Tracing peatland geomorphology:
sediment and contaminant movements
in eroding and restored systems
A thesis submitted to the University of Manchester for the
degree of Doctor of Philosophy in the Faculty of Humanities
2014
Emma Louise Shuttleworth
School of Environment, Education and Development
2
For my grandparents Dick, Bunty, Bill, and Eileen
3
Contents Page
CONTENTS PAGE .................................................................................................. 3
LIST OF FIGURES .................................................................................................. 8
LIST OF TABLES .................................................................................................. 12
ABSTRACT ......................................................................................................... 14
DECLARATION.................................................................................................... 15
COPYRIGHT STATEMENT .................................................................................... 16
ACKNOWLEDGEMENTS ...................................................................................... 17
CHAPTER 1 INTRODUCTION ............................................................................... 19
1.1. Blanket peat: An introduction ........................................................................................... 19
1.1.1. Formation ....................................................................................................................... 19
1.1.2. Distribution ..................................................................................................................... 20
1.1.3. Physical characteristics ................................................................................................... 20
1.1.4. Hydrology ....................................................................................................................... 22
1.1.5. Importance ..................................................................................................................... 23
1.2. Blanket peat degradation .................................................................................................. 25
1.2.1. Pressures ........................................................................................................................ 25
1.2.2. Erosion ............................................................................................................................ 30
1.2.3. Restoration ..................................................................................................................... 36
1.3. Significance of peatland geomorphology ........................................................................... 38
1.4. Research Rationale ............................................................................................................ 39
1.5. Aims .................................................................................................................................. 41
1.5.1. Objectives ....................................................................................................................... 41
4
1.6. Thesis structure ................................................................................................................. 42
1.6.1. Contributions to papers ................................................................................................. 44
CHAPTER 2 METHODOLOGY ............................................................................... 46
2.1. Field Area .......................................................................................................................... 47
2.1.1. The Peak District ............................................................................................................. 47
2.1.2. The Bleaklow Plateau ..................................................................................................... 50
2.1.3. Upper North Grain .......................................................................................................... 52
2.2. Field Techniques ................................................................................................................ 54
2.2.1. Assessing surface Pb storage using field portable XRF (Papers 1, 2, and 4) ................... 54
2.2.2. Suspended sediment sampling (2 and 3) ....................................................................... 56
2.3. Laboratory Techniques ...................................................................................................... 66
2.3.1. Environmental Magnetism (Papers 2 and 3) .................................................................. 66
2.3.2. Deriving Pb content using ICP-OES analysis (papers 2 and 3) ........................................ 68
2.3.3. Organic matter content (Papers 2 and 3) ....................................................................... 70
2.4. Data analysis ..................................................................................................................... 70
2.4.1. Manipulating geospatial data (Papers 2 and 4) .............................................................. 70
2.4.2. Modelling suspended sediment source (Papers 2 and 3) .............................................. 72
CHAPTER 3 ASSESSMENT OF LEAD CONTAMINATION IN PEATLANDS USING FIELD
PORTABLE XRF (PAPER 1) ................................................................................... 75
Abstract ......................................................................................................................................... 75
3.1. Introduction ...................................................................................................................... 76
3.2. Materials and Methods ..................................................................................................... 78
3.2.1. Field Area ........................................................................................................................ 78
3.2.2. Field Survey .................................................................................................................... 79
3.2.3. Laboratory Analysis ........................................................................................................ 80
3.2.4. Moisture correction ....................................................................................................... 81
3.2.5. Statistical analyses .......................................................................................................... 81
3.3. Results ............................................................................................................................... 84
3.3.1. Analysis Time .................................................................................................................. 84
3.3.2. Relationship between in situ and ex situ FPXRF analysis ............................................... 88
5
3.3.3. Relationship between FPXRF and ICP-OES analysis ........................................................ 88
3.4. Discussion .......................................................................................................................... 89
3.4.1. Analysis time .................................................................................................................. 89
3.4.2. Detection Limit ............................................................................................................... 90
3.4.3. Moisture content............................................................................................................ 90
3.4.4. Quality of relationship between FPXRF and acid extraction .......................................... 91
3.4.5. FPXRF as an alternative for acid extractible method ..................................................... 92
3.5. Conclusions and recommendations ................................................................................... 93
3.6. Acknowledgements ........................................................................................................... 94
CHAPTER 4 PEATLAND RESTORATION: CONTROLS ON SEDIMENT PRODUCTION
AND REDUCTIONS IN CARBON AND POLLUTANT EXPORT (PAPER 2) ................... 95
Abstract ......................................................................................................................................... 95
4.1. Introduction ...................................................................................................................... 96
4.2. Materials and Methods ..................................................................................................... 99
4.2.1. Study area....................................................................................................................... 99
4.2.2. Field measurement....................................................................................................... 101
4.2.3. Laboratory analysis ....................................................................................................... 105
4.2.4. Modelling ..................................................................................................................... 107
4.2.5. Material flux calculation ............................................................................................... 113
4.3. Results ............................................................................................................................. 113
4.3.1. Predicted source contributions .................................................................................... 113
4.3.2. Material fluxes through TIMS ....................................................................................... 116
4.4. Discussion ........................................................................................................................ 116
4.4.1. Effect of surface condition on sediment source ........................................................... 116
4.4.2. Effect of surface condition on sediment associated fluxes .......................................... 119
4.4.3. Implications for restoration and further research ....................................................... 120
4.5. Conclusions ..................................................................................................................... 122
4.6. Acknowledgements ......................................................................................................... 122
6
CHAPTER 5 CONTROLS ON THE FLUVIAL EXPORT OF SEDIMENT ASSOCIATED LEAD
AND PARTICULATE CARBON FROM ERODING PEATLANDS (PAPER 3) ................ 123
Abstract ....................................................................................................................................... 123
5.1 Introduction .................................................................................................................... 124
5.2 Field area ......................................................................................................................... 126
5.3 Methods .......................................................................................................................... 127
5.3.1 Field sampling ................................................................................................................... 127
5.3.2 Laboratory analysis ........................................................................................................... 132
5.3.3 Modelling .......................................................................................................................... 133
5.3.4 Statistical analysis ............................................................................................................. 134
5.4 Results ............................................................................................................................. 138
5.4.1 Catchment conditions ....................................................................................................... 138
5.4.2 Predicted source contributions ........................................................................................ 141
5.4.3 Relationship between sediment source and sampling height (Hypothesis 1) .................. 142
5.4.4 Relationship between sediment source and peak Q and SSC .......................................... 144
5.5 Discussion ........................................................................................................................ 145
5.5.1 Testing the hypotheses ..................................................................................................... 145
5.5.2 Organic sediment exhaustion and supply limitation ........................................................ 146
5.5.3 Evidence for a lead-flush .................................................................................................. 147
5.6 Conclusion ....................................................................................................................... 151
5.7 Acknowledgements ......................................................................................................... 152
CHAPTER 6 CONTAMINATED SEDIMENT DYNAMICS IN PEATLAND HEADWATERS
(PAPER 4) ........................................................................................................ 153
Abstract ....................................................................................................................................... 153
6.1. Introduction .................................................................................................................... 153
6.2. Field area ......................................................................................................................... 156
6.3. Methods .......................................................................................................................... 157
6.3.1. Field Survey .................................................................................................................. 157
7
6.3.2. Data Analysis ................................................................................................................ 158
6.4. Results ............................................................................................................................. 160
6.5. Discussion ........................................................................................................................ 164
6.5.1. Catchment .................................................................................................................... 164
6.5.2. Surface type .................................................................................................................. 165
6.5.3. Vegetation cover .......................................................................................................... 166
6.5.4. Wind ............................................................................................................................. 169
6.5.5. Aspect ........................................................................................................................... 170
6.5.6. Gully Depth ................................................................................................................... 171
6.6. Conclusions ..................................................................................................................... 172
6.7. Acknowledgements ......................................................................................................... 173
CHAPTER 7 SUMMARY AND CONCLUSIONS ...................................................... 174
7.1. Peatland sediment dynamics (Overarching aim) .............................................................. 174
7.1.1. Vegetation .................................................................................................................... 174
7.1.2. Sediment preparation .................................................................................................. 175
7.1.3. Meteorological conditions ........................................................................................... 176
7.1.4. Degree of degradation ................................................................................................. 176
7.2. Development of Methods (Objective 1) ........................................................................... 176
7.3. Sediment dynamics at different spatial scales (Objective 2) ............................................ 177
7.4. Implications for Restoration and Management ............................................................... 178
7.5. Further work ................................................................................................................... 179
7.5.1. Extend the use of FPXRF ............................................................................................... 179
7.5.2. Refine and extend use of mixing models ..................................................................... 180
7.5.3. Better understand the controls on sediment and pollutant dynamics ........................ 180
7.6. Tracing peatland geomorphology .................................................................................... 182
REFERENCES .................................................................................................... 183
WORD COUNT: 64,191
8
List of figures
Figure 1.1: Distribution of blanket peat (a) globally, and (b) in the UK (after Lindsay, 1995 and Tallis, 1997). Black areas show where blanket peat has been recorded while grey shading denotes areas with a climate suitable for blanket peat formation. ......................... 20
Figure 1.2: Conceptual model of the relationships between climate change, visitors, ecosystems and wildfire (after McMorrow et al., 2009). All factors listed have the potential to exacerbate/induce erosion. ............................................................................................... 28
Figure 1.3: Effects of weathering at peat surface: (a) desiccation, (b) frost action (needle ice). ......................................................................................................................................... 32
Figure 1.4: Four stages of evolution of hillslope gullies (after Bower, 1960a; adapted from Evans and Warburton, 2007). (a) Initial 'V' shaped incision; (b) 'V' shaped gully to full depth of peat; (c) Flat floored profile as lateral erosion of peat exceeds vertical erosion into mineral substrate; (d) Failure of steep sides and re-vegetation............................................ 33
Figure 1.5: Schematic diagram showing different mechanisms of aeolian transport in dry and wet conditions (after Evans and Warburton, 2007) ....................................................... 34
Figure 1.6: Example of peatland restoration strategies: (a) gully blocking, (b) reseeding bare peat surfaces (source: Moors for the Future). ....................................................................... 37
Figure 1.7: The role of geomorphology in peatland function and material flux (adapted
from Evans and Warburton, 2010). .................................................................................. 38
Figure 1.8: Thesis structure. Numbers in brackets relate to the objectives outlined in Section 1.5., indicating the Section or Chapter where these are addressed. ........................ 43
Figure 2.1 Methodological framework .................................................................................. 46
Figure 2.2: Location map of the Peak District National Park (PDNP). Red star indicates Bleaklow plateau. ................................................................................................................... 48
Figure 2.3: A typical profile of Pb deposition and storage in the Peak District (after Rothwell et al., 2005). ........................................................................................................................... 48
Figure 2.4: Location the Bleaklow Plateau relative to the industrial cities of Manchester and Sheffield. ................................................................................................................................ 49
Figure 2.5: Restoration carried out by MFF Clockwise from top left: Heather brash; spreading lime and fertiliser; plug planting; geojute (source: Moors for the Future Partnership). .......................................................................................................................... 51
Figure 2.6: Erosion-restoration continuum. Top: intact peatland; Middle: actively eroding with little vegetation cover; Bottom: re-vegetated gullies. ................................................... 51
Figure 2.7: a) Location of Upper North Grain (UNG) catchment (starred); b) aerial photograph of UNG catchment. The dense dendritic gully network is clearly visible (Pawson et al., 2008). ........................................................................................................................... 52
9
Figure 2.8: Upper North Grain gully profile showing the exposure of underlying geology at the base of the peat profile (Source: J. J. Rothwell). ............................................................. 53
Figure 2.9: Using the field portable XRF (a) in situ for Paper 4 and (b) ex situ for Paper 1. .. 55
Figure 2.10: Cross-section of two different time integrated mass flux sampler (TIMS) designs as described in: (a) Phillips et al. (2000), and (b) Owens et al. (2006). .................... 57
Figure 2.11: TIMS operating in the field: (a) the original Philips et al. (2000) design, (b) a half-sized Philips et al. (2000) stlyle design, (c) two TIMS based on the original Owens et al. (2006) design.......................................................................................................................... 59
Figure 2.12: Interval plots for parameters which produced significant differences when comparing sediment collected by the original Phillips et al. (2000) and Owens et al., (2006) TIMS designs depicting 95% confidence intervals for the means: (a) mass of sediment retained, (b) ARM, (c) SIRM. .................................................................................................. 64
Figure 2.13: Interval plots for ARM – the only parameter to produce a significant difference when comparing sediment collected by the Owens et al. (2006) TIMS adaptations. ........... 65
Figure 2.14: Interpolated surface Pb concentrations at the field sites studied in Papers 1 and 2 produced using Surfer 8.0 and TAS GIS: (a) degraded, (b) re-vegetated, (c) intact. .... 71
Figure 3.1: Study area. Grey-hatched area denotes location of sampling sites. ................... 79
Figure 3.2: Sequence of statistical analyses carried out to assess the quality of linear relationships. T-test satisfied at 0.05 confidence level.......................................................... 83
Figure 3.3: Coefficients of variation (CV) produced for peat samples containing various concentrations of Pb with increasing ex situ FPXRF analysis time. Superscript a denotes certified reference material. .................................................................................................. 84
Figure 3.4: Linear regressions of logged Pb concentrations (ppm): a) raw in situ and ex situ FPXRF; b) moisture-corrected in situ and ex situ FPXRF; c) ex situ FPXRF and ICP-OES; and d) moisture-corrected in situ and ICP-OES analyses. Regression lines are shown as solid black lines. Outliers removed from the final regression are shown as open circles. Where appropriate, graphs also display a regression line which passes through the origin (dashed black line). The 1:1 line is also shown (grey line). .................................................................. 87
Figure 4.1: Location map. Grey hatched rectangle denotes the position of the field area. .. 99
Figure 4.2: Surface condition at the three field areas: (a) shallow drainage depression at the intact field area; (b) deeply incised gullies with sparse vegetation cover at the eroding field area; (c) application of ’geojute’ at the re-vegetated field area in 2003; (d) the re-vegetated field area today. ................................................................................................................... 101
Figure 4.3: Location of suspended sediment sampling sites at the three field areas: (a) intact, (b) eroding, (c) re-vegetated. Drainage networks (black lines) were derived using TAS GIS (Lindsay, 2005). White dots represent sites where suspended sediment was collected. Red dots represent sampling sites where no suspended sediment was collected. ............ 102
Figure 4.4: Steps for deriving catchment Pb concentrations using the re-vegetated field area as an example: (a) modelled surface Pb concentrations and gully network overlay, (b) final
10
surface map, (c) close up of watershed delineation. White line represents watershed delineation. White line represents watershed, white dot represents TIMS location. ........ 105
Figure 4.5: PCA analysis showing three distinct potential sources of suspended sediment. ............................................................................................................................................. 107
Figure 4.6: Relationship between LOI derived inorganic matter content and modelled contributions from the underlying geology for a selection of suspended sediment samples: (a) includes Xlf in the model, (b) excludes Xlf from the model. The 1:1 line is shown as a dashed line. .......................................................................................................................... 112
Figure 4.7: Modelled relative contributions of individual source types to suspended sediment at the (a) intact, (b) eroding, and (c) re-vegetated field areas. ........................... 115
Figure 4.8: Relative fluxes of (a) POCTIMS and (b) PbTIMS at the three field areas. Fluxes have been given the suffix TIMS to emphasize the study specific nature of the data; the calculated fluxes are only representative of sediment passing through the TIMS, and are not a quantitative estimate at a catchment scale. .............................................................. 117
Figure 5.1: Location of field area. (a) Red star depicts location of Upper North Grain relative to the Bleaklow Plateau; (b) Arial photograph of UNG catchment (after Pawson et al., 2008). The blue star shows the location of the sampling site; the yellow star shows. ....... 127
Figure 5.2: (a) Field installation, securing TIMS to instrument bridge; (b) schematic of the operational setup (not to scale), the topmost trap is drawn in cross section, showing the polystyrene filling. ................................................................................................................ 129
Figure 5.3: Relationship between duration of TIMS inundation and mass of sediment retained. ............................................................................................................................... 130
Figure 5.4: Hypothesised patterns of suspended sediment (SS) composition collected at different stages of the hydrograph should organic or contaminated sediment become limited early in storm events (not to scale). (a) Proportion of organic/contaminated sediment reducing with sampling height; (b) higher proportions of organic/contaminated sediment retained by traps that were topmost during peak flows; (c) higher proportions of organic/contaminated sediment retained by traps that were topmost during predicted peaks in suspended sediment concentration (SSC), prior to peak flows. ............................ 137
Figure 5.5: Duration of TIMS inundation over the five sampling campaigns: (a) Summer 2011, (b) Autumn 2011, (c) Winter 2011, (d) Summer 2012, (e) Autumn 2012. TIMS1 was the uppermost to be installed, TIMS6 was at the bottom of the stack. .............................. 141
Figure 5.6: Modelled relative contributions of individual source types to suspended sediment over the five sampling campaigns: (a) Summer 2011, (b) Autumn 2011, (c) Winter 2011, (d) Summer 2012, and (e) Autumn 2012. .................................................................. 143
Figure 5.7: Desiccated peat collecting on gully floor (Source: J. J. Rothwell). .................... 148
Figure 6.1: Location of the study site. (a) The Bleaklow Plateau in relation to the industrial cities of Manchester and Sheffield. The red start denotes the gullied field area, just north of the Bleaklow summit. The blue star denotes the location of the automatic weather station. (b) View down Catchment 2 from Transect A, showing Transects B to D. Transect markersare spaced at 2 m intervals. .................................................................................... 157
11
Figure 6.2: (a) Schematic depicting mean lead concentrrationsons measured along the four transects (A-D) on the different catchment surface types (figure not to scale); (b) Spread of lead concentrations grouped by surface type. .................................................................... 160
Figure 6.3: Relationship between lead storage on gully floors and distance from gully head. ............................................................................................................................................. 162
Figure 6.4: Interval plots for factors and interactions which produced significant differences when comparing lead storage based on ANOVA depicting 95% confidence intervals for the means. .................................................................................................................................. 163
Figure 6.5: Freshly deposited peat accumulating behind tussocks of Eriophorum on the floor of Catchment 2. ........................................................................................................... 168
Figure 6.6: Schematics depicting possible explanations for the leeward lead enhancement on interfluve surfaces (not to scale). (a) Contaminated material is incrementally moved in a leeward direction across interfluves by rain splash. (b) Surface deflation is exposing different stages of the Pb depositional profile on interfluve surfaces, exposing higher concentrations on the leeward extremes. The dotted black line represented the pre-erosion surface; the black line represents the post-erosion surface; blue arrows represent sediment movement by wind; red lines represent the lead depositional profile. .............................. 170
12
List of tables
Table 1.1 Key physical properties of peat, and examples of the importance of these for understanding geomorphic processes (adapted from Evans and Warburton, 2007) ........... 21
Table 1.2: Characteristics of acrotelm and catotelm (after Ingram, 1978)........................... 23
Table 2.1: Summary of the parameters used to compare the sediment composition collected by the different TIMS designs. ............................................................................... 63
Table 2.2: Result of the t-test employed to compare the original TIMS designs. Significant parameters are given in bold. ................................................................................................ 63
Table 2.3: Result of the ANOVA employed to compare the characteristic of sediment collected by the Owens et al. (2006) TIMS adaptations. Significant parameters are given in bold. ....................................................................................................................................... 65
Table 3.1: Parameters produced by liner regression analysis which are used to assess the relationships between the various analyses. ......................................................................... 82
Table 3.2: Criteria for assigning relationship quality (adapted from Kilbride et al. 2006). ... 82
Table 3.3: Summary of parameters produced by regressions of time dependant FPXRF analysis. .................................................................................................................................. 85
Table 3.4: Statistics and quality levels for raw and moisture-corrected in situ FPXRF analysis in relation to ex situ FPXRF analysis. ...................................................................................... 86
Table 3.5: Statistics and quality levels of FPXRF analysis in relation to ICP-OES analysis. ..... 88
Table 4.1: Summary of the catchment characteristics at the three field areas. ................. 102
Table 4.2: Summary of the characteristics of the four potential sources. .......................... 104
Table 4.3: Kruskal–Wallis H-test results employed to select the fingerprint properties to distinguish the individual source types at the three field areas. ......................................... 110
Table 4.4: The results of the initial DFA employed to select an optimum composite fingerprint to distinguish the individual source types at the three field areas. 100% of the source type samples were classified correctly after the first step. Properties marked with an asterisk (*) were not included in the final model as they were already incorporated as part of the SIRM/ARM ratio......................................................................................................... 110
Table 4.5: The results of the second DFA, omitting χlf, employed to select an optimum composite fingerprint to distinguish the individual source types at the three field areas. 100% of the source type samples were classified correctly after the first step. ................. 112
Table 4.6: Summary of the raw tracer values for the properties incorporated into the optimised mixing model for the SS collected at each sampling site.................................... 114
Table 5.1: Spearman’s rank correlations for duration of TIMS inundation vs. mass of sediment retained for each of the sampling campaigns. Significant parameters are given in bold (95% confidence interval). ........................................................................................... 130
13
Table 5.2: Summary of the characteristics of the three potential sediment sources. ........ 131
Table 5.3: Results of the Kruskal–Wallis H-test and discriminant function analysis employed to select the fingerprint properties to distinguish the individual source types. Kruskal Wallis critical value at 99% confidence = 10.60. *not reported as no more parameters required to discriminate sources. ........................................................................................................... 134
Table 5.4: Summary of the conditions that characterise each sampling campaign. a Sediment collected by the bottom trap (TIMS6) was not truly representative of stormflow conditions, so was omitted from the statistical analysis and not included in this table. ....................... 139
Table 5.5: Spearman’s rank correlations for sampling height vs. modelled proportions of suspended sediment derived from the surface and the peat mass. Significant parameters are given in bold (90% confidence interval). ....................................................................... 142
Table 5.6: Spearman’s rank correlations for ftop and fSSC vs. the modelled proportions of suspended sediment derived from the surface and the peat mass for each of the five sampling campaigns. Significant parameters are given in bold (90% confidence interval). 144
Table 6.1: Summary of selected controls on peatland sediment dynamics. ....................... 155
Table 6.2: Descriptive statistics for each factor tested by the GLM. ................................... 159
Table 6.3: ANOVA results for square-root transformed data. P = probability of factor being zero and ω² = generalized proportion of variance explained. Significant results in bold. .. 161
Table 6.4: Mean lead storage on bare and vegetated surfaces (µg g-1). ............................. 162
Table 6.5: Spearman’s rank correlations for prevailing wind direction vs. Pb storage on the different catchment surfaces. Significant parameters are given in bold (95% confidence interval). ............................................................................................................................... 164
Table 6.6: Spearman’s rank correlations for mean upslope gully depth vs. Pb storage on the gully walls. Significant parameters are given in bold (95% confidence interval). ............... 164
14
ABSTRACT
Peatlands are an important store of soil carbon, play a vital role in global carbon cycling, and can also act
as sinks of atmospherically deposited heavy metals. Large areas of the UK’s blanket peat are significantly
degraded and actively eroding, which negatively impacts carbon and pollutant storage. The restoration of
eroding UK peatlands is a major conservation concern, and over the last decade measures have been
taken to control erosion and restore large areas of degraded peat. In severely eroded peatlands,
topography is highly variable, and an appreciation of geomorphological form and process is key in
understanding the controls on peatland function, and in mitigating the negative impacts of peatland
erosion.
The blanket peats of the Peak District, Southern Pennines, UK, embody many problems and pressures
faced by peatlands globally, and are amongst the most heavily eroded and contaminated in the world.
The near-surface layer of the peat is contaminated by high concentrations of anthropogenically derived,
atmospherically deposited heavy metals, which are released into the fluvial system as a consequence of
widespread erosion. Whilst not desirable, this legacy of lead pollution and its release, offer a unique
opportunity to trace peatland sediment movements and thus investigate the controls on sediment and
contaminant mobility.
A suite of established field, analytical and modelling techniques have been modified and adapted for use
in peatland environments: (i) by incorporating a simple correction for moisture content, field portable
XRF has been shown to be an accurate, cost-effective, and rapid tool for assessing in situ lead
concentrations in wet organic sediments; (ii) a lightweight time integrated mass flux sampler has been
developed for deployment at multiple remote peatland field sites, and has been used to explore spatial
and temporal suspended sediment dynamics; and (iii) sediment source fingerprinting and numerical
mixing models, traditionally used to determine sources of fine sediment in minerogenic systems, have
been used to investigate suspended sediment composition in contaminated organic rich catchments.
These modified methods have been successfully employed in combination to address issues of sediment
and contaminant release.
Several mechanisms and controls have been shown to be important influences on sediment dynamics
and Pb release across a range of spatial and temporal scales: (i) the presence of vegetation is key in
stabilising the peat’s surface and trapping mobilised sediment; (ii) sediment preparation influences the
timing of POC and Pb release; (iii) antecedent water tables may control the timing and the nature of
sediment entering the fluvial system during storm events; and (iv) the degree of degradation influences
both Pb storage and release. At the landscape scale, peatland restoration significantly mitigates sediment
production in eroding peatlands and substantially reduces carbon and pollutant export. At the catchment
scale, sediment preparation and hydrological connectivity are important controls on the magnitude and
timing of sediment and lead fluxes from eroding peatland catchments. At the plot scale, complex small
scale spatial patterns of contaminant storage in eroding headwater catchments can be explained by
interactions between topographic setting and vegetation cover, and the mobilisation of sediment by wind
and water.
This deeper understanding of the multi-scalar dynamics of sediment movements in eroding peatlands is
important in the context of: (i) the release and reworking of legacy contamination in organic rich systems;
(ii) the response of blanket peats to climate change; (iii) informing future restoration strategies that aim
to manage peatland sediment and contaminant fluxes.
15
Declaration
No portion of the work referred to in this thesis has been submitted in support of an application
for another degree or qualification of this or any other university or institute of learning.
Emma Louise Shuttleworth
16
Copyright Statement
i) The author of this thesis (including any appendices and/or schedules to this thesis)
owns certain copyright or related rights in it (the “Copyright”) and s/he has given
The University of Manchester certain rights to use such Copyright, including for
administrative purposes.
ii) Copies of this thesis, either in full or in extracts and whether in hard or electronic
copy, may be made only in accordance with the Copyright, Designs and Patents Act
1988 (as amended) and regulations issued under it or, where appropriate, in
accordance with licensing agreements which the University has from time to time.
This page must form part of any such copies made.
iii) The ownership of certain Copyright, patents, designs, trademarks and other
intellectual property (the “Intellectual Property”) and any reproductions of
copyright works in the thesis, for example graphs and tables (“Reproductions”),
which may be described in this thesis, may not be owned by the author and may be
owned by third parties. Such Intellectual Property and Reproductions cannot and
must not be made available for use without the prior written permission of the
owner(s) of the relevant Intellectual Property and/or Reproductions.
iv) Further information on the conditions under which disclosure, publication and
commercialisation of this thesis, the Copyright and any Intellectual Property and/or
Reproductions described in it may take place is available in the University IP Policy
(see http://documents.manchester.ac.uk/DocuInfo.aspx?DocID=487), in any
relevant Thesis restriction declarations deposited in the University Library, The
University Library’s regulations (see
http://www.manchester.ac.uk/library/aboutus/regulations) and in The University’s
policy on Presentation of Theses
17
Acknowledgements
Firstly, a massive thank you to my supervisors, Martin Evans and James Rothwell for still
taking me on after I turned up an hour late to our first meeting, and for all of their support,
understanding, inspiration, and endless advice over the last five years. To Martin for his
calming influence, unending knowledge, and for helping me see the big picture; for our
shared dislike of bureaucracy and paperwork and our shared enthusiasm for
geomorphology. I’m sure I’ll be asking myself: “WWMD?” (What would Martin do?) for
years to come. To James, my cardi hero, for his attention to the finest of details, and all his
help in my development as a researcher and educator; for the bridge building, kick
sampling, sheep-poop throwing, and Casio wearing; for the windups, the Guinness, the
shots and the putting the world to rights. It’s been a privilege to be your first PhD student.
Equal thanks also go to Simon Hutchinson, my co-author and unofficial third supervisor, for
all of the analytical opportunities, support, cake and guidance that he has provided. I’m
looking forward to many more exciting methodological developments together in the
future.
My eternal undying gratitude goes to my parents, for all of their love and support over the
last 31 ½ years; for putting up with all of the false-starts, dead-ends, U-turns, and late night
phone calls; for trying to hide their glazed looks when I talk about what I’ve been up to for
the last five years; for all of the cat sitting, long-distance visiting, and endless supply of
Marks and Spencer’s food. Mum and Dad – I love you.
Thanks to my fellow postgrads and staff in Geography for all of the comradery, support and
laughs. Special mentions go to Beth Cole and Claire Goulsbra, my fellow Martin’s Angels,
for welcoming me to the department; Lisa Ficklin, Ioanna Tantanasi and Danielle Alderson
the never-ending supply of positivity; to Fiona Smyth, Sarah Hall, Jason Dortch and the
gone-but-not-forgotten Jeff Blackford for always checking in on me; to John Moore, Jon
Yarwood and Pete Ryan for all of their help in the lab, but mainly for their friendship and
support when times were tough; to Helen Wilson and Jonny Darling for cat sitting and help
with the job application; and finally to my fellow writing-uppers, Mark Usher and Jana
Wendler, for making the final few months of incarceration in the Arthur Lewis Building
bearable (and dare I say enjoyable?!).
18
To my Liverpool buds: Sarah Kneen, Jack Dods, and Dan Schillereff; and to the Manchester
escapees: RfC, Fosb and Anna– your friendship has got me through some dark times and for
that I am forever grateful. I’m looking forward to some massive catch-ups in the near
future and to picking up our adventures where we left off before the PhD took over.
Special thanks to Mrs Peppin for introducing me to the wonderful world of soil science;
Peter James for all of his advice and encouragement back in Liverpool that set me off on my
career in academia (especially for the beautiful reference he wrote me for my PhD); and to
the BSG for all of the career benefiting opportunities they have provided, and for making
me feel like a valued member of a supportive research community.
Big thanks also go to all of the fieldwork helpers, cake bakers, silly dancers, brew makers,
cat sitters, and everyone who has helped me make it to the end. You are too numerous to
mention but all equally important.
And finally, to the three most important men in my life: to Jamie Brewster for his support
and encouragement at the beginning, to Toby Cat for his welcome appearance half way
through, and to Gareth Clay for his support and seemingly limitless patience at the end.
Your existence means so much more to me than I can express in mere words. Thank you.
19
Chapter 1 Introduction
1.1. Blanket peat: An introduction
There are 353.4 M ha of peatland worldwide, most of which (85.9%) can be found in the
temperate zone of the northern hemisphere (Moore, 2002). In the British Isles, peat covers
approximately 8% of the total land surface (Francis, 1990). Most of this takes the form of
blanket peat, with the UK and Ireland supporting 15% of the world’s total resource of this
type of land cover (Tallis, 1997). The term blanket bog was first used by Tansley (1939) to
describe widespread ombrotrophic mires which follow the underlying topography like a
blanket (Evans and Warburton, 2007). Being ombrotrophic they receive all of their water
and nutrients from precipitation only (Bragg and Tallis, 2001) and are typically acidic and
nutrient poor. They occur in climates with precipitation excess, on flat or gently sloping
ground, where drainage is impeded by low soil hydraulic conductivity (Ingram, 1982), and
may incorporate former basin mires in topographic low points and areas of raised mire
formed on summits or interfluves (Evans and Warburton, 2007). In the UK, peatlands face
threats from pressures such as climate change, legacy atmospheric pollution, poor
management and anthropogenic disturbance (Bonn et al., 2009). Consequently, over the
last 1000 years a significant proportion of the UK’s blanket peat has become degraded and
is actively eroding.
1.1.1. Formation
Peat is an accumulation of the partly decomposed or un-decomposed remains of plant
material formed where permanently high water tables limit decomposition (Evans, 2009).
Peat formation is a dynamic process, involving upward growth of vegetation which buries
older growth. The buried vegetation dies but waterlogged anaerobic conditions prevail in
the majority of the peat profile due to high water tables so decomposition is retarded. As
such, dead organic matter is able to accumulate; forming thick (several metres) deposits of
partially decomposed organic sediments. These sediments remain rich in organic
compounds and can become sinks of elements such as carbon and nitrogen (Charman,
2002).
20
1.1.2. Distribution
The distribution of blanket mire is closely controlled by climate, requiring a positive water
balance to enable growth. Lindsay (1995) listed the four key environmental conditions
blanket bog formation requires:
>1000 mm rainfall annually;
>160 wet days (days receiving over 1 mm rain) per year;
A mean temperature < 15 °C for warmest month;
Little seasonal variability in temperature.
They are therefore restricted to a few regions with wet, oceanic climate influence, in
latitudes greater than 40˚ north and south with the exception of the area of bog forming in
the Ruwenzori Mountains, Uganda (Figure 1.1a). Although blanket peat occurs widely in
the UK and Ireland (Figure 1.1b), elsewhere it is less extensive. Instances can be found
along the western coastlines of Scandinavia and Canada and Northern Japan, and in a few
locations within similar southern-hemisphere latitudes, including Tasmania, New Zealand
and the Falkland Islands.
Figure 1.1: Distribution of blanket peat (a) globally, and (b) in the UK (after Lindsay, 1995 and Tallis, 1997). Black areas show where blanket peat has been recorded while grey shading denotes areas with a climate
suitable for blanket peat formation.
1.1.3. Physical characteristics
Evans and Warburton (2007) note that the unusual (and heterogeneous) properties and
behaviour of peat pose a great challenge to understanding the erosive processes affecting
the peat’s surface, and that an appreciation of the physical characteristics of peat as an
earth material is required. Some of the key properties and their importance are listed in
Table 1.1.
21
Peat Property Importance Reference
Basic Properties
Water content of peat The water content of peat can vary from about 200% to > 2000% of dry weight. The ability to store large volumes of water is the most striking characteristic of peat.
Hobbs (1986)
Permeability and hydraulic conductivity
Permeability is a fundamental property controlling water movement and consolidation in peat. Permeability decreases markedly with depth with the abrupt transition from the acrotelm (aerated upper surface layers) to the denser catotelm (lower layers). Hydraulic conductivity may vary up to eight orders of magnitude between the layers.
Ingram (1983)
Bulk density The degree of decomposition and peat bulk density are intrinsically related. Decomposition decreases pore size. Bulk densities are low and variable, but tend to increase with increasing depth as underlying peat is compressed by the weight of the overlying layers.
Eggelsmann, et al. (1993) Clymo (1983)
Gas content Gas content in peat may be as large as 5% of the volume. Most of this is free gas which influences permeability, consolidation and loaded pore pressures.
Hanrahan (1954)
Organic (carbon) content A high organic content is an intrinsic property of peat. Typically carbon contents of peat are approximately half the organic matter content which has important implication for terrestrial carbon stores.
Worrall et al. (2003)
Micromorphology of peat Important for water flow and rewetting in peat and secondary compression of the peat mass Mooney et al. (2000)
Hydrogen ion activity and pH Soil water pH is strongly correlated with vegetation and peat type and the chemistry of the water supply. Values range 3.5 to 6. Organic peat acid can be associated with weakened of peat slopes
Söderblom (1974)
Geotechnical Behaviour
Geotechnical behaviour – standard index properties
Standard index (consistency) tests are not easily applied to peat material. Liquid limits are useful in characterising certain types of peat but plasticity tests cannot easily be applied due to a lack of mineral clay.
Hobbs (1986) Carlsten (1993)
Stress-strain – primary and secondary consolidation
By virtue of a very high water content peat is an extremely compressible material. Rapid consolidation is followed by secondary compression which is the dominant process.
Fox and Edil (1996)
Changes in mechanical properties with organic content
No systematic relationship exist between mechanical properties and organic matter – soils behave in a complex manner due to differences in the amount and type of organic matter present
(Farrell et al., 1994).
Flowing properties of peat slurry.
Liquefaction of basal peat deposits, transport of material in peat mass movement runout and transfer in river systems Luukkainen (1992)
Peat Creep Slope instability and surface rupturing Carling (1986)
Shrinkage and desiccation Peat is susceptible to shrinkage due to high water content. Desiccation cracking may promote delivery of surface water to the subsurface hydrological system promoting elevated pore pressures and peat mass failure
Hendrick (1990)
Thermal behaviour Peat and other organic materials behave very differently in the cold: dry peat has a very low thermal conductivity due to high air content; wet saturated peat can have 5 x higher thermal conductivity; whilst frozen peat 28x higher
Seppälä (2004).
Table 1.1 Key physical properties of peat, and examples of the importance of these for understanding geomorphic processes (adapted from Evans and Warburton, 2007)
22
Peat is a complex system comprising solid, liquid and gaseous fractions which interact
through changing position, volume and mass, and as such is not a homogeneous substance
(Egglesman et al., 1993). Evans and Warburton (2007) describe a ‘typical’ peat composition
of 85% water, 8% organic remains, 5% air and 2% mineral material (by volume), but the
physical composition of peat can vary greatly depending on the type of plant material
which makes up the organic matter, and the degree of decomposition. Sphagnum mosses
tend to dominate, but organic remains can also be made up of varying quantities of other
mosses, sedges, and woody Ericaceous shrubs. The leaves and stems of bog mosses such as
Sphagnum are decomposed more rapidly under oxidizing conditions compared to material
derived from harder woody dwarf shrubs, causing variations in the bulk density of the peat
(Moore, 1987). The composition of plant remains and the degree of decomposition also
influence porosity; organic residues both contain pores and form pores, but increased
levels of decomposition, increase the volume of the inter-residual pores and decrease the
volume of the intra-residual pores (Egglesman et al., 1993). The von Post classification (von
Post, 1924), based on a visual inspection of plant remains, humification and water content,
provides a semi-quantitative means of describing the physical and structural properties of
peat.
1.1.4. Hydrology
The hydrology and geomorphology of blanket peats are intimately linked (Evans and
Warburton, 2007), so an appreciation of hydrology is essential to fully understanding
functions and processes in peatland systems (Eggelsman et al., 1993). The very existence of
blanket peatlands is dependent on the local water balance and the unique hydraulic
properties of peat, and there is an extensive body of work on the subject (e.g. Tallis, 1973;
Damman, 1986; Evans et al., 1999; Holden and Burt, 2002a; Holden, 2005). Peatland
growth and development is controlled by local hydrology and the topography of the pre-
existing landscape. Peat begins to accumulate in areas of lower relief and reduced
drainage, and drainage pathways can be inherited from the pre-peat topography (Evans
and Warburton, 2007). Vertical changes in peat properties have a major control on
peatland hydrology and function. Ingram (1978) identified two distinct layers within the
peat profile: upper active ‘acrotelm’ peat layer with a high hydraulic conductivity and
fluctuating water table, and a more inert lower ‘catotelm’ layer, which corresponds to the
permanently saturated main body of peat (Holden, 2005). The key descriptors of this
“diplotelmic mire hypothesis” are summarised in Table 1.2.
23
Runoff plays a key role in the movement of sediment through peatland systems so is of
fundamental importance to peatland geomorphology. Blanket peats are highly productive
of runoff (Holden and Burt, 2002a); high water tables cause the peat to saturate rapidly
and saturation-excess overland flow dominates. Hydrographs produced by stormflow in
peatland systems are often termed ‘flashy’, in reference to a rapid increase in discharge
following the onset of precipitation which produces a sharp peak before a return to levels
just above baseflow soon after the cessation of rainfall (Evans and Warburton, 2007).
Consequently, much of the annual flow-, and therefore sediment-, regimes of peat
catchments are dominated by storm events. Crisp (1966) estimated that 80 to 90 % of
sediment flux occurred within two hours of peak discharge. Fluvial action is the dominant
mechanism controlling sediment flux and erosion in degraded systems, and once gully
networks are established, sediment is rapidly and efficiently evacuated from the system.
Acrotelm Catotelm
Water table Oscillating Continuously saturated
Aerobic status Periodically aerobic Anaerobic
Moisture content Variable Continuously saturated
Hydraulic conductivity High Very low
Exchange of energy and
matter
Rapid Slow
Microbial activity High – aerobic and anaerobic Low - anaerobic
Table 1.2: Characteristics of acrotelm and catotelm (after Ingram, 1978).
1.1.5. Importance
Globally, blanket peatlands support a variety of ecosystem services making them an
important economic, scientific and recreational resource. Humans have managed British
uplands since the Mesolithic period (10 - 4 k yr BC) (Warburton, 2003) and today upland
blanket peats represent some of the most heavily managed environments in Britain. Tallis
(1998) estimated at least 82% of British blanket mire was substantially modified as a result
of management.
1.1.5.1. Resource management
According to Stewart and Lance (1991) over half of UK peat has been drained to improve
grazing and hunting and to prepare land for afforestation. British uplands support around 3
24
million sheep (Natural England, 2009) and 9% has been afforested (Cannell et al., 1993).
Globally, 15 million hectares of peatlands have been drained for timber production, more
than 90% of which has taken place in Fennoscandia and Russia (Moore, 2002). Up to 40% of
English moorland has received some burn management (Worrall et al., 2009), mainly to
optimise red grouse habitat. In England and Wales alone, approximately 400,000 ha of
moorland produce income from grouse shooting contributing £192 M to the UK upland
economy (Ramchunder et al., 2009). British uplands encompass seven National Parks and
nine Areas of Outstanding National Beauty (AONB). As such they are popular destinations
for outdoor recreational activities such as hiking and climbing. These activities benefit not
only visitors to these areas (through improved health and access to nature) but the income
from tourism is also an important contributor to local economies. The headwaters of many
major rivers drain areas of upland blanket peat. Around 70% of Britain’s drinking water
comes from upland catchments (Natural England, 2009). Reservoirs in the Peak District
alone supply 450 million litres of water a day to surrounding urban areas (Bonn et al.,
2009). Although no longer common practice in the UK, peat can be harvested as a source of
fuel. The countries of the former U.S.S.R. account for approximately 95% of the peat mining
world-wide with most of it utilized for electricity (Hartig et al., 1997).
1.1.5.2. Ecology
Blanket peatlands also contain some globally rare plant species (e.g. Scirpus cespitosus,
Erica tetralix, Calluna vulgaris, Eriophorum vaginatum and Molinia caerulea) (Ramchunder
et al., 2009), are an important breeding ground for a diverse mix of bird species (Thompson
et al., 1995) and their small acid streams have distinct invertebrate assemblages (Eyre et
al., 2005). Although blanket peats are often cited as lacking biodiversity, their fauna and
flora are distinctive and many groups are confined to this habitat (Moore, 2002). Where
growing conditions have been conducive, peats can contain a stratigraphic record of fossil
remains of plants which provide information on successional development (e.g. Bradshaw
et al., 2005), changing hydrological conditions (e.g. Reid and Thomas, 2006), changing
nutrient status (e.g. Mighall et al. 2009) and climatic shifts (e.g. Mauquoy et al., 2002),
upon which projections into the future can be based.
1.1.5.3. Carbon storage
Peat-forming ecosystems are essentially unbalanced in their carbon budgets; carbon is
fixed into plant matter during photosynthesis but due to slow decomposition rates in the
catotelm the carbon is ‘locked up’ in the peat mass leading to a net surplus. Peatlands have
25
accumulated C at an average rate of 0.96 M tonnes C yr-1 throughout the Holocene (Worrall
and Evans, 2009), making them a substantial store and potential sink of atmospheric C.
Gorham (1991) estimated that globally peatlands contain about 455 Gt of carbon,
representing 20-30% of all terrestrial carbon on earth and in the UK, peatlands are the
single largest carbon store (Cannell et al. 1993) storing around 3 billion tonnes of carbon.
1.1.5.4. Lead sink
Anthropogenic lead (Pb) pollution has long been recognised as a global phenomenon
dating back more than 3,000 years (Lee and Tallis, 1973). Ombrotrophic peatlands are
highly sensitive to atmospheric deposition (Shotyk et al., 1998), and peatland soils in close
proximity to urban and industrial areas can be contaminated with atmospherically
deposited heavy metals. The strong complexation of Pb to organic matter (OM) (Stevenson
1976; Vile et al. 1999) means that peatlands can represent significant sinks of Pb (Shotyk et
al., 2000; Bindler et al., 2004; Farmer et al., 2005; Rothwell et al. 2007a, 2010a). Peat cores
can be used to reconstruct long-term Pb deposition and pollution histories as peatlands
retain a record of atmospheric metal deposition (e.g. Lee and Tallis 1973; Shotyk et al.,
1998; Marx et al. 2010).
1.2. Blanket peat degradation
Upland environments in the UK face many pressures which in turn drive related changes;
most pressingly by degrading the landscape and increasing the area of peatland affected by
erosion. The consequences of upland degradation are diverse and often detrimental to the
functioning of ecosystem services which subsequently affects the economic value of these
marginal areas (Bonn et al., 2009).
1.2.1. Pressures
1.2.1.1. Climate Change
Blanket peat growth relies on a positive water balance maintained by high rainfall, and so
its distribution is closely controlled by climate (Evans and Warburton, 2007). The 2007 IPCC
report identifying uplands as particularly vulnerable to climate change (Parry et al., 2007),
and there is historical evidence that previous phases of peat erosion may have been
initiated during prolonged dry periods, especially when these are followed by periods of
heavy rainfall (Tallis, 1997). Climate-change scenarios proposed by Hulme et al. (2002)
suggest that in future decades the UK will experience warmer, drier summers and stormier
26
winters. Summer drought will lead to a reduction of water tables and desiccation of the
peat surface. Once desiccated, peat becomes hydrophobic and is hard to re-wet; the
surface layers crack and disaggregate making it available for transportation (Evans and
Warburton, 2007). When combined with increased winter storm activity and runoff, this is
likely to increase the rate of peat erosion.
Clark et al. (2010) assessed the vulnerability of blanket peat to climate change in Great
Britain based on climate and greenhouse gas emission projections detailed by Hulme et al.
(2002), and suggest that there will be a long-term decline in the distribution of actively
growing blanket peat, although it is emphasised that existing peatlands may well persist for
decades under a changing climate. Peatlands have played an important role in sequestering
atmospheric carbon over the Holocene (Yu, 2011) and are currently a large store of carbon
(Gorham, 1991, Joosten, 2009). However, Billett et al., (2010) suggest that current
accumulation rates are some of the lowest seen over the last 150 years. Lloyd and Taylor
(1994) and Waddington et al. (1998) predict an increase in CO2 respiration accompanying
the lower water tables associated with climate change, as more peat will be exposed to
aerobic decomposition, and methane production has been directly related to peatland
water table depth (Worrall et al., 2007). Severe droughts and subsequent re-wetting
cycles can lead to increased carbon losses as dissolved organic carbon (Freeman et al.,
2001) and drought-triggered decomposition cascades, leading to destabilisation of the
carbon stock, could apply to up to 60% of all peatlands (Fenner and Freeman, 2011)
1.2.1.2. Mismanagement
The upland blanket peats of the UK have traditionally been heavily managed for low
density farming, energy, forestry and game rearing (Ballard et al., 2011). Good
management practice can enhance ecosystem services and as such is sympathetic to the
landscape (e.g. traditional hay meadow management for biodiversity), but many peatlands
are subject to management systems that have not always been conducive to carbon
storage (Holden et al., 2007), and unsympathetic management can lead to a reduced plant
cover and the exposure of bare peat.
Moorlands cannot normally sustain large grazing densities; most heather grows only when
grazing is below 2 sheep ha−1 (Holden et al., 2007). The number of sheep in Great Britain
rose from around 8 million in the 1860s to a peak of 44 million in 1993 (Sansom, 1999).
Common Agricultural Policy subsidies in the 1970s and 1980s resulted in increased
stocking; 29% of moors were stocked above sustainable levels in 1977 and by 1987 this had
27
increased to 71% (Holden et al., 2007). The impact of sheep grazing is most directly
expressed through the formation of scars in the surface cover, which lead to local
movement of soil material (Evans, 1997). Reductions in grazing pressure can sometimes
result in recolonisation of eroded scars, but in many peat catchments erosion can often
continue if unchecked by human intervention (Holden et al., 2007).
Fire is used to maintain dwarf shrub habitats, mainly for grouse shooting and, to a lesser
extent, to improve grazing. In most instances, controlled fires are burnt with the wind to
facilitate propagation. These are known as ‘cool burns’ (Crowle and McCormack, 2009),
which if carried out successfully remove vegetation but leave the moss layer intact. When
burning is carried out against the wind (‘back-burning’), the rate of spread is much slower
and a more intense fire is produced (‘hot burn’) (Hobbs and Gimingham, 1987) removing all
vegetation and posing similar threats to the environment as listed for wildfire in Section
1.2.1.4. Yallop et al. (2006) note the long recovery period for this type of burning leaves the
peat surface exposed and at risk of erosion and deflation for up to 7 years.
