Floodplain Hydrology and Biogeochemistry
Charles N. Jones
Dissertation submitted to the faculty of the Virginia Polytechnic Institute and State University in
partial fulfillment of the requirements for the degree of
Doctor of Philosophy
in
Biological Systems Engineering
Durelle T. Scott, Chair
W. Cully Hession
Erich T. Hester
Richard F. Keim
David J. Sample
July 30, 2015
Blacksburg, Virginia
Keywords: river-floodplain connectivity, inundation hydrology, solute fate and transport, ecological
engineering, biogeochemistry, nutrients, dissolved organic matter, stream restoration
Copyright © 2015 C. Nathan Jones
FLOODPLAIN HYDROLOGY AND BIOGEOCHEMISTRY
C. Nathan Jones
ABSTRACT
River-floodplain connectivity is defined as the water mediated transfer of materials and energy
between a river or stream and its adjacent floodplain. It is generally accepted that restoring and/or
enhancing river-floodplain connectivity can reduce the downstream flux of reactive solutes such
as nitrogen (N) and phosphorus (P) and thus improve downstream water quality. However, there
is little scientific literature to guide ecological engineering efforts which optimize river-floodplain
connectivity for solute retention. Therefore, the aim of my dissertation research was to examine
feedbacks between inundation hydrology and floodplain biogeochemistry, with an emphasis on
analyzing variation experienced along the river continuum and the cumulative effects of river-
floodplain connectivity at the basin scale. This was completed through four independent
investigations. Field sites ranged from the Atchafalaya River Basin, the largest river-floodplain
system in the continental US, to the floodplain of a recently restored headwater stream in
Appalachia. We also developed a method to examine river-floodplain connectivity across large-
river networks and applied that methodology to US stream network. Largely, our results highlight
the role floodwater residence time distributions play in floodplain biogeochemistry. In headwater
streams, residence times restrict redox dependent processes (e.g. denitrification) and downstream
flushing of reactive solutes is the dominant process. However, in large-river floodplains, redox
dependent processes can become solute limited because of prolonged residence times and
hydrologic isolation. In these floodplains, the dominant process is often autochthonous solute
accumulation. Further, results from our modeling study suggest large-river floodplains have a
greater impact on downstream water quality than floodplains associated with smaller streams, even
when considering cumulative effects across the entire river network.
iii
To Mom and Dad. This work would not have been possible without your love, support, and
occasional (but well deserved) kick-in-the-ass.
iv
ACKNOWLEDGMENTS
While the African proverb, “It takes a whole village to raise a [gradstudent],” is overly quoted and
probably cliché, it accurately captures my graduate experience at Virginia Tech. I am very thankful
for the comradery and collaborations I experienced in the greater water science community, the
BSE family, and the Hydroecology Lab over the last five years. I would like to think this degree
is not just a piece of paper, but the sum total of our shared experiences and memories.
Specifically, I would like to thank my adviser Dr. Durelle “Scotty” Scott for his patience, shared
insight, and guidance. He continues to freely and enthusiastically share his ideas, collaborations,
and time with our lab group. I’ve been truly blessed to have him as an adviser, and hope to provide
the same caliber of mentoring to my future students.
I would also like to thank my dissertation committee and other collaborators:
Forest and Wetland Ecohydrology Lab: Drs. Richard Keim (Committee Member) and
Brandon Edwards have been incredibly supportive and generous with their ideas,
resources, and time. Both suffered through edits on my first manuscript, and at the very
minimum, I owe their lab one boat anchor.
Environmental Hydraulics Research Group: Dr. Erich Hester (Committee Member) and
Chris Guth were both critical to the execution and success of the StREAM Lab flood
experiments, and Erich’s insights in surface water-groundwater connectivity have been
invaluable to my understanding of river systems.
USGS Hydroecology of Flowing Water: Drs. Jud Harvey and Jesus Gomez-Velez have
been incredible resources and set a very high standard in our recent modeling work. I truly
appreciate their patience and look forward to future collaborations.
VT-BSE: Dr. Cully Hession (Committee Member) has been an integral part of my
education here at tech. He recruited me, was a co-advisor for my first year, and provided
the opportunity to conduct a pilot study of our StREAM Lab floodplain experiment through
the REU program. Dr. David Sample (Committee Member) has also been an integral part
of my experience at tech, providing insight into modeling studies and resources to conduct
those studies.
This list wouldn’t be complete if I didn’t thank the BSE Staff members: Denton Yoder, Laura
Lehman, and Kelly Peeler. None of this work would have been possible without these individual’s
expertise, sense of humor, and hard work.
Finally, I would like to thank my friends and family for their support and prayers these last five
years. Not many people have the support system I do, and during my roughest days, I could always
find someone to provide perspective and encouragement. Also, I want to acknowledge my
beautiful wife, Virginia Jones, for her part in this work. She’s listened to me talk about floodplains
and river networks for hours, proof read most of my papers, and continues to be an amazing friend
and wife.
v
TABLE OF CONTENTS
Table of Contents ............................................................................................................................ v
List of Tables ............................................................................................................................... viii
List of Figures ................................................................................................................................ ix
Chapter 1. Introduction ................................................................................................................... 1
1.1 Food, Water, and Energy Nexus ........................................................................................... 1
1.2 Riverine Systems and Floodplains ........................................................................................ 2
1.3 Ecological Engineering and River-Floodplain Connectivity ................................................ 2
1.4 Works Cited .......................................................................................................................... 4
Chapter 2. Perirheic mixing and biogeochemical processing in flow-through and backwater
floodplain wetlands ......................................................................................................................... 8
2.1 Introduction ........................................................................................................................... 8
2.2 Methods ................................................................................................................................ 9
2.2.1 Site Description .............................................................................................................. 9
2.2.2 Synoptic Sampling Design and Analysis ..................................................................... 11
2.2.3 Residence Time Conceptualization.............................................................................. 12
2.2.4 Statistical Analysis ....................................................................................................... 12
2.3 Results ................................................................................................................................. 14
2.3.1 Ordination Analysis ..................................................................................................... 14
2.3.2 Cluster Analysis of Samples ........................................................................................ 15
2.3.3 Comparison between the Connected and Disconnected Classes ................................. 16
2.4 Discussion ........................................................................................................................... 17
2.4.1 Conceptual Model ........................................................................................................ 17
2.4.2 Classification of Hydrologically Connected and Disconnected Locations .................. 18
2.4.3 Carbon, Nitrogen, and Phosphorus dynamics in the ARB........................................... 19
2.4.4 Temporal and Spatial Variability in Floodplain Processes .......................................... 20
2.4.5 Management Implications ............................................................................................ 20
2.5 Conclusions ......................................................................................................................... 21
2.6 Acknowledgements ............................................................................................................. 21
2.7 Works Cited ........................................................................................................................ 21
vi
Chapter 3. Biogeochemical processing of dissolved organic matter and nitrogen in floodwaters
across a gradient of river-floodplain connectivity within the Atchafalaya River floodplain ....... 26
3.1 Introduction ......................................................................................................................... 26
3.2 Methods .............................................................................................................................. 27
3.2.1 Site Description ............................................................................................................ 27
3.2.2 Synoptic Sampling Design and Analysis ..................................................................... 30
3.3. Results ................................................................................................................................ 31
3.3.1 Hydrologic Tracer of Water ......................................................................................... 31
3.3.2 Reactive Solute Characterization ................................................................................. 32
3.3.3 DOM Characterization ................................................................................................. 32
3.4 Discussion ........................................................................................................................... 36
3.4.1 River-floodplain connectivity ...................................................................................... 36
3.4.2 Biogeochemical Processing of Nitrogen within the Floodplain .................................. 36
3.4.3 Temporal variation in Dissolved Organic Matter ........................................................ 38
3.5. Conclusion ......................................................................................................................... 39
3.6 Works Cited ........................................................................................................................ 39
Chapter 4. Seasonal Variation in Floodplain Biogeochemical Processing in a Restored
Headwater Stream ......................................................................................................................... 45
4.1 Introduction ......................................................................................................................... 46
4.2 Methods .............................................................................................................................. 48
4.2.1 Site Description ............................................................................................................ 48
4.2.2 Experimental Design .................................................................................................... 49
4.2.3 Flood Experiment Analyses ......................................................................................... 49
4.2.4 Reach-Scale Modeling ................................................................................................. 50
4.3 Results and Discussion ....................................................................................................... 50
4.3.1 Inundation Hydrology .................................................................................................. 50
4.3.2 Dissolved Organic Matter Export ................................................................................ 53
4.3.3 Initial Flush of Reactive Solutes .................................................................................. 53
4.3.4 Nitrate Removal ........................................................................................................... 55
4.3.5 Modeled Load Reductions ........................................................................................... 57
4.3.6. Management Implications ........................................................................................... 58
vii
4.4 Acknowledgements ............................................................................................................. 58
4.5 Works Cited ........................................................................................................................ 58
Chapter S4. Supplemental Information ......................................................................................... 65
S4.1 Climatic Data Acquisition ................................................................................................ 65
S4.2 Water Quality Sample Handling and Analysis ................................................................. 65
S4.3 Estimates of Reach-Scale Load Reductions ..................................................................... 65
S4.4 Works Cited ...................................................................................................................... 69
Chapter 5. The effect of hydrogeomorphic gradients and hydroclimatic setting on river-
floodplain connectivity at the continental scale ............................................................................ 71
5.1 Introduction ......................................................................................................................... 71
5.2 Methods .............................................................................................................................. 73
5.3 Results and Discussion ....................................................................................................... 76
5.3.1 Heterogeneity of River-Floodplain Connectivity across USGS Gaging Stations ....... 76
5.3.2 Effect of Catchment Area on River-Floodplain Connectivity ..................................... 78
5.3.3 Effect of Physiography on River-Floodplain Connectivity ......................................... 80
5.4 Conclusion .......................................................................................................................... 80
5.5 Works Cited ........................................................................................................................ 81
Chapter S5. Supplemental Material .............................................................................................. 85
S5.1 Data Acquisition ............................................................................................................... 85
S5.2 Analysis of USGS Gage Data ........................................................................................... 87
S5.2.1 Identifying the threshold for connectivity ................................................................. 87
S5.2.2 Estimation of Sub-daily hydrographs ........................................................................ 88
S5.2.3 Characterization of river-floodplain connectivity ..................................................... 90
S5.3 Application to the National Stream Network ................................................................... 95
S5.4 Works Cited ...................................................................................................................... 98
Chapter 6. Concluding remarks and questions for future study.................................................. 102
6.1 Works Cited ...................................................................................................................... 104
viii
LIST OF TABLES
Table 4-1. Summary of residence time metrics ............................................................................ 50
Table 4-2. Summary of Spiraling metrics and associated statistics from tracer addition. ........... 56
Table S4-1. Regional Regression Coefficient Values .................................................................. 67
Table S4-2. Log Pearson Type 3 Parameters ............................................................................... 67
Table S5- 1. Main variables used in analysis ............................................................................... 86
Table S5-2. Median Flood Volume Ratio by Stream Order and Physiographic Division .......... 94
Table S5-3. Median Storage Index by Stream Order and Physiographic Division ..................... 95
Table S5-4. Spearman Correlation (RS) between FVRand Stream Order ................................... 95
Table S5-5. Estimated annual floodplain storage......................................................................... 96
Table S5-6. Cumulative floodplain storage and length weighted mean floodplain storage ........ 96
Table S5-7. Cumulative length normalized floodplain storage.................................................... 97
ix
LIST OF FIGURES
Figure 2-1. Map of the Southern Atchafalaya River Basin (ARB) .............................................. 10
Figure 2-2. Constituent concentrations in ordination space ......................................................... 13
Figure 2-3. Evolution of the floodwater chemistry over the course of the annual flood pulse .... 14
Figure 2-4. Separation of Connected and Disconnected sites ...................................................... 15
Figure 2-5. Box plots of nitrogen speciation and the DOC:NO3- ratios throughout the flood ... 16
Figure 2-6. Relationship between A.) DOC and δ18O and B.) Fe and SRP ............................... 17
Figure 2-7. Conceptual sketch ..................................................................................................... 18
Figure 3-1. Map displaying the sampling locations ..................................................................... 28
Figure 3-2. Annual Hydrograph and River-Floodplain Hysteresis Plot ......................................29
Figure 3-3. Isotopic content of samples collected over the course of the flood .......................... 32
Figure 3-4. NO3-, TKN, and DOC:TKN ratio over the course of the flood ................................. 33
Figure 3-5. Dissolved carbon concentration and isotopic content across the flood. .................... 34
Figure 3-6. Spectral Indicies of dissolved organic matter (DOM) across the flood pulse ........... 35
Figure 3-7. Byplots showing relationships between reactive solutes. ......................................... 36
Figure 3-8. Conceptual model of swamp channel hydrology and biogeochemistry. ................... 37
Figure 4-1. Conceptual model of processes within the experimental floodplain slough. ............ 47
Figure 4-2. Climatic data measured in the StREAM Lab floodplain. .......................................... 48
Figure 4-3. Inlet and outlet concentrations of dissolved reactive constituents. ........................... 52
Figure 4-4. The total load observed at the inlet (dark) and outlet (light) of the floodplain. ........ 54
Figure 4-5. Areal NO3- uptake ..................................................................................................... 57
Figure S4-1. Map of estimated inundation and flow-inundation relationship ............................. 67
Figure S4-2. Parameterized Flow Duration Curve (FDC) at the StREAM Lab watershed. ........ 68
Figure 5-1. Conceptual model displaying the threshold of connectivity ..................................... 73
Figure 5-2. Flood volume ratio and storage index approximations. ............................................ 76
Figure 5-3. Floodplain storage volume ........................................................................................ 78
Figure 5-4. The summation of length normalized floodplain storage volume............................. 80
Figure S5-1. Relationship between QBKF* and catchment area for 5,981 gaging stations ........... 88
Figure S5-2. Illustration of the Sangal [1983] approximation. .................................................... 84
Figure S5-3. Strong relationship experienced between QIPF and QMDF at Colombia River ......... 90
Figure S5-4. Annual hydrograph from Boulder Creek near Boulder ....................................................... 90
Figure S5-5. Estimate of UBKF at USGS gaging stations ............................................................. 92
Figure S5-6. Conceptualization of the Divided Channel Method (DCM) ................................... 93
Figure 6-1. Conceptual model of large scale controls on net solute removal in floodplains…..102
1
CHAPTER 1. INTRODUCTION
“I recognize the right and duty of this generation to develop and use the natural
resources of our land; but I do not recognize the right to waste them, or to rob, by
wasteful use, the generations that come after us” – Theodore Roosevelt, 1910.
1.1 Food, Water, and Energy Nexus While the world’s population is expected to stabilize in the next 100 years [Lutz et al., 2001; Lutz
et al., 2014], the next generation of decision makers, engineers, and scientists are going to face
monumental challenges that involve complex interactions between social, economic, and
environmental drivers. The common denominator will likely be the interplay between food
security [Godfray et al., 2010], supply of freshwater [Vörösmarty et al., 2000], and energy use and
production [Lewis and Nocera, 2006]. Even today, these variables are drivers of unrest and change
across the world. This is most evident in the recent effects of global climate change [Walther et
al., 2002], which have been widely attributed to the increase in fossil fuel emissions which began
during the Industrial Revolution in the mid-1700s [Hegerl et al., 2007]. The Arab Spring, a series
of major protests, civil uprisings, and even civil wars in the Middle East and northern Africa, is
largely attributed to food shortages caused by global fluctuations in climate, among other
complicated social and economic factors [Davide et al., 2015]. Even here in the relatively stable
US, changes in climate have had adverse effects in the form of both drought and flood events.
During the 2012 drought in the Midwest and Northern Plains, commodity prices increased
anywhere from 2% (dairy) to 16% (beef) because of drought-impacted grain crop [Mallya et al.,
2013]. Ironically, we are simultaneously observing a general intensification of flooding in the
Midwest [Mallakpour and Villarini, 2015] and Eastern US [Villarini and Smith, 2010], the cost of
which will likely be disproportionate. Just in the last decade, flooding damage from Hurricane
Sandy (2012) and Hurricane Katrina (2005) was estimated at $50 Billion and $108 Billion,
respectively [Berg et al., 2013; Brown et al., 2005]. To put these costs into perspective, the cost of
Hurricane Katrina was approximately 4% of federal spending in 2005 [US Senate, 2005].
Another major challenge is balancing the intensification of agriculture with the loss of ecosystem
services. In the next 30 years alone, current research suggests we will have to double food
production to keep up with current consumption rates associated with increases in both population
and prosperity [Foley et al., 2011; Godfray et al., 2010]. Yet, at the current production levels,
agriculture landuse is already a major driver behind environmental change and degradation [Foley
et al., 2005; Power, 2010]. This degradation is particularly relevant when we examine water
quality across the United States, where over 1 million kilometers of our rivers and streams are
considered impaired [USEPA, 2011]. Much of this degradation is associated with the green
revolution and the subsequent threefold increase in agricultural production over the last 50-75
years, which was made possible by the technological advances in the production of nitrogen (N)
and phosphorus (P) fertilizers [Villalba et al., 2008].
Concentrating large amounts of fertilizers on relatively small areas of land to enhance production
essentially decouples the N and P cycles from carbon cycling [Sharpley et al., 2013], leading to
degradation of hydrologically connected aquatic ecosystems [Vitousek et al., 1997]. In undisturbed
systems, N, P, and carbon cycling is tightly coupled and is controlled by biological processing and
ecosystem stoichiometry. However, in agroecosystems elemental ratios are no longer coupled,
often exceeding assimilative thresholds and leading to the “leakage” of inorganic N and P to
2
adjacent waterways [Dodds et al., 2010; Evans-White et al., 2009]. Elevated N and P
concentrations promote excessive algal growth and can lead to hypoxia, or suboxic conditions not
suitable for aquatic life [Rabalais et al., 2002], leading to the degradation of both the stream
network and receiving water bodies. Worldwide, over 400 receiving bodies experience hypoxia
annually, including the economically and culturally import Gulf of Mexico and Chesapeake Bay
[Diaz and Rosenberg, 2011].
1.2 Riverine Systems and Floodplains The ecological integrity and the associated ecosystem services of our riverine systems are directly
linked to the dynamic nature of the hydrologic cycle [Poff et al., 1997] and heterogeneity of our
landscapes [Ward et al., 2002]. Unlike a pipe, riverine systems have multiple methods of
conveying water in both the subsurface and surface environments. Subsurface flowpaths, or
hyporheic flow, can vary from short flowpaths on the order of centimeters to longer flowpaths on
the order of kilometers [Helton et al., 2012]. Hyporheic flow often occurs directly beneath the
stream bed, but can also be found in the subsurface adjacent to the stream [Poole et al., 2008]. As
stream order increases, the complexity and number of hyporheic flowpaths also increases [Stanford
and Ward, 1993], but the flowpaths convey a smaller portion of the total river flow [Wondzell,
2011]. During high flow events (e.g. floods), overbank flow is routed from the channel to adjacent
low-lying land, known as the floodplain [Poole, 2010]. As the stream network shifts from
headwater streams to a larger river, overbank flows transition from short, sporadic flows to longer,
sustained pulses [Tockner et al., 2000]. As river stage dynamically fluctuates, the stream
ecosystem expands and contracts as both hyporheic and overbank flowpaths dynamically connect
and disconnect from the main channel [Stanley et al., 1997]. Many of these alternate flowpaths are
mixing zones between different sources of water [Mertes, 1997], producing ecotonal gradients that
are often both ecologically and economically important [McClain et al., 2003; Vidon et al., 2010].
Of particular interest, floodplains are hotspots for solute retention and removal within riverine
networks. This is largely due to feedbacks between inundation hydrology and biogeochemical
processes that control solute retention and removal within floodplains [Junk et al., 1989; Tockner
et al., 2000]. More specifically, increased residence time of floodwaters increase the potential for
solute retention and removal [Forshay and Stanley, 2005; James et al., 2008]. During inundation,
floodwaters mix with local waters and come into contact with reactive floodplain sediments, often
forming redox gradients at the sediment-water interface which enhance redox-dependent processes
such as denitrification [Mertes, 1997; Noe and Hupp, 2005]. However, both short and prolonged
residence times can restrict redox-dependent processes, suggesting that a balance between river-
floodplain exchange and residence time distributions should be established [James et al., 2008;
Tockner et al., 1999]. However, complicated feedbacks between river-floodplain exchange and
inundation hydrology limit our understanding of how to achieve this balance at both the reach and
basin scale.
1.3 Ecological Engineering and River-Floodplain Connectivity Riverine floodplains provide many ecosystem services, ranging from increased
agriculture/silviculture production and flood peak attenuation, to biogeochemical processing of
excess agrochemicals during flood events [Tockner and Stanford, 2002]. In many human-impacted
basins, the majority of the solute balance occurs during discrete flood pulses when rivers are
hydrologically connected to their adjacent floodplains [Kaushal et al., 2014; Royer et al., 2006],
making river-floodplain connectivity an attractive management variable when engineering
solutions to reduce the downstream flux of reactive solutes (e.g. Mitsch et al. [2001]). The
3
importance of river-floodplain connectivity is evident in recent efforts in the Mid-Atlantic and
North Eastern US associated with the Chesapeake Bay TMDL, where grants for stream restoration
are prioritized through functional assessments which emphasize river-floodplain connectivity
[Harman et al., 2012]. However, there is little scientific literature to guide ecological engineering
efforts which optimize river-floodplain connectivity for solute retention. Therefore, the overall
goal of this dissertation is to further investigate feedbacks between inundation hydrology
and floodplain biogeochemistry, specifically analyzing variation experienced along the
gradient from headwater streams to large rivers and the cumulative effects of river-
floodplain connectivity at the basin scale.
This dissertation is composed of four independent studies that spanned the river continuum,
investigating the effects of residence time, dissolved organic matter content, seasonality, and
hydrogeomorphic conditions on the downstream flux of solutes:
Chapter 2. Perirheic mixing and biogeochemical processing in flow-through and
backwater floodplain wetlands
Chapter 3. Biogeochemical processing of dissolved organic matter and nitrogen in
floodwaters across a gradient of river-floodplain connectivity within the Atchafalaya River
floodplain
Chapter 4. Seasonal variation in floodplain biogeochemical processing in a restored
headwater stream
Chapter 5. The effect of hydrogeomorphic gradients and hydroclimatic setting on river-
floodplain connectivity at the continental scale
“In the space of one hundred and seventy-six years the Lower Mississippi has shortened itself two
hundred and forty-two miles. That is an average of a trifle over one mile and a third per year.
Therefore, any calm person, who is not blind or idiotic, can see that in the Old Oolitic Silurian
Period, just a million years ago next November, the Lower Mississippi River was upwards of one
million three hundred thousand miles long, and stuck out over the Gulf of Mexico like a fishing-
rod. And by the same token any person can see that seven hundred and forty-two years from now
the Lower Mississippi will be only a mile and three-quarters long, and Cairo and New Orleans
will have joined their streets together, and be plodding comfortably along under a single mayor
and a mutual board of aldermen. There is something fascinating about science. One gets such
wholesale returns of conjecture out of such a trifling investment of fact.”
- Life on the Mississippi, Mark Twain
4
1.4 Works Cited Berg, R. J., J. L. Beven Ii, E. S. Blake, J. P. Cangialosi, T. B. Kimberlain, and C. National
Hurricane (2013), United States National Oceanic and Atmospheric Administration's National
Weather Service.
Brown, D. P., R. D. Knabb, C. National Hurricane, and J. R. Rhome (2005), United States
National Oceanic and Atmospheric Administration's National Weather Service.
Davide, N., J. Aled Wynne, and B. Giangiacomo (2015), Quantitative Assessment of Political
Fragility Indices and Food Prices as Indicators of Food Riots in Countries, Sustainability, 7(4).
Diaz, R. J., and R. Rosenberg (2011), Introduction to Environmental and Economic
Consequences of Hypoxia, Int. J. Water Resour. Dev., 27(1), 71-82.
Dodds, W. K., W. H. Clements, K. Gido, R. H. Hilderbrand, and R. S. King (2010), Thresholds,
breakpoints, and nonlinearity in freshwaters as related to management, Journal of the North
American Benthological Society, 29(3), 988-997.
Evans-White, M. A., W. K. Dodds, D. G. Huggins, and D. S. Baker (2009), Thresholds in
macroinvertebrate biodiversity and stoichiometry across water-quality gradients in Central Plains
(USA) streams, Journal of the North American Benthological Society, 28(4), 855-868.
Foley, J. A., et al. (2005), Global Consequences of Land Use, Science, 309(5734), 570-574.
Foley, J. A., et al. (2011), Solutions for a cultivated planet, Nature, 478(7369), 337-342.
Forshay, K. J., and E. H. Stanley (2005), Rapid nitrate loss and denitrification in a temperate
river floodplain, Biogeochemistry, 75(1), 43-64.
Godfray, C. J., J. R. Beddington, I. R. Crute, L. Haddad, L. Davi, J. F. Muir, P. Jule, S.
Robinson, S. M. Thomas, and C. Toulmin (2010), Food Security: The Challenge of Feeding 9
Billion People, Science, 327(5967), 812-818.
Harman, W., R. Starr, M. Carter, K. Tweedy, M. Clemmons, K. Suggs, and C. Miller (2012), A
Function-Based Framework for Stream Assessment and Restoration ProjectsRep. EPA 843-K-12-
006, US Environmental Protection Agency, Office of Wetlands, Oceans, and Watersheds,
Washington, DC.
Hegerl, G., F. Zwiers, P. Braconnot, N. Gillet, and Y. Luo (2007), Understanding and attributing
climate change, Understanding and attributing climate change.
Helton, A. M., G. C. Poole, R. A. Payn, C. Izurieta, and J. A. Stanford (2012), Scaling flow path
processes to fluvial landscapes: An integrated field and model assessment of temperature and
dissolved oxygen dynamics in a river-floodplain-aquifer system, J. Geophys. Res.-Biogeosci.,
117, 13.
5
James, W. F., W. B. Richardson, and D. M. Soballe (2008), Effects of residence time on summer
nitrate uptake in mississippi river flow‐ regulated backwaters, River Research and Applications,
24(9), 1206-1217.
Junk, W. J., P. B. Balyley, and R. E. Sparks (1989), The flood pulse concept in river-floodplain
systems., Can. Spec. Publ. Fish. Aquat. Sci., 106, 110-127.
Kaushal, S. S., P. M. Mayer, and P. G. Vidon (2014), Land use and climate variability amplify
carbon, nutrient, and contaminant pulses: a review with management implications, JAWRA
Journal of ….
Lewis, N. S., and D. G. Nocera (2006), Powering the planet: Chemical challenges in solar energy
utilization, Proceedings of the National Academy of Sciences, 103(43), 15729-15735.
Lutz, W., W. Sanderson, and S. Scherbov (2001), The end of world population growth, Nature,
412(6846), 543-545.
Lutz, W., W. Butz, S. KC, W. Sanderson, and S. Scherbov (2014), Population growth: Peak
probability, Science, 346(6209), 561.
Mallakpour, I., and G. Villarini (2015), The changing nature of flooding across the central
United States, Nature Clim. Change, 5(3), 250-254.
Mallya, G., L. Zhao, X. C. Song, D. Niyogi, and R. S. Govindaraju (2013), 2012 Midwest
Drought in the United States, J Hydrol Eng, 18(7), 737745.
McClain, M. E., et al. (2003), Biogeochemical Hot Spots and Hot Moments at the Interface of
Terrestrial and Aquatic Ecosystems, Ecosystems, 6(4), 301-312.
Mertes, L. A. K. (1997), Documentation and significance of the perirheic zone on inundated
floodplains, Water Resour. Res., 33(7), 1749-1762.
Mitsch, W. J., J. W. Day, J. W. Gilliam, P. M. Groffman, D. L. Hey, G. W. Randall, and N. M.
Wang (2001), Reducing nitrogen loading to the Gulf of Mexico from the Mississippi River
Basin: Strategies to counter a persistent ecological problem, Bioscience, 51(5), 373-388.
Noe, G. B., and C. R. Hupp (2005), Carbon, nitrogen, and phosphorus accumulation in
floodplains of Atlantic Coastal Plain rivers, USA, Ecological Applications, 15(4), 1178-1190.
Poff, N. L., J. D. Allan, M. B. Bain, J. R. Karr, K. L. Prestegaard, B. D. Richter, R. E. Sparks,
and J. C. Stromberg (1997), The natural flow regime, Bioscience, 47(11), 769-784.
Poole, G. C. (2010), Stream hydrogeomorphology as a physical science basis for advances in
stream ecology, Journal of the North American Benthological Society, 29(1), 12-25.
6
Poole, G. C., S. J. O'Daniel, K. L. Jones, W. W. Woessner, E. S. Bernhardt, A. M. Helton, J. A.
Stanford, B. R. Boer, and T. J. Beechie (2008), Hydrologic spiralling: The role of multiple
interactive flow paths in stream ecosystems, River Research and Applications, 24(7), 1018-1031.
Power, A. G. (2010), Ecosystem services and agriculture: tradeoffs and synergies, Philosophical
Transactions of the Royal Society of London B: Biological Sciences, 365(1554), 2959-2971.
Rabalais, N. N., R. E. Turner, and W. J. Wiseman (2002), Gulf of Mexico hypoxia, aka "The
dead zone", Annual Review of Ecology and Systematics, 33.
Royer, T. V., M. B. David, and L. E. Gentry (2006), Timing of Riverine Export of Nitrate and
Phosphorus from Agricultural Watersheds in Illinois: Implications for Reducing Nutrient
Loading to the Mississippi River, Environ. Sci. Technol.
Senate, U. (2005), 2005 Public Budget Database, Fiscal Year 2005 Public Budget Database,
Receipts-Public Budget Database.
Sharpley, A., H. P. Jarvie, A. Buda, L. May, B. Spears, and P. Kleinman (2013), Phosphorus
Legacy: Overcoming the Effects of Past Management Practices to Mitigate Future Water Quality
Impairment, Journal of Environmental Quality, 42(5), 1308-1326.
Stanford, J. A., and J. V. Ward (1993), An Ecosystem Perspective of Alluvial Rivers:
Connectivity and the Hyporheic Corridor, Journal of the North American Benthological Society,
12(1), 48-60.
Stanley, E. H., S. G. Fisher, and N. B. Grimm (1997), Ecosystem expansion and contraction in
streams, Bioscience, 47(7), 427-435.
Tockner, K., and J. A. Stanford (2002), Riverine flood plains: present state and future trends,
Environmental Conservation, 29(3), 308-330.
Tockner, K., F. Malard, and J. V. Ward (2000), An extension of the flood pulse concept,
Hydrological Processes, 14(16-17), 2861-2883.
Tockner, K., D. Pennetzdorfer, N. Reiner, F. Schiemer, and J. V. Ward (1999), Hydrological
connectivity, and the exchange of organic matter and nutrients in a dynamic river-floodplain
system (Danube, Austria), Freshw. Biol., 41(3), 521-535.
USEPA (2011), Watershed Assessment, Tracking & Environmnetal Results, edited.
Vidon, P., C. Allan, D. Burns, T. P. Duval, N. Gurwick, S. Inamdar, R. Lowrance, J. Okay, D.
Scott, and S. Sebestyen (2010), Hot Spots and Hot Moments in Riparian Zones: Potential for
Improved Water Quality Management1, J. Am. Water Resour. Assoc., 46(2), 278-298.
7
Villalba, G., Y. Liu, H. Schroder, and R. U. Ayres (2008), Global Phosphorus Flows in the
Industrial Economy From a Production Perspective, Journal of Industrial Ecology, 12(4), 557-
569.
Villarini, G., and J. A. Smith (2010), Flood peak distributions for the eastern United States,
Water Resources Research, 46(6), W06504.
Vitousek, P. M., H. A. Mooney, J. Lubchenco, and J. M. Melillo (1997), Human Domination of
Earth's Ecosystems, Science, 277(5325), 494-499.
Vörösmarty, C. J., P. Green, J. Salisbury, and R. B. Lammers (2000), Global Water Resources:
Vulnerability from Climate Change and Population Growth, Science, 289(5477), 284-288.
Walther, G.-R., E. Post, P. Convey, A. Menzel, C. Parmesan, T. J. C. Beebee, J.-M. Fromentin,
O. Hoegh-Guldberg, and F. Bairlein (2002), Ecological responses to recent climate change,
Nature, 416(6879), 389-395.
