Predicted reference* values for ne sediment cover in the New Zealand river network.
*Reference is dened by the absence of human land-use impacts. For more information on the source
model see Appendix 6.3.
80 Sediment Assessment Methods • Section 4. Supporting information
Next, the eect of land-use variables (native vegetation cover, impervious surface cover and predicted
nitrogen concentrations as an indicator of land-use intensity) were xed and used as an oset in a
model to predict ne sediment cover in the absence of land use. The resulting boosted regression tree
model had a cross-validation error of 0.63 and explained 42% of the variance in the sediment data. The
small decrease in model performance compared to the former inclusive model provides condence
in the xed model output. The tted functions from this second model were used to predict sediment
cover for each NZREACH based solely on environmental variability, i.e., expected reference values in
the absence of land use (Figure 4-25). As expected, from the tted functions of the former model, high
levels of ne sediment cover were predicted for areas of relatively low slope, high temperature and low
rainfall, low ow and soft geology, for example, Northland, coastal Bay of Plenty, coastal Manawatu and
Hawkes Bay, and plateau areas of upland Otago.
Predicted reference values for ne sediment cover in New Zealand streams ranged from 0% to 100%
with an average value of 7.7%. A summary of modelled contemporary and modelled reference values
for sediment cover in each stream segment based on FENZ 20-level groupings further illustrates the
logical output of the model predictions (Table 4-9). As expected, small coastal to inland streams with
low gradient and low rain days had the highest predicted sediment cover (Groups A-D, F & G); but these
streams have also been subject to the highest land-use pressure and hence have the largest divergence
between predicted contemporary and predicted reference values.
Independent values from the protocol development phase were used to validate the model
predictions. The strongest correlation was between the bankside reach scale estimate of sediment
cover and the modelled observed measure: r = 0.58, p < 0.001 (Figure 4-26). The relationship with all
other estimates of % sediment were signicant (p < 0.001), but not as strong: Bankside run, r = 0.40, n =
244; In-stream visual, r = 0.29, n = 236; Wolman r = 0.42, n = 291).
The validation is good and the correlation is almost as strong as the predictive error of the model (CV
= 0.63). However, it is clear from Figure 4-26 that the model has the potential to both over- and under-
estimate sediment cover at both the high and low ends of the range in values. The linear relationship
suggests the predictive model is likely to overestimate low sediment cover and underestimate high
sediment cover.