Peat is commonly drained in many European countries in order to lower water tables to
improve grazing and hunting or to prepare land for afforestation, and many governments,
including the United States and the United Kingdom, have historically subsidized the
drainage of wetlands to increase national crop yields (Hartig et al., 1997). Fifteen million
hectares of peatlands have been drained for timber production in the boreal and
temperate zones, more than 90% of which has taken place in Fennoscandia and Russia
(Paivanen 1997). Britain is one of the most extensively drained lands in Europe
(Ramchunder et al., 2009) with more than half the agricultural activity in Britain occurring
on land that has been drained (Holden et al., 2004). About 190 000 ha of deep peatland
have been afforested with coniferous plantations since 1945 (Cannell et al., 1993).
However, Stewart and Lance (1983) demonstrated that there was no evidence that
peatland draining fulfils the claims made for it, and as such the economic benefits are very
low and yet the potential environmental effects high (Holden et al., 2004).
1.2.1.3. Anthropogenic disturbance
The countryside is increasingly used for recreation, EU funding has been made available to
support farmers wishing to diversify into activities such as tourism (European Agricultural
Guidance and Guarantee Fund) (Holden et al., 2007). However, walkers and hikers are
among the most widespread and persistent causes of anthropogenic disturbance. Repeated
trampling leads to significant changes in the plant community and in some cases loss of
28
plant cover all together (Pearce-Higgins and Yalden, 1997). The climate change scenarios
discussed in Section 1.2.1.1. predict warmer, longer summers (Hulme et al., 2002) and
milder winters which are likely to increase tourist numbers (McMorrow et al., 2009) leading
to concern at the likely effect of such activities upon the landscape (Pearce-Higgins and
Yalden, 1997).
1.2.1.4. Wildfire
Wildfires can occur accidentally, as a result of arson or when managed burns get out of
control (Bruce, 2002). Unlike well managed burns (which should not burn into the litter or
soil layer), many wildfires burn for longer and at hotter temperatures, deep into peat
profile. This has a range of negative consequences, such as: exposure of the peat surface,
creating long lived fire scars which are vulnerable to erosion (Anderson et al., 2009);
alteration of the structure and hydrology of the acrotelm (Holden et al., 2007); increases in
sediment flux (Tallis, 1987); and increased exports of DOC, CO2 and CH4 (Dawson and Smith,
2007).
Figure 1.2: Conceptual model of the relationships between climate change, visitors, ecosystems and wildfire (after McMorrow et al., 2009). All factors listed have the potential to exacerbate/induce erosion.
29
Wildfires occur more frequent in years of severe drought (McMorrow et al., 2009), and the
projected increases in mean annual temperature may lead to an increase in wildfire
frequency in peatlands (Turetsky et al., 2006). Warmer winters and summers will lengthen
the thermal growing season for plants, producing more biomass to fuel the burn (Running,
2006), and water tables will take longer to recharge after prolonged dry
periods(McMorrow et al., 2009), increasing the length of the fire season. Furthermore,
warmer weather is likely to increase tourist numbers which in turn increases the risk of
outbreaks as visitors are the main ignition source through negligence or arson. The
relationship between climate change, visitors and fire risk is summarised in Figure 1.2.
1.2.1.5. Pollution
There have been significant changes to atmospheric chemistry across the UK over the past
few hundred years (Holden et al., 2007); sources include fossil fuel combustion, heavy
industry and (more recently) vehicle emissions (Rothwell et al., 2005). As blanket mires are
ombrotrophic, they are particularly sensitive to atmospheric deposition.
SO2 (sulphur dioxide) emitted as a by-product of fossil fuel combustion oxidises to H2SO4
(sulphuric acid) and falls as acid rain contributing to the acidification of peatlands (Holden
et al., 2007). This poses several problems including a decline in Sphagnum (Ferguson et al.,
1978; Tallis, 1985 and 1987), limited recolonisation of bare peat areas (Bell, 1973; Mackay
and Tallis, 1996), and the acidification of waters draining from peatland soils (Caporn and
Emmett, 2009). Additionally, Rothwell et al. (2006) demonstrated stored heavy metals can
desorb from contaminated sediment into acidic stream water, potentially further impacting
downstream water resources.
Industrial and agricultural activities contribute to an increasing load of ammonia and
nitrate compounds in the atmosphere (Lee, 1998). Plant species associated with
ombrotrophic habitats are not by nature nutrient-demanding and can easily become
ousted by more competitive species if the supply of nutrients is increased (Moore, 2002),
and elevated levels of nitrogen (and sulphur) pollutants have also been shown to have a
detrimental effect on the survival and growth of Calluna vulgaris seedlings (Mackay and
Tallis, 1996).
30
1.2.2. Erosion
1.2.2.1. Distribution
Figure 1.1b shows the distribution of blanket peat in the UK; between 30 and 74% of the
peat’s surface area is affected by gullying, depending on region (Evans, 2009). Outside of
the UK, peat erosion occurs on a much smaller scale and seems to be largely influenced by
localised disturbance or particular environment impacts. For example:
Campbell et al. (2002) investigated erosion and surface stability in abandoned
milled peatlands in Quebec, Canada;
Foster et al. (1988) describe minor gullying in the peatlands of Labrador and
Sweden;
Hartig et al. (1997) discuss the sensitivity of eastern European peatlands to
projected climate change;
Luoto and Sepala (2000) hypothesise about the involvement of wind action on the
development of 'peat cakes' in Finland.
1.2.2.2. The onset of erosion
There is much uncertainty over the trigger(s) of the onset of peat erosion. Throughout the
latter half of the Twentieth Century, Tallis built up an extensive body of work on the subject
(e.g. Tallis, 1973, 1985, 1987) however there remains no clear agreement on the causes of
blanket peat erosion. Tallis (1997) concludes that the ubiquitous presence of gully systems
in British upland blanket mires suggests the necessity of a wide-ranging mechanism of
formation. However, there may be a variety of triggers working either alone or in
combination.
Sedimentary evidence from several sites around the UK and Ireland indicate dates of
erosion initiation are spread over the last 3000 years (Labadz et al., 1991; Mackay and
Tallis, 1996; Tallis, 1997 and 1998). Some early erosion has been attributed to prehistoric
disturbance, such as forest clearance at the peat margin. This may have altered catchment
hydrology by lowering the base water-level and imparting extra erosive energy, thereby
triggering stream incision and land slippage (Ellis and Tallis, 2001) and the subsequent
headward extension of streams into the peat blanket (Tallis, 1995). Some theories attribute
the onset of major erosion episodes to historic wildfire (Mackay and Tallis, 1996) or historic
human-induced fire (Tallis, 1987).
31
Many gully systems in British blanket peats are several hundred years old, and thus may
pre-date intensive human exploitation (Tallis, 1998) so a purely natural origin for some
gully systems in peat cannot be ruled out. Peat erosion may represent a natural end point
to peat accumulation; peat depths may cross a threshold after which saturated peat
becomes unstable (Tallis, 1998). Some gully systems are thought to have originated from
the linkage and drainage of pool systems (Bower, 1960a; Tallis and Livett, 1994) or the
collapse of pipe systems within the peat (Bower, 1960a; Holden et al., 2006) while shifts in
climate could also be an underlying cause (Stevenson et al., 1990). Tallis (1995) linked the
inception of gully systems in the Southern Pennines to the dry conditions of the Early
Mediaeval Warm Period, (ca. AD 1100–1250) while Stevenson et al., (1990) found evidence
of enhanced peat erosion between AD 1530 and 1690 throughout UK which coincide with
the harsher and wetter conditions of the Little Ice Age (ca. AD 1500–1850).
More recently, intensification of grazing and burning have been implicated as contributing
factors to erosion under the marginal climatic conditions at the edge of the blanket mire
distribution range in the southern Pennines (Bragg and Tallis, 2001). Ellis and Tallis (2001)
also suggest that the effects of human deforestation of upland hillslopes on hydrology in
Britain, further augmented by climatic and modern anthropogenic factors provides a
plausible explanation for blanket mire erosion. Both theories point to a combination of
factors influencing the erosive process.
1.2.2.3. Mechanisms of erosion
Peat erosion is negligible below an established plant cover; bare peat, by contrast, is readily
erodible (Bragg and Tallis 2001). It seems therefore, that bare peat is a prerequisite for
erosion to take place. The exposed uppermost region of the peat mass gradually loses its
structural cohesiveness through frost action and desiccation (Figure 1.3) (Tallis, 1973;
Francis, 1990; Labadz et al., 1991). This loose sediment has a very low density which can be
removed from bare peat surfaces by one of three key mechanisms:
through the action of running water (e.g. Labadz et al., 1991; Evans et al., 2006);
wind (Tallis 1997, Warburton, 2003; Foulds and Warburton, 2007a, 2007b);
chemical oxidation (Francis, 1990; Waddington and McNeil, 2002; Evans and
Warburton, 2005).
32
Figure 1.3: Effects of weathering at peat surface: (a) desiccation, (b) frost action (needle ice).
1.2.2.3.1. Fluvial erosion
Blanket peat catchments are highly productive of surface and near surface runoff through
the acrotelm (Evans et al. 1999; Holden and Burt 2002a). This is the main process by which
peat is removed from bare surfaces and transferred to the fluvial system. Running water is
the dominant mechanism in initial stripping of vegetation cover along drainage lines and
the initiation of gully erosion (Evans and Warburton, 2007). The underlying cause of gully
network development is unclear (see Section 1.2.2.2), but once they begin to develop,
fluvial erosion of the peat is rapid, producing deep, extensive systems. Bower (1960a)
identified four stages in the development of gullies from shallow ‘v’ shaped channels to
flat-floored ‘u’ shaped profiles (Figure 1.4). Bower (1960a) also identified two dominant
patterns of gully erosion. Type I is characterised by frequently branching dendritic channels
with a high drainage density. These are associated with lower slopes less than 5°. Type II
channels are straighter and unbranched, aligned normal to the slope on steeper ground
with a lower drainage density.
33
Figure 1.4: Four stages of evolution of hillslope gullies (after Bower, 1960a; adapted from Evans and Warburton, 2007). (a) Initial 'V' shaped incision; (b) 'V' shaped gully to full depth of peat; (c) Flat floored
profile as lateral erosion of peat exceeds vertical erosion into mineral substrate; (d) Failure of steep sides and re-vegetation.
1.2.2.3.2. Wind erosion
Despite wind erosion being documented as an important factor in the degradation of
upland peat (Bower, 1960a; Radley, 1962, 1965; Barnes, 1963; Evans and Warburton,
2007), aeolian processes of peat erosion have received relatively little attention. This is
particularly surprising given the exposed location of many peatlands, and peat’s low bulk
density when compared to other soil types (Egglesman et al., 1993). However, recently
Warburton (2003) and Foulds and Warburton (2007a and 2007b) have provided the first
quantitative measurements of the rates and process dynamics of the removal of material
by wind action. Warburton (2003) cites wind-assisted splash (Figure 1.5) as the dominant
wind erosion process in peatlands, transporting peat particles over relatively short yet
significant distances (1 to 10 m per event). Foulds and Warburton (2007a) show that
considerable transport can also occur as dry blow of ‘peat dust’ (Figure 1.5). All three
studies show that the dominant direction of peat flux was closely aligned with the
prevailing wind, with fluxes in the direction of the prevailing wind up to 12 times greater
than in the opposing direction (Warburton, 2003; Foulds and Warburton, 2007b).
34
Figure 1.5: Schematic diagram showing different mechanisms of aeolian transport in dry and wet conditions (after Evans and Warburton, 2007)
1.2.2.3.3. Wastage
Peat wastage is a combination of biochemical oxidation, shrinkage, consolidation and
compaction (Francis, 1990). The degree of wastage is difficult to assess, and although
Hutchinson (1980) demonstrated 11–18 mm a−1 of peat surface recession due to oxidation
of lowland peat very little work exists on likely rates in upland systems. However, there is
evidence from upland erosion pin and sediment trap data indicate that exposed peat is
rapidly oxidised, and that wastage losses can be up to 40% (Evans et al., 2006).
1.2.2.4. Consequences
Once initiated, erosion can become self-perpetuating. When exposed, the surface of the
peat becomes vulnerable to erosive processes which can alter the hydrology and structural
cohesiveness of the peat preventing vegetation re-colonising these bare areas. Upon gully
formation, the concentration of flow deepens and widens the channel and the gully can
extend headward, further altering the hydrology of the region and increasing the potential
for further degradation.
35
1.2.2.4.1. Sediment flux
There is increasing awareness of the environmental significance of suspended sediment
with respect to its role as a vector for the transfer of nutrients and contaminants in fluvial
systems (Ballantine et al., 2008; Hatfield and Maher, 2008). Fine sediment is itself
increasingly recognized as a pollutant in its own right, when delivered at accelerated rates
(e.g. Evans and Warburton, 2005; Holliday et al., 2008). Increased volume of sediment can
impact water quality through the deposition of fluvially transported particulate peat in
reservoirs (Evans and Warburton, 2005) and can lead to the loss of filter feeding organisms
(Ramchunder et al., 2009).
1.2.2.4.2. Carbon flux
There has been an increasing recognition of the importance of fluvial systems in the
terrestrial carbon cycle, but there has been limited focus on fluvial geomorphology in
relation to carbon cycling in peatlands (Pawson et al., 2012). Carbon is lost from peatlands
either as CO2 or CH4 produced by the microbial breakdown of organic matter, or via the
fluvial system as dissolved- or particulate- organic carbon (DOC and POC) and dissolved
inorganic carbon (DIC) (Billet et al., 2010). Increased sediment flux will transfer more DOC
and POC to peatland streams. The removal of DOC from water sources already represents
one of the major costs to water treatment in upland Britain (Worrall et al., 2004) so
enhanced levels in the system would exacerbate the problem. Catchment mass balances
have shown that up to 40% of DOC released from peat soils is lost in transit through the
stream network and this loss could be as CO2 to the atmosphere (Worrall et al., 2006) and
there is an indication that some proportion of the POC flux shares the same fate (Worrall et
al., 2009) which would further contribute to “greenhouse gas” induced climate change.
The majority of the work examining fluvial carbon exports from peatlands has focused on
DOC (e.g. Hope et al., 1994; Dawson et al., 2002; Worrall et al., 2004; Billett et al., 2006;
Andersson & Nyberg, 2008), with less attention given to particulate organic carbon (POC)
fluxes (e.g. Pawson et al., 2008, 2012). Particulate carbon can be the most significant vector
for carbon loss from eroding peatland systems (Worrall et al., 2003), and although there is
limited information regarding the fate of fluvial POC, there is evidence to suggest POC from
peatland systems can undergo transformation to DOC in the fluvial environment or become
mineralized to CO2 during periods of floodplain storage (Pawson, 2008; Pawson et al., 2012;
Moody et al., 2013). As POC has the potential to transform to atmospherically active forms
36
of carbon, large fluxes of POC mobilised from eroding peatlands are a potentially important
component of the greenhouse gas balance of these systems.
1.2.2.4.3. Contaminant flux
As with the carbon sink there is some concern that peatlands may be shifting from a sink to
a source of stored contaminants. The importance of suspended sediment in the transport
and biogeochemical cycling of contaminants in the aquatic system is well recognised (e.g.
Hart, 1982; Tipping et al., 2010) and the physical erosion of peat has been highlighted as a
mechanism for the release of significant quantities of lead to surface waters (Rothwell et
al., 2005, 2008a; Shotbolt et al., 2006; Rose et al., 2012). High rates of organic sediment
flux therefore have the potential to transfer significant amounts of stored pollutants to the
aquatic system. In areas where sheet erosion of the surface layer occurs, much higher
contaminant concentrations might be expected as the pollutants are found at or near the
peat surface (Rothwell et al., 2007a). Water table draw down associated with gully erosion
may pose further issues. Industrially derived sulphur particles, become oxidized during
periods of lowered water table, forming soluble sulphates and, ultimately, sulphuric acid
upon dissolution into groundwater when the water table rises again (Tipping et al., 2003).
This could lead to increased desorption of toxic heavy metals from eroded peat particles,
elevating the concentrations of dissolved heavy metals in peatland streams (Rothwell et al.,
2006), posing a threat to the sustainability of aquatic ecosystem (Rhind, 2009) and
potentially compromising downstream water resources (Shotbolt et al., 2006).
1.2.3. Restoration
Given the adverse consequences of peat erosion and the self-perpetuating nature of the
feedback mechanism outlined in Section 1.2.2.4., the physical rehabilitation of peatlands is
of great importance in preventing the spread of erosion to neighbouring areas (Mackay and
Tallis, 1996) and restoring the peat’s functionality (Worrall et al., 2009). Much work on
blanket peat restoration has focussed on mined (e.g. Charman, 2002) and drained (e.g.
Holden et al., 2004) peatlands, with less attention paid to eroded systems (Evans and
Warburton, 2007).
In mined and drained areas, restoring high water tables has been identified as key in
reintroducing mire vegetation and returning peatlands to an actively accumulating state.
The simplest method of restoring high water tables is to dam discharge channels (Moore,
2002), and restoration strategies often incorporate drain or ditch blocking as a means of
37
raising water tables (e.g. Holden et al., 2004). There is evidence to suggest natural re-
vegetation may occur in eroding peatlands when gully sides collapse, effectively blocking
the channel, and trapping water and sediment (Evans and Warburton, 2005) and blocking
in headwater gullies (Figure 1.6a) has proved successful in mimicking the natural process of
gully floor re-vegetation to provide effective reductions in sediment flux (Evans et al.,
2006). However, blocking has met varied success in wider, deeper gullies where side- and
under-cutting can cause blocks to fail (Evans et al., 2005).
Evans and Warburton (2005) identified the re-vegetation of eroded peatlands as a major
control on sediment flux from gullied systems. If natural vegetation is absent and
recolonisation is deemed unlikely, active measures are pursued (Figure 1.6b), often
through reseeding and application of a mulch or heather brash (Holden et al., 2007) to
stabilise the surface and retain surface moisture. Active re-vegetation of bare peat is
encouraged by a series of mitigating restoration techniques:
1. application of lime and fertiliser,
2. re-seeding areas with grass nurse crop,
3. spreading of heather brash, heather bales or geo-textiles creating a protective
cover for the new vegetation.
The aim is to provide suitable habitat conditions for a natural re-colonisation by native
blanket bog species, such as cotton grass and sphagnum species. These re-vegetation
techniques are applied in conjunction with stock removal to help the moorland to recover
(Anderson et al., 2009).
Figure 1.6: Example of peatland restoration strategies: (a) gully blocking, (b) reseeding bare peat surfaces (source: Moors for the Future).
38
1.3. Significance of peatland geomorphology
In intact peatlands, geomorphology is simply a boundary condition, whereby landscape
position influences mire type (Lindsay et al., 1998). However, in severely eroded peatlands,
the development of gully networks produces highly variable topography, and
geomorphological form and process become key controls on peatland function (Evans and
Warburton, 2010). This is illustrated in Figure 1.7.
The mobilisation and removal of peat from bare surfaces by the processes of physical
erosion produces a large flux of particulate organic matter from degraded peatlands via the
fluvial system. Peat is typically around 48% organic carbon (Pawson, 2008) so high fluvial
sediment fluxes from eroding peatlands represent a large loss of POC. Gully erosion further
impacts peatland carbon storage through the augmentation of water tables, which can
increase dissolved and gaseous carbon fluxes (Clay et al., 2012). In addition to the impacts
of erosion on peatland carbon cycling, the physical removal of peat has been found to
Figure 1.7: The role of geomorphology in peatland function and material flux (adapted from Evans and
Warburton, 2010).
39
transfer substantial quantities of stored contaminants to the fluvial system (Rothwell et al.,
2006, 2007b, 2008b). Evans et al. (2006) showed that the re-vegetation of gullies is
effective in limiting sediment flux from eroding peat catchments, and large areas of British
peatland have been restored through a range of techniques aimed at stabilising eroding
surfaces (Cole et al., 2014).
At present, severe and extensive erosion of blanket peat is a phenomenon which is almost
unique to the UK and Ireland, and consequently the role of geomorphological processes
and the need to actively restore degraded systems is unusual (Evans and Warburton, 2010).
However, predicted changes in climate increase the likelihood of peatland physical
instability becoming more widespread. An understanding of geomorphological controls on
sediment release, carbon cycling and contaminant flux is therefore essential to identify and
mitigate the negative impacts of peatland erosion.
1.4. Research Rationale
There is a growing body of work relating to peatland geomorphology, most of which has
been carried out in the UK. Studies by Evans and Warburton (2005) and Evans et al. (2006)
provide comprehensive data on the full range of peat erosion processes from a single
peatland site. However, these studies focus on constructing sediment budgets and
quantifying catchment scale export of organic sediment, and while Rothwell et al. (2007b)
surmised that variability in the Pb content of fluvial sediments was likely due to differences
in catchment erosion processes, to date, there has been no attempt to provide equivalent
data on the mechanisms which control contaminant release and storage.
This thesis seeks to investigate sediment dynamics and Pb release in the severely eroded
and contaminated peatlands of the Peak District National Park (PDNP), South Pennines, UK.
The blanket peats of the PDNP embody many of the problems and pressures outlined in
Section 1.2.1. The Park lies at the southernmost climatic margin for blanket peat growth, is
intensively managed, and the peat is amongst the most heavily eroded and contaminated
in the world. The area also supports a range of ecosystem services, is of cultural and
historical importance, and has seen significant investment into pioneering schemes aimed
at combating the high level of degradation (see Section 2.1.1. for more detail).
Consequently, the peatlands of the PNDP provide valuable insight into erosion dynamics
and contaminant release which may prove analogous for the trajectories of similar
peatland systems globally, and will play a vital role in informing future management
strategies should the physical instability of peatlands become more widespread.
40
Peatland sediment dynamics have been studied at a range of scales, for example:
Evans and Lindsay (2010a and b) investigated the effects of gully erosion across the
whole landscape;
Pawson et al. (2008 and 2012) and Evans et al. (2006) derived organic sediment
fluxes from individual eroding blanket peat catchments;
Labadz et al. (1991) looked at sediment release and composition during single
storm events;
Klove (1998) and Holden and Burt (2002b) used rainfall simulations to study
erosion and sediment delivery mechanisms at the plot scale.
However, Evans and Warburton (2010) highlight that there is often a mismatch between
plot-scale erosion rates and catchment particulate export, and that factors affecting
sediment delivery must be carefully considered. There is evidence to suggest that
contaminated sediment storage and release should also be considered in such a way;
Bindler et al. (2004) and Rothwell et al. (2007a) found substantial spatial heterogeneity in
blanket peat Pb pollution records at both the landscape and within-site scales, and the
magnitude of suspended sediment associated Pb transport has been shown to vary greatly
across eroding blanket peatlands (Rothwell et al., 2010a), and also during and between
storm events in a single catchment (Rothwell et al., 2005, 2007b), indicating that various
external factors are influencing the supply of contaminated sediment .
Previously, information on sediment generation and provenance in peatlands has been
obtained using a range of indirect measurement or monitoring techniques, but these
methods can be hampered by problems of spatial and temporal sampling, and operational
difficulties (Walling et al., 2008). Pawson et al. (2008) also note the need for high resolution
temporal sampling of suspended sediment in peatland systems due to the episodic nature
of sediment flux; however such intensive sampling campaigns can be highly labour
intensive, and the associated logistical problems mean that many manual sampling
strategies can fail to coincide with the main periods of sediment transport (Collins and
Walling, 2004). Further issues arise from the quantity and representativeness of the
samples collected.
The legacy of Pb pollution in the PDNP, whilst undesirable, offers a unique opportunity to
apply a sediment fingerprinting approach, developed for use in minerogenic systems, to
trace sediment movements in peatlands. By exploiting the stored Pb as a tracer of
contaminated sediment, not only can Pb contaminant release be studied in greater detail,
41
but a greater understanding of peatland sediment dynamics can be gained. The technique
can be implemented at a range of scales to determine the key mechanisms influencing
sediment storage and mobility in eroding catchments, assess the factors controlling
contaminated sediment release at the event scale, and monitor the effects that
degradation and restoration have on sediment and contaminant dynamics across the
landscape. To date, no such study has been undertaken, and many of the standard
methods which are commonly used in other settings will have to be adapted for use in
organic systems.
1.5. Aims
A series of interconnected projects use sediment source fingerprinting techniques to build
on our current understanding of peatland geomorphology, sediment dynamics, and
contaminant storage (outlined in Sections 1.1. and 1.2.), with the aim of identifying the key
processes that drive peat erosion and contaminated sediment release, and in turn assess
the efficacy of current restoration practices in the Peak District National Park.
1.5.1. Objectives
The specific aims of each project are outlined in detail in their respective Chapters, but in
order to realise the overarching aim of the thesis, two key objectives are met:
Objective 1: Develop a set of tools to trace sediment movements in eroding peatlands,
including:
a) a rapid, non-destructive, low cost means to quantify near surface Pb
concentrations in situ (see Chapter 3/Paper 1)
b) a compact, lightweight sediment trap which can be deployed across multiple sites
in remote areas (see Section 2.2.2)
c) a quantitative method to evaluate the relative contributions of different potential
sediment sources (see Chapter 4/Paper 2)
Objective 2: Use the tools developed as part of Objective 1 to investigate sediment and
pollution dynamics across a range of spatial scales:
a) Landscape – An assessment of the effectiveness of restoration practices carried out
by the Moors for the Future Partnership on the Bleaklow Plateau in reducing
sediment production, and POC and Pb release. This project was funded by Moors
for the Future Partnership, and is presented in Chapter 4 (Paper 2).
42
b) Catchment – An investigation into the hydrological and geomorphic mechanisms
governing Pb and POC release between and within storm events in the Upper
North Grain catchment. This project builds on the findings of Rothwell et al. (2005)
and is presented in Chapter 5 (Paper 3).
c) Plot – Identification of the key geomorphic controls on contaminated sediment
storage and release in severely degraded, un-restored, headwater gullies of the
Bleaklow Plateau. This project aims to address some of the uncertainties posed by
the landscape and catchment scale studies, and is presented in Chapter 6 (Paper 4).
1.6. Thesis structure
Figure 1.8 summarises the structure of the thesis, and specifies the sections and chapters
which address each of the objectives outlined in Section 1.3. In studying peatland sediment
dynamics at a range of spatial scales, this thesis can be divided into a series of sub-projects,
united by a single overarching aim and interlinked methodologies (see Section 1.3.). Each
of these sub-projects make a unique contribution to peatland science, and thus lend
themselves to being written as a series of individual journal articles. As such, this thesis is
presented in an ‘alternative format’, incorporating chapters that are in a format suitable for
submission for publication in peer-reviewed journals. The articles which make up Chapters
3 and 4 have already been published, and the articles put forward in Chapters 5 and 6 have
been written with specific journals in mind.
43
Figure 1.8: Thesis structure. Numbers in brackets relate to the objectives outlined in Section 1.5., indicating the Section or Chapter where these are addressed.
44
1.6.1. Contributions to papers
As is common in environmental science, the articles presented in Chapters 3 to 6 were
written in collaboration with several co-authors: Emma Shuttleworth (ELS), Martin Evans
(MGE), James Rothwell (JJR), Simon Hutchinson (SMH), Gareth Clay (GDC). The
contributions of the co-authors to each paper is as follows:
Chapter 3/Paper 1, published as: Shuttleworth, E. L., Evans, M. G., Hutchinson, S. M., &
Rothwell, J. J. (2014) “Assessment of Lead Contamination in Peatlands Using Field Portable
XRF” Water, Air, and Soil Pollution 225:1844, DOI 10.1007/s11270-013-1844-2.
ELS designed and co-ordinated the field campaign, conducted all of the lab work, researched and implemented the analytical framework, and wrote and re-drafted the manuscript
MGE provided supervisory advice when setting up the field campaign, and commented on manuscript drafts.
JJR provided supervisory advice when setting up the field campaign, assisted with fieldwork, and commented on manuscript drafts.
SMH provided analytical equipment, taught ELS how to use field portable XRF, assisted with fieldwork, and commented on manuscript drafts.
Chapter 4/Paper 2, published as: Shuttleworth, E.L., Evans, M.G., Hutchinson, S.M., &
Rothwell, J.J. (2014) “Peatland restoration: controls on sediment production and reductions
in carbon and pollutant export” Earth Surface Processes and Landforms DOI:
10.1002/esp.3645
ELS designed and co-ordinated the field campaign, conducted all of the lab work, researched and implemented the analytical framework, adapted the modelling approach used, and wrote and re-drafted the manuscript.
MGE provided supervisory advice when setting up the field campaign and modelling approach, and commented on manuscript drafts.
SMH provided analytical equipment, taught ELS how to use field portable XRF, assisted with fieldwork, and commented on manuscript drafts.
JJR provided supervisory advice when setting up the field campaign and modelling approach, and commented on manuscript drafts.
Chapter 5/Paper 3, in preparation for submission to Hydrological Processes as:
Shuttleworth, E.L., Evans, M.G., & Rothwell, J.J. “Controls on the fluvial export of sediment
associated lead and particulate carbon from eroding peatlands”.
45
ELS designed and co-ordinated the field campaign, conducted all of the lab work, researched and implemented the analytical framework, adapted the modelling approach used, and wrote and re-drafted the manuscript.
MGE provided supervisory advice when setting up the field campaign, modelling approach, and analytical framework, and commented on manuscript drafts.
JJR provided supervisory advice when setting up the field campaign, modelling approach, and analytical framework, assisted with fieldwork, and commented on manuscript drafts.
Chapter 6/Paper 4, in preparation for submission to Catena as: Shuttleworth, E.L., Clay,
G.D., Evans, M.G., Hutchinson, S., & Rothwell, J.J. “Contaminated sediment dynamics in
peatland headwaters”.
ELS designed and co-ordinated the field campaign, conducted all of the lab work, researched and implemented the analytical framework, and wrote and re-drafted the manuscript.
GDC assisted with fieldwork and statistical analysis, and commented on manuscript drafts.
MGE provided supervisory advice when setting up the field campaign and analytical framework, assisted with fieldwork, and commented on manuscript drafts.
SMH provided analytical equipment, assisted with fieldwork, and commented on manuscript drafts.
JJR provided supervisory advice when setting up the field campaign and analytical framework, assisted with fieldwork, and commented on manuscript drafts.
46
Chapter 2 Methodology
This chapter provides a general overview of the field sites and methods that have been
utilised in various combinations throughout the thesis. A suite of standard magnetic,
geochemical, and statistical analyses have been employed, in addition to several methods
which have been developed or adapted from established procedures, to investigate spatial
and temporal patterns of sediment release and storage in contaminated peatland systems.
These include:
i) developing the application of field portable XRF (FPXRF) to assess lead
contamination in wet organic sediments in situ
ii) modifying time integrated mass-flux samplers (TIMS):
i. for easy deployment across multiple sites in remote areas to
investigate suspended sediment (SS) at the landscape scale
ii. to capture sediment at different flow depths to study SS at the
event scale
iii) adapting a numerical mixing model to determine sediment source in
contaminated organic systems.
Figure 2.1 Methodological framework
47
Figure 2.1 shows how the methods and field areas inter-relate in relation to the four papers
presented in Chapters 3 to 6. The specific methodologies for FPXRF and sediment source
fingerprinting are addressed in detail in Papers 1 and 2 respectively. To avoid repetition
these will only be covered in brief in the following chapter. The choice of specific statistical
analyses is also detailed in each paper, and so they are not included here.
2.1. Field Area
2.1.1. The Peak District
The Peak District is an upland area in central and northern England which marks the
southern extent of the Pennines range (Figure 2.2). It is the oldest National Park in Britain
and covers an area of 1438 km2, over a third of which is made up of upland moorlands
which are protected by national and international conservation designations as a Site of
Special Scientific Interest (SSSI), a Special Area of Conservation (SAC), and a Special
Protection Area (SPA) (Bonn et al., 2009). The Peak District National Park (PDNP) supports
many ecosystem services, and its peatlands are subject to the pressures and degradation
discussed in Chapter 1. The Park offers over 3,000 km of public footpaths, including the
Pennine Way and 32 000 ha of open access land (Peak District National Park, 2013), and
65% of the moorland is managed for red grouse shooting (Sotherton et al., 2009). Roughly
a quarter of the UK population lives within an hour’s travel and the Park receives in excess
of 10 million visitor days per year (Global Tourism Solutions, 2009). In addition to tourism,
79% of the land is farmed for livestock, and 55 reservoirs supply 450 million litres of water
per day to the surrounding area (Bonn et al., 2009). The peats of the Peak District store up
to 30-40 Mt of carbon and have the potential to sequester up to 62 000 t of CO2 per year
(Worrall et al., 2009).
The blanket peats of the Peak District are amongst the most heavily eroded and
contaminated in the world. According to Tallis (1997), the area has experienced two major
periods of erosion resulting in the formation of extensive gully systems: circa 550 BP,
caused widespread desiccation and degradation, and circa 250 years ago. Situated in the
heartland of the 19th century English Industrial Revolution, the PDNP is surrounded by the
industrial cities of Leeds, Manchester, Sheffield, Nottingham, Derby, and Stoke on Trent,
and air pollution has been blamed for the poor habitats in the South Pennines for
generations. As early as 1872, Smith concluded that polluted rain was having a serious
effect on vegetation (Caporn and Emmett, 2009). Tallis (1987) linked pollution from the
48
Figure 2.2: Location map of the Peak District National Park (PDNP). Red star indicates Bleaklow plateau.
Industrial Revolution with the demise of Sphagnum in peat cores (most likely due to
sulphur pollution: Ferguson et al., 1978). Today, air quality has improved but the legacy of
atmospheric deposition remains: there are large expanses of bare peat flats, hundreds of
kilometres of the peat’s surface are dissected by gullies, and high concentrations of heavy
metals can be found near the peat’s surface (Figure 2.3). Erosion in is dominated by gullying
and sheet erosion which produces sediment yields in excess of 100 t km-2 a-1 (Labadz et al.,
1991; Evans et al., 2006), and is releasing substantial amounts of sediment associated Pb
into the fluvial system (e.g. Rothwell et al., 2007b, 2007c).
Figure 2.3: A typical profile of Pb deposition and storage in the Peak District (after Rothwell et al., 2005).
49
The area has also been subject to extreme historic environmental change and currently lies
at the southern climatic fringe of blanket peatlands and receives lower rates of
precipitation than any other British upland peats (Tallis, 1997). Average annual rainfall in
the south Pennines is around 1200mm a−1 which is close to the lower limit for active peat
growth in the UK (Evans et al., 2006). This marginality makes these peatlands highly
susceptible to future changes in climate. Hulme et al. (2002) predict that summer
conditions will become warmer and drier and winters become milder and wetter; by the
2020s average summer rainfall is likely to decrease by approximately 10%, and 23-45% by
2080. Increases in the number and severity of storms and summer droughts could result in
further degradation and gully development, which in turn will influence the carbon balance
(Evans et al., 2006; Pawson et al., 2008) the mobilisation of pollutants (Rothwell et al.,
2005, 2007b, 2010b; Tipping et al., 2010), and water table variability, flow pathways and
runoff generation (Daniels et al., 2008; Goulsbra et al., 2014).
As highlighted in Section 1.2., the blanket peats of the PDNP embody many problems and
pressures faced by peatlands worldwide. Consequently, the work carried out in the area is
of great value, offering a unique focus of national and international relevance, and the
PDNP has been a focus of peatland research for several decades. Bower (1960a & b, 1961)
first described the eroded landscape of the Pennines, and from the 1960s onward John
Tallis produced a substantial body of work investigating the timing and causes of the
initiation of this erosion (e.g. Tallis, 1964, 1973, 1985, 1987, 1995, 1997). This study aims to
add to this body of work, focussing on the Bleaklow area of the Southern Pennines (Figure
2.2).
Figure 2.4: Location the Bleaklow Plateau relative to the industrial cities of Manchester and Sheffield.
50
2.1.2. The Bleaklow Plateau
The Bleaklow Plateau (500 – 633 m) is an upland blanket peatland in the PDNP (Figure 2.4).
Peat depths across the plateau vary between 2 and 3 m (Evans and Lindsay 2010a), and
cover an underlying geology composed of sandstone bedrock from the Millstone Grit Series
(MGS) (Wolverson-Cope, 1976) which is overlain in places by fine grained head deposits of
weathered MGS shales (Rothwell et al., 2005). Contemporary vegetation cover is
representative of the mire (bog) communities detailed in the UK National Vegetation
Classifications (NVC), including M19-Calluna vulgaris—Eriophorum vaginatum blanket mire,
M20-Eriophorum vaginatum blanket and raised mire and the M3-Eriophorum angustifolium
bog pool community (Cole et al., 2014). Mean monthly temperatures measured vary
between 12.9 °C (July) and 1.44 °C (February) (2003-2013), annual rainfall is 1020-1840 mm
(2007-2013), and the prevailing wind direction is SSW (195°) (unpublished data recorded at
Upper North Grain). The plateau lies between the industrial cities of Manchester and
Sheffield, and consequently, were subject to substantial deposition of atmospheric
pollutants during the Industrial Revolution. The near-surface layer of the peat is
contaminated by high concentrations of anthropogenically derived, atmospherically
deposited Pb (in excess of 1700 mg kg-1; Paper 1).
Over the last decade there has been a move to actively restore the degraded catchments of
Bleaklow (see paper 2 for more detail). The Moors for the Future (MFF) Partnership,
supported by the UK Heritage Lottery Fund has invested millions of pounds to identify
suitable approaches to control and reverse peatland erosion, which have been pioneered
the Plateau (Figure 2.5). As such, Bleaklow comprises catchments which represent various
stages of the erosion-restoration continuum (Figure 2.6), and has been a focus of recent
research into the impacts of peat erosion. Work has focused on peatland hydrology (e.g.
Daniels et al., 2008; Goulsbra et al., 2014), carbon flux and sequestration (e.g. Pawson et
al., 2008, 2012; Clay et al. 2012), pollutant storage and mobility (e.g. Ferguson et al., 1978;
Hutchinson, 1995; Rothwell et al., 2005), and the effects of restoration (e.g. Dixon et al.,
2013; Cole et al., 2014).
Papers 1, 2, and 4 build on this previous research by mapping surface Pb storage and
determining sediment composition at intact, eroding and restored areas on the Bleaklow
Plateau, and using this information to investigate the controls on particulate carbon
mobilisation, and sediment associated Pb release and storage in eroding and restored
catchments.
51
Figure 2.5: Restoration carried out by MFF Clockwise from top left: Heather brash; spreading lime and fertiliser; plug planting; geojute (source: Moors for the Future Partnership).
Figure 2.6: Erosion-restoration continuum. Top: intact peatland; Middle: actively eroding with little vegetation cover; Bottom: re-vegetated gullies.
52
2.1.3. Upper North Grain
Upper North Grain (UNG) is a small headwater stream that drains a blanket peat covered
catchment on the south eastern edge of the Bleaklow Plateau (Figure 2.7). The catchment
lies between 490 and 541 m OD, covers an area of 0.38 km2, and receives approximately
1200 mm rainfall each year. The vegetation cover is predominantly composed of Calluna
vulgaris, Eriophorum vaginatum, Empetrum nigrum, Erica tetralix, Vaccinium myrtillus, and
patches of Sphagnum spp, and land use is dominated by rough grazing by sheep. The peat
reaches up to 4 m in thickness, contains high concentrations of Pb stored in the near-
surface layer (Rothwell et al., 2005) and overlies the MGS lithology described above. The
catchment is heavily eroded with Bower Type I peat gullies (Bower, 1961). In the upper
reaches gully incision is confined to the peat, but the underlying geology becomes exposed
further downstream (Figure 2.8).
Figure 2.7: a) Location of Upper North Grain (UNG) catchment (starred); b) aerial photograph of UNG catchment. The dense dendritic gully network is clearly visible (Pawson et al., 2008).
With the permission of the National Trust the catchment is used as a field research and
teaching laboratory by the Upland Environments Research Unit (UpERU) at the University
of Manchester. As such, the catchment is heavily instrumented and has been the focus of a
variety of geomorphological and hydrological projects (e.g. Evans et al., 2006; Daniels et al.,
2008; Pawson et al., 2008; Goulsbra et al. 2014). Meteorological conditions are monitored
by Skye automatic weather station (AWS) which records a variety of parameters including
temperature, precipitation and water table depth, as described in Goulsbra et al. (2014).
Stage is monitored using a pressure transducer (Hobo water level logger U20-001-04)
which has been used to determine sediment rating curves (e.g. Evans et al., 2006; Pawson
et al., 2008, 2012). This extensive record of instrumentation has produced a considerable
53
body of work, including much of Rothwell et al.’s recent investigations into Pb release from
eroding peatlands (Rothwell et al., 2005; 2007b, 2007c, 2007d, 2008a, 2008b, 2010b) which
provides a wealth of background data for this thesis.
Paper 3 builds on the work of Rothwell et al. (2005), exploring the mechanisms of Pb
release under storm conditions in the UNG catchment. UNG was also the field site used to
develop the sediment trap outlined in Section 2.2.2.
Figure 2.8: Upper North Grain gully profile showing the exposure of underlying geology at the base of the peat profile (Source: J. J. Rothwell).
54
2.2. Field Techniques
2.2.1. Assessing surface Pb storage using field portable XRF (Papers 1, 2,
and 4)
Near surface Pb storage can vary greatly over short distances in contaminated peatlands
(Bindler et al., 2004; Farmer et al., 2005; Rothwell et al., 2007a), which is further
complicated by gullying and the removal of surface material in heavily degraded areas. The
mixing model outlined in Paper 2 requires precise characterisation of the chemical
properties of potential source materials in order to accurately determine sediment
provenance, and although Rothwell et al. (2007a) recommend the use of 15 samples to
reliably characterise within-region Pb storage in intact peatlands, this was found to be
insufficient at the degraded and restored field sites (Paper 2). There was therefore, a need
to analyse a large number of samples to properly quantify Pb storage in each catchment.
Traditional geochemical analyses of peat often require time consuming sample
preparation, costly reagents, and can result in sample destruction limiting further analysis.
Field portable x-ray fluorescence (FPXRF) continues to gain acceptance in the study of
metal contaminated soil (VanCott et al., 1999; Kalnicky and Singhvi, 2001; Martín Peinado
et al., 2010; Hu et al., 2014) as it allows a large number of samples to be processed in situ
in a relatively short time, giving a high level of detail with little disturbance to the
surrounding area.
XRF analysis is based on the principle of atomic excitation to identify elements by the
characteristic wavelength that they emit when subjected to radiation. When a sample is
irradiated with primary X-rays, inner-orbital electrons in atoms become excited and are
photo-ejected. This leaves the atom in an excited state, with a vacancy in the inner shell of
electrons. During relaxation of the atom, an outer orbital electron fills this vacancy, and
characteristic, secondary X-rays are emitted that possess wavelengths unique to each
element (Johnson et al., 1995). The emission of this secondary X-ray is called fluorescence
(Block et al., 2007). The XRF unit detects the intensity of these secondary X-rays, and
calculates concentration data for the elements of interest in the sample. Water absorbs
and scatters x-rays, lowering precision and accuracy and increasing detection limits (Ge et
al., 2005), and large quantities of light elements such as carbon can lower apparent
concentrations of heavier elements (Löwemark et al., 2011), so until recently FPXRF had
not been used on peat samples due to their high carbon, and in situ moisture content.
55
However, there is evidence to suggest that heavier elements (such as Pb) are less affected
by this interference (Kalnicky and Singhvi, 2001), and samples with high moisture and OM
contents have been successfully analysed by FPXRF (e.g. Bernick et al., 1995; Solo-Gabriele
et al., 2004; Hürkamp et al., 2009a). Independently of this thesis, Shand and Wendler
(2014) assessed the use of FPXRF for the analysis of dried and ground peaty soils in a
laboratory setting and found that the unit they used “gave acceptable data for Pb”. They
also speculate about its value for onsite analysis but did not test their unit in the field.
Paper 1 details the result of a comprehensive study which used a Niton XL3t 900 Handheld
XRF Analyser to compare in situ FPXRF field measurements with ex situ FPXRF and ICP-OES
analysis of the same samples (Figure 2.9). The study successfully demonstrates that after
correcting for moisture content, in situ FPXRF readings are directly comparable with ex situ
readings, and that both display strong linear correlation with ICP-OES results, allowing
comparison with other studies.
The methodology outlined in Paper 1 was used in Paper 2 to assess landscape scale
variations in surface Pb storage, and in Paper 4 to investigate contaminated sediment
storage at the catchment scale.
Figure 2.9: Using the field portable XRF (a) in situ for Paper 4 and (b) ex situ for Paper 1.
56
2.2.2. Suspended sediment sampling (2 and 3)
Rationale 2.2.2.1.
Pawson et al. (2008) note the need for high resolution temporal sampling of SS in peatland
systems due to the episodic nature of organic sediment flux. However, there are several
problems associated with some traditional SS sampling programmes. Manual sampling
strategies can fail to coincide with the main periods of sediment transport, and such
fieldwork is often highly labour intensive (Collins and Walling, 2004). Automatic water
samplers, such as those used by Rothwell et al., (2005), can be costly (especially if deployed
at several sites), require a power source, and are awkward to install in more remote field
areas. Further issues arise from the quantity and representativeness of the sample these
devices typically collect. Phillips et al. (2000) note that sample volumes collected by
commercially available automatic water samplers are often insufficient to yield substantial
quantities of suspended sediment. Obtaining a sufficient amount of sediment for
geochemical analysis involves recovery from a large volume of water (>50 l), initially stored
in situ before transport to the laboratory (Walling, 2005). These issues can constrain site
selection, limit the resolution of sampling intervals, and affect the general operation of the
sampling programme. Instantaneous samples taken at fixed intervals may not take into
account inter- and intra-storm variations in sediment source areas and time-variant inputs
from point-sources (Phillips et al., 2000).
Time integrated mass flux samplers (TIMS) 2.2.2.2.
Phillips et al. (2000) developed a simple time integrated mass sampling device (TIMS) to
collect a representative sediment sample of sufficient mass for the subsequent analysis of
its properties. The sampler is made up of a metre long piece of piping (98 mm ID), sealed at
both ends with inlet and outlet tubes (4 mm ID) passing through the end caps. A funnel
placed over the inlet tube streamlines the body and minimizes disruption to the ambient
flow structure around the opening (Figure 2.10a). The device is filled with clean native
water and secured to the channel bed at the field site where it operates unattended for the
duration of the sampling programme. Water enters the inlet tube at a velocity similar to
that of the ambient flow but, because the main cylinder has a greater cross sectional area
compared with that of the inlet tube, the velocity is reduced by a factor in excess of 600
relative to that of the ambient flow. This reduction in flow velocity induces sedimentation
as the water moves through the main cylinder towards the outlet tube. The sediment is
57
contained within the sampler body is of sufficient mass to satisfy a wide range of laboratory
analyses.
Figure 2.10: Cross-section of two different time integrated mass flux sampler (TIMS) designs as described in: (a) Phillips et al. (2000), and (b) Owens et al. (2006).
TIMS overcome many of the problems outlined above; they are relatively cheap to
construct, actively collect sediment throughout the entire duration of a sampling campaign,
can be left to operate in situ without the aid of a power source, require little maintenance
(Philips et al., 2000; Russell et al., 2000). Field and laboratory tests show the physical
characteristics and chemical composition of the suspended sediment collected by the
sampler are similar to those of instantaneous manual samples collected during the same
period (Philips et al., 2000; Russell et al., 2000; Smith and Owens, 2014) and the device has
been employed in a wide range of suspended sediment studies (e.g. McDowell and
Willcock, 2004; Evans et al., 2006; Hatfield and Maher, 2008; Ballantine et al., 2008;
McDonald et al., 2010; Martínez-Carreras et al., 2012).
An alternative to the Phillips et al. (2000) design is a sampler first used by Owens et al.
(2006) and Petticrew et al. (2006). This is made up of a gravel-filled cylinder, enclosed at
each end by mesh (Figure 2.10b). Flow entering the trap is slowed by the large surface area
of the gravel and suspended sediment is deposited within the gravel pores. Caps are
secured over each end of the tube before removal from the channel and the contents are
later sieved to separate the suspended sediment from the gravel. In the Owens et al.
(2006) study the sampler yielded enough suspended sediment for geochemical, magnetic
and particle size analysis. This design is compact making it easy to install in large numbers
in the field.
58
Operational issues 2.2.2.2.1.