Ward, J. V., K. Tockner, D. B. Arscott, and C. Claret (2002), Riverine landscape diversity,
Freshw. Biol., 47(4), 517-539.
Wondzell, S. M. (2011), The role of the hyporheic zone across stream networks, Hydrological
Processes, 25(22), 3525-3532.
8
CHAPTER 2. PERIRHEIC MIXING AND BIOGEOCHEMICAL
PROCESSING IN FLOW-THROUGH AND BACKWATER
FLOODPLAIN WETLANDS
C. Nathan Jones, Durelle T. Scott, Brandon L. Edwards, Richard. F. Keim
Submitted: March 2014
To: Water Resources Research
Status: Accepted for publication in September 2014
Abstract Inundation hydrology and associated processes control biogeochemical processing in floodplains.
To better understand how hydrologic connectivity, residence time, and intrafloodplain mixing vary
in floodplain wetlands, we examined how water quality of two contrasting areas in the floodplain
of the Atchafalaya River—a flow-through and a backwater wetland—responded to an annual flood
pulse. Large, synoptic sampling campaigns occurred in both wetlands during the rising limb, peak,
and falling limb of the hydrograph. Using a combination of conservative and reactive tracers, we
inferred three dominant processes that occurred over the course of the flood pulse: flushing (rising
limb), advective transport (peak), and organic matter accumulation (falling limb).
Biogeochemistry of the two wetlands was similar during the peak while the river overflowed into
both. However, during the rising and falling limbs, flow in the backwater wetland experienced
much greater residence time. This led to the accumulation of dissolved organic matter and
dissolved phosphorus. There were also elevated ratios of dissolved organic carbon to nitrate in the
backwater wetland, suggesting nitrogen removal was limited by nitrate transported into the
floodplain there. Collectively, our results suggest inclusion of a temporal component into the
perirheic concept more fully describes inundation hydrology and biogeochemistry in large river
floodplain.
2.1 Introduction Hydrologic connectivity of flowing waters and their adjacent floodplains is a key variable
controlling material flux from river basins. During floods, the river and adjacent floodplain become
hydrologically connected [Poole, 2010], allowing for surface exchange of materials between the
floodplain and river network [Junk et al., 1989; Tockner et al., 2000]. Both the physical
environment (e.g., topography and surface roughness) and antecedent conditions (e.g., soil
moisture) affect water flow and therefore control the movement of reactive solutes across the
floodplain surface [Mertes, 1997; Harvey et al., 2003; Welti et al., 2012]. In many floodplains, the
initial wetting of soils releases solutes attached to the soil surface leading to a first flush of reactive
solutes upon inundation [e.g., Bechtold et al., 2003]. These constituents fuel internal heterotrophic
activity and can also be flushed downstream through the river network [Noe et al., 2013]. Where
floodwaters meet locally derived water on the floodplain, an area of mixing known as the perirheic
zone forms, in part characterized by differences in chemical properties of the water [Mertes, 1997].
In many cases, steep redox gradients develop at the soil-water interface and across the perirheic
zone, allowing for rapid nutrient transformation [Forshay and Stanley, 2005; Racchetti et al., 2011;
Scaroni et al., 2011]. For example, during the 2011 flood within the Atchafalaya River Basin,
9
Louisiana USA, periods of high hydrologic connectivity during the flood peak resulted in
significant nitrate (NO3-) removal [Scott et al., 2014].
In addition to hydrologic connectivity, residence time of water in the floodplain is a key variable
that affects biogeochemical processes [Cirmo and McDonnell, 1997]. A combination of regional
climate, local hydrology, topography, and feedbacks associated with internal complexity and river-
floodplain exchange, control residence time [Hughes, 1980; Welti et al., 2012]. Areas of the
floodplain that experience long residence times (e.g., backwaters) also experience the greatest
nitrogen removal rates [Misiti et al., 2011]. However, in many systems nitrogen removal can be
limited by NO3- transport [Forshay and Stanley, 2005; Racchetti et al., 2011]. Long-residence
waters begin to accumulate reactive solutes such as soluble reactive phosphorus (SRP),
ammonium, and labile dissolved organic carbon (DOC), posing a risk for downstream waters
[Banach et al., 2009; Chow et al., 2013]. In contrast, flow-through wetlands often have removal
rates as low as ~10% [McJannet et al., 2012]. Therefore, net floodplain function is a balance
between maximizing nutrient retention and minimizing export of other reactive constituents
[Tockner et al., 1999].
Inundation hydrology and associated processes (e.g., hydrologic connectivity, residence time, and
perirheic mixing) are critical for biogeochemical processing in floodplain ecosystems. Within the
Mississippi River Basin, and other large river systems, it has been proposed to restore many of
these processes to reduce the downstream flux of agrochemicals (e.g. Mitsch et al. [2001]).
However, there are few studies that have examined how these processes and the resulting
biogeochemical responses dynamically vary throughout the course of a flood and across the
floodplain. Furthermore, little information is available on how redox conditions in large floodplain
systems affect nitrogen removal and behavior of other reactive solutes. To better understand the
above processes, we conducted synoptic samplings across two contrasting areas of floodplain—a
flow through wetland and a backwater wetland—over the course of a large annual flood pulse.
Specifically, our goal was to examine differences in the evolution of the floodwaters in sites with
contrasting hydrologic regimes and infer the effect of residence time and hydrologic connectivity
on biogeochemical processes.
2.2 Methods 2.2.1 Site Description
The fifth largest river in North America, the Atchafalaya River flows 275 km though the
Atchafalaya River Basin (ARB)—a 3,391 km2 floodplain contained within levees—to its outlet in
the Gulf of Mexico [Ford and Nyman, 2011]. As a distributary system, 30% of the combined flows
of the Mississippi and Red rivers are routed into the Atchafalaya River.
The southern portion of the ARB, where we conducted this study, is made up of a series of
interconnected channels, canals, lakes, and sloughs that generally flow southward towards the Gulf
of Mexico (GOM) through a floodplain dominated by bottomland hardwoods and cypress-tupelo
swamp (Figure 2-1). Because the ARB is confined by artificial levees on the east and west, the
major water source is the Atchafalaya River, but there are several minor sources including
navigational locks and pumping stations for drainage from adjacent agricultural land. Much of the
permanently inundated portion of the basin is hydrologically isolated from the river [Scott et al.,
2014], and many of these backwater locations are low in dissolved oxygen and provide poor fish
10
habitat [Kaller et al., 2011]. During large flood pulses, as much as 50% of river flow is routed into
the floodplain, reconnecting backwater sites to the main river [Scott et al., 2014]. However, it is
unclear how much floodplain-river exchange occurs after the initial filling of the backwater areas
[BryantMason et al., 2013].
Figure 2-1. Map of the Southern Atchafalaya River Basin (ARB) and the hydrograph at Butte La
Rose (USGS Gage 07381515).
11
The ARB is divided into 13 distinct Water Management Units (WMUs), separated by internal
barriers to flow, as defined by the U.S. Army Corps of Engineers. For this study, the Buffalo Cove
and Henderson WMUs were chosen to represent flow-through and backwater wetland systems,
respectively. Flow exchanges with the main channel are highly variable at both sites over time.
The Buffalo Cove WMU, approximately 23,000 ha, was previously a lake-margin, deepwater
cypress swamp hydrologically connected to Grand Lake before becoming more isolated in recent
decades by sedimentation from the adjacent Atchafalaya River [Tye and Coleman, 1989].
Typically, the Buffalo Cove WMU only receives water through its southern border. However,
during the large 2011 flood when we sampled, river stage was high enough to route water through
cuts in the levees along the northern boundary and allow flow-through hydrologic conditions, or
flow from upstream to downstream. The Henderson WMU, approximately 29,000 ha, is also a
deepwater cypress swamp, but water is kept artificially deep by a low dam on its sole connection
to the Atchafalaya River near its southeastern (downstream) border [Tye and Coleman, 1989].
Representing backwater hydrology, floodwaters both fill and drain the wetland through this
connection, resulting in relatively hydrologically isolated areas of inundation in the northern
portion of the WMU.
2.2.2 Synoptic Sampling Design and Analysis
Three separate, week-long synoptic sampling campaigns were completed: the rising limb (52
samples), peak (39 samples), and falling limb (31 samples) of the 2011 flood. At each site, a 1 L
sample of near-surface water was collected from a boat using a peristaltic pump and filtered
through 0.45 μM Geotech capsule filters. Samples were stored on ice in the dark and were
transported to the laboratory within 24 hours. Subsamples for nutrient and metals analyses were
stored in 250 mL plastic bottles and frozen until analysis. Subsamples for dissolved organic carbon
and fluorescence analysis were stored in precombusted amber glass vials and refrigerated until
analysis within 5 days. Water isotope samples were collected separately and stored in 20 mL glass
vials with no headspace.
Analyses for ammonium (NH4+) and soluble reactive phosphorus (SRP) were completed on a
SEAL Analytical AA3 analyzer using the Berthelot reaction (salicylate alternative) and the
Murphy and Riley method, respectively. Chloride (Cl-), sulfate (SO42-), and nitrate (NO3-) were
measured through ion chromatography using a Dionex ICS-3000. Inductively couple plasma
atomic emission spectrometry was used to measure dissolved metals (Mn2+ and Fe2+). Dissolved
organic carbon (DOC) and total dissolved nitrogen (TDN) were measured using high temperature
oxidative combustion using Shimadzu TOC-V CPH. Dissolved organic nitrogen (DON) was
calculated by subtracting NO3- and NH4+ from TDN. Carbon fluorescence was measured using
Horiba Jobin Yvon Fluoromax-4 following the protocol outlined by Cory et al. [2010]. Excitation-
emission matrices were applied to the Cory and McKnight [2005] Parallel Factor Analysis
(PARAFAC) model to analyze DOM composition. Specifically, the redox coefficient was
estimated as RC = Q2/HQ, where Q2 and HQ are oxidized and reduced quinones of the DOM,
respectively, and large RC indicates oxidized conditions [Cory and McKnight, 2005]. Stable
isotopic composition, δ18O and δ2H (δD), was measured using cavity ring-down laser
spectrometry with a Picarro Water Isotope Analyzer. Deuterium excess (d) represents the deviation
from the meteoric water line (MWL), which indicates fractionation caused by evaporation
[Kendall and McDonnell, 1998].
12
2.2.3 Residence Time Conceptualization
Our analysis investigated both the temporal and spatial variability of hydrologic residence time in
the ARB using a combination of conservative and reactive solutes. Stable isotopes of water have
been used extensively to define sources of water (e.g., Hunt et al. [1998]) and the age or residence
time of water (e.g., McGuire et al. [2005]) in a variety of systems. Because of strong latitudinal
gradients of isotopic composition of precipitation and thus river water across the Mississippi Basin
[Kendall and Coplen, 2001], river water was relatively low in δD and δ18O compared to rainfall-
derived floodplain water in the ARB [Scott et al. 2014]. In addition, δ18O was high in floodplain
water because of evaporative fractionation over time: as water evaporates, the light isotopologues
are preferentially evaporated, which increases δ18O in the residual water [Kendall and McDonnell,
1998]. Because there are two primary sources of water in the ARB, river water and rain water, and
because both fractionation of river water and addition of rainwater both increase δ18O relative to
river water, δ18O is a good proxy for residence time: high δ18O indicates higher residence time
and low δ18O indicates similarity to river water and therefore a shorter residence time in the ARB.
A similar approach was used by Scott et al. [2014] and Kaller et al. [in review], using δD to identify
regions of lower and higher residence times in the ARB for this event.
2.2.4 Statistical Analysis
A combination of ordination and cluster analyses was used to identify the evolution of water
through time, examine gradients from highly connected to hydrologically isolated regions, and
elucidate the effects of hydrologic connectivity and perirheic mixing on biogeochemical
processing. Nonmetric Multidimensional Scaling (NMDS) was chosen for the ordination analysis
for its ability to handle nonnormal and nonlinear datasets [Kruskal, 1964]. Euclidean distance of
the scaled data was used as the dissimilarity measure. The number of iterations in finding the least-
stress ordination was 10,000 and the first axis was rotated to the most dominant gradient. The
number of axes was chosen by minimizing both the dimensionality, the number of axis, and stress,
a goodness-of-fit measure that represents the deformation in the data as it is transformed into
ordination space [Clarke, 1993; Kruskal, 1964]. Linear environmental vectors were fit using least
squares to identify gradients of constituents in ordination space. All analyses were completed in
statistical software R using the Vegan package.
A Hierarchical Cluster Analysis (HCA) was run on the NMDS ordination scores to separate the
ordination into groupings. Ward’s Minimum Variance clustering technique was used in
conjunction with Euclidean distance as the dissimilarity measure [Orloci, 1967; Ward, 1963]. A
Multi-Response Permutation Procedure (MRPP) was then used to determine whether the clusters
were significantly different from each other [Mielke and Berry, 2001]. The MRPP was conducted
on the rank transformed water quality data associated with each cluster instead of the NMDS scores
used to develop the clusters. This was done to fully represent the data structure while also
mimicking the NMDS procedure [McCune et al., 2000].
Once statistically different groups were defined, pairwise Wilcoxon tests were used to test for
differences in the individual chemical analytes between each pair of different groups [Hodges and
Lehmann, 1963]. All Wilcoxon test were used in conjunction with Holm’s correction to control
for family-wise error rates at alpha = 0.05 [Holm, 1979].
13
Figure 2-2. Constituent concentrations in ordination space. Bubble size is proportional to the
concentration of each constituent, and the length and direction of vector arrows indicate the
direction and strength of linear correlation of that constituent projected.
δ18O R2=0.87
NO3_
R2=0.63
SRP R2=0.48
Fe R2=0.66
RC R2=0.72
Cl-
R2=0.54
DOC R2=0.83
DON
R2=0.69
NMDS-1 NMDS-3
SO42-
R2=0.67
A.)
C.)
NH4+
R2=0.35
D.)
E.)
F.)
G.)
H.)
I.)
J.)
B.)
14
2.3 Results 2.3.1 Ordination Analysis
A 3-axis NMDS model best represented the gradients in the dataset. There was high correlation
(R2 = 0.956) and relatively low stress (S=0.10) between the observed and modeled dissimilarity
(Figure 2-2). The gradient in NMDS-1 was associated with -DOC, -DON, -δ18O, +NO3, and +SO42-
(Figure 2-2a-d,j). Because DOM accumulates over time (e.g., Tockner et al. [1999]), electron
acceptors (e.g. NO3- and SO4
2-) typically decrease with increased residence time (e.g., Forshay and
Stanley, [2005]), and enriched (+) isotopic composition indicates evaporation or local rainfall (e.g.,
Scott et al., [2014]), NMDS-1 appears to represent (inverse) residence time. The gradient in
NMDS-2 was associated with -SRP, -Fe2+, and +RC (Figure 2-2e-g). SRP is often released when
redox sensitive ferric soils are reduced (e.g., Banach et al. [2009]). Therefore, NMDS-2 appears
to be an (inverse) redox gradient. Finally, the gradient in NMDS-3 was most closely associated
with -NH4+, -Cl-
, and –SO4- (Figure 2-2h-i); however, these linear gradients are weaker than those
of the previous two groups. Variation of these constituents, specifically with chloride, occurred
during the rising limb, which might have been an initial flush in the connected regions.
Figure 2-3. Evolution of the floodwater chemistry over the course of the annual flood pulse. Axis
labels are qualitative interpretations based on strength of correlation with constituents. Circles
(grey = flow-through, black = backwater) represent individual samples, while density ellipses
represent the location of 50% of samples from the backwater and flow-through wetlands,
respectively, from each synoptic sampling event. This figure corresponds to the first column in
Figure 2-2.
15
There were temporal changes in water quality across the flood event in both wetlands (Figure 2-3).
Both wetlands started the event with water in a relatively high reduced state (low NMDS-2) and
long residence times (low NMDS-1), then moved to shorter residence times and oxidized states
during the peak of the flood (Figure 2-3). Following the peak, residence time increased in each
wetland but redox states remained relatively oxidized. Because the backwater wetland was less
hydrologically connected to the river, it experienced greater extremes in redox conditions and
residence times compared to the flow-through wetland. While full hysteresis is not represented in
this dataset, it appears likely that reducing conditions were returning to those similar to the rising
limb.
2.3.2 Cluster Analysis of Samples
The cluster analysis separated samples from across the two wetlands into two significantly
different, homogeneous categories (Figure 2-4a; p= 0.0001 and a= 0.14). The two-cluster model
accounted for 58% of the variance in the data. Because of differences in isotopic content
(Figure 2-4c), the two clusters were classified as connected and disconnected.
Figure 2-4. Separation of Connected and Disconnected sites. A.) Dendrogram resulting from the
Hierarchal Cluster Analysis (HCA) on NMDS scores. Results separated disconnected, orange,
from connected, blue, sites. B.) Locations of connected and disconnected sites in each of three
synoptic samples. C-G.) Boxplots of floodwater constituents. Grey diamonds indicate mean
concentration in river water [Scott et al. 2014]. Statistical differences between classes are denoted
by alphabetic groups
16
Over the course of the storm, many sampling sites fluctuated between the connected and
disconnected classifications (Figure 2-4b). The greatest spatial variability occurred during the
rising limb, when sites in backwater and flow-through wetlands were primarily categorized as
disconnected and connected, respectively. During the peak, all were classified as connected, and
during the falling limb, sites once again separated into two distinct classes.
2.3.3 Comparison between the Connected and Disconnected Classes
Dissolved constituents varied significantly by connectivity class (Figure 2-4c-g). Disconnected
sites (enriched in δ18O) had significantly higher concentrations of DOC and Fe2+ and lower
concentrations of TDN. There was no significant difference in SRP by connectivity class.
Concentrations of nitrogen species varied by connectivity class and through the course of the
flooding event (Figure 2-5). During the rising limb of the flood, NO3- was significantly higher in
connected sites (Figure 2-5a), but there was no significant difference between classes during the
falling limb. In contrast, DON and the DOC:NO3- ratio were significantly higher in disconnected
sites over the course of the flood. There were no significant differences in NH4+ by connectivity
class over the course of the flood, though the direction of the effects was similar to that of NO3-.
Figure 2-5. Box plots of nitrogen speciation and the DOC:NO3- ratios throughout the flood,
separated by connected (blue) and disconnected (orange) sites as defined by the HCA (Fig 4).
Differences among groups are indicated by letters.
There was a strong, positive relationship between δ18O and DOC (Figure 2-6a), which suggests
residence time was a controlling factor on DOM accumulation in the two wetlands over the course
of the flood. There was separation between connected and disconnected classes, consistent with
the finding that δ18O and DOC were dominant variables in the NMDS and HCA. There was a
positive relationship between Fe2+ and SRP (Figure 2-6b), which suggests linkage by redox
process, consistent with the NMDS. The separation between classes is not as distinct, suggesting
site-specific factors were important for lower concentrations of Fe2+ and SRP, but the highest
concentrations of Fe2+ and SRP occurred in disconnected sites.
B.)
C.)
D.)
A.)
B.)
C.)
D.)
A.)
B.)
C.)
D.)
A.)
B.)
C.)
D.)
A.)
17
Figure 2-6. Relationship between A.) DOC and δ18O and B.) Fe and SRP at connected (blue)
and disconnected (orange) sites.
2.4 Discussion 2.4.1 Conceptual Model
Our results suggest that perirheic exchange, or surficial mixing in the floodplain, and redox
conditions controlled the biogeochemistry of floodwaters during the flood (Figure 2-7). Sites were
disconnected, or hydrologically isolated, when river stage was relatively low, resulting in limited
exchange of water and materials between the river and floodplain. Biogeochemical cycling was
fueled by internal, autochthonous materials, and long residence times led to reducing conditions,
which allowed for the accumulation of organic matter and SRP. In connected sites, local floodplain
water mixed with floodwater, forming a transitional zone where allochthonous materials were
transported from oxic floodwaters and fueled denitrification.
The zone of intrafloodplain mixing, at the interface between floodplain water and floodwaters, is
important for biogeochemical processing and can be termed the biogeochemical perirheic zone.
This is an extension of hydrologic concept of the perirheic zone originally proposed by Mertes
[1997], which is the mixing zone between new water (i.e., river water) and old water (e.g.,
groundwater, flow from small tributary streams, and precipitation). Within the ARB, the main
source of water is the river. However, our results show that the residence time in floodplain
flowpaths changes the composition of that river water, essentially creating floodplain water as a
distinct end-member.
Adding a time dimension to the perirheic concept more fully describes inundation hydrology in
large river floodplains, where backwater residence times are long enough to substantially alter
water quality and solute concentrations. For example, mean water residence time within a
relatively pristine Amazon River floodplain was estimated to be 5 months during the seasonal
flood pulse [Bonnet et al., 2008]. Similarly, in a recently restored floodplain along the Danube
River, Schagerl et al. [2009] found mean water residence times up to 30 days and maximum
residence times up to 90 days on an annual basis. In both cases, the characteristics of the floodplain
water changed over time due to depositional processes and biological activity [Hein et al., 2004;
Bourgoin et al., 2007]. Within the ARB, Scott et al. [2014] found significant differences between
hydrologically isolated sites and river water, indicating biogeochemical processing in the
18
floodplain. The results presented in our study confirm these findings, highlighting the observed
gradient from riverine floodwater to biogeochemically processed floodplain waters.
Figure 2-7. Conceptual sketch illustrating the differences between flow-through (connected) and
backwater (disconnected) locations within a floodplain.
2.4.2 Classification of Hydrologically Connected and Disconnected Locations
The results from the cluster analysis suggest both temporal and spatial variation in hydrologic
conditions (i.e., residence time of water) and biogeochemical processes (Figure 2- 4). We infer
that the connected and disconnected classes represent floodwater and floodplain water,
respectively, as defined by the conceptual model above. The chemistry associated with the
connected cluster most closely resembled river water. This is consistent with the observation that
all sites were classified as connected during the peak of the flood (Figure 4b), with floodwater
biogeochemistry. Conversely, there were both connected and disconnected sites during the rising
and falling limbs. In sites classified as disconnected, the enriched δ18O and elevated DOM suggest
relatively long residence times (Figures 4c-d), while lower TDN and higher Fe2+ suggest reducing
conditions (Figure 2-4e,g). Both longer residence times and reducing redox conditions are
indicative of processed floodplain water [Racchetti et al., 2011]. Although sites were classified in
a binary fashion, they exist on a continuum of residence time. As water moves across the
continuum, the accumulation of DOM and subsequent depletion electron acceptors indicate a shift
19
from allochthonous to autochthonous control of biogeochemical processes [Junk et al., 1989;
Tockner et al., 2000]. While processes were observed within a temporal pattern of variability, the
degree of hydrologic connectivity was greater than normal during the observed flood pulse.
2.4.3 Carbon, Nitrogen, and Phosphorus dynamics in the ARB
Carbon Accumulation in Backwaters
In the ARB, inundation hydrology and the residence time of water in the floodplain are large
controls on local dissolved organic matter. Increased residence time in backwater areas results in
increased DOC (Figure 2-6b). Similar increases of DOM in floodplain wetlands have been
observed in both the pristine Okavango Delta of Botswana [Cawley et al., 2012] and in a restored
floodplain along the Danube in Austria [Reckendorfer et al., 2013]. Increases in DOM can come
from flushing of organic soils [Fellman et al., 2009], leachate from litter-fall [Chow et al., 2013],
and algal biomass [Hein et al., 2004]. While further analysis and characterization is needed to
understand the sources and processing of organic matter in the ARB, the relationship between
DOC and δ18O was one of the strongest signals on the floodwater biogeochemistry and suggests
residence time is an important variable in floodplain DOM dynamics.
Nitrogen Retention and Removal
In floodplain wetlands, the combination of landscape position and inundation hydrology plays a
crucial role in local redox conditions at the sediment-water interface [Welti et al., 2012], thus
controlling internal cycling of nitrogen [Wolf et al., 2013]. During both the rising and falling limbs,
sites classified as disconnected had significantly lower NO3- and significantly higher DOC:NO3
-
ratios. This suggests that NO3- was limiting in the disconnected sites, consumed by both biotic
assimilation and, more importantly, denitrification [Taylor and Townsend, 2010; Scaroni et al.,
2011]. While denitrification rates were not explicitly measured in our study, other studies have
indicated similar results when comparing connected and disconnected floodplain wetlands
[Forshay and Stanley, 2005; Racchetti et al., 2011]. Therefore, while redox conditions in
backwater areas are possibly more favorable for nitrogen removal, the increased flux of NO3- in
less hydrologically isolated sites most likely leads to greater nitrogen loss through denitrification.
Scott et al. [2014] found that during the peak of the 2011 flood in the ARB, the entire basin remove
~37% of NO3- on a daily basis, with over 60% of NO3
- in backwater sites removed. However,
during low flow when there was limited exchange of materials (e.g., NO3-) between the river and
floodplain, the ARB removed less than 10% of NO3- on a daily basis.
Phosphorus Release
In sites classified as disconnected, it appears that SRP was mobilized through redox reactions with
ferric soils (Figure 2-6b). There are three main mechanisms for phosphorus release in wetlands
[Ardon et al., 2010]: mobilization of phosphorus as sediment (e.g. Surridge et al. [2012]),
mineralization of organic phosphorus (e.g. Noe and Hupp [2005]), and desorption of Fe2+-bound
ortho-phosphate (e.g. Aldous et al. [2005]). The last two processes have been found to be
responsible for the downstream flux of SRP from floodplain ecosystems [Baldwin and Mitchell,
2000]. Because flow velocity was low and therefore sediment transport was limited in our
disconnected sites, we conclude the most likely reason for increased SRP is desorption, similar to
Banach et al. [2009], who found that local redox conditions control phosphorus export.
20
2.4.4 Temporal and Spatial Variability in Floodplain Processes
Floodwater biogeochemistry was hysteretic with respect to redox conditions and connectivity over
the course of the flood (Figure 2-3), suggesting there are three dominant processes controlling
floodwater biogeochemistry: flushing (dominant during the rising limb), advective transport
(dominant during the peak), and organic matter accumulation (dominant during the falling limb).
These three processes occurred simultaneously throughout the flood, but the prevalence of each
process varied spatially and temporally depending on local conditions. Tockner et al. [1999, 2000]
suggested that a similar pattern of hysteresis occurs in the Danube floodplain, where the floodplain
transitions from a biotic phase (i.e., disconnected), to primary production phase (i.e., groundwater
connection), and a transport phase (i.e., surface connection). While groundwater flow in the ARB
is restricted [Newman and Keim, 2013], a similar continuum between allochthonous and
autochthonous control appears to occur throughout the flood period.
Because of the timing of the three synoptic events (Figure 2-1), it appears full hysteresis was
expressed from baseflow to baseflow before and after the event, because rising-limb conditions
indicated long residence time in backwater sites in the first sample period. As the ARB transitioned
from the rising limb to the peak of the hydrograph, there was a strong gradient in SRP and Fe2+
(Figure 2-2e-f and Figure 2-3). This suggests that the more hydrologically isolated sites in the
backwater wetland had developed reducing conditions resulting in the mobilizing of SRP during
iron reduction. As these sites became connected, flushing exported SRP. It is likely much of the
flushed SRP then sorbed to suspended sediment in the water column [Owens and Walling, 2002],
potentially increasing the redox potential of the flood waters. Once at the peak, there was much
less variability in water chemistry at all sites, suggesting dominance by river water and low
residence time for biogeochemical processing. During the falling limb, NO3- decreased and
dissolved organic matter increased to concentrations similar to the rising limb. However, iron and
SRP were much lower on the falling limb, suggesting that residence times were not long enough
to favor desorption of iron through redox processes.
2.4.5 Management Implications
In watershed management, it is critical to incorporate strategies and designs that maximize
pollutant removal (e.g., nitrogen through denitrification) while also minimizing the mobilization
of limiting nutrients (e.g., dissolved phosphorus). In the Lower Mississippi River Valley, it has
been proposed to use freshwater diversion and floodplain wetlands to reduce nitrogen loads to the
GOM [Mitsch et al., 2001]. While there is little scientific literature to guide the management of
the large floodplain systems, the Atchafalaya Basin is a highly studies system and can provide
insights that are applicable to other large rivers. This study highlights the need to manage redox
conditions to balance nitrogen removal with reactive solute accumulation. Here, the relatively
disconnected backwater wetland experienced much greater hysteresis in residence time and redox
conditions than the flow-through wetland, indicating a greater accumulation of reactive SRP and
DOM. Because all sites were classified as connected during the flood, it is probable that much of
the accumulated SRP and DOM were flushed downstream, thus increasing the risk of hypoxia to
the GOM. Tockner et al. [1999] found similar results in the restoration of a side arm, or backwater
floodplain wetland, of the Danube River. They suggested that to balance the nutrient retention and
export of DOM it was important to enhance hydrologic connectivity. In the case of the ARB, Scott
et al. [2014] found that greater floodplain-river exchange led to high NO3- removal rates through
perirheic exchange. However, BryantMason et al. [2013] suggested there was limited exchange
21
between WMUs and the Atchafalaya River after the hydrologically isolated WMUs were filled
during the rising limb. Therefore, by increasing hydrologic connectivity between the river and
adjacent floodplain, possibly by creating flow-through hydrologic conditions, NO3- loads to the
GOM could be further reduced.
2.5 Conclusions We examined how hydrologic regime affected floodplain biogeochemistry in the Atchafalaya
River Basin, a large floodplain. Both the flow-through and backwater wetlands experienced
connected and disconnected, or hydrologically isolated, conditions depending on the timing
relative to the flood peak. Sites were classified in a binary fashion over the course of the flood, but
they actually occurred across a gradient between biogeochemically processed floodplain water and
oxic floodwater. As inundated areas transitioned between hydrologically disconnected and
connected over the course of the flood, three main processes were evident: flushing of
hydrologically isolated areas during the rising limb, advective transport during the peak of the
flood, and organic matter accumulation during the falling limb. During the flushing phase,
disconnected sites showed elevated organic matter and SRP. As backwater regions became
hydrologically connected during peak flow, water chemistry throughout the basin was relatively
uniform because advective transport dominated. However, as sites began to disconnect during the
falling limb, floodplain residence times increased and organic matter accumulation began to
dominate hydrologically remote sites. Throughout the flood, disconnected conditions (e.g. areas
dominated by OM accumulation) only occurred in the backwater wetland, suggesting more
biogeochemical processing within the flow-through wetland. These results highlight the
importance of surficial mixing and biogeochemical exchange that occur in the perirheic zone,
especially when managing the export of redox sensitive materials such as nitrogen and phosphorus.
2.6 Acknowledgements This work was supported by NSF grant 1171363 to Virginia Polytechnic Institute and State
University, NSF grant 1141417 to Louisiana State University, and the Institute for Critical
Technology and Applied Science at Virginia Tech.
2.7 Works Cited Aldous, A., P. McCormick, C. Ferguson, S. Graham, and C. Craft (2005), Hydrologic regime
controls soil phosphorus fluxes in restoration and undisturbed wetlands, Restoration Ecology,
13(2).
Ardon, M., S. Montanari, J. L. Morse, M. W. Doyle, and E. S. Bernhardt (2010), Phosphorus
export from a restored wetland ecosystem in response to natural and experimental hydrologic
fluctuations, J. Geophys. Res.-Biogeosci., 115.
Baldwin, D. S., and A. M. Mitchell (2000), The effects of drying and re-flooding on the sediment
and soil nutrient dynamics of lowland river-floodplain systems: A synthesis, Regul. Rivers-Res.
Manage., 16(5), 457-467.
Banach, A. M., K. Banach, E. J. W. Visser, Z. Stepniewska, A. J. M. Smits, J. G. M. Roelofs, and
L. P. M. Lamers (2009), Effects of summer flooding on floodplain biogeochemistry in Poland;
implications for increased flooding frequency, Biogeochemistry, 92(3), 247-262.
Bechtold, J. S., R. T. Edwards, and R. J. Naiman (2003), Biotic versus hydrologic control over
seasonal nitrate leaching in a floodplain forest, Biogeochemistry, 63(1), 53-71.
22
Bonnet, M. P., et al. (2008), Floodplain hydrology in an Amazon floodplain lake (Lago Grande de
Curuaí), J. Hydrol., 349(1–2), 18-30.
Bourgoin, L. M., M.-P. Bonnet, J.-M. Martinez, P. Kosuth, G. Cochonneau, P. Moreira-Turcq, J.-
L. Guyot, P. Vauchel, N. Filizola, and P. Seyler (2007), Temporal dynamics of water and sediment
exchanges between the Curuaí floodplain and the Amazon River, Brazil, J. Hydrol., 335(1–2),
140-156.
