Assessing patterns and processes of landscape change in Okefenokee Swamp, Georgia

MISSING IMAGE

Material Information

Title:
Assessing patterns and processes of landscape change in Okefenokee Swamp, Georgia
Physical Description:
Book
Creator:
Loftin, Cynthia Smith
Publication Date:

Record Information

Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 30025849
oclc - 41875550
System ID:
AA00020959:00001

Full Text










ASSESSING PATTERNS AND PROCESSES OF LANDSCAPE CHANGE
IN OKEFENOKEE SWAMP, GEORGIA












By

CYNTHIA SMITH LOFTIN


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


1998
























Copyright 1998

by

Cynthia Smith Loftin

























To my family: your love, patience, and encouragement made this work possible.

In memory of my friend, Millicent Quammen, who was an inspiration.














ACKNOWLEDGMENTS


I am indebted to many individuals who contributed to this work. My academic

advisory committee provided direction in my graduate studies, which has contributed to

my continuing development as a scientist. The encouragement, tolerance, interest, and

counsel of my advisor, Dr. Wiley Kitchens, permitted me to explore thoroughly my

research topic, which contributed significantly to its successful conclusion. I am grateful

for his continued confidence. Dr. Ramon Littell assisted with design and analyses of the

vegetation species-hydroperiod and seed bank studies; I appreciate his patience, time,

and assistance. Dr. Jack Putz's keen, editorial eye forced me to clarify many aspects of

this document; his thoroughness and frankness are appreciated. Dr. George Tanner and

Dr. Loukas Arvanitis contributed perspectives and suggestions that improved this

dissertation. I want to thank them for their interest and time. Dr. Franklin Percival

attended my defense and provided thoughtful and humorous discussion throughout my

doctoral program; I appreciate his time and encouragement.

Many students, technicians, and friends assisted with field work, compilation of

GIS and spreadsheet databases, and GIS model development and analyses. My sincere

appreciation is extended to C. Depkin, J. Brookshire, P. Owen, D. Evenson, D. O'Neill,

K. Williges, N. Ansay, P. Gonzalez, J. Aufmnuth, J. Kitchens, J. Halblieb, and J. Loftin for

their tireless dedication. B. Raspberry of the USFWS assisted with the GPS topography









survey; without his involvement, the hydrology study would not have been possible. C.

Trowell made his summaries of the Okefenokee Swamp human history readily available;

I want to thank him for bringing many details to my attention. Fellow graduate students

and coop unit employees L. Brandt, J. Silveira, C. Allen, K. Rice, A. Garmestani, W.

Bryant, D. Hughes, and B. Fesler contributed thoughtful discussion, administrative

assistance, and much-needed levity. L. Pearlstine guided me through my initial foray

into the world of GIS, and W. Hyde provided excellent computer systems management; I

am grateful for their patience.

The staff of the Okefenokee National Wildlife Refuge deserve my deepest

gratitude for their assistance. They permitted unlimited access to records and the swamp,

as well as assisted in GPS surveys, water level recorder maintenance, and provided

housing and equipment upon request. S. Reeves, S. Aicher, L. Mallard, J. Burkhart, S.

Davis, R. Phernetton, T. Huling, S. Jones, and C. Thompson, as well as many others, are

dedicated stewards of that natural treasure. Foresight of the USFWS and refuge

managers to recognize that a study of the effects of the Suwannee River sill on the

swamp hydrology and ecology was needed and to provide sufficient funding to address

the salient questions was integral to initiation of this study.

My family continues to give encouragement and love, which have fueled me

throughout this effort; I want to thank Dorothy for her patient, loving care of Steven, and

Kelly for her friendship and super babysitting. Finally, I thank Jim for his understanding,

interest, and love, which enabled me to attack this project, and without whom I could not

have completed it, and Steven, who has shown me another facet of life.
















TABLE OF CONTENTS



ACKNOWLEDGMENTS ................................................ iv

LIST OF TABLES ...................................................... x

LIST OF FIGURES .................................................... xx

ABSTRACT ........................................................ xxviii

CHAPTERS

1 THE OKEFENOKEE SWAMP AND THE SUWANNEE RIVER SILL ...... 1

Ecosystem Hierarchies and Scales ..................................... 5
Driving Functions .............. ................................. 11
Disturbances, Heterogeneity, and Succession .......................... 15
Monitoring Landscape Change ...................................... 19
The Okefenokee Swamp Ecosystem .................................. 20
Human Modification of Okefenokee Swamp .......................... 28

2 DATABASE ORIGIN AND DEVELOPMENT ....................... 34

Data Sources and Extent ........................................... 34
Swamp Water Level Data .......................................... 36
Water Level Recorder Performance ................................. 39
Estimation of Missing Water Level Data ............................. 71
Precipitation Gauge Network Assessment............................. 83
Background ............................................... 83
M ethods .................................................. 87
Results of Precipitation Network Assessment.................... 89
Discussion of Precipitation Network Analysis .................. 106
Estimation of Missing Precipitation Data ............................ 107
Estimation of Evapotranspiration, Inflow, and Outflow Data .............. 108









Swamp Basin Delineation and Characterization ....................... 120
Effects of the Suwannee River Sill on Swamp Water Level Conditions ..... 130
Topography Surface .............................................. 146
Collection of Point Elevation Data .................................. 150
Surface Interpolation ............................................. 160
Topography Surface Description and Trends ......................... 164
Satellite Imagery Classification and Accuracy Assessment ............... 172
Image Preparation ............................................... 173
Image Classification ............................................. 176
Image Classification Accuracy Assessment .......................... 178
Image Classification and Accuracy Results ........................... 193
Interpreting the Accuracy Assessment................................ 199
Applying Image Classification Procedures ........................... 202

3 OKEFENOKEE SWAMP HYDROLOGY MODEL ................... 207

Introduction .................................................... 207
M ethods ....................................................... 209
M odel Objective .......................................... 209
M odel Overview .......................................... 210
Model Data Sources ........................................ 220
Precipitation .............................................. 220
Evapotranspiration ......................................... 221
Creek Inflow Volumes ...................................... 221
River Outflow Volumes ..................................... 222
Water Depth and Topographic Surfaces ....................... 223
Data Surfaces Used for Model Assessment..................... 223
Model Manipulation and Assessment ......................... 225
Wildfires in the Area Affected by the Suwannee River Sill ......... 227
Vegetation in the Area Affected by the Suwannee River Sill ........ 228
R results ........................................................ 229
Area Affected by the Suwannee River Sill ...................... 229
Model Accuracy: 1980-1993 ................................. 229
Model Responses to Sill Manipulations ....................... 267
Regional Hydrologic Trends ................................. 314
System Sensitivities ........................................ 348
Wildfire Occurrence ....................................... 361
Vegetation Change ......................................... 364
D discussion ..................................................... 375
Model Performance ........................................ 375
Effects of the Suwannee River Sill ............................ 379
System Sensitivities ................................. ....... 382









4 LANDSCAPE LEVEL VEGETATION CHANGES IN OKEFENOKEE
SW AM P ................................................. 385

Introduction .................................................... 386
Dynamics of Swamp Vegetation .............................. 386
Sill Affected Vegetation Change .............................. 390
M ethods ....................................................... 391
Logging Tramlines ......................................... 391
Pre-Logging Vegetation (1850-1890) ......................... 395
Post-Logging Vegetation (1952) .............................. 398
17 Years With-Sill and 22 Years Post-Fire Vegetation (1977) ....... 410
30 Years With-Sill and 35 Years Post-Fire (1990) ................ 411
Wildfire Bumrn Area Maps ................................... 411
Map Comparisons ......................................... 415
R results ........................................................ 415
Overall Changes in Vegetation Distributions and Composition ...... 415
Logging Impacts ........................................... 428
Fire and Vegetation Change .................................. 436
Vegetation Changes in the Areas Affected by the Suwannee River
Sill ............................................... 438
D discussion ..................................................... 465

5 FIRE IN OKEFENOKEE SWAMP ................................ 483

Introduction .................................................... 483
M ethods ....................................................... 488
Wildfires and Prescribed Fires ................................ 488
Wildfire Occurrences and Vegetation Types .................... 491
R results ........................................................ 492
Fire Sizes, Frequencies, and Causes .......................... 492
Vegetation Changes Where Fires Occurred ..................... 505
Vegetation Changes Regardless of Fire Occurrence ............... 514
Wildfire and Logging ....................................... 530
Recurrence of Fires ........................................ 531
Fire Occurrence and Water Levels............................. 532
Fire and the Suwannee River Sill ............................ 541
D discussion ..................................................... 555

6 RELATIONSHIPS OF OKEFENOKEE SWAMP VEGETATION
DISTRIBUTIONS AND THE HYDROLOGIC ENVIRONMENT .............. 564

Introduction .................................................... 564
M ethods ....................................................... 568


viii









Vegetation Sampling ....................................... 568
Preparation of Hydrologic Data ............................... 574
Analysis of Vegetation Data ................................. 579
R results ........................................................ 585
Species' Environments ..................................... 585
Species' Environments and Modeled Hydrologic Changes.......... 681
D discussion ..................................................... 681
Species' Associations and the Hydrologic Environment............ 681
Vegetation Changes Due Sill Impoundment Effects .............. 697

7 RESPONSE OF THE OKEFENOKEE SWAMP SEED BANK TO
ALTERATIONS IN THE HYDROLOGIC ENVIRONMENT ...... 702

Introduction .................................................... 702
M ethods ....................................................... 707
Seed Bank Sampling ....................................... 707
Analysis of Seed Bank Emergence Data ....................... 712
Results ........................................................ 713
Species' Responses ........................................ 713
Trends in Response to Hydrologic Conditions .................. 732
D discussion ..................................................... 744
Wetland Seed Bank Composition and Vegetation Community
Dynamics .......................................... 744
Effects of the Suwannee River Sill on the Okefenokee Swamp Seed
Bank .............................................. 757

8 SUMMARY AND CONCLUSIONS ............................... 760

Human Activity in the Okefenokee Swamp .......................... 761
Okefenokee Swamp Hydrology ..................................... 764
Okefenokee Swamp Vegetation ................................... 766
The Okefenokee Swamp Landscape ................................ 768

APPENDIX A SUWANNEE RIVER SILL AUTHORIZATION BY CONGRESS. 770

APPENDIX B COMPUTER MODEL CODE FOR HYDRO-MODEL .......... 772

APPENDIX C VEGETATION TRANSECT LOCATIONS ................. 815

LIST OF REFERENCES ............................................... 819

BIOGRAPHICAL SKETCH .....................................835















LIST OF TABLES


Table pagg

1-1. Human-caused manipulations of Okefenokee Swamp vegetation and
topography occurring during the past 150 years ......................... 29

2-1. Water level recorder elevations and staff corrections, operating period, and
precipitation gauge locations ........................................ 40

2-2. Summary parameters of water level recorders installed at Okefenokee National
Wildlife Refuge during 12-5-1979 through 6-15-1995. Elevations are in
meters above mean sea level. Basin delineation is discussed in the Swamp
Basin Delineation and Characterization section ......................... 66

2-3. Summary of water level and precipitation recorder performance during
12-5-1979 through 6-15-1995 at Okefenokee National Wildlife Refuge ...... 68

2-4. Summary of water level recorder performance prorated to the initial operating
period(1054 days) ................................................ 72

2-5. Best correlation pairs and regression equations used to estimate missing water
level recorder data during 1941-1995 ................................. 75

2-6. Daily average precipitation estimated with measurements made daily, and
approximated with biweekly or monthly calculations of daily averages
during 31 March 1992-3 July 1995 ................................... 88

2-7. Stratum and station variances and covariances of daily precipitation estimates,
averaged by day, biweekly, and monthly during 31 March 1992-3 July 1995.
Symbology is defined in the chapter text.............................. 90

2-8. Relative, spatial, and total variances within the precipitation gauge network,
network accuracy with various recorder densities, and suggested strata
allocation of precipitation gauges .................................... 95









2-9. Best correlation pairs and regression equations used to estimate missing
precipitation recorder data for use in HYDRO-MODEL, during 1930-1993.. 109

2-10. Regression relationships used to estimate river and creek outflow rates from
Okefenokee Swamp during 1930-1993 .............................. 114

2-11. Estimating creek flow into Okefenokee National Wildlife Refuge using water
depth estimates from creek staffs ................................... 116

2-12. Regression relationships used to estimates water depths at staffs in
northwestern creeks from flow measurements at the Suwannee River-Fargo
gauge ......................................................... 117

2-13. Monthly latitude adjustment to account for seasonal radiation in calculation
of Thornthwaite's PE (from Thornthwaite (1948)) ..................... 118
2-14. Regression equations used to estimate missing daily maximum air
temperature at NOAA weather stations around Okefenokee National
W wildlife Refuge ................................................. 119

2-15. Comparison of flow rates measured at the Suwannee River (Fargo) and the St.
Marys River (Moniac) gauges before and after construction of the Suwannee
River Sill, during 1930-1993 ....................................... 132

2-16. Comparison of pre-sill and with-sill biweekly total precipitation volumes at
SCFSP and SCRA during 1930-1995 .............................. 134

2-17. Comparison of evapotranspiration (ET) estimates in the Okefenokee Swamp
area before and after Suwannee River sill construction, 1930-1993 ......... 137

2-18. Comparison of SCFSP and SCRA water surface elevations above mean sea
level (AMSL) before and after construction of the Suwannee River sill,
1941-1995 ..................................................... 141

2-19. Differences in mean monthly precipitation among decades at SCFSP and
SCRA, and 95% confidence intervals. No differences were significant at
a < 0.05. Data were log-normalized before comparisons were made ....... 143

2-20. Differences in mean biweekly water surface elevation among decades at
SCFSP and SCRA, and 95% confidence intervals. Differences marked
with are significant at a < 0.05 .................................... 147









2-21. Elevations and locations of benchmarks established in the Okefenokee
Swamp National Wildlife Refuge and perimeter, and peat and sand surface
elevations above men sea level (AMSL) at each site ................... 152