There is much anecdotal evidence that the Philips style TIMS are not always reliable, and
may not be suitable for deployment in all field settings. When deployed in small peatland
headwater catchments by Rothwell et al. (2010b), two out of ten Philips style TIMS failed to
capture any sediment. Size is also a concern; while the Philips style TIMS are smaller and
lighter than an automatic water sampler, they still measure in excess of 1m in length and
would be cumbersome to deploy in large numbers in remote locations. The internal volume
of the main chamber is 7.5 litres, meaning 7.5 kg of water would be mixed with the SS
retrieved from each TIMS which would be impractical to carry in large numbers. McDowell
and Willcock (2007) used a half-length version where all dimensions were scaled to 50% of
their original length, thereby maintaining the ratio between the diameters of the inlet and
main body, and inducing the same reduction in velocity as the full size model. However,
this smaller version presents a further problem, as the 2mm ID inlet tube could easily
become obstructed by larger peat particles; indeed, the full sized (4mm ID) inlet tube could
also become blocked. A further potential problem arises from the fact that the Phillips style
TIMS is designed to be fully submerged at all times, and so may not be suitable for use in
ephemeral streams. The combination of the small cross sectional area of the inlet, the
comparatively large internal volume of the main body and the speed at which the stage
level increases during storm events could mean that air pressure within a sampler would
prevent it from filling with water.
The Owens et al. style TIMS overcome some of the problems of the Philips design but also
presents some issues of its own. The large mesh capped opening is unlikely to become
blocked and would allow the body of the trap to instantly fill when flow is initiated, and the
smaller size would make it easier to deploy in large numbers. However, the weight of the
gravel would limit the number of TIMS which could be carried to and from remote areas,
and the blunt ended inlet may affect flow dynamics, especially if a number of traps were to
be stacked together or placed side by side. Additionally, the Owens style TIMS has not been
subject to the rigorous testing of the Phillips sampler so the representativeness of the
sediment retained is unknown.
Pilot Study 2.2.2.3.
A field study was conducted to determine which TIMS would be most suitable deployment
in remote ephemeral peatland catchments. Replicas of the original Phillips and Owens
designs were tested, together with TIMS adapted to address some of the concerns outlined
above.
59
TIMS design 2.2.2.3.1.
Two Philips style TIMS were tested:
A full size replica of the original Philips et al. (2000) design (Figure 2.11a).
A half size version of the Philips et al. (2000) TIMS, as used by McDowell and
Willcock (2007), but with inlet and outlet tubes measuring the same as the full size
design (Figure 2.11b). The inlet and outlet tubes were not scaled down due to
concerns over blockage by larger organic particles. Within the main body of the
full- and half- sized Phillips et al. (2000) samplers, the flow velocity is reduced by a
factor in excess of 600, relative to that of the ambient flow. This is by virtue of the
ratio between the cross-sectional areas of the main body and the inlet tube. If the
diameter of the inlet tube was doubled this decrease in velocity would be reduced
by a factor of four, i.e. the flow velocity in the main cylinder would be reduced by a
factor in excess of 150, relative to that of the ambient flow. This would still induce
sedimentation within the main body but it is unclear how representative such a
sample would be.
Figure 2.11: TIMS operating in the field: (a) the original Philips et al. (2000) design, (b) a half-sized Philips et al. (2000) stlyle design, (c) two TIMS based on the original Owens et al. (2006) design.
60
Three TIMS were tested based on the Owens design. These TIMS were constructed from
PVC piping with dimensions: 52 mm (ID) x 0.5 m, capped at each end by 8 mm plastic mesh
(Figure 2.11c), each with a different material enclosed in the main chamber:
Gravel (approx. 16–64 mm) – after Owens et al. (2006)
Polystyrene packing ‘peanuts’ (approx. 20x20x30 mm)– These provide a cost
effective means of reducing mass, but it is unclear if the irregular pitted surface of
the polystyrene would affect the composition of the sediment retained.
Anti-evaporation spheres (20 mm ID) – These would reduce mass and their smooth
surface is unreactive so should not affect sediment composition.
TIMS construction 2.2.2.3.2.
All TIMS were constructed out of PVC pipe. The full size and half size Philips style samplers
were made up of a PVC pipe measuring 98 mm (ID) by 1m and 52 mm (ID) by 0.5m
respectively. Both sizes were sealed at each end by a screw on cap with a 150 mm length of
semi-rigid nylon pneumatic tubing (4 mm ID) threaded through the centre and made
watertight with silicone sealant. Commercially available polyethylene funnels were secured
to the upstream end of each sampler over the inlet tube and sealed with silicone sealant.
Owens style samplers were also based around a 52 mm (ID) by 0.5m length of pipe. Once
filled with gravel or gravel substitute, each end was sealed with 8mm plastic mesh secured
to a pipe coupling. The pipe coupling joins were made water tight using weather resistant
gaffer tape.
Two large cable ties (10 mm wide) were pulled tight around the main body of each
sampler.
Field deployment 2.2.2.3.3.
The TIMS were tested in the lower reaches of the UNG catchment (Chapter 3) where the
stream flows over a bedrock surface, allowing the samplers to be securely fixed to eyelets
which were screwed into the channel bed. The TIMS were positioned near the centre of the
channel with their long axes parallel to the direction of flow and fixed in place by cable ties
looped through the stream bed eyelets and the large cable ties secured around each trap.
The TIMS were deployed for 9 sampling campaigns between March 2010 and August 2011.
Each sampling campaign lasted between two and four weeks. The sediment and water (and
gravel or gravel substitute) retained in each sampler were emptied into new large
61
polythene bags, sealed and returned to the laboratory. Three sampling campaigns did not
yield sufficient sediment for analysis due to lack of precipitation in the catchment.
Laboratory analysis 2.2.2.3.4.
SS samples collected by Owens style TIMS were washed through an 8 mm sieve with
deionised water to separate the sediment from the filling. SS is usually separated from the
excess water by settling and/or centrifusion (e.g. Philips et al. 2000; Rothwell et al., 2010b;
Owens et al., 2012). However, the volume of water associated with the full sized Philips
sampler was not practical to centrifuge, and there were concerns that some of the organic
portion of the sediment would remain in suspension and be lost with the supernatant if
samples were left to settle, so the resulting slurry was oven dried at 40 °C (so as not to
affect the magnetic mineralogy of the samples: Walden et al., 1999) until a constant weight
was achieved.
Once dry, the bulk weight retained by each TIMS was determined before the samples were
gently disaggregated and homogenised by hand using a pestle and mortar. Samples were
then subsampled in triplicate, provided enough SS had been retained. Samples were
analysed for low frequency magnetic susceptibility (χlf), magnetic remanence (ARM and
SIRM), lead (Pb) content, and organic matter content (OM). These analyses are outlined in
detail in Section 2.3.
Statistical Analysis 2.2.2.3.5.
The TIMS designs were assessed in two stages. Firstly, the characteristics of the SS collected
by the original Philips and Owens designs were considered to see if the two different
samplers are comparable. This was achieved using t-tests on each of the parameters tested
(p ≤ 0.05). Secondly, the modified designs were compared to the originals to see if the TIMS
could be better adapted to deployment in remote areas. Differences in the amount and
composition of the sediment collected by each type of TIMS were assessed using a one-way
analysis of variance (ANOVA), followed by Tukey's post-hoc comparison at the 95 % level (p
≤ 0.05).
Results and discussion 2.2.2.3.6.
Descriptive statistics for all of the parameters are presented in Table 2.1.
2.2.2.3.6.1. Comparison of original designs
The results of the t-tests carried out on the sediment collected by the original Philips and
Owens sampler designs are detailed in Table 2.2. The most striking difference lies between
62
the mass of sediment retained (Figure 2.12a). The Philips style sampler did not collect more
than 5g of sample during any of its deployments – less than half of the mass suggested by
Philips et al. (2000) as sufficient to permit detailed geochemical analysis. Indeed, the
paucity of sample limited the analyses that could be performed in this pilot study. In
contrast, the Owens style TIMS collected between 20 and 50g, which would be more than
sufficient to perform multiple analyses.
Both designs retained similar concentrations of OM and Pb, but the SS collected by the
Philips style TIMS had an enhanced magnetic signature. Although χlf and SIRM/ARM
readings were not found to be statistically different, ARM and SIRM values were
significantly higher in SS collected by the Philips style sampler (Figure 2.12b and 2.12c). χlf is
roughly proportional to the concentration of ferrimagnetic minerals present, while the
SIRM/ARM ratio if sensitive to the size of ferrimagnetic grains (see Section 2.3.1. for more
detail). SIRM/ARM values are relatively high, indicating that the magnetic material is
dominated by coarse grained ferrimagnetic material (i.e. the fly ash stored in the near
surface peat; Rothwell et al., 2005). Accordingly, the fact that the sediment collected by
both samplers produce statistically similar χlf and SIRM/ARM values indicates that the
magnetic signature in both sediment samples is derived from similar amounts of coarse
grained ferrimagnetic material (i.e. AIS).
Although not straightforward, several authors (e.g. Dunlop, 1981; Maher, 1988; Oldfield,
1990; Hatfield and Maher, 2008) have shown that there is an inverse relationship between
particle size and ARM and SIRM parameters, so differences in the magnetic properties of
the SS collected by the two TIMS could be caused by particle size effects. Ambient
sediment samples were not collected in conjunction with TIMS deployment so it is unclear
whether the Philips style sampler is preferentially retaining finer material, or if the Owens
style sampler is not efficiently retaining smaller particle sizes.
The more commonly used Philips style sampler does not reliably retain sufficient sediment
for subsequent analyses in peatland streams, so is not suitable for use in this thesis. The
majority of the parameters tested (including those used in the optimised mixing models
detailed in Papers 2 and 3) show no significant difference between the two designs, so the
Owens style sampler appears to be well suited for use in ephemeral peatland catchments.
Like the full sized version, the half sized Philips style sampler also did not retain enough
sediment for analysis so will not be considered further.
63
Philips Owens
Full size Half size
Gravel
Polystyrene packing peanuts
Anti-evaporation
spheres
Weight (g)
Mean 2.4 2.7 37.6 73.4 48.6
Min 0.9 0.4 19.4 18.0 14.3
Max 4.7 6.1 52.4 196.2 142.3
OM (%)
Mean 42.3 38.3 39.8 37.3 27.7
Min 40.6 27.7 10.4 14.9 10.0
Max 45.3 46.2 61.8 66.4 66.0
Pb (mg kg-1)
Mean 116.6 74.6 99.3 92.6 69.8
Min 104.9 63.1 79.7 67.5 45.6
Max 128.4 94.7 129.7 135.5 85.0
χlf
Mean 8.461 4.73 4.351 4.09 2.27
Min 4.938 1.45 1.92 3.36 1.15
Max 12.56 8.76 5.37 8.57 3.30
ARM
Mean 0.85 0.49 0.22 0.30 0.15
Min 0.59 0.18 0.12 0.09 0.09
Max 1.03 0.75 0.37 0.28 0.23
SIRM
Mean 83.05 53.47 21.46 26.84 10.69
Min 66.74 17.58 7.75 5.00 4.95
Max 102.35 82.32 39.16 22.93 20.00
SIRM/ARM
Mean 100.72 107.05 91.25 83.14 68.28
Min 64.67 100.35 63.11 72.49 54.41
Max 120.68 110.49 105.58 101.24 86.21
Table 2.1: Summary of the parameters used to compare the sediment composition collected by the different TIMS designs.
t-value p
Weight 5.521 0.001
OM 0.922 0.409
Pb 1.222 0.346
χlf 2.148 0.064
ARM 5.876 0.001
SIRM 5.727 0.001
SIRM/ARM 0.605 0.567
Table 2.2: Result of the t-test employed to compare the original TIMS designs. Significant parameters are given in bold.
64
Figure 2.12: Interval plots for parameters which produced significant differences when comparing sediment collected by the original Phillips et al. (2000) and Owens et al., (2006) TIMS designs depicting 95% confidence
intervals for the means: (a) mass of sediment retained, (b) ARM, (c) SIRM.
65
F-value p
Weight 0.56 0.583
OM 0.55 0.592
Pb 2.20 0.153
χlf 3.03 0.099
ARM 4.58 0.042
SIRM 3.40 0.079
SIRM/ARM 2.28 0.158
Table 2.3: Result of the ANOVA employed to compare the characteristic of sediment collected by the Owens et al. (2006) TIMS adaptations. Significant parameters are given in bold.
2.2.2.3.6.2. Comparison of Owens style sampler modifications
The results of the ANOVA comparing all Owens style TIMS designs can be found in Table
2.3. ARM is the only parameter to show a significant difference. This difference lies
between the polystyrene and anti-evaporation sphere fillings (t=3.026; p=0.0347), but both
of the alternative fillings were statistically similar to the original gravel filling (Figure 2.13).
These results indicate that the Owens style sampler can easily be adapted to reduce its
mass.
The polystyrene filling was chosen over the anti-evaporation spheres as the composition of
sediment collected by the polystyrene filled traps was generally closer to that collected the
original design (Table 2.1). The polystyrene filled TIMS are also lighter, and substantially
cheaper, allowing multiple TIMS to easily be deployed for minimal outlay of costs.
Figure 2.13: Interval plots for ARM – the only parameter to produce a significant difference when comparing sediment collected by the Owens et al. (2006) TIMS adaptations.
66
2.3. Laboratory Techniques
2.3.1. Environmental Magnetism (Papers 2 and 3)
All substances possess magnetic properties (Walden et al., 1999), and the magnetic
attributes of soils and sediments have been used to characterise potential sources of SS in
a vast number of studies over the last five decades (e.g. Walling et al., 1979; Yu and
Oldfield, 1993; Hutchinson, 1995; Rothwell et al., 2005; Blake et al., 2006; Hatfield and
Maher, 2008). Magnetic analysis is widely applied as it is cheap, simple, rapid, and non-
destructive (Oldfield, 1991), and can produce reliable results from relatively small samples
(Liu et al., 2012).
Coal combustion and many associated heavy industries generate residual, un-combusted
mineral particles, many of which leave the site of combustion as particulate pollutants in
the atmosphere. These are often referred to under the generic term ‘fly ash’ composed of
Spherical Carbonaceous Particles (SCPs) and Inorganic Ash Spheres (IAS) (Oldfield, 2014),
the latter are rich in both magnetite and haematite which give them a distinct magnetic
signature (Winburn et al., 2000) and are referred to as “ferromagnetic spherules” in Papers
2 and 3. These IAS persist in the stratigraphic record and have been cited as a key marker for
human activity and the onset of the Anthropocene (Oldfield, 2014). Concerns have been raised
over the survival of IAS in peatland environments; given the low pH of ombrotrophic bogs,
some magnetite dissolution is likely (Oldfield, 1991). However, Williams (1988) and Clymo
et al. (1990) hypothesise that this dissolution likely distorts rather than entirely destroys
the depositional record, and magnetic measurements of recent, near surface, peats have
been used to provide a good record of the deposition of particulate pollutants emitted as a
result of industrial activity around the world (e.g. Great Britain and Scandinavia - Thompson
and Oldfield, 1986; Canada - Tolonen and Oldfield, 1986; China - Bao et al., 2012).
High concentrations of magnetic minerals are stored in the near-surface layer of peat soils
of the Peak District National Park, and although they have been shown to be variable
(Rothwell and Lindsay, 2007), Rothwell et al. (2005) and Hutchinson (1995) used the
magnetic signature of this pollution record to help distinguish between sediment derived
from the peat’s surface from other catchment sources. Hutchinson (1995) and Rothwell et
al. (2005) demonstrated that subsurface peats are magnetically impoverished, as organic
matter – peat’s main constituent – is diamagnetic, that Millstone Grit sandstone contains
only a small magnetic fraction. Rothwell et al. (2005) go on to discuss the magnetic
properties of the MGS shales and head deposits: shales usually exhibit only weak magnetic
67
remanences, but organic-rich shales, may contain paramagnetic iron sulphides such as
pyrite fine-grained, ferrimagnetic sulphides like greigite or pyrrhotite, and so exhibit quite
strong magnetic susceptibility (e.g. Krs et al., 1992; Snowball and Torii, 1999). Glacial head
deposits at UNG contain a high proportion of weathered MGS shale, and a small quantity
of, coarse, sandy detritus, and so should yield weak-to-moderate magnetic remanence, and
a low SIRM/ARM value indicating the presence of a small quantity of fine-grained
ferrimagnetic material (Rothwell et al., 2005).
Sample preparation 2.3.1.1.
All samples were dried at 40 °C as higher temperatures can affect the magnetic mineralogy
(Smith, 1999). Samples were gently disaggregated in a pestle and mortar and packed into
standard 10 ml pre-weighed (to 0.001g) plastic magnetic sample pots. Once filled the
sample pots were re-weighed .and the mass of the sample was calculated in order to derive
mass specific magnetic parameter values. Where there was not sufficient sample available
to fully fill a sample pot, cling film was used to pack out the pot to immobilise the sample in
the centre of the pot. Cling film is diamagnetic, and so will not influence the final mass
specific magnetic value. Dearing (1994) found that errors can arise if pots are not
completely full. Errors are typically less than 3% if pots are more than 39% full, but if less
than 5% of the pot is full, the mass specific error may be in excess of 15%. As such, values
derived from small sample sizes should be treated with caution.
Magnetic susceptibility 2.3.1.2.
Magnetic susceptibility is roughly proportional to the concentration of ferrimagnetic
minerals, such as magnetite, or the concentration of canted antiferromagnetic minerals,
such as haematite, in the absence of ferromagnetic material (Dearing, 1994). Both of these
minerals are found in IAS (Winburn et al., 2000) so magnetic susceptibility measurements
will indicate the concentration of these particulates.
Low frequency (lf – 0.47 kHz) magnetic susceptibility was measured at room temperature
using a Bartington Instruments Ltd. MS2 meter. The volume specific readings (κ) were
converted to the more commonly quoted mass specific susceptibility values (χ) but dividing
κ values by sample mass (g) and divided by 10 to give units of SI units x 10-6 m3 kg-1
(Dearing, 1994).
Magnetic remanence 2.3.1.3.
Measurements of magnetic remanence (ARM and SIRM) were made with a Molspin
Instruments ‘Minispin’ fluxgate magnetometer (Walden et al., 1999).
68
Anhysteretic remanent magnetization (ARM) 2.3.1.3.1.
ARM is acquired when a sample is subject to a steady or slowly varying field, superimposed
on a decaying alternating field of higher frequency (Dunlop and Argyle, 1997). It is
proportional to the concentration of ferimagnetic grains in the 0.02 to 0.4 µm (stable single
domain or SSD) size range and is particularly sensitive to magnetic grain size, yielding
progressively higher values as the grain size of any magnetic component (particularly
ferrimagnets like magnetite) becomes finer, within the SSD size category (Dunlop, 1981).
ARMs were acquired at room temperature in a peak a.c. demagnetizing field of 100 mT,
with a superimposed d.c. field of 0.1 mT, using a Molspin Instruments’ AF-demagnetizer.
Saturation isothermal remanent magnetization (SIRM) 2.3.1.3.2.
Magnetism resulting from short-term exposure to strong magnetizing fields at room
temperature is referred to as isothermal remanent magnetism (IRM). SIRM is the maximum
remanence that can be produced from short-term exposure to strong magnetizing fields
(1T) (Oldfield et al., 1979). All remanence-carrying magnetic grains will make some
contribution to SIRM; however, even at very low concentrations the ferrimagnetic
component will dominate and Oldfield et al. (1978 and 1979) used SIRM to infer the
presence of AIS. This parameter is also grain-size dependent (Maher, 1988; Oldfield, 1990).
SIRMs were imparted at room temperature in an applied d.c. field of 1 mT, using a Molspin
Instruments’ high-field ‘Pulse Magnetizer’.
The SIRM/ARM Ratio 2.3.1.3.3.
In samples dominated by ferrimagnetic minerals, SIRM/ARM is indicative of relative
magnetic grain-size variations (Hutchinson, 1995). A high ratio of SIRM/ARM indicate the
presence of coarse (multidomain – MD) ferrimagnetic grains, whereas a low ratio of
SIRM/ARM points to fine (single domain – SD) ferrimagnetic grains. The AIS in fly ash have
been shown to produce relatively high SIRM/ARM values (Oldfield et al., 1985).
2.3.2. Deriving Pb content using ICP-OES analysis (papers 2 and 3)
The FPXRF described in Section 2.2.1. was not available at the beginning of this study, so SS
Pb content was determined by inductively coupled plasma-optical emission spectrometry
(ICP-OES). Ideally, the Pb content of all samples should have been measured using the same
equipment, and the early SS samples should have been re-analysed by the XPXRF after it
became available. However, several samples were completely destroyed during ICP-OES
and subsequent analyses so this was not possible, so for parity all SS samples were
69
analysed using ICP-OES and the derived concentrations were made comparable with FPXRF
field measurements using the linear equation detailed in Paper 1.
Pb analysis was carried out on by inductively coupled plasma-optical emission
spectrometry (ICP-OES) using a Perkin Elmer Optima 2100 DV ICP-OES. This technique
requires sediment samples to be brought into solution by acid digestion. There are a
variety of methods of acid digestion and Yaffa and Farmer (2006) note that the lack of
standardisation within the literature makes comparison between studies difficult. Most
methods involve extraction with aqua regia or various concentrations and/or combinations
of nitric acid (HNO3), perchloric acid (HClO4), sulphuric acid (H2SO4), hydrochloric acid (HCl),
and hydrofluoric acid (HF) (Cook et al., 1997; Ure and Davidson, 2002). HF provides the
most complete digestion as it dissolves metals associated with the silicate matrix, but its
use is highly hazardous and such aggressive digestion is not required to digests the
organically bound fraction relevant to this study. Although HNO3 only provides pseudo-
total concentration data, it is often used when the emphasis has been on ‘environmental
pollution’ associated with anthropogenic activities (Komarek et al. 2006; Yafa and Farmer
2006), and has been employed in several studies on Pb contaminated peats in the South
Pennines (e.g. Markert and Thornton 1990; Rothwell et al., 2005, 2008a) so it’s use will
allow comparison with other local studies.
Method 4.1.1.1.
Samples of 0.2g were weighed into Teflon microwave digestion vessels (recorded to
0.001g) and digested in 10ml 15 M HNO3 in using a CEM MARSXpress Microwave System.
Microwave ovens enable the digestion of peat samples under high pressure and high
temperature, speeding up preparation time and reducing the volume of acid required (Le
Roux and De Vleeschouwer, 2010). The samples were ramped to 175 °C over 10 min, held
at 175 °C for a further 15 min, and then cooled (US EPA method 3051A). Upon cooling,
solutions were filtered through Whatman GF/C glass microfibre filter papers to remove any
remaining solid material. The filtrate was then diluted with deionised water to bring it into
analytical range, and stored at 4 °C in sterile polythene tubes prior to ICP-OES analysis.
Where possible, samples were analysed in triplicate, providing the bulk sample was of
sufficient size for subsampling. To ensure accuracy and precision, a certified reference
material (CRM) was also tested with each batch of samples. There is no commercially
available ICP-OES CRM for heavily contaminated peat, so NIST SRM 2704 (Buffalo River
Sediment) was digested and analysed with the peat samples; recovery was always within
70
10% for Pb. Aqueous Pb standard solutions ranging from 10 ppb to 10 ppm were used to
calibrate the ICP-OES. A blank 2% solution of HNO3 was also analysed.
2.3.3. Organic matter content (Papers 2 and 3)
Organic matter and the organic carbon content were used as a tracer parameters in the
mixing model outlined in Section 2.4.2. to aid the differentiation of material sourced from
the peat mass from sediment derived from the underlying geology. The organic matter
content of potential sediment sources and SS was determined using the standard loss on
ignition (LOI) method (e.g. Heiri et al., 2001; Wright et al., 2008) which determines the total
mass loss from a sample during combustion and infers this as loss of organic matter.
Sediment was oven dried at 105 °C for 24 hours, weighed into pre-weighed crucibles to
0.001g and fired in a furnace at 550 °C for 4 hours. The ashed samples were then re-
weighed and the percentage mass loss calculated. LOI accuracy is susceptible to variations
in sample size (Heiri et al., 2001) so, sample size permitting, approx. 2 g of sediment was
ignited. In Paper 2, the percentage of OM was used to estimate the organic carbon (OC)
content of the samples based on the assumption that the carbon content of OM in peat is
approximately 48% (± 1.15; Pawson, 2008).
2.4. Data analysis
2.4.1. Manipulating geospatial data (Papers 2 and 4)
Ordinary Kriging 2.4.1.1.
Surfer 8.0 is a grid based contouring graphics program which interpolates irregularly spaced
XYZ data into a regularly spaced grid which can be used to create contour maps and surface
plots. This package was used to produce geochemical maps based on the moisture
corrected Pb data generated by the FPXRF surface survey detailed in Papers 1 and 2. Maps
were produced using ordinary Kriging techniques with a linear variogram, as recommended
for small data sets with less than 250 observations.
Kriging is a geostatistical gridding method that optimally predicts values using observations
taken at a known location (Cressie, 1990). A variogram is used to fit a model of the spatial
correlation of the observed values. Kriging is thus a form of weighted averaging in which
the weights are chosen such that the error associated with the predictor is less than for any
other linear sum depending on the location of the points used and the covariation in the
variogram (Hemyari and Nofziger, 1987). Ordinary kriging has proved a useful tool for
71
investigating and mapping soil pollution by heavy metals (e.g. Leenaers et al., 1990; Atteia
et al., 1994; Shi et al., 2007; Hani and Pazira, 2011).
TAS GIS 2.4.1.2.
Terrain analysis is the process of extracting information from digital elevation models
(DEMs) (Pike, 2000) and TAS (Terrain Analysis System) is a freely available software
package, designed to perform spatial analysis for hydro-geomorphic applications (Lindsay,
2005). TAS was used to manipulate surface Pb data, and topographic data based on a LiDAR
DEM (2 m ground resolution, 250 mm vertical accuracy) which was flown by the UK
Environment Agency under license to the National Trust in December 2002 and made
available through the Moors for the Future partnership.
Figure 2.14: Interpolated surface Pb concentrations at the field sites studied in Papers 1 and 2 produced using Surfer 8.0 and TAS GIS: (a) degraded, (b) re-vegetated, (c) intact.
72
Catchment delineation 2.4.1.2.1.
The LiDAR DEM was used to define flow pathways, delineate catchment boundaries and
derive catchment areas. These calculations were based on the Quinn et al. (1995)
modification of FD8 flow routing algorithm (Freeman, 1991; Quinn et al. 1991).
Catchment Pb storage 2.4.1.2.2.
The Surfer grid files described in Section 2.4.1.1. were imported into TAS as an ASCII XYZ
vector file. This was converted to a TIN and rasterised to produce a map of Pb surface
storage which could be manipulated in reference to topographic data. Evans and Lindsay
(2010a) derived an algorithm by which the areal extent of erosional gullies was derived by
combining areas of low difference from mean elevation and high positive plan curvature.
The gully extent map produced by Evans and Lindsay (2010a) was combined with the
interpolated Pb surface plot. The resulting maps depict Pb concentration across interfluve
surfaces (Figure 2.14) and were used to calculate surface Pb storage for individual
catchments.
Mean upslope gully depth (MUGD) 2.4.1.2.3.
Rothwell et al. (2010b) found that mean upslope gully depth (MUGD) is a major control on
sediment-associated Pb concentrations and used TAS to produce a map depicting MUGD
for the Bleaklow area. This map was used to aid site selection, and derive MUGD values for
individual study catchments in order to assess relationships between MUGD and modelled
SS sources in the eroding and restored field areas.
2.4.2. Modelling suspended sediment source (Papers 2 and 3)
Over the past few decades there has been growing awareness of the wide-ranging
environmental significance of suspended sediment transport by rivers which has generated
a considerable body of work on the subject, most notably by Des Walling (e.g. Walling et
al., 1979, 1999, 2001, 2002, 2009; Walling and Moorehead, 1987, 1989; Walling, 2005;
Walling and Woodward, 1992, 1995; Collins and Walling 2002, 2004, 2007; Owens and
Walling, 2002, 2003).
The suspended load will commonly represent a mixture of sediment derived from different
locations and from different source types within a catchment. Identifying the source of
suspended sediment is of key importance for understanding fluvial geomorphic process
and systems (e.g. Collins and Walling, 2004). Sediment source can exert a key control on
both the physical and geochemical properties of suspended sediment, which in turn exert a
73
fundamental control over the magnitude of sediment-associated nutrient and contaminant
fluxes (Walling, 2005). Knowledge of the relative contributions from various catchment
sources is therefore an essential precursor to the design and implementation of effective
sediment management aimed at minimizing sediment and nutrient export from the
surrounding catchment (Hatfield and Maher, 2008). Resources could be wasted if control
measures focussed on reducing surface erosion, when most of the sediment transported
through a river system was contributed by channel and gully erosion (Walling, 2005).
Traditionally, information on sediment provenance was obtained using a range of indirect
measurement or monitoring techniques, aimed at either identifying areas from which
sediment is being mobilized or comparing rates of sediment mobilisation from potential
source areas, in order to assess their likely relative contributions which were frequently
hampered by problems of spatial and temporal sampling, operational difficulties and the
costs involved (Walling et al., 2008). Sediment source fingerprinting aims to provide
quantitative information on the relative importance of potential sources of suspended
sediment. It involves collecting a sample of the suspended sediment transported and
comparing its physical or geochemical properties with those of potential sources within the
catchment (Walling, 2013). The fingerprinting approach has been increasingly adopted as a
more direct and reliable means of gathering such information as it provides a simple and
cost-effective means of assembling spatially- and temporally-integrated data (Collins and
Walling, 2004). This method involves two stages:
1. The selection of a suit of physical or chemical properties which clearly differentiate
potential source materials.
2. The comparison of measurements of the same property obtained from suspended
sediment with equivalent values for potential sources, in order to identify the likely
source of that sediment.
Properties which have previously been utilised include: geochemistry (e.g. Douglas et al.,
2003), environmental magnetism (e.g. Oldfield et al., 1985), plant pollen (e.g. Brown, 1985)
and the activity of fallout radionuclides (e.g. Walling and Woodward, 1992). By carefully
selecting the composite fingerprint, and including a substantial number of fingerprint
properties with contrasting origins, environmental behaviour and controls, it is possible to
discriminate between several potential sources and to quantify their relative contributions
to the sediment load of a stream (Walling, 2005). The fingerprinting approach has mainly
been applied to determine sources of fine (< 63 µm) sediment in mineral dominated
74
catchments (e.g. Carter et al., 2003; Devereux et al., 2010; Collins et al., 2010; Smith et al.,
2013), but Hutchinson (1995) and Rothwell et al. (2005) successfully derived fingerprints
for peatland catchment sources, allowing them to distinguish between ‘clean’ subsurface
and contaminated near-surface peat particulates by using a combination of geochemical,
magnetic and radiometric techniques. However, neither study went on to fully quantify the
relative importance of these two sources.
Relative contributions from potential sediment sources can be determined using a set of
linear equations that represent the value of an individual tracer property in sediment as a
function of the sum of the values of that tracer for each source multiplied by the unknown
proportional contribution from each source (Smith and Blake, 2014). This is commonly
referred to as a mixing model. A variety of correction factors are often included to account
for variance in tracer properties of the potential sources, and potential physical and
chemical changes during fluvial erosion and transportation (e.g. Walling, 2013). Tracer data
also often undergoes pre-treatment for particle size and organic matter differences
between source soils and sediment (e.g. Gruszowski et al., 2003; Collins et al., 2012a).
However, there is growing criticism of this kind of adjustment (e.g. Koiter et al., 2013;
Smith and Blake, 2014) and several fingerprinting studies have not applied these correction
factors (e.g. Martinez-Carreras et al., 2010; Evrard et al., 2011). Uncertainty associated with
source properties which display variable characteristics is addressed by utilising Monte
Carlo techniques (e.g. Collins and Walling, 2007; Collins et al., 2010; Martinez-Carreras et
al., 2010) to perform multiple iterations of the mixing model using different possible values
of the source properties. The goodness of fit (GOF) can be assessed using the relative mean
error (RME) (e.g. Collins et al. 2010), based on a comparison of the measured and predicted
property values for each sample. An RME value of <10% is seen as evidence of a
satisfactory fit (Walling, 2013).
Paper 2 describes the mixing model in full, and also outlines how the model was adapted
for use in peatland systems. The mixing model has been employed to investigate landscape
scale differences in sediment composition across an erosion-restoration cycle in Paper 2,
and temporal controls on sediment release throughout storm events in Paper 3.
75
Chapter 3 Assessment of lead contamination in
peatlands using field portable XRF (Paper 1)
This chapter was published as Shuttleworth, E.L., Evans, M.G., Hutchinson, S.M., &
Rothwell, J.J. (2014) “Assessment of Lead Contamination in Peatlands Using Field Portable
XRF” Water, Air, and Soil Pollution 225:1844, DOI 10.1007/s11270-013-1844-2.
Abstract
Ombrotrophic peatlands are highly sensitive to atmospheric heavy metal deposition.
Previous attempts to quantify peatland lead pollution have been undertaken using the
inventory approach. However, there can be significant within-site spatial heterogeneity in
lead concentrations, highlighting the need for multiple samples to properly quantify lead
storage. Field portable x-ray fluorescence (FPXRF) continues to gain acceptance in the study
of contaminated soil, but has not thus far been used to assess peatland lead
contamination. This study compares lead concentrations in surface peat samples from the
South Pennines (UK) derived using: (a) FPXRF in the field; (b) FPXRF in the lab on dried
samples; and (c) ICP-OES analysis. FPXRF field and lab data are directly comparable when
field measurements are corrected for water content; both can be easily used to estimate
acid extractable lead using regression equations. This study is a successful demonstration
of FPXRF as a tool for a time- and cost-effective means of determining the lead content of
contaminated peatlands, which will allow rapid landscape scale reconnaissance, core
logging, surface surveys, and sediment tracing.
Keywords: FPXRF; Organic matter; High moisture content; Pollution; Heavy metals; In situ
measurement; Data quality
76
3.1. Introduction
Anthropogenic lead (Pb) pollution has long been recognised as a global phenomenon
dating back more than 3,000 years. A wide variety of environmental archives have been
used to reconstruct the spatial and temporal patterns of Pb deposition. These include lake
sediments (e.g. Renberg et al., 1994, 2001; Bränvall et al., 2001), ice cores (e.g. Murozumi
et al., 1969; Hong et al., 1994; Zheng et al., 2007); and peat (e.g. Lee and Tallis, 1973;
Shotyk et al., 1998).
Rain-fed (ombrotrophic) peatlands in particular are highly sensitive to atmospheric
deposition (Shotyk, 1998). Peatland soils in close proximity to urban and industrial areas
can be contaminated with atmospherically deposited heavy metals. The strong
complexation of Pb to organic matter (OM) (Stevenson, 1976; Vile et al., 1999) means that
peatlands can represent significant sinks of Pb (Shotyk et al., 2000; Bindler et al., 2004;
Farmer et al., 2005; Rothwell et al., 2007a, 2010a). Peat cores can be used to reconstruct
long-term Pb deposition and pollution histories as peatlands retain a record of atmospheric
metal deposition (e.g. Lee and Tallis, 1973; Shotyk et al., 1998; Marx et al., 2010). Recently,
peatland Pb research has focussed on: the reconstruction of atmospheric Pb inventories
within and between regions (e.g. Weiss et al., 1999; Shotyk et al., 2003; Novak et al., 2003;
De Vleeschouwer et al., 2007; Rothwell et al., 2007a, 2010a); mobility of Pb within the peat
profile (e.g. Mackenzie et al., 1998; Vile et al., 1999; Novak et al., 2011); release of Pb into
the fluvial system (e.g. Tipping et al., 2003; Rothwell et al., 2005, 2007b, 2008a, 2010b;
Shotbolt et al., 2006; Dawson et al., 2010); and the timing and magnitude of mining and
smelting impacts (e.g. Kempter and Frenzel, 2000; Monna et al., 2004; Mihaljevič et al.,
2006; Hürkamp et al., 2009a).
Many peatlands in the UK are actively eroding which has the potential for such catchments
to turn from Pb sinks into sources, and release Pb to the fluvial system (Shotbolt et al.,
2006; Rothwell et al., 2005, 2007b, 2008a; Dawson et al., 2010). Pb can have toxic effects
on terrestrial plants, invertebrates and microorganisms (Tyler et al., 1989); it is known to
have a variety of effects on the human nervous and circulatory systems, and is relatively
toxic at low concentrations (Vile et al., 2000). Quantifying Pb contained in actively eroding
peatlands is vital in order to understand Pb storage and release in such systems, assess
potential ecological damage and risk to human health, and formulate mitigation strategies
(Smith et al., 2005).
77
Previous attempts to quantify peatland Pb pollution using the inventory approach have
found significant within-site spatial heterogeneity in Pb concentrations (Bindler et al., 2004;
Farmer et al., 2005; Rothwell et al., 2007a). In heavily eroded areas, this is further
complicated by gullying and the removal of surface material. This small-scale variability
highlights the need to analyse multiple samples to properly quantify Pb storage.
Conventional geochemical analyses (e.g. acid digestion followed by ICP or AAS analysis) of
peat are often time consuming, costly, and can result in sample destruction limiting further
analysis. Despite relatively high detection limits when compared with lab-based analyses,
field portable x-ray fluorescence (FPXRF) continues to gain acceptance in the study of metal
contaminated soil (VanCott et al., 1999; Kalnicky and Singhvi 2001; Martín Peinado et al.,
2010). It allows a large number of samples to be processed in situ in a relatively short time,
giving a high level of detail with little disturbance to the surrounding area. FPXRF also
offers significant advantages over off-site laboratory analysis in terms of on-site decision-
making and faster turnaround of results. When compared with other analytical methods
(ICP, AAS, etc.), estimates of elemental concentrations in soils made using FPXRF have
given data of acceptable quality (e.g. Shefsky 1997; Kilbride et al., 2006; Makinen et al.,
2006; Radu and Daimond 2009). However, analysis is traditionally restricted to fine,
inorganic material, and there is limited information about the suitability of FPXRF when
analysing other matrices, such as peat.
Dried peat samples are routinely analysed using laboratory-based XRF as compressed
powder or pellets; for organic samples, measurements on powder are generally preferred
as the samples can be used for subsequent analyses (Le Roux and De Vleeschouwer 2010).
However, in situ, peat has a high moisture content (typically 90% or more), which can affect
the accuracy of XRF analysis (Argyraki et al., 1997; U.S. EPA 1998). Water absorbs and
scatters x-rays, thus scattering primary x-rays and reducing the intensity of characteristic x-
rays, resulting in lower precision and accuracy, and increased detection limits (Ge et al.,
2005). The fluorescence signal emitted from the sample surface is a function of the
composition of the sediment. Large quantities of light elements (such as carbon found in
OM) can cause a dilution effect, lowering apparent concentrations of heavier elements
(Löwemark et al., 2011). However, this interference may be less for elements with higher
energy x-ray lines such as Pb (Kalnicky and Singhvi 2001), and samples with a moisture
content significantly higher than 20% have been successfully analysed by FPXRF when
confirmation samples are also analysed ex situ (Bernick et al., 1995; Hürkamp et al., 2009b).
78
Solo-Gabriele et al. (2004) have shown FPXRF arsenic detection is not affected by the
moisture content of treated wood which also has a high OM content.
This study aims to assess the efficacy of FPXRF analysis of Pb-contaminated peat in the
Peak District, southern Pennines, UK. In situ field measurements are compared with
concentration data derived ex situ (on processed samples) by the same FPXRF unit and the
more traditional and widely used method of nitric acid (HNO3) extraction and inductively
coupled plasma optical emission spectroscopy (ICP-OES) analysis. Although HNO3 only
provides pseudo-total concentration data, it is often used when the emphasis has been on
‘environmental pollution’ associated with anthropogenic activities (Komarek et al., 2006;
Yafa and Farmer 2006), and has been employed in several studies on Pb-contaminated
peats in the study area (e.g. Markert and Thornton 1990; Rothwell et al., 2005 2007a 2007b
2008a). The ex situ FPXRF results are compared with the acid digest data to assess the
FPXRF unit’s use as a laboratory tool. The effects of sample moisture content and analysis
time are also considered.
3.2. Materials and Methods
3.2.1. Field Area
The Bleaklow Plateau (505 – 633 m) in the Peak District National Park, Northern England
(Figure 3.1) is characterised by extensive, deep blanket peats. The area lies in close
proximity to the industrial cities of Manchester and Sheffield. Consequently, the near-
surface layer of the blanket peat is contaminated by high concentrations of
anthropogenically derived, atmospherically deposited Pb (in excess of 1600 ppm; Rothwell
et al., 2007a) and the area has been a focus of heavy metal contamination research for
several decades (e.g. Lee and Tallis 1973; Livett et al., 1979; Markert and Thornton 1990;
Jones and Hao 1993; Smith et al., 2005; Rothwell et al., 2007a, 2007b, 2008a, 2010b). The
area supports a range of ecosystem services such as farming, water provision, and
recreation, and lies at the southern climatic boarder of blanket bog distribution (Bonn et
al., 2009). These external pressures have led to severe environmental degradation and
erosion is widespread (Bower 1960b, 1961; Tallis 1985; Bonn et al., 2009); the removal of
material has led to large expanses of exposed peat surfaces intersected by gullies.
79
Figure 3.1: Study area. Grey-hatched area denotes location of sampling sites.
3.2.2. Field Survey
A total of 159 in situ FPXRF Pb measurements were taken to spatially characterise and map
surface Pb concentrations as part of a wider study (Shuttleworth et al., 2012). A handheld
Niton XL3t 900 XRF analyser was used to obtain data across 15.25 ha of peatland in a
gridded pattern, covering a range of surface conditions (severely degraded to intact), over a
three day period. The analyser was internally calibrated using the ‘soil’ function. Analysis
time was set to 120 seconds as recommended by Ridings et al. (2000) and Kilbride et al.
(2006). The accuracy of the method was corroborated by analyses of certified reference
material (CRM). There is no commercially available XRF CRM for heavily contaminated peat
so NCS DC73308 (Chinese stream sediment) was used as this has the most appropriate Pb
concentration of the CRMs available to the study. The relative percent difference (RPD)
between the concentration in the reference material and the concentration measured by
FPXRF was within 10% for Pb. Where necessary, vegetation was removed and the peat’s
surface was lightly compacted by hand in order to present a smooth flat surface to the XRF
sensor (Ridings et al., 2000).
Samples from the top 10-15 mm of each site were collected using a stainless steel palette
knife in order to determine the water content of the peat following lab analysis.
80
3.2.3. Laboratory Analysis
Moisture Content 3.2.3.1.
Surface samples were oven-dried at 40°C until a constant weight was achieved (typically 6
days). The moisture content at each sampling site was calculated based on the difference
between the wet and dry masses of these samples.
Sample preparation 3.2.3.2.
A stratified random subset of 40 samples was selected for ex situ chemical analysis. In
order to represent the full range of Pb concentrations found across the field site, the 159 in
situ measurements were sorted by Pb concentration, divided into quartiles, and 10 samples
were selected at random from each quartile.
There is no standard method for ex situ XRF analysis of organic samples; however, it is
generally agreed that samples need to be dried, ground, and sieved in order to homogenise
the sample and present a fine matrix to the XRF sensor. The oven-dried samples were
therefore ground to a fine powder using an agate ball mill and passed through a 250 μm
sieve (Argyraki et al., 1997; Clarke et al., 1999; Ridings et al., 2000).
Ex situ FPXRF 3.2.3.3.
Each sample was subsampled three times. Subsamples were analysed as loose powders
pressed into sample cups fitted with a 6 µm thick polyester film which provides a thin film
sample window. Sample cups were placed in the FPXRF laboratory sample support
stand and analysed for 120 seconds. NCS DC73308 was again used as CRM. The RPD
between the concentration in the reference material and the concentration
measured by FPXRF was within 10% for Pb.
Analysis time 3.2.3.4.
Six powdered peat samples were selected to represent the range of Pb
concentrations across the field sites. Pb concentrations in this subsample set ranged
from 75 to 1700 ppm. Samples were analysed for 30, 60, 120, 180, 300, and 600 seconds.
Analysis times in excess of 600 seconds were not considered necessary as these would not
be practical for use in the field.
The output from the Niton XL3t 900 FPXRF includes the two-sigma (two standard deviations
from the mean) margin of error for each reading (Niton XL3t 900 Product Specifications).
81
These error values were used to calculate the coefficient of variation (CV) for each reading
and thus give an indication of the precision of the measurements.
Acid extraction (HNO3) 3.2.3.5.
The same subsamples used for ex situ XRF analysis were used for acid extraction; 0.2 g of
each subsample was digested in 10ml 15 M HNO3 in Teflon microwave digestion vessels
using a microwave apparatus (MARS Xpress, CEM). The peat samples were ramped to
175 °C over 10 min, held at 175 °C for a further 15 min, and then cooled (US EPA method
3051A). Upon cooling, solutions were filtered through Whatman GF/C glass microfibre filter
papers to remove any remaining solid material. The filtrate was then diluted with deionised
water to bring it into analytical range, and stored at 4 °C in sterile polythene tubes prior to
analysis. Pb concentrations were determined using ICP-OES (Perkin Elmer Optima 2100
DV). There is no commercially available ICP-OES CRM for heavily contaminated peat, so
NIST SRM 2704 (Buffalo River Sediment) was digested and analysed with the peat samples;
recovery was within 10% for Pb.
3.2.4. Moisture correction
A simple equation was used to account for the dilution effect of the high moisture content
of the peat:
𝑪𝒄 =𝑪𝒇.𝒎𝒘
𝒎𝒅 Equation 3.1
Where: Cc is the corrected Pb concentration, Cf is the raw field measurement of Pb
concentration, mw is the wet mass of the sample, md is the dry mass of the sample.
3.2.5. Statistical analyses
The assessment of the quality of data produced by using the FPXRF in situ is adapted from
similar assessment carried out by Kilbride et al. (2006). Parameters produced by linear
regression analysis were used to assess the strength, precision and accuracy of the
relationship between in situ and ex situ derived data (Table 3.1). The data were then
assigned to one of three quality levels (Table 3.2). The sequence of statistical analyses
which were carried out is outlined in Figure 3.2. Data were log transformed prior to analysis
in order to satisfy the assumption that the residuals of the linear regressions should follow
a normal distribution (Ebdon, 1985). Two samples had to be removed at this stage as they
had produced concentrations below the limit of detection (<LOD) for both in situ and ex
situ measurements and therefore could not be log transformed.
82
Table 3.1: Parameters produced by liner regression analysis which are used to assess the relationships between the various analyses.
Table 3.2: Criteria for assigning relationship quality (adapted from Kilbride et al. 2006).
Parameter Description
R2 Fraction of total variation in the data from one method which is accounted for by its relationship with another method. Can be used as a measure of the strength of the linear association between the two methods.
Relative standard deviation (RSD)
Standard deviation of the sample mean relative to the true mean (i.e. a measure of the dispersion of data points about the linear trend line). Can be used as a measure of precision.
Inferential statistics on linear model parameters
Null hypotheses: c = 0, and m = 1, tested at the 0.05 level, compare the regression model to a y = x relationship. Can be used to assess the accuracy of the tested method.
Data quality level Statistical requirement
Definitive R2 = 0.85 – 1
RSD < 10%
Inferential statistics must indicate the two data sets are statistically similar, i.e. the relationship y = x is accepted.
Quantitative R2 = 0.7 – 1
RSD < 20%
Inferential statistics indicate the two data sets are statistically different, i.e. the relationship y = mx or y = mx + c is accepted.
Qualitative R2 < 0.7
RSD > 20%
Inferential statistics indicate the two data sets are statistically different.
83
Figure 3.2: Sequence of statistical analyses carried out to assess the quality of linear relationships. T-test satisfied at 0.05 confidence level.
84
3.3. Results
3.3.1. Analysis Time
Figure 3.3 shows the effect of increasing analytical time on the CV produced by the FPXRF
when analysing samples containing varying Pb concentrations. The CV decreases as a
power function of analysis time for all samples, producing regressions with power functions
of approximately -0.5 (statistically similar to each other, paired t-test, p=0.025). The details
of each regression are summarised in Table 3.3.
Figure 3.3: Coefficients of variation (CV) produced for peat samples containing various concentrations of Pb with increasing ex situ FPXRF analysis time. Superscript a denotes certified reference material.
85
Pb (ppm) Power function R2
27a -0.50 1.00
75 -0.55 0.75
130 -0.51 0.94
400 -0.51 0.98
650 -0.50 1.00
1130 -0.50 0.99
1700 -0.51 0.99
a Certified Reference Material
Table 3.3: Summary of parameters produced by regressions of time dependant FPXRF analysis.
In the majority of samples, the largest decrease in CV can be seen when analysis time is
increased from 30 to 60 seconds (approximately 38% reduction). A further large decrease
in CV occurs between 60 and 120 seconds of analysis time (approximately 27% reduction).
For analysis times greater than 120 seconds, CV continues to reduce but by progressively
smaller increments with additional time.
