Mississippi State University Scholars Junction

Theses and Dissertations Theses and Dissertations

1-1-2010

The Use Of General Land Office Records And Geographical Information Systems For Restoration Of Native Prairie Patches In The Jackson Prairie Region In

Michael Tobit Gray

Follow this and additional works at: https://scholarsjunction.msstate.edu/td

Recommended Citation Gray, Michael Tobit, "The Use Of General Land Office Records And Geographical Information Systems For Restoration Of Native Prairie Patches In The Jackson Prairie Region In Mississippi" (2010). Theses and Dissertations. 4685. https://scholarsjunction.msstate.edu/td/4685

This Graduate Thesis - Open Access is brought to you for free and open access by the Theses and Dissertations at Scholars Junction. It has been accepted for inclusion in Theses and Dissertations by an authorized administrator of Scholars Junction. For more information, please contact [email protected].

THE USE OF GENERAL LAND OFFICE RECORDS AND GEOGRAPHICAL

INFORMATION SYSTEMS FOR RESTORATION OF NATIVE PRAIRIE

PATCHES IN THE JACKSON PRAIRIE REGION IN MISSISSIPPI

By

Michael Tobit Gray

A Master’s Thesis Submitted to the Faculty of Mississippi State University in Partial Fulfillment of the Requirements for the Degree of Master in Landscape Architecture in the Department of Landscape Architecture

Mississippi State, Mississippi

December 2010 Copywright by

Michael Tobit Gray

2010

THE USE OF GENERAL LAND OFFICE RECORDS AND GEOGRAPHICAL

INFORMATION SYSTEMS FOR RESTORATION OF NATIVE PRAIRIE

PATCHES IN THE JACKSON PRAIRIE REGION IN MISSISSIPPI

By

Michael Tobit Gray

Approved:

Timothy J. Schauwecker Robert F. Brzuszek Assistant Professor of Landscape Associate Professor of Landscape Architecture Architecture (Director of Thesis) (Committee Member)

William Cooke G. Wayne Wilkerson Associate Professor of Geosciences Associate Professor of Landscape (Committee Member) Architecture Graduate Coordinator in the Department of Landscape Architecture George M. Hopper Interim Dean of the College of Agriculture and Life Sciences

Name: Michael Tobit Gray

Date of Degree: December 10, 2010

Institution: Mississippi State University

Major Field: Landscape Architecture

Major Professor: Dr. Timothy J. Schauwecker

Title of Study: THE USE OF GENERAL LAND OFFICE RECORDS AND GEOGRAPHICAL INFORMATION SYSTEMS FOR RESTORATION OF NATIVE PRAIRIE PATCHES IN THE JACKSON PRAIRIE REGION IN MISSISSIPPI

Pages in Study: 170

Candidate for Degree of Master of Landscape Architecture

The Jackson Prairie Region represents a rare, poorly understood and threatened ecosystem. A review of the literature concerning the ecology, physiography and geology of the Region was undertaken, along with a search of General Land Office (GLO) records for locations of historic prairie patches. The prairie patch location data was transcribed, digitized and inserted as a base map in a Geographical Information System (GIS). A set of current, local conditions indicating suitability for conservation or restoration, along with importance weights for each condition, was derived from stakeholder input. A simple additive weighting technique was used to rank the locations in terms of suitability for conservation or restoration. Historic patch locations were visited and the presence of prairie indicator species noted. The use of GLO records and GIS in this way improves the ability of landscape architects to enhance and preserve an imperiled habitat as they design across multiple scales.

Key words: General Land Office records, Geographical Information Systems, Jackson

Prairie Region

DEDICATION

I dedicate this to my truest and dearest friend, my wife, Sally Hatch Gray.

ii

ACKNOWLEDGEMENTS

The author wishes to express his sincere gratitude to many people helped this thesis to become reality. Dr. Timothy J. Schauwecker brought the project to me and gave me invaluable guidance in the conception and direction that it could take. His comments during the writing phase were concise and extremely constructive. Professor Robert F.

Brzuszek taught me a great deal about natural systems and prairies in particular and I hope to continue learning from him throughout my life. Dr. William Cooke, a born teacher, fired my imagination regarding Geographical Information Systems and the fundamental issues surrounding spatial data. Kate Grala was enormously helpful, not only in the automation of the line-generation process, but particularly as a friendly guide to the perplexing intricacies of GIS software operations. Taylor Dearman helped me a great deal with reading scanned manuscripts of the surveys and identifying the useful ones. The people affiliated with the Nature Conservancy of Mississippi, especially Stacey

Shankle, conservation director, Heather Sullivan, Mississippi Natural Heritage Program botanist, and Ken Gordon, U. S. Forest Service botanist, gave me numerous hints, tips, ideas and words of encouragement along the way. I would like especially to thank the anonymous TNC donor who funded the research. Above all I want to thank my wife,

Sally, for her patience and understanding.

iii

TABLE OF CONTENTS

DEDICATION...... ii

ACKNOWLEDGEMENTS...... iii

LIST OF TABLES...... vii

LIST OF FIGURES ...... viii

LIST OF ABBREVIATIONS ...... xii

CHAPTER

I. INTRODUCTION ...... 1

1.1. Purpose of Study...... 1 1.2. Objectives of Study...... 1

II. LITERATURE REVIEW ...... 3

2.1. The Genesis of the Prairie Habitat in North America ...... 3 2.2. Blackland Prairies ...... 9 2.3. The Jackson Prairie Region ...... 10 2.3.1. Geology and of The Jackson Prairie Region...... 11 2.3.2. Vegetation of the Jackson Prairie Region...... 13 2.3.3. Areal Extent of the Jackson Prairie Region ...... 16 2.3.3.1. Introduction...... 16 2.3.3.2. Harper and Hilgard ...... 18 2.3.3.3. Lowe and Fenneman ...... 19 2.3.3.4. Geologic Map of Mississippi...... 22 2.3.3.5. Edwards and Merril et al...... 25 2.3.3.6. Hodgkins et al. and Chapman et al...... 28 2.3.3.7. MARIS...... 37 2.3.3.8. The Mississippi Comprehensive Wildlife Conservation Strategy...... 37 2.4. General Land Office (GLO) Records ...... 39 2.4.1. Historical Background ...... 39

iv 2.4.2. The Use of GLO Records in Research...... 42 2.4.2.1. Introductions ...... 42 2.4.2.2. Barone...... 44 2.5. Suitability Maps...... 48 2.5.1. McHarg ...... 49 2.5.2. Hopkins...... 51 2.5.3. Carr and Zwick ...... 54 2.5.4. Malczewski ...... 58

III. METHODS ...... 60

3.1. Format of Surveys Used for This Study ...... 60 3.2. Establishing the Study Area...... 61 3.3. Transcribing the Data...... 63 3.4. Inserting the GLO Data into a GIS ...... 66 3.5. Establishing Criteria for Suitability ...... 68 3.5.1. Selecting the Datasets ...... 69 3.5.2. Rationales for Selections...... 70 3.5.2.1. State Park Land...... 71 3.5.2.2. Sixteenth Section Land ...... 71 3.5.2.3. National Forest Land ...... 72 3.5.2.4. Transmission Lines ...... 74 3.5.2.5. Land Cover...... 74 3.5.2.6. Distance to Primary Roads...... 74 3.5.2.7. Kernel Density ...... 75 3.5.2.8. Patch Size ...... 75 3.5.3. Establishing Scores and Weights ...... 75 3.5.4. Data Preprocessing...... 78 3.5.4.1. State Park Land...... 78 3.5.4.2. Sixteenth Section Land ...... 80 3.5.4.3. National Forest Land ...... 80 3.5.4.4. Transmission Lines ...... 80 3.5.4.5. Land Cover...... 82 3.5.4.6. Distance to Primary Roads...... 82 3.5.4.7. Kernel Density ...... 83 3.5.4.8. Patch Size ...... 85 3.6. The Map Algebra Equation ...... 85

IV. RESULTS AND DISCUSSION ...... 87

4.1. Introduction...... 87 4.2. Map Algebra Output ...... 87 4.3. GLO Prairies East of the Pearl River ...... 93 4.4. Ground-truthing and the Resilience of Patches ...... 97 4.4.1. The Study Area...... 98

v 4.4.2. The Top Scores East of the Pearl River ...... 101 4.4.3. New Discoveries ...... 105 4.4.4. GLO Patches Persist as Open Land Classes...... 108 4.5. Plat Maps and Survey Notes: Two Georeferencing Strategies ...... 109 4.6. A Heterogeneous Jackson Prairie Region...... 115 4.7. Landscape Architecture: Implications and Applications ...... 116 4.7.1. Design in the Context of Environmental Crises ...... 116 4.7.2. Ecological Design: A Brief History ...... 117 4.7.3. Implications for Design at Different Scales ...... 121 4.7.3.1. Site Specific Design ...... 122 4.7.3.2. Ecological Restoration ...... 124 4.7.3.3. Urban Planning ...... 125 4.7.3.4. Regional Management Decisions ...... 127

V. CONCLUSION ...... 129

WORKS CITED ...... 134

APPENDIX

A. SURVEY TRANSCRIPTION...... 141

B. GROUND-TRUTHING NOTES ...... 168

vi

LIST OF TABLES

3.1 Transcription of GLO data in figure 3.3 ...... 65

3.2 Variables Selected for Suitability Analysis ...... 70

3.3 Weight of Importance Values Assigned by Stakeholders ...... 76

3.4 Stakeholder Scores for Land Cover Classes ...... 77

4.1 Classes of Prairie Patch Resilience Observed during Ground- truthing...... 99

4.2 Variable Scores, Overall Scores and Notes: Top Ten Sites East of Pearl River ...... 103

vii

LIST OF FIGURES

2.1 Extents of tallgrass prairie in North America ...... 8

2.2 A generalized model of the geology, vegetation and topography of a cuesta in southwestern Arkansas, adapted from Foti et al. (2003)...... 11

2.3 Fenneman’s (1938) schematic of the underlying geology of the East Gulf Coastal Plain ...... 12

2.4 Lowe’s (1915) map of Mississippi Regions ...... 20

2.5 Fenneman’s (1938) map “Belts of East Gulf Coastal Plain and a part of the Sea Island Section” ...... 22

2.6 Geologic Map of Mississippi from the Mississippi Department of Environmental Quality (2009) ...... 24

2.7 Edwards’ (1954) Physiography of Hinds County, Mississippi ...... 26

2.8 A context map for Bulletin 126, Newton County Geology and Mineral Resources by Merril et al. (1985) ...... 27

2.9 Forest Habitat Regions Map by Hodgkins et al. (1976) ...... 29

2.10 Detail of Forest Habitat Regions Map by Hodgkins et al. (1976) with county names added and Jackson Prairie Region highlighted ...... 30

2.11 USGS Map of Ecoregions by Chapman, et al. (2004) ...... 33

2.12 Detail of USGS Map of Ecoregions by Chapman, et al. (2004) ...... 34

2.13 Physiographic Regions of Mississippi from MARIS (1994) and the ten counties of the study area...... 36

2.14 Jackson Prairie Habitat as designated in the CWCS published by The Mississippi Museum of Natural Science (2005) ...... 39

viii

2.15 Indian land cessions and the Jackson Rock Formation ...... 41

2.16 Location and distribution of GLO prairie patches in the Jackson Prairie Region as published by Barone using plat maps and as generated for this thesis using field notes...... 47

2.17 Detail of location and distribution of GLO prairie patches in the Jackson Prairie Region as published by Barone using plat maps and as generated for this thesis using field notes ...... 48

3.1 The one hundred and twenty-nine townships of the study area ...... 62

3.2 The 25 missing surveys, 104 found surveys and the 54 which mention prairie ...... 63

3.3 A typical section line survey...... 64

3.4 Vector lines created in ArcGIS from the GLO data ...... 68

3.5 GLO Prairie, MNHP Jackson Prairie inventory points and National Forest Land ...... 73

3.6 GLO Prairie (magenta line) in and around LeFleur’s Bluff State Park (green line) in Jackson...... 79

3.7 ArcMap screenshot showing the necessity of the buffer for the transmission line shapefile ...... 81

3.8 Kernel density of GLO prairie lines with kernel radius of 5000 meters ...... 83

3.9 Kernel density of GLO prairie lines with kernel radius of 10,000 meters ...... 84

3.10 Kernel density of GLO prairie lines with kernel radius of 20,000 meters ...... 84

3.11 Workflow in ArcGIS/ArcMap ...... 86

4.1 Histogram of the standardized scores of the map algebra suitability output ...... 88

4.2 Map display of the maximum scores of the map algebra suitability output ...... 89

ix

4.3 Map display of the minimum scores of the map algebra suitability output ...... 90

4.4 Map display of the classes of scores for the western part of the study area ...... 91

4.5 Map display of the classes of scores for the central part of the study area ...... 92

4.6 Map display of the classes of scores for the eastern part of the study area ...... 93

4.7 Histogram of the suitability scores for GLO prairies east of the Pearl River ...... 95

4.8 Map display of the highest scoring cells east of the Pearl River ...... 96

4.9 Map display of the lowest scoring cells east of the Pearl River ...... 97

4.10 Historic GLO prairie under a cell phone tower on Blossom Hill Road in Scott County. with Schizachyrium scoparium and Asclepias tuberosa ...... 100

4.11 Good quality prairie relict on a natural gas right-of-way on County Road 31 in Eastern Jasper County...... 100

4.12 Top Score GLO prairies east of the Pearl River, MNHP Jackson Prairie inventory sites, aerial images, roads and streams ...... 102

4.13 GLO Prairie with a high score in an old field in Jasper County with Manfreda virginica, Andropogon glomeratus, Andropogon gerardii and Silphium integrifolium...... 104

4.14 Newly discovered historic GLO prairie relict in Jasper County ...... 106

4.15 Rhamnus caroliniana on a newly discovered GLO prairie in Wayne County ...... 107

4.16 A comparison of landcover data in GLO lines and in random points in the Jackson Prairie Region shows that, while most of the prairies may be gone, the locations tend to remain open classes of land...... 109

x 4.17 GLO prairie lines created for this study (magenta), and a plat map from the BLM georefenced to a township layer from MARIS using ArcGIS control points tool ...... 113

4.18 The same plat map used in figure 4.17 with the prairie polygon file provided by Barone...... 114

xi

LIST OF ABBREVIATIONS

CWCS Comprehensive Wildlife Conservation Strategy

GLO General Land Office

GIS Geographical Information System

MARIS Mississippi Automated Research Information System

MNHP Mississippi Natural Heritage Foundation

NRCS Natural Resources Conservation Service

MDWFP Mississippi Department of Wildlife, Fisheries and Parks

MDEQ Mississippi Department of Environmental Quality

SGCN Species of Greatest Conservation Need

USEPA United States Environmental Protection Agency

USFS United State Forest Service

xii

CHAPTER I

INTRODUCTION

1.1 Purpose of the Study

The purpose of this study is to examine the General Land Office (GLO) records of

the Jackson Prairie Region in Mississippi and determine whether they contain

information about the historic locations of prairie patches that might be of use to

organizations and stakeholders involved in the restoration and conservation of native

prairies. The research questions are these: do the GLO records contain sufficient data

about the geographical locations of the patches to be of use to stakeholders? Can the data in the records be translated to map coordinates in a Geographical Information System

(GIS), and, if so, can the data be combined with other publicly available data layers in an

ESRI ArcMap environment in such a way as to identify the most suitable candidates for restoration?

1.2 Objectives of the Study

Several objectives were developed and met in order to achieve the goal of answering the research question. They include:

1. Examine the GLO records for information about Jackson Prairie. 2. Transcribe the information about Jackson Prairie in the GLO records.

1 3. Put the transcribed information into a Geographical Information System (GIS) so that the prairie patches can be georeferenced, mapped, and subjected to multi-criteria analysis. 4. Conduct stakeholder meetings with conservationists to determine the appropriate criteria for analysis. 5. Develop a logical method for the expression of stakeholder input so that the historic patch locations can be ranked in terms of suitability for restoration or conservation in the context of GIS analysis. 6. Visit the actual locations of the GLO prairies and record present conditions. 7. Obtain stakeholder feedback on the results and make suggestions for future research.

2

CHAPTER II

LITERATURE REVIEW

2.1 The Genesis of the Prairie Habitat in North America

The rise of the grassland biome on the North American continent probably commenced in the Miocene-Pliocene transition five to seven million years ago (Axelrod

1985, 164, Anderson 2006, 638). The uplift of the Rocky Mountains created a wedge- shaped rain shadow in the central part of the continent resulting in a climatic, west-to-east

moisture gradient, with drier, shortgrass prairie and scrub in the west and a shifting

mosaic of tallgrass prairie, oak savannah, and forest in the east, covering an area from

present day Canada to Texas and Nebraska to Ohio (Kline 1997, 3). Whittaker (1975)

describes a moisture gradient across the central part of the continent in terms of an

“ecocline” encountered westward from the southern Appalachians to the Rocky

Mountains, characterized by decreasing heights of the grasses as aridity increases (162).

Prairies in North America are therefore traditionally divided into three sectors based on

average height of the grasses, which is a function of precipitation: shortgrass prairies in

the west, tallgrass or “true prairies” in the east, and a mixedgrass prairie in between

(Anderson 2006, 628). In the eastern tallgrass prairie, the historic shifting pattern of

prairie-forest transition was determined by recent fire history and short-term fluctuations

3 in regional climate, with frequent fires favoring prairie and savanna over forest in the drier periods (Kline 1997, 3).

Anderson (2006) associates the rise of the grassland biome with the adaptive development of C4 grasses, a class characterized by their ability to photosynthesis more efficiently during periods of high temperature, high solar radiation and low soil moisture

(633). There are three main biochemical pathways by which plants use photosynthesis to gain energy from carbon: the calvin cycle (C3), the C4-dicarboxylic acid cycle (C4) and crassulacean acid metabolism (CAM) (Mooney 1972, 316). C4 and C3 grasses are intermixed in the tallgrass and mixedgrass types of prairie, with C3 grasses and forbes generally dominant in mesic conditions of the spring and early summer and C4 grasses emerging later and maximizing growth in midsummer (Anderson 2006, 633).

Anderson (2006) acknowledges the critical role played by fire, stressed by

Axelrod (1985), but cites extensively several recent examinations of the importance of other factors such as topography and the dynamics of air masses in the production of a prairie “peninsula” extending into Illinois and northern Indiana, an area in which the climate generally supports forests. Anderson (2006) also uses recent studies in his discussion of the role played by mammalian grazers. The role played by bison has only been understood in the past two decades, when grasslands large enough to support a reasonable population capable of grazing in ways that simulate historic conditions became available for study (638). Anderson (2006) cites literature suggesting that the presence of bison and fire together may increase species richness, since frequent fires, especially annual burns, favor C4 grasses at the expense of C3 plants, an effect countered by the bison’s preference the C4 grasses for grazing, causing the less palatable C3 plants

4 to maintain their abundance. Anderson (2006) also cites studies indicating that bison grazing patterns increase diversity in species of birds and grasshoppers as well, although the affects of fire on invertebrates in general is varied and an increasing number of entomologists are expressing concern that insects are being harmed by current prescribed burning practices. Anderson (2006) discusses modifications to current practices to mitigate this effect, including leaving portions unburned as refugia for fire sensitive species.

Axelrod (1985) stresses that fire, not climate, determines the pattern of forest and grassland in the Middle West. He notes that grasslands are unstable when in contact with woody vegetation and are therefore often invaded by forests on the eastern and northern extents of their range, and by woodlands on the southern and southwestern extents (165).

Axelrod (1985) points out that, in the absence of fire, forest spreads at the expense of prairie, whereas fires can restrict or even destroy forests (166).

Axelrod admires Sauer (1950) for his well-reasoned rejection of the notion that regional grasslands are controlled by climate. Noting the anthropogenic aspect of fire in the landscape, Sauer (1950) refers to fire-determined grasslands as “chiefly a cultural phenomenom.” He describes them as “ecologic assemblages…not in climax nor in equilibrium,” meaning that they do not persist in the absence of disturbance in the way a

“climax” forest does, as the final stage in a series of successive types of cover, but are disturbance dependant. Topography and climate play a role, but only to the extent that they favor fire: smooth or rolling topography and a climate that allows short periods of dry weather during which the ground cover dries out (20). Pyne (1982) defines a “fire

5 climax” as a certain type of vegetative cover that is maintained by a fire regime and that,

in the absence of fire, would give way to other cover types.

Whittaker (1975) acknowledges this dynamic characteristic of grasslands, that is,

its distribution as arid, seemingly anomalous, patches in areas where, as Cowels (1928)

points out, “rain falls on grass and trees alike,” by giving them special treatment in his

graphed pattern of world biome types as a function of mean annual precipitation (x-axis)

and mean annual temperature (y-axis). Tropical rainforest occupies the warmest, wettest extreme while tundra lies in the colder, drier end of the graph. Whittaker (1975) encircles grasslands with a dot and dash line indicating a wide range of environments in which

either grassland or one of the other types dominated by woody vegetation may form the

vegetative cover of different areas. He notes that in these areas continental climates, soil

effects, and fire effects can shift the balance between woodland, shrubland, and grassland

types (167).