2-22. Peat thickness values used to estimate peat depth by vegetation type, to
supplement the coverage of estimated peat depths where data gaps exist .... 165

2-23. Composition and area of classes in the Okefenokee Swamp satellite image
classification ................................................... 179

2-24. Vegetation species found in ground-truthed sites used in the satellite image
classification ................................................... 182

2-25. Error matrix for the 11-class satellite image classification, within 10 pixels
(100 m) of ground truth sample point location. Rows are reference data;
columns are classification data. Cell values are number of sample points ... 184

2-26. Error matrix for the 13-class satellite image classification, within 10 pixels
(100 m) of ground truth sample point location. Rows are reference data;
columns are classification data. Cell values are number of sample points. .. 186

2-27. Error matrix for the 17-class satellite image classification, within 10 pixels
(100 m) of ground truth sample point location. Vegetation class number
refers to Table 2-23. Rows are reference data; columns are classification
data. Cell values are number of sample points ........................ 188

2-28. Error matrix for the 22-class satellite image classification, within 10 pixels
(100 m) of ground truth sample point location. Vegetation class number
refers to Table 2-23. Rows are reference data; columns are classification
data. Cell values are number of sample points ....................... 190

2-29. Categorical Kappa coefficients (K,,), user's and producer's accuracies for
classes with > 18 ground-truthed sites in the 11-, 13-, 17-, and 22-class
classifications within 10 pixels (100 m) of sample location. .............. 195

2-30. Class confusions in the 11-, 13-, 17-, and 22-class classifications within 10
pixels (100m) of sample location, for classes with user's accuracy <80%.
The class with the most frequent error is underlined ................... 197

2-31. Error matrix of classes in the swamp-islands-and-uplands map that are
combined with those of the 17-class map. Cell values are number of
samples re-classified in the swamp-islands-and-uplands classification from
classes in the 17-class map ......................................... 200









3-1. Manning's roughness coefficients used in HYDRO-MODEL to adjust surface
water flow rates over various substrates (adapted from Ward (1996)). ..... 219

3-2. Best HYDRO-MODEL settings and check data format for stations in
Okefenokee Swamp during 1941-1993 model simulations ................ 260

3-3. Comparison of check station data and best model output, 1980-1993 ........ 268

3-4. Summary statistics of recorder data and model output at check stations during
1980-1993 ..................................................... 271

3-5. Water depth ranges for hydroperiod group delineations ................... 282

3-6. Comparisons of changes in water depths and hydroperiod group frequencies in
with-sill and no-sill model simulations, 1941-1993. Stations with poor
with-sill model versus recorder agreement in 1980-1993 are omitted ........ 283

3-7. Comparison of changes in growing season and non-growing season
hydroperiod group frequencies in with-sill and no-sill model simulations,
1941-1993. Stations with poor with-sill versus recorder agreement in
1980-1993 are omitted ............................................ 315

3-8. Comparison of water depths and changes at the south Sill Gate and other check
stations throughout the Okefenokee Swamp during 1980-1993 with-sill,
no-sill, and no-outflow model simulations ............................. 318

3-9. Vegetation changes occurring in areas affected by the sill and burned during
1855-1993 ..................................................... 365

3-10. Composition of vegetation during 1952, 1977, and 1990 throughout the
Okefenokee Swamp, the floodplain sill impoundment impact area, and the
Cypress Creek watershed area. All calculations were made with 6-class
vegetation maps with a minimum mapping unit of 320 m; comparison areas
for the sill area of impact (AOI) are clipped to match the area interpreted
from 1952 photography, and reported values are % of the vegetation in each
category inside and outside the sill AOI during the specified interval ........ 371

3-11. Rate of change in vegetation composition during 1952-1977 and 1977-1990
throughout the Okefenokee Swamp, the floodplain sill impoundment affected
area, and the Cypress Creek watershed area. All calculations are made
with 6-class vegetation maps with a minimum mapping unit of 320 m;
comparison areas for the sill area of impact (AOI) are clipped to match the
area interpreted from 1952 photography, and reported values are % of the









vegetation category change occurring in the specified interval. Overall
change refers to that occurring in the entire swamp, including the sill AOI,
during that interval. Bare Ground-Urban and Open Water classes were not
interpreted in the 1977 map, and are omitted from these comparisons. ..... 373

4-1. Sources of pre-logging survey notes used to create the pre-logging vegetation
map of Okefenokee Swamp ........................................ 396

4-2. Vegetation class descriptions and merges created for comparing maps of
Okefenokee Swamp vegetation distributions during 1990, 1977, 1952, and
before logging occurred (1850-1890) ................................. 399

4-3. Date groupings for wildfire map sets and for vegetation distribution
comparisons .................................................... 414

4-4. Okefenokee Swamp vegetation composition estimated from pre-logging
surveys conducted during 1850-1890 ................................ 417

4-5. Okefenokee Swamp vegetation composition estimated from an 11 May 1990
SPOT satellite image ............................................. 419

4-6. Okefenokee Swamp vegetation composition estimated from 1952 black and
white aerial photography ......................................... 421

4-7. Estimated Okefenokee Swamp vegetation composition compiled from 1977
color-infrared photography interpreted by McCaffrey and Hamilton (1980)
with a minimum mapping unit of 320 m ............................. 422

4-8. Landscape level vegetation changes occurring in Okefenokee Swamp during
1850-1951. Minimum mapping unit for the comparison is 240 m. Reported
values are % of the vegetation class in 1850 occurring in the specified class
in 1952 ....................................................... 425

4-9. Landscape-level vegetation changes occurring in Okefenokee Swamp during
1952-1977. Minimum mapping unit for the comparison is 320 m. Reported
values are % of the vegetation class in 1952 occurring in the specified class
in 1977 ....................................................... 426

4-10. Landscape level vegetation changes occurring in Okefenokee Swamp during
1977-1990. Minimum mapping unit for the comparison is 320 m. Reported
values are % of the vegetation class in 1977 occurring in the specified class
in 1990 ....................................................... 427









4-11. Estimated composition of logging tramline areas before logging occurred,
recorded in surveys conducted during 1850-1890 ...................... 429

4-12. Estimated composition during 1952 of areas previously logged ............. 430

4-13. Estimated composition during 1977 of areas previously logged ............. 431

4-14. Estimated composition during 1990 of areas previously logged ............. 432

4-15. Proportions of the entire swamp and logged areas that remained in persistent
vegetation types between intervals, and the predominant type of
replacement where changes occurred during 1952-1977 and 1977-1990 ..... 434

4-16. Vegetation changes occurring during 1952-1977 in areas logged during
1890-1942. Minimum mapping unit for the comparison is 320 m. Values
are % of the vegetation class in 1952 occurring in the specified class
in 1977 ........................................................ 435

4-17. Vegetation changes occurring during 1977-1990 in areas logged during
1890-1942. Minimum mapping unit for the comparison is 320 m. Values
are % of the vegetation class in 1977 occurring in the specified class
in 1990 ........................................................ 437

4-18. Vegetation types that burned after 1855 and before 1952, and the types of
vegetation that occurred in the burned areas in 1952 ................... 439

4-19. Logging tramline fuel load estimates ................................. 440

4-20. Fuel load composition for fires occurring during 1855-1951, 1952-1976,
and 1977-1990 ................................................. 441

4-21. Proportion of wildfires in logged and unlogged tramline areas ............. 442

4-22. Vegetation types that burned during 1954-1955, and the types of vegetation
that occurred in the burned areas by 1977 ............................ 443

4-23. Vegetation types that burned after 1955 and before 1990, and the types of
vegetation that occurred in the burned areas in 1990 ................... 444

4-24. Vegetation changes occurring during 1952-1990 in the river floodplain area
most likely affected by the sill's impoundment and in the Cypress Creek









watershed area. Minimum mapping unit for the comparison is 240 m.
Values are % of the vegetation class in 1952 occurring in the specified
class in 1990 ................................................... 446

4-25. Vegetation changes occurring during 1952-1990 in the floodplain area
most likely affected by the sill's impoundment and in the Cypress Creek
watershed area. Classes from the 1990 map have not been grouped; values
are % of the vegetation class in 1952 occurring in the specified class of the
ungrouped map in 1990. Minimum mapping unit for the comparison
is 240 m ........................................................44 8

4-26. Vegetation changes occurring during 1977-1990 in the floodplain area most
likely affected by the sill's impoundment effects and the Cypress Creek
watershed area. Neither map consisted of grouped classes; values are % of
the vegetation class in the ungrouped 1977 map in the specified class of the
ungrouped map in 1990. Minimum mapping unit for the comparison
is 240 m ....................................................... 452

4-27. Vegetation changes occurring during 1952-1977 in the floodplain area
most likely affected by the sill's impoundment effects and the Cypress
Creek watershed. The 1977 map did not consist of grouped classes; values
are % of the vegetation class in the 1952 map in the specified class of the
ungrouped map in 1977. Minimum mapping unit for the comparison
is 320 m ....................................................... 460

4-28. Vegetation changes occurring in Okefenokee Swamp during 1855-1990.
Values are % of the prelogging vegetation in the specified class in each
class during 1990. Minimum mapping unit for the 1990 map is 10 m;
interpretation of the prelogging survey notes is on a much greater scale,
from summarization of narratives and observations .................... 466

4-29. Vegetation changes occurring in Okefenokee Swamp during 1977-1990.
Values are % of the vegetation from the specified 1977 class changing to
the specified vegetation type by 1990. Minimum mapping unit for the
m aps is 320 m ................................................... 472

5-1. Fuel models used in fuel load calculations, from Anderson (1982) ............ 493

5-2. Summary of wildfires in the Okefenokee Swamp National Wildlife Refuge
area, 1855-1993 .................................................. 495

5-3. Vegetation in 1990 in areas that burned during 1954-1955 .................. 511









5-4. Vegetation that burned during 1990-1993 .............................. 512

5-5. Types of vegetation changes in 1952-1977, and their proportions in area
burned and not burned during 1855-1952, 1952-1977, and 1977-1990 ...... 515

5-6. Types of vegetation changes in 1977-1990, and their proportions in area
burned and not burned during 1952-1955, 1977-1990, and 1990-1993 ...... 521

5-7. Composition of logging harvest during 1909-1927, from Hopkins (1947) and
Izlar (1984) ..................................................... 530

5-8. Proportion of burned areas that repeatedly burned ....................... 533

5-9. Spearman rank order correlation comparisons (r, P) of wildfire size, water
depths, and wildfire cause, for wildfires occurring during 1941-1993 ....... 534

5-10. Wildfires occurring in the Suwannee River floodplain and Cypress Creek
watershed areas affected by the sill, water level conditions when the fires
ignited, and the water level condition that would have been required to
create impounded surface water and arrest the spread of these fires ........ 551

6-1. Structural zone types recognized along sampled topographic/hydrologic
gradients in Okefenokee Swamp .................................... 572

6-2. Recorders and nearest survey benchmarks used to estimate water surface
elevations at vegetation transects during 1960-1995 .................... 576

6-3. Inundation depth classes defined for analysis of species occurrence in
hydrologic environments .......................................... 578

6-4. Hydrologic environments during 1962-1995 of species occurring in vegetation
sample plots during 1993-1994. Water depth conditions (DC) are described
in the table footnote ............................................. 593

6-5. t-test (mean water depth) and Wilcoxon rank-sum (percent of interval in each
depth class) comparisons of hydrologic environments during 1962-1995
where species were present and absent in vegetation sample plots during
1993-1994 ..................................................... 607

6-6. Significant parameters in the species-environment multiple regression models
using all herbaceous and woody species sample data (logit-transformed)
regardless of species presence ........................................... 637


xvii








6-7. Significant parameters in the species-environment multiple regression models
using all herbaceous and woody species sample data regardless of species
presence, when inundation depths are shallow (0 < depth <0.30 m). ...... 640

6-8. Significant parameters in the species-environment multiple regression models
using understory, shrub, and tree sample data where species are present ..... 643

6-9. Significant parameters in the species-environment multiple regression models
using herbaceous and woody species data where species are present and
inundation depths are shallow (0 < depth < 0.30 m) ..................... 646

6-10. Associations of vegetation species based on average and standard deviation of
daily water depths measured or estimated at transect sample sites during
1962-1995. Values in column headings are the ranges of average water
depths the ranges of standard deviations ............................. 648

6-11. Associations of vegetation species based on similar 3-dimensional plots of
modeled species occurrence and daily depth-inundation duration
relationships at transect sample sites during 1962-1995. ................. 679

6-12. Predicted changes in biweekly water depth range (from depth classes) and
variability by swamp region, predicted by the swamp hydrology model with
sill removal (summarized from Figure 3-18)........................... 682

7-1. Species germinating in the seed bank samples, and their distributions among
areas, seasons, and treatments ..................................... 714

7-2. Standing vegetation species absent from the Okefenokee Swamp seed bank
samples, but present in plots of established vegetation.................... 719

7-3. Average number of species in the established vegetation and in the seed bank
from structural zones throughout Okefenokee Swamp ................... 723

7-4. Modeled parameters and their significance in predicting responses of seed
bank species to area, structural zone type, and treatment................. 726

7-5. Hydrologic environments where seed bank species are found in Okefenokee
Swamp, and areas of maximum species abundances in seed bank samples
and established vegetation ........................................ 733

7-6. Effects of season on response of seed bank samples (counts) collected from
hydrologic zone types and areas of Okefenokee Swamp .................. 736


xviii








7-7. Germination and dispersal characteristics of Okefenokee Swamp seed bank
species, summarized from field observation, seed bank samples,
Porcher (1995), and Conti and Gunther (1984) ......................... 738
















LIST OF FIGURES


Figure paM

1-1. The four ecosystem functions and their relationships to the amount of stored
capital and degree of connectedness, from Holling (1986) ................. 8

1-2. Interactions of temporal and spatial scales and the processes shaping a
landscape and its components (adapted from Holling (1992)) .............. 10

1-3. Interactions of disturbances and system potential energy levels leading to
instability and restructuring (adapted from Forman and Godron (1986)) ...... 12

1-4. Location of the Okefenokee Swamp and Okefenokee Swamp National
W wildlife Refuge .................................................. 21

