Citation
Ecohydrological Study of Watersheds within the Military Installation in Fort Benning, Georgia

Material Information

Title:
Ecohydrological Study of Watersheds within the Military Installation in Fort Benning, Georgia
Creator:
BHAT, SHIRISH ( Author, Primary )
Copyright Date:
2008

Subjects

Subjects / Keywords:
Watersheds ( jstor )
Streams ( jstor )
Water quality ( jstor )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Shirish Bhat. Permission granted to University of Florida to digitize and display this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Embargo Date:
5/1/2005
Resource Identifier:
436098663 ( OCLC )

Downloads

This item has the following downloads:

bhat_s ( .pdf )

bhat_s_Page_050.txt

bhat_s_Page_068.txt

bhat_s_Page_053.txt

bhat_s_Page_094.txt

bhat_s_Page_080.txt

bhat_s_Page_002.txt

bhat_s_Page_109.txt

bhat_s_Page_067.txt

bhat_s_Page_126.txt

bhat_s_Page_085.txt

bhat_s_Page_019.txt

bhat_s_Page_051.txt

bhat_s_Page_026.txt

bhat_s_Page_097.txt

bhat_s_Page_089.txt

bhat_s_Page_058.txt

bhat_s_Page_088.txt

bhat_s_Page_090.txt

bhat_s_Page_017.txt

bhat_s_Page_033.txt

bhat_s_Page_015.txt

bhat_s_Page_049.txt

bhat_s_Page_042.txt

bhat_s_Page_066.txt

bhat_s_Page_031.txt

bhat_s_Page_006.txt

bhat_s_Page_055.txt

bhat_s_Page_101.txt

bhat_s_Page_010.txt

bhat_s_Page_084.txt

bhat_s_Page_004.txt

bhat_s_Page_075.txt

bhat_s_Page_057.txt

bhat_s_Page_086.txt

bhat_s_Page_003.txt

bhat_s_Page_061.txt

bhat_s_Page_034.txt

bhat_s_Page_062.txt

bhat_s_Page_041.txt

bhat_s_Page_105.txt

bhat_s_Page_024.txt

bhat_s_Page_038.txt

bhat_s_Page_073.txt

bhat_s_Page_013.txt

bhat_s_Page_114.txt

bhat_s_Page_096.txt

bhat_s_Page_001.txt

bhat_s_Page_014.txt

bhat_s_Page_079.txt

bhat_s_Page_116.txt

bhat_s_Page_098.txt

bhat_s_Page_012.txt

bhat_s_Page_107.txt

bhat_s_Page_122.txt

bhat_s_Page_048.txt

bhat_s_Page_054.txt

bhat_s_Page_029.txt

bhat_s_Page_087.txt

bhat_s_Page_091.txt

bhat_s_Page_117.txt

bhat_s_Page_102.txt

bhat_s_Page_111.txt

bhat_s_Page_082.txt

bhat_s_Page_045.txt

bhat_s_Page_060.txt

bhat_s_Page_074.txt

bhat_s_Page_065.txt

bhat_s_Page_095.txt

bhat_s_Page_071.txt

bhat_s_Page_022.txt

bhat_s_Page_028.txt

bhat_s_Page_047.txt

bhat_s_Page_103.txt

bhat_s_Page_077.txt

bhat_s_Page_093.txt

bhat_s_Page_119.txt

bhat_s_Page_072.txt

bhat_s_Page_070.txt

bhat_s_Page_009.txt

bhat_s_Page_123.txt

bhat_s_pdf.txt

bhat_s_Page_056.txt

bhat_s_Page_076.txt

bhat_s_Page_016.txt

bhat_s_Page_081.txt

bhat_s_Page_118.txt

bhat_s_Page_120.txt

bhat_s_Page_011.txt

bhat_s_Page_108.txt

bhat_s_Page_127.txt

bhat_s_Page_023.txt

bhat_s_Page_115.txt

bhat_s_Page_059.txt

bhat_s_Page_113.txt

bhat_s_Page_064.txt

bhat_s_Page_110.txt

bhat_s_Page_052.txt

bhat_s_Page_035.txt

bhat_s_Page_125.txt

bhat_s_Page_027.txt

bhat_s_Page_092.txt

bhat_s_Page_046.txt

bhat_s_Page_025.txt

bhat_s_Page_078.txt

bhat_s_Page_124.txt

bhat_s_Page_100.txt

bhat_s_Page_104.txt

bhat_s_Page_063.txt

bhat_s_Page_121.txt

bhat_s_Page_030.txt

bhat_s_Page_099.txt

bhat_s_Page_043.txt

bhat_s_Page_037.txt

bhat_s_Page_036.txt

bhat_s_Page_083.txt

bhat_s_Page_069.txt

bhat_s_Page_007.txt

bhat_s_Page_044.txt

bhat_s_Page_106.txt

bhat_s_Page_032.txt

bhat_s_Page_039.txt

bhat_s_Page_018.txt

bhat_s_Page_005.txt

bhat_s_Page_040.txt

bhat_s_Page_112.txt

bhat_s_Page_020.txt

bhat_s_Page_021.txt

bhat_s_Page_008.txt


Full Text












ECOHYDROLOGICAL STUDY OF WATERSHEDS WITHIN THE MILITARY
INSTALLATION IN FORT BENNING, GEORGIA













By

SHIRISH BHAT


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


2005

































Copyright 2005

by

Shirish Bhat
































This work is dedicated to my parents, Tara and Narmada Bhat, and uncle Devendra Bhat.















ACKNOWLEDGMENTS

First of all, I would like to thank Dr. Jennifer Jacobs, for her constant guidance,

encouragement, patience and continuous support over the past five years. Her

enthusiasm for research and quest for excellence have left an everlasting impression in

my mind. To me, she has been more than an advisor, and this research would not have

been possible without her. Secondly, I would like to thank Dr. Kirk Hatfield for being on

my committee and offering me guidance in the later part of my research. I would also

like to thank Dr. Wendy Graham, Dr. Ramesh Shrestha, and Dr. Ramesh Reddy for being

on my committee and their invaluable suggestions throughout the research work. I would

like to thank Dr. Richard Lowrance and Randall Williams of USDA-ARS, Georgia,

whose contributions in this research have been tremendous. I deeply benefited from all

the long hours of fruitful discussions with them on a multitude of topics.

I would like to thank Hugh Westbury at Fort Benning, Georgia, for his coordination

efforts during the field trips, Dwight Dindial and Charles Campbell for their efforts in the

collection of the field data, and Phil Harmer for his contribution in the laboratory.

I wish to extend my gratitude to Lewis and Kim Bryant, and Donna Rowland, who

have been very helpful during my difficult days at UF. I also wish to extend my gratitude

to Subarna and Pramila Malakar for their care and support. I wish to extend my thanks to

Mr. Binod Palikhe for his help during my academic pursuit. I would like to extend my

thank to Debra Carol, Carol Hipsley, Sonja Lee, Doretha Ray, and all the Civil

Engineering staff for their help during all these years. I would also like to thank my









friends Deepak Singh, Shashi Shrestha, Hemant Belbase, Rishi Bhattarai, Dipendra Piya,

and Anand Bastola, and colleagues Nebiyu Tiruneh, Aniruddha Guha, Qing Sun, Mark

Newman, Ali Sedighi, Chris Brown, Brent Whitfield, Gerard Ripo, Haki Klammler and

others for their encouragements and moral support.

I would like to thank my parents and all the family members for their constant love

and encouragement. They have allowed me to pursue whatever I wanted in life. Without

their guidance and affection, it would have been impossible for me to advance my

education.
















TABLE OF CONTENTS

page


A C K N O W L E D G M E N T S ................................................................................................. iv

LIST OF TABLES ....................................................... ............ ....... ....... ix

LIST OF FIGURES ............................... ... ...... ... ................. .x

ABSTRACT .............. .......................................... xi

CHAPTER

1 GENERAL INTRODUCTION ...................................................... .....................

O bj ectiv es .................................................................. .................................. .5
D issertation O organization .............................................. ........ .............................

2 ECOLOGICAL INDICATORS IN FORESTED WATERSHEDS IN FORT
BENNING, GEORGIA: RELATIONSHIP BETWEEN LAND USE AND
STR EA M W A TER Q U A LITY ......................................................... .....................9

Introduction ...................................... .................................. ......... .... .. 9
Study A rea .................................. ............................ .... ...... ........ 12
M ethods of Study .................. .................................... ........... ............ .. 13
D description of W atersheds ................................................................... ............... 13
Characterization of Disturbance Categories.................................................. 14
Collection and Analysis of Stream Samples ....................................... .......... 15
Statistical A n aly ses........... ........................................................ .. ... .... ... ... 15
R results ................................................................................ ............... 16
Watershed Physical Characteristics.................... ..... ........................ 16
W ater Q quality Param eters......................................................... ............... 19
Effects of D disturbance Categories .................................................................. 21
Relationship between Watershed Physical Characteristics and Water Quality
P aram eters .............................................................................................. ........22
D discussion ....................................................................................................... ....... 29
Conclusion and Recommendations.................... ....... ........................... 33

3 HYDROLOGIC INDICES OF WATERSHED SCALE MILITARY IMPACTS IN
F O R T B E N N IN G G A ...................................................................... ...................35

In tro d u ctio n ...................................... ................................................ 3 5
vi









Flow Regim es and Hydrologic Indices.................................... ....................... 37
Effects of Flow Magnitude on Stream Ecology ................................................40
Effects of Flow Duration on Stream Ecology ............................................40
Effects of Flow Frequency on Stream Ecology................................................41
Effects of Flow Rate of Change on Stream Ecology .....................................42
D ata C collection ...................................................... ................. 42
W atershed Characteristics ............................................................................42
Precipitation and Streamflow Data ..................... .... ............... 45
M methods .............................. .....................................4 5
Annual-Based Hydrologic Indices ............................. ..... ...................... 45
Storm -Based Hydrologic Indices ............................................. ............... 48
Statistical A naly ses........... ........................................................ .. ... .... .... 48
R e su lts .......... ... ....... ........ ......... ........... ............................. 4 8
W atershed Disturbance Characteristics............... .............................................48
Annual-Based Hydrologic Indices ............................. ..... ...................... 50
Storm -B ased H ydrologic Indices ........................ .... ... ................. .... 51
Relationship between Military Land Management and Storm-Based
H ydrologic Indices ...... ....... .... ....... ...... ...... ......... .............. .. 52
Discussion .................. .......... ..................... .......... .. 54
Conclusion and Recommendations.................... ....... ........................... 58

4 PREDICTION OF NITROGEN LEACHING FROM FRESHLY FALLEN
LEAVES: APPLICATION OF RIPARIAN ECOSYSTEM MANAGEMENT
M O D E L (R E M M ) ............................................................................ ................ .. 59

In tro d u ctio n ........................................................................................................... 5 9
B asic Concepts of REM M ................................................. ........................... 62
Hydrology component of REMM .............. ......................... ....................63
Nutrient component of REM M ................................. ................ ............... 64
A application of R EM M ........................................................................... ............... 66
Study Area .................................... ................................ .........66
D ata and Input Param eters........................................................ ............... 68
M odel C alib ration .......... ..... .......................................................... .... .... ... ..7 1
Sensitivity A naly sis ............. ........................................................ ... .... .... 73
R e su lts ...........................................................................................7 3
Hydrology ...... ......... .. ........... ..................73
N itrog en ...................................................................................................74
S en sitiv ity A n aly sis ....................................................................................... 7 7
Discussion ....................... ................................ 80
Conclusion and Recommendations............................. .......... 83

5 SUMMARY AND CONCLUSION ............. ............................................ 85

APPENDIX

A ANNUAL-BASED INDICES DEFINITIONS AND CALCULATION
P R O C E D U R E S .................................................................................................... 8 8









B STORM-BASED INDICES DEFINITIONS AND CALCULATION
P R O C E D U R E S ................................................................................ ................ .. 9 5

C NITROGEN TRANSFORMATIONS IN REMM ...................................................98

D PARAMETERS AND RATE CONSTANTS USED IN REMM ............................99

L IST O F R E F E R E N C E S ...................................................................... ..................... 102

BIOGRAPHICAL SKETCH ........................................................................115
















LIST OF TABLES


Table page

2-1. Physical characteristics of study watersheds in Fort Benning, Georgia. ..............17

2-2. Pearson correlation coefficients between watershed characteristics. ....................18

2-3. t-test results for differences in mean values of watershed physical characteristics
and w ater quality param eters. .............................................................................21

2-4. Pearson correlation coefficients between watershed characteristics and water
quality param eters. ....................... ...................... .................... .. ..... 23

2-5. Stepwise multiple regression models for water quality parameters ........................24

3-1. Definitions of terms used to describe hyetograph and response hydrograph...........39

3-2. Physical characteristics of study watersheds. ...................................................44

3-3. Summary of the recommended hydrologic indices for 'perennial runoff type of
store a m s................................. ........................................................... ............... 4 6

3-4. Summary of hydrologic indices used in the Storm-Based Hydrologic Indices. ......49

3-5. Pearson correlation coefficients between watershed physical characteristics and
event based hydrologic indices.. ......................... .............................................. 53

3-6. Stepwise multiple regression models for event based hydrologic indices. ...........54

4-1. REMM model parameters values by the riparian zone for the Bonham-South
watershed in Fort Benning, Georgia. ............................................ ............... 67

4-2. Soil and vegetation nutrient pools for Bonham-South watershed's riparian area.....69

4-3. Model simulated annual hydrologic budget for the riparian areas in
B onham -South w atershed.. ............................. .... ...................................... 69

4-4. Sensitivity of modeled streamflow and TKN based on +/- 10% change in model
parameters for Bonham-South watershed. .................................... .................80
















LIST OF FIGURES


Figure p

2-1 Study w atersheds ................................................ .............. ..... 14

2-2 Box plots of water quality parameters ......................................... ...............20

2-3 Relationships between military land and water quality parameter. .......................25

2-4 Relationships between road density and water quality parameter .........................26

2-5 Relationships between number of roads crossing streams and water quality
param eter............................................................................................... 27

2-6 Relationships between disturbance index and water quality parameter .................28

3-1 Terms used to describe hyetograph and response hydrograph.............................39

3-2 Study w atersheds ................................................ .............. ..... 43

4 -1 S tu dy area .......................................................................................................... . 6 6

4-2. Comparison of REMM simulated daily flow with the observed daily flow for
the B onham -South w atershed........................................................ ............... 74

4-3. Observed canopy cover and TKN concentrations each month for the
B onham -South w atershed. ............................................. .............................. 75

4-4. Monthly simulated TKN concentration and simulated leaf mass in the Bonham-
South 's riparian area............. ...................................... ................ .. .... ..... .. 75

4-5. Monthly simulated TKN concentration and simulated nitrogen present in live
leaves the Bonham-South's riparian canopy. ........................... ................................ 76

4-6. Observed TKN concentrations during the events.....................................................78

4-7. Surface runoff comparison of the observed and simulated TKN masses..................79

4-8. Total streamflow comparison of the observed and simulated TKN masses..............79















Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

ECOHYDROLOGICAL STUDY OF WATERSHEDS WITHIN THE MILITARY
INSTALLATION IN FORT BENNING, GEORGIA

By

Shirish Bhat

May 2005

Chair: Kirk Hatfield
Cochair: Jennifer Jacobs
Major Department: Civil and Coastal Engineering

Relationships among watershed physical characteristics and water quality

parameters were explored for seven watersheds in Fort Benning, Georgia, using statistical

analyses to identify chemical indicators of ecological changes. Correlations were

identified among the indicators and watershed physical characteristics. Regression

results suggested that pH, chloride, total phosphorus, total Kjeldahl nitrogen, total

organic carbon, and total suspended solids are useful indicators of watershed physical

characteristics that are susceptible to perturbations.

The magnitude, frequency, duration, timing, and rate of change of hydrologic

conditions regulate ecological processes in aquatic ecosystems. Analysis of 26 non-

redundant hydrologic indices showed undisturbed watershed produced higher magnitude

and more frequent low- and high-flows as compared to the disturbed watershed.

Eighteen storm-based indices, grouped into four flow components, were proposed to

statistically characterize hydrologic variation among different watersheds. Results









showed that these storm-based indices might be used as surrogates to the indices derived

from long-term data. Statistical analysis showed that the watershed physical

characteristics such as military training land, road density, and the number of roads

crossing streams predicted hydrologic indices such as storm-based baseflow index,

bankfull discharge, response lag, and time of rise well.

Riparian vegetation has an important role in altering water quality in the forested

watersheds. The leaching of organic or mineral products on the forest floor provides

potential additional effects on water quality. In these areas, nutrients are released into the

fresh water systems due to the leaching and decomposition of vegetation litter. The

release of nutrients from plant litters prior to decomposition may be an important aspect

of characterizing stream water quality. To explore nitrogen leaching in a riparian area

during litterfall, Riparian Ecosystem Management Model developed by USDA-ARS has

been applied. The simulated total Kjeldahl nitrogen masses in the study watershed were

close to the observed values. The model effectively captured the trends of litter mass

accumulation in the riparian area and subsequent high concentrations and masses during

those periods.














CHAPTER 1
GENERAL INTRODUCTION

Ecohydrology is the interdisciplinary research field in which hydrology and

ecology come together. It may be defined as the study of the functional relations between

hydrology and biota at the watershed scale (Zalewski, 2000). Resource managers and

land use planners seek ecohydrological knowledge enabling them to design effective land

use plans and water management strategies.

Water, soil, and plant cover are fundamental components that determine the

productivity of the land. The successional stages in evolution of ecosystems depend on

climatic and hydrologic conditions and on nutrient availability. The unique composition

of plants and animals determines the ability to retain water and nutrients within the

system (Zalewski et al., 1997). The amount of water and its quality in the aquatic

environment are guided mostly by climate. Water and nutrient cycling in terrestrial

systems are tightly linked with each other and with human activity. The growth and

activity of human populations have increased the input of nutrients to terrestrial and

aquatic ecosystems (Schindler and Bayley, 1993; Vitousek et al., 1997). Various forms

of environmental perturbations affect quality and quantity of inland waters by increasing

runoff, erosion, sedimentation, and pollution.

Modifications of land use affect the water quality, residence time, surface runoff,

soil moisture, evaporation, and ground water. For example, an increase in urbanization

and non-forest land increased nitrogen concentration in the streams (Osborne and Wiley,

1988; Sponseller et al., 2001). Land-use practices can potentially affect and modify









channels along the flow paths in a landscape and may enhance or disrupt runoff

(Nakamura et al., 2000). Forest cutting can increase the frequency and volume of debris

slides due to dead and decayed tree roots that contribute to soil strength on marginally

stable sites (Sidle et al., 1985).

Water quality is highly variable, from place to place and from time to time, even

within a particular ecosystem. It is dependent on many factors, both natural and as a

consequence of human activities. In both cases, the quality of water is further affected by

the soils and vegetation over and through which it passes. Nutrients are released into the

freshwater systems due to the decomposition of the vegetation litter and modify the water

quality. Riparian vegetation has an important role in altering water quality as it functions

as a source or a sink of various organic and mineral components (Cirmo and McDonnell,

1997).

Agricultural practices and urbanization are widely recognized human impacts on

water resources. A specific type of human impact is caused at military installations.

Military training within a watershed can affect the drainage patterns, vegetation, and

soils. The impact is due to troop maneuvers and large tracked and wheeled vehicles that

traverse thousands of hectares in a single training exercise (Quist et al., 2003). The

impacts of such activities range from minor soil compaction and lodging of standing

vegetation to severe compaction and complete loss of vegetation cover in areas where

training is concentrated (Wilson, 1988; Milchunas et al., 1999). The resultant impacts are

evident both in the stream hydrology and the stream ecosystems. Potential impacts on

terrestrial and aquatic ecosystems include disruption in soil density and water content

(Helvey and Kochenderfer, 1990), addition of sediment, nutrients, and contaminants in









aquatic ecosystems (Gjessing et al., 1984), and impairment of natural habitat

development,and woody debris dynamics in forested floodplain streams (Piegay and

Landon, 1997). Military roads can significantly alter hillslope hydrology by

redistributing soil and rock materials on slopes and increasing the rate of debris slide

initiation. Roads also directly change the hydrology by intercepting shallow groundwater

flow paths, diverting the water along the roadway and routing it to stream at road

crossings. Road crossings can intercept groundwater drainage networks and collect

groundwater from upslope areas, diverting it into drainage ditches as surface water

(Wemple et al., 1996). Road crossings commonly found in military training areas may

act as barriers to the movement offish and other aquatic habitats (Fumiss et al., 1991).

Given the nature of military land use, management or military testing and training,

military land managers face the conflicting demands of balancing the primary military

mission with legal requirements to protect land and water quality (Milchunas et al.,

1999).

The military impacts can result in significant disruptions to the water and nutrient

cycles in the freshwater ecosystems and water resources. A few studies in the past have

addressed the effects of military training on terrestrial and aquatic ecosystems. For

example, Wilson (1988) found that the tank traffic resulted in a significant loss of native

species, increased abundance of introduced species, and increased bare soil at a training

site located in Manitoba. Milchunas et al. (1999) examined the effects of military

vehicles on plant communities and soil characteristics in Pinon Canyon Maneuver Site,

Colorado. Whitecotton et al. (2000) examined the impact of foot traffic from military

training on soil bulk density, infiltration rate, and aboveground biomass. Recently, Quist









et al. (2003) conducted a study in the Fort Riley Military Reservation, Kansas, to study

the effects of military use on terrestrial and aquatic communities.

Understanding human impacts in many landscapes needs the identification of

critical landscape elements and analysis of landscape pattern change (O'Neill et al.,

1997). Generalized response to military training includes the reduction in native and

perennial grasses, abundance of introduced species, and an increase in bare soil. Clearly,

these responses will alter biogeochemical and hydrological processes, which regulate

nutrient and water dynamics. However, impact of landscape changes to water quality and

stream is not well understood (Wand et al., 2001).

In order to maintain and improve water quality, there is an increasing need to

understand the relationships among watershed land use and stream ecosystems (Wang et

al., 2001). In parallel, perturbation induced effects on water quantity are critical to

stream ecology.

One of the many approaches to study the functional interrelations between the

hydrology and the stream biota is streamflow characterization and classification. This

approach develops hydrologic indices that account for characteristics of streamflow

variability that are biologically relevant (Olden and Poff, 2003). Characterization of

streams through development of ecologically relevant hydrologic indices is based on

long-term streamflow data. However, past studies overlooked the importance of storm-

flow data for the development of such hydrologic indices. Flow characteristics are

important where changes in land use are anticipated and where alterations to the flow

regimes need to be assessed.









The leaching of organic or mineral products from the forest floor provides potential

additional effects on water quality. Fluxes of nitrogen through the riparian zone are

intrinsically linked to water movement, both over and through the soil, and are also

strongly influenced by biological processes occurring in that zone. Nitrogen and organic

carbon dynamics in riparian zones are closely interrelated. While many of the factors

that can potentially influence nitrogen and carbon fluxes through riparian zones are

broadly known, there is presently incomplete quantitative information on the relative

importance of the flushing of nitrogen from freshly fallen leaves during precipitation

events.

Objectives

The United State's Department of Defense (DOD) policy has established

ecosystem management as its approach to manage the military lands by maintaining and

improving the sustainability and biological diversity of terrestrial and freshwater

ecosystems while supporting human needs, including the DOD missions. In order to

identify critical deficiencies and research opportunities on ecosystem management

problems on defense installations, the Strategic Environmental Research and

Development Program (SERDP) of DOD initiated the SERDP Ecosystem Management

Program (SEMP) in December 1997. The objectives of SEMP were to (1) establish long-

term research sites on DOD lands for military-relevant ecosystem research, (2) conduct

ecosystem research and monitoring activities relevant to DOD requirements and

opportunities, and (3) facilitate the integration of results and findings of research into

DOD ecosystem management practices.

The goal of this research is to illustrate how an ecohydrological approach could be

used to advance our ability to predict the effects of anthropogenic perturbations on water-









vegetation-nutrient interactions in the military installation at Fort Benning, Georgia. The

issues addressed in this research encompass many of the main scientific challenges in the

military installations' ecohydrology that include the effects of military related

perturbations on stream water quality and quantity, and the value of studying nutrient

dynamics in the riparian corridors of such regions. The specific objectives of this

research include (1) identification and examination of the statistical relationships among

water quality parameters and the watershed physical characteristics in low-nutrient

watersheds, (2) identification and development of hydrologic indices that characterize the

impact of military land management on watersheds, and (3) investigation of the effects of

nitrogen leaching from freshly fallen leaves on nutrient dynamics in a riparian area.

Dissertation Organization

Each chapter of this dissertation, except Chapters 1 and 5, is written as a self-

contained individual paper focusing on a topic that has not been addressed before.

Contributions are in the areas of water quality and land use within a military installation

(Chapter 2), development of storm-based hydrologic indices (Chapter 3), and watershed

scale nutrient leaching from a riparian area (Chapter 4). Chapter 5 summarizes and

concludes the research work

Chapter 2 is an original contribution in the effects of military activities related land

use on surface water quality. The major results from Chapter 2 are that the

concentrations of total organic carbon, total Kjeldahl nitrogen, total suspended solids, pH,

and total phosphorus in the stream show the greatest susceptibility to direct effects of

military activities. This chapter also identifies significant statistical relationships among

the water quality parameters and the military land uses. These relationships provide the









guidance for maintaining the surface water quality within the Fort Benning military

installation.

Chapter 3 is an original contribution in the ecohydrology that presents both annual-

based and storm-based methods for determining hydrologic indices that are of ecological

importance. Detailed descriptions of the methods for determining these indices and their

significance in aquatic ecosystems are described in this chapter. To statistically

characterize the hydrologic variation among different watersheds, 32 annual-based

hydrologic indices are analyzed. Eight out of 32 annual-based indices are recommended

to use for management practices within the Fort Benning military installation. As a new

approach to characterize the streamflow variability, 18 storm-based indices are proposed.

These indices are grouped into magnitude, frequency, duration, and rate of change of

flow. Storm-based indices are compared with annual-based indices. The storm-based

methodology provides guidance for measurements of appropriate indices within the

military installation.

Chapter 4 is an original contribution to the water quality function of the riparian

area. Nitrogen leaching from freshly fallen leaves in a riparian area during the

precipitation events is quantified. The observed nitrogen masses in the stream during the

precipitation events are compared with the model simulated values. Riparian Ecosystem

Management Model (REMM) is used to quantify the nitrogen in the riparian area. In this

chapter, details of the nitrogen leaching from the freshly fallen leaves are explored. The

results showed that the model effectively captured the trends of litter mass accumulation

in the riparian area and subsequent high concentrations during those periods. Analysis






8


showed that the simulated total Kjeldahl nitrogen masses during the precipitation events

were close to the observed masses.

Chapter 5 summarizes and concludes the research work. This chapter contains

recommendations to improve management of the water resources within the Fort Benning

military installation. Future research needs are also outlined in this chapter.














CHAPTER 2
ECOLOGICAL INDICATORS IN FORESTED WATERSHEDS IN FORT BENNING,
GEORGIA: RELATIONSHIP BETWEEN LAND USE AND STREAM WATER
QUALITY

Introduction

Ecological monitoring is essential to protect ecological health and integrity. As

human activity alters land cover, degradation of water resources begins in the upland

areas of a watershed. The first step toward effective ecological monitoring and

assessment is to realize that the ultimate goal is to measure and evaluate the

consequences of human actions on ecological systems. Human activities that alter land

use eventually affect biogeochemical processes that influence water quality and alter

ecological processes.

The National Research Council of the United States recently conducted a critical

evaluation of indicators used to monitor ecological changes from either natural or

anthropogenic causes. During recent decades, efforts have been increasing to develop

reliable and comprehensive environmental indicators because of growing environmental

concerns (National Research Council, 2000). Indicators rapidly and effectively

communicate system status. Ecological indicators help to elucidate both the effects of

human activities and natural processes. They can also help to assess future implications

of these factors on ecosystem integrity. Once indicators identify areas or elements of the

environment that are under stress, successful management of problems can be measured

relative to both interim targets and long-term goals.









Indicators that relate key ecological responses to human perturbations provide

useful tools to better understand ecological effects and their monitoring and management.

A suite of indicators ranging from microbiologic to landscape metrics is necessary to

capture the full spatial, temporal, and ecological complexity of impacts (Dale et al.,

2002). Evaluation of representative indices across major physical gradients (e.g., soils,

geology, land use, water quality and quantity) can signal early environment change and

help diagnose the cause of an environmental problem.

Understanding human impacts in many landscapes needs the identification of

critical landscape elements and analysis of landscape pattern change (O'Neill et al.,

1997). Attention has refocused on relationships among watershed characteristics and

stream water quality (Johnson et al., 1997). In order to maintain and improve water

quality, there is an increasing need to understand the relationships among watershed land

use and stream ecosystems (Wang et al., 2001).

Land use provides information about ecosystem function and characterizes the

extent and diversity of ecosystem types. National Research Council (2000) has

recommended land use as one of the most effective indicators for ecological assessment.

Hydrologists and aquatic ecologists have long known that the pathway by which water

reaches to a stream or lake has a major effect on water quality. Early studies on the

physical (Harrel and Dorris, 1968) and chemical (Hynes, 1960) characteristics of

watersheds focused on the influence of geomorphic characteristics such as drainage area,

gradient, and stream order on turbidity, dissolved oxygen concentration, and temperature.

Many recent studies examine the influence of terrestrial ecosystems on stream or wetland

water quality (Richards and Host, 1994; Richards et al., 1997). Many other studies have









found relationships between land use and concentrations of nutrients in streams (e.g.,

Hunsaker and Levine, 1995; Johnes et al., 1996; Bolstad and Swank, 1997). Watershed

properties constrain in-stream physicochemical and biotic features. Richards et al. (1996)

showed that ecosystems could be influenced by land use at regional or broad geographic

scales. Osborne and Wiley (1988) found that the distance of urban land cover from the

stream effectively predicts stream nitrogen and phosphorus concentrations.

Within a military installation context, land managers are challenged to use the land

for military training purposes in a manner that is both ecologically sound and meets

military mission requirements (Garten Jr et al., 2003). Lands can suffer a slow

degradation if over-utilized by long-term human activities. The heavy vehicles used in

mechanized military training cause disturbance of soil structure and can change the

physical properties of the soil (Iverson et al., 1981). In rangelands, tracked vehicle traffic

affects the hydrological characteristics (Thurow et al., 1993). Trampled vegetation,

vehicle tracks through undisturbed area, and erosion caused by the overuse of trails are

some examples of the visible degradation to a landscape caused by military training

exercises. Few studies have developed predictive relationships among watershed

physical characteristics and surface water chemistry specific to military land use and low-

nutrient systems.

In the coastal plain of the Apalachicola-Chattahoochee-Flint (ACF) river basin,

cropland and silvicultural land in upland areas is separated from streams by relatively

undisturbed riparian flood plain and wetland habitats (Frick et al., 1998). This is in

contrast to many intensively farmed areas of the United States where wetlands have been

drained, channelized or filled, and little or no riparian buffers remain between cropland









and streams. Frick et al. (1998) reported that the lower nutrient concentrations in streams

within the ACF river basin could partially be attributed to wetland buffer areas, and

minimal use of pesticides as compared to other areas of the United States. Other studies

in the southeastern coastal plain watersheds (e.g., Lowrance 1984; Lowrance et al., 1992;

Perry et al., 1999; Fisher et al., 2000) focused primarily on the agricultural impacts, and

urbanization on stream water quality. The contribution of areas affected by military

training to nutrient discharges, specifically in Fort Benning watersheds, is yet to be

quantified.

This paper identifies and examines the statistical relationships among water quality

parameters and the watershed physical characteristics in seven low-nutrient watersheds

located in the Fort Benning military installation, Georgia. It is hypothesized that surface

water quality parameters can be used as indicators of ecological changes in watersheds.

Study Area

The Fort Benning Army Installation occupies approximately 73,503 ha in

Chattahoochee, Muscogee, and Marion Counties of Georgia and Russell County of

Alabama (Figure 2-1). The climate at Fort Benning is humid and mild. Rainfall in this

region occurs regularly throughout the year. July and August are the warmest months

with average daily maximum and minimum temperatures of 37 and 15C. An average

daily maximum and minimum temperature of 15.5 and -1IC are reported in the coldest

months, January and February. Annual precipitation averages 1050 mm with October

being the driest month (Dale et al., 2002). Most of the precipitation occurs in the spring

and summer as a result of thunderstorms. Heavy rains are typical during the summer but

can occur in any month. Snow accounts for less than 1% of the annual precipitation.









Fort Benning is located within the southern Appalachian Piedmont and Coastal

Plains. The northern boundary of the installation lies along a transition zone between the

Piedmont and Upper Coastal Plain. The soils in the area are dominated by loamy sand

with some sandy loam. Following establishment of the installation in 1918, with

subsequent additions in 1941, we see that heavy training impacts only selected, mostly

upland, portions of the installation. Many areas are maintained as safety buffers, and

have little military use. Timber management includes harvesting and thinning. The

loblolly and longleaf pine forests are subjected to regular low-level fires for management

purposes (Dale et al., 2002).

Methods of Study

Description of Watersheds

The study watersheds, Bonham-1 and Bonham-2, Bonham, Little Pine Knot, Sally,

Oswichee, and Randall (named for the creek which drains the watershed), within Fort

Benning represent a range of region's soils, topography, land use, and vegetation

communities (Figure 2-1). These watersheds have a heterogeneous land cover

predominantly consisting of either forested or open areas. Forested areas are broadly

characterized as mixed pine and hardwoods or pine that are mostly 30-50 years old with

the soils in A-horizon range approximately 1-10 cm in depth (Garten Jr et al., 2003).

Open areas are either military, brush, or managed wildlife openings. Other cover

includes upland and bottomland hardwood forests. The military openings are clear-cut

parcels of land dominated by grass and bare soil that are used as military training

grounds. The brush openings consist of tall grass and immature hawthorn. The wildlife

openings are natural openings in the forests that are vegetated primarily by grass. Land










impacts due to heavy military activities (e.g., infantry, artillery, wheeled, and tracked

vehicle training) occur only in selected portions in these watersheds.


Figure 2-1. Study watersheds, Bonham-1, Bonham-2, Bonham, Little Pine Knot,
Randall, Sally, and Oswichee, in the Fort Benning military installation. Also
shown are stream network and sampling locations.

Characterization of Disturbance Categories

A disturbance index (DIN) was defined to characterize the watersheds at Fort

Benning into two disturbance categories: low-impact and high-impact. A DIN is the sum

of the area of bare ground on slopes greater than 3 degrees and on roads, as a proportion

of the total watershed area (Maloney et al., in press). The road areas were estimated by

multiplying their length by the measured average width of 20 m. Percentage of bare

ground was determined by using TM imagery and slope was derived from digital


&Sampihi, Locafflea
z fr efgFaulns

OM anm1

MI Lfttlo Me Knot
RaRndadt
O~wmchee
rFot oonuing


6 a 6 id 16 P3113MUN49









elevation maps. The TM imagery and digital elevation maps were obtained from

Strategic Environmental Research and Development Plan (SERDP)'s Ecosystem

Management Project (SEMP) database. Watersheds having a disturbance index from 0 to

11% are designated as low-impacted watersheds. High-impacted watersheds have

disturbance indices greater than or equal to 11%.

Collection and Analysis of Stream Samples

Surface water quality data were collected at seven streams biweekly from October

2001 to November 2002, and monthly thereafter to September 2003. Water samples

were collected in high-density polyethylene bottles. Bottles were soaked in de-ionized

water and rinsed with sample water prior to collection. The filtration was conducted at

the sampling sites using 0.45 |tm pore size polyethersulfone membranes. Filtered sample

was used to determine chloride (Cl) concentration, whereas raw sample was used for total

suspended solids (TSS) determination. Unfiltered samples for analyzing total Kjeldahl

nitrogen (TKN), total phosphorus (TP), and total organic carbon (TOC) were acidified

using double distilled sulfuric acid. The stream water pH, conductivity, and temperature

were measured at the time of sampling. All samples were kept cool in an icebox,

transported to the Soil and Water Science Department laboratory, University of Florida,

and refrigerated until analyzed. All samples were analyzed using standard methods

(American Public Health Association, 1992).

Statistical Analyses

The Pearson's correlation coefficients were calculated to examine the strength and

significance of the relationships between a watershed physical characteristic and a water

quality parameter. Two-sample t-tests were performed at 5% level of significance to test

whether mean values of watershed physical characteristics and water quality parameters









differ between low- and high-disturbance watersheds. Characteristics showing

significant correlations with a water quality parameter were considered for stepwise

multiple linear regression models. Only variables having less than or equal to 0.05

significance level were retained in the regression models.

Results

Watershed Physical Characteristics

The watersheds' physical characteristics are summarized in Table 2-1. Most of the

watersheds are highly vegetated (70% or more) except Oswichee (38%) with the majority

characterized by pine and mixed pine and hardwoods. Deciduous forest typically covers

only a small percentage of these watersheds. However, Bonham-1 consists of 27% of

deciduous forest. The study watersheds range from less than 1 to 84 km2. The

topographic characteristics of study watersheds are typical of forested watersheds of

southeastern coastal plain (Lowrance, 1992; Perry et al., 1999). Average elevations vary

from 104 to 148 m above mean sea level. Maximum slopes vary from 4 to 6 degrees.

Sandy soils are common in most of the study watersheds. However, loamy soils cover

most of the Sally and Oswichee watersheds. Bottomlands comprise 6 to 20% of the

watershed. The military training extent (0 to 6%) is relatively small. Total bare lands in

these watersheds comprise 9 to 21% of the watershed area, of which 1 to 8% of the total

area is unpaved roads and trails. This extent and variability of military training and bare

land are typical of the entire Fort Benning installation.

Some watershed characteristics are strongly correlated at significance level of 0.05

or lower (Table 2-2). Significant positive correlations exist for pine with the bottomland

wetlands, deciduous vegetation with stream density, number of roads crossing streams

with road length and percent of loam, and disturbance index with percent bare land.









Negative correlations were found for mixed vegetation with percentage of loamy soil and

number of roads crossing streams (NRC), military areas with normalized difference

vegetative index (NDVI) and stream density, and DIN, and road density with NRC.