McHarg’s (1992) account of the genesis of prairies is more poetic than most, yet

concurs on the broader elements of the story as presented elsewhere in the literature. He

associates the emergence of grasslands worldwide with the rise of the angiosperms in

Jurassic times. The embryonic plant encapsulated in a fleshy shell was more enduring and

mobile than the naked seed or the older spore, allowing the new class of plants, especially

grasses, to spread and cover more areas of the planet. This gave rise to large herbivores,

fleet-footed and far-ranging prey to predators. Among the predators were humans, who

possessed a “tool far more powerful than required.” McHarg (1992) goes on:

This new and devastating tool was fire. It was no novelty in the prairie – lightning-caused fire was common and indeed the prairie climax was a response

6 to it. But the induced fire of the hunter was more frequent than the natural occurrence. The prairies were burned to drive the bison and deer, the mammoth and mastodon into closed valleys or over precipices. This was a time of climatic adversity, threatening to the creatures as well as to men. It is thought that it was the combination of human hunters and a hostile climate that resulted in the extinction of this first great human inheritance in North America, the prairie herbivores. Firelike the grasses spread, firelike the herds of grazing animals swept to exploit the prairies – and it was the fire of the aboriginal hunter that hastened or accomplished their extinction. This was the first major impact of man on the continent during the aboriginal occupation. (67)

Although the beginning of the rise of the grassland biome in North America can be dated to the Miocene-Pliocene transition 5 million years ago (Axelrod 1985, 195,

Anderson 2006, 638), the extents of forest, woodland and grassland areas have fluctuated greatly with changes in climate over the intervening millennia. Woodland reached extensively into the great plains as recently as 12,000 years ago, at the beginning of the

Holocene epoch. Prairies reached their current extent sometime in the late quaternary, or around the time of the altithermal, as recently as 5,000 years ago (Axelrod 1985, 196). A striking fact about the Central Plains is their lack of endemic species, while the bordering forests and woodlands have numerous endemic species and some endemic genera. This is further evidence that the regional grassland biome attained its present extent chiefly in post-glacial time (166). Packard and Mutel (1997) provide a map of this extent (Figure

2.1). Since most of the native prairie habitat has now disappeared, this map represents areas where prairie restoration can be potentially considered rather than a depiction of the

general presence of prairie.

7

Figure 2.1

Extents of tallgrass prairie in North America

From the perspective of Landscape Architecture and Land Planning, it is tempting

to speculate on the use of fire by humans to maintain a certain quality in the landscape in

order to make it suitable for a particular use (such as hunting) as a primordial land

management practice. The practice may be older than agriculture, and is done on a larger

scale. Pyne (1982) writes extensively on the cultural phenomenon of fire, noting that

humans have since prehistory managed to transport, ignite and otherwise control fire in ways unthinkable regarding other powerful natural forces such as tornadoes, floods and

8 windstorms. He argues that mankind is the primary source of ignition of fire worldwide, and the most significant modifier of the fire environment, and that the exclusion of fire from a landscape is as ecologically powerful as its application, causing dramatic changes in recent history for plant assemblages in the western part of North America.

Most of the original prairie ecosystem of North America has been lost to agriculture and forest encroachment. Less than one percent of the pre-European settlement prairie survives today (Anderson 2002). Jordan (1997) points out that by the time Americans began to care about prairie restoration, the old prairies were almost gone.

Furthermore, an early challenge for practitioners of prairie restoration was that, in contrast to other kinds of ecosystems, such as the forests of the Northeast, prairies, once destroyed, do not come back on their own. Prairies must be re-created, a complicated process characterized by Jordan (1997) as involving ongoing and increasingly sophisticated and self-conscious attempts (xiii). The destruction of the prairies on the continent of North America is described by Jordan (1997) as a “cataclysmic change” occurring within a single life span: “[a]t no other time in history had an entire landscape been changed so dramatically and in so short a time” (xiv).

2.2 Blackland Prairies

The term “blackland prairies” refers to assemblages of scattered prairie patches that occur in belted formations in Texas, Louisiana, Arkansas, Mississippi and .

The belted pattern of prairies in these states can be seen in the map by Packard and Mutel

(1997, Figure 2.1). The term “blackland” refers to the deep mantle of black soil high in organic matter which occurs over a substrate of Cretaceous chalk or marl. Blackland

9 prairies are generally dominated by Andropogon gerardii, A. scoparious, Sorghastrum nutans, and Tripsacum dactyloides (Foti 1989, 23). Because the climate of the Gulf

Coastal Plain is subtropical and the dominant cover is forest, the prairie patches are considered anomalies (Moran 1997). Blackland prairies are physiographically, botanically and ecologically one of the least understood systems in North America

(Leidolf and McDaniel 1998, Foti 1989, Gordon 1989). Heavy agricultural pressure has reduced the number and size of native prairies to the degree that very little remains today, mostly in the form of small patches (0.5 to 5.0 ha) in abandoned pastures, along roadsides, in power line rights-of-way and on public and private natural areas (Peacock and Schauwecker 2003, 4). If ignored, these relict patches may continue to degrade, and the southeastern part of the prairie biome in North America could be lost.

2.3 The Jackson Prairie Region

The Jackson Prairie Region is a belt varying from ten to thirty miles wide, from

Madison and Hinds counties, which include the capital city of Jackson and the surrounding metro area, extending east and southeast and crossing into Alabama at the border of Clarke and Wayne Counties. Earlier descriptions place the western edge at the cities of Vicksburg and Yazoo City (Lowe 1915, 47). General Land Office (GLO) records indicate that the prairie openings extended through the Hills Physiographic

Region of Yazoo County and even into the alluvial soils of the Delta, occupying a total of

19,555 hectares (Barone 2005a). The areal extent of the region is inconsistently described in the literature, a matter discussed in detail below.

10 Most of the known prairie relicts in the region lie within the boundaries of the

Bienville National Forest in Scott, Newton, Smith and Jasper counties (Gordon and

Wiseman 1989, 4). Harrell Prairie Hill was established by the National Forest Service in

1971, and was the only protected prairie relict in the region until the United States Forest

Service (USFS) cleared and began management of six other prairie habitats in 2006.

2.3.1 Geology and Soils of The Jackson Prairie Region

Blackland prairies occur on distinctive calcareous substrates, often on the faces of asymmetrical ridges called cuestas (Figure 2.2), chalk or marl outcrops mantled with only a thin layer of soil that may be black but is more commonly olive (Foti et al. 2003, 96).

Figure 2.2

A generalized model of the geology, vegetation and topography of a cuesta in southwestern Arkansas, adapted from Foti et al. (2003).

11 The Jackson Prairie Region, unlike most blackland prairie belts, is of Eocene rather than Cretaceous origin, having been deposited in a neritic or deep marine environment of the Mississippi embayment approximately 38 million years ago (Moran et al. 1997). Fenneman (1938, 69) demonstrates how the tilted strata relate to the various physiographic regions of the East Gulf Coastal Plain (Figure 2.3). The soil is generally well-drained, alkaline and largely comprised of eroded Maytag soils derived from marls of Yazoo clay situated on gently to moderately sloping uplands (Mississippi Museum of

Natural Science 2005).

Figure 2.3

Fenneman’s (1938) schematic of the underlying geology of the East Gulf Coastal Plain

The name “Jackson” is given not only to the prairie region but to the underlying rock group as well. Although the placement of the physiographic label in this figure implies that the Jackson Prairie Region is additionally underlain by Vicksburg and Claiborne formations, the region is in fact entirely underlain by the Jackson rock group. In his description of this figure Fenneman notes that “not all of the cuestas shown in the section are prominent throughout the area.” 12 Moran et al. (1997) collected soil samples from excavation pits on four prairie

relicts in the Jackson Prairie Region. All four sites contained soils classified as Vertisols

with “very fine and fine, smectitic, thermic Chromic Hapluderts.” According to Buol et

al. (1989):

Typical properties of Vertisols are: clay texture; strong granular structure in the upper 6 to 20 inches; calcareous to neutral reaction; gilgai relief; high shrink-swell potentials; plastic consistency; calcium (Ca) and/or magnesium (Mg) as the dominant cations; montmorillonite as the major clay mineral; calcareous parent material; thick sola; low chromas; high organic matter contents; little weathering; and tallgrass or savanna vegetation.

2.3.2 Vegetation of the Jackson Prairie Region

Jackson Prairie is the least studied vegetation type in Mississippi and may be one

of the rarest natural community types in North America (Gordon and Wiseman 1989, 3).

Only a few floristic inventories have been published (Barone and Hill 2007). Lowe

(1921) lists nineteen herbaceous species characteristic of the Black Belt prairies and

thirty-four characteristic of those of the Jackson Belt. Of the two lists, only four species

are common to both (33, 48-9). Jones (1971) recorded the more typical or common

species of the Jackson Prairie Region. They include Andropogon gerardii, A.

gloemeratus, A. virginicus, Apocynum cannibinum, Asclepis veridis, A. viridiflora, Aster

patens, A. pilosus, Berchemia scandens, Bromus commutatus, Bulelia lycioides, Carex debilis, Centaurea cyanus, Centrosema viginianum, Cocculus carolinus, Cornus asperifolia, Crataegus spp., Crotalaria sagittalis, Croton monanthogynus, Desmanthus illnoensis, Desmodium marilandicum, Echinacea purpurea, Elymus vigincus, Eupatorium coelestinum, Euphorbia hirta, Gentiana villosa, Gleditsia tiracanthos, Helenium

13 autumnale, Heliopsis minor, Ilex deciduas, Ipomoea pandurata, Kuhnia eupatoroides,

Lathyrus hirsutus, Lythrum salicaria, Melilotus alba, Monarda citriodora, M. fistulosa,

Myosotis verna, Neptunia lutea, Oenothera speciosa, Panicum spp., Penstemon laevigatus, Petalostemon candidus, P. purpureus, Phalaris canariensis, Ratibida pinnata,

Robinia pseudo-acacia, Rudbeckia hirta, Ruellia humilis, Sabatia angularis, Scutellaria parvula, Solidago altissima, Sorghastrum nutans, Strophostyles umbellate, Tridens flavus, Verbena halei, V. rigida, and Vitis aestivalis.

The effort to create an inventory of all the relict prairies in the Jackson Prairie

Region, organized by the Mississippi Natural Heritage Program (MNHP), commenced in the 1980s (Gordon and Wiseman 1989). Written by USFS botanists, the original report concentrated on the Bienville National Forest. Fifty-four relict prairies, located through a process of field observation and analysis of National High Altitude Photography (NHAP) color infrared photography, are listed in the MNHP report (5). The MNHP has made the inventory available in the form of an ArcGIS shapefile. A “shapefile” is a proprietary

Environmental Systems Research Institute, Inc. (ESRI) GIS file format that includes a series of linked files that detail the map projection, database attributes and geometry. The shapefile of the prairie patch inventory used for this study features a total of seventy-three sites. The official MNHP inventory for Jackson Prairie remnants has since grown to eighty-nine locations (Ken Gordon, March 5, 2009, e-mail message to author).

The MNHP 1989 Bienville National Forest Prairie Survey (Gordon and Wiseman

1989) does not include a list of typical prairie species. Gordon and Wiseman use a list of prairie indicators in their scoring system used to rank the locations. Each prairie was given a point for the presence of each of the following: Petalostemon candidum, P.

14 purpureum, Echinacea purpurea, Ratibida pinnata, Andropogon scoparius, Andropogon gerardii, Sorghastrum nutans, Aster novae-angliae, Asclepias hirtella, Spiranthes magnicamporum, and Silphium species. Points were subtracted for the presence of two species considered weeds in the prairies: Melilotus alba and Gaura species. The report notes the presence of Juniperis virginiana, in many of the remnants. The report also notes the absence of two prairie indicator species, Silphium laciniatum and Boutelona curtipendula, from the fifty-four locations, even though both are known to occur in the region (10). The latter has been found in one of the seventy-three locations listed in the shapefile versions of the MNHP inventory.

Gordon and Wiseman (1989) describe the Jackson Prairie as “vegetationally similar” to the Black Belt. Barone and Hill (2007) collected and identified grasses and forbs from five Jackson Prairie and fourteen Black Belt prairie remnants. Treating the two belts as essentially the same, they identify 198 species, 168 of them native. Seven of the species were found at all or all but one site and these were “generally widespread species often observed along roadsides or old fields”. Species common at prairie remnants but generally rare elsewhere included Ratibida pinnata, Dalea candida, Dalea purpurea, Neptunia lutea, Rudbeckia hirta, Silphium laciniatum and Symphyotrichum ericoides (227). Comparing their inventory to the lists of characteristic prairie species published in the early part of the twentieth century, including Lowe (1921), Barone and

Hill (2007) assert that the species that have disappeared in the interim are associated with mesic and sandy conditions. They conclude that in the effort to conserve and restore the endangered blackland prairie ecosystem, “special attention should be given to locate and

15 preserve any existing remnants that are on mesic or sandy soils, as these sites may

contain some of the missing species” (231).

Some woody species characteristic of Jackson Prairie include Rhamnus

lanceolata, Cercis Canadensis, Cornus drumundii, Fraxinus spp. and Diospyros viginiana. Rare woody species found in the Jackson prairie include Quercus

oglethorpensis, hawthorns: Crataegus ashei, C. triflora, and C. meridianalis. The region

is also home to the Jackson Prairie crayfish, Procambarus barbiger, a candidate species

of the Endangered Species Act (Elsen and Wieland, 2003).

2.3.3 Areal Extent of The Jackson Prairie Region

2.3.3.1 Introduction

The areal extent of the Jackson Prairie region has been variously described over

the years, with the land between the Pearl and Big Black rivers, or portions of Hinds and

Madison Counties, either included or excluded depending on the source of the map. The

dispute arises because features of four identified physiographic regions, Jackson Prairie,

Loess Hills, North Central Hills and South Central Hills, are present in the area.

Fenneman, in his introduction Physiography of the Eastern United States (1938), asserts

that the term “physiography” represents an overlapping of the fields of geology and

geography. The data used to organize the land into physiographic regions are the product

of geological processes. The ultimate descriptions based on these data “should be the

basis of geography” (v). Fenneman acknowledges that the boundaries can differ for each

science and that even within the single science of physiography “two men, or the same

16 man at different times, may wish to emphasize different elements of the picture and, accordingly, to divide the total area in a different manner” (vi). The following survey of attempts to circumscribe and describe the Jackson Prairie Region perhaps could be read as a catalogue of various efforts of individuals to do exactly that.

The Jackson Prairie Region has been described as a “Soil Region” (Lowe 1915,

164), a “Physiographic Region” (MARIS Technical Center 1994), one of a number of

“Belts of The East Gulf Coastal Plain” (Fenneman 1938, 68), a “Physiographic Province”

(Merril 1985, 16), a “Forest Habitat Region” (Hodgkins, Cannon and Miller, 1976), an

“Ecoregion” (Chapman, et al. 2004) and a “habitat type” (Mississippi Museum of Natural

Science, 2005). The earliest maps of the prairies were in the form of plat maps created by the deputy surveyors of the General Land Office (GLO) in the 1820s and 1830s. The

GLO surveyors were principally concerned with an accurate establishment of section, quarter section, township and range boundaries for the state of Mississippi, prairie being one of a number of features in the landscape that were recorded. It wasn’t until Barone

(2005a, 2005b) digitized and stitched together the GLO plat maps using ArcView software that the entirety of the belted pattern of pre-European settlement prairie formations in the GLO records became visible. Barone’s digitized plat maps reveal historical records of prairies in some unexpected places, including prairies located on alluvial soils close to the Loess Hills on the eastern edge of the Delta (Barone 2005a,

146).

17 2.3.3.2 Harper and Hilgard

Harper (1857), noted that the timbered lands, on “the orange sand of the Miocene

rocks,” are interspersed with prairies where the “white carbonate of lime constitutes the

upper stratum” (163). Harper does not use the term “Jackson Prairie,” but describes an

arrangement of prairie patches in Wayne, Clarke, Jasper, Smith, Scott and Rankin

Counties. Harper, then, does not specifically say that Madison and Hinds Counties

contain prairies. His geologic map of the state shows southwestern Madison County and

all of Hinds County as Eocene, in contrast to the counties to the north, but since he

describes the entire southern half of the state as Eocene, the Jackson Prairie Region is not

recognized as a distinctive geologic belt. Harper’s map shows prairie openings in the

Jackson Belt and the Black Belt, and, in a designation that anticipates Axelrod (1985), he describes them on the map and in the text as “Postpliocene,” or a much more recent formation.

Hilgard (1860), also a State Geologist, describes a “Central Prairie Region” that includes the counties Harper mentioned above and adds Madison, Hinds and Yazoo counties (330), extending the range to the north and west and bringing the description more in line with the observations in the GLO records. Hilgard notes that the prairies are

“not generally the prevalent feature of its surface,-not even to the extent to which this is the case in the Northeastern Prairie Region,” by which he means the Black Belt (330, emphasis his). Hilgard’s (1860) observation is consistent with present day literature: prairie openings in the Jackson Prairie region are anomalous patches in a forested matrix and as such are smaller in size and more sparsely distributed than those of the Black Belt

Region. Hilgard recognizes that prairie openings exist in Hinds and Madison Counties,

18 which he describes as “interspersed with patches of prairie, and modified more or less, in

different localities, by the admixture of the calcareous, and mostly very clayey, materials

of the Tertiary” (332).

2.3.3.3 Lowe and Fenneman

E.N. Lowe, director of the Mississippi State Geological Survey in the early part of

the twentieth century, described the Jackson Prairie Region as “a belt varying from 10 to

30 miles wide…in the extreme western part from the longitude of Jackson to the bluff

hills; eastward it narrows to the Alabama line” (Lowe 1915, 47). Lowe’s map (164,

Figure 2.4) divides the state into “Soil Regions” with names and boundaries that

correspond with Fenneman’s belts. Lowe’s map and text make clear that he considers portions of Madison and Hinds counties to be part of the Jackson Prairie Region. He designates the northwest half of Madison County as North Central Plateau, and his

Jackson Prairie Belt covers the southwest half of the county, extending into the northern fourth of Hinds County. The remaining three-fourths of Hinds County is designated as

Long Leaf Pine Hills. For Lowe, the Loess Hills do not exist in Hinds or Madison

County.

19

Figure 2.4

Lowe’s (1915) map of Mississippi Soil Regions

20 Fenneman (1938, 68) does not include county boundaries in his map entitled

“Belts of the East Gulf Coastal Plain and a part of the Sea Island section” (Figure 2.5), but the Yazoo, Big Black and Pearl Rivers are easily recognizable. Since the Big Black and Pearl Rivers form much of the boundary of Hinds and Madison counties, Fenneman’s

belief that the Jackson Prairie Belt exists in these counties is demonstrated. An interesting

difference between Fenneman (1938) and Lowe (1915) is that while Lowe observes the

Jackson Prairie as crossing the Big Black River and extending into Yazoo County,

Fenneman marks the Big Black River as the approximate northwestern boundary. The

difference may simply be the result of a difference in scale: Lowe is mapping the state of

Mississippi while Fenneman is mapping the Gulf Coastal Plain. For whatever reason, the boundary has shifted southward and eastward slightly between Lowe (1915) and

Fenneman (1938). This southward and eastward movement in the boundary will progress

further with Merril, et al. (1985), and further still with Hodgkins et al. (1976) and

Chapman et al. (2004), discussed below.

21 Figure 2.5

Fenneman’s (1938) map “Belts of East Gulf Coastal Plain and a part of the Sea Island Section”

2.3.3.4 Geologic Map of Mississippi

A geologic map of Mississippi published by the Division of Geology in the

Mississippi Department of Environmental Quality (DEQ) shows the belted patterns of formations and rock units in the state (Thompson 2009, Figure 2.6). The Pleistocene loessal deposits are shown as a band striking north-south, generally perpendicular to the orientation of the geologic belts. Geologists have mapped the older strata underneath the loess, presumably because loessal areas are subject to erosion and deep incision and the older strata can be easily observed in stream channels. Lowe (1915) describes the loess hills as bluffs on the western edge, 150 to 250 feet above the delta. The loess itself

“varies in depth from 30 to 90 feet at the edge of the bluff, thinning eastward, to the east

22 margin of the region, where it feathers out” (33). Fenneman (1938) describes a loess belt

5 to 15 miles wide that stretches from the Ohio River to the Gulf of Mexico and also notes that the thickness of the loess “decreases rapidly toward the east” where it loses its

“distinctively calcareous” character and is more accurately described as brown yellow loam (80). The edge of the loess as shown on Thompson’s 2009 map, then, is actually a gradient of one type of land form into another, giving rise to inconsistencies about the western edge of the Jackson Prairie Region among the maps reviewed here.

