Predicting the Downstream Impacts of Headwater Burial in Kemptville Creek Watershed (Ontario) Using the Swat Model

PREDICTING THE DOWNSTREAM IMPACTS OF HEADWATER BURIAL IN KEMPTVILLE CREEK WATERSHED (ONTARIO) USING THE SWAT MODEL

Emily Bronwyn Su

A thesis submitted to the Department of Geography and Planning in conformity with the requirements for the degree of Master of Science

Queen’s University Kingston Ontario, Canada (May 2020)

Abstract

Headwater streams can compose up to 80% of the total stream length within a watershed but remain most susceptible to stream burial during land use change or management. Rerouting streams into tile drains, canals, non-perforated pipes and paving over, are all examples of stream burial. This study explores the importance of delineating headwater streams within a watershed hydrological network and the downstream impacts of headwater burial. We used the Soil and

Water Assessment Tool (SWAT) model to predict changes in discharge, chlorophyll a, DO and sediment for downstream ecosystems, following headwater stream burial. Within the study area of Kemptville Creek subwatershed of the Rideau River Watershed in Kemptville and surrounding area, Ontario, Canada, 908 headwater stream segments totaling 443km of stream length were delineated from a 10-hectare threshold. Model calibration was significantly improved with the inclusion of delineated headwater streams, increasing R2 by 0.49. Headwater streams were confirmed with field observations in 70% of 30 selected headwater stream sites. Headwater streams were classified based on land use disturbance coefficients and potential downstream organic subsidy contribution. Each headwater class saw 100% stream burial through conversion into irrigation canals using SWAT model. Headwater stream burial was also applied to all headwater streams as one collective group at 25, 50, 75 and 100%. Ontario Benthic

Biomonitoring Network (OBBN) samples from Rideau Valley Conservation Authority were assessed using the Family Biotic Index (FBI) OBBN metric and Shannon-Weiner Diversity

Index. Predicted discharge, chlorophyll a, dissolved oxygen (DO) and sediment inputs were compared against the OBBN community structure, at 7 sample sites. Headwater stream burial results indicated significant decreases in discharge, chlorophyll a, DO and sediment to downstream ecosystems. Headwater stream burial within wetlands representing 33% of the headwaters within Kemptville Creek, had the most significant impact on downstream discharge,

ii chlorophyll a, DO and sediment. Changes for these variables correlated with several benthic macroinvertebrate families, which supported predictions of community structure shifts following headwater stream burial. Results from this work highlight the importance of headwater streams to ecohydrology within watersheds and contributes to the understanding of ecosystem response as hydrologic regimes are altered following stream burial.

iii

Acknowledgements

First and foremost, I wish to thank Dr. Neal A. Scott. As my supervisor Dr. Scott introduced me to systems modelling and guided me through a rewarding learning process that has evolved my passion for ecosystem management and holistic thinking. I would also like to thank Dr. Joseph D.

White who provided his expertise on the Soil and Water Assessment Tool and the biological component of my project. I wish to thank Jennifer Lamoureux from Rideau Valley Conservation

Authority for partnering with me and allowing me to utilize data for my project. I would also like to thank Les Stanfield for his guidance and insight on research surrounding headwater streams.

Financial assistance for my research was provided by Queen’s University and NSERC through the Discovery Grant program and the NSERC LEADERS CREATE program.

And finally, thank you to my husband, family and friends for supporting and encouraging and having patience throughout my research.

Table of Contents

Abstract ...... ii Acknowledgments...... iv List of Tables ...... viii List of Figures ...... x List of Abbreviations ...... xii

Chapter 1 – Introduction ...... 1 1.1 Background ...... 1 1.2 Importance of Headwater Streams ...... 2 1.3 Surrounding Land use Impacts ...... 2 1.4 Modelling Hydrological Systems ...... 4 1.5 Research Objectives ...... 5 1.6 Study Area ...... 6 1.7 Thesis Outline ...... 7

Chapter 2 – Literature Review ...... 8 2.1 Introduction ...... 8 2.2 Drainage Systems...... 8 2.3 Functions of Headwater Streams ...... 9 2.3.1 Physical Structures of Headwater Streams ...... 9 2.3.2 Chemical Functions of Headwater Streams ...... 10 2.3.3 Biological Functions of Headwater Streams...... 11 2.4 Modelling Watersheds ...... 12 2.5 References ...... 15

Chapter 3 – Simulating downstream impacts of headwater stream burial for Kemptville Creek watershed predicted by SWAT model ...... 20 3.1 Abstract ...... 20 3.2 Introduction ...... 21 3.3 Material and Methods ...... 23 3.3.1 Study Area ...... 23 3.3.2 SWAT Model ...... 26 3.3.4 Data ...... 28 3.3.5 SWAT Model Calibration and Validation ...... 31 3.3.6 SUFI-2...... 33 3.3.7 Headwater Stream Delineation Confirmation ...... 37 3.3.8 Scenarios ...... 38 3.4 Results ...... 40 3.4.1 Headwater Stream Delineation Confirmation ...... 40 3.4.2 Calibration and Validation ...... 41 3.4.3 Scenario Results ...... 46 3.5 Discussion ...... 48 3.5.1 Prediction Headwater Streams Using SWAT ...... 48

3.6 Conclusions ...... 50 3.7 References ...... 52

Chapter 4 – Predictions of downstream benthic macroinvertebrate communities by analysis of chlorophyll a, sediment and DO following headwater stream burial ...... 55 4.1 Abstract ...... 55 4.2 Introduction ...... 56 4.2.1 Chlorophyll α ...... 58 4.2.2 DO ...... 58 4.2.3 Sediment ...... 59 4.3 Materials and Methods ...... 60 4.3.1 Overview ...... 60 4.3.2 Predicting downstream subsides from headwaters ...... 61 4.3.2.1 LDI ...... 61 4.3.3 Benthic Macroinvertebrates Community Structure ...... 63 4.3.3.1 Shannon-Weiner Diversity Index ...... 66 4.3.3.2 The Family Biotic Index ...... 67 4.3.4 Scenarios ...... 68 4.4 Results ...... 68 4.4.1 LDI ...... 69 4.4.2 Benthic Macroinvertebrate Community Structure ...... 69 4.4.2.1 Shannon-Weiner Diversity Index ...... 69 4.4.2.2 The Family Biotic Index ...... 70 4.4.3 Scenario Results ...... 71 4.4.3.1 LDI ...... 77 4.4.3.2 Predictions for Benthic Community Structure ...... 79 4.5 Discussion ...... 82 4.6 Conclusions ...... 85 4.7 References ...... 86

Chapter 5 – Conclusions ...... 89 5.1 Summary ...... 89 5.2 Limitations ...... 90 5.3 Future Directions ...... 91

Appendix A: Observations for 30 predicted headwater streams looking for visible path of flow and active flow; 50%,53%,40% and 37% of predicted headwaters were actively flowing during observation in March, April, May and June 2019, respectively ...... 94 Appendix B: LDI headwater stream classification for Kemptville Creek watershed ...... 95

Appendix C: Tolerance Values for benthic macroinvertebrates found within Kemptville Creek watershed in part of the Family Biotic Index (Bode et al., 1996; Hauer and Lamberti, 1996; Hilsenhoff, 1988; Plafkin et al.,1989) ...... 96

Appendix D: LDI scenario t-test results for each OBBN sample site ...... 98

Appendix E: BC, KC and MC OBBN sample sites and LDI classified headwaters that had the greatest impact on each site ...... 100

Appendix F: Benthic macroinvertebrate families which resulted in greater than >50% correlation with discharge ...... 104

Appendix G: Benthic macroinvertebrate family count vs base scenario and LDI group scenario that had the greatest impact on chlorophyll a, DO, and sediment concentrations for each OBBN site 36 during May and October from 2003 – 2013 ...... 105

vii

List of Tables

Table 2.1: Consistent and inconsistent responses generally associated with urbanization. Adapted from Walsh et al., 2005 (increases decreases ) ...... 14 Table 3.1: Land use class description and distribution assigned to KC watershed ...... 30 Table 3.2: Elevation bands and distribution assigned to KC watershed ...... 30 Table 3.3: Canadian Soil class with US soil class assigned to KC watershed ...... 30 Table 3.4: Additional parameter inputs based on data availability ...... 31 Table 3.5: Parameter ranges, fitted values and identifier codes (r = existing parameter value is multiplied by (1+ a given value), v = existing parameter value is to be replaced by a given value for the 34 parameters chosen for sensitivity analysis availability ...... 35 Table 3.6: Conducted scenarios with change applied, degree of change, adjusted parameters and specific parameter inputs ...... 38 Table 3.7: Calibration and Validation results for p-factor, r-factor, R2, NS and p-bias ...... 41 Table 3.8: T-tests results for each scenario against base scenario discharge results ...... 46 Table 4.1: Land use disturbance coefficients with rationale (Stanfield and Kilgour, 2012) ...... 61 Table 4.2: Predicted terrestrial-aquatic subsidy rank (H = herbaceous plants, T= trees, LG = long grasses, W = woody material, WETN/WETF = wetland non forested and wetland forested, FRST/FRSE/FRSD = forest mixed, forest coniferous, forest deciduous, RNGE = range, PAST = pasture, AGRR= agricultural row crops, SWRN = south western range, URBN = urban) ...... 61 Table 4.3: Land use disturbance coefficients (Stanfield and Kilgour, 2012) and total aquatic- terrestrial subsidy potential values with primary land uses ...... 62 Table 4.4: Water quality evaluation using the family-level biotic index (Hilsenhoff, 1988) ...... 67 Table 4.5: Scenario type and change to watershed with adjusted parameter input values (ICANL = irrigation canal, rte = route file in SWAT input data file) ...... 68 Table 4.6: Percent headwaters falling within LDI and TASP classification ...... 69 Table 4.7: Shannon-Weiner diversity index results for OBBN Sites for 2013 Species richness (S), Species Diversity (H), Species equitability (EH), Maximum Species Diversity (Hmax) ...... 70 Table 4.8 FBI results for BC (Barnes Creek), KC (Kemptville Creek) and MD (Mud Creek) samples taken during spring and fall 2013 (FBI = Hilsenholff’s Family Biotic Index) ...... 70

Table 4.9: Results for scenarios 1-8 indicating changes in focus variables and whether change is significant ...... 71 Table 4.10: Results for scenarios 5-18 indicating LDI scenario that resulted in most significant change for chlorophyll a, DO and sediment loading at each OBBN site ...... 77 Table 4.11: Highlighted LDI group scenarios that induced the greatest percent change for OBBN sites following headwater stream burial ...... 78

viii

Table 4.13: Summary of correlation analysis between discharge, chlorophyll a, DO and sediment loading and each of the top 6 families at OBBN sites. Sites that presented with Pearson correlation >.50 noting sample time as spring (S) and/or spring (F) ...... 80

List of Figures

Figure 1.1: Schematic of tile drainage for agricultural crop ...... 4 Figure 3.1: Kemptville Creek watershed within RVCA jurisdiction, with single gauging station location ...... 24 Figure 3.2: SWAT water flow path schematic ...... 28 Figure 3.3: Scatterplot of residuals vs predicted discharge illustrating primary outliers during January, April and December, and highlighting the importance of snow water content ...... 32 Figure 3.4 SWAT-CUP interface execution (Rouholahnejad et al., 2012) ...... 33 Figure 3.5: A. Parameter values vs objective function showing sample distribution and parameter sensitivity for the parameters: SURLAG, CH_K2, ALPHA_BF, RCHRG_DP, SMFMX, SOL_Z. GW_DELAY, SOL_AWC, SNO50COV, OV_N, SFTMP, LAT_TTIME, GW_REVAP, EPCO, EVRCH B. Parameter values vs objective function showing sample distribution and parameter sensitivity for the parameters: CN2, TIMP, SOL_BD, SMTMP, SMFMN, GWQMN,CH_N2, SHALLST, ESCO, SNOCOVMX, SOL_ZMX, SOL_K, DDRAIN, TDRAIN, SOL_ALB, BLAI, ALPHA_BNK ...... 37 Figure 3.6: Photographic documentation of delineated headwater verification site 350 Jig St Oxford Mills, Ontario, picture taken facing W for DNT10; from left to right: March, April, May and June of 2019 ...... 40

Figure 3.7: Observed versus predicted monthly discharge for uncalibrated model without SWAT predicted headwater streams (R2 0.18); uncalibrated model with SWAT predicted headwater streams (R2 0.72); calibrated model without SWAT predicted headwater streams (R2 0.27); calibrated model with SWAT predicted headwater streams (R2 0.76) ...... 42

Figure 3.8: Observed and predicted monthly discharge for uncalibrated model with SWAT predicted headwater streams (R2 0.72); uncalibrated model without SWAT predicted headwater streams (R2 0.18); calibrated model with SWAT predicted headwater streams (R2 0.76); calibrated model without SWAT predicted headwater streams (R2 0.27) ...... 44

Figure 3.9: Scatterplot of residuals vs predicted discharge illustrating primary outliers during January, March, April, and December ...... 45 Figure 3.10: Discharge with precipitation (light blue) for the year 2000 for scenarios 4 and 8 where 100% tile-drained headwaters (dark blue) and irrigation canal (yellow) headwaters were implemented ...... 47 Figure 3.11: Discharge with precipitation (light blue) for the year 2000 for scenarios 4 and 8 where 100% irrigation canal (yellow) headwaters were implemented ...... 47

Figure 3.12: Scenarios 5-8 where 25, 50, 75 and 100% of the headwaters were converted to irrigation canals, temporal extent of the watershed proportions: 17, 29, 40 and 48% are for 25, 50, 75 and 100% headwater streams respectively ...... 48

Figure 4.1: Meandering stream displaying the morphological attributes riffle and pool (adapted from the Ontario Ministry of Natural Resources 2003 ...... 63 Figure 4.2: RVCA OBBN 2003-2013 sample site location within stream segments of Barnes Creek (BC), Kemptville Creek (KC) and Mud Creek (MC) for Kemptville Creek watershed .....64 Figure 4.3: RVCA OBBN sample sites Barnes Creek (BC), Kemptville Creek (KC) and Mud Creek (MC) with surrounding land use Wetland non-forested (WETN) Wetland forested (WETF) Water (WATR) Urban (URBN) Southwestern Range (SWRN) Range (RNGE) Pasture (PAST) Hay (HAY) Mixed Forest (FRST) Deciduous Forest (FRSD) Coniferous Forest (FRSE) Corn (CORN) Alfalfa (ALFA) ...... 65 Figure 4.4: Percent change in discharge from base scenario readings with each headwater loss scenario (Year 2003) ...... 72 Figure 4.5: Discharge readings following headwater stream burial in increasing increments of 25% illustrating significant decreases (Year 2003) ...... 73

Figure 4.6: Chlorophyll α concentration readings following headwater stream burial in increasing increments of 25%, biological threshold indicated at 4 µg/L (>4 µg/L is lethal) (Year 2003) .....74

Figure 4.7: Percent change in Chlorophyll α concentration from base scenario readings with each headwater loss scenario (Year 2003) ...... 74

Figure 4.8: DO concentration readings following headwater stream burial in increasing increments of 25%; biological threshold indicated at 4mg/L; <4mg/L is lethal to aquatic life (WHO,1984; IJC,1977) (Year 2003) ...... 75

Figure 4.9: Percent change in DO concentration from base scenario readings with each headwater loss scenario (Year 2003) ...... 75

Figure 4.10: Sediment concentration readings following headwater stream burial in increasing increments of 25% indicating fluctuations above biological threshold (Year 2003) ...... 76 Figure 4.11: Percent change in Sediment concentration from base scenario readings with each headwater loss scenario (Year 2003) ...... 76

List of Abbreviations

ALPHA_BNK Baseflow factor for bank storage BC Barnes Creek monitoring site(s) BLAI Maximum potential leaf area index CDEM Canadian Digital Elevation Model CFSR Climate Forecast System Reanalysis CN2 Initial SCS runoff curve number for moisture condition CH_K2 Effective hydraulic conductivity in main channel alluvium (mm/hr) CH_N2 Manning’s roughness coefficient “n” value for the main channel CANSIS Canadian Soil Information Services DDRAIN Depth to the subsurface drain (mm) DEM Digital elevation model DO Dissolved Oxygen EPCO Plant uptake compensation factor ESCO Soil evaporation compensation factor EVRCH Reach evaporation adjustment factor FBI Family Biotic Index GDRAIN Drain tile lag time (hours) GIS Global information systems GW_DELAY Groundwater delay time GWQMN Threshold depth of water for return flow GW_REVAP Groundwater “revap” coefficient HRU Hydrological Response Unit IJC International Joint Commission KC Kemptville Creek monitoring site(s) LAT_TIME Lateral flow travel time (days) MC Mud Creek monitoring site(s) MNR Ministry of Natural Resources

xii

MNRF Ministry of Natural Resources and Forestry MUSLE Universal Soil Loss Equation NCEP National Centers for Environmental Prediction OBBN Ontario Benthic Biomonitoring Network OMAFRA Ontario Ministry of Agriculture and Rural Affairs OV_N Manning’s roughness coefficient “n” value for overland flow RCHRG_DP Deep aquifer percolation fraction REVAPMN Threshold depth of water in the shallow aquifer for “revap” or percolation to the deep aquifer to occur (mm (H2O) RVCA Rideau Valley Conservation Authority SFTMP Snowfall temperature (°C)

SHALLST Initial depth of water in the shallow aquifer (mm H2O)

SMTMP Snowmelt base temperature (°C)

SMFMN Melt factor for snow on June 21 (mm/°C-day)

SMFMX Melt factor for snow on December 21 (mm/°C-day)

SNOCOVMX Minimum snow water content that corresponds to 100% snow cover (mm H2O) SNO50COV Fraction of snow volume represented by SNOCOVMX that corresponds to 50% snow cover (mm H2O)

SOL_ALB Moist soil albedo. The ratio of the amount of solar radiation reflected by a body to the amount incident upon it, expressed as a fraction. The value for albedo should be reported when the soil is at or near field capacity

SOL_AWC Available water capacity of the soil layer (mm H2O/mm soil)

SOL_BD Moist bulk density (Mg/m3 or g/cm3)

SOL_K Saturated hydraulic conductivity (mm/hr)

SOL_Z Depth from soil surface to bottom of layer (mm)

SOL_ZMX Maximum rooting depth of soil profile (mm)

SURLAG Surface runoff lag coefficient

xiii

SWAT Soil and Water Assessment Tool

TASP Terrestrial-aquatic organic subsidy potential

TIMP Snowpack temperature lag factor WHO World Health Organization

xiv

Chapter 1 – Introduction 1.1 Background

Land use change to residential, commercial and industrial areas often involves the modification of drainage systems. Although, hydrological systems are visible all around us, larger waterways and waterbodies evoke a heightened need for preservation and conservation, while smaller waterways, such as headwater streams, often go unnoticed (Elmore and

Kaushal 2008). Headwaters are more often subject to land use change than more defined and continuously flowing higher order streams (Elmore and Kaushal 2008).

Headwater streams are the smallest and most abundant streams within a watershed system and include small periodically drying wetlands. When headwater flow paths come together, they form second order streams which eventually flow into third and then fourth order streams. Today you will find headwater streams located within various landscapes that are commonly integrated or buried within agricultural and urban drainage systems.

Recent research has shown that headwater streams contribute to the maintenance of chemical, biological, and physical properties of downstream ecosystems (Masese et al (2017;

Zhang et al., 2018; Rodrigues-Filho et al., 2017, Weber,2018). But as headwater streams become affected by land use change, their ability to maintain the chemical, biological and physical functions are altered. Understanding the impact of headwater stream burial within various landscapes on downstream ecosystems, will contribute to the development of better planning policies that support watershed health. Despite what is known about headwater streams, little experimental work has explored the impacts of headwater stream loss within multiple land use types and the impacts on downstream hydrology and biology.

1.2 Importance of Headwaters Streams

Headwater streams are unique, having a greater exchange of materials between adjacent terrestrial systems (Kielstra et al., 2017; Little and Altermatt 2018). Headwater streams are known to have ill-defined channels, which allows water flow to connect with bank material such as grasses, rocks or bare earth. Greater terrestrial connectivity allows for transport of various subsidies downstream and throughout the watershed (Wallace, 2015).

Detritus, woody material, nutrients and invertebrates, are some of the terrestrial subsidies that feed downstream food webs (Gomi et al., 2002).

Wipfli and Gregovich (2002) found that fishless headwater streams in southeastern

Alaska provided enough subsidies to downstream networks to meet the energy needs of 100-

2000 young of the year salmonids per km of salmonid-bearing stream. The connection that headwater streams have with the adjacent terrestrial land contributes chemical, biological and physical elements to higher order streams and ultimately larger waterways and water bodies.

Headwater streams are not always actively flowing and are therefore classified as intermittent or ephemeral. Intermittent streams, also termed “gaining streams”, have channels below the water table but will stop flowing during dry periods. Ephemeral streams, also termed “losing streams”, flow in response to precipitation and their channels are above the water table. The change in fluvial dynamics within headwater streams, brings various levels of exchange of subsides between terrestrial and aquatic environments. Exchange across habitat boundaries is a fundamental element in the study of food web dynamics (Polis et al., 1997) and important to maintaining ecosystem integrity.

1.3 Surrounding Land use Impacts

During land use change, headwater streams are often subject to stream burial, which involves a watercourse being redirected into tile-drains, canals, buried within non-perforated

2 pipes and/or filled in and paved over (Roy et al., 2009; Irwin, 1997). Identifying where headwater streams exist and monitoring their activity, brings further understanding of how they contribute to a watershed. Analysis on how headwater stream burial affects overall health of the watershed is important for establishing improved management practices within industries closely connected to with watersheds.

The agricultural industry often manages excess water in crop production areas through the implementation of tile drainage. The drainage allows crop roots to grow deeper to access water sources and ultimately increases crop yield without over saturation and rot. Streams that might have cut across agricultural land are essentially removed, and any water flow then becomes a part of the tile-drainage process (Figure 1.1). Changing the path of flow can also influence element concentrations in surface waters. Elevated nutrient levels are often a result of agricultural land use, and can lead to increased algal blooms and eutrophication, which ultimately creates hypoxic environmental conditions for aquatic wildlife.

Land use change to residential, commercial and industrial areas often involve stream burial (Roy et al.,2009). Headwaters have been most susceptible to stream burial during land use change partly because they have not been adequately mapped (Hansen, 2001; Colson et al., 2008). However, even with improved technology, characterizing hydrologic permanence

(ephemeral, intermittent) is challenging (Roy et al., 2009). Stream networks weave throughout various land use types, making them unavoidable during land use change and management. Continued efforts toward headwater stream mapping and monitoring will further the preservation of these important features.

Figure 1.1 Schematic of tile drainage for agricultural crop

1.4 Modelling Hydrological Systems

In spite of the importance of headwater streams, field experiments quantifying their direct impacts are difficult to implement. Modelling is an excellent method to experiment and make predictions, avoiding irreversible impacts that experiments in the real environment can cause.