Table 2-1. Physical characteristics of study watersheds in Fort Benning, Georgia.
Acronyms BON-1, BON-2, BON, LPK, OSW, Ran, and SAL represent
Bonham-1, Bonham-2, Bonham, Little Pine Knot, Randal, and Sally,
respectively.
Watersheds


Physical Characteristics
Topography
Area, km2
Average Elevation, m
Average Slope, degree
Vegetation
Pine, %
Deciduous, %
Mixed, %
Wetland, %
Military Land, %
NDVI
Soil
Sandy loam, %
Loamy sand, %
Road
Road Length, km
Road Density, km/km2
Stream
Stream Length, km
Stream Density, km/km2
Stream Order
Other
No. of Roads Crossing
Streams
Bare Land, %
Disturbance Index, %


BON-1 BON-2 BON


0.76
121.8
5.46


28
27
39
6
0
0.36


78
9


2.21
133.5
4.89


30
6
50
8
6
0.30


69
9


12.73
125.5
5.04


40
8
22
9
5
0.32


69
31


3.6 11.4 51.6
4.8 5.1 4.1


2.6
3.4
2



1
1
11


3.9
1.7
2



2
11
21


29.1
2.3
4



13
11
19


LPK OSW RAN SAL


18.01
146.3
5.32


41
2
34
17
2
0.34


72
28


56.6
3.1


43.3
2.4
4



11
4
11


83.39
104.2
4.48


26
0
5
7
5
0.35


24
73


196.6
2.4


170.6
2.1
5



55
4
9


74.38
136.8
4.57


58
3
9
20
3
0.34


68
32


415.1
3.1


323.5
2.4
6



43
4
10


25.31
136.8
5.42


48
12
15
10
2
0.36


49
51


97.6
3.8


65.2
2.6
4



21
7
15












Table 2-2. Pearson correlation coefficients between watershed characteristics. Characteristics are acronymed as follows: Pine forest
(PIN), Deciduous forest (DCD), Mixed forest (MXD), Wetland (WET), Military land (MIL), Sandy Soil (SND), Loamy
Soil (LOM), Road Length (RDL), Road Density (RDN), Stream Density (STD), Normalized Difference Vegetative Index
(NDVI), No. of Roads Crossing Streams (NRC), % Bare Land (PBL), and Disturbance Index (DIN). *, **, and ***
indicates significance at or below 0.05, 0.01, and 0.001 probability levels, respectively.
PIN DCD MXD WET MIL SND LOM RDL RDN STD NDVI NRC PBL
DCD -0.25
MXD -0.43 0.40
WET 0.96*** -0.33 -0.42
MIL -0.20 -0.65 0.00 -0.18
SND 0.21 0.44 0.67 0.17 -0.30
LOM 0.06 -0.50 -0.85* 0.14 0.15 -0.93***
RDL 0.63 -0.47 -0.74 0.50 0.05 -0.28 0.41
RDN -0.26 0.65 0.81* -0.34 -0.01 0.63 -0.81* -0.64
STD 0.01 0.84* 0.04 -0.04 -0.92*** 0.34 -0.24 -0.15 0.17
NDVI 0.10 0.41 -0.48 0.07 -0.80* -0.30 0.41 0.22 -0.39 0.74
NRC 0.23 -0.58 -0.90** 0.19 0.21 -0.76* 0.83* 0.81* -0.86** -0.27 0.33
PBL 0.00 -0.28 0.25 0.09 0.72 0.04 -0.11 -0.34 0.40 -0.67 -0.79* -0.30
DIN -0.10 0.05 0.52 -0.07 0.55 0.30 -0.42 -0.52 0.71 -0.44 -0.77* -0.58* 0.93***









Water Quality Parameters

Water quality parameters in the study watersheds varied over the sampling period

and among watersheds. Variability of water quality parameters among watersheds is

observed (Figure 2-2). Mean pH in the study watersheds ranged from 4.2 to 7.0. Mean

conductivity ranged from 16.4 to 44.5 [[S/cm. Mean temperatures varied from 17.5 to

20.8C. Low concentrations of TP and TKN were observed in all the watersheds under

study as compared to forested watersheds in the southeastern coastal plain watersheds

(Lowrance et al., 1984), and across the United States (Meader and Goldstein, 2003;

Fisher et al., 2000). Concentrations of TKN, TP, and Cl were often below the detection

limit. Mean concentrations of TP varied widely, ranging from 0.003 to 0.020 mg/L and

TKN varied from 0.20 to 0.35 mg/L; TOC from 1.35 to 3.33 mg/L; Cl from 1.46 to 4.13

mg/L; and TSS from 4.15 to 10.30 mg/L. As depicted in Figure 2-2, each stream

exhibited distinct water quality signatures with the exception of temperature and TSS.

Seasonal variations in the water quality parameters are responsible for much of this

observed variability among watersheds.

Stream pH fluctuated more during June and July and was elevated during

December through February. Conductivity values showed slight fluctuations from May

to July. On multiple occasions, high conductivity values were observed in the Randall

stream. Chloride, TKN, and TP showed distinct seasonal patterns. The concentrations of

these parameters were low from June to September, and high from March to May and

from October to December. In contrast, TOC peaked from August to October and again

from March to July. Higher concentrations of TSS were observed from July to

September in all the streams.














10 -
9-



S6-
a

4-
3-
2-

90-






so -
S70 -




10 -

0.107






0.0B2

0.00-


g4


2

1-


1 2 h 4 15
Wasehed ID


6 7


35-


20-

15
10-
E-' 5


1.0



gS
405
H


00_


10 -





C
u





70
70 -
60-

S30-
30 -
20 -
10 -
0..


1 2 3 W 5M
Watershed I0


Figure 2-2. Box plots of water quality parameters. Each plot consists of outliers, most
extreme data, 75th, 50th, and 25th percentile values. Watershed IDs
represent- 1: Bonham-1, 2: Randall, 3: Oswichee, 4: Little Pine Knot, 5: Sally,
6: Bonham, and 7: Bonham-2.


T yT


*


4*a
*



N 4






: ^ l ll


6 7


I I I I I I I










$~$-~=B










f d e~B,









Et3 6Q









Effects of Disturbance Categories

Table 2-3 presents the t-test results of the comparison of physical characteristics

and water quality parameters based on the watershed disturbance level. The only

physical characteristic having a significant difference (c = 0.04) between these two

groups was DIN. While the low-impact watersheds tended to have higher chemical

Table 2-3. t-test results for differences in mean values of watershed physical
characteristics and water quality parameters. indicates significance at or
below 0.05 probability level. NS indicates non-significant difference at the
0.05 probability level.
Low-Impacted High-Impacted
Mean SD Mean SD
Watershed Characteristics
Pine, % 38.3 14.8 39.3 9.0 NS
Deciduous, % 8.0 12.7 8.7 3.1 NS
Mixed, % 21.8 17.2 29 18.5 NS
Wetland, % 14.5 9.5 16 7.2 NS
Military Land, % 2.4 2.0 4.3 2.3 NS
Sand, % 60.5 24.7 62.3 11.5 NS
Loam, % 35.5 26.9 30.3 21.0 NS
NDVI 0.35 0.01 0.32 0.03 NS
Road Length, km 168 184 53.5 43.2 NS
Road Density, km/km2 3.3 1.0 4.3 0.7 NS
Stream Density, km/km2 2.6 0.6 2.2 0.4 NS
No. of Roads Crossing Streams 27.5 25.6 12 9.5 NS
Disturbance Index, % 10.2 0.9 18.3 3.1 *
Water Quality Parameters
pH 5.6 1.3 4.5 0.3 NS
Temperature, C 19.3 0.9 18 0.3 NS
Conductivity, [iS/cm 25.9 13.3 20.4 2.5 NS
TKN, mg/L 0.3 0.03 0.2 0.05 NS
TP, mg/L 0.011 0.006 0.007 0.003 NS
TOC, mg/L 2.9 0.4 2.1 0.7 NS
Cl, mg/L 2.4 0.5 1.8 0.3 NS
TSS, mg/L 9.1 2.2 4.8 0.5 *









concentrations than the high-impact watersheds, only TSS showed significant difference

(a = 0.03). Even though the results showed no significant statistical differences at a

confidence level of 95%, the t-test results of all water quality parameters, except

conductivity and TP, showed significant differences at 80% confidence interval between

high- and low-impacted watersheds. The relatively small sample size and natural

variability among watersheds may have limited the ability to discern significance

differences.

Relationship between Watershed Physical Characteristics and Water Quality
Parameters

Correlation and regression analyses were performed to identify relationships among

the watershed physical characteristics and the water quality parameters. Table 2-4 shows

that each water quality parameter had a significant relationship with one or more

watershed physical characteristics (Table 2-4). The correlation results show that

decreasing mixed vegetation increased pH and TP. Sandy and loamy soils had opposite

effects on TP. An increase in sandy soil decreased TP, whereas an increase in loamy soil

increased TP. Increasing military land decreased TOC. Temperature, pH, conductivity,

and Cl increased as the road length increased. The number of roads crossing streams had

positive correlations on pH and TP. Percent bare land was negatively correlated with

TOC and TSS. Disturbance index was negatively correlated with TKN and TOC.

Graphical relationships provide insight into nonlinear relationships that exist

between indicators and response variables. Figures 2-3 to 2-6 show some of the most

striking relationships between watershed characteristics directly and/or indirectly affected

by management of military lands and their effect on water chemistry. As the % military

land increases, the TKN, TOC, and TSS decrease in a linear fashion. However,









Table 2-4. Pearson correlation coefficients between watershed characteristics and water
quality parameters. Characteristics are acronymed as follows: Pine forest
(PIN), Deciduous forest (DCD), Mixed forest (MXD), Wetland (WET),
Military land (MIL), Sandy Soil (SND), Loamy Soil (LOM), Road Length
(RDL), Road Density (RDN), Stream Density (STD), Normalized Difference
Vegetative Index (NDVI), No. of Roads Crossing Streams (NRC), % Bare
Land (PBL), and Disturbance Index (DIN). *, **, and *** indicates
significance at or below 0.05, 0.01, and 0.001 probability levels, respectively.
pH Temperature Conductivity TKN TP TOC Cl TSS
PIN 0.36 0.78 0.54 0 -0.09 0.15 0.56 -0.14
DCD -0.47 -0.37 -0.40 -0.12 -0.33 0.39 0.13 0.42
MXD -0.79* -0.55 -0.38 -0.39 -0.83* -0.24 -0.49 -0.11
WET 0.24 0.62 0.44 0.11 -0.11 0.18 0.37 -0.29
MIL 0.09 0 -0.10 -0.50 0.10 -0.87** -0.47 -0.53
NDVI 0.32 0.11 0.10 0.60 0.46 0.80* 0.55 0.54
PBL -0.45 -0.25 -0.51 -0.72 -0.42 -0.81* -0.68 -0.87**
DIN -0.65 -0.36 -0.60 -0.84* -0.63 -0.76* -0.65 -0.72
SND -0.50 -0.02 0.14 -0.19 -0.78* 0.15 0.06 0.15
LOM 0.57 0.16 0.03 0.39 0.81* 0.07 0.07 -0.14
RDL 0.94*** 0.95*** 0.82* 0.19 0.57 0.09 0.79* 0.30
RDN -0.75* -0.42 -0.54 -0.72 -0.78* -0.35 -0.32 -0.15
NRC 0.93*** 0.59 0.51 0.36 0.90** 0.06 0.47 0.19
STD -0.14 -0.11 -0.01 0.39 -0.04 0.82* 0.44 0.65

disturbance index may operate as a threshold indicator of pH and TSS where pH

decreases in response to a relatively low level DIN while the TSS threshold for DIN

impact is somewhat higher. No significant relationships are found between the water

quality parameters and extent of pine and deciduous forest. Similarly, wetland showed

no effect on these parameters.

Stepwise multiple regressions identified relationships between the water quality

parameters and watershed physical characteristics that are susceptible to the disturbances

(Table 2-5). A statistically significant regression model was found for every water

quality parameter. Prediction of pH variability among watersheds is particularly well

captured by pine forest and road length. The regression relationships indicate that all of









the water quality parameters depend on at least one aspect of military management.

Several water quality parameters, Cl, TP, TOC, TSS, and TKN, depend only on

management aspects of the military installation. For example, Cl depends on change in

military land and road length. Total phosphorus is strongly related to the number of

roads crossing streams. However, the influence of vegetation and soils characteristics is

clearly important in pH, conductivity, and temperature. For example, conductivity

appeared to be well captured by area covered by sandy soil and the number of roads


Table 2-5. Stepwise multiple regression models for water quality parameters. pH is
unitless, temperature is measured in degrees centigrade, conductivity is
measured in mS/cm, TP, TKN, TOC, Cl, and TSS are measured in mg/L.
Pine forest (PIN), Military land (MIL), Sandy Soil (SND), Loamy Soil
(LOM), Road Length (RDL), Road Density (RDN), No. of Roads Crossing
Streams (NRC), Percent Bare Land (PBL), and Disturbance Index (DIN) are
the independent variables retained in the regression analyses. *, **, and ***
indicates significance at or below 0.05, 0.01, and 0.001 probability levels,
respectively.
Water Quality Independent Variables Retained and
Parameters Regression Equations R2


pH

Cl


TP


Conductivity


Temperature

TOC


TSS


Pine, Road Length
5.50 0.0382 PIN + 0.00924 RDL
Military, Road Length
2.19 + 0.145 MIL + 0.00344 RDL
No. of Roads Crossing Streams
0.00451 + 0.000236 NRS
Sandy Soil, No. of Roads Crossing Streams
- 24.2 + 0.552 SND + 0.666 NRS
Loamy Soil, Road Density
26.1 0.051 LOM 1.49 RDN
Military Land
3.45 0.274 MIL
Percent of Bare Land
11.2 0.648 PBL
Disturbance Index
0.445 0.0109 DIN


0.98***


0.90**


0.81**


0.80*


0.77*

0.76**


0.76**


TKN


0.70*
















R = 0.00
T


R2 = 4E-06


-S





-R = 0.01






I


R =0.22





I I


R = 0.01


R2 = 0.76


2
0


60

- 40

S20

0
C


0.025
0.020
0.015
0.010
0.005
0.000


Military (%)


Figure 2-3. Relationships between military land and water quality parameter. Vertical
bars represent standard errors.


I I


-R = 0.27




I I
m--j


Military (%)
















R = 0.55


8
6
4

2
0


60

40

S20
S0
C
i o


0.025
0.020
0.015
0.010
0.005
0.000


R2 = 0.29


RO = 0.51





-I I


R2 = 0.11



C1




R = 0.02




I I

0 2 4 6

Road Density (km/km2)


Figure 2-4. Relationships between road density and water quality parameter. Vertical
bars represent standard errors.


- I I I


R2 = 0.17




-2
-^





SR = 0.61








R =0.10


tt I




0 2 4 6
Road Density ( )
Road Density (kmn/km )


-




-
















8 -
6-
4-
2-
0


60 _
S40 -
20 -
0

C0

0.030 -
S0.020 -
g
0.010
0.000


4-

S2-

S-
0


R = 0.86


R = 0.35


R = 0.81








R = 0.22



S30 -
5 20
10
0-


S0.6

0.4

0.2

0.0



6


2
0


20 -
S15-
g 10 -
S5-
0


0 20 40 60


Number of Roads
Crossing Streams


R = 0.25


R2 = 0.12


2 0.00
- R = 0.00


I-----


R2 = 0.03






0 20 40 60


Number of Roads
Crossing Streams


Figure 2-5. Relationships between number of roads crossing streams and water quality
parameter. Vertical bars represent standard errors.


-


B














R2 = 0.42
if


R = 0.13




I I


R2 = 0.40


R2 = 0.42


R = 0.36
4-7


25
20 -
15 -
10 -
5
0-

0.5
0.4
0.3
0.2 -
0.1
0.0


R = 0.57


F~k^


0 10 20 30

Disturbance Index (%)


0 10 20 30

Disturbance Index (%)


Figure 2-6. Relationships between disturbance index and water quality parameter.
Vertical bars represent standard errors.


S= 0.69


60

S40

20

0
u


0.025
0.020
0.015 -
0.010 -
0.005
0.000


-4


R2 = 0.52


20 -
15
10
5
0


+


.









crossing streams, whereas temperature was captured by loamy sand soil and road density.

Overall, the regression models show that it is possible to quantify the effects of watershed

physical characteristics on the water quality parameters.

Discussion

Variations in stream water chemistry among the study watersheds reflect

differences in biogeochemical reactions occurring in the watersheds. The results of this

study indicate that even in low impact watersheds, physical characteristics may be used to

explain variations in stream water chemistry and, by inference, the relative watershed

disturbance levels. The observed variability in many of the chemical parameters studied

in these watersheds can be attributed to physical characteristics of the watersheds or land

management patterns within the watersheds as evidenced by the road network, forestry

practices, and military training.

Diverse human activities interact to affect conditions in watersheds and water

bodies. Sites of interest can be grouped and placed on a gradient according to activities

and their effects. The results of this study suggest that the vegetation type, road length,

number of roads crossing streams, and disturbance index are important predictors of

water quality variability. Vegetation cover was related to stream pH and TP (mixed

forest) and conductivity (pine forest). However, deciduous forest cover was not related

to any of the water quality parameters suggesting a limited effect in organic matter and

nutrient production and variability as observed among Fort Benning watersheds.

However some other subtle landscape changes resulted in relatively larger impacts on

water quality parameters. Low-impact watersheds tend to produce higher concentrations

of nutrients in the streams. This can be attributed to the availability of more soil organic









matter and the rapid biogeochemical processes occurring in the low-impact watersheds as

compared to the high-impact watersheds.

It is extremely difficult to capture all aspects of human influence in a single graph

or statistical test. However, sometimes meaningful chemical patterns can be lost by

excessive dependence on the outcome of menu-driven statistical tests (Karr and Chu,

1999). Figures 2-3 to 2-6 depict several different aspects of stream's chemical conditions

against several measures of human influences, such as military land use, disturbance

index, number of roads crossing streams, and road density. The distribution of circles in

most of these figures illustrate that a chemical metric indicates little about a condition

simply because it does not correlate strongly with a single surrogate of that condition.

However, where the relationship between human influence and stream's chemical

response is strong, statistics and graph agree.

The correlation tests identify linear relationships between a chemical response and

watershed characteristics. Weak statistical correlations observed in these analyses may

have missed important chemical patterns. For example, nonlinear patterns were observed

for forest types, bare land (not shown here), and disturbance index (Figure 2-6). The

plots in Figure 2-6 show a step-function for TSS and pH. The scatter of this dataset

shows little or no statistical significance, but can be interpreted chemically. For TSS,

those watersheds having a disturbance index of 11% or lower had a higher level than

those with a greater disturbance index.

When a number of variables interact to influence water quality conditions, it may

be difficult to explain observed variability in a single plot against one dimension of

human influence (Figure 2-4). Chemical responses were plotted against the road









densities for various watersheds. The Pearson correlation coefficient for TP was

significant capturing human influence on this chemical parameter. The response of TP is

visibly distinguished from others. A similar discussion is true for military land with TOC

(Figure 2-3); number of roads crossing streams with pH and TP (Figure 2-5); and

disturbance index with TKN (Figure 2-6).

The relationships between water quality parameters and physical characteristics

indicate that disturbances in low nutrient forested environments decrease some chemical

signatures. Watersheds with more roads, e.g., Randall and Oswichee, have relatively

high pH, conductivity, and Cl compared to the watersheds with fewer roads. Watersheds

with a small portion of military land, e.g., Bonham-1, Sally, and Little Pine Knot, have

relatively high TOC concentrations. In contrast, watersheds characterized by higher road

densities, e.g., Bonham and Bonham-2, had low TP concentrations. Higher disturbance

index, similar to the road density, showed lower TKN and TOC concentrations in the

streams. Mixed vegetation, road length, percent of bare land, DIN, and number of roads

crossing streams were able to capture most of the variability in water quality parameters.

In a watershed scale study conducted in Ontario, Canada, Sliva and Williams

(2001) found a negative correlation of forested land cover with TSS and chloride. In

contrast, Johnson et al. (1997) showed a positive relationship of forest with TSS in a

study of landscape influence on water chemistry in the Saginaw Bay watershed of central

Michigan. Johnson et al.'s (1997) results indicated that row crop agriculture had the

highest effect on total nitrogen, nitrate, and total dissolved solids. They also observed

that urban and forest areas were positively correlated good predictors of TSS, whereas

row crop agriculture was positively correlated with total nitrogen. Basnyat et al. (1999)









reported a positive association of TSS with agricultural practices in the Fish River

watershed, Alabama. The Fort Benning installation is characterized by relatively low

variability in forest cover and suggests, in contrast to other studies, that neither TSS nor

Cl may be related to forest cover under existing land management practices. Instead,

Fort Benning's road extent and percent bare land are better predictors of TSS and Cl.

Most studies identified urban land use as a dominant factor causing elevated total

nitrogen and nitrate concentrations in the streams (Hill, 1981; Osborne and Wiley, 1988).

Sponseller et al. (2001) found positive correlation of total inorganic nitrogen with

percentage of non-forested land in southwestern Virginia watersheds. A negative

correlation of TKN to DIN, in this study, is consistent with studies (e.g., Sponseller et al.,

2001; Hunsaker and Levine, 1995; Johnson et al., 1997) that have shown stream nitrogen

concentrations to be good predictors of non-forest area at the watershed scale.

In a study of 101 watersheds in New Zealand, Close and Davis-Colley (1990)

found that between 60 and 80% of the variance in conductivity, total nitrogen, and nitrate

was accounted for by landscape factors including geology and land use. However, in that

same study, landscape factors accounted for only 50% of the variance in ammonia and

phosphorus species. Their results parallel those found at Fort Benning in that no strong

relationships were observed between land use and TP. The strong negative correlation of

loamy sand soil with TP in Fort Benning is consistent with Hill's (1981) study, conducted

in a sandy loam region similar to portions of Fort Benning watersheds that reported

negative correlations between phosphorous concentrations and abandoned farmland and

forest.









Most variations in stream water chemistry are driven by climatic and biotic factors

and are therefore largely governed by the processes that are taking place in the terrestrial

part of the watershed such as natural or human induced vegetation cover changes

(Semkin et al., 1994). Our results show interactions among landscape factors and water

quality indicators. Results also indicate that it is possible to observe the response of these

water quality parameters to physical attributes of watersheds. The importance of water

quality parameters in the present study appeared to be attributable to the perturbations

related to military training and associated parameters within a watershed as these

parameters clearly captured the changes in physical parameters that are more sensitive to

such kind of influences.

Conclusion and Recommendations

Sometimes a single variable can capture and integrate multiple sources of

influence. More often, a small number of ecological attributes provide reliable signals

about ecological condition. Water chemistry prediction using watershed physical

characteristics in this study showed mixed results compared to the other investigations.

However, most of the watershed physical characteristics used in our analysis did explain

the variability in water quality parameters. This study documented strong relationships

between certain watershed physical characteristics that are more susceptible to human

induced perturbations, specifically military related disturbances, and water quality

parameters in military installation at Fort Benning. Watersheds with more roads crossing

streams tended to produce more TP. Total Kjeldahl nitrogen and TOC variations were

well captured by DIN and extent of military land, respectively. Total suspended solids'

variability, on the other hand, was captured by the percent of bare land within a

watershed. Road length captured most of the variability in pH and Cl. Conductivity and









temperature values were dependent on soil types and road characteristics. The variations

in stream water chemistry are largely attributable to disturbance levels and the types of

biogeochemical reactions occurring in the watersheds. Regression results suggest that

TOC, TKN, and TSS were useful indicators of watershed physical characteristics as they

are more susceptible to direct effects of military activities. Although pH, conductivity,

and TP showed good correlations with the road length, these parameters indicated strong

but indirect influence of military training activities on watersheds.

Foreseeing a single indicator of water quality that would be sensitive to all kinds of

perturbations in the watershed is extremely difficult. Ability to detect perturbations can

be related to spatial and temporal scales. It is necessary to recognize the effects of

natural disturbances on ecosystem structure and functioning. It is suggested that

priorities for determining ecological indicators specific to water quality should include

(1) development of framework to determine proper reference states within watershed

against which to detect loss of ecosystem health, (2) broadening our knowledge of

ecosystem sensitivity to perturbations of varying intensity, spatio-temporal distribution,

and type, and (3) development of suites of indicators necessary to detect the broadest

spectrum of perturbations in watersheds.














CHAPTER 3
HYDROLOGIC INDICES OF WATERSHED SCALE MILITARY IMPACTS IN
FORT BENNING, GA

Introduction

A goal of stream flow characterization and classification is to develop hydrologic

indices that account for characteristics of streamflow variability that are biologically

relevant (Olden and Poff, 2003). Broadly, indices are attributes that respond in a known

way to a disturbance i.e., they relate key ecological responses to human activities. Index

identification is based on the goals and objectives set for a particular ecosystem or region.

A good index should be sensitive to stressors, biologically and socially relevant, broadly

applicable to many stressors and sites, diagnostic of the particular stressor causing the

problem, measurable, interpretable, and not redundant with other measured indices

(Cairns et al., 1993).

Ecologically relevant hydrologic indices developed in the past not only characterize

particular regions, but also quantify flow characteristics that are sensitive to various

forms of human perturbations. For example, early studies on hydrological indicators

focused on variation of mean daily flow to study the pattern of fish in Illinois and

Missouri (Horwitz, 1978). In Great Britain, Moss et al. (1987) used average flow

conditions to predict macro-invertibrate fauna of unpolluted streams and Townsend et al.

(1987) examined persistence of community structure for benthic invertebrates. In arid

regions of southwest United States, Minckley and Meffe (1987) studied effects of short-

term flood frequency in stream fish communities. Poff and Ward (1989) used long-term









discharge records (17-81 years) of 78 streams from across the continental United States

to develop a general quantitative characterization of streamflow variability. Similarly,

Jowett and Duncan (1990) studied skewness in flows and peak discharges in relation to

in-stream habitat and biota in New Zealand.

More recent investigations have begun to focus on examining suites of hydrologic

indices that are ecologically relevant to quantify hydrologic regimes. These studies

report numerous such indices. For example, Poff and Allan (1995) studied stream fish

assemblage for 34 sites in Wisconsin and Minnesota in conjunction with long-term

stream flow variability and predictability as well as frequency and predictability of high

and low flow extremes. In the process of deriving ecologically relevant hydrologic

indices, Clausen and Biggs (1997) identified thirty-four hydrological variables from daily

flow records at eighty-three New Zealand sites. The authors related these variables to

benthic biota including periphyton and invertibrate species richness and diversity. Wood

et al. (2000) reported the importance of hydrological conditions in explaining the

ecological role when the authors studied the changes in macro-invertibrate community in

response to flow variations in the Little Stour River in the United Kingdom. Pettit et al.

(2001) described a method for assessing seasonality and variability of natural flows and

their influence on riparian vegetation in two contrasting river systems in western

Australia.

To isolate core flow variables for ecological studies, it is important to know not just

the ecological relevance of the variables, but also the interrelationships among the

variables in order to avoid redundancy in the analyses (e.g., Clausen and Biggs, 2000;

Olden and Poff, 2003). Hydrologic indices have been criticized for being overly









simplified and lacking adequate biological relevance. Stream ecologists are now facing

difficulty in choosing appropriate and relevant ones from the available suit of indices.

For example, the Indicators of Hydrologic Alteration (Richter et al., 1996) approach is

commonly used for characterizing human modification of flow regimes, yet it contains 64

statistics (32 measures of central tendency and 32 measures of dispersion), many of

which are inter-correlated (Olden and Poff, 2003).

To date, characterization of streams or regions through determination and

development of ecologically relevant hydrologic indices are based on long-term stream

flow data. However, given the multitude of methods to characterize stream flow, past

studies overlooked the value of storm flow data for the development of such ecologically

relevant hydrologic indices. Flow characteristics are especially important where changes

in land use are anticipated and where alterations to the flow regime need to be assessed.

The primary objective of the study is to identify hydrologic indices that

characterize the impact of military land management on watersheds in Fort Benning,

Georgia. Towards this end, this study investigates both storm and annual hydrographs. it

is hypothesized that, in addition to annual-based indices, storm-based hydrologic indices

are indicative of alteration in stream ecology. Here, a suite of event based hydrologic

indices is proposed. Storm-based and the annual-based indices are calculated and used to

compare and contrast impacted watersheds with a reference watershed. Additionally,

specific military land management practices are used to predict storm-based indices.

Flow Regimes and Hydrologic Indices

To assess hydrologic alterations within an ecosystem, Richter et al. (1996)

developed a method to compute representative, multi-parameter suite of hydrologic

characteristics that are of ecological relevance, commonly known as Indicators of









Hydrologic Alteration (IHA). Olden and Poff (2003) comprehensively reviewed

currently available hydrologic indices for characterizing streamflow regimes and

recommended non-redundant indices for various stream types that may differ in major

aspects of ecological organization. Poff (1996) provides a comprehensive catalog of the

stream types for small to mid-size relatively undisturbed streams, classified according to

variation in ecologically relevant hydrological characteristics, in continental United

States. The assessment of IHA as well as other studies (e.g., Poff and Allan, 1995;

Clausen and Biggs, 1997; Wood et al., 2000; Pettit et al., 2001; Olden and Poff, 2003) to

identify hydrologic indices is based on long-term flow data.

An alternative approach to identify ecologically relevant hydrologic indices is to

conduct an assessment based on the storm hydrograph. This approach is useful when

long-term data for a particular stream or region are not available, when significant data

gaps exist, or coincident records are not available. Storm hydrographs are traditionally

described by characteristics including peak flow, total volume of direct runoff, and

duration as shown in Figure 3-1. The time characteristics of the hydrograph and its

relationship to the precipitation event are presented in Table 3-1. Towards the goal of

characterizing the spatial variations of hydrologic conditions using storm-based indices

that are ecologically relevant as well as sensitive to human influences, the ecological

function of hydrologic characteristics that are relevant to storm-based hydrologic indices

are considered.

Examination of the storm hydrographs reveals numerous potential indices. A set of

18 storm-based ecologically relevant hydrologic indices that characterize variation in

water condition in individual watershed is proposed. Included proposed indices are














Precipitation
[Li



















Figure 3-1. Terms used to describe hyetograph and response hydrograph. Refer to Table
3-1 for the definitions of the terms.


Table 3-1. Definitions of terms used to describe hyetograph and response hydrograph.
Time instants Time durations











tpo= beginning of precipitation Tw = te tpo= duration of water input
tpc = centroid of precipitation TrI= tqO tpo= response lag
tpe = end of precipitation T tpkr- tqO time of rise

tqO= beginning of hydrograph rise T1p tpk tpO = lag-to-peak
tpk = time of peak discharge Tic tqc tpc = centroid lag
tq = centroid of response hydrograph Tb tqe qO time base
t = end of response hydrograph T= te tp = time of concentration


response factor (ratio of direct runoff depth to precipitation depth), baseflow index (ratio

of baseflow volume to total volume during an event), peak discharge, dimensionless

fife, fGt tsc A












Figure 3-1. Termsed to describe hyetograph and response hydrolag (T), time of rise (Tr), and time base (Tb),
3-1bankfull dischargfor the and the slopes of rising and falling limb of the hydrographs. A flowterms.
Table 3-1. Definitions of terms used to describe hyetograph and response hydrograph.
Time instants Time durations
tpo = beginning of precipitation T, = tp, tpo = duration of water input
tp, = centroid of precipitation Tri = tqo tpo = response lag
tpe = end of precipitation Tr = tpk tqo = time of rise
tqo = beginning of hydrograph rise TIP = tpk tpo = lag-to-peak
tpk = time of peak discharge T1c = tq, tpc = centroid lag
tq, = centroid of response hydrograph Tb = tq, tqo = time base
tq, = end of response hydrograph T, = tq, tpe = time of concentration

response factor (ratio of direct runoff depth to precipitation depth), baseflow index (ratio

ofbaseflow volume to total volume during an event), peak discharge, dimensionless

numbers related to hydrograph response lag (TrI), time of rise (Tr), and time base (Tb),

bankfull discharge and the slopes of rising and falling limb of the hydrographs. A flow









with 1.67-year return interval is often recognized as bankfull discharge (Poff, 1996). The

eighteen ecologically relevant hydrologic indices were divided into four components of

hydrologic regimes, magnitude, frequency, duration, and rate of change, to statistically

characterize hydrologic variation.

Effects of Flow Magnitude on Stream Ecology

This group includes 6 parameters (mean and coefficient of variation) related to

response factor, baseflow index, and the peak discharge (qpk). The magnitude of the

water condition at any given time is a measure of the availability or suitability of habitat

and defines such habitat attributes as wetted area or habitat volume, or the position of a

water table relative to wetland or riparian plant rooting zones (Richter et al., 1996). High

flows maintain ecosystem productivity and diversity. For example, high flows remove

and transport fine sediments that would otherwise fill the interstitial spaces in productive

gravel habitats (Beschta and Jackson, 1979). Floods import woody debris into the

channel (Keller and Swanson, 1979), where it creates new, high-quality habitat (Moore

and Gregory, 1988; Wallace and Benke, 1984). Floodplains and wetlands provide

important nursery grounds for fish and export organic matter and organisms back into the

main channel (Junk et al., 1989; Welcomme, 1992). The scouring of floodplain soils

revives habitat for plant species that germinate only on barren, wetted surfaces that are

free of competition (Scott et al., 1996) or that require access to shallow water tables

(Stromberg et al., 1997).

Effects of Flow Duration on Stream Ecology

The 6 parameters in this group measure the duration of all the events considered in

the analysis. The parameters included in this group are the mean, and coefficient of

variation of dimensionless indices Tri/Tic, Tr/Tic, and Tb/Tic, where TrI, Tr, Tb, and T1i









corresponding to the response lag, time of rise, time base, and centroid lag of a storm

hydrograph, respectively. The duration is the period of time associated with a specific

storm condition that determines the differences in tolerance to prolonged flooding in

riparian plants (Chapman et al., 1982). Changes in duration of inundation can alter the

abundance of plant cover types (Auble et al., 1994). For example, increased duration of

inundation has contributed to the conversion of grassland to forest along a regulated

Australian River (Bren, 1992). For aquatic invertebrates and fishes prolonged flows of

particular levels can also be damaging (Closs and Lake, 1996). Whether a particular life-

cycle phase can be completed or the degree to which stressful effects such as inundation

of a flood plain can accumulate may be assessed from the duration of time over which a

specific water condition exists (Richter et al., 1996).

Effects of Flow Frequency on Stream Ecology

The two parameters in this group measure the number of occurrences of the

magnitude of the stream flow condition with respect to bankfull discharge. These

numbers of occurrences of bankfull discharges are reported as percentages of the total

events under analysis. Measures of exceedance of bankfull conditions have greater

ecological importance. These flows regulate numerous ecological processes within

riparian as well as flood plain areas. Frequent, moderately high flows effectively

transport sediment through the channel (Leopold et al., 1964). Movement of sediment

and organic resources such as detritus and attached algae revive the biological

community and allow many species with fast life cycles and good colonizing ability to re-

establish (Fisher, 1983). Consequently, the composition and relative abundance of

species that are present in a stream often reflect the frequency and intensity of high flows

(Meffe and Minckley, 1987; Schlosser, 1985).









Effects of Flow Rate of Change on Stream Ecology

This group includes 4 parameters, mean rate of change in peak discharge in rising

limb and falling limb of the storm hydrograph and their variabilities. Flow conditions'

rate of change, or flashiness, refers to the rate at which flow changes from one magnitude

to another, can influence species persistence and coexistence. At the extremes, flashy

streams have rapid rates of change, whereas stable streams have slow rates of change

(Poff et al., 1997). The rate of change in water conditions may be tied to the stranding of

certain organisms along the water's edge or in ponded depressions, or the ability of plant

roots to maintain contact with phreatic water supplies (Richter et al., 1996). Non-native

fishes generally lack the behavioral adaptations to avoid being displaced downstream by

sudden floods (Minckley and Deacon, 1991). Meffe (1984) documented that a native

fish, the Gila topminnow (Poeciliopsis occidentalis), was locally extirpated by the

introduced predatory mosquitofish (Gambusia affinis) in locations where natural flash

floods were regulated by upstream dams, but the native species persisted in naturally

flashy streams.

Data Collection

Watershed Characteristics

The Fort Benning study watersheds (Figure 3-2), Bonham-1 and Bonham-2,

Bonham, Little Pine Knot, and Sally Branch (named for the creek/stream that drains the

watershed), represent a range of the region's soils, topography, land use, and vegetation

communities (Table 3-2). In addition to standard watershed characteristics, the military

training land area and a dimensionless military disturbance index (DIN) were calculated

for each watershed. The military land use include fire ranges, maneuver training areas

for light, amphibious, and heavy training, air field, artillery and mortar firing points,
































15 3 12


Figure 3-2. Study watersheds, Bonham-1, Bonham-2, Bonham, Little Pine Knot,
Randall, Sally Branch, and Oswichee, in the Fort Benning military
installation.

airborne drop zones, specialized non-live-fire training assets, and duded impact areas. A

DIN is the sum of area of bare ground on slopes greater than 3 degrees and on roads as a

proportion of the total watershed area (Maloney et al., in press). Additional details

regarding the study area are found in Bhat et al., Ecological indicators in forested

watersheds in Fort Benning, GA: Relationship between land use and stream water

quality, submitted to Ecological Indicators; hereinafter referred to as Bhat et al.,

submitted manuscript).

According to variations in ecologically relevant hydrological characteristics that are

based on flow variability and predictability, and low- and high-flow extremes, the









Table 3-2. Physical characteristics of study watersheds. Acronyms BON-1, BON-2,
BON, LPK, and SAL represent the streams (or watersheds) Bonham-1,
Bonham-2, Bonham, Little Pine Knot, and Sally Branch, respectively.
Physical Characteristics BON-1 BON-2 BON LPK SAL
Topography
Area, km2 0.76 2.21 12.73 18.01 25.31
Average Elevation, m 121.8 133.5 125.5 146.3 136.8
Average Slope, degree 5.46 4.89 5.04 5.32 5.42
Vegetation
Pine, % 28 30 40 41 48
Deciduous, % 27 6 8 2 12
Mixed, % 39 50 22 34 15
Wetland, % 6 8 9 17 10
Military Training Land, % 0 6 5 2 2
NDVI 0.36 0.30 0.32 0.34 0.36
Soil
Sandy loam, % 78 69 69 72 49
Loamy sand, % 9 9 31 28 51
Road
Road Length, km 3.6 11.4 51.6 56.6 97.6
Road Density, km/km2 4.8 5.1 4.1 3.1 3.8
Stream
Stream Length, km 2.6 3.9 29.1 43.3 65.2
Stream Density, km/km2 3.4 1.7 2.3 2.4 2.6
Stream Order 2 2 4 4 4
Other
No. of Roads Crossing Streams 1 2 13 11 21
Bare Land, % 1 11 11 4 7
Disturbance Index, % 11 21 19 11 15

streams in the study watersheds are classified as 'perennial runoff (Poff, 1996). As the

present study focuses on the impacts of military training within an ecosystem as well as

the effects of these impacts on the stream biota, a reference watershed was identified to

contrast with watersheds impacted by military training activities. Approximately 94% of

the area of Bonham-1 watershed is covered by forest. Mechanized military activities are









not conducted in this watershed as compared to 2-6% of the total area of other watersheds

used for the same purpose. Also, the watershed has a small percentage of bare land,

limited roads, and only one road crossing the stream. Hence, Bonham-1 was used as a

reference watershed for this study.

Precipitation and Streamflow Data

Streamflow and precipitation were measured from January 2000 to December

2003. Precipitation was measured by twelve tipping bucket rain gauges distributed

throughout the study area. Watershed precipitation was determined by areal weighting

using the Thiessen polygon method. Daily discharge values for Bonham-1 and Bonham-

2 were calculated from ten-minute continuous stage records using rating curves. Stream

stage and velocity were measured half-hourly for Bonham, Little Pine Knot, and Sally

Branch. These data were used to calculate daily discharges using the area-velocity

method.

Methods

Annual-Based Hydrologic Indices

The annual-based approach uses multi-year streamflow records to define a series of

ecologically relevant hydrologic indices. These indices may be used to characterize

intra-annual variation in water conditions, analyze temporal variations, and compare

impacts of alteration among watersheds. For this assessment, nonredundant yet

biologically significant hydrologic indices are adapted as per the recommendations of

Olden and Poff (2003). Table 3-3 lists these indices for the perennial runoff type of






46


Table3-3. Summary of the recommended hydrologic indices for 'perennial runoff type of
streams (after Olden and Poff, 2003) used in the Annual-Based Hydrologic
Indices. BON-1, and BON-2 represent Bonham-1 and Bonham-2,
respectively. BON-1 is reference (REF) and BON-2 is (IMP) watershed. To
maintain the consistency to the past studies, the symbols presenting the
hydrologic indices in this table are kept as the same as those used in Olden
and Poff (2003). Data types A, M, and D stand for annual, monthly, and daily
discharge data, respectively. For definitions and method of calculation of the
indices, refer to Appendix A.


Flow components and name
of the hydrologic index
Magnitude of flow events
Averageflow conditions
Variability in December flow
Mean annual runoff


Symbol Units


MA26 unitless


BON-1 BON-2 Data
(REF) (IMP) type


0.70 0.47


MA41 m3/s/km2 0.00397


0.00504


Spreads in daily flows
Low flow conditions
Baseflow index


Mean of annual minimum flows
Median of annual minimum flows
High flow conditions


MAlO 1/m3/s


81.4 43.5


ML17 unitless 0.12 0.57
ML14 unitless 0.17 0.58
ML16 unitless 0.17 0.79


High flow volume


Mean maximum flow in May
Median of annual maximum flows
Frequency of flow events
Low flow conditions
Frequency of low flow spells
Variability in low flood pulse count
High flow conditions
High flood pulse count
Three times median flow frequency
Seven times median flow frequency


days/no.
MH23 of years

MH8 m3/s


0.016 0.017


MH14 unitless 63.4


FL3 year 1 0.75
FL2 unitless 0.31


FH4 year
FH6 year


FH7 year 6


0
0.26


10.25 2.50


0 A









Table 3-3. Summary of the recommended hydrologic indices for 'perennial runoff type
of streams (after Olden and Poff, 2003) used in the Annual-Based Hydrologic
Indices (Continued).