1-5. Hypothetical community changes occurring with peat accumulation in the
absence of disturbance in Okefenokee Swamp (adapted from
Hamilton (1982)) ................................................ 26

2-1. Water level and precipitation recorder locations in the Okefenokee Swamp
during 1941-1995 ................................................. 38

2-2. Daily water surface elevation above mean sea level (AMSL) during 1941-May
1995 recorded at locations in the Okefenokee National Wildlife Refuge, GA.. 42

2-3. Recorder distribution for daily measurement of precipitation at 11 stations in
Okefenokee Swamp ............................................... 98

2-4. Recorder distribution for daily measurement of precipitation at 14 stations in
Okefenokee Swamp ............................................... 99

2-5. Recorder distribution for biweekly measurement of precipitation at 11 stations
in Okefenokee Swamp ............................................ 100

2-6. Recorder distribution for biweekly measurement of precipitation at 14 stations
in Okefenokee Swamp ............................................ 101

xx








2-7. Recorder distribution for monthly measurement of precipitation at 11 stations
in Okefenokee Swamp ............................................ 102

2-8. Recorder distribution for monthly measurement of precipitation at 14 stations
in Okefenokee Swamp ............................................ 103

2-9. Recorder locations for daily measurement of air temperature and surface water
inflows in Okefenokee Swamp ..................................... 113

2-10. Variance contours for water depths recorded daily at locations in the
Okefenokee Swamp during 1992-1995 ............................... 121

2-11. Water level recorder locations and hydrologic basins in Okefenokee
Sw am p ........................................................ 123

2-12. Trends in water level fluctuations in the Okefenokee Swamp hydrologic
basins ........................................................ 124

2-13. Monthly average evapotranspiration estimated in the Okefenokee Swamp
area during 1930-1993 ............................................ 126

2-14. Average monthly precipitation estimated at sites in the Okefenokee Swamp
area during 1930-1993 ............................................ 127

2-15. Average daily inflow estimated for northwestern creeks entering Okefenokee
Swamp during 1930-1993 ......................................... 128

2-16. Average daily outflow estimated for creeks and rivers exiting Okefenokee
Swamp during 1930-1993 ......................................... 129

2-17. Average monthly precipitation and water surface elevation reported by 5-year
intervals at SCFSP, during 1941-1994 ................................ 149

2-18. Locations surveyed and extracted from USGS 1994 1:24,000 topographic
maps for development of the Okefenokee Swamp topographic surface ...... 161

2-19. Correction surface used to adjust the swamp topography map, generated from
the difference between the actual and interpolated surface elevation data .... 162

2-20. Peat and sand surface topography in Okefenokee Swamp. Darker areas are
lower in elevation above mean sea level ............................. 163

2-21. Estimated thickness of peat over the basement sands in Okefenokee
Swam p ........................................................ 166

xxi









2-22. Estimated sand surface elevation above mean sea level under the surface
peat in Okefenokee Swamp ........................................ 167

2-23. Elevation gradients recorded along transects in Okefenokee Swamp prairies .. 168

2-24. Creeks, rivers, lakes, and canoe trails where surface water flow occurs in the
Okefenokee Swamp .............................................. 170

2-25. Surface water drainage patterns and underlying topographic gradients in
Okefenokee Swamp ............................................ 171

2-26. Merging 10 m pixel panchromatic and 20 m pixel multispectral imagery to
create a multispectral image with 10 m pixel resolution ................. 175

3-1. Processing area for the Okefenokee Swamp HYDRO-MODEL .............. 211

3-2. Flowchart of Okefenokee Swamp HYDRO-MODEL components ........... 212

3-3. Neighborhood search in HYDRO-MODEL to determine direction and amount
of water to move in each cell and time interval ....................... 214

3-4. HYDRO-MODEL menu interface for setting user-defined parameters ........ 215

3-5. Locations of zones used in HYDRO-MODEL to distribute inflowing water
across the landscape .............................................. 217

3-6. Location of zone used in HYDRO-MODEL to extract outflowing water from
the Suwannee River floodplain ..................................... 218

3-7. Locations of water level recorders used to assess HYDRO-MODEL
performance .................................................... 224

3-8. Estimated topographic surface representing the pre-sill peat surface elevations.
Dark areas are low in elevation ..................................... 226

3-9. Estimated area of impact of the Suwannee River sill on the Okefenokee
Swamp hydrologic environment during various water level conditions, and
regions that may experience head reversals when water levels are high ...... 230

3-10. Estimated recorder data and model output from the "with-sill" and "no-sill"
simulations for 1980-1993 ......................................... 231


xxii









3-11. Estimated recorder data and model output from stations with poor model
performance in "with-sill" and "no-sill" simulations for 1980-1993 ........ 251

3-12. Inverse-distance-weighted, contoured estimates of increases in average semi-
monthly water surface elevations (m) at recording stations, attributed to the
Suwannee River sill during 1980-1993 ............................... 276

3-13. Comparison of semi-monthly water surface elevations at Cypress Creek and
Sapp Prairie under increasing water level conditions in the sill gate area
during 1980-1993 ............................................... 277

3-14. Comparison of average, semi-monthly water surface elevations in the
Cypress Creek watershed under low, average, high, and very high water
levels in the sill gate area during 1980-1993 .......................... 278

3-15. Locations of topographic highs in the Suwannee River floodplain near the
Suwannee River sill and Cypress Creek .............................. 280

3-16. Comparison of average, semi-monthly water surface elevations in the
Sweetwater Creek watershed under low, average, high, and very high water
levels in the sill gate area during 1980-1993 ........................... 281

3-17. Changes in most frequent hydroperiod groups during 1980-1993, with sill
removal. Numbers represent the most frequent with-sill hydroperiod groups
(see Tables 3-5 and 3-6) versus most frequent no-sill hydroperiod groups.
Areas with significant change are marked with *. Average semi-monthly
water depth decrease with sill removal is noted in ( ) .................... 294

3-18. Changes in hydroperiod depth group frequencies with and without the sill
during 1941-1993, by decade intervals ............................... 295

3-19. Comparison of semi-monthly water surface elevations at recorder stations
and the sill gate during 1980-1993, in "with-sill" and "no-sill" model
simulations. Data are arranged to illustrate the change in water levels
at the selected stations, relative to increasing water depth at the sill ......... 335

3-20. Estimated average biweekly flow rates at Suwannee River (Fargo) and St.
Marys River (Moniac) during 1980-1993 ............................ 345

3-21. Locations of topographic change in the Suwannee River floodplain within the
Okefenokee Swamp and southwest of the Suwannee River sill ............. 347

3-22. Inverse-distance-weighted, contoured estimates of increases in average semi-
monthly water surface elevations (m) at recording stations, attributed to


xxiii









Suwannee River outflow retained in the swamp during 1980-1993 model
sim ulations .................................................... 349

3-23. Manipulations of estimated evapotranspiration rates and responses of the
model at recorder stations during 1980-1993 ........................... 351

3-24. Areas affected by the Suwannee River sill at various water level conditions,
and burned by wildfires during 1960-1993 ............................. 362

4-1. General effects of fire and logging disturbances on Okefenokee Swamp
vegetation types, adapted from Hamilton (1982) ........................ 387

4-2. Autogenic succession and conditions that drive succession in the Okefenokee
Swamp, adapted from Hamilton (1982) ............................. 389

4-3. Locations of logging railroads tramliness) and buffer used to approximate
logged area .................................................... 394

4-4. Estimated pre-logging (1850-early 1900s) vegetation in Okefenokee Swamp,
and approximate routes of 19th and 20th century surveyors .............. 397

4-5. Photo interpretation results of Okefenokee National Wildlife Refuge
vegetation during 1952 ........................................... 409

4-6. Scanned, transformed version of map created by McCaffrey and Hamilton
(1980) of Okefenokee Swamp vegetation during 1977 ................... 412

4-7. Vegetation distributions delineated in a classification of 1990 SPOT satellite
im agery ....................................................... 413

4-8. Areas of vegetation change occurring during 1952-1990, and regions where
the swamp hydrologic environment has been affected by the sill. .......... 447

4-9. Dominant successional processes occurring in the Okefenokee Swamp
landscape during 1977-1990. Although all vegetation types were present at
both times, dominant community types at each time changed between years,
as indicated by types in rectangles (1977) and ellipses (1990). Arrows and
numbers indicate changes under various conditions. Greatest increase was
in cypress-bay-blackgum. Mixed blackgum swamp was a minor type in
both years (modified from Hamilton (1982)) ......................... 476

5-I. Number and cause of wildfires reported in the Okefenokee National Wildlife
Refuge area during 1855-1936 ...................................... 497


xxiv









5-2. Total area burned in the Okefenokee National Wildlife Refuge area by
wildfires during 1855-1936 ........................................ 498

5-3. Total area burned by wildfires during 1855-1936, excluding the large fires of
1931-1932 ..................................................... 499

5-4. Total area burned in the Okefenokee National Wildlife Refuge by wildfires
during 1936-1993 ................................................ 500

5-5. Number and causes of wildfires reported in the Okefenokee National Wildlife
Refuge during 1937-1993 ......................................... 501

5-6. Total area burned by wildfires during 1937-1993, excluding those in
1954-1955 ..................................................... 502

5-7. Number and causes of wildfires preceding (1855-1936) and following
(1937-1993) Okefenokee National Wildlife Refuge establishment ......... 503
5-8. Total area burned and causes of wildfires occurring in Okefenokee Swamp by
intervals during 1855-1993 ........................................ 504

5-9. Number and causes of wildfires in Okefenokee Swamp during 1855-1993 ..... 506

5-10. Total area burned by wildfires in Okefenokee Swamp during 1855-1993 ..... 507

5-11. Prescribed burning compartments in the Okefenokee National Wildlife
R efuge ........................................................ 508

5-12. Water levels at Stephen C. Foster State Park (SCFSP) and Suwannee Canal
Recreation Area (SCRA), and total prescribed burning area, during
1973-1993 ..................................................... 509

5-13. Water levels at Stephen C. Foster State Park (SCFSP) and Suwannee Canal
Recreation Area (SCRA), and the number of wildfires reported monthly .... 536

5-14. Water levels at Stephen C. Foster State Park (SCFSP) and Suwannee Canal
Recreation Area (SCRA), and area burned by wildfires ................. 542

5-15. Total area of Okefenokee National Wildlife Refuge burned by prescribed
fires during 1973-1993 ............................................ 547

5-16. Total area of Okefenokee National Wildlife Refuge burned by prescribed
fires and wildfires by intervals during 1855-1993 ...................... 548


xxv









5-17. Locations of wildfires in Okefenokee National Wildlife Refuge, during
1855-1959 and 1960-1993. Extent of each fire is indicated by a black
outline ........................................................ 549

5-18. Hypothesized return frequency, duration, extent, and intensity of drought and
wildfires in Okefenokee Swamp, and the extent and permanency of
subsequent vegetation changes ..................................... 560

6-1. Locations of vegetation transects sampled during 1993-1994 in Okefenokee
Sw am p ........................................................ 569

6-2. Schematic diagram of the placement of understory, overstory, and shrub plots,
and tree belts along a vegetation transect sampled during 1993-1994 in the
Okefenokee Swamp .............................................. 570

6-3. Example interpretation of axes (Graph 1) and curvatures (Graph 2) on
3-dimensional plots of model-predicted abundances of species with
flooding depth and duration ........................................ 583

6-4. Average daily water depths (1962-1995) for species recorded at Okefenokee
Swamp sample sites during 1993-1994 .............................. 586

6-5. Average daily water depths along vegetation transects sampled in Okefenokee
Swamp during 1993-1994 .......................................... 591

6-6. Trends in abundances of all species occurring along an exposure gradient
during 1993-1994 in Okefenokee Swamp ............................ 592

6-7. Hydrologic conditions where species occurred at greatest abundance
(90-100% maximum density or percent cover) ........................ 621

6-8. Distribution of sample points in observed and model-predicted relationships
between species abundance (1993-1994) and inundation depth and duration
(1962-1995) .................................................... 650

7-1. General scheme of seed bank sample collection sites, relative to other
vegetation sample plots, along transects in Okefenokee Swamp ............ 708

7-2. Seed bank emergence experiment sample layout in greenhouse with
continuous swamp water irrigation system ........................... 710


xxvi









7-3. Counts of species and germinated seeds in seed bank samples, compared to
species counts in the standing vegetation within the structural zone at the
collection site. Structural zones are described in Table 6-1. .............. 721

7-4. Vegetation structural zones arranged with increasing duration of deep water
flooding (> 0.30 m; dotted line). Duration of exposed conditions is plotted
(solid line), and variance in average daily water depth for each zone type is
indicated. Species generally associated with each zone type in the seed
bank and established vegetation are listed in Tables 7-4 and 7-6 ........... 756


xxvii














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

ASSESSING PATTERNS AND PROCESSES OF LANDSCAPE CHANGE
IN OKEFENOKEE SWAMP, GEORGIA

By

Cynthia Smith Loftin

December 1998

Chairperson: Dr. Wiley M. Kitchens
Major Department: Wildlife Ecology and Conservation

The Okefenokee Swamp is one of the largest freshwater wetlands in the world.

Currently protected and managed as a national wilderness area and national wildlife

refuge, the swamp has a history of human-caused manipulation and modification. The

swamp landscape is dynamic; vegetation compositions and distributions continually

change as the hydrologic environments change. These dynamics are driven by natural

processes such as peat accumulation and wildfire, as well as the artificial manipulations

of the recent past.

The Suwannee River sill was constructed following extensive wildfires during

1954-1955, with the intent of protecting the swamp and surrounding uplands from effects

of wildfires. During subsequent years, concern was raised that the dam might be

adversely affecting the swamp ecology by extending periods of inundation, increasing


xxviii









water depths, and subsequently affecting swamp vegetation. Delineating the effects of

the Suwannee River sill on the swamp hydrologic environment and vegetation

distributions, in the process of exploring relationships among driving functions and

landscape responses, was a purpose of this dissertation research.