BryantMason, A., Y. J. Xu, and M. A. Altabet (2013), Limited capacity of river corridor wetlands
to remove nitrate: A case study on the Atchafalaya River Basin during the 2011 Mississippi River
Flooding, Water Resources Research, 49(1), 283-190.
Cawley, K. M., P. Wolski, N. Mladenov, and R. Jaffe (2012), Dissolved Organic Matter
Biogeochemistry Along a Transect of the Okavango Delta, Botswana, Wetlands, 32(3), 475-486.
Cirmo, C. P., and J. J. McDonnell (1997), Linking the hydrologic and biogeochemical controls of
nitrogen transport in near-stream zones of temperate-forested catchments: a review. J. Hydrol.,
199(1), 88-120.
Chow, A. T., J. N. Dai, W. H. Conner, D. R. Hitchcock, and J. J. Wang (2013), Dissolved organic
matter and nutrient dynamics of a coastal freshwater forested wetland in Winyah Bay, South
Carolina, Biogeochemistry, 112(1-3), 571-587.
Clarke, K. R. (1993), Nonparametric multivariate analyses of changes in community structure,
Aust. J. Ecol., 18(1), 117-143.
Cory, R. M., and D. M. McKnight (2005), Fluorescence spectroscopy reveals ubiquitous presence
of oxidized and reduced quinones in dissolved organic matter, Environ. Sci. Technol., 39(21),
8142-8149.
Cory, R. M., M. P. Miller, D. M. McKnight, J. J. Guerard, and P. L. Miller (2010), Effect of
instrument-specific response on the analysis of fulvic acid fluorescence spectra, Limnology and
Oceanography-Methods, 8, 67-78.
Fellman, J. B., E. Hood, R. T. Edwards, and D. V. D'Amore (2009), Changes in the concentration,
biodegradability, and fluorescent properties of dissolved organic matter during stormflows in
coastal temperate watersheds, J. Geophys. Res.-Biogeosci., 114.
Ford, M., and J. A. Nyman (2011), Preface: an overview of the Atchafalaya River, Hydrobiologia,
658(1), 1-5.
Forshay, K. J., and E. H. Stanley (2005), Rapid nitrate loss and denitrification in a temperate river
floodplain, Biogeochemistry, 75(1), 43-64.
Harvey, J. W., M. H. Conklin, and R. S. Koelsch (2003), Predicting changes in hydrologic
retention in an evolving semi-arid alluvial stream, Advances in Water Resources, 26(9), 939-950.
Hein, T., C. Baranyi, W. Reckendorfer, and F. Schiemer (2004), The impact of surface water
exchange on the nutrient and particle dynamics in side-arms along the River Danube, Austria, Sci
Total Environ, 328(1-3), 207-218.
Hodges, J. L., Jr., and E. L. Lehmann (1963), Estimates of Location Based on Rank Tests, The
Annals of Mathematical Statistics, 34(2), 598-611.
Holm, S. (1979), A simple sequentially rejective multiple test procedure, Scandinavian Journal of
Statistics, g, 383-386.
23
Hughes, D. A. (1980), Floodplain inundation - processes and relationships with channel discharge,
Earth Surf. Process. Landf., 5(3), 297-304.
Hunt, R. J., T. D. Bullen, D. P. Krabbenhoft, and C. Kendall (1998), Using stable isotopes of water
and strontium to investigate the hydrology of a natural and a constructed wetland, Ground Water,
36(3), 434-443.
Junk, W. J., P. B. Balyley, and R. E. Sparks (1989), The flood pulse concept in river-floodplain
systems., Can. Spec. Publ. Fish. Aquat. Sci., 106, 110-127.
Kaller, M. D., W. E. Kelso, B. T. Halloran, and D. A. Rutherford (2011), Effects of spatial scale
on assessment of dissolved oxygen dynamics in the Atchafalaya River Basin, Louisiana,
Hydrobiologia, 658(1), 7-15.
Kaller, M. D., R. F. Keim, B. L. Edwards, A. R. Harlan, T. Pasco, W. E. Kelso, and D. A.
Rutherford. Aquatic vegetation mediates the relationship between hydrologic connectivity and
water quality in a managed floodplain, Atchafalaya River, Louisiana, USA. In review,
Ecohydrology
Kendal, C., and J. McDonnel (1998), Isotope Tracers in Catchment Hydrology, Elsevier, The
Neatherlands.
Kendall, C., and T. B. Coplen (2001), Distribution of oxygen-18 and deuterium in river waters
across the United States, Hydrological Processes, 15(7), 1363-1393.
Kruskal, J. B. (1964), Multidimensional scaling: a numerical method, Psychometrika, 29, 115-129.
Lane, R. R., J. W. Day, and B. Thibodeaux (1999), Water quality analysis of a freshwater diversion
at Caernarvon, Louisiana, Estuaries, 22(2A), 327-336.
McCune, B., R. Rosentreter, J. M. Ponzetti, and D. C. Shaw (2000), Epiphyte habitats in an old
conifer forest in Western Washington, USA, Bryologist, 103(3), 417-427.
McGuire, K. J., J. J. McDonnell, M. Weiler, C. Kendall, B. L. McGlynn, J. M. Welker, and J.
Seibert (2005), The role of topography on catchment-scale water residence time, Water Resources
Research, 41(5).
McJannet, D., J. Wallace, R. Keen, A. Hawdon, and J. Kemei (2012), The filtering capacity of a
tropical riverine wetland: II. Sediment and nutrient balances, Hydrological Processes, 26(1), 53-
72.
Mertes, L. A. K. (1997), Documentation and significance of the perirheic zone on inundated
floodplains, Water Resour. Res., 33(7), 1749-1762.
Mielke, P. W., and K. J. Berry (2001), Permutation Methods: A Distance Function Approach. .
Misiti, T. M., M. G. Hajaya, and S. G. Pavlostathis (2011), Nitrate reduction in a simulated free-
water surface wetland system, Water Research, 45(17), 5587-5598.
Mitsch, W. J., J. W. Day, J. W. Gilliam, P. M. Groffman, D. L. Hey, G. W. Randall, and N. M.
Wang (2001), Reducing nitrogen loading to the Gulf of Mexico from the Mississippi River Basin:
Strategies to counter a persistent ecological problem, Bioscience, 51(5), 373-388.
Newman, A. E., and R. F. Keim (2013), Mesoscale connectivity through a natural levee,
Hydrology and Earth System Sciences, 17(2), 691-704.
Noe, G. B., and C. R. Hupp (2005), Carbon, nitrogen, and phosphorus accumulation in floodplains
of Atlantic Coastal Plain rivers, USA, Ecological Applications, 15(4), 1178-1190.
24
Noe, G. B., C. R. Hupp, and N. B. Rybicki (2013), Hydrogeomorphology Influences Soil Nitrogen
and Phosphorus Mineralization in Floodplain Wetlands, Ecosystems, 16(1), 75-94.
Oksanen J., F. G. Blanchet, R. Kindt, P. Legendre, P.R. Minchin, R. B. O'Hara, G.L. Simpson, P.
Solymos, M.H.H.Stevens, H. Wagner (2013). vegan: Community Ecology. Package. R package
version 2.0-10. http://CRAN.R-project.org/package=vegan
Orloci, L. (1967), An Agglomerative method for classification of plant communities, Journal of
Ecology, 55, 208-212.
Owens, P. N., and D. E. Walling (2002), The phosphorus content of fluvial sediment in rural and
industrialized river basins, Water Research, 36(3), 685-701.
Poole, G. C. (2010), Stream hydrogeomorphology as a physical science basis for advances in
stream ecology, Journal of the North American Benthological Society, 29(1), 12-25.
R Development Core Team (2012). R: A language and environment for statistical computing. R
Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL http://www.R-
project.org/
Racchetti, E., M. Bartoli, E. Soana, D. Longhi, R. R. Christian, M. Pinardi, and P. Viaroli (2011),
Influence of hydrological connectivity of riverine wetlands on nitrogen removal via denitrification,
Biogeochemistry, 103(1-3), 335-354.
Reckendorfer, W., A. Funk, C. Gschopf, T. Hein, and F. Schiemer (2013), Aquatic ecosystem
functions of an isolated floodplain and their implications for flood retention and management, J.
Appl. Ecol., 50(1), 119-128.
Reddy, K. R., and W. H. Patrick (1984), Nitrogen transformations and loss in flooded soils and
sediments, Crc Critical Reviews in Environmental Control, 13(4), 273-309.
Schagerl, M., I. Drozdowski, D. G. Angeler, T. Hein, and S. Preiner (2009), Water age - a major
factor controlling phytoplankton community structure in a reconnected dynamic floodplain
(Danube, Regelsbrunn, Austria), Journal of Limnology, 68(2), 274-287.
Scaroni, A., J. Nyman, and C. Lindau (2011), Comparison of denitrification characteristics among
three habitat types of a large river floodplain: Atchafalaya River Basin, Louisiana, Hydrobiologia,
658(1), 17-25.
Scott, D. T., R. F. Keim, C. N. Jones, B. Edwards, D. E. Kroes, J. R. White, and J. A. Nyman
(2014), Floodplain biogeochemical processing of floodwaters during the Mississippi River flood
of 2011, Water Resources Research. In Press.
Surridge, B. W. J., A. L. Heathwaite, and A. J. Baird (2012), Phosphorus mobilisation and
transport within a long-restored floodplain wetland, Ecol Eng, 44, 348-359.
Taylor, P. G., and A. R. Townsend (2010), Stoichiometric control of organic carbon-nitrate
relationships from soils to the sea, Nature, 464(7292), 1178-1181.
Tockner, K., F. Malard, and J. V. Ward (2000), An extension of the flood pulse concept,
Hydrological Processes, 14(16-17), 2861-2883.
Tockner, K., D. Pennetzdorfer, N. Reiner, F. Schiemer, and J. V. Ward (1999), Hydrological
connectivity, and the exchange of organic matter and nutrients in a dynamic river-floodplain
system (Danube, Austria), Freshw. Biol., 41(3), 521-535.
25
Tye, R. S., & Coleman, J. M. (1989). Evolution of Atchafalaya lacustrine deltas, south-central
Louisiana. Sedimentary Geology, 65(1), 95-112.
Ward, J. H. (1963), Hierarchical grouping to optimise an objective function, Journal of the
American Statistical Association, 58, 236-244.
Welti, N., E. Bondar-Kunze, G. Singer, M. Tritthart, S. Zechmeister-Boltenstern, T. Hein, and G.
Pinay (2012b), Large-scale controls on potential respiration and denitrification in riverine
floodplains, Ecol Eng, 42, 73-84.
Wolf, K. L., G. B. Noe, and C. Ahn (2013), Hydrologic Connectivity to Streams Increases Nitrogen
and Phosphorus Inputs and Cycling in Soils of Created and Natural Floodplain Wetlands, Journal
of Environmental Quality, 42(4), 1245-1255.
26
CHAPTER 3. BIOGEOCHEMICAL PROCESSING OF
DISSOLVED ORGANIC MATTER AND NITROGEN IN
FLOODWATERS ACROSS A GRADIENT OF RIVER-
FLOODPLAIN CONNECTIVITY WITHIN THE
ATCHAFALAYA RIVER FLOODPLAIN
C. Nathan Jones, Durelle T. Scott, Brandon L. Edwards, Richard. F. Keim
Submitted: anticipated, August 2015
To: Water Resources Research
Status: Draft
Abstract In large river systems, biogeochemical processing in floodplains can reduce the downstream flux
of contaminants. Thus, careful management of river-floodplain connectivity in large river systems
provides a potentially valuable conduit for mitigating downstream effects of harmful solutes. In
this study, we examined dissolved organic matter and nitrogen across spatial and temporal
gradients of river-floodplain connectivity in a highly connected region of the Atchafalaya River
floodplain. We completed three synoptic sampling campaigns across the rising limb, peak, and
falling limb of the 2013 annual flood pulse. Inundation hydrology and organic matter
biogeochemistry were characterized using hydrologic tracers and spectral indices. Samples were
collected in secondary canals at locations representing a gradient of river-floodplain connectivity.
During the rising and falling limb, isotopic ratios were relatively uniform across the sampling
gradient, suggesting the main source of water in the secondary channels was riverine and there
was little interaction with hydrologically isolated backwaters. In contrast, water isotopes
experienced much great variation during the peak, indicating enhanced river-floodplain
connectivity and biogeochemical interaction between riverine and backwater sources. While
gradients in redox sensitive solutes were observed, relatively low DOC to NO3- ratios suggest
denitrification was not NO3- limited. Further, increases in both dissolved inorganic carbon and the
fluorescence index over the course of the flood suggest that subsidies from floodwater and seasonal
increases in temperature enhanced biogeochemical processing within secondary channels.
3.1 Introduction Most of the world’s large rivers have been significantly impacted by anthropogenic modifications
[Vitousek et al., 1997]. Managed hydrology (e.g. Leopold, 1968), flood control (e.g. Sparks et al.
[1998]), and channelization (e.g. Galat et al. [1998]) have largely reduced river-floodplain
connectivity in many large river systems [Tockner and Stanford, 2002]. River-floodplain
connectivity drives water mediated exchange of material and facilitates retention and removal of
reactive solutes such as nitrogen and phosphorus [Pringle, 2003; Tockner and Stanford, 2002].
Thus, the loss or restriction of hydrologic connectivity between large rivers and their floodplains
exacerbates the downstream flux of harmful solutes typically associated with environmental
degradation (e.g. nitrogen and eutrophication; Rabalais et al. [2002]).
27
One potentially attractive solution to mitigate the flux of harmful solutes is managing hydraulic
control structures to optimize, or in some cases reestablish, river-floodplain connectivity in large
river systems (e.g. Lane et al., [1999]; James et al., [2008]; Welti et al. [2012]). Because the rate
of biogeochemical processing (e.g. denitrification) in floodplains often exceeds the rate of riverine
solute delivery [Forshay and Stanley, 2005], solute retention is often transport limited [Racchetti
et al., 2011; Jones et al., 2014] and can be increased through enhanced river-floodplain
connectivity [James et al., 2008]. Increased river-floodplain connectivity significantly reduces
downstream NO3- loads [Scott et al., 2014].
While recent evidence suggests managing river-floodplain connectivity is a viable avenue to
reduce downstream flux of solutes (e.g. Mitsch et al., [2001]), there is sparse scientific literature
to guide managers and decision makers. Largely, this is due to the complexity and heterogeneity
of ecosystem processes that occur in large river floodplains [Hughes, 1980; Scown and Thoms,
2015]. Within the Atchafalaya River Basin, a highly managed large river floodplain, previous
research has shown that floodwater residence times strongly influence floodplain biogeochemistry
and thus downstream water quality (e.g. Lambou et al. [1983]; Sabo et al. [1999], Shen et al.
[2012]; Jones et al. [2014]; Scott et al. [2014]). Interestingly, these studies, along with studies in
other large river floodplains (e.g. Mladenov et al. [2005]; Fellman et al. [2013]; Hedges et al.
[2000]; Zurbrugget al. [2013]), suggest that floodplains are typically sources of dissolved organic
matter (DOM) during periods of river-floodplain connectivity. This is of particular interest,
because DOM cycling fuels biogeochemical processing in floodplains (e.g. Hein et al. [2003]) and
in turn imparts a strong signal on downstream water quality (e.g. Shen et al. [2012]).
To better understand DOM cycling during periods of river-floodplain connectivity, and the
associated effect on biogeochemical processing of floodwaters, we conducted large synoptic
sampling campaigns across a hydrologically connected region of the Atchafalaya River Basin
(ARB) over an annual flood pulse. Specifically, the goal of our study was to characterize temporal
and spatial variation in DOM quality over the course of the annual flood and infer the effect of
DOM and river-floodplain connectivity on biogeochemical processing of nitrogen.
3.2 Methods 3.2.1 Site Description
This study was conducted in the Atchafalaya River Basin (ARB) in southern Louisiana, USA. The
Atchafalaya River is the fifth largest river in North America. It starts at the convergence of the
Red and Mississippi Rivers, where it receives 30% of the combined flows of both through the US
Corp of Engineers Old River Control Structure. The river is constrained by guide levees over
approximately the upper half of its length and then flows through an interconnected system of
bayous, canals, and floodplain lakes to the Gulf of Mexico (GOM). In general, the upper reach is
incised and largely disconnected from much of the floodplain except during very large events
[Kroes and Kraemer, 2013]. The lower basin is characterized by a network of lateral secondary
channels which connect riverine flow to the floodplain. The ARB is contained by flood protection
levees on the east and west, and has internal hydraulic control structures that subdivide the basin
into 13 water management units. The basin transitions downstream from agriculture and
bottomland hardwood forest to cypress tupelo swamp. Much of the southern portion of the basin
is near permanently inundated yet hydrologically isolated from downstream transport during
normal flow conditions [Jones et al., 2014]. During annual flood pulses, up to 50% of flow routed
28
into the ARB is contained within backwater regions [Scott et al., 2014; Kroes et al., 2015], but it
unclear how much river-floodplain connectivity and exchange occurs after the initial filling of
backwater areas in each WMU [BryantMason et al., 2013].
Figure 3-1. Sampling locations within the Atchafalaya River Basin in southern Louisiana, USA.
29
Samples were collected in the Flat Lake WMU, a hydrologically connected region in the east-
central portion of the lower ARB (Figure 3-1). The WMU is bordered by the Atchafalaya River
on the west and the Intercoastal Waterway (ICWW) on the East, both of which flow southward
towards to the GOM and are connected by a series of canals, bayous, and floodplains lakes
(Figure 3-1). Collectively, these secondary channels act as preferential flowpaths of floodwaters
during annual flood events [Kroes and Kraemer, 2013], facilitating enhanced river-floodplain
connectivity between the main channel and hydrologically isolated backwater regions. Generally,
as floodwaters are routed from the main channel into backwater regions during the rising limb,
areas previously hydrologically isolated or disconnected become connected forming areas of
surficial mixing, or perirheic zones (e.g. Jones et al., [2014]; Mertes, [1997]). During the falling
limb, water slowly drains back to main channel allowing for prolonged hydrologic connectivity
between secondary channels and backwater regions, as illustrated by the counterclockwise
hysteresis between main channel and floodplain stage in Figure 3-2b [Hughes, 1980; Rudorff et
al., 2014]. However, because of the heterogeneous nature of the floodplain surface and the
associated inundation hydrology, secondary channels experience a variety of river-floodplain
connectivity conditions [Jones et al., 2014]. Therefore, to characterize the spatial and temporal
variation of hydrologic connectivity within this system, we chose 16 sampling locations to
represent a gradient of river-floodplain connectivity from the main channel to hydrologically
isolated backwater region. Endmembers included the Atchafalaya River (e.g. most connected) and
Murphy Lake (e.g. most disconnected), as defined by previous studies (e.g. Scott et al., [2014];
Hupp et al., [2008]). Similar to Scott et al., [2014], endmember selection is confirmed by the
gradient in isotopic content during the rising limb synoptic sampling campaign (Figure 3-3b).
Figure 3-2. A.) Hydrograph displaying the 2013 flood pulse experienced in the Atchafalaya River
main channel (USGS Gage 07381490). The blue circle, orange square, and red diamond denote
the rising limb, peak, and falling limb synoptic sampling campains, repsectively. B.) Plot
displaying counterclockwise hysterses exerpienced between the main channel flow and floodplain
stage (USGS Gage 073815963). Blue, orange, and red arrows correspond to rising limb, peak, and
falling limb segments of the hydrograph.
30
3.2.2 Synoptic Sampling Design and Analysis
Sample Collection and Handling. Three synoptic sampling campaigns were conducted over the
course of the 2013 flood pulse: the rising limb (February), the peak (June), and the falling limb
(August, Figure 3-2a). These synoptic sampling campaigns mirror our sampling of the 2011
Atchafalaya Flood (e.g. Jones et al. 2014], but examines a more hydrologically connected portion
of the ARB. At each site, surface water samples were collected at a depth of 0.5 m and filtered
through a 0.45 µM Geotech capsule filter using a peristaltic pump. Samples were stored in 250 mL
plastic bottles for nutrients; 40 mL precombusted amber glass vials for dissolved organic carbon
(DOC), dissolved inorganic carbon (DIC), carbon isotopes, and fluorescence analysis; and 40 mL
glass vials with no head space for water isotope analysis. All samples were stored on ice and
transported to the laboratory within 24 hr. Once at the lab, nutrient samples and carbon samples
were stored at -20⁰C and 5⁰C, respectively, until analysis.
Water Isotope Analysis. We utilized water isotopes as hydrologic tracers to better characterize
mixing patterns over the course of the flood. Isotopes ratios of both of oxygen and hydrogen (δ18O
and δD) were measured using a Picarro Water Isotope Analyzer. Deuterium excess (d), a measure
of evaporative fractionation in water isotopes, was estimated by the deviation from the meteoric
water line [Kendall and McDonnell, 2006]. The isotopic composition at each sampling location is
a mixture of two water sources: locally derived rainwater and regionally derived floodwater [Scott
et al., 2014]. Because the Mississippi and Red Rivers drain a large portion of the United States,
latitudinal gradients in precipitation strongly influence the floodwater signal and are responsible
for depleted values of δ18O and δD [Kendall and Coplen, 2001]. In contrast, locally derived
rainwaters that are stored in backwater regions are highly enriched because of preferential
evaporation of lighter isotopes. Because of the differences between the regional water and locally
derived water, water isotopes are an ideal tracer that can be utilized to examine sources and
residence time of water within the Atchafalaya Basin [Jones et al., 2014; Scott et al., 2014].
Reactive Solute Analysis. We characterized dissolved N, DOC, and DIC to illustrate spatial and
temporal variation in biogeochemical processing over the course of the flood. NO3-concentrations
were measured through ion chromatography using a Dionex ICS-300 analyzer. Dissolved organic
carbon (DOC), dissolved inorganic carbon (DIC), and total dissolved nitrogen (TDN) were
measured using high temperature oxidative combustion using a Shimadzu TOC-CPH analyzer.
Total Kjeldahl nitrogen (TKN) was estimated by subtracting NO3- from TDN and assumed to
represent the sum of ammonia (NH4+) and dissolved organic nitrogen (DON). Here, NO3
-
represents the mobile form of N (Figure 3-3a), TKN represents the reactive or labile proportion of
dissolved N, and the ratio of DOC to NO3- is an indicator of NO3- limitation for denitrification
and other biogeochemical processes (e.g. Taylor and Townsend [2010]).
Carbon Isotope Analysis. To better understand DOM processing, and its effect on
biogeochemical processing of floodwater N, we measured stable isotopes of dissolved carbon
(δ13C-DIC and δ13C-DOC) and a variety of spectral indices over the course of the flood. δ13C-DOC
and δ13C-DIC were measured using an Aurora TOC analyzer and Picarro carbon isotope analyzer
in series, where the Aurora TOC analyzer oxidized the dissolved carbon to CO2 and the picarro
measured isotopic ratios through ring-down laser spectrometry. Both δ13C-DOC and δ13C-DIC are
tracers of dissolved carbon sources. δ13C-DOC is typically utilized to distinguish differences
between DOM parent material, where C3 and C4 plant’s isotopic composition range from -23 to -
31
35‰ and -10 to -14‰, respectively [Cerling et al, 1993]. Within the Atchafalaya, potential sources
of DOC include flushing of organic soils (e.g., Fellman et al. [2009]), leachate from litterfall (e.g.
Chow et al. [2013]), and decomposition of algal biomass (e.g. Hein et al. [2003]) [Jones et al.,
2014], all of which have overlapping isotopic signatures [France, 1996]. Sources of DIC could
potentially include the byproducts from respiration (e.g. Welti et al. [2012]), carbonate dissolution
(Raymond and Cole [2003]), and atmospheric exchange of CO2 (e.g. Bianchi et al. [2013]).
Because atmospheric CO2- isotopic composition is nearly constant at -8‰ [Craig, 1953],
metabolism of plant derived DOM (low δ13C) depletes δ13C-DIC while photosynthesis typically
increases δ13C-DIC by introducing enriched atmospheric CO2 [Finlay, 2003].
Spectral Indices. We utilized the spectral indices based on carbon fluorescence and absorbance
to better understand differences in DOM composition and quality over the course of the flood.
Carbon fluorescence and absorbance was measured using a Horiba Jobin Yvon Fluoromax-4 and
a spectrometer following methods outlined by Corey et al. [2010]. Specific UV Absorbance at
254nm (SUVA) is a measure of aromaticity, where large values represent highly aromatic
molecules and are typically assumed to be of terrestrial origin [Weishaar et al., 2003]. Across the
US stream network, SUVA values in rivers and streams typically range from 2-5 L mg-1 m-1, where
values become more homogenous with larger watershed area because of increased residence time
and processing [Creed et al., 2015]. Similarly, we examined the fluorescent component of DOM
across the flood pulse using fluorescence spectrometry and the associated fluorescence index (FI)
and redox index (RI). The FI is used to examine the source of fulvic acid component of DOM,
where microbial derived sources of DOM have relatively high values (e.g. 1.74 in the McMurdo
Dry Valleys) and vegetative sources have relatively low values (e.g. 1.24 in the Suwannee River)
[Hood et al., 2003]. The RI describes the redox state of the fluorescent fraction of DOM and was
calculated using excitation-emission matrices and the Cory and McKnight [2005] Parallel Factor
Analysis (PARAFAC) model [Miller and McKnight, 2010]. Specifically, the RI was estimated as
the ratio of oxidized and reduced quinones identified using the Cory and McKnight [2005]
PARAFAC model, where larger RI values suggest oxidized DOM [Jones et al., 2014].
Statistical Analysis. To infer temporal variation in floodplain biogeochemistry over the flood, a
pairwise Wilcoxon test was used to test for differences in individual chemical analytes across the
three synoptic sampling events [Hodges and Lehmann, 1963]. All Wilcoxon tests were used in
conjunction with a Bonferroni correction to control for family wise error rates [Benjamini and
Hochberg, 1995], and significance was evaluated at p=0.05.
3.3. Results 3.3.1 Hydrologic Tracer of Water
The isotopic signal experienced across the flood was similar to the variation characterized in
previous flood events in other portions of the ARB (e.g. Scott et al. [2014]; Jones et al. [2014]; M
Kaller et al. [2015]): the rising and falling limb experienced relatively enriched values of δ18O and
δD, while the peak experienced more depleted values (Figure 3-2). The isotopic composition at
each sampling location is a mixture of two water sources: locally derived rainwater and regionally
derived floodwater [Scott et al., 2014]. Because the Mississippi and Red Rivers drain a large
portion of the United States, latitudinal gradients in precipitation strongly influence the floodwater
signal and are responsible for depleted values of δ18O and δD [Kendall and Coplen, 2001]. In
contrast, locally derived rainwater are highly enriched because of preferential evaporation of
32
“light” isotopologues [Kendall and McDonnell, 2006]. Because of the differences between the
regional water and locally derived water, water isotopes are an ideal tracer that can be utilized to
examine sources and residence time of water within the Atchafalaya Basin [Jones et al., 2014;
Scott et al., 2014].
Figure 3-3. Isotopic content of samples collected over the course of the flood.Isotopic content of
samples collected over the course of the flood. Blue, orange, and red points represent samples from
the rising limb, peak, and falling limb of the hydrograph, respectively. In boxplots, black and
brown points correspond to backwater swamp and river endmembers, and statistical differences
are denoted by alphanumeric groups.
3.3.2 Reactive Solute Characterization
In this analysis, we characterized observation of NO3-, TKN, and DOC:TKN ratios to demonstrate
biogeochemical processing experienced within the floodplain over the course of the flood. There
were significantly higher concentrations of nitrate during the peak and falling limb of the
hydrograph. Further, during all three synoptic events, higher NO3- concentrations were in river
water than backwater endmembers, which suggests there is a redox gradient associated with the
degree of hydrological connectivity (e.g. Scott et al. [2014]). Similarly, TKN values were
significantly higher during the peak than rising limb, but the falling limb did not show significant
differences between the other two synoptics. Increased connectivity to backwater and the flushing
of accumulated solutes is likely the cause of the increased TKN values and suggests a pulse of
labile solutes (e.g. NH4+ and DON) occurred during the peak (e.g. BryantMason et al. [2013]).
Finally, the DOC to NO3- ratio was significantly less during the peak and falling limb of the
hydrograph, suggesting relative NO3- limitation of denitrification during the rising limb of the
hydrograph (e.g. Jones et al. [2014]).
3.3.3 DOM Characterization
Previous research has shown that residence time within backwater regions of the ARB strongly
influences DOC concentrations (e.g. Lambou et al. [1983]; Shen et al. [2012]; Jones et al. [2014];
Scott et al. [2014]), but little is known about the DOM composition and its effect on associated
biogeochemical processes. Here, we utilized a combination of DOC/DIC ratios (Figures 5a-b),
stable isotope of DOC and DIC (Figures 5c-d), and spectral indices of DOM (Figure 3-6) to further
characterize the fate and transport of DOM within the ARB.
33
Figure 3-4. A.) NO3
-,B.) TKN, and C.) DOC:TKN ratio over the course of the flood. Black and
brown points represent river and backwater swamp endmembers.
DOC concentrations generally decreased over the course of the flood, where rising limb
concentrations were significantly higher than the peak and falling limb concentrations
(Figure 3-5a). Potential sources of DOC include flushing of organic soils (e.g., Fellman et al.
[2009]), leachate from litterfall (e.g. Chow et al. [2013]), and algal biomass (e.g. Hein et al.
[2003]). Inversely, DIC concentrations increased over the course of the flood, where all three
synoptic samplings were significantly different (Figure 3-5b). Sources of DIC could potentially
include the byproducts from respiration (e.g. Welti et al. [2012]), carbonate dissolution (Raymond
and Cole [2003]), and atmospheric exchange of CO2 (e.g. Bianchi et al. [2013]).
In this analysis, we characterized observation of NO3-, TKN, and DOC:TKN ratios to demonstrate
biogeochemical processing experienced within the floodplain over the course of the flood. There
34
were significantly higher concentrations of nitrate during the peak and falling limb of the
hydrograph. Further, during all three synoptic events, higher NO3- concentrations were in river
water than backwater endmembers, which suggests there is a redox gradient associated with the
degree of hydrological connectivity (e.g. Scott et al. [2014]). Similarly, TKN values were
significantly higher during the peak than rising limb, but the falling limb did not show significant
differences between the other two synoptics. Increased connectivity to backwater and the flushing
of accumulated solutes is likely the cause of the increased TKN values and suggests a pulse of
labile solutes (e.g. NH4+ and DON) occurred during the peak (e.g. BryantMason et al. [2013]).
Finally, the DOC to NO3- ratio was significantly less during the peak and falling limb of the
hydrograph, suggesting relative NO3- limitation of denitrification during the rising limb of the
hydrograph (e.g. Jones et al. [2014]).
Figure 3-5. Dissolved organic and inorganic carbon concentration and isotopic content across the
flood. Black and brown points represent river and backwater swamp endmembers.
We utilized the stable isotopes of DOC and DIC to further characterize the sources and fate of
DOM across the sampling events. The rising limb had statistically lower δ13C-DOC than the peak
and falling limbs. δ13C-DOC values during the rising limb are similar to previously measured in
the lower Mississippi Basin (-28‰; Wang et al. [2004]), and enrichment of δ13C-DOC from the
rising limb to peak and falling limb suggest a shift in DOM sources, potentially shifting from
aquatic sources to greater influence of DOM derived in the floodplain [Fellman et al., 2013;
Zurbrugg et al., 2013]. Inversely, δ13C-DIC decreased over the course of the flood, where all three
synoptic samplings were statistically different. Further, there appears to have been a (decreasing)
gradient in δ13C-DIC from floodwaters to backwater across all three events. Rising limb δ13C-DIC
values are similar to typical river water draining from weathered soils (e.g. -10 to -12 ‰, Mook
35
and Tan [1991]). Depletion of δ13C-DIC during the peak and falling limb synoptics, and also along
the gradient from river to backwater sites, suggest an increase in biogeochemical processing across
both spatial and temporal gradients (e.g. Wachniew [2006]).
Figure 3-6. Specific ultra violet absorbance (SUVA), fluorescence index (FI), and redox index
(RI) from across the flood pulse.