Flow components and name
of the hydrologic index
Duration of flow events
Low flow conditions
Variability in annual minima of
90-day means of daily discharge
Variability in low flow pulse duration
Variability in annual minima of
1-day means of daily discharge
High flow conditions
Means of 30-day maxima of daily discharge
Variability in high flow pulse duration
Flood free days
Timing of flow events
Average flow conditions
Constancy
Seasonal predictability of flooding
High flow conditions
Seasonal predictability of non-flooding
Rate of change in flow events
Average flow conditions
Variability in reversals (Positive)
Variability in reversals (Negative)
Change of flow (Decreasing)

Change of flow (Increasing)


BON-1
Symbol Units (REF)


DL10
DL17


unitless 0.78
unitless 0.09


BON-2
(IMP)


Data
type


0.32 D/M/A


0.07


DL6 unitless 0.60 0.28 D/M/A


DH13
DH16
DH24


unitless
unitless
days


TAl unitless
TA3 unitless


5.59
0.006
56.0


0.28
0.70


1.84
0.005
61.5


D/M/A
A
A


0.44
0.84


TH3 unitless 0.40 0.41


RA9
RA9
RA7

RA6


unitless
unitless
m3/s

m3/s


0.04
0.06


0.07
0.12


0.1136 0.0324

0.1119 0.0496


streams found in the study region. The definitions and the methods used to calculate

these indices are listed in Appendix A. In this study, the annual-based indices are used to


compare and contrast the Bonham-1 and Bonham-2 watersheds.









Storm-Based Hydrologic Indices

The storm-based approach uses the storm hydrographs to determine the indices.

Hydrograph separation was used to identify distinct storm events. 44-100 storm events

from 2001 to 2003 were used to calculate the response factor, baseflow index,

dimensionless indices (Trl/Tic, Tr/Tic, and Tb/Tic), watershed area (A) scaled peak

discharge (qpk/A), bankfull discharges, and the rate of change of peak discharges in rising

and falling limbs for five watersheds, where Bonham-1 is the reference watershed (Table

3-4). Appendix B lists the indices, definitions, and the methods of calculation.

Statistical Analyses

The ANOVA and the Tukey's multiple comparison tests were performed to

determine the differences between reference watershed and the impacted watersheds'

mean values of indices. The percent of forest extent, military training land, road density,

the number of roads crossing streams, bare land fraction, and disturbance index were

considered to assess the effects of disturbance on hydrology and the potential impact on

stream ecology. Pearson's correlation coefficients were calculated to examine the

strength and significance of the relationships between a watershed physical characteristic

and storm-based hydrologic index. Stepwise multiple linear regressions were performed

to identify relationships between an index and management related watershed

characteristics. Only variables having p-values less than or equal to 0.05 were retained in

the regression models.

Results

Watershed Disturbance Characteristics

The watersheds' physical characteristics are summarized in Table 3-2. Most of the

watersheds are highly vegetated (70% or more) with the majority characterized by pine









Table 3-4. Summary of hydrologic indices used in the Storm-Based Hydrologic Indices.
BON-1, BON-2, BON, LPK, and SAL represent the streams (or watersheds)
Bonham-1, Bonham-2, Bonham, Little Pine Knot, and Sally Branch,
respectively. BON-1 is reference (REF) and other watersheds are impacted
(IMP). represents the mean values of the indices of the impacted watersheds
that are different at significance level of 0.05 from the mean values of
reference watershed as confirmed by Tukey's multiple comparison test. For
definitions and method of calculation of the indices, refer to Appendix A.


Flow components and name
of the hydrologic index
Magnitude
Mean response factor
Variability in response factor
Mean baseflow index
Variability in baseflow index
Mean peak discharge
Variability in peak discharge
Frequency
Bankfull discharge
Two times bankfull discharge
Duration
Duration of time base
Variability in time base
Duration of response lag
Variability in response lag
Duration of time of rise
Variability in time of rise
Rate of change

Mean slope of rising limb
Variability in rising slopes

Slope of falling limb
Variability in falling slopes


REF


IMP


Symbol Units BON-1 BON-2 BON LPK SAL


MMRF
MyRF
MMBF
MvBF

MMPD
MvPD


FIFD
F2FD


DMTB
DvTB
DMRL
DvRL
DMTR
DvTR


RpPD
RpvPD

RNPD


unitless
unitless
unitless
unitless

m3/s/km2
unitless


0.030
1.09
0.73
0.19

0.077
2.0


0.010
1.45
0.86*
0.11

0.025*
2.3


% 90
% 82


unitless
unitless
unitless
unitless
unitless
unitless


m3/s/hr
unitless

m3/s/hr


0.20
2.2

0.050


0.060*
1.18
0.95*
0.04

0.030*
1.2


2
0


1.2*
0.6
1.1*
0.6
0.5
1.0


0.06*
2.4

0.020


0.10
1.4

0.070


0.001*
0.89
0.90*
0.07
0.001*
1.0


29
10


1.3*
0.5
0.8
0.5
0.5
0.9


0.01*
1.4

0.001*


0.020
1.30
0.95*
0.04
0.014*
1.4


6
0


1.2*
0.5
1.3*
0.4
0.5
0.7


0.04*
1.6

0.020


RNVPD unitless 2.4 2.0 1.5 1.2 1.2









and mixed pine and hardwoods. Deciduous forest typically covers only a small

percentage of these watersheds. However, Bonham-1 consists of 27% of deciduous

forest. The study watersheds range from less than 1 to 25 km2. Average elevations and

maximum slopes are relatively constant. Sandy and loamy soils are common in most of

the study watersheds. The military training land (0 to 6%) is relatively small, but varies

among watersheds. Total bare lands in these watersheds comprise 11 to 21% of the

watershed area, of which 1 to 8% of the total area is unpaved roads and trails. This extent

and variability of military training and bare land are typical of the entire Fort Benning

installation. While road density is relatively comparable among watersheds, the number

of roads crossing streams varies from 1 to 21. Military training land, bare land, and

disturbance index are positively correlated at significance level of 0.05 or lower.

Annual-Based Hydrologic Indices

Table 3-3 summarizes annual-based analysis results for Bonham-1 (reference) and

Bonham-2 (impacted) watersheds. The average flow conditions revealed that the mean

annual flow (MA41) is higher in the impacted watershed. However, the reference

watershed has higher flow variability for both annual (MAlO) and December month

(MA26) periods. The average flow event timing is more constant (TAl) and predictable

(TA3) in the impacted watershed.

The impacted watershed maintained a higher magnitude of minimum flows as

depicted by the baseflow index (ML17). This result is consistent across other low flow

indices (ML14 and ML16) and the findings that the impacted watershed had no low flow

spells (FL3) and lower variability of low flow pulse counts (FL2). Higher coefficient of

variation in annual minima of 90-day (DL10) and 1-day (DL6) means of daily discharge

for the reference watershed and similar variability in low flow pulse duration (DL17)









were found for both watersheds. While low flow values vary more for the reference

watershed, once the flow goes low, it stays low for same duration in both watersheds.

The reference watershed produces higher magnitude flow and thus maintains higher

median flow (MH14) during events. However, in the month of May, mean of the

maximum monthly flows (MH8) were similar for both the watersheds. During high flow

conditions, the reference watershed crosses a threshold of seven times the median annual

daily flow volume (MH23). In the impacted watershed, these floods never occurred. The

high flood pulse counts (FH4) and the frequency of floods (FH6 and FH7) in the reference

watershed are more than the impacted watershed. However, on a few occasions, the flow

crossed a lower threshold of 3 times median frequency of flood (FH6) in the impacted

watershed. The 30-day floods (DH13) went higher and stayed high (DH16) in the

reference watershed. Periods between floods (DH24) are similar for both watersheds, that

is, they have approximately two months of flood free days a year and of comparable

predictability (TH3).

Storm-Based Hydrologic Indices

The results of storm-based hydrologic assessment are summarized in Table 3-4.

Analysis of variance tests indicated that the mean values of storm-based indices except

the time of rise differ at the significance level of 0.05. For a number of indices, the

reference watershed exhibits distinct behavior as compared to the impacted watersheds.

Tukey's multiple comparison tests indicate that the indices other than time of rise and the

rate of change in falling limb in the reference watershed were significantly different from

the impacted watersheds. The reference watershed is characterized by a relatively low

baseflow index (MMBF) with significantly higher (MMPD) and more variable (MvPD)

peak discharge. During events, 90% of the total events produced greater than bankfull









discharge (F1FD) in the reference watershed indicating a highly connected system as

compared to 2-47% in impacted watersheds.

Storm flows consistently lasted longer and responded faster to rain events in the

reference watershed (DMTB and DvTB). Once the stream responded, the time of rise

(MMTR) was similar in all the watersheds. The reference watershed's combination of

fast response and high peak discharge results in a rapidly increasing rising limb (RpPD)

as compared to impacted watersheds.

Relationship between Military Land Management and Storm-Based Hydrologic
Indices

Correlation and regression analyses were performed to determine the relationship

among the watershed physical characteristics and the storm-based hydrologic indices.

Table 3-5 indicates that 7 key storm-based hydrologic indices are significantly related to

military land management. Increased military training land, bare land, and the

disturbance index will increase the time of rise as well as the variability in the time base.

Increasing the road density increases the variability in the time base and the rate of

change of rising limb. Increasing the number of roads crossing streams increases the

storm response lag, but decreases the time base. Results also show that an increase in the

number of roads crossing streams decreases the variability in the rate of change of falling

limb. No effects on hydrologic indices were identified for forestry management

practices.

Stepwise multiple correlations characterized the response of storm-based indices to

military impacts (Table 3-6). The greatest impact of land management is found with

statistically significant predictive models for indices of time base, response lag, and time









Table 3-5. Pearson correlation coefficients between watershed physical characteristics
and event based hydrologic indices. Characteristics are acronymed as Forest
(FOR), Military Training Land (MIL), Road Density (RDN), No. of Roads
Crossing Streams (NRC), % Bare Land (PBL), and Disturbance Index (DIN).
and ** indicates significance at or below 0.05, and 0.01 probability levels,
respectively.
Hydrologic Indices FOR MIL RDN NRC PBL DIN
MMRF 0.25 0.15 0.15 0.13 0.27 0.31
MvRF 0.40 0.59 0.71 -0.05 0.67 0.81
MMBF -0.62 0.44 -0.56 0.83 0.68 0.41
MyBF 0.63 -0.39 0.61 -0.85 -0.63 -0.35
MMPD 0.83 -0.37 0.64 -0.62 -0.47 -0.19
MvPD 0.66 0.21 0.96** -0.73 0.04 0.34
F1FD 0.57 -0.40 0.57 -0.85 -0.65 -0.39
F2FD 0.69 -0.43 0.62 -0.82 -0.64 -0.36
DMTB 0.63 -0.17 0.77 -0.91* -0.42 -0.12
DvTB -0.50 0.98** 0.14 0.08 0.96** 0.90*
DMRL -0.23 0.06 -0.54 0.97** 0.41 0.18
DvRL 0.21 0.59 0.84 -0.78 0.33 0.57
DMTR -0.39 0.99** 0.32 -0.17 0.89* 0.90*
DvTR -0.86 0.67 -0.45 0.35 0.68 0.47
RpPD 0.79 -0.34 0.59 -0.55 -0.42 -0.16
RpvPD 0.63 0.13 0.91* -0.78 -0.06 0.24
RNPD 0.42 0.10 0.35 -0.13 0.14 0.26
RNVPD 0.70 -0.07 0.85 -0.88* -0.29 0.03

of rise. Military training land, road density, and the number of road crossing streams


were the three management variables that impacted storm responses.









Table 3-6. Stepwise multiple regression models for event based hydrologic indices.
Military Training Land (MIL), Road Density (RDN), and No. of Roads
Crossing Streams (NRC), are the independent variables retained in the
regression analyses. and ** indicates significance at or below 0.05, and
0.01 probability levels, respectively.
Hydrologic indices Independent variables retained and R2 (adj)
regression equations
MMBF 0.729 + 1.91 MIL + 0.00968 NRC 0.94*
MvPD 1.21 + 0.666 RDN 0.88**
F1FD 0.929 6.97 MIL 0.039 NRC 0.94*
DMTB 1.8 0.0352 NRC 0.77*
DvTB 0.418 + 3.05 MIL 0.95**
DMRL 0.483 + 0.0393 NRC 0.92**
DMTR 0.454 + 0.837 MIL 0.97**
RNVPD 2.20 0.0563 NRC 0.70*

Discussion

Results representing annual-based average flow conditions showed higher

magnitudes of MA26, MA41, and MA10 in the reference watershed as compared to the

impacted one. In addition, these flows are less constant and less predictable over the

years as compared to the impacted watershed. Aquatic communities can show distinct

differences to changes in velocity and reduction in bed gradient, and associated fining of

bed sediments (Clausen and Biggs, 2000). Periphyton and benthic invertebrates are

particularly sensitive to different velocities and bed sediment size/stability (Minshall,

1984; Biggs, 1996). Clausen and Biggs (1997) found that invertebrate species richness

and periphyton biomass changed based on flow. These changes are likely related to

riparian vegetation on the amount of leaf litter input (Vannote et al. 1980; Davies-Colley

and Quinn, 1998).









The higher magnitude of the annual-based low flow indices (ML17, ML14, and

ML16) in the impacted watershed is attributed to higher groundwater input as compared

to the reference watershed. This increased input likely reflects the reduction of

interception characteristic of the military land uses (Bryant et al., in press). The relative

magnitude of low flows is likely to have important influences on biota through the

intensity of habitat destruction associated with drying during low flows. In this study, the

lower magnitudes of low flows as depicted by lower baseflow index (ML17) and smaller

low flow pulses (FL3) suggest that the long periods of low flow condition are more likely

in the reference watershed. Higher variability in the duration during low flow condition

(DL10, DL17, and DL6) in the reference watershed supports likelihood of such long

periods of low flow conditions. Long periods of low flow conditions and higher

variability in these conditions may provide selective pressure for specific life history

characteristics such as invertibrate aestivation and egg diapause, and physiological

tolerance to low dissolved oxygen (Williams and Hynes, 1977).

The higher magnitude of the annual-based high flow indices (MH23 and MH14) and

peak discharges (MMPD and MvPD) in the reference watershed as compared to the

impacted watershed suggests the likelihood of habitat regeneration associated with

sediment transport and floodplain inundation during high flows. For high flow events,

the degree of riverbed communities' disturbance is strongly related to degree of bed

movement (e.g., Biggs et al., 1999). Dissolved inorganic nitrogen and phosphorus

concentrations in rivers are strongly negatively correlated with specific yield and high

flow magnitude among watersheds, and among years within watersheds (e.g., Biggs and

Close, 1989; Close and Davies-Colley, 1990; Grimm and Fisher, 1992). This reflects the









degree of flushing of the nutrients mineralized through organic matter breakdown in the

soil profile and leachate from the underlying substrata.

Floods or high flow conditions are also important in influencing community

structure. The results based on the annual flow in this study clearly show that the

reference watershed produced more frequent high flows (FH4, FH6, and FH7). Similarly,

the results from storm-based analysis show that the frequency of discharges equaling or

exceeding bankfull (FiFD and F2FD) is higher in the reference watershed. Floods are

widely viewed as reset mechanisms (Resh et al., 1988), and flood-related mortality to

lotic organisms can result either directly from scouring, crushing, or downstream export

of individuals (Minckley and Maffe, 1987) or indirectly from food resources loss

(Hanson and Waters, 1974). As flood frequency increases, some invertebrates in the

reference watershed actively migrate either into the substratum or to quieter backwaters

to avoid sudden floods. Floods have been shown to regulate community structure by

facilitating local coexistence between asymmetrically competitive algal species (Power

and Stewart, 1987) and invertibrate species (Hemphill and Cooper, 1983) and between an

exotic fish predator and its native, relatively flood-resistant prey (Meffe, 1984). The

result in this study showed the consistency of storm-based F1FD and F2FD with annual-

based FH4, FH6, and FH7 indices suggesting F1FD and F2FD may be used as alternative

indices.

As indicated by the higher annual-based mean duration of high flow condition

(DH13) and the duration of storms (DMTB) in the reference as compared to the impacted

watershed, it is apparent that water resides in the reference watershed for a longer period

of time during high flow condition. However, the DMRL is smaller for the reference









watershed suggesting a quick rise in hydrograph. This can be attributed to a better

connectivity of riparian areas to the stream. This connectivity is important as the nutrient

concentrations can strongly control autotrophic production during inter-flood periods

(e.g., Biggs, 2000). Also, individual high flow events greatly reduce the biomass and

change the species composition of periphyton (e.g., Biggs and Stokseth, 1996), and

invertebrates (e.g., Cobb et al., 1992).

Predictive relationships were identified for storm-based hydrologic indices based

on watershed scale military management and suggest that the key variables related to

hydrologic alteration are military training land, road density, and the number of roads

crossing streams. Military training within a watershed can modify annual and storm

generated runoff due to changes in drainage, vegetation, and soils. The impact is due to

troop maneuvers and large, tracked and wheeled vehicles that traverse thousands of

hectares in a single training exercise (Quist et al., 2003). These activities' impacts,

ranging from minor soil compaction and lodging of standing vegetation to severe

compaction and complete loss of vegetation cover in areas with concentrated training use

(Wilson, 1988; Milchunas et al., 1999), are evident in the stream hydrographs and

suggest impacts to stream ecosystems. Disruption in soil density and water content

(Helvey and Kochenderfer, 1990), addition of sediment, nutrients, and contaminants in

aquatic ecosystems (Gjessing et al., 1984), and impairment of natural habitat

development and woody debris dynamics in forested floodplain streams (Piegay and

Landon, 1997) are potential impacts on terrestrial and aquatic ecosystems. In a recent

study in the same watersheds in Fort Benning, Houser et al. (2004) found that gross

primary productivity, respiration, and the benthic organic matter were low in the stream









corresponding to the impacted watershed (higher disturbance index) as compared to the

reference one (lower disturbance index). Road crossings commonly found in military

training areas may act as barriers to the movement of fishes and other aquatic animals

(Fumiss et al., 1991). Roads also directly change the hydrology by intercepting shallow

groundwater flow paths, diverting the water along the roadway and routing it efficiently

to streams at crossings (Wemple et al, 1996). This can cause or contribute to changes in

timing and routing of runoff (Jones and Grant, 1996).

Conclusion and Recommendations

In the present study, a reference watershed was compared to impacted watersheds

using annual and storm-based hydrologic indices within the Fort Benning military

installation. The results suggest a subset of the hydrologic indices recommended by

Olden and Poff (2003) are necessary for the Fort Benning streams. It is recommended

that mean annual runoff and its spread, baseflow index, high flow volume, frequency of

low flow spells, high flood pulse count, variability in annual minima of 90-day means of

daily discharges, and constancy be used in stream ecology studies in the Fort Benning

streams to identify disturbances relative to a reference watershed at a watershed scale.

The indices based on storm data may be used to augment the annual indices or as

surrogate to those indices. Storm-based magnitude and variability in peak discharge,

baseflow index, and the bankfull discharge were consistent with the results from annual-

based analysis. With respect to the potential influence of the frequency and duration

aspects of the flow regimes on the stream ecology, duration and variability of time base,

and duration of response lag are identified as critical hydrologic indices. The military

management practices, military training land, road density, and the number of road

crossing streams, were found to significantly affect these indices.














CHAPTER 4
PREDICTION OF NITROGEN LEACHING FROM FRESHLY FALLEN LEAVES:
APPLICATION OF RIPARIAN ECOSYSTEM MANAGEMENT MODEL (REMM)

Introduction

Forest canopy litterfall initiates the major pathway for recycling of nutrients from

plant to soil (Bubb et al., 1998). Nutrients are returned to the soil via litterfall and

throughfall fluxes in forested ecosystem (Ukonmanaho and Starr, 2001). Litterfall,

comprised predominantly of leaf litter, is usually the most important nutrient source

(Herbohn and Congdon, 1998). Litter quantity, litter decomposition and nutrient release

patterns are factors for understanding nutrient cycling in forest ecosystems (Rogers,

2002). Litterfall is primarily caused by natural senescence of leaves. Forest litterfall

quantity and composition varies among tree species, stand age and development, and

reflects environmental conditions, particularly water and nutrient availability (Binkley,

1986; Polglase and Attiwill, 1992). The amount and distribution of litterfall through time

are also affected by season, rainfall amount and distribution, and wind speed (Crockford

and Richardson, 1998).

Fluxes of dissolved organic matter are an important vector for the movement of

nutrients both within and between ecosystems (Cleveland et al., 2004). In many forest

ecosystems, more than half of the nitrogen is soil solution is in organic form (Qualls et

al., 1991). Hydrology plays an important role in both production and mobilization of

dissolved organic matter in the forest floor (Park and Matzner, 2003). A substantial

amount of potentially soluble organic matter exists in an adsorbed phase (Christ and









David, 1996). The amount of percolating water mobilizes these soluble organic matters

(Tipping et al., 1999).

Past studies have shown that the dissolved organic matter flux increases when the

percolating water passes through the forest floor (McDowell and Likens, 1988; Qualls et

al., 1991; Currie et al., 1996; Michalzik et al., 2001). In a recent study in India, Singh et

al. (1999) reported that the rapid mass loss of leaf litter occurred during the rainy season

and the rate of litter mass loss was positively correlated with the cumulative rainfall. In

an early study, Bocock et al. (1960) observed a rapid leaching from oak and ash leaf

litter, and suggested that the weight loss during the first month was largely due to a

physical loss by leaching. The primary sources of dissolved nutrients are considered to

be leaching of substance from fresh litter and the product of decomposition of the plant

litters (Qualls et al., 1991). Water-soluble nutrients are more easily leached from the leaf

litter of deciduous species, including birch, than from coniferous species, including Scots

pine (Pinus sylvestris) and Norway spruce (Picea abies) (Harris and Safford, 1996;

Hongve, 1999).

Depending on the species, leaf litter may release from 5 to 30% of original dry

weight as dissolved organic matter within 24 hours (Cummins, 1974). In a woodland

stream ecosystem in Massachusetts, McDowell and Fisher (1976) found that leaves lose

weight rapidly by abiotic leaching of soluble constituents for 1-3 days and decay more

slowly thereafter as a result of microbial decomposition. Gosz et al. (1973) estimated

15% mass loss for yellow birch leaves (Betula lutea) due to leaching within first month

on the forest floor. Heath et al. (1966) reported a 26% weight loss from birch leaves (B.









alba) that remained for two months on the forest floor. Nykvist (1961) found an 8%

weight loss in B. verrucosa after 24 hours of leaching in distilled water.

Annual fluxes of nitrogen in litterfall are greater than other nutrients regardless of

forest type but the amounts vary among forest types (Ukonmaanaho and Starr, 1999).

Dissolved organic nitrogen is the major form of nitrogen in stream water draining from

many mature forest watersheds, comprising about 95% of the total nitrogen (Qualls et al.,

1991). Nitrogen content of a hardwood forest litterfall on the Hubbard Brook

experimental forest comprised approximately 37% of the total nutrients (Gosz et al.,

1972). In a low nutrient system at the Coweeta Hydrologic Laboratory in North Carolina,

where carbon (C) to nitrogen (N) ratios in freshly fallen litter varied from 100 to 220,

Qualls et al. (1991) reported that the output of dissolved organic nitrogen from the forest

floor was 28% of the nitrogen input from the litterfall.

The Riparian Ecosystem Management Model (REMM) was developed by the U.S.

Department of Agriculture's (USDA) Agricultural Research Service (ARS) to

characterize the role of the riparian area on stream water quality (Lowrance et al., 2000;

Altier et al., 2002). In REMM, the riparian areas' water quality functions for control of

N, phosphorus (P), and sediment transport into surface waters are simulated and

analyzed. This model has primarily been tested and applied on high-nutrient riparian

buffers adjacent to agricultural fields in the past (e.g., Inamdar et al., 1999a; Inamdar et

al., 1999b; Lowrance et al., 2000).

The complex dynamics related to the nutrients in a watershed is particularly of

interest in low nutrient systems as the release of nutrients from plant litters prior to

decomposition may be an important aspect of characterizing stream water quality. A









watershed scale study related to surface water quality was conducted in the Fort Benning

military installation located in southwest Georgia. Experimental data from Fort Benning

shows an increase in total Kjeldahl nitrogen concentrations following litterfall. The

release of nitrogen is relatively rapid as compared to that released from decomposition of

organic matter. It is hypothesized that after leaves fall, the first precipitation events can

be expected to result immediate nitrogen leaching and a corresponding increase in

nitrogen levels in the stream water. The goal of this study is to determine whether freshly

fallen leaves in a riparian area are a significant source of nitrogen in low nutrient system.

The specific objectives of this study are to quantify nitrogen in streamflow using REMM

and verify the modeled results with the observed data.

Basic Concepts of REMM

Originally, REMM was developed to simulate buffer systems that meet

specifications recommended by the USDA Forest Service and the USDA Natural

Resource Conservation Service as a national standard (Natural Resource Conservation

Service, 1995). Three zones parallel to the stream characterize the riparian system. Thus

REMM simulates a three-zone buffer. The riparian area's soil is characterized in three

layers. In these soil layers, vertical and lateral movement of water and associated

dissolved nutrients takes place. The top two soil layers correspond with the soil horizons

A and B, respectively. Depth of the root zone or a restricting soil horizon defines the

third soil layer. A litter layer that interacts with surface runoff covers the soil layer 1.

The model takes upland water output supplied by the user and calculates loadings of

water, nutrients, sediment, and carbon based on actual area of the zones of a riparian area

(Lowrance et al., 2000).









The model input data consist of the physical conditions of the riparian area,

external loadings of water and nutrients to the riparian area, soil, vegetation, and soil

information. Altier et al. (2002) documents a detailed description of the algorithms,

equations, and parameters used in the model.

Hydrology component of REMM

REMM measures water stores and fluxes using a daily water balance. The water

balance includes interception, evapotranspiration, infiltration, vertical drainage, surface

runoff, subsurface lateral flow, upward flux from water table, and deep seepage.

Interception losses occur in the vegetation canopy and litter layer. Canopy interception is

an exponential function of the canopy storage capacity and the amount of daily rainfall,

and is simulated using a modified form of the Thomas and Beasley (1986) equation.

Precipitation falling through the canopy (throughfall) is subject to litter interception.

Similar to canopy interception, litter interception is determined by the mass of the litter at

any given time and the litter storage capacity.

Infiltration at the soil surface depends on the total depth of water. This depth of

water is the sum of the throughfall depth and overland flow depth from the upslope zone

or field. If the sum of incoming upland runoff and throughfall depth is less than the

infiltration capacity of the riparian soil, runoff infiltrates at the rate of application.

During high intensity storms, the runoff and throughfall rate may exceed the infiltration

rate. Surface runoff from the riparian buffer occurs when total water depth exceeds the

infiltration depth. In such conditions, model simulates the infiltration using a modified

form of the Green-Ampt equation (Stone et al., 1994).

When soil moisture exceeds the field capacity, vertical drainage from a soil layer

occurs. This vertical drainage also depends on the available water storage capacity of the









receiving layer. The rate of drainage is controlled by the lesser of the vertical hydraulic

conductivities of the draining and receiving layer. Vertical unsaturated conductivity is

simulated as a function of the soil moisture content using Campbell's equation (Campbell,

1974). When a water table builds up over the restricting soil layer, subsurface lateral

movement occurs. Lateral water movement is simulated using Darcy's equation.

Saturation excess overland flow occurs when soil profile is completely saturated.

Nutrient component of REMM

Simulation of nutrients in REMM is based on the Century Model (Parton et al.,

1987). The C cycle is fundamental for simulation of all organic matter dynamics and

many nutrient cycling processes in REMM. Release and immobilization of N is in

proportion to transformations of C. Soil and litter C is divided into residue (woody

debris, leaf litter, and roots) and humus (soil organic matter) pools. Each C pool has an

associated mineralization rate, transformation efficiency, and a specific range of C to N

ratios (Inamdar et al., 1999b). A first order rate equation with respect to C determines the

mineralization of C from each pool. Litter from leaves, stems, branches, coarse roots,

and fine roots are allocated into a readily decomposable (metabolic) or a resistant

(structural) residue pool based on lignin to N ratio (Pastor and Post, 1986). Each pools

decomposes at different rates.

REMM assumes a fixed fraction of C in fresh plant residue. Dissolved C from

metabolic residue and active humus pools differ by litter and soil layer (Altier et al.,

2002). The sum of C in throughfall and incoming runoff is mixed with C on the ground

surface to determine dissolved and adsorbed concentrations. Dissolved C moves with

surface runoff, subsurface lateral flow, and vertical drainage (Lowrance et al., 2000).









The amount of dissolved C carried out of the litter or soil layer is a function of the water

volume and the dissolved C concentration.

Nitrogen is input to the riparian area through precipitation, surface and subsurface

water flow, and adsorbed to incoming sediment carried by surface flow. Presence of

nitrogen in different forms and the associated degree of physical and chemical

stabilization influences its availability for microbial transformations and plant uptake.

Effective C to N ratios influencing residue decomposition are calculated in the model as a

function of the content of N in the litter as well as the content of available inorganic N in

the soil. As decomposition of the litter takes place, C and N are released (Lowrance et

al., 2000).

Inorganic nitrogen, ammonium and nitrate, are both available for immobilization

into soil organic matter. Immobilization of nitrate occurs only after all available

ammonium has been used. Ammonium may be in solution or adsorbed to the soil matrix.

Nitrification is calculated with a first-order rate equation influenced by temperature, soil

moisture, and pH. Temperature and soil moisture are variables, recalculated each day but

pH is a constant for each zone and layer combination. With increasing amounts of

substrate, the rate of nitrification is lagged to represent a delayed microbial response.

The calculation of denitrification is a function of the interaction of factors representing

the anaerobic condition, temperature, nitrate, and mineralizable C. Denitrification mostly

occurs as water-filled pore space gets above 60% (Inamdar et al., 1999b). Nitrogen

transformation processes are shown in Appendix C.










Application of REMM

Study Area

The study area consisted of a second-order watershed, Bonham-South (221 ha), in

the Fort Benning military installation, Georgia (Figure 4-1). A detailed description of

this site along with the instrumentation and the type of data collected is provided in


Figure 4-1. Study area. Bonham-South is a second-order watershed within the Fort
Benning military installation in Georgia.

Bryant et al., in press and Bhat et al., in review. The riparian area covers approximately

2.5% of the total watershed area. The riparian length is 3,500 m. Based on the visual

inspection in the watershed, a 12 m wide riparian buffer along the stream is considered.


0 0.25 0.5 1
I iMC I Ih









This riparian buffer is divided into three equal zones (Zone 1, Zone 2, and Zone 3) of 4 m

each. Corresponding soil layers (Soil layer 1, Soil layer 2, and Soil layer 3) are

considered in each zone. A selective list of riparian area and hydrological parameters and

their corresponding values are provided in Table 4-1. The riparian area consists primarily

Table 4-1. REMM model parameters values by the riparian zone for the Bonham-South
watershed in Fort Benning, Georgia. The dimension of riparian area from the
upland to the stream perpendicular to the stream is referred to as width, and
the distance along the stream is referred to as length. The parameter values
are identical for Zone 1, Zone 2, and Zone 3 except for the slope. The
superscripts 1, 2, and 3 of the slope values represent the Zone 1, Zone 2, and
Zone 3, respectively.
Parameters Units Values
Riparian zone length m 3,500
Riparian zone width m 4
Slope % 3.01, 3.82, 4.23
Manning's n unitless 0.06
Total soil profile thickness m 3.3
Individual soil horizon thickness m
Soil layer 1 0.3
Soil layer 2 1.0
Soil layer 3 2.0
Saturated hydraulic conductivity cm/h
Soil layer 1 15
Soil layer 2 13
Soil layer 3 1
Porosity cm/cm
Soil layer 1 0.40
Soil layer 2 0.35
Soil layer 3 0.35
Wilting point cm/cm
Soil layer 1 0.07
Soil layer 2 0.07
Soil layer 3 0.07
Riparian area ha 4.2
Field drainage area (surface) ha 216.8
Field drainage area (subsurface) ha 221.0









of hardwoods (approximately 60%) that are mostly scrub oak (Q. rubra), sweetbay (M.

virginiana), sweet gum (L. styraciflua), water oak (Q. nigra), willow oak (Q. phellos),

red maple (A. rubrum), and swamp tupelo (N. aquatica. Mixed species including

southern scrub oak (Q. rubra) and yellow pine (P. ponderosa) cover approximately 21%

of the riparian area, whereas loblolly (P. taeda) and longleaf (P. palustris) pine cover

approximately 19%. The riparian area soils are predominantly sandy loam. The uplands

are mostly on loamy sand.

Data and Input Parameters

REMM operates at a hillslope scale. To apply the model at watershed scale, it is

necessary to convert the watershed into an equivalent hillslope by adjusting the geometry

of watershed. As the riparian buffer has specified length (parallel to the stream) and

width (perpendicular to the stream), upland contributing field width of the watershed was

adjusted to equate the total area of the watershed.

REMM requires soil and vegetation nutrient pools for the riparian area of the

watershed to initialize the model (Table 4-2). Required field data to run the model

includes daily surface and subsurface runoff from the uplands to the riparian area (Table

4-3). The model also requires daily weather data and parameter values representing

topographic, soil, and vegetative conditions within the riparian area. Strategic

Environmental Research and Development Plan's (SERDP) Ecosystem Management

Project (SEMP) continuously records the atmospheric data within the military

installation. Daily weather data were obtained from SEMP data repository that included

precipitation amount and duration, maximum and minimum air temperatures, solar

radiation, wind speed, and dew point temperature. Daily streamflow for Bonham-South









Table 4-2. Soil and vegetation nutrient pools for Bonham-South watershed's riparian
area. The litter and soil pools of nutrient consist of metabolic, structural,
active, slow, and passive humus pools. The values in litter layer as well as
individual soil layers in each zone are presented. Vegetation pools consist of
nutrients in leaves, stems, branches, coarse roots, fine roots, and heartwoods.
Total values of nutrients are reported in the vegetation pool. The parameter
values are identical for Zone 1, Zone 2, and Zone 3.
Parameters Depth Bulk density Carbon Nitrogen Biomass
(m) (g/cm3) (kg/ha) (kg/ha) (kg/ha)
Litter and Soil Litter 0.01 1.0 18,100 1,134
Soil layer 1 0.30 1.4 29,280 655
Soil layer 2 1.00 1.5 24,640 644
Soil layer 3 2.00 1.5 11,560 252
Total 83,580 2,685
Live vegetation 31,390 753 78,477


Table 4-3. Model simulated annual hydrologic budget for the riparian areas in Bonham-
South watershed. and ** represents the surface and subsurface flow input to
the zone 3 from uplands. Values are rounded off to nearest whole numbers.
Year
Units 2000 2001 2002 2003 4-yr average
Precipitation mm/yr 475 736 731 1,014 739


Surface flow in*
Subsurface flow in**
Observed stream flow (total)
Simulated surface flow
Simulated subsurface flow
Simulated stream flow (total)


mm/yr
mm/yr
mm/yr
mm/yr
mm/yr
mm/yr


253
258
6
244
250


was calculated from 10-minute continuous stage records using rating curves and area-

velocity method.

Typically, REMM uses daily surface runoff depth generated from the runoff

collected by surface runoff samplers at the upland-Zone 3 interface. REMM uses the

hydraulic gradient upland wells and Zone 3 wells, saturated thickness at Zone 1, and the

saturated hydraulic conductivity to compute subsurface flow loading to the buffer. Due









to the lack of measured surface and subsurface flow from the upland to the riparian area,

it is assumed that the baseflow fraction in the measured streamflow is a contribution of

upland subsurface flow to the riparian area. Therefore, measured streamflow from

January 2000 through December 2003 was partitioned into baseflow and surface runoff

using constant slope base flow separation technique (McCuen, 1998). Surface runoff

from the upland to the riparian area was not considered (Table 4-3). Parameters that

describe the riparian area dimensions, soil, and vegetation characteristics were derived

from measured data and previously published literature (Inamdar et al., 1999a, Inamdar et

al., 1999b, Lowrance et al., 2000; Garten Jr., et al., 2003). Important parameters related

to litter and soil layer are listed in Appendix D. Monthly canopy cover in the riparian

area was determined by direct measurement with a Model-A spherical densiometer using

the method outlined by Lemmon (1956) from June 2001 to September 2003.

Measured water and soil nutrient concentrations in the study area included the total

Kjeldahl nitrogen (TKN), nitrate, and ammonium in precipitation, streamflow, soil water,

and shallow groundwater. Two transects in the riparian area near the outlet of the

watershed were considered for groundwater and soil water monitoring. Groundwater

monitoring wells (1-3 m deep) and adjacent tension lysimeters at 20 and 60 cm depths

were positioned on both sides of the stream along the riparian transects. Stream water,

groundwater, and soil water samples were collected from October 2001 to September

2003; biweekly from October 2001 to November 2002 and monthly thereafter. Bonham-

South stream was sampled during 16 storm events between September 2002 and

September 2003 using an event triggered ISCO sampler that was programmed to collect









hourly samples based on the flow depth. Bhat et al. (unpublished manuscript), describes

the sample analysis procedures for different chemical constituents.

As REMM simulations are performed on a daily time step, and observed TKN

concentrations are of hourly time step, comparisons between the two were based on the

TKN masses produced during the events that lasted 24 hours or longer. The model

simulates the masses of dissolved and particulate organic nitrogen, dissolved ammonium

and nitrate in surface and subsurface flows on a daily basis. The mass of simulated TKN

is the sum of dissolved and particulate organic nitrogen and dissolved ammonium.

Carbon dissolves from metabolic residue and active humus pools. In REMM, incoming

C from precipitation, surface, and subsurface flow is assumed to be in dissolved form.

Carbon available to be dissolved from the active humus pool is a fraction of the total C

present in the same pool. This fraction of set at 0.31 based on an estimate by McGill et al.

(1981) for the proportion of dying bacteria in metabolic form. The total amount of

dissolved C is the sum of metabolic C residue and the dissolved C fraction in the active

humus pool. In REMM, stoichiometric relationships are assumed between C and N in the

organic matter. As C is transformed, corresponding N is also transformed. A Freundlich

isotherm method determines the amount of dissolved ammonium from inorganic pool of

N in REMM. Observed masses of TKN during storm events are calculated by

multiplying the concentrations by the volume of water.

Model Calibration

REMM's hydrology and nitrogen components were evaluated using data collected

in this study from riparian area as well as data from the literature. Data collected in

nearby watersheds in Fort Benning, Georgia were used to initialize the C and N in the

soil pools (Garten Jr. et al., 2003). The litter C and N pools are based on the literature









values (Silveira, M.L., Reddy, K.R. (Department of Soil and Water Science, University

of Florida), Comerford, N.B. (Department of Soil and Water Science, University of

Florida). Litter decomposition and soluble organic carbon and nitrogen release in a

forested ecosystem. Unpublished manuscript). Soil and vegetation nutrient pools for the

riparian area of the watershed are listed in Table 4-2. Bonham-South's soil layer

thickness, soil porosity, saturated hydraulic conductivity, clay content, carbon decay rate,

and denitrification rate were calibrated.

Model calibration involved the comparison of the simulated streamflow and TKN

output with the measured values. While calibrating hydrology and nutrient components

of the model, parameters such as soil and litter C and N pools, which are based on

literature values, were fixed. Other fixed parameters included riparian length and width,

surface and subsurface draining area. The remaining parameters including soil layer

thickness, saturated hydraulic conductivity, soil porosity, clay content, carbon decay rate,

and denitrification rate in each zone and soil layers of the riparian buffer were adjusted to

match the observed and predicted results.