23

Figure 2.6

Geologic Map of Mississippi from the Mississippi Department of Environmental Quality (2009)

24 2.3.3.5 Edwards and Merril et al.

A 1954 master’s thesis by Douglas Edwin Edwards, The Physiography of Hinds

County, repeats Lowe’s view that the Jackson Prairie Region extends into Yazoo County

(36). Edwards (1954) divides the portion of the Jackson Prairie Region in Hinds County into two sub-regions: the typical Jackson Prairie sub-region and the Vicksburg type

Prairie sub-region, according to differences in geological formation, a slight difference in topography, and the fact that the two sub-regions are divided by a “major landform, the

Forest Hill cuesta” (Figure 2.7). Edwards’ physiographic map also includes a group of polygons labeled “prairies as mapped in 1820.” Presumably these are derived from GLO plat maps. The notion of a “Vicksburg” prairie belt or sub-region corresponds to the

Vicksburg strata in Thompson (2009) and is repeated by Merril et al. (1985, Figure 2.8).

Merril et al. (1985) and Edwards (1954) describe a Jackson Prairie Region that extends west of the Pearl River into Madison and Hinds Counties.

25

Figure 2.7

Edwards’ (1954) Physiography of Hinds County, Mississippi

26

Figure 2.8

A context map for Bulletin 126, Newton County Geology and Mineral Resources by Merril et al (1985)

Merril et al (1985) show 12 “Physiographic Provinces.” The categories “Paleozoic Bottoms” in the northeast corner and the “Vicksburg Hills” just to the south of the Jackson Prairie Region are added to Lowe’s 1915 list.

27 2.3.3.6 Hodgkins et al. and Chapmen et al.

Hodgkins, et al., (1976) divide Mississippi into five provinces, which in turn are

divided into 18 regions (Figure 2.9). Some of these regions are then divided into different

types. Although their map is titled “Forest Habitat Regions from Satellite Imagery,” the

researchers stress in the supplement to the map (Hodgkins et al., 1979) that theirs is a

physiographic classification system (17). Hodgkins, et al. (1976) place the Jackson

Prairie Region in the Hilly Coastal Plain Province. Their map marks the Pearl River as the northwestern boundary of the Jackson Prairie Region, so that the Region does not reach into Madison or Hinds counties (Figure 2.10). Almost all of Madison and Hinds

Counties are here classified as in the Deep Loess Plains Region of the Deep Loess

Province, the exceptions being a band in the northwestern part of Madison County (The

Lower Loam Hills Region in the Hilly Coastal Plain Province) and a section along the

southern boundary of Hinds County (The Southern Loam Hills Region in the Middle

Coastal Plain Province). No mention is made in the map notes or in the supplemental bulletin of prairie as a potential land cover in the Jackson Prairie Region, rather it is

generally described as very good for loblolly pine and good for hardwoods, with alkaline

areas that are good for pastures and hay fields.

28

29

Figure 2.9

Forest Habitat Regions Map by Hodgkins et al (1976)

Figure 2.10

Detail of Forest Habitat Regions Map by Hodgkins et al (1976) with county names added and Jackson Prairie Region highlighted.

In terms of geographical area, Hodgkins, et al. (1976) present the narrowest of any version of the Jackson Prairie Region in any of the maps considered for this study.

This is due to the fact that they divide what is commonly viewed as the Jackson Prairie

Region into Jackson Hills and Jackson Prairie Regions. They describe the Jackson Hills

Region (27 in Figure 2.10) as a “long, ruggedly hilly divide, of deep sandy and loamy soil, extending throughout much of the length of the so-called ‘Jackson Prairie.’” The

Jackson Prairie Region is described simply as “the portion of the ‘Jackson Prairie’ not

30

buried under the sandy Jackson Hills. The topography is gently rolling, and land clearing

is extensive” The Jackson Prairie Region is divided into “Loessal,” and “Non-Loessal” types (28A and 28B respectively in Figure 2.10). The distinction does not indicate two

different kinds of prairies: the “Non-Loessal” Jackson Prairie is characterized as having

extensive areas of deep clays that are mostly acid to depths of 3 feet or more, very

suitable for loblolly pine, moderately suitable for hardwoods, and areas of alkaline clay

generally used for pasture. Acid alluvial clays and alkaline alluvial clays commonly used

for pasture are also mentioned. “Loessal” prairies on the other hand, are described as the

portion of the ‘thin-loess’ belt where the loess mantle covers Jackson Prairie material.

Where erosion has removed the loess mantle, soils are “approximately as described for

28A” or the “Non-Loessal” type. Hodgkins et al. (1976), then, essentially describe a

narrow band of calcareous marl overlain to the south by the sandy “Jackson Hills” and to

the west by loessal soils of increasing depth as one moves westward, with no prairie

habitat west of the Pearl River.

The United States Geological Survey (USGS) has published a map of ecoregions

of Mississippi (Chapman, et al., 2004, Figures 2.11, 2.12). Chapman, et al. (2004) define

ecoregions as areas of regional similarity in ecosystems and natural resources. The

approach, as outlined in the notes that accompany the map, is “based on the analysis of

the spatial patterns and the composition of biotic and abiotic phenomena that affect or

reflect differences in ecosystem quality and integrity. These phenomena include geology,

physiography, vegetation, climate, soils, land use, wildlife, and hydrology.” The United

States Environmental Protection Agency (USEPA) has developed a Roman numeral

hierarchical scheme for the classification of ecoregions. Mississippi contains four level

31

III and 21 level IV ecoregions. The general pattern follows that of the other maps

discussed here. Like the Forest Habitats map, the Ecoregions of Mississippi use the Pearl

River as the northwestern boundary of the Jackson Prairie and locate none of the region in Madison or Hinds Counties. The ecoregions map differs slightly from the Forest

Habitat Regions Map in the configuration of the boundary between the Loess Plains and the Southern Hilly Gulf Coastal Plain in northern Madison County, but agrees with the classification of most of the two counties as being Loess Plains.

32

Figure 2.11

USGS Map of Ecoregions by Chapman, et al. (2004)

33

Figure 2.12

Detail of USGS Map of Ecoregions by Chapman, et al. (2004).

The ten counties of the study area are emphasized here with red outlines and labels. From the poster legend: 74a: Bluff Hills; 74b: Loess Plains; 65d: Southern Hilly Gulf Coastal Plain; 65r: Jackson Prairie; 65q: Burhstone/Lime Hills.

34

The notes that accompany the USGS map of Ecoregions of Mississippi describe it

as the product of a number of agencies, including the USFS, the USEPA and the Natural

Resources Conservation Service (NRCS) working together to develop a common

framework of ecological classifications. This process requires awareness of the

differences in conceptual approaches and mapping methodologies applied by the different

agencies. The notes assert that “as each of these frameworks is further refined, their

differences are becoming less discernable.” Agencies are reaching a consensus on how to

draw the lines on the land. It is possible that these two maps, Forest Habitats and

Ecoregions, and the interagency cooperation involved in their creation, led the

Mississippi Automated Resource Information System (MARIS) Technical Center to

remove Madison and Hinds counties from the Jackson Prairie Region in their map of the

Physiographic Regions of Mississippi (Figure 2.13). Such a designation is based on

reasonable assessments of what particular features (in this case, those associated with

loess plains) are dominant in the landscape. Land classification in the presence of a fuzzy

boundary is not entirely a deterministic process, however, and landscape architects and land planners working in these counties should be aware of the presence of prairies in the historic record in these counties and the inclusion of these counties as part of the Jackson

Prairie Region by previous observers.

35

Figure 2.13

Physiographic Regions of Mississippi from MARIS (1994) and the ten counties of the study area

36

2.3.3.7 MARIS

The MARIS Technical Center is responsible for the management of the on-line

spatial data warehouse for the state of Mississippi. As a digital file available online, their

physiographic map has accompanying metadata describing its origins, which states that

the “coverage was compiled from a number of data sources that describe the state's physiographic regions. It was drawn by MTC staff and select representatives of other state agencies based on forest habitat regions, soil associations, and major land resource areas” (MARIS, 1994).

Two features of the MARIS map of physiographic regions are distinct from all other maps. One is that the Jackson Prairie Region is bisected by the North Central Hills

Region in eastern Jasper County. The other involves the way MARIS classifies Madison and Hinds counties: rather than designate them as Loess or Jackson Prairie, or draw the line in the middle as might be indicated by Thompson (2009), Merril et al. (1985) or

Edwards (1954), MARIS (1994) eliminates both Loess and Jackson Prairie from these two counties (except for the western edge of Hinds County), and replaces them with the

North and South Central Hills Regions.

2.3.3.8 The Mississippi Comprehensive Wildlife Conservation Strategy

The Mississippi Comprehensive Wildlife Conservation Strategy (CWCS) was designed to serve as a comprehensive guide for fish and wildlife conservation statewide for the next half century and encompasses a broad set of conservation strategies for wildlife and fish species and their key habitats (Mississippi Museum of Natural Science,

2005). The authors of this document, for the most part researchers associated with the

37

Mississippi Department of Wildlife and Fisheries (MDWFP) and the Mississippi Natural

Heritage Foundation (MNHP), emphasize a more biotic element of the picture than

would those who are interested in physiographic classifications. The CWCS document

views habitats chiefly as the essential element by which animals of special concern, or, as

the document writers term them, species of greatest conservation need (SGCN) can be

preserved in the wild. The habitat maps are numerous and, given their mission, designed

to cast a broad net and overlap each other, a feature not shared by other maps here. In their depiction of the Jackson Prairie Region, the authors of the CWCS simply borrow the shape of the Jackson Rock Group from the geological map of Mississippi (Figure 2.14).

For the CWCS, it doesn’t matter how much loess overlays the western counties or how much sand is deposited on the southern edge: if the geological map says Jackson Group, the potential for Jackson Prairie habitat exists in that area.

38

Figure 2.14

Jackson Prairie Habitat as designated in the CWCS published by The Mississippi Museum of Natural Science (2005)

2.4 General Land Office (GLO) Records

2.4.1 Historical Background

The General Land Office (GLO) and the survey system of township and range was established and refined through a series of acts of the United States Congress in the early part of the nineteenth century (Fitz 1832). The rectangular survey system subdivided the land into township (analogous to latitude, or parallel) and range

(analogous to longitude, or meridian) lines that were theoretically six miles apart. This was an improvement over the English “metes and bounds” system which was based on key landmarks such as wooden posts, large rocks, trees, buildings and the like, since such 39

surveys became invalid as landmarks were altered or destroyed (Bragg 2004, 169). Each

township was divided into 36 sections of one square mile each. Each section was divided into 160-acre quarter sections, which in turn were divided into 40-acre quarters. These parcels were the basic units of homestead purchases.

GLO Surveys of the state of Mississippi were for the most part completed in phases immediately following land cession treaties with the Choctaw and Chickasaw

Indian tribes. Between 1786 and 1830, nine treaties were signed between the United

States government and the Choctaw Nation. Four of these, Fort Adams 1801, Mount

Dexter 1805, Doak’s Stand 1820 and Dancing Rabbit Creek 1830, were land cessions

involving areas of what is now the state of Mississippi (Mississippi Band of Choctaw

Indians 2010). The Treaty of Dancing Rabbit Creek resulted in the removal of greater

part of the Choctaw Nation from the state and their relocation in Oklahoma. The Jackson

Prairie Region occupies parts of three separate treaty areas, Mount Dexter 1805, Doak’s

Stand 1820 and Dancing Rabbit Creek 1830, and so was surveyed in three separate

phases (Figure 2.15).

40

Figure 2.15

Indian land cessions and the Jackson Rock Formation

41 The map used in Figure 2.15 was adapted by the author to include the Jackson

Rock Formation polygon (green) derived from a geologic map obtained from MARIS

(1994). The Formation traverses land ceded in the treaties of Doak’s Stand (1820, pink),

Dancing Rabbit Creek (1836, blue), and Mount Dexter (1805, yellow). Knowledge of these dates facilitates the locating of the correct survey manuscript among the set. The

original map was obtained from TNGenWeb Project, an online source for genealogical

research which contains maps of Indian land cessions for seven southeastern states.

According to the TNGenWeb Project, the maps are from the original book Indian Land

Cessions in the United States, Compiled by Charles C. Royce. It was published in The

18th Annual Report of the Bureau of American Ethnology -- 1896-’97, Vol II,

Smithsonian Institution, printed by the Government Printing Office, 1899.

2.4.2 The Use of GLO Records in Research

2.4.2.1 Introduction

The GLO surveyors have been viewed as early pioneers. “With the fur traders, the

miners and the early settlers, they formed the vanguard of a civilization” (Agnew 1941,

369). In addition to preparing plat maps for settlers, the deputy surveyors were collecting

geographical data about the newly acquired land. Gideon Fitz (1832), Surveyor-General,

instructs the deputy surveyors of Mississippi to

…notice the shape of water courses, ponds, prairies &c as he goes along the lines, and in passing through the sections, by which he may be enabled to lay down in the maps, a good eye draft of all those objects with the connections of roads and paths. He should have the form of a Township, with the square sections in his field book, and while on the ground, represent as well as he can judge by

42 the eye, on such maps, all the objects above mentioned, and the windings of swamps, or low land along the water courses of note (16).

The limitations and biases of GLO records are discussed by Bragg (2002) and

Bourdo (1956). Bourdo (1956) emphasizes that vegetative data derived from the recorded size and spatial dispersion of witness trees needs to be interpreted according to instructions to surveyors regarding which trees to select and blaze, and that these instructions changed over time. Bourdo (1956) acknowledges that some surveys in the

Michigan GLO records were found to be the fraudulent product of surveyors rushing to meet a deadline, but that the errors were corrected over time. Despite these limitations and biases, GLO records represent a relatively reliable record of land cover at the time of

European contact (Bragg 2002, Bourdo 1956, Whitney 1986). Since this thesis does not

use vegetative data, focusing instead on point data records of prairie margins, the bias

discussed in much of the literature (the selection of witness trees by the individual

surveyor) does not apply.

There are many examples of researchers using GLO data in order to understand

landscape features at the time of European settlement. Researchers in central Iowa used

them to obtain landscape-level information on tree canopy composition, providing a

context and framework for inferences made while compiling reference data for restoring

ecologically rare tallgrass oak savannas (Asbjornsen et al. 2005). Researchers have used

GLO records to determine the composition and disturbance pattern of pine forests in

northern lower Michigan, and the results were correlated with soils that produce

vegetation highly susceptible to fire (Whitney 1986). Alice Simms Jones and E. Gibbs

Patton created a method for using GLO records to determine tree density to argue for the

43 existence of a “natural prairie” in Sumter County, Alabama in response to Erhard

Rostlund, who had insisted that the Alabama Black Belt was not a unique and distinctive vegetation zone (Jones and Patton, 1966; Rostlund 1957). Don C. Bragg (2002, 2004) writes extensively about how the records were produced, and derives from them information about settlement patterns and prairie patch locations in Ashley County,

Arkansas. GLO data has been combined with archeological evidence to determine forest conditions in the Tombigbee Natural Forest in the absence of human occupation (Peacock et al. 2008).

2.4.2.2 Barone

John A. Barone (2005a, 2005b) used plat maps created by GLO surveyors to examine the presence and distribution of blackland prairies in Mississippi and Alabama.

Writing in the Journal of the Mississippi Academy of Sciences, Barone (2005a) describes a process of examining GLO plat maps from microfilm records at the Mississippi

Department of Archives and History and paper copies from the Office of The Secretary of State. Those records that showed prairies were scanned to create JPEG files, which in turn were georeferenced using the GIS software program ArcMap, resulting in 265 polygons totaling 19,555 hectares. This result is only for the Jackson Belt and the alluvial plain of western Mississippi. It does not include the northeastern prairies of the Black

Belt. Barone acknowledges some inaccuracies in the GLO plat maps, such as prairies documented by other sources that do not appear in the GLO records, or polygons that are cut off by a section line that do not appear in the adjacent section even though they should.

44 Barone published his digitized plat maps of the northeastern, or Black Belt, prairies in the journal Castanea (Barone 2005b). This article includes an impressive amount of historical observations of Black Belt prairie and concentrates on Rostlund

(1957), demonstrating that he misrepresented a lot of his source material to make it seem that the prairie region consisted of sparsely scattered patches and the trees were dominant. Concerning the utility of GLO-generated data, Barone mentions some of the same accuracy issues noted in the Mississippi Academy of Sciences article and attributes these to the fact that “some teams of surveyors created plat maps that were highly detailed while others recorded very few features of the landscape, apart from rivers and streams” (179). Elsewhere he concludes “because of inattentiveness by some teams of surveyors, many prairies, especially smaller ones, are unrecorded and likely lost to history” (180). By no means are the GLO records a definitive inventory of pre-European settlement prairie patches, in fact, we are sure that it is not.

The difference between Barone’s method for georeferencing GLO records of prairie patches and the method used in this thesis involves the source of the data. Features on the land were noted by the surveyors according to distances from section corners.

These notations amount to x and y coordinates in the township and range grid pattern, and the surveyors used these notes to generate the plat maps. The method employed by this thesis involves input of the coordinates from the field notes into the GIS software program ArcMap. Barone’s method involves importing JPEG files, or plat map images, into a GIS one section at a time. These image files are then aggregated into a larger map.

When the ArcMap vector polygon file created by Barone was requested for the purpose of comparison with the line files generated from the notes, Barone graciously

45 shared it with the writer of this thesis. Three main issues are evident when Barone’s polygons are compared with the vector lines generated for this thesis using field notes.

First, somewhere in the process of creating a large map (twenty-two counties including both prairie belts and the alluvial outliers) from a collection of one-mile-square plat maps, some spatial accuracy was lost. This can be discerned by the fact that rectilinear lines in an otherwise irregular polygon indicate an instance in which a prairie extending across section lines was recorded in one section but not in another. In such cases, the rectilinear edges should align with section lines, but they do not, indicating spatial inaccuracy. Second, not every prairie patch recorded in the field notes was included in the plat maps, resulting in the appearance of a number of patches in this thesis that are not included in Barone’s set. Third, Barone had access to township surveys that were not available or not found in the set of surveys used for this thesis, resulting in the appearance of a number of patches in Barone’s set that are not included in this thesis.

Generally speaking, however, a comprehensive correspondence in the placement and distribution patterns of the two sets is discernable (Figures 2.16, 2.17).

46

Figure 2.16

Location and distribution of GLO prairie patches in the Jackson Prairie Region as published by Barone using plat maps and as generated for this thesis using field notes

47

Figure 2.17

Detail of location and distribution of GLO prairie patches in the Jackson Prairie Region as published by Barone using plat maps and as generated for this thesis using field notes

2.5 Suitability Maps

One objective of this study is to develop a process by which the prairie patches found in the GLO records can be ranked in terms of suitability for conservation based on stakeholder input. What follows is a quick review of the literature on suitability mapping and multicriteria decision making that forms the theoretical basis for the analysis.

The use of map overlays, sometimes called sieve mapping, can be traced back to the nineteenth century and the work of Charles Eliot, Warren Manning and the firm of

48 Olmsted, Olmsted and Elliot. While developing a plan for the town of Billerica,

Massachusetts, the team created “sun maps” by drawing the different map layers, hydrology, soils, vegetation and topography, on cloth and mounting them in a window.

The process is described in some detail along with a survey of the development of map overlays for suitability in the twentieth century in To Heal The Earth: Selected Writings of Ian McHarg (1998) by Ian L. McHarg and Frederick R. Steiner (203-206).

2.5.1 McHarg

Ian L. McHarg (1920-2001) is often credited with formulating the theoretical foundations and basic processes used in the creation of modern suitability maps. Born in

Scotland, McHarg received professional degrees in both landscape architecture and city planning from Harvard University. He taught landscape architecture at the University of

Pennsylvania, where his popular course “Man and Environment” influenced the environmental movement of the 1960’s and 1970s. Frederick Steiner, dean and chair of the school of Architecture, University of Texas at Austin, asserts that McHarg’s book

Design with Nature, published in 1969, “not only changed design and planning, but influenced fields as diverse as geography and engineering, forestry and environmental ethics, soils science and ecology…Environmental impact assessment, new community development, coastal zone management, brownfields restoration, zoo design, river corridor planning, and ideas about sustainability and regenerative design all display the influence of Design with Nature” (Steiner 2004).