Finally, we utilized SUVA, FI, and RI indices to characterize the DOM composition and quality
over the course of the flood. We observed SUVA in the typical range for flowing waters (e.g. 2-5
L mg-1 m-1) during the rising and falling limbs. However, similar to other observation in large river
floodplains (e.g. Mladenov et al. [2005]), SUVA vales were much greater at the peak indicating a
greater dominance of increase in terrestrial derived DOM likely due to the enhanced river-
floodplain connectivity.
36
Similarly, we examined the fluorescent component of DOM across the flood pulse using
fluorescence spectrometry and FI and RI indices (Figures 6b-c). Our FI observations fall in the
range typically observed in large rivers (e.g., 1.4-1.5; McKnight et al. [2001]), but increased over
the course of the flood suggesting an increase in microbial derived DOM (Figure 3-6b), suggesting
an increase in heterotrophic activity over the course of the flood. While we observed relatively
little temporal change in RI over the course of the flood, there appears to be a consistent separation
between the river and backwater endmembers (Figure 3-6c).
Figure 3-7. Linear relationships between A.) SUVA and δD, B.) NO3
- and DOC concentrations,
and C.) FI and DIC concentration.
3.4 Discussion 3.4.1 River-floodplain connectivity
Within our study area, secondary channels were highly connected to the river over the course of
the flood pulse (Figure 3-8). During the rising and falling limb, we observed minimal variation in
δD (Figure 3-3b), suggesting the river was the dominant source of water at the sampling locations.
However, during the peak, we observed greater variation and significantly enriched δD, suggesting
mixing between regional floodwaters and processed floodplain water. Connectivity across
hydrologically isolated portions of the floodplain, both permanently and seasonally inundated
areas, is further confirmed by significant increase in SUVA and TKN during the peak (Figures 4b
and 6a), suggesting a greater influence of allochthonous DOM likely originating from previously
disconnected locations. Similar patterns have been observed in other large rivers, where the flood
pulse resulted in an export of DOM to fuel downstream biogeochemical processing (e.g. the
Danube in Europe, Reckendorfer et al. [2013]; the pristine Okavango Delta in Batswana, Mladenov
et al. [2005]; and the Amazon River in South America, Junk et al. [1989]). While this pulse of
potentially reactive DOM fuels biogeochemical processing, it can also lead to excess emissions of
CO2 [Bianchi et al., 2013].
3.4.2 Biogeochemical Processing of Nitrogen within the Floodplain
Previous studies have observed NO3- limited denitrification within the ARB (e.g. Jones et al.
[2014]) and across other medium to large river floodplains (e.g. Forshay and Stanley [2005];
Racchetti et al. [2011]). Largely, this is due to a lack of hydrologic exchange and perirheic mixing
37
between rivers and their adjacent floodplains [James et al., 2008; Jones et al., 2014]. In contrast,
we did not observe NO3- limitation in our study area, likely due to the high degree of hydrologic
connectivity at the sampling locations. There was a significant linear relationship between DOC
and NO3- (R2=0.52, p<0.05, Figure 3-7b), and relatively low DOC: NO3
- elemental ratios
(Figure 3-4c), both suggesting denitrification was not NO3- limited over the course of the flood
pulse [Taylor and Townsend, 2010]. Further, a redox gradient from the river to backwater
endmembers is evident in both the RI and NO3- data suggesting NO3
- removal through
denitrification (Figure 3-4a and 3-6c). Cumulatively, these measures, along evidence from
previous studies at this site (e.g. Scott et al. [2014]), suggest elevated nitrogen removal due to
increased river-floodplain connectivity. However, Scaroni et al. [2010] found relatively
homogenous (potential) denitrification rates across the ARB and indicated organic matter lability
was another key variable that could limit denitrification within the ARB.
Figure 3-8. Conceptual model of swamp channel hydrology and biogeochemistry evolution over
the course of an annual flood pulse. A.) Secondary channel hydrograph components, B.) DOM
38
flux, and C.) biogeochemical processing. D.) River-floodplain connectivity during the rising limb,
peak, and falling limb of a typical flood pulse.
3.4.3 Temporal variation in Dissolved Organic Matter
Conceptually, we separate DOM into two distinct categories: highly processed allochthonous
DOM and microbial derived autochthonous DOM. The sources of allochthonous DOM include
both main channel and hydrologically disconnected regions of the floodplain and is characterized
by high aromaticity and low lability. In contrast, the autochthonous DOM is derived from internal
cycling within the secondary channel and is largely composed of labile aliphatic compounds.
Because the Atchafalaya River is composed of the Mississippi and Red rivers, riverine DOM
quality and content should be relatively uniform across the sampling period because of increased
residence time and processing associated with transport in large river systems [Creed et al., 2015;
Rosemond et al., 2015]. Duan et al. [2007] noted that this is particularly true in the lower
Mississippi River, where much of the river is hydrologically disconnected from its floodplain, as
opposed to other large systems such as the Amazon (e.g. Hedges et al. [2000]) and the Zambezi
(e.g. Zurbrugg et al. [2013]). Therefore, as river water is routed into the ARB, conceptualized as
a control volume, any changes to DOM composition are likely do to internal cycling (e.g.
production of autochthonous DOM) and river-floodplain connectivity (e.g. exchange of
allochthonous DOM with the floodplain). SUVA and δ13C-DOC signals (Figures 5c and 6a) both
experience significant changes over the course of the flood, indicating changes in a shift of
dominance of river derived DOM to floodplain derived DOM. Alone, the shift in δ13C-DOC only
suggests a shift in DOM source, largely because of overlap of isotopic content in potential
endmembers [France, 1996]. However, the simultaneous and significant increase in SUVA
suggest allochthonous DOM was the dominant source during the peak of the hydrograph.
It also appears heterotrophic activity and production of autochthonous DOM increased over the
course of the flood because of increases in DIC concentrations (Figures 3-5b and 3-7c), increases
in FI (Figures 3-6b and 3-7c), and decrease in δ13C-DIC (Figure 3-5d). Increased biogeochemical
processing can partially be explained by seasonal differences in temperature, which have been
shown to drive spatial and temporal redox gradients within the ARB [Kaller et al., 2011]. It can
also be explained by nutrient subsidies from the flood pulse (e.g., Junk et al. [1989]; Tockner et
al. [2000]), which have been shown to fuel internal cycling within the ARB [Jones et al., 2014]
and other large river floodplains like the Amazon [Hedges et al., 2000]. While the FI index values
fall within the expected range for large river systems (e.g. McKnight et al. [2001]), they increased
over the course of the flood suggesting a greater influence of autochthonous DOM. Similarly, the
increase in DIC and subsequent depletion of δ13C-DIC suggest increases in respiration and
heterotrophic activity.
These results highlight the hysteretic nature of floodwater biogeochemistry in large river
floodplains, where the floodplains fluctuate between three dominant processes: flushing (rising
limb), advective transport (peak), and organic matter accumulation (falling limb) [Jones et al.,
2014]. More specifically, our study highlights the increased potential for nitrogen retention during
the flood recession because of enhanced production of labile DOM to fuel biogeochemical
processes such as denitrification.
39
3.5. Conclusion Building on previous research (e.g. Jones et al., [2014]; Scott et al., [2014]), we examined the fate
and transport of dissolved organic matter (DOM) and nitrogen (N) within secondary channels of a
highly connected portion of the Atchafalaya River floodplain. Through three synoptic sampling
campaigns across the rising limb, peak, and falling limb of the hydrograph, we were able to capture
spatial and temporal gradients of inundation hydrology and floodplain biogeochemical processing.
Primary observations included:
The secondary channels within the Atchafalaya River floodplain were largely
dominated by river water during the rising and falling limbs of the annual flood
pulse, but local floodplain water was more dominant during the peak when river-
floodplain connectivity was enhanced.
The Atchafalaya River floodplain was a major source of DOM during the flood,
exporting highly processed aromatic DOM downstream.
Seasonal increases in temperature and nutrient subsidies from the flood pulse
facilitated increases in heterotrophic activity during the falling limb of the annual
flood pulse.
Because of the unique setting of the Atchafalaya River Basin (ARB), we had the opportunity of
examining and isolating the effects of a large-river floodplain on biogeochemical processing of
DOM and N. These findings will be important as managers begin to consider reestablishing river-
floodplain connectivity in areas like the lower Mississippi River Valley, where river-floodplain
connectivity has been eliminated through engineering activities.
3.6 Works Cited Benjamini, Y., and Y. Hochberg (1995), Controlling the false discovery rate - a practical and
powerful approach to multiple testing, J. R. Stat. Soc. Ser. B-Methodol., 57(1), 289-300.
Bianchi, T. S., F. Tigreros, S. A. Lewis, M. Shields, H. J. Mills, D. Butman, C. Osburn, P.
Raymond, C. G. Shank, and S. F. DiMarco (2013), Enhanced transfer of terrestrially derived
carbon to the atmosphere in a flooding event, Geophys. Res. Lett., 40(1), 116-122.
BryantMason, A., Y. Jun Xu, and M. A. Altabet (2013), Limited capacity of river corridor wetlands
to remove nitrate: A case study on the Atchafalaya River Basin during the 2011 Mississippi River
Flooding, Water Resources Research 49, 283–190, doi:10.1029/2012WR012185.
Cerling, T. E., Y. Wang, and J. Quade (1993), Expansion of C4 ecosystems as an indicator of
global ecological change in the late Miocene, Nature, 361(6410), 344-345.
Craig, H. (1953), The geochemistry of the stable carbon isotopes, Geochimica et Cosmochimica
Acta, 3(2), 53-92.
Chow, A. T., J. N. Dai, W. H. Conner, D. R. Hitchcock, and J. J. Wang (2013), Dissolved organic
matter and nutrient dynamics of a coastal freshwater forested wetland in Winyah Bay, South
Carolina, Biogeochemistry, 112(1-3), 571-587.
Cory, R. M., and D. M. McKnight (2005), Fluorescence spectroscopy reveals ubiquitous presence
of oxidized and reduced quinones in dissolved organic matter, Environ. Sci. Technol., 39(21),
8142-8149.
40
Cory, R. M., M. P. Miller, D. M. McKnight, J. J. Guerard, and P. L. Miller (2010), Effect of
instrument-specific response on the analysis of fulvic acid fluorescence spectra, Limnology and
Oceanography-Methods, 8, 67-78.
Cook, R. L., J. E. Birdwell, C. Lattao, and M. Lowry (2009), A Multi-method comparison of
Atchafalaya Basin surface water organic matter samples, Journal of Environmental Quality, 38(2),
702-711.
Creed, I. F., et al. (2015), The river as a chemostat: fresh perspectives on dissolved organic matter
flowing down the river continuum, Can. J. Fish. Aquat. Sci.
Duan, S., T. S. Bianchi, A. M. Shiller, K. Dria, P. G. Hatcher, and K. R. Carman (2007), Variability
in the bulk composition and abundance of dissolved organic matter in the lower Mississippi and
Pearl rivers, Journal of Geophysical Research: Biogeosciences (2005–2012), 112(G2).
Fellman, J. B., E. Hood, R. T. Edwards, and D. V. D'Amore (2009), Changes in the concentration,
biodegradability, and fluorescent properties of dissolved organic matter during stormflows in
coastal temperate watersheds, J. Geophys. Res.-Biogeosci., 114.
Fellman, J. B., N. E. Pettit, J. Kalic, and P. F. Grierson (2013), Influence of stream-floodplain
biogeochemical linkages on aquatic foodweb structure along a gradient of stream size in a tropical
catchment, Freshwater Science, 32(1), 217-229.
Finlay, J. (2003), Controls of streamwater dissolved inorganic carbon dynamics in a forested
watershed, Biogeochemistry, 62(3).
Ford, M., and J. A. Nyman (2011), Preface: an overview of the Atchafalaya River, Hydrobiologia,
658(1), 1-5.
Forshay, K. J., and E. H. Stanley (2005), Rapid nitrate loss and denitrification in a temperate river
floodplain, Biogeochemistry, 75(1), 43-64.
France, R. L. (1996), Scope for use of stable carbon isotopes in discerning the incorporation of
forest detritus into aquatic foodwebs, Hydrobiologia, 325(3), 219-222.
Galat, D. L., L. H. Fredrickson, and D. D. Humburg (1998), Flooding to restore connectivity of
regulated, large river wetlands natural and controlled flooding as complementary processes along
the lower Missouri River, BioScience, 48(9), 721-733.
Harrison, M. D., A. J. Miller, P. M. Groffman, P. M. Mayer, and S. S. Kaushal (2014), Hydrologic
controls on nitrogen and phosphorous dynamics in relict oxbow wetlands adjacent to an urban
restored stream, JAWRA Journal of the American Water Resources Association, 50(6), 1365-
1382.
Hedges, J. I., et al. (2000), Organic matter in Bolivian tributaries of the Amazon River: A
comparison to the lower mainstream, Limnol. Oceanogr., 45(7), 1449-1466.
Hein, T., C. Baranyi, G. J. Herndl, W. Wanek, and F. Schiemer (2003), Allochthonous and
autochthonous particulate organic matter in floodplains of the River Danube: the importance of
hydrological connectivity, Freshw. Biol., 48(2), 220-232.
Hodges, J. L., Jr., and E. L. Lehmann (1963), Estimates of location based on rank tests, The Annals
of Mathematical Statistics, 34(2), 598-611.
41
Hood, E., D. M. McKnight, and M. W. Williams (2003), Sources and chemical character of
dissolved organic carbon across an alpine/subalpine ecotone, Green Lakes Valley, Colorado Front
Range, United States, Water Resources Research, 39(7).
Hughes, D. A. (1980), Floodplain inundation - processes and relationships with channel discharge,
Earth Surf. Process. Landf., 5(3), 297-304.
Hughes, F. M. R. (1997), Floodplain biogeomorphology, Prog. Phys. Geogr., 21(4), 501-529.
Hupp, C., C. Demas, D. Kroes, R. Day, and T. Doyle (2008), Recent sedimentation patterns within
the central Atchafalaya Basin, Louisiana, Wetlands, 28(1), 125-140.
James, W. F., W. B. Richardson, and D. M. Soballe (2008), Effects of residence time on summer
nitrate uptake in mississippi river flow‐regulated backwaters, River Research and Applications,
24(9), 1206-1217.
Jones, C. N., D. T. Scott, B. L. Edwards, and R. F. Keim (2014), Perirheic mixing and
biogeochemical processing in flow-through and backwater floodplain wetlands, Water Resources
Research, 50(9), 7394-7405.
Junk, W. J., P. B. Balyley, and R. E. Sparks (1989), The flood pulse concept in river-floodplain
systems., Can. Spec. Publ. Fish. Aquat. Sci., 106, 110-127.
Kaller, M., R. Keim, B. Edwards, A. Raynie Harlan, T. Pasco, W. Kelso, and D. Allen Rutherford
(2015), Aquatic vegetation mediates the relationship between hydrologic connectivity and water
quality in a managed floodplain, Hydrobiologia, 1-13.
Kaller, M. D., W. E. Kelso, B. T. Halloran, and D. A. Rutherford (2011), Effects of spatial scale
on assessment of dissolved oxygen dynamics in the Atchafalaya River Basin, Louisiana,
Hydrobiologia, 658(1), 7-15.
Kaushal, S. S., P. M. Mayer, and P. G. Vidon (2014), Land use and climate variability amplify
carbon, nutrient, and contaminant pulses: a review with management implications, JAWRA
Journal of the American Water Resources Association, 50(3), 585-614.
Kendall, C., and T. B. Coplen (2001), Distribution of oxygen-18 and deuterium in river waters
across the United States, Hydrological Processes, 15(7), 1363-1393.
Kendall, C., and J. J. McDonnell (2006), Isotope Tracers in Catchment Hydrology, Elsevier,
Amsterdam, The Netherlands.
Kreiling, R. M., J. P. Schubauer-Berigan, W. B. Richardson, L. A. Bartsch, P. E. Hughes, J. C.
Cavanaugh, and E. A. Strauss (2013), Wetland management reduces sediment and nutrient loading
to the upper Mississippi River, Journal of Environmental Quality, 42(2), 573-583.
Kroes, D. E., and T. F. Kraemer (2013), Human-induced stream channel abandonment/capture and
filling of floodplain channels within the Atchafalaya River Basin, Louisiana, Geomorphology,
201, 148-156.
Kroes, D. E., E. R. Schenk, G. B. Noe, and A. J. Benthem (2015), Sediment and nutrient trapping
as a result of a temporary Mississippi River floodplain restoration: The Morganza Spillway during
the 2011 Mississippi River Flood, Ecol Eng., 82, 91-102.
Lane, R. R., J. W. Day, and B. Thibodeaux (1999), Water quality analysis of a freshwater diversion
at Caernarvon, Louisiana, Estuaries, 22(2A), 327-336.
42
Leopold, L. B. (1968), Hydrology for urban land planning: A guidebook on the hydrologic effects
of urban land use, US Government Printing Office.
Llewellyn, D. W., G. P. Shaffer, and N. J. Craig (1996), A decision‐support system for prioritizing
restoration sites on the Mississippi River alluvial plain, Conservation Biology 10(5), 1446-1455.
McKnight, D. M., E. W. Boyer, P. K. Westerhoff, P. T. Doran, T. Kulbe, and D. T. Andersen
(2001), Spectrofluorometric characterization of dissolved organic matter for indication of
precursor organic material and aromaticity, Limnol. Oceanogr., 46(1), 38-48.
Miller, M. P., and D. M. McKnight (2010), Comparison of seasonal changes in fluorescent
dissolved organic matter among aquatic lake and stream sites in the Green Lakes Valley, Journal
of Geophysical Research, 115(G1).
Mitsch, W. J., J. W. Day, J. W. Gilliam, P. M. Groffman, D. L. Hey, G. W. Randall, and N. M.
Wang (2001), Reducing nitrogen loading to the Gulf of Mexico from the Mississippi River Basin:
Strategies to counter a persistent ecological problem, Bioscience, 51(5), 373-388.
Mladenov, N., D. M. McKnight, P. Wolski, and L. Ramberg (2005), Effects of annual flooding on
dissolved organic carbon dynamics within a pristine wetland, the Okavango Delta, Botswana,
Wetlands, 25(3), 622-638.
Mook, W. G., and F. C. Tan (1991), Stable carbon isotope ratios in rivers and estuaries, in
Biogeochemistry of Major World Rivers, edited by E. T. Degens, S. Kemp and J. E. Richey, pp.
245-264, Wiley, New York.
Noe, G. B., and C. R. Hupp (2007), Seasonal variation in nutrient retention during inundation of a
short-hydroperiod floodplain, River Research and Applications, 23(10), 1088-1101.
Pringle, C. (2003), What is hydrologic connectivity and why is it ecologically important?,
Hydrological Processes, 17(13), 2685-2689.
Rabalais, N. N., R. E. Turner, and W. J. Wiseman (2002), Gulf of Mexico hypoxia, aka "The dead
zone", Annual Review of Ecology and Systematics, 33.
Racchetti, E., M. Bartoli, E. Soana, D. Longhi, R. R. Christian, M. Pinardi, and P. Viaroli (2011),
Influence of hydrological connectivity of riverine wetlands on nitrogen removal via denitrification,
Biogeochemistry, 103(1-3), 335-354.
Raymond, P. A., and J. J. Cole (2003), Increase in the export of alkalinity from North America's
largest river, Science, 301(5629), 88-91.
Reckendorfer, W., A. Funk, C. Gschopf, T. Hein, and F. Schiemer (2013), Aquatic ecosystem
functions of an isolated floodplain and their implications for flood retention and management, J.
Appl. Ecol., 50(1), 119-128.
Rosemond, A. D., J. P. Benstead, P. M. Bumpers, and V. Gulis (2015), Experimental nutrient
additions accelerate terrestrial carbon loss from stream ecosystems, Science. 347(6226), 1142-
1145.
Royer, T. V., M. B. David, and L. E. Gentry (2006), Timing of riverine export of nitrate and
phosphorus from agricultural watersheds in Illinois: implications for reducing nutrient loading to
the Mississippi River, Environ. Sci. Technol. 40(13). 4126-4131.
43
Rudorff, C. M., J. M. Melack, and P. D. Bates (2014), Flooding dynamics on the lower Amazon
floodplain: 1. Hydraulic controls on water elevation, inundation extent, and river‐floodplain
discharge, Water Resources Research, 50(1), 619-634.
Sabo, M. J., C. F. Bryan, W. E. Kelso, and D. A. Rutherford (1999), Hydrology and aquatic habitat
characteristics of a riverine swamp: II. Hydrology and the occurrence of chronic hypoxia,
Regulated Rivers: Research & Management, 15(6), 525-544.
Scaroni, A. E., C. W. Lindau, and J. A. Nyman (2010), Spatial variability of sediment
denitrification across the Atchafalaya River basin, Louisiana, USA, Wetlands, 30(5), 949-955.
Scott, D. T., R. F. Keim, B. L. Edwards, C. N. Jones, and D. E. Kroes (2014), Floodplain
biogeochemical processing of floodwaters in the Atchafalaya River Basin during the Mississippi
River flood of 2011, Journal of Geophysical Research: Biogeosciences, 119(4), 2013JG002477.
Scown, M. W., and M. C. Thoms (2015), An index of floodplain surface complexity, Hydrology
and ….
Shen, Y., C. G. Fichot, and R. Benner (2012), Floodplain influence on dissolved organic matter
composition and export from the Mississippi—Atchafalaya River system to the Gulf of Mexico,
Limnol. Oceanogr.
Sparks, R. E., J. C. Nelson, and Y. Yin (1998), Naturalization of the flood regime in regulated
rivers, Bioscience.
Taylor, P. G., and A. R. Townsend (2010), Stoichiometric control of organic carbon-nitrate
relationships from soils to the sea, Nature, 464(7292), 1178-1181.
Tockner, K., and J. A. Stanford (2002), Riverine flood plains: present state and future trends,
Environmental Conservation, 29(3), 308-330.
Tockner, K., F. Malard, and J. V. Ward (2000), An extension of the flood pulse concept,
Hydrological Processes, 14(16-17), 2861-2883.
Tockner, K., D. Pennetzdorfer, N. Reiner, F. Schiemer, and J. V. Ward (1999), Hydrological
connectivity, and the exchange of organic matter and nutrients in a dynamic river-floodplain
system (Danube, Austria), Freshw. Biol., 41(3), 521-535.
Vitousek, P. M., H. A. Mooney, J. Lubchenco, and J. M. Melillo (1997), Human Domination of
Earth's Ecosystems, Science, 277(5325), 494-499.
Wachniew, P. (2006), Isotopic composition of dissolved inorganic carbon in a large polluted river:
The Vistula, Poland, Chemical Geology.
Wang, X. C., R. F. Chen, and G. B. Gardner (2004), Sources and transport of dissolved and
particulate organic carbon in the Mississippi River estuary and adjacent coastal waters of the
northern Gulf of Mexico, Marine Chemistry, 89(1-4), 241-256.
Weishaar, J. L., G. R. Aiken, B. A. Bergamaschi, M. S. Fram, R. Fujii, and K. Mopper (2003),
Evaluation of specific ultraviolet absorbance as an indicator of the chemical composition and
reactivity of dissolved organic carbon, Environ. Sci. Technol., 37(20), 4702-4708.
Welti, N., E. Bondar-Kunze, G. Singer, M. Tritthart, S. Zechmeister-Boltenstern, T. Hein, and G.
Pinay (2012), Large-scale controls on potential respiration and denitrification in riverine
floodplains, Ecol Eng, 42, 73-84.
44
Zurbrugg, R., S. Suter, M. F. Lehmann, B. Wehrli, and D. B. Senn (2013), Organic carbon and
nitrogen export from a tropical dam-impacted floodplain system, Biogeosciences, 10(1), 23-38.
.
45
CHAPTER 4. SEASONAL VARIATION IN FLOODPLAIN
BIOGEOCHEMICAL PROCESSING IN A RESTORED
HEADWATER STREAM
C. Nathan Jones, Durelle T. Scott, Christopher Guth, Erich T. Hester, W. Cully Hession
Submitted: May 2015
To: Environmental Science & Technology
Status: In revision
Abstract Stream and river restoration activities have recently begun to emphasize the enhancement of
biogeochemical processing within river networks through the restoration of floodplain wetlands.
It is generally accepted that this practice removes pollutants such as nitrogen and phosphorus
because the increased contact time of nutrient-rich floodwaters with reactive floodplain sediments.
Our study examines this assumption in the floodplain of a recently restored, low-order stream
through five seasonal experiments. During each experiment, a floodplain wetland was artificially
inundated for three hours. Both the net flux of dissolved nutrients and nitrogen uptake rate were
measured during each experiment. The floodplain was typically a source of dissolved phosphorus
and dissolved organic matter, a sink of NO3-, and variable source/sink of ammonia. NO3
- uptake
rates were relatively high when compared to riverine uptake, especially during the spring and
summer experiments. However, when scaled up to the entire 1 km restoration reach with a simple
inundation model, less than 0.5-1.5% of the annual NO3- load would be removed because of the
short duration of river-floodplain connectivity. These results suggest that restoring river-floodplain
connectivity is not necessarily an appropriate best management practice for nutrient removal in
low-order streams with legacy soil nutrients and provide further guidance restoration activities in
the Mid-Atlantic and North Eastern US.
Photo Credit: W. Cully Hession
46
4.1 Introduction Stream and river restoration is a rapidly growing industry within the United States1. Largely, this
growth is a response to the almost 1 million km of impaired streams across the United States and
regulatory measures associated with the Clean Water Act and Endangered Species Act.2, 3
Traditional restoration objectives include streambank stabilization, riparian/instream habitat
enhancement, and more recently, a functional lift of the degraded stream ecosystem.4, 5 Here,
functional lift refers to enhancing ecosystem function as a whole, not just focusing on a single
aspect of restoration (e.g. bank stabilization), in order to maximize both ecosystem health and
services. Creating functional lift within the stream can include restoring hydrologic connectivity
between the channel and adjacent landscape.6 Specifically, connectivity between rivers and their
adjacent floodplains provide many ecosystem services such as floodwater storage, increased
ecosystem productivity, and increased biogeochemical processing of floodwaters.7 The latter is of
interest in many urban and agricultural watersheds, where water quality impairment is associated
with storm water management and legacy agricultural practices, respectively.8
Because of their transitional and dynamic nature, floodplains are hotspots for biogeochemical
activity9, 10 and can be both a source and sink of nutrients such as nitrogen (N) and phosphorus (P).
In a typical floodplain found in temperate climates, nutrient retention and removal is controlled by
a combination of antecedent moisture condition, biogeochemical processing rates, and the
residence time of floodwaters within the floodplain (Figure 4-1).11, 12 During the interflood period,
or the time between flood events, floodplain sediments can accumulate dissolved organic matter
(DOM),13 soluble reactive phosphorus (SRP),14 and ammonia (NH4+).15 Antecedent soil moisture
conditions control the export and processing of these constituents once the floodplain is
reconnected to the adjacent river channel.16 If the floodplain is relatively dry prior to inundation,
accumulated nutrients can be flushed downstream, where the NH4+ is likely transformed to
NO3- through nitrification17 and SRP likely spirals downstream through cycles of sorption and
desorption with suspended sediments.18 When the floodplain is inundated prior to hydrologic
connection with the river, an area of mixing known as the perirheic zone is established between
the existing floodplain water and river water.19 In many cases, steep redox gradients form across
the perirheic zone and at the sediment-water interface, allowing for rapid transformation of
reactive solutes through redox processes, facilitating the transition of floodwaters to
biogeochemically processed floodplain water.12
In addition to effects associated with antecedent moisture condition, biotic processing and growing
season play an important role in nutrient processing capacity of floodplains.20, 21 Biotic
assimilation, or the conversion of inorganic nutrients to organic nutrients, in benthic sediments is
typically attributed to algal and vegetative communities in floodplains.22 In forested floodplain
systems where water and/or light are limiting, biotic assimilation can account for less than 10% of
added nitrate removal (NO3-).23 In contrast, in highly productive backwater floodplain lakes,
assimilation can account for 76% of NO3- removal.24 In both cases, the floodplain is acting as
temporary sink of N, which will be available for internal cycling during the interflood period and
potentially removed from the floodplain as particulate N through a combination decay processes
(e.g., mineralization) and hydrologic export.25 Seasonal differences in biotic uptake are strongly
coupled with seasonal patterns in water availability, temperature, and solar intensity,21, 26 where
differences between seasonal processing rates can result in N load reductions ranging from 0.05%
to 60%, respectively, within the same floodplain wetland.27
47
Biogeochemical processing is also strongly tied to the residence times of floodwaters,28, 29 or the
amount of time floodwaters are contained within a floodplain. In medium to large riverine systems,
residence times within floodplains can be long enough to develop conditions ideal for
biogeochemical processing.30 However, in headwater floodplains, short residence times restrict
redox dependent processes such as denitrification,20 and nutrient export is largely controlled by
the balance of biotic uptake and soil flushing processes. This suggests that the variable solute
source/sink characteristic observed headwater floodplains20, 31, 32 is largely dependent on seasonal
variables such as temperature and antecedent soil moisture. The observed variability of
biogeochemical processing across studies is inconsistent with recent efforts in the US Mid-Atlantic
and Northeast to promote river-floodplain connectivity in restoration projects in order to reduce
the downstream flux of nitrogen (N) and phosphorus (P), especially in restoration sites located in
headwater catchments. Therefore, our overall objective was to further investigate mechanisms
associated with N and P processing in a headwater floodplain over the course of an entire year and
to provide the restoration community further guidance to optimize nutrient removal through river-
floodplain connectivity. Specifically, the goals of this study were to examine the first flush of
dissolved nutrients during the initial wetting of soils; quantify the seasonal variation in source/sink
characteristics of the floodplain; and estimate N load reduction associated with river-floodplain
connectivity at the reach scale.
Figure 4-1. Conceptual Model of processes within the experimental floodplain slough. A.) Plan
view of the experimental slough highlighting the inlet and outlet structures, the three sampling
cross sections (XS-1, XS-2, XS-3), and the mixing zone between floodwater (dark grey area)
pumped into the slough and local water (light grey area) likely derived from a combination of
groundwater inputs, rainwater, and previous inundation events. Dashed lines represent
conceptualized flowpaths: advective flow and transient storage dominated flowpaths, respectively.
B.) Cross-sectional view of the inundated floodplain displaying potential mechanisms that control
nutrient cycling and retention within the experimental slough.
48
4.2 Methods 4.2.1 Site Description
The study site is an abandoned slough within the floodplain of a third-order stream (Stroubles
Creek) in the Ridge and Valley physiographic province of southwestern Virginia, USA and within
the Virginia Tech Stream Restoration, Education, and Management Lab (StREAM Lab). The
contributing watershed is approximately 15 km2 which is 84% urban/residential landcover, 13%
agriculture, and 3% forest.33 The contributing area within the stream reach is predominantly row
crop agriculture, and the site itself is located along a recently restored (c. 2009) reach of stream.
The slough has an approximate surface area of 450 m2 and the flowpath length approximately 60
m (Figure 4-1a). Previous to restoration, the channel was severely incised and riparian productivity
was altered by agricultural grazing activities.34 Post restoration and cattle removal, the site has
been inundated 2-3 times a year and is dominated by reed canary grass.
Figure 4-2. Climatic data measured in the StREAM Lab floodplain. The dark grey area represents
the volumetric soil water content (Vw), the blue line is the annual hydrograph (Q), the light grey
bars represent monthly evapotranspiration (ET), the dark grey points represent mean daily
temperature (Temp), dark grey bars represent the rainfall hyetograph (Precip), and circles and
squares represent the timing of saturated and dry floods, respectively.
Hydrology within the floodplain slough is largely controlled by both evapotranspiration (ET) and
a shallow confining layer of clay. Water sources include periodic inundation from the channel,
groundwater, and direct precipitation. Previous modelling studies suggest ET drives seasonal
differences in the floodplain water table, where soils are saturated during winter/spring when ET
rates are minimal, and the water table is drawn down in the summer when ET rates are high during
49
the growing season.35 This pattern was experienced during the study period, where soils were
saturated from late fall through mid-summer (Figure 4-2), although this patter was exacerbated by
a wetter-than-normal spring and early summer. Further, the shallow clay layer acts as a confining
layer, which restricts surface water-groundwater interactions and water loss to the subsurface
during the flood experiments. More information about the StREAM Lab and surface water-
groundwater interaction at the site can be found at streamlab.bse.vt.edu and in the companion study
(Hester et al., Surface water-groundwater exchange processes in an experimental floodplain along
a restored headwater stream, submitted 2015),36 respectively.