Fox (1981) recommends mean biased error (MBE) and mean absolute error (MAE)

to measure the difference between observed and model predicted values. Mean biased

error is calculated as the average error between predicted and observed values

accumulated over the total number data points. Mean absolute error considers the

absolute values of the errors. In the present study, MBE and MAE are modified to

calculate the percent difference of modeled streamflow and TKN mass from the observed

values. Mean bias difference (MBD) calculates the average difference accumulated over

the total number of events, expressed as a percent of the observed value. Mean absolute









difference (MAD), expressed also as a percent, considers the absolute values of the

differences. Mean absolute difference takes negative values and replaces them with their

absolute values. Observed and modeled streamflow was also compared using the Nash-

Sutcliffe efficiency. The Nash-Sutcliffe efficiency criteria is based on the normalized

least square objective function that evaluates the sum of the squares of flow residuals

(Nash and Sutcliffe, 1970).

Sensitivity Analysis

A sensitivity analysis was performed to determine the effects of key hydrological,

soil, and vegetation parameters on streamflow and TKN fluxes. The parameters were

canopy cover fraction, riparian zone width, soil layer thickness, maximum carbon decay

rate for litter and humus, maximum denitrification rate, soil saturated hydraulic

conductivity, soil porosity, and soil clay content. Each parameter was changed by +10%

and -10% from the values used as the best estimates for the calibration simulations.

Results

Hydrology

Simulated daily flows for the study watershed closely matched the observed flows

(Figure 4-2). Over the 4-year simulation period, MBD and MAD between observed and

predicted streamflow were 2% and 12%, respectively. The model had Nash-Sutcliffe

efficiency of 80%. REMM tends to underestimate the streamflow during low flow and

overestimate during the storms. Approximately 1% of the streamflow was the

contribution from surface runoff generated in the riparian area. Majority of streamflow

output was through the subsurface (Table 4-2).










2.0- 0
1.8 -
1.650
1.6 ------- Simulated
S1.4 Observed
1.2ate
SPrecipitation
0 6


















correspond to simulated leaf fall (Figure 4-4). Modeled TKN concentrations were higher.




















during high leaf fall months. From October 2001 to December 2003, modeled leaf fall
o l 0.41
0.2
0 .0 .......



Date

Figure 4-2. Comparison of REMM simulated daily flow with the observed daily flow for
the Bonham-South watershed.

There was considerable variability between and within years for measured

precipitation and streamflow data during study period. It is evident from Figure 4-2 that

an increase in the average flow during the year 2003. This increase was consistent with

the variation in precipitation during the study period. Overall, model simulations for

streamflow can be considered good for the calibrated parameters.

Nitrogen

Observed stream TKN concentrations during the study period showed a strong

correspondence with the leaf fall (Figure 4-3). Approximately 50% of the canopy in the

riparian area dropped during November-January each year. Higher TKN concentrations

were observed during the same period of time. The simulated TKN concentrations also

correspond to simulated leaf fall (Figure 4-4). Modeled TKN concentrations were higher

during high leaf fall months. From October 2001 to December 2003, modeled leaf fall

totaled 12,200 kg/ha, which corresponds to 5,400 kg/ha/yr. As depicted in the figure,











100 1.2
0 1.0
80 a
70 -
)- 0.8
50 CanopyCover -0.6
S40 TN
30 \ 0
S20 0.2
10
0 . 0 .0




Date

Figure 4-3. Observed canopy cover and TKN concentrations each month for the
Bonham-South watershed.


3000

2500

2000

1500

S1000

500

0


r -"CC] C"-] C-] C C']""- C10 mmC C mm i C



Date


Figure 4-4. Monthly simulated TKN concentration and simulated leaf mass in the
Bonham-South's riparian area.

simulated leaf fall began as early as June each year and completed by the end of

December-January. As observed in the riparian area of the watershed, leaves were

present in the canopy year-round. For example, during the 2001-2002 dormant season,









approximately 45% of the canopy was still covered with leaves. Simulated N present in

the live vegetation showed an inverse relation with the simulated TKN concentrations in

the stream (Figure 4-5).


80 1.6
:70 -mE Simulated leafN 1.4
70 1.4
---- Simulated TIKN
5 0 1.0
.* -- : ,*
40 .. 0.8
30 -- 0.6!
S20 0.4
10 0.2
0 -----0.0
ooOooooooooooooooooooOOoooo
II I I I I I I I I I I I I I I I I


Date

Figure 4-5. Monthly simulated TKN concentration and simulated nitrogen present in live
leaves the Bonham-South's riparian canopy.

A careful analysis of model simulation results showed that a daily total

precipitation of 6 mm or more produced surface runoff in the riparian area. This surface

runoff is responsible for leaching and transporting TKN from the leaf litter in the riparian

zone to the stream. Due to the irregular nature of surface runoff generation, the TKN

variability during such events, the frequencies of observed data, and REMM's ability to

simulate on a daily basis, storm events lasting 24 hours or longer were selected for further

study. The masses of TKN were calculated for two different scenarios. First, total mass

of TKN during an event was calculated for the total streamflow. Second, baseflow was

separated from the total flow, and TKN mass corresponding to surface runoff was

determined. The observed and simulated TKN masses during individual storm events









were compared for both the scenarios. Out of the 16 storm samples collected in the study

area, 6 storms between October 2002 and May 2003 were suitable for the comparison

with the daily output from REMM (Figure 4-6). The simulated TKN masses produced in

surface runoff during the storms were comparable to the observed values (Figure 4-7).

MBD and MAD between the observed and the simulated TKN masses were 8% and 23%,

respectively. As the majority of the flow in the stream is subsurface flow, the effect of

subsurface flow in carrying the TKN mass was also analyzed. The result showed that

MBD and MAD between observed and simulated total TKN masses were 17% and 24%,

respectively (Figure 4-8).

Sensitivity Analysis

The changes in streamflow and TKN outputs for the -10% to +10% parameter

changes ranged from 0 to 1% and 0 to 7%, respectively (Table 4-4). Changes in soil

porosity did not have significant effects on the streamflow and TKN. A decrease in soil

porosity of 10% led to a total streamflow increase of only 0.5%, whereas the increase in

TKN was only 0.6%. Increase in clay content decreased the TKN output by 0.7%.

Increase in saturated hydraulic conductivity by 10% did not affect the streamflow but

increased TKN by 7.2%. Increasing soil layer thickness slightly reduced the streamflow,

but reduced TKN by 5.4%.

























momn m oom
Storm on 10/15102 begin ng 11
Storm on 10/15102 beginning 11 AM


I 0 0



00 0 0


0 0 0 0 00 0 0 0
000002000 0

Storm beginning 02/16/2003 6:30 AM


0 0



0-0

0 c


D 0i D C I D 0 0 00 0 a


Storm on 04/24/2003 beginning 8:30 PM


0

21

4

6
'*g S


0

2

4

6 .
CI
4J


0

2

4

6
C.
=


0000000000000
0000000000000
-Storm on 11-11002 beginning 7 AM
Stonn on 11/11/2002 beginning 7 AM


2

4

6 H
' 1


Storm egiing 0222/03 4:30 AM
Stonn beginning 02/22103 4:30 AM


0

2

4

6 .,
CI
Hi


0



4


ud


0000000000000000

CeMn] en 0 -en-en]- in- en-nM
Storm on 05/22/2003 beginning 9:30 PM


Figure 4-6. Observed TKN concentrations during the events. Vertical bars represent
precipitation, solid lines represent the streamflow, and the hollow circles
represent the TKN concentrations.


20

I15

S10

0.5


00


2.5

2.0-

a|1.5

S1.0
0 .
0.5
Sn


20


15


10


0 5


0 0


rZ


U.1
















- 0.4

0.3

0.2


C".] C"-] C\1 CA"] C" CA]
-' I ,-' CA] C1 ".]
O C1 C-O .

Storm Events


Figure 4-7. Comparison of the observed and simulated TKN masses during the events.
These masses represent the contribution from the surface runoff only.



0.8
0.7 E Observed
0.6 Simulated
0.5

C 0.4


0.2
0.1
0.0
ci c] enm m m en
o o o o o o
o o o o o o
C'i C-i C\1 C-i C\1 C-i

0< C\.1 CA] n

Storm Events


Figure 4-8. Comparison of the observed and simulated TKN masses during the events.
These masses represent the contribution from the total flow.










Table 4-4. Sensitivity of modeled streamflow and TKN based on +/- 10% change in
model parameters for Bonham-South watershed.


Parameters
Canopy cover (+)
Canopy cover (-)
Soil layer thickness (+)
Soil layer thickness (-)
Denitrification rate (+)
Denitrification rate (-)
Carbon decay rate (+)
Carbon decay rate (-)
Saturated hydraulic conductivity (+)
Saturated hydraulic conductivity (-)
Soil porosity (+)
Soil porosity (-)
Clay content (+)
Clay content (-)


Discussion

Over the four years, the simulated average annual surface runoff that is generated in

the riparian area and contributed to the streamflow was less than 1% of the annual

observed precipitation (Table 4-3). The simulated surface runoff generated in the

riparian buffer in the study watershed is small compared to earlier studies done in a

similar watershed in Georgia. A study conducted by Shirmohammadi et al. (1984) in

Little River watershed in Georgia reported approximately 4-12% contribution of annual

precipitation to surface runoff Inamdar et al. (1999a) reported the average annual

surface runoff contribution to streamflow was approximately 8% of the annual

precipitation for Gibbs Farm site within the Little River watershed. Surface runoff

contribution to streamflow in our study watershed may not be significant in terms of


Percentage change
Stream flow
-0.9
0.9
-0.1
0.1
0.0
0.0
0.0
0.0
0.0
0.0
-0.4
0.5
0.0
0.0


TKN
2.3
-0.2
-5.4
5.7
0.0
0.0
0.9
-0.5
7.2
-0.2
-1.5
0.6
-0.7
0.4









hydrology, but the modeled results suggest that it is capable of transporting nitrogen from

the litter and top soil layer to the stream (Figure 4-7). Subsurface contribution to the

streamflow for the Bonham-South watershed was approximately 20% of the average

annual precipitation (Table 4-3). The simulated subsurface contribution in this study is

close to the range of 14-22% reported by Shirmohammadi et al. (1984) for Little River

watershed.

The riparian forest community found along Bonham-South watershed is

representative of the Southeastern riparian forest (Shure and Gottaschalk, 1985;

Lowrance et al., 2000). Slash pine (P. elliotii), longleaf pine (P. palustris), black gum

(N. sylvatica), sweet gum (L. styraciflua), scrub oak (Q. rubra), sweetbay (M

virginiana), water oak (Q. nigra), willow oak (Q. phellos), red maple (A. rubrum), and

swamp tupelo (N. aquatic) are often dominant canopy species in these riparian forests.

The modeled total leaf fall in the study watershed from October 2001 to December 2003

was approximately 12,200 kg/ha, which corresponds to 5,400 kg/ha/yr. The modeled

average litter mass produced during the same period was approximately 7,000 kg/ha/yr.

Approximately 77% of the litter mass were leaves. Contribution of the modeled leaf fall

to the total litter production falls in the neighborhood of 80% reported by Meentemeyer et

al. (1982), and within the range of 72-84% reported by Shure and Gottaschalk (1985).

The simulated litter mass produced in the study area is higher than the average 5,000

kg/ha for a mixed hardwood forest in the Northeastern U.S. (Gosz et al., 1972) and is

comparable with the results found in the Southeastern U.S. Mulholland (1981) reported

litter production of 6,100 kg/ha in a small stream swamp in eastern North Carolina. A

bottomland hardwood forest and a cypress-tupelo stand located in Louisiana produced









litter masses of 5,750 kg/ha and 6,200 kg/ha, respectively (Conner and Day Jr., 1976).

Brinson et al. (1980) reported a value of 6,500 kg/ha for an alluvial swamp forest in

North Carolina.

Simulated results showed that the nitrogen content in the litter during the study

period averaged 77 kg/ha of which approximately 59 kg/ha was the contribution from the

leaves alone. This value is less than 83.4 kg/ha of nitrogen contribution from the

deciduous leaves in the Northeastern U.S. (Gosz et al., 1972). As observed in the riparian

area, 45% canopy was present even during the dormant seasons because of the presence

of approximately 19% of conifers and 21% of mixed species. Modeled results also

support the presence of the carbon in the leaves and the biomass in the canopy. The

modeled result showed that on an average approximately 527 kg/ha of carbon was

present in the canopy during the dormant season. This value of carbon suggests the

presence of approximately 22 kg/ha of nitrogen in the canopy during the same time

(Figure 4-5). Higher concentrations of TKN during the leaf fall and the inverse relation

with the nitrogen present in the canopy support the hypothesis of flushing of nitrogen

from the leaf litter.

The simulated TKN concentrations follow the trend of leaf mass accumulation in

the riparian area suggesting a higher rate of release of TKN from freshly fallen litter

during the fall and early spring. Brinson et al. (1980) reported higher nutrient fluxes

following litterfall in an alluvial swamp forest in North Carolina. In a study of nutrient

content of litterfall in the Southeastern U.S., Gosz et al. (1972) reported that

approximately 56% of the total nutrient was released during September through

December when the majority of litterfall (50%) occurred. A study conducted in cypress









swamp forest in Florida, Schlesinger (1978) reported that in the month of November

when 56% litterfall occurred, 45% of nitrogen was released from the system.

Higher TKN concentrations were observed in the Bonham-South stream during the

precipitation events (Figure 4-6). The storm events presented in Figure 4-6 were long

enough to calculate the masses of TKN during the storms to compare REMM's simulated

daily values. Comparisons of TKN masses produced during the events were comparable

to REMM's simulated values. The masses of TKN during the storms were calculated for

two different scenarios. When modeled TKN mass contribution of surface runoff that is

produced in the riparian zone was compared with the observed value in the stream, MBD

and MAD for the six storms were 8% and 23%, respectively. Mean biased difference

value is less reliable as there is the risk that large outliers cancel each other out. As MAD

takes negative values and replaces them with their absolute values, large outliers are de-

emphasized; hence it is less sensitive to extreme values. A MAD of 23% in the surface

runoff suggests that the model was able to produce comparable amount of TKN as

observed during the storm. The comparison between observed and simulated TKN mass

from the total flow showed comparable MBD (17%) and MAD (24%). On average, 43%

of the observed total TKN mass was contributed by surface runoff. This result suggests

that the surface runoff during the precipitation events is a major source of leaching

nitrogen from the forest floor and transporting it to the stream.

Conclusion and Recommendations

Originally, REMM was developed to operate at a hillslope scale. The results in this

study suggested that given appropriate upland inputs for a site, REMM can be used at

different scenarios of riparian area width, length of zones, vegetation type, and soil

properties. The trend and magnitude of the observed streamflow for the study watershed









was effectively simulated by REMM. The model, however, overestimated the

streamflow during high flow and underestimated during low flow periods. The

hydrologic budget showed a good agreement between observed and predicted

streamflow.

The simulated litter and corresponding leaf masses in the study watershed were

comparable to the values reported in the literature. Due to the model's longer simulation

time step, a meaningful comparison of simulated and observed TKN was possible only

through the masses produced during storms that were equal to or longer than a day.

Comparison of TKN masses during six different storm events showed similar values both

in the surface runoff and the total flow. The results supported the hypothesis of nitrogen

leaching from freshly fallen leaves during the precipitation events. These results

provided further insight into the nutrient dynamics of the riparian area. The model

simulations respond as expected to precipitation and vegetation patterns over the study

period. Results clearly indicated that the presence of fresh leaf litter in the riparian area

increases the TKN concentration, and hence mass, in the stream. The model effectively

captured the trends of leaf mass accumulation in the riparian area and subsequent high

concentration and mass during those periods. With the present version of REMM, the

comparison between observed and simulated values is possible only on a daily basis.

However, given the fact that the precipitation events and the sampling frequencies are

often shorter, simulation results using a smaller time scale would be useful. Therefore, it

is recommended to modify REMM from its present version of daily time step to a smaller

time step, preferably hourly, for more effective and meaningful interpretation and

evaluation of riparian nutrient flushing.














CHAPTER 5
SUMMARY AND CONCLUSION

The National Research Council of the United States has proposed that reliable and

comprehensive environmental indicators be developed to monitor ecological changes

from natural and anthropogenic causes. The military installation in Fort Benning,

Georgia offered a unique opportunity to study military impacts on water quality and

quantity. Ecohydrological approaches were used to relate the effects of anthropogenic

perturbations on water-vegetation-nutrient interactions in the Fort Benning watersheds.

In this research, statistical relationships among water quality parameters and the

watershed physical characteristics in low-nutrient watersheds were identified and

examined. This research also identified and developed hydrologic indices that

characterize the impact of military land management on watershed, and investigated the

riparian corridor's role on water quality by quantifying the nitrogen leaching from freshly

fallen leaves.

Relationships among watershed physical characteristics and water quality

parameters in the study watersheds and the regression analysis showed that pH, chloride,

total phosphorus, total Kjeldahl nitrogen, total organic carbon, and total suspended solids

are useful indicators of watershed physical characteristics that are susceptible to

perturbations. A comparison of the results between a reference and impacted watershed

in terms of hydrologic indices that are derived from long-term daily flow as well as the

storm-based data showed a clear distinction between the watersheds in terms of

hydrological flow regimes. Storm-based magnitude of baseflow index, magnitude and









variability of peak discharge, and the frequency of bankfull discharge were consistent

with the results from annual-based analysis. Results showed that these storm-based

indices might be used as surrogates to the indices derived from long-term data. The

analysis identified the relationships between the extent of military training land, road

density, and the number of roads crossing streams with the storm-based baseflow index,

bankfull discharge, response lag, and time of rise. For the low nutrient systems of this

study, seasonal and storm variations in water quality were found to be strongly influenced

by precipitation events which caused nitrogen leaching from recently fallen leaf litter and

increased nitrogen levels in the stream water.

This study showed that the signatures of military alterations to watershed landscape

are detectable in the water quality and the flow regime. The observed alterations to these

regimes suggest impacts to aquatic ecosystems. The water quality related indicators and

hydrologic indices presented in this research provide specific measures of stream water

quality and instream flow, respectively, that respond to military impacts. Each indicator

is linked to one or more readily measured watershed scale factors. The indicators could

be useful in predicting effects of military land management and evaluating restoration

activities on the quality as well as the quantity of the water.

This research has several implications for the Fort Benning military installation.

The results of this study indicated that baseflow sampling of water quality can be used to

assess the military training impacts on stream water quality. The water quality indicators

identified in this study provide measures of the current watershed conditions. The water

quality indicators identified in this study provide measures of the current watershed

conditions. Changes to stream water quality due to military training within the Fort









Benning military installation can be identified by collecting long-term and routine stream

water quality data and comparing indicator values to current conditions. Future activities

should develop indicator thresholds beyond which improved management and restoration

practices should be implemented. As nutrients are one major factor controlling the

quality of the receiving aquatic ecosystems, the thresholds should be based on the

ecosystems impacts.

The U.S. Army manages approximately 4.8 x 106 hectares of land for military

training, and it has developed the Land Condition Trend Analysis (LCTA) program to

systematically monitor terrestrial impacts from military training and to support the

mitigation and remediation of severely impacted training lands (Quist et al., 2003).

Although the magnitude of the training would differ from one installation to the other, the

nature of the impact remains similar. Military training typically increases soil bulk

densities and compaction, decreases infiltration, diminishes plant growth, degrades water

quality, and affects water quantity. Therefore, the research findings in this study can be

implemented for management and restoration practices to other military installations

across the U.S. to minimize the impacts of the military training. A common finding

between the water quality and quantity signatures is that military training impacts differ

from those found due to agriculture and urbanization. In the former, water quality

degradation impacts typically manifest as high nutrient loads. However, in military

installations, the loss of topsoil and vegetation results in waters that have significantly

lower nutrient levels. In a similar fashion, the hydrologic regimes in watersheds with

significant military impacts suffer from a loss of variability, but exhibit higher annual

discharge values.















APPENDIX A
ANNUAL-BASED INDICES DEFINITIONS AND CALCULATION PROCEDURES


Symbol


Definition


Method


MA26 Coefficient of variation
in monthly flows for
December


MA41 Mean annual flow
divided by watershed
area


Calculate mean monthly flow for December (q ), i
= 1,..., n;, n =no. of years
Calculate mean of all December flows,
- n
QD qd /n
1=1
Calculate standard deviation (SD) of qd

Calculate coefficient of variation, CV= SD/QD

Calculate mean yearly flow (q), i = 1,..., n =
no. ofyears
Divide q, by watershed area (A)


MAlO Ranges in daily flows
divided by median daily
flows (where range in
daily flows is the ratio of
20th/80th percentiles in
daily flows across all
years)


ML17 Seven-day minimum
flow divided by mean
annual daily flows


Calculate Q/A = qYl /n,i

of years

Combine daily flows of all years
Calculate 20th percentile (Q,20)
Calculate 80th percentile (Qps,)
Calculate median flow (Qm)
Calculate R= Q,,/Q,,o
Calculate R/ Qm


1,..., n;, n= no.


Calculate mean yearly flow (q ), i = 1,..., n;, n
no. ofyears
Calculate 7-day minimum ( q7m, ) flow
Calculate ratios R, = (q7mn)' / q
n
Calculate R,l /n
1=1




Full Text

PAGE 1

ECOHYDROLOGICAL STUDY OF WATE RSHEDS WITHIN THE MILITARY INSTALLATION IN FORT BENNING, GEORGIA By SHIRISH BHAT A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2005

PAGE 2

Copyright 2005 by Shirish Bhat

PAGE 3

This work is dedicated to my parents, Tara and Narmada Bhat, and uncle Devendra Bhat.

PAGE 4

iv ACKNOWLEDGMENTS First of all, I would like to thank Dr. Je nnifer Jacobs, for her constant guidance, encouragement, patience and continuous support over the past five years. Her enthusiasm for research and quest for excellence have left an everlasting impression in my mind. To me, she has been more than an advisor, and this research would not have been possible without her. Secondly, I would like to thank Dr. Kirk Hatfield for being on my committee and offering me guidance in the la ter part of my research. I would also like to thank Dr. Wendy Graham, Dr. Ramesh Shrestha, and Dr. Ramesh Reddy for being on my committee and their invalu able suggestions throughout th e research work. I would like to thank Dr. Richard Lowrance and Randall Williams of USDA-ARS, Georgia, whose contributions in this research have b een tremendous. I deeply benefited from all the long hours of fruitful discussions with them on a multitude of topics. I would like to thank Hugh Westbury at Fo rt Benning, Georgia, for his coordination efforts during the field trips, Dwight Dindial and Ch arles Campbell for their efforts in the collection of the field data, and Phil Harmer for his contribution in the laboratory. I wish to extend my gratitude to Lewi s and Kim Bryant, and Donna Rowland, who have been very helpful during my difficult days at UF. I also wish to extend my gratitude to Subarna and Pramila Malakar for their care and support. I wi sh to extend my thanks to Mr. Binod Palikhe for his help during my academic pursuit. I would like to extend my thank to Debra Carol, Carol Hipsley, S onja Lee, Doretha Ray, and all the Civil Engineering staff for their help during all thes e years. I would also like to thank my

PAGE 5

v friends Deepak Singh, Shashi Shrestha, Hemant Belbase, Rishi Bhattarai, Dipendra Piya, and Anand Bastola, and colleagues Nebiyu Tiruneh, Aniruddha Guha, Qing Sun, Mark Newman, Ali Sedighi, Chris Brown, Brent Wh itfield, Gerard Ripo, Haki Klammler and others for their encourag ements and moral support. I would like to thank my pare nts and all the family member s for their constant love and encouragement. They have allowed me to pursue whatever I wanted in life. Without their guidance and affection, it would have been impossible for me to advance my education.

PAGE 6

vi TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES.............................................................................................................ix LIST OF FIGURES.............................................................................................................x ABSTRACT....................................................................................................................... xi CHAPTER 1 GENERAL INTRODUCTION....................................................................................1 Objectives..................................................................................................................... 5 Dissertation Organization.............................................................................................6 2 ECOLOGICAL INDICATORS IN FORESTED WATERSHEDS IN FORT BENNING, GEORGIA: RELATIONS HIP BETWEEN LAND USE AND STREAM WATER QUALITY....................................................................................9 Introduction................................................................................................................... 9 Study Area..................................................................................................................12 Methods of Study........................................................................................................13 Description of Watersheds..................................................................................13 Characterization of Di sturbance Categories........................................................14 Collection and Analysis of Stream Samples.......................................................15 Statistical Analyses..............................................................................................15 Results........................................................................................................................ .16 Watershed Physical Characteristics.....................................................................16 Water Quality Parameters....................................................................................19 Effects of Disturbance Categories.......................................................................21 Relationship between Watershed Physical Characteristics and Water Quality Parameters........................................................................................................22 Discussion...................................................................................................................29 Conclusion and Recommendations.............................................................................33 3 HYDROLOGIC INDICES OF WATERSHED SCALE MILITARY IMPACTS IN FORT BENNING, GA...............................................................................................35 Introduction.................................................................................................................35

PAGE 7

vii Flow Regimes and H ydrologic Indices.......................................................................37 Effects of Flow Magnitude on Stream Ecology..................................................40 Effects of Flow Duration on Stream Ecology.....................................................40 Effects of Flow Frequency on Stream Ecology...................................................41 Effects of Flow Rate of Change on Stream Ecology...........................................42 Data Collection...........................................................................................................42 Watershed Characteristics...................................................................................42 Precipitation and Streamflow Data......................................................................45 Methods......................................................................................................................45 Annual-Based Hydrologic Indices......................................................................45 Storm-Based Hydrologic Indices........................................................................48 Statistical Analyses..............................................................................................48 Results........................................................................................................................ .48 Watershed Disturbance Characteristics...............................................................48 Annual-Based Hydrologic Indices......................................................................50 Storm-Based Hydrologic Indices........................................................................51 Relationship between Military Land Management and Storm-Based Hydrologic Indices...........................................................................................52 Discussion...................................................................................................................54 Conclusion and Recommendations.............................................................................58 4 PREDICTION OF NITROGEN LEACHING FROM FRESHLY FALLEN LEAVES: APPLICATION OF RIPARIAN ECOSYSTEM MANAGEMENT MODEL (REMM)......................................................................................................59 Introduction.................................................................................................................59 Basic Concepts of REMM..........................................................................................62 Hydrology component of REMM.......................................................................63 Nutrient component of REMM...........................................................................64 Application of REMM................................................................................................66 Study Area...........................................................................................................66 Data and Input Parameters...................................................................................68 Model Calibration................................................................................................71 Sensitivity Analysis.............................................................................................73 Results........................................................................................................................ .73 Hydrology............................................................................................................73 Nitrogen...............................................................................................................74 Sensitivity Analysis.............................................................................................77 Discussion...................................................................................................................80 Conclusion and Recommendations.............................................................................83 5 SUMMARY AND CONCLUSION...........................................................................85 APPENDIX A ANNUAL-BASED INDICES DE FINITIONS AND CALCULATION PROCEDURES..........................................................................................................88

PAGE 8

viii B STORM-BASED INDICES DEFI NITIONS AND CALCULATION PROCEDURES..........................................................................................................95 C NITROGEN TRANSFORMATIONS IN REMM.....................................................98 D PARAMETERS AND RATE CO NSTANTS USED IN REMM..............................99 LIST OF REFERENCES.................................................................................................102 BIOGRAPHICAL SKETCH...........................................................................................115

PAGE 9

ix LIST OF TABLES Table page 2-1. Physical characteristics of study wa tersheds in Fort Benning, Georgia. ................17 2-2. Pearson correlation coefficients between watershed characteristics. ......................18 2-3. t-test results for differences in mean values of watershed phys ical characteristics and water quality parameters. ..................................................................................21 2-4. Pearson correlation coefficients be tween watershed characteristics and water quality parameters....................................................................................................23 2-5. Stepwise multiple regression models for water quality parameters..........................24 3-1. Definitions of terms used to de scribe hyetograph and response hydrograph............39 3-2. Physical characteristics of study watersheds.............................................................44 3-3. Summary of the recommended hydrologic i ndices for 'perennial runoff' type of streams......................................................................................................................46 3-4. Summary of hydrologic i ndices used in the Storm-Ba sed Hydrologic Indices. ......49 3-5. Pearson correlation coe fficients between watershed physical characteristics and event based hydrologic indices................................................................................53 3-6. Stepwise multiple regression models for event based hydrologic indices. ............54 4-1. REMM model parameters values by th e riparian zone for the Bonham-South watershed in Fort Benning, Georgia........................................................................67 4-2. Soil and vegetation nutrient pools for Bonham-South watersheds riparian area.....69 4-3. Model simulated annual hydrologic budget for the riparian areas in Bonham-South watershed........................................................................................69 4-4. Sensitivity of modeled streamflow and TKN based on +/10% change in model parameters for Bonham-South watershed................................................................80

PAGE 10

x LIST OF FIGURES Figure page 2-1 Study watersheds......................................................................................................14 2-2 Box plots of water quality parameters......................................................................20 2-3 Relationships between military land and water quality parameter..........................25 2-4 Relationships between road dens ity and water quality parameter...........................26 2-5 Relationships between number of ro ads crossing streams and water quality parameter..................................................................................................................27 2-6 Relationships between disturbance index and water quality parameter...................28 3-1 Terms used to describe hy etograph and response hydrograph.................................39 3-2 Study watersheds......................................................................................................43 4-1 Study area................................................................................................................. 66 4-2. Comparison of REMM simulated daily fl ow with the observed daily flow for the Bonham-South watershed...................................................................................74 4-3. Observed canopy cover and TKN c oncentrations each month for the Bonham-South watershed........................................................................................75 4-4. Monthly simulated TKN concentrati on and simulated leaf mass in the BonhamSouths riparian area.................................................................................................75 4-5. Monthly simulated TKN concentration and simulated nitrogen present in live leaves the Bonham-Souths riparian canopy............................................................76 4-6. Observed TKN concentrations during the events......................................................78 4-7. Surface runoff comparison of the observed and si mulated TKN masses..................79 4-8. Total streamflow comparison of the observed and si mulated TKN masses..............79

PAGE 11

xi Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy ECOHYDROLOGICAL STUDY OF WATE RSHEDS WITHIN THE MILITARY INSTALLATION IN FORT BENNING, GEORGIA By Shirish Bhat May 2005 Chair: Kirk Hatfield Cochair: Jennifer Jacobs Major Department: Civil and Coastal Engineering Relationships among watershed physical characteristics and water quality parameters were explored for seven watersheds in Fort Benning, Geor gia, using statistical analyses to identify chemical indicators of ecological changes. Correlations were identified among the indicators and watershe d physical characteristics. Regression results suggested that pH, ch loride, total phosphorus, tota l Kjeldahl nitrogen, total organic carbon, and total suspended solids are useful indicators of watershed physical characteristics that are suscep tible to perturbations. The magnitude, frequency, duration, timi ng, and rate of change of hydrologic conditions regulate ecological processes in aquatic ecosystems. Analysis of 26 nonredundant hydrologic indices showed undistur bed watershed produced higher magnitude and more frequent lowand high-flows as compared to the disturbed watershed. Eighteen storm-based indices, grouped into four flow components, were proposed to statistically characterize hydrologic varia tion among different watersheds. Results

PAGE 12

xii showed that these storm-based indices might be used as surrogates to the indices derived from long-term data. Statistical analys is showed that the watershed physical characteristics such as military training land, road density, and the number of roads crossing streams predicted hydrologic indice s such as storm-based baseflow index, bankfull discharge, response lag, and time of rise well. Riparian vegetation has an important role in altering water quality in the forested watersheds. The leaching of organic or mi neral products on the forest floor provides potential additional effects on water quality. In these areas, nutrients are released into the fresh water systems due to the leaching a nd decomposition of vegetation litter. The release of nutrients from plan t litters prior to decomposition may be an important aspect of characterizing stream water quality. To e xplore nitrogen leaching in a riparian area during litterfall, Riparian Ecosystem Ma nagement Model developed by USDA-ARS has been applied. The simulated total Kjeldahl nitrogen masses in the study watershed were close to the observed values. The model effectively captured the trends of litter mass accumulation in the riparian area and subse quent high concentrations and masses during those periods.

PAGE 13

1 CHAPTER 1 GENERAL INTRODUCTION Ecohydrology is the interdisciplinary research field in which hydrology and ecology come together. It may be defined as the study of the functional relations between hydrology and biota at the watershed scale (Zalewski, 2000). Resource managers and land use planners seek ecohydrological knowle dge enabling them to design effective land use plans and water management strategies. Water, soil, and plant cover are funda mental components that determine the productivity of the land. The successional stag es in evolution of ecosystems depend on climatic and hydrologic conditions and on nutri ent availability. The unique composition of plants and animals determines the ability to retain water and nutrients within the system (Zalewski et al., 1997). The amount of water and its quali ty in the aquatic environment are guided mostly by climate. Water and nutrient cycling in terrestrial systems are tightly linked with each other and with human activity. The growth and activity of human populations have increased the input of nutrients to terrestrial and aquatic ecosystems (Schindler and Bayley, 1993; Vitousek et al., 1997). Various forms of environmental perturbations affect quality and quantity of inland waters by increasing runoff, erosion, sedimentation, and pollution. Modifications of land use affect the wate r quality, residence tim e, surface runoff, soil moisture, evaporation, and ground water. For example, an increase in urbanization and non-forest land increased nitrogen concen tration in the streams (Osborne and Wiley, 1988; Sponseller et al., 2001). Land-use prac tices can potentially affect and modify

PAGE 14

2 channels along the flow paths in a la ndscape and may enhance or disrupt runoff (Nakamura et al., 2000). Forest cutting can increase the frequency and volume of debris slides due to dead and decayed tree roots that contribute to soil strength on marginally stable sites (Sidle et al., 1985). Water quality is highly variable, from place to place and from time to time, even within a particular ecosystem. It is depe ndent on many factors, both natural and as a consequence of human activities. In both cases, the quality of water is further affected by the soils and vegetation over and through which it passes. Nutrients are released into the freshwater systems due to the decomposition of the vegetation litter and modify the water quality. Riparian vegetation ha s an important role in alteri ng water quality as it functions as a source or a sink of va rious organic and mineral components (Cirmo and McDonnell, 1997). Agricultural practices an d urbanization are widely recognized human impacts on water resources. A specific type of human impact is caused at military installations. Military training within a wa tershed can affect the draina ge patterns, vegetation, and soils. The impact is due to troop maneuvers a nd large tracked and wheeled vehicles that traverse thousands of hectares in a single tr aining exercise (Quist et al., 2003). The impacts of such activities range from minor soil compac tion and lodging of standing vegetation to severe compaction and complete loss of vegetation cover in areas where training is concentrated (Wilson, 1988; Milchunas et al., 1999). The resultant impacts are evident both in the stream hydrology and the stream ecosystems. Potential impacts on terrestrial and aquatic ecosys tems include disruption in so il density and water content (Helvey and Kochenderfer, 1990), addition of sediment, nutrients, and contaminants in

PAGE 15

3 aquatic ecosystems (Gjessing et al., 1984) and impairment of natural habitat development,and woody debris dynamics in forested floodplain streams (Piegay and Landon, 1997). Military roads can signi ficantly alter hillslope hydrology by redistributing soil and rock materials on slope s and increasing the ra te of debris slide initiation. Roads also directly change th e hydrology by intercepting shallow groundwater flow paths, diverting the water along the roadway and routing it to stream at road crossings. Road crossings can intercept groundwater drai nage networks and collect groundwater from upslope areas, diverting it into drainage ditches as surface water (Wemple et al., 1996). Road crossings co mmonly found in military training areas may act as barriers to the movement of fish and other aquatic ha bitats (Furniss et al., 1991). Given the nature of military land use, ma nagement or military testing and training, military land managers face the conflicting demands of balancing the primary military mission with legal requirements to protect land and water qualit y (Milchunas et al., 1999). The military impacts can result in significant disruptions to the water and nutrient cycles in the freshwater ecosystems and water resources. A few studies in the past have addressed the effects of military training on terrestrial and aquatic ecosystems. For example, Wilson (1988) found that the tank traffi c resulted in a significant loss of native species, increased abundance of in troduced species, and increased bare soil at a training site located in Manitoba. Milchunas et al. (1999) examined the effects of military vehicles on plant communities and soil char acteristics in Pinon Canyon Maneuver Site, Colorado. Whitecotton et al. (2000) examined the impact of foot traffic from military training on soil bulk density, infiltration rate and aboveground biomass. Recently, Quist

PAGE 16

4 et al. (2003) conducted a study in the Fort Riley Military Re servation, Kansas, to study the effects of military use on terrestrial and aquatic communities. Understanding human impacts in many la ndscapes needs the identification of critical landscape elements and analysis of landscape pattern change (OÂ’Neill et al., 1997). Generalized response to military tr aining includes the reduction in native and perennial grasses, abundance of introduced species, and an in crease in bare soil. Clearly, these responses will alter biogeochemical a nd hydrological processes, which regulate nutrient and water dynamics. However, impact of landscape changes to water quality and stream is not well understood (Wand et al., 2001). In order to maintain and improve water quality, there is an increasing need to understand the relationships among watershed land use and stream ecosystems (Wang et al., 2001). In parallel, perturbation induced effects on water quantity are critical to stream ecology. One of the many approaches to study th e functional interrelations between the hydrology and the stream biota is streamflow characterization and classification. This approach develops hydrologic indices that account for characteristics of streamflow variability that are biologically relevant (Olden and Poff, 2003). Characterization of streams through development of ecologically relevant hydrologic indices is based on long-term streamflow data. However, past studies overlooked the importance of stormflow data for the development of such hydrol ogic indices. Flow characteristics are important where changes in land use are anticipated and where alterations to the flow regimes need to be assessed.

PAGE 17

5 The leaching of organic or mineral products from the forest floor provides potential additional effects on water quality. Fluxes of nitrogen through the riparian zone are intrinsically linked to water movement, bot h over and through the soil, and are also strongly influenced by biological processes occurring in that zone. Nitrogen and organic carbon dynamics in riparian zones are closely interrelated. While ma ny of the factors that can potentially influence nitrogen a nd carbon fluxes through riparian zones are broadly known, there is presently incomplete quantitative information on the relative importance of the flushing of nitrogen from freshly fallen leaves during precipitation events. Objectives The United StateÂ’s Department of De fense (DOD) policy has established ecosystem management as its approach to manage the military lands by maintaining and improving the sustainability and biological diversity of terrestrial and freshwater ecosystems while supporting human needs, including the DOD missions. In order to identify critical deficiencies and rese arch opportunities on ecosystem management problems on defense installations, the St rategic Environmental Research and Development Program (SERDP) of DOD in itiated the SERDP Ecosystem Management Program (SEMP) in December 1997. The objectiv es of SEMP were to (1) establish longterm research sites on DOD lands for military-relevant ecosystem research, (2) conduct ecosystem research and monitoring activ ities relevant to DOD requirements and opportunities, and (3) facilitate the integration of results and findings of research into DOD ecosystem management practices. The goal of this research is to illustra te how an ecohydrological approach could be used to advance our ability to predict the effects of anth ropogenic perturbations on water-

PAGE 18

6 vegetation-nutrient interactions in the military installation at Fort Benning, Georgia. The issues addressed in this research encompass many of the main scientific challenges in the military installationsÂ’ ecohydrology that in clude the effects of military related perturbations on stream wate r quality and quantity, and the value of studying nutrient dynamics in the riparian corridors of such regions. The specific objectives of this research include (1) identification and examin ation of the statistical relationships among water quality parameters and the watershed physical characteristics in low-nutrient watersheds, (2) identification and development of hydrologic i ndices that characterize the impact of military land management on watershe ds, and (3) investigati on of the effects of nitrogen leaching from freshly fallen leaves on nutrient dynamics in a riparian area. Dissertation Organization Each chapter of this dissertation, except Chapters 1 and 5, is written as a selfcontained individual paper focusing on a topi c that has not been addressed before. Contributions are in the areas of water qual ity and land use within a military installation (Chapter 2), development of storm-based hydr ologic indices (Chapter 3), and watershed scale nutrient leaching from a riparian area (Chapter 4). Chapter 5 summarizes and concludes the research work Chapter 2 is an original contribution in th e effects of military activities related land use on surface water quality. The major results from Chapter 2 are that the concentrations of total organi c carbon, total Kjeldahl nitroge n, total suspended solids, pH, and total phosphorus in the str eam show the greatest suscepti bility to direct effects of military activities. This chapte r also identifies significant statistical relationships among the water quality parameters and the military land uses. These relationships provide the

PAGE 19

7 guidance for maintaining the surface water quality within the Fort Benning military installation. Chapter 3 is an original contribution in the ecohydrol ogy that presents both annualbased and storm-based methods for determini ng hydrologic indices that are of ecological importance. Detailed descriptions of the me thods for determining th ese indices and their significance in aquatic ecosystems are descri bed in this chapter. To statistically characterize the hydrologic variation among different watersheds, 32 annual-based hydrologic indices are analyzed. Eight out of 32 annual-based indices are recommended to use for management practices within the Fort Benning military installation. As a new approach to characterize the streamflow vari ability, 18 storm-based indices are proposed. These indices are grouped into magnitude, fr equency, duration, and rate of change of flow. Storm-based indices are compared w ith annual-based indices. The storm-based methodology provides guidance for measurements of appropriate i ndices within the military installation. Chapter 4 is an original c ontribution to the water quality function of the riparian area. Nitrogen leaching from freshly falle n leaves in a riparian area during the precipitation events is quantified. The observe d nitrogen masses in the stream during the precipitation events are compared with the m odel simulated values. Riparian Ecosystem Management Model (REMM) is used to quantify the nitrogen in the ripa rian area. In this chapter, details of the nitroge n leaching from the freshly falle n leaves are explored. The results showed that the model effectively captured the trends of litter mass accumulation in the riparian area and subsequent high con centrations during those periods. Analysis

PAGE 20

8 showed that the simulated total Kjeldahl nitrogen masses during the precipitation events were close to the observed masses. Chapter 5 summarizes and concludes the re search work. This chapter contains recommendations to improve management of the water resources within the Fort Benning military installation. Future research n eeds are also outlined in this chapter.