The Soil and Water Assessment Tool (SWAT) model bridges disciplines and supports rigorous evaluation of hypotheses about watershed changes under various scenarios (Arnold, 1998). The

SWAT model has been used to analyze phosphorus mobility in different landscapes and cropping systems (Michaud et al., 2007), examine soil water patterns under different grazing intensities within small watersheds (Mapfumo et al., 2004), determine relative magnitude and soil moisture variability in topsoil (Deliberty and Legates 2003), analyze water savings with crop changes and improved irrigation efficiency (Santhi et al. 2006) as well as identify how the changes in agricultural practices impact streamflow (Masih et al., 2011). Kalcic et al. (2015) used SWAT model to evaluate no-till cereal rye cover crops, filter strips, grassed waterways, created wetlands, and restored prairie habitats in two west-central Indiana watersheds with tile

4 drainage systems. Kalcic et al. (2015) found that with spatial optimization techniques, pollutants could be reduced by approximately 60%. The SWAT model can execute a multitude of scenarios and was chosen for this study to simulate stream burial through tile drainage and canal implementation. Modelling these changes allows for predictions of how headwater stream burial might impact the chemical, physical and biological properties of downstream ecosystems.

1.5 Research Objectives

The goal of this study was to predict changes for downstream ecosystems following headwater stream burial within Kemptville Creek watershed. Specific objectives of this study were:

(1) Evaluate SWAT model’s ability to predict headwater stream location and extent through calibration and validation of stream network discharge and examine how burial (simulated using tile drainage and canals) alters stream discharge.

(2) Determine the impact of headwater stream burial in different land use types on downstream biota by analysing chlorophyll a, dissolved oxygen (DO) and sediment against benthic macroinvertebrate communities.

Objectives were achieved by conducting an extensive analysis of secondary data using a range of sources including online peer reviewed journals, and local conservation authority watershed reports. A quantitative data assessment using modelling was also executed using catchment and local scale data provided by Rideau Valley Conservation Authority (RVCA), the

Ministry of Natural Resources and Forestry (MNRF), the Ministry of Natural Resources (MNR), the Ontario Ministry of Agriculture and Rural Affairs (OMAFRA), the Canadian Digital

Elevation Model (CDEM) and the Canadian Soil Information Services (CANSIS). This research

5 analysis was comprised of both ground and satellite data that was integrated into the SWAT (Soil and Water Assessment Tool) model.

1.6 Study Area

Within the RVCA jurisdiction there are 9 watersheds, Kemptville Creek was chosen as the watershed of focus because it has the greatest percentage of agricultural land. Kemptville

Creek has a drainage area of 456 square km, and 60 km of Rideau River. At present, RVCA has mapped 434 km of stream tributaries belonging to 6 catchments: Town of Kemptville, Barnes

Creek, Oxford Mills, North Branch, South Branch, and Mud Creek. There are four small lakes within Kemptville Creek watershed: Cranberry Lake, Atkins Lake, Mud lake and Lissons Lake.

The regional climate of Kemptville Creek watershed is described as cold and temperate.

Precipitation records indicate total precipitation averaging 868 mm per year, February having the lowest rainfall (56mm) and August having the highest (85mm). Temperature ranges from an average of 20oC during the summer (June – September) and -10oC during the winter (December

– March). Snow accumulation during winter months has reached 60 mm in water equivalent.

The warmest month of the year is July, averaging 20.6oC and January is the coldest averaging -

9.9oC (RVCA, 2013).

The North Grenville municipality's population is projected to increase by 12,901 people by 2028, a growth rate in excess of the provincial average (MNGOP, 2010). Considerable pressure has been exerted on the municipality for residential development. Low density residential development will take place in hamlets, then into existing rural residential subdivisions and finally into traditional rural areas (agricultural uses, forestry uses). With plans for residential area expansion and agricultural development, land use change is inevitable. In

preparation for land use change, understanding how stream burial impacts the watershed

becomes even more important.

1.7 Thesis Outline

The outline of this thesis is as follows: Chapter 2 includes a literature review of drainage systems, how we recognize headwater streams, the physical, chemical and biological functions of headwater streams, as well as a review of the modelling program used for this project. Chapter 3 is a manuscript which focuses on SWAT model’s ability to predict headwater stream location and contribution to discharge over a 30-year period, and the impact of burial on hydrology. Chapter 4 is a manuscript which looks at how headwater stream burial influences stream discharge, chlorophyll a, sediment and DO. 14 scenarios based on headwaters organized into land use disturbance and potential downstream subsidy transfers are conducted to predict changes to benthic macro invertebrate communities and the cumulative effects on ecosystem integrity for downstream ecosystems. Lastly, Chapter 5 contains conclusions from this research and suggested steps towards future research.

Chapter 2 – Literature Review

2.1 Introduction

For over 150 years, drainage systems have been part of agricultural practice to reduce excess water within soil and increase agricultural productivity. These systems can be surface and/or subsurface drainage such as open canals, tile drains and/or buried non-perforated pipes and are integrated with natural watercourses (Irwin, 1997). Watercourses integrated with drainage systems are often smaller, less visible and ill-defined headwater streams. The role of headwaters streams in watershed hydrology that changes with land use change or development has been getting increased attention (Sigler et al., 2018; Marttila et al., 2018; Fork et al., 2018;

Spence et al., 2018). Drainage systems within agricultural land can be changed from open drains to enclosed drains; increasing surface area for productivity, land value, soil erosion control, and/or to remove the need for regulatory compliance with buffer zones and setbacks (Sadler

Richards, 2005). As agricultural drainage systems are changed or implemented, watercourses that are a part of the drainage system or within the surrounding area can also be affected.

Understanding how drainage systems influence natural continuous or intermittent watercourses could help mitigate watershed impacts. This literature review will cover drainage systems and the physical, chemical and biological functions of headwater streams at the watershed scale.

2.2 Drainage Systems

Headwaters modified by stream burial are still part of the hydrologic cycle, but the modifications can alter the rate at which water percolates through soil. In areas dominated by urban land use, smaller perennial streams are the only natural aquatic ecosystems to remain

(Bourassa et al., 2017). In agricultural areas, water entering a crop with tile drainage percolates down through the soil to the tile-drainage which collects and transports the water out of the soil

8 and into a water canal through piping. As drainage systems are improved or changed, the local conservation authority in partnership with the Ministry of Natural Resources, evaluates the drainage system for potential impacts on fish communities. Evaluation is based on the Drain

Review Protocol issued by Fisheries and Oceans Canada (Fisheries and Oceans Canada 1998;

Fisheries and Oceans Canada 1986). Channelized stream maintenance can involve removal of riparian vegetation and or dredging to improve hydraulic capacity by removing accumulated sediment. If the drainage system is deemed to have critical habit to fish communities, changes will not take place. Due to a lack of research on headwaters there is little evidence to inhibit stream burial and changes to pre-existing drainage systems.

2.3 Functions of Headwater Streams

2.3.1 Physical Structures of Headwater Streams

Alterations to natural drainage features can change the relationship between ground water and surface water and ultimately affect stream flow regime (Stromberg et al., 2007). As surface water flow decreases, the aquatic-terrestrial connection also decreases and can reduce streambank herbaceous species (Stromberg et al. 2007). Canal and tile-drained streams do not have the heterogeneity of natural stream systems. They lack point bars, pools and riffles, which would otherwise provide habitat and encourage species diversity (Shields et al., 1994). Pools and riffles are important habitat for benthic macroinvertebrates. Riffles provide highly oxygenated areas as stream discharge flows over shallow and rocky stream segments. Pools offer refuge area with deep waters and cooler temperatures. Increases or decreases in stream discharge would affect these areas of important habitat. The artificial drainage created through culverts, and channelization, decreases water storage, increases discharge variability and reduces habitat complexity (King et al., 2014).

2.3.2 Chemical Functions of Headwater Streams

Headwaters influence the quantity and quality of water downstream (Minshall et al.,

1985; Vannote et al., 1980; Freeman et al. 2007). As development continues to increase, the magnitude and duration of surface water runoff increases with impermeable surfaces

(Csicsairova et al., 2020). Increase flows can also alter stream channel morphology, enlarging the channel as it accommodates high flows (Anim et al., 2018). Change in channel morphology from flooding, can lead to unstable and unvegetated banks, muddy channel beds and debris accumulations (Leopold, 1968). During high precipitation events, water flowing through piped streams would have significantly less infiltration due to the constructed impermeable substrate.

Although headwater streams can have low flow, they contribute substantial amounts of subsidies downstream (Argerich et al 2016) but little is known about how burial within land use types alter the transport of chemicals and nutrients to downstream ecosystems (Sadler Richards, 2004).

The bi-directional lateral connectivity between a stream channel and floodplain are essential in establishing a connection between biota and floodplain habitats (Junk et al., 1989;

Power et al., 1995) and contributing to nutrient transformations. Nutrient spiraling is an integral process which takes place between stream orders and nitrogen and phosphorus cycling are pivotal in biogeochemistry. Quantities of nutrients are chemically transformed, decreasing loading in receiving waters (Sadler Richards, 2004). In natural headwater streams, nutrient spiraling is performed by stream bed algal and microbial populations. Invertebrate and fish populations then feed on the algae and microbes, ingesting the nutrients to then be released back into the ecosystem. The rate of stream discharge shapes stream morphology and determines the rate of instream subsidy transfer. The rate at which nutrients and sediment are delivered

10 downstream changes residence time and ultimately affects how organisms can utilize these subsidies (Alexander et al., 2000; Peterson et al., 2001).

2.3.3 Biological Functions of Headwater Streams

The terrestrial-aquatic relationship that headwaters provide is complex and can be interpreted through analysis of fluvial processes and species diversity (Hauer and Lorang, 2004).

Terrestrial-aquatic linkages support deposition of sediments, the shaping of vegetation communities (Friedman and Lee, 2002) and bank soil wetting (Shentsis and Rosenthal, 2003).

Benthic macroinvertebrate communities are also supported by terrestrial-aquatic linkages and are often found within the specific habitats of headwaters (Clarke et al., 2010). This group of organisms contribute to downstream secondary productivity as a primary source of food for fish

(Bourassa, 2017). As indicator species they are widely used to determine water quality and environmental conditions. Changes in benthic macroinvertebrate community composition can be indication of an influential anthropogenic activity such as urbanization (Wang and Kanehl,

2003), and channel regulation (Fleituch, 2003; Lepori and Malmqvist, 2007). Higher benthic macroinvertebrate diversity is indicative of oxygen rich environments (Evans-White et al., 2009) and decreased diversity often results from pesticide and fertilizer application (Rasmussen et al.,

2013).

The ‘Urban stream syndrome’ predicts decreases in less tolerant taxa to be associated with canalization and increases of impervious substrate (Walsh et al. 2005). Impervious substrate and canals would create barriers to organic subsidies often entering streams from bank material.

Wetlands within headwater catchments have historically been converted to irrigated farmland, removing wetland habitat and endangering aquatic and riparian species (Stromberg et al., 2004; Cowley, 2006). As natural headwaters become buried, existing habitat and food web functions change for both headwater and downstream ecosystems. Partial or complete barriers

11 are created and can prevent fish and macroinvertebrate species from colonizing a stream segment or wetland. These partial or complete barriers limit the aquatic-terrestrial connectivity which can reduce the subsidization of organic detritus from riparian vegetation (Wallace, 2015). Organic matter contributes nutrients critical for cellular processes and development in organisms, organisms that then become nutrients for other organisms higher on the food chain. Decreases in organic matter input could affect nutrient transformation rates and fluxes and subsequently aquatic trophic structure (Martinez et al., 2013).

2.4 Modelling Watersheds

The collection of headwater stream data by Conservation Authorities has only recently been integrated into regular monitoring programs, which involves water samples for temperature, conductivity, dissolved oxygen, turbidity, pH, major nutrient and contaminant sources, as well as looking for channel hardening, dredging, scouring, erosion or barrier evidence and tracking depth, wetted width, hydraulic head and location . For just over a decade, this data has been collected to produce more accurate assessments on watershed health. In consideration of the degree to which most headwaters have already been diverted into pipes, canals, culverts, or filled in, it is important to take caution with further research to protect the headwaters that remain and the approach that is used to understand how they contribute to fundamental hydrological processes at various scales. Identifying the approach that holds the most merit in recognizing these driving factors during hydrological alteration is in constant pursuit (Pearce 2003).

Although similar responses have been seen in streams following land use change such as urbanization, the response can vary in magnitude based on the degree of land use change.

Consistent and inconsistent responses from urbanization have been documented for higher order

12 streams (Pearce 2003) (Table 2.1), a similar approach in response analysis should be executed for headwaters in various land use types.

Spatially, watershed-scale models are categorized as lumped, semi-distributed, or distributed models. Lumped models consider a watershed as a single unit with parameters and variables averaged over this unit (Dwarakish and Ganasri, 2015). Semi-distributed models like

SWAT, divide the watershed into sub-units (HRUs). Distributed models divide the watershed into smaller grid cells of spatial heterogeneity. Spatial variability is best represented by semi- distributed and distributed models (Dwaraksih and Ganasri, 2015). Hydrological processes that take place within a model can be classified as empirical, conceptual or physically based.

Empirical models lack physical processes such as runoff, infiltration, evapotranspiration and changes in vegetation on hydrological processes (Dwarakish and Ganasri, 2015). Conceptual models use simplifications of the complex processes where physically based models explicitly represent the spatial variability of land use characteristics. SWAT model incorporates the semi- distributed physically based structure to include spatial variability and explicit representation of hydrological processes and the land surface characteristics that influence them. Ecosystem response to hydrological alteration is a collection of consistent and inconsistent symptoms.

Incorporating a multitude of ecosystem response variables into one study approach is captured by modelling.

Table 2.1 Consistent and inconsistent responses generally associated with urbanization. Adapted from Walsh et al., 2005 (increases decreases )

Feature Consistent Response Inconsistent Response (can both increase and decrease) Hydrology Frequency of overland flow Baseflow magnitude Frequency of erosive flow Magnitude of peak flow Lag time to peak flow

Channel morphology Channel width Sedimentation Pool Depth Scour Sinuosity Channel complexity Water Chemistry Nutrients (N, P) Suspended sediments Contaminants Temperature Ecology Tolerant Invertebrates Algal biomass Sensitive invertebrates and fishes Leaf decomposition Fish abundance/biomass Oligotrophic diatoms Nutrient uptake

2.5 References

Alexander, R.B., Smith, R.A and Schwarz, G.E. 2000. Effect of Stream Channel Size on the Delivery of Nitrogen to the Gulf of Mexico. Nature, 403(6771):758-761. Anim, D. O., Fletcher, T. D., Pasternack, G. B., Vietz, G. J., Duncan, H. P., and Burns, M. J. 2019. Can catchment-scale urban stormwater management measures benefit the stream hydraulic environment? Journal of Environmental Management, 233: 1–11. Argerich A, Haggerty R, Johnson SL, Wondzell SM, Dosch N, Corson-Rikert H, Ashkenas LR, Pennington R, Thomas CK. 2016.Comprehensive multiyear carbon budget of a temperate headwater stream. Journal Geophysical Research, 121:1306–1315. Arnold, J. G., P. M. Allen, R. Muttiah, and G. Bernhardt. 1995. Automated baseflow separation and recession analysis techniques. Ground Water, 33(6): 1010-1018. Arnold, J. G., Srinivasan, R., Muttiah, R. S., and Williams, J. R. 1998. Large area hydrologic modeling and assessment part I: Model development. Journal of the American Water Resources Association, 34(1): 73-89. Bourassa, A.L, Fraser, L., Beisner, B.E .2017. Benthic macroinvertebrate and fish metacommunity structure in temperate urban streams. Journal of Urban Ecology, 3:1–14. Brown, C.D., Van Beinum, W. 2009. Pesticide transport via sub-surface drains in Europe. Environmental Pollution, 157: 3314–3324. Canadian Soil Information Services (CANSIS). 2013. Government of Canada. [Online] < http://sis.agr.gc.ca/cansis/soils/on/soils.html> Accessed 2018. Clarke, A., Mac Nally, R., Bond, N. R. and Lake, P. S. 2010. Conserving macroinvertebrate diversity in headwater streams: The importance of knowing the relative contributions of α and β diversity. Diversity and Distributions, 16:725–736. Colson, T.P., Gregory, J. D., Mitasova, H., and Nelson, S.A.C. 2008.Topographic and soil maps do not accurately depict headwater stream networks. National Wetlands Newsletter 30(3):25–28. Cornelissen, T., Diekkrüger, B., and Giertz, S. 2013. A comparison of hydrological models for assessing the impact of land use and climate change on discharge in a tropical catchment. Journal of Hydrology, 498: 221–236. Csicsaiova, R., Marko, I., Stanko, S., Skultetyova, I., and Hrudka, J. 2020. Impact of stormwater runoff in the urbanized area. Earth and Environmental Science, 444: 1-5. DeLiberty TL and Legates DR. 2003. Interannual and seasonal variability of modelled soil moisture in Oklahoma. International Journal of Climatology, 23: 1057–1986. Dwarakish, G. S., and Ganasri, B.P. 2015: Impact of land use change on hydrological systems: A review of current modeling approaches. Cogent Geoscience, 1, 1115691 Evans-White M.A., Dodds W.K., Huggins D.G., Baker D.S. 2009. Thresholds in macroinvertebrate biodiversity and stoichiometry across water-quality gradients in Central Plains (USA) streams. Journal of North American Benthological Society, 28:855–868. Elmore, A.J., Kaushal, S.S.2008. Disappearing headwaters: patterns of stream burial due to urbanization. Frontiers in Ecology and the Environment, 6:308–312 Fleituch, T. 2003. Structure and functional organization of benthic invertebrates in a regulated stream. International Review of Hydrobiology, 88:332–344. Fisheries and Oceans Canada.1986. The Department of Fisheries and Oceans Policy for the Management of Fish Habitat. Ottawa, ON, Department of Fisheries and Oceans, Communications Directorate.

Fisheries and Oceans Canada.1998. A Decision Framework for the Determination and Authorization of Harmful Alteration, Disruption or Destruction (HADD) of Fish Habitat. Ottawa ON, Fisheries and Oceans Canada, Communications Directorate Fork, M. L., Blaszczak, J. R., Delesantro, J. M. and Heffernan, J. B. 2018.Engineered headwaters can act as sources of dissolved organic matter and nitrogen to urban stream networks. Limnology and Oceanography Letters, 3:215–224. Freeman, M. C., Pringle, M.C., and Jackson, C.R. 2007. Hydrologic connectivity and the contribution of stream headwaters to ecological integrity at regional scales. Journal of the American Water Resources Association, 43:5–14. Friedman J.M. and Lee V.J. 2002. Extreme floods, channel change, and riparian forests along ephemeral streams. Ecological Monographs, 72:409–425. Gaynor, J., Tan, C., Drury, C., Welacky, T., Ng, H., Reynolds, W. 2002. Runoff and drainage losses of atrazine, metribuzin, and metolachlor in three water management systems. Journal of Environmental Quality, 31:300–308. Green, C. H., M. D. Tomer, M. Di Luzio, and J. G. Arnold. 2006. Hydrologic evaluation of the Soil and Water Assessment Tool for a large tile‐drained watershed in Iowa. Transaction of the ASABE, 49(2): 413– .422 Gomi, T., Sidle R.C. and Richardson, J.S. 2002. Understanding processes and downstream linkages of headwater systems. BioScience, 52(10): 905-916. Hansen, W. F. 2001. Identifying stream types and management implications. Forest Ecology and Management 143:39–46. Hargreaves, G.H., Samani, Z.A., 1985. Reference crop evapotranspiration from 26 temperature. Applied Engineering in Agriculture, 1 (2), 96-99 Hauer F.R. and Lorang M.S. 2004. River regulation decline of ecological resources, and potential for restoration in a semi-arid lands river in the western USA. Aquatic Sciences, 66:388–401 Irwin, R. W. 1979. Field survey of drainage benefits in Ontario. 1-13. St. Joseph, MI, American Society of Agricultural Engineers. American Society of Agricultural Engineers Winter Meeting. Ref Type: Conference Proceeding Jourdonnais, J. H. and Hauer, F.R. 1993. Electrical frequency control and its effects on flow and river ecology in the Lower Flat- head River, Montana. Rivers, 4(2):132–145 Junk, W. J., Bayley, P.B., and Sparks, R.E. 1989. The flood pulse concept in river-floodplain systems, pp. 110-127. In: Dodge, D. P. (ed.) Proceedings of the International Large River Symposuim. Canadian Special Publication of Fisheries and Aquatic Sciences 106: 629 pp Kalcic, M., I. Chaubey, and J. Frankenberger, 2015. Defining Soil and Water Assessment Tool (SWAT) Hydrologic Response Units (HRUs) by Field Boundaries. International Journal of Agricultural and Biological Engineering, 8(3): 69–80 Kielstra, B. W., Arnott, S. E. and Gunn, J. M. 2017. Subcatchment deltas and upland features influence multiscale aquatic ecosystem recovery in damaged landscapes. Ecological Applications, 27:2249–2261 King, K.W., Fausey, N.R., Williams, M.R. 2014. Effect of subsurface drainage on streamflow in an agricultural headwater watershed. Journal of Hydrology, 519: 438–445 Little, C. J. and Altermatt, F. 2018 Landscape configuration alters spatial arrangement of terrestrial-aquatic subsidies in headwater streams. Landscape Ecology, 33:1519–1531. Leopold, L.B. 1968. Hydrology for urban land planning – a guidebook on the hydrologic effects of urban land us. U.S Geological Survey, Circular. 554, 18pp.