4.2.2 Experimental Design
We conducted five experimental floods over the course of one year to capture seasonal differences
in floodplain biogeochemical processes: April 8 (spring), June 29 (early summer), August 30 (late
summer), November 11 (fall), and February 7 (winter) across 2013 and 2014. Each experimental
flood lasted 3 hours to simulate natural overbank flood events, where quasi steady-state flow
conditions were achieved in the first two hours and the third hour was used to conduct a nitrogen
uptake experiment. Stream water was pumped into the floodplain slough using a Berkeley B3-
ZRMS irrigation pump at a flow rate of approximately 24 L·s-1. Inlet and outlet flows were
measured using an ultrasonic Fuji M-flow flowmeter and a 7.62 cm parshall flume with an Onset
HOBO Pressure Transducer, respectively. We collected water quality samples at the inlet (e.g., the
pump) at 30 min intervals and at three locations (XS1, XS2, and XS3) along the centerline of flow
in the slough at 15 min intervals (Figure 4-1a). During the first 15 min after water arrived at each
sampling location, we collected samples at 5 min intervals in order to capture the “first flush”
phenomena. At hour two of each experiment, we instantaneously added a solution of NaNO3 and
NaCl tracers to the inlet of the slough. As the tracer slug moved through the site, we collected
samples at 1-3 min intervals at XS2 and XS3. Here, it is important to highlight we pumped stream
water representative of baseflow, not flood flow, into the experimental slough. Because of
differences in background solute concentrations and sediment load, some biogeochemical
processes (e.g. denitrification) may have altered kinetics and processes associated with sediment
deposition (e.g. total phosphorus removal) could not be accurately represented. Further
information about sample analysis and handling can be found in the associated supplemental
material.
4.2.3 Flood Experiment Analyses
We characterized surface-water hydraulics using first-arrival and steady-state residence time
metrics. The first-arrival time represents the time between the pump starting and the start of water
flow at XS3 (e.g., the end of the slough), while steady-state residence time represents the time
from injection of the tracer to the time when the peak of the conservative solute breakthrough
curve passed XS3. Note, the tracer occurred at hour two during quasi steady-state flow conditions.
For each experimental flood, we estimated nutrient retention (or export) using a mass balance in
order to examine the source/sink characteristics of the floodplain slough. Inlet flux was calculated
using measured flow and solute concentrations at the pump, while outlet flux was calculated using
measured flow and solute concentrations at the outlet and XS3, respectfully. Here, we assumed
solute concentrations at XS3 represented solute concentrations at the outlet because of the
development of concentrated flow and proximity between the two locations
We estimated NO3- uptake using the Tracer Analysis for Spiraling Curve Characterization
(TASCC) methodology.37 While this method was designed for stream systems, it has been recently
50
used within alluvial wetlands where advective flow dominates.38 Essentially, this method utilizes
the difference in the breakthrough curves of the conservative (Cl-) and nonconservative (NO3-)
tracers to estimate instantaneous spiralling metrics (e.g., Newbold et al.,1982)39 and also to
estimate background spiraling metrics through regression. Here, spiraling metrics refer to kinetic
parameters typically used to characterize solute dynamics in streams such as uptake velocity,
uptake length, and areal uptake. The calculated areal uptake (Utot, µg N m-2 min-1) and uptake
velocity (Vtot , m min-1) are applied to Michaelis-Menten kinetics model, and relationships are
developed between NO3- concentrations and spiraling metrics.
To elucidate differences in nitrogen processing across seasons and floodplain location, we
compared results from the five TASCC injections. We completed a nonparametric multivariate
analysis of variance (NPMANOVA)40 where the significance of seasonal variables (e.g., season,
soil temperature, volumetric water content, and background NO3- concentrations), location (e.g.,
cross section), and interaction between seasonal variables and location were tested. We then used
a robust analysis of covariance (robust ANCOVA)41 in a pairwise fashion to test differences in
Michaelis-Menten kinetic models developed for the two cross sections during each event. We
completed all calculations with R Statistical Software42 using the Vegan43 and WRS44 packages.
4.2.4 Reach-Scale Modeling
We estimated annual NO3- removal for the associated 1 km restoration reach using a simple well
mixed-reactor model that incorporated measured areal uptake, a simple inundation model derived
from raster analysis, and a synthetic flow record based on regional regression based on USGS
gaging data and landuse characteristics. Further model details can be found in the supplementary
material. Here, it is important to highlight the parsimonious and conservative nature of the
presented model. Feedbacks between river-floodplain interactions and biogeochemical processing
are very complex and at this point, poorly characterized.12, 45, 46 However, our model estimates
potential load reductions utilizing simple measures and provides a rough estimate of
biogeochemical processing within the StREAM Lab floodplain. While we acknowledge the
uncertainty associated with the resulting load reduction estimates, the estimates provide an
approximate characterization of load reductions in low-order floodplains that can be applied to
restoration projects in the Mid-Atlantic Ridge and Valley.
Table 4-1. Summary of residence time metrics
Flood First-arrival
Time (min)
Steady-State
Residence Time (min)
Spring 12.5 28
Summer 1 12.5 29
Summer 2 31.1 27
Fall 35.0 51
Winter 9.5 38
4.3 Results and Discussion 4.3.1 Inundation Hydrology
Topography, antecedent moisture conditions, and vegetative roughness controlled inundation
hydrology within the floodplain slough. Generally, the flood progression was similar across all
five floods: the slough would fill and then begin to drain back into the adjacent stream at the natural
51
outlet point down gradient of XS3 (Figure 4-1a). When examining the first-arrival times of the
experimental floods, two groups are apparent: saturated and dry floods. The spring, early summer,
and winter floods all had relatively short first-arrival times (9.5 – 12.5 min) and were inundated
prior to pumping (i.e. saturated floods), whereas the late summer and fall floods had relatively
long first-arrival times (>30 min) and the slough was empty prior to the experiment (i.e. dry
floods). A discussion of how antecedent moisture and vegetation control surface water-
groundwater exchange and water storage in the slough can be found in Hester et al., 2015.36 Similar
patterns in flooding have been documented in the Nyack River floodplain in the western US, where
individual sloughs were characterized as hydrologic facets47 and were the scalable unit used to
model inundation hydrology.48 However, the filling and overflow of the slough does not entirely
characterize water movement within the experimental slough. The lag in conservative tracer signal
(Figure 4-3a), along with visual indicators (e.g. dye tracer and vegetation disturbance), suggest
that a mixing zone between the flood water and local water formed over the course of each
saturated flood, potentially indicating multiple surface-water flowpaths exist within the slough.
Similar to patterns observed by Mertes19 in large river systems, inundation hydrology and mixing
patterns observed in the floodplain were also controlled by antecedent inundation conditions (e.g.,
wet vs dry floods). During the saturated floods, two dominant flowpaths were conceptualized
(Figure 4-1a). Flow through the deepest portion of the slough, where local water was present prior
to flooding, was largely dominated by transient storage and experienced relatively long residence
times. A second primary flowpath formed on the inside of the floodplain slough, which essentially
bypassed the existing local water and was dominated by advective transport (Figure 4-1a). These
flowpaths were conceptualized through visual inspection of both dye tracers and vegetation
disturbance. During these experimental floods, it took 60-75 min for the slough to become
completely mixed as seen in the effluent NO3- signal (Figure 4-3b), suggesting the persistence of
the perirheic zone at the boundary of the two flowpaths. In contrast, the dry floods displayed
relatively uniform effluent signal (Figure 4-3a) and prolonged first-arrival times (Table 4-1)
because the lack of local water prior to flooding. With the available data, it is unclear if a perirheic
zone formed over the course of either dry flood. However, the fall flood had a much longer steady-
state residence time, suggesting relatively uniform flow when compared to the late summer flood
and could be attributed to greater matting of senescing vegetation within the local water column.36
Floodplain conductivity of surface-waters has been described as a measure of the floodplains
ability to route floodwaters, and is a function of the storm hydrograph, floodplain topography, and
vegetative roughness45. However, floodplains exist as a mosaic of habitats49 and therefore
experience a heterogeneous distribution of floodplain conductivity. Within the experimental
slough, the two conceptualized flowpaths experience differences in vegetation density.
Specifically, the deeper portion of the slough has much denser vegetation than the shallow portion,
potentially leading to greater mixing and reduced water velocities at the sediment water-interface.
Therefore, in addition to a decrease in conductance associated with the perirheic zone, the deeper
flowpath could also have decreased conductance because of increased vegetative roughness. As
vegetation conditions varied across the five experimental floods, it is likely the relative
contribution of the two surficial flowpaths also varied. These differences in inundation hydrology,
along with differences in the roughness due plant communities, help explain the difference in
biogeochemical processing between the 5 experimental floods.
52
Figure 4-3. Inlet and outlet concentrations of dissolved reactive constituents during the five
inundation experiments. Average inlet concentrations are representative of instream solute
concentrations and are shown using bar charts on the left, where error bars represent standard error
from the mean. Outlet concentrations from the first 120 min of each flood are displayed on the
right. Red (plus sign), green(cross), yellow (triangle), orange (hollow circle), and blue (diamond)
are associated with the spring, early summer, late summer, fall, and winter floods, respectively;
outer circles and squares denote saturated and dry experimental floods, respectively.
53
4.3.2 Dissolved Organic Matter Export
Floodplains are an important source of DOM to riverine networks.11 DOC export from the
floodplain slough was fairly consistent throughout all five floods, where there was a pronounced
first flush of DOC during each flood (Figure 4-3e and 4-4e). Patterns reflective of two distinct
processes appear when analyzing the DOC concentration within the floodplain effluent: dilution
and flushing. The saturated floods experienced a dilution of DOC, where mixing between stream
water and local water is apparent in the gradual decline in effluent concentration (Figure 4-3e). In
contrast, the dry floods experienced flushing of DOC from the soil surface within the slough,50
where the signal exhibited a rapid decline in effluent concentration. The experimental slough was
a relatively large source of DOC during the saturated floods, where effluent DOC mass was 30-
60% greater than the influent mass. However, the dry floods exported relatively little or no DOC
export. Our results are consistent with other observations of accumulation of DOC in inundated
floodplains18, and could be attributed to decay of algal biomass25 and plant material.13 Because of
the dominance of reed canary grass within the study slough, it is likely a dominant source of
organic matter and autochthonous nutrients.51 Further, our results suggest inundation hydrology,
and more specifically, antecedent inundation conditions controls DOC export and processing
within the experimental slough (Figure 4-4e).
4.3.3 Initial Flush of Reactive Solutes
Similar to other studies, the experimental slough was a consistent net source of SRP during each
experimental flood. Potential sources of SRP include organic matter, likely derived from reed
canary grass decay51; nutrient rich sediments and particulate organic matter deposition from
previous floods14,52; and the legacy of over 100 years of cattle grazing at the site. Possible
mechanisms responsible for SRP loss from floodplains include the oxidation of DOM,53 reduction
of iron-phosphate sediment,54 and desorption of loosely held SRP- at the soil surface52. However,
it is important to note that we did not capture depositional processes during the flood experiments,
and thus, were unable to characterize total phosphorus removal and sorption/desportion of SRP
associated with newly deposited sediments.
Here, we propose that both hydrologic and biotic processes control SRP export from the slough.
In contrast to the DOC signal, the dry floods had relatively large SRP first flushes when compared
to the saturated floods (Figure 4-3 and 4-4). During the interflood period (e.g., no overbank or
artificial flow), organic matter at the soil surface would be available for mineralization and the
production of loosely held SRP;14 thus, this mechanism is likely a dominant source of SRP during
the dry floods. While the saturated floods did not experience a strong first flush, they were still net
sources of SRP. This is likely attributed to redox conditions that developed during prolonged
inundation, where ferrous soils release SRP when reduced,55 however, thorough analysis of soil
chemistry was not conducted across the study slough to identify the persistance of ferrous soils.
In addition to hydrologic control (e.g., dry vs saturated antecedent inundation conditions), SRP
loads into and out of the floodplain varied greatly with season (Figure 4-4d) and are likely linked
to biological uptake within the floodplain.20 During the spring flood, the growing season was just
beginning when P is in high demand. Thus, available SRP was likely being utilized for plant
biological activity. Plant and microbial uptake has been widely observed in streams and associated
riverine wetlands,56 and periphyton can act as a temporary sink that completely masks
54
Figure 4-4. The total load observed at the inlet (dark) and outlet (light) of the floodplain slough
during the 5 inundation experiments. Dry floods are denoted by bars with hatching, and both the
relative difference (%) and mass difference (g) between the loads is also displayed.
55
P signals.57 This is a possible explanation of the minimal SRP load experienced within the spring
flood, and is in contrast to the fall flood, where much of the plant material was senescing and
possibly contributed to the large net export of SRP through mineralization of newly senesced
material. Other restored floodplains with legacy nutrients have been shown to be sources of SRP
for 20+ years post restoration.58
Over the course of the five experimental floods, the slough was a variable source/sink of NH4+.
During the fall flood, there was an observable first flush of both NH4+ and SRP (Figures 3b-c),
suggesting elevated mineralization of DOM. However, there were no apparent seasonal or
hydrologic drivers behind the NH4+ signal observed in the other four floods. This is consistent with
investigation of other headwater floodplains in both the Mid-Atlantic20 and upper Mississippi
River Valley,32 where total export varied with season and by event, respectively. NH4+ processing
within the floodplain environments is relatively complicated because of the lability of NH4+, its
affinity for sorption, and its tendency to nitrify in oxic environments.17 Noe et al., 201314 examined
mineralization rates across a gradient of floodplains in the Mid-Atlantic and found that nitrification
was dominant in headwater floodplains while ammonification was dominant in larger river
floodplains. While nitrification was not explicitly measured in this study, variability in nitrification
rates could explain the variable source/sink pattern experienced within the experimental slough.
4.3.4 Nitrate Removal
The floodplain was typically a sink of NO3-, where net removal ranged from 2 to 26%
(Figure 4-4b). Similar to the DOC effluent signal, perirheic mixing was evident as the NO3--rich
stream water mixed with the NO3- limited local water during the saturated floods (Figure 4-3b).
When comparing DOC and NO3- effluent concentrations, the DOC signal reached equilibrium
more quickly, suggesting NO3- removal through both biotic uptake and denitrification. Forshay
and Stanley23 observed similar removal of NO3- when river water mixed with local water in a
backwater floodplain, where rapid NO3- loss was attributed to denitrification. Denitrification is a
well-documented process within riverine floodplains, with recorded removal rates ranging from
undectable59 to greater than 75%29 removal of the total riverine NO3- load. Denitrification is
controlled by the amount of NO3- transported into the floodplain,30 the availability of carbon,11 and
the residence time of floodwaters.48 In many backwater floodplains, denitrification is often limited
by the amount of NO3- transported into floodplain environment, suggesting limited exchange
between channel and floodplain.60 In contrast, in environments where river-floodplain interaction
is greater and advective flow dominates in the floodplain environment, denitrification is often
limited by short residence times.59 This was likely the case in our study slough, where mean
residence times ranged from 27 to 51 min, which greatly contrasts floodplain dominated systems
like the Amazon and Danube Rivers where mean residences times can be months.61, 62
Areal uptake (Figure 4-5) was a function of both seasonality, and to a lesser extent, heterogeneity
of flowpaths within the floodplain. The NPMANOVA model accounted for 52% of variation
(R2=0.52, p<0.001) in the relationship between total areal uptake and NO3- concentration across
all five tracer experiments at both cross sections. Spatially, sampling location (e.g., XS2 vs XS3)
only accounted for 1.7% of the observed variation (p=0.01). This suggests while sampling location
was a statistically significant predictor, it is practically unimportant and the observed nitrate uptake
signal was relatively homogeneous at the spatial resolution of our sampling. Therefore, spiraling
metrics were calculated using data from both XS2 and XS3. Seasonal variables accounted for 33%
56
of variation (p<0.001) in the model, and upon closer inspection, soil moisture, background NO3-
concentration, and soil temperature account for 15%,10%, and 2.5% of the observed variation
(p<0.001), respectively. Antecedent moisture conditions affect floodwater residence times (e.g.
dry vs saturated floods), can limit or enhance biological activity and uptake21, and control
mechanisms associated with the first flush (e.g. oxidation of organic matter)16. Additionally,
background NO3- concentrations have been shown to exert influence on NO3
- uptake rates, where
higher NO3- concentrations typically lead to greater NO3
- uptake but lower net removal.63, 64
However, because soil moisture and background NO3- concentration are confounding variables, it
is difficult to isolate their individual effects on the NO3- uptake with the available data. For
example, the two late summer and fall floods were both dry floods and experiences similar
background NO3- concentrations. However, the uptake rates were significantly different,
suggesting there was a shift in dominant processes affecting export.
Table 4-2. Summary of Spiraling metrics and associated statistics from tracer addition.
Experiment Background Uptake Length Michaelis-Menten Model Groups
Wilcox ANCOVA
Sw,amb
(m) Std Error
(m) p-Value
Umax (µg N m-2 min-1)
Km (µg N m-2 min-1)
p-Value
Spring 116 21.4 <0.001 2865 39.9 <0.001 A Early Summer 258 25.9 <0.001 1256 46.7 <0.001 B Late Summer 157 7.9 <0.001 2135 70.2 <0.001 C
Fall 319 44.2 <0.001 1255 123 <0.001 B Winter 304 176 0.12 1483 73.9 0.002 B
Utilizing a Wilcox ANCOVA in a pairwise fashion, the five floods separated into three statistically
different groups: 1) spring, 2) late summer, and 3) winter, early summer, and fall (Table 4-2).
Increased biological uptake is expected in the spring and summer and conversely muted uptake in
the fall and winter because of seasonal differences in temperature and its effects on the biological
communities. However, this pattern is confounded by the low uptake rates experienced in the early
summer floods. Potentially, the low uptake rates measured in the early summer flood were related
to elevated NH4+ export (Figure 4-4d), where highly labile NH4
+ satisfied N-demand within the
slough and reduced rates of NO3- uptake. Similarly, Noe and Hupp20 did not find seasonal patterns
in NO3- export from a small floodplain system, suggesting confounding factors such as nitrification
and organic matter availability/lability also controlled the export of NO3- from the floodplain.
When compared to typical instream areal NO3- uptake,65 NO3
- uptake rates were relatively high
within the slough during the five experimental floods (Figure 4-5). In a recent national assessment
of instream NO3- uptake kinetics, the LINXII study measured total instream NO3
- uptake using N15
additions across 56 different streams in North America. Represented by the dark line in Figure 4-5,
the regression model of LINXII uptake rates is lower than the rates measured across the five
experimental floods. This could partially be explained by differences in measurement techniques,
where the bulk injection alters instream kinetics because of the large increase in NO3-
concentrations.66 Further, the background NO3- concentration of the stream water used to inundate
the floodplain (Figure 4-3) was not necessarily representative of floodwater NO3- concentrations.
However, the response to the NO3- addition was minimal during all five experimental floods (e.g.,
muted rising limbs in the Michaelis-Menten curves), suggesting biogeochemical processing within
57
the floodplain was not limited by NO3- , where the current N pool is derived from internal cycling
of nutrients and potentially legacy N from the 100+ years of grazing at the site.51, 67 Increased
biogeochemical processing within the floodplain can be explained by increased contact between
the floodplain soils and floodwater, increase in availability of labile organic matter, and also
increase in biological uptake.
Figure 4-5. Areal NO3- uptake calculated for spring (red, plus sign), early summer (green, cross),
late summer (yellow, triangle), fall (orange, circle), and winter (blue, diamond) experimental
floods. Points represent individual samples and lines represent Michaelis-Menten Kinetic models
developed for each flood, respectively. The black dashed line represents expected total instream
uptake based on data taken from the 58 streams across North America in the LINXII study.65
Statistical differences denoted in Table 4-2 are displayed using letters A-C.
4.3.5 Modeled Load Reductions
Annual and storm NO3- load reductions were minimal when measured uptake was extrapolated to
the restored floodplain across the synthetic flow record. For individual events, NO3- removal
ranged from 8.6 to 17.6 kg of NO3-N removed per storm, or 7.5 to 10.3% removal of the total
storm load. However, this resulted in annual removal ranging from 0 to 139 kg of NO3-N removed,
or 0 to 1.5% of the annual load of NO3-. Mean individual storm and annual load reductions were
10% and 0.60%, respectively. The relatively simple model is conservative, in that it over estimates
inundation time through a rectangular hydrograph, it assumes relatively high NO3- concentrations
during storm flow (e.g., no dilution affects were accounted for), and the maximum observed areal
uptake (spring, 2000 µg N m-2 min-1) was used for the estimate of NO3- removal. This suggests the
restored floodplain is ineffective in removal of instream NO3- and can be explained by both limited
river-floodplain connectivity and short residence times associated with short-hydroperiod
floodplains. Roley et al., 201263 found similar load reduction associated with two-stage ditch
located in the Midwestern US, where NO3- uptake was relatively high but hydrologic connectivity
and residence time were relatively low. They went on to conclude that to maximize nitrogen
removal, floodplain restoration design should enhance removal of NO3- from waters draining to
the site (e.g. upslope water). In addition, it is also important to note the modeled stream reach in
58
our study represents a relatively small section of stream, and riparian buffers/wetlands have been
shown to remove significant loads of NO3- when the entire basin is taken into account.68 Here, if
we assume that 1 km of stream removes 10% of the NO3- load for individual storms, then it would
take approximately 10 km of restored floodplain to reduce the entire nitrate load being delivered
to the StREAM Lab for an average storm. While this is a vast oversimplification of
biogeochemical processing of floodwaters within floodplains, it does highlight the importance of
the cummulative effects of riparian/floodpalin systems along the stream cooridor.
4.3.6. Management Implications
Recent efforts by the Chesapeake Bay Program (CBP) have emphasized re-establishing
river-floodplain connectivity to improve downstream water quality, essentially making river-
floodplain connectivity a best management practice (BMP) for the stream restoration industry in
the eastern US.4 Here, we do not want to discount ecosystem services derived from restoration
efforts in headwater streams, the water quality benefits riparian zones and floodplains provide by
processing upslope water, or even the cumulative influence of floodplains across the river network.
However, results from studies in the Upper Midwest,32 Mid-Atlantic Piedmont,20, 27 and now the
Appalachian Ridge and Valley (i.e. this study) show that individual floodplains associated with
small to medium size streams can actually be net sources of nutrients. Therefore, these results
cumulatively suggest that restoring river-floodplain connectivity is not always an appropriate BMP
for nutrient removal, and site specific conditions such as annual inundation duration and legacy
landuse should be considered when optimizing river-floodplain connectivity for nutrient removal
and retention. Potentially, these finding would be most useful during site selection and
prioritization phase of restoration design, where multiple objectives and design parameters are
often balanced. Further, our results highlight the potential for flushing of reactive SRP from
floodplain wetlands, and more specifically, stream-wetland complexes often used for mitigation
banking and sediment removal.69 While these engineered wetlands undoubtedly remove sediment
attached P and N through deposition and denitrification,70 respectively, designers must also
balance flushing of allochthonous SRP and other reactive solutes.
4.4 Acknowledgements We want to thank the students in the StREAM Lab REU Program (NSF-EEC-REU 1156688) for
their insights and assistance in piloting and conducting the artificial floods. Specifically, Katy
Hofmeister, Tyler Weiglein, and Dylan Cooper were critical to the success of these experiments.
Further, we thank the National Science Foundation (ENG-CBET-1066817), and the Virginia Tech
Institute for Critical Technology and Applied Sciences (ICTAS) for their support. Opinions
expressed here are those of the authors and not necessarily those of the NSF or ICTAS.
4.5 Works Cited 1. Bernhardt, E. S.; Palmer, M. A.; Allan, J. D.; Alexander, G.; Barnas, K.; Brooks, S.; Carr, J.;
Clayton, S.; Dahm, C.; Follstad-Shah, J.; Galat, D.; Gloss, S.; Goodwin, P.; Hart, D.; Hassett, B.;
Jenkinson, R.; Katz, S.; Kondolf, G. M.; Lake, P. S.; Lave, R.; Meyer, J. L.; O'Donnell, T. K.;
Pagano, L.; Powell, B.; Sudduth, E. Ecology - Synthesizing US river restoration efforts. Science
2005, 308 (5722), 636-637.
2. Lave, R.; Doyle, M.; Robertson, M. Privatizing stream restoration in the US. Soc Stud Sci 2010,
40 (5), 677-703.
59
3. USEPA Watershed Assessment, Tracking & Environmnetal Results.
http://www.epa.gov/waters/ir/
4. Harman, W.; Starr, R.; Carter, M.; Tweedy, K.; Clemmons, M.; Suggs, K.; Miller, C. A function-
based framework for stream assessment and restoration projects; EPA 843-K-12-006; US
Environmental Protection Agency, Office of Wetlands, Oceans, and Watersheds, Washington,
DC: 2012.
5. Bernhardt, E. S.; Sudduth, E. B.; Palmer, M. A.; Allan, J. D.; Meyer, J. L.; Alexander, G.;
Follastad-Shah, J.; Hassett, B.; Jenkinson, R.; Lave, R.; Rumps, J.; Pagano, L. Restoring rivers
one reach at a time: Results from a survey of US river restoration practitioners. Restoration
Ecology 2007, 15 (3), 482-493.
6. Kondolf, G. M.; Boulton, A. J.; O'Daniel, S.; Poole, G. C.; Rachel, F. J.; Stanley, E. H.; Wohl,
E.; Bang, A.; Carlstrom, J.; Cristoni, C.; Huber, H.; Koljonen, S.; Louhi, P.; Nakamura, K. Process-
based ecological river restoration: Visualizing three-dimensional connectivity and dynamic
vectors to recover lost linkages. Ecol Soc 2006, 11 (2), 5.
7. Tockner, K.; Stanford, J. A. Riverine flood plains: Present state and future trends. Environmental
Conservation 2002, 29 (3), 308-330.
8. Kaushal, S. S.; Groffman, P. M.; Band, L. E.; Shields, C. A.; Morgan, R. P.;Palmer, M. A.;Belt,
K. T.;Swan, C. M.;Findlay, S. E. G.;Fisher, G.T. Interaction between urbanization and climate
variability amplifies watershed nitrate export in Maryland, Environ. Sci. Technol. 2008, 42 (16),
5872-5878.
9. Vidon, P.; Allan, C.; Burns, D.; Duval, T. P.; Gurwick, N.; Inamdar, S.; Lowrance, R.; Okay,
J.; Scott, D.; Sebestyen, S. Hot spots and hot moments in riparian zones: potential for improved
water quality management1. J. Am. Water Resour. Assoc. 2010, 46 (2), 278-298.
10. McClain, M. E.; Boyer, E. W.; Dent, C. L.; Gergel, S. E.; Grimm, N. B.; Groffman, P. M.;
Hart, S. C.; Harvey, J. W.; Johnston, C. A.; Mayorga, E.; McDowell, W. H.; Pinay, G.
Biogeochemical hot spots and hot moments at the interface of terrestrial and aquatic ecosystems.
Ecosystems 2003, 6 (4), 301-312.
11. Tockner, K.; Pennetzdorfer, D.; Reiner, N.; Schiemer, F.; Ward, J. V. Hydrological
connectivity, and the exchange of organic matter and nutrients in a dynamic river-floodplain
system (Danube, Austria). Freshw. Biol. 1999, 41 (3), 521-535.
12. Jones, C. N.; Scott, D. T.; Edwards, B. L.; Keim, R. F. Perirheic mixing and biogeochemical
processing in flow-through and backwater floodplain wetlands. Water Resour. Res. 2014, 50 (9),
7394-7405.
13. Chow, A. T.; Dai, J. N.; Conner, W. H.; Hitchcock, D. R.; Wang, J. J. Dissolved organic matter
and nutrient dynamics of a coastal freshwater forested wetland in Winyah Bay, South Carolina.
Biogeochemistry 2013, 112 (1-3), 571-587.
14. Noe, G. B.; Hupp, C. R.; Rybicki, N. B. Hydrogeomorphology influences soil nitrogen and
phosphorus mineralization in floodplain wetlands. Ecosystems 2013, 16 (1), 75-94.
60
15. Wolf, K. L.; Noe, G. B.; Ahn, C. Hydrologic connectivity to streams increases nitrogen and
phosphorus inputs and cycling in soils of created and natural floodplain wetlands. Journal of
Environmental Quality 2013, 42 (4), 1245-1255.
16. Baldwin, D. S.; Mitchell, A. M. The effects of drying and re-flooding on the sediment and soil
nutrient dynamics of lowland river-floodplain systems: A synthesis. Regul. Rivers-Res. Manage.
2000, 16 (5), 457-467.
17. Jicha, T. M.; Johnson, L. B.; Hill, B. H.; Regal, R. R.; Elonen, C. M.; Pearson, M. S. Spatial
and temporal patterns of nitrification rates in forested floodplain wetland soils of upper Mississippi
River pool 8. River Research and Applications 2013, 30 (5), 650-662.
18. Owens, P. N.; Walling, D. E. The phosphorus content of fluvial sediment in rural and
industrialized river basins. Water Research 2002, 36 (3), 685-701.
19. Mertes, L. A. K. Documentation and significance of the perirheic zone on inundated
floodplains. Water Resour. Res. 1997, 33 (7), 1749-1762.
20. Noe, G. B.; Hupp, C. R. Seasonal variation in nutrient retention during inundation of a short-
hydroperiod floodplain. River Research and Applications 2007, 23 (10), 1088-1101.
21. Bechtold, J. S.; Edwards, R. T.; Naiman, R. J, Biotic versus hydrologic control over seasonal
nitrate leaching in a floodplain forest. Biogeochemistry 2003, 63 (1), 53-71.
22. Vymazal, J. Removal of nutrients in various types of constructed wetlands. Sci Total Environ
2007, 380 (1–3), 48-65.
23. Forshay, K. J.; Stanley, E. H. Rapid nitrate loss and denitrification in a temperate river
floodplain. Biogeochemistry 2005, 75 (1), 43-64.
24. Friedrich, J.; Dinkel, C.; Grieder, E.; Radan, S.; Secrieru, D.; Steingruber, S.; Wehrli, B.
Nutrient uptake and benthic regeneration in Danube Delta lakes. Biogeochemistry 2003, 64 (3),
373-398.
25. Hein, T.; Baranyi, C.; Herndl, G. J.; Wanek, W.; Schiemer, F. Allochthonous and
autochthonous particulate organic matter in floodplains of the River Danube: the importance of
hydrological connectivity. Freshw. Biol. 2003, 48 (2), 220-232.
26. Kadlec, R. H.; Reddy, K. Temperature effects in treatment wetlands. Water Environ. Res. 2001,
543-557.
27. Harrison, M.; Groffman, P.; Mayer, P.; Kaushal, S. Nitrate removal in two relict oxbow urban
wetlands: a 15N mass-balance approach. Biogeochemistry 2012, 111 (1-3), 647-660.
28. Welti, N.; Bondar-Kunze, E.; Singer, G.; Tritthart, M.; Zechmeister-Boltenstern, S.; Hein, T.;
Pinay, G. Large-scale controls on potential respiration and denitrification in riverine floodplains.
Ecol Eng 2012, 42, 73-84.
61
29. Scott, D. T.; Keim, R. F.; Edwards, B. L.; Jones, C. N.; Kroes, D. E. Floodplain biogeochemical
processing of floodwaters in the Atchafalaya River Basin during the Mississippi River flood of
2011. J. Geophys. Res.-Biogeosci. 2014, 119 (4), 537-546.
30. Racchetti, E.; Bartoli, M.; Soana, E.; Longhi, D.; Christian, R. R.; Pinardi, M.; Viaroli, P.
Influence of hydrological connectivity of riverine wetlands on nitrogen removal via denitrification.
Biogeochemistry 2011, 103 (1-3), 335-354.
31. Huber, B.; Luster, J.; Bernasconi, S. M.; Shrestha, J.; Pannatier, E. G. Nitrate leaching from
short-hydroperiod floodplain soils. Biogeosciences 2012, 9 (11), 4385-4397.
32. Kreiling, R. M.; Schubauer-Berigan, J. P.; Richardson, W. B.; Bartsch, L. A.; Hughes, P. E.;
Cavanaugh, J. C.; Strauss, E. A. Wetland management reduces sediment and nutrient loading to
the upper Mississippi River. J. Environ. Qual. 2013, 42 (2), 573-583.