Data collected at various spatial and temporal scales were examined to identify

the sill's effects. A water level recorder network was spatially linked with a global

positioning system survey, and the resultant topographic surface and hydrologic data

were included in a grid-cell based hydrology model to track water movement throughout

the swamp. Model simulations illustrated swamp water level fluctuations before and

after the sill was in place, and predicted recent hydrologic history in the sill's absence, as

well as sensitivities of swamp hydrology to altered evapotranspiration rates. Model

simulations also predicted that the sill was affecting about 18% of the swamp area with

increased inundation depths and durations, and vegetation change attributed to the sill

was limited to this area.

Vegetation dynamics were also assessed at several scales, with remote sensing

techniques, species-hydroperiod descriptions, and seed bank analysis and hydrologic

manipulation. Current vegetation distributions are artifacts of historic logging and recent

lack of fire, and also show sensitivity to local hydrologic environments. Inundation

depth and hydroperiod create hydropattemrns that influence species distributions. The

swamp landscape is an expression of local dynamics, coupled with landscape-level

processes such as fire, drought, and extensive historic logging occurring at multiple

temporal scales.


xxix













CHAPTER 1
THE OKEFENOKEE SWAMP AND THE SUWANNEE RIVER SILL



The Okefenokee Swamp is a 200,000 ha freshwater wetland in Southeast Georgia

and Northeast Florida. The landscape was relatively undisturbed by American

explorers and settlers until the end of the nineteenth century, when it was subjected to

draining, timber harvest, and mining. Protection and preservation of the landscape and

remaining resources were goals in 1937, when the swamp became part of the National

Wildlife Refuge system. The Suwannee River Sill, constructed in 1960 across the main

outflow channel of the Suwannee River where it exited the swamp, was also intended to

protect and preserve the swamp. Built in response to fires that burned across the swamp

and into the surrounding landscape during 1954-1955, the sill was to impound water in

the Okefenokee Swamp to keep similar fires from igniting and burning in the swamp.

During the 30 years following construction of the sill, refuge managers, biologists, and

the public exploring the swamp noticed changes in the composition and distributions of

vegetation communities throughout the swamp. Were the changes indicating that the sill

was affecting swamp vegetation, or was it undergoing natural successional processes?

Concern for the health of the swamp ecosystem began to emerge. During 1989 the

conditions of the sill gate structures were reviewed and found to be unstable, and in need

of repair. Should the sill gates and impoundment berm be repaired, modified, or









2

destroyed? Was the Suwannee River Sill responsible for altering swamp vegetation, or

were the perceived changes artifacts of the observers' temporal and spatial scales?

Rather than "protecting" the swamp, was the sill damaging the wetland by disrupting the

natural hydrologic environment and subsequently the vegetation community dynamics?

Addressing these questions presents an opportunity to examine the Okefenokee

Swamp landscape composition and structure, and identify processes that create and

maintain this structure. Hydrology is a primary driving function of all wetlands, and the

hydrologic regime, principally hydroperiod, determines wetland type (Mitsch and

Gosselink 1986). Many wetlands are also shaped by fire, and fire suppression may

compromise wetland integrity (Mitsch and Gosselink 1986). In many wetland systems

fire and the hydrologic regime are intricately linked; periodic droughts create conditions

favorable for burning. Fires occur, potentially altering site environments (e.g., soil

composition, site elevation, and hydrologic features), and subsequent species

composition. Alterations of frequencies, intensities, and extent of these processes (fire

and the hydrologic regime) can modify landscape composition and structure (DeAngelis

and White 1994). Human activity has disrupted Okefenokee Swamp hydrology and fire

regimes. Hierarchy theory suggests that the extent of these disruptions depends on the

organizational level of the swamp ecosystem that is normally affected by these processes,

and the relative importance of the affected driving function in maintenance of the system

hierarchy.

The purposes of this work are to describe the spatial, hydrologic environment of

the Okefenokee Swamp, to identify changes in vegetation community distributions since











the sill's construction and their probable causes, and to examine the swamp landscape

structure and driving functions in the context of hierarchy and succession theories. To

achieve these goals it was necessary to analyze the swamp vegetation and shaping

functions from several spatial and temporal scales. Hydrologic monitoring and

topographic surveying at locations throughout and surrounding the swamp provided data

for describing the swamp hydrology. Remote sensing and ground truthing provided

landscape-level vegetation distribution information. Transects across topographic

gradients provided relational data among species occurrences, hydrologic features, and

site conditions. Seed bank composition, source, and response to hydrologic regimes

imposed in controlled greenhouse conditions suggested species germination sensitivities

and potential responses to changing hydrologic conditions. Wildfire and prescribed

burning records provided a spatial history of fire to compare with hydrologic and

vegetation distribution information. Pre-logging surveys implied vegetation distributions

resulting from natural successions, and logging records, historic aerial photography, and

recent satellite imagery provided an indication of the extent and duration of logging

impacts. A spatial hydrology model was used to estimate the spatial extent of the sill's

influence on the swamp hydrologic environment. And, a geographical information

system (GIS) was used to identify the spatial relationships of all of these components to

the sill and to current vegetation community distributions and hydrologic features,

elucidating the sill's effects on the swamp ecosystem.

This dissertation is arranged in 8 chapters. The first chapter discusses the

application of theories of hierarchy, scale, and succession to dynamics in the wetland











landscape. A description of the Okefenokee Swamp ecosystem, history of human

influence, and discussion of the history and intended purpose of the Suwannee River Sill

are also contained in the chapter. Chapter 2 describes the acquisition and management

of data included in development of the spatial hydrology model (precipitation,

evapotranspiration, flow, and topography), vegetation distribution and species-

hydroperiod associations (satellite image classification and accuracy assessment, aerial

photography interpretation, and hydroperiod calculations), and precipitation and water

level recorder network accuracy assessments. Swamp hydrologic conditions during the

period, 1941-1995, are summarized in Chapter 2. Development, implementation, and

assessment of a swamp hydrology model (HYDRO-MODEL) are detailed in Chapter 3.

An assessment of the impact of the Suwannee River Sill on the swamp hydrologic

environment based on the HYDRO-MODEL output is made in Chapter 3. Changes in

vegetation community distributions since before logging occurred were detected with

comparisons of pre-logging survey notes, post-logging aerial photography, and satellite

imagery interpretations; these results are summarized in Chapter 4. Chapter 5 details the

wildfire and prescribed burning history; interactions of fire history and swamp hydrology

with vegetation changes are identified. Hydrologic environments associated with swamp

vegetation species during 1962-1995 are described in Chapter 6, and seed bank

composition and response to experimentally altered hydrology are detailed in Chapter 7.

A synthesis of swamp vegetation succession in response to hydrologic alterations,

logging, and wildfire management, and the role of these functions in shaping the swamp

landscape is presented in Chapter 8.









5

This introductory chapter provides the theoretical basis of this wetland landscape

analysis. First, the hierarchical structure of the components and processes of a landscape

and the effect of observer scale on recognizing this organization are addressed.

Thereafter, the processes that structure the wetland landscape (driving functions of fire

and hydrology, and the "disturbances" they create), and the system's responses

(perceived homogeneity and heterogeneity of the landscape, succession, and the

resiliencies inherent in all systems) are discussed, as are techniques for studying

landscape change (GIS, spatial modeling). Finally, a description of the Okefenokee

Swamp environment, and a history of human impacts on the system are presented. The

chapter concludes with a discussion of the Suwannee River Sill history and the primary

questions directing this research.


Ecosystem Hierarchies and Scales



Landscapes are the expression of multitudes of components and processes

interacting at varying temporal and spatial scales. A hierarchical arrangement of these

components and processes is the framework of ecosystem organization (O'Neill et al.

1989a, 1989b). Ecosystem predictability and stability are dependent on preserving the

processes and components occurring at multiple spatial and temporal scales that have

resulted in the expressed system structure (O'Neill et al. 1989, O'Neill 1989, Holling

1987, 1986, Allen 1987, Urban et al. 1987, Allen and Starr 1982). Hierarchy theory

(Allen and Starr 1982) recognizes nesting of system functions and properties with finer











scale. This nesting arrangement provides the framework in which an ecosystem is

structured (Allen and Wyleto 1983). Interactions may occur among and within levels in

the hierarchy, and the outcomes of these interactions may be predictable, until a

disruption in the system's usual processes may lead to development of a system of

different components. Eventually a new hierarchical framework develops, which may

not contain the same components and may result in a different but stable system,

components, and processes; a new stability domain develops, as the system reorganizes

in response to these changes (Holling 1987, 1986, 1973). For example, each individual

plant in a wetland cycles through a period of germination, growth, reproduction, and

senescence, and many individuals are in various parts of this cycle at any time. An

individual may live its entire life under a fairly constant, predictable environment. Some

individuals and species in some years, however, will encounter limiting environments,

such as extreme drought, which may eliminate them from the standing vegetation in the

landscape. In the altered environment other species find the conditions suitable for their

growth and survival. Thus, environmental modification (e.g.,long-term drought) can

result in changes in species composition and changes driven by processes occurring at a

scale and hierarchical level greater than the individual, i.e.,the wetland or the region. A

new hierarchy results, with different species and environmental conditions, and possibly

driven by different controlling functions. The original species may be present in the seed

bank, however, and may again become part of the standing vegetation, given a return to

conditions suitable for germination and maturation. The sustainable system depends on

this type of adaptive cycle, whereby the system develops, becomes stable, undergoes










disruption, reorganizes, and has the potential to return to its original design or reorganize

into another level in the hierarchy (Figure 1-1) (Holling 1987, 1986). A heterogeneous

landscape indicates that a hierarchy of processes is operating at different spatial and

temporal scales (O'Neill 1989). A connectivity among levels of the hierarchy that

responds to variabilities of the systems' processes and components at various spatial

extents is essential to the system's sustainability (Holling 1995, Allen and Wyleto 1983).

Stability perceived at the landscape level is a function of dynamics acting locally

at a small scale, as well as at a greater extent. Thresholds exist whereby changes in the

system components and processes disrupt the function of the system; these changes

might be at any level of the system's hierarchy. The effect on the landscape could be

innocuous, such as elimination of single individuals from a large population of great

extent, or could result in the restructuring of the entire system by removing entire species

or communities. A system's complexity is defined by the boundaries of the multiple

levels and the interacting relationships among them. In general, large structures and low

frequency processes occupy high levels of hierarchy and affect multiple layers of the

system, whereas small structures and high frequency processes are low in the hierarchy

and have limited effects (Ahl and Allen 1996). Predictability in the behavior of the

system and recognition of the underlying, hierarchical model of the system's design and

controlling processes result from observation at multiple levels of organization or scales

(Ahl and Allen 1996).

The study of ecological processes requires selection of appropriate data resolution

and extent. Perception and interpretation of the landscape, its patterns, driving















4 Renewal
* Accessible carbon,
nutrients, and energy







1 Exploitation
r-strategy
Pioneer
Opportunist


Low -- ORGANIZATION


Conservation 2
*K-strategy
* Climax
* Consolidation


I


"Creative"
Destruction
*Fire
* Storm
*Pest
* Senescence


-- High


Weak-lo CONNECTEDNESS Strong








Figure 1-1. The four ecosystem functions and their relationships to the amount of stored
capital and degree of connectedness, from Holling (1986).


I e










processes, composition, and changes are dependent on the scale of observation (Holling

1992, O'Neill et al. 1989a, 1989b, Forman and Godron 1986, Allen and Wyleto 1983)

(Figure 1-2). For example, spatial scale is important to determining perceived effects of

disturbance in a landscape; what appears as disruption at a local scale may actually be

maintenance of the landscape mosaic at a regional scale (Risser 1991). Fire may

maintain the landscape mosaic by changing the distributions of communities in

landscape, but it may be a disturbance if it completely eliminates the potential return of

the species in the landscape. Fine-scale measurement narrows scope and restricts extent;

as data resolution becomes more coarse and extent increases, scope increases and the

range of potential values for a particular landscape variable and its controlling processes

also increase. The observer defines a measurement scale when the objectives are stated; -7

a particular question posed by an observer defines the scale and hierarchy of interest.

Selecting an inappropriate scale may lead to misinterpretation of patterns and driving

processes. Identifying the appropriate data scale for detecting structure in a landscape

and determining the processes that shape that landscape are fundamental to recognizing

the landscape's hierarchical organization. Assessment of the effects of human-induced,

landscape-level perturbations on a complex ecosystem requires integrating information

from multiple disciplines and data resolutions. Analyses of system response to

perturbations must address these interacting components, processes, and scales

(Meentemeyer and Box 1987). Understanding how landscape pattern recognition

correlates with scale facilitates compilation of information across scales (Farmer and























Duration of
Disturbance
or
Landscape
Component
(Log Years)


10,000 years

1000 years

100 years

10 years

Year
Month
Day
Hour


Minute


Space (Log Kilometers)


Figure 1-2. Interactions of temporal and spatial scales and the processes shaping a landscape and its components
(adapted from Holling (1992)).









11

Adams 1991, Musick and Grover 1991). Viable management of an ecosystem ultimately

depends on recognition of and continued interactions of these multiple-resolution

components (Soule 1985).


Driving Functions



Spatial pattern in a landscape is the expression of interactions of current or

historic processes and system components, and may determine the structure and function

of the future landscape (Forman and Godron 1986). These determinant or controlling

processes are the drivingfunctions of the landscape. The dynamics of fire and hydrology

in a wetland are driving functions that result in a shifting mosaic of communities across

the landscape in various stages of succession, with current species composition reflecting

a combination of inter- and intraspecific interactions and the recent driving

environmental influence. Current compositions and distributions of communities in the

landscape offer indications of the driving processes influencing the landscape in the past.

Metastability (equilibrium) of the landscape may increase in the absence of disturbance

(e.g.,fire), allowing the successional sequence to progress and requiring a greater degree

of disturbance over time to disrupt the equilibrium (Forman and Godron 1986) (Figure 1-

3). Distribution of species in the landscape, and knowledge of species' sensitivities to

environmental conditions may elucidate the processes exerting greatest control on the

landscape structure.



