Lepori F. and Malmqvist B. 2007. Predictable changes in trophic community structure along a spatial disturbance gradient in streams. Freshwater Biology, 52:2184–2195 Mapfumo, E., D. S. Chanasyk, and W. D. Willms. 2004. Simulating daily soil water under foothills fescue grazing with the Soil and Water Assessment Tool model (Alberta, Canada). Hydrological Processes, 18(3): 2787‐2800 Martı´nez, A., Larrañaga, A., Pe´rez, J., Descals, E., Basaguren, A., Pozo, J. 2013. Effects of pine plantations on struc- tural and functional attributes of forested streams. Forest Ecology Management, 310: 147–155. Marttila, H., Karjalainen, S., Kuoppala, M., Nieminen, M., Ronkanen, A., Klöve, B., and Hellsten, S. 2013. Elevated nutrient concentrations in headwaters affected by drained peatland. Science of the Total Environment, 643: 1304–1313. Masese, F. O., Salcedo-Borda, J. S., Gettel, G. M., Irvine, K. & McClain, M. E. 2017. Influence of catchment land use and seasonality on dissolved organic matter composition and ecosystem metabolism in headwater streams of a Kenyan river. Biogeochemistry, 132:1– 22. Masih, I., Maskey, S., Uhlenbrook, S., Smakhtin, V., 2011. Impact of upstream changes in rain- fed agriculture on downstream flow in a semi-arid basin. Agricultural Water Management, 100 (1): 36–45. Michaud, A.R., Beaudin, I., Deslandes, J., Bonn, F., and Madramootoo, C.A. 2007. SWAT- predicted influence of different landscape cropping system alterations on phosphorus mobility within Pike River watershed of south-western Quebec. Canadian Journal of Soil Science, 87(3):329-344. Me, W., Abell, J.M. and Hamilton, D.P. 2015. Effects of hydrologic conditions on SWAT model performance and parameter sensitivity. Hydrology and Earth System Sciences, 19: 4127- 4147 Minshall, G.W., Cummins, K.W., Peterson, R.C., Cushing, C.E., Bruns, D.A., Sedell, J.R., and Vannote, R.L.1985. Developments in stream ecosystem theory. Canadian Journal of Fisheries and Aquatic Sciences, 42:1045-1055. Neitsch, S. L., Arnold, J. G., Kiniry, J. R., and Williams, J. R. 2011. Soil and water assessment tool theoretical documentation version 2009. Texas Water Resources Institute. Ontario Ministry Agriculture, Food and Rural Affairs (OMAFRA). 2004. Details on Drainage Program Changes. Ontario Ministry Agriculture, Food and Rural Affairs Peterson, B.J., Wollheim, W.M., Mulholland, P.J., Webster, J.R., Meyer, J.L., Tanks, J.L., Marti, E., Bowden, W.B., Valett, H.M., Hershey, A.E., McDowell, W.H., Dodds, W.K., Hamiton, S.K., Gregory, S and Morrall, D.D. 2001. Control of nitrogen export from watersheds by headwater streams. Science, 292:86-90. Power, M. E., Sun, A., Parker, G., Dietrich, W.E., and Wootton, J.T. 1995. Hydraulic food-chain models. BioScience, 45(3):159–167. Polis, G.A., Anderson, W.B., and R.D. Holt. 1997. Toward and integration of landscape and food web ecology: the dynamics of spatially subsidized food webs. Annual Review of Ecology and Systematics, 28:289–316. Rasmussen, J. J., McKnight, U.S., Loinaz, M.C., Thomsen, N.I., Olsson, M.E., Bjerg, P.L., Binning, P.J. Kronvang, B. 2013. A catchment scale evaluation of multiple stressor effects in headwater streams. Science of the Total Environment, 442:420–431. Rodrigues-Filho, C.A.S., Gurgel-Lourenço, R. C., Lima, S. M. Q., de Oliveira, E. F. and Sánchez-Botero, J. I. 2017.What governs the functional diversity patterns of fishes in the

headwater streams of the humid forest enclaves: environmental conditions, taxonomic diversity or biotic interactions? Environmental Biology of Fishes, 100: 1023–1032. Roy, A. H., Dybas, A. L., Fritz, K. M., and Lubbers, H. R. 2009. Urbanization impacts the extent and hydrologic permanence of headwater streams in a Midwestern US metropolitan area. Journal of the North American Benthological Society 28:911–928. Sadler Richards, J. 2004. A Review of the Enclosure of Watercourses in Agricultural Landscapes and River Headwater Functions. 1-41. Sarnia, ON, Fisheries and Oceans Canada. Santhi, C., Srinivasan, R., Arnold, J.G. and Williams, J.R. 2006.A modeling approach to evaluate the impacts of water quality management plans implemented in a watershed in Texas. Environmental Modelling and Software, 21(8):1141–1157. Shields F.D., Knight S.S., and Cooper C.M. 1994. Effects of channel incision on base flow stream habitats and fishes. Environmental Management, 18:43–57 Shentsis I. and Rosenthal E. 2003. Recharge of aquifers by flood events in an arid region. Hydrological Processes, 17: 695–712. Sigler, W.A., Ewing, S.A., Jones, C.A., Payn, R.A., Brookshire, E.N.J., Klassen, J.K., Jackson- Smith, D., Weissmann, G.S. 2017. Connections among soil, ground, and surface water chemistries characterize nitrogen loss from an agricultural landscape in the upper Missouri River basin. Journal of Hydrology, 556:247–261. Skaggs, R. W., Breve, M. A. and Gilliam, J. W. 1994. Hydrologic and water quality impacts of agricultural drainage. Critical Reviews in Environmental Science and Technology, 24: 1- 32. Smiley Jr., P.C., King, K.W., and Fausey, N.R., 2014. Annual and seasonal differences in pesticide mixtures within channelized agricultural headwater streams in central Ohio. Agricultural Ecosystems and the Environment, 193: 83–95. Smith, D.R., King, K.W., Johnson, L., Francesconi, W., Richards, P., Baker, D., and Sharpley, A.N. 2015. Surface runoff and tile drainage transport of phosphorus in the midwestern United States. Journal of Environmental Quality, 44: 495–502. Stromberg, J. C., Beauchamp, V. B., Dixon, M. D., Lite, S. J. and Paradzick, C. 2007. Importance of low-flow and high-flow characteristics to restoration of riparian vegetation along rivers in arid south-western United States. Freshwater Biology, 52: 651–679. Vannote R.L., Minshall G.W., Cummins K.W., Sedell J.R. and Cushing C.E.1980. The River Continuum Concept. Canadian Journal of Fisheries and Aquatic Sciences, 37: 130–137. Wallace JB, Eggert SL, Meyer JL, Webster JR. 2015. Stream invertebrate productivity linked to forest subsidies: 37 stream-years of reference and experimental data. Ecology, (96):1213– 1228. Wang L. and Kanehl P. 2003. Influences of watershed urbanization and instream habitat on macroinvertebrates in cold water streams. Journal of the American Water Resources Association, 39: 1181–1196. Weber, G., Christmann, N., Thiery, A. C., Martens, D. and Kubiniok, J. 2018.Pesticides in agricultural headwater streams in southwestern Germany and effects on macroinvertebrate populations. Science of the Total Environment, 619–620:638–648. Williams, J. R. and Berndt, H. D. 1977. Sediment yield prediction based on watershed hydrology. Transactions of the ASAE, 6:1100-1104. Williams, M., King, K., and Fausey, N. 2015. Contribution of tile drains to basin discharge and nitrogen export in a headwater agricultural watershed. Agriculture and Water Management, 158: 42–50.

Wipfli, M.S., and D.P. Gregovich. 2002. Export of invertebrates and detritus from fishless headwater streams in southeastern Alaska: implications for downstream salmonid production. Freshwater Biology 47: 957 – 969. Zhang, W., Chen, D. and Li, H. 2018.Spatio-temporal dynamics of water quality and their linkages with the watershed landscape in highly disturbed headwater watersheds in China. Environmental Science and Pollution Research, 25: 35287–300.

Chapter 3 – Hydrological impacts of headwater stream burial for Kemptville Creek watershed as predicted by SWAT model

3.1 Abstract

Understanding hydrological processes at a catchment scale is challenging, especially when data are limited. The ability to model stream networks inclusive of headwater streams is important to predict and assess cumulative impacts for downstream ecosystems. For the Rideau

River Watershed, Kemptville and surrounding region in Ontario, Canada, headwater streams were delineated at a 10-hectare threshold, adding 908 stream segments to the previously mapped and verified stream network used by Rideau Valley Conservation Authority (RVCA). Thirty of the additional stream segments were chosen based on stream accessibility for in situ verification, with results showing 70% accuracy. Of those confirmed streams, 53% were actively flowing during peak flow season. Using the Soil and Water Assessment Tool (SWAT), monthly stream flow was simulated for different scenarios associated with headwater stream location and removal for Kemptville Creek watershed of Rideau River Watershed. Uncalibrated model discharge with and without predicted headwater streams were compared, showing that headwater streams account for an additional 54% of the variability in streamflow. Calibrated SWAT simulations showed moderate correlation with predicted monthly discharge for Kemptville Creek

(R2 =0.76). Calibration proved difficult for parameters governing snowmelt thus peak flow periods were not simulated effectively. Eight scenarios were conducted to implement headwater stream burial through the addition of tile drainage and irrigation canals in increments of 25%.

Headwater stream burial by canal implementation induced a significant decrease in monthly discharge (p < 0.05) relative to base scenario. Monthly discharge with 100% canaled headwater streams decreased monthly average streamflow from 5.38 m3/s to 0 m3/s. Headwater stream

20 burial by tile drainage implementation revealed no change in monthly discharge (p < 0.05). With continued headwater stream research and in situ data collection, model performance will improve and generate predictions with even greater reliability.

3.2. Introduction

Active headwater streams can provide significant conveyance of water in watersheds, approximately 70% of the mean-annual water volume of second order streams (Alexander et al.,

2007. Headwater streams can compose up to 80% of the total stream length (Gomi et al., 2002;

Meyer et al., 2003) within a watershed but remain most susceptible to stream burial during land use change. Changes in land use can affect the hydrology of the landscape but predicting these changes can be challenging with topography, vegetation, and soil composition affecting stream discharge (Skaggs et al., 1994). The function of headwater streams and the impacts during land use change or development have been getting increased attention (Sigler et al., 2018; Marttila et al., 2018; Fork et al., 2018). Identifying where and to what capacity headwater streams exist is still unknown for many regions. Headwater streams are unique, having a greater exchange of materials between adjacent terrestrial systems (Kielstra et al., 2017; Little and Altermatt 2018).

Headwater streams are known to have ill-defined channels, which allows water flow to connect with bank material such as grasses, rocks or bare earth. Greater terrestrial connectivity allows for transport of various subsidies downstream and throughout the watershed (Wallace, 2015). The capacity and conveyance of a drainage network greatly influences patterns of discharge during extreme rainfall events (Robinson and Rycroft, 1999) and during freshet (Hall, 2016). Freshet is an annual hydrologic event providing a large proportion of water influx to the watershed’s outlet flow. Timing of freshet flow is linked to winter precipitation and spring temperatures.

Watercourses that are part of or within proximity to agricultural drainage and irrigation systems can be affected by the change in soil drainage rate, water retention and pumping in irrigation canals, and runoff patterns. Drainage systems are implemented to remove excess water from crop soil, increase surface area for productivity, increase land value, control soil erosion, and/or to remove the need for regulatory compliance with buffer zones and setbacks (Sadler

Richards, 2004). Understanding how agricultural drainage and irrigation systems influence natural continuous and/or intermittent watercourses could help mitigate watershed impacts.

The SWAT (Soil and Water Assessment Tool) model is a spatially explicit model that brings together data on climate, soil, land use, vegetation, land management and topography, to analyze the impact of land use and other spatial changes on hydrological systems within agricultural basins (Arnold et al., 1995). Originally developed by the USDA Agricultural

Research Service (USDA-ARS) for providing hydrologic data for watersheds with no monitoring data, it continues to be improved with capabilities advancing its performance under varying scenarios and increased spatial detail. It is an efficient physically based model that enables long- term prediction with variables that can be isolated under different scenarios. The SWAT model couples with a GIS (geographic information system) interface lending to a simplified process of watershed discretization and parameter assignment.

Headwaters within agricultural landscapes are often intertwined with industrial complexes, making it difficult for conservation authorities to monitor both the natural processes and management practices that affect them. Objectives for this study were: (1) calibrate and validate the SWAT model with headwater streams delineated at a threshold of 10 hectares reduced from a

735 hectare threshold; (2) compare uncalibrated and calibrated model discharge with and without predicted headwater streams (3) implement headwater stream burial through 4 tile drainage scenarios in increasing increments of 25% headwater burial and 4 canal implementation

22 scenarios in increasing increments of 25% headwater burial and (4) compare predicted monthly discharge for each stream burial scenario.

3.3. Materials and Methods

3.3.1 Study Site

Kemptville Creek is 1 of 9 watersheds within the Rideau River watershed monitored by the Rideau Valley Conservation Authority (RVCA) (Figure 3.1). Some studies for Kemptville

Creek watershed have included climate change impacts on extreme floods (Seidu et al. 2012) and microarray assessments of virulence, antibiotic and heavy metal resistance (Unc et al. 2012).

Kemptville Creek has a drainage area of 456 square km, and 60 km of Rideau River. There are

434 km of stream tributaries belonging to 6 catchments: Town of Kemptville, Barnes Creek,

Oxford Mills, North Branch, South Branch, and Mud Creek. Kemptville Creek has four lakes:

Cranberry Lake, Atkins Lake, Mud Lake and Lissons Lake. Kemptville Creek watershed is also experiencing rapid land use change, transitioning from long established family farms and small businesses to new country estates and commercial property (RVCA, 2013). This makes it an ideal study area to study the impact of headwater burial on hydrological processes using SWAT.

The regional climate of Kemptville Creek watershed is described as cold and temperate.

– March). Snow accumulation during winter months has reached 60 mm in water equivalent.

The warmest month of the year is July, averaging 20.6oC and January is the coldest averaging -

9.9oC.

Figure 3.1. Kemptville Creek watershed within RVCA jurisdiction, with single gauging station location

As a region dominated by agriculture, various tillage techniques and drainage systems are scattered throughout the landscape. The dominant soil type is Farmington which is a well- drained soil with an intermediate available water storage capacity (4-5 cm) within the control section. Soil texture is silt loam with a moderate to very strong calcareous parent material (6 –

40% CaCO3 equivalent) (CANSIS, 2013). Kemptville Creek watershed has relatively little topography, with slope not exceeding 8%.

Kemptville Creek has only one control structure, the Oxford Mills sluice gate dam, located 15 km upstream of the watershed’s outlet. Logs are used to raise or lower the upstream water level depending on the flow conditions. History of dam construction and changes in backwater area and volumes is not available.

Increases in urban development can influence potential flood risk because of the increase in impervious substrate (Leopold, 1968) and so it’s important to assess the impact of land use change on headwater streams, and how these changes might impact hydrological processes to predict changes with future development. The municipality's population (16,451) is projected to increase by 12,901 people by 2028, a growth rate in excess of the provincial average (The

Corporation of The Municipality of North Grenville, 2018). Considerable pressure has been exerted on the municipality for residential development. With plans for residential area expansion and agricultural development, land use change is inevitable. The municipality’s official plan includes land use change from agricultural, idle land and forested areas to residential areas (MNGOP, 2010). In preparation for land use change, understanding how headwater stream burial impacts the watershed becomes even more important.

3.3.2 SWAT Model

Using the SWAT model (Arnold et al., 1995), Kemptville Creek stream network was delineated at a threshold of 10 hectares. Decreasing the auto-selected threshold from 735 hectares to 10, established an additional 908 stream segments. Simulated discharge data for these additional stream segments were examined using the SWAT Output Viewer (Yu, 2014) to ensure the streams were exhibiting the general behavior of ephemeral and intermittent headwater streams. Initial calibration of the model was performed manually, examining residuals for predicted discharge data. This process also contributed to objective 2, where uncalibrated and calibrated model discharge with and without the predicted headwater streams were compared.

Calibration and validation were also performed using SWAT-CUP which is an automated calibration and validation program using the same interface as SWAT. Scenarios outlined in objective 3 were executed with adjustments made to stream routing and land management files.

Parameter inputs were adjusted for 25, 50, 75 and 100% of delineated headwater streams for each of the 8 scenarios implementing headwater stream burial. The project master Access file was the primary input location for adjustments made with each scenario. Changes were rewritten to the ArcSWAT database and model runs were executed over a 30-year period from 1984 to

2013 based on available data. Results of each scenario were compared to the calibrated base simulation inclusive of predicted headwater streams. Analysis was performed using coefficient of determination (R2), Nash-Sutcliffe (NS) model efficiency coefficient (Nash and Sutcliffe,

1970) and t-tests approximation of degrees of freedom.

SWAT operates using the water balance equation (1), representing the hydrological processes: runoff, percolation, evapotranspiration and lateral surface flow.

SWt = SW + Σ(Rday – Qi – Ea – Pi – QRi) (1)

Where SWt is final soil water content, SW is the initial soil water content, t is the time measured in days, Rday is the amount of precipitation, Qi is the amount of surface runoff, Ea is the amount of evapotranspiration, Pi is the amount of percolation, and QRi is the amount of return flow.

Surface runoff is estimated using the SCS (Soil Conservation Service) formula (2)

(Arnold et al., 1995).

2 Q = {0 (P-Ia) /P-Ia+S for P≤Ia for P> Ia (2)

Where Q is equal to runoff, P is equal to rainfall, I is equal to initial abstraction, and S is equal to potential maximum retention after runoff begins. Percolation is based on a storage and crack flow model to represent macropore flow through desiccated soils (Arnold et al., 1995). A kinematic storage model controls the lateral surface flow while evapotranspiration is modeled after methods described by Hargreaves and Samani (1985). A modified version of Williams and

Berndt’s (1977) Universal Soil Loss Equation (MUSLE), contributes to the estimated sediment yield (3). With detailed parameterization of the SWAT model, it is assumed that default values for assigned soil types will perform effectively and account for watershed-specific dynamics.

0.56 Y = 11.8 * (Q *qp) K*LS*C*P (3)

Where Y equals sediment yield, Q equals runoff volume, and qp = peak runoff rate. K, LS, C and

P are the USLE factors representing erodibility, slope length-gradient, crop management and erosion control, respectively.

The SWAT model operates based on hydrologic response units (HRUs). These HRUs are defined by land use, soil type and land management. These units receive weather conditions from known or simulated climate data and generate response operations based on those inputs. Runoff is generated and routed through each HRU and enters the reach within the HRU to then contribute to total watershed/watershed runoff. As water is routed through each HRU, channel roughness, geometry, bank storage and soil conditions determine how runoff is generated.

SWAT also models lateral flow based on kinematic storage generated from slope, soil porosity, water content, field capacity, recharge and hydraulic conductivity (Neitsch et al. 2011). Water loss within SWAT is modeled by infiltration, evapotranspiration, canopy interception and abstraction or transmission loss (Figure 3.2).

Figure 3.2. SWAT water flow path schematic adapted from Neitsch et al., 2011

3.3.4 Data

Input data included land use, elevation, soil type and climate files. Kemptville Creek watershed stream network was delineated by burning in a previously established stream network

(Ontario Ministry of Natural Resources and Forestry-Provincial Mapping Unit, 2012) to a 10-m

28 resolution Digital Elevation Modal (DEM) obtained from Canadian Digital Elevation Model

(CDEM, 2011) and extending the network based on an analysis area of 10 hectares reduced from the default threshold of 735 hectares. This smaller area of analysis allowed for more headwater streams or drainage areas to be delineated based on elevation. The dam operated as a flow- through dam during the model runs. Kemptville Creek watershed land use classes, elevation bands and soil classes are listed with percent cover in Tables 3.1, 3.2 and 3.3, respectively. Land use data was obtained from the Ecological Land Classification (ELC) vector by Ontario Ministry of Natural Resources and Forestry (OMNR, 2008). Soil class distribution was obtained from the

Ontario Ministry of Natural Resource (OMNR, 2011) soil survey complex vector. Details for each soil type were collected from the Canadian Soil Information Services (CANSIS). Five soil parameters required by SWAT model were not available through CANSIS (HYDGRP = hydro group, ANION_EXCL = fraction of porosity, SOL_CRK = potential or maximum crack volume,

SOL_ALB = moist soil albedo, USLE_K = universal soil loss equation top layer erodibility) thus calculations were made to determine these values and assumptions were made if calculations were not applicable. Canadian soil types were matched as closely as possible to US soils within the SWAT model module. Using a Canadian study site requires additional field samples to obtain accurate values for required soil parameters; unfortunately, field samples were not available for this study. Using the DEM, the maximum of 5 elevation bands were selected. The selection of multiple elevation bands allows for narrower ranges of slope classes when dividing the watershed into similar areas of elevation. Weather data was obtained from the National

Centers for Environmental Prediction (NCEP, 2014) Climate Forecast System Reanalysis

(CFSR, 2014).

Table 3.1. Land use class description and distribution assigned to KC watershed

Land use Class SWAT Description Percent Cover ALFA Agricultural Alfalfa Field 3 CORN Agricultural Corn Field 8 FRSD Deciduous Forest 7 FRSE Forest 10 FRST Mixed Forest 5 HAY Agricultural Hay Field 15 PAST Pasture 5 RNGE Range Grasses 1 SWRN Range Open Idle Land 0.03 URBN Urban 6 WATR Open Water 0.3 WETN Wetland 36 WETF Forested Wetland 3

Table 3.2. Elevation bands and distribution assigned to KC watershed

Elevation Bands Percent Slope Percent Cover 1 0-2 65 2 2-4 31 3 4-6 3 4 6-8 1 5 8-9999 0

Table 3.3. Canadian soil class with US soil class assigned to KC watershed

Canadian Soil Class US Soil Class Percent Cover NORTHGOWER SHALLOW AQUENTS 0.2 NORTHGOWER BLASDELL 0.5 NAPANEE CLAVERACK 0.002 RUBICON DUANE 3.3 ROCKLAND FARMINGTON 0.1 MUCK FLUVAQUENTS 21.8 MATILDA HAVEN 4.0 GRENVILLE IRA 14.1% LYONS LYONS 5.6% ALLENDALE NAUMBURG 8.1% FARMINGTON SUN 37.2% OSGOODE TACONIC 2.9% MANOTICK TIOGA 0.6% CARP WALLKILL 0.3% WATER WATER 1.1% HINCHINBROOKE WAYLAND 0.3%

The SWAT model allows the user to run simulations with additional parameter inputs

based on available data for the study site. Kemptville Creek included additional inputs beyond

the required land use, soil type, elevation and weather data. Additional parameters were included

in model set-up and are found in Table 3.4.

Table 3.4. Additional parameter inputs based on data availability

Project Variable Input Module Source Tile Drainage installation from 1981-2014 Operations (OMR, 2019) Lakes Ponds (Google Earth, 2019; Bathymetry Maps) Road Sweeping Management (Leeds and Grenville, 2019) Fertilizer Management (OMAFRA, 2014)

Natural lakes occurring in the watershed included for the Cranberry, Atkins, Mud, and

Lissons Lake with model lake inputs including surface area derived from Google Earth imagery

and bathymetry maps were used for depth. Pre-existing tile drainage covers an area of 4.56km2

and falls within 22 HRUs. Road sweeping was concentrated within urban areas and was

scheduled during the month of May on an annual basis and was included in this study relative to

potential resulting sediment influx.

The watershed delineation process outlined an area of 36,737 ha with 1634 (HRUs) and

one gauging station within HRU 118. Each land use class, elevation band and soil class influence

the components of the water balance within each HRU and collectively the watershed.

3.3.5 SWAT Model Calibration and Validation

Initial calibration of the model was performed manually within ArcMap, allowing for a

greater understanding of hydrological processes within SWAT. With examination of residuals of

predicted discharge data, major outliers during the months of April, August and October were

identified (Figure 3.2). For the study region, these time periods exhibit significant high and low

31 flows. Outliers during these months suggested that the variables influencing freshet, and evapotranspiration needed further calibration.