33. Jin, S.; Yang, L.; Danielson, P.; Homer, C.; Fry, J.; Xian, G. A comprehensive change detection
method for updating the National Land Cover Database to circa 2011. Remote Sens Environ 2013
(132), 159-175.
34. Thomspson, T. W.; Hession, W. C.; Scott, D. T. The StREAM Lab at Virginia Tech. Resource
Magazine 2012, 8-9.
35. Azinheira, D. L.; Scott, D. T.; Hession, W.; Hester, E. T. Comparison of effects of inset
floodplains and hyporheic exchange induced by in-stream structures on solute retention. Water
Resour. Res. 2014, 50 (7), 6168-6190.
36. Hester, E.T.; Guth, C.R.; Scott, D.T.; Jones, C.N. Surface water-groundwater exchange
processes in an experimental floodplain along a restored headwater stream. In Review. 2015.
37. Covino, T. P.; McGlynn, B. L.; McNamara, R. A. Tracer Additions for Spiraling Curve
Characterization (TASCC): Quantifying stream nutrient uptake kinetics from ambient to
saturation. Limnology and Oceanography-Methods 2010, 8 (9), 484-498.
38. Wollheim, W. M.; Harms, T. K.; Peterson, B. J.; Morkeski, K.; Hopkinson, C. S.; Stewart, R.
J.; Gooseff, M. N.; Briggs, M. A. Nitrate uptake dynamics of surface transient storage in stream
channels and fluvial wetlands. Biogeochemistry 2014, 120 (1-3), 239-257.
39. Newbold, J. D.; O'Neill, R. V.; Elwood, J. W.; Winkle, W. V. Nutrient spiralling in streams:
implications for nutrient limitation and invertebrate activity. Am. Nat. 1982, 120 (5), 628-652.
40. Anderson, M. J. A new method for non-parametric multivariate analysis of variance. Austral
Ecol 2001, 26 (1), 32-46.
41. Wilcox, R. Robust multivariate regression when there is heteroscedasticity. Commun Stat
Simulat. 2009, 38 (1), 1-13.
42. R Core Development Team. R: A language and environment for statistical computing, R
Foundation for Statistical Computing: Vienna, Austria, 2013.
62
43. Oksanen, J.; Blanchet, F. G.; Kindt, R.; Legendre, P.; Minchin, R. R.; O'Hara, R. B.; Simpson,
G. L.; Solymos, P.; Stevens, M. H. H.; Wagner, H. Vegan: Community Ecology Package.
Retreived from: http://CRAN.R-project.org/package=vegan: 2013.
44. Wilcox, R.R., & Schönbrodt, F.D. The WRS package for robust statistics in R (version 0.24).
Retrieved from http://r-forge.r-project.org/projects/wrs/: 2014.
45. Hughes, D. A. Floodplain inundation - processes and relationships with channel discharge.
Earth Surf. Process. Landf. 1980, 5 (3), 297-304.
46. Knight, D. River flood hydraulics: Theoretical issues and stage-discharge relationships. River
basin modelling for flood risk mitigation 2006, 17, 301-334.
47. Jones, K. L.; Poole, G. C.; O'Daniel, S. J.; Mertes, L. A. K.; Stanford, J. A. Surface hydrology
of low-relief landscapes: Assessing surface water flow impedance using LIDAR-derived digital
elevation models. Remote Sens Environ 2008, 112 (11), 4148-4158.
48. Helton, A. M.; Poole, G.; Payn, r.; Izurieta, C.; Stanford, J. Relative influences of the river
channel, floodplain surface, and alluvial aquifer on simulated hyrologic residence time in a
montane river floodplain. Geomorphology 2014, 205, 17-26.
49. Thorp, J. H.; Thoms, M. C.; Delong, M. D. The riverine ecosystem synthesis: biocomplexity
in river networks across space and time. River Research and Applications 2006, 22(2),123-147.
50. Fellman, J. B.; Hood, E.; Edwards, R. T.; D'Amore, D. V. Changes in the concentration,
biodegradability, and fluorescent properties of dissolved organic matter during stormflows in
coastal temperate watersheds. J. Geophys. Res.-Biogeosci. 2009, 114 (G1).
51. Kreiling, R.; De Jager, N.; Swanson, W.; Strauss, E.; Thomsen, M., Effects of flooding on ion
exchange rates in an upper Mississippi River floodplain forest impacted by herbivory, invasion,
and restoration. Wetlands 2015, 1-8.
52. Dupas, R.; Gruau, G.; Gu, S.; Humbert, G.; Jaffrézic, A.; Gascuel-Odoux, C., Groundwater
control of biogeochemical processes causing phosphorus release from riparian wetlands. Water
Research 2015, 84, 307-314.
53. Noe, G. B.; Hupp, C. R. Carbon, nitrogen, and phosphorus accumulation in floodplains of
Atlantic Coastal Plain rivers, USA. Ecological Applications 2005, 15 (4), 1178-1190.
54. Ardon, M.; Montanari, S.; Morse, J. L.; Doyle, M. W.; Bernhardt, E. S. Phosphorus export
from a restored wetland ecosystem in response to natural and experimental hydrologic fluctuations.
J. Geophys. Res.-Biogeosci. 2010, 115 (G4).
55. Reckendorfer, W.; Funk, A.; Gschopf, C.; Hein, T.; Schiemer, F. Aquatic ecosystem functions
of an isolated floodplain and their implications for flood retention and management. J. Appl. Ecol.
2013, 50 (1), 119-128.
56. Hoffmann, C. C.; Kjaergaard, C.; Uusi-Kamppa, J.; Hansen, H. C. B.; Kronvang, B.
Phosphorus retention in riparian buffers: review of their efficiency. Journal of Environmental
Quality 2009, 38 (5), 1942-1955.
63
57. Jarvie, H. P.; Sharpley, A. N.; Scott, J. T.; Haggard, B. E.; Bowes, M. J.; Massey, L. B. Within-
River Phosphorus Retention: Accounting for a Missing Piece in the Watershed Phosphorus Puzzle.
Environ. Sci. Technol. 2012, 46 (24), 13284-13292.
58. Surridge, B. W. J.; Heathwaite, A. L.; Baird, A. J. Phosphorus mobilisation and transport
within a long-restored floodplain wetland. Ecol Eng 2012, 44, 348-359.
59. McJannet, D.; Wallace, J.; Keen, R.; Hawdon, A.; Kemei, J. The filtering capacity of a tropical
riverine wetland: II. Sediment and nutrient balances. Hydrological Processes 2012, 26 (1), 53-72.
60. BryantMason, A.; Xu, Y.-J.; Altabet, M. A. Limited capacity of river corridor wetlands to
remove nitrate - A case study on the Atchafalaya River Basin during the 2011 Mississippi River
flooding. Water Resourc. Res. 2012, 49(1), 283-290.
61. Bonnet, M. P.; Barroux, G.; Martinez, J. M.; Seyler, F.; Moreira-Turcq, P.; Cochonneau, G.;
Melack, J. M.; Boaventura, G.; Maurice-Bourgoin, L.; León, J. G.; Roux, E.; Calmant, S.; Kosuth,
P.; Guyot, J. L.; Seyler, P. Floodplain hydrology in an Amazon floodplain lake (Lago Grande de
Curuaí). J. Hydrol. 2008, 349 (1–2), 18-30.
62. Schagerl, M.; Drozdowski, I.; Angeler, D. G.; Hein, T.; Preiner, S. Water age - a major factor
controlling phytoplankton community structure in a reconnected dynamic floodplain (Danube,
Regelsbrunn, Austria). Journal of Limnology 2009, 68 (2), 274-287.
63. Roley, S. S.; Tank, J. L.; Stephen, M. L.; Johnson, L. T.; Beaulieu, J. J.; Witter, J. D. Floodplain
restoration enhances denitrification and reach-scale nitrogen removal in an agricultural stream.
Ecol. Appl. 2012, 22 (1), 281-297.
64. Mulholland, P. J.; Helton, A. M.; Poole, G. C.; Hall, R. O.; Hamilton, S. K.; Peterson, B. J.;
Tank, J. L.; Ashkenas, L. R.; Cooper, L. W.; Dahm, C. N.; Dodds, W. K.; Findlay, S. E. G.;
Gregory, S. V.; Grimm, N. B.; Johnson, S. L.; McDowell, W. H.; Meyer, J. L.; Valett, H. M.;
Webster, J. R.; Arango, C. P.; Beaulieu, J. J.; Bernot, M. J.; Burgin, A. J.; Crenshaw, C. L.;
Johnson, L. T.; Niederlehner, B. R.; O'Brien, J. M.; Potter, J. D.; Sheibley, R. W.; Sobota, D. J.;
Thomas, S. M., Stream denitrification across biomes and its response to anthropogenic nitrate
loading. Nature 2008, 452, (7184), 202-U46.
65. Hall, R. O.; Tank, J. L.; Sobota, D. J.; Mulholland, P. J.; O'Brien, J. M.; Dodds, W. K.; Webster,
J. R.; Valett, H. M.; Poole, G. C.; Peterson, B. J.; Meyer, J. L.; McDowell, W. H.; Johnson, S. L.;
Hamilton, S. K.; Grimm, N. B.; Gregory, S. V.; Dahm, C. N.; Cooper, L. W.; Ashkenas, L. R.;
Thomas, S. M.; Sheibley, R. W.; Potter, J. D.; Niederlehner, B. R.; Johnson, L. T.; Helton, A. M.;
Crenshaw, C. M.; Burgin, A. J.; Bernot, M. J.; Beaulieu, J. J.; Arango, C. P. Nitrate removal in
stream ecosystems measured by N-15 addition experiments: Total uptake. Limnol. Oceanogr.
2009, 54 (3), 653-665.
66. Earl, S. R.; Valett, H. M.; Webster, J. R. Nitrogen spiraling in streams: comparisons between
stable isotope tracer and nutrient addition experiments. Limnol. Oceanogr. 2007, 52, 1718-1723.
67. Covino, T.; McGlynn, B.; McNamara, R. Land use/land cover and scale influences on in-
stream nitrogen uptake kinetics. J. Geophys. Res.-Biogeosci. 2012, 117 (G2).
64
68. Clement, J. C.; Aquilina, L.; Bour, O.; Plaine, K.; Burt, T. P.; Pinay, G. Hydrological flowpaths
and nitrate removal rates within a riparian floodplain along a fourth-order stream in Brittany
(France). Hydrol Process 2003, 17 (6), 1177-1195.
69. Filoso, S.; Smith, S. M. C.; Williams, M. R.; Palmer, M. A., The efficacy of constructed
stream–wetland complexes at reducing the flux of suspended solids to Chesapeake Bay. Environ.
Sci. Technol. 2015, 49, (15), 8986-8994.
70. Hernandez, M. E.; Mitsch, W. J., Denitrification in created riverine wetlands: Influence of
hydrology and season. Ecol Eng 2007, 30, (1), 78-88.
65
CHAPTER S4. SUPPLEMENTAL INFORMATION
S4.1 Climatic Data Acquisition Climatic Data was collected at the site using the StREAM Lab weather station that has been
deployed since 2011. Hourly evapotranspiration (ET) was obtained using standard procedures
associated with the Bowen Ratio estimation69,70. Relevant measurements were taken hourly using
CS300-QD solar radiation sensors, HC2S3-QD air temperature and relative humidity probes,
CS106 barometric pressure sensor, and a WindSonic4-QD 2-D sonic wind sensor. Hourly rainfall
was recorded using a TE25-QD tipping bucket. Volumetric water content (%) was measured
within the slough using a Decagon GS3 soil moisture probes approximately 10 cm below the soil
surface. Stage in Stroubles Creek was recorded using a stilling basin and Campbell Scientific
CS450 pressure transducer and converted to flow using a 2-stage rating curve developed for the
site. Climatic data can be viewed real time at http://streamlab.bse.vt.edu/.
S4.2 Water Quality Sample Handling and Analysis All water quality samples were collected in 1 L plastic bottles and filtered in the field within 30
min of collection using a peristaltic pump and 0.45 µM Geotech capsule filter. Subsamples for
nutrient analysis were stored in the dark at 4C in 250 mL plastic bottles, while subsamples for
DOC analysis were stored in the dark at 4C in 40 mL pre-combusted amber glass vials. Processed
samples were transported to the VT-BSE Water Quality Lab, stored at 4C, and analyzed within 5
days. Analyses for NH4+ and SRP were completed using a SEAL Analytical AA3 analyzer using
the Berthelot reaction (saliciylate alternative) and the Murphy and Riley method, respectively; Cl-
and NO3- were analyzed using Dionex ICS-3000 ion chromatograph; and DOC and TDN were
measured using high temperature oxidative combustion with a Shimadzu TOC-V CPH.
S4.3 Estimates of Reach-Scale Load Reductions Annual NO3
- removal for the associated restoration reach was estimated using a simple well mixed
reactor model that incorporated measured areal uptake, a simple inundation model, and the
development of synthetic flow record. Here, it is important to highlight the parsimonious and
conservative nature of the presented model. Feedbacks between river-floodplain interactions and
biogeochemical processing are very complex and at this point, poorly characterized18, 44-45.
However, our model estimates potential load reductions utilizing simple measures and provides a
back-of-the-envelope estimate of biogeochemical processing within the StREAM Lab floodplain.
While we acknowledge the uncertainty associated with the resulting load reduction estimates, the
estimates provide an approximate characterization of load reductions in low-order floodplains that
can be applied to restoration projects in the Mid-Atlantic Ridge and Valley.
To estimate load reductions, the StREAM Lab reach was conceptualized as a completely mixed
reactor across a synthetic flow record. For individual storms, mass removal rate (MRemoval , µg N
m-2 min-1) was estimated simply by:
66
removal tot i stormM U A d Equation S4-1
where Utot is total areal uptake, Ai is the estimated area of inundation, and dstorm is the estimated
storm duration. Utot is defined by the highest observed areal uptake observed during the during the
tracer experiments (2800 µg N m-2 min-1). The restored reach, and associated floodplain, serves
as the modeled control volume. For simplicity, Cin was held constant at 100 µM NO3-. While this
concentration is consistent for storm flow observations in the StREAM Lab, it does not take into
account seasonal differences in NO3- concentrations or dilution typically observed during storm
events. Therefore, it is important to highlight that our model estimates characterize potential
removal.
Ai was defined using a simple inundation model developed using methods similar to Jones et al.,
2008.46 A 3 meter digital elevation model (DEM) was developed for the site utilizing airborne
LiDAR data provided from the Town of Blacksburg. The DEM was then normalized to stream
slope by subtracting a “stream-slope” raster from the DEM. The stream slope raster was developed
by using inverse-distance weighting interpolation of surface-water elevations and applied across
the extent of the original DEM. Then, conditional raster algebra was utilized to identify inundated
areas for a given stage (Figure S4-1a), and stage was converted to flow using a rating curve
developed for the site. This resulted in relationship between Ai and Q (Figure S4-1b).
Figure S4-1. (Left) Map of estimated inundation experienced across the StREAM Lab Floodplain
at a streamflow of 8 cms. Note, the study site is denoted at the bottom of the page. (Right) Plot
displaying the relationship between inundation area (Ai) and streamflow (Q).
Because the StREAM Lab has a relatively short flow record (4 yrs), it was necessary to develop a
synthetic flow record to better characterize potential removal in the floodplain. This was done by
developing a flow duration curve (FDC) of mean daily flow (QMDF) from regional statistics,
67
developing a 1000 yrs of synthetic flow record, and finally identifying overbank events during that
record. Using methods outlined by Ssegane et al., 2013 71, regional regression equations were
developed to estimate the FDC experienced at the StREAM Lab gage. To parameterize the
regression estimates, the USGS stream gaging network72 and associated watershed attributes from
the NHDplus database73 were utilized. More specifically, gages within the Ridge and Valley
Physiographic Region74 with watersheds less than 140 km2 were chosen for the analysis (n=227).
After optimization, regression variables were watershed area (WS), percent pasture landcover as
defined by NLCD 200132 (pasture), and mean annual rainfall (precip):
31 2
0[ ] [ ] [ ]MDFQ WS pasture precip Equation S4-2
Table S4-1. Regional Regression Coefficient Values
% Exceedence β0 β1 β2 β3 R2 P
99 -10.42 1.00 -0.05 1.82 0.69 <0.001
95 -7.35 0.96 -0.04 1.28 0.67 <0.001
90 -5.85 0.94 -0.03 1.02 0.66 <0.001
80 -4.58 0.91 -0.03 0.78 0.64 <0.001
70 -4.42 0.88 -0.02 0.72 0.61 <0.001
60 -4.92 0.56 -0.02 0.76 0.59 <0.001
50 -4.90 0.83 -0.02 0.73 0.54 <0.001
40 -4.94 0.80 -0.01 0.71 0.50 <0.001
30 -5.6 0.76 -0.01 0.78 0.46 <0.001
20 -6.26 0.73 -0.01 0.84 0.42 <0.001
10 -7.51 0.69 -6.2e-4 0.98 0.36 <0.001
5 -9.81 0.68 0.01 1.27 0.32 <0.001
1 -13.58 0.65 0.01 1.75 0.25 <0.001
where β0 , β1, β2, and β3 are regression coefficients defined for exceedence intervals 0.99, 0.95,
0.90, 0.80, … , 0.10, 0.05, 0.01 (Table S4-1). Then, the regression equations were applied to the
StREAM Lab gage’s watershed and a Log Pearson type III (LPIII) distribution was fit to the results
using a maximum-likelihood fitting technique75 (Table S4-2, Figure S4-2).
Table S4-2. Log Pearson Type 3 Parameters
Shape (α) 1.51
Rate (β) 7.65
68
Figure S4-2. Parameterized Flow Duration Curve (FDC) at the StREAM Lab watershed. Grey
points result from regional regression, while the dashed line is the Log Pearson Type III
distribution fit to points at specific exceedence levels.
Then, the 1000 yr flow record was synthesized by resampling the parametrized LPIII distribution
using a random sampling technique developed by Ahrens and Dieter 1982.76 Here, it is important
to note that the synthetic record consists of QMDF, which does accurately capture overbank storm
events in low-order streams. 77-79
Similar to methods developed by Tan et al., 2007 82, the continuous flow record was discretized
into discrete subdaily hydrographs. Here, it is assumed the volume of water represented by
integrating QMDF across 24 hours equals the volume of stormflow and baseflow across that same
day:
MDF BF SFQ dt Q dt Q dt Equation S4-3
where QBF and QSF represent baseflow and stormflow, respectively. QBF was defined as 0.2 cms,
which is consistent with baseflow observed during the summer. QSF was estimated using a
rectangular hydrograph assumption and known relationship between instantaneous peak flow
(QIPF) and QMDF77, 80, 81:
IPF s MDFQ C Q Equation S4-4
( ) 1MDF BF
storm
s MDF
Q Q dayd
C Q
Equation S4-5
where CS is the steepness index defined by the relationship between QMDF and QIPF at the StREAM
Lab gage.
Bankful flow (QBKF), was used to identify individual overbank events over the course of the
synthetic flow record. Then, the relative NO3- removal was calculated as the ratio of Mremoval and
the total NO3- load and tabulated for individual storm events:
69
Re% movalStorm
in
MRemoval
QC
Equation S4-6
Then, relative annual NO3- load reductions were tabulated across the synthetic flow record by
calculating the ratio of Mremoval, across the storm events in the respective year, to total annual load
experienced across the respective year:
Re
%
moval
year
Annual
in
M
RemovalQC
Equation S4-7
S4.4 Works Cited 1. Irmak, S., NEBFLUX. ASABE 2010, (53), 1098-1115.
2. Campbell Scientific Inc., C. S. Bowen Ratio Instrumentation Manual. Logan, UT, 1998.
3. Jones, C. N.; Scott, D. T.; Edwards, B. L.; Keim, R. F. Perirheic mixing and biogeochemical
processing in flow-through and backwater floodplain wetlands. Water Resour. Res. 2014, 50 (9),
7394-7405.
4. Hughes, D. A. Floodplain inundation - processes and relationships with channel discharge.
Earth Surf. Process. Landf. 1980, 5 (3), 297-304.
5. Knight, D. River flood hydraulics: Theoretical issues and stage-discharge relationships. River
basin modelling for flood risk mitigation 2006, 17, 301-334.
6. Jones, K. L.; Poole, G. C.; O'Daniel, S. J.; Mertes, L. A. K.; Stanford, J. A. Surface hydrology
of low-relief landscapes: Assessing surface water flow impedance using LIDAR-derived digital
elevation models. Remote Sens Environ 2008, 112 (11), 4148-4158.
7. Ssegane, H.; Amatya, D. M.; Tollner, E. W.; Dai, Z.; Nettles, J. E. Estimation of Daily
Streamflow of Southeastern Coastal Plain Watersheds by Combining Estimated Magnitude and
Sequence. J. Am. Water Resour. As. 2013, 49 (5), 1150-1166.
8. USGS National Water Information System. http://nwis.usgs.gov (accessed 2014).
9. Horizon Systems. National Hydrography Dataset Plus: Documentation. 2007
10. Fenneman, N. M.; Johnson, D. W. Physiographic divisions of the conterminous. In USGS, Ed.
Reston, VA, 1946.
11. Jin, S.; Yang, L.; Danielson, P.; Homer, C.; Fry, J.; Xian, G. A comprehensive change detection
method for updating the National Land Cover Database to circa 2011. Remote Sens Environ 2013
(132), 159-175.
12. Venables, W. N.; Ripley, B. D. Modern Applied Statistics with S. Fourth edition. Springer:
2002.
70
13. Ahrens, J. H.; Dieter, U. Generating gamma variates by a modified rejection technique.
Communications of the ACM 1982, 25, 47-54.
14. Langbein, W. Peak discharge from daily records. Water Resour. Bull 1944, 145.
15. Fill, H. D.; Steiner, A. A. Estimating instantaneous peak flow from mean daily flow data. J
Hydrol Eng 2003, 8 (6), 365-369.
16. Tan, K.-S.; Chiew, F. H. S.; Grayson, R. B. A steepness index unit volume flood hydrograph
approach for sub-daily flow disaggregation. Hydrol. Process. 2007, 21 (20), 2807-2816.
17. Fuller, W. E. Flood flows. Trans. Am. Soc. Civ. Eng. 1914, 77, 564-617.
71
CHAPTER 5. THE EFFECT OF HYDROGEOMORPHIC
GRADIENTS AND HYDROCLIMATIC SETTING ON RIVER-
FLOODPLAIN CONNECTIVITY AT THE CONTINENTAL
SCALE
C. Nathan Jones, Durelle T. Scott, Jesus D. Gomez-Velez, Judson W. Harvey
Submitted: anticipated, August 2015
To: Geophysical Research Letters
Status: Under internal review
Abstract Hydrologic connectivity between rivers and their adjacent floodplains control many ecological
processes that are consequential at catchment to continental scales; however little work has been
done to characterize river-floodplain connectivity beyond site specific studies. We developed a
method to examine the effects of contributing catchment size and physiography on river-floodplain
connectivity across the continental United States. Utilizing flow data from over 6,800 USGS
gages, we estimated transient storage volume in floodplains for individual floods across each
gage’s flow record. Floodplain transient storage varied from 5 to 50% of the total flood volume
for individual events, and there were no significant patterns between this proportion and
physiography or catchment area, highlighting the variability of flooding experienced across the
US. However, floodplain transient storage increased with stream order when generalized to the US
river network, both when examining individual reaches and cumulative storage across each stream
order. Further, floodplain transient storage was greatest in the southeastern coastal plain.
Collectively, our results highlight the importance of river-floodplain connectivity in large, lowland
rivers.
5.1 Introduction Feedbacks between regional to continental scale hydrologic and biogeochemical processes control
the timing and magnitude of water and material flux from large catchments. Because of the
complexity and heterogeneity of hydrologic processes at this scale, observations must be paired
with modeling efforts to better understand the effects of perturbations such as urbanization and
climate change. Recent efforts include estimations of nutrient flux from large river basins
[Alexander et al., 2000; Woodward et al., 2012], global and continental fluxes of riverine CO2
[Butman and Raymond, 2011; Raymond et al., 2013], and most recently, basin wide estimates of
the effects of hyporheic connectivity on solute retention and transport [Gomez-Velez and Harvey,
2014; Kiel and Cardenas, 2014]. While these contemporary modeling efforts highlight many
important processes, they typically focus on baseflow or median streamflow conditions. This
neglects discrete flood pulses, where hydrologic connection between the landscape and river
network is enhanced, and as a result, downstream flux of solutes often increases. This is
particularly relevant in catchments dominated by urban and agricultural landuses, where the
majority of the annual solute flux is accounted for during periods of elevated flow [Kaushal et al.,
2014; Royer et al., 2006].
72
Flood pulses drive many ecosystem processes, ranging from controlling ecosystem composition
and structure to the exchange of energy and materials between rivers and their adjacent landscape
[Junk et al., 1989; Stanley et al., 1997; Tockner et al., 2000]. The latter is of interest for
downstream water quality, because floodplains serve as both a source and sink of materials within
a catchment [Noe and Hupp, 2007; Roley et al., 2012; Scott et al., 2014]. During overbank events,
large amounts of nutrient rich sediments and particulate organic matter are deposited on floodplain
surfaces [Curran and Hession, 2013; Hupp, 2000]. In systems that experience high rates of
deposition, these materials are often buried and effectively removed from the river network [Battin
et al., 2009; Ricker and Lockaby, 2015]. In other systems, these depositional materials provide
resources for enhanced biological processing and internal biogeochemical cycling [Welti et al.,
2012; Wolf et al., 2013], potentially increasing the biogeochemical processing of floodwaters and
improvement of downstream water quality [Mitsch et al., 2001; Tockner and Schiemer, 1997].
However, the extent of biogeochemical processing in floodplains is largely controlled by
feedbacks between redox dependent processes (i.e., denitrification) and inundation hydrology
[Helton et al., 2012; Jones et al., 2014]. For example, headwater floodplains are often a source of
nutrients because of the short residence time of floodwaters within the floodplain [Kreiling et al.,
2013; Noe and Hupp, 2007; Roley et al., 2012], while large river-floodplain systems are often a
sink because of prolonged residence times and its effect on redox conditions [Forshay and Stanley,
2005; Racchetti et al., 2011; Scott et al., 2014].
While there are many site specific investigations of feedbacks between inundation hydrology and
biogeochemistry in floodplains (e.g. Kreiling et al. [2013]; Noe and Hupp [2007]; Scott et al.
[2014]), few studies have attempted to quantify the cumulative effects of floodplains on material
expert from large river basins. Mertes [1997] hypothesized that interactions between hydroclimatic
setting and catchment area control inundation hydrology, suggesting that as rivers transitioned
from headwater streams to large rivers, they also shifted from a dominance of mixing in the
hyporheic zone to mixing in the perirheic zone. This hypothesis is partially confirmed by recent
modeling efforts, where hyporheic exchange decreased with increasing stream size and was shown
to be significant source of biogeochemical processing potential [Gomez-Velez and Harvey, 2014].
However, little work has been done to examine variation in river-floodplain interactions across
hydrogeomorphic gradients or hydroclimatic settings; and because river-floodplain interactions
drive biogeochemical processing of floodwaters (e.g., Junk et al., [1989]; Tockner et al. [2000]),
characterizing these large scale controls is an important first step in understanding the effects river-
floodplain connectivity on downstream water quality.
In this letter, we present a methodology to examine river-floodplain connectivity across large
catchments. We apply this methodology to the continental US and quantify the effects of
hydroclimatic setting and hydrogeomorphic gradients on river-floodplain connectivity. Broadly,
river-floodplain connectivity is defined as the exchange of energy and materials between the river
and floodplain [Pringle, 2003; K. Tockner and Stanford, 2002]. Here, we characterize river-
floodplain connectivity by estimating floodplain transient storage volume (VFP), or the volume of
transient storage experienced within a given reach’s floodplain, through (1) the analysis of 20 years
of flow data from over 6,800 USGS gaging station and (2) the application of observed variability
to the entire US stream network.
73
5.2 Methods
Figure 5-1. A.) Rating curve from the Atchafalaya River near Simmesport, LA (USGS Gage
07381515) and a conceptual sketch of a river cross section highlighting the assumption that the
rating curve’s breakpoint represents the threshold for river-floodplain connectivity, or QBKF*. B.)
USGS gages utilized in this study from the GAGESII database across the different physiographic
regions. C.) An annual hydrograph highlighting the total flood volume (VT) as delineated by the
estimate of QBKF*. D.) Plot displaying the relationship between QBKF* and catchment area across
the gaging network. Colors correspond to physiographic regions displayed in B.
2.1 Characterization of Floodplain Transient Storage
Using flow data from over 6,800 USGS gaging stations [USGS, 2014] and the NHDplus database
[Horizon Systems, 2006], we characterized transient storage experienced in the floodplain for
individual floods across each gage’s flow record and then generalize the associated
characterization (i.e., VFP) to the US stream network. In this letter, we provide a brief accounting
of the methods used to accomplish this; however, a more thorough description of the methods and
associated assumptions can be found in the supporting information. Our analysis of the flow data
hinges on (i) identification of individual overbank floods, (ii) application of modified divided
channel method (DCM) to separate flood flow into channel and floodplain components, and (iii)
estimation of the transient volume of those components: channel volume (VC) and VFP. Then, using
information derived from the flow data analyses, we estimate annual floodplain transient storage
volume (S) across the US stream network based on both stream size and physiographic regions.
74
For each gage, we established the threshold for river-floodplain connectivity using a piecewise
regression analysis of paired stage-discharge measurements (Figure 5-1a). The breakpoint between
regression segments often has physical meaning [Bacon and Watts, 1971], and in this application,
we assume it represents the point of incipient flooding, or the minimum flow where river-
floodplain connectivity occurs (QBKF*). This is consistent with hydrogeomorphic theory, which
suggests that there is an abrupt change in cross sectional geometry when stream flow transitions
from channel flow to overbank flooding [Hupp, 2000; Williams, 1978; Wolman, 1955].
Uncertainty associated with this estimate of QBKF* is largely associated with the hysteretic nature
of flood flows and river stage [Domeneghetti et al., 2012; Hughes, 1980; Knight, 2006], flow
measurement error [Kennedy, 1984; Rantz, 1982], and non-stationarity in stream/floodplain
bathymetry [Slater et al., 2015]. While this level of uncertainty may be unsuitable for site level
studies or engineering applications, we believe our estimation of QBKF* provides an acceptable
level of certainty to characterize river-floodplain connectivity at the continental scale. This is
illustrated by the emergence of power-law scaling between estimated QBKF* and catchment area
across gages utilized in this study (Figure 5-1d). The existence of this relationship does not provide
validation of the modeled QBKF* values, but it is consistent with other observations of regional and
continental scale power-law relationships between measured bankfull flow and catchment area
(e.g. Bieger et al., [2015]). See the supplemental material for more information.
We utilized a modification of the DCM (e.g. Chow [1959]) to estimate VFP for individual flood
events across each gage’s record from 1986 to 2006. Because USGS flow data is typically stored
as mean daily flow (QMDF), subdaily hydrographs were discretized using methods similar to Sangal
[1983] and Tan et al. [2007]. This allowed for the inclusion of floods with durations less than 24
hours, which is critical when characterizing floodplain transient storage in headwater streams.
Then, we identified individual overbank events using QBKF* as a threshold across each gages flow
record, respectively, and estimated the total flood volume (VT) for each flood by integrating under
the flood hydrograph with respect to time ( , Figure 5-1c). Then, we used the DCM
to divide flood flow into channel and floodplain components, where VT is assumed to be the sum
of VFP and VC. Similar to VT, we approximated VC by integrating estimated channel velocity (UC)
with respect to the cross sectional area of the channel control volume (AC) and flood duration (
). UC is assumed to be equivalent to bankful velocity and is estimated using
relationships developed by Bjerklie [2007]. AC was estimated using downstream hydraulic
geometry relationships developed by Wilkerson et al. [2014] and application of the DCM active
channel divisions. Finally, we calculated VFP by subtracting VC from VT. This process was
conducted for individual floods identified across each gage’s flow record from 1986 to 2006. See
the supplemental material for more additional information.