PAGE 21

9 CHAPTER 2 ECOLOGICAL INDICATORS IN FORESTED WATERSHEDS IN FORT BENNING, GEORGIA: RELATIONSHIP BETWEE N LAND USE AND STREAM WATER QUALITY Introduction Ecological monitoring is esse ntial to protect eco logical health and integrity. As human activity alters land cover, degradati on of water resources begins in the upland areas of a watershed. The first step to ward effective ecological monitoring and assessment is to realize that the ultima te goal is to measure and evaluate the consequences of human actions on ecological systems. Human activities that alter land use eventually affect biogeochemical pro cesses that influence water quality and alter ecological processes. The National Research Council of the Unite d States recently conducted a critical evaluation of indicators used to monitor ecological changes from either natural or anthropogenic causes. During recent decades, efforts have been increasing to develop reliable and comprehensive environmental i ndicators because of growing environmental concerns (National Research Council, 2000) Indicators rapidly and effectively communicate system status. Ecological indicato rs help to elucidate both the effects of human activities and natural processes. They can also help to asse ss future implications of these factors on ecosystem integrity. Once i ndicators identify areas or elements of the environment that are under st ress, successful management of problems can be measured relative to both interim targets and long-term goals.

PAGE 22

10 Indicators that relate ke y ecological responses to hu man perturbations provide useful tools to better underst and ecological effects and thei r monitoring and management. A suite of indicators ranging from microbiol ogic to landscape metrics is necessary to capture the full spatial, temporal, and ecol ogical complexity of impacts (Dale et al., 2002). Evaluation of representative indices across major physical gradients (e.g., soils, geology, land use, water quality and quantity) can signal early environment change and help diagnose the cause of an environmental problem. Understanding human impacts in many la ndscapes needs the identification of critical landscape elements and analysis of landscape pattern change (OÂ’Neill et al., 1997). Attention has refocused on relations hips among watershed characteristics and stream water quality (Johnson et al., 1997). In order to maintain and improve water quality, there is an increasing need to unders tand the relationships among watershed land use and stream ecosystems (Wang et al., 2001). Land use provides information about ecosy stem function and characterizes the extent and diversity of ecosystem types. National Research Council (2000) has recommended land use as one of the most eff ective indicators for ecological assessment. Hydrologists and aquatic ecologists have l ong known that the pathway by which water reaches to a stream or lake has a major effect on water quality. Early studies on the physical (Harrel and Dorris, 1968) and ch emical (Hynes, 1960) characteristics of watersheds focused on the influence of geomorph ic characteristics such as drainage area, gradient, and stream order on turbidity, disso lved oxygen concentration, and temperature. Many recent studies examine the influence of terrestrial ecosystems on stream or wetland water quality (Richards and Host, 1994; Richards et al., 1997). Many other studies have

PAGE 23

11 found relationships between land use and concen trations of nutrients in streams (e.g., Hunsaker and Levine, 1995; Johnes et al ., 1996; Bolstad and Swank, 1997). Watershed properties constrain in-s tream physicochemical and biotic f eatures. Richards et al. (1996) showed that ecosystems could be influenced by land use at regional or broad geographic scales. Osborne and Wiley (1988) found that the distance of urban land cover from the stream effectively predicts stream n itrogen and phosphorus concentrations. Within a military installation context, land managers are challenged to use the land for military training purposes in a manner that is both ecologically sound and meets military mission requirements (Garten Jr et al., 2003). Lands can suffer a slow degradation if over-utilized by long-term human activities. The heavy vehicles used in mechanized military training cause disturbance of soil structure and can change the physical properties of the soil (Iverson et al., 1981). In rangelands, tracked vehicle traffic affects the hydrological characteristics (T hurow et al., 1993). Trampled vegetation, vehicle tracks through undisturbed area, and er osion caused by the overuse of trails are some examples of the visible degradation to a landscape caused by military training exercises. Few studies have develope d predictive relationships among watershed physical characteristics and surface water chem istry specific to military land use and lownutrient systems. In the coastal plain of the Apalachicol a-Chattahoochee-Flint (ACF) river basin, cropland and silvicultural la nd in upland areas is separa ted from streams by relatively undisturbed riparian flood plai n and wetland habitats (Frick et al., 1998). This is in contrast to many intensively farmed areas of the United States where wetlands have been drained, channelized or filled, and little or no riparian buffers remain between cropland

PAGE 24

12 and streams. Frick et al. (1998) reported that the lower nutri ent concentrations in streams within the ACF river basin could partially be attributed to wetland buffer areas, and minimal use of pesticides as compared to ot her areas of the United States. Other studies in the southeastern coastal plain watershe ds (e.g., Lowrance 1984; Lowrance et al., 1992; Perry et al., 1999; Fisher et al., 2000) focused primarily on th e agricultural impacts, and urbanization on stream water quality. The contribution of areas affected by military training to nutrient discharges, specifically in Fort Benning watersheds, is yet to be quantified. This paper identifies and examines the st atistical relationships among water quality parameters and the watershed physical charact eristics in seven low-nutrient watersheds located in the Fort Benning military installati on, Georgia. It is hypot hesized that surface water quality parameters can be used as indicat ors of ecological changes in watersheds. Study Area The Fort Benning Army Installation occupies approximately 73,503 ha in Chattahoochee, Muscogee, and Marion C ounties of Georgia and Russell County of Alabama (Figure 2-1). The climate at Fort Be nning is humid and mild. Rainfall in this region occurs regularly throughout the year. July and August are the warmest months with average daily maximum and mi nimum temperatures of 37 and 15oC. An average daily maximum and minimum temperature of 15.5 and -1oC are reported in the coldest months, January and February. Annual pr ecipitation averages 1050 mm with October being the driest month (Dale et al., 2002). Mo st of the precipitation occurs in the spring and summer as a result of thunderstorms. H eavy rains are typical during the summer but can occur in any month. Snow accounts for less than 1% of the annual precipitation.

PAGE 25

13 Fort Benning is located within the sout hern Appalachian Piedmont and Coastal Plains. The northern boundary of the instal lation lies along a transi tion zone between the Piedmont and Upper Coastal Plain. The so ils in the area are dominated by loamy sand with some sandy loam. Following establis hment of the installation in 1918, with subsequent additions in 1941, we see that heavy training imp acts only selected, mostly upland, portions of the installa tion. Many areas are maintain ed as safety buffers, and have little military use. Timber mana gement includes harves ting and thinning. The loblolly and longleaf pine fore sts are subjected to regular lo w-level fires for management purposes (Dale et al., 2002). Methods of Study Description of Watersheds The study watersheds, Bonham-1 and Bonham2, Bonham, Little Pine Knot, Sally, Oswichee, and Randall (named for the creek wh ich drains the watershed), within Fort Benning represent a range of regionÂ’s so ils, topography, land use, and vegetation communities (Figure 2-1). These watersheds have a heterogeneous land cover predominantly consisting of either forested or open areas. Forest ed areas are broadly characterized as mixed pine and hardwoods or pine that are mostly 30-50 years old with the soils in A-horizon range approximately 1-10 cm in depth (Garten Jr et al., 2003). Open areas are either military, brush, or managed wildlife openings. Other cover includes upland and bottomland hardwood forest s. The military openings are clear-cut parcels of land dominated by grass and bare soil that are used as military training grounds. The brush openings consist of tall grass and immature hawthorn. The wildlife openings are natural openings in the forests that are vegetated primarily by grass. Land

PAGE 26

14 impacts due to heavy military activities (e .g., infantry, artillery, wheeled, and tracked vehicle training) occur only in sele cted portions in these watersheds. Figure 2-1. Study watersheds, Bonham-1, Bonham-2, Bonham, Little Pine Knot, Randall, Sally, and Oswichee, in the Fort Benning military installation. Also shown are stream network and sampling locations. Characterization of Disturbance Categories A disturbance index (DIN) was defined to characterize the watersheds at Fort Benning into two disturbance categories: lowimpact and high-impact. A DIN is the sum of the area of bare ground on slopes greater than 3 degrees and on roads, as a proportion of the total watershed area (Mal oney et al., in press). The road areas were estimated by multiplying their length by the measured average width of 20 m. Percentage of bare ground was determined by using TM imagery and slope was derived from digital

PAGE 27

15 elevation maps. The TM imagery and digi tal elevation maps were obtained from Strategic Environmental Research and Development Plan (SERDP)Â’s Ecosystem Management Project (SEMP) database. Waters heds having a disturba nce index from 0 to 11% are designated as low-impacted waters heds. High-impacted watersheds have disturbance indices greater than or equal to 11%. Collection and Analysis of Stream Samples Surface water quality data were collected at seven streams biweekly from October 2001 to November 2002, and monthly therea fter to September 2003. Water samples were collected in high-density polyethylene bo ttles. Bottles were soaked in de-ionized water and rinsed with sample water prior to collection. The filtration was conducted at the sampling sites using 0.45 m pore size polyethersulfone membranes. Filtered sample was used to determine chloride (Cl) concentr ation, whereas raw sample was used for total suspended solids (TSS) determination. Unfiltered samples for analyzing total Kjeldahl nitrogen (TKN), total phosphorus (TP), and total organic carbon (TOC) were acidified using double distilled sulfuric acid. The stream water pH conductivity, and temperature were measured at the time of sampling. A ll samples were kept cool in an icebox, transported to the Soil and Water Science Depa rtment laboratory, University of Florida, and refrigerated until analyze d. All samples were analyzed using standard methods (American Public Health Association, 1992). Statistical Analyses The PearsonÂ’s correlation coefficients were calculated to examine the strength and significance of the rela tionships between a watershed physic al characteristic and a water quality parameter. Two-sample t-tests were pe rformed at 5% level of significance to test whether mean values of watershed physical characteristics and wate r quality parameters

PAGE 28

16 differ between lowand high-disturbance watersheds. Characteristics showing significant correlations with a water quality parameter were considered for stepwise multiple linear regression models. Only variables having less than or equal to 0.05 significance level were retained in the regression models. Results Watershed Physical Characteristics The watershedsÂ’ physical characteristics are summarized in Table 2-1. Most of the watersheds are highly vegetated (70% or mo re) except Oswichee (38%) with the majority characterized by pine and mixed pine and hard woods. Deciduous forest typically covers only a small percentage of these watersheds However, Bonham-1 consists of 27% of deciduous forest. The study watersheds range from less than 1 to 84 km2. The topographic characteristics of study watersheds are typical of forested watersheds of southeastern coastal plain (Lowrance, 1992; Pe rry et al., 1999). Average elevations vary from 104 to 148 m above mean sea level. Ma ximum slopes vary from 4 to 6 degrees. Sandy soils are common in most of the study watersheds. However, loamy soils cover most of the Sally and Oswichee watersheds. Bottomlands comprise 6 to 20% of the watershed. The military training extent (0 to 6%) is relatively small. Total bare lands in these watersheds comprise 9 to 21% of the wa tershed area, of which 1 to 8% of the total area is unpaved roads and trails. This extent and variability of military training and bare land are typical of the entir e Fort Benning installation. Some watershed characteristic s are strongly correlated at significance level of 0.05 or lower (Table 2-2). Significant positive corr elations exist for pine with the bottomland wetlands, deciduous vegetation with stream density, number of roads crossing streams with road length and percent of loam, and disturbance index with percent bare land.

PAGE 29

17 Negative correlations were found for mixed vege tation with percentage of loamy soil and number of roads crossing streams (NRC), military areas with normalized difference vegetative index (NDVI) and stream density, and DIN, and road density with NRC. Table 2-1. Physical character istics of study watersheds in Fort Benning, Georgia. Acronyms BON-1, BON-2, BON, LPK, OSW, Ran, and SAL represent Bonham-1, Bonham-2, Bonham, Little Pine Knot, Randal, and Sally, respectively. Watersheds Physical Characteristics B ON-1BON-2BON LPK OSW RAN SAL Topography Area, km2 0.762.2112.73 18.0183.39 74.3825.31 Average Elevation, m 121.8 133.5 125.5 146.3 104.2 136.8 136.8 Average Slope, degree 5.464.895.04 5.324.48 4.575.42 Vegetation Pine, % 28 30 40 41 26 58 48 Deciduous, % 27 6 8 2 0 3 12 Mixed, % 39 50 22 34 5 9 15 Wetland, % 6 8 9 17 7 20 10 Military Land, % 0 6 5 2 5 3 2 NDVI 0.360.300.32 0.340.35 0.340.36 Soil Sandy loam, % 78 69 69 72 24 68 49 Loamy sand, % 9 9 31 28 73 32 51 Road Road Length, km 3.6 11.4 51.6 56.6 196.6 415.1 97.6 Road Density, km/km2 4.8 5.1 4.1 3.1 2.4 3.1 3.8 Stream Stream Length, km 2.6 3.9 29.1 43.3 170.6 323.5 65.2 Stream Density, km/km2 3.4 1.7 2.3 2.4 2.1 2.4 2.6 Stream Order 2 2 4 4 5 6 4 Other No. of Roads Crossing Streams 1 2 13 11 55 43 21 Bare Land, % 1 11 11 4 4 4 7 Disturbance Index, % 11 21 19 11 9 10 15

PAGE 30

18Table 2-2. Pearson correlation co efficients between watershed ch aracteristics. Characteristic s are acronymed as follows: Pine forest (PIN), Deciduous forest (DCD), Mixed forest (MXD), We tland (WET), Military land (MIL), Sandy Soil (SND), Loamy Soil (LOM), Road Length (RDL), Road Density (RDN), Stream Density (STD), Normalized Difference Vegetative Index (NDVI), No. of Roads Crossing Stream s (NRC), % Bare Land (PBL), and Di sturbance Index (DIN). *, **, and *** indicates significance at or below 0.05, 0.01, and 0.001 probability levels, respectively. PIN DCD MXD WET MIL SND LOM RDL RDN STD NDVI NRC PBL DCD -0.25 MXD -0.43 0.40 WET 0.96*** -0.33 -0.42 MIL -0.20 -0.65 0.00 -0.18 SND 0.21 0.44 0.67 0.17 -0.30 LOM 0.06 -0.50 -0.85* 0.14 0.15 -0.93*** RDL 0.63 -0.47 -0.74 0.50 0.05 -0.28 0.41 RDN -0.26 0.65 0.81* -0.34 -0.01 0.63 -0.81* -0.64 STD 0.01 0.84* 0.04 -0.04 -0.92***0.34 -0.24 -0.15 0.17 NDVI 0.10 0.41 -0.48 0.07 -0.80* -0.30 0.41 0.22 -0.39 0.74 NRC 0.23 -0.58 -0.90**0.19 0.21 -0.76* 0.83* 0.81*-0.86**-0.27 0.33 PBL 0.00 -0.28 0.25 0.09 0.72 0.04 -0.11 -0.34 0.40 -0.67 -0.79* -0.30 DIN -0.10 0.05 0.52 -0.07 0.55 0.30 -0.42 -0.52 0.71 -0.44 -0.77* -0.58*0.93***

PAGE 31

19 Water Quality Parameters Water quality parameters in the study wa tersheds varied over the sampling period and among watersheds. Variab ility of water quality parameters among watersheds is observed (Figure 2-2). Mean pH in the study watersheds ranged from 4.2 to 7.0. Mean conductivity ranged from 16.4 to 44.5 S/cm. Mean temperatures varied from 17.5 to 20.8oC. Low concentrations of TP and TKN were observed in all the watersheds under study as compared to forested watersheds in the southeastern coas tal plain watersheds (Lowrance et al., 1984), and across the Un ited States (Meader and Goldstein, 2003; Fisher et al., 2000). Concentr ations of TKN, TP, and Cl we re often below the detection limit. Mean concentrations of TP vari ed widely, ranging from 0.003 to 0.020 mg/L and TKN varied from 0.20 to 0.35 mg/L; TOC fr om 1.35 to 3.33 mg/L; Cl from 1.46 to 4.13 mg/L; and TSS from 4.15 to 10.30 mg/L. As depicted in Figure 2-2, each stream exhibited distinct water quality signatures with the exception of temperature and TSS. Seasonal variations in the wa ter quality parameters are re sponsible for much of this observed variability among watersheds. Stream pH fluctuated more during J une and July and was elevated during December through February. Conductivity values showed slight fluctuations from May to July. On multiple occasions, high conductiv ity values were observed in the Randall stream. Chloride, TKN, and TP showed distin ct seasonal patterns. The concentrations of these parameters were low from June to September, and high from March to May and from October to December. In contrast, TOC peaked from August to October and again from March to July. Higher concentrati ons of TSS were observed from July to September in all the streams.

PAGE 32

20 Figure 2-2. Box plots of water quality parameters Each plot consists of outliers, most extreme data, 75th, 50th, and 25th percentile values. Watershed IDs represent1: Bonham-1, 2: Randall, 3: Os wichee, 4: Little Pine Knot, 5: Sally, 6: Bonham, and 7: Bonham-2.

PAGE 33

21 Effects of Disturbance Categories Table 2-3 presents the t-test results of the comparison of physical characteristics and water quality parameters based on the watershed disturbance level. The only physical characteristic havi ng a significant difference ( = 0.04) between these two groups was DIN. While the low-impact wate rsheds tended to have higher chemical Table 2-3. t-test results for differences in mean values of watershed physical characteristics and water quality parame ters. indicates significance at or below 0.05 probability level. NS indi cates non-significant difference at the 0.05 probability level. Low-Impacted High-Impacted Mean SD Mean SD Watershed Characteristics Pine, % 38.3 14.8 39.3 9.0 NS Deciduous, % 8.0 12.7 8.7 3.1 NS Mixed, % 21.8 17.2 29 18.5 NS Wetland, % 14.5 9.5 16 7.2 NS Military Land, % 2.4 2.0 4.3 2.3 NS Sand, % 60.5 24.7 62.3 11.5 NS Loam, % 35.5 26.9 30.3 21.0 NS NDVI 0.35 0.01 0.32 0.03 NS Road Length, km 168 184 53.5 43.2 NS Road Density, km/km2 3.3 1.0 4.3 0.7 NS Stream Density, km/km2 2.6 0.6 2.2 0.4 NS No. of Roads Crossing Streams 27.5 25.6 12 9.5 NS Disturbance Index, % 10.2 0.9 18.3 3.1 Water Quality Parameters pH 5.6 1.3 4.5 0.3 NS Temperature, oC 19.3 0.9 18 0.3 NS Conductivity, S/cm 25.9 13.3 20.4 2.5 NS TKN, mg/L 0.3 0.03 0.2 0.05 NS TP, mg/L 0.011 0.006 0.007 0.003 NS TOC, mg/L 2.9 0.4 2.1 0.7 NS Cl, mg/L 2.4 0.5 1.8 0.3 NS TSS, mg/L 9.1 2.2 4.8 0.5

PAGE 34

22 concentrations than the high-impact waters heds, only TSS showed significant difference ( = 0.03). Even though the resu lts showed no significant sta tistical differences at a confidence level of 95%, the t-test results of all water quality parameters, except conductivity and TP, showed significant differe nces at 80% confiden ce interval between highand low-impacted watersheds. The relatively small sample size and natural variability among watersheds may have lim ited the ability to discern significance differences. Relationship between Watershed Physic al Characteristics and Water Quality Parameters Correlation and regression analyses were performed to identify relationships among the watershed physical characteristics and the water quality paramete rs. Table 2-4 shows that each water quality parameter had a significant relationship with one or more watershed physical characteristics (Table 24). The correlation results show that decreasing mixed vegetation increased pH and TP. Sandy and loamy soils had opposite effects on TP. An increase in sandy soil decr eased TP, whereas an increase in loamy soil increased TP. Increasing military land decrea sed TOC. Temperature, pH, conductivity, and Cl increased as the road length increase d. The number of roads crossing streams had positive correlations on pH and TP. Percent bare land was negatively correlated with TOC and TSS. Disturbance index was nega tively correlated with TKN and TOC. Graphical relationships provi de insight into nonlinear relationships that exist between indicators and response variables. Fi gures 2-3 to 2-6 show some of the most striking relationships between watershed characte ristics directly and/or indirectly affected by management of military lands and their e ffect on water chemistry. As the % military land increases, the TKN, TOC, and TSS decr ease in a linear fashion. However,

PAGE 35

23 Table 2-4. Pearson correlation coefficients between watershed characteristics and water quality parameters. Characteristics are acronymed as follows: Pine forest (PIN), Deciduous forest (DCD), Mixed forest (MXD), Wetland (WET), Military land (MIL), Sandy Soil (SND), Loamy Soil (LOM), Road Length (RDL), Road Density (RDN), Stream Density (STD), Normalized Difference Vegetative Index (NDVI), No. of Ro ads Crossing Streams (NRC), % Bare Land (PBL), and Disturbance Inde x (DIN). *, **, and *** indicates significance at or below 0.05, 0.01, and 0.001 probability levels, respectively. pH TemperatureConductivityTKNTP TOC Cl TSS PIN 0.36 0.78 0.54 0 -0.09 0.15 0.56 -0.14 DCD -0.47 -0.37 -0.40 -0.12 -0.33 0.39 0.13 0.42 MXD -0.79* -0.55 -0.38 -0.39 -0.83* -0.24 -0.49 -0.11 WET 0.24 0.62 0.44 0.11 -0.11 0.18 0.37 -0.29 MIL 0.09 0 -0.10 -0.50 0.10 -0.87** -0.47 -0.53 NDVI 0.32 0.11 0.10 0.60 0.46 0.80* 0.55 0.54 PBL -0.45 -0.25 -0.51 -0.72 -0.42 -0.81* -0.68 -0.87** DIN -0.65 -0.36 -0.60 -0.84*-0.63 -0.76* -0.65 -0.72 SND -0.50 -0.02 0.14 -0.19 -0.78* 0.15 0.06 0.15 LOM 0.57 0.16 0.03 0.39 0.81* 0.07 0.07 -0.14 RDL 0.94*** 0.95***0.82* 0.19 0.57 0.09 0.79*0.30 RDN -0.75* -0.42 -0.54 -0.72 -0.78* -0.35 -0.32 -0.15 NRC 0.93*** 0.59 0.51 0.36 0.90**0.06 0.47 0.19 STD -0.14 -0.11 -0.01 0.39 -0.04 0.82* 0.44 0.65 disturbance index may operate as a thres hold indicator of pH and TSS where pH decreases in response to a relatively low level DIN while the TSS threshold for DIN impact is somewhat higher. No signifi cant relationships are found between the water quality parameters and extent of pine and deciduous forest. Similarly, wetland showed no effect on these parameters. Stepwise multiple regressions identified relationships between the water quality parameters and watershed physical characteristic s that are susceptible to the disturbances (Table 2-5). A statistical ly significant regression mode l was found for every water quality parameter. Prediction of pH variab ility among watersheds is particularly well captured by pine forest and road length. The regression relationships indicate that all of

PAGE 36

24 the water quality parameters depend on at least one aspect of military management. Several water quality parameters, Cl, TP, TOC, TSS, and TKN, depend only on management aspects of the military installati on. For example, Cl depends on change in military land and road length. Total phosphorus is strongly related to the number of roads crossing streams. However, the influen ce of vegetation and soils characteristics is clearly important in pH, c onductivity, and temperature. For example, conductivity appeared to be well captured by area cove red by sandy soil and the number of roads Table 2-5. Stepwise multiple regression models for water quality parameters. pH is unitless, temperature is measured in degrees centigrade, conductivity is measured in mS/cm, TP, TKN, TOC, Cl, and TSS are measured in mg/L. Pine forest (PIN), Military land (MIL), Sandy Soil (SND), Loamy Soil (LOM), Road Length (RDL), Road Density (RDN), No. of Roads Crossing Streams (NRC), Percent Bare Land (PBL ), and Disturbance Index (DIN) are the independent variables retained in the regression analyses. *, **, and *** indicates significance at or belo w 0.05, 0.01, and 0.001 probability levels, respectively. Water Quality Independent Variables Retained and Parameters Regression Equations R2 Pine, Road Length pH 5.50 – 0.0382 PIN + 0.00924 RDL 0.98*** Military, Road Length Cl 2.19 + 0.145 MIL + 0.00344 RDL 0.90** No. of Roads Crossing Streams TP 0.00451 + 0.000236 NRS 0.81** Sandy Soil, No. of Roads Crossing Streams Conductivity 24.2 + 0.552 SND + 0.666 NRS 0.80* Loamy Soil, Road Density Temperature 26.1 – 0.051 LOM – 1.49 RDN 0.77* Military Land TOC 3.45 – 0.274 MIL 0.76** Percent of Bare Land TSS 11.2 – 0.648 PBL 0.76** Disturbance Index TKN 0.445 0.0109 DIN 0.70*

PAGE 37

25 Figure 2-3. Relationships between military la nd and water quality parameter. Vertical bars represent standard errors.

PAGE 38

26 Figure 2-4. Relationships between road densit y and water quality parameter. Vertical bars represent standard errors.

PAGE 39

27 Figure 2-5. Relationships betw een number of roads crossing streams and water quality parameter. Vertical bars represent standard errors.

PAGE 40

28 Figure 2-6. Relationships between disturba nce index and water quality parameter. Vertical bars represent standard errors.

PAGE 41

29 crossing streams, whereas temperature was cap tured by loamy sand soil and road density. Overall, the regression models s how that it is possible to qu antify the effects of watershed physical characteristics on th e water quality parameters. Discussion Variations in stream water chemistry among the study watersheds reflect differences in biogeochemical reactions occurrin g in the watersheds. The results of this study indicate that even in low impact waters heds, physical characteristics may be used to explain variations in stream water chemistry and, by infere nce, the relative watershed disturbance levels. The observed variability in many of the chemical parameters studied in these watersheds can be attributed to physic al characteristics of the watersheds or land management patterns within the watersheds as evidenced by the road network, forestry practices, and military training. Diverse human activities interact to aff ect conditions in watersheds and water bodies. Sites of interest can be grouped a nd placed on a gradient according to activities and their effects. The results of this study suggest that th e vegetation type, road length, number of roads crossing streams, and dist urbance index are impor tant predictors of water quality variability. Vegetation cover was related to stream pH and TP (mixed forest) and conductivity (pine forest). Howe ver, deciduous forest cover was not related to any of the water quality parameters sugge sting a limited effect in organic matter and nutrient production and vari ability as observed among Fort Benning watersheds. However some other subtle landscape changes resulted in relatively larger impacts on water quality parameters. Low-impact waters heds tend to produce higher concentrations of nutrients in the streams. This can be attrib uted to the availability of more soil organic

PAGE 42

30 matter and the rapid biogeochemical processes occurring in the low-impact watersheds as compared to the high-impact watersheds. It is extremely difficult to capture all asp ects of human influence in a single graph or statistical test. However, sometimes m eaningful chemical patterns can be lost by excessive dependence on the outcome of me nu-driven statistical tests (Karr and Chu, 1999). Figures 2-3 to 2-6 depict several different aspects of streamÂ’s chemical conditions against several measures of human influences, such as military land use, disturbance index, number of roads crossing streams, and ro ad density. The distribution of circles in most of these figures illustra te that a chemical metric i ndicates little about a condition simply because it does not correlate strongly with a single surrogate of that condition. However, where the relationship between human influence and streamÂ’s chemical response is strong, statistics and graph agree. The correlation tests identify linear relati onships between a chemical response and watershed characteristics. Weak statistica l correlations observed in these analyses may have missed important chemical patterns. For example, nonlinear patterns were observed for forest types, bare land (not shown here ), and disturbance inde x (Figure 2-6). The plots in Figure 2-6 show a step-function for TSS and pH. The scatter of this dataset shows little or no statistical si gnificance, but can be interp reted chemically. For TSS, those watersheds having a dist urbance index of 11% or lower had a higher level than those with a greater disturbance index. When a number of variable s interact to influence wa ter quality conditions, it may be difficult to explain observed variability in a single plot agai nst one dimension of human influence (Figure 2-4). Chemical responses were plotted against the road

PAGE 43

31 densities for various watersheds. The P earson correlation coefficient for TP was significant capturing human influence on this ch emical parameter. The response of TP is visibly distinguished from others. A similar discussion is true for military land with TOC (Figure 2-3); number of road s crossing streams with pH and TP (Figure 2-5); and disturbance index with TKN (Figure 2-6). The relationships between water quality pa rameters and physical characteristics indicate that disturbances in low nutrient fo rested environments decrease some chemical signatures. Watersheds with more roads, e.g., Randall and Oswichee, have relatively high pH, conductivity, and Cl compared to the wa tersheds with fewer roads. Watersheds with a small portion of military land, e.g., B onham-1, Sally, and Little Pine Knot, have relatively high TOC concentrations. In contra st, watersheds characterized by higher road densities, e.g., Bonham and Bonham-2, had low TP concentrations. Higher disturbance index, similar to the road density, showed lower TKN and TOC concentrations in the streams. Mixed vegetation, road length, percen t of bare land, DIN, and number of roads crossing streams were able to capture most of the variability in water quality parameters. In a watershed scale study conducted in Ontario, Canada, Sliva and Williams (2001) found a negative correlation of forested land cover with TSS and chloride. In contrast, Johnson et al. (1997) showed a posi tive relationship of forest with TSS in a study of landscape influence on water chemistr y in the Saginaw Bay watershed of central Michigan. Johnson et al.Â’s (1997) results in dicated that row crop agriculture had the highest effect on total nitrogen, nitrate, and total dissolved solids. They also observed that urban and forest areas were positively correlated good predictors of TSS, whereas row crop agriculture was positively correlated with total nitrogen. Basnyat et al. (1999)

PAGE 44

32 reported a positive association of TSS with agricultural practices in the Fish River watershed, Alabama. The Fort Benning inst allation is characterized by relatively low variability in forest cover and suggests, in co ntrast to other studies that neither TSS nor Cl may be related to forest cover under ex isting land management pr actices. Instead, Fort BenningÂ’s road extent and percent bare land are better predictors of TSS and Cl. Most studies identified urban land use as a dominant factor caus ing elevated total nitrogen and nitrate concentrations in th e streams (Hill, 1981; Osborne and Wiley, 1988). Sponseller et al. (2001) found positive correlation of tota l inorganic nitrogen with percentage of non-forested land in southw estern Virginia watersheds. A negative correlation of TKN to DIN, in this study, is consistent with studie s (e.g., Sponseller et al., 2001; Hunsaker and Levine, 1995; Johnson et al ., 1997) that have shown stream nitrogen concentrations to be good pr edictors of non-forest area at the watershed scale. In a study of 101 watersheds in New Zealand, Close and Davis-Colley (1990) found that between 60 and 80% of the varian ce in conductivity, total nitrogen, and nitrate was accounted for by landscape factors including geology and land use. However, in that same study, landscape factors accounted for onl y 50% of the varian ce in ammonia and phosphorus species. Their result s parallel those found at Fort Benning in that no strong relationships were observed between land use and TP. The strong ne gative correlation of loamy sand soil with TP in Fort Benning is consistent with HillÂ’s (1981) study, conducted in a sandy loam region similar to portions of Fort Benning waters heds that reported negative correlations between phosphorous c oncentrations and abandoned farmland and forest.

PAGE 45

33 Most variations in stream water chemistry are driven by climatic and biotic factors and are therefore largely governed by the processes that are taking place in the terrestrial part of the watershed such as natural or human induced vegetation cover changes (Semkin et al., 1994). Our results show in teractions among landscape factors and water quality indicators. Results also indicate that it is possible to observe the response of these water quality parameters to physical attributes of watersheds. The importance of water quality parameters in the pres ent study appeared to be attr ibutable to the perturbations related to military training and associated parameters within a watershed as these parameters clearly captured the changes in physic al parameters that are more sensitive to such kind of influences. Conclusion and Recommendations Sometimes a single variable can capture and integrate multiple sources of influence. More often, a small number of ecological attributes pr ovide reliable signals about ecological condition. Water chemis try prediction using watershed physical characteristics in this study showed mixed re sults compared to the other investigations. However, most of the watershed physical char acteristics used in our analysis did explain the variability in water quality parameters. This study documented strong relationships between certain watershed physical characteri stics that are more susceptible to human induced perturbations, specifically military related disturbances, and water quality parameters in military installation at Fort Be nning. Watersheds with more roads crossing streams tended to produce more TP. Total Kjeldahl nitrogen and TOC variations were well captured by DIN and extent of military land, respectively. Total suspended solidsÂ’ variability, on the other hand, was captured by the percent of bare land within a watershed. Road length captured most of the variability in pH and Cl. Conductivity and

PAGE 46

34 temperature values were depende nt on soil types and road char acteristics. The variations in stream water chemistry are largely attribut able to disturbance le vels and the types of biogeochemical reactions occurring in the wa tersheds. Regression results suggest that TOC, TKN, and TSS were usef ul indicators of watershed phy sical characteristics as they are more susceptible to direct effects of military activities. Although pH, conductivity, and TP showed good correlations with the road length, these parameters indicated strong but indirect influence of military training activities on watersheds. Foreseeing a single indicator of water quality that would be sensitive to all kinds of perturbations in the watershed is extremely difficult. Ability to detect perturbations can be related to spatial and temporal scales. It is necessary to re cognize the effects of natural disturbances on ecosystem structur e and functioning. It is suggested that priorities for determining ecological indicat ors specific to water quality should include (1) development of framework to determine proper reference states within watershed against which to detect loss of ecosystem health, (2) broadening our knowledge of ecosystem sensitivity to perturbations of va rying intensity, spatio -temporal distribution, and type, and (3) development of suites of indicators necessa ry to detect the broadest spectrum of perturbations in watersheds.