Samples with lower Pb concentrations produce larger CVs than those with higher Pb
concentrations regardless of analysis time. All peat regressions tend towards CVs of <1.5%
after 600 seconds. There is little difference in the regressions produced by samples
containing Pb concentrations greater than 400 ppm. Samples containing lower Pb
concentrations have markedly larger CVs; the difference is especially pronounced when
analysis time is shorter.
86
a Figures in parentheses are superfluous as relationship quality has already been decided.
Table 3.4: Statistics and quality levels for raw and moisture-corrected in situ FPXRF analysis in relation to ex situ FPXRF analysis.
N Residuals normal? R2 RSD c Accept H0
(c=0) M Accept H0
(m = 1) Quality level
Raw 40 Yes 0.86 10.50 (-1.71)a (No) (0.74) (No) Quantitative
Corrected 40 No (0.92) (6.03) (-0.48) (Yes) (0.96) (Yes) Invalid test
Corrected, outliers removed
38 Yes 0.96 3.95 -0.44 Yes 0.96 Yes Definitive
87
Figure 3.4: Linear regressions of logged Pb concentrations (ppm): a) raw in situ and ex situ FPXRF; b) moisture-corrected in situ and ex situ FPXRF; c) ex situ FPXRF and ICP-OES; and d) moisture-corrected in situ
and ICP-OES analyses. Regression lines are shown as solid black lines. Outliers removed from the final regression are shown as open circles. Where appropriate, graphs also display a regression line which passes
through the origin (dashed black line). The 1:1 line is also shown (grey line).
88
3.3.2. Relationship between in situ and ex situ FPXRF analysis
The statistical criteria and quality levels for the in situ data are shown in Table 3.4.
Raw data 3.3.2.1.
There is a strong relationship between raw in situ and ex situ data (R2 = 0.86); however, the
in situ readings significantly underestimate the ex situ derived concentrations (Figure 3.4a)
and data points are spread out around the trend line (RSD = 10.5%). The raw in situ data
therefore only achieved a quantitative quality level.
Moisture-corrected data 3.3.2.2.
Moisture contents ranged from 43.32 to 87.59 % at the eroded site, and 76.97 to 85.24 %
at the intact site. Correcting the in situ data for moisture content produced two significant
outliers (highlighted in Figure 3.4b). Removing these samples (see discussion for
justification) from the regression made little difference to the linear model, which then
produced a valid definitive quality level (R2 = 0.96, RSD = 2.38%). Figure 3.4b shows the
linear regression and the ideal (y = x) correlation between the moisture-corrected in situ
measurements and ex situ data.
n Residuals normal?
R2 RSD C Accept
H0
(c=0)
M Accept H0
(m = 1)
Quality level
Ex situ 40 Yes 0.99 1.75 -0.63 No (1.06)a
(Yes) Quantitative
In situ corrected
38 Yes 0.94 4.59 -1.02 No (1.02) (Yes) Quantitative
a Figures in parentheses are superfluous as relationship quality has already been decided.
Table 3.5: Statistics and quality levels of FPXRF analysis in relation to ICP-OES analysis.
3.3.3. Relationship between FPXRF and ICP-OES analysis
The statistical criteria and quality levels for the XRF data are shown in Table 3.5.
Ex situ 3.3.3.1.
There is an excellent relationship between ex situ XRF and ICP-OES data (R2 = 0.99, RSD =
1.75%); however, FPXRF readings overestimate ICP-OES derived Pb values at all but the
lowest concentrations of Pb (Figure 3.4c). These two methods therefore exhibit a strong
quantitative relationship.
89
In situ 3.3.3.2.
The moisture-corrected in situ data with outliers removed was used to assess the
relationship between in situ readings and ICP-OES derived data. This produces a strong
relationship (R2 = 0.94, RSD = 4.59%). XRF readings overestimate high ICP-OES derived Pb
concentrations while underestimating lower concentrations (Figure 3.4d) leading to a
quantitative relationship.
3.4. Discussion
3.4.1. Analysis time
It has been widely reported that additional analysis time increases the precision of FPXRF
units (e.g. Hou et al., 2004; Raab et al., 2005; Kilbride et al., 2006; Block et al., 2007). This
holds true for the analysis of peat samples, as demonstrated by the reduction in CV with
additional analysis time (Figure 3.3).
The effect of additional analysis time is most pronounced in samples containing lower Pb
concentrations. In samples containing Pb concentrations of 400 ppm Pb and above, the
level of precision is similarly high across all analysis times (CVs < 4%) despite the large
range of Pb concentrations (up to 1700 ppm). In samples containing less than 400 ppm Pb,
this high level of precision is lost. At lower Pb concentrations, there is marked
improvement in precision as analysis time is increased to 120 seconds, after which
precision continues to improve but additional time makes less of an impact. In all samples
analysed, CVs are less than 5% after 120 seconds, which is regarded as ‘excellent’ precision
by Martín Peinado et al. (2010).
R2 values for all samples are in excess of 0.98 (Table 3.3) with the exception of the peat
samples with two lowest Pb concentrations (130 ppm, R2 = 0.93; 75 ppm, R2 = 0.75).
However, the CRM which contains 27 ppm Pb produces a strong regression (R2 = 1.00), so it
is unlikely that the strength of the relationship is affected solely by Pb concentration.
Sturgeon (2000), Hou et al. (2004), and Löwemark et al. (2011) have reported that the
lighter elements that make up OM (carbon, hydrogen and nitrogen) can have a diluting
effect, resulting in lower relative concentrations of elements of interest. Therefore, it is
probable that the high OM content of peat is affecting the FPXRF’s performance when
analysing samples containing lower Pb concentrations.
In order to achieve high precision results across all Pb concentrations, analysis conditions
should be maximised in order to meet the requirements of low Pb concentrations (Kilbride
90
et al., 2006). However, sample throughput for the FPXRF is highly affected by analysis time
(Kilbride et al., 2006), and the FPXRF has been designed for use in the field where lengthy
analysis times and sample preparation (such as OM removal) are not practical. Although
the results indicate that precision is improved by longer analysis times in samples
containing lower Pb concentrations (as also found by Sterling et al., 2000; Hou et al., 2004;
Raab et al., 2005), CVs fall below the 5% recommended by Martín Peinado et al. (2010)
after only 120 seconds of analysis time. Increasing analysis time has little effect on
concentration reading; therefore, a count time of 120 seconds is appropriate for the
analysis of Pb-contaminated peat. This is in agreement with the recommendations of
Ridings et al. (2000) and Kilbride et al. (2006) for FPXRF analysis of soil.
3.4.2. Detection Limit
Where ‘<LOD’ is reported as a result, the error column on the FPXRF output contains the
estimated detection limit for that measurement rather than the error (USEPA 2008).
Eighteen field readings reported Pb as <LOD with detection limits ranging from 1.37 to 1.97
ppm. When corrected for moisture content, this becomes 4.05 to 8.67 ppm. The lowest
valid in situ Pb concentration recorded was 9.17 ppm (when corrected for moisture)
indicating that the detection limit for Pb when analysing peat in situ is approximately 9
ppm. Two ex situ measurements recorded Pb as <LOD with detection limits ranging from
1.37 to 1.84 ppm. This indicates that the operational detection limit of the unit is
approximately 2 ppm for Pb after 120 seconds of analysis time. However, Raab et al. (2005)
recommend that results which are near the detection limit should be discarded, and that
the detection limit should be defined as 3.3 times higher. This would increase the detection
limit of the unit to 7 ppm after 120 seconds of analysis time. This is similar to the detection
limit found by Raab et al. (2005) for Pb in alluvial soils, and compares favourably with ICP-
OES analysers which have detection limits ranging from <1 to 25 ppm for Pb, after solid
concentrations have been calculated from the concentration of the analysed solution
(Perkin Elmer, 2011; EAG, 2007).
3.4.3. Moisture content
The raw in situ FPXRF data display a strong linear relationship with ex situ derived values.
However, the field-based readings significantly underestimated the concentrations
produced after samples were processed for ex situ analysis (Figure 3.4a). Raw in situ data
also do not produce a very precise estimate of ex situ FPXRF data, as shown by the high RSD
(10.50%; Table 3.4) and the spread of data points about the regression line (Figure 3.4a).
91
This is likely due to the high moisture content of peat in situ scattering primary x-rays and
absorbing the characteristic secondary x-rays of Pb (Hou et al., 2004). This effectively
dilutes the secondary x-ray signal received by the XRF detector, decreasing the apparent Pb
concentration (Kalnicky and Singhvi, 2001).
Correcting for the effects of sample moisture produced two significant outliers (Figure
3.4b). Although attempts were made to collect the same peat that was measured in situ,
the samples were extracted from a larger area of the peat’s surface than is analysed by the
FPXRF sensor. Pb deposition and storage in peat has been shown to be heterogeneous
(Bindler et al., 2004) so the Pb content of the samples analysed ex situ may not be truly
representative of the in situ analysis. Outliers may also be indicative of the ‘nugget’ effect
as described by Kalnicky and Singhvi (2001) where the sample may have contained a small
particle of analyte (in this case Pb) which is only picked up by either the in situ or ex situ
measurement. It was therefore deemed acceptable to remove the outliers from the linear
regression.
Correcting for moisture content produces in situ concentrations directly comparable to ex
situ derived data. This also significantly increases the precision of the estimate (RSD =
3.95%) and the strength of the relationship (R2 = 0.96) producing definitive quality data.
This indicates that by applying the moisture correction, in situ field data are of the same
concentration and standard as data produced under controlled lab conditions.
Despite achieving this definitive relationship, the moisture-corrected in situ data slightly
underestimates concentrations from ex situ analysis. In situ, peat typically has a very loose
and heterogeneous structure (Van Asselen and Roosendaal, 2009) presenting an open
matrix to the XRF sensor which will reduce the apparent concentration of Pb (US EPA,
1998).
3.4.4. Quality of relationship between FPXRF and acid extraction
Both ex situ and moisture-corrected in situ FPXRF Pb concentrations correlate strongly with
ICP-OES derived Pb values (R2 = 0.99 and 0.94 respectively). However, neither method
produces a definitive relationship with ICP-OES (Table 3.5).
Ex situ FPXRF analysis consistently overestimates ICP-OES concentrations (Figure 3.4c). This
is because HNO3 only gives pseudo-total concentrations as it digests the organically bound
fraction (Rothwell et al., 2010a), while the FPXRF measures the total element concentration
(Boyle, 2000). The discrepancy is more pronounced in samples containing higher
92
concentrations of Pb. These higher Pb concentrations are found in the upper 10 to 15 cm of
the peat profile (Rothwell et al., 2007a), and have been found to have a different isotopic
signature to the lower Pb concentrations found further down the profile due to differences
in source; surface Pb bears similar isotopic ratios to imported Pb ores and leaded petrol,
while the isotopic signature further down is similar to Pb found in local ores and British coal
(Rothwell et al., 2010a). Anthropogenic Pb is, on the whole, more labile than background
Pb and so is extractable by HNO3 (Teutsch et al., 2001; Komárek et al., 2006); however, it is
possible that Pb from different sources is differentially available to HNO3 digestion.
The relationship is similar for corrected in situ FPXRF data, but the correlation is slightly less
strong, due to the uncertainty caused by the scattering effects of moisture (Ge et al., 2005)
and the loose peat density (US EPA 1998).
3.4.5. FPXRF as an alternative for acid extractible method
For comparison with existing research on Pb-contaminated peatlands, both in situ and ex
situ FPXRF measurements can be easily used to estimate acid extractable Pb concentrations
using regression equations.
FPXRF’s main advantage over ICP-OES is the rapid turnaround of results. Processing the 40
samples (120 subsamples) for ICP-OES took eight labour intensive days in the lab (plus the
three field days and sample drying time). This was partially due to the length of time it
takes to digest, filter, dilute and analyse samples, and partly due to restrictions on the
number of samples that can be processed at any one time. By contrast, the most intensive
of the three days of field surveying yielded 81 FPXRF readings across an area of
approximately 6 ha, and while the subsequent drying caused a delay in obtaining corrected
results, it required little extra lab work. If a soil moisture field sensor was also used in situ
this would negate the need for sample collection, giving a fast turnaround on a large
number of readings. Even when the FPXRF is used ex situ, processing and analysing the 120
subsamples took four labour intensive days (plus the initial six days of drying), effectively
halving the amount of time spent on lab analysis.
When used ex situ, not only does the minimal sample pretreatment significantly reduce
analysis time, it also negates the need to use and dispose of harmful chemicals and does
not require additional equipment (e.g. microwave, filtration) reducing running costs to the
project. Moreover, the ex situ method is non-destructive, leaving samples available for
further analysis. The use of FPXRF in situ causes minimal disturbance to the surrounding
93
area which is especially important in environmentally delicate areas such as eroding
peatlands. The production of rapid, real time Pb concentration data in situ can aid on site
decision-making. Again this is pertinent to peatlands where Pb deposition and storage is
heterogeneous and Pb concentrations can vary over relatively small distances (Bindler et
al., 2004; Rothwell et al., 2007a).
3.5. Conclusions and recommendations
FPXRF shows great promise as a tool for a rapid and cost-effective means of determining
the Pb content of contaminated peatlands. Pb concentrations derived by correcting in situ
readings for moisture content correlate strongly with results obtained using the unit ex
situ. While analysis time may influence the precision of data produced by FPXRF analysis,
CVs fall below 5% after only 120 seconds of analysis time and increasing analysis time has
little effect on concentration readings. One hundred and twenty seconds is deemed to be a
sufficient analysis time for use both in situ and ex situ. Both in situ and ex situ FPXRF
readings correlate strongly with ICP-OES. The resulting linear regression gives a conversion
equation to express FPXRF readings as estimated ICP-OES results, for comparison with
other studies.
As a cautionary note, this study was limited to one area of the Peak District so it is unclear
if these correlations hold true for all Pb-contaminated peatlands which may contain
different levels or sources of Pb contamination, or where the peat may have different
physical properties (e.g. density, moisture, and OM content). Confirmatory analysis on a
subset of samples representing a range of Pb concentrations by ICP-OES would be advisable
alongside FPXRF analysis (c.f. Kilbride et al., 2006), to verify the relationship in other
peatlands and organic rich soils.
FPXRF makes accurate field analysis possible, allowing a large number of measurements to
be taken over a relatively short timeframe. FPXRF could be applied in a variety of contexts;
for example, detailed Pb profiles could be rapidly determined on multiple peat cores to
reconstruct more robust Pb inventories; detailed Pb surface surveys may be conducted on
contaminated peatlands; and the method could be used in combination with sediment
fingerprinting techniques to chemically characterise potential sources of suspended
sediment, and trace the origin and fate of Pb-contaminated sediment through peatland
systems.
94
3.6. Acknowledgements
We thank The University of Manchester for the provision of a Graduate Teaching
Studentship (to E. L. Shuttleworth). We are grateful to The National Trust and United
Utilities for allowing work to be carried out at the study sites and to the University of
Manchester and Moors for the Future who provided funding for analytical costs. Thanks
also go to John Moore, Jonathan Yarwood, and Laurie Cunliffe for their assistance in the
lab, and to Jason Dortch for his help with constructing figures. Finally, we would like to
thank the reviewers for their helpful comments and suggestions.
95
Chapter 4 Peatland restoration: controls on sediment
production and reductions in carbon and pollutant export
(Paper 2)
This chapter has been published as Shuttleworth, E.L., Evans, M.G., Hutchinson, S.M., &
Rothwell, J.J. (2014) “Peatland restoration: controls on sediment production and reductions
in carbon and pollutant export” Earth Surface Processes and Landforms DOI:
10.1002/esp.3645
Abstract
Peatlands are an important store of soil carbon, and play a vital role in global carbon
cycling, and when located in close proximity to urban and industrial areas, can also act as
sinks of atmospherically deposited heavy metals. Large areas of the UK’s blanket peat are
significantly degraded and actively eroding which negatively impacts carbon and pollutant
storage. The restoration of eroding UK peatlands is a major conservation concern, and over
the last decade measures have been taken to try to control erosion and restore large areas
of degraded peat. This study utilises a sediment source fingerprinting approach to assess
the effect of restoration practices on sediment production, and carbon and pollutant
export in the Peak District National Park, southern Pennines (UK). Suspended sediment was
collected using time integrated mass flux samplers (TIMS), deployed across three field
areas which represent the surface conditions exhibited through an erosion-restoration
cycle: (i) intact (ii) actively eroding, and (iii) recently re-vegetated. Anthropogenic pollutants
stored near the peat’s surface have allowed material mobilised by sheet erosion to be
distinguished from sediment eroded from gully walls. Re-vegetation of eroding gully
systems is most effective at stabilising interfluve surfaces, switching the locus of sediment
production from contaminated surface peat to relatively ‘clean’ gully walls. The
stabilisation of eroding surfaces reduces particulate organic carbon (POC) and lead (Pb)
fluxes by two orders of magnitude, to levels comparable with those of an intact peatland,
thus maintaining this important carbon and pollutant store. The re-vegetation of gully
floors also plays a key role in decoupling eroding surfaces from the fluvial system, and
further reducing the flux of material. These findings indicate that the restoration practices
96
have been effective over a relatively short timescale, and will help target and refine future
restoration initiatives.
Key words: Sediment fingerprinting; Sediment source tracing; Organic sediment; Upland
erosion; Re-vegetation; POC; Lead
4.1. Introduction
The peatlands of the northern hemisphere hold an estimated 455 billion tonnes of soil
carbon comprising over 30% of global soil carbon storage (Batjes, 1996). Changes in the
uptake or release of soil carbon to the atmosphere may significantly affect atmospheric
carbon concentrations; the stability of peatlands is therefore a significant concern (Evans
and Warburton, 2010). In the UK, peatlands face threats from pressures such as climate
change, legacy atmospheric pollution, poor management and anthropogenic disturbance
(Bonn et al., 2009). Consequently, over the last 1000 years a significant proportion of the
UK’s blanket peat has become degraded and is actively eroding. Currently, this degree of
erosion is unusual compared to the global peatland setting; however, the
Intergovernmental Panel on Climate Change identifies peatlands as particularly vulnerable
to future land use and climate change (Parry et al., 2007). The potential for peatland
desiccation and permafrost melt due to predicted climate warming means that the physical
instability of peatlands may become more widespread in the future (Evans and Warburton,
2010). There is therefore a pressing need to understand and efficiently mitigate the
impacts of peatland erosion.
Carbon is lost from peatlands either as CO2 or CH4 produced by the microbial breakdown of
organic matter, or via the fluvial system as dissolved- or particulate- organic carbon (DOC
and POC) and dissolved inorganic carbon (DIC) (Billett et al., 2010). There has been an
increasing recognition of the importance of fluvial systems in the terrestrial carbon cycle,
but there has been limited focus on fluvial geomorphology in relation to carbon cycling in
peatlands (Pawson et al., 2012). The majority of the work examining fluvial carbon exports
from peatlands has focused on DOC (e.g. Hope et al., 1994; Dawson et al., 2002; Worrall et
al., 2004; Billett et al., 2006; Andersson & Nyberg, 2008), with less attention given to
particulate organic carbon (POC) fluxes (e.g. Pawson et al., 2008, 2012). Particulate carbon
can be the most significant vector for carbon loss from eroding peatland systems (Worrall
et al., 2003), and although there is limited information regarding the fate of fluvial POC,
97
there is evidence to suggest POC from peatland systems can undergo transformation to
DOC in the fluvial environment or become mineralized to CO2 during periods of floodplain
storage (Pawson, 2008; Pawson et al., 2012; Moody et al., 2013). As POC has the potential
to transform to atmospherically active forms of carbon, large fluxes of POC mobilised from
eroding peatlands are an important component of the greenhouse gas balance of these
systems.
In addition to their role in the terrestrial carbon cycle, peatlands can be sinks of
atmospherically deposited toxic heavy metals (Shotyk et al., 1997). These pollutants are a
legacy of past industrial activity and are often found at or near the peat surface (Rothwell
et al., 2007a). The importance of suspended sediment in the transport and biogeochemical
cycling of contaminants in the aquatic system is well recognised (e.g. Hart, 1982; Tipping et
al., 2010) and the physical erosion of peat has been highlighted as a mechanism for the
release of significant quantities of lead to surface waters (Rothwell et al., 2005, 2008a,
2008b; Shotbolt et al., 2006; Rose et al., 2012). This poses a threat to the sustainability of
aquatic ecosystem (Rhind, 2009) and could compromise downstream water resources
(Shotbolt et al., 2006).
Restoration of eroding UK peatlands has been a major conservation concern for several
decades (e.g. Tallis and Yalden, 1983; Wheeler et al., 1995; Gorham and Rochefort, 2003,
Dixon et al. 2013). Recently, there has been a move to actively restore large areas of peat
using a range of techniques such as the re-vegetation of gully walls and interfluves through
stabilisation with textiles and seeding (Cole et al., 2014). Previous work by Evans et al.
(2006) has shown that the re-vegetation of gullies is effective in limiting sediment flux from
eroding peat catchments by trapping mobilised sediment within the gully system, but, little
is known about the sources of sediment still entering the fluvial system in restored
catchments. Sediment source exerts a fundamental control on the volume and the physical
and geochemical properties of the sediment entering the fluvial system (Walling et al.,
2001), which in turn will control the magnitude of sediment-associated carbon and
pollutant fluxes. Identifying the source of suspended sediment is of key importance for
understanding fluvial geomorphic process and systems, and therefore is an essential
precursor to the design and implementation of effective sediment management (Hatfield
and Maher, 2008). In the context of restoration, resources could be misdirected if control
measures were focussed on reducing surface erosion, when most of the sediment
transported through a system was contributed by channel and gully erosion (Walling,
2005).
98
Previously, information on suspended sediment provenance in peatlands has been
obtained using a range of indirect measurement or monitoring techniques, aimed at either
identifying areas from which sediment is being mobilized or comparing rates of sediment
mobilisation from potential source areas, in order to assess their likely relative
contributions (e.g. field observations, erosion pins, Gerlach troughs; Evans and Warburton,
2007). However, these methods can be hampered by problems of spatial and temporal
sampling, and operational difficulties (Walling et al., 2008). A fingerprinting approach has
been widely adopted as a means of identifying the sources of sediment (e.g. Oldfield et al.,
1985; Collins et al., 1997; Collins and Walling, 2004; Smith et al., 2013), as it can provide a
relatively simple and cost-effective means of assembling spatially- and temporally-
integrated data (Collins and Walling, 2004). This approach involves first selecting a suite of
physical or chemical properties which clearly differentiate potential source materials. The
characteristics of the source materials can then be compared with measurements of the
same properties obtained from suspended sediment using a numerical mixing model, to
identify the likely source of that sediment. This mixing model approach has been used to
determine sources of fine sediment in a variety of settings: agricultural catchments
(Gruszowski et al., 2003; Collins et al., 2010), urban environments (Carter et al., 2003;
Devereux et al., 2010), forest environments (Motha et al., 2003), and in burnt catchments
(Wilkinson et al., 2009; Smith et al., 2013), but has not previously been applied to organic
rich upland systems. Hutchinson (1995) and Rothwell et al., (2005) successfully derived
fingerprints for peatland catchment sources, allowing them to distinguish between ‘clean’
subsurface peat and contaminated near-surface peat, but neither study went on to apply a
mixing model.
This paper aims to assess the effectiveness of restoration practices on reducing sediment
production in an eroding peatland in the Bleaklow area of the Peak District National Park,
southern Pennines, UK. For the first time, a fingerprinting approach employing a numerical
mixing model has been applied to a peatland setting in order to understand sediment
dynamics, which in turn provides information on POC and Pb release pre- and post-
restoration. Time integrated suspended sediment samples have been collected at field
areas which represent the three surface conditions exhibited through an erosion-
restoration cycle: intact, actively eroding, and re-vegetated.
99
4.2. Materials and Methods
4.2.1. Study area
The Bleaklow Plateau (500 – 633 m) is an upland blanket peatland in the Peak District
National Park (PDNP) in the Southern Pennines, UK (Figure 4.1). Peat depths across the
plateau vary between 2 and 3 m (Evans and Lindsay 2010b), and cover an underlying
geology composed of sandstone bedrock from the Millstone Grit Series (MGS) (Wolverson-
Cope, 1976) which is overlain in places by fine grained head deposits of weathered MGS
shales (Rothwell et al., 2005). The plateau lies between the industrial cities of Manchester
and Sheffield, the heartland of the 19th century English Industrial Revolution. Consequently,
the blanket peats here are amongst the most contaminated in the world, and the near-
surface layer of the peat is contaminated by high concentrations of anthropogenically
derived, atmospherically deposited Pb (in excess of 1700 mg kg-1; Shuttleworth et al.,
2014a).
Figure 4.1: Location map. Grey hatched rectangle denotes the position of the field area.
Anthropogenic and climatic pressures have led to severe environmental degradation across
the plateau and erosion is widespread (Tallis 1985; Bonn et al. 2009). As such, the area has
been a focus of peatland research for several decades. Bower (1960a & b, 1961) first
described the eroded landscape of the Pennines and much work has been carried out by
Tallis (e.g. 1964, 1985, 1997) investigating the timing and causes of the initiation of this
erosion. More recently work aimed at understanding the erosion has focused on peatland
100
hydrology (e.g. Daniels et al., 2008; Goulsbra et al., 2014), carbon flux and sequestration
(e.g. Pawson et al., 2008, 2012; Clay et al. 2012), pollutant storage and mobility (e.g.
Ferguson et al., 1978; Hutchinson, 1995; Rothwell et al., 2005), and the effects of
restoration (e.g. Dixon et al., 2013; Cole et al., 2014).
The Moors for the Future (MFF) Partnership was set up in 2003 to combat the degradation
in the Peak District, supported by the UK Heritage Lottery Fund. The partnership comprises
governmental bodies, non-governmental institutions and the three water companies based
in the Peak District, and aims to identify suitable approaches for restoring some of the
degraded and eroded moorland found in the area. MFF has invested upwards of £13
million to control and reverse peatland erosion, a considerable proportion of which has
been devoted to the Bleaklow Plateau. The restoration work on Bleaklow involved the
planting of a nurse crop of grasses, and the application of lime and fertiliser. Geojute, a
geotextile mesh, was also used to stabilise the peat’s surface, and cut heather brash was
applied to create a microclimate, and prevent loss of seeds and further surface erosion.
Finally, species such as bilberry, crowberry and cotton grass were planted directly into the
peat (Cole et al., 2014).
This study focuses on three sites across the Bleaklow plateau, representing various states
of disturbance: (i) intact (ii) actively eroding, and (iii) recently re-vegetated. Due to
restrictions in the availability of suitable field areas, it was not possible to replicate field
area conditions, leading to a pseudo-replicated design when considering site variation. The
intact field area (to the authors’ knowledge) has never been subject to the heavy erosion
that affects/has affected the other two sites and so acts as a control. The peat surface is
fully vegetated and is drained by a series of shallow depressions which are mostly
vegetated, with some exposed peat on channel banks (Figure 4.2a). The actively eroding
field area is heavily degraded and dissected with gullies, ranging from shallow ephemeral
headwaters to deeply incised channels which have cut into the underlying geology.
Vegetation is sparse; some gully floors, sides and interfluves are vegetated but bare peat is
prevalent (Figure 4.2b). The re-vegetated site encompasses a series of wide, deeply incised,
gullies which drain the plateau to the west. MFF began restoration measures in the area in
2003 (Figure 4.2c), and today the area has almost full vegetation cover on interfluves and
gully floors, with small channels cutting down to the underlying geology in some gullies.
Bare peat is still exposed on some gully walls (Figure 4.2d).
101
Figure 4.2: Surface condition at the three field areas: (a) shallow drainage depression at the intact field area; (b) deeply incised gullies with sparse vegetation cover at the eroding field area; (c) application of ’geojute’ at
the re-vegetated field area in 2003; (d) the re-vegetated field area today.
4.2.2. Field measurement
Suspended sediment sampling 4.2.2.1.
Composite suspended sediment (SS) samples were collected using time-integrated mass
flux samplers (TIMS) as described by Owens et al. (2006). This variant of TIMS was chosen
over the more widely used TIMS developed by Phillips et al. (2000) owing to their smaller
size and larger inlet making them more suitable for use in small headwater streams with
intermittent flow. Field tests showed there was no significant difference in the properties
of SS collected by the two different designs (Shuttleworth, unpublished data). TIMS were
constructed from PVC piping with dimensions: 52 mm (ID) x 0.5 m, capped at each end by 8
mm plastic mesh. For the purposes of this study, the gravel filling used by Owens et al.
(2006) was replaced with polystyrene packing ‘peanuts’ to reduce mass and allow multiple
TIMS to be easily deployed in remote areas. Flow through the field areas can be ephemeral
in nature so TIMS were deployed where there was visual evidence of concentration of flow.
No gullies are present at the intact field area, so sampling sites were chosen in the field by
observations of surface flow pathways. See Figure 4.2a for an example of a typical drainage
102
Field area Successful
sampling sites
Sites where no sediment captured
MUGD (m)
Mean surface Pb (μg g-1)
Catchment area (m2)
(a) Intact 5 7 n/a 290 – 401 838 - 7259
(b) Eroding 11 0 0.59 - 1.76 32 – 642 95 - 858
(c) Re-vegetated 6 8 0.73 - 1.18 178 – 366 484 - 7183
Table 4.1: Summary of the catchment characteristics at the three field areas.
Figure 4.3: Location of suspended sediment sampling sites at the three field areas: (a) intact, (b) eroding, (c) re-vegetated. Drainage networks (black lines) were derived using TAS GIS (Lindsay, 2005). White dots represent sites where suspended sediment was collected. Red dots represent sampling sites where no
suspended sediment was collected.
103
depression where TIMS were deployed at the intact field area. The TIMS were fixed in place
using wooden stakes driven into the channel beds with their long axes parallel to the
direction of flow. Water and entrained SS passed through the pore spaces of the
polystyrene filling and flow velocity within the body of the TIMS was reduced thus
encouraging sediment deposition (Owens et al., 2012).
A total of 11 TIMS were deployed at the eroding site, 15 at the re-vegetated site and 16 at
the intact site (Figure 4.3). Mean upslope gully depth (MUGD) has been shown to be a
control on sediment-associated Pb concentrations (Rothwell et al., 2010b) so sampling sites
at the eroding and re-vegetated field areas were chosen using a gully depth map developed
by Evans and Lindsay (2010b) as a guide. Similar ranges of MUGD were sampled at each
field area (Table 4.1).
The TIMS were deployed for five sampling campaigns between October 2010 and January
2012, each lasting approximately ten weeks. At the end of each deployment, the
polystyrene filling and sediment retained in each sampler were emptied into large
polythene bags, sealed and returned to the laboratory where they were stored at 4 °C prior
to analysis. Due to the ephemeral nature of flow in these catchments and the variable
amounts of sediment transported, several samplers which were deployed at the re-
vegetated and intact sites did not collect samples during every campaign, and some did not
yield any sediment at all (Figure 4.3, Table 4.1).
Source identification and sampling 4.2.2.2.
Rothwell et al. (2005) identified four distinct catchment materials that potentially
contribute to the SS load in the Bleaklow area: near-surface ‘dirty’ peat, subsurface ‘clean’
peat, Millstone Grit sandstone, and periglacial head deposits. Samples of these four sources
were collected from the three field areas in October 2012. Near-surface peat samples
(n=15) where extracted using a gouge corer and comprised a homogenised mixture of the
upper 10 cm of each core. The remaining three sources were sampled using a plastic
trowel. Subsurface peat (n=10) was collected from gully walls at a depth of approximately
1 m below the peat’s surface. Any friable material on the gully wall was first cleared to
avoid contamination by surface derived sediment which may have fallen from above.
Samples of Millstone Grit (n=10) and periglacial head deposits (n=7) were collected from
exposures of the underlying geology at the base of deeply incised gullies. The
characteristics of the four potential sediment sources are summarised in Table 4.2.
104
Pb concentrations at the peat’s surface were found to vary over two orders of magnitude
from 32 to 1034 µg g-1. Rothwell et al. (2007a) recommend the use of 15 samples to reliably
characterise within-region Pb storage, based on data obtained from intact sites in the
Bleaklow area. However, varying rates of surface erosion at the eroding and re-vegetated
field areas may have further complicated Pb storage: exposing higher concentrations of Pb
stored below the peat’s surface or removing the polluted surface layer all together. Surface
Pb storage therefore required more extensive characterisation.
Surface peat Subsurface peat Millstone grit Head deposits
(n=15) (n=10) (n=10) (n=7)
OC Mean 450.33 491.27 6.99 22.32
(g kg-1) Max 470.59 496.45 10.60 36.99
Min 409.69 481.69 3.78 12.72
Pb Mean 486.86 <LOD 20.25 50.13
(µg g-1) Max 1033.88 <LOD 52.03 108.15
Min 31.81 <LOD 0.46 19.00
χlf Mean 7.80 0.02 0.11 0.39
(10-6 m2 g-1) Max 14.16 0.22 0.35 0.61
Min 3.43 -0.09 -0.04 0.18
SIRM Mean 7051.76 2.80 101.10 89.54
(10-5 Am2 g-1) Max 8799.69 6.41 185.30 187.11
Min 4058.36 1.09 74.63 51.34
ARM Mean 37.37 0.15 1.52 1.38
(10-5 Am2 g-1) Max 54.99 0.22 3.67 2.19
Min 20.55 0.06 0.45 0.67
SIRM/ARM Mean 192.83 19.32 79.71 65.36
Max 243.43 42.84 164.14 92.05
Min 160.02 5.03 41.48 49.02
Table 4.2: Summary of the characteristics of the four potential sources.
4.2.2.2.1. Surface survey
A handheld Niton XL3t 900 XRF analyser was used to assess surface Pb concentrations
across the three field areas following the method outlined in Shuttleworth et al. (2014a).
Measurements were taken in a 50 m gridded pattern across each of the field areas. Large
variations in Pb concentrations were observed over very small distances at the eroding field
area, so more intensive sampling was conducted around each study catchment. The water
content at each sampling location was also determined and used to correct for the dilution
effect of the high moisture content of the peat (c.f. Shuttleworth et al., 2014a).
105
The moisture corrected Pb data were used to produce geochemical maps using ordinary
Kriging techniques in Surfer 8.0 (eroding, n=109; re-vegetated, n=66; intact, n=40). The
Surfer grid files were then imported into TAS GIS (a freely available software package for
performing spatial analysis operations for hydro-geomorphic applications; Lindsay, 2005),
and overlaid with the gully network for each site (c.f. Evans and Lindsay, 2010b). The
resulting maps depict interpolated Pb concentration across interfluve surfaces and were
used to calculate surface Pb storage in each field area (Table 4.1) and in each catchment
(Figure 4.4).
Figure 4.4: Steps for deriving catchment Pb concentrations using the re-vegetated field area as an example: (a) modelled surface Pb concentrations and gully network overlay, (b) final surface map, (c) close up of
watershed delineation. White line represents watershed delineation. White line represents watershed, white dot represents TIMS location.
4.2.3. Laboratory analysis
Walling (2005) stresses the importance of using a composite fingerprinting approach to
sediment source ascription, incorporating multiple properties into the model which are
capable of discriminating between several potential sources. A variety of physical and
chemical properties have been successfully used to discriminate between potential
sediment sources (see Walling (2005, 2013) and Davis and Fox (2009) for a full review). The
range of properties available in peat is limited, as fluctuating water tables and changing
redox conditions can affect the mobility of many elements. However, Hutchinson (1995)
and Rothwell et al. (2005) have used the anthropogenic contamination in the Bleaklow area
106
to distinguish material eroded from the peat’s surface from peat mobilised from gully walls.
Both Hutchinson (1995) and Rothwell et al. (2005) used a suite of magnetic measurements
to infer the presence of high concentrations of anthropogenically derived, coarse-grained,
ferromagnetic spherules stored near the peat’s surface. Rothwell et al. (2005) also used Pb
concentrations as another indicator of material derived from the peat’s surface; Pb has a
high affinity to organic matter and has been identified as the least mobile heavy metal in
wetland ecosystems (Farmer et al., 2005; Novak et al., 2011). The organic carbon (OC)
content was also considered to help distinguish material sourced from peat from the
underlying geology. Evans et al. (2013) found annual POC losses of between 1.6 and 3.8% in
exposed peat samples isolated from the main peat mass, so any changes in the organic
properties of the SS entrained in the TIMS over the ten week sampling period would be
negligible.
Sample preparation 4.2.3.1.
SS samples were washed through an 8 mm sieve with deionised water to separate the
sediment from the polystyrene. The resulting slurry was oven dried at 40 °C (so as not to
affect magnetic mineralogy of the samples; Walden et al., 1999) until a constant weight
was achieved. Once dry, samples were gently disaggregated by hand using a pestle and
mortar and homogenised. Samples were then subsampled in triplicate, provided enough SS
had been collected to allow this.
Source samples were oven dried, disaggregated and homogenised as above.
Analysis 4.2.3.2.
Magnetic susceptibility measurements (χlf and χhf) were made with a Bartington
Instruments Ltd. MS2 meter (Dearing, 1994) and measurements of magnetic remanence
(ARM and SIRM) were made with a Molspin Instruments ‘Minispin’ fluxgate magnetometer
(Walden et al., 1999). ARMs were acquired in a peak a.c. demagnetizing field of 100 mT,
with a superimposed d.c. field of 0.1 mT, using a Molspin Instruments’ AF-demagnetizer.
SIRMs were imparted in an applied d.c. field of 1 mT, using a Molspin Instruments’ high-
field ‘Pulse Magnetizer’. Pb concentrations were determined using inductively coupled
plasma - optical emission spectrometry (ICP-OES) using a HNO3 digest, as outlined in
Shuttleworth et al. (2014a). The amount of organic matter (OM) was determined using the
loss on ignition (LOI) method at 550 °C for 4 h. The OM content was converted to OC
estimates based on the assumption that the majority of suspended load is organic material
107
and that the carbon content is approximately 48% of the suspended organic load (Pawson,
2008).
4.2.4. Modelling
Fingerprinting 4.2.4.1.
An initial investigation of the source data using principal components analysis (PCA) can be
seen in Figure 4.5. Millstone Grit sandstone and periglacial head deposits map onto the
same area of the biplot. For the purposes of this study there is no need to distinguish
between these two minerogenic sources so they have been combined into one source
group and will be referred to collectively as “underlying geology”. The PCA shows that
samples from the three sources can be clearly distinguished from each other using only the
few discriminatory properties tested for, indicating that SS provenance in the field areas
could be modelled using a multivariate mixing model, first proposed by Collins et al. (1997).
Figure 4.5: PCA analysis showing three distinct potential sources of suspended sediment.
108
The mixing model 4.2.4.2.
Relative contributions from potential sediment sources were modelled using a set of linear
equations that represent the value of an individual tracer property in sediment as a
function of the sum of the values of that tracer for each source multiplied by the unknown
proportional contribution from each source (Smith and Blake, 2014). Two linear boundary
conditions were imposed on the mixing model to ensure that the relative contributions
from the individual sediment sources are non-negative and that these contributions sum to
unity (Collins et al., 2010). Solutions were obtained by minimising the sum of squares of the
weighted relative errors associated with the equations:
∑ {(𝐶𝑖 − (∑ 𝑃𝑠𝑆𝑠𝑖𝑚𝑠=1 𝑆𝑉𝑠𝑖))/𝐶𝑖}
2𝑊𝑖
𝑛𝑖=1 Equation 4.1
where: Ci = concentration of fingerprint property (i) measured in the suspended sediment
sample; Ps = the modelled optimised percentage contribution from source category (s); Ssi =
mean concentration of fingerprint property (i) measured in source category (s); SVsi =
weighting representing the within-source variability of fingerprint property (i) in source
category (s); Wi = tracer discriminatory weighting; n = number of fingerprint properties
comprising the optimum composite fingerprint; m = number of sediment source categories.
A variety of correction factors are often included in the mixing model to account for
variance in tracer properties of the potential sources and thus the discriminatory power of
each tracer, and potential physical and chemical changes during fluvial erosion and
transportation. Tracer-specific weightings to account for within-source variation (SVsi) and
each tracer’s discriminatory power (Wi) were included in the optimised model to address
the heterogeneity in the pollution signal across the peat’s surface. Incorporating these
weightings made little to no difference to the modelled outputs for most of the sampling
sites, but in catchments where Pb storage is low, their inclusion improved the goodness of
fit of the model (see next paragraph) and prevented overestimation of inputs from the
underlying geology (see section 6.2.4.3.1). SVsi was derived for each property, based on the
coefficients of variation of that property for each potential source (Collins et al., 2012b). Wi
was based on information on the relative discriminatory efficiency of each individual tracer
included in any given composite fingerprint provided by the results of the DFA (Collins et
al., 2010). In the case of Pb, catchment specific weightings were derived. Tracer data also
109
often undergoes pre-treatment for particle size and organic matter differences between
source soils and sediment (e.g. Gruszowski et al., 2003; Collins et al., 2012a, 2012b).
However, there is growing criticism of this kind of adjustment (e.g. Koiter et al., 2013;
Smith and Blake, 2014) and several fingerprinting studies have not applied these correction
factors (e.g. Walling et al., 1999; Martínez-Carreras et al., 2010; Evrard et al., 2011). Due to
the organic nature of the catchments in this study, pre-treatment for particle size and
organic matter were not included in the mixing model.
Uncertainty in source apportionment results was determined using a Monte Carlo sampling
framework using the median and median absolute deviation (MAD) as location and scale
estimators (c.f. Collins et al., 2012b) to generate random deviates for the fingerprint
properties of source and sediment samples for 3000 iterations. The goodness of fit (GOF) of
the mixing model outputs for every SS sample were tested by comparing the measured
fingerprint property concentrations in the SS with the corresponding values predicted by
the model (i.e. the relative mean error) (Collins et al., 2010):
𝐺𝑂𝐹 = 1 − [1
𝑛∑ {(𝐶𝑖 − (∑ 𝑃𝑠𝑆𝑠𝑖
𝑚𝑠=1 𝑆𝑉𝑠𝑖))/𝐶𝑖}
2𝑊i
𝑛𝑖=1 ] Equation 4.2
Property selection 4.2.4.3.
The two-stage statistical procedure outlined by Collins and Walling (2002) was used to
determine the discrimination of the potential sediment sources.
The Kruskal–Wallis H-test was used to examine the ability of individual properties to
explicitly distinguish between samples of surface peat, subsurface peat and underlying
geology. This provided a basis for eliminating redundant fingerprint properties. Although
there are limitations to using ratios in mixing models (Walling, 2005), the SIRM/ARM ratio
has been included as Hutchinson (1995) and Rothwell et al. (2005) found it to be key in
identifying the presence of material originating from the peat’s surface. The ratio is
sensitive to magnetic grain size (Walden et al., 1999), so can be used to infer the presence
of anthropogenically derived, coarse-grained, ferromagnetic spherules found near the
peat’s surface (Oldfield et al., 1978). The results of the Kruskal–Wallis H-test are presented
in Table 4.3. All six properties included in the test passed for each of the field areas and so
were entered stage two of the statistical verification.
110
Area Intact Eroding Re-vegetated
Property H-value p-value H-value p-value H-value p-value
OC
30.048 0.000
30.041 0.000
30.041 0.000
Pb
30.692 0.000
28.456 0.000
26.532 0.000
χlf
25.382 0.000
25.382 0.000
25.382 0.000
SIRM
29.56 0.000
29.56 0.000
29.56 0.000
ARM
29.808 0.000
30.041 0.000
30.041 0.000
SIRM/ARM 26.319 0.000 25.821 0.000 26.319 0.000
Critical value at 99% confidence = 10.60
Table 4.3: Kruskal–Wallis H-test results employed to select the fingerprint properties to distinguish the individual source types at the three field areas.
Field area Step Property selected
Wilks’ lambda
Intact 1 OC 0.0015
2 χlf < 0.0001
3 SIRM/ARM < 0.0001
4 Pb < 0.0001
Eroding 1 OC 0.0023
2 SIRM/ARM 0.0004
3 χlf 0.0002
4 Pb 0.0002
5 ARM* 0.0001
Re-vegetated 1 OC 0.0021
2 SIRM/ARM 0.0003
3 Pb 0.0002
4 χlf 0.0001
5 SIRM* 0.0001
Table 4.4: The results of the initial DFA employed to select an optimum composite fingerprint to distinguish the individual source types at the three field areas. 100% of the source type samples were classified correctly after the first step. Properties marked with an asterisk (*) were not included in the final model as they were
already incorporated as part of the SIRM/ARM ratio.
111
Stepwise Discriminant Function Analysis (DFA) was used to further assess the
discriminatory power of the tracer properties that passed the Kruskal–Wallis H-test (Collins
and Walling, 2002). DFA identifies an optimum source fingerprint that comprises the
minimum number of tracer properties that provide the greatest discrimination between
the analysed source materials based on the minimisation of Wilks' lambda (Smith and
Blake, 2014). According to the DFA, OC, the SIRM/ARM ratio, χlf and Pb can be used to
distinguish between the sources at the three field areas (Table 4.4). ARM and SIRM were
also included in the DFA output for the eroding and re-vegetated site respectively. These
were not included in the final model as they were already incorporated as part of the
SIRM/ARM ratio, and their influence on the reduction of Wilks’ Lambda is minimal.
4.2.4.3.1. Property revision
Initial runs of the model incorporating the four tracer properties identified by the DFA,
overestimated contributions from the underlying geology and produced low GOF values
(GOF >0.90 in only 3 out of 45 runs). The proportion of inorganic material in the SS samples,
determined from the results of the LOI analysis, was used as an indication of the maximum
possible input from the underlying geology. When the modelled contributions from the
underlying geology were compared with the data derived from LOI (Figure 4.6a) the model
over estimated inputs from the underlying geology in the majority of SS samples, especially
those containing low concentrations of inorganic material (< 12%). Similar overestimations
occurred if any of the magnetic concentration indicators (χlf, ARM, SIRM) were
incorporated into the model. It is possible that the anthropogenically derived magnetic
spherules do not behave conservatively during fluvial transport. Due to their fine nature,
some spherules could be “flushed out” of the organic sediment either during transport or
after deposition in the TIMS, reducing their concentration and thus SS χlf measurements.
This would give sediment derived from the peat’s surface χlf values closer to that of the
underlying geology. ARM and SIRM values would be similarly affected, but the SIRM/ARM
ratio should remain constant; any magnetic spherules present would still dominate the
magnetic grain size signature of surface derived peat as its main constituent (OM) is
diamagnetic (i.e. displays only weak or negative magnetic behaviour; Walden et al., 1999).
The DFA was recalculated, omitting the magnetic concentration parameters. According to
the revised DFA, a combination of OC, Pb and SIRM/ARM is still capable of correctly
classifying 100% of the source samples at each of the three field areas (Table 4.5). Using
only these three properties in the model reduced the estimated contribution of the
underlying geology, making it more realistic in relation to the LOI data (Figure 4.6b), and
112
also improved the GOF of the model in most cases (GOF increased by between 0.02 and
0.66 in 38 out of 45 runs; GOF >0.90 in 39 out of 45 runs). The final optimised model
therefore used OC, Pb and SIRM/ARM as tracers to determine the contributions of the
potential sources to the SS load. A summary of the raw tracer values for these three
properties for the SS collected at each sampling site can be found in Table 4.6.
Figure 4.6: Relationship between LOI derived inorganic matter content and modelled contributions from the underlying geology for a selection of suspended sediment samples: (a) includes Xlf in the model, (b) excludes
Xlf from the model. The 1:1 line is shown as a dashed line.
Field area Step Property selected
Wilks’ lambda
Intact 1 OC 0.0015
2 Pb 0.0001
3 SIRM/ARM 0.0001
Eroding 1 OC 0.0023
2 SIRM/ARM 0.0004
3 Pb 0.0003
Re-vegetated 1 OC 0.0021
2 SIRM/ARM 0.0003
3 Pb 0.0002
Table 4.5: The results of the second DFA, omitting χlf, employed to select an optimum composite fingerprint to distinguish the individual source types at the three field areas. 100% of the source type samples were
classified correctly after the first step.
113
4.2.5. Material flux calculation
Several studies have highlighted that the TIMS designed by Phillips et al. (2000), which
works on a similar principle the TIMS employed in this study whereby flow velocity is
slowed to encourage deposition within the sampling chamber, underestimate the actual
sediment load (e.g. Phillips et al., 2000; Perks et al., 2013; Smith and Owens, 2014). The
Owens et al. (2006) TIMS design used in this study has not been so rigorously tested, but it
is reasonable to assume that it will also have less than 100% trapping efficiency. In light of
this, the absolute material fluxes for each catchment could not be determined. However,
Perks et al. (2013) suggest that sediment collected by TIMS is suitable for characterising
relative temporal and spatial patterns of SS associated fluxes. The three field areas were in
close enough proximity that the TIMS would have been subject to the same hydrological
conditions and will have been active for similar periods of time. Therefore, the relative
fluxes of material through the TIMS can be calculated and used to compare the magnitude
of OC and Pb export at each of the three field areas. The relative fluxes of material through
the TIMS were calculated using the following equation:
𝐹𝑖 =𝐶𝑖×𝑚
𝐴×𝑡 Equation 4.3
Where: F = the flux of material (i) through the TIMS; Ci = the concentration of material (i)
measured in the suspended sediment; m is the total mass of the suspended sediment; A is
catchment area; t = sampling time. See Table 4.6 for concentration data and mass of SS.