With each simulation, outputs were carefully examined in order to identify parameters that may not be accounted for or that needed adjustment. Knowledge of the hydrological cycle and water balance equation guided changes to these parameters. Snowfall and water equivalents were parameters of focus and where key outliers were identified (Figure 3.3). Further calibration was required for snowfall to resolve some of the major outliers occurring during January, April and

December.

120 April 100 January December 80 January February 60 December March

Snowfall Snowfall (mm) 40 January December 20 January

December April 0 April April April -15 -10 -5 0 5 10 15 20 • Residuals

Figure 3.3. Scatterplot of residuals for the parameter snowfall, illustrating primary outliers during January, April and December, highlighting the importance of snow water content

Manual calibration was very time consuming but essential to understanding how SWAT operates. Although educationally beneficial, manual calibration did not achieve the desired R2.

Using the autocalibration publicly accessible application SWAT Calibration and Uncertainty

Procedures (SWAT-CUP) (Figure 3.4), model calibration and validation was performed. SWAT-

CUP uses SUFI-2 (Sequential Uncertainty Fitting), PSO (Particle Swarm Optimization), GLUE

(Generalized Likelihood Uncertainty Estimation), ParaSol (Parameter Solution) and MCMC

(Markov chain Monte Carlo) procedures that conduct sensitivity analysis, calibration, validation and uncertainty analysis (Eawag, 2009). SUFI-2 was chosen for calibration and validation with this study. A full description of SUFI-2 is presented by Abbaspour et al. (2004).

Figure 3.4. SWAT-CUP interface execution adapted from (Rouholahnejad et al., 2012)

3.3.6 SUFI-2

The SUFI-2 procedure was chosen for its ability to quickly identify parameters through its sampling and searching technique. It quantifies the output uncertainty by the 95% prediction uncertainty (95PPU). In capturing the multivariate uniform distribution, it accounts for uncertainties for all sources by parameter uncertainty in the hydrological model. Application of SUFI-2 was used with the Nash–Sutcliffe (NS) coefficient calculated as follows:

푛 표푏푠 푠푖푚 2 ∑ (푌푖 −푌푖 ) 푖=1 푁푆 = 1 − 푛 2 (4) 표푏푠 푚푒푎푛 ∑ (푌푖 −푌 ) 푖=1

where n is the number of the observed data points, and yobs and ysim represent the observation and

m mean model simulation with parameters at time i respectively, and y is the average value of the observations.

The use of NS coefficient quantifies a model’s simulation ability to predict variable outcomes and the overall goodness-of-fit. The closer the NS value is to 1 the more accurate the model predictions are. It is a relevant technique as it minimizes the number of model runs needed for calibration and validation while producing good prediction uncertainty ranges that capture the data points within the prediction uncertainty envelope generated during each simulation. This was important given the computational demand of the model for Kemptville Creek watershed.

The model was calibrated and validated during the periods 1984–1994, and 1995–2013 respectively. Calibrating for the water balance at watershed level, captures the variability of

HRUs generated by SWAT during the delineation process. HRUs are key features of SWAT that hold unique land use/management and soil attributes in a single response unit. To capture the variability of climate data in the calibration process, observed data included at least one wet, average and dry year. Data for calibration and validation were obtained from the Kemptville

Creek hydrometric gauge station 02LA006 Latitude: 44° 59' 39'' N Longitude: 75° 39' 44'' W operated by Environment Canada and covering a gross drainage area of 411km2 and for the time period of 1984-2013. The Kemptville Creek gauge station is in HRU 118 of the delineated

SWAT watershed network.

The SUFI-2 algorithm operates by initially assuming large parameter uncertainty which allows for the observed data to be captured within the 95 percent uncertainty envelope. With each iteration during calibration, the parameter ranges are updated through sensitivity analysis,

95 percent confidence intervals and correlation. Parameter ranges continue to decrease and isolate the best simulation. Calibration and validation variables of focus were P-factor, R-factor,

R2, NS and PBIAS.

The P-factor represents the percentage of observed data within the 95 percent uncertainty envelope (95PPU), while the R-factor represents the thickness of the 95PPU envelope. PBIAS

34 illustrates the average tendency of the simulated data to be greater or smaller than the observed data (Gupta et al. 1999). Calibration results should indicate that most of the model simulations are captured within the 95PPU envelop and that the envelop be small. There are no exact goal numbers for R and P-factors but the greater the values the more confidence you can have in your model results. Gupta et al. (1999) suggests that P-factor be greater than 70% and R-factor be closer to 1 for discharge.

The SUFI-2 procedure was run for 50 simulations based on computer CPU capability.

Having a higher number of simulations allowed for the calibration program to narrow in on the best fit parameter values for Kemptville Creek watershed. 34 parameters were manually chosen during calibration over 4 iterations. More parameters allowed SUFI-2 to perform sensitivity analysis as a guidance as to which parameters to exclude for the following iterations (Table 3.5).

Figure 3.5 illustrates the distribution of the sampling points, giving an idea of parameter sensitivity based on whether there is a visible trend as parameter values increase.

Table 3.5. Parameter ranges, fitted values and identifier codes (r = existing parameter value is multiplied by (1+ a given value), v = existing parameter value is to be replaced by a given value for the 34 parameters chosen for sensitivity analysis

Parameter Description Min and Fitted Value Identifier Max Range code ALPHA_BNK Baseflow factor for bank storage 0-1 0.55 r BLAI Maximum potential leaf area index 0-12 0.35 r CN2 Initial SCS runoff curve number 35-98 0.35 r for moisture condition CH_K2 Effective hydraulic conductivity in 0.025-250 142.5 r main channel alluvium (mm/hr) CH_N2 Manning’s roughness coefficient -0.01-0.3 2 r “n” value for the main channel DDRAIN Depth to the subsurface drain 0-1200 0.29 r (mm) EPCO Plant uptake compensation factor 0-1 0.45 v ESCO Soil evaporation compensation 0-1 0.25 v factor EVRCH Reach evaporation adjustment 0.5-1 0.55 v factor

GDRAIN Drain tile lag time (hours) 0-100 675.9 r GW_DELAY Groundwater delay time 0-500 0.12 r

GWQMN Threshold depth of water for 0-5000 3250 r return flow GW_REVAP Groundwater “revap” coefficient 0.02-0.2 0.12 r LAT_TIME Lateral flow travel time (days) 0-180 3 r OV_N Manning’s roughness coefficient 0.01-30 0.25 r “n” value for overland flow RCHRG_DP Deep aquifer percolation fraction 0-1 0.65 r REVAPMN Threshold depth of water in the 0-500 275 r shallow aquifer for “revap” or percolation to the deep aquifer to occur (mm (H2O) TIMP Snowpack temperature lag factor 0-1 0.46 v TDRAIN 0-72 -.13 r SFTMP Snowfall temperature(°C) -5-5 SHALLST Initial depth of water in the 0-1000 3900 r shallow aquifer (mm H2O) SMTMP Snowmelt base temperature(°C) -5-5 1.9 r SMFMX Melt factor for snow on June 21 0-10 6.5 v (mm/°C-day) SMFMN Melt factor for snow on December 0-10 5.5 v 21 (mm/°C-day) SNOCOVMX Minimum snow water content that 0-500 22.5 r corresponds to 100% snow cover (mm H2O) SNO50COV Fraction of snow volume 0-1 100 v represented by SNOCOVMX that SOL_ALB corresponds to 50% snow cover 0-0.25 0.5 r (mm H2O) SOL_AWC Available water capacity of the 0-1 0.2 r soil layer (mm H2O/mm soil) SOL_BD Moist bulk density (Mg/m3 or 0.9-2.5 0.05 r g/cm3) SOL_K Saturated hydraulic conductivity 0-2000 -0.15 r (mm/hr) SOL_Z Depth from soil surface to bottom 0-3500 0.25 r of layer (mm) SOL_ZMX Maximum rooting depth of soil 0-3500 3325 r profile (mm) SURLAG Surface runoff lag coefficient 0.05-24 2.5 r

Iterations were run until the change for R2 and NS were smaller than 0.02. At this point parameters were set to “fitted values” by choosing the recommended best simulation. After

36 selecting the best simulation, SUFI-2 run.bat was executed and model parameters were set to the fitted parameter values. Table 3.5 summarizes the calibrated parameters, their final fitted values

and calculated statistics based on simulated and observed monthly discharge.

Objective Function Objective

Parameter Values A. B. Figure 3.5 A. Parameter values vs objective function (R2) showing sample distribution and parameter sensitivity for the parameters: SURLAG, CH_K2, ALPHA_BF, RCHRG_DP, SMFMX, SOL_Z. GW_DELAY, SOL_AWC, SNO50COV, OV_N, SFTMP, LAT_TTIME, GW_REVAP, EPCO, EVRCH B. Parameter values vs objective function showing sample distribution and parameter sensitivity for the parameters: CN2, TIMP, SOL_BD, SMTMP, SMFMN, GWQMN,CH_N2, SHALLST, ESCO, SNOCOVMX, SOL_ZMX, SOL_K, DDRAIN, TDRAIN, SOL_ALB, BLAI, ALPHA_BNK.

3.3.7 Headwater Stream Delineation Confirmation

Confirmation of SWAT headwater stream delineation was performed prior to running the scenarios. Ground truths were obtained by visiting 30 of the delineated headwater streams during

March, April, May and June 2019. The 30 headwater streams selected for ground truthing, were chosen based on road accessibility or visibility. At each confirmation site, the presence of a distinct path of flow, and water flow were noted as well as photographic documentation. The temperature for March 16th, 2019 ranged from -2 to 0 degrees Celsius, with snow and ice still present on the ground. Three more sets of ground truths were obtained in April, May and June

2019, to compare observations. March and April were of importance due to spring melt flow and the ephemeral and intermittent nature of headwater streams.

3.3.8 Scenarios

Eight different scenarios were executed based on the specific headwater burial implementation. Headwater burial was represented with changes to stream route and management (.rte and .mgt in SWAT model) input parameters. Implementing tile drainage as a burial type involved the parameter inputs for depth to sub-surface drain (mm), time to drain soil to field capacity (hrs) and drain tile lag time (hrs) (DDRAIN, TDRAIN, GDRAIN).

Implementing irrigation canals within headwaters, involved turning on the canal parameter (ICANL) within SWAT. The canal parameter restricts outflow for each stream segment that it is applied to and does not include calculations for canal seepage unless the

SWAT code is manually adjusted. Irrigation from canal water is specified through the parameter inputs: depth of irrigation, time and source and can be scheduled or automated irrigation.

Automated irrigation is triggered by crop water stress or soil moisture loss as a function of actual and potential ET. Irrigation is only active if specified by the user. In this study, no irrigation was selected. The SWAT model does not permit the removal of a stream segment once it has been delineated, so implementing stream burial by adjusting parameters within the routing module

(ICANAL) was a good alternative to simulating headwater loss. Essentially, canal implementation allowed for complete headwater loss within the watershed. Details of each scenario with different stream burial and percent headwater stream loss are outlined in Table 3.6.

Table 3.6. Conducted scenarios with change applied, degree of change, adjusted parameters and specific parameter inputs

Scenario Temporal Extent of Adjusted SWAT Input Value 411 km2 parameter(s) 1 Tile drain enclosure /25% 67.93 (17%) DDRAIN, TDRAIN, 1000,48,24 headwater streams GDRAIN (.mgt) 2 Tile drain enclosure /50% 118.68 (29%) DDRAIN, TDRAIN, 1000,48,24 headwaters streams GDRAIN (.mgt)

3 Tile drain enclosure /75% 163.68 (40%) DDRAIN, TDRAIN, 1000,48,24 headwaters streams GDRAIN (.mgt) 4 Tile drain enclosure 197.9 (48%) DDRAIN, TDRAIN, 1000,48,24 /100% headwaters streams GDRAIN (.mgt) 5 Irrigation canal /25% 67.93 (17%) ICANAL (.rte) 1 headwaters streams 6 Irrigation canal /50% 118.68 (29%) ICANAL (.rte) 1 headwaters streams 7 Irrigation canal /75% 163.68 (40%) ICANAL (.rte) 1 headwaters streams 8 Irrigation canal /100% 197.9 (48%) ICANAL (.rte) 1 headwaters streams

When tile-drainage is added to an HRU, the calculation is as follows:

TTlag = tilelag /24 (5) where TTlag is the lateral flow travel time (days) and tilelag is the tile drain lag time (hrs). When tile drainage is not present within an HRU, the calculation is as follows:

TTlag = 10.4 x Lhill /Ksat,max (6) where TTlag is the lateral flow travel time (days), Lhill is the hillslope length (m), and Ksat,max is the top soil horizon saturated hydraulic conductivity (mm/hr).

Tile drainage takes place within an HRU when the perched water table rises above the specified depth of the tile-drains. Water entering the tile drains on a given day is calculated as follows:

Tilewtr = (hwtbl – hdrain/ hwtbl) x (SW – FC) x (1-exp[-24/tdrain]) if hwtbl > hdrain (7) where tilewtr is water removed from the soil on a given day by tile drainage (mm H2O), hwtbl is the height of the water table above the impervious layer (mm), hdrain is the height of the tile drain above the impervious layer (mm), SW is soil water (mm H2O), FC is the field capacity of the profile (mm H2O), and tdrain is the time required for the soil to drain to field capacity (hrs)

(Neitsch et al., 2011).

When canals are included in SWAT, HRU runoff and drainage are intercepted by the

canal. The equation involves the ratio of runoff and drainage received by the canals to total

runoff and drainage in HRU.

푄 µ = 푟푒푡푒푛푡푖표푛 (8) 푅+퐷푟푎푖푛푎푔푒

where Qretention is the total is the water retained in the canal, R is the volume of runoff (m3) and

Drainage(m3) is drainage generated in the HRU.

As precipitation falls within the model, peak runoff rates are calculated using the time of

concentration taken from the sum of overland flow time and channel flow time. Overland and

channel flow are determined from derivations of the Manning’s equation (Manning, 1891).

Concentration time is the duration of total runoff from entire watershed to reach the outlet. Soil

characteristics contribute to field capacity, wilting point and available water capacity

calculations. Soil profiles based on soil type input determine soil water and water movement.

Penman-Monteith equation (Monteith, 1965) was chosen for potential evapotranspiration

calculation using mean daily parameter values. Transpiration, soil sublimation, canopy storage

contribute to actual evapotranspiration during model runs.

3.4 Results

3.4.1 Headwater Stream Delineation Confirmation

Figure 3.6. Photographic documentation of delineated headwater verification site 350 Jig St Oxford Mills, Ontario, picture taken facing W; from left to right: March, April, May and June of 2019 40

In-field verification concluded that 50%, 53%, 40% and 37% of the delineated headwater streams were actively flowing during the months of March, April, May and June 2019, respectively (Appendix A). During the sample period, daily mean downstream flows ranged from 11 to 24.4 m3/s. 70% of the headwater streams chosen for confirmation had channels where

SWAT had predicted them to be. The 908 headwater stream segments delineated by SWAT at the 10-hectare threshold, more than doubled (877 km up from 434km) the total length of stream tributaries currently verified by RVCA, suggesting that approximately 443 km of streams may be unaccounted for. Figure 3.6 illustrates the change in flow for the headwater stream confirmed at

350 Jig St. Oxford Mills. The image captured for June 2019 (far right) illustrates how headwaters are often ill-defined features and difficult to identify during low flow months.

3.4.2 Calibration and Validation

Calibration increased R2 results by 0.04 from 0.72 to 0.76, with predicted headwaters and

0.09 from 0.18 to 0.27 without predicted headwaters (Table 3.7).

Table 3.7. Calibration and Validation results for p-factor, r-factor, R2, NS and p-bias Mean (m3/s) Period Obs. Pred. Time Step p-factor r-factor R2 NS p-bias Calibration 4.64 4.34 Monthly 0.54 0.79 0.76 0.72 6.6 Validation 4.64 4.83 Monthly 0.53 0.63 0.72 0.71 -3.9

Overall, the inclusion of headwater streams through the calibration process increased the

R2, illustrating the importance of headwater streams contribution to predicted downstream discharge (Figure 3.7).

Uncalibrated without predicted Uncalibrated with predicted headwaters headwaters 40

) 35 /s

3 35 30 30 25 25 20 20 15 15 10 10 5 5

Observed Discharge (m Observed 0 0 0 10 20 30 40 0 10 20 30 40

• Predicted Observed

Predicted Discharge (m3/s) Calibrated without predicted headwaters Calibrated with predicted headwaters

40 40

) 35 35

/s 3 30 30

25 25 arge (m arge 20 20

15 15

10 10

5 5

Observed Disch Observed 0 0 0 10 20 30 40 0 10 20 30 40

• Predicted Observed

Predicted Discharge (m3/s)

Figure 3.7. Observed versus predicted monthly discharge for uncalibrated model without SWAT predicted headwater streams (R2 0.18); uncalibrated model with SWAT predicted headwater streams (R2 0.72); calibrated model without SWAT predicted headwater streams (R2 0.27); calibrated model with SWAT predicted headwater streams (R2 0.76) The increased accuracy established with the inclusion of headwaters, supports the critical

contribution of water from headwaters to the overall hydrologic yield of the basin. Looking at

Figure 3.7 and 3.8, we can see that the uncalibrated model without predicted headwaters

overestimated peak flows (R2 = 0.18). The uncalibrated model with the inclusion of predicted

42 headwater streams has a huge bias indicating it does not simulate peak flows successfully. This bias is most likely a result of the uncalibrated parameters that govern peak flows during spring thaw. These parameters control snowfall and snowmelt which in turn would control flow for intermittent headwaters.

The results for R2 are consistent with previous studies (Moriasi et al., 2005; Ahmed,

2010; Qiao et al., 2015). Moriasi et al., 2005 presents a summary of modelling study results, concluding that NS values for stream discharge average 0.79 and 0.63 for calibration and validation respectively. For SWAT model results specifically, NS values > 0.65 are considered very good for calibration and validation (Saleh et al., 2000).

Uncalibrated without predicted headwaters 40

2000 2010 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2001 2002 2003 2004 2005 2006 2007 2008 2009 2011 2012 2013 Uncalibrated with predicted headwaters 40

1984 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 1985 Calibrated without predicted headwaters

/s)

3 30

1995 2004 2013 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1996 1997 1998 1999 2000 2001 2002 2003 2005 2006 2007 2008 2009 2010 2011 2012 1984

Monthly Discharge (m MonthlyDischarge Calibrated with predicted headwaters 40

1990 2003 1984 1985 1986 1987 1988 1989 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013

Predicted Observed Date (yrs) Figure 3.8. Observed and predicted monthly discharge for uncalibrated model without SWAT predicted headwater streams (R2 0.18); uncalibrated model with SWAT predicted headwater streams (R2 0.72); calibrated model without SWAT predicted headwater streams (R2 0.27); calibrated model with SWAT predicted headwater streams (R2 0.76) Calibration (1984-1994) Validation (1995-2013)

Overestimates of discharge were common during the months of March, April, and

December, falling within the top 10% of negative residuals (Figure 3.9). April had the highest count of overestimates, 19% more than August and 31% more than May. Underestimates of discharge were common during the months of January, March and December, falling with the top 10% of positive residuals. December and January had the highest count underestimates, both making up 54% of the total underestimates and having 5% more than March. Based on residuals, parameters governing snow (TIMP, SFTMP, SMTMP, SMFMX, SMFMN, SNOVOCMX and

SNO50COV) were examined during calibration.

April 35 March April 30

April /s) 3 April 25 April April April April April April April 20 April March March March April 15 MarchApril March March April April March March

Predicted Discharge (m March March March August 10 December January November February August December 5 December December January December 0 -16 -11 -6 -1 4 9 Residuals

Figure 2.9. Scatterplot of simulated residuals vs predicted discharge illustrating primary outliers during January, March, April and December

3.4.3 Scenario Results

Headwater stream burial implemented during the 4 tile-drainage scenarios

(25,50,75,100%), did not produce a change in discharge for Kemptville Creek watershed.

Headwater stream burial by canal implementation at increasing increments of 25% did produce a significant difference (Table 3.8).

Table 3.8. T-tests results for each scenario against base scenario discharge results

Scenario p(T<=t) one-tail p(T<=t) two-tail Conclusion 1-25% tile drained 0.5 1 Insignificant difference 2-50% tile drained 0.5 1 Insignificant difference 3-75% tile drained 0.5 1 Insignificant difference 4-100% tile drained 0.5 1 Insignificant difference 5-25% irrigation canals <0.001 <0.001 Significant difference 6-50% irrigation canals <0.001 <0.001 Significant difference 7-75% irrigation canals <0.001 <0.001 Significant difference 8-100% irrigation canals <0.001 <0.001 Significant difference

Implementing tile drainage for 100% of the headwaters gave the same discharge pattern as seen with the base scenario (Figure 3.10). Irrigation canal implementation for 100% of the headwaters drastically decreased discharge (Figure 3.11). During precipitation events, discharge is more variable compared to tile-drained headwaters. Tile-drainage appears to maintain drainage at a constant rate for those periods of moderate precipitation, while irrigation canals cause more fluctuations in discharge but at a drastically reduced rate.

45 0 40 5

35 10 /s

3 30 15 25 20

20 25 mm

15 30 Discharge Discharge m 10 35 5 40 0 45

Months 100% tile Base Precipitation

Figure 3.10. Discharge for base scenario (dotted black line) with precipitation (light blue) for the year 2000 for scenarios 4 and 8 where 100% tile-drained headwaters (dark blue) was implemented

0.25 0 5 0.2 10

/s 15 3 0.15 20

25 mm 0.1

30 Dischare Dischare m 0.05 35 40 0 45

Months 100% canals Precipitation

Figure 3.11. Discharge with precipitation (light blue) for the year 2000 for scenarios 8 where 100% irrigation canal (yellow) headwaters were implemented

Figure 3.12 illustrates the decrease in discharge as irrigation canals are implemented in increments of 25, 50, 75 and 100%. 25% headwater loss represented 17% of the watershed area, produced an average of 7% of the base scenario discharge for the year 2003.

8% 7% 6% 5% 4%

3% Discharge 2% 1%

0% MeanProportion of Base Scenario 17% 29% 40% 48% Proportion of Watershed

Figure 3.12. Scenarios 5-8 where 25, 50, 75 and 100% of the headwaters were converted to irrigation canals for the year 2003, temporal extent of the watershed proportions: 17, 29, 40 and 48% are for 25, 50, 75 and 100% headwater streams respectively

3.5 Discussion

3.5.1 Predicting Headwater Streams using SWAT

The SWAT model predicted the general trends of observed discharge for Kemptville

Creek watershed but there were inconsistencies with predicted peak and base flows.

Overestimates of discharge were found predominantly during April and underestimates were found predominately during December. These inconsistencies are understandable given the limited data for the parameterization and calibration process. The underprediction of peak flows during the initial simulation prior to calibration captured in Figure 3.8, indicate that under the default parameter settings the model is not capturing vital snow melt processes of this region. As headwaters were removed for the second simulation of the uncalibrated model, we saw over predicted peak flows and an absence of low flows. Inconsistencies in model discharge are accepted based on the underlying variables within the watershed that were not accounted for.