To better describe the observed variation in the gaging data, we developed 2 metrics: the flood
volume ratio (FVR) and the storage index (SI). The FVR is the ratio of the summation of VT and
the total volume of streamflow 2006
1986Qdt , respectively, experienced across each gage’s record
from 1986 to 2006 (Equation 5-1). This is a cumulative measure of flooding and generally
describes the upstream hydrology and the hydroclimatic setting of each gage. In contrast, SI is a
ratio of VFP and VT for individual floods and provides a discrete measure of river-floodplain
connectivity that provides an approximation of the proportion of flood flow experienced in the
T FloodV Q dt
C C CV U dA dt
75
floodplain (Equation 5-2). For simplicity, the median SI from each gage was utilized in subsequent
calculations. To better understand patterns associated with stream order and physiography, we
found median values of FVR and SI for each physiographic province across the observed stream
orders. Then, we used Spearmen Correlation (e.g. Spearman [1904]) to test if there was a
significant effect of stream order on median FVR and median SI, respectively, for each
physiographic division.
2006
1986
TVFVR
Qdt
(5-1)
(5-2)
2.2 Application to the US Stream Network
In order to gain perspective on continental scale patterns in floodplain transient storage, we applied
the metrics obtained from gaging station analysis to the US stream network. While the USGS
gaging station coverage is extensive, individual gages only serve as point measurements, and
analysis of river-floodplain connectivity across the entire network is needed to fully understand
spatial variations associated with both physiography and catchment area. Thus, gages were
subdivided into their respective physiographic domains (R) and stream order (ω), as defined by
Fenneman and Johnson [1946] and Strahler [1957]; and, annual floodplain transient storage
volume (SR,ω) was approximated for each subdivision as the product of median observed floodplain
transient storage volume ( ) and median number of events annually ( )Fn .
,R FP FS V n (5-3)
In order to generalize the floodplain transient storage metrics to the US stream network, values of
SR,ω were applied to individual stream reaches with corresponding R and ω values. The NHDplus
stream network and database was used to define network geometry and ω for individual stream
reaches. To summarize the effects of catchment area on river-floodplain connectivity, we
calculated cumulative floodplain transient storage volume by stream order ( ,RS S
) and
length (lreach) weighted mean floodplain transient storage volume for each stream order (
, /R reach reachS S l l
). Similarly, to summarize the patterns associated with physiography
and river-floodplain connectivity, we calculated the length normalized cumulative floodplain
transient storage volume across each physiographic province ( , /R R reach
R
S S l ).
FP
T
VSI
V
FPV
76
5.3 Results and Discussion 5.3.1 Heterogeneity of River-Floodplain Connectivity across USGS Gaging Stations
Figure 5-2. A.) The flood volume ratio (FVR) and B.) median storage index (SI) approximated
from USGS flow records. Lines represent median index values experienced at specific steam
orders across the different physiographic regions, while the 10-90% quantile interval from across
the entire dataset are displayed in grey.
There was considerable variability of river-floodplain connectivity experienced across the USGS
gaging stations, as measured by the both the SI and FVR (Figure 5-2). These results are consistent
with recent efforts to understand patterns and mechanisms associated with flood peak distributions
across the US (e.g. Mallakpour and Villarini [2015]; Slater et al. [2015]; Villarini et al. [2009]),
where variability in flooding is typically attributed to heterogeneity of upstream hydrology and the
temporal variability of hydroclimatic conditions. Because of economic and social consequences of
flooding, the majority of these analyses focus on the distribution of flood peaks and “mixtures” of
flood generating mechanisms (e.g. Hirschboeck [1988]; Murphy [2001]; Villarini and Smith
[2010]). Our analysis represents a significant departure from these studies, in that we examined
flood volumes as opposed to peak flow, which has consequences for solute transport and ecological
function. Here, we believe further investigation is needed to fully realize the benefits of
investigations of distributions of integrative flood volume.
When examining the FVR in relation to stream order, two groups of physiographic divisions
became apparent: rainfall and snowmelt dominated regions as displayed in blue and orange,
77
respectively, in Figures 1b and 2. These opposing hydroclimatic regimes are large scale controls
that influence feedbacks between hydrogeomorphic and ecological processes [Montgomery,
1999]. The Appalachian Highlands, Atlantic Plain, and Interior Highlands physiographic divisions
all had increasing and statistically significant correlations between median FVR and stream order.
Similarly, in an analysis of flood generating mechanisms at USGS gages across the Eastern US,
Villarini and Smith [2010] found that flood peak distribution parameters were strongly tied to basin
area and flood generating mechanisms were typically associated with atmospheric flooding (e.g.
organized convective thunderstorms, tropical cyclones, etc.). In contrast, the Intermontane
Plateaus, Laurentian Upland, Pacific Mountains, and Rocky Mountains all experienced decreasing
or statistically insignificant trends between FVR and stream order. Here, we attribute this trend to
the overwhelming proportion of the annual water budget experienced during seasonal snowmelt
runoff events, or in the case of drylands of the Intermontane Plateaus, seasonal monsoonal events.
Interestingly, the Interior Plains physiographic domain experiences a positive, yet statistically
insignificant trend. We attribute this to the gradient from snowmelt and rain on snow/ice to summer
convective thunderstorm flood generating mechanisms experienced from north to south in the
physiographic domain [Villarini et al., 2011]. Differences in these two hydroclimatic regimes
ultimately control variability in river-floodplain connectivity experienced across the two groups
of physiographic divisions.
Our examination of median SI highlights the temporal and spatial variability of transient storage
volume (e.g. VFP) experienced across different floodplain environments. Because of computational
and measurement constraints, few studies estimate the water budget of floodplains during
overbank events (e.g. Alsdorf et al. [2005]; Scott et al. [2014]); yet, this is a critical variable in
understanding potential biogeochemical processing within floodplain environments from site to
catchment scales. Our estimation of SI provides a coarse approximation of river-floodplain
exchange, with median values ranging from 5 to 50% across physiographic regions. This is
consistent with flume studies of two stage channels (e.g., Ackers [1993]), where VFP ranged from
approximately 10-50% of VT and was a function of mean water levels in both channel and
floodplain control volumes, respectively. While SI estimates provide an important first step in
understanding the role of transient storage in floodplain across the river network, SI does not
explain mixing processes that regulate residence time of floodwater and ultimately control
depositional and biogeochemical processes [Hughes, 1980; Jones et al., 2014].
78
5.3.2 Effect of Catchment Area on River-Floodplain Connectivity
Figure 5-3. A.) Map of estimated floodplain transient storage volume in 1st to 8th order streams
across the US. Note, rivers greater than 8th order (e.g. Lower Mississippi River) were not included
in this analysis and are displayed in grey. B.) Cumulative floodplain transient storage volume
across the stream network. C.) Length weighted mean floodplain transient storage volume across
the stream network. Diamonds are based on estimates from median values of floodplain transient
storage volume assigned to the stream network, while bars represent the 10th and 90th quantiles of
floodplain transient storage volume.
79
When applied to the US stream network, our results show that river-floodplain connectivity
generally increases with stream order (Figure 5-3). This is consistent with hydrogeomorphic
theory, which suggests that as river networks transition from headwater streams to large rivers,
flooding transitions from short episodic pulses to longer, sustained floods [Junk et al., 1989]. This
pattern in flooding leads to annual inundation times ranging from hours in many headwater streams
(e.g. Harrison et al. [2014]) to months in large river systems (e.g. Scott et al. [2014]). Therefore,
it is intuitive and unsurprising that transient storage in floodplains would increase with increasing
stream order (Figure 5-3c). When examining cumulative SR,ω across the river network
(Figure 5-3b), floodplain transient storage volume generally increases with stream order. This
highlights the importance of river-floodplain connectivity in large river systems.
In many ways, patterns of river-floodplain connectivity experienced along the river corridor are
the inverse of patterns associated with surface water-groundwater (SW-GW) connectivity [Mertes,
1997]. The hyporheic zone is often defined as the subsurface mixing zone where water and
materials are exchanged between surface water and groundwater [Gooseff, 2010; White, 1993;
Winter et al., 1998]. Similarly, the perirheic zone is defined as the surface mixing zone where
water and materials are exchanged between the channel and floodplain [Jones et al., 2014; Mertes,
1997]. While the number and heterogeneity of hyporheic flowpaths generally increases with
increasing stream size [Stanford and Ward, 1993], hyporheic flowpaths generally convey a smaller
portion of total flow with increasing stream size [Wondzell, 2011]. Because of the dendritic nature
of stream networks, and the overwhelming number of small order streams when compared to
higher order streams, this gradient in SW-GW connectivity highlights the importance of hyporheic
flow in headwater streams at the catchment scale [Gomez-Velez and Harvey, 2014; Kiel and
Cardenas, 2014]. Paired with our results, these studies cumulatively suggest that as the river
corridor shifts from headwater streams to large rivers, mixing processes shift from a dominance of
hyporheic flowpaths and SW-GW connectivity to perirheic flowpaths and river-floodplain
connectivity.
Feedbacks between residence times of floodwaters and river-floodplain exchange largely control
biogeochemical processing within floodplains [Jones et al., 2014]. Here, we hypothesize that
similar to the observed increase in transient storage volume with stream order, there is also an
increase in floodwater residence time with increasing stream size. In headwater floodplains with
relatively short residence times (e.g. hours), the dominant biogeochemical mechanism is flushing
and the floodplain is a source of reactive solutes [Harrison et al., 2014; Kreiling et al., 2013; Noe
and Hupp, 2007; Roley et al., 2012]. In contrast, biogeochemical cycling in medium to large rivers
is often suppressed because of prolonged residence times and transport limitation [Forshay and
Stanley, 2005; Racchetti et al., 2011]. Therefore, further investigation is needed to understand
biogeochemical processing potential of floodplains across large catchments.
80
5.3.3 Effect of Physiography on River-Floodplain Connectivity
Figure 5-4. The summation of length normalized floodplain transient storage volume for each
physiographic province.
Finally, we see distinct patterns of river-floodplain connectivity across the landscape (Figure 5-4).
When observing the length normalized cumulative floodplain transient storage volume across each
physiographic province, we find that floodplain transient storage volume is most prevalent in the
Coastal Plain and Central Lowlands. In the coastal plain, this is largely driven by a combination
of expansive floodplains formed during the pluvial period 18,000 to 10,000 years ago [Hupp, 2000]
and regular occurrence of tropical cyclones [Villarini and Smith, 2010]. This is in contrast to high
gradient streams with confined valleys typically found in Rocky Mountains and the relatively dry
climate experienced across the southwestern US. Similar to patterns seen in the gaging station time
series, our results confirm that the contemporary hydroclimatic setting is a dominant variable
affecting river-floodplain connectivity.
5.4 Conclusion We developed a methodology to characterize river-floodplain connectivity across large river
basins. We applied this methodology to the US River network and characterized the effects
contributing catchment area and physiography on floodplain transient storage volume. The
analysis consisted of quantifying floodplain transient storage volume at over 6,800 gages across
the US; then, we generalized information about river-floodplain connectivity to the greater river
network. Largely, our results suggest:
1. There is considerable spatial heterogeneity and temporal variability associated with
floodplain transient storage volume experienced across the USGS gaging stations.
81
However, floodplain transient storage volume accounted for 5 to 50% of the total flood
volume for most flood events.
2. Floodplain transient storage volume increased along the river corridor. Our results suggest
that annual floodplain transient storage volume increased with stream order for both
individual stream reaches and cumulatively across the river network.
3. The Coastal Plain, Great Plains, and Central Lowlands are hotspots for river-floodplain
connectivity. This is due to their hydroclimatic setting and geomorphic form.
Our analysis represents an important first step in quantifying the cumulative effects of floodplains
on downstream hydrology and water quality. We characterized river-floodplain connectivity by
estimating the volume of transient storage in the floodplain. However, in order to understand
reactive transport processes, it will be critical to quantify residence times of floodwaters within
floodplains and also the exchange rates between the river and floodplain (e.g. Harvey et al. [2013]).
Our results suggest that river-floodplain connectivity is an important variable in the fate and
transport of materials, especially in medium to large river systems, and should be included in
contemporary modeling effort to understand regional to continental scale processes.
5.5 Works Cited Ackers, P. (1993), Stage-discharge functions for two-stage channels: The impact of new
research, J. Inst. Water. Env. Man., 7(1), 52-61.
Alexander, R. B., R. A. Smith, and G. E. Schwarz (2000), Effect of stream channel size on the
delivery of nitrogen to the Gulf of Mexico, Nature, 403(6771), 758-761.
Alsdorf, D., T. Dunne, J. Melack, L. Smith, and L. Hess (2005), Diffusion modeling of
recessional flow on central Amazonian floodplains, Geophys. Res. Lett., 32(21).
Bacon, D. W., and D. G. Watts (1971), Estimating the Transition between Two Intersecting
Straight Lines, Biometrika, 58(3), 525-534.
Battin, T. J., S. Luyssaert, L. A. Kaplan, A. K. Aufdenkampe, A. Richter, and L. J. Tranvik
(2009), The boundless carbon cycle, Nat Geosci, 2(9), 598-600.
Bjerklie, D. M. (2007), Estimating the bankfull velocity and discharge for rivers using remotely
sensed river morphology information, J. Hydrol., 341(3–4), 144-155.
Butman, D., and P. A. Raymond (2011), Significant efflux of carbon dioxide from streams and
rivers in the United States, Nat Geosci, 4(12), 839-842.
Chow, V. T. (1959), Open-channel hydraulics, McGraw-Hill.
Curran, J. C., and W. C. Hession (2013), Vegetative impacts on hydraulics and sediment
processes across the fluvial system, J. Hydrol., 505(0), 364-376.
Fenneman, N. M., and D. W. Johnson (1946), Physiographic divisions of the conterminous,
edited by USGS, Reston, VA.
Forshay, K. J., and E. H. Stanley (2005), Rapid nitrate loss and denitrification in a temperate
river floodplain, Biogeochemistry, 75(1), 43-64.
82
Gomez-Velez, J. D., and J. W. Harvey (2014), A hydrogeomorphic river network model predicts
where and why hyporheic exchange is important in large basins, Geophys. Res. Lett., 41(18),
6403-6412.
Gooseff, M. N. (2010), Defining hyporheic zones–advancing our conceptual and operational
definitions of where stream water and groundwater meet, Geography Compass, 4(8), 945-955.
Harrison, M. D., A. J. Miller, P. M. Groffman, P. M. Mayer, and S. S. Kaushal (2014),
Hydrologic Controls on Nitrogen and Phosphorous Dynamics in Relict Oxbow Wetlands
Adjacent to an Urban Restored Stream, J. Am. Water Resour. As., 50(6), 1365-1382.
Harvey, J. W., J. Böhlke, M. A. Voytek, D. Scott, and C. R. Tobias (2013), Hyporheic zone
denitrification: Controls on effective reaction depth and contribution to whole‐ stream mass
balance, Water Resour. Res., 49(10), 6298-6316.
Helton, A. M., G. C. Poole, R. A. Payn, C. Izurieta, and J. A. Stanford (2012), Scaling flow path
processes to fluvial landscapes: An integrated field and model assessment of temperature and
dissolved oxygen dynamics in a river-floodplain-aquifer system, J. Geophys. Res.-Biogeosci.,
117, 13.
Hirschboeck, K. K. (1988), Flood Hydroclimatology, in Flood Geomorphology, p. 27-49, John
Wiley & Sons, New York.
Horizon-Systems, (2006), National Hydrography Dataset Plus, Retreived from
http://www.horizon-systems.com/NHDPlus/index.php
Hughes, D. A. (1980), Floodplain inundation - Processes and relationships with channel
discharge, Earth Surf. Process. Landf., 5(3), 297-304.
Hupp, C. R. (2000), Hydrology, geomorphology and vegetation of Coastal Plain rivers in the
south-eastern USA, Hydrological Processes, 14(16-17), 2991-3010.
Jones, C. N., D. T. Scott, B. L. Edwards, and R. F. Keim (2014), Perirheic mixing and
biogeochemical processing in flow-through and backwater floodplain wetlands, Water. Resour.
Res., 50(9), 7394-7405.
Junk, W. J., P. B. Balyley, and R. E. Sparks (1989), The flood pulse concept in river-floodplain
systems., Can. Spec. Publ. Fish. Aquat. Sci., 106, 110-127.
Kaushal, S. S., P. M. Mayer, and P. G. Vidon (2014), Land use and climate variability amplify
carbon, nutrient, and contaminant pulses: a review with management implications, J. Am. Water
Resour. As., 50(3), 585-614.
Kiel, B. A., and M. B. Cardenas (2014), Lateral hyporheic exchange throughout the Mississippi
River network, Nat Geosci, 7(6), 413-417.
Kreiling, R. M., J. P. Schubauer-Berigan, W. B. Richardson, L. A. Bartsch, P. E. Hughes, J. C.
Cavanaugh, and E. A. Strauss (2013), Wetland Management Reduces Sediment and Nutrient
Loading to the Upper Mississippi River, J. Environ. Qual., 42(2), 573-583.
Mallakpour, I., and G. Villarini (2015), The changing nature of flooding across the central
United States, Nature Clim. Change, 5(3), 250-254.
Mertes, L. A. K. (1997), Documentation and significance of the perirheic zone on inundated
floodplains, Water Resour. Res., 33(7), 1749-1762.
83
Mitsch, W. J., J. W. Day, J. W. Gilliam, P. M. Groffman, D. L. Hey, G. W. Randall, and N. M.
Wang (2001), Reducing nitrogen loading to the Gulf of Mexico from the Mississippi River
Basin: Strategies to counter a persistent ecological problem, Bioscience, 51(5), 373-388.
Montgomery, D. R. (1999), Process domains and the river continuum, J. Am. Water Resour.
Assoc., 35(2), 397-410.
Murphy, P. J. (2001), Evaluation of mixed-population flood-frequency analysis, J. Hydrol. Eng.,
6(1), 62-70.
Noe, G. B., and C. R. Hupp (2007), Seasonal variation in nutrient retention during inundation of
a short-hydroperiod floodplain, River Research and Applications, 23(10), 1088-1101.
Pringle, C. (2003), What is hydrologic connectivity and why is it ecologically important?, Hydro.
Process., 17(13), 2685-2689.
Racchetti, E., M. Bartoli, E. Soana, D. Longhi, R. R. Christian, M. Pinardi, and P. Viaroli (2011),
Influence of hydrological connectivity of riverine wetlands on nitrogen removal via
denitrification, Biogeochemistry, 103(1-3), 335-354.
Raymond, P. A., et al. (2013), Global carbon dioxide emissions from inland waters, Nature,
503(7476), 355-359.
Ricker, M., and B. G. Lockaby (2015), Soil Organic Carbon Stocks in a Large Eutrophic
Floodplain Forest of the Southeastern Atlantic Coastal Plain, USA, Wetlands, 35(2), 291-301.
Roley, S. S., J. L. Tank, M. L. Stephen, L. T. Johnson, J. J. Beaulieu, and J. D. Witter (2012),
Floodplain restoration enhances denitrification and reach-scale nitrogen removal in an
agricultural stream, Ecol. Appl., 22(1), 281-297.
Royer, T. V., M. B. David, and L. E. Gentry (2006), Timing of Riverine Export of Nitrate and
Phosphorus from Agricultural Watersheds in Illinois: Implications for Reducing Nutrient
Loading to the Mississippi River, Environ. Sci. Technol., 40(13), 4126-4131.
Sangal, B. P. (1983), Practical Method of Estimating Peak Flow, J Hydraul Eng-Asce, 109(4),
549-563.
Scott, D. T., R. F. Keim, B. L. Edwards, C. N. Jones, and D. E. Kroes (2014), Floodplain
biogeochemical processing of floodwaters in the Atchafalaya River Basin during the Mississippi
River flood of 2011, J. Geophys. Res.-Biogeosci., 119(4), 537-546.
Slater, L. J., M. B. Singer, and J. W. Kirchner (2015), Hydrologic versus geomorphic drivers of
trends in flood hazard, Geophys. Res. Lett., 2014GL062482.
Spearman, C. (1904), The Proof and Measurement of Association between Two Things, Am. J.
Psychol., 15(1), 72-101.
Stanford, J. A., and J. V. Ward (1993), An Ecosystem Perspective of Alluvial Rivers:
Connectivity and the Hyporheic Corridor, J. N. Am. Benthol. Soc., 12(1), 48-60.
Stanley, E. H., S. G. Fisher, and N. B. Grimm (1997), Ecosystem expansion and contraction in
streams, Bioscience, 47(7), 427-435.
Strahler, A. N. (1957), Quantitative analysis of watershed geomorphology, Eos, Transactions
American Geophysical Union, 38(6), 913-920.
84
Tan, K.-S., F. H. S. Chiew, and R. B. Grayson (2007), A steepness index unit volume flood
hydrograph approach for sub-daily flow disaggregation, Hydro. Process., 21(20), 2807-2816.
Tockner, K., and F. Schiemer (1997), Ecological Aspects of the Restoration Strategy for a River-
Floodplain System on the Danube River in Austria, Global Ecol. Biogeogr., 6(3/4), 321-329.
Tockner, K., and J. A. Stanford (2002), Riverine flood plains: present state and future trends,
Environ. Conserv., 29(3), 308-330.
Tockner, K., F. Malard, and J. V. Ward (2000), An extension of the flood pulse concept, Hydro.
Process., 14(16-17), 2861-2883.
United Station Geological Survey, (2014), National Water Information System, Retreived from
waterdata.usgs.gov/nwis.
Villarini, G., and J. A. Smith (2010), Flood peak distributions for the eastern United States,
Water Resour. Res., 46, W06504, doi:10.1029/2009WR008395.
Villarini, G., F. Serinaldi, J. A. Smith, and W. F. Krajewski (2009), On the stationarity of annual
flood peaks in the continental United States during the 20th century, Water Resour. Res., 45,
W08414, doi:10.1029/2008WR007645
Villarini, G., J. A. Smith, M. L. Baeck, and W. F. Krajewski (2011), Examining Flood
Frequency Distributions in the Midwest U.S., J. Am. Water Resour. Assoc., 47(3), 447-463.
Welti, N., E. Bondar-Kunze, G. Singer, M. Tritthart, S. Zechmeister-Boltenstern, T. Hein, and G.
Pinay (2012), Large-scale controls on potential respiration and denitrification in riverine
floodplains, Ecol Eng, 42, 73-84.
White, D. S. (1993), Perspectives on Defining and Delineating Hyporheic Zones, J. N. Am.
Benthol. Soc., 12(1), 61-69.
Wilkerson, G. V., D. R. Kandel, L. A. Perg, W. E. Dietrich, P. R. Wilcock, and M. R. Whiles
(2014), Continental-scale relationship between bankfull width and drainage area for single-
thread alluvial channels, Water Resour. Res., 50(2), 919-936.
Williams, G. P. (1978), Bank-full discharge of rivers, Water Resour. Res., 14(6), 1141-1154.
Winter, T. C., J. W. Harvey, O. L. Franke, and W. M. Alley (1998), Ground water and surface
water: A single resource, U.S. Geol. Surv. Circ., 1139, 79 pp.
Wolf, K. L., G. B. Noe, and C. Ahn (2013), Hydrologic Connectivity to Streams Increases
Nitrogen and Phosphorus Inputs and Cycling in Soils of Created and Natural Floodplain
Wetlands, J. Environ. Qual., 42(4), 1245-1255.
Wolman, M. G. (1955), The natural channel of Brandywine creek, Pennsylvania, United States
Geological Survey Professional Paper, 271, 56.
Wondzell, S. M. (2011), The role of the hyporheic zone across stream networks, Hydro.
Process., 25(22), 3525-3532.
Woodward, G., et al. (2012), Continental-Scale Effects of Nutrient Pollution on Stream
Ecosystem Functioning, Science, 336(6087), 1438-1440.
85
CHAPTER S5. SUPPLEMENTAL MATERIAL
The overall goal of this study was to explore the effects of geomorphic gradients and hydroclimatic
setting on hydrologic connectivity between rivers and floodplains across the continental US.
Generally, hydrologic connectivity is defined as the water-mediated exchange of matter and energy
between the river and its adjacent landscape [Pringle, 2003; Tockner and Stanford, 2002].
Operationally, connectivity between rivers and their adjacent floodplain, or river-floodplain
connectivity, can be described by the duration of connection, the exchange and flux of water
between the floodplain and river, and the residence time of floodwaters within the floodplain
[Jones et al., 2014]. However, because of the heterogeneous nature of a typical floodplain surface
and its effects on inundation hydrology, the latter variables are difficult to measure at the site level
[Hughes, 1980], thus limiting regional to continental scale assessments of floodplain connectivity.
This includes recent efforts to understand biogeochemical cycles at the network scale (e.g.
Alexander et al., [2000]; Ensign and Doyle, [2006]; Gomez-Velez and Harvey, [2014]). Below, we
propose a simple methodology to analyze river-floodplain connectivity, as characterized by an
estimate of transient storage volume in the floodplain, at the continental scale and apply it across
the US river network.
Our methodology can be broken into two main components: (1) the analysis of 20 years of flow
data from over 6,800 USGS gaging station and (2) the application of the observed variability to
the entire US stream network. Our analysis of the flow data hinges on (i) identification of
individual overbank floods, (ii) application of modified divided channel method (DCM) to separate
flood flow into channel and floodplain components, and (iii) estimation of the volume of those
components: channel volume (VC) and VFP. Then, using information derived from the flow data
analyses, we estimate annual floodplain transient storage (S) across the US stream network based
on both stream size and physiographic regions. Table S5-1 displays definitions of pertinent
variables.
S5.1 Data Acquisition We utilized data from both the USGS stream gaging network [USGS, 2014] and the NHDplus
geospatial database [Horizon Systems, 2014] for this analysis. The USGS stream gaging network
consists of over 20,000 gaging stations that continuously collect a variety of parameters describing
streamflow, water quality, and various other environmental conditions. We chose the gaging
stations identified within GAGESII database (n=6785) for this analysis. These sites are located
along perennial streams or rivers, have a flow record of at least 20 years, and occur within a
watershed that can be reliably delineated [Falcone et al., 2010]. We obtained stage-discharge field
measurements, a corrected stage-discharge relationship, and a 20 year flow record (1986-20006)
for each gage from the National Water Information System (http://waterdata.usgs.gov/nwis).
The NHDplus geodatabase combines data from the National Hydrography Dataset (NHD), the
National Elevation Dataset (NED), and the National Watershed Boundary Dataset (WBD) [McKay
et al., 2012]. The base unit within the geodatabase is the flow line, which represents a relatively
short reach of stream or river. For each flow line, the database contains information about the
individual reach and contributing watershed such as landuse, climatic data, and annual runoff
estimates. For this analysis, we utilized channel slope (SCH), contributing catchment area (AWS),
and Strahler stream order (ω) from the NHDplus geodatabase.
86
Table S5-1. Main variables used in analysis
Variable Definition Description Units
Q -- streamflow [L3T-1]
S -- stage [L]
QBKF* -- threshold for flooding [L3T-1]
QMDF -- mean daily flow [L3T-1]
QIPF , , 1 , 14
2
MDF n MDF n MDF nQ Q Q estimated instantaneous peak flow [L3T-1]
tF -- duration of a flood event [T]
nF -- number of overbank events in a given water year [-]
wBKF
0.191
0.462
2.18 , ln( ) 1.6
1.41 , ln 1.6
WS WS
WS WS
A A
A A
bankful width [L]
AWS -- contributing catchment area [L2]
UBKF 0.31 0.321.37 CHS bankful velocity [LT-1]
SCH -- channel slop [LL-1]
λ 1.1210.2 BKFw meander wavelength [L]
VT Qdt flood volume [L3]
VC * BKFBKFQ dt U dAdt ∬ channel volume [L3]
VFP F CV V floodplain volume [L3]
FVR 2006
1986
TV
Qdt
flood volume ratio [LL-1]
SI FP
T
V
V storage index [LL-1]
ω -- Strahler stream order [-]
R -- physiographic division or province [-]
SR,ω FP fV n estimated annual floodplain transient storage
volume [L3]
Sω ,RS
cumulative floodplain transient storage volume
across a given stream order [L3]
S
,R reach
reach
S l
l
length weighted mean floodplain transient
storage volume across a given stream order [L3]
SR , /R reach
R
S l length normalized cumulative floodplain
transient storage volume for a given
physiographic province
[L3 L-1]
*Note, “~” and “-“ above letters denotes median and mean values, respectively.
87
S5.2 Analysis of USGS Gage Data S5.2.1 Identifying the threshold for connectivity
In order to isolate individual storms across a gage’s flow record, we identified the threshold for
flooding through a piecewise regression analysis, or breakpoint analysis, of paired stage-discharge
measurements. The piecewise regression model consists of two or more general linear models with
each model bound to a specific range of the independent variable. A breakpoint is a point of
transition at the intersection between general linear models that may be interpreted to have physical
meaning, such as a shift in regime [Bacon and Watts, 1971]. For this application, we are utilizing
a two linear segment model, where the breakpoint is interpreted as characterizing bankfull
discharge (QBKF*) when incipient flooding is occurring. The piecewise regression model was
optimized utilizing an iterative procedure developed by Muggeo [2003] and implemented using R
statistical software [R Development Core Team, 2012].
Equation S5-1 displays the linear model chosen to represent stage-discharge measurement pairs.
Here, Q(S) is rating curve function where Q is streamflow and S is measured stage, F1 and F2
represent general linear models, I(S) is a binary function, and c is the breakpoint or threshold stage.
At each gage, we evaluated linear, exponential (log-transformed data), and power (log-log
transformed data) models, and we selected the model with the minimum Root Mean Square Error
(RMSE). The resulting model was considered adequate if: the slopes of the two segments were
significantly different (Davies Test, p<0.05), and if the estimated breakpoint was greater than the
minimum recorded S of the published USGS stage-discharge rating curve. Of the over 6,500 gages
analyzed, 5,981 produced adequate breakpoints.
1 2
1,( )
0,
( ) ( ) ( ) ( ) [1 ( )],Q S F S I S F S I
CI
S
S
SS
C
Equation S5- 1
While the proposed technique provides an adequate method to estimate trends in river-floodplain
connectivity at the continental scale, there is considerable uncertainty associated in estimating
QBKF* utilizing stage-discharge measurements. Hydraulic geometry measures such as cross-
sectional area (AXS), S, and wetted perimeter (WP) exhibit a dramatic shift in relation to stream
flow (Q) as the flow regime shifts from in-channel flow to overbank or flood flows [Hupp, 2000;
Williams, 1978; Wolman, 1955]. Williams [1978] suggested that this shift was most evident in the
relationship between Q and AXS, but could be observed in the relationship between Q and S in areas
where floodplains persist. This is evident in methods utilized by the USGS to develop streamflow
rating curves of compound channels [Rantz, 1982], where breakpoints in the stage-discharge
relationship represent dramatic shifts in hydraulic geometry. The relationship between Q and S is
imperfect in natural channels [Chow, 1959]. Stage-discharge relationships often exhibit hysteretic
behavior over the course of individual storm-hydrographs because of hydraulic factors such as
backwatering, seasonal differences in vegetative roughness, and unsteady flow [Domeneghetti et
al., 2012; Hughes, 1980; Knight, 2006]. As a result, there is often subjectivity in the analysis of
stage-discharge rating curves that allows for experience and site-specific knowledge [Kennedy,
1984]. Other independent variables such as mean velocity and water surface slope are often utilized
as independent variables in rating curves to minimize the effect of hysteresis in the stage-discharge
relationships [Kennedy, 1984; Rantz, 1982]. Further complicating matters, extreme high flows are
88
often estimated through an extrapolation from the indirect slope-area technique, which relies on
the parameterization of the Manning Equation [NRCS, 2012]. Because of their location on the
stage-discharge graph, these extreme events can exhibit substantial leverage in the pairwise
regression analysis, potentially imposing propagation error on the estimated breakpoint.
Figure S5-1. A.) Relationship between QBKF* and catchment area for 5,981 USGS gaging stations
across the continental US. B.) Relationship between bankful depth (dBKF* ) and catchment area.
The black lines represents the relationships derived from data in this study, while the orange line
represents a relationship developed by Beiber et al., [2015].
While these uncertainties in quantifying QBKF* might be unsuitable for site level hydrologic studies
or engineering applications, we believe our estimation of QBKF* provides an acceptable level of
certainty to characterize flooding at the continental scale. This is illustrated by the emergence of
power-law scaling between estimated QBKF* and catchment area across gages utilized in this study
(Figure S5-4a). The existence of this relationship does not provide validation of the modeled QBKF*
values, but it is consistent with other observations of regional and continental scale power-law
relationships between measured bankfull flow and catchment area. More specifically, we
compared the downstream hydraulic geometry relationship developed by Bieger et al., [2015] for
the continental US (n=1,310 sites) to results from our study. Here, bankful depth (dBKF*) was
estimated by subtracting minimum observed stage (Smin) from the stage corresponding to the
estimated breakpoint (SBKF*). Even with the added uncertainty of estimating dBKF*, the
Beiger et al.,[2015] model accounted for a significant amount of variation in the dBKF* values
(p<0.01), providing further evidence the breakpoint analysis adequately defined downstream
hydraulic geometry relationships.