4.3. Results
4.3.1. Predicted source contributions
Figure 4.7 shows the mean modelled contributions of each source type for the SS collected
at each of the sampling sites across the three field areas during the five sampling
campaigns. The contribution of sources varies across all sampling sites and between the
three field areas. The dominant sediment source at the intact field area is surface peat,
accounting for 69 - 91 % of the total SS load. The remaining SS is made up of 9 - 30 %
subsurface peat, and at one site there is an input from the underlying geology (9 %).
114
Site n
OC (g kg-1)
Pb (µg g-1)
SIRM/ARM
Suspended sediment (g)
mean max min mean max min mean max min mean max min
Intact 1 4
427 471 356
224 281 135
185 234 160
4.41 8.90 1.50
2 3
476 479 469
285 331 221
168 177 160
1.20 1.81 0.80
3 2
476 476 475
310 365 255
188 223 153
8.30 10.70 5.90
4 2
458 462 455
378 390 365
127 140 113
1.18 3.60 0.04
5 2
475 477 473
309 333 285
169 181 151
4.55 5.00 4.10
Eroding
1 4
480 481 476
152 198 112
158 175 140
2.76 5.60 0.08
2 3
474 483 460
212 309 131
194 263 158
0.84 2.52 0.02
3 5
479 486 458
187 302 140
157 225 132
1.93 2.83 0.43
4 4
475 482 463
109 141 70
144 220 101
3.14 7.10 1.15
5 4
401 481 186
131 165 71
170 184 157
13.93 22.40 6.48
6 5
476 484 466
116 151 74
173 197 137
27.08 46.70 2.66
7 5
412 479 198
121 194 51
178 198 156
36.33 58.90 7.35
8 5
486 493 481
183 275 70
173 191 143
19.30 42.40 1.95
9 5
485 493 481
203 322 109
170 209 144
17.19 42.80 2.35
10 4
430 476 358
112 205 59
167 182 156
10.71 22.80 2.35
11 4
195 336 84
114 154 45
173 202 143
21.51 35.60 3.44
Revegetated
1 2
479 479 479
95 108 67
134 215 52
0.72 1.76 0.10
2 3
374 496 172
155 217 31
148 245 54
0.39 0.87 0.02
3 2
474 483 466
190 239 154
79 88 63
0.09 0.20 0.02
4 3
356 477 90
92 171 17
146 162 122
3.15 7.40 0.71
5 2
59 85 15
26 38 13
121 170 61
78.78 96.68 62.95
6 3 468 475 461 155 222 108 163 257 68 0.50 0.90 0.13
Table 4.6: Summary of the raw tracer values for the properties incorporated into the optimised mixing model for the SS collected at each sampling site.
115
Figure 4.7: Modelled relative contributions of individual source types to suspended sediment at the (a) intact, (b) eroding, and (c) re-vegetated field areas.
116
Relative source contributions at the eroding and re-vegetated field areas are much more
varied. At the eroding site surface peat is the dominant source of sediment at 6 out of the
11 sampling sites, subsurface peat contributes the most to the SS load at 4 sites, and while
the underlying geology contributes 0 - 17% at these sites, it is the main SS constituent at
site 11 (57%). At the re-vegetated field area, surface peat is the main contributor to SS at
only one site; subsurface peat dominates the SS load (50 – 55%) at three sites while the
underlying geology is the primary source of sediment (91 %) at one site.
4.3.2. Material fluxes through TIMS
The fluxes presented here are the relative fluxes of material passing through the TIMS at
each sampling site and are not quantitative estimates at the catchment scale. They have
been derived in order to compare the magnitude of OC and Pb export at each of the three
field areas. To highlight the study-specific nature of these data, fluxes will be referred to as
POCTIMS and PbTIMS. OC will now be referred to as POC (particulate organic carbon) to
distinguish SS associated carbon from other fluvial OC fluxes (e.g. DOC). The fluxes of PbTIMS
and POCTIMS are presented in Figure 4.8.
There are marked differences in the magnitude and variability of POCTIMS and sediment
associated PbTIMS fluxes at the three field areas. Fluxes at the eroding field area are much
greater than at the intact and re-vegetated sites, and vary over several orders of
magnitude; POCTIMS ranges from 0.75 to 113.76 mg m-2, and PbTIMS ranges from 0.10 to
48.01 µg m-2 per 10 week sampling period. Mean fluxes at the intact and re-vegetated
areas are not statistically different to each other at the 95% level (two tailed t-test, POC: α=
0.239; Pb: α = 0.051), but PbTIMS is slightly higher at the intact area and POCTIMS is more
variable at the re-vegetated field area (Levene's Test, α=0.003).
4.4. Discussion
4.4.1. Effect of surface condition on sediment source
The majority of SS mobilised at the intact site comes from the peat’s surface. This is
unsurprising as the drainage network here comprises a series of shallow depressions where
little subsurface peat is exposed, limiting it as a potential contributor to the SS load. There
is a small contribution (9%) to SS from the underlying geology at one sampling site which is
unexpected as there was no exposure of this source observed upstream of the TIMS.
117
Figure 4.8: Relative fluxes of (a) POCTIMS and (b) PbTIMS at the three field areas. Fluxes have been given the suffix TIMS to emphasize the study specific nature of the data; the calculated fluxes are only representative
of sediment passing through the TIMS, and are not a quantitative estimate at a catchment scale.
Holden and Burt (2002a) found evidence at Moor House in the North Pennines that piping
can connect the peat’s surface with the underlying substrate so it is possible that pipes
could be delivering material from the underlying geology to the surface to contribute to the
SS load. Holden et al. (2012) show that pipe networks can also transport substantial
amounts of POC, so some of the subsurface peat found in the SS at the intact site could
have been delivered to the surface by pipe flow, rather than being eroded from the
vegetated drainage channels.
SS composition in the eroding and re-vegetated areas is more varied due to the
concentration of flow in deeply incised gullies. Subsurface peat is exposed on gully walls
and floors, and in the deeper gullies, material from the underlying geology can also be
mobilised. Consequently, there is a larger pool of sediment available to mix with material
sourced from the surface than at the intact field area. At the eroding site, the surface is still
the main contributor to SS at the majority of the sampling sites, but subsurface peat is
dominant at four out of the eleven sites studied. At the re-vegetated site subsurface peat is
the dominant source at three out of the six catchments while the material from the surface
is the main contributor at only one sampling site.
118
Despite efforts to select sampling sites in the eroding and re-vegetated areas with similar
MUGD, it was not possible to keep this constant. To ensure that the dominant organic
sediment source (surface versus subsurface peat) was not simply a function of gully depth,
the proportions of surface derived sediment were normalised by MUGD. The normalised
proportion of surface material entering the system at the eroding site is significantly higher
at the eroding field area (two tailed t-test, α= 0.016). Rothwell et al. (2010b) found a strong
negative relationship between MUGD and SS associated Pb concentration (i.e. proportion
of sediment derived from the peat’s surface) due to conservative mixing of contaminated
and ‘clean’ peat particulates as sediment moves down eroding gully walls. Strong
correlation between MUGD and modelled surface input can be found at the re-vegetated
site (r=-0.884, p=0.010). However, this relationship does not hold at the eroding site; there
is a moderate negative relationship between MUGD and modelled surface input but this is
not statistically significant at the 95% level (r=-0.44, p=0.198).
The role of vegetation in stabilising the peat’s surface and reducing sediment production is
well established. In Rothwell et al.’s (2010b) study, the relationship between MUGD and SS-
associated Pb concentrations was derived in catchments where interfluve surfaces were
well vegetated and gully walls were bare, so minimal material would be mobilised from the
surface and the main source of SS would have been from gully walls. The Pb in the SS would
have been sourced from the top of gully walls where the contaminated layer is exposed.
There is a considerable amount of bare peat exposed on interfluve surfaces at the eroding
site, presenting a larger surface area of contaminated material to localised sheet erosion,
thus contributing more contaminated material to the SS load. The strong relationship
between MUGD and modelled surface input at the re-vegetated area indicates that
sediment production is similar to that discussed by Rothwell et al. (2010b); i.e. material
mobilised from gully walls (whether clean or contaminated peat) is the main source of SS.
The relationship between MUGD and surface input has implications as to what the
modelled contributions reveal about the sources of organic sediment production at the
eroding and re-vegetated field areas. In eroding catchments, the relative contributions
differentiate between material mobilised from the surface and gully walls, while at the re-
vegetated field area, they provide information on the relative inputs of contaminated and
clean peat eroded from gully walls.
119
4.4.2. Effect of surface condition on sediment associated fluxes
The most striking distinction between the three field areas is the absolute magnitude of the
sediment associated fluxes; average POCTIMS and PbTIMS at the eroding site are two orders of
magnitude greater than those of the re-vegetated and intact sites (Figure 4.8).
Both POCTIMS and PbTIMS are closely tied to sediment production. In the case of POCTIMS,
fluxes are linked to the physical removal of peat from any source. Vegetation cover is a key
control on sediment production; the dense moorland vegetation that covers intact blanket
peatlands protects the peat’s surface from erosive processes, inhibiting sediment
mobilisation and SS yield (Evans and Warburton, 2007). This theory holds true at the intact
field area. Although the peat’s surface is the main source of sediment (see section 6.4.1),
containing high OC concentrations, the full vegetation cover restricts sediment production
so POCTIMS is low. By contrast, the sparse vegetation cover at the eroding field area offers
little protection, resulting in high rates of peat erosion and high POCTIMS values.
Mean POCTIMS at the re-vegetated field area is statistically similar to that of the intact field
area (see section 6.3.2). This is surprising as there is still a substantial amount of bare peat
exposed on some gully walls which will be subject to erosive processes. However, there is a
growing body of work (e.g. Evans and Warburton, 2005, 2010; Evans et al., 2006)
highlighting the importance of the role of vegetation in filtering organic particles from
overland flow, thus reducing connectivity between the erosional surfaces and channels.
Evans et al. (2006) cite gully floor vegetation as a key control on this. It is therefore likely
that the re-established vegetation on gully floors at the re-vegetated field area is
intercepting any POC mobilised from bare gully walls, and that this decoupling of eroding
surfaces from the fluvial system is reducing POCTIMS fluxes to those of the intact peatland.
In contrast to POCTIMS, PbTIMS fluxes are linked to the mobilisation of contaminated surface
peat only. Again, PbTIMS is low at the intact field area despite the majority of sediment being
sourced from the contaminated surface. As with POCTIMS this is due to the thick surface
vegetation inhibiting sediment production from this source. Mean PbTIMS at the re-
vegetated field area is of the same order of magnitude to that of the intact field area, but is
slightly lower due to the larger pool of clean material sourced from gully walls diluting Pb
contaminated sediment. The eroding field area produces considerably higher PbTIMS fluxes
than the other two field areas. As discussed in section 6.4.1, the lack of vegetation on
interfluve surfaces means that contaminated surface peat is a substantial contributor to
the SS load, releasing large amounts of Pb contaminated sediment into the fluvial system.
120
Both fluxes vary at all three field areas. Some variability would be expected over the 16
month sampling period, owning to temporal fluctuations in sediment supply and the
ephemeral nature of flow. Fluvial processes are the dominant mechanism controlling
sediment flux in eroding peatlands, both in the mobilisation and transport of material
through the system. Evans and Warburton (2007) also cite weathering (via frost action and
desiccation) as a key control on sediment ‘preparation’, producing bare peat surfaces
mantled with a layer of loose friable material, and the availability of such easily mobilised
material has been linked to sediment exhaustion in peatland catchments (Francis, 1990;
Labadz et al., 1991). Individual periods of TIMS deployment will have experienced a range
of antecedent conditions, in addition to variations in the timing and intensity of rainfall
which will have influenced the volume and nature of sediment collected. Spatial variations
in vegetation cover and Pb storage in individual catchments will further affect sediment
production and thus the amount and composition of material collected by the TIMS. This
intra-site heterogeneity is most pronounced at the eroding field area, causing POCTIMS and
PbTIMS to span several orders of magnitude.
4.4.3. Implications for restoration and further research
The difference between POCTIMS fluxes at the eroding and re-vegetated field areas, and the
similarity in fluxes at the re-vegetated and intact field areas, indicate that the MFF’s
restoration efforts on the Bleaklow Plateau have been effective. Organic sediment yields at
the re-vegetated field area have been reduced to those comparable to an intact peatland, a
process that Evans and Warburton (2007) hypothesised would take thousands of years to
occur naturally. The strong relationship between MUGD and modelled surface input at the
re-vegetated field area indicates that gully walls are the main source of SS post-restoration.
When viewed in combination with the low PbTIMS flux from the re-vegetated field area, we
can infer that the newly established vegetation has been most successful in stabilising the
surface of interfluves and limiting SS production from this source. The re-vegetation of gully
floors is also likely playing an important role in reducing connectivity between erosive
surfaces and the fluvial system, further reducing sediment associated fluvial fluxes. Future
improvements in restoration efforts should therefore additionally concentrate on
stabilising gully walls to further limit sediment production.
Carbon cycling 4.4.3.1.
POC release has been highlighted as an important part of the peatland carbon balance
(Evans and Warburton, 2010, Evans and Lindsay, 2010a, Pawson et al., 2012) so the
121
reduced POC flux following restoration will have a significant impact of the overall export of
carbon from the restored peatland system. The newly established vegetation has stabilised
the peat’s surface, preventing the physical removal of POC and thus preserving this store of
carbon. Reducing the amount of POC entering, and being stored in, the fluvial system also
limits the conversion of POC to DOC and CO2, further restricting carbon release.
From the data generated by this study, it has not been possible to determine the relative
importance of surface stabilisation and the trapping efficiency of gully floor vegetation in
reducing POC export. As noted above, bare gully walls are still likely to be a source of
sediment production, yet this does not seem to be reflected in the POCTIMS flux. POC
intercepted by vegetation at the slope-channel interface, and stored on gully floors has the
potential to oxidise to CO2 and contribute to the overall greenhouse gas emissions from the
area (Pawson et al., 2012; Evans et al., 2013). Further research into the magnitude and
longevity of POC storage by gully floor vegetation is needed to fully understand the impact
of restoration on the overall carbon balance.
Pb export 4.4.3.2.
Pb export at the eroding field area is two orders of magnitude greater than at the intact
field area. Previous research has demonstrated that the erosion of the contaminated near-
surface layer of peat in the Bleaklow area is releasing considerable amounts Pb into the
fluvial system (Rothwell et al., 2005, 2007b, 2008a, 2008b). The re-vegetation of eroding
gullies has decreased sediment associated Pb export to levels comparable to, and in some
cases below, those of the intact field area. This reduction of peatland Pb release will restrict
the amount of Pb entering drinking water reservoirs situated downstream of the Bleaklow
Plateau (Shotbolt et al., 2006), and will limit the potential for in-stream dissolution of
particulate bound Pb (Rothwell et al., 2008b).
High modelled inputs of sediment sourced from the surface of eroding catchments, and the
breakdown of the relationship between MUGD and modelled surface input at the eroding
site indicate that Rothwell et al. (2010b) could have underestimated SS Pb concentrations
in gullied areas where bare peat is exposed on interfluve surfaces. Future attempts to
model Pb release from contaminated peatlands should take surface vegetation cover into
account, and these bare areas should be the focus of peatland restoration as a matter of
priority to reduce sediment associated Pb export.
122
4.5. Conclusions
This study represents the first assessment of the effect of peatland restoration on sediment
production at the landscape scale. It is also the first study to successfully apply a numerical
mixing model to obtain quantitative estimates of the relative importance of different
sediment sources in peatland catchments.
Surface condition plays an important role in determining the main locus of sediment
production in peatlands. In intact peatlands, sediment is mostly derived from the surface,
while sediment generated in actively eroding areas is sourced from both the surface and
gully walls. Following re-vegetation, gully walls become the dominant source of sediment.
The re-vegetation of eroding gullies significantly reduces sediment production, decreasing
sediment associated fluxes by two orders of magnitude. A decade after re-vegetation
efforts commence, POC and Pb fluxes are reduced to levels comparable to those of an
intact peatland.
The findings of this study have immediate practical implications relating to upland
management and the control of peatland erosion. The results demonstrate the importance
of vegetation in reducing sediment yield, and provide a strong theoretical justification for
the re-vegetation techniques which have been pioneered by Moors for the Future. This is
of significant benefit to the maintenance of ecosystem services in areas of eroding peat,
and is of particular importance to the management of upland peatland carbon balances
and downstream water quality.
4.6. Acknowledgements
We thank The University of Manchester for the provision of a Graduate Teaching
Studentship (to E. Shuttleworth). We are grateful to The National Trust and United Utilities
for allowing work to be carried out at the study sites and to the University of Manchester
and Moors for the Future who provided funding for analytical costs. Thanks also go to John
Moore, Jonathan Yarwood and Laurie Cunliffe for their assistance in the lab, and to Gareth
Clay for his help proof reading the manuscript. Special thanks Simon Pulley for sharing his
macros and his assistance in fine tuning the mixing model. Finally, we would like to thank
the two anonymous reviewers for their comments which helped improve the paper.
123
Chapter 5 Controls on the fluvial export of sediment
associated lead and particulate carbon from eroding
peatlands (Paper 3)
This chapter is in preparation for submission to Hydrological Processes as Shuttleworth,
E.L., Evans, M.G., & Rothwell, J.J. “Controls on the fluvial export of sediment associated
lead and particulate carbon from eroding peatlands”.
Abstract
Large areas of the UK blanket peat are significantly degraded and are actively eroding due
to climatic and anthropogenic perturbations. Blanket peats are an important store of
carbon and the near-surface layer of many of the UK’s peatlands is also contaminated with
industrially derived atmospherically deposited contaminants such as lead (Pb). The stability
of peatlands is therefore important for the preservation of this carbon and limiting
contaminant mobilisation; however, little is known about the relative contributions of
various sediment sources entering the fluvial system. Previous work has identified rapid
changes in sediment composition during storm events, indicating organic sources become
supply limited but the underlying spatiotemporal controls on sediment mobilisation and
export are unclear. A sediment source fingerprinting approach was employed to determine
changes in sediment provenance during storm events and provide information on the
mechanisms of sediment exhaustion in the Peak District National Park, southern Pennines
(UK). Suspended sediment was collected using time integrated mass flux samplers (TIMS),
across a range of flow conditions in an eroding blanket peat catchment. Suspended
sediment carried at the beginning of storm hydrographs was relatively enriched in organic
matter, confirming the accepted model of organic sediment exhaustion during these
events. A flushing of high concentrations of lead early in storm events is evident under
certain conditions. The supply and composition of suspended sediment is controlled by: (i)
the physical availability of erodible organic sediment produced through weathering; (ii) the
degree of hydrological connectivity.
124
Key Words: sediment fingerprinting; sediment source tracing; organic sediment; upland
erosion; contaminant release; POC; weathering; hydrological connectivity
5.1 Introduction
Upland blanket peatlands cover 8% of the land area of the UK (Tallis, 1997), and support a
variety of ecosystem services including water supply, leisure activities, biodiversity and
carbon sequestration making them an important economic, scientific and recreational
resource. However, over the last 1000 years a significant proportion of the UK’s blanket
peat has become degraded and is actively eroding as a consequence of climate change,
legacy atmospheric pollution, poor management and anthropogenic disturbance (Bonn et
al., 2009). Extensive erosion potentially compromises the ability of the peatlands to
maintain ecosystem functions, and can negatively impact downstream water quality,
landscape stability, and carbon and contaminant storage.
Peat soils are the single largest carbon reserve in the UK (Cannell et al., 1993) storing up to
50 % of the UK’s soil carbon (Milne and Brown, 1997). However, the widespread
degradation of upland blanket peatlands threatens the integrity of this important carbon
store, and there is evidence that UK peatlands are becoming net sources of carbon to the
environment (Janssens et al., 2005). The majority of work examining fluvial carbon exports
from peatlands has focused on dissolved organic carbon (DOC) (e.g. Hope et al., 1994; Worrall
et al., 2004) with less attention given to particulate organic carbon (POC) fluxes, but extensive
deep gully systems, incised into the peat’s surface have been shown to be a major source of
particulate carbon loss in peatland streams (Evans et al., 2006; Pawson et al., 2012). The fate of
this POC is largely unknown but there is evidence to suggest it has the potential to transform to
atmospherically active DOC in the fluvial environment (Pawson et al., 2012) and as such POC
release has important implications for downstream water quality and climate change (Worrall
et al., 2004; Moody et al., 2013).
Blanket peats can also be sinks for atmospherically deposited heavy metals (Vile et al.,
1999; Farmer et al., 2005, Rothwell et al., 2010a); consequently, the near surface layer of
peatlands in close proximity to urban and industrial areas can contain high concentrations
of anthropogenically derived lead (Pb) (Livett et al., 1979; Shotyk et al., 1997; Rothwell et
al., 2007a). There is growing concern over the release of Pb from contaminated upland
125
peat catchments, particularly in the dissolved phase (e.g. Tipping et al., 2003; Vinogradoff
et al., 2005; Tipping et al., 2010), but there has been less focus on the export of sediment
bound Pb, despite particulates being recognised as the major vector for metal transport in
the fluvial system (Horowitz, 1995). The physical erosion of peat has been highlighted as a
mechanism for the release of substantial quantities of lead contaminated sediment to
surface waters (Rothwell et al., 2005 and 2007b; Shotbolt et al., 2006; Rose et al., 2012),
and Rothwell et al., 2008b) suggest that interactions between contaminated particulates
and the water column may elevate dissolved metal concentrations. Consequently, metal
release from eroding peatlands poses a threat to the sustainability of aquatic ecosystems
(Rhind, 2009) and has the potential to compromise downstream drinking water resources
(Shotbolt et al., 2006).
Rothwell et al. (2005) found that a large proportion of Pb contaminated sediment was
transported during the early stages of a storm event in a polluted catchment, as an initial
‘lead-flush’. However, Rothwell et al. (2007b) did not find any further evidence of a lead-
flush in subsequent storm events sampled in the same catchment, despite all of the events
displaying similar discharge characteristics. Sediment supply is a key control on the timing
and magnitude of sediment export in peatland systems, and so may also influence
contaminant release. Tallis (1973) first discussed the role of weathering in ‘preparing’
sediment for removal from the bare peat surfaces. Processes such as frost action and
desiccation destroy the structure of the surficial peat (Luoto and Sepälä, 2000), and
produce a superficial layer of friable material which is easily mobilised compared to the
more resistant cohesive peat underneath. Francis (1990) and Labadz et al., 1991) linked this
idea of sediment preparation to observations of organic sediment exhaustion in peatland
systems, and intra-storm supply limitation is often cited as the driver behind the positive
hysteresis displayed in the relationship between suspended sediment concentration (SSC)
and discharge (Q) (Labadz et al., 1991; Yue, 2005; Evans et al., 2006; Pawson et al., 2008).
There is historical evidence that previous phases of peat erosion may have been initiated
when prolonged dry periods were followed by periods of heavy rainfall (Tallis, 1997).
Current scenarios for future climate change suggest that UK peatlands may be subject to
drier summers and wetter, stormier winters (Hulme et al., 2002), which could not only
initiate additional phases of erosion, but also intensify sediment production and export,
thus exacerbating existing degradation. This increased instability has the potential to
negatively affect peatland C balances and increase the flux of Pb contaminated sediments
from peatland catchments (Rothwell et al., 2005, Evans and Warburton, 2010). It is
126
therefore vital to better understand the processes which drive sediment release in eroding
peatland systems.
This paper aims to find further evidence of the ‘lead-flush’ documented by Rothwell et al.
(2005), and provide further information on the mechanisms of organic sediment
exhaustion in an eroding blanket peat catchment in the southern Pennines, UK. Novel
suspended sediment (SS) sampling techniques and a sediment source fingerprinting
approach outlined in Shuttleworth et al. (2014b) have been employed to determine
changes in sediment provenance during storm events. Antecedent conditions in the
catchment were also monitored to investigate the controls on sediment production and
release.
5.2 Field area
Upper North Grain (UNG) is a small headwater stream that drains a blanket peat covered
catchment on the south eastern edge of the Bleaklow Plateau in the Peak District, South
Pennines, UK (Figure 5.1). The catchment lies between 490 and 541 m OD, covers an area
of 0.38 km2, and receives approximately 1200 mm rainfall each year. Land use in the
catchment is dominated by rough grazing by sheep. The peat, which reaches up to 4 m in
thickness, overlies a bedrock of interbedded sandstones and shales of the Millstone Grit
Series (Wolverson-Cope, 1976). Some of the bedrock of the catchment is overlain by
periglacial head deposits, which are derived from weathered sandstones and shales. High
concentrations of Pb are stored in the near-surface layer of the peat, a legacy of
atmospheric deposition during the 19th Century English Industrial Revolution (Rothwell et
al., 2007b). The catchment is heavily eroded with Bower Type I peat gullies (Bower, 1961).
In the upper reaches gullying is confined to incision into the peat, but the underlying
geology becomes exposed further downstream. Meteorological conditions in the
catchment are monitored by Skye automatic weather station (AWS) (described by Goulsbra
et al., 2014) which records a variety of parameters including temperature, precipitation and
water table depth. The catchment has been the focus of a range of geomorphological and
hydrological research providing background data for this study (e.g. Evans et al., 2006;
Rothwell et al., 2005 and 2007b; Daniels et al., 2008; Goulsbra et al., 2014).
127
Figure 5.1: Location of field area. (a) Red star depicts location of Upper North Grain relative to the Bleaklow Plateau; (b) Arial photograph of UNG catchment (after Pawson et al., 2008). The blue star shows the location of the sampling site; the yellow star shows.
5.3 Methods
5.3.1 Field sampling
5.3.1.1 Suspended sediment
Pawson et al. (2008) note the need for high resolution temporal sampling of SS in peatland
systems due to the episodic nature of organic sediment flux. However, such intensive
sampling campaigns can be highly labour intensive and the associated logistical problems
mean that many manual sampling strategies can fail to coincide with the main periods of
sediment transport (Collins and Walling, 2004). Automatic water samplers (such as those
used by Rothwell et al., 2005) can be costly and difficult to install in more remote field
areas. Further issues arise from the quantity and representativeness of the sample these
devices typically collect. Furthermore, SS sample volumes collected by automatic water
samplers may not yield sufficient quantities of sediment for subsequent analyses (Phillips et
al., 2000). Time integrated mass-flux samplers (TIMS) offer a low-cost, low-tech alternative
that overcome the problems outlined above, as they are can be left in the field to
continuously capture sediment over the course of several weeks, and yield a composite
sample which is representative of the entire sampling period (e.g. Phillips et al., 2000;
Owens et al., 2006). TIMS have been successfully deployed in eroding and restored
peatland systems to derive spatial trends in sediment export (Shuttleworth et al., 2014b).
128
The sampling site was located in the lower reaches of the catchment (Figure 5.1b) where an
instrument ‘bridge’ that spans the channel is secured to the bedrock stream bed (Figure
5.2a). Six TIMS were fixed to the bridge uprights at regular height intervals with their long
axes parallel to the direction of flow (Figure 5.2b) in order to capture sediment transported
a range of discharge conditions. The TIMS employed in this study were based on the design
described by Owens et al. (2006), used by Shuttleworth et al. (2014b) to collect SS in
eroding and restored peatland catchments. TIMS were constructed from PVC piping (52
mm (ID) x 0.5 m), and capped at each end by 8 mm plastic mesh (Figure 5.2b). The gravel
filling was replaced with polystyrene packing ‘peanuts’ (c.f. Shuttleworth et al., 2014b).
Water and entrained SS pass through the pore spaces between the polystyrene, slowing
flow within the body of the TIMS and encouraging sediment deposition. Ephemeral
streamflow (ES) sensors, adapted from those developed by Goulsbra et al. (2009) were
fixed to the underside of each TIMS to monitor the duration that each TIMS was active. At
the end of deployment, the polystyrene filling and sediment retained in each sampler were
emptied into large polythene bags, sealed and returned to the laboratory where they were
stored at 4 °C prior to analysis.
5.3.1.1.1 Testing the method
As the TIMS have not been deployed in this vertical context before, it is important to
ensure that the traps were all working in a similar manner (i.e. no trap were preferentially
collecting sediment due to greater trapping efficiency). If this is the case then there should
be a positive relationship between the total mass of sediment collected and the duration
each trap was active.
Figure 5.3 and Table 5.1 show the relationship between total mass of sediment collected in
each trap in relation to the duration that each trap was inundated. Overall, when the data
for the five sampling campaigns are combined, there is a strong positive relationship
(Spearman’s rank: ρ=0.672, p=0.000), indicating that sediment capture increases with time
inundated. This shows that traps that were active for longer periods of time yielded a
greater mass of sediment, as would be expected of traps which were constructed to the
same design specifications.
129
Figure 5.2: (a) Field installation, securing TIMS to instrument bridge; (b) schematic of the operational setup (not to scale), the topmost trap is drawn in cross section, showing the polystyrene filling.
130
Figure 5.3: Relationship between duration of TIMS inundation and mass of sediment retained.
2011 2012
Summer Autumn Winter Summer Autumn All
ρ 1.000 0.771 0.829 0.600 0.657 0.672
Sig. 0.000 0.036 0.021 0.104 0.078 0.000
Table 5.1: Spearman’s rank correlations for duration of TIMS inundation vs. mass of sediment retained for each of the sampling campaigns. Significant parameters are given in bold (95% confidence interval).
Positive correlation is also evident when looking at these two variables for each individual
sampling campaign, but the relationship is weaker and less significant in the two 2012 data
sets. The Summer 2012 sampling period covered four flashy high intensity storm events,
flow was quick to rise and quick to return to near baseflow conditions, so the middle four
(out of six) traps were all active for a similar length of time and retained similar masses of
sediment (Figure 5.3) making any relation between mass retention and time less
pronounced.
131
5.3.1.2 Potential sources of suspended sediment
Rothwell et al. (2005) identify four distinct catchment materials that potentially contribute
to the SS load in the UNG catchment: near-surface ‘dirty’ peat, subsurface ‘clean’ peat,
Millstone Grit sandstone, and periglacial head deposits. Shuttleworth et al. (2014b) found it
difficult to differentiate the two minerogenic components when applying a mixing model to
SS collected nearby on the Bleaklow Plateau, referring to them collectively as “underlying
geology”. This term has also been adopted for this study as there is no need to distinguish
between these two minerogenic sources.
Grab samples of the three potential sources were collected from around the catchment
using a plastic trowel. Peat samples were taken from gully walls; near-surface (n=9) from the
upper 10 cm, and subsurface (n=10) from approximately 1 m below the peat’s surface. Any
friable material on the gully wall was first cleared to avoid contamination by surface-derived
Surface peat Subsurface peat Underlying geology
(n=9) (n=10) (n=10)
OC mean 453.71 491.60 13.92
(g kg-1) max 489.46 496.26 25.74
min 421.22 484.35 4.76
Pb mean 585.15 <LOD 32.40
(µg g-1) max 933.88 <LOD 92.31
min 307.26 <LOD 0.98
x lf mean 7.80 0.00 0.24
(10-6 m2 g-1) max 14.16 0.18 0.61
min 3.43 -0.08 -0.04
SIRM mean 6637.78 3.02 92.23
(10-5 Am2 g-
1) max 12116.68 5.71 187.11
min 2016.52 1.11 18.28
ARM mean 36.06 0.15 1.29
(10-5 Am2 g-
1) max 73.99 0.21 3.19
min 10.28 0.06 0.40
SIRM/ARM mean 192.62 20.81 70.31
max 238.23 36.82 105.45
min 163.29 5.22 38.25
Table 5.2: Summary of the characteristics of the three potential sediment sources.
132
sediment which may have fallen from above (c.f. Shuttleworth et al., 2014b). Samples of the
underlying geology (n=10) were collected from exposures at the base of deeply incised gullies.
The characteristics of the four potential sediment sources are summarised in Table 5.2.
5.3.2 Laboratory analysis
5.3.2.1 Sample preparation
SS samples were washed through an 8 mm sieve with deionised water to separate the
sediment from the polystyrene. The resulting slurry was oven dried at 40 °C (so as not to
affect magnetic mineralogy of the samples; Walden et al., 1999) until a constant weight
was achieved. Once dry, samples were weighed, gently disaggregated by hand using a
pestle and mortar, and homogenised. Samples were then subsampled in triplicate,
provided enough SS had been collected to allow this.
Source samples were oven dried, disaggregated and homogenised as above.
5.3.2.2 Analysis
The range of properties available for use in sediment source fingerprinting in peat is
limited, as peat is primarily composed of organic matter, and fluctuating water tables and
changing redox conditions affect the mobility of many elements. However, Hutchinson
(1995), Rothwell et al. (2005), and Shuttleworth et al. (2014b) have made inferences about
the relative contributions of surface and subsurface peat to SS in the Bleaklow area, using
the anthropogenic contamination stored near the peat’s surface as a fingerprint. All three
studies use a suite of magnetic measurements to infer the presence of high concentrations
of anthropogenically derived, coarse grained, ferromagnetic spherules in at the peat’s
surface. Rothwell et al. (2005) and Shuttleworth et al. (2014b) also use Pb concentrations
as another indicator of material derived from the peat’s surface due to its high affinity to
organic matter and lack of mobility in wetland ecosystems (Farmer et al., 2005; Novak et
al., 2011). Shuttleworth et al., 2014b) include the organic carbon (OC) content of SS to help
distinguish material sourced from the peat mass from the underlying geology.
Magnetic susceptibility measurements (χlf and χhf) were made with a Bartington
Instruments MS2 meter (Dearing, 1994) and measurements of magnetic remanence (ARM
and SIRM) were made with a Molspin Instruments ‘Minispin’ fluxgate magnetometer
(Walden et al., 1999). Pb concentrations were determined using inductively coupled
plasma - optical emission spectrometry (ICP-OES) using a HNO3 digest, as outlined in
133
Shuttleworth et al. (2014a). The amount of organic matter (OM) was determined using the
loss on ignition (LOI) method at 550 °C for 4 h.
5.3.3 Modelling
5.3.3.1 Discriminating catchment sediment sources
A two-stage statistical procedure outlined by Collins and Walling (2002) was used to
determine the ability of the properties measured to distinguish between the three
catchment sources. Firstly, the Kruskal–Wallis H-test was used to examine the ability of
individual properties to explicitly differentiate samples of the three catchment sources, and
provide a basis for eliminating redundant fingerprint properties. Secondly, stepwise
Discriminant Function Analysis (DFA) was used to further assess the discriminatory power
of the tracer properties that passed the Kruskal-Wallis H-test. Despite limitations to using
ratios in mixing models (Walling, 2005), the SIRM/ARM ratio has been included in the
statistical analyses as Hutchinson (1995) and Rothwell et al. (2005) found it to be a key
parameter in detecting the presence of material derived from the peat’s surface, and
Shuttleworth et al. (2014b) successfully incorporated the parameter into their optimised
mixing model. The results of the statistical analyses are presented in Table 5.3. All 6
properties passed the Kruskal-Wallis H-test and were entered into the DFA. According to
the DFA, OM, the SIRM/ARM ratio, and Pb content can be used to distinguish between the
three catchment sources, which is in keeping with the parameters used by Shuttleworth et
al. (2014b).
5.3.3.2 Apportioning sediment sources
The relative contributions of the individual spatial source units and corresponding
uncertainties were determined using the numerical mass balance model developed by
(Collins et al., 1997), which has been applied to peatland catchments before to assess the
impact of restoration practices on sediment production, and is outlined in detail by
Shuttleworth et al. (2014b). Uncertainty was determined using a Monte Carlo sampling
framework using the median and median absolute deviation (MAD) as location and scale
estimators (c.f. Collins et al., 2012b). The goodness of fit (GOF) of the mixing model outputs
was tested by comparing the measured fingerprint property concentrations in the SS with
the corresponding values predicted by the model (i.e. the relative mean error) (Collins et
al., 2010). Pre-treatment for particle size and organic matter have not been included in the
mixing model (c.f. Shuttleworth et al., 2014b). Tracer-specific weightings to account for
within-source variation (SVsi); and each tracer’s discriminatory power (Wi) have been
134
Kruskal Wallis Test DFA
Property
H-value p-value
Wilks’ lambda
OC
30.041 0
0.0023
Pb
30.696 0
0.0004
SIRM/ARM 25.821 0
0.0002
Xlf
25.482 0
*
SIRM
29.560 0
*
ARM 30.052 0 *
Table 5.3: Results of the Kruskal–Wallis H-test and discriminant function analysis employed to select the fingerprint properties to distinguish the individual source types. Kruskal Wallis critical value at 99%
confidence = 10.60. *not reported as no more parameters required to discriminate sources.
included in the optimised model to address the heterogeneity in tracer properties in the
potential sources, particularly in the near surface pollution signal (based on the derivations
outlined in Collins et al., 2010 and Collins et al., 2012 respectively).
5.3.4 Statistical analysis
Runoff from peatlands is typically ‘flashy’, with storm hydrographs showing a rapid increase
in Q following the onset of precipitation, sharp peaks, and a return to levels just above
baseflow soon after the cessation of rainfall (Evans and Warburton, 2007). The vertical
profile of time integrated sediment samples will therefore represent a composite of
sediment carried during the rising and receding limb of the hydrograph, so in order to
investigate evidence of a Pb-flush and organic sediment exhaustion, the nature of the
discharge-sediment relation must be considered.
Suspended sediment concentrations (SSC) tend to rise rapidly, peak and fall rapidly early in
storm events, but there is some variation in the relative timing of the discharge-sediment
response (Labadz et al. (1991); SSC often peaks before Q (positive hysteresis), but may
continue to rise after Q begins to fall (negative hysteresis). In small catchments, positive
hysteresis is normally expected (Labadz et al. (1991), and Rothwell et al. (2007b) found that
SSC and Q either peaked simultaneously or displayed positive hysteresis in most of the
storm events they studied in the UNG catchment.
In order to seek evidence of a Pb-flush and organic sediment exhaustion, a series of
hypotheses have been constructed. Each successive hypothesis takes into account an
additional level of complexity posed by the SSC-Q relation which may influence the
composition of the SS collected through the vertical profile. The hypotheses will be tested
using the Spearman's rank correlation coefficient (ρ).
135
Hypothesis 1a: There is a positive relationship between the modelled proportion of peat-
derived sediment and sampling height.
Hypothesis 1b: There is a positive relationship between the modelled proportion of
surface-derived sediment and sampling height.
Rationale: Sediment captured in the upper traps will only be composed of material carried
during high flows early in storm events (the ‘flashy’ peak of the hydrograph), while
sediment collected in the lower traps will contain a composite of material carried early in
the event and later as flow subsides (i.e. the receding limb of the hydrograph). If the supply
of peat-derived sediment were to become exhausted or limited early in storm events, the
modelled SS composition in the upper traps should be relatively enriched in peat-derived
material. Material captured in the lower traps should contain a higher proportion of
material sourced from the underlying geology, which would become the dominant
sediment source if the two peat sources were to become exhausted (Figure 5.4a).
Similarly, any evidence of an early lead-flush would be more apparent in sediment captured
in the upper traps, while the lower traps would contain a higher proportion of ‘clean’
material deposited later in the event.
Hypothesis 2a: There is a positive relationship between the modelled proportion of peat-
derived sediment and the number of times each trap was the topmost during a storm
event.
Hypothesis 2b: There is a positive relationship between the modelled proportion of
surface-derived sediment and the number of times each trap was the topmost during a
storm event.
Rationale: During the majority of phases of inundation, stage did not rise high enough to
activate the top sediment trap (Figure 5.5). Therefore, any relative enrichment in surface-
or peat- derived sediment carried early in the storm will have been captured in the highest
trap to be inundated during that particular event (Figure 5.4b). The more instances a trap
were the topmost (ftop), the higher the proportion of surface- or peat-derived material it
should contain.
Hypothesis 3a: There is a positive relationship between the modelled proportion of peat-
derived sediment and the number of times each trap was the topmost during a predicted
peak in SSC.
136
Hypothesis 3b: There is a positive relationship between the modelled proportion of
surface-derived sediment and the number of times each trap was the topmost during a
predicted peak in SSC.
Rationale: A lag is often introduced to account for hysteresis when modelling SSC-Q
relationships, and several studies have incorporated a 30 minute lag between peak SSC and
peak Q into the construction of sediment rating curves in the UNG catchment (e.g. Evans et
al., 2006; Rothwell et al., 2007c; Pawson et al., 2008). Consequently the relationship
between sediment source and the frequency that each trap was topmost during a potential
peak in SSC (fSSC) 30 min before peak Q has also been considered. Any relative enrichment
in surface- or peat- derived sediment carried early in the storm will have been captured in
the highest trap to be inundated during predicted peaks in SSC (Figure 5.4c).
137
Figure 5.4: Hypothesised patterns of suspended sediment (SS) composition collected at different stages of the hydrograph should organic or contaminated sediment become limited early in storm events (not to scale). (a)
Proportion of organic/contaminated sediment reducing with sampling height; (b) higher proportions of organic/contaminated sediment retained by traps that were topmost during peak flows; (c) higher
proportions of organic/contaminated sediment retained by traps that were topmost during predicted peaks in suspended sediment concentration (SSC), prior to peak flows.
138
5.4 Results
5.4.1 Catchment conditions
The conditions which characterise each sampling campaign are summarised in Table 5.4.
While classic flashy flow regimes were exhibited for most of the phases of inundation,
there are several examples where conditions within the catchment effected the stream’s
response to rainfall, such as water table recharge following prolonged dry periods (e.g.
Summer 2011: Figure 5.5a), and fluctuations in rainfall intensity throughout some storms,
resulting in multiple peaked hydrographs (e.g. Autumn and Winter 2011: Figure 5.5b and
5.5c).
Summer 2011 followed an exceptionally warm dry spring (hottest spring temperatures
recorded in catchment between 2003 and 2013, rain 24% below 2003-2013 catchment
average) and was the driest of the five sampling campaigns with a mean daily rainfall of less
than 4mm, (September 2011 rainfall 30% below 2003-2013 catchment average) and water
table was supressed bellow 5 cm for 24% of the sampling period. This was followed by the
wettest sampling campaign, Autumn 2011, which had a mean daily rainfall of 13 mm and
water table within 5 cm of the surface for 97.2% of the time, and rainfall was 45% above
the October catchment average. Winter 2011 was the only sampling campaign to be
affected by frost. Although ground frost was not monitored, needle ice was observed in the
catchment during Winter 2011. Based on air temperature recorded at the AWS, an
estimated 17 days of the sampling period could have been affected by ground frost.
November rainfall was the second lowest on record, and much of the precipitation fell as
snow.
2012 was the wettest year on recorded at UNG. Summer 2012 experienced in excess of
150% average catchment rainfall, with June rainfall more than double the monthly average,
and the sampling period was dominated by four heavy storm events. The two 2012
sampling campaigns experienced higher rainfall intensities than 2011. Autumn 2012
experienced the highest intensity storm recorded throughout all of the sampling
campaigns, with maximum intensity reaching 11.3 mm hr-1.
139
Sampling season
Summer 2011
Autumn 2011
Winter 2011
Summer 2012
Autumn 2012
Duration
4 weeks 2 weeks 4 weeks 3 weeks 4 weeks
Events
Number of inundation events
3 7 5 6 12
Number of multiple peaks
0 3 1 2 2
Number of times topmost trap
a 1 0 (0) 1 (1) 1 (1) 2 (0) 1 (0)
(top during predicted peak in SSC) 2 1 (0) 2 (2) 0 (0) 2 (3) 0 (1)
3 1 (0) 2 (1) 1 (1) 0 (0) 2 (0)
4 0 (2) 1 (2) 0 (0) 0 (1) 4 (6)
5 1 (0) 1 (2) 3 (3) 2 (2) 5 (6)
Antecedent conditions
Average temperature of the preceding season (°C)
7.0 11.4 9.3 5.9 11.6
Total rainfall of the preceding season (mm)
178.7 329.6 370.1 372.2 643.2
Seasonal deviation from average
Temperature (°C; long-term seasonal average 2003-2013)
-0.9 1.5 -0.3 -0.8 -1.2
Rainfall (mm; long-term seasonal average 2007-2013)
-77.8 -35.0 6.3 235.8 70.2
Water table (cm depth; during sampling period)
Max water table
-19.5 -9.0 -5.5 -15.2 -18.8
Mean water table
-4.9 -0.3 -1.2 -2.5 -3.7
Temp (°C; during sampling period)
Mean
11.4 8.3 4.3 11.0 10.8
Max
19.7 13.5 10.5 20.3 21.2
Min
6.5 0.1 -1.3 6.4 2.4
Rainfall (mm; during sampling period)
Total rainfall (mm)
101.91 181.8 107.11 168.91 155.75
Mean daily (mm)
3.6 13.0 3.8 8.0 5.6
Max intensity of events
<3mm hr-1
0 2 0 2 3
3-6mm hr-1
3 4 5 0 6
6-9mm hr-1
0 1 0 4 0
>9mm hr-1
0 0 0 0 1
Snow lying/frost evident
No No Yes No No
Table 5.4: Summary of the conditions that characterise each sampling campaign. a
Sediment collected by the bottom trap (TIMS6) was not truly representative of stormflow conditions, so was omitted from the
statistical analysis and not included in this table.
140
141
Figure 5.5: Duration of TIMS inundation over the five sampling campaigns: (a) Summer 2011, (b) Autumn 2011, (c) Winter 2011, (d) Summer 2012, (e) Autumn 2012. TIMS1 was the uppermost to be installed, TIMS6
was at the bottom of the stack.
5.4.2 Predicted source contributions
Figure 5.6 shows the mean modelled contributions of each source type to the SS collected
at different flow depths during each of the five sampling campaigns. The composition of SS
collected in all traps is dominated by the underlying geology with modelled inputs ranging
from 61 to 91%. Material derived from the peat’s surface makes up between 3 and 15% of
the total SS load while subsurface peat contributes 1 to 32%. Overall, the organic
component (total of surface and subsurface peat combined: 9 to 39 %) is low for a peatland
catchment but falls into the lower end of the ranges reported by Rothwell et al. (2007b)
and Shuttleworth et al. (2014b): 9 to 77% and 9 – 100% respectively.
The Summer and Winter 2011 sampling campaigns (Figs 5.6a and 5.6c) show evidence that
SS collected in the upper traps contains higher proportions of material derived from the
142
contaminated surface, while the Autumn 2012 sampling campaign appears to have
captured the opposite trend (Figure 5.6e). Generally, higher proportions of material
derived from the underlying geology can be found in SS collected by lower traps during the
three 2011 sampling campaigns; however, SS collected in the topmost trap during Summer
and Autumn 2011 contains substantially higher modelled proportions of material from the
underlying geology than the trap immediately below (Figures 5.6a and 5.6b). Again, the
Autumn 2012 sampling campaign shows the reverse pattern, with the highest contributions
from the underlying geology found in the top three traps.
5.4.3 Relationship between sediment source and sampling height
(Hypothesis 1)
The relationships between sampling height and the modelled proportions of material
derived from the surface and the peat mass are presented in Table 5.5. There is very little
correlation between sampling height and modelled sediment source. The only significant
relationship is between trap height and the proportion of surface-derived material for the
Winter 2011 sampling campaign, which displays very strong positive correlation at the 99%
level. If the significance level is lowered to 90%, there are strong negative relationships
between sampling height and both surface and peat-derived sediment collected during
Autumn 2012 (the opposite direction to the hypothesis).
2011 2012
Summer Autumn Winter Summer Autumn
Surface peat ρ 0.800 0.334 0.928 -0.031 -0.714
Sig. 0.100 0.259 0.004 0.477 0.055
Total peat ρ 0.800 -0.029 0.580 0.000 -0.714
Sig. 0.100 0.479 0.114 0.500 0.055
Table 5.5: Spearman’s rank correlations for sampling height vs. modelled proportions of suspended sediment derived from the surface and the peat mass. Significant parameters are given in bold (90% confidence
interval).
143
Figure 5.6: Modelled relative contributions of individual source types to suspended sediment over the five sampling campaigns: (a) Summer 2011, (b) Autumn 2011, (c) Winter 2011, (d) Summer 2012, and (e) Autumn
2012.