Potential energy w|
Most
in the domain Metastable Least
40* VMetastable

jf|LDisturbance ^
4- Unstable Equilibrium
4 Metastable Equilibrium Most Stable
Stability Domain


Increasing disturbance energy required to achieve
an unstable equilibrium and reach a stability domain
of higher potential energy

Figure 1-3. Interactions of disturbances and system potential energy levels leading to instability and restructuring
(adapted from Forman and Godron (1986)).










Hydrology is a driving force in shaping the function and structure of wetlands.

Hydrologic conditions such as water depth, flood duration and frequency, and water flow

patterns influence both physical and biotic processes. Primary productivity,

decomposition of plant material, nutrient cycling and availability, and vegetation

composition are to some degree controlled by hydrology. Temporal constancy of a

wetland's hydrology may be the dominant factor determining its biotic composition

(Mitsch and Gosselink 1986). Increased flood duration may lower plant species richness

as flood-intolerant species are eliminated, while decreased flooding or more frequent

drawdowns may promote nutrient cycling, decomposition, and primary productivity.

Modifications of a wetland's hydrologic regime can alter the species

composition, distribution, and productivity. Prediction of vegetation changes must

consider relationships between hydrology and ecophysiology of individual species (Leck

et al. 1989). Some plant species respond to flooding with inhibition of seed production

and germination, retarded shoot, cambial, and root growth, arrested reproductive growth,

and death. Wetland plants have mechanisms to acclimate to stresses of inundation, such

as reduced gaseous exchange in a flooded environment. Formation of adventitious roots,

aerenchyma tissue, hypertrophy of stem lenticels, secondary roots, and formation of

knees or pneumatophores may be structural changes occurring in response to flood stress,

to increase exchange of oxygen and waste products (Kozlowski 1984a, 1984b). A plant's

age, duration of flooding, and the nature of the floodwater influence its response. If

flooding persists species are replaced by flood-tolerant wetland species (Kozlowski

1984a, 1984b), usually resulting in a community with lower species diversity.










Elimination of flooded conditions may encourage return of the displaced species.

Deuver (1988) and Duever et al. (1987) demonstrated that hydroperiod is a determinant

of the distribution and composition of freshwater wetland communities in Florida.

Disruptions in a wetland's hydrologic environment could lead to landscape-level

structural changes in the wetland. Hydrology is a high-level controlling process in the

wetland system hierarchy; its disruption could lead to a new hierarchical framework for

the wetland system.

In addition to hydrology, fire is an environmental force that may shape a wetland

over time. In southern stillwater swamps fire plays an important function in sculpting

ecosystem structure (Ewel 1990). Occurrence of these fires is affected by seasonal

cycling of hydrology and periodic droughts. During drier periods fires combust living

and standing dead vegetation, litter, and normally saturated layers of peat. Although

rapid community replacement might occur when mild fires leave many species alive to

resprout, intense fires can eliminate all of the standing vegetation and possibly the peat.

The seed bank in the exposed peat or sediment becomes the initial regenerative source,

supplemented by seeds dispersed from adjacent sources. At this time species that can not

germinate in inundated conditions can become established, provided the burned peat

surface remains exposed. A similar response may occur when normally submerged

substrate is exposed by drought. If new species establishing during dry periods are not

tolerant of inundated conditions, they may be eliminated when water levels rise. With

artificially extended inundation and limited fire disturbance, establishment of

inundation-tolerant species occurs, slowly changing the local vegetation community











composition and structure and ultimately the landscape to a long-hydroperiod system.

Intentional removal of fire while maintaining natural hydrologic processes can also

restructure the landscape, by replacing fire-dependent species with those intolerant of

fire.


Disturbances. Heterogeneity. and Succession



Changes in landscape structure may result from influences of the driving,

functional processes that historically shaped the landscape. Changes may also reflect

recent alterations in those processes that shape levels in the system's hierarchy. Whether

changes in landscape structure and composition are perceived as disruptive to the system

depends on the observer's scale, objectives, the type and intensity of the disturbance, and

the system's evolution. A disturbance creates unsuitable conditions for some component

of the system; its effects may be at various scales, favor some species and eliminate

others, and may be essential in a system's maintenance. A true disturbance is a type of

disruption absent from a system's evolutionary history (Rapport et al. 1985). Periodic

fire and drought are driving processes in a wetland where species have evolved under

their influence. The effects of fire and drought may not be disruptive to the system

overall; the persistence of some species in the landscape may be dependent on occasional

fire or drought to improve conditions for germination, remove competitors, or modify

conditions that promote succession. Removal of these disturbances, which are driving

functions in many wetlands (e.g.,fire or hydrologic cycling, such as drought-flooding











cycles), may be more detrimental to a species that evolved in a system maintained by

periodic disruptions. For example where fire is removed, competitive advantage of fire-

intolerant species may permit displacement of those dependant on fire. Artificially long

hydroperiods and excessive water depths resulting from impoundment eliminate

germination and reduce survival of species not adapted to those conditions. Unnatural

disruptions on an ecosystem may affect biological diversity and processes with which the

system evolved, which could ultimately damage the health of the system (Soule 1985).

Ultimately, an altered stability domain is reached when a system's resilience to these

perturbations is exceeded (Holling 1995, 1987, 1986).

Spatial heterogeneity in communities across a landscape reflects species sorting

in a spatially diverse biotic and abiotic landscape (Milne 1991). Heterogeneity is scale-

dependent; what appears heterogeneous at one scale is homogeneous at another

(Meentemeyer and Box 1987). Landscape homogeneity may express a synergy of

functions, which at a smaller scale appears to create heterogeneity (Meentemeyer and

Box 1987). Although homogeneous landscapes are thought to enhance the spread of

disturbance, heterogeneity may also exacerbate effects of disturbance by increasing

exposure of the landscape interior through percolation (Risser 1991). Change in

heterogeneity with spatial scale may reflect the functional organization of and shaping

processes in the landscape (Musick and Grover 1991). A holistic approach to studying

ecosystem responses recognizes that the effects of disturbance and change may occur

over varying temporal and spatial scales, and differentially influence individual system

components.










Environmental modification and changes in plant community distributions and

composition in the landscape co-occur. The environmental change might be in response

to stochastic events, such as fire or extreme weather, or caused by the landscape's

occupants, such as peat accumulation, chemical soil modification, or shading. A suite of

species will be adapted to the general conditions of the geographic region, i.e.,weather

and geologic history will determine the potential species pool for the area. Which
/
species are present in the standing vegetation will depend on the propagule source,

competitive interactions among species, and environmental limitations of the site. While

an individual occupies a site, it gradually modifies the site's characteristics, so that

conditions become less suitable for itself and more favorable for other species. The site

will undergo changes in species composition as the physical characteristics of the site are

modified, eventually altering community structure and ultimately modifying the

landscape. This change, or succession, in species composition driven by physiographic

and biotic agents was first described in detail by Cowles (1911). Clements (1916)

observed specific associations of species occurring in predictable sequences in

colonization of a landscape; these seral stages terminated in a climax community

specific to the system. He believed the climax community was an expression of the

system, and not driven by changes caused by individual species. Disruption of the

successional sequence by disturbance returned the entire species group, or sere, to an

earlier seral stage, and the sequence of change would repeat. This idea was challenged

by Gleason (1926) who recognized that a succession of species may occupy a site, but

questioned that the sequence and association were predetermined by the system. He









18

believed that the expression of species at a site reflected the available propagule pool and

variations of the environment, and could be altered by the species present, so that

seemingly similar sites could be occupied by different individuals, species, and

associations of species. Change in the composition of the site might occur, but he

believed that it was not necessarily by predetermined associations of species; the

response was of the individual, not the association. Gleason believed disturbance

prohibited a true climax community from developing. Perhaps there is an acceptable

compromise between these approaches. Succession is a phenomenon of the individual

and species, not the community, and results from differential life histories, adaptations

along environmental gradients, and competition among species. Change in an

individual's environment that exceeds its tolerances may lead to occupation by other

individuals and species. These changes operate at multiple spatial and temporal scales

that may be complementary or independent. The selection of species that may occur is

limited by adaptations to the environment; this gives the appearance of an association of

species in a community type, but it is on individuals, not the group, that the environment

exerts control.

Although disturbance might appear to disrupt an ecosystem, it can also be

considered a driver of the succession continuum. Disturbance usually adjusts conditions

to those earlier in the continuum; a cycle of disturbance, occupation, modification,

development, and repeated disturbance develops. A system's response to the

disturbance depends on the severity of the disruption and the degree of system

complexity and development (Holling 1987, Allen and Wyleto 1983). Systems respond











to disturbance with a period of release and reorganization; a longer state of

disequilibrium follows disturbance of later succession communities before

reorganization occurs because more older systems may be less resilient to disturbance

(Holling 1987). However, the succession sequence repeats in response to the

disturbance, unless the system has been unnaturally altered. Disruptions in components

and functions that the altered system experiences may prevent it from developing the

same hierarchical structure; response of the system to disturbances may then be

unpredictable and lead to an alternative stability domain (Holling 1987, Forman and

Godron 1986).


Monitoring Landscape Change



Changes in plant communities as they occur within the landscape can be

monitored using remote sensing and geographical information systems (GIS).

Frequently, remote sensing provides historic data unavailable in any other format.

Remote sensing provides data at various temporal and spatial scales at a cost lower than

required by traditional field censusing techniques, which are used to validate

interpretations of remotely sensed data with information at greater resolution. These data

can be combined with other site features (e.g.,water chemistry, hydrologic parameters,

topography, soil type) in a spatially referenced database. Estimation of missing data with

interpolation may be necessary to provide complete spatial coverage, and data scale must

be comparable among variables. Cartographic modeling techniques can be used with










these data to describe relationships among parameters and changes occurring, describe

how landscape structure influences responses to perturbations, manipulate landscape

features, and predict spatial effects of these manipulations at various scales (Turner and

Gardner 1991). Limitations of the data and GIS techniques must be recognized,

however, so that the influence of data and model scale on interpretations of results is

understood (Haines-Young and Chopping 1996, Meentemyer and Box 1987).


The Okefenokee Swamp Ecosystem



The Okefenokee Swamp is a complex of forested uplands and freshwater

wetlands covering approximately 1670 km2 of lower Atlantic Coastal Plain in Ware,

Clinch, Charlton, and Echols Counties, Georgia, and Baker County, Florida (Figure 1-4).

Approximately 80% of the swamp is within the Okefenokee Swamp National Wildlife

Refuge. The geologic origin of the swamp is debated; the traditional theory is that the

swamp basin began to form during the Yarmouth Interglacial (200,000 years ago) when a

coastal lagoon became separated from the Atlantic Ocean by a sand bar, known today as

Trail Ridge (Carver et al. 1986, Cohen 1973b). During the thousands of years following

this isolation, the seawater evaporated and organism remains and salts were removed by

water and wind. Climatic changes occurring during the last glaciation brought increased

precipitation, which collected in the lagoon basin and provided an environment suitable

for freshwater wetland plants. Peat began to accumulate 6,500 years ago, as decay of

plant remains was delayed by continuous flooding which created anaerobic, acidic

























SSavannah




Tallahassee Jacksonville



Okefenokee Swamp
National Wildlife Refuge







Figure 1-4. Location of the Okefenokee Swamp and Okefenokee Swamp National
Wildlife Refuge.










conditions (Cohen 1973b). Peat accumulation continues today and is punctuated by

periods of extreme drought, when peat is removed by fire and oxidation. An alternative

theory initiates basin formation approximately 12,000 years ago; wind scoured the area,

creating a depression where intercepted rainfall and surface runoff accumulated, and

decreasing hydraulic head and outflow velocity increased retention time of standing

water (Parrish 1971). Wetland plants eventually invaded, and the basin filled with

accumulating peat (Rykiel and Parrish 1979, Rykiel 1977, Parrish 1971).

The swamp's watershed (3702 km2) includes 3 drainage basins (Brook and Hyatt

1985, Hyatt 1984, Rykiel 1977). The Suwannee River carries 85% of the exiting flow

from the western swamp; the St. Marys River (11%) and Cypress and Sweetwater Creeks

(4%) account for the remainder exiting the southern third of the swamp (Rykiel 1977).

Groundwater exchange is minimal (Brook and Hyatt 1985, Hyatt and Brook 1984). The

Suwannee River sill was constructed in 1960 to intercept part of the Suwannee River

discharge from the swamp; the low, earthen dam was intended to impound water in the

swamp to protect it from drought, and to control the initiation and spread of wildfires

within and beyond refuge borders (Chapter 742, Public Law 81-810, 70 Statute 668).

Discharge from the swamp via the Suwannee River and variability of flow into the St.

Marys River decreased during 1960-1986, following construction of the sill, whereas

flow into the St. Marys River increased (Yin and Brook 1992b, Yin 1990). Water enters

the swamp as precipitation (70%) and surface drainage of uplands along the western and

eastern boundaries (Rykiel 1977). Water levels are generally lowest during April-May,

when evapotranspiration demands are high and seasonal precipitation is low, and










October-November due to low precipitation (Chapter 2). Most rainfall occurs during

June-September. During periods of normal hydrology, when peat is continuously

saturated, swamp water depths average 0.7 m (Finn and Rykiel 1979). During the 25

years following construction of the Suwannee River sill, water depths levels during

droughts were estimated to be 11 cm higher than during pre-sill droughts (Yin nd Brook

1992b, Yin 1990, Finn and Rykiel 1979).

Several vegetation communities occur in the Okefenokee Swamp. Prairies are

found where peat layers are thick over depressions in the basement topography (Cohen et

al. 1984, Cohen 1974, 1973a, 1973b) and cover approximately 8% of the swamp.

Vegetation communities include shallow emergent prairies of yellow-eyed grass (Xyris

spp.) and Walter's sedge (Carex walteriana) and deeper rooted or floating aquatic

macrophytes (fragrant water lily, Nymphaea odorata, and golden club, Orontium

aquaticum). Forested areas of pond cypress (Taxodium acsendens), titi (Cyrilla

racemiflora), hurrahbush (Lyonia lucida), loblolly bay (Gordonia lasianthus), and

dahoon holly (Ilex cassine) cover 57% of the swamp. Forested uplands of slash pine

(Pinus elliottii), longleaf pine (P. palustris), saw-palmetto (Serenoa repens), and

gallberry (Ilex glabra) occur on the remaining area of sandy islands and ridges (5%).