One distinct missing variable within Kemptville Creek watershed, is the presences of beavers. Beavers create restricted outflow in many segments of the stream network within

Kemptville Creek (RVCA, 2013). Efforts have been made to remove beaver dams in recent years, but it is hard to account for the unpredictability of the beaver and the impacts they have on stream discharge and pooling. Smaller wetland areas can also be categorized as headwaters.

Ephemeral ponds and other lowland areas are significant contributors to flow regimes and biodiversity. During high precipitation events, wetlands act as retention areas. Wetlands can hold large volumes of water, making them significant flood control elements but difficult to access and measure.

Although the percentage of predicted headwater streams confirmed as actively flowing was only 53% during peak flow season, 70% of the predicted headwater streams locations were confirmed in situ. The error in headwater stream location would affect scenario results; 25% stream burial would effectively account for only 17.5% of the total headwaters. Percent bias

(Gupta et al. 1999) results indicate both negative and positive values representing over and underestimates, respectively. PBIAS results fall within the satisfactory range of ±25% for streamflow (Moriasi et al., 2007)

Performance of the modeled monthly discharge for Kemptville Creek watershed was satisfactory. Results for NS were >0.65, above the ideal threshold based on monthly timescale simulation results using the SWAT model (Lin et al., 2017). Although the watersheds’ response to tile-drainage was inconsistent with current knowledge of tile-drainage impacts, the effects of tile-drainage on headwaters streams specifically is not well understood. The equation for tile- drainage indicates that tile drainage takes place within an HRU when the perched water table rises above the specified depth of the tile-drains. Tile drains were implemented at a depth of

1000 mm (DDRAIN) which is standard for Ontario (Crabbe et al., 2012). With no observed

49 changes in discharge following tile drain implementation, it suggests that with the well-drained soil type within the study site that the water table did not rise above the depth of the tile drains.

Headwaters play a critical role in ensuring flow and recharge for hydrological systems.

We saw that with the removal of headwater streams, discharge drastically changed. Covering a large percentage of surface area within most watershed systems, headwaters have the greatest connection with available water. Changes in peak and low flows contribute to changes in flow velocity and flood potential (Karr, 1991). In areas where there is poorly drained soil, peak flows are more likely to decrease with subsurface drainage as a result of decreased runoff. Studies show that there has been up to 60% decrease in peak flows during high precipitation events where subsurface drainage is present (Skaggs et al.,1994). Results suggest that soils for this study area have better soil structure and a greater capacity to store water with the addition of tile- drainage removing gravitational water and creating more porous soil. As drainage systems become part of watershed flow regimes, it is essential that they can carry out hydrological functions to a degree that maintains ecological integrity downstream. Results from the conducted tile drainage implementation scenarios suggest watershed discharge is not affected by tile drainage. Although discharge did not change with the implementation of tile drains, stream flow through artificial drainage systems alters habitat connectivity ultimately removing potential refuge areas and subsidy transport.

3.6. Conclusions

Simulating discharge for Kemptville Creek watershed evaluated the performance of the

SWAT model for watersheds with ungauged headwater streams and single variable calibration.

The hydrological connectivity established by the SWAT model, provided further knowledge of existing and potential stream networks at a finer scale. Reducing the stream network delineation

50 threshold by 98.6% produced 908 additional headwater streams with ground truthing confirming

70% location accuracy. Calibration proved difficult for the critical parameters controlling snow fall and spring thaw with residual scatter plots highlighting the importance of accounting for water in snowfall. Simulation of the uncalibrated model inclusive of headwaters produced significant underestimates of discharge during peak flow periods. Results confirmed the importance of headwater streams’ contributing to whole-watershed discharge and that with burial by canal impoundments for 25% of the headwaters, a 94% decrease in discharge would be observed for the entire watershed. Many watershed studies are single disciplinary focused, leading to watershed management decisions being made without thorough understanding of ecosystem processes. The SWAT model provides a detailed system approach with multi variable inputs producing multi disciplinary results and analysis at watershed and watershed scales. This study illustrates SWAT model’s ability to predict headwater stream location and function in a watershed with limited data, as well as highlights the importance of headwater streams conveying water across the landscape to larger ecosystems.

3.7. References

Abbaspour, K.C., Johnson C.A., and Van Genuchten, M.T. 2004. Estimating uncertain flow transport parameters using a sequential uncertainty fitting procedure. Vadose Zone Journal, 3:1340-1352. Abbaspour, K.C .2015. SWAT Calibration and Uncertainty Programs Alexander, R.B., Boyer, E.W., Smith, R.A., Schwarz, G.E., and Moore, R.B., 2007. The role of headwater streams in downstream water quality. Journal of American Water Resource Association, 43: 41 – 59. Ahmed, F. A. 2012.Hydrologic Model of Kemptville Basin-Calibration and Extended Validation. Water Resource Management. 26:2583–2604. Arnold, J. G., P. M. Allen, R. Muttiah, and G. Bernhardt. 1995. Automated baseflow separation and recession analysis techniques. Ground Water, 33(6): 1010-1018. Canadian Soil Information Services (CANSIS). 2013. Government of Canada. [Online] < http://sis.agr.gc.ca/cansis/soils/on/soils.html> Accessed 2018. Crabbe, P., Lapen, D., Clark, H., Sunohara, M., and Liu, Y. 2012. Economic benefits of controlled tile drainage: Watershed evaluation of beneficial management practices, South Nation basin, Ontario. Water Quality Research Journal of Canada, 47(1):30-41. DeLiberty T.L., and Legates, D.R. 2003.Interannual and seasonal variability of modelled soil moisture in Oklahoma. International Journal of Climatology, 23: 1057–1986. Eawag. 2009. SWAT-CUP. Dübendorf, Switzerland: Swiss Federal Institute of Aquatic Science and Technology. Fork, M.L., Blaszczak, J.R., Delesantro, J.M. and Heffernan, J.B. 2018. Engineered headwaters can act as sources of dissolved organic matter and nitrogen to urban stream networks. Limnology and Oceanography Letters, 3:215–224. Green, C.H., Tomer, M.D., Di Luzio, M. and Arnold, J.G. 2006. Hydrologic evaluation of the Soil and Water Assessment Tool for a large tile‐drained watershed in Iowa. Transactions of the ASABE, 49(2): 413‐422. Gupta, H.V., Sorooshian, S., and Yapo, P.O.1999. Status of automatic calibration for hydrologic models: Comparison with multilevel expert calibration. Journal of Hydrologic Engineering, 4(2): 135-143. Hall, J. 2016. High Freshets and Low-Lying Farms: Property Law and St. John River Flooding in Colonial New Brunswick. The Dalhousie Law Journal 39, 1: 195–219. Hargreaves, G.H., and Samani, Z.A.1985. Reference crop evapotranspiration from 26 temperature. Applied Engineering in Agriculture, 1 (2): 96-99. Kalcic, M., Chaubey, I. and Frankenberger, J. 2015. Defining Soil and Water Assessment Tool (SWAT) Hydrologic Response Units (HRUs) by Field Boundaries. International Journal of Agricultural and Biological Engineering, 8(3): 69-80. Karr, J.R.1991. Biological Integrity: A Long-Neglected Aspect of Water Resource Management. Ecological Applications, 1 (1): 66-84. Leopold, L. B.1968., Hydrology for urban planning—A guidebook on the hydrologic effects of urban land use, Tech. Rep. 556, 18 pp., U.S. Geol. Surv., Washington, D. C Lin, F., Chen, X., and Yao, H. 2017. Evaluating the use of Nash-Sutcliffe efficiency coefficient in goodness-of-fit measures for daily runoff simulation with SWAT. Journal of Hydrologic Engineering, 22(11): 05017023. Manning, R.1891. "On the flow of water in open channels and pipes". Transactions of the Institution of Civil Engineers of Ireland, 20: 161–207.

Mapfumo, E., Chanasyk, D.S. and Willms, W.D. 2004. Simulating daily soil water under foothills fescue grazing with the Soil and Water Assessment Tool model (Alberta, Canada). Hydrology Processes, 18(3): 2787‐2800. Marttila, H., Karjalainen, S., Kuoppala, M., Nieminen, M., Ronkanen, A., Klöve, B., and Hellsten, S. 2013. Elevated nutrient concentrations in headwaters affected by drained peatland. Science of the Total Environment, 643: 1304–1313. Masih, I., Maskey, S., Uhlenbrook, S., Smakhtin, V., 2011. Impact of upstream changes in rain- fed agriculture on downstream flow in a semi-arid basin. Agricultural Water Management, 100 (1): 36–45. Michaud, A.R., Beaudin, I., Deslandes, J., Bonn, F., and Madramootoo, C.A. 2007. SWAT- predicted influence of different landscape cropping system alterations on phosphorus mobility within Pike River watershed of south-western Quebec. Canadian Journal of Soil Science, 87(3):329-344. The Corporation of The Municipality of North Grenville. 2018. Municipality of North Grenville Official Plan. [Online] < https://www.northgrenville.ca/work/building-planning-and- development/planning-development/planning-division/official- plan?highlight=WyJvZmZpY2lhbCIsInBsYW4iLCJvZmZpY2lhbCBwbGFuIl0=> Accessed 2018. Monteith, J. L.1965. "Evaporation and environment". Symposia of the Society for Experimental Biology, 19: 205–224. Moriasi, D.N., Arnold, J.G., Van Liew, M.W., Bingner, R.L., Harmel R.D., and Veith, T.L. 2007. Model evaluation guidelines for systematic quantification of accuracy in watershed simulations. American Societty of Agricultural and Biological Engineers, 50(3):885-900 Nash, J.E., and Sutcliffe, J.V., 1970. River flow forecasting through conceptual models part I – a discussion of principles. Journal of Hydrology, 10 (3): 282–290. Neitsch, S.L., Arnold, J.G., Kiniry, J.R., and Williams, J.R. 2011. Soil and Water Assessment Tool Theoretical Documentation Version 2009. Grassland, Soil and Water Research Laboratory – Agricultural Research Service Blackland Research Center. Texas Water Resources Institute Technical Report No. 406. Ontario Ministry Agriculture, Food and Rural Affairs (OMAFRA).2004. Details on Drainage Program Changes. Ontario Ministry Agriculture, Food and Rural Affairs. Qiao, L., Zou, C. B., Will, R. E. and Stebler, E. 2015. Calibration of SWAT model for woody plant encroachment using paired experimental watershed data. Journal of Hydrology, 523: 231–239. Rideau Valley Conservation Authority. 2013.Kemptville Creek Subwatershed Report 2013. [Online] < https://watersheds.rvca.ca/watersheds/kemptville-creek/watershed-report- kemptville-creek> Accessed 2018. Robinson, M., and Rycroft, D.W.1999. The impact of drainage on stream flows. Agronomy Monograph, 38:767–800. Rouholahnejad, E., Abbaspour, K.C., Vejdani, M., Srinivasan, R., Schulin, R., Lehmann, A. 2012. Modelling freshwater availability using SWAT model at a catchment-scale in Ivory Coast. Environmental Modelling and Software, 31:28–36. Sadler Richards, J. 2004. A Review of the Enclosure of Watercourses in Agricultural Landscapes and River Headwater Functions. 1-41. Sarnia, ON, Fisheries and Oceans Canada Saleh, A, J. G. Arnold, P. W. Gassman, L. M. Hauk, W. D. Rosen- thal, J. R. Williams, and A. M. S. MacFarland. 2000. Application of SWAT for the upper North Bosque River watershed. Transaction of the ASAE, 43(5):1077-1087.

Seidu, O., Ramsay, A., and Nistor, I. 2012.Climate change impacts on extreme floods I: combining imperfect deterministic simulations and non-stationary frequency analysis, Natural Hazards, 61: 647– 659. Sigler, W.A., Ewing, S.A., Jones, C.A., Payn, R.A., Brookshire, E.N.J., Klassen, J.K., Jackson- Smith, D., and Weissmann, G.S. 2018. Connections among soil, ground, and surface water chemistries characterize nitrogen loss from an agricultural landscape in the upper Missouri River Basin. Journal of Hydrology, 556: 247–261. Skaggs, R.W., Breve, M.A. and Gilliam, J.W. 1994. Hydrologic and water quality impacts of agricultural drainage. Critical Reviews in Environmental Science and Technology, 24: 1- 32. Unc, A., Zurek, L., Peterson, G., Narayanan, S., springthorpe, S.V/. Sattar, S.A. 2012. Microarray assessment of virulence, antibiotic, and heavy metal resistance in an agricultural watershed creek. Journal of Environmental Quality, 41(2): 534-543 Williams, J.R. and Berndt, H.D., 1977. Sediment yield prediction based on watershed hydrology. Transactions of the American Society of Agricultural and Biological Engineers, 20(6): 1100–1104. Yu, Zhiqiang. 2014.Application of water quality modelling to develop and evaluate scenarios of land use change and beneficial management practice implementation in targeted watersheds of the Red and Assiniboine river basins. SWAT CUP. (Red-Assiniboine Project - Phase 2). Agriculture and Agri-Food Canada (AAFC): National Science and Engineering Council Visiting Fellowship.

Chapter 4 – Predicting the impact of headwater stream burial on downstream benthic

macroinvertebrate communities using SWAT model

4.1 Abstract

Burial of headwater streams can have downstream impacts that could influence benthic macroinvertebrate communities. Here, we predict changes in water quality parameters over a 30- year period that could influence benthic macroinvertebrate community structure for Kemptville

Creek watershed following headwater stream burial using the Soil and Water Assessment Tool

(SWAT) model (Arnold et al., 1995). This catchment has undergone (and still undergoing) significant land use change but data on headwater stream loss is not available. Changes in chlorophyll a, dissolved oxygen (DO) and sediment concentrations were assessed for 18 headwater stream loss scenarios. Headwater streams were classified using land use disturbance coefficients (Stanfield and Kilgour, 2012) and identified for potential cross-habitat resource exchange for downstream delivery. Classification established 14 groups of headwater streams.

Headwater streams were also randomly divided into 25, 50, 75 and 100% of the total headwaters.

Conversion of headwater streams to irrigation canals was used to simulate the loss of headwater streams. Without the input of an irrigation schedule, water entering the hydrological response unit (HRU) with an irrigation canal present will have significantly restricted outflow and essentially hold water within the canal. Canal implementation for the 14 land use disturbance and cross-habitat resource exchange scenarios led to a decline in chlorophyll a (kg), dissolved oxygen (DO) (kg) and sediment (tons), with the loss of wetland headwaters having the most significant impact (P<0.05). Implementation of irrigation canals for all headwaters decreased monthly average watershed discharge from 5.38 m3/s to 0 m3/s. With the average 100% decrease in discharge seen with 25% headwater loss relative to the base scenario without headwater loss, chlorophyll α (µg/L) concentrations initially increased during spring melt and then decreased as

55 discharge diminished. DO (mg/L) and sediment (mg/L) concentrations decreased. Increased chlorophyll α and decreased DO and sediment concentrations following 25% headwater stream loss, indicates that even at a small percentage loss, headwater stream burial impacts water quality for downstream ecosystems. Benthic macroinvertebrate data collected in 2013 illustrated a community structure indicative of ‘fair’ water quality for BC and ‘good’ water quality for KC and MD. Changes in water quality following headwater stream burial were predicted to transform benthic macroinvertebrate communities by reducing less tolerant families such as

Baetidae and increasing tolerant families such as Chironomidae. Overall, based on the benthic macroinvertebrate data and simulated model results, water quality ranking based on FBI would change from ‘fairly substantial pollution likely’ and ‘some organic pollution probable’ to

‘substantial pollution likely’ and ‘fairly substantial pollution likely’.These transitions would in turn influence how subsidies transferred downstream are utilized by different functional feeding groups and subsequently influence predatory fish species. The study identifies the importance of headwater streams in maintaining watershed discharge, and how headwater burial can lead to a decline in water quality parameters that are valuable to maintaining diverse and equitable ecosystems.

4.2. Introduction

Headwaters are largely responsible for the transport of allochthonous material, some of which has been processed into fine particulates by benthic macroinvertebrates (Wipfli and

Musslewhite, 2004). Headwater streams include intermittent and ephemeral streams with little to no channel or bank definition. Wetlands can also be part of headwaters, especially when they exhibit periodic drying periods. Headwaters have a greater terrestrial-aquatic connectivity due to their ill-defined banks, allowing organic material such as grasses, leaves and soil to be swept into

56 the water and transferred downstream (Kielstra et al., 2017; Little and Altermatt 2018). Input of allochthonous material from headwaters offers high exports of carbon and nutrients and supports secondary productivity despite the headwaters’ ephemeral and intermittent characteristics

(Argerich et al. 2016; Peterman et al. 2008). Leaf litter transforms stream food-web dynamics

(Marcarelli et al. 2011) and can vary based on leaf species, richness and diversity, which influences detritus breakdown rate (Little and Altermatt, 2018).

Headwaters provide excellent habitat and refuge areas for many species of benthic macroinvertebrates. Although headwaters provide habitat for invertebrate species, they are likely to have smaller populations because they don’t have upstream recruitment by drift (Dosskey et al., 2010). Headwater streams are the first stage of downstream recruitment, increasing downstream species richness as stream segments merge into higher order streams.

Samples of benthic macroinvertebrate species within stream segments have been taken at Barnes,

Kemptville, and Mud Creeks sample sites as part of the Ontario Benthic Biomonitoring Network

(OBBN) (Jones et al., 2004). The OBBN is widely used as an indicator metric for the ecological health of streams, wetlands and lakes. This metric uses benthic aquatic invertebrate taxa as indication of water quality, based on species composition. Streams are frequently buried during the process of urbanization, sometimes to the extent that the total length of buried streams exceed that of open streams (Beaulieu et al., 2014). As headwater streams become buried and/or restricted from conveying water, the terrestrial-aquatic resource exchange no longer benefits downstream ecosystems. The decrease in drift and movement of organic material entering the stream changes, ultimately changing the trophic structure of aquatic food webs (Dosskey et al.,

2010).

4.2.1 Chlorophyll a

Aquatic food webs are fueled by allochthonous material and primary productivity.

Chlorophyll a is indicative of primary productivity and a key water quality measurement when determining the health of an aquatic ecosystem (Beaulieu et al., 2014). Increases in chlorophyll a have been found to reduce the amount of macrophytes in aquatic environments (Uwadiae, 2016) associated with eutrophication in which microalgae proliferate into blooms, leading to hypoxic conditions. Dense planktonic algae biomass within an aquatic environment, reduces plant heterogeneity and thus available food resources and habitat for aquatic organisms. Benthic macroinvertebrate diversity subsequently declines, and more pollution-tolerant species increase in abundance. Changes in the vegetative structure of streams, ultimately changes the community structure of benthic macroinvertebrates.

4.2.2 DO

Substrate heterogeneity within riffles and pools of stream segments offer unique habitat for invertebrate production and are important during benthic macroinvertebrate assessments

(Jones et al., 2004). Stream flow over rocks and hard instream substrate creates riffles which are highly oxygenated and preferred habitat for certain species of benthic macroinvertebrates. The rate of stream discharge contributes to riffle and deep pool formation as well as stream sinuosity.

Drainage characteristics of surficial geology also contributes to rate of discharge, determining rate of water infiltration and runoff. Decreased rates of discharge would reduce DO levels produced by riffles and deeper, cooler pools. Increases in chlorophyll a would also lead to eutrophication and hypoxic conditions resulting in more pollution-tolerant taxa (Rosenberg and

Resh 1993, Jiang et al. 2011).

4.2.3 Sediment

Riverine productivity can be negatively impacted by increased sediment concentrations.

In addition to decreases in primary production, increased sediment loading can smother benthic macroinvertebrate communities (Aspray et al., 2016). In agricultural catchments, sediments can arise from livestock access to streams and reduce riparian vegetation causing bank erosion.

Increased sediment can lead to decreases in invertebrate life cycle, size, and species with gill respiration, filtering and grazing survival mechanisms (Larsen et al., 2001). Burrowing, fine detritus feeding and tegumental respiration are adaptations some invertebrate species possess to survive habitats dominated by fine substrata (Townsend et al., 2008) and those without these adaptations would most likely not survive increases in fine substrata. The increase in sediment concentration affects oxygen, and nutrient concentrations (Bilotta and Brazier, 2008) as well as the survival characteristics (foraging and mobility) of benthic macroinvertebrates. The aquatic- terrestrial connectivity that headwater streams have can also create higher sediment concentrations as a result of ill-defined channels and overland flow during higher precipitation events and freshet.

Given the potentially important role of headwater streams on biodiversity, my work explores the following Objectives: (1) To classify headwater streams into groups based on the land use distribution index (LDI) originally developed by Morrison et al. (2006) and identify potential energy contributions to downstream ecosystems (Stanfield and Kilgour, 2012) (2)

Evaluate Barnes Creek, Kemptville Creek, and Mud Creek OBBN sites’ benthic macro invertebrate structure based on Shannon-Weiner Diversity Index (Shannon and Weaver, 1949), and the Family Biotic Index (FBI) (Hilsenhoff, 1982); (3) Implement headwater stream burial through 4 canal implementation scenarios in increasing increments of 25% headwater burial and

14 LDI headwater stream burial scenarios based on headwater subsidy potential and LDI

59 assignment; (4) compare predicted chlorophyll α, DO and sediment concentrations for each stream burial scenario conducted over a 10-year period from 2003 to 2013 (5) compare simulated changes in water quality with headwater burial to benthic macroinvertebrate community structure changes over time.

4.3. Materials and Methods

4.3.1 Overview

This study focuses on the biological impact of headwater burial in Kemptville Creek watershed following headwater stream burial and predicts changes in benthic macroinvertebrate communities. These predicted changes are then verified using long-term field data. Using headwater streams delineated at a threshold of 10 hectares (Su, 2020 Chapter 3), headwaters were organized and divided into 14 groups based on the land use disturbance coefficient index

(LDI) originally developed by Morrison et al. (2006) and were identified for potential energy contribution to downstream ecosystems (Stanfield and Kilgour, 2012). OBBN data were analyzed using the Shannon-Weiner Diversity Index (Shannon and Weaver, 1949), and the

Family Biotic Index (FBI) (Hilsenhoff, 1982) metric. The previously calibrated and validated

Soil and Water Assessment Tool (SWAT) (Arnold et al., 1995) model used in Chapter 3 was used for this study. In order to simulate headwater loss within the watershed, irrigation canal implementation was chosen as the applied technique within SWAT model. Irrigation canals were implemented for each LDI headwater group as well as for 25, 50, 75 and 100% of all headwater streams.