144
5.4.4 Relationship between sediment source and peak Q and SSC
The relationship between sediment source and the frequency that each trap was the
topmost to be inundated (ftop) is presented in Table 5.6. The relationship between sediment
source and the frequency each trap was topmost to be inundated during the predicted
peak in SSC (fSSC) is also presented. In some instances, prolonged periods of light rain (< 3
mm hr-1) maintained flow through the bottom trap following a storm event, or raised stage
slightly above base level to activate the bottom trap only. As a result the sediment
collected in this trap is not truly representative of stormflow conditions, rather a mix of
stormflow and elevated baseflow, and as such is not suitable to look for evidence of
organic sediment exhaustion or a Pb-flush, so ftop and fSSC for the bottom trap were omitted
from the statistical analysis.
All correlations are only significant at the 90% level. The data from all of the 2011 sampling
campaigns produce a strong positive relationship between ftop and the proportion of
sediment derived from the two peat sources (i.e. SS collected in traps which were the
topmost during individual events is relatively enriched in organic material). There is also
very strong positive correlation between ftop and the proportion of surface-derived material
in SS collected during the Summer 2011 sampling campaign. There is no significant
correlation between either of the sediment sources and ftop during the 2012 sampling
campaigns, but both display strong positive correlation between fSSC and the proportion of
peat-derived sediment.
2011 2012
Summer n=4 Autumn n=6 Winter n=6 Summer n=6 Autumn n=6
ftop fSSC ftop fSSC ftop fSSC ftop fSSC ftop fSSC
Surface peat ρ 0.894 -0.949 0.484 -0.623 0.423 0.423 0.211 0.033 0.029 0.213
Sig. 0.053 0.026 0.165 0.093 0.202 0.202 0.344 0.475 0.478 0.343
Total peat ρ 0.894 -0.738 0.638 0.278 0.626 0.626 0.297 0.708 0.029 0.638
Sig. 0.053 0.131 0.087 0.297 0.092 0.092 0.284 0.058 0.478 0.087
Table 5.6: Spearman’s rank correlations for ftop and fSSC vs. the modelled proportions of suspended sediment derived from the surface and the peat mass for each of the five sampling campaigns. Significant parameters
are given in bold (90% confidence interval).
145
5.5 Discussion
5.5.1 Testing the hypotheses
Hypothesis 1: Looking for evidence of sediment exhaustion in material captured at high
flows
The lack of correlation between sampling height and modelled sediment source (Table 5.5)
could indicate one of two things:
1. Organic sediment does not become exhausted during storm events and a lead-flush
only occurred during the Winter 2011 sampling campaign,
2. Hypothesis 1 does not test a suitable model for sediment transport in the
catchment.
The former of these two interpretations is unlikely to be the case, as the statistical analysis
relating to Hypothesis 3 does show evidence of organic sediment exhaustion during all five
sampling campaigns (discussed in more detail below), and organic sediment exhaustion
would be expected based on previous research (Francis, 1990; Labadz et al., 1991). This
hypothesis may have produced more significant correlation if the traps had only captured
sediment over the course of a single storm, or if all storm events had had similar
characteristics and produced a uniform set of hydrographs. However, as is often the reality
of field sampling, this was not the case, and during multiple storms traps collected
sediment from different stages of the hydrograph. As such, this hypothesis presented an
over simplistic model of flow generation and sediment transport in the catchment.
Hypotheses 2 and 3: Looking for evidence of sediment exhaustion in material captured at
peak Q and peak SSC
The positive relationships between the proportion of peat-derived sediment and ftop and
fSSC (Table 5.6) provide consistent evidence that sediment carried early in storm events is
relatively enriched in organic matter. This provides further evidence that organic sediment
becomes exhausted/limited during storm events in peatland catchments. As Pb has a
strong affinity for organic matter (Rothwell et al., 2007b), we can infer that if a lead-flush
was to have occurred, evidence would have been captured in a similar way (i.e. Summer
2011). The fact that the 2011 source data correlates with ftop while the 2012 correlates with
fSCC highlights the need to consider the SSC-Q relationship when investigating sediment
provenance during storm flow.
146
5.5.2 Organic sediment exhaustion and supply limitation
Tallis (1973), Francis (1990) and Labadz et al. (1991) have highlighted the importance of
sediment ‘preparation’ in relation to organic sediment supply in peatland catchments.
Freshly exposed peat is fibrous, cohesive and resistant to water erosion, while weathering
by either frost action or desiccation produces a superficial friable layer on bare peat
surfaces which is readily mobilised and rapidly depleted (Evans and Warburton 2007). The
positive relationships between the proportion of peat-derived sediment, and ftop and fSSC
(Table 5.6) for all of the sampling campaigns, suggests that sediment carried early in the
majority of the storm events was relatively enriched in organic matter. This indicates that
organic sediment exhaustion in the UNG catchment must be occurring predominantly at
the event scale, despite storms often occurring in quick succession, which would not allow
time for surface preparation or weathering between events. Evans and Warburton (2007)
note that it is very rare that the friable layer is completely removed during a single storm
event, and Carling, (1983), Labadz et al. (1991) and Yang (2005) also found that substantial
sediment transport can occur throughout a series of storms in peatland catchments,
indicating that the supply of organic sediment becomes limited over the course of a storm,
rather than fully exhausted. Klove (1998) also observed this intra-storm sediment
exhaustion in field experiments on mined peat, and suggests that this may be associated
with a decrease in the erosive power of rainsplash as the peat’s surface wets up, and the
incision of rills into the more resistant unweathered peat below.
Further evidence of organic sediment supply limitation in the UNG catchment can be seen
in the relative proportions of sediment collected by the topmost traps for the Summer
2011 and Autumn 2011 sampling periods (Figure 5.6a and 5.6b). Although, all of the 2011
data correlates with ftop indicating there was no lag between peak SS organic matter
concentration and peak Q, in both cases, the SS collected by the topmost trap is composed
of substantially higher proportions of material from the underlying geology than the trap
below. During the Summer 2011 sampling period, peak flow only reached the topmost trap
once, 9 hours after the preceding peak in Q (Figure 5.5a). Labadz et al. (1991) found
evidence of that sediment supply can become limited between storms which occur in close
temporal proximity in peatland catchments, so it is likely that the first storm mobilised
most of the readily available friable material (evidenced in the high proportions of peat-
derived material collected in TIMS3 - the topmost trap for this event: Figure 5.5a and Figure
5.6a), limiting the amount available for transport during the successive storm. Similarly,
peak flow also only reached the topmost trap once during the Autumn 2011 sampling
147
campaign, this time after a prolonged period of rain of variable intensities which did not
produce a typically flashy hydrograph (Figure 5.5b), so it is possible that the supply of
organic sediment had become limited before the top trap became active.
The correlation between fSSC and the proportion of peat-derived sediment during the 2012
sampling campaigns indicates that sediment was entering the fluvial system more quickly in
2012 than in 2011. 2012 was the wettest year that has been recorded at UNG between
2003 and 2013, with total annual rainfall 38% higher than the catchment average, Summer
and Autumn 2012 experienced some of the highest seasonal rainfall recorded in the
catchment, with June rainfall more than double the monthly average, and the two 2012
sampling campaigns also experienced higher rainfall intensities than 2011 (Table 5.4).
Goulsbra et al. (2014) cite catchment wetness as a key control on the connectivity in
peatland systems, and Holden and Burt (2002a) found that rain splash was an important
agent of disturbance and entrainment of bare peat as particles. The wetter conditions
during 2012 may have maintained an expanded drainage network, linking ephemeral
gullies to the main channel, so at the onset of heavy rain, organic sediment was quickly
exported to the main channel, before peak Q.
5.5.3 Evidence for a lead-flush
Only the Summer 2011 and Winter 2011 sampling campaigns show any evidence of an
initial flush of contaminated surface-derived sediment early in storm events. The very
strong positive correlation between ftop and the proportion of surface-derived material in
sediment collected during the Summer 2011 sampling campaign (ρ=0.894, p=0.053),
suggests that material carried at peak flow, early in the storm events, contained a higher
proportion of Pb contaminated material than sediment which was transported as flow
subsided. While no such relationship is evident in the Winter 2011 data, there is a very
strong relationship between the proportion of surface-derived sediment and sampling
height (ρ=0.928, p=0.004) indicating that sediment transported at high flows was relatively
enriched in Pb compared to sediment collected at lower flow conditions. The upper traps
were only inundated on one occasion during the Winter 2011 sampling campaign, so the
lead-flush may have been restricted to this one high flow event.
Rothwell et al. (2005) propose that a combination of high rainfall intensity (8-12 mm hr-1)
early in the storm event, coupled with a high water table, which would have generated
surface or near-surface runoff, were responsible for the initial Pb-flush that they observed.
However, during the Summer and Winter 2011 sampling campaigns rainfall did not exceed
148
6 mm hr-1, while the other three sampling campaigns all contained at least one higher
energy event (Table 5.4). Furthermore, the water table was at or within 5 cm of the surface
at the onset of the majority of storm events throughout all of the sampling campaigns,
which would have quickly generated surface or near-surface flow. This suggests that other
factors must be driving the Pb-flush in the UNG catchment, and that these may change
through time. Sections 5.5.3.1. and 5.5.3.2. explore the catchment conditions that may
have contributed to the sediment characteristics observed during the Summer and Winter
2011 sampling campaigns.
5.5.3.1 Summer 2011
Rothwell (2006) found that sediment production from the contaminated surface layer is
highest during Summer months, and that sediments collected from the base of gully walls
is relatively enriched in Pb by a factor of 4 during this period, compared to winter months.
Rothwell (2006) also observed that this sediment enriched with surface-derived material
collects on gully floors during dry periods (Figure 5.7), and is subsequently washed through
the system when it rains. Summer 2011 was the driest of all of the sampling campaigns and
followed an exceptionally warm dry spring (Table 5.4) which would have desiccated
exposed peat surfaces producing the friable material noted by Francis (1990) and Tallis
(1973), and will likely have been enriched in Pb (c.f. Rothwell, 2006).
Figure 5.7: Desiccated peat collecting on gully floor (Source: J. J. Rothwell).
149
There were only three phases of substantially elevated stage (higher than the bottom trap)
during the sampling period, and the top three traps were only inundated during two storms
which occurred in quick succession towards the end of the sampling period (Figure 5.5a).
The topmost trap (TIMS2) collected a higher proportion of surface-derived material than
the one immediately below (TIMS3) despite only being inundated during the second of
these two closely spaced storm events. The SS captured by TIMS2 also contained less peat-
derived sediment overall, indicating that there may have been a shift in the dominant
sediment source between the two storms. Two key mechanisms could have contributed to
this: supply limitation and network expansion.
As discussed in Section 5.5.2., sediment supply can become limited between storms which
occur in close temporal proximity in peatland catchments (Labadz et al., 1991). The first
storm will have mobilised the friable material from the gully walls, and flushed any
desiccated sediment which had collected on gully floors through the system (Rothwell,
2006). This is evidenced in the high proportions of surface and subsurface peat collected in
TIMS3 (the topmost trap for this event). During the second storm, this ready supply of
organic material will have become limited in the main channel, but may not yet have been
mobilised from the upper reaches of the catchment. Hydrological connectivity is
increasingly being recognised as a spatiotemporal control on biogeochemical cycling and
material fluxes (Pringle, 2003), particularly in agricultural and urban settings (e.g.
Wigington et al., 2005; Heathwaite et al., 2005; Berg et al., 2008), and Goulsbra et al.
(2014) suggest that hydrological connectivity within the UNG catchment may be an
important control on sediment and contaminant release, with antecedent water table
depth controlling the rate of network expansion. Prior to the two storms, the water table at
the weather station had been supressed to 149 mm by six days of warm dry weather and
was recharged by a prolonged period of light rain leading up to the two storm events
(Figure 5.6a). Although the water table was within 10 mm of the surface at the AWS
immediately preceding the first storm, it may have taken longer to recharge elsewhere in
the catchment, particularly in the upper reaches which are heavily dissected by Bower Type
I gullies (Bower, 1961). The catchment will have progressively ‘wet up’ during the course of
this extended period of rainfall, connecting the more distal headwater gullies to the main
channel.
Rothwell et al. (2010b) demonstrate that suspended sediments exported from catchments
with a shallow mean upslope gully depth (i.e. headwaters) have a much higher lead content
than those exported from catchments with a deep mean upslope gully depth. In shallower
150
gullies, relatively uncontaminated deep peat layers represent a smaller proportion of the
exposed gully wall, providing less potential for dilution of peat particulates sourced from
the contaminated surface as they move down the gully walls (Evans and Warburton, 2005).
Thus, although the overall supply of organic material in the catchment had been reduced
by the first storm, it is likely that sediment sourced from these newly connected headwater
gullies will have been released into the main channel during the second storm, contributing
to the high proportion of surface material carried during peak flow and captured in TIMS2.
5.5.3.2 Winter 2011
Similar to the Summer 2011 sampling campaign, the two uppermost traps were only
inundated once during Winter 2011, and third trap was only inundated on one further
occasion (Figure 5.5c; Table 5.4). These three traps contain significantly higher amounts of
material derived from the peat’s surface than the bottom three traps (Figure 5.6c).
However, the conditions in the catchment were very different to Summer 2011: while the
beginning of the sampling period was unusually dry, wetter conditions prevailed towards
the end, and snow was the main form of precipitation. This was also the only sampling
period to be affected by frost.
Frost heave and needle ice formation have been shown to play a major role in the
preparation of erodible material (Tallis, 1973; Evans and Warburton, 2007) as these
processes destroy the structure of the surficial peat (Luoto and Sepälä, 2000) producing a
fluffy loose texture which is easily dislodged and transported. Francis (1990) notes that
frost heave preferentially affects previously loosened peat, and as discussed above,
Rothwell et al. (2010b) found that surficial sediment on gully walls is a mixture of clean
peat and material that has fallen from the contaminated layer above, so this redeposited
sediment will be particularly prone to frost action. An estimated 17 days of the sampling
period could have been affected by ground frost, which would have produced this frost
‘fluff’ on gully walls.
The upper three traps, which contain some of the highest proportions of surface-derived
material, were only inundated during a relatively small precipitation event (max intensity,
4.04 mm hr-1, total rainfall 7.25mm). Stage began to rise before the onset of precipitation
and coincided with a rise in temperature from 1 to 7.2 °C, indicating that this phase of
inundation may have been largely driven by runoff from snowmelt rather than
precipitation. Snowmelt can have a significant effect on catchment runoff and sediment
production as the water stored in the snowpack is released (Evans at al., 1999; Lana-
151
Renault et al., 2011), and has been shown to be a vector for heavy metal release into the
fluvial system (Erel and Patterson, 1994; Lin and Herber, 1997; Rember and Trefry, 2004).
Any frost loosened material would have been left undisturbed by the relatively low volume
and intensity precipitation events which characterised the rest of the sampling period,
some of which fell as snow, further reducing the amount of runoff produced. However, the
large volume of runoff generated by melting which produced enough Q to activate the
topmost trap, would have easily mobilised the loose redeposited material and flushed
contaminated sediment through the system, as evidenced by the elevated proportion of
surface-derived material collected by the upper traps during this event.
5.6 Conclusion
This study represents the first use of time integrated mass flux samplers (TIMS) and
sediment source fingerprinting to explore the mechanisms of organic sediment and heavy
metal release at the event scale.
This method has produced evidence that suspended sediment carried at the beginning of
storm events is relatively enriched with peat-derived material. This confirms the accepted
model of organic sediment exhaustion during the course of storm events, and that organic
sediment transport becomes limited between storms which occur in quick succession,
reducing the amount of organic matter available for transport during successive storms.
The timing of this organic sediment exhaustion is linked to catchment wetness and rainfall
intensity.
The contaminated surface layer of the peat is releasing Pb into the fluvial system
throughout the year, but a flushing of Pb early in storm events is only evident under certain
conditions. Sediment ‘preparation’ by either desiccation or frost action is a key precursor
for a ‘lead-flush’. Hydrological connectivity and snowmelt may also play an important role
in transferring contaminated sediment from ephemeral headwater gullies to the main
channel.
These findings have implications for future sediment release under predicted changes in
climate and land management practices. Projected warmer drier summers may enhance Pb
release. Water table drawdown may become more prevalent and prolonged, disconnecting
ephemeral gullies from the main channel for longer periods allowing desiccated
contaminated material to build up on gully floors which would eventually be flushed
through the system in high concentrations. Alternatively, restoration efforts which aim to
152
elevate water table, such as gully blocking, would increase hydrological connectivity, better
linking headwaters which are more productive of contaminated sediment to the main
channel.
5.7 Acknowledgements
We would like to thank The University of Manchester for the provision of a Graduate
Teaching Studentship (to E. Shuttleworth) and for funding for analytical costs. We are
grateful to The National Trust and United Utilities for allowing work to be carried out at the
study sites. Thanks also go to John Moore and Jonathan Yarwood for their assistance in the
lab, to all those who helped in the field, particularly Jeff Blackford for his assistance in
setting up the field kit, to Simon Hutchinson for the loan of magnetics kit, and to Gareth
Clay for his help proof reading the manuscript.
153
Chapter 6 Contaminated sediment dynamics in peatland
headwaters (Paper 4)
This chapter is in preparation for submission to Catena as Shuttleworth, E. L., Clay, G. D.,
Evans, M. G., Hutchinson, S. M., & Rothwell, J. J. “Contaminated sediment dynamics in
peatland headwaters”.
Abstract
The near-surface layer of the blanket peats of the Peak District National Park, southern
Pennines, UK, is severely contaminated with high concentrations of anthropogenically
derived, atmospherically deposited lead (Pb). These peats are severely degraded, and there
is increasing concern that erosion is releasing considerable quantities of this legacy
pollution into surface waters. However, there is also evidence that a substantial proportion
of contaminated surface sediment may be stored elsewhere in the catchment. This study
uses the Pb contamination stored near the peat’s surface as a fingerprint to trace
contaminated sediment dynamics in three severely degraded headwater catchments, to
investigate the mechanisms that control Pb release and storage in peatland systems.
Erosion is exposing high concentrations of Pb on interfluve surfaces, and substantial
amounts of reworked contaminated material is stored on other catchment surfaces. A
variety of mechanisms have been shown to control Pb release and storage on the different
surfaces, including: (i) wind action on interfluves; (ii) the aspect of gully walls, and (iii) gully
depth. Vegetation also plays an important role in retaining contaminated sediment on all
surfaces.
Key words: Peat; Lead; Erosion; Deposition; Wind; Weathering; Vegetation
6.1. Introduction
The near surface layer of peatlands in close proximity to urban and industrial areas can be
contaminated with atmospherically deposited heavy metals (Vile et al., 1999; Rothwell et
al., 2010a) and as such, peatlands can represent significant sinks of anthropogenically
derived Pb (Shotyk et al. 2000; Farmer et al. 2005; Rothwell et al. 2007a). Many blanket
peats in the UK are substantially degraded as a result of climate change, pollution, and
154
mismanagement (Ferguson et al., 1978; Holden et al., 2007; Bonn et al., 2009).
Consequently there is concern that erosion is releasing substantial quantities of Pb
contaminated sediment into the fluvial system (Yang et al., 2002; Shotbolt et al., 2006;
Rothwell et al., 2007a, 2008b, 2010b), and that the removal of contaminated surface
material is reducing the ability of a peatland to act as a long-term sink for atmospherically
deposited contaminants (Rothwell et al., 2008b).
Over the last three decades there has been a move to restore areas of degraded peatland
(Tallis and Yalden, 1983; Wheeler et al., 1995; Gorham and Rochefort, 2003), and recent
peatland Pb research has focussed on understanding the processes involved in the release
of contaminated sediment, in order to inform effective management strategies:
Rothwell et al. (2005, 2007a, and 2008b) have shown that peat erosion is releasing
substantial amounts of Pb into the fluvial system in both particulate and dissolved
phases.
Tipping et al. (2003) and Lucassen et al. (2002) modelled Pb release under varying
acidification and drought scenarios.
Shotbolt et al. (2006) used reservoir sediments to determine fluxes of heavy metals
from contaminated catchments, and highlight downstream issues for water quality
and aquatic ecosystems.
Shuttleworth et al. (2014a) developed the use of FPXRF in peatlands to assess
contamination across wide spatial scales.
Rothwell et al. (2010b) and Shuttleworth et al. (2014b) developed models to
determine landscape scale patterns of sediment associated Pb release into the
fluvial system.
Near surface Pb storage has been shown to be highly heterogeneous due to spatial
variability in atmospheric Pb deposition (Bindler et al., 2004; Farmer et al., 2005; Rothwell
et al., 2007a), and can be further complicated by removal of surface material from exposed
surfaces in degraded areas (Shuttleworth et al., 2014b). This complexity presents a
significant challenge when modelling catchment Pb fluxes. Rothwell et al. (2010b) derived a
strong relationship between gully depth and sediment-associated lead concentrations, but
Shuttleworth et al. (2014b) found that this relationship broke down in areas of severely
degraded peat where a higher proportion of material derived from the contaminated
surface was entering the fluvial system. Shuttleworth (unpublished data) also found that in
some headwater catchments suspended sediment Pb concentrations exceed those stored
155
on interfluve surfaces, indicating that there may be substantial storage of Pb contaminated
sediment elsewhere in the catchment. A deeper understanding of the mechanisms of Pb
release and storage is therefore required to better quantify contaminant export from
eroding peat systems.
Evans and Warburton (2005) and Evans et al. (2006) provide the first comprehensive data
on the full range of peat erosion processes from a single peatland site. However, these
studies focus on constructing sediment budgets and quantifying catchment scale export of
organic sediment, and while Rothwell et al. (2007b) surmised that variability in the Pb
content of fluvial sediments was likely due to differences in catchment erosion processes,
to date, there has been no attempt to provide equivalent data on the mechanisms which
control Pb release and storage. This paper uses the Pb contamination stored near the
peat’s surface as a fingerprint to trace contaminated sediment movement and storage in
three severely degraded peat headwater catchments. Pb concentrations on different
catchment surfaces (interfluves, gully walls and channel floors) are quantified, and patterns
in Pb storage are investigated in reference to established peatland geomorphic theory
(Table 6.1) to identify the key controls on contaminated sediment dynamics.
Control Mechanism Effect Study
Vegetation Protects surface; Traps mobilised sediment
Reduces sediment production; Reduces sediment movement
Evans and Warburton (2005); Shuttleworth et al. (2014b)
Weathering Frost heave/needle ice; Desiccation
Prepares' surface; produces readily mobilised sediment
Tallis (1973); Francis (1990); Labadz et al. (1991); Luoto and Sepälä (2000)
Erosive process
Fluvial; Aeolian; Mass movement
Mobilises sediment; deposits sediment
Holden and Burt (2002b); Foulds and Warburton (2007a, b); Warburton et al. (2004)
Degree of degradation
Surface removal; Gullying
Controls sediment associated Pb concentrations
Shuttleworth et al. (2014b); Rothwell et al. (2010b)
Table 6.1: Summary of selected controls on peatland sediment dynamics.
156
6.2. Field area
The Bleaklow Plateau is an area of upland blanket peat that makes up part of the Peak
District National Park (PDNP) in the Southern Pennines, UK (Figure 6.1a). The plateau lies
between 500 and 633 m asl, and is situated between the industrial cities of Manchester
and Sheffield. Peat depths of up to 3 m (Evans and Lindsay 2010a) overlie a sandstone
bedrock from the Millstone Grit Series (MGS) (Wolverson-Cope, 1976) and fine grained
head deposits of weathered MGS shales (Rothwell et al., 2005). Mean monthly
temperatures vary between 12.9 °C (July) and 1.44 °C (February) (2003-2013), annual
rainfall is 1020-1840 mm (2007-2013), and the prevailing wind direction is SSW (195°)
(unpublished AWS data). The peat is amongst the most degraded and contaminated in the
world. Anthropogenic and climatic pressures have caused widespread erosion (Bower,
1960a, 1906b, 1961; Tallis 1985; Bonn et al., 2009), and Shuttleworth et al. (2014a)
recorded Pb concentrations in excess of 1700 mg kg-1 exposed at the peat’s surface; a
legacy of atmospheric deposition during the English Industrial Revolution.
Bleaklow has been the focus of a multi-million pound restoration initiative (Shuttleworth et
al., 2014b), but this study concentrates on three headwater catchments in an actively
eroding area of the plateau to the north of Bleaklow Head (Figure 6.1b). This area has been
purposefully left in its degraded state to act as a baseline for comparison with restored
areas. Consequently the field site has been the focus of recent research into carbon
release, pollutant dynamics and peatland restoration (e.g. Clay et. al., 2012; Dixon et al.,
2013; Cole et al., 2014; Shuttleworth et al., 2014a and 2014b).
The three headwater catchments are typical of the area, with steep walled gullies with
depths varying from around 1m at the gully heads to 3-4 m at the gully mouths. Vegetation
cover is sparse and bare peat is prevalent. Any vegetation present on interfluve surfaces is
composed of a mixture of low lying shrubs (Calluna vulgaris, Erica tetralix, Vaccinium
myrtillus) and cotton grass (Eriophorum vaginatum) and is likely a composite of the original
pre-gullying vegetated surface and some newly-established vegetation. Vegetation on gully
walls and floors is dominated by cotton grass, which is interpreted to have established on
these surfaces post-disturbance. A few ericaceous shrubs are a present on gully walls which
appear to have originated at the peat’s surface but have been transported downslope
during localised slope failure.
157
Figure 6.1: Location of the study site. (a) The Bleaklow Plateau in relation to the industrial cities of Manchester and Sheffield. The red start denotes the gullied field area, just north of the Bleaklow summit. The blue star denotes the location of the automatic weather station. (b) View down Catchment 2 from Transect A, showing Transects B to D. Transect markersare spaced at 2 m intervals.
6.3. Methods
6.3.1. Field Survey
Surface Pb concentrations were measured at 1 m intervals along four parallel transects
which ran perpendicular to the three gullies (Figure 6.1b). A total of 188 readings were
taken using a handheld Niton XL3t 900 XRF analyser following the method outlined in
Shuttleworth et al. (2014a). Where necessary, vegetation was removed and the peat’s
surface was lightly compacted by hand in order to present a smooth flat surface to the XRF
sensor (c.f. Ridings et al. 2000). There is no commercially available XRF Certified Reference
Material (CRM) for heavily contaminated peat so NCS DC73308 (Chinese stream sediment)
was used as this has the most appropriate Pb concentration of the CRMs available to the
158
study. The relative percent difference (RPD) between the concentration in the reference
material and the concentration measured by the FPXRF was within 10% for Pb. Samples
from the top 10-15 mm of each sampling point were collected using a stainless steel
palette knife in order to determine the water content to correct for the dilution effect of
the high moisture content of the peat (c.f. Shuttleworth et al., 2014a).
6.3.2. Data Analysis
Surface, catchment and vegetation effects 6.3.2.1.
A general linear model (GLM) approach based on an analysis of variance (ANOVA) was
employed to determine the statistical significance of the influence of three factors and
their interactions on Pb storage. Pb concentrations on the different catchment surfaces
were compared to investigate the relative amounts of contaminated sediment stored on
each surface type. Heterogeneous aerial surface deposition (Bindler et al., 2004; Rothwell
et al., 2007a) may influence the level of contamination within each catchment which would
restrict the amount of contaminated sediment available for redistribution on the different
surfaces, so this was also considered. Finally, the presence or absence of vegetation was
also included, as vegetation and sediment dynamics are closely linked in blanket peats
(Evans and Warburton; 2005; Evans et al., 2006; Shuttleworth et al., 2014b).
Initial investigation of the data showed that Pb storage on northwest (NW) facing gully
walls was significantly different to Pb storage on southeast (SE) facing walls (2-tailed t-test,
α=0.000) so these were included in the model as two separate surface types. In doing so,
the power of the model’s ability to explain the variance in the data increased from 12% to
16%. Pb data within each factor were tested for normality (Anderson-Darling) and equality
of variance (Levene); any factors that failed were square root transformed and retested.
After transformation, all bar two factors met the required criteria. While datasets failing to
meet the assumptions of ANOVA are not ideal, Rutherford (2001) states that
interpretations of GLM-ANOVA models remain robust with moderate amounts of
assumption violation provided the factor level sample sizes are greater than five, as is the
case with this model (Table 6.2). Tukey’s pairwise comparison was applied post hoc, in
order to assess where the significant differences lie. All relationships were tested at the 95
% level (p ≤ 0.05). The magnitude of the effects of each significant factor and interaction
were calculated using a generalised ω2 (Olejnik and Algina, 2003).
159
Directional trends 6.3.2.2.
Foulds and Warburton (2007a) found that the dominant direction of peat flux was closely
aligned with the prevailing wind, and Warburton (2003) and Foulds and Warburton (2007b)
showed that sediment fluxes in the direction of the prevailing wind can be up to 12 times
greater than in the opposing direction. Prevailing wind direction measured at the
automatic weather station situated on the south east edge of the Bleaklow Plateau (Figure
6.1a), was SSW (195°), so any directional trends in Pb contaminated sediment storage
caused by wind action should be evident in a roughly north – south alignment. Pb storage
on gully walls and floors may be influenced by gully depth. Rothwell et al. (2010b) showed
that gully wall Pb concentrations decrease with depth below the interfluve surface, and
that suspended sediment Pb values varied inversely with MUGD as contaminated and
‘clean’ peat particulates mix when sediment moves down the faces of eroding walls.
The nonparametric Spearman’s rank correlation coefficient was employed to test for
directional trends in the un-transformed data (Hollander and Wolfe, 1973). Pb
concentrations were tested against their corresponding northing value as a proxy for wind
direction. The effect of MUGD was assessed on gully walls by considering the relationship
between Pb and the depth of the sampling point below the interfluve surface (c.f. Rothwell
et al., 2010b). The gully depth map developed by Evans and Lindsay (2010a) was not of a
suitable resolution to derive MUGD for the gully floor sampling points so the relationship
between Pb concentrations on gully floors and distance from the gully head was tested
based on the assumption that in headwaters, gully depth rapidly increases with distance
from gully head.
Factor Level n Mean R.S.D. max min
Catchment 1 48 124 95.7 502 < LOD
2 69 222 116 1660 < LOD
3 71 213 91.5 1010 < LOD
Surface type Tops 75 245 81.3 1660 < LOD
NW facing walls 43 80.5 121 382 < LOD
SE facing walls 37 207 60.4 555 < LOD
Floors 33 209 119 1010 < LOD
Vegetation cover Bare 137 191 103 1010 < LOD
Vegetated 51 200 120 1660 < LOD
Table 6.2: Descriptive statistics for each factor tested by the GLM.
160
Figure 6.2: (a) Schematic depicting mean lead concentrrationsons measured along the four transects (A-D) on the different catchment surface types (figure not to scale); (b) Spread of lead concentrations grouped by
surface type.
6.4. Results
Figure 6.2 shows Pb storage on the different catchment surfaces, Table 6.2 summarises the
Pb concentrations which characterise each of the levels that make up the factors tested by
the GLM, and Table 6.3 summarises the results of the ANOVA. All surfaces store substantial
amounts of Pb contaminated sediment, but this storage is highly variable. Concentrations
across the field site range from below the limit of detection to 1660 g kg-1, and Pb
161
Source p ω²
Catchment 0.015 0.033
Surface 0.000 0.172
Vegetation cover 0.642 0.000
Catchment*Surface 0.062 0.026
Catchment*Vegetation cover 0.227 0.002
Surface*Vegetation cover 0.003 0.043
Total
0.276
Table 6.3: ANOVA results for square-root transformed data. P = probability of factor being zero and ω² = generalized proportion of variance explained. Significant results in bold.
contaminated and clean sediment can be found in all catchments and on all surface types.
Variability is evident both between and within the catchments and surface types; all bar
one relative standard deviation (RSD) is in excess of 80% (Table 6.2). Interfluve surfaces
contain the highest Pb concentrations (mean 244 g kg-1, max 1660 g kg-1) but there is also
considerable Pb storage on gully floors and walls.
The total explanatory power of the model is only 27.6% (Table 6.3). This is a reflection of
the complexity pattern of Pb storage displayed in Figure 6.2, and is likely in part a product
of the interference of directional relationships. Surface type was found to be a significant
control on Pb storage, explaining 17.2 % of the variation in the data. Post-hoc testing shows
Pb storage on NW facing walls is significantly less than on SE facing walls (p=0.002) and
interfluve surfaces (p=0.000), but no other significant differences were identified between
surface types (Figure 6.3b). Catchment is also significant, explaining 3 % of the variation in
the data (Table 6.3); Pb storage in Catchment 1 is significantly lower than that of
Catchment 3 (p=0.011; Figure 6.3a). The interaction between surface type and vegetation
cover was also found to be significant, explaining 4 % of the variation in the data, but
vegetation cover alone was not (Table 6.3). Bare NW facing walls contain significantly lower
Pb concentrations than all other bare surfaces (p<0.001), and vegetated interfluve surfaces
(p=0.000) and vegetated SE facing walls (p=0.005). Vegetated gully floors also contain
significantly lower Pb concentration than vegetated interfluve surfaces (p=0.008). Although
there is no statistically significant difference in Pb storage between bare and vegetated
areas on individual surfaces, there are some interesting relationships which should be
noted. Pb concentrations tend to be higher under vegetation on interfluve surfaces and on
NW facing gully walls; mean Pb storage on bare interfluve surfaces is 67% of the mean
vegetated value, while mean Pb storage on bare NW facing walls is only 34% of the mean
concentration found under vegetation (Table 6.4; Figure 6.3c). In contrast, Pb
concentrations are substantially lower under vegetation on gully floors than on bare areas.
162
Mean Pb storage on bare gully floors is almost 3 times higher than the mean vegetated
concentration (Table 6.4; Figure 6.3c). This indicates that vegetated and bare areas affect
contaminant storage in different ways on the different surface types, and may explain why
the GLM did not identify vegetation as a significant factor.
There is weak yet significant positive correlation between northing and Pb concentration
on interfluve surfaces and NW facing walls (Table 6.5), but no such relation is apparent on
gully floors or SE facing walls. There is a significant negative relationship between Pb
storage and distance from gully head on gully floors (ρ = -0.359, p = 0.011, 1 tailed, Figure
6.4), but on gully walls there is no correlation between Pb concentration and MUGD (Table
6.6).
Bare Vegetated
All Surfaces 191 200
Interfluves 223 338
NW facing walls 51.2 156
SE facing walls 215 192
Gully floors 281 99.4
Table 6.4: Mean lead storage on bare and vegetated surfaces (µg g-1
).
Figure 6.3: Relationship between lead storage on gully floors and distance from gully head.
163
Figure 6.4: Interval plots for factors and interactions which produced significant differences when comparing lead storage based on ANOVA depicting 95% confidence intervals for the means.
164
n ρ p
Interfluves 78 0.325 0.004
Floors 33 0.150 0.406
NW facing walls 43 0.322 0.035
SE facing walls 37 0.238 0.156
Table 6.5: Spearman’s rank correlations for prevailing wind direction vs. Pb storage on the different catchment surfaces. Significant parameters are given in bold (95% confidence interval).
n ρ p
All 81 0.027 0.407
SE facing 38 0.079 0.319
NW facing 43 0.042 0.395
Table 6.6: Spearman’s rank correlations for mean upslope gully depth vs. Pb storage on the gully walls. Significant parameters are given in bold (95% confidence interval).
6.5. Discussion
The statistical analyses have highlighted several factors which influence Pb storage within
the headwater catchments. The following sections discuss these results in reference to the
controls summarised in Table 6.1.
6.5.1. Catchment
Catchment 1 contains significantly less Pb than the other two catchments (Figure 6.3a), and
although the interaction between catchment and surface type was not shown to be
significant, Pb storage on the different surface types in Catchment 1 is substantially lower
when compared to corresponding surface types in the other two catchments (Figure 6.2b).
Rothwell et al. (2007a) showed that the near surface record of Pb deposition in the
southern Pennines can vary both horizontally and vertically over relatively short distances;
peak Pb values can occur at different depths in the peat profile, and can vary by up to 1000
µg g-1 over only a few hundred metres. There are several possible explanations for this
variability, including: differences in peat accumulation rates (e.g. Mighall et al., 2002),
spatial heterogeneity in atmospheric deposition (e.g. Norton et al., 1997), spatial and
temporal variation in plant community (e.g. Bindler et al., 2004), and varying rates of
decomposition (Biester et al., 2003). The observed variations in Pb storage on interfluve
surfaces at the field site are likely due a combination of these factors, and Shuttleworth et
165
al. (2014b) have also demonstrated that surface Pb concentrations can be influenced by
the depth of material removed by erosion in degraded systems.
The amount of Pb stored in the near surface layer of peat will then affect the Pb content of
any sediment that is mobilised from interfluve surfaces that is subsequently re-deposited
on other catchment surfaces (e.g. walls and floors). Overall catchment Pb storage will
therefore depend on the amount of Pb that was initially deposited and stored, the severity
of subsequent erosion, and the efficiency of sediment removal from the catchment.
However, within the constraints of this study it is not possible to assess whether lower
levels of Pb storage in Catchment 1 are due to initial lower deposition and retention rates,
or a greater degree of erosion than experienced by the other two catchments.
6.5.2. Surface type
Although there is great variation in Pb storage across all of the different catchment
surfaces, Pb concentrations on SE facing gully walls and gully floors are statistically similar
to the levels found on the surface of interfluves (Table 6.3). Maximum Pb storage on
interfluve surfaces (1660 µg g-1) is comparable to the highest values recorded by Shotyk et
al. (2000) in Gola di Lago, Switzerland (1527 µg g-1) and by Zhulidov et al. (1997) in Fenno-
Scandia tundra in Russia (1650 µg g-1), and the maximum Pb concentrations found on SE
facing gully walls and gully floors (555 and 1010 µg g-1) exceed many previously reported
maximum near surface Pb concentrations found in peatlands around the world (e.g. 479 µg
g-1 in Bozi Dar, Czech Republic: Vile et al., 2000; 400 in Lochnagar, Scotland: Yang et al.,
2001).
Rothwell et al. (2007b) found that re-deposited fluvial sediment (on floodplains and trash-
lines) elsewhere in the Peak District contained some Pb, but concentrations of these re-
worked sediments were one or two orders of magnitude lower than those stored near the
peat’s surface. Similarly, gully walls have been shown to store Pb contaminated sediment,
with Pb concentrations reducing downslope as contaminated material mixes with ‘clean’
sediment below the contaminated layer (Rothwell et al., 2010b). However, there is no
evidence of Pb concentrations decreasing with distance down gully walls (Table 6.6), and
the high Pb concentrations recorded on gully floors and SE facing gully walls greatly exceed
those recorded in reworked sediment by Rothwell et al. (2007b). This indicates that in
headwater catchments, the contaminated surface layer is releasing substantial amounts of
contaminated sediment, which is subsequently stored on gully walls and floors, and that
166
these surfaces are important areas of deposition and storage. The factors controlling
storage on these surfaces are discussed in Sections 6.5.3. and 6.5.5.
NW facing gully walls also store some Pb, but they are the ‘cleanest’ of the four surface
types, containing significantly lower Pb concentrations than the other three surfaces (Table
6.2). The differences in Pb storage between NW and SE facing gully walls indicates that
aspect plays an important role in determining the amount of sediment storage on gully
walls and is explored further in Section 6.5.5.
6.5.3. Vegetation cover
The presence of vegetation has been shown to greatly influence sediment storage in
peatlands (Tallis and Yalden, 1983; Evans and Warburton, 2005; Evans et al., 2006;
Shuttleworth et al., 2014b), and although the vegetation cover alone was not found to
control Pb storage at the field site (mean Pb storage on bare and vegetated surfaces only
differs by 9 µg g-1 – Table 6.4), the interaction between surface type and vegetation cover is
significant.
The lowest Pb concentrations overall are found on bare NW facing gully walls and
vegetated gully floors, while the highest concentrations are found on vegetated interfluve
surfaces and bare gully floors (Figure 6.3c; Table 6.4), indicating that the presence or
absence of vegetation cover doesn’t have a consistent effect on Pb storage on the different
surface types. Vegetation cover does not contribute to any statistically significant
differences in Pb storage on individual surface types, likely due to the high variability in Pb
values recorded on each surface (Table 6.2), but there are some marked differences in
mean Pb storage on bare and vegetated surfaces on three of the four surface types (Table
6.4; Figure 6.3c). Higher mean Pb concentrations are found under vegetation than on bare
areas on NW facing gully walls and interfluve surfaces, while the opposite relationship is
evident on gully floors, where mean Pb storage is highest on bare areas. This indicates that
vegetation may be influencing Pb storage in different ways on the different surface types.
Interfluve surfaces 6.5.3.1.
In the Peak District, peak Pb concentrations of up to 1650 µg g-1 can typically be found
between 5 and 10 cm below the peat’s surface, surface Pb concentrations in excess of 300
µg g-1 are common, and Pb contamination is minimal below depths of 30 cm (Rothwell et
al., 2007a, 2007b). Shuttleworth et al. (2014b) found that there is relatively little variation
in Pb concentrations across the surface of intact areas of peatland (290 – 400 µg g-1), while
167
Pb concentrations on interfluves in degraded areas range from 10s to 1000s µg g-1. In intact
peatlands vegetation inhibits surface recession but exposed peat is highly susceptible to
erosion (Tallis, 1997; Evans and Warburton, 2007), and Shuttleworth et al. (2014b) surmise
that varying rates of surface erosion in degraded areas expose different stages of the Pb
deposition profile at the peat’s surface.
The observed differences in Pb storage on vegetated and bare interfluve surfaces in the
study catchments are likely also a product of differing rates of surface lowering. Bare areas
on interfluves will be vulnerable to erosive processes, and will have been stripped of some
(at some sampling sites, all) of the contaminated layer, resulting in surface Pb
concentrations which range from below the FPXRF limit of detection to 1050 µg g-1, while
any vegetation present on interfluves will have protected the underlying peat from erosion,
preserving the contaminated layer of sediment below. The highly variable levels of Pb
exposed on bare areas average out to a mean substantially lower than the mean Pb
concentrations preserved under vegetation (Table 6.4).
Gully walls 6.5.3.2.
The record of atmospheric Pb deposition is limited to the upper 30 cm of the peat profile
(Rothwell et al., 2007a, 2007b). When gully walls are cleared of superficial friable material,
Pb concentrations are negligible (Shuttleworth et al., 2014b), so any Pb enriched material
found on gully walls is interpreted to be reworked sediment derived from the near surface
contaminated layer.
Similar to interfluves, Pb storage is also considerably higher under vegetation than on bare
areas on NW facing walls. However, the vegetation on NW facing gully walls is influencing
Pb storage in a different way than on interfluves; rather than preserving an intact
contaminated surface, vegetation on gully walls will intercept contaminated peat particles
which have been mobilised from the peat’s surface as they move downslope (Evans and
Warburton, 2005; Rothwell et al., 2010b). The vegetation will then protect this redeposited
sediment from subsequent erosion (as described in Section 6.5.3.1.) while any
contaminated sediment deposited on bare walls is easily remobilised, leading to a relative
enrichment of Pb contamination under gully wall vegetation.
No such relationship is evident on SE facing gully walls where Pb concentrations are similar
on bare and vegetated faces. This provides further evidence that aspect is controlling
sediment storage on gully walls (see Section 6.5.5).
168
Gully floors 6.5.3.3.
Vegetated gully floors and floodplains have been shown to be important areas of sediment
deposition in peatlands (Evans et al., 2005; Rothwell et al., 2007b), and gully floor
vegetation is often cited as pivotal in reducing connectivity between eroding surfaces and
the fluvial system (e.g. Evans and Warburton, 2005; Crowe et al., 2008; Molina et al., 2009).
Ostensibly, it is therefore surprising that mean Pb concentrations are three times higher on
bare peat than on vegetated surfaces, indicating greater storage of contaminated sediment
on bare surfaces. However, Evans and Warburton (2005) note that sediment is not
efficiently transported across vegetated alluvial fans, and indicate that sediment deposition
may be limited to the upstream extremity of vegetated surfaces, so contaminated
sediment may not have reached the vegetated sampling sites on gully floors. Figure 6.5
shows freshly deposited peat building up behind tussocks of Eriophorum (cotton grass) on
the floor of Catchment 2, indicating that vegetation on gully floors may be encouraging
upstream deposition in a similar manner to gully blocks (Evans et al., 2005). Bare areas of
peat on gully floors therefore represent significant deposition of reworked material derived
from the contaminated surface layer.
It is also possible that the sampling method may have not fully captured the Pb storage on
vegetated areas of gully floors. Surface vegetation was cleared in order to present a
smooth flat surface to the FPXRF sensor (c.f. Ridings et al., 2000) but Eriophorum produces
a dense root network which would have remained in situ, and may have ‘diluted’ the FPXRF
Pb reading.
Figure 6.5: Freshly deposited peat accumulating behind tussocks of Eriophorum on the floor of Catchment 2.
169
6.5.4. Wind
The positive correlation between Pb concentration and northing value indicates that Pb
storage increases in a northerly direction on interfluve surfaces, in the leeward direction of
the prevailing wind (SSW, 195°). Aeolian processes have received relatively little attention
in the study of peatland erosion, but Foulds and Warburton (2007a) found that the
dominant direction of peat flux was closely aligned with the prevailing wind, and
Warburton (2003) and Foulds and Warburton (2007b) showed that sediment fluxes in the
direction of the prevailing wind can be up to 12 times greater than in the opposing
direction. Wind erosion dominated by rainsplash transports peat particles over relatively
short distances (1 to 10 m per event), while transport of ‘peat dust’ under dry conditions
can be much greater (in excess of 50 m). Under these conditions individual peat particles
can move hundreds of metres in the direction of the prevailing wind over the course of a
year (Warburton, 2003). As such, wind action may be driving the observed pattern of
leeward Pb enhancement on interfluve surfaces by removing contaminated material from
the windward extreme of interfluve surfaces and re-depositing it downwind (Figure 6.6a).
Alternatively, surface deflation may be exposing different stages of the Pb depositional
profile on interfluve surfaces, exposing higher concentrations on the leeward extremes
(Figure 6.6b).
Within the constraints of the study it is difficult to say which hypothesis is the most likely
scenario. Warburton (2003) cites wind-assisted splash as the dominant wind erosion
process in peatlands, implying a gradual enhancement of Pb as contaminated material is
progressively moved in a leeward direction across interfluves. However, considerable
transport can also occur as dry blow (Foulds and Warburton 2007a) which could be
evacuating substantial amounts of contaminated sediment from the windward extremes of
interfluves. Contaminated sediment could either be redeposit on other surface types within
the catchment, potentially contributing to some of the observed patterns of Pb
enhancement (e.g. the correlation between NW facing gully wall Pb and wind direction), or
the contaminated sediment may be removed from the catchment altogether and be
redeposited ‘off site’. Further investigation is required to determine whether the observed
leeward pattern of surface Pb enhancement is driven by freshly deposited material or
exposure of the peak in Pb deposition preserved in the peat profile (c.f. Rothwell et al.,
2007a).
170
Figure 6.6: Schematics depicting possible explanations for the leeward lead enhancement on interfluve surfaces (not to scale). (a) Contaminated material is incrementally moved in a leeward direction across
interfluves by rain splash. (b) Surface deflation is exposing different stages of the Pb depositional profile on interfluve surfaces, exposing higher concentrations on the leeward extremes. The dotted black line represented the pre-erosion surface; the black line represents the post-erosion surface; blue arrows
represent sediment movement by wind; red lines represent the lead depositional profile.
6.5.5. Aspect
Pb storage is significantly higher on SE-facing than on NW-facing walls (Figure 6.2; Table
6.2), indicating that aspect plays an important role in determining the mechanisms behind
sediment dynamics on gully walls in eroding peatlands. Sediment ‘preparation’ is often
cited as an important control on sediment production and mobilisation on bare peat
surfaces (e.g. Tallis, 1973; Francis, 1990; Labadz et al. 1991); freshly exposed peat is fibrous,
cohesive and resistant to water erosion, while weathering produces a superficial friable
layer on bare peat surfaces which is readily mobilised and rapidly depleted (Evans and
Warburton 2007). Francis (1990) noted that surface recession was greatest on southwest
facing areas of bare peat, stressing the importance of desiccation-related phenomena,
while Birnie (1993) reported maximum erosion on northerly aspects suggesting that greater
frost frequency was an important factor.
171
NW facing gully walls are more prone to disturbance by frost heave and needle ice
formation which destroy the structure of the surficial peat (Luoto and Sepälä, 2000)
producing a fluffy loose texture which is easily dislodged and transported. Francis (1990)
notes that frost heave preferentially affects previously loosened peat, so any redeposited
contaminated sediment will be particularly prone to frost action and easily removed. Thus
Pb storage is low on NW facing gully walls. Evans and Warburton (2007) cite running water
at the dominant mechanism in mobilising sediment on gully walls; however, the significant
correlation between Pb concentrations on NW facing slopes and prevailing wind direction
(Table 6.5) indicates that rainsplash may also influence sediment movement on these
surfaces (Warburton, 2003). Despite this low Pb storage, it should be noted that any
vegetation present on NW-facing walls traps and protects some contaminated sediment
from subsequent mobilisation (Section 6.5.3.2.).