McHarg’s method, used in a case study in Design with Nature to propose the alignment of the Richmond Parkway on Staten Island, involves stacking a series of

49 transparent maps on a light table. Individual physical constraints, such as slope, surface

drainage, bedrock foundation, etc, were represented by gray areas on the maps. Areas of

high value, such as water value, forest value, recreation value, etc., were also colored

gray. When all the maps were complete and stacked on a light table, the area of lightest

tone was determined as the proper alignment route. The method was considered a

departure from the traditional cost-benefit analysis because of the broadness of its scope:

McHarg argued effectively for a comprehensive and complex set of “values” inherent in

ecological processes that planners and highway engineers had never considered:

The method proposed here is an attempt to remedy deficiencies in route- selection method. It consists, in essence, of identifying both social and natural processes as social values. We will agree that land and building values do reflect a price value system, we can also agree that for institutions that have no market value there is still a hierarchy in values…There are comparative measures of water quantity and quality, soil drainage characteristics. It is possible to rank forest or marsh quality, in terms of species, numbers, age and health in order of value. Wildlife habitats, scenic quality, the importance of historic buildings, recreational facilities can all be ranked…[A]ny alignment that transects areas of high social values and also incurs penalties in heightened construction costs will represent a maximum social cost solution. The alternative is always to be sought – an alignment that avoids areas of high social costs and incurs the least penalties in construction costs and creates new values. The basis of the method is constant for all case studies – that nature is interacting process, a seamless web, that it is responsive to laws, that it constitutes a value system with intrinsic opportunities and constraints to human use (McHarg 1992, 33-34).

Other case studies presented in Design with Nature to demonstrate the execution of the method include an optimum land use plan for a suburb of Baltimore called The Valleys, suitability maps for conservation, recreation and urbanization for

Staten Island, and a land use suitability map for the entire Potomac River Basin.

50 2.5.2 Hopkins

As the McHargian method of stacking maps of physical constraints and social values gained acceptance it was refined in response to those who pointed out lapses in the validity of some of the assumptions involved in the evaluation processes. Hopkins (1977) argues that McHarg’s method essentially amounts to an addition equation: evaluating relative shades of gray constitutes summing values from the individual maps in the stack, a process Hopkins terms “ordinal combination.” The process is mathematically invalid because it assumes that all the numbers in the equation are on an interval scale and that the distance (intervals) between the various ranks are equal. Furthermore, since the operation of overlaying maps, or ordinal combination, is equivalent to addition, an assumption that the ratings of each factor are independent is implied (389). If the relative suitability of one factor is dependent on another, as can be the case sometimes between soil type and slope for example, simply aggregating all the values fails to account for this relationship. Therefore, “ordinal combination is not a good method for generating suitability maps because of the implied addition of ordinal scale numbers and because of the implied independence of factors” (Hopkins 1977, 390).

Hopkins (1977) outlines a set of procedures to address these problems. His linear combination method solves the problem of unequal intervals by multiplying each factor by a weight of importance, or a proportional score, resulting in output in a single measurement scale with a common unit. His nonlinear combination method solves the problem of assumed independence by analyzing all costs and impacts for all factors combined, an analytical rather than mathematical process, the output of which could be expressed in each homogeneous area created by the overlay. The problem with the

51 nonlinear method is that the costs and impacts of these interactions are generally not known. His factor combination method is a similar way to get around the independence of factors problem: combine all the factors before the calculation is made, resulting in a composite map of areas that are homogeneous relative to all factors. Then a suitability evaluation can be made for each area relative to each specific combination of factors. Of course, this method is characterized by a loss of efficiency and is best used for studies involving only a few factors. Hopkins points out that factor combination is the method used by McHarg in his plan for The Valleys: two factors, forest cover and topography, were used to generate five combinations: valley floors, unforested valley walls, forested valley walls, forested plateau, and unforested plateau. Forest cover apparently did not apply to any valley floor areas in the study.

Another method proposed by Hopkins (1977) is cluster analysis, which “explicitly identifies homogeneous regions by successively pairing the most similar sites or groups of sites, based on an index of similarity across a set of factors.” Cluster analysis is similar to the factor combination method in that at some point an aggregate rating system based on implicit judgment (rather than mathematical processes) must be applied. One intriguing potential of the method is that it allows for measures of variation in factors within clusters as a measure of suitability. “Diversity within a homogeneous region is often a more useful measure of suitability for particular uses than a modal type or specified range,” asserts Hopkins. “For example, a region with diverse slopes makes a good site for a planned unit development; a region with all flat land or all steep slopes does not” (394).

52 Hopkins (1977) proposes a “rules of combination” method, which is very similar

to the factor combination method, the difference being that rules of combination uses

verbally expressed explicit rules about particular factor combinations, rather than implicit

judgments regarding each possible combination. Hopkins (1977) points out that, while

most people associate McHarg’s method with the ordinal combination method, “many of his studies are more accurately described in terms of rules of combination.” This method can be taken a step farther: once the combinations of types from each subset of strongly interdependent factors are rated as combinations, these combinations can be ranked, with each lower order combination treated as an integral whole. This is referred to as hierarchical combination. Both methods, rules of combination and hierarchical combination, have the advantages of handling independence of variables, explicit identification of regions and explicit determination of rankings (396).

Hopkins (1977) demonstrates the invalidity and insufficiencies of the three mathematical combination methods. The ordinal combination method is invalid because of its assumptions and its inability to handle interdependence. The linear combination method should only be used when the variables in question can be determined to be truly independent. The nonlinear combination is problematic because the full range of costs and impacts for the mathematical combinations is usually not known. Factor combination and cluster analysis can be used in certain situations, but these methods do not compare favorably with other methods in terms of costs and benefits. Hopkins concludes that “for most studies, the best approach is to use the linear and nonlinear combination methods as a first stage, followed by rules of combination” (397).

53 2.5.3 Carr and Zwick

Smart Land-Use Analysis: The LUCIS Model (2007) by Margaret H. Carr and

Paul D. Zwick is a case study of a land-use suitability mapping project involving nine counties in central Florida. The book is intended to present a defensible land-use analysis method to the GIS community and also to serve as a textbook for an advanced-level GIS class. The Land Use Conflict Identification Strategy (LUCIS) method analyzes land in terms of three uses: urban, agriculture and conservation, and seeks to identify regions where these uses may come in conflict. The method could be run using any GIS-based software but is presented and described through the use of ModelBuilder, a graphic programming environment embedded within ArcGIS in which the automatic and sequential execution of a series of selected operations can be implemented.

Carr and Zwick (2007) summarize the problems that arise when an analyst simply aggregates or sums a set of values in a stack of maps, namely that datasets that don’t share a common unit of measurement, or datasets wherein independence between variables is erroneously assumed, lead to incorrect results. Carr and Zwick’s presentation of and solution to the problem is quite a bit simpler than that of Hopkins (1977) and amounts to basically making sure that all datasets used are of the appropriate level of measurement. Carr and Zwick (2007) deal with independence of variables through the application of a weighted overlay, or rank of importance factor, on the features in the datasets and on the datasets themselves. Their data analysis methods are adapted from the textbook Exploring Geographic Information Systems by Nicolas Chrisman (2002).

Following is a summary of Carr and Zwick’s treatment of Chrisman’s four rules.

54 The enumeration method preserves the values from the input layers in a table, the

result being a clear and distinct set of unique attribute combinations. One drawback to the

method is that the number of combinations can become quite large. The method is useful because structured query language (SQL) can be used to create a selection of feature outputs based on specific input combinations.

The dominance rule involves the selection of a single value that is preferred over all other values. The selection is defined by external rules rather than by combinations of values. The simplest example is exclusionary screening, also known as sieve mapping, where the datasets are queried and the result is a binary output where 1 = true (included) and 0 = not true (excluded). Exclusionary screening requires the use of the OR operator in a query. For example, if a given cell in grid one is greater than seven, OR of the coincident cell in grid two equals four, OR if the coincident cell in grid three equals six, then select, otherwise exclude. Another example of the dominance rule is exclusionary ranking. The datasets must first be ranked according to a common value scheme, usually

1 to 5 or 1 to 9. The input datasets are then summed and the exclusionary rule applies to the cells that did or did not achieve a predetermined value threshold. This only works on truly independent inputs. A third example of the dominance rule is exclusionary selection, where the input datasets are not ranked but remain as measured values. An example of this would be a map of distance to work combined with a map of cost of rent.

The two values are summed and any cell values above a certain threshold are deemed undesirable.

The contributory rule refers to a set of operations in which values from one input contributes to the output results without regard to values from other inputs. With nominal

55 data, the values can be simply counted, generating simple statistics such as the mode and variety. This operation is called voting tabulation and differs from the dominance rule in that the statistic or value is determined from the internal values of the inputs and not from an external rule. Obviously the applications of this operation are limited because it assumes equal importance among the inputs. An example would be an output cell generated by ten layers of input data: if seven of the ten inputs are ranked suitable and three unsuitable, the result would be misleading if any one of the three unsuitable values are associated with relatively important variables. Carr and Zwick (2007) explain that

Chrisman gets around this problem with a weighted voted operation, in which the features in the inputs are weighted by importance before the tabulation. Since the importance ranking is ordinal, there is no way to determine how much more important one feature is than another. There is no guarantee that a very important ranking is five, seven or ten times more important than a less important one. Because of this, Carr and

Zwick (2007) explain that Chrisman recommends using weighted voting only to identify the best and worst (highest and lowest) factors.

The LUCIS method employs different rules for combining spatial data depending on the type of data in the input sets, but Chrisman’s fourth rule, the interaction rule, is applied most often in the case study due to the fact that, unlike the contributory rule, it does the best job of considering interactions between factors. The interaction rule accomplishes this through the use of a weighted overlay system. Each input dataset is assigned utility factors within it, based on an organized hierarchy of criteria, which in turn are derived from a stated set of goals and objectives. Once the utility factors are assigned, the raster layer is referred to as a single utility assignment (SUA). Multiple

56 SUAs are combined, again using a weighted overlay, to create a simple multiple utility

assignment (MUA). These MUAs are combined, again using weighted overlay, to create

the final map surface. Following this description of the interaction rule, Carr and Zwick

(2007) present a series of methods for determining the utility factors: modeler

assignment, group voting, modified delphi process and pairwise comparison. With the exception of the modeler assignment, where the modeler simply assigns the utility values, these methods all involve some degree of stakeholder or outside expert input.

The case study that makes up the greater part of Smart Land Use Analysis (2007), the creation of development, agriculture and conservation land use suitability maps for nine counties in north-central Florida, consists of the establishment of a detailed set of goals and objectives, the creation of work-flow charts and a series of steps by which input maps were processed and transformed into SUAs and MUAs in an ArcGIS and

ModelBuilder environment according to the rules for combining spatial data as outlined above.

In conclusion, Carr and Zwick (2007) are not as concerned as Hopkins (1977) with the specific details of the issues involving mathematical validity in land suitability analysis. Carr and Zwick (2007) stress that the modeler must be aware of validity issues, but levels of measurement conflicts can essentially be resolved through reclassification: interval data can be transformed to nominal data by grouping operations for example.

Nominal data can not strictly be transformed into interval data, but it can at least be ranked in terms of suitability. The validity issues associated with assumed independence of variables can be dealt with by such operations as the weighted overlay. The LUCIS method, then, does not “solve” the validity problem, but it works around them in a

57 defensible way, and, most importantly, through the use of tools available in the ESRI software suite, specifically ArcGIS and ModelBuilder.

2.5.4 Malczewski

GIS and Multicriteria Decision Making (1999) by Jacek Malczewski is a thorough and systematic guide to using raster data in a GIS. Malczewski covers much of

the same ground as Hopkins (1977) and Carr and Zwick (2007). Unlike Carr and Zwick,

Malczewski’s work is not designed around a case study or a particular software package,

rather it concentrates on specific mathematical and conceptual aspects of raster data and

how such data can be transformed into information and used in making decisions or

determining suitability based on criteria.

Malczewski (1999) provides an explanation of how GIS works that complements

the literature discussed here. His terminology is different but the larger pattern of his

system parallels that of Carr and Zwick (2007) in that it involves establishing goals and

objectives, selecting evaluation criteria, determining weights and applying decision rules.

Malcewski provides a method for standardizing scores called the maximum score

procedure (117) whereby all scores are divided by the maximum score, resulting in a set of values that retain the relative order of magnitude of the original scores. For example, the scores 3, 15 and 26 can be standardized by dividing each score by 26, obtaining standardized scores of 0.115, 0.577 and 1.0 respectively. The maximum score will always be 1.0 in this system. The relative order of magnitude is unchanged: 3/15 = 0.115/0.577,

3/26 = 0.115/1 and 15/26 = 0.577/1. This linear scale transformation of scores was used in this study.

58 Malczewski’s (1999) Simple Additive Weighting Method (199) is essentially the same as the map algebra process described by Carr and Zwick (2007). The criteria (or map layers) are defined and assigned weights of relative importance. Each layer is multiplied by its corresponding weight, and the products are all summed resulting in a set of overall performance scores. The cells (Malcewski calls them “alternatives”) with the highest performance scores are determined to be most suitable. This is the process used in this study.

59

CHAPTER III

METHODS

3.1 Format of Surveys Used for This Study

Issues regarding the validity of GLO records in general are discussed in section

2.4.2.1 of the Literature Review. Mississippi State University has obtained scanned copies of the hand-written GLO surveys and field notes. They are formatted as over sixty thousand individual .tif files recorded on fifteen compact discs. One disc also contains an fdn viewer, a piece of software that contains an index of townships categorized by districts and listed by file numbers. Issues pertaining to the legibility and the organization of the set of files caused the author to decide to examine the files in their entirety, with help from the fdn viewer software, his thesis advisor and a fellow student, taking notes along the way, searching for those that were in the study area. Since, as discussed in section 2.4.1, the surveyors used different baseline meridians following different Indian treaties, resulting in the numbers for the townships repeating in different parts of the state, it was sometimes unclear at first whether a particular township noted in the manuscript is in the study area or not. By learning the dates of the treaties and looking for corresponding dates on the surveys a determination was made more quickly (Figure

2.15). The Bureau of Land Management has since solved the problem of duplicate townships by putting numerals in front of the original number. For example, township 5

60

north range 2 east in the Choctaw district is now called 25N2E. Township 5 north range

2 east in the Washington district is now called 55N2E. In the handwritten manuscripts, both surveys are referred to as T5NR2E. T5NR2E in the Choctaw district, surveyed in

1821, is in the study area, whereas T5NR2E in the Washington district, surveyed in 1802, is not.

3.2 Establishing the Study Area

The ambiguity regarding the western extent of the Jackson Prairie Region is discussed in subsection 2.3.3 of the literature review. The region as indicated by the

MARIS (1994) map was deemed too restrictive in light of the historical record. The researcher opted for a more inclusive set of townships, and used the geologic extent of the Jackson Rock Group (Figure 2.6) and the map of the Jackson Prairie created by

Merril, et al (1985, Figure 2.8), as guides. One hundred and twenty-nine townships were selected for the study area (Figure 3.1).

61

Figure 3.1

The one hundred and twenty-nine townships of the study area

The survey notes for 104 of these 129 townships were found in the set of discs. Of these, fifty-four had some mention of prairie in the survey (Figure 3.2).

62

Figure 3.2

The 25 missing surveys, 104 found surveys and the 54 which mention prairie

3.3 Transcribing the Data

Surveyors marked township boundaries, section lines and quarter section points by dragging lengths of chain across the land, blazing trees, called “witness trees” or bearing trees,” and setting posts. Margins of prairies, and branches of streams, rivers and swamps were noted by their distance, in chains, from the corner of origin. In order to georeference these features, four elements are necessary: the corner of origin, the direction in which the surveyor is travelling, the distance to the margin measured, and whether the chain used for the measurement is 33 or 66 feet long. Each chain has 100 links. If the shorter chain is used, one mile equals 160 chains. Most of the surveys used in 63

the study were created with a 66 foot chain, causing one mile to be measured at 80

chains. Although each section boundary is considered to be one mile long, often the measurements were a number of links more or less than 80. In the example given here

(Figure 3.3), the line between sections 2 and 11 is measured at 79 chains, 85 links, or 15

links short of 80 chains. The surveyor has measured a mile ten feet shorter than expected

(15 x 0.66 = 9.9).

Figure 3.3

A typical section line survey

The surveyor in this example (Figure 3.3) is moving east on the line between sections 2 and 11 in Township 3 North, Range 12 East. He enters a prairie at 3 chains and exits the prairie at 10 chains. He sets the quarter section post at 40 chains and notes his bearing trees. He enters another prairie at 65 chains and does not exit it. The section

64

corner post is in the prairie, and the distances to the bearing trees are noticeably greater

(52 and 130 links, or 34 and 86 feet) than they were for the quarter section post (32 and

62 links, or 21 and 42 feet), which was not in prairie. The bearing tree information is

interesting, but was not included in the transcription. Recording bearing tree data is time-

consuming and ultimately unnecessary for the generation of prairie margin locations in

the GIS. The identifying file number for the survey page was included in the

transcription, so the file could be referenced later if the data was questionable. In some

cases, mistakes were made: the section number 2 in this example was originally

transcribed as 3, and the error was caught when the GIS was unable to generate the line

due to the fact that sections 3 and 11 do not share a boundary. The notes on land cover

were also included in the transcription, even though they were not used in the GIS. The

transcription for the survey in Figure 3.3 is shown in Table 3.1.

Table 3.1

Transcription of GLO data in figure 3.3

btw total chains in page # direction sections enter p exit p mile comments 45752 E 2,11 3 10 80 Land undula ting very open Post Oak and Black Jack woods land 45752 E 2,11 65 80 80 clay soil

The transcription also included notations of section or township lines where the prairies were too small and numerous to measure or the habitat graded continuously from prairie to woodland. These lines were near other observed prairies and the notes included

65

observations such as “this line thinly timbered with scrubby oak.” or “interspersed with prairies.”

Also included in the transcription are surveys of townships in the study area that did not contain references to prairie, with the notation that they did not, as well as surveys in the township area that were deemed unavailable after significant efforts were made to locate them in the files. In this way, all the townships of the study area were accounted for in the transcription. The transcription is included as an addendum to this thesis.

3.4 Inserting the GLO Data into a GIS

The ArcMap tool by which the transcribed data could be used to generate

“distance to” and “distance from” coordinates in ArcGIS was developed by Kate Grala, a researcher in the Department of Geosciences at Mississippi State University. In order to create the lines representing prairie, the generation of points of entry and exit on section lines was necessary. Each set of entry and exit points on a section line required a discrete point of origin. Essentially each section corner in the study area had to be designated as a point with a unique identifier so that it could act as a point of origin. This was achieved by using the standard deviation of the four section numbers associated with the particular corner, in combination with the identifying township and range information. This process of point generation can create interior section corners only. Since section corners along township boundary lines are shared between townships, they were processed differently.

In order to generate these points, it was necessary to consider each township as possessing only northern and eastern boundaries: southern and western boundaries were given away to adjacent townships. The tool then generated points on township northern

66

and eastern boundaries by selecting the nearest interior section corner and moving one mile north or east as required.

Once the points of origin were generated in this way, entrance and exit measurements were converted to “distance to” and “distance from” values. Chain and link measurements were converted to meters. The cardinal direction of the surveyor’s path was included in the operation and vector lines were generated according to these directions and distances. The lines were checked for anomalies. Those that had been transcribed improperly generally appeared well outside the study area. At this stage it was critically important that the file numbers of the scanned survey pages were properly recorded. In all, less than a dozen transcribed notations required correction. The corrected data was then processed by the model, resulting in a final set of 323 vector lines (Figure

3.4).

67

Figure 3.4

Vector lines created in ArcGIS from the GLO data

3.5 Establishing Criteria for Suitability

The goal of ranking the GLO prairies in terms of suitability for restoration or conservation was achieved by analyzing a set of features and conditions at those same locations. The first objective was to select a number of datasets, the “stack of maps” that is the basis of the McHargian suitability analysis. The second objective was to establish a set of weights, scores and mathematical operations to apply to the data in order to execute the analysis.

68

3.5.1 Selecting the Datasets

Stakeholder input was sought in order to establish which datasets would be selected and evaluated in the creation of the suitability map. Stacey Shankle, Director of

Conservation for the Nature Conservancy of Mississippi, and Dr. Timothy Schauwecker, assistant professor of Landscape Architecture at Mississippi State University, were consulted regarding the identification of potential stakeholders. Mr. Shankle assembled a group of people representing organizations which The Nature Conservancy of Mississippi considers partners in conservation. A meeting with the stakeholders was held in January of 2009 to discuss the selection of datasets. In addition to Mr. Shankle and the thesis author, the attendees of that meeting were:

Glenda Clardy, biologist with Natural Resource Conservation Service. Ken Gordon, botanist with the United States Forest Service Rick Hamrick, biologist with Mississippi Department of Wildlife, Fisheries and Parks Heather Sullivan, botanist with The Mississippi Natural Heritage Program Sherry Surrette, Mississippi Natural Heritage Program Coordinator Lisa Yager, Research and Collections Coordinator, Mississippi Museum of Natural Science.

It was broadly concluded at this meeting that the best candidates for restoration are lands that are not developed and are free from the threat of development. Public land was for this reason considered a priority. The potential of The Environmental Quality

Incentive Program (EQIP), Wildlife Habitat Incentives Program (WHIP), and other Farm

Bill conservation programs as funding mechanisms was discussed. The relative ease of restoring a forested site, which would have fewer invasive exotics present than an agricultural site versus the potential of agricultural sites, which are the targets of the funding mechanisms mentioned above, was also discussed. Of agricultural uses, pasture

69

was agreed to be the easiest to restore, with native prairie indicator species likely present

along with the exotics. Patch density, species dispersal distances and the notion that the identification of a species of concern that utilized prairie habitat could be used to prioritize restoration of patches identified as clusters were also considered as possible features the stakeholders would have liked to have represented in the datasets.