Dense shrub thickets of titi, hurrahbush, and fetterbush (Leucothoe racemosa), covered

with a blanket of bamboo greenbriar (Smilax laurifolia) and Walter's greenbriar (S.

walteri) fill the remaining swamp interior (29%). Much of the western portion of the

swamp, where mixed forests of pond cypress, loblolly bay, and blackgum (Nyssa

sylvatica v. biflora) historically predominated, was logged during 1900-1930 (Izlar










1984). This area currently supports stands of shrubs and hardwoods, with little cypress

regeneration (Hamilton 1984, 1982).

The classic model of hydrarch succession (development of a terrestrial forest

climax community from an open water body) directed by autogenic processes (Mitsch

and Gosselink 1986) is only partially applicable to the swamp. The topography

facilitates collection of surface water in the swamp, and periodic droughts expose the

accumulated peat, allowing oxidation and decline in the surface elevation. Site

elevations are raised and hydroperiods altered when accumulated peat is not periodically

exposed and oxidized, creating more favorable conditions for species less tolerant of

flooding. However, in the swamp's history this exchange of species and apparent

"progression" have frequently been disrupted when drought, fire, and subsequent species

changes occur, and the wetland landscape mosaic is maintained (Hopkins 1947).

Palynological studies suggest that overall plant composition has been similar to

the current species composition since peat layers began to accumulate (Cohen et al. 1984,

Rich 1984a, 1984b, 1979, Cohen 1975, 1974, 1973a, 1973b). However, spatial

distribution of these communities has varied, as is indicated in peat deposits (Cohen et

al.l984, Rich 1984a, 1984b, 1979, Cohen 1975, 1974, 1973a, 1973b). Hydrologic

variations and fire interact to direct succession in the swamp (Roelle and Hamilton 1990,

Hamilton 1984, 1982, Deuver and Riopelle 1984, Duever 1982, 1979, Rykiel 1977).

Many species occurring in the swamp are adapted to nutrient-poor, saturated conditions.

Okefenokee Swamp surface waters contain <1% of the system's nutrients; 59-98% of the

Ca, Mg, Na, and K are found in the system's standing vegetation, and the remainder is










encumbered in slowly decomposing peat (Rykiel 1977). Exposure of the peat surface

during drought hastens peat decomposition and bacterial cycling (Murray and Hodson

1985), making nutrients more available for use (Schoenberg and Oliver 1988, Bosserman

1983a, 1983b, Flebbe 1983), as do fires which may accompany extended drought. Most

of the swamp has developed from open prairie to shrub bog to cypress or bay forest

during some period of the past 6000 years, with undisturbed intervals varying from

decades to hundreds of years. Drought, peat accumulation, and battery formation reduce

the apparent water level, which permits succession of flood-intolerant woody species to

occur. A progression from prairie to cypress swamp to broadleaved evergreen or mixed

cypress swamp occurs in the absence of disturbance as peat accumulates (Hamilton 1984,

1982) (Figure 1-5).

As indicated by layers of charcoal in peat deposits, fire has checked succession in

the swamp since peat began to accumulate thousands of years ago (Cohen et al. 1984,

Cohen 1975, 1974, 1973a). Fire retards the progression of prairie to wooded swamp or

returns the vegetation to and earlier stage. Certain vegetation communities such as

cypress are frequently associated with concentrations of charcoal in the peat, suggesting

a susceptibility to fire, especially during droughts (Cohen et al. 1984, Cohen 1975, 1974,

1973a). The central, deep-peat prairies have never been completely succeeded to cypress

forest, possibly because they are topographic lows that have maintained conditions too

saturated for forest species, or severe fire has burned the area frequently enough to retard

expansion of woody species. Fires which bum the surface peat remove fire-intolerant

plants but usually do not kill shrubs and large trees rooted in deep peat or sand beneath









Open Water

Aquatic Prairie
+
Herbaceous Prairie

Shrub Prairie

ess-Bay Shrub
Shrub-Bay


Cypress-Bay Forest


Gum-


Bay-Forest


)-Gum-Cypress


Cypress Forest


Bay-Forest


Figure 1-5. Hypothetical community changes occurring with peat accumulation in the absence of disturbance in Okefenokee
Swamp (adapted from Hamilton (1982)).


Shrub-Cypnr









27
shallow peat (Cypert 1973, 1961). More severe fires that burn to the deep, sub-peat sand

layer are disjunct and rarely occur. Deep lakes occurring in the eastern swamp may have

resulted from hot fires that burned through the accumulated peat and into the underlying

sand. Prairies result when severe fires remove peat and woody root systems, preventing

reestablishment of existing woody vegetation (Cypert 1973, 1961) due to lowered

topographic surface and increased inundation depth. The light, surface fires which

historically occurred frequently are probably more important than the infrequent,

widespread, severe fires in maintaining the mosaic of existing vegetation associations

(Roelle and Hamilton 1990). The manipulations of the swamp vegetation composition

and hydrology during the past two centuries and current fire management have affected

fire frequency and occurrence across the swamp (see Human Modification of

Okefenokee Swamp section).

The Okefenokee Swamp landscape structure is affected by vegetation community

succession. Swamp vegetation is determined by the hydrologic environment; disturbance

history; species pool; propagule distribution and establishment requirements; potential

longevity of propagules, juveniles, and adults; and, interactions of these features.

Although species composition and abundance vary from site to site, there is a limited

number of species occurring in the swamp, each with a certain range of life history

requirements and environmental tolerances. Thus only a certain suite of species are

likely to occur, and their presence in the landscape is mediated by inter- and intraspecific

interactions, as well as other environmental processes. The swamp is maintained as a

metastable equilibrium, where species are fluctuating between a competitive equilibrium









28

(maintaining the appearance of stability) and disequilibrium (species replacement), with

intervening periods of changing communities in response to disturbance of greater

intensity on a larger spatial scale. In the swamp's pre-modem (pre-1890) history this

disturbance has been periods of drought and fire, creating a moving mosaic of vegetation

communities in different stages of development in the landscape (Hamilton 1984, 1982,

Rykiel 1977). Hydrologic alterations, logging, and changes in the burning regime are

perturbations with the potential to disrupt development of this mosaic and affect future

swamp structure.



Human Modification of Okefenokee Swamp



Humans have inhabited the Okefenokee Swamp for at least the past 4,000 years,

and have lived in the Okefenokee Swamp area for 10-12,000 years (Trowell 1984a,

1984b), and have variously modified the swamp, particularly during the 20th century

(Trowell 1994, 1989a, 1989b, 1988a, 1988b, 1987, 1984) (Table 1-1). The region's

name is derived from a Seminole-Creek Nation word, Oke-fin-o-cau, meaning "land of

the trembling earth" (McQueen and Mizell 1926), in reference to the floating islands

found throughout the swamp. The swamp was surveyed by Mansfield Torrance in 1850

and many others in the following years, and was purchased from the State of Georgia by

the Suwannee Canal Company in 1890 (Trowell 1994, 1989a, 1989b, 1988a, 1988b,

1984a, 1984b). The Suwannee Canal was excavated during 1890-1897 to drain the

swamp to create an agricultural district; the effort failed, and in 1904 the land was











Table 1-1. Human-caused manipulations of Okefenokee Swamp vegetation and
topography occurring during the past 150 years.

Type of When Manipulation Probable Scale
Manipulation Occurred of Effect
Prescribed Burning, Arson pre-settlement to present local to swamp-wide
Dredging, Peat Mining late 19th century to mid- local peat removal,
20th century regional hydrologic effect
Logging late 19th century to mid- locally intensive,
20th century regionally scattered
Impoundment 1960-1962 construction, local to regional effects
(Suwannee River sill) 1962 fully operational depending on seasonal
water levels
Boat Trail Cutting and 20th century local to regional effects on
Maintenance submerged, emergent, and
nearby terrestrial
______________ __ vegetation


purchased by the Hebard Lumber Company. Marketable cypress, pine, and hardwoods

were removed from the swamp and processed at local sawmills for shipment throughout

the Southeast. Logging operations ceased in 1937 when the property was purchased by

the United States government and added to the National Wildlife Refuge system. Peat

mining in the Northeast swamp ceased in the 1950s, and the refuge was designated a

national wilderness area in 1974 (Trowell 1989c, Fortson 1961).

During 1954-1955 the region experienced a severe drought and nearly 80% of the

swamp was burned by wildfires (Hamilton 1984, 1982). Many of these fires began in the

surrounding uplands, spread into the swamp where the peat slowly burned, and returned

to the perimeter uplands. Neighboring landowners sustained significant property loss









30

from these fires. There was great interest in protecting the swamp and surrounding lands

from future fires; a law (Appendix A) was enacted by the United States Congress in 1956

to require construction of a dam, the Suwannee River sill,

to protect the natural features and the very substantial
public values represented in the Okefenokee National
Wildlife Refuge, Georgia, from disastrous fires..., and for
the purpose of safeguarding the forest resources on more
than four hundred thousand acres of adjoining lands
recently damaged by wildfires originating in or sustained
by the desiccated peat deposits in the Okefenokee Swamp.
(Chapter 742, Public Law 81-810, 70 Statute 668, pages
781-782).

A perimeter road that would permit access to remote areas for fire control and serve as a

fire break to spreading fires was also required by the law. In 1962 construction of the sill

berm and closure of the 2 spillway gates were completed. The berm spans 7.2 km across

the exiting flow of the Suwannee River and averages 35.5 m above mean sea level and 3-

4 m above the surrounding Suwannee River floodplain; a ditch borders its entire length

to the east. The original south gate collapsed in 1979 and was replaced; the north gate is

the original structure. Although the gates are maneuverable, they remain closed to

maximize the impoundment.

Apparent changes in vegetation composition of the Okefenokee Swamp during

1960-1990 precipitated concern that the Suwannee River sill and the Okefenokee

National Wildlife Refuge fire management policy were permanently altering the

swamp's ecology (Roelle and Hamilton 1990). Severe drought and fire have occurred in

the Okefenokee Swamp at approximately 20-year intervals during the past 150 years

(Cypert 1973, 1961). The Suwannee River sill was constructed to prevent recurrence of










fires during these droughts. During 1962-1990 extensive fires did not occur in the

swamp. This may have been the result of the Refuge's fire management policy rather

than the impoundment effects of the sill. Yin and Brook (1992b) and Yin (1990) found

that the amount of water retained by the sill during severe drought (11 cm) was not

enough to counteract an extreme drawdown (1-1.5 m during 1954-1955) due to drought.

In fact, scattered fires during 1990 and 1993 suggest that the sill had not eliminated fire

in that region. Thus, the sill was performing as it was intended (i.e.,to suppress fires)

only in its localized area during periods of average hydrologic conditions, temporally and

spatially extending hydroperiod beyond the local area during intervening years when

water levels were generally higher (Roelle and Hamilton 1990), and not retaining a

substantial amount of water during extended periods of below average rainfall.

Extending flooding by impounding runoff and stream flow may reduce water

level variation that normally occurs with precipitation (Finn and Rykiel 1979). Finn and

Rykiel (1979) compared pre- and with-sill water levels measured at the Camp Cornelia

boat basin 29 km east of the sill, and reported an increase (10-13 cm) in average monthly

water level after sill construction. Yin and Brook (1992b) and Yin (1990) measured an

increase in average storage and a decrease in discharge. The higher water level behind

the sill decreases the gradient approaching the sill, reducing flow and pooling the water

(Finn and Rykiel 1979), especially during periods of above average rainfall. If the sill is

extending periods of high water, it may be altering the landscape by affecting vegetation

succession. Decreased fire frequency and extent may be encouraging woody vegetation

to invade prairies during the occasional drier periods, hastening succession to cypress or








32

bay swamp, and eliminating the mosaic of vegetation and the associated biodiversity in a

landscape historically perpetuated by periodic disturbances (Hamilton 1984, 1982).

When this study was initiated in late 1991, the Suwannee River sill had

deteriorated since its construction and was in need of repair. The uncertainty of the sill's

effects on the hydrology and vegetation of the swamp raised questions of whether the sill

should be opened, repaired as a fixed height weir, or replaced with a controllable

structure. Effects of the sill on vegetation communities within the landscape needed to

be documented and predicted effects of future hydrologic management alternatives

analyzed so that the refuge hydrology could be effectively managed. This dissertation

research identifies the spatial extent of the Suwannee River sill's modification of swamp

hydrology, and spatial changes in vegetation composition since the sill was constructed.

Probable causes of the vegetation changes are proposed, and several hydrology

management options and their effects on swamp vegetation composition are investigated.

The following guiding questions are addressed in these dissertation chapters:

1) Have vegetation community distributions changed since

the Suwannee River sill was constructed? If so, where have

these changes occurred? Have fire frequency and distribution

changed during this period?

2) Are swamp vegetation community composition and distribution

correlated with hydroperiod and water depth?









33

3) What are the potential responses of the Okefenokee Swamp

vegetation communities and landscape to future sill modification

and hydrologic manipulation?














CHAPTER 2
DATA BASE ORIGIN AND DEVELOPMENT





Data Sources and Extent




Determining the effects of the Suwannee River Sill on the hydrology and

vegetation of Okefenokee Swamp required diverse point and spatial data. The origin,

management, and quality assessments of data used in the swamp hydrology model (see

Chapter 3) and vegetation change analysis (see Chapter 4) are detailed in this chapter.