S5.2.2 Estimation of Sub-daily hydrographs
Practical estimation of basin-scale flood characteristics across the historic flow record utilizes
records of mean daily flow (QMDF). Floods of short duration are common in low order streams and
ideally, sub-daily flow measurements would be utilized to characterize overbank floods in streams
where the majority of floods are often less than 24 hours. Several methods have been developed
A.) B.)
89
to estimate instantaneous peak flow (QIPF) from QMDF records [Fill and Steiner, 2003]. The first
utilizes regional regression techniques to estimate the relationship between QMDF and QIPF for a
specific physiographic region or larger basin (e.g., Fuller [1914]; Taguas et al. [2008]). While
regional regression techniques have proven to be an effective estimators of QIPF, they do not
provide any information about the shape of the hydrograph and are not related to the prevailing
hydrologic conditions. In contrast, the second technique estimates QIPF utilizing consecutive QMDF
values (e.g., Langbein [1944]; Sangal [1983]; Fill and Steiner [2003]). Of particular interest,
Sangal [1983] utilizes a triangular hydrograph assumption, where the QIPF is the result of
geometric calculations that assume the area under the triangular hydrograph is equal to the area
under daily hydrograph (Figure S5-2 and Equation S5-2). Tan et al. [2007] combined these
approaches and found a strong linear relationship between estimated and actual QIPF in gages
where the triangular hydrograph assumption was appropriate.
, , 1 , 14
2
MDF n MDF n MDF n
IPF
Q Q QQ
Equation S5- 2
Figure S5-2. Illustration of the Sangal [1983] approximation. (Left) Plot of QMDF over the course
of three days. (Right) The same hydrograph where Equation S5-2 was applied to estimate QIPF.
Here, it is important to note Equation S5-2 is based on the assumption that the volumes represented
by V1 and V2 are equivalent.
For the purposes of this study, the Sangal [1983] triangular hydrograph approximation was used
in order to more accurately represent each gage’s historic flow record. At each gage, the
appropriateness of this correction was tested by developing a linear relationship between QIPF and
QMDF and testing for significance (p<0.05). Note, because annual maximum flows are recorded
within the USGS record, 20 pairs of QIPF and QMDF were available from 1986-2006 to develop this
relationship. As illustrated by data from the Colombia River (USGS Gage 12399500) in
Figure S5-3, most gages exhibit a strong linear relationship between QIPF and QMDF. However,
gages without a significant relationship between QIPF-QMDF were omitted form the analysis. Then,
the triangular hydrograph correction was applied to all days where QMDF was less than QBKF*. This
90
allowed for the identification of floods where QMDF was less than QBKF*, thus providing a better
representation of river-floodplain connectivity in low-order streams.
Figure S5-3. Strong relationship experienced between QIPF and QMDF at the Michigan River near
the Fort Collins, CO (USGS Gage 12399500, catchment area 4 km). Of the 6,800 gages tested,
only 100 produced QIPF -QMDF relationships that were not statistically significant (p<0.05).
S5.2.3 Characterization of river-floodplain connectivity
Figure S5-4. Annual hydrograph from Boulder Creek near Boulder, CO across the 1997-1998 water
year. As delineated by QBKF*, the number of floods (nF) would be equal to 1 for this water year, the
duration of that storm (tF) would be 14 days, and the total flood volume (VT) would approximately 41.5
million m3.
Annual hydrograph characterization
To characterize river-floodplain connectivity across USGS gaging sites, measures of flood
duration (tF), annual number of events (nF), and total flood volume (VT) were estimated for
individual overbank floods across each gage’s flow record from 1986 to 2006. As illustrated by
91
Figure S5-4, discrete overbank floods were identified using QBKF* as a threshold for river-
floodplain connectivity. For each identified overbank flood, values of tF and VT represent the time
stream flow was greater than QBKF* and the volume of stream flow during that time as defined by
Equation S5-3, respectively.
, T FQdtV dt t Equation S5- 3
Floodplain and channel volume characterization
To better characterize river-floodplain connectivity and the potential for biogeochemical
processing at individual gages, we estimated the volume of transient storage in the floodplain, or
volume of floodplain water (VFP), and the volume of channel flow (VC) across each flood’s
hydrograph. The estimation of VFP and VC are based on a combination of estimated bankfull width
(wBKF), bankfull velocity (UBKF), and the application of the divided channel method (DCM).
To estimate the bankfull width (wBKF) at each gage, we took advantage of empirical relationships
relating wBKF to each gage’s catchment area (AWS). Originally applied to natural channels by
Leopold and Maddock [1953], hydraulic geometry measures such as bankfull width (wBKF), mean
bankfull depth (yBKF), and suspended sediment load can often be predicted using power-law
relationships where QBKF or contributing catchment area (AWS) are independent variables [Dunne
and Leopold, 1978; Hey and Thorne, 1986; Wilkerson et al., 2014]. Of particular interest, wBKF is
often utilized as a dependent variable because it can be measured remotely (e.g. Allen and Pavelsky
[2015]) and is less sensitive to measurement error when compared to other hydraulic geometry
measures such as bankfull stage or cross-sectional area [He and Wilkerson, 2011; Wahl, 1977].
For the purposes of this study, wBKF was estimated using relationships developed by Wilkerson et
al. [2014] (Equation S5-4). These relationships are ideal because they are based on a large data set
(n=1018) that were developed from sites at or near USGS gaging stations across the US.
0.191
0.462
2.18 , ln( ) 1.6
1.41 , ln 1.6 B
WS WS
WS WS
KF
A Aw
A A
Equation S5- 4
To estimate UBKF, we utilized empirical relationships where channel parameters associated with
energy regulation (e.g. channel slope, SCH, and meander wavelength, λ) serve as independent
variables. Unlike other hydraulic geometry measures, AWS is not a strong predictor of UBKF [Singh,
2003]. This is because downstream velocity tends to be constant, or gradually increase, as AWS
increases [Leopold, 1953]. For use in remote sensing applications, Bjerklie [2007] derived two
relationships estimating UBKF. The first highlights the connection between meanders and flow
resistance, and similar to Manning’s Equation, estimates UBKF as a function of water surface slope,
modified meander wavelength, and a fitting factor “m” that represents an arbitrary fraction of the
meander length. The derivation of the theoretical equation provides justification for the second
model, an empirical model that relates UBKF with SCH and λ (Equation S5-5). The model is based
on geomorphic surveys and flume studies completed by Church and Rood [1983] and Leopold et
al. [1960], respectively. Bjerklie [2007] goes on to suggest that in the absence of remotely sensed
data, wavelength could be estimated using relationships presented in Leopold and Wolman [1964]
and, more preferable, Williams [1986] (Equation S5-6).
92
0.31 0.321.37 CHBKFU S Equation S5- 5
1.1210.2 BKFw Equation S5- 6
Figure S5-5. Estimate of UBKF at USGS gaging stations using relationship developed by
Bjerklie [2007].
Figure S5-5 displays estimates of UBKF from Equations S5-5 and S5-6. Because Equation S5-5 is
based on a relatively small sample size (n=78), it is less robust than the relationships presented in
Equation S5-4. However, the estimates of UBKF are fairly constant across stream order, which is
consistent with general hydrogeomorphic theory (e.g. Leopold, [1953]). Another alternative to
estimating UBKF would be to divide QBKF* by an estimate of AXS at bankful conditions derived from
continental scale hydraulic geometry relationships (e.g. Andreadis et al. [2013]). While this
approach could potentially provide an adequate estimate of UBKF, error in the estimate would be
compounded by the uncertainty associated with the calculation of QBKF*.
One dimensional flow in compound channels is often modeled with DCM, where flow is calculated
separately in the channel and floodplains compartments [Chow, 1959; Knight, 2006]. The classic
method presented by Chow [1959] does not account for the exchange of momentum or matter.
However, other methods (e.g. Bousmar and Zech [1999]; Knight and Demetriou [1983]; Myers
[1978]) have improved upon this and are extensively reviewed by Knight [2006]. While we are
not modeling one dimensional flow here, we are conceptualizing floods by separating the active
flood flows as occurring in both channel and floodplain compartments (Figure S5-6).
93
Figure S5-6. (Bottom) Conceptualization of the Divided Channel Method (DCM) where the active
storm channel cross section is subdivided into floodplain and channel compartments. (Top) Cross
section displaying the subdivisions of the channel into bankful channel and bankful excess
compartments.
Flood flow is defined conceptually as the sum of channel flow (QC) and transient storage in the
floodplain. Integrating across a single overbank flood hydrograph, VT is the sum of channel flood
volume, VC, and floodplain flow volume, VFP:
T C FPV V V Equation S5- 7
To find VC, it is necessary to further subdivide the channel compartment into the bankful channel
and bankful excess compartments (Figure S5-6). VC is the sum of total volume of flow in the
bankful channel (VBKF) and a bankfull excess compartment that remains associated with the
channel (VBFE).
C BKF BFEV V V
Equation S5- 8 Here, VBKF is calculated by integrating the estimated bankfull flow (QBKF*) over the course of
the storm using:
* BKF BKF
V Q dt Equation S5- 9
Similarly, VBKE is calculated by integrating UBKF across the bankfull excess cross sectional area
over the course of the individual storm:
94
BFE BKFV U dAdt∬
Equation S5- 10
The DCM method’s relatively simple geometry specifies the cross sectional area of the excess
bankfull compartment as equivalent to the product of the bankful width (wBKF) and the stage excess
of bankful (Δy). Equation S5-10 is further simplified by assuming UC is equivalent to UBKF:
BFE BKF BKFV U w y t Equation S5- 11
Once VT and VC are estimated, we calculated VFP by rearranging Equation S5-7 and simply
subtracting VC from VT.
Flood Volume Ratio and Storage Index characterization
Finally, two indices are calculated to characterize river-floodplain connectivity across each gage’s
flow record: the flood volume ratio (FVR) and the storage index (SI). The FVR provides an
understanding of each gages upstream hydrologic conditions and is the ratio of the sum of discrete
estimates of VT and total streamflow experienced across each gages flow record:
2006
1986
TVFVR
Qdt
Equation S5- 12
In contrast, the SI characterizes river-floodplain connectivity experienced during individual floods
and is the ratio of VFP and VT :
FP
T
VSI
V
Equation S5- 13
To explore patterns associated with both physiography and contributing catchment area, we
summarized FVR and SI by physiographic divisions (R) and stream order (ω), as defined by
Fenneman and Johnson [1946] and Strahler [1957] (Tables S5-2 and S5-3). Note, because each
gage experiences a range of SI, the median SI from each gage was used to tabulate values across
the different R and ω subdivisions.
Table S5-2. Median Flood Volume Ratio by Stream Order (ω) and Physiographic Division (R)
Stream
Order
Appalachian
Highlands
Atlantic
Plain
Interior
Highlands
Interior
Plains
Intermontane
Plateaus
Laurentian
Upland
Pacific
Mountains
Rocky
Mountains
1 0.34 0.34 0.18 0.39 0.58 NA 0.56 0.64
2 0.30 0.40 0.22 0.31 0.69 0.26 0.54 0.69
3 0.31 0.38 0.28 0.36 0.47 0.48 0.51 0.66
4 0.36 0.45 0.27 0.34 0.53 0.39 0.50 0.64
5 0.42 0.39 0.24 0.34 0.46 0.34 0.57 0.55
6 0.57 0.63 0.26 0.40 0.48 0.41 0.58 0.54
7 0.48 0.59 0.60 0.52 0.52 NA 0.47 0.57
8 NA 0.72 NA 0.32 NA NA NA NA
95
Table S5-3. Median Storage Index by Stream Order (ω) and Physiographic Division (R)
Stream
Order
Appalachian
Highlands
Atlantic
Plain
Interior
Highlands
Interior
Plains
Intermontane
Plateaus
Laurentian
Upland
Pacific
Mountains
Rocky
Mountains
1 0.26 0.20 0.28 0.23 0.18 NA 0.17 0.08
2 0.22 0.20 0.13 0.20 0.13 0.15 0.23 0.17
3 0.27 0.24 0.21 0.18 0.12 0.13 0.26 0.12
4 0.25 0.23 0.16 0.20 0.16 0.17 0.22 0.12
5 0.24 0.22 0.26 0.19 0.16 0.19 0.21 0.13
6 0.27 0.24 0.25 0.20 0.15 0.11 0.20 0.12
7 0.24 0.19 0.14 0.21 0.15 NA 0.16 0.08
8 NA 0.13 NA 0.19 NA NA NA NA
Then, we used the Spearmen Correlation (e.g. Spearman [1904]) to test if there was a significant
effect of stream order on median FVR and median SI, respectively, for each physiographic
division. We did not find any significant correlations between stream order and SI, but did find
significant correlations between stream order and FVR for a subset of the physiographic divisions
(Table S5-4).
Table S5-4. Spearman Correlation (RS) between FVR
and Stream Order by Physiographic Division
Physiographic Division RS p-value
Appalachian Highlands 0.83 0.02**
Atlantic Plain 0.90 <0.01**
Interior Highlands 0.72 0.07*
Interior Plain 0.27 0.52
Intermontane Plateaus -0.56 0.19
Laurentian Upland 0.31 0.61
Pacific Mountains -0.29 0.53
Rocky Mountains -0.81 0.03**
S5.3 Application to the National Stream Network We then applied the metrics obtained from the USGS gaging stations to the US stream network in
order to gain perspective on continental scale patterns in river-floodplain connectivity. While the
USGS gaging station coverage is extensive, individual gages only serve as point measurements,
and analysis of river-floodplain connectivity across the entire network is needed to fully
understand spatial variations associated with both physiography and catchment area. Thus, gages
were subdivided into their respective physiographic division (R) and stream order (ω), as defined
by Fenneman and Johnson [1946] and Strahler [1957]. We tabulated median values for floodplain
volume (VFP) and annual number of events (nf) for each subdivision, and then we estimated annual
floodplain transient storage (SR,ω):
,R FP fS V n Equation S5-14
96
The results for each subdivision can be found in Table S5-5. We then applied estimates of SR,ω
from each subdivision to flow lines of matching properties within the NHDplus flow network. For
example, the median d for gages along 4th Order streams in the Atlantic Plain was 2 days. All flow
lines of 4th order streams in the Atlantic plain were then assigned d = 2 days. Note, because there
was a lack of USGS gaging stations along 8th order streams in most physiographic divisions, we
utilized median values from all gages along 8th order streams and applied them to 8th order
subdivisions.
Table S5-5. Estimated annual floodplain transient storage (106 m3) by Stream Order (ω) and
Physiographic Division (R)
Strea
m
Order
Appalachi
an
Highlands
Atlant
ic
Plain
Interior
Highlan
ds
Interi
or
Plains
Intermonta
ne
Plateaus
Laurenti
an
Upland
Pacific
Mountai
ns
Rocky
Mountai
ns
1 0.01 0.04 1.19 0.04 0.02 NA 0.02 0.01
2 0.05 0.08 0.02 0.05 0.01 0.01 0.04 0.03
3 0.18 0.22 0.28 0.14 0.03 0.10 0.25 0.06
4 0.63 1.11 0.82 0.57 0.11 0.47 0.77 0.13
5 2.17 2.97 2.82 1.61 0.33 2.25 2.82 0.43
6 8.60 14.37 3.36 3.67 1.40 2.28 2.81 1.11
7 23.24 31.67 7.40 17.83 1.05 NA 41.81 1.42
8 55.23 55.23 55.23 55.23 55.23 55.23 55.23 55.23
To summarize river-floodplain connectivity across hydrogeomorphic gradients (i.e. stream order),
we estimated cumulative floodplain transient storage (Sω) and length weighted mean floodplain
transient storage
( S ) using SR,ω values from the entire NHDplus network (Table S5-6):
Table S5-6. Cumulative floodplain transient storage (Sω) and length weighted mean floodplain
transient storage ( S ) across stream order from the US stream network.
Stream Order
(ω)
Cumulative Floodplain
Transient storage
(1012 m3)
Length Weighted Mean
Floodplain Transient Storage
(106 m3)
1 0.73 0.52
2 0.25 0.40
3 0.40 1.19
4 0.79 4.22
5 1.23 11.5
6 2.52 44.5
7 3.44 134
8 1.09 110
97
,RS S
Equation S5- 15
,R reach
reach
S l
Sl
Equation S5- 16
Finally, we summarized river-floodplain connectivity across physiographic provinces by
calculating the cumulative length normalized floodplain transient storage (SR) values across each
physiographic province (Table S5-7):
, /R R reach
R
S S l Equation S5- 17
Note, here we used physiographic province, as opposed to physiographic division, to examine
regional differences in river-floodplain connectivity. As a subset of physiographic divisions, the
provinces provided a more appropriate resolution to differentiate variation in river-floodplain
connectivity across regions.
Table S5-7. Cumulative length normalized floodplain transient storage (SR) across
physiographic provinces
Physiographic Province Cumulative Floodplain transient storage
(x109 m3/km)
Coastal Plain 9.89
New England 1.36
St. Lawrence Valley 0.08
Valley and Ridge 3.64
Adirondack 0.27
Appalachian Plateaus 2.73
Blue Ridge 0.53
Central Lowland 5.31
Piedmont 4.35
Superior Upland 0.22
Interior Low Plateaus 0.64
Ozark Plateaus 1.41
Ouachita 0.75
Great Plains 8.77
Middle Rocky Mountains 0.09
Southern Rocky Mountains 0.02
Wyoming Basin 0.08
Northern Rocky Mountains 0.19
Basin and Range 0.29
Colorado Plateaus 0.63
Cascade-Sierra Mountains 0.34
Columbia Plateau 0.23
Pacific Border 2.03
Lower Californian 0.01
98
S5.4 Works Cited Alexander, R. B., R. A. Smith, and G. E. Schwarz (2000), Effect of stream channel size on the
delivery of nitrogen to the Gulf of Mexico, Nature, 403(6771), 758-761.
Allen, G. H., and T. M. Pavelsky (2015), Patterns of river width and surface area revealed by the
satellite-derived North American River Width (NARWidth) dataset, Geophys. Res. Lett.,
2014GL062764.
Andreadis, K. M., G. J. P. Schumann, and T. Pavelsky (2013), A simple global river bankfull
width and depth database, Water Resources Research, 49(10), 7164-7168.
Bacon, D. W., and D. G. Watts (1971), Estimating the Transition between Two Intersecting
Straight Lines, Biometrika, 58(3), 525-534.
Bieger, K., H. Rathjens, P. M. Allen, and J. G. Arnold (2015), Development and Evaluation of
Bankfull Hydraulic Geometry Relationships for the Physiographic Regions of the United States,
JAWRA Journal of the American Water Resources Association, n/a-n/a.
Bjerklie, D. M. (2007), Estimating the bankfull velocity and discharge for rivers using remotely
sensed river morphology information, J. Hydrol., 341(3–4), 144-155.
Bousmar, D., and Y. Zech (1999), Momentum transfer for practical flow computation in
compound channels, Journal of hydraulic engineering, 125(7), 696-706.
Chow, V. T. (1959), Open-channel hydraulics, McGraw-Hill.
Church, M., and K. Rood (1983), Catalogue of alluvial river channel regime data, first ed. ,
Department of Geography, University of British Colombia, Vancouver, Canada.
Domeneghetti, A., A. Castellarin, and A. Brath (2012), Assessing rating-curve uncertainty and
its effects on hydraulic model calibration, Hydrology and Earth System Sciences, 16(4), 1191-
1202.
Dunne, T., and L. Leopold (1978), Water in environmental planning, New York, 566-580.
Ensign, S. H., and M. W. Doyle (2006), Nutrient spiraling in streams and river networks, Journal
of Geophysical Research: Biogeosciences, 111(G4), G04009.
Falcone, J. A., D. M. Carlisle, D. M. Wolock, and M. R. Meador (2010), GAGES: A stream gage
database for evaluating natural and altered flow conditions in the conterminous United States,
Ecology, 91(2), 621-621.
Fenneman, N. M., and D. W. Johnson (1946), Physiographic divisions of the conterminous,
edited by USGS, Reston, VA.
99
Fill, H. D., and A. A. Steiner (2003), Estimating instantaneous peak flow from mean daily flow
data, J Hydrol Eng, 8(6), 365-369.
Fuller, W. E. (1914), Flood flows., Trans. Am. Soc. Civ. Eng. , 77, 564-617.
Gomez-Velez, J. D., and J. W. Harvey (2014), A hydrogeomorphic river network model predicts
where and why hyporheic exchange is important in large basins, Geophys. Res. Lett., 41(18),
2014GL061099.
He, L., and G. V. Wilkerson (2011), Improved Bankfull Channel Geometry Prediction Using
Two-Year Return-Period Discharge1, JAWRA Journal of the American Water Resources
Association, 47(6), 1298-1316.
Hey, R., and C. Thorne (1986), Stable Channels with Mobile Gravel Beds, Journal of Hydraulic
Engineering, 112(8), 671-689.
Horizon Systems (2014), NHDplus Dataset, accessed [December, 2014], at URL
[http://www.horizon-systems.com/nhdplus/]
Hughes, D. A. (1980), Floodplain Inundation - Processes And Relationships With Channel
Discharge, Earth Surf. Process. Landf., 5(3), 297-304.
Hupp, C. R. (2000), Hydrology, geomorphology and vegetation of Coastal Plain rivers in the
south-eastern USA, Hydrological Processes, 14(16-17), 2991-3010.
Jones, C. N., D. T. Scott, B. L. Edwards, and R. F. Keim (2014), Perirheic mixing and
biogeochemical processing in flow-through and backwater floodplain wetlands, Water Resources
Research, 50(9), 7394-7405.
Kennedy, E. J. (1984), Discharge ratings at gaging stations: U.S. Geologial Survey Techniques
of Water-Resources Investigations, book 3, chap A10, edited.
Knight, D. (2006), River flood hydraulics: theoretical issues and stage-discharge relationships,
River basin modelling for flood risk mitigation, 17, 301-334.
Knight, D., and J. Demetriou (1983), Flood Plain and Main Channel Flow Interaction, Journal of
Hydraulic Engineering, 109(8), 1073-1092.
Langbein, W. (1944), Peak discharge from daily records, Water Resour. Bull, 145.
Leopold, L. B., and T. Maddock (1953), The hydraulic geometry of stream channels and some
physiographic implications.
Leopold, L. B. (1953), Downstream change of velocity in rivers, American Journal of Science,
251(8), 606-624.
100
Leopold, L. B., and M. G. Wolman (1964), Fluvial processes in geomorphology, edited, San
Francisco: WH Freeman.
Leopold, L. B., R. A. Bagnold, M. G. Wolman, and L. Brush (1960), Flow resistance in sinuous
or irregular channels, paper presented at The Physics of Sediment Transport by Wind and Water,
ASCE.
Mallakpour, I., and G. Villarini (2015), The changing nature of flooding across the central
United States, Nature Clim. Change, 5(3), 250-254.
McKay, L., T. Bondelid, T. Dwald, J. Johnson, R. Moore, and A. Rea (2012), NHDPlus Version
2 User Guide.
Muggeo, V. M. R. (2003), Estimating regression models with unknown break-points, Stat Med,
22(19), 3055-3071.
Myers, W. (1978), Momentum transfer in a compound channel, Journal of Hydraulic Research,
16(2), 139-150.
NRCS. (2012), National Engineering Handbook: Part 630—Hydrology, USDA Soil Conservation
Service: Washington, DC, USA.
Pringle, C. (2003), What is hydrologic connectivity and why is it ecologically important?,
Hydrological Processes, 17(13), 2685-2689.
Pierson, M. S., B. J. Rosenbaum, L. D. McKay, and T. G. Dewald (2008), Strahler Stream Order
and Strahler Calculator Values in NHDPlusRep.
R Development Core Team (2012), R: A Language and Enviroment for Statistical Computing, R
Found For Stat. Compput, Vienna. [Available at http://ww.R-project.org/]
Rantz, S. (1982), Measurement and computation of streamflow; Volume 1, measurement of stage
and discharge; Volume 2, computation of discharge, US Geological Survey water-supply paper,
2175, 631.
Sangal, B. P. (1983), Practical Method of Estimating Peak Flow, J Hydraul Eng-Asce, 109(4),
549-563.
Singh, V. P. (2003), On the theories of hydraulic geometry, International Journal of Sediment
Research, 18(3), 196-218.
Taguas, E. V., J. L. Ayuso, A. Pena, Y. Yuan, M. C. Sanchez, J. V. Giraldez, and R. Pérez
(2008), Testing the relationship between instantaneous peak flow and mean daily flow in a
Mediterranean Area Southeast Spain, CATENA, 75(2), 129-137.
101
Tan, K.-S., F. H. S. Chiew, and R. B. Grayson (2007), A steepness index unit volume flood
hydrograph approach for sub-daily flow disaggregation, Hydrological Processes, 21(20), 2807-
2816.
Tockner, K., and J. A. Stanford (2002), Riverine flood plains: present state and future trends,
Environmental Conservation, 29(3), 308-330.
USGS (2014), National Water Information System data available on the World Wide Web
(Water Data for the Nation), accessed [December, 2014], at URL [waterdata.usgs.gov/nwis.]
Wahl, K. L. (1977), Accuracy of channel measurements and implications in estimating
streamflow characteristics, Journal of Research of the Us Geological Survey, 5(6), 811-814.
Wilkerson, G. V., D. R. Kandel, L. A. Perg, W. E. Dietrich, P. R. Wilcock, and M. R. Whiles
(2014), Continental-scale relationship between bankfull width and drainage area for single-
thread alluvial channels, Water Resources Research, 50(2), 919-936.
Williams, G. P. (1978), Bank-full discharge of rivers, Water Resources Research, 14(6), 1141-
1154.
Williams, G. P. (1986), River meanders and channel size, J. Hydrol., 88(1–2), 147-164.
Wolman, M. G. (1955), The natural channel of Brandywine creek, Pennsylvania, United States
Geological Survey Professional Paper, 271, 56.
102
CHAPTER 6. CONCLUDING REMARKS AND QUESTIONS
FOR FUTURE STUDY
In the presented research, floodwater residence time emerged as a dominant variable controlling
the fate and transport of solutes within floodplain systems (Figure 6-1). At the extremes, short
residence times restrict biogeochemical processes, whereas prolonged residence times lead to
solute limitation. In both of these cases, dominant biogeochemical process shift from removal and
retention to export (i.e., flushing) and accumulation of reactive solutes, respectively. Intrinsically,
river networks experience this gradient in processes (e.g. flushing-retention-accumulation) as the
network transitions from relatively short, sporadic floods in headwater catchments to longer,
seasonal flood pulses in large-river systems. However, there are confounding factors at the site
level that also control distributions of floodwater residence times and thus the dominant
biogeochemical process.
Figure 6-1. Conceptual model of large scale controls on net solute removal in floodplains. After
James et al., [2008].
In the StREAM Lab floodplain, representative of the headwater stream endmember of river-
floodplain connectivity, we measured relatively high rates of nitrate uptake in the spring and late
summer. However, because inundation period is so short (<4 hrs/yr), annual load reductions were
minimal at the reach scale. Further, because of the legacy of agriculture landuse at this site, the
experimental floodplain was a net source of reactive solutes, suggesting flushing was a dominant
biochemical mechanism during inundation events. While our findings do not diminish the value
of small floodplains’ ability to process upslope water, or the cumulative effects of small
floodplains across large-river basins, our results did suggest that the restoration industry should
focus on enhancing river-floodplain connectivity in medium to large streams when optimizing
restoration projects for solute removal.
103
On the other end of the spectrum, we observed a dominance of both solute accumulation and
retention in the Atchafalaya River floodplain. At hydrologically isolated locations, there were
increased concentrations of reactive solutes suggesting accumulation, whereas in more connected
areas, reduced solute concentrations suggested enhanced retention and removal. Here, we observed
increased river-floodplain connectivity in reference to flow-through conditions (as compared to
backwater conditions) and preferential flow of floodwaters in swamp channels. Our results, along
with a handful of other studies, suggest river-floodplain connectivity could be further optimized
for solute retention within the Atchafalaya River Basin.
When examining the cumulative effects of river-floodplain connectivity across larger-river basins,
it is clear that large-river floodplains have greater capacity to process floodwaters than smaller
headwater floodplains, both at the site level and when taken cumulatively across large river basins.
This suggest that ecological engineering efforts should focus on enhancing river-floodplain
connectivity in medium to large-rivers when optimizing for nutrient removal. However, there is
uncertainty in how to move forward and implement optimization strategies. Below, we present
three potential areas of study that build off of the research presented in this dissertation:
Characterization of river-floodplain exchange: At this point, there are relatively few studies
that characterize river-floodplain exchange at the reach scale (e.g. Rudorff et al., 2014; Helton et
al., 2014). However, developing an accurate water balance for individual floodplains is critical in
understanding floodwater residence times and the associated biogeochemical processes (e.g.
Hughes, 1980). Largely, this has not occurred due to computational limitations associated with
2D hydraulic models and the difficulty of measuring floodplain microtopography and diffuse flow
associated with overbank flooding. However, recent advances in both computing power and
LiDAR should encourage new modeling efforts to simulate floodplain inundation and flow.
Ecological Engineering in the Atchafalaya River Basin: Currently, several studies have been
completed on river-floodplain connectivity within the Atchafalaya River Basin, and loosely, the
consensus is that nitrogen removal could be increased through enhanced river-floodplain
connectivity. I believe the next logical progression is to model maximum load reduction,
experimentally increase river-floodplain connectivity in collaboration with the Corps of Engineers,
and further characterize organic matter fate and transport.
Inclusion of floodplains in continental scale network modeling efforts: We are currently on the
cusp of a new understanding of solute fate and transport within large river basins (Creed et al.,
2015). However, contemporary analyses typically focus on baseflow or median flow conditions
(e.g. Gomez-Velez and Harvey, 2014; Kiel and Cardenas, 2014; Woodward et al., 2012), while
the majority of the solute balance often occurs during flood events (e.g. Royer et al., 2006) when
rivers are typically connected to their adjacent floodplains. Therefore, it is critical that floodplains
be included in any large-scale understanding of solute flux in riverine systems.
“We shall never achieve harmony with the land, anymore than we shall achieve absolute justice
or liberty for people. In these higher aspirations the important thing is not to achieve but to strive.”
― Aldo Leopold, Round River: From the Journals of Aldo Leopold
104
6.1 Works Cited Creed, I. F., et al. (2015), The river as a chemostat: fresh perspectives on dissolved organic
matter flowing down the river continuum, Can. J. Fish. Aquat. Sci.
Gomez-Velez, J. D., and J. W. Harvey (2014), A hydrogeomorphic river network model predicts
where and why hyporheic exchange is important in large basins, Geophys. Res. Lett., 41(18),
2014GL061099.
Helton, A. M., G. C. Poole, R. A. Payn, C. Izurieta, and J. A. Stanford (2014), Relative
influences of the river channel, floodplain surface, and alluvial aquifer on simulated hydrologic
residence time in a montane river floodplain, Geomorphology, 205, 17-26.
Hughes, D. A. (1980), Floodplain inundation - processes and relationships with channel
discharge, Earth Surf. Process. Landf., 5(3), 297-304.
James, W. F., W. B. Richardson, and D. M. Soballe (2008), Effects of residence time on summer
nitrate uptake in mississippi river flow‐ regulated backwaters, River Research and Applications,
24(9), 1206-1217.
Royer, T. V., M. B. David, and L. E. Gentry (2006), Timing of Riverine Export of Nitrate and
Phosphorus from Agricultural Watersheds in Illinois: Implications for Reducing Nutrient
Loading to the Mississippi River, Environ. Sci. Technol. 40(13). 4126-4131.
Rudorff, C. M., J. M. Melack, and P. D. Bates (2014), Flooding dynamics on the lower Amazon
floodplain: 1. Hydraulic controls on water elevation, inundation extent, and river‐floodplain
discharge, Water Resources Research, 50(1), 619-634.
Woodward, G., et al. (2012), Continental-Scale Effects of Nutrient Pollution on Stream
Ecosystem Functioning, Science, 336(6087), 1438-1440.