Another criteria in the selection of datasets, given the limited scope of the project,

was that the data be publicly available, complete and GIS-ready. Eight variables were

selected (Table 3.2). Six of them required acquisition of outside datasets and two

obtained information about the patches through ArcGIS Spatial Analyst operations.

Table 3.2

Variables Selected for Suitability Analysis

Variable File File Type 1 State Park Land State Parks Vector 2 16th Secti on Land Town ship sections Vector 3 National Forest Land National Forest owned land Vector 4 Transmission Lines Transmission lines Vector 5 Land Cover Land Cover Raster 6 Distance to Primary Roads Primary Roads Vector 7 Kernel Density Spatial Analyst Raster 8 Patch Size Spatial Analyst Raster

3.5.2 Rationales for Selections

Following is a brief description of why each of the six datasets and the two

Spatial Analyst operations could be considered relevant to the project of natural

blackland prairie restoration and conservation.

70

3.5.2.1 State Park Land

State parks are one of the three categories of publicly owned land present in the study area. Public lands were discussed in the stakeholder meeting. Public lands are of interest to stakeholders because they are already under management directed by an identifiable entity. The rationale assumes that state park managers would be receptive to conservation efforts if made aware of the historic presence of a natural prairie.

3.5.2.2 Sixteenth Section Land

The Mississippi State Constitution decrees that all sixteenth section lands be used to generate revenues for the county school systems. County board of education members may seem an unlikely partner for prairie conservation, but a precedent does exist in

Oktibbeha County, Mississippi, at the community of Osborn, near the town of Starkville, for cooperation between conservationists and board of education members that features a public education component to restoration efforts. In the late nineties a group of local high school teachers became interested in a relict blackland prairie on 16th section land as

a field trip destination with educational potential. Eventually faculty members at

Mississippi State University were included in the creation of a management plan for the

site (Wiygul et al. 2003). Management efforts, consisting mostly of suppression of

Juniperis virginiana by volunteers, are ongoing. The group of volunteers holds a 99-year

lease on the 140 acre site (Victor Maddox, personal communication). The rationale

assumes that 16th section land has been subject to less development historically and that

county school boards in the study area would be receptive to conservation efforts in the

same way as those in Oktibbeha County have been.

71

3.5.2.3 National Forest Land

As mentioned in section 2.3 of the Literature Review, Harrell Prairie Hill was established in the Bienville National Forest as a managed native prairie in 1971. Harrell

Prairie Hill was the only managed prairie in the Jackson Prairie Region until 2008, when the United States Forest Service (USFS) began management of six other sites. As mentioned in section 2.3.2 of the Literature Review, botanists working for the USFS have been developing an inventory of native prairie patches since the 1980s (Gordon and

Wiseman 1989). The database of the inventory is managed by the Mississippi Natural

Heritage Program, (MNHP), which provided a copy, in the form of an ArcMap file, containing a total of seventy-four points, for this study. The spatial relationship between

USFS land, the MNHP inventory points and the GLO lines derived from the survey notes is shown in Figure 3.5.

72

Figure 3.5

GLO Prairie, MNHP Jackson Prairie inventory points and National Forest Land.

Based on comments made at the stakeholder meeting, The Nature Conservancy of

Mississippi considers the USFS to be a strong and active partner in conservation. Ken

Gordon, the Forest Service botanist who co-authored the 1989 report on the prairie relict inventory, participated in the stakeholder meeting and the survey. The rationale assumes that stakeholders would be able to take advantage of the interest in conservation already

demonstrated by the USFS.

73

3.5.2.4 Transmission Lines

Transmission lines indicate long-term ownership and a management pattern that

suppresses both development and woody vegetation. Transmission lines have played an

important role in the sixteenth section prairie near Osborne, Mississippi mentioned above.

3.5.2.5 Land Cover

Any management goal must consider existing land cover. Comments in the stakeholder meeting indicated that agricultural uses and forests are associated with high suitability for restoration and conservation, while developed areas are associated with low suitability.

3.5.2.6 Distance to Primary Roads

The rationale for this variable assumes that prairie is a natural habitat that is more likely to persist in areas with less human activity and disturbance. It also assumes that human activity and disturbance decreases as distance to primary roads increase.

Secondary roads and county roads were not considered, since they are distributed fairly evenly over the area. The variable is meant to act as a proxy measurement of

“remoteness” in the landscape, and assumes that the more “remote” a spot is, the more suitable it would be for restoration. Since prescribed burns are a common prairie management practice, and smoke from these burns creates a problem for traffic, increased distance to primary roads could be assumed to decrease the difficulty of prescribed burning.

74

3.5.2.7 Kernel Density

The rationale for the density variable is that patches which can be determined to

be part of a cluster of patches may be of greater value to pollinators and may facilitate

species dispersal. A patch that is considered unsuitable by other criteria may nevertheless

be of some value if it is part of a cluster and can therefore facilitate species dispersal.

Conceptually, this variable links the goal of prairie restoration and conservation with the

larger goal of preserving regional biodiversity. The rationale assumes that the Kernel

Density operator in Spatial Analyst is an acceptable tool for measuring the degree to

which any particular line in the set of GLO prairie lines can be considered part of a

cluster of lines. The process for calculating Kernel Density is discussed below in the

section on data preprocessing.

3.5.2.8 Patch Size

Conservationists suggest that biodiversity is greater when patches are large rather

than small (Helmund and Smith 2006, 64). Some species prefer edges while others prefer

interior spaces. Larger patches have more interior habitat and therefore more species

biodiversity, making them more valuable to conservationists.

3.5.3 Establishing Scores and Weights

Scores and weights for variables were established in order to address the problems

of validity of calculations discussed in section 2.5 in the literature review. The methods

of Carr and Zwick (2007) were considered in the effort to establish a common scale of measurement for the eight variables and to express mathematically the relative

75

importance of the variables in the evaluation. A survey of the attendees of the January,

2009 meeting was conducted by email. The stakeholders were asked to rank the eight variables in terms of importance to conservation in such a way that the sum of all ranks

would be one hundred. The point of having the ranks sum to one hundred was so that

stakeholders would be forced to consider relative values of importance. If one variable

were twice as important as another, or equal to five others combined in the stakeholder’s

mind, the process of dividing a limited set of points among the variables was designed to

encourage the stakeholder to consider those ratios. Four responses were received. In one

case, the stakeholder’s weights summed to 55, so the weights were normalized by

multiplying each by 1.818 (100/55 = 1.818). The variables with the weights are shown in

Table 3.3.

Table 3.3

Weight of Importance Values Assigned by Stakeholders

Rounded Variable S1 S2 S3 S4 Averages Averages It is on state park land 10 5 10 14.544 9.886 10 It is on US Forest Service land 25 20 25 18.18 22.045 22 It is on 16th section land 5 5 5 12.726 6.9315 7 Transmission lines are present 5 5 5 14.544 7.386 7 It is far from primary roads 10 15 5 5.454 8.8635 9 Current land use / land cover 20 15 25 9.09 17.2725 17 Patch size 15 15 15 10.908 13.977 14 Proximity to other known or potential patches. 10 20 10 14.544 13.636 14 TOTAL 100 100 100 99.99 99.9975 100

Stakeholders were also asked to give scores of 1 to 10, with 10 being most suitable for conservation, to the classes of land cover in the land cover grid. The results of

those scores are shown in Table 3.4. 76

Table 3.4

Stakeholder Scores for Land Cover Classes

Rounded Land Cover S1 S2 S3 S4 Averages Averages Water 1 1.25 1 1 1.0625 1 Developed: Residential 1 0 1.25 4 1.5625 1 Developed: Commercial 2 0 0 3 1.25 1 Barren Transitional 7 10 6.25 7 7.5625 8 Forest: Deciduous 8 8.75 8.75 6 7.875 8 Forest: Evergreen 10 6.25 3.75 6 6.5 7 Forest: Mixed 8 7.5 6.25 6 6.9375 7 Herbaceous: Pasture 6 8.75 10 10 8.6875 10 Herbaceous: Row Crops 4 7.5 7.5 8 6.75 7 Herbaceous: Urban Recreational 5 5 10 7 6.75 7 Wetlands: Woody 2 0 2.5 0 1.125 1

A score of 1 to 10 was used for each of the eight variables. The first four, state

park lands, 16th section lands, USFS land and the presence of transmission lines, are

binary categories: the condition described either exists at the cell location or it does not.

All four conditions are positively associated with suitability for conservation, so they are

scored as 1 (the condition does not exist at the cell location) or 10 (the condition does

exist). Distance to primary roads, patch size and proximity to other patches are

continuous data variables. Distance to primary roads has a potential zero value and the other two do not, but they are all measured at equal intervals. These three datasets were

transformed into categorical variables by using Jenk’s natural breaks, five classes. Jenk’s

method identifies relatively large gaps in the values by examining all possible groups in

the dataset with the goal of minimizing the variance within groups and maximizing the

variance between groups (ESRI 2004). Once the breaks in the datasets were assigned by

the ArcGIS operator, these sets were reclassified so that the old values of 1,2,3,4 and 5

77

were transformed to new values of 2,4,6,8 and 10. In this way, each variable was measured at the same scale. Theoretically, the maximum output value for any cell in the set after the calculation is 1000: a perfect ten on variable multiplied by the importance weights. In reality such a score is impossible because a positive score for each of the three categories dealing with public lands excludes the possibility of the other two.

3.5.4 Data Preprocessing

The GLO prairie lines were converted from vector to raster at one hundred meter resolution. One hundred meter resolution was deemed reasonable for ease of processing for a study area that includes parts of ten counties. In addition, the Land Cover grid used in the suitability analysis also utilizes 100 meter resolution. The new, raster version of the

GLO prairie line file was named [all_p61]. A field was added to the attribute table in which the value for all the cells was given as (1).

NAD 1983 Transverse Mercator was chosen as the projected coordinate system.

All data sets not already in NAD 1983 Transverse Mercator were reprojected. Following is a description of how each dataset or map layer was processed for the map algebra equation.

3.5.4.1 State Park Land

A shapefile of state parks was obtained from MARIS. Only two state parks exist in the study area, Roosevelt State Park in Morton and LeFleur’s Bluff State Park in

Jackson. The GLO records show a prairie in LeFleur’s Bluff State Park, which is also the

78

site of the Mississippi Museum of Science and the Mississippi Natural Heritage Program, where two of the stakeholders work (Figure 3.6).

Figure 3.6

GLO Prairie (magenta line) in and around LeFleur’s Bluff State Park (green line) in Jackson.

The state park shapefile was used as an extraction mask on [all_p61], resulting in a raster file of this single example. This raster was reclassified with “NoData” as zero and then added to [all_p61] in raster calculator, a tool in the ArcGIS Spatial Analyst suite of tools that allows mathematical operations to be performed on raster grids, resulting in a file with values (1, 2). These values were replaced with (1, 10), and the output was renamed [st_prks].

79

3.5.4.2 Sixteenth Section Land

A shapefile of all sections in Mississippi was obtained from MARIS. The “select by attributes” function was used to select only sixteenth sections, and a new layer was created from the selection. A 100 meter buffer was applied to the sections and used as a mask to extract the cells from [all_p61] which are along sixteenth section lines. The buffer was necessary because the sixteenth section boundary alone does not catch some cells that lie immediately outside the polygon even though they are in fact associated with that section line. Just as with state parks, the extracted cells were then reclassified with

“no data” as zero so they could be added back to [all_p61] in raster calculator, and the new values were reclassified as (1,10) and the output was renamed [sec_16].

3.5.4.3 National Forest Land

A shapefile of National Forest-owned land was obtained from MARIS. The raster for the suitability analysis was created in exactly the same way as the state parks raster, by extracting the cells from [all_p61], reclassifying “no data” as zero, adding them back to [allPp61] and reclassifying the output as (1, 10). The output file was renamed [usfs].

3.5.4.4 Transmission Lines

A polyline shapefile of transmission lines was obtained from MARIS. The preprocessing was exactly the same as that of the sixteenth section lands. A buffer was necessary because the polyline alone was an insufficient mask for the selection of cells not exactly coincidental to the line but close enough to be affected (Figure 3.7).

80

Figure 3.7

ArcMap screenshot showing the necessity of the buffer for the transmission line shapefile.

Figure 3.7 shows that the vector file (red line) for the transmission line is about 50 meters off the center of the power cut in the aerial image. The hundred meter buffer on the transmission line allowed the extraction to select cells along the GLO prairie (green) running parallel to it. If the transmission line without the buffer had been used as an extraction mask, only one of the (light blue) cells it passes through on the north end would have been selected. The GLO prairie vector shapefile (blue line) extends south out of the power cut and out of the influence of the buffer, since the transmission line turns to the southwest.

81

As with the preceding three files, the transmission line cells were reclassified with

“no data” as zero, then added to [all_p61] in raster calculator, and the output was reclassified with values of (1, 10) and renamed [tran_line].

3.5.4.5 Land Cover

A raster file of land cover classes was obtained from MARIS. The file [all_p61] was used as the mask to extract the land cover class values to the prairie lines. The values were then reclassified according to the scores given by the stakeholders and renamed

[recl_ext_nlcd].

3.5.4.6 Distance to Primary Roads

A polyline file of primary roads was obtained from MARIS. In the Spatial

Analyst options window, the analysis mask was set to the study area file and the cell size was set at 100. The straight line distance function was opened in Spatial Analyst and the primary roads file loaded in the window, resulting in an output raster file in which every cell contained a value representing the distance to the nearest primary road from that cell.

The file [all_p61] was used as an extraction mask to obtain a file of prairie lines containing distances to primary roads. This file was classified using Jenk’s natural breaks, five classes, and then reclassified with values (2,4,6,8,10) representing the five classes. The file was then renamed [recl_ pri_rd].

82

3.5.4.7 Kernel Density

Conceptually, the kernel density function in spatial analyst places a smoothly curved surface, or kernel, over each line. Its value is greatest on the line and diminishes by degrees as distance from the line increases, reaching zero at the distance of the search radius from the line. The density is calculated by adding all the values for all the kernels over all the lines. The Kernel density operator was opened in the Spatial Analyst suite of tools. The prairie vector lines were used as the input map and the cell size was set at 100.

The kernel radius can be adjusted. A larger radius tends to display regional variations at the expense of local variation. A smaller radius does the opposite: local variation is displayed but regional variation is lost. A kernel radius of 5000 meters (Figure 3.8) was compared with radii of 10,000 meters (Figure 3.9) and 20,000 meters (Figure 3.10).

Figure 3.8

Kernel density of GLO prairie lines with kernel radius of 5000 meters

83

Figure 3.9

Kernel density of GLO prairie lines with kernel radius of 10,000 meters

Figure 3.10

Kernel density of GLO prairie lines with kernel radius of 20,000 meters

The grid created with the kernel density of 10,000 meters was selected as showing the maximum amount of both local and regional variation. The file [all_p61] was used as a mask to extract the cell values, and the resulting output was classified using Jenk’s

natural breaks, five classes, and then reclassified with values (2,4,6,8,10) representing the

five classes. The file was then renamed [rcl_dens].

84

3.5.4.8 Patch Size

Transects through prairie patches recorded in the nineteenth century do not reveal the actual areas of the patches, but it is reasonable to assume that longer transects are

generally associated with larger areas. In order to join contiguous lines together and

acknowledge lines that run across multiple sections, the raster file was converted to a

vector polygon file and then the aggregate tool was used to combine contiguous

polygons. Once the area for these polygons was calculated, the file was converted to

raster with area as the value field. The raster grid was classified using Jenk’s natural

breaks, five classes, then reclassified with values (2,4,6,8,10) and renamed [recl_area].

3.6 The Map Algebra Equation

Each raster data layer was multiplied by the importance weight derived from the

stakeholder input surveys and then all eight were summed. The equation entered in the

raster calculator is as follows:

([st_prks] * 10) + ([usfs] * 22) + ([sec_16] * 7) + ([tran_line] * 7) + ([recl_pri_rd]

* 9) + ([recl_ext_nlcd] * 17) + ([recl_area] * 14) + ([rcl_dens] * 14)

A Work Flow Chart was created to display the treatment of each variable in the

ArcGIS environment (Figure 3.11).

85

Figure 3.11

Workflow in ArcGIS/ArcMap

86

CHAPTER IV

RESULTS AND DISCUSSION

4.1 Introduction

This study is the most comprehensive examination of the Jackson Prairie Region to date. Out of the components of this study, the literature review, the examination of the

GLO records, the mapping of the data, the ranking of patches through map algebra or simple additive weighting by stakeholder input criteria, map algebra output and the ground-truthing visits to the historic locations, some general patterns of observation arise.

4.2 Map Algebra Output

The map resulting from the equation contained cell values, or scores, ranging from 91 to 538. These scores were standardized by dividing all values by 538, producing a top score of 1, a minimum score of 0.17 and a mean score of 0.63 (Figure 4.1)

87

Figure 4.1

Histogram of the standardized scores of the map algebra suitability output.

In order to display the top score GLO prairie lines, the map algebra output raster was reclassified again with the four lower scoring bins given values of “no data” and the

resulting raster was converted to vector lines so it would appear at the scale necessary to

display the entire study area (Figure 4.2). The cells that scored lowest were processed for

display in the same way (Figure 4.3). Finally the study area was divided into eastern,

central and western portion so that the spatial distribution of the full range of the five

classes could be displayed (Figures 4.4, 4.5 and 4.6).

88

Figure 4.2

Map display of the maximum scores of the map algebra suitability output.

89

Figure 4.3

Map display of the minimum scores of the map algebra suitability output.

90

Figure 4.4

Map display of the classes of scores for the western part of the study area.

91

Figure 4.5

Map display of the classes of scores for the central part of the study area.

92

Figure 4.6

Map display of the classes of scores for the eastern part of the study area.

4.3 GLO Prairies East of the Pearl River

Most of the high-scoring cells are in the western part of the study area, in central

Madison County, where, as discussed in the Literature Review, the Jackson Prairie

Physiographic Region grades into the Loess Bluff and the North Central Hills

93

Physiographic Regions. The GLO surveyors recorded some of the largest and most densely-clustered prairies in this region and much of the land is in agricultural use: three variables that were given high weight of importance values in the calculation. Madison

County is intensively developed. Most of the non-agricultural lands are occupied by the subdivision and commercial strip landscape matrix of the suburbs of the state capitol,

Jackson, just to the south in Hinds County. The low scores of the minimum group can be generally attributed to the variables patch size (they are small) and the developed class of lands characteristic of the Jackson urban core.

Site visits to Madison County during the ground-truthing phase of the research

(discussed in section 4.4 below) revealed scant instances of prairie indicators and no examples of intact, natural prairie of any size. In order to gain a better understanding of the suitability range in the counties east of the Pearl River, that is, the Jackson Prairie

Region as designated by MARIS (2004), Hodgkins et al (1976) and Chapman et al

(2004), the cells in Madison and Hinds Counties were removed from the analysis and the scores were reclassified again using Jenk’s natural breaks, five classes. These new scores range from 91 to 432 and were standardized by dividing each by the maximum score. The histogram (Figure 4.7) shows a minimum, mean score and normal distribution similar to that found for the entire study area. The classes with maximum and minimum scores were converted to vector lines and displayed (Figures 4.8 and 4.9). The low scores appear to be associated with the small patch size and developed land use patterns characteristic of western Rankin County, whereas the maximum scores were probably derived from agricultural land use patterns and the presence of the Bienville National Forest in the four

94

central counties. Interestingly, the cluster of prairies in the far eastern end of the region was not represented in either the maximum or minimum score categories.

Figure 4.7

Histogram of the suitability scores for GLO prairies east of the Pearl River.

95

Figure 4.8

Map display of the highest scoring cells east of the Pearl River.

96

Figure 4.9

Map display of the lowest scoring cells east of the Pearl River.

4.4 Ground-truthing and the Resilience of Patches

In order to meet the goal of determining the utility of GLO records for prairie conservation in the Jackson Prairie Region, the actual locations of a number of the GLO prairies were visited and conditions were observed. The results of these observations have implications for the resilience of prairie patches in the Jackson Prairie Region.

Resilience is a term applied with growing frequency in the fields of planning and ecology. It refers to the ability of either a natural or an urban area to return to a state of equilibrium and relative stability after a major disturbance or disruption (Steiner 2006).

Prairies depend on a pattern of disturbance, especially fire, in order to maintain stability 97

and persist in the landscape (Anderson 2006, Axelrod 1985, Sauer 1950). The settlement of the Jackson Prairie Region by homesteaders in the nineteenth century brought an end

to the use of fire by humans as a management tool in the area. This historic fact, along

with the pressures of development, is often cited in the literature as the reason for the

disappearance of natural prairies from the landscape. The ground-truthing phase of the

research indicates that development pressure is high, but that in the absence of

development over half of the historic GLO prairie locations have prairie indicators and a

third are relict prairies.