PAGE 47

35 CHAPTER 3 HYDROLOGIC INDICES OF WATERSHED SCALE MILITARY IMPACTS IN FORT BENNING, GA Introduction A goal of stream flow char acterization and classificati on is to develop hydrologic indices that account for charac teristics of streamflow vari ability that are biologically relevant (Olden and Poff, 2003). Broadly, indi ces are attributes th at respond in a known way to a disturbance i.e., they relate key eco logical responses to human activities. Index identification is based on the goals and objectiv es set for a particular ecosystem or region. A good index should be sensitive to stressors, biologically a nd socially relevant, broadly applicable to many stressors and sites, diagnos tic of the particular stressor causing the problem, measurable, interpretable, and not redundant with other measured indices (Cairns et al., 1993). Ecologically relevant hydrologic indices deve loped in the past not only characterize particular regions, but also quantify flow ch aracteristics that are sensitive to various forms of human perturbations. For exampl e, early studies on hydrological indicators focused on variation of mean daily flow to study the pattern of fish in Illinois and Missouri (Horwitz, 1978). In Great Britain, Moss et al. (1987) used average flow conditions to predict macro-invertibrate fa una of unpolluted streams and Townsend et al. (1987) examined persistence of community structure for benthi c invertebrates. In arid regions of southwest United St ates, Minckley and Meffe (1987) studied effects of shortterm flood frequency in stream fish commun ities. Poff and Ward (1989) used long-term

PAGE 48

36 discharge records (17-81 years) of 78 streams from across the continental United States to develop a general quantitativ e characterization of streamflow variability. Similarly, Jowett and Duncan (1990) studied skewness in flows and peak discharges in relation to in-stream habitat and biota in New Zealand. More recent investigations have begun to focus on examining suites of hydrologic indices that are ecologically relevant to quantify hydrologic regimes. These studies report numerous such indices. For example, Poff and Allan (1995) studied stream fish assemblage for 34 sites in Wisconsin and Minnesota in conjunction with long-term stream flow variability and predictability as well as frequency and predictability of high and low flow extremes. In the process of deriving ecologically relevant hydrologic indices, Clausen and Biggs (1997) identified th irty-four hydrological variables from daily flow records at eighty-three New Zealand site s. The authors related these variables to benthic biota including periphy ton and invertibrate species richness and diversity. Wood et al. (2000) reported the importance of hydrological conditions in explaining the ecological role when the authors studied the changes in macro-inver tibrate community in response to flow variations in the Little St our River in the United Kingdom. Pettit et al. (2001) described a method for assessing seasona lity and variability of natural flows and their influence on riparian vegetation in two contrasting river systems in western Australia. To isolate core flow variables for ecological studies, it is important to know not just the ecological relevanc e of the variables, but also the interrelationships among the variables in order to avoi d redundancy in the analyses (e.g., Clausen and Biggs, 2000; Olden and Poff, 2003). Hydrologic indices have been criticized for being overly

PAGE 49

37 simplified and lacking adequate biological relevance. Stre am ecologists are now facing difficulty in choosing appropriate and relevant ones from the available suit of indices. For example, the Indicators of Hydrologic Al teration (Richter et al., 1996) approach is commonly used for characterizi ng human modification of flow regimes, yet it contains 64 statistics (32 measures of central tenden cy and 32 measures of dispersion), many of which are inter-correlated (Olden and Poff, 2003). To date, characterization of streams or regions through determination and development of ecologically relevant hydrol ogic indices are based on long-term stream flow data. However, given the multitude of methods to characterize stream flow, past studies overlooked the value of storm flow data for the development of such ecologically relevant hydrologic indices. Flow characteri stics are especially important where changes in land use are anticipated and where alterations to the flow regime need to be assessed. The primary objective of the study is to identify hydrologi c indices that characterize the impact of military land ma nagement on watersheds in Fort Benning, Georgia. Towards this end, this study inve stigates both storm and annual hydrographs. it is hypothesized that, in addition to annual-ba sed indices, storm-based hydrologic indices are indicative of alteration in stream ecology. Here, a suite of event based hydrologic indices is proposed. Storm-based and the annua l-based indices are calculated and used to compare and contrast impacted watersheds with a reference wate rshed. Additionally, specific military land management practices ar e used to predict storm-based indices. Flow Regimes and Hydrologic Indices To assess hydrologic altera tions within an ecosystem Richter et al. (1996) developed a method to compute representa tive, multi-parameter suite of hydrologic characteristics that are of ecological rele vance, commonly known as Indicators of

PAGE 50

38 Hydrologic Alteration (IHA). Olden and Poff (2003) comprehensively reviewed currently available hydrologic indices for characterizing streamflow regimes and recommended non-redundant indices for various stream types that may differ in major aspects of ecological organization. Poff ( 1996) provides a comprehensive catalog of the stream types for small to mid-size relatively undisturbed streams, classified according to variation in ecologically rele vant hydrological characterist ics, in continental United States. The assessment of IHA as well as other studies (e .g., Poff and Allan, 1995; Clausen and Biggs, 1997; Wood et al., 2000; Pet tit et al., 2001; Olden and Poff, 2003) to identify hydrologic indices is ba sed on long-term flow data. An alternative approach to identify ecol ogically relevant hydrol ogic indices is to conduct an assessment based on the storm hydrogr aph. This approach is useful when long-term data for a particular stream or re gion are not available, when significant data gaps exist, or coincident r ecords are not available. Stor m hydrographs are traditionally described by characteristics including peak flow, total volume of direct runoff, and duration as shown in Figure 3-1. The time characteristics of the hydrograph and its relationship to the precipitation event are pr esented in Table 3-1. Towards the goal of characterizing the spatial vari ations of hydrologic conditions using storm-based indices that are ecologically relevant as well as sensitive to human infl uences, the ecological function of hydrologic characteri stics that are relevant to storm-based hydrologic indices are considered. Examination of the storm hydrographs reveal s numerous potential indices. A set of 18 storm-based ecologically relevant hydrologi c indices that characterize variation in water condition in individual watershed is proposed. Included proposed indices are

PAGE 51

39 Figure 3-1. Terms used to describe hyetogr aph and response hydrograph. Refer to Table 3-1 for the definitions of the terms. Table 3-1. Definitions of terms used to describe hyetograph and response hydrograph. Time instants Time durations tp0 = beginning of precipitation Tw = tpe tp0 = duration of water input tpc = centroid of precipitation Trl = tq0 tp0 = response lag tpe = end of precipitation Tr = tpk tq0 = time of rise tq0 = beginning of hydrograph rise Tlp = tpk tp0 = lag-to-peak tpk = time of peak discharge Tlc = tqc tpc = centroid lag tqc = centroid of response hydrograph Tb = tqe tq0 = time base tqe = end of response hydrograph Tc = tqe tpe = time of concentration response factor (ratio of direct runoff depth to precipitation depth) baseflow index (ratio of baseflow volume to total volume during an event), peak discharge, dimensionless numbers related to hydrograph response lag (Trl), time of rise (Tr), and time base (Tb), bankfull discharge and the slopes of rising and falling limb of the hydrographs. A flow

PAGE 52

40 with 1.67-year return interval is often recognized as bankfu ll discharge (Poff, 1996). The eighteen ecologically relevant hydrologic indices were divide d into four components of hydrologic regimes, magnitude, frequency, durati on, and rate of change, to statistically characterize hydrologic variation. Effects of Flow Magnitude on Stream Ecology This group includes 6 parameters (mean and coefficient of variation) related to response factor, baseflow index, and the peak discharge (qpk). The magnitude of the water condition at any given time is a measure of the availability or suitability of habitat and defines such habitat attri butes as wetted area or habita t volume, or the position of a water table relative to wetland or riparian plant rooting zone s (Richter et al., 1996). High flows maintain ecosystem productivity and di versity. For example, high flows remove and transport fine sediments that would otherw ise fill the interstitial spaces in productive gravel habitats (Beschta and Jackson, 1979). Floods import woody debris into the channel (Keller and Swanson, 1979), where it cr eates new, high-quality habitat (Moore and Gregory, 1988; Wallace and Benke, 1984). Floodplains and wetlands provide important nursery grounds for fish and export organic matter and organisms back into the main channel (Junk et al., 1989; Welcomme 1992). The scouring of floodplain soils revives habitat for plant species that germin ate only on barren, wetted surfaces that are free of competition (Scott et al., 1996) or th at require access to shallow water tables (Stromberg et al., 1997). Effects of Flow Duration on Stream Ecology The 6 parameters in this group measure the duration of all the ev ents considered in the analysis. The parameters included in th is group are the mean, and coefficient of variation of dimensionless indices Trl/Tlc, Tr/Tlc, and Tb/Tlc, where Trl, Tr, Tb, and Tlc

PAGE 53

41 corresponding to the response la g, time of rise, time base, and centroid lag of a storm hydrograph, respectively. The duration is the period of time associated with a specific storm condition that determines the differe nces in tolerance to prolonged flooding in riparian plants (Chapman et al., 1982). Ch anges in duration of inundation can alter the abundance of plant cover types (Auble et al., 1994). For example, increased duration of inundation has contributed to the conversion of grassland to forest along a regulated Australian River (Bren, 1992). For aquatic i nvertebrates and fishes prolonged flows of particular levels can also be damaging (Clo ss and Lake, 1996). Whether a particular lifecycle phase can be completed or the degree to which stressful effects such as inundation of a flood plain can accumulate may be asse ssed from the duration of time over which a specific water condition exists (Richter et al., 1996). Effects of Flow Frequency on Stream Ecology The two parameters in this group measure the number of occurrences of the magnitude of the stream flow condition with respect to bankfull discharge. These numbers of occurrences of bankf ull discharges are reported as percentages of the total events under analysis. Measures of exceedance of bankfull conditions have greater ecological importance. These flows regulat e numerous ecological processes within riparian as well as flood pl ain areas. Frequent, modera tely high flows effectively transport sediment through the channel (Leopo ld et al., 1964). Movement of sediment and organic resources such as detritus a nd attached algae revive the biological community and allow many species with fast li fe cycles and good col onizing ability to reestablish (Fisher, 1983). Consequently, th e composition and relative abundance of species that are present in a stream often reflect the freque ncy and intensity of high flows (Meffe and Minckley, 1987; Schlosser, 1985).

PAGE 54

42 Effects of Flow Rate of Change on Stream Ecology This group includes 4 parameters, mean rate of change in peak discharge in rising limb and falling limb of the storm hydrograph and their variabilities. Flow conditionsÂ’ rate of change, or flashiness, refers to the rate at which flow cha nges from one magnitude to another, can influence species persistence and coexistence. At the extremes, flashy streams have rapid rates of change, whereas stable streams have slow rates of change (Poff et al., 1997). The rate of change in wa ter conditions may be tie d to the stranding of certain organisms along the water's edge or in ponded depressions, or the ability of plant roots to maintain contact with phreatic water supplies (Richter et al., 1996). Non-native fishes generally lack the be havioral adaptations to avoi d being displaced downstream by sudden floods (Minckley and Deacon, 1991). Meffe (1984) documented that a native fish, the Gila topminnow ( Poeciliopsis occidentalis ), was locally extirpated by the introduced predatory mosquitofish ( Gambusia affinis ) in locations where natural flash floods were regulated by upstream dams, but the native species pers isted in naturally flashy streams. Data Collection Watershed Characteristics The Fort Benning study watersheds (F igure 3-2), Bonham-1 and Bonham-2, Bonham, Little Pine Knot, and Sally Branch (n amed for the creek/stream that drains the watershed), represent a range of the regionÂ’ s soils, topography, land use, and vegetation communities (Table 3-2). In addition to stan dard watershed characteristics, the military training land area and a dimensionless military disturbance index (DIN) were calculated for each watershed. The military land use include fire ranges, maneuver training areas for light, amphibious, and heavy training, air fi eld, artillery and mort ar firing points,

PAGE 55

43 Figure 3-2. Study watersheds, Bonham-1, Bonham-2, Bonham, Little Pine Knot, Randall, Sally Branch, and Oswichee, in the Fort Benning military installation. airborne drop zones, speciali zed non-live-fire training asse ts, and duded impact areas. A DIN is the sum of area of bare ground on slope s greater than 3 degrees and on roads as a proportion of the total watershed area (Maloney et al., in press). Additional details regarding the study area are f ound in Bhat et al., Ecologica l indicators in forested watersheds in Fort Benning, GA: Relations hip between land use and stream water quality, submitted to Ecological Indicators; hereinafter referred to as Bhat et al., submitted manuscript). According to variations in ecologically re levant hydrological char acteristics that are based on flow variability and predictability, and lowa nd high-flow extremes, the

PAGE 56

44 Table 3-2. Physical charact eristics of study watersheds. Acronyms BON-1, BON-2, BON, LPK, and SAL represent the streams (or watersheds) Bonham-1, Bonham-2, Bonham, Little Pine Knot, and Sally Branch, respectively. Physical Characteristics BON-1 BON-2 BON LPK SAL Topography Area, km2 0.76 2.21 12.73 18.01 25.31 Average Elevation, m 121.8 133.5 125.5 146.3 136.8 Average Slope, degree 5.46 4.89 5.04 5.32 5.42 Vegetation Pine, % 28 30 40 41 48 Deciduous, % 27 6 8 2 12 Mixed, % 39 50 22 34 15 Wetland, % 6 8 9 17 10 Military Training Land, % 0 6 5 2 2 NDVI 0.36 0.30 0.32 0.34 0.36 Soil Sandy loam, % 78 69 69 72 49 Loamy sand, % 9 9 31 28 51 Road Road Length, km 3.6 11.4 51.6 56.6 97.6 Road Density, km/km2 4.8 5.1 4.1 3.1 3.8 Stream Stream Length, km 2.6 3.9 29.1 43.3 65.2 Stream Density, km/km2 3.4 1.7 2.3 2.4 2.6 Stream Order 2 2 4 4 4 Other No. of Roads Crossing Streams 1 2 13 11 21 Bare Land, % 1 11 11 4 7 Disturbance Index, % 11 21 19 11 15 streams in the study watersheds are classified as ‘perennial runoff’ (Poff, 1996). As the present study focuses on the impacts of military training within an ecosystem as well as the effects of these impacts on the stream biot a, a reference watershed was identified to contrast with watersheds impacted by military training activities. Approximately 94% of the area of Bonham-1 watershed is covered by forest. Mechan ized military activities are

PAGE 57

45 not conducted in this watershed as compared to 2-6% of the total area of other watersheds used for the same purpose. Also, the wate rshed has a small percentage of bare land, limited roads, and only one road crossing the stream. Hence, Bonham-1 was used as a reference watershed for this study. Precipitation and Streamflow Data Streamflow and precipitation were m easured from January 2000 to December 2003. Precipitation was measured by twelve tipping bucket rain gauges distributed throughout the study area. Watershed precip itation was determined by areal weighting using the Thiessen polygon method. Daily discharge values for Bonham-1 and Bonham2 were calculated from ten-minute continuous stage records using rating curves. Stream stage and velocity were measured half-hour ly for Bonham, Little Pine Knot, and Sally Branch. These data were used to calculat e daily discharges us ing the area-velocity method. Methods Annual-Based Hydrologic Indices The annual-based approach uses multi-year st reamflow records to define a series of ecologically relevant hydrologic indices. Th ese indices may be used to characterize intra-annual variation in wate r conditions, analyze tempor al variations, and compare impacts of alteration among watersheds. For this assessment, nonredundant yet biologically significant hydrol ogic indices are adapted as per the recommendations of Olden and Poff (2003). Table 3-3 lists these indices for the perennial runoff type of

PAGE 58

46 Table3-3. Summary of the recommended hydrologi c indices for 'perennial runoff' type of streams (after Olden and Poff, 2003) us ed in the Annual-Based Hydrologic Indices. BON-1, and BON-2 represent Bonham-1 and Bonham-2, respectively. BON-1 is reference (REF ) and BON-2 is (IMP) watershed. To maintain the consistency to the past studies, the symbols presenting the hydrologic indices in this table are kept as the same as those used in Olden and Poff (2003). Data types A, M, a nd D stand for annual, monthly, and daily discharge data, resp ectively. For definitions and method of calculation of the indices, refer to Appendix A. Flow components and name BON-1BON-2 Data of the hydrologic index SymbolUnits (REF) (IMP) type Magnitude of flow events Average flow conditions Variability in December flow MA26 unitless 0.70 0.47 M Mean annual runoff MA41 m3/s/km20.00397 0.00504 A Spreads in daily flows MA10 1/m3/s 81.4 43.5 D Low flow conditions Baseflow index ML17 unitless 0.12 0.57 A Mean of annual minimum flows ML14 unitless 0.17 0.58 A Median of annual minimum flows ML16 unitless 0.17 0.79 A High flow conditions High flow volume MH23 days/no. of years 9.2 0 A Mean maximum flow in May MH8 m3/s 0.016 0.017 M Median of annual maximum flows MH14 unitless 63.4 4.9 A Frequency of flow events Low flow conditions Frequency of low flow spells FL3 year –1 0.75 0 A Variability in low flood pulse count FL2 unitless 0.31 0.26 A High flow conditions High flood pulse count FH4 year –1 6 0 A Three times median flow frequency FH6 year –1 10.25 2.50 A Seven times median flow frequency FH7 year –1 6 0 A

PAGE 59

47 Table 3-3. Summary of the re commended hydrologic i ndices for 'perennial runoff' type of streams (after Olden and Poff, 2003) used in the Annual-Based Hydrologic Indices (Continued). Flow components and name BON-1 BON-2 Data of the hydrologic index SymbolUnits (REF) (IMP) type Duration of flow events Low flow conditions Variability in annual minima of 90-day means of daily discharge DL10 unitless0.78 0.32 D/M/A Variability in low flow pulse duration DL17 unitless0.09 0.07 A Variability in annual minima of 1-day means of daily discharge DL6 unitless0.60 0.28 D/M/A High flow conditions Means of 30-day maxima of daily dischargeDH13 unitless5.59 1.84 D/M/A Variability in high flow pulse duration DH16 unitless0.006 0.005 A Flood free days DH24 days 56.0 61.5 A Timing of flow events Average flow conditions Constancy TA1 unitless0.28 0.44 D Seasonal predictability of flooding TA3 unitless0.70 0.84 M High flow conditions Seasonal predictability of non-flooding TH3 unitless0.40 0.41 M Rate of change in flow events Average flow conditions Variability in reversals (Positive) RA9 unitless0.04 0.07 D Variability in reversals (Negative) RA9 unitless0.06 0.12 Change of flow (Decreasing) RA7 m3/s 0.1136 0.0324 D Change of flow (Increasing) RA6 m3/s 0.1119 0.0496 D streams found in the study region. The defi nitions and the methods used to calculate these indices are listed in Appendix A. In th is study, the annual-based indices are used to compare and contrast the Bonham-1 and Bonham-2 watersheds.

PAGE 60

48 Storm-Based Hydrologic Indices The storm-based approach uses the stor m hydrographs to determine the indices. Hydrograph separation was used to identify distinct storm events. 44-100 storm events from 2001 to 2003 were used to calculat e the response factor, baseflow index, dimensionless indices (Trl/Tlc, Tr/Tlc, and Tb/Tlc), watershed area (A) scaled peak discharge (qpk/A), bankfull discharges, and the rate of change of peak discharges in rising and falling limbs for five watersheds, where Bonham-1 is the reference watershed (Table 3-4). Appendix B lists the indices, defi nitions, and the methods of calculation. Statistical Analyses The ANOVA and the TukeyÂ’s multiple comparison tests were performed to determine the differences between referen ce watershed and the impacted watershedsÂ’ mean values of indices. The percent of fore st extent, military training land, road density, the number of roads crossing streams, bare land fraction, and dist urbance index were considered to assess the effects of distur bance on hydrology and th e potential impact on stream ecology. PearsonÂ’s correlation coeffi cients were calculated to examine the strength and significance of the relationships between a wa tershed physical characteristic and storm-based hydrologic index. Stepwise multiple linear regressions were performed to identify relationships between an i ndex and management related watershed characteristics. Only variables having p-valu es less than or equal to 0.05 were retained in the regression models. Results Watershed Disturbance Characteristics The watershedsÂ’ physical characteristics ar e summarized in Table 3-2. Most of the watersheds are highly vegetated (70% or mo re) with the majority characterized by pine

PAGE 61

49 Table 3-4. Summary of hydrologic indices used in the St orm-Based Hydrologic Indices. BON-1, BON-2, BON, LPK, and SAL repr esent the streams (or watersheds) Bonham-1, Bonham-2, Bonham, Little Pine Knot, and Sally Branch, respectively. BON-1 is reference (REF ) and other watersheds are impacted (IMP). represents the mean values of the indices of the impacted watersheds that are different at significance le vel of 0.05 from the mean values of reference watershed as confirmed by TukeyÂ’s multiple comparison test. For definitions and method of calculation of the indices, refer to Appendix A. Flow components and name REF IMP of the hydrologic index Symbol Units BON-1BON-2BON LPK SAL Magnitude Mean response factor MMRFunitless0.0300.010 0.060* 0.001*0.020 Variability in response factor MVRFunitless1.09 1.45 1.18 0.89 1.30 Mean baseflow index MMBFunitless0.73 0.86* 0.95* 0.90*0.95* Variability in baseflow index MVBFunitless0.19 0.11 0.04 0.07 0.04 Mean peak discharge MMPDm3/s/km20.0770.025* 0.030* 0.001*0.014* Variability in peak discharge MVPDunitless2.0 2.3 1.2 1.0 1.4 Frequency Bankfull discharge F1FD % 90 47 2 29 6 Two times bankfull dischargeF2FD % 82 32 0 10 0 Duration Duration of time base DMTBunitless1.9 1.7 1.2* 1.3* 1.2* Variability in time base DVTB unitless0.4 0.6 0.6 0.5 0.5 Duration of response lag DMRLunitless0.5 0.6 1.1* 0.8 1.3* Variability in response lag DVRL unitless0.8 1.4 0.6 0.5 0.4 Duration of time of rise DMTRunitless0.5 0.5 0.5 0.5 0.5 Variability in time of rise DVTR unitless0.5 0.8 1.0 0.9 0.7 Rate of change Mean slope of rising limb RPPD m3/s/hr 0.20 0.06* 0.10 0.01*0.04* Variability in rising slopes RPVPDunitless2.2 2.4 1.4 1.4 1.6 Slope of falling limb RNPD m3/s/hr 0.0500.020 0.070 0.001*0.020 Variability in falling slopes RNVPDunitless2.4 2.0 1.5 1.2 1.2

PAGE 62

50 and mixed pine and hardwoods. Deciduous forest typically covers only a small percentage of these watersheds. Howeve r, Bonham-1 consists of 27% of deciduous forest. The study watersheds range from less than 1 to 25 km2. Average elevations and maximum slopes are relatively constant. Sa ndy and loamy soils are common in most of the study watersheds. The military training land (0 to 6%) is relatively small, but varies among watersheds. Total bare lands in these watersheds comprise 11 to 21% of the watershed area, of which 1 to 8% of the total area is unpaved roads and trails. This extent and variability of military training and bare land are typical of the entire Fort Benning installation. While road density is relativ ely comparable among watersheds, the number of roads crossing streams varies from 1 to 21. Military training land, bare land, and disturbance index are positively correlated at significance level of 0.05 or lower. Annual-Based Hydrologic Indices Table 3-3 summarizes annual-based analysis results for Bonham-1 (reference) and Bonham-2 (impacted) watersheds. The average flow conditions revealed that the mean annual flow (MA41) is higher in the impacted wate rshed. However, the reference watershed has higher flow va riability for both annual (MA10) and December month (MA26) periods. The average flow event timing is more constant (TA1) and predictable (TA3) in the impacted watershed. The impacted watershed maintained a higher magnitude of minimum flows as depicted by the baseflow index (ML17). This result is consistent across other low flow indices (ML14 and ML16) and the findings that the impacted watershed had no low flow spells (FL3) and lower variability of low flow pulse counts (FL2). Higher coefficient of variation in annual minima of 90-day (DL10) and 1-day (DL6) means of daily discharge for the reference watershed and similar va riability in low flow pulse duration (DL17)

PAGE 63

51 were found for both watersheds. While low fl ow values vary more for the reference watershed, once the flow goes low, it stays lo w for same duration in both watersheds. The reference watershed produces higher magn itude flow and thus maintains higher median flow (MH14) during events. However, in the month of May, mean of the maximum monthly flows (MH8) were similar for both the watersheds. During high flow conditions, the reference watershed crosses a threshold of seven times the median annual daily flow volume (MH23). In the impacted watershed, these floods never occurred. The high flood pulse counts (FH4) and the frequency of floods (FH6 and FH7) in the reference watershed are more than the impacted watershed. However, on a few occasions, the flow crossed a lower threshold of 3 tim es median frequency of flood (FH6) in the impacted watershed. The 30-day floods (DH13) went higher and stayed high (DH16) in the reference watershed. Periods between floods (DH24) are similar for both watersheds, that is, they have approximately two months of flood free days a year and of comparable predictability (TH3). Storm-Based Hydrologic Indices The results of storm-based hydrologic a ssessment are summarized in Table 3-4. Analysis of variance tests indi cated that the mean values of storm-based indices except the time of rise differ at th e significance level of 0.05. Fo r a number of indices, the reference watershed exhibits distinct behavior as compared to the impacted watersheds. TukeyÂ’s multiple comparison tests indicate that the indices other than time of rise and the rate of change in falling limb in the refere nce watershed were signi ficantly different from the impacted watersheds. The reference wa tershed is characterized by a relatively low baseflow index (MMBF) with significantly higher (MMPD) and more variable (MVPD) peak discharge. During events, 90% of the total events produced greater than bankfull

PAGE 64

52 discharge (F1FD) in the reference watershed indi cating a highly connected system as compared to 2-47% in impacted watersheds. Storm flows consistently last ed longer and responded faster to rain events in the reference watershed (DMTB and DVTB). Once the stream responded, the time of rise (MMTR) was similar in all the watersheds. The reference watershedÂ’s combination of fast response and high peak discharge resu lts in a rapidly increasing rising limb (RPPD) as compared to impacted watersheds. Relationship between Military Land Ma nagement and Storm-Based Hydrologic Indices Correlation and regression analyses were performed to determine the relationship among the watershed physical characteristics and the storm-based hydrologic indices. Table 3-5 indicates that 7 ke y storm-based hydrologic indices are significantly related to military land management. Increased military training land, bare land, and the disturbance index will increase th e time of rise as well as the variability in the time base. Increasing the road density increases the vari ability in the time base and the rate of change of rising limb. Increasing the numbe r of roads crossing streams increases the storm response lag, but decreases the time base. Results also show that an increase in the number of roads crossing streams decreases the variability in the rate of change of falling limb. No effects on hydrologic indices were identified for forestry management practices. Stepwise multiple correlations characterized the response of storm-based indices to military impacts (Table 3-6). The greatest impact of land management is found with statistically significant predictive models for indices of time base, response lag, and time

PAGE 65

53 Table 3-5. Pearson correlation coefficients between waters hed physical characteristics and event based hydrologic indices. Char acteristics are acronymed as Forest (FOR), Military Training Land (MIL), Road Density (RDN), No. of Roads Crossing Streams (NRC), % Bare Land (PBL ), and Disturbance Index (DIN). and ** indicates significance at or below 0.05, and 0.01 probability levels, respectively. Hydrologic Indices FOR MIL RDN NRC PBL DIN MMRF 0.25 0.15 0.15 0.13 0.27 0.31 MVRF 0.40 0.59 0.71 -0.05 0.67 0.81 MMBF -0.62 0.44 -0.56 0.83 0.68 0.41 MVBF 0.63 -0.39 0.61 -0.85 -0.63 -0.35 MMPD 0.83 -0.37 0.64 -0.62 -0.47 -0.19 MVPD 0.66 0.21 0.96** -0.73 0.04 0.34 F1FD 0.57 -0.40 0.57 -0.85 -0.65 -0.39 F2FD 0.69 -0.43 0.62 -0.82 -0.64 -0.36 DMTB 0.63 -0.17 0.77 -0.91* -0.42 -0.12 DVTB -0.50 0.98** 0.14 0.08 0.96** 0.90* DMRL -0.23 0.06 -0.54 0.97** 0.41 0.18 DVRL 0.21 0.59 0.84 -0.78 0.33 0.57 DMTR -0.39 0.99** 0.32 -0.17 0.89* 0.90* DVTR -0.86 0.67 -0.45 0.35 0.68 0.47 RPPD 0.79 -0.34 0.59 -0.55 -0.42 -0.16 RPVPD 0.63 0.13 0.91* -0.78 -0.06 0.24 RNPD 0.42 0.10 0.35 -0.13 0.14 0.26 RNVPD 0.70 -0.07 0.85 -0.88* -0.29 0.03 of rise. Military training land, road dens ity, and the number of road crossing streams were the three management variables that impacted storm responses.

PAGE 66

54 Table 3-6. Stepwise multiple regression models for event based hydrologic indices. Military Training Land (MIL), Road Density (RDN), and No. of Roads Crossing Streams (NRC), are the inde pendent variables retained in the regression analyses. and ** indica tes significance at or below 0.05, and 0.01 probability levels, respectively. Hydrologic indices Independent variables retained and R2 (adj) regression equations MMBF 0.729 + 1.91 MIL + 0.00968 NRC 0.94* MVPD 1.21 + 0.666 RDN 0.88** F1FD 0.929 6.97 MIL 0.039 NRC 0.94* DMTB 1.8 0.0352 NRC 0.77* DVTB 0.418 + 3.05 MIL 0.95** DMRL 0.483 + 0.0393 NRC 0.92** DMTR 0.454 + 0.837 MIL 0.97** RNVPD 2.20 – 0.0563 NRC 0.70* Discussion Results representing annual-based av erage flow conditions showed higher magnitudes of MA26, MA41, and MA10 in the reference watershed as compared to the impacted one. In addition, these flows are less constant and le ss predictable over the years as compared to the impacted waters hed. Aquatic communities can show distinct differences to changes in velocity and reducti on in bed gradient, and associated fining of bed sediments (Clausen and Biggs, 2000). Periphyton and benthic invertebrates are particularly sensitive to diffe rent velocities and bed sedime nt size/stability (Minshall, 1984; Biggs, 1996). Clausen and Biggs (1997) found that invertebrate species richness and periphyton biomass changed based on flow These changes are likely related to riparian vegetation on the amount of leaf li tter input (Vannote et al. 1980; Davies-Colley and Quinn, 1998).

PAGE 67

55 The higher magnitude of the annual-based low flow indices (ML17, ML14, and ML16) in the impacted watershed is attributed to higher groundwater input as compared to the reference watershed. This increased input likely reflects the reduction of interception characteristic of the military land uses (Bryant et al., in press). The relative magnitude of low flows is likely to have important influences on biota through the intensity of habitat destruction associated wi th drying during low flows. In this study, the lower magnitudes of low flows as de picted by lower baseflow index (ML17) and smaller low flow pulses (FL3) suggest that the long periods of low flow condition are more likely in the reference watershed. Higher variabil ity in the duration during low flow condition (DL10, DL17, and DL6) in the reference watershed supports likelihood of such long periods of low flow conditions. Long pe riods of low flow conditions and higher variability in these conditions may provide se lective pressure for specific life history characteristics such as inve rtibrate aestivation and e gg diapause, and physiological tolerance to low dissolved oxygen (Williams and Hynes, 1977). The higher magnitude of the a nnual-based high flow indices (MH23 and MH14) and peak discharges (MMPD and MVPD) in the reference watershed as compared to the impacted watershed suggests the likelihood of habitat regenerati on associated with sediment transport and floodplain inundation during high flows. For high flow events, the degree of riverbed communitiesÂ’ disturba nce is strongly related to degree of bed movement (e.g., Biggs et al., 1999). Di ssolved inorganic nitrogen and phosphorus concentrations in rivers are strongly negativ ely correlated with specific yield and high flow magnitude among watersheds, and among years within watersheds (e.g., Biggs and Close, 1989; Close and Davies-Colley, 1990; Gr imm and Fisher, 1992). This reflects the

PAGE 68

56 degree of flushing of the nutrients minera lized through organic matter breakdown in the soil profile and leachate from the underlying substrata. Floods or high flow conditions are also important in influencing community structure. The results based on the annual flow in this study clearly show that the reference watershed produced more frequent high flows (FH4, FH6, and FH7). Similarly, the results from storm-based an alysis show that the frequenc y of discharges equaling or exceeding bankfull (F1FD and F2FD) is higher in the refere nce watershed. Floods are widely viewed as reset mechanisms (Resh et al., 1988), and flood-related mortality to lotic organisms can result either directly from scouring, crushing, or downstream export of individuals (Minckley and Maffe, 1987) or indirectly from food resources loss (Hanson and Waters, 1974). As flood freque ncy increases, some invertebrates in the reference watershed actively migrate either in to the substratum or to quieter backwaters to avoid sudden floods. Floods have been shown to regulate community structure by facilitating local coexistence between asymme trically competitive algal species (Power and Stewart, 1987) and invertibrate species (Hemphill and Cooper, 1983) and between an exotic fish predator and its native, rela tively flood-resistant prey (Meffe, 1984). The result in this study showed th e consistency of storm-based F1FD and F2FD with annualbased FH4, FH6, and FH7 indices suggesting F1FD and F2FD may be used as alternative indices. As indicated by the higher annual-base d mean duration of high flow condition (DH13) and the duration of storms (DMTB) in the reference as compared to the impacted watershed, it is apparent that water resides in the referen ce watershed for a longer period of time during high flow condition. However, the DMRL is smaller for the reference

PAGE 69

57 watershed suggesting a quick ri se in hydrograph. This can be attributed to a better connectivity of riparian areas to the stream. This connectivity is important as the nutrient concentrations can strongly control auto trophic production during inter-flood periods (e.g., Biggs, 2000). Also, indivi dual high flow events grea tly reduce the biomass and change the species composition of peri phyton (e.g., Biggs a nd Stokseth, 1996), and invertebrates (e.g., Cobb et al., 1992). Predictive relationships we re identified for storm-ba sed hydrologic indices based on watershed scale military management and suggest that the key variables related to hydrologic alteration are military training land, road density, and the number of roads crossing streams. Military training within a watershed can modify annual and storm generated runoff due to changes in drainage, vegetation, and soils. The impact is due to troop maneuvers and large, tracked and wheeled vehicles that traverse thousands of hectares in a single training exercise (Quist et al., 2003). These activitiesÂ’ impacts, ranging from minor soil compaction and l odging of standing vegetation to severe compaction and complete loss of vegetation cove r in areas with concentrated training use (Wilson, 1988; Milchunas et al., 1999), are evident in the stream hydrographs and suggest impacts to stream ecosystems. Di sruption in soil density and water content (Helvey and Kochenderfer, 1990), addition of sediment, nutrients, and contaminants in aquatic ecosystems (Gjessing et al., 1984) and impairment of natural habitat development and woody debris dynamics in forested floodplain streams (Piegay and Landon, 1997) are potential impacts on terrestria l and aquatic ecosystems. In a recent study in the same watersheds in Fort Be nning, Houser et al. (2 004) found that gross primary productivity, respiration, and the benthi c organic matter were low in the stream

PAGE 70

58 corresponding to the impacted watershed (highe r disturbance index) as compared to the reference one (lower disturbance index). Road crossings commonly found in military training areas may act as barriers to the move ment of fishes and other aquatic animals (Furniss et al., 1991). Roads also directly change the hydrology by intercepting shallow groundwater flow paths, diverting the water along the roadway and routing it efficiently to streams at crossings (Wemple et al, 1996). This can cause or contribute to changes in timing and routing of runoff (Jones and Grant, 1996). Conclusion and Recommendations In the present study, a reference watershe d was compared to impacted watersheds using annual and storm-based hydrologic in dices within the Fort Benning military installation. The results s uggest a subset of the hydr ologic indices recommended by Olden and Poff (2003) are necessary for the Fort Benning streams. It is recommended that mean annual runoff and its spread, base flow index, high flow volume, frequency of low flow spells, high flood pulse count, variab ility in annual minima of 90-day means of daily discharges, and constancy be used in stream ecology studies in the Fort Benning streams to identify disturbances relative to a reference waters hed at a watershed scale. The indices based on storm data may be us ed to augment the annual indices or as surrogate to those indices. Storm-based ma gnitude and variability in peak discharge, baseflow index, and the bankfull discharge were consistent with the results from annualbased analysis. With respect to the potent ial influence of the frequency and duration aspects of the flow regimes on the stream ecology, duration and vari ability of time base, and duration of response lag are identified as critical hydrologic indices. The military management practices, military training la nd, road density, and the number of road crossing streams, were found to sign ificantly affect these indices.

PAGE 71

59 CHAPTER 4 PREDICTION OF NITROGEN LEACHING FROM FRESHLY FALLEN LEAVES: APPLICATION OF RIPARIAN ECOSYSTEM MANAGEMENT MODEL (REMM) Introduction Forest canopy litterfall initiat es the major pathway for recycling of nutrients from plant to soil (Bubb et al., 1998). Nutrients are returned to the soil via litterfall and throughfall fluxes in forested ecosystem (U konmanaho and Starr, 2001). Litterfall, comprised predominantly of leaf litter, is usually the most important nutrient source (Herbohn and Congdon, 1998). Litter quantity, li tter decomposition and nutrient release patterns are factors for unde rstanding nutrient cy cling in forest ecosystems (Rogers, 2002). Litterfall is primarily caused by natura l senescence of leaves Forest litterfall quantity and composition varies among tree species, stand age and development, and reflects environmental conditions, particularly water and nutrient availability (Binkley, 1986; Polglase and Attiwill, 1992). The am ount and distribution of litterfall through time are also affected by season, rainfall amount and distribution, and wind speed (Crockford and Richardson, 1998). Fluxes of dissolved organic matter are an important vector for the movement of nutrients both within and between ecosyste ms (Cleveland et al., 2004). In many forest ecosystems, more than half of the nitrogen is soil solution is in or ganic form (Qualls et al., 1991). Hydrology plays an important role in both pr oduction and mobilization of dissolved organic matter in the forest floor (Park and Matzner, 2003). A substantial amount of potentially soluble organic matter exists in an adsorbed phase (Christ and

PAGE 72

60 David, 1996). The amount of percolating wate r mobilizes these sol uble organic matters (Tipping et al., 1999). Past studies have shown that the dissolv ed organic matter flux increases when the percolating water passes thr ough the forest floor (McDowell and Likens, 1988; Qualls et al., 1991; Currie et al., 1996; Mich alzik et al., 2001). In a rece nt study in India, Singh et al. (1999) reported that the ra pid mass loss of leaf litter o ccurred during the rainy season and the rate of litter mass loss was positively corr elated with the cumulative rainfall. In an early study, Bocock et al. (1960) observed a rapid leaching from oak and ash leaf litter, and suggested that the weight loss during the first month was largely due to a physical loss by leaching. The primary sources of dissolved nutrients are considered to be leaching of substance from fresh litter and the product of decomposition of the plant litters (Qualls et al., 1991). Water-soluble nutri ents are more easily leached from the leaf litter of deciduous species, includi ng birch, than from conifer ous species, including Scots pine ( Pinus sylvestris ) and Norway spruce ( Picea abies ) (Harris and Safford, 1996; Hongve, 1999). Depending on the species, leaf litter may release from 5 to 30% of original dry weight as dissolved organic matter within 24 hours (Cummins, 1974). In a woodland stream ecosystem in Massachusetts, McDowell and Fisher (1976) found that leaves lose weight rapidly by abiotic leaching of solubl e constituents for 1-3 days and decay more slowly thereafter as a result of microbial decomposition. Gosz et al. (1973) estimated 15% mass loss for yellow birch leaves ( Betula lutea ) due to leaching within first month on the forest floor. Heath et al. (1966) re ported a 26% weight loss from birch leaves ( B.

PAGE 73

61 alba ) that remained for two months on the fo rest floor. Nykvist (1961) found an 8% weight loss in B. verrucosa after 24 hours of leaching in distilled water. Annual fluxes of nitrogen in li tterfall are greater than ot her nutrients regardless of forest type but the amounts vary among fore st types (Ukonmaanaho and Starr, 1999). Dissolved organic nitrogen is the major form of nitrogen in stream water draining from many mature forest watersheds, comprising abou t 95% of the total nitrogen (Qualls et al., 1991). Nitrogen content of a hardwood fo rest litterfall on the Hubbard Brook experimental forest comprised approximatel y 37% of the total nutrients (Gosz et al., 1972). In a low nutrient system at the Cow eeta Hydrologic Laborator y in North Carolina, where carbon (C) to nitrogen (N ) ratios in freshly fallen litter varied from 100 to 220, Qualls et al. (1991) reported that the output of dissolved or ganic nitrogen from the forest floor was 28% of the nitrogen input from the litterfall. The Riparian Ecosystem Management Mode l (REMM) was developed by the U.S. Department of AgricultureÂ’s (USDA) Ag ricultural Research Service (ARS) to characterize the role of the riparian area on stream water quality (Lowrance et al., 2000; Altier et al., 2002). In REMM, th e riparian areasÂ’ water qual ity functions for control of N, phosphorus (P), and sediment transport into surface waters are simulated and analyzed. This model has primarily been te sted and applied on high-nutrient riparian buffers adjacent to agricultural fields in th e past (e.g., Inamdar et al., 1999a; Inamdar et al., 1999b; Lowrance et al., 2000). The complex dynamics related to the nutrien ts in a watershed is particularly of interest in low nutrient systems as the releas e of nutrients from plant litters prior to decomposition may be an important aspect of characterizing stream water quality. A

PAGE 74

62 watershed scale study related to surface water quality was conducted in the Fort Benning military installation located in southwest Georgia. Experimental data from Fort Benning shows an increase in total Kjeldahl nitrogen concentrations following litterfall. The release of nitrogen is relatively rapid as comp ared to that released from decomposition of organic matter. It is hypothesi zed that after leaves fall, the first precipitation events can be expected to result immediate nitroge n leaching and a corresponding increase in nitrogen levels in the stream water. The goa l of this study is to determine whether freshly fallen leaves in a riparian area are a significant source of nitrogen in low nutrient system. The specific objectives of this study are to quantify nitrogen in streamflow using REMM and verify the modeled results with the observed data. Basic Concepts of REMM Originally, REMM was developed to si mulate buffer systems that meet specifications recommended by the USDA Forest Service and the USDA Natural Resource Conservation Service as a national standard (Natural Re source Conservation Service, 1995). Three zones para llel to the stream characterize the riparian system. Thus REMM simulates a three-zone buffer. The ripari an areaÂ’s soil is characterized in three layers. In these soil layers, vertical a nd lateral movement of water and associated dissolved nutrients takes place. The top two soil layers correspond with the soil horizons A and B, respectively. Depth of the root z one or a restricting soil horizon defines the third soil layer. A litter layer that interact s with surface runoff covers the soil layer 1. The model takes upland water output supplied by the user and calculates loadings of water, nutrients, sediment, and carbon based on actual area of the zones of a riparian area (Lowrance et al., 2000).

PAGE 75

63 The model input data consist of the phys ical conditions of the riparian area, external loadings of water and nutrients to the riparian area, soil, vegetation, and soil information. Altier et al. (2002) documents a detailed description of the algorithms, equations, and parameters used in the model. Hydrology component of REMM REMM measures water stores and fluxes us ing a daily water balance. The water balance includes interception, evapotranspiration, infiltration, verti cal drainage, surface runoff, subsurface lateral flow, upward flux from water table, and deep seepage. Interception losses occur in th e vegetation canopy and litter la yer. Canopy interception is an exponential function of the canopy storage capacity and th e amount of daily rainfall, and is simulated using a modified form of the Thomas and Beasley (1986) equation. Precipitation falling through the canopy (throughfall) is subjec t to litter interception. Similar to canopy interception, litter intercepti on is determined by the mass of the litter at any given time and the litter storage capacity. Infiltration at the soil surface depends on the total depth of water. This depth of water is the sum of the throughfall depth and overland flow depth from the upslope zone or field. If the sum of incoming upland r unoff and throughfall depth is less than the infiltration capacity of the ri parian soil, runoff infiltrates at the rate of application. During high intensity storms, the runoff and throughfall rate may ex ceed the infiltration rate. Surface runoff from the riparian buffe r occurs when total water depth exceeds the infiltration depth. In such conditions, model simulates the infiltra tion using a modified form of the Green-Ampt equation (Stone et al., 1994). When soil moisture exceeds the field capaci ty, vertical drainage from a soil layer occurs. This vertical drainage also depends on the available water storage capacity of the

PAGE 76

64 receiving layer. The rate of drainage is co ntrolled by the lesser of the vertical hydraulic conductivities of the draining a nd receiving layer. Vertical unsaturated conductivity is simulated as a function of the soil moisture content using Campbell's equation (Campbell, 1974). When a water table builds up over the restricting soil layer, subsurface lateral movement occurs. Lateral water movement is simulated using DarcyÂ’s equation. Saturation excess overland flow occurs when soil profile is completely saturated. Nutrient component of REMM Simulation of nutrients in REMM is base d on the Century Model (Parton et al., 1987). The C cycle is fundamental for simu lation of all organic matter dynamics and many nutrient cycling processes in REMM. Release and immobilization of N is in proportion to transformations of C. Soil a nd litter C is divided into residue (woody debris, leaf litter, a nd roots) and humus (soil organic ma tter) pools. Each C pool has an associated mineralization rate, transformation efficiency, and a specific range of C to N ratios (Inamdar et al., 1999b). A first order rate equation with respect to C determines the mineralization of C from each pool. Litter from leaves, stems, branches, coarse roots, and fine roots are allocated into a readily decomposable (metabolic) or a resistant (structural) residue pool based on lignin to N ratio (Pastor and Post, 1986). Each pools decomposes at different rates. REMM assumes a fixed fraction of C in fr esh plant residue. Dissolved C from metabolic residue and active humus pools diffe r by litter and soil la yer (Altier et al., 2002). The sum of C in throughfall and inco ming runoff is mixed with C on the ground surface to determine dissolved and adsorbed concentrations. Dissolved C moves with surface runoff, subsurface lateral flow, and ve rtical drainage (Lowrance et al., 2000).