Precipitation, evapotranspiration, surface water inflow and outflow, water surface

elevation, and water depth data are components of the swamp hydrology model. The

hydrology model describes the swamp surface water environment during 1941-1993, the

duration of the complete, concurrent weather and water level data. Summaries for

Suwannee Canal Recreation Area (SCRA) and Stephen C. Foster State Park (SCFSP)

include the complete period of record for these stations, 1941-1995. Historic, daily data

were available for 1930-1991 from gauges monitoring some of the model parameters;

additional data were collected during 1991-1995 from gauges installed in 1991-1992 to

supplement the recorder network. Descriptive statistics for precipitation,









35
evapotranspiration, and surface water flow are calculated from the 1930-1995 data. Due

to recorder discontinuity, malfunction, or removal, the daily records for these parameters

during 1930-1995 were incomplete. Regression equations between correlated recorders

estimated missing data to provide a more complete data record for the hydrology model.

Descriptions of hydrology dataset management and the swamp hydrologic environment

are discussed in the Swamp Water Level Data, Estimation of Missing Water Level Data,

General Swamp Water Level Conditions, and the Suwannee River Sill's Effects on River

Flow and SCRA and SCFSP Water Level Conditions sections of this chapter.

Swamp water level variation was monitored with daily data from a water level

recorder network. Network design redundancies and discrepancies affect the accuracy of

the water level estimates. Identifying the best design, to improve efficiency and accuracy

of the recorder network for the intended purpose, is discussed in the Water Level

Recorder Performance section of this chapter.

Most of the water in the swamp enters as direct precipitation (Yin and Brook

1992b, Yin 1990, Hyatt 1984, Blood 1981, Rykiel 1977). Estimations of spatial

contributions of rainfall to the swamp water budget rely on data gathered at precipitation

recorder stations. Accuracy of the recording network should be quantified so that the

accuracy and limitations of the precipitation estimates can be identified. A rainfall

variation and recorder network design analysis are discussed in the Precipitation Gauge

Network Analysis section of this chapter.

Swamp topographic surface elevation is a component of the swamp hydrology

model. Although 1994, 1:24,000 USGS topographic maps exist for the swamp region,










the data scale is insufficient for directing water movement across the slight gradient of

the swamp at the hydrology model scale (500 x 500 m cell size). A more resolute

topographic map was developed with a Global Positioning Systems (GPS) survey; this

survey also permitted referencing the network of recorders to a common reference

(elevation above mean sea level) and identifying their true location in the landscape

within centimeters. The data collection and interpolation procedures used to create the

swamp topographic surface map, and the development of swamp peat and sand surface

profiles, are discussed in the Topography Map Development Section of this chapter.

A base map of current vegetation was needed to identify changes in vegetation

community composition and distribution occurring in the landscape since the sill was

constructed. SPOT multispectral and panchromatic satellite imagery were used to

produce this base map; changes in swamp vegetation community distributions were

identified by comparing maps created from interpretations of aerial photographs of the

swamp taken in 1952 (7 years pre-sill) to those from 1977 (15 years with-sill) and the

1990 (28 years with-sill) base map. The procedures and accuracy of the satellite imagery

classification are detailed in the Satellite Imagery Classification and Accuracy

Assessment section of this chapter. Details of the photointerpretation and change

assessment procedures are included in Chapter 4.

Swamp Water Level Data


Swamp water level data were compiled from several sources. The longest

duration records were from staff gauges installed in Billy's Lake (at SCFSP) and the










SCRA boat basin during 1941; readings were made several times monthly at both

stations until 1950, when daily readings were begun at SCRA. Daily readings were not

made at SCFSP until 1968. Steven's chart recorders with float gauges were installed at

SCRA in 1979 and SCFSP in 1980. Chart recorders were also installed at 11 other sites

in 1979-1980 (Figure 2-1). Elevations of these recorders were referenced to staff gauges

installed on site, and the reference elevation was transit-surveyed to perimeter USGS

benchmarks during the early 1980s. During 1980-1991, 3 recorders were removed from

the network, and those that remained were not regularly maintained, resulting in an

incomplete record of daily water surface elevation. In 1992 the gauging network was

examined, broken gauges were repaired, and an additional 13 gauges (Omnidata, Inc.

digital recorders with Delta pressure transducers and WaterMark reference staff gauges)

were installed throughout the swamp (Figure 2-1). Elevation of the reference staff at

each recording station was related to a permanently established benchmark located

within 500 m of the recorder. Location and elevation of these benchmarks were surveyed

(see discussion of topography map development in this chapter) relative to Universal

Transverse Mercator (UTM) zone 17 grid X and Y location and NAVD27 elevation

projection of mean sea level, so that water surface measurements could be compared

spatially. The digital gauges recorded water elevation once daily (hourly readings

averaged every 24 hours); data were retrieved from the recorders every 4-6 weeks during

1992-1995. Daily water surface elevation recorded on the accumulated historic charts

(1980-1991) and those retrieved quarterly from chart recorders during 1992-1995 were

digitized and corrected to the reference benchmark elevation. Records from each station











N sill
SRdUge Boundary
V Islands
1Islands

^ \ Suwam ee Canal

SMain Channel, Suwannee River
___ Sill


7 0 7 14


Figure 2-1. Water level and precipitation recorder locations in the Okefenokee Swamp during 1941-1995.










were compiled into spreadsheets; intervals of missing data were identified and

correlations among recorders examined to identify regression equations to use in missing

data estimation (see Recorder Correlations and Missing Data Estimation).

Water Level Recorder Performance

There were 26 gauges recording water elevation continuously or daily during

1979-1995 for varying lengths of time. Elevation of reference staffs above mean sea

level, corrections to historic reference staff data, period of record, and days in operation

are indicated in Table 2-1. The interpreted data from each recorder and staff during the

periods of operation are illustrated in Figure 2-2. Estimated missing data are included in

the plots to approximate a complete record for 1941-1995. Water level data were

estimated for all recorder stations from SCRA and SCFSP staff data during 1941-1979.

Estimates were calculated for 1-80% of the station water level data for 1980-1995;

descriptive statistics of each station's record are listed in Table 2-2.

During 1992-1995 when the water level recorder network density was highest, 23

gauges were working for 20-99% of the interval (Table 2-3). In most cases recorder

malfunction could be attributed to mechanical failure due to interference by wildlife or

refuge visitors, or due to insufficient maintenance of recording equipment. During 1992-

1995 digital recorders were most reliable, although one chart recorder operated for 91%

of the interval. The poorer performance of chart recorders can be attributed to their age.

Most of the stations had been deployed since early 1980. Solar rechargeable batteries

caused problems with 3 digital units, and insufficient charges on non-rechargeable

batteries were responsible for missing data on other units. Over the 694-5672 days of












Table 2-1. Water level recorder elevations and staff corrections, operating period, and
precipitation gauge locations.


Ground
Elevation
Staff Above
Station Type" Start Date End Date Correction Mean Sea
(m) Level at
_________ _______ ________ _______ _______ Staff (m)


Sill (Brown
Trail)

Chase Prairie

Territory Prairie

SCFSP

SCRA

Double Lakes

Gannett Lake

Seagrove Lake

Moonshine
Ridge

Suwannee Creek

Soldier's Camp

Sapp Prairie

Kingfisher
Landing

Coffee Bay

Billy's Lake

Suwannee River

Sweetwater
Creek

Cypress Creek

Floyd's Prairie

Suwannee Creek


chart, wl, p


chart, wl, p

chart, wl, p

chart, wl, p

chart, wl, p

chart, wl, p

chart, wl, p

chart, wl, p

chart, wl, p


chart, wl, p

chart, wl, p

chart, wl, p

chart, wl, p


digital, wl, p

digital, wl

digital, wl

digital, wl


digital, wl

digital, wl, p

digital, wl


2-1-1980


5-6-1980

5-6-1980

2-1-1980

9-1-1979

5-21-1980

6-4-1980

12-5-1979

3-5-1982


5-22-1980

4-11-1980

4-18-1980

1-11-1980


4-1-1992

4-16-1992

5-7-1992

4-2-1992


4-3-1992

4-2-1992

4-17-1992


6-7-1995


4-10-1995

3-13-1995

2-22-1995

7-3-1995

2-22-1995

1-19-1995

2-16-1995

5-16-1994


10-17-1982

3-5-1982

3-9-1988

6-14-1995


6-14-1995

6-13-1995

5-31-1995

6-1-1995


6-1-1995

6-1-1995

7-20-1995


+34.75


+32.24

+30.32

-0.009

+36.14

+36.18

-0.63

+31.11

-1.13


+35.97

+33.83

+35.97

+37.38


+35.77

+33.58

+33.77

+33.84


+32.94

+34.86

+36.66


34.03


36.21

36.66

34.12

36.08

37.35

36.53

36.50

35.62


36.02

34.0

35.20

36.54


36.69

33.6

33.62

34.04


33.95

35.27

36.02









Table 2-1--continued.


Ground
Elevation
Staff Above
Station Type Start Date End Date Correction Mean Sea
(m) Level at
____________Staff (m)
Sapling Prairie digital, wl, p 2-4-1993 2-22-1995 +36.35 36.61
Durdin Prairie digital, wl, p 4-1-1992 6-14-1995 +36.68 36.98
Honey Prairie digital, wl, p 6-16-1992 12-15-1994 +35.84 36.11
Chesser Prairie digital, wl 2-4-1993 6-14-1995 +35.94 35.68
Sapp Prairie digital, wl, p 2-4-1993 12-15-1994 +35.70 35.20

Craven's digital, wl,p 4-1-1992 5-31-1995 +34.95 35.15
Hammock
SCFSP staff, wl 1-4-1941 2-22-1995 -0.009 34.12
SCRA staff, wl 1-4-1941 7-3-1995 +36.14 36.08

SData are recorded daily by automated (chart, digital) systems or refuge personnel (staff), and
stations monitor daily water surface elevations (wl), precipitation (p), or both.









42
37 -
E Billy's Lake
-j 36-
C0
< 35-

0
> 34-
0)
w
33 -- |--|

1941 1944 1947 1950 1953 1956 1959
37 -
E
-. 36-
C/)

35-
C
*._
0
> 34 -
a)
LUJ
33 ------
1960 1963 1966 1969 1972 1975 1978
37 -
E
(/,
-. 36-

< 35-
C
0
m
CU
> 34-

33------
1980 1983 1986 1989 1992 1995

Year



Figure 2-2. Daily water surface elevation above mean sea level (AMSL) during 1941-May
1995 recorded at locations in Okefenokee National Wildlife Refuge, GA.









43
37.0
"E Chase Prairie

-j
CO 36.5-


0
36.0-
()
w
uJ
35.5 ------
1941 1944 1947 1950 1953 1956 1959
37.0 -
E
-J
CO 36.5-

C
0
I 36.0-

LU
w
35.5 i ---
1960 1963 1966 1969 1972 1975 1978
37.0 -





S36.0-
0
S36.0-
(U

35.5 -- |--
1980 1983 1986 1989 1992 1995

Year


Figure 2-2-continued.










37.5

E- 37.0
-J
C0
2 36.5
c
o 36.0
0'%
a 35.5
35.0
35.0


1941 1944 1947 1950 1953 1956 1959


37.5

37.0

36.5

36.0

35.5

35.0


37.5

37.0

36.5

36.0

35.5

35.0


1960 1963 1966 1969 1972 1975 1978


1980 1983 1986 1989 1992 1995

Year


Figure 2-2-continued.


E
-1

(0
,0
4i






E
CO
0
<
4-'
LU






-J
Co


.0

(U










38-
E

C37 -


0
36-
'p
a)
35 -I-
1941
38--
E
-j
i 37 -


0
= 36-
a)

35
1960
38--


I 37 -


0
= 36-
m

w
35 i
1980


I I I I I I
1944 1947 1950 1953 1956 1959


I I I I I 1
1963 1966 1969 1972 1975 1978


I I I 1 1
1983 1986 1989 1992 1995


Year


Figurem 2-2-continued,


Coffee Bay











E
-J
,_i
Co


0
(D





E
JQ)






.J

0
CU

m
0



-J




E
CU
.I;




0)
w
'?


37


36


35.


34-



37


36 -


35-


34-



37-


36-


35-


34-


1960 1963 1966 1969 1972 1975 1978
1


I I IIII
1980 1983 1986 1989 1992 1995

Year


Figure 2-2--continued,


1941 1944 1947 1950 1953 1956 1959












E 35-
-J
CO
34-

0
.,

C 33-
(D
0
uJ
32 -



E 35-
-J
Cl)
S34-

0
C 33-

w
JC

32 -



E 35
-J
Co
|34-

0
I 33-


32 -


I I 1 1 I I I
1960 1963 1966 1969 1972 1975 1978


I I I I 1 1
1980 1983 1986 1989 1992 1995

Year


Figure 2-2-continued.


Cypress Creek


-I I I I I 1I1
1941 1944 1947 1950 1953 1956 1959










38.0-

37.8 -

37.6 -

37.4-

37.2-

37.0-

38.0 -

37.8 -

37.6 -

37.4-

37.2-

37.0 -

38.0 -

37.8-

37.6-

37.4-

37.2-

37.0 -


1941 1944 1947 1950 1953 1956 1959


I I I I I I I1
1960 1963 1966 1969 1972 1975 1978


I I I I 1 1
1980 1983 1986 1989 1992 1995

Year


Fimgure 2-2--continued.









37.6


37.4 -


37.2-


37.0 -

37.6 -


37.4 -


37.2-


37.0 -

37.6


37.4-


37.2-


37.0


1 I I I I I I I _
1941 1944 1947 1950 1953 1956 1959


I I I I I 1 1
1960 1963 1966 1969 1972 1975 1978


1980 1983 1986 1989 1992 1995
1980 1983 1986 1989 1992 1995


Year


Figure 2-2-contmnued.


I Durdin Prairie










37.0 -

E 36.5- Floyd's Prairie
-i
CO
2 36.0 i |

o 35.5-

: 35.0
w
34.5------i
1941 1944 1947 1950 1953 1956 1959
37.0 -

E 36.5-
-i
co
2 36.0-

o 35.5-

o 35.0
LLJ
34.5-I--Ii-
1960 1963 1966 1969 1972 1975 1978
37.0 -

E 36.5-
-J
2 36.0-

o 35.5-

o 35.0-

34.5 f i ------i
1980 1983 1986 1989 1992 1995

Year


Fiure 2-2-continued.