4.3.2 Predicting downstream subsidies from headwaters

4.3.2.1 LDI

To predict potential downstream subsidies provided by the headwaters of Kemptville

Creek watershed, each headwater stream was first assigned a value based on the surrounding land use type. These values are based on the potential energy contribution to downstream ecosystems (Stanfield and Kilgour, 2012). The assigning of values for each headwater stream was derived from the study by Stanfield and Kilgour (2012) which used an adjusted land use disturbance index (LDI) originally developed by Morrison et al. (2006). Impervious substrate values provided by Arnold and Gibbons (1996), Shuster et al. (2005) and Stanfield and Kilgour

(2006) were used to set the Di (disturbance coefficient) for urban land use areas. Di values

(between 0 and 1) for each land use/land cover category are based on the expected magnitude of effect on sediment transport and flow regimes (Stanfield and Kilgour 2012) (Table 4.1).

Table 4.1. Land use disturbance coefficients with rationale (Stanfield and Kilgour, 2012) Land use Di Rationale Water/wetlands 0 No negative effect anticipated All Forests 0.01 Some surface runoff due to access roads Idle Lands 0.03 Tend to be meadows with heaver vehicle use Pasture 0.04 Low-level use by animals, nutrient runoff Gravel pits 0.1 Interception of groundwater, compaction Cropland 0.1 Nutrient, soil loss, compaction Intensive pasture 0.15 Intensive cattle use, high tractor use Urban 0.3 Parks and open areas: heavily cultivated Roads 0.9 Impervious areas

In addition to the Di values, a ranking was assigned for the terrestrial-aquatic subsidy potential (TASP). The TASP rank was determined through percent of herbaceous plants, trees, long grasses and woody material, within each land use area. If the potential of having herbaceous plants, trees, long grasses and woody material was >50% for a particular land use,

61 then each subsidy category was marked as “Yes”. The TASP ranked from very low, low, medium, high and very high terrestrial-aquatic organic subsidy potential (Table 4.2).

Table 4.2. Predicted terrestrial-aquatic subsidy rank (H = herbaceous plants, T= trees, LG = long grasses, W = woody material, WETN/WETF = wetland non forested and wetland forested, FRST/FRSE/FRSD = forest mixed, forest coniferous, forest deciduous, RNGE = range, PAST = pasture, AGRR= agricultural row crops, SWRN = south western range, URBN = urban)

SWAT Land use H T LG W TASP WETN/WETF Yes Yes Yes Yes Very High FRST/FRSE/FRSD Yes Yes Yes Yes Very High RNGE Yes No Yes No Medium PAST Yes No Yes No Medium AGRR No No Yes No Low SWRN No No Yes No Low URBN No No Yes No Low

During the stream network delineation stage of SWAT model setup, 908 headwater streams were produced from the DEM. Based on percent land use area within each HRU, an additional land use class was assigned to each headwater to add additional information on dominant land use classes within that HRU. It was also valuable to visually observe how the headwater stream routed through the HRU and different land-use types to identify potential subsidy contribution. Therefore, both land use Di values were totaled to create a new LDI value

(Table 4.3).

Table 4.3. Land use disturbance coefficients (Stanfield and Kilgour, 2012) and total aquatic-terrestrial subsidy potential values with primary land uses

LDI TASP Primary Land uses 0 Very High Wetland, Wetland 0.01 Very High Wetland, Forest 0.02 Very High Forest, Forest 0.03 High Wetland, Range 0.04 High Pasture, Wetland 0.05 High Forest, Pasture 0.06 Medium Range, Range 0.08 Medium Pasture, Pasture 0.1 High Agricultural, Wetland 0.11 High Forest, Agricultural 0.13 Medium Agricultural, Range

0.14 Medium Agricultural, Pasture 0.4 Medium Urban, Agricultural

4.3.3 Benthic Macroinvertebrate Community Structure

Benthic macroinvertebrate communities are structured based on biotic and abiotic sensitivity. The biotic index system is composed of indicator species that require specified chemical and physical conditions. Analyzing the numbers, behaviour, or simply the presence/absence of certain species of benthic macroinvertebrates within a sample provides indication of current environmental conditions (Jones et al., 2004).

RVCA has been collecting benthic macroinvertebrate samples for Kemptville Creek watershed at 7 different sample sites since 2003. These sites are located within Barnes Creek

(sites 36, 45 and 71), Kemptville Creek (sites 64, 176 and 1092) and Mud Creek (site 1133) catchments of the watershed (Figure 4.2 and 4.3). The sample process takes place in the spring and spring and involves the rapid bioassessment technique using a semi-quantitative collecting method called Kick-and-Sweep (Jones et al., 2004). This protocol involves samples from two riffle locations and one pool within the stream site (Figure 4.1). The process involves the use of a

D-net that is placed close to the stream bottom, downstream of the individual who is instream kicking up the stream bottom substrate. Kicking encourages any benthic macroinvertebrates to

63 be swept into the net located in front of the individual’s feet. Using the collected invertebrates, 4 ecological assessments were performed for each collection site.

Figure 3.1. Meandering stream displaying the morphological attributes riffle and pool (adapted from the Ontario Ministry of Natural Resources 2003)

Figure 4.2. RVCA OBBN 2003-2013 sample site locations within stream segments of Barnes Creek (BC), Kemptville Creek (KC) and Mud Creek (MC) for Kemptville Creek watershed

Figure 4.3. RVCA OBBN sample sites Barnes Creek (BC), Kemptville Creek (KC) and Mud Creek (MC) with surrounding land use Wetland non-forested (WETN) Wetland forested (WETF) Water (WATR) Urban (URBN) Southwestern Range (SWRN) Range (RNGE) Pasture (PAST) Hay (HAY) Mixed Forest (FRST) Deciduous Forest (FRSD) Coniferous Forest (FRSE) Corn (CORN) Alfalfa (ALFA)

4.3.3.1 Shannon-Weiner Diversity Index

The Shannon-Wiener diversity index (Shannon and Weaver, 1949) measures species diversity by incorporating information on the relative abundance of different species. Species richness is number of species in a community and species equitability is the proportion of the species within the community. The equation is calculated as follows:

푅 퐻′ = − ∑ 푝푖 ln (푝푖) (9) 푖=1 where pi is the proportion of species i in the community and R is the total number of species in the data set. The maximum value of H is calculated as follows:

퐻푚푎푥 = ln S (10)

66 where S is the number of species in the community. Hmax will be reached when all species are equally represented in the community. The equitability is measured as:

퐸퐻 = 퐻/퐻푚푎푥 (11)

As the number of species increases, and the proportion of each species within the community becomes more even, H increases (Magurran, 1988). The Shannon-Weiner diversity index was calculated from the OBBN data collected at the 7 sites during spring and spring for the year

2013.

4.3.3.2 The Family Biotic Index

Hilsenhoff (1982) created the Family Biotic Index (FBI) from his modified Biotic Index.

Species of benthic arthropod community were generalized into family-level tolerance values.

Values range from 0 being very intolerant to 10 being highly tolerant. The levels are based on organic pollution tolerance (Table 4.4).

Table 4.4. Water quality evaluation using the family-level biotic index (Hilsenhoff, 1988)

Family Biotic Index Water Quality Degree of Organic Pollution 0.00-3.75 Excellent Organic pollution unlikely 3.76-4.25 Very Good Possible slight organic pollution 4.26-5.00 Good Some organic pollution probable 5.01-5.75 Fair Fairly substantial pollution likely 5.76-6.50 Fairly poor Substantial pollution likely 6.51-7.25 Poor Very substantial pollution likely 7.26-10.00 Very poor Severe organic pollution likely

Benthic macroinvertebrates found within OBBN sites for Kemptville Creek were assigned tolerance values from Hilsenhoff’s FBI where available (Appendix D) and the FBI was calculated using the following equation:

(푛 푥 푎 ) FBI = ∑ 푖 푖 (12) 푁 where ni equals the number of individuals within a taxa, ai equals the tolerance value of a taxon and N equals the total number of organisms in the sample.

4.3.4 Scenarios

Four different scenarios were executed to simulated headwater stream burial at increasing increments of 25%. Headwater burial was represented with changes to stream route (applied in the .rte files in SWAT) input parameters. Implementing irrigation canals where headwater streams were delineated was executed by turning on the canal parameter (ICANL.rte) within the subbasin of the headwater stream. The irrigation canal parameter was chosen to simulate stream burial because it restricts outflow for each stream segment to which it is applied to. Irrigation canals implementation was also applied with each LDI scenario, isolating the identified headwaters within each LDI group (Table 4.5).

Table 4.5. Scenario type and change to watershed with adjusted parameter input values (ICANL = irrigation canal, rte = route file in SWAT input data file) Scenario Burial Type / percent headwaters changed Adjusted SWAT parameter(s) Input Value 1 Irrigation canal /25% ICANAL (.rte) 1 2 Irrigation canal /50% ICANAL (.rte) 1 3 Irrigation canal /75% ICANAL (.rte) 1 4 Irrigation canal /100% ICANAL (.rte) 1 5-18 Irrigation canal /100% for each LDI groups ICANAL (.rte) 1

With each scenario, discharge, chlorophyll a, DO and sediment were evaluated for changes from the base simulation.

4.4. Results

SWAT demonstrated its ability to predict realistic outcomes for increased headwater stream burial within Kemptville Creek watershed. Patterns of observed monthly discharge data for

Kemptville Creek were captured in the flow series simulated by SWAT. Calibration vs. validation data agreed well with model results (R2 values at 0.76 and 0.73 respectively) (Su et al.,

2020 Chapter 3). Chlorophyll a, sediment and DO were not calibrated and validated and thus results for these variables are best observed for changes over time.

4.4.1 LDI

Of the 908 headwater streams, the greatest percentage of headwater streams fell within wetlands at 33% of all headwaters (LDI value of 0). Urban/agricultural followed at 14% (LDI value of 0.4) and agricultural/wetland areas were 3rd also at 14% (LDI value of 0.1) (Table 4.6).

Mapped distribution of all LDI grouped headwaters can be found in Appendix C.

Table 4.6. Percent headwaters falling within LDI and TASP classification

LDI TASP Percent Headwater Primary Land uses Headwaters Area (ha) 0 Very High 33% 106166.5 Wetland, Wetland 0.01 Very High 13% 6857.66 Wetland, Forest 0.02 Very High 9% 3274.78 Forest, Forest 0.03 High 1% 277.46 Wetland, Range 0.04 High 3% 1347.38 Pasture, Wetland 0.05 High 2% 532.55 Forest, Pasture 0.06 Medium 0% 141.28 Range, Range 0.08 Medium 1% 451.74 Pasture, Pasture 0.1 High 14% 7207.56 Agricultural, Wetland 0.11 High 10% 3574.93 Forest, Agricultural 0.13 Medium 1% 347.93 Agricultural, Range 0.14 Medium 2% 525.60 Agricultural, Pasture 0.4 Medium 14% 39318.32 Urban, Agricultural

4.4.2 Benthic Macroinvertebrate Community Structure

4.4.2.1 Shannon-Weiner Diversity Index

Shannon-Weiner diversity index results varied with each site (Table 4.7). Site 36, 64, and

176 had higher diversity during spring, while sites 71, 1092 and 1133 had higher diversity during spring. Site 45 was omitted from this assessment because samples were missed during 2013.

Benthic macroinvertebrates communities for site 64 and 76 were distributed most equitably during spring (both sites EH = 0.83). Site 176 had the highest diversity overall (H = 2.81).

Shannon-Weiner diversity index values that fall between 1 -2 give indication that the

69 environment is moderately polluted (Adu et al., 2016). On average, KC had the highest diversity followed by BC then MC.

Table 4.7. Shannon-Weiner diversity index results showing species richness (S), species diversity (H), species equitability (EH), maximum species diversity (Hmax) for spring (May) and fall samples (October)

OBBN site S H EH Hmax Average H BC 36 spring 30 2.26 0.67 3.40 36 fall 22 2.16 0.63 3.10 45 spring NA NA NA NA 2.03 45 fall NA NA NA NA 71 spring 18 1.68 0.50 2.90 71 fall 26 2.01 0.60 3.26 KC 64 spring 28 2.66 0.83 3.33 64 fall 25 2.47 0.77 3.22 176 spring 32 2.81 0.83 3.47 2.51 176 fall 27 2.43 0.71 3.30 1092 spring 31 2.24 0.66 3.43 1092 fall 30 2.42 0.71 3.40 MC 1133 spring 17 1.60 0.56 2.83 1.83 1133 fall 25 2.05 0.64 3.22

4.4.2.2 Family Biotic Index

The FBI analysis resulted in BC having ‘fairly substantial pollution likely’ and KC and

MD having ‘some organic pollution likely’ (Table 4.8).

Table 4.8. FBI results for BC (Barnes Creek), KC (Kemptville Creek) and MD (Mud Creek) samples taken during spring and fall 2013 (FBI = Hilsenholff’s Family Biotic Index)

OBBN site FBI Water Quality Overall Organic Pollution Average Average Organic for FBI Index Prediction Water Quality Pollution Prediction BC 36 spring 5.86 Fairly Poor Substantial pollution likely 36 fall 5.44 Fair Fairly substantial pollution likely 45 spring NA NA NA Fair Fairly substantial 45 fall NA NA NA pollution likely 71 spring 4.76 Good Some organic pollution probable 71 fall 5.79 Fairly Poor Substantial pollution likely KC 64 spring 4.81 Good Some organic pollution probable 64 fall 4.87 Good Some organic pollution probable 176 spring 5.80 Fairly Poor Substantial pollution likely Good Some organic 176 fall 4.96 Good Some organic pollution probable pollution probable 1092 spring 4.67 Good Some organic pollution probable

1092 fall 4.23 Very Good Possible slight organic pollution MC 1133 spring 4.45 Good Some organic pollution probable Good Some organic 1133 fall 5.05 Fair Fairly substantial pollution likely pollution probable

4.4.3 Scenario Results

All scenarios were compared to the original base simulation. Results for scenarios 1-4 were obtained from the watershed gauging station located in HRU 118. Results for the year 2013 revealed significant changes for discharge, chlorophyll a, DO, and sediment (Table 4.9). The year 2013 was chosen based on the most current data available. One year was isolated to illustrate seasonal fluctuations.

Table 4.9. Results for scenarios 1-8 indicating changes in focus variables and whether change is significant Scenario p(T<=t) one-tail Conclusion Discharge 25% canals <0.001 Significant difference 50% canals <0.001 Significant difference 75% canals <0.001 Significant difference 100% canals <0.001 Significant difference

Chlorophyll a 25% canals <0.001 Significant difference 50% canals <0.001 Significant difference 75% canals <0.001 Significant difference 100% canals <0.001 Significant difference

DO 25% canals <0.001 Significant difference 50% canals <0.001 Significant difference 75% canals <0.001 Significant difference 100% canals <0.001 Significant difference

Sediment Loading 25% canals <0.001 Significant difference 50% canals <0.001 Significant difference 75% canals <0.001 Significant difference 100% canals <0.001 Significant difference

Changes in chlorophyll α, DO and sediment indicate that even with a 25% reduction in headwater streams, significant changes in all variables would result. At 25% headwater stream burial, discharge saw an average of 100% decrease from the base scenario 11 months of the year

(Figure 4.4). September resulted in a 49% decrease from the base scenario. For all 4 scenarios

(25,50,75 and 100% headwater loss), the month of September was the only month that resulted in a smaller percentage decrease in discharge.

-20%

-40%

-60%

-80% Percent Percent Change -100%

-120%

Month

25% 50% 75% 100%

Figure 4.4. Discharge readings following headwater stream burial in increasing increments of 25% illustrating significant decreases (Year 2003)

20 /s 3 15

5 Discharge Discharge m 0

Month

Base 25% 50% 75% 100%

Figure 4.5. Percent change in discharge from base scenario readings with each headwater loss scenario (Year 2003) As a result of the decrease in discharge (Figure 4.4), chlorophyll α saw an increase in concentration above biological threshold of 20µg/L with 25% and 50% stream loss during spring freshet (Figure 4.6). Chlorophyll a thresholds can range from 10-15μg/L for sensitive aquatic species such as salmonids and 15-20 μg/L for other waters not likely inhabiting salmonids

(Pilgrim et al., 2001). As the headwater stream loss increased to 50%, the concentration dropped by 38% but remained greater than the base scenario readings during spring freshet (Figure 4.7).

Once the headwater stream burial reached 75%, readings for chlorophyll α fell below the base scenario readings by 63%. The base scenario readings were found to remain greater than the readings for all headwater stream loss scenarios during the months from June through to

February.

30 g/L

µ 25 a a 20

Chlorophyll 10

Month

Biological Threshold Base 25% 50% 75% 100%

Figure 4.6. Chlorophyll a concentration readings following headwater stream burial in increasing increments of 25%, (Year 2003) indication biological threshold of 20µg/L (Pilgrim et al., 2001) 1200% 1000% 800% 600% 400%

200% Percent Percent Cahnge 0% -200%

Month 25% 50% 75% 100%

Figure 4.7. Percent change in Chlorophyll a concentration from base scenario readings with each headwater loss scenario (Year 2003) Also influenced by discharge, DO readings decreased for the months of March through to

June, October, November and December (Figure 4.8). Reduced DO readings during these months can also be associated with increased chlorophyll α. DO readings did not fall below the

74 biological threshold < 4mg/L but had up to a 46% decrease from the base scenario during March following 50% headwater stream loss (Figure 4.8 and 4.9).

14 12

10 8

DO DO mg/L 6 4

2 0

Month Base 25% 50% 75% 100% Biological Threshold

Figure 4.8. DO concentration readings following headwater stream burial in increasing increments of 25%; biological threshold indicated at 4mg/L; <4mg/L is lethal to aquatic life (WHO,1984;IJC,1977) (Year 2003)

10% 0% -10% -20% -30%

Percent Percent Change -40% -50%

Month

25% 50% 75% 100%

Figure 4.9. Percent change in DO concentration from base scenario readings with each headwater loss scenario (Year 2003) Sediment readings decreased in succession following headwater stream burial. Visually,

Figure 4.10 illustrates the decreased variability that was found with the base scenario throughout

75 the year. The four scenarios displayed a similar pattern in sediment concentration fluctuations throughout the year but are more stable compared to the base scenario. The most drastic decrease in sediment concentration was found with scenario 4 during 100% headwater loss producing a

92.25% average monthly decrease from the base scenario (Figure 4.11).

10 9 8 7 6 5 4 3

Sediment Sediment mg/L 2 1 0

Month

Base 25% 50% 75% 100%

Figure 4.10. Sediment concentration readings following headwater stream burial in increasing increments of 25% (Year 2003)

0% -20% -40% -60%

Month -80% -100% -120%

Month

25% 50% 75% 100%

Figure 4.11 Percent change in sediment concentration from base scenario readings with each headwater loss scenario (Year 2003)

4.4.3.1 LDI

Results for LDI scenarios are indicated in Table 4.10 with t-test results for each variable

in Appendix D. Four primary LDI groups had the most significant impact on the 3 variables for

each OBBN site. Site 176 was the only site that showed no change from the base scenario

readings for DO.

Table 4.10. Results for scenarios 5-18 indicating LDI scenario that resulted in most significant change for chlorophyll a, DO and sediment loading at each OBBN site

OBBN Site Most significant p (<0.05) LDI Scenario Primary Land uses Percent of HW SWAT Sub. for variable associated with LDI within KC Chlorophyll a DO Sediment watershed Loading BC 36 LDI 0 LDI 0 LDI 0 Wetland 33% 45 LDI 0 LDI 0 LDI 0 Wetland 33% 71 LDI 0.4 LDI 0.4 LDI 0.4 Urban, Agriculture 14%

KC 64 LDI 0.11 LDI 0.11 LDI 0.11 Forest, Agriculture 10% 176 LDI 0 - LDI 0.4 Urban Agriculture 14% 1092 LDI 0.01 LDI 0.01 LDI 0.01 Wetland, Forest 13%

MC 1133 LDI 0.01 LDI 0 LDI 0 Wetland 33%

Scenario 5 involved all headwater streams characterized as LDI 0 converted to irrigation

canals. This produced a significant decrease in chlorophyll α, DO and sediment for OBBN

sample sites 36 and 45. Chlorophyll α at site 176 and DO and sediment at site 1133 had the most

significant decrease with scenario 5. Headwaters characterized under LDI 0 made up 33% of the

total headwaters within Kemptville Creek watershed and fell under the primary land use type of

wetlands. OBBN sample sites 36, 45, 76 and 1133 are located within land use types: urban,

urban/ agricultural, urban/forest and pasture/agricultural hay, respectively. Although, it may

appear reasonable to assume that the greater percentage of headwater stream burial would have

the most significant impact, this was not the case for all OBBN samples sites. Chlorophyll a

showed a significant decrease under scenario 14 for site 64 within KC. Scenario 14 involved

implementing irrigation canals for all headwater streams characterized as LDI 0.11 representing

10% of total headwaters. OBBN sample site 1092 for KC produced significant decreases in

chlorophyll a when scenario 7 was executed. Scenario 7 involved implementing irrigation canals

for all headwaters characterized as 0.01 representing 13% of total headwaters (Table 4.11).

Table 4.11. Highlighted LDI group scenarios that induced the greatest percent change for OBBN sites following headwater stream burial LDI TASP Percent Headwater Primary Land uses Headwaters Area (ha) 0 Very High 33% 106166.5 Wetland, Wetland 0.01 Very High 13% 6857.66 Wetland, Forest 0.02 Very High 9% 3274.78 Forest, Forest 0.03 High 1% 277.46 Wetland, Range 0.04 High 3% 1347.38 Pasture, Wetland 0.05 High 2% 532.55 Forest, Pasture 0.06 Medium 0% 141.28 Range, Range 0.08 Medium 1% 451.74 Pasture, Pasture 0.1 High 14% 7207.56 Agricultural, Wetland 0.11 High 10% 3574.93 Forest, Agricultural 0.13 Medium 1% 347.93 Agricultural, Range 0.14 Medium 2% 525.60 Agricultural, Pasture 0.4 Medium 14% 39318.32 Urban, Agricultural

The most significant changes in DO and sediment were seen with scenarios 5, 6, 11, and 14.

DO and sediment decreased under all scenarios. DO for sample site 176 for KC was not affected

under any of the scenarios.

Although the most significant scenario results for the variables being studied were decreases, site 176 and 1092 had increases in sediment following scenario 16 and 8, respectively.

Site 176 resulted in a 79% monthly average increase in sediment concentration following scenario

16 which involved LDI group 0.13 representing 1% of the headwaters and within agricultural/range land use types. Site 1092 resulted in a 13% monthly average increase in

78 sediment concentration with scenario 8 which involved LDI group 0.03 representing 1% of the headwaters and within wetland/range land use types.

Comparison of discharge at the gauging station location in subbasin 118 revealed the most significant difference after scenario 18 was executed (P(T<=t) one-tail = 1.85E-41).

Scenario 18 represented 14% of the total headwater streams within Kemptville Creek watershed characterized under LDI 0.4 and falling within the primary land use type of urban and agricultural.

Urban and agricultural land use types are commonly associated with higher runoff values. LDI headwaters groups 0 and 0.4 in proximity to BC sample sites with direction of flow indicated can be found in Appendix E. Upstream flow is routed primarily through wetland and urban area.