PAGE 77

65 The amount of dissolved C carried out of the litter or soil layer is a f unction of the water volume and the dissolved C concentration. Nitrogen is input to the riparian area through precipitation, surface and subsurface water flow, and adsorbed to incoming sediment carried by surface flow. Presence of nitrogen in different forms and the asso ciated degree of physical and chemical stabilization influences its availability for microbial transformations and plant uptake. Effective C to N ratios influencing residue d ecomposition are calculated in the model as a function of the content of N in the litter as we ll as the content of available inorganic N in the soil. As decomposition of the litter take s place, C and N are released (Lowrance et al., 2000). Inorganic nitrogen, ammonium and nitrate, are both available for immobilization into soil organic matter. Immobilization of nitrate occurs only after all available ammonium has been used. Amm onium may be in solution or adsorbed to the soil matrix. Nitrification is calculated with a first-orde r rate equation influen ced by temperature, soil moisture, and pH. Temperature and soil moisture are variables, recalculated each day but pH is a constant for each zone and laye r combination. With increasing amounts of substrate, the rate of nitrif ication is lagged to represent a delayed microbial response. The calculation of denitrificati on is a function of the interac tion of factors representing the anaerobic condition, temperatur e, nitrate, and mineralizable C. Denitrification mostly occurs as water-filled pore space gets a bove 60% (Inamdar et al., 1999b). Nitrogen transformation processes are shown in Appendix C.

PAGE 78

66 Application of REMM Study Area The study area consisted of a second-order watershed, Bonham-South (221 ha), in the Fort Benning military installation, Georgia (Figure 4-1). A detailed description of this site along with the instrumentation and the type of data collected is provided in Figure 4-1. Study area. Bonha m-South is a second-order wa tershed within the Fort Benning military installation in Georgia. Bryant et al., in press and Bh at et al., in review. The ri parian area covers approximately 2.5% of the total watershed area. The ripari an length is 3,500 m. Based on the visual inspection in the watershed, a 12 m wide riparian buffer along the stream is considered.

PAGE 79

67 This riparian buffer is divided into three e qual zones (Zone 1, Zone 2, and Zone 3) of 4 m each. Corresponding soil layers (Soil laye r 1, Soil layer 2, and Soil layer 3) are considered in each zone. A selective list of riparian area and hydrological parameters and their corresponding values are provided in Table 4-1. The riparian area consists primarily Table 4-1. REMM model parameters values by the riparian zone for the Bonham-South watershed in Fort Benning, Georgia. The dimension of riparian area from the upland to the stream perpendicular to th e stream is referred to as width, and the distance along the stream is referred to as length. The parameter values are identical for Zone 1, Zone 2, and Zone 3 except for the slope. The superscripts 1, 2, and 3 of the slope va lues represent the Zone 1, Zone 2, and Zone 3, respectively. Parameters Units Values Riparian zone length m 3,500 Riparian zone width m 4 Slope % 3.01, 3.82, 4.23 Manning's n unitless 0.06 Total soil profile thickness m 3.3 Individual soil horizon thickness m Soil layer 1 0.3 Soil layer 2 1.0 Soil layer 3 2.0 Saturated hydraulic conductivity cm/h Soil layer 1 15 Soil layer 2 13 Soil layer 3 1 Porosity cm/cm Soil layer 1 0.40 Soil layer 2 0.35 Soil layer 3 0.35 Wilting point cm/cm Soil layer 1 0.07 Soil layer 2 0.07 Soil layer 3 0.07 Riparian area ha 4.2 Field drainage area (surface) ha 216.8 Field drainage area (subsurface) ha 221.0

PAGE 80

68 of hardwoods (approximately 60%) that are mostly scrub oak ( Q. rubra ), sweetbay ( M. virginiana ), sweet gum ( L. styraciflua ), water oak ( Q. nigra ), willow oak ( Q. phellos ), red maple ( A. rubrum ), and swamp tupelo ( N. aquatica ). Mixed species including southern scrub oak ( Q. rubra ) and yellow pine ( P. ponderosa ) cover approximately 21% of the riparian area, whereas loblolly ( P. taeda ) and longleaf ( P. palustris ) pine cover approximately 19%. The riparian area soils are predominantly sandy loam. The uplands are mostly on loamy sand. Data and Input Parameters REMM operates at a hillslope scale. To a pply the model at watershed scale, it is necessary to convert the waters hed into an equivale nt hillslope by adju sting the geometry of watershed. As the riparian buffer has sp ecified length (parallel to the stream) and width (perpendicular to the stream), upland c ontributing field width of the watershed was adjusted to equate the total area of the watershed. REMM requires soil and vegetation nutrient pools for the riparian area of the watershed to initialize the model (Table 42). Required field da ta to run the model includes daily surface and subsurface runoff from the uplands to the riparian area (Table 4-3). The model also requires daily weat her data and paramete r values representing topographic, soil, and vegetative conditions within the riparian area. Strategic Environmental Research and Development PlanÂ’s (SERDP) Ecosystem Management Project (SEMP) continuously records the atmospheric data within the military installation. Daily weather data were obtaine d from SEMP data re pository that included precipitation amount and duration, maximum and minimum air temperatures, solar radiation, wind speed, and dew poi nt temperature. Daily st reamflow for Bonham-South

PAGE 81

69 Table 4-2. Soil and vegetation nutrient pool s for Bonham-South watershedÂ’s riparian area. The litter and soil pools of nutrient consist of metabolic, structural, active, slow, and passive humus pools. Th e values in litter layer as well as individual soil layers in each zone ar e presented. Vegetation pools consist of nutrients in leaves, stems, branches, coar se roots, fine roots, and heartwoods. Total values of nutrients are reported in the vegetation pool. The parameter values are identical for Zone 1, Zone 2, and Zone 3. Parameters DepthBulk densityCarbon Nitrogen Biomass (m) (g/cm3) (kg/ha) (kg/ha) (kg/ha) Litter and Soil Litter 0.01 1.0 18,100 1,134 Soil layer 1 0.30 1.4 29,280 655 Soil layer 2 1.00 1.5 24,640 644 Soil layer 3 2.00 1.5 11,560 252 Total 83,580 2,685 Live vegetation 31,390 753 78,477 Table 4-3. Model simulated annual hydrologic budget for the riparian areas in BonhamSouth watershed. and ** represents the surface and subsurface flow input to the zone 3 from uplands. Values are rounded off to nearest whole numbers. Year Units 2000200120022003 4-yr average Precipitation mm/yr 475 736 731 1,014 739 Surface flow in* mm/yr Subsurface flow in** mm/yr 119 134 143 253 162 Observed stream flow (total) mm/yr 121 137 147 258 166 Simulated surface flow mm/yr 2 5 4 6 4 Simulated subsurface flow mm/yr 106 118 130 244 149 Simulated stream flow (total) mm/yr 109 124 133 250 154 was calculated from 10-minute continuous stag e records using rating curves and areavelocity method. Typically, REMM uses daily surface r unoff depth generated from the runoff collected by surface runoff samplers at the upland-Zone 3 interface. REMM uses the hydraulic gradient upland wells and Zone 3 we lls, saturated thickness at Zone 1, and the saturated hydraulic conductivity to compute s ubsurface flow loading to the buffer. Due

PAGE 82

70 to the lack of measured surface and subsurface flow from the upland to the riparian area, it is assumed that the baseflow fraction in th e measured streamflow is a contribution of upland subsurface flow to the riparian area. Therefore, measured streamflow from January 2000 through December 2003 was parti tioned into baseflow and surface runoff using constant slope base flow separa tion technique (McCuen, 1998). Surface runoff from the upland to the riparian area was not considered (Table 4-3). Parameters that describe the riparian area dimensions, soil, and vegetation character istics were derived from measured data and previously published literature (Inamdar et al., 1999a, Inamdar et al., 1999b, Lowrance et al., 2000; Ga rten Jr., et al., 2003). Im portant parameters related to litter and soil layer are listed in Appendix D. Monthly canopy cover in the riparian area was determined by direct measurement with a Model-A spherical densiometer using the method outlined by Lemmon (1956) fr om June 2001 to September 2003. Measured water and soil nutrient concentra tions in the study area included the total Kjeldahl nitrogen (TKN), nitrate, and ammonium in precipitation, streamflow, soil water, and shallow groundwater. Two transects in the riparian area near the outlet of the watershed were considered for groundwater and soil water monitoring. Groundwater monitoring wells (1-3 m deep) and adjacent tension lysimeters at 20 and 60 cm depths were positioned on both sides of the stream along the riparian transects. Stream water, groundwater, and soil water samples were co llected from October 2001 to September 2003; biweekly from October 2001 to Novemb er 2002 and monthly thereafter. BonhamSouth stream was sampled during 16 st orm events between September 2002 and September 2003 using an event triggered ISCO sampler that was programmed to collect

PAGE 83

71 hourly samples based on the flow depth. Bhat et al. (unpublished manuscript), describes the sample analysis procedures for different chemical constituents. As REMM simulations are performed on a daily time step, and observed TKN concentrations are of hourly time step, co mparisons between the two were based on the TKN masses produced during the events that lasted 24 hours or longer. The model simulates the masses of dissolved and partic ulate organic nitrogen, dissolved ammonium and nitrate in surface and subsurface flows on a daily basis. The mass of simulated TKN is the sum of dissolved and particulate organic nitrogen and dissolved ammonium. Carbon dissolves from metabolic residue a nd active humus pools. In REMM, incoming C from precipitation, surface, and subsurface flow is assumed to be in dissolved form. Carbon available to be dissolved from the ac tive humus pool is a fr action of the total C present in the same pool. This fraction of set at 0.31 based on an estimate by McGill et al. (1981) for the proportion of dying bacteria in metabolic form. The total amount of dissolved C is the sum of metabolic C residue and the dissolved C fraction in the active humus pool. In REMM, stoichiometric relati onships are assumed between C and N in the organic matter. As C is transformed, corres ponding N is also transformed. A Freundlich isotherm method determines the amount of di ssolved ammonium from inorganic pool of N in REMM. Observed masses of TKN during storm events are calculated by multiplying the concentrations by the volume of water. Model Calibration REMMÂ’s hydrology and nitrogen components we re evaluated using data collected in this study from riparian area as well as data from the literature. Data collected in nearby watersheds in Fort Benning, Georgia we re used to initialize the C and N in the soil pools (Garten Jr. et al., 2003) The litter C and N pools ar e based on the literature

PAGE 84

72 values (Silveira, M.L., Reddy, K.R. (Departmen t of Soil and Water Science, University of Florida), Comerford, N.B. (Department of Soil and Water Science, University of Florida). Litter decomposition and soluble organic carbon and nitr ogen release in a forested ecosystem. Unpublished manuscript) Soil and vegetation nutrient pools for the riparian area of the watershed are liste d in Table 4-2. Bonham-SouthÂ’s soil layer thickness, soil porosity, saturate d hydraulic conductivity, clay content, carbon decay rate, and denitrification rate were calibrated. Model calibration involved the comparison of the simulated streamflow and TKN output with the measured values. While calibrating hydrology a nd nutrient components of the model, parameters such as soil and litter C and N pools, which are based on literature values, were fixed. Other fixed pa rameters included ripa rian length and width, surface and subsurface draining area. The remaining parameters including soil layer thickness, saturated hydr aulic conductivity, soil porosity, clay content, carbon decay rate, and denitrification rate in each zone and soil la yers of the riparian buffer were adjusted to match the observed and predicted results. Fox (1981) recommends mean biased error (MBE) and mean absolute error (MAE) to measure the difference between observed and model predicted values. Mean biased error is calculated as the average erro r between predicted and observed values accumulated over the total number data points. Mean absolute error considers the absolute values of the errors. In the present study, MBE and MAE are modified to calculate the percent difference of modele d streamflow and TKN mass from the observed values. Mean bias difference (MBD) calculates the average difference accumulated over the total number of events, e xpressed as a percent of the obs erved value. Mean absolute

PAGE 85

73 difference (MAD), expressed also as a percen t, considers the absolute values of the differences. Mean absolute difference takes ne gative values and repla ces them with their absolute values. Observed and modeled streamflow was also compared using the NashSutcliffe efficiency. The Nash-Sutcliffe ef ficiency criteria is based on the normalized least square objective function that evaluates the sum of the squares of flow residuals (Nash and Sutcliffe, 1970). Sensitivity Analysis A sensitivity analysis was performed to de termine the effects of key hydrological, soil, and vegetation parameters on streamflow and TKN fluxes. The parameters were canopy cover fraction, riparian zone width, soil layer thic kness, maximum carbon decay rate for litter and humus, maximum denitr ification rate, soil saturated hydraulic conductivity, soil porosity, and so il clay content. Each pa rameter was changed by +10% and –10% from the values used as the best estimates for the ca libration simulations. Results Hydrology Simulated daily flows for the study waters hed closely matched the observed flows (Figure 4-2). Over the 4-year simulati on period, MBD and MAD between observed and predicted streamflow were 2% and 12%, re spectively. The model had Nash-Sutcliffe efficiency of 80%. REMM tends to underest imate the streamflow during low flow and overestimate during the storms. Approxima tely 1% of the streamflow was the contribution from surface runoff generated in th e riparian area. Majority of streamflow output was through the subsurface (Table 4-2).

PAGE 86

74 Figure 4-2. Comparison of REMM simulated dail y flow with the observed daily flow for the Bonham-South watershed. There was considerable variability be tween and within years for measured precipitation and streamflow data during study pe riod. It is evident from Figure 4-2 that an increase in the average flow during the ye ar 2003. This increase was consistent with the variation in precipitation during the st udy period. Overall, model simulations for streamflow can be considered good for the calibrated parameters. Nitrogen Observed stream TKN concentrations during the study period showed a strong correspondence with the leaf fa ll (Figure 4-3). Approximate ly 50% of the canopy in the riparian area dropped during November-January each year. Higher TKN concentrations were observed during the same period of time. The simulated TKN concentrations also correspond to simulated leaf fall (Figure 4-4). Modeled TKN concentr ations were higher during high leaf fall months. From Oct ober 2001 to December 2003, modeled leaf fall totaled 12,200 kg/ha, which corresponds to 5,400 kg/ha/yr. As depicted in the figure,

PAGE 87

75 Figure 4-3. Observed canopy cover and T KN concentrations each month for the Bonham-South watershed. Figure 4-4. Monthly simulated TKN concen tration and simulated leaf mass in the Bonham-SouthÂ’s riparian area. simulated leaf fall began as early as J une each year and completed by the end of December-January. As observed in the ripa rian area of the watershed, leaves were present in the canopy year-round. For example, during the 2001-2002 dormant season,

PAGE 88

76 approximately 45% of the canopy was still cove red with leaves. Simulated N present in the live vegetation showed an inverse relation with the simulated TKN concentrations in the stream (Figure 4-5). Figure 4-5. Monthly simulated TKN concentration and simulate d nitrogen present in live leaves the Bonham-SouthÂ’s riparian canopy. A careful analysis of model simulati on results showed that a daily total precipitation of 6 mm or more produced surface r unoff in the riparian area. This surface runoff is responsible for leaching and transporting TKN from the leaf li tter in the riparian zone to the stream. Due to the irregular nature of surface runoff generation, the TKN variability during such events, the frequencie s of observed data, and REMMÂ’s ability to simulate on a daily basis, storm events lasti ng 24 hours or longer were selected for further study. The masses of TKN were calculated for tw o different scenarios. First, total mass of TKN during an event was calculated for th e total streamflow. Second, baseflow was separated from the total flow, and TKN mass corresponding to surface runoff was determined. The observed and simulated T KN masses during indivi dual storm events

PAGE 89

77 were compared for both the scenarios. Out of the 16 storm samples collected in the study area, 6 storms between October 2002 and Ma y 2003 were suitable for the comparison with the daily output from REMM (Figure 4-6) The simulated TKN masses produced in surface runoff during the storms were comparab le to the observed values (Figure 4-7). MBD and MAD between the observed and the simulated TKN masses were 8% and 23%, respectively. As the majority of the flow in the stream is subsurface flow, the effect of subsurface flow in carrying the TKN mass was also analyzed. The result showed that MBD and MAD between observed and simula ted total TKN masses were 17% and 24%, respectively (Figure 4-8). Sensitivity Analysis The changes in streamflow and TKN out puts for the –10% to +10% parameter changes ranged from 0 to 1% and 0 to 7%, respectively (Table 4-4). Changes in soil porosity did not have significant effects on th e streamflow and TKN. A decrease in soil porosity of 10% led to a total streamflow increase of only 0.5%, wh ereas the increase in TKN was only 0.6%. Increase in clay c ontent decreased the TKN output by 0.7%. Increase in saturated hydraul ic conductivity by 10% did not affect the streamflow but increased TKN by 7.2%. Increasing soil layer thickness slightly redu ced the streamflow, but reduced TKN by 5.4%.

PAGE 90

78 Figure 4-6. Observed TKN concentrations dur ing the events. Vertical bars represent precipitation, solid lines represent th e streamflow, and the hollow circles represent the TKN concentrations.

PAGE 91

79 Figure 4-7. Comparison of the observed and simulated TKN masses during the events. These masses represent the contribution from the surface runoff only. Figure 4-8. Comparison of the observed and simulated TKN masses during the events. These masses represent the contribution from the total flow.

PAGE 92

80 Table 4-4. Sensitivity of modeled streamfl ow and TKN based on +/10% change in model parameters for Bonham-South watershed. Percentage change Parameters Stream flow TKN Canopy cover (+) -0.9 2.3 Canopy cover (-) 0.9 -0.2 Soil layer thickness (+) -0.1 -5.4 Soil layer thickness (-) 0.1 5.7 Denitrification rate (+) 0.0 0.0 Denitrification rate (-) 0.0 0.0 Carbon decay rate (+) 0.0 0.9 Carbon decay rate (-) 0.0 -0.5 Saturated hydraulic conductivity (+) 0.0 7.2 Saturated hydraulic c onductivity (-) 0.0 -0.2 Soil porosity (+) -0.4 -1.5 Soil porosity (-) 0.5 0.6 Clay content (+) 0.0 -0.7 Clay content (-) 0.0 0.4 Discussion Over the four years, the simulated average annual surface runoff that is generated in the riparian area and contribu ted to the streamflow was less than 1% of the annual observed precipitation (Table 4-3). The simulated surface runoff generated in the riparian buffer in the study watershed is sm all compared to earlier studies done in a similar watershed in Georgia. A study conducted by Shirmohammadi et al. (1984) in Little River watershed in Georgia reported approximately 4-12% contribution of annual precipitation to surface runoff. Inamdar et al. (1999a) reported the average annual surface runoff contribution to streamflow was approximately 8% of the annual precipitation for Gibbs Farm site within th e Little River watershed. Surface runoff contribution to streamflow in our study wate rshed may not be significant in terms of

PAGE 93

81 hydrology, but the modeled results suggest that it is capable of transpor ting nitrogen from the litter and top soil layer to the stream (Figure 4-7). Subsurface contribution to the streamflow for the Bonham-South watershe d was approximately 20% of the average annual precipitation (Table 4-3). The simula ted subsurface contribution in this study is close to the range of 14-22% reported by Shirmohammadi et al. (1984) for Little River watershed. The riparian forest community f ound along Bonham-South watershed is representative of the Sout heastern riparian forest (Shure and Gottaschalk, 1985; Lowrance et al., 2000). Slash pine ( P. elliotii ), longleaf pine ( P. palustris ), black gum ( N. sylvatica ), sweet gum ( L. styraciflua ), scrub oak ( Q. rubra ), sweetbay ( M. virginiana ), water oak ( Q. nigra ), willow oak ( Q. phellos ), red maple ( A. rubrum ), and swamp tupelo ( N. aquatica ) are often dominant canopy species in these riparian forests. The modeled total leaf fall in the study watershed from October 2001 to December 2003 was approximately 12,200 kg/ha, which corres ponds to 5,400 kg/ha/yr. The modeled average litter mass produced during the same period was approximately 7,000 kg/ha/yr. Approximately 77% of the litter mass were le aves. Contribution of the modeled leaf fall to the total litter pr oduction falls in the ne ighborhood of 80% reported by Meentemeyer et al. (1982), and within the range of 72-84% reported by Shure and Gottaschalk (1985). The simulated litter mass produced in th e study area is higher than the average 5,000 kg/ha for a mixed hardwood forest in the No rtheastern U.S. (Gosz et al., 1972) and is comparable with the results found in the Sout heastern U.S. Mul holland (1981) reported litter production of 6,100 kg/ha in a small str eam swamp in eastern North Carolina. A bottomland hardwood forest and a cypress-t upelo stand located in Louisiana produced

PAGE 94

82 litter masses of 5,750 kg/ha and 6,200 kg/ha, respectively (Conner and Day Jr., 1976). Brinson et al. (1980) reported a value of 6,500 kg/ha for an alluvial swamp forest in North Carolina. Simulated results showed that the nitroge n content in the li tter during the study period averaged 77 kg/ha of which approximate ly 59 kg/ha was the contribution from the leaves alone. This value is less than 83.4 kg/ha of nitrogen contribution from the deciduous leaves in the Northeastern U.S. (Gos z et al., 1972). As observed in the riparian area, 45% canopy was present even during the dormant seasons because of the presence of approximately 19% of conifers and 21% of mixed species. Modeled results also support the presence of the carbon in the leaves and the biomass in the canopy. The modeled result showed that on an average approximately 527 kg/ha of carbon was present in the canopy during the dormant season. This value of carbon suggests the presence of approximately 22 kg/ha of n itrogen in the canopy during the same time (Figure 4-5). Higher concentr ations of TKN during the leaf fall and the inverse relation with the nitrogen present in the canopy support the hypothesis of flushing of nitrogen from the leaf litter. The simulated TKN concentrations follow the trend of leaf mass accumulation in the riparian area suggesting a higher rate of release of TKN from freshly fallen litter during the fall and early spring. Brinson et al. (1980) reported higher nutrient fluxes following litterfall in an alluvial swamp forest in North Carolina. In a study of nutrient content of litterfall in the Southeastern U.S., Gosz et al. (1972) reported that approximately 56% of the total nutrien t was released during September through December when the majority of litterfall ( 50%) occurred. A study conducted in cypress

PAGE 95

83 swamp forest in Florida, Schlesinger (1978) reported that in th e month of November when 56% litterfall occurred, 45% of nitr ogen was released from the system. Higher TKN concentrations were observed in the Bonham-South stream during the precipitation events (Figure 46). The storm events presen ted in Figure 4-6 were long enough to calculate the masses of TKN during the storms to compare REMMÂ’s simulated daily values. Comparisons of TKN masses pr oduced during the events were comparable to REMMÂ’s simulated values. The masses of TKN during the storms were calculated for two different scenarios. When modeled TKN mass contribution of surface runoff that is produced in the riparian zone was compared with the observed value in the stream, MBD and MAD for the six storms were 8% and 23%, respectively. Mean biased difference value is less reliable as there is the risk th at large outliers cancel each other out. As MAD takes negative values and repl aces them with their absolute values, large outliers are deemphasized; hence it is less sensitive to extr eme values. A MAD of 23% in the surface runoff suggests that the model was able to produce comparable amount of TKN as observed during the storm. The comparis on between observed and simulated TKN mass from the total flow showed comparable MB D (17%) and MAD (24%). On average, 43% of the observed total TKN mass was contributed by surface runoff. This result suggests that the surface runoff during the precipitat ion events is a major source of leaching nitrogen from the forest floor and transporting it to the stream. Conclusion and Recommendations Originally, REMM was developed to operate at a hi llslope scale. The results in this study suggested that given appropriate upla nd inputs for a site, REMM can be used at different scenarios of riparian area width, length of zones, vegetation type, and soil properties. The trend and ma gnitude of the observed streamflow for the study watershed

PAGE 96

84 was effectively simulated by REMM. The model, however, overestimated the streamflow during high flow and underest imated during low flow periods. The hydrologic budget showed a good agreemen t between observed and predicted streamflow. The simulated litter and corre sponding leaf masses in the study watershed were comparable to the values reported in the lite rature. Due to the modelÂ’s longer simulation time step, a meaningful comparison of simu lated and observed TKN was possible only through the masses produced during storms that were equal to or longer than a day. Comparison of TKN masses during six different storm events showed similar values both in the surface runoff and the total flow. Th e results supported the hypothesis of nitrogen leaching from freshly fallen leaves during the precipitation events. These results provided further insight into the nutrient dynamics of the riparian area. The model simulations respond as expected to precip itation and vegetation pa tterns over the study period. Results clearly indicated that the presen ce of fresh leaf litter in the riparian area increases the TKN concentrati on, and hence mass, in the stream. The model effectively captured the trends of leaf mass accumulation in the riparian area and subsequent high concentration and mass during those periods. With the present ve rsion of REMM, the comparison between observed and simulated va lues is possible only on a daily basis. However, given the fact that the precipitation events and the sampling frequencies are often shorter, simulation results using a smalle r time scale would be useful. Therefore, it is recommended to modify REMM from its presen t version of daily time step to a smaller time step, preferably hourly, for more eff ective and meaningful interpretation and evaluation of riparian nutrient flushing.

PAGE 97

85 CHAPTER 5 SUMMARY AND CONCLUSION The National Research Council of the Unite d States has proposed that reliable and comprehensive environmental indicators be developed to monitor ecological changes from natural and anthropogenic causes. The military installation in Fort Benning, Georgia offered a unique opportunity to study military impacts on water quality and quantity. Ecohydrological approaches were used to relate the effects of anthropogenic perturbations on water-vegetationnutrient interactions in th e Fort Benning watersheds. In this research, statisti cal relationships among water quality parameters and the watershed physical characteristics in low-nutrient watersheds were identified and examined. This research also identifi ed and developed hydrologic indices that characterize the impact of military land mana gement on watershed, and investigated the riparian corridorÂ’s role on water quality by quantifying the nitrogen leaching from freshly fallen leaves. Relationships among watershed physical characteristics and water quality parameters in the study watersheds and the re gression analysis showed that pH, chloride, total phosphorus, total Kjeldahl nitrogen, tota l organic carbon, and total suspended solids are useful indicators of wate rshed physical characteristics that are susceptible to perturbations. A comparison of the results between a reference and impacted watershed in terms of hydrologic indices that are derived from long-term daily flow as well as the storm-based data showed a clear distinct ion between the watersheds in terms of hydrological flow regimes. Storm-based ma gnitude of baseflow index, magnitude and

PAGE 98

86 variability of peak discharge, and the frequency of bankfu ll discharge were consistent with the results from annual-based analysis Results showed that these storm-based indices might be used as surrogates to the indices derived from long-term data. The analysis identified the relationships between the extent of military training land, road density, and the number of road s crossing streams with the storm-based baseflow index, bankfull discharge, response la g, and time of rise. For the low nutrient systems of this study, seasonal and storm variations in water quality were found to be strongly influenced by precipitation events which caused nitrogen leaching from recently fallen leaf litter and increased nitrogen levels in the stream water. This study showed that the signatures of military alterations to watershed landscape are detectable in the water qual ity and the flow regime. The observed alterations to these regimes suggest impacts to aquatic ecosystems The water quality related indicators and hydrologic indices presented in this research provide specific measur es of stream water quality and instream flow, respectively, that respond to military impacts. Each indicator is linked to one or more readily measured wa tershed scale factors. The indicators could be useful in predicting effects of military land management and evaluating restoration activities on the quality as well as the quantity of the water. This research has several implications fo r the Fort Benning military installation. The results of this study indi cated that baseflow sampling of water quality can be used to assess the military training impacts on stream water quality. The water quality indicators identified in this study provide measures of the current watershed conditions. The water quality indicators identified in this stu dy provide measures of the current watershed conditions. Changes to stream water quality due to military training within the Fort

PAGE 99

87 Benning military installation can be identified by collecting long-term and routine stream water quality data and comparing indicator valu es to current conditions Future activities should develop indicator thresholds beyond wh ich improved management and restoration practices should be implemented. As nutri ents are one major factor controlling the quality of the receiving aqua tic ecosystems, the thresholds should be based on the ecosystems impacts. The U.S. Army manages approximately 4.8 106 hectares of land for military training, and it has developed the Land Cond ition Trend Analysis (LCTA) program to systematically monitor terrestrial impact s from military training and to support the mitigation and remediation of severely imp acted training lands (Quist et al., 2003). Although the magnitude of the training would diffe r from one installation to the other, the nature of the impact remains similar. Military training typically increases soil bulk densities and compaction, decrea ses infiltration, diminishes pl ant growth, degrades water quality, and affects water quantit y. Therefore, the research findings in this study can be implemented for management and restoration practices to other military installations across the U.S. to minimize the impacts of the military training. A common finding between the water quality and quantity signature s is that military training impacts differ from those found due to agri culture and urbanization. In the former, water quality degradation impacts typically manifest as high nutrient loads. However, in military installations, the loss of topsoil and vegetation results in waters that have significantly lower nutrient levels. In a similar fashion, the hydrologic regimes in watersheds with significant military impacts suffer from a lo ss of variability, but exhibit higher annual discharge values.

PAGE 100

88 APPENDIX A ANNUAL-BASED INDICES DEFINITIONS AND CALCULATION PROCEDURES Symbol Definition Method MA26 Coefficient of variation in monthly flows for December Calculate mean monthly flow for December (diq ), i = 1,Â…, n ; n = no. of years Calculate mean of all December flows, DQ=n qn i di/1 Calculate standard deviation ( SD ) of diq Calculate coefficient of variation, CV = SD /DQ MA41 Mean annual flow divided by watershed area Calculate mean yearly flow (yiq), i = 1,Â…, n; n = no. of years Divide yiqby watershed area (A) Calculate n A q A Qn i yi y/ /1 i = 1,Â…, n; n = no. of years MA10 Ranges in daily flows divided by median daily flows (where range in daily flows is the ratio of 20th/80th percentiles in daily flows across all years) Combine daily flows of all years Calculate 20th percentile (20pQ) Calculate 80th percentile (80 pQ) Calculate median flow (mQ) Calculate R = 20pQ/80pQ Calculate R / mQ ML17 Seven-day minimum flow divided by mean annual daily flows Calculate mean yearly flow (yiq), i = 1,Â…, n; n = no. of years Calculate 7-day minimum (min 7q)i flow Calculate ratios Ri = (min 7q)i /yiq Calculate n Rn i i/1

PAGE 101

89 ML14 Mean of the lowest annual daily flow divided by median annual daily flow averaged across all years Extract lowest yearly flow (min ,iq), i = 1,Â…, n; n = no. of years Calculate n q Qn i i/1 min min Calculate median flow (mQ ) of all years Calculate mQ Q/min ML16 Median of the lowest annual daily flows divided by median annual daily flows averaged across all years Extract lowest yearly flow (min ,iq ), i = 1,Â…, n ; n = no. of years Calculate median (medQ) of min iq Calculate median flow (mQ ) of all years Calculate m medQ Q/ MH23 Mean of the high flow volume (calculated as the area between the hydrograph and the upper threshold during the high flow event) divided by median annual daily flow across all years. The upper threshold is defined as 7 times median annual flow Calculate yearly median flow (med iq,), i = 1,Â…, n ; n = no. of years Calculate 7 times med iq, Calculate the flow volume (ViQ ) above 7 times med iq, value Calculate n Q Qn i Vi V/1 Calculate median flow (mQ ) of all years Calculate m VQ Q/ MH8 Mean of the maximum monthly flows for the month of May Extract maximum flow for May (miq ), i = 1,Â…, n ; n = no. of years Calculate mean of all May maximum flows, MayQ= n qn i mi/1 MH14 Median of the highest annual daily flow divided by the median annual daily flow averaged across all years Calculate median flow (miQ ), i = 1,Â…, n ; n = no. of years Calculate n Q Qn i mi m/1 Extract highest daily flow (hiQ ) for each year Calculate median of highest daily flow (medhQ ) Calculate medhQ /mQ

PAGE 102

90 FL3 Total number of low flow spells (threshold equal to 5% of mean daily flow) divided by the record length in years Combine daily flows of all years Calculate mean daily flow (meanQ ) Determine the threshold of 0.05*meanQ Determine the numbers of low flow counts below 0.05*meanQ (lowN ) Calculate lowN /n, n = record length in years FL2 Coefficient of variation in Fl1, where Fl1 is low flood pulse count, which is the number of annual occurrences during which the magnitude of flow remains below a lower threshold. Hydrologic pulses are defined as those periods within a year in which the flow drops below the 25th percentile (low pulse) of all daily values for the time period Calculate 25th percentile value for each year (i pQ25), i = 1,Â…, n ; n = no. of years Determine the numbers of low flow pulses below 25th percentile for each year (i lN25) Calculate mean of low flow pulses, n N Nn i i l l/1 25 25 Calculate standard deviation ( SD ) of i lN25 Calculate CV = SD /25 lN FH4 High flood pulse count is the number of annual occurrences during which the magnitude of flow remains above an upper threshold where the upper threshold is defined as 7 times median daily flow, and the value is represented as an average instead of a tabulated count Calculate yearly median flow (med iq,), i = 1,Â…, n; n = no. of years Calculate 7 times med iq, Determine the number of high flood pulses above 7 times med iq, (hiN7) Calculate n N Nn i hi h/1 7 7 FH6 Mean number of high flow events per year using an upper threshold of 3 times median flow over all years Calculate yearly median flow (med iq,), i = 1,Â…, n; n = no. of years Calculate 3 times med iq, Determine the number of high flood pulses above 3 times med iq, (hiN3) Calculate n N Nn i hi h/1 3 3

PAGE 103

91 FH7 Mean number of high flow events per year using an upper threshold of 7 times median flow over all years Same as FH4 DL10 Coefficient of variation in annual minima of 90day means of daily discharge Determine 90-day minimum flow (iQmin 90) for each year, i = 1,Â…, n; n = no. of years Calculate n Q Qn i i/1 min 90 min 90 Calculate SD of iQmin 90 Calculate CV = min 90/Q SD DL17 Coefficient of variation in low flow pulse durations Calculate 25th percentile value for each year (i pQ25),i = 1,Â…, n; n = no. of years Determine duration between low flow pulses below 25th percentile for each year (i lD25) Calculate mean duration of low flow pulses, n D Dn i i l l/1 25 25 Calculate standard deviation (SD) of i lD25 Calculate CV = SD/25 lD DL6 Coefficient of variation in annual minima of 1day means of daily discharge Determine 1-day minimum flow (iQmin 1) for each year, i = 1,Â…, n; n = no. of years Calculate n Q Qn i i/1 min 1 min 1 Calculate SD of iQmin 1 Calculate CV = min 1/Q SD DH13 Mean annual 30-day maximum divided by median flow Determine 30-day maximum flow (iQmax 30) for each year, i = 1,Â…, n; n = no. of years Calculate n Q Qn i i/1 max 30 max 30 Calculate median flow (miQ),i = 1,Â…, n; n = no. of years

PAGE 104

92 Calculate n Q Qn i mi m/1 Determine mQ Q /max 30 DH16 Coefficient of variation in high flow pulse durations Calculate 75th percentile value for each year (i pQ75), i = 1,Â…, n; n = no. of years Determine duration between high flow pulses above 75th percentile for each year (i hD75) Calculate mean duration of high flow pulses, n D Dn i i h h/1 75 75 Calculate standard deviation (SD) of i hD75 Calculate CV = SD/75 hD DH24 Mean annual maximum number of 365 days over all water years during which no floods occurred over all years Determine flow of magnit ude exceeding a return interval of 1.67 years based on log-normal distribution (yriQ67 1), i = 1,Â…, n; n = no. of years Determine the maximum number of 1.67-year flow non-exceedance days each year (i yrDmax 67 1) Calculate n D Dn i i yr yr/1 max 67 1 max 67 1 TA1 Constancy [Colwell (1974)] Standardize the daily flow values by median flow, and express as natural logarithm values Divide the log-transforme d flow values into 4-6 classes so that these classes cover all the observed flow values Create a matrix of total number of days in a year (columns, t) by number of classes (rows, s) Define the column totals (jX ), row totals (iY), and the grand total (Z) as s i ij jN X1, t j ij iN Y1, and i i j j j ij iY X N Z where Nij is the number of cycles for which the flow is in class i and time j. Determine uncertainty with respect to time t j j jZ X Z X X H1log ) (

PAGE 105

93 Determine uncertainty with respect to class s i i iZ Y Z Y Y H1log ) ( Determine uncertainty with respect to the interaction of time and scale j ij ij iZ N Z N XY Hlog ) ( Determine predictability (P) with the range (0, 1) s X H XY H Plog ) ( ) ( 1 where s is the number of classes Determine the constancy ) ( C with range (0, 1) s Y H C log ) ( 1 TA3 Maximum proportion of all floods over the period of record that fall in any one of six 60-day ‘seasonal’ windows Determine flow of magnit ude exceeding a return interval of 1.67 years based on log-normal distribution (yrQ67 1) over the period of record Divide every year into six 60-day ‘seasonal’ windows Determine the maximum number of yrQ67 1exceedance days (max 67 1 yrD ) over period of record in any one of six 60-day ‘seasonal’ windows Determine the total number of yrQ67 1exceedance days over all years in one of six 60-day seasonal window (yrtotD67 1) Calculate the ratio (max 67 1 yrD/yrtotD67 1) TH3 Maximum proportion of the year (number of days/365) during which no floods have ever occurred over the period of record Determine flow of magnit ude exceeding a return interval of 1.67 years based on log-normal distribution (yrQ67 1) over the period of record Determine the maximum number of yrQ67 1nonexceedance days (yrnonD67 1) over period of record Determine the total number of yrQ67 1nonexceedance days over period of record (yrtotD67 1) Calculate the ratio (yrnonD67 1/yrtotD67 1)

PAGE 106

94 RA9 Coefficient of variation in number of negative and positive changes in water conditions from one day to the next Determine number of rises from one day to the other each year (RiN ), i = 1,…, n ; n = no. of years Determine number of falls each year (FiN ) Calculate n N Nn i Ri R/1 Calculate SDR of RiN Calculate n N Nn i Fi F/1 Calculate SDF of FiN Calculate CVR = SDR /RN Calculate CVF = SDF /FN RA7 Median of difference between natural logarithm of flows between two consecutive days with decreasing flow Transform the daily flow values for each year using natural logarithm, Qi = ln Qi, i = no. of days For all ln Qs each year, determine [ln Qi – ln Qi+1] Separate the values of decreasing flow differences for each year Determine the median of decreasing flow differences each year (dmjQ ), j = 1, …, n; n = no. of years Calculate n Q Qn j dmj dm/1 RA6 Median of difference between natural logarithm of flows between two consecutive days with increasing flow Transform the daily flow values for each year using natural logarithm, Qi = ln Qi, i = no. of days For all ln Qs each year, determine [ln Qi – ln Qi+1] Separate the values of increasing flow differences for each year Determine the median of increasing flow differences each year (imjQ ), j = 1, …, n; n = no. of years Calculate n Q Qn j imj im/1