SE facing slopes are more sheltered from frost action in winter but more prone to drying
and desiccation during summer. They also face the prevailing wind (195°) and are more
exposed to aeolian action, so it is perhaps surprising that contaminated sediment appears
to be accumulating on these surfaces (Figure 6.2; Table 6.2). Desiccated surface crusts
develop over extended periods of dry weather, and while the aggregates derived from
these crusts have a very low density and are easily transported (Evans and Warburton,
2007), initially crusts increase the threshold for entrainment and peat will only be detached
and entrained if surface roughness increases, or shear velocities are high (Foulds and
Warburton, 2007a). As such, any desiccation crusting on SE facing gully walls may serve to
protect the surface from wind and water action, allowing contaminated sediment to build
up.
6.5.6. Gully Depth
Pb concentrations in gully floor sediments decrease with distance from gully head (Figure
6.4), supporting the concept that clean peat makes up a greater proportion of sediment
with distance downstream. As gullies deepen, relatively uncontaminated ‘clean’ peat
represents a larger proportion of the exposed gully wall. The surface derived Pb signal
becomes progressively diluted as contaminated and clean and particulates mix (Rothwell et
al., 2010b).
Despite the evidence from gully floors that the proportion of clean to contaminated
sediment increases with gully depth, there is no relationship between MUGD and Pb
storage on gully walls (Table 6.6). This is similar to the findings of Shuttleworth et al.
172
(2014b) which also did not show any relationship between MUGD and the proportion of
suspended sediment derived from the peat’s contaminated surface. Rothwell et al. (2010b)
derived the MUGD-sediment associated Pb relationship in catchments where interfluve
were well vegetated, so any Pb would only have been sourced from the contaminated layer
exposed at the top of gully walls. In the study catchments, no such relationship is evident
due to the prevalence of bare interfluve surfaces, which will be releasing large volumes of
contaminated material (Shuttleworth et al., 2014b), masking any effect of an increasing
pool of ‘clean’ peat as gullies deepen.
6.6. Conclusions
The legacy of atmospheric Pb deposition stored near the peat’s surface has been
successfully employed as a fingerprint to trace contaminated sediment movement and
storage in degraded peat headwater catchments. Erosion is exposing high concentrations
of Pb on interfluve surfaces, and is mobilising substantial amounts of contaminated
material. Pb contaminated sediment is stored on all catchment surfaces.
Pb concentrations are highly variable on all of the catchment surfaces, and a variety of
mechanisms control Pb release and storage on the different surfaces. Complex small scale
spatial patterns of contaminant storage can be explained by interactions between
topographic setting and vegetation cover, and the mobilisation of sediment by wind and
water:
1. Vegetation plays an important role in retaining contaminated sediment on all
surfaces.
2. Wind erosion is driving patterns of Pb storage on interfluve surfaces.
3. Aspect is key in controlling sediment preparation and Pb storage on gully walls
4. Gully depth influences Pb concentrations found in gully floor sediments.
With regards to peatland restoration, this study provides further evidence that vegetation
plays a key role in stabilising the peat’s surface and trapping mobilised sediment, thus
reducing contaminant export. There is a significant amount of sediment storage in the gully
system which may affect previous estimates of particulate carbon and contaminant loss
from eroding peatlands. Wind has also been highlighted as vector for contaminated
sediment transport, a fraction that is as yet unaccounted for in estimates of peatland Pb
export. Fine metal-laden airborne particulates have been show to affect human respiratory
health in urban environments (e.g. Voutsa and Samara, 2002; Fernandez Espinosa et al.,
173
2002) where sediment associated Pb contamination is not as severe as that found in the
near surface peats of the Peak District. Consequently, the airborne component of the Pb
budget in eroded peatland could be quite significant in a toxicological terms given the
number of people that visit the Peak District (in excess of 10 million visitors days per year;
Global Tourism Solutions, 2009), and contaminated surfaces prone to aeolian action should
be stabilised as a matter of priority.
6.7. Acknowledgements
We would like to thank The University of Manchester for the provision of a Graduate
Teaching Studentship (to E. Shuttleworth) and for funding for analytical costs. We are
grateful to The National Trust and United Utilities for allowing work to be carried out at the
study sites. Thanks also go to Jack Dods and Ioanna Tantanasi for their help in the field.
174
Chapter 7 Summary and Conclusions
The legacy of Pb contamination stored near the peat’s surface in the Peak District has
provided a unique opportunity to study sediment dynamics in eroding and restored
systems. A suite of standard techniques has been modified and adapted for use in peatland
environments, and these have been successfully employed in combination to address
issues of sediment and contaminant release at a range of scales.
7.1. Peatland sediment dynamics (Overarching aim)
Throughout this thesis, certain mechanisms and controls have been shown to be important
influences on sediment dynamics and Pb release across a range of scales.
7.1.1. Vegetation
Vegetation and sediment production are closely linked in eroding blanket peatlands.
Vegetation plays an important role in stabilising the peat’s surface and trapping mobilised
sediment, thus influencing Pb storage and the export of Pb and POC.
At the plot scale, surface Pb concentrations can provide information about the effect that
vegetation has on sediment storage on different surface types (Paper 4). On interfluves,
vegetated surfaces generally contain higher Pb concentrations than those measured on
areas of bare peat; the vegetation protects the underlying peat from erosive processes,
preserving the polluted near surface layer, while contaminated material is more easily
removed from bare surfaces. Elevated Pb levels can also be observed under vegetation on
eroding gully walls, indicating that the vegetation is trapping contaminated sediment that
has been eroded from above. On gully floors, Pb values are lower under vegetation than on
bare surfaces. Whilst appearing to contradict the pattern observed on interfluve surfaces
and gully walls, sediment is not efficiently transported across vegetated surfaces (Evans
and Warburton, 2005), so sediment deposition may be limited to the upstream extremity
of vegetation gully floors. Bare areas represent freshly deposited reworked material which
will be transported further through the system during subsequent storms.
The stabilising effect of vegetation is clearly demonstrated at the landscape scale (Paper 2).
In gullied systems where bare peat is prevalent, all surfaces actively contribute to the
suspended load, while in catchments where there is full vegetation cover on interfluve
surfaces, bare gully walls are the main locus of sediment production. Full vegetation cover
on interfluve surfaces effectively shuts down surface sediment production and reduces Pb
175
and POC export. Conversely, in severely degraded areas where there is little vegetation
cover on interfluve surfaces, POC and Pb fluxes are significantly higher as there is a larger
area of exposed sediment available for mobilisation.
The presence of vegetation on gully floors is also an important factor in reducing Pb and
POC export. Despite the prevalence of bare peat on gully walls in re-vegetated gullies, Pb
and POC fluxes are similar to those of an intact peatland. The near-full vegetation cover on
gully floors in these re-vegetated catchments is likely trapping material mobilised from
gully walls, reducing connectivity between erosive surfaces and the fluvial system.
7.1.2. Sediment preparation
Sediment preparation plays a role in the timing of POC and Pb release. Within site
variations in Pb and POC fluxes were observed at all Bleaklow field sites (Paper 2). The
different sampling campaigns will have experienced a range of antecedent conditions and
thus differing levels of sediment preparation over the 16 months of data collection, leading
to temporal fluctuations in sediment supply which would have influenced the volume and
nature of sediment collected. Sediment carried at the beginning of storms in the UNG
catchment is relatively enriched with peat derived material (Paper 3), corroborating the
findings of Francis (1990) and Labadz et al. (1991): that the organic sediment supply
becomes limited during the course of storm events in peatland systems. Both desiccation
and frost action are key precursors for a flushing of Pb contaminated sediment through the
gully system at the beginning of a storm event (Paper 3). This phenomenon was first
observed by Rothwell et al. (2005), and was only found to occur after an extended dry
period which allowed dry Pb contaminated material to build up on gully floors, and a period
when frost loosened material was flushed through the system by snowmelt.
Aspect has been shown to be a key control on how sediment is ‘prepared’ and stored on
gully walls (Paper 4). North west facing walls which are more exposed to frost action
contain very low Pb concentration, indicating that this type of weathering efficiently
prepares sediment for removal from the wall’s surface. In contrast, the surface of south
east facing gully walls, which are more sheltered from frost action in winter but more
prone to desiccation during summer, contain high Pb values. This indicates that these
surfaces are storing contaminated material mobilised from the peat’s surface, and contrary
to the findings of Paper 3, desiccation crusting may serve to protect the newly deposited
surface from erosion, allowing Pb contaminated sediment to build up.
176
7.1.3. Meteorological conditions
Antecedent water tables influence the timing and the nature of sediment entering the
fluvial system during storm events. Paper 3 suggests that the rate of network expansion is a
temporal control on the release of Pb enriched material from shallow ephemeral
headwaters into the main channel, seemingly confirming the hypothesis put forward by
Goulsbra et al. (2014): that catchment wetness may impact sediment flux by controlling the
level of hydrological connectivity. Additionally, rainfall intensity has been highlighted as an
important factor in determining the timing of sediment transfer to the main channel (Paper
3). Pb storage increases in the leeward direction of the prevailing wind, indicating that wind
action influences sediment dynamics on interfluve surfaces (Paper 4). However, the exact
mechanism is as yet unclear (see section 9.5).
7.1.4. Degree of degradation
The degree of degradation influences both Pb storage and release. There is relatively little
variation in Pb concentrations found across the surface of an intact peatland, while
concentrations across degraded and re-vegetated areas range across several orders of
magnitude. Differing rates of erosion have exposed the peak concentrations of Pb stored
below the peat’s surface or removed the polluted surface layer all together (Papers 2 and
4).
Surface condition also plays an important role in determining the dominant source of
suspended sediment (SS) (Paper 2) and the Pb storage on gully walls (Paper 4). Rothwell et
al. (2010b) found that in catchments with vegetated interfluves, gully depth controls the Pb
content of SS and patterns of Pb storage on gully walls, due to conservative mixing of
contaminated and clean peat as sediment moves down the face of gully walls. However,
Papers 2 and 4 show that in catchments where interfluve surfaces are actively eroding, no
such relationship is evident as there is greater availability and mobility of material derived
from the contaminated surface. In severely degraded areas SS is sourced from both
interfluve surfaces and gully walls, whereas in gullied areas where interfluve surfaces are
stable (i.e. following re-vegetation) gully walls become the dominant source of sediment.
7.2. Development of Methods (Objective 1)
Methodologically, this thesis represents a study of firsts. It has seen the first use of field
portable X-ray Fluorescence (FPXRF) to assess in situ Pb concentrations in wet organic
sediments; the first use of time integrated mass flux samplers (TIMS) to explore landscape
177
scale sediment dynamics in peatlands, and to investigate sediment release at the event
scale; and the first successful application of a numerical mixing model to obtain
quantitative estimates of the relative importance of different sediment sources in peatland
catchments. The key findings and developments are as follows:
1) Pb concentrations derived by correcting in situ FPXRF readings for moisture content
correlate strongly with results obtained by processing samples ex situ, and in situ
results can be easily converted using simple linear regression equations for
comparison with existing studies (Paper 1). As such, FPXRF provides a cost-effective
and rapid tool for assessing Pb contamination in peatlands, and has been used to
assess landscape- and plot-scale contaminant storage in in Papers 2 and 4.
2) The TIMS first described by Owens et al. (2006) can be adapted for deployment at
multiple remote field sites by replacing the standard gravel filling with a light-
weight polystyrene alternative (Chapter 4). The sediment retained by the Owens
style TIMS is of similar composition to that collected by the more commonly used
design described in Philips et al. (2000), and also overcomes some operational
difficulties encountered by the Philips style sampler (e.g. lack of sediment
retention, unreliable in ephemeral flow), providing a reliable sediment trap for use
in organic systems (Papers 2 and 3). These TIMS can also be used to assess event
scale temporal trends by using ephemeral streamflow (ES) sensors, to monitor the
duration of TIMS activity (Paper 3).
3) Sediment source fingerprinting and numerical mixing models which are techniques
traditionally used to determine sources of fine sediment in systems dominated by
minerogenic material, can be applied to the investigation of SS composition in
contaminated organic rich upland catchments. By exploiting the pollutants as a
distinctive fingerprint of surface derived material by careful selection of a set of
conservative tracers, sediment mobilised from interfluve surfaces can be
distinguished from material eroded from gully walls. This approach can be used to
look at both spatial (Paper 2) and temporal (Paper 3) variations in SS composition.
7.3. Sediment dynamics at different spatial scales (Objective 2)
When looking at sediment dynamics at the landscape scale, the presence or absence of
vegetation appears to be the dominant control on sediment and contaminant release
(Paper 2). Pb and C export following re-vegetation is comparable to an intact peatland,
while fluxes are two orders of magnitude greater in areas with little or no vegetation cover.
178
At the catchment scale, the supply of sediment dictates SS composition (Paper 3). This is
controlled by the physical availability of erodible organic sediment produced through
weathering, and the degree of hydrological connectivity which governs the time scale at
which ephemeral headwaters release higher Pb concentrations become linked to the main
channel.
The plot scale analysis detailed in Paper 4 highlights the variety of mechanisms controlling
Pb release and storage on different catchment surfaces: Wind erosion is driving patterns of
Pb storage on interfluve surfaces, aspect is key in controlling sediment preparation and Pb
storage on gully walls, and MUGD/distance from gully head influences Pb concentrations
found in gully floor sediments. Vegetation plays an important role in retaining sediment on
all surfaces.
7.4. Implications for Restoration and Management
Paper 2 is the first time-integrated assessment of the effect of peatland re-vegetation on
sediment production at the landscape scale, and provides a strong theoretical justification
for the re-vegetation techniques which have been pioneered by Moors for the Future in the
Bleaklow area. Vegetation has been shown to reduce sediment production by stabilising
interfluve surfaces, and reduce sediment and pollutant export by decreasing connectivity
between the erosional surfaces and channels.
The findings of this thesis have implications for peatland management practices and
provide information that will help target future initiatives to counter erosion, and reduce
carbon and pollutant release:
Paper 2 suggests that Rothwell et al. (2010b) underestimated SS Pb concentrations
in gullied areas where bare peat is exposed on interfluve surfaces, and these bare
areas should be the focus of peatland restoration as a matter of priority to reduce
sediment associated Pb export.
Papers 2 and 4 highlight the role of gully floor vegetation in intercepting sediment.
POC intercepted by vegetation at the slope-channel interface, and stored on gully
floors has the potential to oxidise to CO2 and contribute to the overall greenhouse
gas emissions from the area so further research into the magnitude and longevity
of POC storage by gully floor vegetation is needed to fully understand the impact of
restoration on the overall carbon balance.
179
Paper 2 indicates that gully walls become the dominant sediment source post-
restoration, and Papers 3 and 4 highlight the importance of sediment preparation
as a key control on the mobilisation of contaminated sediment from gully walls.
Future improvements in restoration efforts should therefore additionally
concentrate on stabilising gully walls to further limit sediment production (e.g.
Geojute – Parry et al., 2014)
While increasing catchment wetness is essential in order to restore ecological
function, Paper 3 suggests that restoration efforts which aim to elevate water table
(thus increasing hydrological connectivity and better linking headwaters to the
main channel) may allow more contaminated sediment to enter the main channel.
This potentially negative side effect should be considered and accounted for in
future restoration initiatives.
Paper 3 indicates that Pb is preferentially released in pulses following dry or frosty
conditions which may have implications for downstream water quality.
7.5. Further work
7.5.1. Extend the use of FPXRF
FPXRF shows great promise as a tool for a rapid and cost-effective means of determining
the Pb content of contaminated peatlands. The method outlined in Paper 1 should be
applicable to the study of Pb concentrations in any contaminated peatland setting
following a brief confirmatory analysis, and could be applied to analyses outside of the
scope of this thesis, such as rapid, in field core logging and the construction of detailed Pb
inventories. There is also scope to extend its use to other contaminated waterlogged
environments such as salt marshes (Williams et al., 1994) and estuaries (Pan and Wang,
2012).
Shand and Wendler (2014) found that FPXRF produced satisfactory results when analysing
copper concentrations in ombrotrophic peat and Kneen (unpublished data) has used FPXRF
to determine relative changes in the concentration of silicon and titanium in peat profiles
as an indicator of anthropogenic activity, which suggests that there is potential to extend
the range of elements that can be analysed. This may require a different internal
calibration to that supplied with the instrument (Shand and Wendler, 2014). In order to
further improve the application of FPXRF to peat samples, certified reference materials
180
specific to peatland contamination would be a useful addition to the current range that are
available.
7.5.2. Refine and extend use of mixing models
Some magnetic parameters proved to be unreliable fingerprint properties for use in
peatland sediment source mixing models, as their inclusion overestimated the proportion
of material derived from underlying geology (Paper 2). The Inorganic Ash Spheres (IAS)
that give the near surface peat its distinctive magnetic signature may not behave
conservatively when they are eroded and subsequently transported through the fluvial
system. Consequently, IAS may be flushed out of the organic sediment during transport,
reducing the magnetic susceptibility of surface derived sediments to values closer to those
produced by the underlying geology. Lab based mixing experiments and the use of scanning
electron microscopy could help determine the behaviour of IAS during fluvial transport, in
order to assess the suitability of magnetic analyses in peatland sediment source ascription.
Other parameters could also be incorporated into the mixing model to help constrain
fingerprints. The range of properties available in peat is limited, as fluctuating water tables
and changing redox conditions can affect the mobility of many elements. Pb was used as a
fingerprint in this thesis as it has long been identified as the least mobile heavy metal in
peatland environments due to it high affinity to organic matter (Farmer et al., 2005);
however, recent studies by Rothwell et al. (2010b) and Novak et al. (2011) found that
copper and zinc are also relatively immobile in in peat profiles, and so these elements may
be suitable for inclusion in subsequent peatland fingerprinting studies.
The refined model could then be applied to the study of sediment dynamics in other
polluted and degraded peatland systems such as mined areas or those affected by fire. The
methodology outlined in Paper 2 could be extended to the long term monitoring of newly
implemented restoration programs, assessing the erosion-restoration cycle from start to
finish in order to better understand how re-vegetation and other practices, reduce POC and
contaminant release over time.
7.5.3. Better understand the controls on sediment and pollutant dynamics
This thesis has generated a substantial amount of evidence to support many existing
hypotheses relating to the factors which control sediment and pollutant dynamics in
eroding peatlands. However, some of the conclusions which have been drawn are
181
speculative as it was not possible to fully investigate all avenues of enquiry in the scope of
this study, and further examination of specific mechanisms is required.
7.5.3.1. Wind
The patterns of Pb storage on interfluve surfaces presented in Paper 4 indicates that the
prevailing wind influences sediment storage and removal on interfluve surfaces; however,
the exact mechanism is unknown, and two possible hypotheses are proposed (see paper 4)
which require further investigation to determine whether surface Pb concentrations are
from freshly deposited material or part of the Pb depositional profile described in Rothwell
et al. (2005 and 2007a).
The proportion of material carried as dry blow may affect estimates of Pb export from
catchments. Previous work on Pb export has focussed on Pb bound to particulates or in the
dissolved phase transported by fluvial system (Rothwell et al. 2007b, 2007c and 2008a), but
the wind transported fraction is as yet unaccounted for.
Fine metal-laden airborne particulates have been shown to affect human respiratory health
in urban environments (e.g. Voutsa and Samara, 2002; Fernandez Espinosa et al., 2002)
where sediment associated Pb contamination is not as severe as that found in the near
surface peats of the Peak District. Consequently, the airborne component of the Pb budget
in eroded peatlands could be significant in toxicological terms given the number of people
that visit the Peak District (in excess of 10 million visitor days per year; Global Tourism
Solutions, 2009).
7.5.3.2. Hydrological connectivity
Goulsbra et al. (2014) hypothesis that the degree of hydrological connectivity is a spatial
control on sediment supply and storm sediment flux, and Paper 3 makes inferences about
catchment wetness and the timing of sediment and Pb release but an integrated study is
required to properly investigate the link between the rate of network expansion and the
timing of sediment and contaminant release.
7.5.3.3. Sediment preparation
Many studies cite the importance of sediment preparation as a control on sediment supply
(e.g. Tallis, 1973; Tallis and Yalden, 1983; Francis, 1990; Labadz et al., 1991; Rothwell,
2006), and Papers 3 and 4 provide further evidence for this, but there has been relatively
little work detailing direct observations of the processes involved. The relative importance
of frost and desiccation is still unknown, and despite the observations of Klove (1998) and
182
Holden and Burt (2002b) the exact mechanisms driving organic sediment limitation (rather
than exhaustion) during storm events are unclear. Paper 4 also suggests that aspect is
controlling the nature of sediment preparation of gully walls, which then impacts sediment
and pollution storage, and requires further investigation.
7.6. Tracing peatland geomorphology
Under current climate change projections, the peatlands of the UK, and indeed the world,
face an uncertain future. Peatlands support a range of ecosystem services; most
importantly they represent one of the largest terrestrial carbon stores. Understanding and
preserving these systems is therefore of vital importance.
Evans and Warburton (2007) noted that the understanding of the geomorphology of
peatlands lagged behind the understanding of peatland ecology and hydrology. This thesis
has demonstrated how these three spheres of peatland science are intimately linked, and
has furthered our understanding of the geomorphic controls on sediment and pollutant
dynamics in eroding peatland systems.
This thesis has provided the first quantitative assessment of sediment and pollution
dynamics through the erosion-restoration cycle, and has highlighted the key role of
vegetation in restricting sediment production and pollutant export. However, there is an
ongoing and pressing need for additional empirical data to guide peatland management
practices and validate restoration initiatives.
The application of sediment source tracing techniques to eroding peatlands provides a
powerful new approach to aid our understanding of the geomorphic controls on these
systems.
183
References
Anderson, P., Buckler, M. and Walker, J. (2009) Moorland restoration: potential and
progress. In: A. Bonn, T.E.H. Allott, K. Hubacek and J. Stewart (Editors), Drivers of
Environmental Change in Uplands. Routledge, Abingdon, Oxon.
Andersson, J.O., and Nyberg, L. (2008) Spatial variation of wetlands and flux of dissolved
organic carbon in boreal headwater streams. Hydrological Processes 22: 1965-1975
Argyraki, A., Ramsey, M. H., Potts, P. J. (1997). Evaluation of portable X-ray fluorescence
instrumentation for in situ measurements of lead on contaminated land. Analyst 122(8),
743-749.
Atteia, O., Dubois, J. P. & Webster, R. (1994) Geostatistical analysis of soil contamination in
the Swiss Jura. Environmental Pollution, 86: 315-327
Ballantine, D.J., Waling, D.E., Collins, A.L., Leeks, G.J.L. (2008) The phosphorus content of
fluvial suspended sediment in three lowland groundwater-dominated catchments. Journal
of Hydrology 357: 140-151.
Bao, K., Xing, W., Yu, X., Zhao, H., McLaughlin, N., Lu, X., Wang, G. (2012) Recent
atmospheric dust deposition in an ombrotrophic peat bog in Great Hinggan Mountain,
Northeast China. Science of the Total Environment 431: 33-45
Barnes, F.A. (1963) Peat erosion in the Southern Pennines: Problems of interpretation. East
Midlands Geographer 3: 216-222
Batjes, N.H. (1996) Total carbon and nitrogen in the soils of the world. European Journal of
Soil Science 47: 151-163. DOI: 10.1111/j.1365-2389.1996.tb01386.x
Bell, J.N.B. (1973) The effect of prolonged low concentrations of sulphur dioxide on the
growth of two moss species. Journal of Bryology, 7: 444 - 445.
Berg, M., Trang, P.T.K., Stengel, C., Buschmann, J., Viet, P.H., Van Dan, N., Giger, W.,
Stuben, D., (2008) Hydrological and sedimentary controls leading to arsenic contamination
of groundwater in the Hanoi area, Vietnam: The impact of iron-arsenic ratios, peat, river
bank deposits, and excessive groundwater abstraction. Chemical Geology, 249(1-2): 91-112.
Bernick, M. B., Kalnicky, D. J., Prince, G., Singhvi, R. (1995) Results of field-portable X-ray-
fluorescence analysis of metal contaminants in soil and sediment. Journal of Hazardous
Materials, 43(1-2) 101-110.
184
Biester, H., Martinez Cortizaz, A., Birkenstock, S., Kilian, R. (2003) Effect of peat
decomposition and mass loss on historic mercury records in peat bogs from Patagonia.
Environmental Science and Technology 37, 32-39.
Billett, M.F., Charman, D.J., Clark, J.M., Evans, C.D., Evans, M.G., Ostle, N.J., Worrall, F.,
Burden, A., Dinsmore, K.J., Jones, T., McNamara, N.P., Parry, L., Rowson, J.G., Rose, R.
(2010) Carbon balance of UK peatlands: current state of knowledge and future research
challenges. Climate Research 45:13-29
Billett, M.F., Deacon, C.M., Palmer, S.M., Dawson, J.J.C., Hope, D. (2006) Connecting
organic carbon in stream water and soils in a peatland catchment. Journal of Geophysical
Research-Biogeosciences 111. DOI: 10.1029/2005jg000065
Bindler, R., Klarqvist, M., Klaminder, J., Forster, J. (2004) Does within-bog spatial variability
of mercury and lead constrain reconstructions of absolute deposition rates from single peat
records? The example of Store Mosse, Sweden. Global Biogeochemical Cycles 18(3).
GB3020
Birnie R.V. (1993) Erosion rates on bare peat surfaces in Shetland. Scottish Geographical
Magazine 109(1); 12-17
Blake, W.H., Wallbrink, P.J., Doerr, S.H., Shakesby, R.A., Humphreys, G.S. (2006) Magnetic
enhancement in wildfire-affected soil and its potential for sediment-source ascription.
Earth Surface Processes and Landforms 31: 249-264
Block, C. N., Shibata, T., Solo-Gabriele, H. M., Townsend, T. G. (2007) Use of handheld X-ray
fluorescence spectrometry units for identification of arsenic in treated wood.
Environmental Pollution 148(2), 627-633.
Bonn, A., Allott, T.E.H., Hubacek, K., Stewart, J. (2009) Drivers of Environmental Change in
Uplands. Routledge: Abingdon, Oxon
Bower, M.M., (1960a) Peat erosion in the Pennines. Advancement of Science, 64: 323-331
Bower, M.M. (1960b) The erosion of blanket peat in the Southern Pennines. East Midlands
Geographer 2: 22-33
Bower, M.M. (1961) The distribution of erosion in blanket peat bogs in the Pennines.
Transactions of the Institute of British Geographers 29: 17-30
185
Boyle, J. F. (2000). Rapid elemental analysis of sediment samples by isotope source XRF.
Journal of Paleolimnology 23(2) 213-221.
Bradshaw, R.H.W., Wolf, A. and Moller, P.F. (2005) Long-term succession in a Danish
temperate deciduous forest. Ecography, 28(2): 157-164.
Bragg, O.M. and Tallis, J.H. (2001) The sensitivity of peat-covered upland landscapes.
Catena, 42(2-4): 345-360.
Brännvall, M.-L., Bindler, R., Emteryd, O. and Renberg, I. (2001). Four thousand years of
atmospheric lead pollution in northern Europe: a summary from Swedish lake sediments.
Journal of Paleolimnology, 25, 421–35.
Brown, A.G., 1985. The potential use of pollen in the identification of suspended sediment
sources. Earth Surface Processes and Landforms, 10(1): 27-32.
Bruce, M.A. (2002) Country report for the United Kingdom. Baltex Fire 2000. The Second
Baltic Seminar and Exercise in Forest Fire and Information and Resources Exchange 2000.
Campbell, D.R., Lavoie, C. and Rochefort, L. (2002) Wind erosion and surface stability in
abandoned milled peatlands. Canadian Journal of Soil Science, 82(1): 85-95.
Cannell, M.G.R., Dewar, R.C. and Pyatt, D.G. (1993) Conifer plantations on drained
peatlands in Britain - a net gain or loss of carbon. Forestry, 66(4): 353-369.
Caporn, S.J.M, and Emmett, B.A. (2009) Threats from air pollution and climate change to
upland systems: past, present and future. In: Bonn, A., Allott, T.E.H., Hubacek, K., Stewart,
J. (2009) Drivers of Environmental Change in Uplands. Routledge: Abingdon, Oxon.
Carling, P.A. (1983) Particulate dynamics dissolved and total load in two small basins,
Northern Pennines, UK. Hydrological Sciences Journal 28(3); 355-375
Carling, P.A. (1986) Peat slides in Teesdale and Weardale, Northern Pennines, July 1983:
description and failure mechanisms. Earth Surface Processes and Landforms, 11: 193– 206.
Carlsen, P. (1993) Peat - geotechnical properties and up to date method of design and
construction. State-of-the-art-report. 215. Linkoping: Swedish Geotechnical Institute.
Carter, J., Owens, P.N., Walling, D.E., Leeks, G.J.L. (2003) Fingerprinting suspended
sediment sources in a large urban river system. Science of the Total Environment 314-316:
513-534. DOI: 10.1016/s0048-9697(03)00071-8
186
Charman, D. (2002) Peatlands and environmental change. Wiley, Chichester
Clark, J.M., Gallego-Sala, A.V., Allott, T.E.H., Chapman, S.J., Farewell, T., Freeman, C., House,
J.I., Orr, H.G., Prentice, I.C., Smith, P., (2010) Assessing the vulnerability of blanket peat to
climate change using an ensemble of statistical bioclimatic envelope models. Climate
Research, 45(1): 131-U462.
Clark, S., Menrath, W., Chen, M., Roda, S., Succop, P. (1999). Use of a field portable X-Ray
fluorescence analyzer to determine the concentration of lead and other metals in soil
samples. Annals of Agricultural and Environmental Medicine: AAEM, 6(1) 27-32.
Clay, G.D., Dixon, S., Evans, M.G., Rowson, J.G., Worrall, F. (2012) Carbon dioxide fluxes and
DOC concentrations of eroding blanket peat gullies. Earth Surface Processes and Landforms
37: 562-571. DOI: 10.1002/esp.3193
Clymo, R.S. (1983) Peat. In: A.J.E Gore (ed.) Ecosystems of the World Volume 4A: Mires:
Swamp, Bog, Fen and Moor. Elsevier.
Clymo, R.S., Oldfield, F., Appleby, P.G., Pearson, G.W., Ratnesar, P., Richardson, N. (1990)
The Record of Atmospheric Deposition on a Rainwater-Dependent Peatland Philosophical
Transactions of the Royal Society of London. Series B, Biological Sciences 327:331-338
Cole, B., McMorrow, J., Evans, M.G. (2014) Empirical Modelling of Vegetation Abundance
from Airborne Hyperspectral Data for Upland Peatland Restoration Monitoring. Remote
Sensing 6: 716-739. DOI: 10.3390/rs6010716
Collins, A.L., Walling, D.E. (2002) Selecting fingerprint properties for discriminating
potential suspended sediment sources in river basins. Journal of Hydrology 261: 218-244
Collins, A.L., Walling, D.E. (2004) Documenting catchment suspended sediment sources:
problems, approaches and prospects. Progress in Physical Geography 28: 159-196. DOI:
10.1191/0309133304pp409ra
Collins, A.L., Walling, D.E., (2007) Sources of fine sediment recovered from the channel bed
of lowland groundwater-fed catchments in the UK. Geomorphology, 88(1-2): 120-138.
Collins, A.L., Walling, D.E., Leeks, G.J.L. (1997) Fingerprinting the origin of fluvial suspended
sediment in larger river basins: Combining assessment of spatial provenance and source
type. Geografiska Annaler Series a-Physical Geography 79A: 239-254
187
Collins, A.L., Walling, D.E., Webb, L., King, P. (2010) Apportioning catchment scale sediment
sources using a modified composite fingerprinting technique incorporating property
weightings and prior information. Geoderma 155: 249-261. DOI:
10.1016/j.geoderma.2009.12.008
Collins, A.L., Zhang, Y., McChesney, D., Walling, D.E., Haley, S.M., Smith, P. (2012a)
Sediment source tracing in a lowland agricultural catchment in southern England using a
modified procedure combining statistical analysis and numerical modelling. Science of the
Total Environment 414: 301-317. DOI: 10.1016/j.scitotenv.2011.10.062
Collins, A.L., Zhang, Y., Walling, D.E., Grenfell, S.E., Smith, P., Grischeff, J., Locke, A.,
Sweetapple, A., Brogden, D. (2012b) Quantifying fine-grained sediment sources in the River
Axe catchment, southwest England: application of a Monte Carlo numerical modelling
framework incorporating local and genetic algorithm optimisation. Hydrological Processes
26: 1962-1983. DOI: 10.1002/hyp.8283
Cook, J.M., Gardner, M.J., Griffiths, A.H., Jesseps, M.A., Ravenscroft, J.E., Yates,R. (1997)
The Comparability of Sample Digestion Techniques for the Determination of Metals in
Sediments. Marine Pollution Bulletin 34(8): 637-644
Cressie, N. (1990) The Origins of Kriging. Mathematical Geology 22(3):239-252
Crisp, D.T. (1966) Input and output of minerals for an area of Pennine moorland –
importance of precipitation, drainage, peat erosion and animals. Journal of Applied Ecology
3(2): 327-329
Crowe, S.K., Evans, M.G., Allott, T.E.H. (2008) Geomorphological controls on the re-
vegetation of erosion gullies in blanket peat: implications for bog restoration. Mires and
Peat 3, 1 - 14.
Crowle, A. and McCormack, F. (2009) Condition of upland terrestrial habitats. In: A. Bonn,
T.E.H. Allott, K. Hubacek and J. Stewart (Editors), Drivers of Environmental Change in
Uplands. Routledge, Abingdon, Oxon.
Damman, A.W.H. (1986) Hydrology, development, and biogeochemistry of ombrogenous
peat bogs with special reference to nutrient relocation in a western Newfoundland bog.
Canadian Journal of Botany-Revue Canadienne De Botanique, 64(2): 384-394.
188
Daniels, S.M., Agnew, C.T., Allott, T.E.H., Evans, M.G. (2008) Water table variability and
runoff generation in an eroded peatland, South Pennines, UK. Journal of Hydrology 361:
214-226. DOI: 10.1016/j.jhydrol.2008.07.042
Davis, C.M., Fox, J.F. (2009) Sediment Fingerprinting: Review of the Method and Future
Improvements for Allocating Nonpoint Source Pollution. Journal of Environmental
Engineering-Asce 135: 490-504. DOI: 10.1061/(asce)0733-9372(2009)135:7(490)
Dawson, J. J. C., Tetzlaff, D., Carey, A. M., Raab, A., Soulsby, C., Killham, K., Meharg, A. A.
(2010). Characterizing Pb Mobilization from Upland Soils to Streams Using Pb-206/Pb-207
Isotopic Ratios. Environmental Science and Technology, 44(1) 243-249.
Dawson, J.J.C. and Smith, P. (2007) Carbon losses from soil and its consequences for land-
use management. Science of the Total Environment, 382(2-3): 165-190.
Dawson, J.J.C., Billett, M.F., Neal, C., Hill, S. (2002) A comparison of particulate, dissolved
and gaseous carbon in two contrasting upland streams in the UK. Journal of Hydrology 257:
226-246. DOI: 10.1016/s0022-1694(01)00545-5
De Vleeschouwer, F., Gerard, L., Goormaghtigh, C., Mattielli, N., Le Roux, G., Fagel, N.
(2007). Atmospheric lead and heavy metal pollution records from a Belgian peat bog
spanning the last two millenia: Human impact on a regional to global scale. Science of the
Total Environment, 377(2-3) 282-295.
Dearing, J.A. (1994) Environmental Magnetic Susceptibility: Using the Bartington MS2
System. Chi Publishing: Kenilworth.
Devereux, O.H., Prestegaard, K.L., Needelman, B.A., Gellis, A.C. (2010) Suspended-sediment
sources in an urban watershed, Northeast Branch Anacostia River, Maryland. Hydrological
Processes 24: 1391-1403. DOI: 10.1002/hyp.7604
Dixon, S., Qassim, S., Rowson, J., Worrall, F., Evans, M., Boothroyd, I., Bonn, A. (2013)
Restoration effects on water table depths and CO2 fluxes from climatically marginal blanket
bog. Biogeochemistry: 1-18. DOI: 10.1007/s10533-013-9915-4
Douglas, G., Palmer, M. and Caitcheon, G., 2003. The provenance of sediments in Moreton
Bay, Australia: a synthesis of major, trace element and Sr-Nd-Pb isotopic geochemistry,
modelling and landscape analysis. Hydrobiologia, 494(1-3): 145-152.
Dunlop, D. J., (1981) The rock magnetism of fine particles. Physics of the Earth and
Planetary Interiors, 26:1-26
189
Dunlop, D.J., Argyle, K.S. (1997) Thermoremanence, anhysteretic remanence and
susceptibility of submicron magnetites: Nonlinear field dependence and variation with
grain size Journal of Geophysical Research: Solid Earth 102(B9):20199-20210
EAG (2007). ICP-OES and ICP-MS Detection Limit Guidance [Online].
http://www.eaglabs.com/documents/icp-oes-ms-detection-limit-guidance-BR023.pdf.
Accessed 20 February 2013.
Ebdon, D. (1985). Statistics in Geography: A Practical Approach - Revised with 17 Programs.
Oxford: Blackwell.
Egglesman, R., Heathwaite, A.L., Gross-Braukmann, G., Kuster, E., Naucke, W., Schich, M.,
Schwikle, V. (1993) Physical processes and properties of mires. In: Heathwaite, A.L.,
Gottlich, Kh. (Eds.). Mires, Process, Exploration and Conservation
Ellis, C.J. and Tallis, J.H. (2001) Climatic control of peat erosion in a North Wales blanket
mire. New Phytologist, 152(2): 313-324.
Erel, Y., Patterson, C.C., (1994) Leakage of industrial lead into the hydrocycle. Geochimica
Et Cosmochimica Acta, 58(15): 3289-3296.
Evans, C., Allott, T.E.H., Billett, M.F., Burden, A., Chapman, P., Dinsmore, K., Evans, M.G.,
Freeman, C., Goulsbra, C., Holden, J., Jones, D., Jones, T., Moody, C., Palmer, S.M., Worrall,
F. (2013) Greenhouse gas emissions associated with non-gaseous losses of carbon - fate of
particulate and dissolved carbon. Centre for Ecology and Hydrology, Bangor.
Evans, D.J., Gibson, C.E., Rossell, R.S. (2006) Sediment loads and sources in heavily modified
Irish catchments: A move towards informed management strategies. Geomorphology 79:
93-113. DOI: 10.1016/j.geomorph.2005.09.018
Evans, M., Lindsay, J. (2010a) High resolution quantification of gully erosion in upland
peatlands at the landscape scale. Earth Surface Processes and Landforms 35: 876-886. DOI:
10.1002/esp.1918
Evans, M., Lindsay, J. (2010b) Impact of gully erosion on carbon sequestration in blanket
peatlands. Climate Research 45: 31-41. DOI: 10.3354/cr00887
Evans, M., Warburton, J. (2005) Sediment budget for an eroding peat-moorland catchment
in northern England. Earth Surface Processes and Landforms 30: 557-577. DOI:
10.1002/esp.1153
190
Evans, M.G. (2009) Natural changes in upland landscapes. In: A. Bonn, T.E.H. Allott, K.
Hubacek and J. Stewart (Editors), Drivers of Environmental Change in Uplands. Routledge,
Abingdon, Oxon, pp. 13-33.
Evans, M.G. and Warburton, J. (2007) Geomorphology of Upland Peat. Blackwell, Oxford,
262 pp.
Evans, M.G., Allott, T.E.H., Holden, J., Flitcroft, C., Bonn, A. (2005) Understanding gully
blocking in deep peat. Moors for the Future Report No. 4, Castleton, Derbyshire.
Evans, M.G., Burt, T.P., Holden, J. and Adamson, J.K. (1999) Runoff generation and water
table fluctuations in blanket peat: evidence from UK data spanning the dry summer of
1995. Journal of Hydrology, 221(3-4): 141-160.
Evans, M.G., Warburton, J. (2010) Peatland Geomorphology and Carbon Cycling.
Geography Compass 4: 1513-1531. DOI: 10.1111/j.1749-8198.2010.00378.x
Evans, M.G., Warburton, J., Yang, J. (2006) Sediment budgets for eroding blanket peat
catchments: Global and local implications of upland organic sediment budgets
Geomorphology 79(1); 45-57
Evans, R. (1997) Soil erosion in the UK initiated by grazing animals - A need for a national
survey. Applied Geography, 17(2): 127-141.
Evans, R. (1998) The erosional impacts of grazing animals. Progress in Physical Geography,
22(2): 251-268.
Evrard, O., Navratil, O., Ayrault, S., Ahmadi, M., Némery, J., Legout, C., Lefèvre, I., Poirel, A.,
Bonté, P., Esteves, M. (2011) Combining suspended sediment monitoring and fingerprinting
to determine the spatial origin of fine sediment in a mountainous river catchment. Earth
Surface Processes and Landforms 36: 1072-1089. DOI: 10.1002/esp.2133
Eyre, M.D., Pilkington, J.G., McBlane, R.P. and Rushton, S.P. (2005) Macroinvertebrate
species and assemblages in the headwater streams of the River Tyne, northern England in
relation to land cover and other environmental variables. Hydrobiologia, 544: 229-240.
Farmer, J.G., Graham, M.C., Bacon, J.R., Dunn, S.M., Vinogradoff, S.I., MacKenzie, A.B.
(2005) Isotopic characterisation of the historical lead deposition record at Glensaugh, an
organic-rich, upland catchment in rural NE Scotland. Science of the Total Environment 346:
121-137. DOI: 10.1016/j.scitotenv.2004.11.020
191
Farrell C.A., Doyle G.J. (2003) Rehabilitation of industrial cutaway Atlantic blanket bog in
County Mayo, North-West Ireland. Wetlands Ecology and Management, 11: 21-35
Fenner, N., Freeman, C. (2011) Drought-induced carbon loss in peatlands. Nature
Geoscience, 4(12): 895-900.
Ferguson, P., Lee, J.A. and Bell, J.N.B. (1978) Effects of sulfur pollutants on growth of
sphagnum species. Environmental Pollution, 16(2): 151-162.
Fernández Espinosa, A.J., Rodrıguez, M.R., Barragán de la Rosa, F.J., Jiménez Sánchez, J.C.
(2002) A chemical speciation of trace metals for fine urban particles Atmospheric
Environment 36(5); 773-780
Foster, D.R., Wright, H.E., Thelaus, M. and King, G.A. (1988) Bog development and landform
dynamics in central Sweden and south eastern Labrador, Canada. Journal of Ecology, 76(4):
1164-1185.
Foulds, S.A., Warburton, J. (2007a) Wind erosion of blanket peat during a short period of
surface desiccation (North Pennines, Northern England). Earth Surface Processes and
Landforms, 32(3): 481-488.
Foulds, S.A., Warburton, J. (2007b) Significance of wind-driven rain (wind-splash) in the
erosion of blanket peat. Geomorphology, 83(1-2): 183-192.
Fox, P.J. Edil, T.B. (1996) Effects of stress and temperature on secondary compression of
peat. Canadian Geotechnical Journal, 33: 405-415.
Francis, I.S. (1990) Blanket peat erosion in a mid-wales catchment during 2 drought years.
Earth Surface Processes and Landforms, 15(5): 445-456.
Freeman, C., Evans, C.D., Monteith, D.T., Reynolds, B., Fenner, N. (2001) Export of organic
carbon from peat soils. Nature, 412(6849): 785-785.
Freeman, G.T. (1991) Calculating catchment area with divergent flow based on a regular
grid. Computers and Geosciences, 17:413-422
Ge, L. Q., Lai, W. C., Lin, Y. C. (2005) Influence of and correction for moisture in rocks, soils
and sediments on in situ XRF analysis. X-Ray Spectrometry, 34(1) 28-34.
Global Tourism Solutions (2009) Peak District National Park and Influence Area, STEAM
Report 2009. Available at
192
http://www.peakdistrict.gov.uk/__data/assets/pdf_file/0018/321642/STEAM-2009.pdf
Last accessed 19th September 2014.
Gorham, E. (1991) Northern peatlands - Role in the carbon-cycle and probable responses to
climatic warming. Ecological Applications, 1(2): 182-195.
Gorham, E., Rochefort, L. (2003) Peatland restoration: A brief assessment with special
reference to Sphagnum bogs. Wetlands Ecology and Management 11: 109-119. DOI:
10.1023/a:1022065723511
Goulsbra, C., Evans, M., Lindsay, J. (2014) Temporary streams in a peatland catchment:
pattern, timing, and controls on stream network expansion and contraction. Earth Surface
Processes and Landforms 39(6): 790-803. DOI: 10.1002/esp.3533
Goulsbra, C.S., Lindsay, J.B., Evans, M.G., (2009) A new approach to the application of
electrical resistance sensors to measuring the onset of ephemeral streamflow in wetland
environments. Water Resources Research, 45(9): W09501.
Gruszowski, K.E., Foster, I.D.L., Lees, J.A., Charlesworth, S.M. (2003) Sediment sources and
transport pathways in a rural catchment, Herefordshire, UK. Hydrological Processes 17:
2665-2681. DOI: 10.1002/hyp.1296
Hani, A., Pazira, E. (2011) Heavy metals assessment and identification of their sources in
agricultural soils of Southern Tehran, Iran. Environmental Monitoring and Assessment
176:677-691
Hanrahan, E. T. (1954). An investigation of physical properties of peat. Geotechnique 4 (3)
108–121
Hart, B. (1982) Uptake of trace metals by sediments and suspended particulates: a review.
Hydrobiologia 91-92: 299-313. DOI: 10.1007/bf02391947
Hartig, E.K., Grozev, O. and Rosenzweig, C. (1997) Climate change, agriculture and wetlands
in Eastern Europe: Vulnerability, adaptation and policy. Climatic Change, 36(1-2): 107-121.
Hatfield R.G., Maher B.A. (2008) Suspended sediment characterization and tracing using a
magnetic fingerprinting technique: Bassenthwaite Lake, Cumbria, UK. Holocene 18: 105-
115. DOI: 10.1177/0959683607085600
193
Heathwaite, A.L., Quinn, P.F., Hewett, C.J.M. (2005) Modelling and managing critical source
areas of diffuse pollution from agricultural land using flow connectivity simulation Journal
of Hydrology 304 (1-4) 446-461 DOI: 10.1016/j.jhysrol.2004.07.043
Heiri, O., Lotter, A.F., Lemcke, G (2001) Loss on ignition as a method for estimating organic
and carbonate content in sediments: reproducibility and comparability of results. Journal of
Paleolimnology 25:101-110
Hemyari, P., Nofziger, D.L. (1987) Analytical Solution for Punctual Kriging in one Dimension.
Soil Science Society of America Journal 51(1): 268-269
Hobbs, N.B. (1986) Mire morphology and the properties and behaviour of some British and
foreign peats. Quarterly Journal of Engineering Geology and Hydrogeology (19): 7-80.
Hobbs, R.J. and Gimingham, C.H. (1987) Vegetation, fire and herbivore interactions in
heathland. Advances in Ecological Research, 16: 87-173.
Holden, J., (2005) Peatland hydrology and carbon release: why small-scale process matters.
Philosophical Transactions of the Royal Society A 363(1837): 2891-2913
Holden, J. and Burt, T.P. (2002a) Laboratory experiments on drought and runoff in blanket
peat. European Journal of Soil Science, 53(4): 675-689.
Holden, J., Burt, T.P. (2002b) Infiltration, runoff and sediment production in blanket peat
catchments: implications of field rainfall simulation experiments. Hydrological Processes
16(13); 2537-2557
Holden, J., Chapman, P.J. and Labadz, J.C. (2004) Artificial drainage of peatlands:
hydrological and hydrochemical process and wetland restoration. Progress in Physical
Geography, 28(1): 95-123.
Holden, J., Evans, M.G., Burt, T.P. and Horton, M. (2006) Impact of land drainage on
peatland hydrology. Journal of Environmental Quality, 35(5): 1764-1778.
Holden, J., Shotbolt, L., Bonn, A., Burt, T.P., Chapman, P.J., Dougill, A.J., Fraser, E.D.G.,
Hubacek, K., Irvine, B., Kirkby, M.J., Reed, M.S., Prell, C., Stagl, S., Stringer, L.C., Turner, A.,
Worrall, F. (2007) Environmental change in moorland landscapes. Earth-Science Reviews,
82(1-2): 75-100.