With observation of the LDI headwaters groups 0.01, 0.11 and 0.4 in proximity to KC

sample site locations, we see that headwater groups 0.01, and 0.11 are surrounded by the primary

land use of wetland, and agriculture, respectively (Appendix E). LDI headwaters group 0 in

proximity to MC sample site illustrates upstream flow routing through a large area of wetland

and urban area (Appendix E).

4.4.3.2 Predictions for Benthic Community Structure

Benthic macroinvertebrate dominant family counts were plotted for visual analysis and

reference for predictions (Appendix G). Correlation observed between discharge, chlorophyll a,

DO and sediment was assessed to provide a general idea of what benthic macroinvertebrate

families might be influenced by these variables.

Analysis of Pearson’s correlation coefficient (Pearson, 1895) showed that changes in chlorophyll a, DO, and sediment correlated with the abundance of some of the families found at sites within BC, KC and MC. Table 4.12 captures Pearson correlation values where changes in discharge, chlorophyll α, DO, and sediment could explain >50% of the abundance seen with

79 specific benthic macroinvertebrate families during spring and spring from 2003-2013. Correlation analysis also revealed that chlorophyll a correlated with more benthic macroinvertebrate families than DO and sediment.

Table 4.12. Summary of Pearson correlation analysis between discharge, chlorophyll a, DO and sediment loading and each of the top 6 families at OBBN sites. Sites that presented with Pearson correlation >.50 noting sample time as spring (S) and/or spring (F). Family Family Order Discharge Chlorophyll a DO Sediment Common Name Asellidae Crustaceans Isopoda 0.69 (36 F) - - - Baetidae Mayflies Ephemeroptera 0.54 (64 S) 0.70 (1133 F) 0.69 (1133 0.55 (64 F) F) 0.57 (36 S) 0.53 0.57 (71 F) (1092 S) Brachycentridae Caddisflies Trichoptera 0.84 (1133 F) 0.83 (1133 - F) Caenidae Mayflies Ephemeroptera - - - Chironomidae True Flies Diptera - 0.56 (36 S) - - 0.56 (36 F) - Elmidae Beetles Coleoptera - - - Ephemerellidae Mayflies Ephemeroptera - 0.54 (176 - S) Heptageniidae Mayflies Ephemeroptera - - - Hyalellidae Crustaceans Isopoda - - - Hydroptilidae Caddisflies Trichoptera - - - Hydropsychidae Caddisflies Trichoptera - 0.53 (71 S) 0.59 0.53 (71 F) (1092 F) Leptophlebiidae Mayflies Ephemeroptera - - - Limnephilidae Caddisflies Trichoptera 0.67 (36 S) - - Naididae Segmented Haplotaxida 0.61 (1092 S) 0.72 (45 S) - - Worms 0.53 (45 F) - - 0.58 (71 F) Simuliidae True Flies Diptera 0.65 (1092 S) - - Tubificidae Segmented Haplotaxida - - - Worms

OBBN were also tested for correlation between discharge and family abundance during

spring and spring from 2003-2013. Discharge correlated with the abundance of families

Asellidae, Baetidae and Naididae at >50% for sites 36 during the spring, 65 and 1092 during the

spring (Appendix F).

From observed correlation between family abundance and chlorophyll a, we can predict that with increased headwater stream burial within wetland areas, sites 36, 45, 71, 1092, and

1133 would project shifts in family structure. Families found to have >50% of their abundance explained by chlorophyll α concentration were from the orders Ephemeroptera, Trichoptera,

Haplotaxida, and Diptera. These are more sensitive taxa and are dependent on filter feeding, collecting and gathering. Increases in chlorophyll α concentrations would most likely impact survival characteristics as a result of benthic and sestonic algae blooms.

Decreases in DO would impact the families: Baetidae, Brachycentridae, Ephemerellidae and Hydropsychidae. These families belong to the Trichoptera, Ephemeroptera and Diptera orders and have lower tolerance values to environmental stress. Some of these families were found to have >50% of their abundance explained by DO readings for the sites: 36, 71, and 1133.

More significant is that all the sites saw decreases in DO below 4mg/L which is the biological threshold and lethal to aquatic life (WHO,1984; IJC,1977).

Fluctuations in sediment concentration was found to explain >50% of the abundance for the families Beatidae and Hydropsychidae (Ephemeroptera) for sites 64 and 1092, respectively.

Sediment reduction is usually seen as a positive effect, but sediment as fine particulate matter can reduce solar penetration to stream bottoms and restrict the growth of algae (Bilotta et al.,

2008). Coupled with the increases in chlorophyll α concentrations, sediment reduction could increase epiphytic algae growth and shift benthic macroinvertebrate community structure away from more sensitive Ephemeroptera to more algae eating Scraper/Grazers such as

Glossiphoniidae, Helicopsychidae, and Heptageniidae. With decreased sediment, less fine particulate matter would alleviate filtering effort and less time would be spent foraging

81 benefiting filter feeders such as Hydropsychidae and Simuliidae productivity favouring increases in abundance of members of these aquatic insect orders.

Cumulatively, results for discharge, chlorophyll α, DO and sediment would most likely lead to the formation of a eutrophic environment. Eutrophy would impact diversity, percent of tolerant species present for all sites. Diversity would decrease with more tolerant taxa survival and less tolerant taxa death, primary function feeding groups would be dominated by Scrapers and Grazers with shifts of available food sources to primarily benthic and sestonic algae.

4.5 Discussion

Human dominated land use types are the major influence on increasing pollution tolerant benthic macroinvertebrate taxa and feeding group distribution (Walsh et al. 2001; Lamberti and

Berg 1995). With the implementation of irrigation canals to simulate headwater loss, we observed decreases in important water quality variables. During high precipitation events and freshet, high flows contribute to increases in subsidies delivered downstream. Headwater stream overland flow transports fine sediments, and other organic matter that contribute to primary productivity downstream. Organic nitrogen (ON) concentrations are elevated during high-flow periods which is consistent with streams dominated by snowmelt (Kaushal and Lewis, 2003).

Higher nitrogen concentrations in turn influence chlorophyll α. With ON from shallow flow paths such as headwaters, it is transformed by sunlight and in-stream processing and photochemical oxidation, establishing bioavailable ON and thus driving in-stream respiration

(Scott et al., 2007).

Headwater loss for the land use types wetland/wetland, urban/agricultural, forest/agricultural, and wetland/forest, resulted in the greatest decrease in discharge, contributing

82 to increased chlorophyll α concentrations. DO concentration decreased with all headwater stream loss scenarios underlining the importance of headwaters contribution to discharge downstream.

Decreases in DO are most likely a result of decreased flow-induced turbulence which normally enhances oxygen dissolution in water (Ohimain et al., 2008). Sediment concentrations had varying results and were dependant on the location of the headwaters being buried. Increases in sediment concentrations seen with headwater loss within wetland/range, and agricultural/range land use types suggests that wetland/range and agricultural/headwaters upstream of site 176 and

1092, could have been acting as sinks for sediment loading. Sediment increases would induce physiological stress and habitat degradation for benthic macroinvertebrate communities

(Newcombe and MacDonald, 1991). Filter feeding benthic macroinvertebrates would be overwhelmed with increased feeding efforts and respiration rates. Headwater stream subsides that are converted by benthic macroinvertebrates from long term storage in sediment into food and ultimately food to downstream consumers, would be lost.

In consideration of the most influential LDI groups for each OBBN site, results did not match up with designated TASP values. The loss of wetland/wetland headwaters having the most impact is consistent with this LDI group having the highest TASP rank (1) as well as for wetland/forest headwaters (0.995) but urban/agricultural and forest/agricultural had TASP ranks that were much lower (0.695 and 0.175, respectively). At first glance LDI scenario results appeared to be correlated with headwater area in that the greater the area the headwaters LDI group covered the more significant the impact for downstream ecosystems but with further analysis we see that LDI group 0.1 representing a higher percentage (14%) of the headwaters covering an area of 7207.56 ha, did not have a significant impact on OBBN sites.

Wetland/wetland LDI group 0, had the most significant impact on chlorophyll α, DO and sediment concentrations for both OBBN sites and overall watershed outlet readings. Wetlands

83 are complex habitats with a high degree of inter- and intra-site variability in retaining, cycling and exporting nutrients (Johnes et al., 2020). Watersheds dominated by forested wetlands have higher biogeochemical transformation rates including mineralization, nitrification and denitrification (Scott et al., 2007).

Initial increases in chlorophyll α concentrations and decreases in DO and sediment following the incremental headwater stream burial illustrates the importance of cumulative water volume in individual stream segments in support of downstream ecosystems. Reinforced is the idea that the amount of water that headwaters convey across the landscape is more valuable in shaping downstream ecosystem than the landscape from which they originate. Discharge rate is important for benthic macroinvertebrate drift and thus community establishment downstream

(Dosskey et al., 2010). Discharge also contributes to bank soil wetting (Shentsis and Rosenthal,

2003) stimulating soil microbe activity which impacts vegetation composition (Klironomos,

2002). Shifts in benthic macroinvertebrate communities would also influence terrestrial predator species such as amphibians, insectivore birds and small rodents.

The loss of headwater streams from any land use area would shorten the distance for benthic macroinvertebrate drift potential and remove refuge habitat from predatory fish. Benthic macroinvertebrates would be forced downstream and into deeper larger waters where they might become prey more readily. The use of synthetic water quality information involved in this study did not allow for complete confidence in simulated water quality results, but tangible relationships were illustrated between water quality variables and benthic macroinvertebrate family abundance. Improving the frequency of water quality sampling and the range of variables sampled for, would contribute to further evaluation of the relationships seen in this study.

4.6. Conclusions

Irrigation canal implementation successfully simulated scenarios that artificially removed headwaters from the watershed stream network. Isolating groups of headwater streams based on land use disturbances and potential contribution of downstream subsidies, identified the more significant headwater groups to specific stream sites. Pairing model predictions with a multi- metric bioassessment approach brought insight into future benthic macroinvertebrate community structure.

The loss of headwater streams through stream burial can be categorized as a loss in habitat connectivity. Habitat connectivity is a fundamental principle of population biology across varying spatial and temporal scales (Meffe and Carroll, 1997) and important for the overall health of watershed systems. Stream burial fragments a watershed habitat, limiting upstream contribution to downstream populations. Headwater streams require increased monitoring to advance understanding. Accounting for up to 80% of most stream networks, headwater features set the stage for downstream health.

4.7 References

Adu, B.W., Oyeniyi, E.A. 2019.Water quality parameters and aquatic insect diversity in Aahoo stream, southwestern Nigeria. Journal of Basic and Applied Zoology, 80: 15 Argerich A., Haggerty R., Johnson S.L, Wondzell S.M, Dosch N., Corson-Rikert H., Ashkenas L.R, Pennington R, and Thomas CK. 2016. Comprehensive multiyear carbon budget of a temperate headwater stream. Journal of Geophysical Research, 121:1306–1315. Arnold, J. G., Allen, P.M., Muttiah, R. and Bernhardt. G. 1995. Automated baseflow separation and recession analysis techniques. Ground Water, 33(6): 1010-1018. Aspray, K.L., Holden, J., Ledger, M.E., Mainstone, C.P and Brown, L.E. 2016. Organic sediment pulses impact rivers across multiple levels of ecological organization. Ecohydrology, 10:1855-1870. Auta, Y.I., Arimoro, F.O., Mohammed, A. Z. and Ayanwale, A.V. 2016. Monitoring seasonal changes in environmental variables and Chironomid diversity at Shiroro Lake, Niger state, Nigeria. International Journal of Applied Biological Research, 7(1): 36-47. Barber-James, H.M., Gattolliat, J.L., Sartori, M. and Hubbard, M.D.2008. Global diversity of mayflies (Ephemeroptera, Insecta) in freshwater. Hydrobiologia, 595: 339–350. Beaulieu, J.J., Mayer, P.M, Kaushal, S.S., Pennino, M.P, Arango, C.P, Balz, D.A., Fritz, K.M., Hill, B.H., Elonen, C.M., and Santo Domingo, J.W. 2014. Effects of urban stream burial on organic matter dynamics and reach scale nitrate retention. Biogeochemistry, 121: 107–126. Biggs, B. J. F 2000. New Zealand periphyton guide-lines: detecting, monitoring and managing of enrichment of streams: background and guidelines. Ministry for the Environment, Wellington, New Zealand. Bilotta, G. S., and Brazier, R. E. 2008. Understanding the influence of suspended solids on water quality and aquatic biota. Water Research, 42: 2849–2861. Bode, R.W., Novak, M.A., and Abele, L.E. 1996. Quality Assurance Work Plan for Biological Stream Monitoring in New York State. NYS Department of Environmental Conservation, Albany, NY. 89p. Broekhuizen, N., Parkyn, S.and Miller, D. 2001. Fine sediment effects on feeding and growth in the invertebrate grazers Potamopyrgus antipodarum (Gastropoda, Hydrobiidae) and Deleatidium sp (Ephemeroptera, Leptophlebiidae). Hydrobiologia, 457: 125-132. Dodds, W K., J. R. Jones, J.R. and Welch, E.B. 1998. Suggested classification of stream trophic state: distributions of temperate stream types by chlorophyll, total nitrogen, and phosphorus. Water Research, 32:1455-1462. Dosskey, M.G., Vidon, P., Gurwick, N.P., Allan, C.J., Duval, T.P., Lowrance, R., 2010. The role of riparian vegetation in protecting and improving chemical water quality in streams. Journal of the American Water Resources Association, 46: 261-277. Elosegi, A., Dıez, J.R. & Pozo, J. 2007.Contribution of dead wood to the carbon flux in forested streams. Earth Surface Processes and Land-forms, 32: 1219–1228. Gebrehiwot, M., Awoke, A., Beyene, A., Kifle, D., Triest, L., 2017. Macroinvertebrate community structure and feeding interactions along a pollution gradient in Gilgel Gibe watershed, Ethiopia: implications for biomonitoring. Limnologica, 62: 68–76. Hauer, F.R., and Lamberti, G.A. 1996. Methods in Stream Ecology. Academic Press. ISBN: 0- 12-332906-X. 696pp. Hilsenhoff, W.L. 1982. Using a biotic index to evaluate water quality in streams: Wisconsin Department of Natural Resources Technical Bulletin no. 132, 22 p.

Hilsenhoff, W.L. 1988. Rapid field assessment of organic pollution with a family-level biotic index. Journal of North American Benthological Society, 7(1):65-68. International Joint Commission (IJC). 1977. New and Revised Specific Water Quality Objectives. Great Lakes Water Quality Board Report, 11-18. Jiang, P. H., Y. D. Cui, H. J. Wang, and H. Z. Wang. 2011.Macrozoobenthos communities and bio-assessment of lakes in Hanyang District. Resources and Environment in the Yangtze Basin, 20:525–533. Jones C, Somers KM, Craig B, Reynoldson TB. 2004. Ontario Benthos Biomonitoring Network protocol manual, version 1.0. Ontario Ministry of Environment. Klironomos, J. N. 2002..Feedback with soil biota contributes to plant rarity and invasiveness in communities. Nature, 417: 67–70. Lamberti, G. A., and M. B. Berg. 1995. Invertebrates and other benthic features as indicators of environmental change in Juday Creek, Indiana. Natural Areas Journal, 15(3):249–258. Little CJ, and Altermatt F. 2018. Species turnover and invasion of dominant freshwater invertebrates alter biodiversity–ecosystem function relationship. Ecological Monographs. Magurran, A.E. 1988. Ecological diversity and its measurement. — Princeton University Press, Princeton, New Jersey. Marcarelli, A.M., Baxter, C.V., Mineau, M.M. and Hall, R. O. 2011.Quantity and quality: unifying food web and ecosystem perspectives on the role of resource subsidies in freshwaters. Ecology, 92: 1215-1225. Meffe, G.K. and C.R. Carroll. 1997. Principles of Conservation Biology, 2nd edition. Sinauer and Associates. Morrison M.M, Marcot B.G, Mannan RW. 2006. Wildlife-Habitat Relationships: Concepts and Applications. Island Press: Washington, DC; 493. Newcombe, C.P., MacDonald, D.D., 1991. Effects of suspended sediments on aquatic ecosystems. North American Journal of Fisheries Management, 11: 72–82. Pilgrim, K., D. Sanders, and T. Dupuis. 2001. Relationship between chlorophyll a and beneficial uses. CH2M Hill, Boise, Idaho. 10 p. Ohimain EI, Imoobe OT, Dorcas DS. 2008.Changes in water physico-chemical properties following the dredging of an oil well access canal in the Niger Delta. World journal of agricultural sciences, 4(6):752-758. Ontario Ministry Agriculture, Food and Rural Affairs (OMAFRA). 2004. Details on Drainage Program Changes. Ontario Ministry Agriculture, Food and Rural Affairs. Pearson, K.1895. "Notes on regression and inheritance in the case of two parents," Proceedings of the Royal Society of London, 58 :240–242. Pechmann. J. H. K. and Scott, D.E., Gibbons,. J. W., and Semlitsch, R. D. 1989. Influence of wetland hydroperiod on diversity and abundance of metamorphosing juvenile amphibians. Wetlands Ecology and Management, 1: 3-1. Peterman, W.E, Crawford, .JA, Semlitsch R.D.2008. Productivity and significance of headwater streams: population structure and biomass of the black-bellied salamander (Desmognathus quadramaculatus). Freshwater Biology, 53:347–357. Plafkin, J.L., Barbour, M.T., Porter, K.D., Gross, S.K., and Hughes, R.M. 1989. Rapid bioassessment protocols for use in streams and rivers: benthic macroinvertebrates and fish. EPA/444/4–89–001. Office of Water, US Environmental Protection Agency, Wash. Rosenberg, D.M., and Resh V.H.1993.Freshwaterbiomonitoring and benthic macroinvertebrates. Chapman and Hall, London, UK

Scott, D., J. Harvey, R. Alexander, and G. Schwarz .2007. Dominance of organic nitrogen from headwater streams to large rivers across the coterminous United States, Global Biogeochemical Cycles, 21:1-8. Shannon, C.E. and Weaver, W.1949.The mathematical theory of communication. University of Illinois Press, Urbana. Stanfield W., Kilgour B.W. 2013.How proximity of land use affects stream fish and habitat. River Research and Applications, 29:891–905. Su, E.B.2020. Hydrological impacts of headwater stream burial for Kemptville Creek watershed as predicted by SWAT model. (Unpublished MSc. Thesis). Queen’s University, Kingston, ON. Townsend CR, Uhlmann SS, Matthaei CD. 2008. Individual and combined responses of stream ecosystems to multiple stressors. Journal of Applied Ecology, 45: 1810–1819. Uwadiae, R.E. 2016. Benthic macroinvertebrate community and chlorophyll a (chl-a) concentration in sediment of three polluted sites in the Lagos Lagoon, Nigeria. Journal of Applied Sciences and Environmental Management, 20(4): 1147-1155. Vannote R.L., Minshall G.W., Cummins K.W., Sedell J.R. and Cushing C.E. 1980. The River Continuum Concept. Canadian Journal of Fisheries and Aquatic Sciences, 37: 130–137 Walsh, C. J., A. Sharpe, P. F. Breen, and J. A. Sonneman. 2001. Effects of urbanization on streams of the Melbourne region, Victoria, Australia. I. Benthic macroinvertebrate communities. Freshwater Biology, 46:535–551. Walsh, C. J., Roy, A., Feminella, J.W., Cottingham, P.D., Groffman, P.M., and Morgan, R.P. 2005. ‘The Urban Stream Syndrome: Current Knowledge and the Search for a Cure’, Journal of the North American Benthological Society, 24: 706–23. Wipfli M.S. and Musslewhite J. 2004. Density of red alder (Alnus rubra) in headwaters influences invertebrate and detritus subsidies to downstream fish habitats in Alaska. Hydrobiologia, 520: 153–163. WHO (World Health Organization).1984. International Standard for Drinking Water (4th Ed), Geneva, 327 pp.

Chapter 5 – Conclusions

5.1 Summary

The SWAT model was evaluated for it’s ability to predict headwater location and total monthly discharge for Kemptville Creek watershed. Predictions were made with limited input data and single variable calibration and validation. Scale of delineation was reduced from 735 hectares to 10, to extend the stream network and include predicted headwaters up to the watershed boundary. Headwaters predicted by SWAT increased total stream length by 102%, adding 443km of headwaters within Kemptville Creek watershed. In field confirmation resulted in 70% accuracy of headwater stream location. Adding the 443km of predicted headwaters, improved calibration by 64% (from R2 = 0.27 to R2 = 0.76).

Manually calibrating the model identified that parameters governing snow water were critical to Kemptville Creek discharge. Major outliers seen through residual assessments, brought further clarity on timing of snow fall and spring thaw. Snow fall and melt temperature, minimum snow water content corresponding to 100% and 50% snow cover and melt factor parameters needed further calibration. Manual and auto calibration resulted in R2 and NS values that were ideal for the SWAT model (R2 = 0.76 and NS = 0.72).

Headwater loss scenarios by tile drainage produced no change from base scenario discharge while headwater loss by irrigation canal produced a significant change in discharge (>90%). The magnitude of discharge lost within headwater regions also left these streams with little to no water for aquatic habitat. Decreases in discharge following headwater loss, indicated that headwaters convey a significant volume of water and when discharge is restricted by burial, downstream water chemistry changes.

At 100% headwater stream loss by irrigation canal, concentrations of DO, chlorophyll a and sediment, decreased in downstream ecosystems. These variables were found to correlate

89 with families of benthic macroinvertebrates currently inhabiting downstream sites within

Kemptville Creek watershed. Decreases in these variables would establish a less favourable environment for families of benthic macroinvertebrates that have less tolerance to environmental stress. Predicted shifts in benthic macroinvertebrate communities in response to changes in DO, chlorophyll a and sediment, subsequently impact food web dynamics within and beyond the watershed.

Rehabilitating and sustaining aquatic ecosystems should be inclusive of headwater regions.

If headwater burial continues to take place during land use development, then as seen with this study, even a small percentage of burial within headwater regions can result in significant change for discharge, DO, chlorophyll a and sediment at a watershed scale.

Understanding the function of headwaters at the catchment scale is essential to implementing more successful land management plans. Using models like SWAT to predict headwater location followed by in field confirmation, are the first steps to discovering cumulative hydrological processes within a watershed.

5.2 Limitations

Lack of soil data likely limited the accuracy of some results, particularly related to stream water nutrient fluxes. Soil types were implemented based on US soils by matching US soils as closely as possible to soil characteristics of the Canadian soils within the study site. The use of

US soil classes as a substitute for Canadian soils may have cause errors during model simulations. These errors could have been associated with the different texture classes and how

SWAT interprets soil water percolation and evaporation. Although textures were matched as closely as possible, soil parameters such as soil crack can be greatly influenced by the percentage of clay. Slight differences in bulk density, soil conductivity, soil albedo for example, may have

90 led to more or less water moving through the soil and overall watershed system which impacts discharge results.