4.4.1 The Study Area

Fifty-eight sites across the study area were visited and notes and photographs

were taken. Sites were selected based on their accessibility from public roads. Of the ten

counties in the study area, all but Yazoo and Newton Counties were visited. Twenty-nine

of the sites, half those visited, were determined to be so intensively managed, generally

as residences, farms or pine plantations, that any vegetative expression of natural habitat,

including prairie, was suppressed. Of the remaining half, those classed as forested or

weedy areas that appeared to be lightly managed or abandoned, 12 (41%) showed no sign

of prairie and were indistinguishable from the surrounding pine or mixed forest matrix, 8

(28%) contained indicator species, mostly in unmowed road right-of-way margins, and 9

(31%) were recognizable prairie relicts containing indicator species over a large area,

generally a half-acre or more. Four of these relicts were on the MNHP inventory list, so

the other five can be considered discoveries of new patches. This information is presented

in Table 4.1.

98

Table 4.1

Classes of Prairie Patch Resilience Observed during Ground-truthing

Abandoned or lightly maintained Intensively No Few Relict managed indicators indicators prairie On the MNHP inventory 1 4 Not on the MNHP inventory 29 12 7 5 TOTALS 29 12 8 9

The fact that prairie indicator species were observed on over half of the sites that were managed in a way as to allow some native flora, and that one third of the sites in this category could be classified as actual relicts, is encouraging for those who wish to use GLO records to locate historic blackland prairies. Generally these relicts showed evidence of infrequent mowing: one was a cell phone tower, (Figure 4.10) another was a natural gas line right-of-way (Figure 4.11), another was a seemingly abandoned pasture

(Figure 4.13, next subchapter). The “light-touch” management that characterizes these sites could be applied by conservationists to other sites in the set. The site pictured in

Figure 4.10 is on the MNHP inventory, but the site pictured in Figure 4.11 is not, making it a new discovery.

99

Figure 4.10

Historic GLO prairie under a cell phone tower on Blossom Hill Road in Scott County. with Schizachyrium scoparium and Asclepias tuberosa

Figure 4.11

Good quality prairie relict on a natural gas right-of-way on County Road 31 in Eastern Jasper County

100

4.4.2 The Top Scores East of the Pearl River

In order to gain a better understanding of the top score class of GLO prairies east of the Pearl River, The Mississippi Natural Heritage Program Jackson Prairie Inventory sites, primary, secondary and county road layers and a mosaic of aerial images of the counties were added to the GIS. The ten highest scoring sites that appeared to be accessible from roads were selected for ground validation (Figure 4.12). Two of these sites turned out to be too far from public roads to be evaluated. Notes were taken on the other eight, and the scores for the individual raster layers in the map algebra equation were examined in order to note exactly how their high scores were obtained (Table 4.2).

Three sites on this list contained numerous indicators and had good potential for restoration. One of these three, number nine on the list, is an old field in Jasper County is not included in the MNHP inventory, making it a new discovery (Figure 4.13).

101

102

Figure 4.12

Top Score GLO prairies east of the Pearl River, MNHP Jackson Prairie inventory sites, aerial images, roads and streams

Table 4.2

Variable Scores, Overall Scores and Notes: Top Ten Sites East of Pearl River

16th State USFS Tranmission Land Patch Site MNHP Section Parks Land Lines Cover Density Size Score Note 1 1 1 1 1 10 8 8 448 Indicators in unmowed road ROW 2 1 1 1 1 10 8 8 448 Corn fields 3 YES 1 1 10 1 7 4 2 441 Great diversity, good relict 4 1 1 10 1 7 4 4 487 Forest, some liatris 103 5 1 1 10 1 7 6 2 469 Forest, some liatris 6 1 1 10 10 8 6 4 532 Forest? Not accessible Forest with many openings, 7 YES 1 1 10 1 8 2 2 502 indicators 8 1 1 1 1 7 8 6 454 Trans line ROW, exotics 9 1 1 1 1 10 8 8 430 Numerous indicators, old field 10 1 1 1 1 10 8 8 430 Not accessed, private land

Figure 4.13

GLO Prairie with a high score in an old field in Jasper County with Manfreda virginica, Andropogon glomeratus, Andropogon gerardii and Silphium integrifolium

The implication for management that arises from our investigation of the ten top

scoring sites east of the Pearl River compared with our investigation of 59 sites spread

across the entire study area is that the high scores are not necessarily correlated with good

ecological conditions. The stakeholders were interested in ecological conditions, but the author was not able to find a spatial data set, variable or sampling design to quantify ecological conditions. The eight variables used in the analysis represent the author’s best attempt to express the wishes of the stakeholders with publicly available data. The eight

104

variables generally reflect the stakeholders’ concern with funding mechanisms, which is

certainly an important consideration. If a site scores highly based on the map algebra

output and also is in good ecological condition (as is the case with sites 3, 7 and 9 in

Table 4.2 above), the site is a good candidate for restoration. Prairie patches of good ecological quality that did not score so highly (such as in Figures 4.11 and 4.12) were also found in the process. Had data layers representing the locations of cellular phone towers and natural gas line right-of-ways been available, that data would have been included in the stakeholder survey and in the map algebra equation, and those patches in those figures would have obtained higher scores.

4.4.3 New Discoveries

As mentioned above, five of the relict prairies found during the ground-truthing phase are not included on the MNHP inventory, making them new discoveries. They are recorded in Appendix B: Ground-truthing Notes as numbers 31, 32, 36, 41 and 58. Each of these sites is briefly described in this section.

Number 31 is an old field near Souinlovey Creek on the east side of County Road

2414 in Northern Jasper County (Figure 4.13).

Number 32 (Figure 4.14) is 700 meters south of Number 31 and could be considered the same spot except that it is on the other side of the road and there is a cultivated field between them. This site is a cedar glade on a ridge at a higher elevation than Number 31. Observed indicators include Cornus drumundii, Diospyros virginiana,

Andropogon gerardii and Silphium spp.

105

Figure 4.14

Newly discovered historic GLO prairie relict in Jasper County

Number 36 is a natural gas right of way on County Road 31 in Eastern Jasper

County (Figure 4.11). Observed species include Asclepius tuberose, Liatris spicata,

Diospyros virginiana and Silphium spp.

106

Number 41 is near Shiloh Creek on Matherville-Frost Bridge Road in Wayne

County (Figure 4.15). Observed species include Ratibida pinnata, Rhamnus caroliniana, and Ruellia humilis.

Number 58 is on the south side of Mudline Road in the Southeastern Corner of

Scott County in the Bienville National Forest. Observed species include Asclepius tuberosa, Ratibida pinnata and Schizachyrium scoparium.

Figure 4.15

Rhamnus caroliniana on a newly discovered GLO prairie in Wayne County.

The discovery of five new patches in good ecological condition that can be confirmed as being prairie in the early nineteenth century is an important result of the

107

method used in this research. The locations of the new discoveries were shared with The

Nature Conservancy of Mississippi as point data in an ArcGIS shapefile.

4.4.4 GLO Patches Persist as Open Land Classes

An examination of the relationship between the historic GLO prairie locations and

the landcover class grid also generated implications for prairie patch resilience in the

Jackson Prairie Region. This examination revealed that GLO prairie locations are more

likely than randomly selected locations to be classified as open land cover types. In order

to compare the two data layers, a series of points at 90 meter intervals was created on the

GLO lines, resulting in the generation of 2166 points. 2166 points were then randomly generated in the Jackson Prairie Region, which was defined by aggregating the polygons of the surface geology on which all the GLO prairie points lie. Land cover values were

extracted from the land cover raster to both sets of points. A dramatic difference between

the two sets in the distribution of the land cover classes that represent open types of cover

was observed (Figure 4.16).

108

Figure 4.16

A comparison of landcover data in GLO lines and in random points in the Jackson Prairie Region shows that, while most of the prairies may be gone, the locations tend to remain open classes of land.

4.5 Plat Maps and Survey Notes: Two Georeferencing Strategies

The difference between a map created in ArcGIS using transcribed GLO data as done in this study and one created by digitizing plat map boundaries (Barone 2005a) is discussed in section 2.4.2.2. Since plat maps are created from survey notes, one would expect greater fidelity between maps created by the two methods. Some additional discussion regarding the inconsistencies is included here to aid future researchers.

Barone (2005a) does not go into detail about how the polygons were georeferenced, except to say that each plat map was georeferenced individually in

ArcMap. The aggregate map, shared with this researcher by Barone via email, lacks 109

alignment in all directions with the township and section boundary file from MARIS, and

therefore also with the GLO prairie lines generated for this study, since those lines are based on the MARIS data. The Barone polygons are also inconsistent with aerial images obtained from MARIS which show the location of many township and section lines by the configuration with roads, windrows and field margins. The lack of correspondence between the Barone polygons and these two data sources could be the result of a failure on the part of this researcher to properly reproject the Barone polygons.

Even if the spatial correspondence of the Barone polygons with the township and section boundaries were corrected, however, such a correction would not account for the fact that some of the patches present on either of the two sets of prairie patch locations are absent from the other. This is related to the formatting of the manuscripts of the surveys themselves, which is discussed in section 3.1. It is quite reasonable to attribute the discrepancy between the map created using the method described in this thesis and that created by Barone (2005a) to less the result of any “mistake” made by a researcher and more the result of the lack of a comprehensive and authoritative set of original GLO surveys. Barone (2005a) examined plat maps either as microfilm (at the Department of

Archives and History, Jackson Mississippi) or as paper copies of the originals (at the office of the Secretary of State, Jackson, Mississippi). A similar discrepancy with those files and the fifteen compact discs of over 64,000 pages of scanned (presumably identical) manuscripts used in this study was found in Gordon and Wiseman’s (1989)

Bienville National Forest Survey Report, which quotes the survey of the line between sections 25 and 26 in Township 6 North, Range 8 East, which runs through Harrell

Prairie Hill, as describing “rolling prairie with scattered pine and crabapple thickets.”

110

The corresponding survey in the records used for this study does not mention prairie on

that line or anywhere else in the township. When asked if he could clarify the

discrepancy, Mr. Gordon replied

From what I have seen, there may be multiple copies of the reports and since this predated copy machines, they were transcribed by hand. Since the process was laborious enough, the "extra" stuff was left out of the copies. I'm guessing because I know what I saw. I went to the Secretary of State's office in Jackson and they brought out original copies. The one you cite was not the only one with extra stuff. I remember "old indian field" and a couple of times people's names (settlers) who had cabins. I wish that I had written down the surveyor's name and more details. Secretary of State and others using these records are most interested in the "numbers" to plot lines and they may also tend to gravitate to the newer (cleaner) copies and put the older ones back in the back. I remember the man who was helping me talking about fungus spores and other health hazards. (Ken Gordon, April 27, 2009, e-mail message to author)

It seems that an exhaustive and definitive set of GLO records for the state of

Mississippi does not exist. Finding no reference to prairie in my survey of T8N, R6E

does not necessarily mean that no prairie was noted by the original surveyor. In other

words, “errors” in the surveys are more likely to be errors of omission rather than errors of commission. Although instances of the production of fraudulent surveys are known

(Bourdo 1956), such errors were usually discovered and corrected. It seems less likely

that a deputy surveyor would go to the trouble of recording the existence of a prairie that

did not in fact exist, and more likely that inattentive or rushed surveyors at times fail to

observe the feature, or, having recorded it, fail to preserve the data through the iterative

process of generating plat maps from field notes.

Therefore, examination of plat maps and field notes combined leads to a more

complete set of polygons. The method used in this thesis generated 323 vector lines

derived from GLO survey point data. These lines accurately correspond with plat maps 111

downloaded from the Bureau of Land Management (BLM) website, http://www.glorecords.blm.gov/. The plat map for Township 8 North, Range 2 East in

Madison County was added to the GIS and georefenced by dragging the township corners to those of a township boundary file obtained from MARIS. The prairie polygons in the

BLM map correspond with those lines derived from survey point data for this study

(Figure 4.17). The lines also indicate prairie in areas not designated on the plat map, specifically between sections 4 and 5 and between sections 24 and 25. This may be due to the fact that, as discussed above, not all the data in the notes was necessarily included in the plat map. Or it could be that in these cases the prairies were too numerous and scattered to be mapped but were included in the transcription, as discussed in section 3.3.

112

Figure 4.17

GLO prairie lines created for this study (magenta), and a plat map from the BLM georefenced to a township layer from MARIS using ArcGIS control points tool.

Barone’s polygons, by contrast, provide more information on the shape of the

prairies, but miss the two small prairies on the boundary between sections 4 and 7 (Figure

4.18). On the other hand, his polygons provide us with the other half of the prairie on the eastern boundary, in the adjacent township, confirmed by the transcribed data in the prairie line associated with sections 24 and 25 (Figure 4.18), yet missing from the BLM plat map. This means that the nineteenth-century technician who drew the plat map for this township included every prairie indicated in the notes except the ones between

113

sections 4 and 5 and between 24 and 25, and that, while we may be uncertain about the

line between 4 and 5, we know that the line between 24 and 25 is a true patch because it

corresponds with a patch in the adjacent eastern township.

Figure 4.18

The same plat map used in figure 4.17 with the prairie polygon file provided by Barone. Barone misses the two small prairies between sections 4 and 7 but adds the other half of the prairie in the adjacent prairie to the east, which is indicated by the GLO prairie lines created for this study but missing from the BLM plat map.

Multiple primary sources lead to a more accurate picture overall. As of this writing, the BLM website has a limited number of plat maps in the Jackson Prairie

Region available for download. As GLO data becomes more organized and readily

114

available, on both the state and national levels, studies of this kind can be conducted more efficiently and effectively.

4.6 A Heterogeneous Jackson Prairie Region

Section 4.4 indicates that natural prairies are at least somewhat resilient in the

landscape. The resilience varies across the region, however. As noted in section 4.3,

ground-truthing found no intact relicts and very few indicators in Madison and Hinds

Counties. The confusion about the physiographic classification of these counties is

discussed in detail in section 2.3.3. West of the Pearl River, the influence of the loess

layer on the physiography and vegetation becomes greater as one moves west. Prairies in

this part of the belt are different in a way that has not been addressed in the literature.

The Jackson Prairie is indeed more heterogeneous than the literature

acknowledges. Edwards (1954) mentions “Jackson type” and “Vicksburg type” prairies,

but for him distinction is physiographic and topographic. He has nothing to say about

possible edaphic, botanical or ecological differences. Hodgkins et al. (1976) distinguish

between loessal and non-loessal prairie types, but indicate that the loessal prairies are

essentially the same as the non-loessal ones in the rest of the region, the difference being

that they appear on specific marl outcrops in an area otherwise dominated by loessal

soils.

The results of this study show that the prairies west of the Pearl River are

different than those in the rest of the Jackson Prairie Region in fundamental ways. The

GIS analysis shows that they are larger and more densely clustered. The ground-truthing

shows that they have not persisted over time, and therefore could be said to be less

115

resilient. The results, along with evidence from Barone (2005a), indicate that they are simply too large and widespread to be considered as confined to anomolous marl outcrops, but rather that they are situated on loess. One stakeholder, Heather Sullivan,

MNHP botanist, suggests that loessal soils are prairie soils, but that succession occurs more rapidly on loessal soils, so abandoned areas are forested quickly. Another reason no relicts were found west of the Pearl River could be that loessal prairies are situated on soil that is more productive than their marl-based eastern counterparts. These large prairies were put into agricultural production early and remain so to this day. The essential floristic and ecological differences between loess and Jackson type prairies should be studied further in order to more effectively restore prairies in this part of the region.

4.7 Landscape Architecture: Implications and Applications.

4.7.1 Design in the Context of Environmental Crisis

A georeferenced study of historical natural prairie patches such as this can be used by planners, ecologists and landscape architects to assist in the preservation of regional biodiversity and thereby mitigate the effects of the ongoing environmental crisis.

There are differences of opinion about the depth of the crisis, as well as about exactly how to assess the overall health and integrity of natural systems (Hellmund and Smith

2006, 6). Some assert that we are in the midst of the sixth great mass extinction in the history of the earth (May and McLean 2007, Noss 2007, Wilson 1999). Planners, ecologists, landscape architects and those working in allied fields are increasingly aware

116

of the growing coalescence of problems that accompany land degradation, population

growth, water quality impairments, fertile soil erosion, biodiversity loss, and spread of

huge, urban areas (Dramstad et al. 1996). The factor that contributes most to the crisis of

the collapse of biodiversity is loss of habitat (Hellmund and Smith 2006, Wilson 1999,

Kihslinger, et al. 2008). Because of this, an inventory and assessment of imperiled

species or habitats on site, and a plan to reduce or eliminate threats to such from

development, is an important step in the landscape design process, including the process

as described by Farr (2008), discussed in some detail below.

4.7.2 Ecological Design: A Brief History

As awareness of environmental issues has increased in the culture at large,

landscape architects and planners have responded with changes in theory and practice.

The discussion of “ecological” or “sustainable” design has contributed significantly to the

theoretical orientation of the discipline of landscape architecture over the past fifty years

(Swaffield 2002). Ecological design can be described as an effort to not only “design

with nature” in the McHargian sense, that is, to design in a way that is in harmony with

the ongoing ecological processes of the landscape matrix, but to also use ecological

processes as an aesthetic idiom of sorts that conveys meanings and cultural values to the

users of the space.

It is difficult to trace the true origins of such an idea, but its roots run through the history of civilization. Artists, philosophers and designers have looked to nature as a

guide for centuries. Jellicoe (1987, 70) describes how tenth-century Chinese garden

designers, inspired by landscape paintings and Taoist philosophy about man’s

117

r el ati o ns hi p t o n at ur e, r e pr es e nt e d n at ur al el e m e nts i n t h eir cr e ati o ns. W h e n C hi n es e

landscape paintings beca me popular in Europe i n t h e fift e e nt h a n d si xt e e nt h c e nt uri es,

they influenced European garden design al o n g wit h artisti c a n d p h il os o p hi c al vi e w p oi nts a b o ut n at ur e (J elli c o e 1 9 8 7, 2 2 3). N at ur alisti c f or ms d e pi ct e d i n l a n ds c a p e p ai nti n gs i ns pir e d t h e E n glis h G ar d e ni n g S c h o ol, r e pr es e nt e d b y Willi a m K e nt, C a p a bilit y Br o w n a n d H u m p hr y R e pt o n, i n its r ej e cti o n of t h e hi ghly for mal and sy m metric al p att er ns of L e

N ôtr e i n t h e ei g ht e e nt h c e nt ur y. T h e d esi g n ers of t h e E n gl ish Gardening School vie wed t h eir t as k as t h at of “i m p r o vi n g ” n at ur al c h ar a ct eristi c s alr e a d y pr es e nt i n a sit e, characteristics their Baroque predecessors had ignored. Alexander Pope (1688-1744), an ad mirer of the English Gardening School, cha m pi o n e d t h e i d e a t h at N at ur e its elf h as a r ol e t o pl a y i n t h e pr o c ess of d esi g ni n g l a n ds c a p es i n his E pistl e t o B urli n gt o n ( q u ot e d i n

Rogers 2001, 236):

C o ns ult t h e g e ni us of t h e pl a c e i n all, T h at t ells t h e w at er or t o ris e, or f all, Or h el ps t h e a m biti o us hill t h e h e a v e ns t o s c al e, Or s c o o ps i n cir cli n g t h e atr es t h e v al e, C alls i n t h e c o u ntr y, c a t c h es o p e ni n g gl a d es, J oi ns willi n g w o o ds a n d v ari es shades fro m shades, N o w br e a ks, or n o w dir e cts t h e i nt e n di n g li n es: P ai nts as y o u pl a nt, a n d as y o u w or k, d esi g ns.

Alt h o u g h t h e pr o d u cts of t h e E n glis h G ar d ening School could be described as n at ur alisti c, es p e ci all y i n c o ntr ast t o w h at ca me before, their meanings and purposes were still pi ct ur es q u e i n n at ur e. T h e pri n ci pl e i nt e nti o ns of such constructions were the cr e ati o n of vi e ws. P o p e is q u ot e d as s a yi n g “ All g ar d e ni n g is l a n ds c a p e- p ai nti n g. J ust li k e a landscape hung up” ( Rogers 2001, 236). The subj e cts of s u c h p ai nti n gs w er e a gr ari a n,

1 1 8

rather than natural, landscapes, and as such were manipulated in their topography and vegetation to serve agrarian (chiefly sheep farming), in addition to aesthetic, purposes.