194
Holden, J., Smart, R.P., Dinsmore, K.J., Baird, A.J., Billett, M.F., Chapman PJ. (2012) Natural
pipes in blanket peatlands: major point sources for the release of carbon to the aquatic
system. Global Change Biology 18: 3568-3580. DOI: 10.1111/gcb.12004
Hollander, M., Wolfe, D.A. (1973) Nonparametric Statistical Methods. John Wiley & Sons:
New York
Holliday, V.J., Warburton, J. and Higgitt, D.L. (2008) Historic and contemporary sediment
transfer in an upland Pennine catchment, UK. Earth Surface Processes and Landforms,
33(14): 2139-2155.
Hong, S.M., Candelone, J.-P., Patterson, C.C. and Boutron, C.F. (1994). Greenland ice
evidence of hemispheric lead pollution 2 millennia ago by Greek and Roman civilizations.
Science, 265, 1841–43.
Hope, D., Billett, M.F., Cresser, M.S. (1994) A review of the export of carbon in river water -
fluxes and processes. Environmental Pollution 84: 301-324. DOI: 10.1016/0269-
7491(94)90142-2
Horowitz, A.J. (1995) The Use of Suspended Sediment and Associated Trace Elements in
Water Quality Studies. IAHS Special Publication No. 4. IAHS Press: Wallingford; 18–52.
Hou, X. D., He, Y. H., Jones, B. T. (2004). Recent advances in portable X-ray fluorescence
spectrometry. Applied Spectroscopy Reviews, 39(1) 1-25.
Hu, W., Huang, B., Weindorf, D.C., Chen, Y (2014) Metals analysis of agricultural soils via
portable X-ray fluorescence spectrometry Bulletin of Environmental Contamination and
Toxicology 92(4): 420-426
Hulme, M., Jenkins, G.J., Lu, X., Turnpenny, J.R., Mitchell, T.D., Jones, R.G., Lowe, J.,
Murphy, J.M., Hassell, D., Boorman, P., McDonald, R., Hill, S. (2002) Climate change
scenarios for the United Kingdom: The UKCIP02 scientific report. Norwich: Tyndall Centre
for Climate Change Research, 120pp
Hürkamp, K., Raab, T., Völkel , J. (2009a). Lead pollution of floodplain soils in a historic
mining area - age, distribution and binding forms. Water, Air and Soil Pollution 201(1-4),
331-345
Hürkamp, K., Raab, T., Völkel, J. (2009b). Two and three-dimensional quantification of lead
contamination in alluvial soils of a historic mining area using field portable X-ray
fluorescence (FPXRF) analysis. Geomorphology 110 28-36.
195
Hutchinson, S.M. (1995) Use of magnetic and radiometric measurements to investigate
erosion and sedimentation in a British upland catchment. Earth Surface Processes and
Landforms 20: 293-314
Ingram, H.A.P. (1978) Soil layers in mires - function and terminology. Journal of Soil Science,
29(2): 224-227.
Ingram, H.A.P. (1982) Size and shape in raised mire ecosystems - a geophysical model.
Nature, 297(5864): 300-303.
Ingram HAP. (1983) Hydrology. In: Ecosystems of the World 4A, Mires: swamp, bog, fen and
moor, Gore AJP. (Ed). Elsevier: Oxford; 67-158.
Janssens, I. A., A. Freibauer, B. Schlamadinger, R. Ceulemans, P. Ciais, A. J. Dolman, M.
Heimann, G. J. Nabuurs, P. Smith, R. Valentini and E. D. Schulze (2005) The carbon budget
of terrestrial ecosystems at country-scale – a European case study. Biogeosciences 2(1): 15-
26.
Johnson, B., Leetham, J., Linton, K. (1995) Effective XRF field screening of lead in soil.
Available at http://info.ngwa.org/gwol/pdf/950161762.PDF. Last accessed 20th September
2014.
Jones, J. M., Hao, J. (1993). Ombrotrophic peat as a medium for historical monitoring of
heavy-metal pollution. Environmental Geochemistry and Health 15(2-3), 67-74.
Joosten H. (2009) The global peatland CO2 picture: Peatland status and drainage related
emissions in all countries of the world. Wetlands International, Ede
Kalnicky, D. J., Singhvi, R. (2001) Field portable XRF analysis of environmental samples.
Journal of Hazardous Materials, 83(1-2), 93-122.
Kempter, H., Frenzel, B. (2000). The impact of early mining and smelting on the local
tropospheric aerosol detected in ombrotrophic peat bogs in the Harz, Germany. Water Air
and Soil Pollution 121(1-4), 93-108.
Kilbride, C., Poole, J., Hutchings, T. R. (2006). A comparison of Cu, Pb, As, Cd, Zn, Fe, Ni and
Mn determined by acid extraction/ICP-OES and ex situ field portable X-ray fluorescence
analyses. Environmental Pollution 143(1) 16-23.
Klove, B. (1998) Erosion and sediment delivery from peat mines. Soil & Tillage Research 45:
199-216
196
Koiter, A.J., Owens, P.N., Petticrew, E.L., Lobb, D.A. (2013). "The behavioural characteristics
of sediment properties and their implications for sediment fingerprinting as an approach
for identifying sediment sources in river basins." Earth-Science Reviews 125(0): 24-42.
Komarek, M., Chrastny, V., Ettler, V., Tlustos, P. (2006) Evaluation of extraction/digestion
techniques used to determine lead isotopic composition in forest soils. Analytical and
Bioanalytical Chemistry 385: 1109-1115
Krs, M., Novak, F., Krsova, M., Pruner, P., Kouklíková, L. and Jansa, J., 1992. Magnetic
properties and metastability of greigite-smythite mineralization in brown-coal basins of the
Krusné hory Piedmont, Bohemia. Physics of the Earth and Planetary Interiors, 70, 273-287
Labadz, J.C., Burt, T.P. and Potter, A.W.R. (1991) Sediment yield and delivery in the blanket
peat moorlands of the southern Pennines. Earth Surface Processes and Landforms, 16(3):
255-271.
Lana-Renault, N., Alvera, B., Garcia-Ruiz, J.M., (2011) Runoff and Sediment Transport during
the Snowmelt Period in a Mediterranean High-Mountain Catchment. Arctic Antarctic and
Alpine Research, 43(2): 213-222.
Le Roux, G., De Vleeschouwer, F. (2010) Preparation of peat samples for inorganic
geochemistry used as palaeoenvironmental proxies. Mires and Peat 7 (Article 04): 1-9
Lee, J. A., Tallis, J. H. (1973) Regional and historical aspects of lead pollution in Britain.
Nature 245(5422) 216-218.
Lee, J.A. (1998) Unintentional experiments with terrestrial ecosystems: ecological effects of
sulphur and nitrogen pollutants. Journal of Ecology, 86(1): 1-12.
Leenaers, H., Okx, J. P. & Burrough, P. A. (1990). Employing elevation data for efficient
mapping of soil pollution on flood plains. Soil Use & Management., 6, 105-114.
Lin, Z.X., Herbert, R.B., (1997) Heavy metal retention in secondary precipitates from a mine
rock dump and underlying soil, Dalarna, Sweden. Environmental Geology, 33(1): 1-12.
Lindsay, J.B. (2005) The Terrain Analysis System: a tool for hydrogeomorphic applications.
Hydrological Processes 19: 1123–1130. DOI: 10.1002/hyp.5818
Lindsay, R. (1995) Bogs: The ecology, classification and conservation of ombotrophic mires.
Scottish National Heritage, Edinburgh, 119 pp.
197
Liu, Q., A. P. Roberts, J. C. Larrasoaña, S. K. Banerjee, Y. Guyodo, L. Tauxe, and F. Oldfield
(2012), Environmental magnetism: Principles and applications. Reviews of Geophysics 50:
RG4002
Livett, E. A., Lee, J. A., Tallis, J. H. (1979) Lead, zinc and copper analyses of British blanket
peats. Journal of Ecology, 67(3), 865-891.
Lloyd, J., Taylor, J.A. (1994) On the temperature-dependence of soil respiration. Functional
Ecology, 8(3): 315-323.
Lowemark, L., Chen, H. F., Yang, T. N., Kylander, M., Yu, E. F., Hsu, Y. W., Lee, T. Q., Song, S.
R., Jarvis, S. (2011) Normalizing XRF-scanner data: A cautionary note on the interpretation
of high-resolution records from organic-rich lakes. Journal of Asian Earth Sciences, 40(6)
1250-1256.
Lucassen, E.C.H.E.T., Smolders, A.J.P., Roelofs, J.G.M. (2002) Potential sensitivity of mires to
drought, acidification and mobilisation of heavy metals: the sediment S/(Ca+Mg) ratio as a
diagnostic tool. Environmental Pollution 120: 635–646.
Luoto, M., Seppala, M., (2000) Summit peats ('peat cakes') on the fells of Finnish Lapland:
continental fragments of blanket mires? Holocene, 10(2): 229-241.
Luukkainen, V. M. (1992) The flowing properties of peat slurry. In Proceedings of the 9th
International Peat Congress, Uppsala, Sweden, 1992. Special Edition of the International
Peat Journal 2(3): 137–45.
Mackay, A.W., Tallis, J.H. (1996) Summit-type blanket mire erosion in the forest of
Bowland, Lancashire, UK: Predisposing factors and implications for conservation. Biological
Conservation, 76(1): 31-44.
MacKenzie, A. B., Logan, E. M., Cook, G. T., Pulford, I. D. (1998). Distributions, inventories
and isotopic composition of lead in Pb-210-dated peat cores from contrasting
biogeochemical environments: Implications for lead mobility. Science of the Total
Environment 223(1) 25-35.
Maher, B.A., (1988) Magnetic properties of some synthetic sub-micron magnetites. Journal
of Geophysical Research 94: 83-96.
Makinen, E., Korhonen, M., Viskari, E. L., Haapamaki, S., Jarvinen, M., Lu, L. (2006).
Comparison of XRF and FAAS methods in analysing CCA contaminated soils. Water Air and
Soil Pollution 171(1-4), 95-110.
198
Markert, B., Thornton, I. (1990). Multielement analysis of an English peat bog soil. Water
Air and Soil Pollution, 49(1-2) 113-123.
Martin Peinado, F., Morales Ruano, S., Bagur Gonzalez, M. G., Estepa Molina, C. (2010) A
rapid field procedure for screening trace elements in polluted soil using portable X-ray
fluorescence (PXRF). Geoderma 159(1-2), 76-82.
Martínez-Carreras, N., Udelhoven, T., Krein, A., Gallart, F., Iffly, J.F., Ziebel, J., Hoffmann, L.,
Pfister, L., Walling, D.E. (2010) The use of sediment colour measured by diffuse reflectance
spectrometry to determine sediment sources: application to the Attert River catchment
(Luxembourg). Journal of Hydrology 382, 49–63.
Marx, S. K., Kamber, B. S., McGowan, H. A., Zawadzki, A. (2010) Atmospheric pollutants in
alpine peat bogs record a detailed chronology of industrial and agricultural development on
the Australian continent. Environmental Pollution 158(5) 1615-1628.
Mauquoy, D., Engelkes, T., Groot, M.H.M., Markesteijn, F., Oudejans, M.G., van der Plicht,
J., van Geel, B (2002) High-resolution records of late-Holocene climate change and carbon
accumulation in two north-west European ombrotrophic peat bogs. Palaeogeography
Palaeoclimatology Palaeoecology 186(3-4): 275-310.
McDonald, D. M., Lamoureux, S. F. & Warburton, J. (2010). Assessment of a time-integrated
fluvial suspended sediment sampler in a high arctic setting. Geografiska Annaler. Series A,
Physical Geography, 92, 225 - 235
McDowell, R.W., Wilcock, R.J. (2004) Particulate Phosphorus Transport within Stream Flow
of an Agricultural Catchment Journal Environmental Quality 33(6): 2111-2121
McDowell, R.W., Wilcock, R.J., (2007) Sources of sediment and phosphorus in stream flow
of a highly productive dairy farmed catchment. Journal Environmental Quality 36(2), 540-
548
McMorrow, J., Lindley, S., Aylen, J., Cavan, G., Albertson, K., Boys, D. (2009) Moorland
wildfire risk, visitors and climate change: patterns, prevention and policy. In: A. Bonn, T.E.H.
Allott, K. Hubacek and J. Stewart (Editors), Drivers of Environmental Change in Uplands.
Routledge, Abingdon, Oxon, pp. 544.
Mighall, T.M., Grattan, J.P., Timberlake, S., Lees, J.A., Forsyth, S. (2002) An atmospheric
pollution history for lead-zinc mining from the Ystwyth Valley, Dyfed, mid-Wales, UK as
199
recorded by an upland blanket peat. Geochemistry: Exploration, Environment, Analysis 2,
175-184.
Mighall, T.M., Timberlake, S., Foster, I.D.L., Krupp, E. and Singh, S. (2009) Ancient copper
and lead pollution records from a raised bog complex in Central Wales, UK. Journal of
Archaeological Science, 36(7): 1504-1515.
Mihaljevic, M., Zuna, M., Ettler, V., Sebek, O., Strnad, L., Golias, V. (2006). Lead fluxes,
isotopic and concentration profiles in a peat deposit near a lead smelter (Pribram, Czech
Republic). Science of the Total Environment, 372(1), 334-344.
Milne, R., T. A. Brown (1997) Carbon in the Vegetation and Soils of Great Britain. Journal of
Environmental Management 49(4): 413-433.
Molina, A., Govers, G., Cisneros, F., Vanacker, V. (2009) Vegetation and topographic
controls on sediment deposition and storage on gully beds in a degraded mountain area.
Earth Surface Processes and Landforms, 34(6): 755-767.
Monna, F., Galop, D., Carozza, L,. Tual, M., Beyrie, A., Marembert, F., Chateau, C., Dominik,
J., Grousset, F. (2004). Environmental impact of early Basque mining and smelting recorded
in a high ash minerogenic peat deposit. Science of the Total Environment, 327(1-3) 197-214.
Moody, C.S., Worrall, F., Evans, C.D., Jones, T. (2013) The rate of loss of dissolved organic
carbon (DOC) through a catchment. Journal of Hydrology 492: 139-150
Mooney, S.J., Holden, N.M., Ward, S.M., Collins, J.F. (2000) The use of quantified structural
descriptors to physically characterise stockpiled milled peat of different genetic origins.
Geoderma 95 15–32
Moore, P.D. (2002) The future of cool temperate bogs. Environmental Conservation, 29(1):
3-20.
Motha, J.A., Wallbrink, P.J., Hairsine, P.B., Grayson, R.B. (2003). Determining the sources of
suspended sediment in a forested catchment in southeastern Australia. Water Resources
Research 39. DOI: 10.1029/2001wr000794
Murozumi, M., Chow, T.J., Patterson, C.C. (1969). Chemical concentrations of pollutant lead
aerosols, terrestrial dusts and sea salts in Greenland and Antarctic snow strata. Geochimica
et Cosmochimica Acta, 33, 1247–94.
200
Natural England (2009) Vital Uplands - Lake District pilot Upland Vision. London: Natural
England
Norton, S.A., Evans, G.C., Kahl, J.S. (1997) Comparison of Hg and Pb fluxes to hummocks
and hollows of ombrotrophic Big Heath Bog and to Nearby Sargent Mt. Pond, Maine, USA.
Water, Air and Soil Pollution 100, 271-288.
Novak, M., Emmanuel, S., Vile, M. A., Erel, Y., Veron, A., Paces, T., Wieder, R. K., Vanecek,
M., Stepanova, M., Brizova, E., Hovorka, J. (2003). Origin of lead in eight central European
peat bogs determined from isotope ratios, strengths, and operation times of regional
pollution sources. Environmental Science and Technology, 37(3), 437-445.
Novak, M., Zemanova, L., Voldrichova, P., Stepanova, M., Adamova, M., Pacherova, P.,
Komarek, A., Krachler, M., Prechova, E. (2011). Experimental Evidence for
Mobility/Immobility of Metals in Peat. Environmental Science and Technology, 45(17),
7180-7187. DOI: 10.1021/es201086v
Oldfield, F (1991) Environmental magnetism – a personal perspective. Quaternary Science
Reviews 10: 73-85
Oldfield, F (2014) Can the magnetic signatures from inorganic fly ash be used to mark the
onset of the Anthropocene? The Anthropocene Review 1-11, DOI:
10.1177/2053019614534402
Oldfield, F. (1990) Magnetic measurements of recent sediments from Big Moose lake,
Adirondack Mountains, NY, USA. Journal of Paleolimnology 4(2): 93-101
Oldfield, F., Brown, A., Thompson, R. (1979) The effect of microtopography and vegetation
on the catchment of airborne particles measured by remanent magnetism. Quaternary
Research 12(3): 326-332
Oldfield, F., Maher, B.A., Donoghue, J., Pierce, J. (1985). Particle-size related, mineral
magnetic source sediment linkages in the Rhode River catchment, Maryland, USA. Journal
of the Geological Society 142: 1035-1046
Oldfield, F., Maher, B.A., Donoghue, J., Pierce, J. (1985). Particle-size related, mineral
magnetic source sediment linkages in the Rhode River catchment, Maryland, USA. Journal
of the Geological Society 142: 1035-1046
201
Oldfield, F., Thompson, R., Barber, K.E. (1978). Changing Atmospheric Fallout of Magnetic
Particles Recorded in Recent Ombrotrophic Peat Sections. Science 199: 679-680. DOI:
10.1126/science.199.4329.679-a
Olejnik, S., Algina, J. (2003) Generalized eta and omega squared statistics: Measures of
effect size for some common research designs. Psychological Methods, 8(4): 434-447.
Owens, P.N. and Walling, D.E. (2002) Changes in sediment sources and floodplain
deposition rates in the catchment of the River Tweed, Scotland, over the last 100 years:
The impact of climate and land use change. Earth Surface Processes and Landforms, 27(4):
403-423.
Owens, P.N. and Walling, D.E., (2003) Temporal changes in the metal and phosphorus
content of suspended sediment transported by Yorkshire rivers, UK over the last 100 years,
as recorded by overbank floodplain deposits. Hydrobiologia, 494(1-3): 185-191.
Owens, P.N., Blake, W.H., Giles, T.R., Williams, N.D. (2012) Determining the effects of
wildfire on sediment sources using 137Cs and unsupported 210Pb: the role of landscape
disturbances and driving forces. Journal of Soils and Sediments 12: 982–994. DOI
10.1007/s11368-012-0497-x
Owens, P.N., Blake, W.H., Petticrew, E.L. (2006) Changes in sediment sources following
wildfire in mountainous terrain: a paired–catchment approach, British Columbia, Canada.
Water, Air, and Soil Pollution: Focus 6: 637–645
Päivänen, J., (1997) Forested mires as a renewable resource: toward a sustainable forestry
practice. In: Northern Forested Wetlands: Ecology and Management [Trettin, C.C., M.F.
Jurgensen, D.F. Grigal, M.R. Gale, and J.K. Jeglum (eds.)]. CRC Lewis Publishers, Boca Raton,
FL and New York, NY, USA; London, United Kingdom; and Tokyo, Japan, pp. 27-44.
Pan, K., Wang, W-X., (2012) Trace metal contamination in estuarine and coastal
environments in China Science of the Total Environment 412-422; 3-16
Parry, L.E., Holden, J., Chapman, P.J. (2014) Restoration of blanket peatlands. Journal of
Environmental Management 133; 193-205
Parry, M.L., Canziani, O.F., Palutikof, J.P., van der Linden, P.J., Hanson, C.E. (2007) Climate
Change 2007: Impacts, Adaptation and Vulnerability. Contribution of Working Group II to
the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. In Fourth
Assessment Report of the Intergovernmental Panel on Climate Change, Parry, M.L.,
202
Canziani, O.F., Palutikof, J.P., van der Linden, P.J., Hanson, C.E. (eds). Cambridge University
Press: Cambridge, UK.
Pawson, R.R. (2008) Assessing the role of particulates in the fluvial organic carbon flux from
eroding peatland systems. In School of Environment and Development. University of
Manchester, Manchester. Unpublished.
Pawson, R.R., Evans, M.G., Allott, T. (2012) Fluvial carbon flux from headwater peatland
streams: significance of particulate carbon flux. Earth Surface Processes and Landforms 37:
1203-1212. DOI: 10.1002/esp.3257
Pawson, R.R., Lord, D.R., Evans, M.G., Allott, T.E.H. (2008) Fluvial organic carbon flux from
an eroding peatland catchment, southern Pennines, UK. Hydrology and Earth System
Sciences 12: 625-634
Peak District National Park (2013) Peak District National Park State of the Park. Available at
http://www.peakdistrict.gov.uk/microsites/sopr/welcoming. Last accessed 20th September
2014.
Pearce-Higgins, J.W. and Yalden, D.W. (1997) The effect of resurfacing the Pennine Way on
recreational use of blanket bog in the Peak District National Park, England. Biological
Conservation, 82(3): 337-343.
Perkin Elmer (2011). Atomic Spectroscopy: A Guide to Selecting the Appropriate Technique
and System [Online].
http://www.perkinelmer.co.uk/PDFs/Downloads/BRO_WorldLeaderAAICPMSICPMS.pdf.
Accessed 20 February 2013.
Perks, M.T., Warburton, J., Bracken, L. (2013) Critical assessment and validation of a time-
integrating fluvial suspended sediment sampler. Hydrological Processes. DOI:
10.1002/hyp.9985
Petticrew, E.L., Ownes, P.N., Giles, T.R. (2006) Wildfire effects on the quantity and
composition of suspended and gravel-stored sediments. Water, Air and Soil Pollution: Focus
6: 647-656
Phillips, J.M., Russell, M.A., Walling, D.E. (2000) Time-integrated sampling of fluvial
suspended sediment: a simple methodology for small catchments. Hydrological Processes
14, 2589–2602.
203
Pringle, C., (2003) What is hydrologic connectivity and why is it ecologically important?
Hydrological Processes, 17(13): 2685-2689.
Quinn PF, Beven KJ, Lamb R, (1995) The ln(a/tanb) index: how to calculate it and how to
use it within the TOPMODEL framework. Hydrological Processes 9: 161–182.
Quinn, P.F., Beven, K.J., Chevallier, P., Planchon, O. (1991) The prediction of hillslope flow
paths for distributed hydrological modelling using digital terrain models Hydrological
Processes, 5:59-79
Raab, T,. Hürkamp, K., Völkel, J. (2005). Detection and Quantification of Heavy Metal
Contamination in Alluvial Soils of Historic Mining Areas by Field Portable X-ray Fluorescence
(FPXRF) Analysis. In: Proceedings of International Conference on Problematic Soils 25-27
May 2005. Eastern Mediterranean University, Famagusta, N. Cyprus.
Radley, J. (1962) Peat erosion on the high moors of Derbyshire and West Yorkshire. East
Midlands Geographer 3(17): 40-50
Radley, J. (1965) Significance of major moorland fires. Nature 205: 1254-1259
Radu, T., Diamond, D. (2009). Comparison of soil pollution concentrations determined
using AAS and portable XRF techniques. Journal of Hazardous Materials 171(1-3) 1168-
1171.
Ramchunder, S.J., Brown, L.E. and Holden, J. (2009) Environmental effects of drainage,
drain-blocking and prescribed vegetation burning in UK upland peatlands. Progress in
Physical Geography, 33(1): 49-79.
Reid, E. and Thomas, M.F. (2006) A chronostratigraphy of mid- and late-Holocene slope
evolution: Creagan a' Chaorainn, Northern Highlands, Scotland. Holocene, 16(3): 429-444.
Rember, R.D., Trefry, J.H., (2004) Increased concentrations of dissolved trace metals and
organic carbon during snowmelt in rivers of the Alaskan Arctic. Geochimica Et
Cosmochimica Acta, 68(3): 477-489.
Renberg, I., Bindler, R. and Brännvall, M.-L. (2001). Using the historical atmospheric lead-
deposition record as a chronological marker in sediment deposits in Europe. The Holocene,
11 (5), 511–516.
Renberg, I., Wik-Persson, M. and Emteryd, O. (1994). Pre-industrial atmospheric lead
contamination detected in Swedish lake sediments. Nature, 368, 323–26.
204
Rhind, S.M. (2009) Anthropogenic pollutants: a threat to ecosystem sustainability?
Philosophical Transactions of the Royal Society B: Biological Sciences 364: 3391-3401. DOI:
10.1098/rstb.2009.0122
Ridings, M., Shorter, A. J., Smith, J. B. (2000) Strategies for the investigation of
contaminated sites using field portable X-ray fluorescence (FPXRF) techniques.
Communications in Soil Science and Plant Analysis, 31(11-14) 1785-1790.
Rose, N.L., Yang, H., Turner, S.D., Simpson, G.L. (2012) An assessment of the mechanisms
for the transfer of lead and mercury from atmospherically contaminated organic soils to
lake sediments with particular reference to Scotland, UK. Geochimica Et Cosmochimica
Acta 82: 113-135
Rothwell, J.J., Robinson, S.G., Evans, M.G., Yang, J., Allott, T.E.H. (2005) Heavy metal release
by peat erosion in the Peak District, southern Pennines, UK. Hydrological Processes 19:
2973-2989. DOI: 10.1002/hyp.5811
Rothwell, J.J. (2006) Fluvial export of heavy metals from contaminated and eroding
peatland catchments, southern Pennines, UK. The University of Manchester. PhD thesis
Rothwell, J.J., Evans, M.G. and Allott, T.E.H. (2006) Sediment-water interactions in an
eroded and heavy metal contaminated peatland catchment, Southern Pennines, UK. Water
Air and Soil Pollution: Focus, 6: 669-676.
Rothwell, J.J., Evans, M.G., Lindsay, J.B., Allott, T.E.H. (2007a) Scale-dependent spatial
variability in peatland lead pollution in the southern Pennines, UK. Environmental Pollution
145: 111-120. DOI: 10.1016 /j.envpol.2006.03.037
Rothwell, J.J., Evans, M.G., Allott, T.E.H. (2007b) Lead contamination of fluvial sediments in
an eroding blanket peat catchment. Applied Geochemistry 22: 446-459. DOI:
10.1016/j.apgeochem.2006.11.002
Rothwell, J.J., Evans, M.G., Daniels, S.M., Allott, T.E.H. (2007c) Baseflow and stormflow
metal concentrations in streams draining contaminated peat moorlands in the Peak District
National Park. Journal of Hydrology 341: 90-104
Rothwell, J.J., Evans, M.G., Liddaman, L.C., Allott, T.E.H. (2007d) The role of wildfire and
gully erosion in particulate lead export from contaminated peatland catchments in the
southern Pennines. Geomorphology 88: 276-284
205
Rothwell, J.J., Lindsay, J.B. (2007) Mapping contemporary magnetic mineral concentrations
in peat soils using fine-resolution digital terrain data Catena 70: 465-474
Rothwell, J.J., Evans, M.G., Daniels, S.A., Allott, T.E.H. (2008a) Peat soils as a source of lead
contamination to upland fluvial systems. Environmental Pollution 153: 582-589. DOI:
10.1016/j.envpol.2007.09.009
Rothwell, J.J., Evans, M.G., Allott, T.E.H. (2008b) In-stream processing of sediment-
associated metals in peatland fluvial systems. Water Air and Soil Pollution 187: 53-64. DOI:
10.1007/s11270-007-9496-8
Rothwell, J. J., Taylor, K. G., Chenery, S. R. N., Cundy, A. B., Evans, M. G., Allottt, T. E. H.
(2010a). Storage and Behavior of As, Sb, Pb, and Cu in Ombrotrophic Peat Bogs under
Contrasting Water Table Conditions. Environmental Science and Technology, 44(22), 8497-
8502.
Rothwell, J.J., Lindsay, J.B., Evans, M.G., Allott, T.E.H. (2010b) Modelling suspended
sediment lead concentrations in contaminated peatland catchments using digital terrain
analysis. Ecological Engineering 36: 623-630. DOI: 10.1016/j.ecoleng.2008.10.010
Running, S.W. (2006) Is global warming causing more, larger wildfires? Science, 313(5789):
927-928.
Russell, M.A., Walling, D.E., Hodgkinson, R.A. (2000) Appraisal of a simple sampling device
for collecting time-integrated fluvial suspended sediment samples. International Association
of Hydrological Sciences: Wallingford
Rutherford A. (2001) Introducing ANOVA and ANCOVA: a GLM Approach. Sage, London
Sansom, A.L. (1999) Upland vegetation management: The impacts of overstocking. Water
Science and Technology, 39(12): 85-92.
Shand, C.A., Wendler, R. (2014) Portable X-ray fluorescence analysis of mineral and organic
soils and the influence of organic matter. Journal of Geochemical Exploration 143; 31-42
Shefsky, S. (1997). Comparing field portable x-ray fluorescence (XRF) to laboratory analysis
of heavy metals in soil [Online]. http://www.clu-
in.org/download/char/dataquality/sshefsky02.pdf. Accessed 14 January 2013.
206
Shi, J. C., Wang, H. Z., Xu, J. M., Wu, J. J., Liu, X. M., Zhu, H. P., et al. (2007). Spatial
distribution of heavy metals in soils: A case study of Changxing, China. Environmental
Geology, 52, 1–10
Shotbolt, L., Hutchinson, S., Thomas, A. (2006) Sediment stratigraphy and heavy metal
fluxes to reservoirs in the Southern Pennine Uplands, UK. Journal of Paleolimnology 35:
305-322. DOI: 10.1007/s10933-005-1594-2
Shotyk, W,. Blaser, P., Grunig, A., Cheburkin, A. K. (2000). A new approach for quantifying
cumulative, anthropogenic, atmospheric lead deposition using peal cores from bogs: Pb in
eight Swiss peat bog profiles. Science of the Total Environment 249(1-3) 281-295.
Shotyk, W., Goodsite, M. E., Roos-Barraclough, F., Frei, R. Heinemeier, J., Asmund, G.,
Lohse, C., Hansen, T. S. (2003). Anthropogenic contributions to atmospheric Hg, Pb and As
accumulation recorded by peat cores from southern Greenland and Denmark dated using
the 14C "bomb pulse curve". Geochimica Et Cosmochimica Acta, 67(21), 3991-4011.
Shotyk, W., Norton, S.A., Farmer, J.G. (1997) Summary of the workshop on peat bog
archives of atmospheric metal deposition. Water, Air, and Soil Pollution 100: 213-219
Shotyk, W., Weiss, D., Appleby, P. G., Cheburkin, A. K., Frei, R., Gloor, M., Kramers, J. D.,
Reese, S., Van der Knaap, W. O. (1998) History of atmospheric lead deposition since 12,370
C-14 yr BP from a peat bog, Jura Mountains, Switzerland. Science 281(5383) 1635-1640.
Shuttleworth, E. L., Evans, M. G., Rothwell, J. J., Hutchinson, S. M. (2012). Impacts of
erosion and restoration on POC flux and pollutant mobilisation in the peatlands of the Peak
District National Park, UK. EGU General Assembly 2012. Vienna, Austria.
Shuttleworth, E.L., Evans, M.G., Hutchinson, S.M., Rothwell, J.J. (2014a) Assessment of Lead
Contamination in Peatlands Using Field Portable XRF. Water, Air, and Soil Pollution 225: 1-
13. DOI: 10.1007/s11270-013-1844-2
Shuttleworth, E.L., Evans, M.G., Hutchinson, S.M., Rothwell, J.J. (2014b) Peatland
restoration: controls on sediment production and reductions in carbon and pollutant
export. Earth Surface Processes and Landforms DOI: 10.1002/esp.3645
Smith, E. J., Hughes, S., Lawlor, A. J., Lofts, S., Simon, B. M., Stevens, P. A., Stidson, R. T.,
Tipping, E., Vincent, C. D. (2005). Potentially toxic metals in ombrotrophic peat along a 400
km English-Scottish transect. Environmental Pollution 136(1) 11-18.
207
Smith, H.G., Blake, W.H. (2014) Sediment fingerprinting in agricultural catchments: A
critical re-examination of source discrimination and data corrections. Geomorphology 204:
177-191. DOI: http://dx.doi.org/10.1016/j.geomorph.2013.08.003
Smith, H.G., Blake, W.H., Owens, P.N. (2013) Discriminating fine sediment sources and the
application of sediment tracers in burned catchments: a review. Hydrological Processes 27:
943-958. DOI: 10.1002/hyp.9537
Smith, J. (1999) An introduction to the magnetic properties of natural materials. In: Walden
J, Oldfield F, Smith JP (eds). Environmental magnetism: a practical guide. Technical Guide,
No 6. Quaternary Research Association, London.
Smith, T.B., Owens, P. N. (2014) Flume- and field-based evaluation of a time-integrated
suspended sediment sampler for the analysis of sediment properties. Earth Surface
Processes and Landforms. 39(9):1197-1207 DOI: 10.1002/esp.3528
Snowball, I.F. & Torii, M. (1999) Incidence and significance of ferrimagnetic iron sulphides
in Quaternary studies. In: B.A. Maher and R. Thompson (eds.) Quaternary Climates and
Magnetism, Cambridge University Press, pp 199-230.
Solderblom, R. (1974) Organic matter in Swedish clays and its importance for quick clay
formation. Proceedings of the Swedish Geotechnical Institute 26
Solo-Gabriele, H. M., Townsend, T. G., Hahn, D. W., Moskal, T. M., Hosein, N., Jambeck, J.,
Jacobi, G. (2004) Evaluation of XRF and LIBS technologies for on-line sorting of CCA-treated
wood waste. Waste Management 24(4), 413-424.
Sotherton, N., May, R., Ewald, J., Fletcher, K., Newborn, D. (2009) Managing uplands for
game and sporting interest: an industry perspective, In: Bonn, A., Allott, T.E.H., Hubacek, K.,
Stewart, J. (2009) Drivers of Environmental Change in Uplands. Routledge: Abingdon, Oxon
Sterling, D. A., Lewis, R. D., Luke, D. A., Shadel, B. N. (2000). A portable x-ray fluorescence
instrument for analyzing dust wipe samples for lead: Evaluation with field samples.
Environmental Research, 83(2) 174-179.
Stevenson, A.C., Jones, V.J. and Battarbee, R.W. (1990) The cause of peat erosion - a
paleolimnological approach. New Phytologist, 114(4): 727-735.
Stevenson, F. J. (1976) Stability-constants of Cu2+, Pb2+, and Cd2+ complexes with humic
acids. Soil Science Society of America Journal, 40(5), 665-672.
208
Stewart, A.J.A. and Lance, A.N., (1991) Effects of moor-draining on the hydrology and
vegetation of northern Pennine blanket bog. Journal of Applied Ecology, 28(3): 1105-1117.
Sturgeon, R. E. (2000) Current practice and recent developments in analytical methodology
for trace element analysis of soils, plants, and water. Communications in Soil Science and
Plant Analysis, 31(11-14) 1479-1512.
Tallis, J.H. (1997) Peat erosion in the Pennines: The badlands of Britain. Biologist, 44(1):
277-279.
Tallis, J.H. (1985) Mass movement and erosion of a southern Pennine blanket peat. Journal
of Ecology, 73(1) 283-315.
Tallis, J.H., Yalden, D.W. (1983) Peak District moorland restoration project phase II report:
Re-vegetation trials. Bakewell: Peak Park Joint Planning board.
Tallis, J.H. (1964) Pre-peat vegetation of Southern Pennines. New Phytologist 63: 363-373
Tallis, J.H. (1973) Studies on Southern Pennine peats .5. Direct observations on peat erosion
and peat hydrology at Featherbed Moss, Derbyshire. Journal of Ecology, 61(1): 1-22.
Tallis, J.H. (1987) Fire and flood at Holme Moss - Erosion processes in an upland blanket
mire. Journal of Ecology, 75(4): 1099-1129.
Tallis, J.H. (1995) Climate and erosion signals in British blanket peats: The significance of
Racomitrium lanuginosum remains. Journal of Ecology 83(6): 1021-1030
Tallis, J.H. (1997) Peat erosion in the Pennines: The badlands of Britain. Biologist 44: 277-
279
Tallis, J.H. (1998) Growth and degradation of British and Irish blanket mires. Environmental
Reviews, 6(2): 81-122.
Tallis, J.H., Livett, E.A. (1994) Pool-and-hummock patterning in a southern pennine blanket
mire .1. Stratigraphic profiles for the last 2800 years. Journal of Ecology, 82(4): 775-788.
Tansley, A. (1939) The British Isles and their vegetation. Cambridge University Press,
Cambridge.
Teutsch, N., Erel, Y., Halicz, L., Banin, A. (2001). Distribution of natural and anthropogenic
lead in Mediterranean soils. Geochimica Et Cosmochimica Acta, 65(17) 2853-2864.
209
Thompson, D.B.A., Macdonald, A.J., Marsden, J.H. and Galbraith, C.A. (1995) Upland
heather moorland in Great-Britain - A review of international importance, vegetation
change and some objectives for nature conservation. Biological Conservation, 71(2): 163-
178.
Thompson, R., Oldfield, F. (1986). Environmental Magnetism. London: Allen and Unwin
Tipping, E., Rothwell, J.J., Shotbolt, L., Lawlor, A.J. (2010) Dynamic modelling of
atmospherically-deposited Ni, Cu, Zn, Cd and Pb in Pennine catchments (northern England).
Environmental Pollution 158: 1521-1529
Tipping, E., Smith, E. J., Lawlor, A. J., Hughes, S., Stevens, P. A. (2003) Predicting the release
of metals from ombrotrophic peat due to drought-induced acidification. Environmental
Pollution 123(2) 239-253.
Tolonen, K., Oldfield, F., (1986) The record of magnetic-mineral and heavy metal deposition
at Regent Street Bog, Fredricton, New Brunswick, Canada. Physics of the Earth and
Planetary Interiors. 42, 57–66.
Turetsky, M.R., Harden, J.W., Friedli, H.R., Flannigan, M., Payne, N., Crock, J., Radke, L.
(2006) Wildfires threaten mercury stocks in northern soils. Geophysical Research Letters,
33(16): 6.
Tyler, G., Pahlsson, A. M. B., Bengtsson, G., Baath, E., Tranvik, L. (1989). Heavy-metal
ecology of terrestrial plants, microorganisms and invertebrates - a review. Water Air and
Soil Pollution, 47(3-4) 189-215.
Ure, A.M., Davidson, C.M. (2002) Chemical speciation in the environment, Second edition.
Oxford: Blackwell Science
USEPA (1998). Environmental Technology Verification Report. Field Portable X-ray
Fluorescence Analyzer. Metorex X-MET 920-P and 940, EPA/600/R-97/146.: United States
Environmental Protection Agency.
USEPA (2008). Basic XRF Concepts. Advanced Design Application, Data Analysis for Field-
Portable XRF. United States Environmental Protection Agency.
Van Asselen, S., & Roosendaal, C. (2009). A new method for determining the bulk density of
uncompacted peat from field settings. Journal of Sedimentary Research, 79, 918-922
210
Van Cott, R. J., McDonald, B. J., Seelos, A. G. (1999). Standard soil sample preparation error
and comparison of portable XRF to laboratory AA analytical results. Nuclear Instruments
and Methods in Physics Research Section A - Accelerators Spectrometers Detectors and
Associated Equipment, 422(1-3), 801-804.
Vile, M. A., Wieder, R. K., Novak, M. (1999) Mobility of Pb in Sphagnum-derived peat.
Biogeochemistry, 45(1), 35-52.
Vile, M. A., Wieder, R. K., Novak, M. (2000) 200 years of Pb deposition throughout the
Czech Republic: Patterns and sources. Environmental Science and Technology, 34(1) 12-21
Vinogradoff, S. I., Graham, M. C., Thornton, G. J. P., Dunn, S. M., Bacon, J. R., Farmer, J. G.
(2005) Investigation of the concentration and isotopic composition of inputs and outputs of
Pb in waters at an upland catchment in NE Scotland. Journal of Environmental Monitoring
7(5): 431-444.
Voutsa, D., Samara, C. (2002) Labile and bioaccessible fractions of heavy metals in the
airborne particulate matter from urban and industrial areas Atmospheric Environment
36(22); 3583-3590
Waddington, J. M., McNeil, P. (2002) Peat oxidation in an abandoned cutover
peatland. Canadian Journal of Soil Science 82:279–286.
Waddington, J.M., Griffis, T.J., Rouse, W.R. (1998) Northern Canadian wetlands: Net
ecosystem CO2 exchange and climatic change. Climatic Change, 40(2): 267-275.
Walden, J., Oldfield, F., Smith, J.P. (1999) Environmental Magnetism: A Practical Guide.
Quaternary Research Association: London
Walling, D. (2013) The evolution of sediment source fingerprinting investigations in fluvial
systems. Journal of Soils and Sediments 13: 1658-1675. DOI: 10.1007/s11368-013-0767-2
Walling, D.E. (2005) Tracing suspended sediment sources in catchments and river systems.
Science of the Total Environment 344: 159-184. DOI: 10.1016/j.scitotenv.2005.02.011
Walling, D.E. and Woodward, J.C. (1992) Use of radiometric fingerprints to derive
information on suspended sediment sources. In: Erosion and Sediment Transport
Monitoring Programmes in River Basins, ed. Bogen, J., Walling, D.E. and Day, T.J, 153-
243.IAHS,
211
Walling, D.E., Bogen, J. and Porto, P. (2009) Special Issue: Sediment Sources and Sediment
Delivery under Environmental Change Foreword. Catena, 79(3): 179-180.
Walling, D.E., Collins, A.L., Stroud, R.W. (2008) Tracing suspended sediment and particulate
phosphorous sources in catchments. Journal of Hydrology 350: 274-289.
DOI:10.1016/j.jhydrol.2007.10.047
Walling, D.E., Moorehead, P.W. (1987) Spatial and Temporal Variation of the Particle-Size
Characteristics of Fluvial Suspended Sediment. Geografiska Annaler. Series A, Physical
Geography 69(1): 47-59
Walling, D.E., Moorehead, P.W. (1989) The particle size characteristics of fluvial suspended
sediment: an overview. Hydrobiologia 176/177: 125-149
Walling, D.E., Owens, P.N., Leeks, G.J.L., (1999) Fingerprinting suspended sediment sources
in the catchment of the River Ouse, Yorkshire, UK. Hydrological Processes 13, 955–975.
Walling, D.E., Peart, M.R., Oldfield, F., Thompson, R. (1979) Suspended sediment sources
identified by magnetic measurements Nature 281: 110-113
Walling, D.E., Russell, M.A., Hodgkinson, R.A. and Zhang, Y. (2002) Establishing sediment
budgets for two small lowland agricultural catchments in the UK. Catena, 47(4): 323-353.
Walling, D.E., Russell, M.A., Webb, B.W. (2001) Controls on the nutrient content of
suspended sediment transported by British rivers. Science of the Total Environment 266:
113-123
Walling, D.E., Woodward, J.C. (1995) Tracing suspended sediment sources in river basins: a
case study of the River Culm, Devon, UK. Marine and Freshwater Research 46: 327-336.
Warburton, J. (2003) Wind-splash erosion of bare peat on UK upland moorlands. Catena,
52(3-4): 191-207.
Warburton, J., Holden, J., Mills, A.J. (2004) Hydrological controls of surficial mass
movements in peat. Earth Science Reviews 67(1-2): 139-156.
Weiss, D., Shotyk, W., Appleby, P. G., Kramers, I. D., Cheburkin, A. K. (1999). Atmospheric
Pb deposition since the industrial revolution recorded by five Swiss peat profiles:
Enrichment factors, fluxes, isotopic composition, and sources. Environmental Science and
Technology, 33(9) 1340-1352.
212
Wheeler, B.D., Shaw, S.C., Fojt, W.J., Robertson, R.A. (1995) Restoration of temperate
wetlands. Wiley: Chichester
Wigington, P.J., Moser, T.J., Lindeman, D.R., (2005) Stream network expansion: a riparian
water quality factor. Hydrological Processes, 19(8): 1715-1721.
Wilkinson, S.N., Wallbrink, P.J., Hancock, G.J., Blake, W.H., Shakesby, R.A., Doerr, S.H.
(2009) Fallout radionuclide tracers identify a switch in sediment sources and transport-
limited sediment yield following wildfire in a eucalypt forest. Geomorphology 110: 140-151.
DOI: http://dx.doi.org/10.1016/j.geomorph.2009.04.001
Williams, T. M., (1988) The geochemistry and magnetic mineralogy of Holocene strata in
acid basins: Case studies of granitic catchments in the Galloway and Rannoch moor regions
of Scotland, Ph.D. thesis, Univ. of Edinburgh, Edinburgh, Scotland.
Williams, T.P., Bubb, J.M., Lester, J.N. (1994) Metal accumulation within salt marsh
environments: A review Marine Pollution Bulletin 28(5); 277-290
Winburn, R.S., Lerach, S.L., McCarthy, G.J., Grier, D.G., Cathcart, J.D.. (2000) Quantification
of ferrite spinel and hematite in magnetically enriched fractions. JCPDS-International
Centre for Diffraction Data 2000, Advances in X-ray Analysis 43: 350–255.
Wolverson-Cope, F. (1976) Geology Explained in the Peak District. David and Charles: Devon
Worrall, F., Burt, T., Adamson, J. (2004) Can climate change explain increases in DOC flux
from upland peat catchments? Science of the Total Environment 326: 95-112. DOI:
10.1016/j.scitotenv.2003.11.022
Worrall, F., Burt, T.P. and Adamson, J. (2006) The rate of and controls upon DOC loss in a
peat catchment. Journal of Hydrology, 321(1-4): 311-325.
Worrall, F., Burt, T.P., Adamson, J., Reed, M., Warburton, J., Armstrong, A., Evans, M.
(2007) Predicting the future carbon budget of an upland peat catchment. Climatic Change,
85(1-2): 139-158.
Worrall, F., Evans, M.G. (2009) The carbon budget of upland peat soils. In: Bonn, A., Allott,
T.E.H., Hubacek, K. and Stewart, J., 2009. Drivers of Environmental Change in Uplands.
Abingdon: Routledge, 544 pp.
213
Worrall, F., Evans, M.G., Bonn, A., Reed, M.S., Chapman, D., Holden, J. (2009) Can carbon
offsetting pay for upland ecological restoration? Science of the Total Environment 408(1);
26-36
Worrall, F., Reed, M., Warburton, J., Burt, T. (2003) Carbon budget for a British upland peat
catchment. Science of the Total Environment 312: 133-146. DOI:
dx.doi.org/10.1016/S0048-9697(03)00226-2
Worrall, F., Reed, M.S., Warburton, J. and Burt, T. (2003) Carbon budget for a British upland
peat catchment. Science of the Total Environment, 312(1-3): 133-146.
Wright, A.L., Wang, Y., Reddy, K.R. (2008) Loss-on-Ignition Method to Assess Soil Organic
Carbon in Calcareous Everglades Wetlands. Communications in Soil Science and Plant
Analysis 39: 3074-3083
Yafa, C., Farmer, J. G. (2006) A comparative study of acid-extractable and total digestion
methods for the determination of inorganic elements in peat material by inductively
coupled plasma-optical emission spectrometry. Analytica Chimica Acta, 557(1-2) 296-303.
Yallop, A.R., Thacker, J.I., Thomas, G., Stephens, M., Clutterbuck, B., Brewer, T., Sannier, D.
(2006) The extent and intensity of management burning in the English uplands. Journal of
Applied Ecology, 43(6): 1138-1148.
Yang, H., Rose, N.L., Battarbee, R.W., Boyle, J.F., (2002) Mercury and Lead Budgets for
Lochnagar, a Scottish Mountain Lake and Its Catchment. Environmental Science &
Technology, 36(7): 1383-1388.
Yang, H., Rose, N.L., Boyle, J.F., Battarbee, R.W. (2001) Storage and distribution of trace
metals and spheroidal carbonaceous particles (SCPs) from atmospheric deposition in the
catchment peats of Lochnagar, Scotland. Environmental Pollution 115 (2), 231-238.
Yang, J. (2005) Monitoring and modelling sediment flux from a blanket peat catchment in
the Southern Pennines. Unpublished PhD theses. School of Geography, University of
Manchester
Yu Z., (2011) Holocene carbon flux histories of the world’s peatlands: Global carbon-cycle
Implications. The Holocene, 21: 761-774
Yu, L, Oldfield, F (1993) Quantitative sediment source ascription using magnetic
measurements in a reservoir‐catchment system near Nijar, SE Spain. Earth Surface
Processes and Landforms 18(5):441-454
214
Zheng J., Shotyk W., Krachler M., Fisher D. (2007). 15,800 years of atmospheric lead
deposition on Devon Ice Cap, Nunavut, Canada: natural and anthropogenic enrichments,
isotopic composition, and predominant sources. Global Biogeochemical Cycles, 21, GB2027.
Zhulidov, A.V., Headley, J.V., Robarts, R.D., Nikanorov, A.M., Ischenko, A.A., Champ, M.A.,
(1997) Concentrations of Cd, Pb, Zn, and Cu in contaminated wetlands of the Russian Artic.
Marine Pollution Bulletin 35 (7-12), 252-259.