Data on snow within the region was also limited and with the potential of snow water varying temporally, calibration for snow parameters was challenging.

Higher resolution spatial data specifically for the DEM input, would support hydrological delineation within the model and improved the accuracy of headwater location and morphology.

Headwater stream confirmation was limited due to the required land access to observed and measure active headwaters. In field verification involved only 3% of the 908 SWAT predicted headwater streams based on visibility or accessibility from road crossings. With access granted by landowners, further headwater stream confirmation would help improve understanding of headwater function within agricultural lands.

Chlorophyll a, DO, and sediment concentrations were not calibrated or validated within the original model set-up, therefore values produced by the model were ignored. However, the importance of models, like SWAT is to fill in knowledge gaps and based on outcomes for these variables following headwater burial, we can conclude that the biological diversity of benthic macroinvertebrates would decrease with less sensitive families increasing in abundance.

Sustaining diversity for benthic macroinvertebrates subsequently sustains diversity for predator species.

5.3 Future Directions

Headwater stream research continues to become more prevalent, and numerous studies have already proven the importance of headwaters for specific study areas. Even if data are limited, hydrological models such as SWAT have great potential for supporting watershed and watershed decisions on environmental planning and development. Although, studies have shown improved results when SWAT is coupled with other models, this study has shown that depending on what

91 variable outputs you are focusing on, SWAT on its own can provide insights into watershed function.

While the SWAT model performed well despite the lack of data for the study area, important variables that could have influenced chlorophyll a, DO and sediment loading results were not addressed such as total phosphorous input and water temperature.

This study is executed over the time period from 1984 to 2013, this time period was chosen based on data availability. This 30-year period would have seen land use change within the watershed and most likely the growth of residential and commercial area. Changes in land use were not implemented except for the implementation of tile-drainage installment over the study period. The degree of land use change that took place within Kemptville and surrounding area over the 30-year period was uncertain. Speculation based on the available data used in this study would suggest that the conversion from agricultural land to commercial and residential land more than likely took place over this time period and was probably focused within the eastern boarders of Kemptville based on it’s proximity to Ottawa, Ontario. If most of the land use change did take place within Kemptville, the impact on discharge measurements may not have been significant because the gauging station lies upstream (south) of Kemptville. The fact that land use is static for the 30-years during simulations is not realistic and changes not accounted for could have potentially contributed to the result of a lower NS and R2 during calibration and validation.

The addition of two new gauging stations positioned within the North and South Branch subbasins would isolate and identify processes taking place within these systems. These two subbasins converge and form the main channel within Kemptville Creek watershed. Individual calibration using discharge data for each branch would bring further knowledge of processes

92 influencing fluxes seen with discharge, chlorophyll a, sediment and DO and potential links relative to location within the watershed.

Although there are gaps in available data for Kemptville Creek watershed that might have improved calibration and the confidence in scenario outcomes, there are some important observations that support the ability to make reasonable predictions of headwater stream delineation using the SWAT model. Calibration and validation also confirmed that SWAT model simulates monthly discharge quite well and without the use of additional model extensions or code manipulation.

Changes to any system are often a result of cumulative effects and thus it is difficult to isolate one variable as the cause of change. Although, the results from modelling headwater stream burial with Kemptville Creek have shown that headwater loss has a significant impact on downstream ecosystems, it is important to increase confidence in these results by expanding the knowledge of the study area. Data collection continues to contribute to advancing knowledge of natural systems and resources all over the world but where data collection is absent, natural systems are often suffering. With the use of models such as SWAT, reliable predictions can be made for hydrological systems to prevent and protect them before they are negatively impacted.

Appendix A

Observations for 30 predicted headwater streams looking for visible path of flow and active flow; 50%,53%,40% and 37% of predicted headwaters were actively flowing during observation in March, April, May and June 2019, respectively

Stream Land use Distinct Path March Flow April Flow May Flow June Flow 8 AG No Yes Yes Yes Yes 9 AG Yes Yes No No No 10 AG Yes Yes Yes Yes Yes 11 SWMP/AG Yes Yes Yes Yes Yes 12 AG Yes Yes No No No 15 FRST/RES Yes Yes Yes Yes Yes 16 SWMP No No No No No 17 SWMP Yes Yes Yes Yes Yes 18 AG Yes Yes No No No 19 AG No No No No No 34 FRST No No Yes Yes No 38 AG No No Yes Yes No 39 AG Yes No Yes Yes Yes 40 AG Yes No Yes Yes Yes 41 AG/RES No No No No No 42 AG Yes Yes No No No 43 AG Yes Yes Yes Yes No 44 AG No No No No No 45 AG Yes Yes Yes Yes Yes 46 FRST No No No No No 47 AG No No No No No 48 AG Yes Yes Yes Yes Yes 49 AG No No No No No 50 AG No No No No No 51 AG/RES No Yes Yes Yes Yes 52 AG/RES No Yes No No No 55 Industrial Yes No Yes Yes Yes 56 Industrial Yes No Yes Yes No 57 Trail No No No No No 58 Cemetery Yes Yes Yes Yes No

Appendix B

LDI headwater stream classification for Kemptville Creek watershed

Appendix C

Tolerance Values for benthic macroinvertebrates found within Kemptville Creek watershed in part of the Family Biotic Index (Bode et al., 1996; Hauer and Lamberti, 1996; Hilsenhoff, 1988; Plafkin et al.,1989) Phylum Class Order Family Tolerance Value Arthropoda Insecta Crustacea Gammaridae 4 Diptera Athericidae 2 Ceratopogonidae 6 Chironomidae 6 Dolichopodidae 4 Ephydridae 6 Empididae 6 Muscidae 6 Psychodidae 10 Simuliidae 6 Syrphidae 10 Tabanidae 6 Tipulidae 3 Coleoptera Dryopidae 5 Psephenidae 4 Ephemeroptera Ameletidae 0 Baetidae 4 Caenidae 7 Ephemerellidae 1 Ephemeridae 4 Heptageniidae 4 Leptophlebiidae 2 Lestidae 9 Oligoneuriidae 0 Lepidoptera Pyralidae 5 Megaloptera Corydalidae 0 Sialidae 4 Ordonata Aeshnidae 3 Calopterygidae 5 Coenagrionidae 9 Corduliidae 5 Gomphidae 1

Libellulidae 9 Sphaeriidae 8 Plecoptera Capniidae 1 Chloroperlidae 1 Helicopsychidae 3 Leuctridae 0 Nemouridae 2 Perlidae 1 Perlodidae 2 Taeniopterygidae 2 Trichoptera Brachycentridae 1 Glossosomatidae 0 Hydropsychidae 4 Hydroptilidae 4 Lepidostomatidae 1 Limnephilidae 4 Odontoceridae 0 Oligoneuriidae 2 Philopotamidae 3 Polycentropodidae 6 Psychomyiidae 2 Uenoidae 3 Malacostraca Amphipoda Hyalellidae 8 Isopoda Asellidae 8 Cnidaria Hydrozoa Anthoathecata Hdridae 5 Mollusca Heterobranchia Gastropoda Lymnaeidae 6 Physidae 8 Bivalvia Sphaeriida Sphaeriidae 8 Platyhelminthes Turbellaria 4

Appendix D

LDI scenario t-test results for each OBBN sample site

OBBN Sample Significant Difference Significant Difference Significant Difference Site SWAT for Chlorophyll A for DO for Sediment Loading Subbasin 36 LDI 0 LDI 0 LDI 0 BARNES P(T<=t) one-tail P(T<=t) one-tail P(T<=t) one-tail CREEK 1.12E-09 1.25E-51 4.66E-40 P(T<=t) two-tail P(T<=t) two-tail P(T<=t) two-tail 2.25E-09 2.51E-51 9.32E-40

45 LDI 0 LDI 0 LDI 0 BARNES P(T<=t) one-tail P(T<=t) one-tail P(T<=t) one-tail CREEK 1.58E-09 2.74E-55 2.37E-42 P(T<=t) two-tail P(T<=t) two-tail P(T<=t) two-tail 3.16E-09 5.47E-55 4.75E-42

64 LDI 0.11 LDI 0.11 LDI 0.11 KEMPTVILLE P(T<=t) one-tail P(T<=t) one-tail P(T<=t) one-tail CREEK 1.04E-20 6.93E-48 3.25E-37 P(T<=t) two-tail P(T<=t) two-tail P(T<=t) two-tail 2.08E-20 1.39E-47 6.49E-37

71 - LDI 0.2 LDI 0.2 BARNES P(T<=t) one-tail P(T<=t) one-tail CREEK 0.005201 4.1E-36 P(T<=t) two-tail P(T<=t) two-tail 0.010402 8.21E-36

176 LDI 0.2 - LDI 0.2 KEMPTVILLE P(T<=t) one-tail P(T<=t) one-tail CREEK 1.05E-26 1.47E-16 P(T<=t) two-tail P(T<=t) two-tail 2.1E-26 2.94E-16

1092 LDI 0.01 LDI 0.01 LDI 0.01 KEMPTVILLE P(T<=t) one-tail P(T<=t) one-tail P(T<=t) one-tail CREEK 4.17E-10 1.86E-54 7.85E-35 P(T<=t) two-tail P(T<=t) two-tail P(T<=t) two-tail 8.33E-10 3.72E-54 1.57E-34

1133 LDI 0 LDI 0 LDI 0 MUD CREEK P(T<=t) one-tail P(T<=t) one-tail P(T<=t) one-tail 5.14E-10 8.35E-53 5.73E-35 P(T<=t) two-tail P(T<=t) two-tail P(T<=t) two-tail 1.03E-09 1.67E-52 1.15E-34

Appendix E

BC OBBN sample sites and LDI classified headwaters that had the greatest impact on site

100

KC OBBN sample sites and LDI classified headwaters that had the greatest impact on OBBN site

101

KC OBBN sample sites and LDI classified headwaters that had the greatest impact on OBBN site

102

MC OBBN sample sites and LDI classified headwaters that had the greatest impact on OBBN site

103

Appendix F

Benthic macroinvertebrate families which resulted in greater than >50% correlation with discharge

100 12

90 10 80

8 /s 60 3

50 6

Taxa Taxa Count 40

4 Discharge m 30

20 2 10

0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Date

Asellidae Baetidae Naididae 36 Fall 64 Spring 1092 Spring

104

Appendix G

Benthic macroinvertebrate family count vs base scenario and LDI 0 scenario chlorophyll a concentration for site 36 of KC during May and October from 2003 – 2013 May 400 1.6

350 1.4

300 1.2 g/L

250 1 µ ɑ 200 0.8

Taxa Taxa Count 150 0.6

100 0.4 Chlorophyll

50 0.2

0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year

Chironomidae Simuliidae Baetidae Elmidae Naididae Limnephilidae Base 36 LDI 0

October 700 0.6

600 0.5

500

0.4 µg/L

400 ɑ 0.3

300 Taxa Taxa Count 0.2

200 Chlorophyll

100 0.1

0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year

Chironomidae Asellidae Hyalellidae Elmidae Naididae Hydropsychidae Base 36 LDI 0

105

Benthic macroinvertebrate family count vs base scenario and LDI 0 scenario DO concentration for site 36 of KC during May and October from 2003 – 2013

May 400 12 350 10 300 8 250

200 6 mg/L 2 2

150 DO Taxa Taxa Count 4 100 2 50 0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year

Chironomidae Simuliidae Baetidae Elmidae Naididae Limnephilidae Base 36 LDI 0 Biological Threshold

October 700 12

600 10 500 8 400

6 mg/L

300 2 DO

Taxa Taxa Count 4 200 100 2 0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year

Chironomidae Asellidae Hyalellidae Elmidae Naididae Hydropsychidae Base 36 LDI 0 Biological Threshold

106

Benthic macroinvertebrate family count vs base scenario and LDI 0 scenario sediment concentration for site 36 of KC during May and October from 2003 – 2013

May 400 12

350 10 300 8 250

200 6

Taxa Taxa Count 150

4 Sediment mg/L 100 2 50

0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year

Chironomidae Simuliidae Baetidae Elmidae Naididae Limnephilidae Base 36 LDI 0

October 700 9

8 600 7 500 6

400 5

300 4 Taxa Taxa Count

3 Sediment mg/L 200 2 100 1

0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year

Chironomidae Asellidae Hyalellidae Elmidae Naididae Hydropsychidae Base 36 LDI 0

107

Benthic macroinvertebrate family count vs base scenario and LDI 0 scenario chlorophyll a concentration for site 45 of KC during May and October from 2003 – 2013

May 400 1.6 350 1.4 300 1.2

250 1 µg/L ɑ ɑ 200 0.8

150 0.6 Taxa Taxa Count

100 0.4 chlorophyll 50 0.2 0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year

Chironomidae Simuliidae Baetidae Elmidae Naididae Asellidae Base 45 LDI 0

October 600 0.5 0.45 500 0.4 0.35 400 g/L

0.3 µ ɑ 300 0.25 0.2 Taxa Taxa Count 200

0.15 chlorophyll 0.1 100 0.05 0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year

Chironomidae Hydropsychidae Tubificidae Elmidae Asellidae Naididae Base 45 LDI 0

108

Benthic macroinvertebrate family count vs base scenario and LDI 0 scenario DO concentration for site 45 of KC during May and October from 2003 – 2013

May 400 12 350 10 300 8 250

200 6 mg/L 2 2

150 DO Taxa Taxa Count 4 100 2 50 0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year

Chironomidae Simuliidae Baetidae Elmidae Naididae Asellidae Base 45 LDI 0 Biological Threshold

October 600 12

500 10

400 8

300 6 mg/L

2 2 DO Taxa Taxa Count 200 4

100 2

0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year

Chironomidae Hydropsychidae Tubificidae Elmidae Asellidae Naididae Base 45 LDI 0 Biological Threshold

109

Benthic macroinvertebrate family count vs base scenario and LDI 0 scenario sediment concentration for site 45 of KC during May and October from 2003 – 2013

May 400 9

350 8

300 7 6 250 5 200 4

Taxa Taxa Count 150

3 Sediment Sediment mg/L 100 2 50 1 0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year

Chironomidae Simuliidae Baetidae Elmidae Naididae Asellidae Base 45 LDI 0

October 600 9 8 500 7 400 6 5 300 4

Taxa Taxa Count 200 3 Sediment Sediment mg/L 2 100 1 0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year

Chironomidae Hydropsychidae Tubificidae Elmidae Asellidae Naididae Base 45 LDI 0

110

Benthic macroinvertebrate family count vs base scenario and LDI 0.4 scenario chlorophyll a concentration for site 71 of KC during May and October from 2003 – 2013

May 500 4.5 450 4 400 3.5

350 g/L

3 µ 300 2.5 250 2 200 Taxa Taxa Count 1.5

150 Chlorophyllɑ 100 1 50 0.5 0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year

Chironomidae Simuliidae Baetidae Elmidae Hydropsychidae Hydroptilidae Base 71 LDI 0.4

October 500 0.7 450 0.6 400

350 0.5 µg/L

300 0.4 ɑ 250

200 0.3 Taca Taca Count

150 0.2 Chlorophyll 100 0.1 50 0 0 BC 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year

Chironomidae Tubificidae Hydropsychidae Elmidae Asellidae Naididae Base 71 LDI 0.4

111

Benthic macroinvertebrate family count vs base scenario and LDI 0.4 scenario DO concentration for site 71 of KC during May and October from 2003 – 2013

May 500 12 450 400 10 350 8 300

250 6 mg/L 2 2

200

DO Taxa Taxa Coun 150 4 100 2 50 0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year

Chironomidae Simuliidae Baetidae Elmidae Hydropsychidae Hydroptilidae Base 71 LDI 0.4 Biological Threshold

October 500 12 450 10 400 350 8 300

250 6 mg/L 2 2

200 DO Taxa Taxa Count 4 150 100 2 50 0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year

Chironomidae Tubificidae Hydropsychidae Elmidae Asellidae Naididae Base 71 LDI 0.4 Biological Threshold

112

Benthic macroinvertebrate family count vs base scenario and LDI 0.4 scenario sediment concentration for site 71 of KC during May and October from 2003 – 2013

May 500 9 450 8 400 7 350 6 300 5 250 4 200 Taxa Taxa Count 3 150 Sediment mg/L 100 2 50 1 0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year

Chironomidae Simuliidae Baetidae Elmidae Hydropsychidae Hydroptilidae Base 71 LDI 0.4

October 500 9 450 8 400 7 350 6 300 5 250 4 200 Taxa Taxa Count 3 150 Sediment mg/L 100 2 50 1 0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year

Chironomidae Tubificidae Hydropsychidae Elmidae Asellidae Naididae Base 71 LDI 0.4

113

Benthic macroinvertebrate family count vs base scenario and LDI 0.11 scenario chlorophyll a concentration for site 64 of KC during May and October from 2003 – 2013

May 300 6

250 5 g/L 200 4 µ 150 3

100 2 Taxa Taxa Count

50 1 Chlorophyllɑ 0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year

Chironomidae Simuliidae Baetidae Asellidae Hyalellidae Elmidae Base 64 LDI 0.11

October 350 8

300 7 6

250 g/L µ

5 ɑ 200 4 150

3 Taxa Taxa Count 100

2 Chlorophyll 50 1 0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year

Chironomidae Baetidae Hyalellidae Heptageniidae Caenidae Ephemerellidae Base 64 LDI 0.11

114

Benthic macroinvertebrate family count vs base scenario and LDI 0.11 scenario DO concentration for site 64 of KC during May and October from 2003 – 2013

May 300 12

250 10

200 8

150 6 mg/L

2 2 DO Taxa Taxa count 100 4

50 2

0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year

Chironomidae Simuliidae Baetidae Asellidae Hyalellidae Elmidae Base 64 LDI 0.11 Biological Threshold

October 350 12

300 10 250 8 200

6 mg/L

150 2 DO

Taxa Taxa Count 4 100 50 2 0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year

Baetidae Chironomidae Hyalellidae Heptageniidae Caenidae Ephemerellidae Base 64 LDI 0.11 Biological Threshold

115

Benthic macroinvertebrate family count vs base scenario and LDI 0.11 scenario sediment concentration for site 64 of KC during May and October from 2003 – 2013

May 300 9 8 250 7 200 6 5 150 4

Taxa Taxa Count 100 3 Sediment Sediment mg/L 2 50 1 0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year

Chironomidae Simuliidae Baetidae Asellidae Hyalellidae Elmidae Base 64 LDI 0.11

October 350 9 8 300 7 250 6 200 5

150 4

Taxa Taxa Count 3 100 Sediment mg/L 2 50 1 0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year

Baetidae Chironomidae Hyalellidae Heptageniidae Caenidae Ephemerellidae Base 64 LDI 0.11

116

Benthic macroinvertebrate family count vs base scenario and LDI 0.01 scenario chlorophyll a concentration for site 1092 of KC during May and October from 2003 – 2013

May 400 0.6

350 0.5 300

0.4 g/L

250 µ

200 0.3

Taxa Taxa Count 150 0.2 100 chlorophyllɑ 0.1 50

0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year

Chironomidae Simuliidae Baetidae Elmidae Hyalellidae Hydropsychidae Base 1092 LDI 0.01

October 400 0.16

350 0.14

300 0.12 g/L

250 0.1 µ ɑ ɑ 200 0.08

Taxa Taxa Count 150 0.06

100 0.04 chlorophyll

50 0.02

0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year

Chironomidae Hyalellidae Baetidae Elmidae Hydropsychidae Leptophlebiidae Base 1092 LDI 0.01

117

Benthic macroinvertebrate family count vs base scenario and LDI 0.01 scenario DO concentration for site 1092 of KC during May and October from 2003 – 2013

May 400 12

350 10 300 8 250

200 6 mg/L

2 2 DO

Taxa Taxa Count 150 4 100 2 50

0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year

Chironomidae Simuliidae Baetidae Elmidae Hyalellidae Hydropsychidae Base 1092 LDI 0.01 Biological Threshold

October 400 12

350 10 300 8 250

200 6 mg/L

2 2 Axis Axis Title 150 DO 4 100 2 50

0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Axis Title

Chironomidae Hyalellidae Baetidae Elmidae Hydropsychidae Leptophlebiidae Base 1092 LDI 0.01 Biological Threshold

118

Benthic macroinvertebrate family count vs base scenario and LDI 0.01 scenario sediment concentration for site 1092 of KC during May and October from 2003 – 2013

May 400 3.5

350 3 300 2.5 250 2 200 1.5

Taxa Taxa Count 150 1 Sediment mg/L 100

50 0.5

0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year

Chironomidae Simuliidae Baetidae Elmidae Hyalellidae Hydropsychidae Base 1092 LDI 0.01

October 400 2.5

350 2 300

250 1.5

200

Taxa Taxa Count 1

150 Sediment Sediment mg/L

100 0.5 50

0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year

Chironomidae Hyalellidae Baetidae Elmidae Hydropsychidae Leptophlebiidae Base 1092 LDI 0.01

119

Benthic macroinvertebrate family count vs base scenario and LDI 0 scenario chlorophyll a concentration for site 1133 of MC during May and October from 2003 – 2013

May 400 70

350 60 300 50

250 µg/L 40 ɑ 200 30

Taxa Taxa Count 150

20 chlorophyll 100

50 10

0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year

Chironomidae Naididae Asellidae Hydroptilidae Simuliidae Limnephilidae Base 1133 LDI 0

October 700 40

600 35 30

500 g/L

25 µ 400 20 300

Taxa Taxa Count 15 200 10 chlorophyllɑ

100 5

0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year

Chironomidae Baetidae Simuliidae Hyalellidae Leptophlebiidae Brachycentridae Base 1133 LDI 0

120

Benthic macroinvertebrate family count vs base scenario and LDI 0 scenario DO concentration for site 1133 of MC during May and October from 2003 – 2013

May 400 120 350 100 300 80 250

200 60 mg/L 2 2

150 DO Taxa Taxa Count 40 100 20 50 0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year

Chironomidae Naididae Asellidae Hydroptilidae Simuliidae Limnephilidae Base 1133 LDI 0 Biological Threshold

October 700 200 180 600 160 500 140

400 120

100 mg/L 300 2

80 DO Taxa Taxa Count 200 60 40 100 20 0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year

Chironomidae Baetidae Simuliidae Hyalellidae Leptophlebiidae Brachycentridae Base 1133 LDI 0 Biological Threshold

121

Benthic macroinvertebrate family count vs base scenario and LDI 0 scenario sediment concentration for site 1133 of MC during May and October from 2003 – 2013

May 400 3.5

350 3 300 2.5 250 2 200 1.5

150 Taxa Taxa Count

1 sediment mg/L 100 50 0.5 0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year

Chironomidae Naididae Asellidae Hydroptilidae Simuliidae Limnephilidae Base 1133 LDI 0

October 700 3

600 2.5 500 2 400 1.5 300

Taxa Taxa Count 1 200 sediment mg/L

100 0.5

0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year

Chironomidae Baetidae Simuliidae Hyalellidae Leptophlebiidae Brachycentridae Base 1133 LDI 0

122