PAGE 107

95 APPENDIX B STORM-BASED INDICES DEFINITIONS AND CALCULATION PROCEDURES Index Symbol Definition Methods MMRF Mean value of the response factor Calculate precipitation depth for each event (iP ), i = 1,Â…, n ; n = no. of events Create hydrograph Separate baseflow Calculate DRO depth after de ducting the baseflow portion from the hydrograph for each event (iD ) Calculate response factor for each event, i i iP D RF / Calculate the mean, n RF RFn i i/1 MVRF Coefficient of variation in response factors Calculate RF Calculate standard deviation ( SD ) of RFi Calculate _/RF SD CV MMBF Mean value of baseflow index Create hydrograph Separate baseflow Determine baseflow volume (Vb) and total volume (Vt) for each events Calculate BFi = (Vb/Vt,)i, i = 1,Â…,n; n = no. of events Calculate the mean, n BF BFn i i/1 MVBF Coefficient of variation in baseflow index Calculate BF Calculate (SD) of BFi Calculate _/ BF SD CV MMPD Mean value of peak discharges divided by the watershed area Determine ( qpk)i of the event, i = 1,Â…, n ; n = no. of events Normalize ( qpk)i by the watershed area, A qi pk/ ) ( Calculate mean, _/ A qpk = n i i pkn A q1/

PAGE 108

96 MVPD Coefficient of variation in peak discharges Calculate _/ A qpk Calculate (SD) of A qi pk/ ) ( Calculate CV = SD / _/ A qpk F1FD Percentage of peak discharge equals bankfull discharge Determine bankfull discharge Count peak discharges equal to bankfull discharge for all events and express the count as a percentage of total events F2FD Percentage of peak discharge 2 times above bankfull discharge Determine bankfull discharge Count peak discharges e qual to 2.0 times bankfull discharge for all events a nd express the count as a percentage of total events DMTB Mean value of time base divided by the watershed response time Determine time base for each hydrograph Normalize the time base by the response time (Tbi), i = 1,Â…,n; n = no. of events Calculate mean, n T Tn i bi b/1 DVTB Coefficient of variation in time base Calculate bT Calculate (SD) of (Tbi) Calculate CV = SD/_ bT DMRL Mean value of response lag divided by the watershed response time Determine response lag for each hydrograph Normalize the response lag by the response time (Trli), i = 1,Â…,n; n = no. of events Calculate mean, n T Tn i rli rl/1 DVRL Coefficient of variation in response lag Calculate rlT Calculate (SD) of (Trli) Calculate CV = SD/_ rlT DMTR Mean value of time of rise divided by the watershed response time Determine time of rise for each hydrograph Normalize the time of rise by the response time (Tri), i = 1,Â…,n; n = no. of events Calculate mean, n T Tn i ri r/1

PAGE 109

97 DVTR Coefficient of variation in time of rise Calculate rT Calculate (SD) of (Tri) Calculate CV = SD/_ rT RPPD Mean rate of change in peak discharge in rising limb Determine (qpk)i of the event, i = 1,Â…,n; n = no. of events Normalize (qpk)i by the watershed area, A qi pk/ ) ( Normalize the time of rise by the response time (Tri), i = 1,Â…,n; n = no. of events Calculate the ratio of A qi pk/ ) ( to (Tri) Calculate the mean of the ratios RPVPD Coefficient of variation in the rate of change in peak discharge in rising limb Calculate mean of RPPDi Calculate (SD) of RPPDi Calculate CV = SD/mean of RPPDi RNPD Mean rate of change in peak discharge in falling limb Determine (qpk)i of the event, i = 1,Â…,n; n = no. of events Normalize (qpk)i by the watershed area, A qi pk/ ) ( Normalize the time of rise by the response time (Tri), i = 1,Â…,n; n = no. of events Normalize the time base by the response time (Tbi) Calculate the difference (Tbi Tri) Calculate the ratio of A qi pk/ ) ( to (Tbi Tri) Calculate the mean of the ratios RNVPD Coefficient of variation in the rate of change in peak discharge in falling limb Calculate mean of RNPDi Calculate (SD) of RNPDi Calculate CV = SD/mean of RNPDi

PAGE 110

98 APPENDIX C NITROGEN TRANSFORMATIONS IN REMM

PAGE 111

99 APPENDIX D PARAMETERS AND RATE CONSTANTS USED IN REMM Parameters Units Values Litter layer parameters (all zones) Layer depth cm 1 Evaporation factor unitless 4 Evaporation constant unitless -0.45 Litter moisture mm 2 Litter bulk density g/cm3 1 Ammonium adsorption coefficient (a) unitless 1.102 Ammonium adsorption coefficient (b) unitless 0.956 Litter pH unitless 6.2 Litter C structural pool kg/ha 2613 Litter C metabolic pool kg/ha 436 Litter C active pool kg/ha 5 Litter C slow pool kg/ha 3000 Litter C passive pool kg/ha 1900 Litter C lignin kg/ha 9100 Litter ammonium pool kg/ha 1.34 Litter nitrate pool kg/ha 0.24 Litter N structural pool kg/ha 86 Litter N metabolic pool kg/ha 3 Litter N active pool kg/ha 1 Litter N slow pool kg/ha 273 Litter N passive pool kg/ha 173

PAGE 112

100 Parameters Units Values Layer 1 Layer 2 Layer 3 Soil layer parameters (all zones) Pore size distribution index unitless 0.38 0.38 0.38 Bubbling pressure head cm 1.8 1.8 5.6 Starting moisture content cm/cm 0.20 0.31 0.35 Saturated conductivity cm/hr 15 11 1 Sand content % 75 60 51 Silt content % 16 9 12 Clay content % 9 31 38 Bulk density g/cm3 1.4 1.5 1.5 Start carbon structural pool kg/ha 1114 135 135 Start carbon metabolic pool kg/ha 1787 215 215 Start carbon active pool kg/ha 400 400 200 Start carbon slow pool kg/ha 12000 12000 6000 Start carbon passive pool kg/ha 7600 7600 3800 Start carbon lignin pool kg/ha 3360 1680 560 Start nitrogen ammonium pool kg/ha 0.001 0.001 0.001 Start nitrogen nitrate pool kg/ha 0.001 0.001 0.001 Start nitrogen structur al pool kg/ha 1 1 1 Start nitrogen metabolic pool kg/ha 81 10 10 Start nitrogen active pool kg/ha 16 8 8 Start nitrogen slow pool kg/ha 303 157 157 Start nitrogen passive pool kg/ha 191 81 81

PAGE 113

101 Rate constants Values Carbon release rates (litter and soil layers) Metabolic residue pool 0.60 Structural residue pool 0.15 Active humus pool 0.02 Slow humus pool 0.005 Passive humus pool 0.00002 Denitrification rates Litter 0.02 Soil layer 1 0.02 Soil layer 2 0.01 Soil layer 3 0.002

PAGE 114

102 LIST OF REFERENCES Altier, L.S., Lowrance, R.R., Williams, R.G., In amdar, S.P., Bosch, D.D., Sheridan, J.M., Hubbard, R.K., Thomas, D.L. 2002. Riparian Ecosystem Management Model (REMM): simulator for ecological processe s in riparian zones. United States Department of Agriculture, Agricultural Research Service, Conservation Research Report 46. American Public Health Association. 1992. Standard Methods for the Examination of Water and Wastewater, 20th ed ition. Washington, D.C. Auble, G.T., Friedman, J.M., Scott, M.L. 1994. Relating riparian vegetation to present and future streamflows. Ecol ogical Applications 4, 544-554. Band, L.E., Tague, C.L., Groffman, P., Belt, K. 2001. Forest ecosystem processes at the watershed scale: hydrological and ecological controls of nitrogen export. Hydrological Processes 15: 2013-2028. Basnyat, P., Teeter, L., Flynn, K., Lockaby, B. 1999. Relationships between landscape characteristics and nonpoint source pollu tion inputs to coas tal estuaries. Environmental Management 23: 539-549. Berg, B., Staaf, H. 1981. Leaching, accumulation and release of nitrogen in decomposing litter. In Clark, F.E., Rossw all, T. (editors), Terrestrial Nitrogen Cycles. Ecological Bulle tins (Stockholm) 33: 163-178. Bernal, S., Butturini, A., Nin, E., Sabater, F., Sabater, S. 2003. Leaf litter dynamics and nitrous oxide emission in a mediterranean riparian forest: implications for soil nitrogen dynamics. Journal of Environmental Quality 32: 191-197. Beschta, R.L., Jackson, W.L. 1979. The intrus ion of fine sediments into a stable gravel bed. Journal of the Fisheries Re search Board of Canada 36: 207-210. Bhat S., Jacobs, J.M. (Department of Civil Engineering, University of New Hampshire), Hatfield, K. (Department of Civil and Coastal Engineering, University of Florida), Prenger, J. Ecological i ndicators in forested waters heds in Fort Benning, GA: relationship between land use and stream water quality. Ecol ogical Indicators. Accepted pending revision. Biggs, B.J.F. 1996. Hydraulic habitat of plan ts in streams. Regulated Rivers: Research and Management 12: 131-144.

PAGE 115

103 Biggs, B.J.F. 2000. Eutrophication of stream s and rivers: dissolved nutrient-chlorophyll relationships for benthic algae. Jour nal of the North American Benthological Society 19: 17-31. Biggs, B.J.F., Close, M.E. 1989. Periphyton biomass dynamics in gravel bed rivers: the relative effects of flow s and nutrients. Freshwater Biology 22: 209-231. Biggs, B.J.F., Smith, R.A., Duncan, M.J. 1999. Velocity and sediment disturbance of periphyton in headwater streams: biomass and metabolism. Journal of the North American Benthological Society 18: 144-163. Biggs, B.J.F., Stokseth, S. 1996. Hydraulic ha bitat preferences for pe riphyton in rivers. Regulated Rivers: Research and Management 12: 251-261. Binkley, D. 1986. Forest Nutrition Ma nagement. John Wiley & Sons, New York. Bocock, K.L., Gilbert, O., Capstick, C.K., Twinn, D.C., Waid, J.S., Woodman, M.J. 1960. Changes in leaf litter when placed on the surface of soils with contrasting humus types. I. Losses in dry weight of oak and ash leaf litte r. Journal of Soil Science 11:1-9. Bolstad, P.V., Swank, W.T. 1997. Cumulativ e impacts of land use on water quality in a southern Appalachian watershed. Jo urnal of American Water Resources Association 33: 519-533. Bosch, D.D., Sheridan, J.M., Lowrance, R.R. 1996. Hydraulic gradients and flow rates of a shallow coastal plain aquifer in a forest ed riparian buffer. Transactions of the American Society of Agricu ltural Engineers 39: 865-871. Bren, L.J. 1992. Tree invasion of an interm ittent wetland in relation to changes in the flooding frequency of the River Murray, Aust ralia. Australian Journal of Ecology 17: 395-408. Brinson, M.M., Bradshaw, H.D., Holmes, R. N., Elkins, Jr., J.B. 1980. Litterfall, stemflow, and throughfall nutrient fluxes in an alluvial swamp forest. Ecology 61: 827-835. Bryant, M.L., Bhat, S., Jacobs, J.M. In Press. Characterization and modeling of throughfall temporal variability for forest communities in the southeastern U.S. Journal of Hydrology. Bubb, K.A., Xu, Z.H., Simpson, J.A., Saffi gna, P.G. 1998. Some nutrient dynamics associated with litterfall and litter decomposition in hoop pine plantations of southeast Queensland, Australia. Fore st Ecology and Management 110: 343-352. Cairns, Jr., J., McCormick, P.V., Neiderlehn er, B.R. 1993. A proposed framework for developing indicators for ecosystem h ealth. Hydrobiologia 263: 1-44.

PAGE 116

104 Chapman, R.J., Hinckley, T.M., Lee, L.C., Te skey, R.O. 1982. Impact of water level changes on woody riparian and wetland commu nities. Vol. 10. Kearneysville (WV), US Fish & Wildlife Servi ce. Publication No. OBS-82/83. Campbell, G.S. 1974. A simple method for determining unsaturated conductivity from moisture retention data. Soil Science 117: 311-314. Christ, M.J., David, M.B. 1996. Dynamics of extractable orga nic carbon in Spodosol forest floors. Soil Biology and Biochemistry 28: 1171-1179. Cirmo, C.P., McDonnell, J.J. 1997. Linki ng the hydrologic and biogeochemical controls of nitrogen transport in near-stream zone s of temperate-forested catchments: a review. Journal of Hydrology 199: 88-120. Clausen, B., Biggs, B.J.F. 1997. Relations hips between benthic biota and hydrological indices in New Zealand streams. Freshwater Biology 38: 327-342. Clausen, B., Biggs, B.J.F. 2000. Flow i ndices for ecological studies in temperate streams: groupings based on covarian ce. Journal of Hydrology 237: 184-197. Cleveland, C.C., Neff, J.C., Townsend, A. R., Hood, E. 2004. Composition, dynamics, and fate of leached dissolved organic matter in terrestrial ecosystems: results from a decomposition experiment. Ecosystems 7: 275-285. Close, M.E., Davies-Colley, R.J. 1990. Base flow chemistry in New Zealand rivers 2. influence of environmental factors. New Zealand Journa l of Marine and Freshwater Research 24: 343-356. Cole, D.W., Rapp, M. 1981. Elemental cycling in forest ecosystems. In Reichle, D.E. (editor). Dynamic Properties of Forest Ecosystems. International Biological Programme 23. Cambridge University Press, London, pp. 341-409. Closs, G.P., Lake, P.S. 1996. Drought, diffe rential mortality and the coexistence of a native and an introduced fish species in a southeast Australian intermittent stream. Environmental Biology of Fishes 47: 17-26. Cobb, D.G., Galoway, T.D., Flannagan, J.F. 1992. Effects of discharge and substrate stability on density and species composition of stream insects. Canadian Journal of Fisheries and Aquatic Sciences 49: 1788-1795. Colwell, R.K. 1974. Predictability, constanc y, and contingency of periodic phenomena. Ecology 5: 1148-1153. Conner, W.H., Day, Jr., J.W. 1976. Productiv ity and composition of a baldcypress-water tupelo site and a bottomland hardwood s ite in a Louisiana swamp. American Journal of Botany 63: 1354-1364.

PAGE 117

105 Crockford, R.H., Richardson, D.P. 1998. Litte rfall, litter and associated chemistry in a dry sclerophyll eucalypt fore st and a pine plantation in southeastern Australia: 1. litterfall and litter. Hydrologi cal Processes 12: 365-384. Cummins, K.W. 1974. Structure and functi on of Stream ecosystems. BioScience 24: 631-641. Currie, W.S., Aber, J.D., McDowell, W.H., B oone, R.D., Magill, A.H. 1996. Vertical transport of dissolved organic C and N unde r long-term N amendments in pine and hardwood forests. Biogeochemistry 35: 471-505. Dale, V.H., Beyeler, S.C. 2001. Challenge s in the development and use of ecological indicators. Ecologica l Indicators 1: 3-10. Dale, V.H., Beyeler, S.C., Jackson, B. 2002. Understory vegetation indicators of anthropogenic disturbance in longleaf pine forests at Fort Benning, Georgia, USA. Ecological Indicators 1: 155-170. Davies-Colley, R.J., Quinn, J.M. 1998. Str eam lighting in five regions of North Island, New Zealand: control by channel size and riparian vegetation. New Zealand Journal of Marine and Fres hwater Research 32: 591-606. Fisher, S.G. 1983. Succession in streams. In Barnes, J.R., Minshall, G.W. (editors), Stream Ecology: Application and Testing of General Ecological Theory. Plenum Press, New York. Fisher, D.S., Steiner, J.L., Endale, D. M., Stuedemann, J.A., Schomberg, H.H., Franzluebbers, A.J., Wilkinson, S.R. 2000. The relationship of land use practices to surface water quality in the Upper Oc onee watershed of Georgia. Forest Ecology and Management 128: 39-48. Fox, D.G. 1981. Judging air quality model performance: a summary of the AMS workshop on dispersion model performance. Bulletin of Americ an Meteorological Society 62: 599-609. Frick, E.A., Hippe, D.J., Buell, G.R., C ouch, C.A., Hopkins, E.H., Wangsness, D.J., Garrett, J.W. 1998. Water quality in the Apalachicola-Chattahoochee-Flint river basin, Georgia, Alabama, and Florida, 1992–95. U.S. Geological Survey Circular1164. Furniss, M.J., Roeloffs, T.D., Yee, C.S. 1991. Road construction and maintenance. In Meehan, W.R. (editor). Influences of forest and rangeland management on salmonid fishes and their habitats. Speci al publication 19. American Fisheries Society, Bethesda, Maryland. pp. 297-323. Garten, Jr., C.T., Ashwood, T.L., Dale, V.H. 2003. Ef fect of military training on indicators of soil quality at Fort Benning, Georgia. Ecological Indicators 3: 171179.

PAGE 118

106 Gjessing, E., Lygren, E., Berglind, L., Gulb randsen, T., Skanne, R. 1984. Effect of highway runoff on lake water quality. Science of the Total Environment 33: 247257. Gosz, J.R., Likens, G.E., Bormann, F.H. 1972. Nutrient content of litter fall on the Hubbard Brook experimental forest. Ecology 53: 769-784. Gosz, J.R., Likens, G.E., Borman. F.H. 1973. Nutrient release from decomposing leaf and branch litter in the Hubbard Broo k forest, New Hampshire. Ecological Monographs 443:173-191. Grimm, N.B., Fisher, S.G. 1992. Responses of arid-land streams to changing climate. In Firth, P. and Fisher, S.G. (editors). Global Climate Change and Freshwater Ecosystems. Springer, New York, pp. 211-233. Hanson, D.L., Waters, T.F. 1974. Recovery of standing crop and production rate of a brook trout population in a flood-damaged st ream. Transactions of American Fisheries Society 103: 431-439. Harrel, R.C., Dorris, T.C. 1968. Stream order, morphometry, physio-chemical conditions, and community structure of benthic microinvertibrates in an interminnent stream system. American Naturalist 80: 221-251. Harris, M.M., Safford, L.O. 1996. Effects of season and four tree species on soluble carbon content in fresh and decomposing litter of temperate forests. Soil Science, 161: 130-135. Heath, G.W., Arnold, M.K., Edwards, C.A. 1966. Studies in leaf litter breakdown. I. Breakdown rates of leaves of differe nt species. Pedobiologia 6: 1-12. Helvey, J.D., Kochenderfer, J.N. 1990. Soil density and moisture content on two unused forest roads during first 30 months after construction. Research paper NE-629. U.S. Forest Service. Northeast Fo rest Experiment Station, Broomhall, Pennsylvania. Hemphill, N., Cooper, S.D. 1983. The eff ect of physical disturbance on the relative abundances of two filter-feed ing insects in a small str eam. Oecologia 58: 378-383. Herbohn, J.L., Congdon, R.A. 1998. Ecosys tem dynamics at disturbed and undisturbed sites in North Queensland wet tropical rain fo rest. III. nutrient returns to the forest floor through litterfall. Journa l of Tropical Ecology 14: 217-229. Hill, A.R. 1981. Streams phosphorus exports from catchments with contrasting land uses in southern Ohio. Water Resources Bulletin 17: 627-634. Hongve, D. 1999. Production of dissolved or ganic carbon in forested catchments. Journal of Hydrology 224: 91-99.

PAGE 119

107 Horwitz, R.J. 1978. Temporal variability patterns and the dist ributional patterns of stream fishes. Ecological Monographs 48: 307-321. Houser, J.N. (Oak Ridge National Laboratory, Oak Ridge, TN), Mulholland, P.J. (Oak Ridge National Laboratory, Oak Ridge, TN), Maloney, K.O. Catchment disturbance and stream metabolism: patte rns in stream respiration and gross primary production along a gradient of upl and soil and vegetation disturbance. Unpublished manuscript. Hunsaker, C.T., Levine, D.A. 1995. Hierar chical approaches to the study of water quality in rivers. BioScience 45: 193-203. Hynes, H.B.N. 1960. The Biology of Pollu ted Waters, Liverpool University Press, Liverpool. Inamdar, S.P., Sheridan, J.M., Williams, R.G., Bosch, D.D., Lowrance, R.R., Altier, L.S., Thomas, D.L. 1999a. Riparian Ecosystem Management Model (REMM): I. testing of the hydrologic com ponent for a Coastal Plain riparian system. American Society of Agricultural Engineers 42: 1679-1689. Inamdar, S.P., Lowrance, R.R., Altier, L. S., Williams, R.G., Hubbard, R.K. 1999b. Riparian Ecosystem Management Model (REM M): II. testing of the water quality and nutrient cycling component for a Coas tal Plain riparian system. American Society of Agricultural Engineers 42: 1691-1707. Iverson, R.M., Hinckley, B.S., Webb, R.M., Hallet, B. 1981. Physical effects of vehicular disturbance on arid la ndscapes. Science 212: 915-917. Johnes, P., Moss, B., Phillips, G. 1996. Th e determination of total nitrogen and total phosphorous concentrations in freshwater s from land use, stock headage and population data: testing a model for us e in conservation and water quality management. Freshwater Biology 36: 451-473. Johnson, L.B., Richards, C., Host, G.E., Art hur, J.W. 1997. Landscape influences on water chemistry in midwestern stream ecosystems. Freshw ater Biology 37: 193208. Jones, J.A., Grant, G.E. 1996. Cumulative effects of forest harvest on peak streamflow in the western cascades of Oregon. Water Resources Research 32: 959-974. Jowett, I.G., Duncan, M.J. 1990. Flow va riability in New Zealand rivers and its relationship to in-stream habitat and biot a. New Zealand Journal of Marine and Freshwater Research 24: 305-317. Junk, W.J., Bayley, P.B., Sparks, R.E. 1989. The flood pulse concept in river-floodplain systems. Canadian Special Publication of Fisheries and Aquatic Sciences 106: 110-127.

PAGE 120

108 Karr, J.R., Chu, E.W. 1999. Restori ng Life in Running Waters: Better Biological Monitoring. Island Press, Washington, D.C. Keller, E.A., Swanson, F.J. 1979. Effects of large organic material on channel form and fluvial processes. Earth Surface Processes and Landforms 4: 351-380. Lemmon, P. E. 1956. A spherical densiomete r for estimating forest overstory density. Forest Science 2: 314-320. Leopold, L.B., Wolman, M.G., Miller, J.P. 1964. Fluvial processes in geomorphology. W.H. Freeman and Sons, San Francisco. Lowrance, R. 1992. Groundwater nitrate and denitrification in a coastal plain riparian forest. Journal of Environmental Quality 21: 401-405. Lowrance, R.R., Altier, L.S., Williams, R.G., Inamdar, S.P., Sheridan, J.M., Bosch, D.D., Hubbard, R.K., Thomas, D.L. 2000. REMM: the riparian ecosystem management model. Journal of Soil and Water Conservation 55: 27-34. Lowrance, R., Todd, R.L., Asmussen, L.E. 1984. Nutrient cycli ng in an agricultural watershed: II. Streamflow and artificia l drainage. Journa l of Environmental Quality 13: 27-32. Maloney, K.O., Mulholland, P.J., Feminella, J.W. In Press. Influence of catchment-scale military land use on physicochemical condi tions in small Southeastern Plains streams (USA). Environmental Management. McCuen, R.H. 1998. Hydrologic Analysis a nd Design. Pearson Prentice Hall, Upper Saddle River, NJ. McDowell, W.H., Fisher, S.G. 1976. Autumn al processing of dissolved organic matter in a small woodland stream ecosystem. Ecology 57: 561-569. McDowell, W.H., Likens, G.E. 1988. Origin composition, and flux of dissolved organic carbon in the Hubbard Brook Valley. Ecological Monographs 58: 177-195. McGill, W.B., Hunt, H.W., Woodmansee, R.G ., Reuss, J.O. 1981. PHOENIX, a model of the dynamics of carbon and nitrogen in gr assland soils. In Clark, F.E., Rosswall, T. (editors). Terrestrial Nitrogen Cycles. Ecological Bulletin (Stockholm) 33: 49115. Meador, M.R., Goldstein, R.M. 2003. A ssessing water quality at large geographic scales: relations among land use, water phys icochemistry, riparian condition, and fish community structure. Envi ronmental Management 31: 504-517. Meentemeyer, V., Box, E.O., Thompson, R. 1982. World patterns and amounts of terrestrial plant litter pro duction. Bioscience 32:125-128.

PAGE 121

109 Meffe, G.K. 1984. Effects of abiotic dist urbance on coexistence of predator and prey fish species. Ecology 65: 1525-1534. Meffe, G.K., Minckley, W.L. 1987. Persiste nce and stability of fi sh and invertebrate assemblages in a repeatedly disturbed S onoran Desert stream. American Midland Naturalist 117: 177-191. Meyer, J.L., Wallace, J.B. Eggert, S.L. 1998. Leaf litter as a source of dissolved organic carbon in streams. Ecosystems 1: 240-249. Michalzik, B., Kalbitz, K., Park, J.H., Soli nger, S., Matzner, E. 2001. Fluxes and concentrations of dissolved organic car bon and nitrogen – a synthesis for temperate forests. Biogeochemistry 52: 173-205. Milchunas, D.G., Schultz, K.A., Shaw, R.B. 1999. Plant community response to disturbance by mechanized military maneuve rs. Journal of Environmental Quality 28: 1533-1547. Minckley, W.L., Deacon, J.E. 1991. Battle Against Extinction: Native Fish Management in the American West. University of Arizona Press, Tucson, AZ. Minckley, W.L., Meffe, G.K. 1987. Diffe rential selection by flooding in stream-fish communities of the arid American Southwes t. In Matthews, W.J., Heins, D.C. (editors). Community and Evolutionary Ecology of North American Stream Fishes. University of Oklahoma Press, Norman, OK. Minshall, G.W. 1984. Aquatic insect-substr atum relationships. In Resh, V.H. and Rosenberg, D.M. (editors). The Ecology of Aquatic Insects. Praeger Scientific, pp. 358-400. Moore, K.M.S., Gregory, S.V. 1988. Res ponse of young-of-the-year cutthroat trout to manipulations of habitat structure in a sma ll stream. Transactions of the American Fisheries Society 117: 162-170. Moss, D., Furse, M.T., Wright, J.F., Armitage, P.D. 1987. The prediction of the macroinvertebrate fauna of unplotted runni ng-water sites in Gr eat Britain using environmental data. Freshwater Biology 17: 41-52. Mulholland, P.J. 1981. Organic carbon flow in a swamp-stream ecosystem. Ecological Monographs 51: 307-322. Nakamura, F., Swanson, F.J., Wondzell, S.M. 2000. Disturbance regimes of stream and riparian systemsa disturbance cascade pe rspective. Hydrological Processes 14: 2849-2860. Nash, J.E., Sutcliffe, V. 1970. River flow forecasting through conceptual models, I. a discussion of principles. J ournal of Hydrology 10: 282-290.

PAGE 122

110 National Research Council (NRC). 2000. Ecolog ical Indicators for the Nation. National Academy Press, Washington, D.C. Natural Resources Conservation Service (NRCS). 1995. Riparian Forest Buffer, 391. Model State Standard and General Speci fications. NRCS Watershed Science Institute. Seattle, WA. Nykvist, N. 1961. Leaching and decomposition of litter. III. experime nts on leaf litter of Betula verrucosa. Oikos 12: 249-263. Ohte, N., Tokuchi, N., Katsuyama, M., Hobara, S., Asano, Y., Koba, K. 2003. Episodic increases in nitrate concentrations in st reamwater due to the partial dieback of a pine forest in Japan: runoff generation pr ocesses control seasona lity. Hydrological Processes 17: 237-249. Olden, J.D., Poff, N.L. 2003. Redundanc y and the choice of hydrologic indices for characterizing streamflow regimes. Rive r Research and Applications 19: 101-121. OÂ’Neill, R.V., Hunsaker, C.T., Jones, K.B., Riitters, K.H., Wickham, J.D., Schwartz, P.M., Goodman, I.A., Jackson, B.L., Baillargeon, W.S. 1997. Monitoring environmental quality at the lands cape scale. BioScience 47: 513-519. Osborne, L., Wiley, M. 1988. Empirical re lationships between land use/cover patterns and stream water quality in an agricultura l watershed. Journal of Environmental Management 26: 9-27. Park, J., Matzner, E. 2003. Controls on the release of dissolv ed organic carbon and nitrogen from a deciduous forest fl oor investigated by manipulations of aboveground litter inputs and water fl ux. Biogeochemistry 66: 265-286. Parton, W.J., Schimel, D.S., Cole, C.V., Oj ima, D.S. 1987. Analysis of factors controlling soil organic matter levels in gr eat plains grasslands Soil Society of America Journal 51: 1173-1179. Pastor, J., Post, W.M. 1986. Influence of c limate, soil moisture, and succession on forest carbon and nitrogen cycles. Biogeochemistry 2: 3-27. Perkins, H.F., Hook, J.E., Barbour, N.W. 1986. Soil characteristics of selected areas of the Coastal Plain experiment stations, University of Georgia. Research Bulletin 346. Perry, C.D., Vellidis, G., Lowrance, R., Thom as, D.L. 1999. Watershed-scale water quality impacts of riparian forest management. Journal of Water Resources Planning and Management 125: 117-125. Pettit, N.E., Froend, R.H., Davies, P.M. 2001. Identifying the natural flow regime and the relationship with ripari an vegetation for two contrasting western Australian rivers. Regulated Rivers: Res earch and Management 17: 201-215.

PAGE 123

111 Petts, G.E. (editor). 1984. Impounded Rivers : Perspectives for Ecological Management. Wiley, Chichester. Piegay, H., Landon, N. 1997. Promoting ecolo gical management of riparian forest on the Drome River, France. Aquatic C onservation: Marine and Freshwater Ecosystems 7: 287-304. Poff, N.L. 1996. A hydrogeogr aphy of unregulated streams in the United States and an examination of scale-dependence in some hydrological descript ors. Freshwater Biology 36: 101-121. Poff, N.L., Allan, J.D. 1995. Functional or ganization of stream fish assemblages in relation to hydrological vari ability. Ecology 76: 606-627. Poff, N.L., Allan, J.D., Bain, M.B., Karr, J.R ., Prestegaard, K.L., Richter, B.D., Sparks, R.E., Stromberg, J.C. 1997. The natu ral flow regime: a paradigm for river conservation and restoration. BioScience 47: 769-784. Poff, N.L., Ward, J.V. 1989. Implications of streamflow variability and predictability for lotic community structure: a regional anal ysis of streamflow patterns. Canadian Journal of Fisheries and A quatic Sciences 46: 1805-1818. Polglase, P.J., Attiwill, P.M. 1992. Nitroge n and phosphorus cycling in relation to stand age of Eucalyptus regnans F. Mu ell. I. return from plant to soil in litterfall. Plant and Soil 142: 157-166. Power, M.E., Stewart, A.J. 1987. Disturba nce and recovery of an algal assemblage following flooding in an Oklahoma stream. American Midland Naturalist 117: 333-345. Qualls, R.G., Haines, B.L., Swank, W.T. 1991. Fluxes of dissolv ed organic nutrients and humic substances in a deci duous forest. Ecology 72: 254-266. Quist, M.C., Fay, P.A., Guy, C.S., Knapp, A.K., Rubenstein, B.N. 2003. Military training effects on terrestrial and aquatic communities on a grassland military installation. Ecological Applications 13: 432-442. Resh, V.H., Brown, A.V., Covich, A.P., Gurt z, M.E., Li, H.W., Mi nshall, G.W., Reice, S.R., Sheldon, A.L., Wallace, J.B., and Wissma r, R. 1988. The role of disturbance in stream ecology. Journal of the Nort h American Benthological Society 7: 433455. Richards, C., Haro, R.J., Johnston, L.B., Ho st, G.E. 1997. Watershed and reach-scale properties as indicators of macroinvertebrate species trai ts. Freshwater Biology 37: 219-230. Richards, C., Host, G.E. 1994. Examining land use influences on stream habitats and macroinvertebrates: a GIS approach. Water Resources Bulletin 30: 729-738.

PAGE 124

112 Richards, C., Johnston, L.B., Host, G.E. 1996. Assessing the influence of landscapescale watershed features on phys ical habitat and stream biot a. Canadian Journal of Fisheries and Aquatic Sciences 53: 295-311. Richter, B.D., Baumgartner, J.V., Powell, J., Braun, D.P. 1996. A method for assessing hydrologic alteration within ecosystems. Conservation Biology 10: 1163-1174. Rogers, H.M. 2002. Litterfall, decomposition and nutrient release in a lowland tropical rain forest, Morobe Province, Papua New Guinea. Journal of Tropical Ecology 18: 449-456. Schiemer, F., Zalewski, M., Thorpe, J.E. (e ditors). 1995. The Importance of AquaticTerrestrial Ecotones for Freshwater Fish. Developments in Hydrobiology. Volume 105. Kluwer, Dordrecht. Schindler, D., Bayley, S. 1993. The biosphe re as an increasing sink for atmospheric carbon: estimates from increasing nitroge n deposition. Global Biogeochemistry Cycles 7: 717-734. Schlesinger, W.H. 1978. Co mmunity structure, dynamics and nutrient cycling in the Okefenoke cypress swamp forest. Ecological Monographs 48: 43-65. Schlosser, I.J. 1985. Flow regime, juvenile abundance, and the a ssemblage structure of stream fishes. Ecology 66: 1484-1490. Scott, M.L., Friedman, J.M., Auble, G.T. 1996. Fluvial processes and the establishment of bottomland trees. Geomorphology 14: 327-339. Semkin, R.G., Jeffries, D.S., Clair, T. A. 1994. Hydrochemical methods and relationships for study of stream output from small catchments. In Molden, B. and Cerny, J. (editors). Biogeochemistry of Small Catchments: A Tool for Environmental Research. John Wiley and Sons, Chichester, pp. 163-188. Sheridan, J.M., Lowrance, R.R., Henry, H.H. 1996. Surface flow sampler for riparian studies. Applied Engineering in Agriculture 12: 183-188. Shirmohammadi, A.J., Sheridan, J.M., Knisel W.G. 1984. An approximate method for partitioning daily streamflow data. Journal of Hydrology 74: 335-354. Shure, D.J., Gottschalk, M.R. 1985. Litte r-fall patterns within a floodplain forest. American Midland Naturalist 114: 98-111. Sidle, R.C., Pearce, A.J., OÂ’Loughlin, C.L. 1985. Soil mass movementinfluence of natural factors and land use. Americ an Geophysical Union Water Resources Monograph 11, American Geophysical Union, Washington, D.C.

PAGE 125

113 Silveira, M.L., Reddy, K.R. (Department of Soil and Water Science, University of Florida), Comerford, N.B. (Department of Soil and Water Science, University of Florida). Litter decomposition and soluble organic carbon and nitr ogen release in a forested ecosystem. Unpublished manuscript. Singh, K.P., Singh, P.K., Tripathi, S.K. 1999. Litterfall, litter decomposition and nutrient release patterns in four native tree species raised on coal mine spoil at Singrauli, India. Biology and Fertility of Soils 29: 371-378. Sliva, L., Williams, D.D. 2001. Buffer z ones versus whole catchment approaches to studying land use impact on river water quality. Water Research 35: 3462-3472. Sponseller, R.A., Benfield, E.F., Valett, H.M. 2001. Relationships between land use, spatial scale and stream microinvertibra te communities. Fr eshwater Biology 46: 1409-1424. Stone, J. J., Hawkins, R.H., Shirley, E.D. 1994. An approximate form of the Green Ampt infiltration equation. ASCE Journa l of Irrigation and Dr ainage 120: 128-137. Stromberg, J.C., Fry, J., Patten, D.T. 1997. Marsh development afte r large floods in an alluvial, arid-land rive r. Wetlands 17: 292-300. Tipping, E., Woof, C., Rigg, E., Harrison, A. F., Inneson, P., Taylor, K., Benham, D., Poskitt, J., Rowland, A.P., Bol, R., Harkness, D.D. 1999. Climatic influences on the leaching of dissolved organic ma tter from upland UK moorland soils investigated by a field mani pulation experiment. Envir onment International 25: 8395. Thomas, D.L., Beasley, D.B. 1986. A phys ically-based forest hydrology model. I: development and sensitivity of component s. Transactions of the ASAE 29: 962972. Thurow, T.L., Warren, S.D., Carlson, D.H. 199 3. Tracked vehicle tr affic effects on the hydrologic characteristics of central Te xas rangeland. Transactions of the American Society of Agricu ltural Engineers 36: 1645-1650. Ukonmaanaho, L., Starr, M. 2001. The im portance of leaching from litter collected in litterfall traps. Environmental Monitoring and Assessment 66: 129-146. Vannote, R.L., Minshall, G.W., Cummins, K.W., Sedell, J.R., Cushing, C.E. 1980. The river continuum concept. Canadian Journa l of Fisheries and Aquatic Sciences 37: 130-137. Vitousek, P.M., Aber, J.D., Howarth, R.W., Likens, G.E., Matson, P.A., Schindler, D.W., Schlesinger, W.H., Tilman, D.G. 1997. Human alteration of the global nitrogen cycle: sources and consequences. Ecological Applications 7: 737-750.

PAGE 126

114 Wallace, J.B., Benke, A.C. 1984. Quantifica tion of wood habitat in subtropical coastal plains streams. Canadian Journal of Fisheries and Aquatic Sciences 41: 1643-1652. Wang, L., Lyons, J., Kanehl, P., Bannerman R. 2001. Impacts of urbanization on stream habitat and fish across multiple spatial scales. Environmental Management 28: 255-266. Welcomme, R.L. 1992. River conservation futu re prospects. In Boon, P.J., Calow, P., and Petts, G.E. (editors). River Cons ervation and Management. John Wiley and Sons. New York. Wemple, B.C., Jones, J.A., Grant, G.E. 1996. Channel network extension by logging roads in two basins, western Cascades, Oregon. Water Resources Bulletin 32: 1195-1207. Whitecotton, R.C.A., David, M.B., Darmody, R. G., Price, D.L. 2000. Impact of foot traffic from military training on soil a nd vegetation properties. Environmental Management 26: 697-706. Wilson, S.D. 1988. The effects of military tank traffic on prairie: a management model. Environmental Management 12: 397-403. Wood, P.J., Agnew, M.D., Petts, G.E. 2000. Flow variations and macroinvertebrate community responses in a small groundwat er-dominated stream in south-east England. Hydrological Processes 14: 3133-3147. Zalewski, M. 1999. Minimising the risk a nd amplifying opportunitie s for restoration of shallow reservoirs. Hydr obiologia 395/396: 107-114. Zalewski, M. 2000. Ecohydrologythe sc ientific background to use ecosystem properties as management tools toward sustainability of water resources, Ecological Engineering 16: 1-8. Zalewski, M., Janauer, G.A., Jolankai, G. 1997. Ecohydrology: a new paradigm for the sustainable use of aquatic resources. In Conceptual Background, Working Hypothesis, Rationale and Scientific Guid elines for the Implementation of the IHPV Projects 2.3/2.4. UNESCO, Paris T echnical Documents in Hydrology No. 7.

PAGE 127

115 BIOGRAPHICAL SKETCH Shirish Bhat was born in Pokhara, Nepal, on January 12, 1968. He graduated with a bachelorÂ’s degree in science from Tribhuvan University, Nepal, in 1989. He graduated with his second bachelorÂ’s degree in civil engineering from Regional Engineering College, Warangal, India, in 1994. He work ed as a civil engineer for the Canadian Center for International Studies and Coope ration (CECI) in Nepal from July 1994 to August 1995. He also worked as a civil e ngineer consultant for the Department of National Parks and Wildlife Conservation, G overnment of Nepal, from September 1995 to August 1997. In Fall 1997, he began gradua te studies in the Department of Civil and Environmental Engineering at South Dakota State Universi ty, Brookings, South Dakota, USA, and obtained his masterÂ’s degree (the sis option) in Summer 1999. Shirish joined the Department of Civil and Coastal Engineerin g at the University of Florida, Florida, USA, in the Fall of 1999 and will obtain his Ph.D. degree in Spring 2005 under the supervision of Dr. Jennifer Jac obs and Dr. Kirk Hatfield. His primary research interests include land use impacts on surface water qua lity and quantity, riparian area nutrient dynamics, biogeochemistry of wetlands, and water resources management. He is a student member of the American Ge ophysical Union (AGU) and the ASCE.