In America, Thomas Jefferson (1743-1825) proclaimed the virtues of the agrarian landscape as the ideal setting for the new democracy. Rogers (2001, 267) points out that, whereas Rousseau’s vision of a more perfect human society was inspired by visions of

Rome, Sparta and Geneva, Jefferson, responding to experiences of crowded industrial towns in England and mob behavior in Paris, favored the ideal of “a virtuous nation-state over a virtuous city-state.” The Louisiana Purchase of 1803, having procured for the country such a great expanse of sparsely-populated land, offered the opportunity to expand Jefferson’s vision of Virginia as a manifestation of a pastoral Golden Age —in which men preserve their freedom by transforming fruitful wilderness into independent farms— to a continental scale (Rogers 2001, 267).

The vision of the American wilderness as an Edenic landscape, free of the taint of corruption and decadence of the Old Country, capable of instilling virtue in its denizens as they propagate new social structures based on liberty and equality, influenced the writings of Ralph Waldo Emerson (1803-1882), Nathaniel Hawthorne (1804-1864), and

Walt Whitman (1819-1892), among others, and therefore can be associated, directly and indirectly, with the efforts of American landscape architects such as Andrew Jackson

Downing (1815-1852), Frederick Law Olmstead (1822-1903), the architect Frank Lloyd

Wright (1867-1959) and the Dutch-born landscape architect Jens Jensen (1860-1961), as they promulgated an American aesthetic through the use of forms derived from an indigenous ecological setting. The notion that Nature is the supreme artist, that artistic achievement involves translating or imitating qualities, modalities and processes found in

119

nature, was therefore already an established trope in philosophy and the arts when Ian

McHarg first wrote Design with Nature (1992) in 1967. McHarg’s achievement,

discussed in section 2.5.1, is the assignment of a social value to the processes of nature,

and the development of a landscape design method that maximized the potentials of and

minimized threats to such values.

The concept was developed further by Allan Ruff, director of Landscape Studies at the University of Manchester, U.K., 1987-97, who translated ideas developed during his exploration of Dutch parks into a broad sketch of an ecological design method in an essay, originally published in 1982, entitled “An Ecological Approach to Landscape

Design.” After enumerating a series of critical points, Ruff concludes that

[a] landscape developed along these ecological lines will serve to create a powerful aesthetic form that can both reflect and affect positive environmental change. Recognition of this fact can make landscape design more vital, the profession of landscape design more useful and the management of durable and dwindling resources and fragile ecosystems more publicly important (Ruff 2002, 177).

Ann Whiston Spirn championed the idea that urban design be more responsive to ecological processes in The Granite Garden: Urban Nature and Human Design (1984), in which she describes in detail the ongoing and largely unnoticed ecological processes already occurring in the urban setting. She portrays a “ubiquitous” nature, an integrated totality that “embraces” the city. She goes on:

To seize opportunities inherent in the city’s natural environment, to see beyond short-term costs and benefits, to perceive the consequences of the myriad, seemingly unrelated actions that make up daily city life, and to coordinate thousands of incremental improvements, a fresh attitude to the city and the

120

molding of its form is necessary. The city must be recognized as part of nature and designed accordingly (5).

John T. Lyle (2002) links ecological design to aesthetic principles and deep

meaning in “Can floating Seeds Make Deep Forms?” originally published in 1991. Lyle

(2002) argues that ecological design produces “deep” rather than “shallow” forms because it is rooted in the “underlying complex and elegant ecosystematic order of nature

as the essential and fundamental inspiration for design.” Philosophically this seems to

come very close to the notions of Alexander Pope and the English Gardening School.

Lyle and his contemporaries, however, reject the limitations imposed by the picturesque

motivation and its preoccupation with the visual experience. Catherine Howett (2002)

explicates this rejection thoroughly in “Systems, Signs and Sensibilities,” originally

published in 1987. Even though, as Lyle (2002) points out, actual ecological processes are so complex that they generally can only be perceived by trained ecologists, current practitioners, quoted extensively by Howett (2002), seek to translate ecological principles into a new landscape aesthetic that engages participants in an experience of dynamic environmental forces.

4.7.3: Implication for Design at Different Scales

Given the fact that the discipline has demonstrated a commitment to incorporate ecological ideas more fully into the design process, this thesis has implications for the theory of landscape architecture and its practice in the Jackson Prairie Region at (at least) four different scales: site specific design, ecological restoration, urban planning, and regional management decisions.

121

4.7.3.1 Site Specific Design

The awareness of the presence of an imperiled ecological habitat, even one lost to history, is important to those who make decisions at site specific scale. The presence of calcareous soils and prairie indicator species should be considered as presenting a unique set of opportunities and constraints to any landscape architect working in or near any of the southeastern prairie belts. The degree to which any site is proximal or coincident to the pattern of GLO prairies can add value to efforts to conserve or celebrate that part of our ecological heritage. Increasing availability of the georeferenced data of the General

Land Office (manuscripts, plat maps, and their digitized equivalents available online from sources such as the Bureau of Land Management) in a GIS environment will facilitate the efforts of ecological designers to preserve what is left and restore what is lost. It is the hope of this researcher that an examination of the GLO data, either as notes or as plat maps, would become a standard step in the inventory and analysis process of all landscape architects at all scales.

The Crosby Arboretum in Picayune, Mississippi is celebrated as being the only facility of its kind to display native plants in situ, so as to express relationships among species and between species and setting, as opposed to a landscaped collection of individual plants in mulched beds. The Crosby Arboretum strives for scientific accuracy in addition to aesthetic experience in conveying to its visitors a set of values and meanings related to the ecology of the piney woods component of the Pearl River drainage basin (Brzuszek 2007). When design work began in 1979, the long, flat site, formerly a pine plantation and strawberry farm that had suffered hurricane damage and been abandoned, had almost no scientific value and little aesthetic merit (Johnson and

122

Frankel, 1991). The ecological possibilities of the site were stretched to include eight distinct natural vegetation types by manipulating two important environmental determinants: fire and moisture (Brzuszek 1990). The manipulation of moisture has resulted in a range of habitats from aquatic to pond margins to bogs to dry forest, woodland and savanna. Similarly the patterns of fire frequency have established a vegetative spatial gradient from forest to woodland/savanna to grassland.

The creation of a Jackson Prairie Region version of the Crosby Arboretum is entirely possible given what can be brought to light through research such as is included in this thesis. As with Crosby, a successful design would likely depend on moisture and fire frequency gradients to put into motion the interplay of components. In the GLO notes, in Township 3 North, Range 10 East, in Jasper County, between sections 16 and

21, the surveyor exits a swamp in enters prairie directly. The note is included in

Appendix A: Survey Transcription. It is also included as stop 25 in Appendix B, Ground

Truthing Notes. The researcher was unable to gain access to the exact spot because it was on gated land far from the road, but the GLO note does indicate that prairies existed on slopes adjacent to swamps. The importance of this feature is reinforced by Barone and

Hill (2007), discussed in section 2.3.2, who assert that the plant species that have disappeared from blackland prairie species inventories over the past hundred years are generally associated with wet, sandy conditions, indicating that wet, sandy prairies should be a conservation priority.

A Jackson Prairie Region version of the Crosby Arboretum, therefore, should include a wet component, a feature that does not currently exist on any of the US Forest

Service prairie restoration sites mentioned below. It should also include the surrounding

123

pine or mixed hardwood forest and demonstrate the interactions of the habitats. A full

palette of woody and herbaceous species can be derived from the literature cited in this

thesis. A successful design, therefore, would not focus entirely on the opening itself but

would utilize the dynamic interactions with the edges and neighboring habitats from wet

to dry and from those that are burned periodically and those that are not.

4.7.3.2 Ecological Restoration

Prairie patches in the Jackson Prairie are a rare and poorly understood habitat

(Gordon 1989). The continued degradation and eventual disappearance of blackland

prairies is a real possibility (Peacock and Schauwecker 2003, 4) and would be a significant setback to efforts to preserve regional biodiversity. Temperate grassland is identified as a major habitat type by The Nature Conservancy and as such is included as

part of their “Conservation by Design” management strategy which has as its goal the

effective conservation of places that represent at least ten percent of every major habitat

type on Earth by 2015 (The Nature Conservancy 2006). By mapping historic prairie patch

locations, ranking them in terms of stakeholder criteria and identifying relict prairies in

the region, this thesis constitutes a significant contribution to regional biodiversity

conservation.

An ongoing collaborative effort between The Nature Conservancy of Mississippi

and the United States Forest Service to restore prairies on six sites in the Bienville

National Forest is known as the Stewardship Prairies Project. A test of the effectiveness

of triclopyr (trade name Garlon) on the suppression of regrowth of felled woody plants

was conducted by the writer of this thesis and Dr. Timothy Schauwecker by executing

124

pre-treatment and post-treatment surveys of woody plants in observation plots in the six

sites. The test allowed the thesis writer to observe the emergence of prairie vegetation on formerly forested patches over a period of two years (2008-2010). With proper management, the Stewardship Prairies Project could be a successful restoration. The potential for prairie at these sites was determined by deficient timber yields, the presence of indicator species, and soil tests (Gloria Nelson, personal communication). GLO data,

such as that used in this thesis, could be added to the set of variables in the selection of

future restoration sites. None of the six Stewardship Prairies Project sites are indicated in

the GLO data, but only one of the six is contiguous with a section line. GLO data does indicate several historic prairie patches along section lines in the two townships in which

the six restoration sites are located.

4.7.3.3 Urban Planning

In addition to recommendations for regional management patterns and ecological restoration projects, a thesis such as this has implications for the design of cities. Douglas

Farr’s Sustainable Urbanism: Urban Design with Nature (2008) is an urban design

handbook conceptually based on the newly inaugurated Leadership in Energy and

Environmental Design - Neighborhood Development (LEED-ND) rating system. The

United States Green Building Council (USGBC) first developed the LEED system in

1995 to encourage environmental or green building practices in the construction of

buildings. The Neighborhood Development standards, drafted in 2002 in partnership with

the Congress of New Urbanism and the National Resource Defense Council, represents a

shift towards a more comprehensive view of green building. As a result, the sizeable base

125

of practicing LEED Accredited professionals and the USGBC are “well positioned to be

a virtual green army of sustainable urbanists” (Farr 2008, 36).

The protection of imperiled species and ecological communities is a prerequisite

of LEED-ND certification under the Smart Location and Linkage heading (U. S. Green

Building Council 2007). To achieve certification, the project must comply with an

approved Habitat Conservation Plan (HPC) under the Endangered Species Act for each

identified species or community. If no HPC exists, the developer must coordinate with

the state’s Natural Heritage Program or fish and wildlife agency to perform adequate

surveys of ecological conditions of the project. If imperiled species or communities are

identified, the developer must work with a qualified biologist or conservation organization to implement a buffer and eliminate or mitigate threats from the development.

Neither Farr (2008) nor the LEED-ND prerequisites (2007) address the issue of lost habitats evident in the historical record that could potentially be restored. The results of this thesis shows GLO prairie patches scattered throughout the Jackson Metro Area.

The existence of this record and the locations of the lost patches ought to be known by every landscape architect and urban planner in the Jackson Prairie Region. As such, the information in this study presents those who plan and design the built environment with a set of possible constraints to be considered at the inventory and analysis phase. Of course, once identified and either buffered or integrated into a use pattern, a restored prairie relict can be considered an opportunity during the design phase.

126

4.7.3.4 Regional Management Decisions

The pattern of historic prairie patch distribution at a regional landscape level

should be considered by conservationists, especially in terms of the distributions

implications for species dispersal and other patch dynamics. Landscape Architects must

continue to take on regional scale problems in order to maintain a place at the table,

alongside biologists, engineers and policy makers, when such implications are discussed

and regional policies are decided.

Glenda Clardy, biologist with the Natural Resource Conservation Service (NRCS)

and one of the stakeholders in this study, suggested that research such as this could be

used to form the basis of the designation of Jackson Prairie as a regional target habitat of

the Wildlife Habitat Incentives Program (WHIP), which could fund conservation efforts

on agricultural land (personal communication). Since landscape architects are well-suited

to work within the nexus of ecological and social values, they could contribute

significantly to the establishment of what importance GLO data would play, relative to soil types, vegetative cover or other variables, in the justification of the spatial pattern of this funding mechanism. In other words, a defensible decision rule matching the funds with the spatial distribution of the target habitat would be necessary in order to explain

why one landowner qualifies for the funds and another does not, and landscape architects

should expect to participate, along with professionals in allied fields, in the creation of

such a rule.

Another regional scale implication of this study involves maintenance patterns

along road and utility rights of way. Some good quality habitats were discovered in these

areas during the ground-truthing phase of this study. The Mississippi Department of

127

Transportation (MDOT) and various utilities maintain extensive corridors in the Jackson

Prairie Region. Establishing maintenance patterns compatible with their uses and prairie restoration should not be difficult.

128

CHAPTER V

CONCLUSION

Survey transcriptions from the General Land Office (GLO) records of Mississippi were found to be useful in providing locations of historic prairie patches that, when combined with publicly available landscape-level data, can aide conservationists in the restoration and conservation of this imperiled habitat. Limitations to this method include that fact that GLO surveys are restricted to section and township lines only. They do not provide data about conditions in the section interior. Another limitation is the lack of a variable that is capable of quantifying ecological conditions in the computation of suitability. Since the ecological conditions of the particular sites are unknown, assignment of suitability values had to be computed without regard to this important consideration. The development of such a variable is difficult to achieve, since GIS processes require, for the most part, data acquired from remote sensing, and the most defensible way to evaluate the condition of a prairie patch is to observe the presence of indicator species, which must be done on site. A third limitation is the incomplete and disorganized state of the GLO transcripts themselves, discussed in sections 3.1 and 4.5.

This limitation leads to errors of omission rather than errors of commission.

Blackland prairies of the Gulf Coastal Plain are rare, poorly understood and threatened ecological assemblages. Of these typically belted formations of prairie

129

patches, the Jackson Prairie Region in Mississippi is perhaps the most poorly understood and most threatened of all. Under the direction of a grant funded by The Nature

Conservancy (TNC) of Mississippi and with the guidance of Dr. Timothy Schauwecker, the researcher exhaustively searched the literature to gain an understanding of the ecology, physiography and geology of the Jackson Prairie Region. A thorough search of

General Land Office (GLO) records for precise locations of pre-European settlement prairie patches in the region was also undertaken. Entrance and exit points of prairie patches along section lines were transcribed, and the data was digitized and georeferenced using ArcGIS software in an automated process developed by Kate Grala, a researcher in the Department of Geosciences, Mississippi State University. The automation tool allowed the researcher to use the prairie patch location data from the

GLO records as a base map in a Geographical Information System (GIS). Input from stakeholders, representatives of TNC’s partners in conservation, was applied in the selection of a set of conditions that would cause the historical patch locations to be considered suitable for conservation or restoration. These conditions were added to

ArcGIS base map as data layers and were assigned weights of importance according to stakeholder input. A simple additive weighting technique was used in the ArcGIS map algebra function in order to rank the historic patch locations in terms of suitability for conservation or restoration. A significant portion of the locations ranked as highly suitable are in Madison County in an area characterized by loessal soils and considered by some researchers to be outside the Jackson Prairie Region. Further study on the differences between loessal and non-loessal prairie patches is recommended. Fifty-eight historic patch locations were visited in 2009. Of these sights, 29 (50%) were so

130

intensively managed that all evidence of prairie was suppressed. Of the remaining 29 locations, 8 (28%) contained indicator species and 9 (31%) were identifiable relict prairie patches. Of these, five are not on an inventory of known prairie patches maintained by the Mississippi Natural Heritage Foundation, indicating that these are new discoveries of previously unrecorded prairie patches in the region.

The process of deriving data from GLO records for the purpose of identifying the historical pattern of distribution of a rare habitat has implications for landscape architecture. The ability to create spaces that preserve and enhance our ecological heritage at multiple scales (site specific, municipal, regional) is improved by our ability to use GIS software and current spatial data to analyze and interpret georeferenced historical data. As Carr and Zwick (2007) point out, “the cumulative reality of individual land-use decisions is rarely, if ever, understood.” Land use change often occurs one parcel at a time. When aggregated changes are reviewed over intervals of five or ten years, unanticipated and rapid transformation of large areas can be revealed.

Geographical Information Systems (GIS) are becoming more useful every year in helping planners and stakeholders gain an understanding of the information derived from spatial data. This understanding is of course limited by the data and by the questions, goals and objectives which drive the research. The myriad of small changes that, when combined, lead to fragmentation and ultimately the destruction of natural habitats will continue to be a critical topic of study for those wishing to pass an intact, functioning biosphere to future generations.

Having said that, much of the most important work to be done in the coming decades can’t be quantified in columns and rows. It lies beyond our computer screens and

131

hard drives and involves our ability to actually see the totality of what our environment is

telling us, and to carefully use that information as we make decisions about multiple

human uses of land. Ecologists necessarily focus their attention on individual components

of systems and the processes, flows of energy and mass, by which the components are

related one to the other. Once a system is conceived or modeled in the human mind, or in

a diagram on a page or in a modeling software package, its relationship to other systems,

the interplay between the carbon cycle and the hydrological cycle for example, must also

be addressed. Even for the specialist, comprehension of these processes across multiple

temporal and spatial scales is difficult. If ecological designers wish to derive aesthetic norms and cultural values from the study of these systems (and I believe the project to be well worth the effort), such that people coming and going, stopping, working, living and playing in these environments, begin to get a sense of their place in the greater system, to truly, as Annie Dillard (1998) puts it “explore the neighborhood, view the landscape, to discover at least where it is we have been so startlingly set down, if we can’t learn why,” they must not only master quantitative analysis but be willing to aggressively and immediately move beyond it. Many blackland prairie patches were abandoned once they were determined to be unsuitable for farming. Aldo Leopold (1949) asserts:

If the land mechanism as a whole is good, then every part is good, whether we understand it or not. If the biota in the course of aeons, has built something we like but do not understand, then who but a fool would discard seemingly useless parts? To keep every cog and wheel is the first precaution of intelligent tinkering.

Good designers are master recyclers, and, like the medieval alchemist, continuously seek new possibilities for and evaluations of what has been cast aside, of

132

“the stone the builders rejected” (Psalm 118:22 [New Oxford Annotated Bible]). Every

component in a design is important. A garden can be viewed as a microcosm of a world.

As we endeavor to care for the inhabitants of the earth by caring for the earth itself, we should study individual components and larger integrated systems with equal diligence.

As we work to translate the values and meanings inherent in these systems back to those

who use the land, we obtain something approaching wisdom.

133

WORKS CITED

Agnew, Dwight L. 1941. The government surveyor as a pioneer. The Mississippi Valley Historical Review 28 no. 3: 369-382.

Anderson, Roger C. 2006. Evolution and origin of the central grassland of North America: climate, fire and mammalian grazers. Journal of the Torrey Botanical Society 133.4: 626-647.

Asbjornsen, H., L. A. Brudvig, C. M. Mabry, C. W. Evans and H. M. Karnitz. 2005. Defining reference information for restoring ecologically rare tallgrass oak savannas in the midwestern United States. Journal of Forestry 103 no. 7: 335-50.

Axelrod, Daniel I. 1985. Rise of the grassland biome, central North America. The Botanical Review 51 no. 2 (April-June): 163-201.

Barone, John A. 2005a. The historical distribution of prairies in the Jackson Prairie Belt and in western Mississippi. Journal of the Mississippi Academy of Sciences 50 no. 2: 144-48.

———. 2005b. Historical Presence and distribution of prairies in the Black Belt of Mississippi and Alabama. Castanea 70 no. 3: 170-183.

Barone, John A. and Jovonn G. Hill. 2007. Herbaceous flora of blackland prairie remnants in Mississippi and western Alabama.” Castanea 72 no. 4: 226-34.

Bourdo, Eric A. Jr. 1956. A review of the General Land Office Survey and of its use in quantitative studies of former forests. Ecological Society of America 47 no. 4: 754-768.

Bragg, Don C. 2002. Checklist of major plant species in Ashley County, Arkansas noted by General Land Office surveyors. Journal of the Arkansas Academy of Sciences 56.

———. 2004. “General Land Office Surveys as a source for Arkansas history: the example of Ashley County.” The Arkansas Historical Quarterly 18 no. 2: 166- 84.

134 Brzuszek, Robert. 1990. Establishing Spatial Patterns for the Beech-Magnolia Exhibit in the Crosby Arboretum of Picayune, Mississippi. Master’s thesis, Louisiana State University.

Brzuszek, Robert. 2007. Are they getting it? Visitors respond to the Crosby Arboretum’s ecological aesthetic. Landscape Architecture May: 78-85.

Buol, S. W., F. D. Hole, and R. J. McCracken. 1989. Soil Genesis and Classification. 3rd ed. Ames IA: Iowa State University Press. Quoted in L.P. Moran, D.E. Pettry, R.E Switzer, S.T. McDaniel and R.G. Wieland. 1997. Soils of native prairie remnants in the Jackson Prairie Region of Mississippi. Bulletin 1067. Mississippi Agricultural and Forestry Experiment Station, Mississippi State, MS.

Bureau of Land Management General Land Office Records. http://www.glorecords.blm.gov/ (accessed December 3, 2009).