37.5-


37.0 -


36.5 -


36.0-

37.5 -


37.0 -


I I I I I I I
1960 1963 1966 1969 1972 1975 1978


1980 1983 1986 1989 1992 1995


Year


Figure 2-2--continued.


1941 1944 1947 1950 1953 1956 1959


36.5 -H


36.0

37.5


37.0


36.5


36.0










I


E
-J
(I)
CD
<
0.


oU





E
-J
co
<=

0
-W


w






-j
-o
O
_e







0
LU
-W
w
uj


38.0 -


37.5 -


37.0 -


36.5 -


36.0 -


38.0 -


37.5-


37.0-


36.5-


36.0-


38.0 -


37.5-


37.0-


36.5-


36.0-


1960 1963 1966 1969 1972 1975 1978


I I


I I -------I---I
1980 1983 1986 1989 1992 1995

Year


Figure 2-2-continued


Honey Prairie


I 1 4I I I I 1 5 I5
1941 1944 1947 1950 1953 1956 1959










38.0-
13 Kingfisher Landing

-J
CO
< 37.5-
0
(D
W 37.0

1941 1944 1947 1950 1953 1956 1959
38.0-
E
-J
CO
< 37.5-
0


W 37.0

1960 1963 1966 1969 1972 1975 1978
38.0 -


-J
CO
< 37.5-
0

4)
W 37.0

1980 1983 1986 1989 1992 1995

Year


Figure 2-2-continued.











E
- 36.0-



0
c '

> 35.5-
4)
uLJ

1941 1944 1947 1950

E
...j 36.0-
C,)



> 35.5-
uJ
U)
w

1960 1963 1966 1969

E
j 36.0-



0
> 35.5-
u)


1980 1983 1986 1

Year


II 1
1953 1956 1959


I I I
1972 1975 1978


I I I
989 1992 1995


Figure 2-2-continued.


Moonshine Ridge











E
-J

C/

0


LU



w

I-,
E

-o

CO
CU

0)





Col
_i






C
0
LU
m





w


38.0-


37.5 -


37.0 -


36.5 -


36.0-


38.0 -


37.5 -


37.0 -


36.5 -


36.0 -


38.0 -


37.5-


37.0-


36.5-


36.0 -


1980 1983 1986 1989 1992 1995

Year


Figure 2-2--=continued.


1941 1944 1947 1950 1953 1956 1959


I I I 1 I I I 1
1960 1963 1966 1969 1972 1975 1978











37.0 -


36.5 -


36.0-


35.5-



37.0 -


36.5 -


36.0-


35.5 -



37.0-


36.5 -


36.0-


35.5-


1960
1960


I I I I 1 1
1963 1966 1969 1972 1975 1978


I I I


I I I I 1 1
1980 1983 1986 1989 1992 1995

Year


Figure 2-2--continue,


1941 1944 1947 1950 1953 1956 1959










37 -
EA
.. 36-
0o
35-l
0c
> 34-

LUJ
w
334-------
1941 1944 1947 1950
37-
E
..j 36-
co

< 35-
c
0
> 34-
Cu

33 -
1960 1963 1966 1969
37-


S...j36-


< 35-
c
0
S34-
CJ
34)
w
33---
1980 1983 1986 1

Year


II I 1
1953 1956 1959


1972 1975 1978
1972 1975 1978


I I I
989 1992 1995


Figure 2-2-continued,


SCFSP










37.5-

E SCRA
.j 37.0-
C,)

< 36.5-
c
0
.U
> 36.0-


35.5 -1i -- -
1941 1944 1947 1950 1953 1956 1959
37.5 -
E
-j 37.0-
U)

S36.5-
c
0
> 36.0-
()
LLu
35.5 ------
1960 1963 1966 1969 1972 1975 1978
37.5
E
.-j 37.0-

< 36.5-
c
0
-.9
> 36.0-

35.5-
1980 1983 1986 1989 1992 1995

Year


Figure 2-2-continued.









37.5 -
E Seagrove Lake
-j 37.0-
C,,


j) i
< 36.5-
c
0
CU
> 36.0-
()
J 35.5 -

1941 1944 1947 1950 1953 1956 1959
37.5 -
E
..j 37.0-

< 36.5-
a
0
> 36.0-
a)
w
35.5- 1-----------

1960 1963 1966 1969 1972 1975 1978
37.5-

..j 37.0-


C,)
< 36.5-
0

CU
> 36.0-

35.5 I-
1980 1983 1986 1989 1992 1995

Year


Figure 2-2--continued.








60
37 -
E Sill
.j36-

< 35-
c
0
> 34 v '
>4)
w
33-
33,, -- ------] ------ | --------

1941 1944 1947 1950 1953 1956 1959
37 -
E
S36-

< 35
a
0
< 35- f^


> 34-
0
LJU
33------
1960 1963 1966 1969 1972 1975 1978
37 -
E
.j 36-
CO
< 35
0
> 34
w
3i 3 3

1980 1983 1986 1989 1992 1995

Year


Figure 2-2--continued.









61

E 35.0 Soldier's Camp

-J
Co
S34.5

o 34.0


[w 33.5


1941 1944 1947 1950 1953 1956 1959

E 35.0-
-o
S34.5-

0 34.0-

a)
M

U- 33.5-


1960 1963 1966 1969 1972 1975 1978


E 35.0-
.-i


C')*
(34.5-

o 34.0-


w 33.5


1980 1983 1986 1989 1992 1995

Year


Figure 2-2--continued.










38.5 -
38.0 -
37.5 -
37.0 -
36.5 -
36.0 -
35.5 -


38.5 -
38.0-
37.5-
37.0 -
36.5-
36.0-
35.5-


38.5-
38.0-
37.5-
37.0-
36.5-
36.0-
35.5-


1 I I I I I I I
1941 1944 1947 1950 1953 1956 1959


I I I I 1 I I 1
1960 1963 1966 1969 1972 1975 1978


I I I I I I
1980 1983 1986 1989 1992 1995


Year


Figure 2-2-continued.


Suwannee Creek










38 -
37-
36-
35-
34-
33-
32-
31 -

38 -
37-
36-
35-
34-
33-
32-
31 -

38 -
37-
36-
35-
34-
33-
32-
31 -


1941 1944 1947 1950 1953 1956 1959
1941 1944 1947 1950 1953 1956 1959


II I I I I 1
1960 1963 1966 1969 1972 1975 1978


I I I I I 1
1980 1983 1986 1989 1992 1995


Year


Figure 2-2-continued,


Suwannee River












E 35.0-
-J
C)

C
I34.5-


O 34.0-
W

33.5 -



E 35.0-
-i
Co
34.5-

0
S34.0-
uJ
0
33.5 -



S35.0-
-J



0
34.5 -

,o
( 34.0-

33.5 -
33.5-


1I I I I I I 1
1941 1944 1947 1950 1953 1956 1959


1960 1963
1960 1963


1966
1966


1969 1972 1975 1978
1969 1972 1975 1978


1980 1983 1986 1989 1992 1995

Year


Figure 2-2-continued


Sweetwater Creek









65
37.5-
E Territory Prairie
-J
U) 37.0-


0
= 36.5-


36.0 ------
1941 1944 1947 1950 1953 1956 1959
37.5-
E
-j
U) 37.0-


0
S36.5-
'I)

36.0 i- ----
1960 1963 1966 1969 1972 1975 1978
37.5 -
E
-j
C.) 37.0-


0
= 36.5-

ILl
36.0 ----
1980 1983 1986 1989 1992 1995

Year


Figure 2-2-continued.












Table 2-2. Summary parameters of water level recorders installed at Okefenokee
National Wildlife Refuge during 12-5-1979 through 6-15-1995. Elevations are in
meters above mean sea level. Basin delineation is discussed in the Swamp Basin
Delineation and Characterization section.


Mean Daily Variance in Minimum Maximum
Basin and Water Surface Daily Water Daily Water Daily Water
Station' Elevation Surface Surface Surface
Elevation Elevation Elevation


Northwest
Basin

Suwannee
Creek (digital)
Suwannee
Creek (chart)

Floyd's Prairie

Sapling Prairie

Suwannee
River

Billy's Lake

SCFSP

Sill (Brown
Trail)

Craven's
Hammock

Northeast
Basin

Kingfisher
Landing

Double Lakes

Durdin Prairie

Central Basin

SCRA

Seagrove Lake


36.96


36.39


35.64

37.08

34.65


35.03

34.95

35.08


35.51





37.55


37.54

37.35


36.52


36.63


0.12


0.04


0.02

0.02

0.27


0.05

0.08

0.23


0.08





0.02


0.02

0.004


0.05

0.04


36.13


35.77


35.26

36.74

33.31


34.53

34.07

33.29


35.05





37.25


37.16

37.17


35.72

36.00


38.07


36.76


36.03

37.36

35.39


35.52

36.04

36.10


36.11





37.86


37.86

37.49


37.04

37.10









Table 2-2-continued.


Mean Daily Variance in Minimum Maximum
Basin and Water Surface Daily Water Daily Water Daily Water
Station* Elevation Surface Surface Surface
Elevation Elevation Elevation


Chase Prairie

Gannett Lake

Territory
Prairie

Chesser Prairie

Coffee Bay

Sweetwater
Creek

Honey Prairie

Southeast
Basin

Moonshine
Ridge

Soldier's Camp

Southwest
Basin

Sapp Prairie
(chart)

Sapp Prairie
(digital)

Cypress Creek


36.49

36.78

36.90


36.63

36.81

34.56


37.15




35.81


34.18




36.61


36.47


34.14


0.01

0.03

0.03


0.02

0.04

0.02


0.01




0.01


0.05




0.01


0.02


0.06


36.75


36.17

36.37


36.26

36.13

34.11


36.86




35.55


33.70




36.36


36.13


33.35


"Operating interval duration for each station is included in Table 2-1.


36.02

37.17

37.31


36.91

37.08

35.07


37.31




36.01


34.59




36.91


36.72


34.69











Table 2-3. Summary of water level and precipitation recorder performance during
12-5-1979 through 6-15-1995 at Okefenokee National Wildlife Refuge.


Duration of Duration of Proportion of
Type of Data Recorder Recorder Days
Station Collected Installment Operation Recorder
(days) (days) Functioning
________ _______ ________ ________ Properly


Chase Prairie

Double Lakes

Gannett Lake

Kingfisher
Landing

Moonshine
Ridge

Sapp Prairie
(chart)

SCFSP

SCRA

Seagrove
Lake

Sill (Brown
Trail)

Soldier's
Camp

Suwannee
Creek (chart)
Territory
Prairie

Billy's Lake

Chesser
Prairie

Coffee Bay


water level

water level

water level

water level


water level


water level


water level

water level

water level


water level


water level


water level


water level


water level

water level


water level


5519

5672

5490

5635


4851


2882


5672

5672

5672


5614


694


878


5519


1156

862


1172


4358

4343

3843

3553


2277


1482


4071

5017

4328


2649


681


854


3748


1141

791


967


83









Table 2-3--continued.


Duration of Duration of Proportion of
Type of Data Recorder Recorder Days
Station Collected Installment Operation Recorder
(days) (days) Functioning
_______ _______ ________ __Properly


Craven's
Hammock

Cypress Creek

Durdin Prairie

Floyd's
Prairie

Honey Prairie

Suwannee
River

Sapp Prairie
(digital)

Sapling
Prairie

Suwannee
Creek
(digital)

Sweetwater
Creek

Craven's
Hammock

Coffee Bay

Durdin Prairie
Floyd's
Prairie
Honey Prairie

Sapling
Prairie


water level


water level

water level

water level


water level

water level


water level


water level


water level



water level


precipitation


precipitation

precipitation

precipitation


precipitation

precipitation


1171


1169

1102

1170


1101

1170


862


862


1155



1170


1162


1171

1176

1156


924

749


1039


1036

963

1117


224
1060


677


639


975



883


997


960

1051

1103


236

635


89

87

95


20

91


79


74


84



75


86


82

89

95


26

85









Table 2-3--continued


Duration of Duration of Proportion of
Type of Data Recorder Recorder Days
Station Collected Installment Operation Recorder
(days) (days) Functioning
_________ _______ Properly


Sapp Prairie
(digital)

Suwannee
River

SCFSP

Double Lakes

SCRA

Chase Prairie

Seagrove
Lake

Kingfisher
Landing

Gannett Lake

Territory
Prairie

Sill (Brown
Trail)

Moonshine
Ridge

Suwannee
Creek (chart)

Soldier's
Camp
Sapp Prairie
(chart)


precipitation


precipitation


precipitation

precipitation

precipitation

precipitation

precipitation


precipitation


precipitation

precipitation


precipitation


precipitation


precipitation


precipitation


precipitation


I L I I


97


684


1126


5501

5375

5784

5454

5553


5634


5344

5425


5605


4458


878


704


2888


663


1033


2407

3469

4034

3650

3342


3017


3853

3523


2798


2398


766


552


1161












chart recorder operation, 46% operated for >75% of the installation period. Over the

installation period of digital recorders (862-1172 days), 85% functioned for >75% of

the interval. If the operation period is pro-rated to the same length for both recorder

types (first 1054 days after installation), 86% of the chart recorders were operating

for >75% of the interval; 81% of the digital recorders had similar performance

(Table 2-4). These performance ratings should be considered in management of the

monitoring network. The initial performance of the chart recorders surpasses that of

the digital equipment; their longevity is proven; the record is continuous (not point

observations by time intervals); and, if data retrieval and station maintenance are

regular, data management procedures can be as automated as that for digital

recorders. New Steven's chart recorders and platforms should be considered for

replacement of old instrumentation, especially for remote, seldom-visited sites.

Most of the existing units are experiencing failure due to decaying installation

platforms, not necessarily due to failure of the recording equipment. Repairs on the

chart instruments can generally be made in place without the diagnostic equipment

needed for digital units. The digital units should be located in the more accessible

locations, since maintenance frequency is generally higher, recorders are less

reliable, and diagnoses are more difficult.

Estimation of Missing Water Level Data


The swamp hydrology model requires starting water depths throughout the

swamp and a dataset of bi-weekly, average water depths for model calibration.