Carr, Margaret H. and Paul D. Zwick. 2007. Smart Land-Use Analysis: The LUCIS Model. Redlands, CA: The ESRI Press.

Chapman, S. S., G. E. Griffith, J. M. Omernik, J. A. Comstock, M. C. Beiser, and D. Johnson. 2004. Ecoregions of Mississippi, (color poster , descriptive text, summary tables and photographs) Reston, Virginia: U.S. Geological Survey.

Chrisman, N. 2002. Exploring Geographic Information Systems. 2nd ed. New York: John Wiley & Sons, Inc.

Cowels, H.C. 1928. Persistence of prairies. Ecology 9: 380-382.

Dillard, Annie. 1998. Pilgrim at Tinker Creek. New York: HarperCollins Publishers, Inc.

Dramstad, Wenche E., James D. Olsen and Richard T. T. Forman. 1996. Landscape Ecology Principles in Landscape Architecture and Land Use Planning. Washington, DC: Island Press.

Edwards, Jr., Douglas Edwin. 1954. The Physiography of Hinds County, Mississippi. Master’s Thesis. Mississippi State University.

Elsen, Dean and Ronald Wieland. 2003. Rediscovery and management of prairie remnants of the Bienville National Forest, east-central Mississippi. In Blackland Prairies of the Gulf Coastal Plain: Nature, Culture and Sustainability, ed. Evan Peacock and Timothy Schauwecker, 239-235. Tuscaloosa, Alabama: The University of Alabama Press.

ESRI. 2004. ArcGIS 9 Desktop Help. Redlands, California.

135 Farr, Douglas. 2008. Sustainable Urbanism: Urban Design with Nature. Hoboken, NJ: John Wiley & Sons, Inc.

Fenneman, Nevin M. 1938. Physiography of Eastern United States. New York: McGraw- Hill Book Company, Inc.

Fitz, Gideon. 1832. Instructions for Surveying in the State of Mississippi. Natchez, Mississippi: R. Semple for the Surveyor General.

Foti, Thomas L. 1989. Blackland prairies of southwestern Arkansas. Proceedings Arkansas Acadamy of Science 43: 23-28.

Foti, Thomas L, Scott Simon, Douglas Zollner and Meryl Hattenbach. 2003. Blackland prairie landscapes of southwestern Arkansas. In Blackland Prairies of the Gulf Coastal Plain: Nature, Culture and Sustainability. ed. Evan Peacock and Timothy Schauwecker, 48-63. Tuscaloosa, Alabama: The University of Alabama Press.

Gordon, Ken L. and J.B. Wiseman, Jr. 1989. Bienville National Forest prairie survey final report. Museum Technical Report No. 7. Mississippi Department of Wildlife, Fisheries and Parks, Museum of Natural Science. Jackson, MS.

Harper, L. 1857. Preliminary Report on the Geology and Agriculture of the State of Mississippi. Jackson, MS: E. Barksdale, State Printer.

Helmund, Paul Cawood and Daniel Somers Smith. 2006. Designing Greenways: Sustainable Landscapes for Nature and People. Washington: Island Press.

Hilgard, E. 1860. Report on The Geology And Agriculture of The State of Mississippi. Jackson, MS: E. Barksdale, State Printer.

Hodgkins, Earl J., Timothy K. Cannon and W. Frank Miller. 1976. Forest habitat regions from satellite imagery: states of Alabama and Mississippi (poster and descriptive text) Department of Forestry, Alabama Agricultural Experiment Station, Auburn University. Department of Forestry, Mississippi Agricultural and Forestry Experiment Station, Mississippi State University.

Hodgkins, Earl J., M.S. Golden and W.F. Miller. 1979. Forest Habitat Regions and Types on a Photomorphic-Physiographic Basis: A Guide to Forest Site Classification in Alabama and Mississippi. Southern Cooperative Series No. 210, Alabama Agricultural Experiment Station, Auburn University, Mississippi State University.

Hopkins, Lewis D. 1977. Methods for generating land suitability maps: a comparative evaluation. Journal of the American Planning Association 43 no. 4 (October): 386-400.

136 Howett, Catherine. 2002. Systems, signs and sensibilities. In Theory in Landscape Architecture: A Reader. ed. Simon Swaffield, 108-116. Philadelphia: University of Pennsylvania Press.

Jellicoe, Geoffrey and Susan Jellicoe. 1987. The Landscape of Man. London: Thomas and Hudson, Ltd.

Johnson, Jory and Felice Frankel. 1991 Modern Landscape Architecture: Redefining the Garden. New York: Abbeville Press.

Jones, Alice Simms and E. Gibbs Patton. 1966. Forest, “prairie” and soils in the Black Belt of Sumter County, Alabama, in 1832. Ecology 47: 75-80.

Jones, S.B. Jr. 1971. A virgin prairie and a virgin loblolly pine stand in central Mississippi. Castanea 36: 223-225.

Jordan III, William R. 1997. Forward to The Tallgrass Restoration Handbook for Prairies, Savannas and Woodlands. Ed. Stephen Packard and Cornelia F. Mutel. Covelo, California: Island Press.

Kihslinger, Rebecca L., Jessica Wilkinson and James McElfish. 2008. Biodiversity corridors. In Sustainable Urbanism: Urban Design with Nature by Douglas Farr, 120-121. Hobokon, NJ: John Wiley & Sons, Inc.

Kline, Virginia M. 1997. Orchards of oak and a sea of grass. In The Tallgrass Restoration Handbook for Prairies, Savannas and Woodlands. Ed. Stephen Packard and Cornelia F. Mutel. Covelo, California: Island Press.

Leidolf, A. and Sidney McDaniel. 1998. A Floristic study of Black Prairie plant communities at sixteen section prairie, Oktibbeha County, Mississippi. Castanea 63 no. 11: 51-62.

Leopold, Aldo. 1949. A Sand County Almanac: And Sketches Here and There. Oxford. Oxford University Press, Inc.

Lowe, E. N. 1915. Mississippi: Its Geology, Geography, Soils and Mineral Resources. Jackson, Mississippi: Tucker Printing House.

Lowe, E. N. 1921. Plants of Mississippi: A List of Flowering Plants and Ferns. Jackson, MS: Hederman Bros.

Lyle, John. 2002. Can floating seeds make deep forms? In Theory in Landscape Architecture: A Reader. ed. Simon Swaffield, 188. Philadelphia: University of Pennsylvania Press.

137 Malczewski, Jacek. 1999. GIS and Multicriteria Decision Analysis. New York: John Wiley and Sons, Inc.

MARIS. 1994. Physiographic Regions. GIS data layer map of the physiographic regions of Mississippi available for download at http://www.maris.state.ms.us/HTM/DownloadData/Statewide-Theme.html

May, Robert McCredie and Angel R. McLean. 2007. Theoretical Ecology: Principles and Applications. Oxford: Oxford University Press.

McHarg, Ian L. 1992. Design With Nature: 25th Anniversary Edition. New York: John Wiley and Sons, Inc.

McHarg, Ian L. and Frederick R. Steiner, eds. 1998. To Heal The Earth: Selected Writings of Ian L. McHarg. Washington, D.C.: Island Press.

Merrill, Robert K., James J. Sims, Jr., Delbert E. Gann, and Kenneth J. Liles. 1985. Newton County Geology and Mineral Resources: Bulletin 126. Jackson, Mississippi: Mississippi Department of Natural Resources.

Mississippi Band of Choctaw Indians. Treaties between the United States and the Choctaw Nation. http://www.choctaw.org/History/Treaties/Treaties.html (accessed March 29, 2010).

Mississippi Museum of Natural Science. 2005. Mississippi’s Comprehensive Wildlife Conservation Strategy. Jackson, Mississippi: Mississippi Department of Wildlife, Fisheries and Parks, Mississippi Museum of Natural Science.

Mooney, H. A. 1972. The carbon balance of plants. Annual Review of Ecology and Systematics 3: 315-346.

Moran, L. P., D. E. Pettry, R. E Switzer, S. T. McDaniel and R. G. Wieland. 1997. Soils of native prairie remnants in the Jackson Prairie Region of Mississippi. Bulletin 1067. Mississippi Agricultural and Forestry Experiment Station, Mississippi State, MS.

Noss, Reed F. 2007. Values are a good thing in conservation biology. Conservation Biology 21 no. 1: 18-20.

Packard, S. and C. F. Mutel, eds. 1997. The Tallgrass Restoration Handbook for Prairies, Savannas and Woodlands. Covelo, California: Island Press.

Peacock, Evan, John Rodgers, Kevin Bruce and Jessica Gray. 2008. Assessing the pre- modern tree cover of the Ackerman Unit, Tombigbee National Forest, North

138 Central Hills, Mississippi, using GLO survey notes and archeological data. Southeastern Naturalist 7 no. 2: 245-66.

Peacock, Evan and Timothy Schauwecker. 2003. The nature, culture and sustainability of blackland priaires. In Blackland Prairies of the Gulf Coastal Plain: Nature, Culture and Sustainability. ed. Evan Peacock and Timothy Schauwecker, 1-7. Tuscaloosa, Alabama: The University of Alabama Press.

Pyne, Stephen J. 1982. Fire in America: A Cultural History of Wildland and Rural Fire. Seattle and London: University of Washington Press.

Rogers, Elizabeth Barlow. 2001. Landscape Design: A Cultural and Architectural History. New York: Harry N. Abrams, Inc.

Rostlund, Erhard. 1957. The myth of a natural prairie belt in Alabama: an interpretation of historical records. Annals of the Association of American Geographers 47: 392- 411.

Ruff, Allan. 2002. An ecological approach. In Theory in Landscape Architecture: A Reader. ed. Simon Swaffield, 175-177. Philadelphia: University of Pennsylvania Press.

Sauer, Carl O. 1950. Grassland climax, fire and man. Journal of Range Management 3, no. 1 (January): 16-21.

Spirn, Ann Whiston. 1984. The Granite Garden: Urban Nature and Human Design. New York: BasicBooks.

Steiner, Frederick. 2004. Healing the earth: the relevance of Ian McHarg’s work for the future. Philosophy and Geography 7 no. 1 (February): 141-149.

———. 2006. Metropolitan resilience: The role of universities in facilitating a sustainable metropolitan future. In Toward a Resilient Metropolis: The Role of State and Land Grant Universities in the 21st Century, ed. Arthur C. Nelson, Barbaa L. Allen and David L. Trauger, 1-18. Alexandria, VA: Metropolitan Institute at Virginia Tech.

———. 2008. The Living Landscape: An Ecological Approach to Landscape Planning. Washington, DC: Island Press.

Swaffield, Simon. 2002. Part V: Ecological design and the aesthetics of sustainability. In Theory in Landscape Architecture: A Reader. ed. Simon Swaffield. 171. Philadelphia: University of Pennsylvania Press.

139 The Nature Conservancy. 2006. Conservation by Design: A Strategic Framework for Mission Success. Arlington, VA.

TNGenWeb Project. Indian land cessions in the American Southeast. http://www.tngenweb.org/cessions/ (accessed March 29, 2010).

Thompson, David E. 2009. Geologic map of Mississippi, (color map with legend). Mississippi Office of Geology, MDEQ.

U.S. Green Building Council. 2007. LEED for Neighborhood Development Rating System: Pilot Version. U. S. Green Building Council.

Whitney, Gordon G. 1986. Relation of Michigan’s presettlement pine forests to substrate and disturbance history. Ecology 67 no. 6: 1548-1559.

Whittaker, Robert H. 1975. Communities and Ecosystems. New York: Macmillan Publishing Company, Inc.

Wilson, E. O. 1999. The Diversity of Life. New York: W. W. Norton & Company, Inc.

Wiygul, Sherril, Kay Krans, Richard Brown and Victor Maddox. 2003. Restoration of a prairie remnant in the Black Belt of Mississippi. In Blackland Prairies of the Gulf Coastal Plain: Nature, Culture and Sustainability. ed. Evan Peacock and Timothy Schauwecker, 254-261. Tuscaloosa, Alabama: The University of Alabama Press.

140

APPENDIX A

SURVEY TRANSCRIPTION

141

KEY: page #: The number of the .tif file in the digital record of scanned survey manuscripts T: Township. All the Townships in the Jackson Prairie Region are north of their baseline (although they are not all north of the same baseline) R: Range E/W: Whether the Range is east or west. Ranges west of the Choctaw Meridian number 1 or 2. Ranges in the extreme eastern end of the Jackson Belt are numbered westward from the St. Stephens Meridian in Alabama and their numbers range from 5 to 9. D: The direction in which the surveyor is traveling when he executes the survey. btw sect: The two sections between which the line is surveyed. If the survey is on a townhip boundary, causing one of the sections to be in a different township, the notations in this column are not standardized. They were standardized later when the data was prepared for GIS automation. enter: the distance from the corner to the point at which the prairie was entered in chains. exit: the distance from the corner to the point at which the prairie was exited in chains. total chains in mile: whether the mile was measured as 80 or 160 chains. comments: Comments about the conditions of the land are transcribed from the manuscripts and original spelling and punctuation is generally preserved. Comments about the condition of the manuscripts or how they are organized are by the author. NO P = no prairie mentioned in the survey. NA = survey not available.

total chains page # T R E/W D btw sect enter exit in mile comments 24403 1 7 E NO P 8557 1 8 E NO P 1997 1 9 E NO P 1 10 E NA 1 10 E Not in fdn viewer Mostly unreadable but seems generally 48825 1 11 E forested 45570- 45615 1 12 E NO P 1 13 E NA 1 13 E Not in fdn viewer 11561 1 14 E MAP shows p in 15,16 Land gently rolling soil 11596 1 14 E N 15,16 25 50 80 2nd rate pine oak + 142 hickory

fdn finder shows file 60318 is followed by 61781, therefore: Not 60536- available, these file #s 60603 1 15 E out of sequence or lost Land broken soil 2nd rate Pine Oak Hickory + 12857 1 16 E S 33,34 3 14 80 Blackjack Land broken soil 2nd rate Pine Oak Hickory + 12857 1 16 E S 33,34 26 32 80 Blackjack Land broken soil 2nd 12858 1 16 E E 27,34 69 74 80 rate Pine Oak Hickory Land broken soil 2nd rate Pine Oak + 12875 1 16 E E 23,26 71 80 80 Blackjack Land broken soil 2nd 12877 1 16 E N 34,35 47 73 80 rate Pine Oak Hickory Land rolling soil 2nd rate Pine Oak + 12878 1 16 E E 26,35 33 80 80 Blackjack Land level soil 2nd rate 12880 1 16 E E 25,36 0 4 80 Oak + Blackjack Land level soil 2nd rate 12880 1 16 E E 25,36 70 80 80 Oak + Blackjack Land rolling 2nd rate soil Pine Oak + 12881 1 16 E N 25,26 0 60 80 Blackjack Land rolling 2nd rate soil Pine Oak + 12881 1 16 E N 25,26 73 80 80 Blackjack Land rolling 2nd rate soil Pine Oak + 12882 1 16 E E 24,25 76 80 80 Blackjack Land rolling soil 2nd 12883 1 16 E N 23,24 0 5 80 rate Pine Oak + Hickory Land rolling soil 2nd 12884 1 16 E E 13,24 44 50 80 rate Pine Oak + Hickory 12922 1 16 E MAP showing prairies Land rolling soil 2nd 6845 1 17 E W 30,31 0 6 80 rate Pine Oak + Hickory MAP but does not show p recorded on 6845. seems to sho p btw 6908 1 17 E 29,28 This is a narrow township on the eastern edge of the 1751 1 18 E E 20,29 42 51 80 state 7034- 7273 2 6 E NA

143 24694 2 7 E NO P 8599- 8642 2 8 E NO P 11464- 11527 2 9 E NO P 49869 2 10 E NO P 49537 2 11 E NO P 45675- 45727 2 12 E NO P 20769 2 13 E NO P 2 14 E NO P 60604- this pages are skipped 60635 2 15 E on disc 14 13464 2 16 E NO P 5414- 5448 2 17 E NO P Level poor land ash woods and bushy prairie 5-20-09: moved this line to T2 so in toby_pnts so it will 35464 3 1 W N 32,33 0 160 160 show up in T3 first half good soil post oak red ash hickory. Residue poor bushy 35464 3 1 W N 28,29 0 160 160 prairie surface wvy first half level good soil tall oak and hickory the latter poor soil 35465 3 1 W N 16,17 67 80 160 blackjack growth the first 130 ch in poor prairie deeply set with small bushes, residue ash timber level poor 35479 3 1 W E 21,28 0 130 160 land 10809 3 5 E NO P 21810 3 5 E NO P 21822 3 6 E NO P 24986 3 7 E N 13,14 12 25 No p all pine and oak. Very little information in 15264- this survey, very brief 15305 3 8 E notes. 46247- No p, pine and pine- 46310 3 9 E oak stiff cane through the swamp, residue 2nd rate prairie land. Swamp overflow 1 to 3 feet timber oak gum hickory beach (sic) [at 45 he leaves Talla 49700 3 10 E E 16,21 45 76 80 Halla swamp and 144 enters p directly.]

Black soil, 2nd rate land, little or no timber 49713 3 10 E E 10,15 6 58 80 in the line. Prairie, 2nd rate land. Timber princepally post 49714 3 10 E N 9,10 0 80 80 oak 2nd rate land, mostly prairie. Timber post 49716 3 10 E E 3,10 20 60 80 oak and blackjack 49406 3 11 E NO P Before entering prairie hilly, timber described as described on last mile In prairie soil good a few scattering Black 12123 3 12 E E 3,T4R12.34 55 80 80 Jacks Gently rolling land ***** timber Black Jack, post 12123 3 12 E E 2,T4R12.35 0 30 80 oak ** pine ********* crosses asochattahiccah then atahomahiccah, but on Jasper07image looks like the same creek. Image shows this p patch as a small opening in the middle of a large expanse of 45736 3 12 E E 8,17 72 80 80 woods coming south between 8,9, does not mentin p but has no tree to mark corner. Land broken sandy soil growth pine post oak blackjack and 45741 3 12 E E 9,16 0 8 80 hickory first half mile undulationg good 2nd rate. In jasper07 grid this looks like a really 45741 3 12 E S 16,17 0 20 80 nice spot! 45751 3 12 E S 2,3 0 10 80 Land poor prairie interspersed with Black 45751 3 12 E S 2,3 38 74 80 Jack and Post Oak 45752 3 12 E E 2,11 3 10 80 Land undulating very 45752 3 12 E E 2,11 65 80 80 open Post Oak and 145 Black Jack woods land clay soil 45763 3 12 E N 11,12 20 35 80 Land undulating interspersed with pine and Post Oak Poor pararie cly soil no tree 45763 3 12 E N 11,12 60 80 80 to mark section 11 on 45764 3 12 E E 1,12 65 80 80 45765 3 12 E N 1,2 56 80 80 45767 3 12 E MAP with P Waving prairie, small ???? 2nd rate mostly 21341 3 13 E E 17,20 5 80 80 poor 3 14 E NO P 60636- 60674 3 15 E NA land undulating pararia soil black clay 2nd quality thinly timbered with post oak + Black 45760 3 12 E E 13,24 0 80 80 Jack Wavy land, very poor, blackjack, bushes, and prairie, 5-20-09: checked the survey page, moved this to 24472 4 1 W N 2,3 0 160 160 thinly timbered 4 1 E NA 21770 4 2 E NO P 4 3 E NA 4 3 E Not in fdn viewer 11234 4 4 E 11234 is nothing 4 5 E NA 4 5 E Not in fdn viewer 4 6 E NA 4 6 E Not in fdn viewer readable but on first glance seems pretty 7988 4 7 E swampy 9463-9574, 15737- 8643- 15430, 17256-17366: 8701 4 8 E NO P 46473 4 9 E E 6,7 15 40 80 46504 4 9 E E 23,26 36 54 80 Poor pine black+post oak woods - after leaving low ground and 12106 4 9 E E 5,T5R9.32 34 44 80 prairie

146 2nd mile of EBL of T4R10E 1st half mile as described on last- last half is prairie good 12124 4 10 E N 25,30 of R11 47 75 80 land 49740 4 10 E NO P prairie with scattered timber low and scrubby [4th mile EBL of 12128 4 11 E N 18,13 0 80 80 T4R11E] corner sections 2,3,10,11 are in the edge of the prairie. 1st half hilly poor sandy land, residue level 2nd rate land small timber 49844 4 11 E E 3,10 0 80 80 post oak and blackjack intersect at sections 2,3,10,11….soil 2nd rate land thinly timbered oak and 49845 4 11 E S 2,3 0 80 80 blackjack 40: raised mound 1/4 section in small prairie…..level 2nd rate land -- first half prairie residue small Black 49846 4 11 E E 2,11 0 80 80 Jack growth First half level 2nd rate soil oak + Black Jack Residue Poor sandy soil gently undulationg small timber Post oak + 49847 4 11 E S 10,11 0 5 80 Black Jack Poor sandy soil Black Jack bushes with wome small pine gently 49848 4 11 E E 11,14 20 35 80 undulating [no p marked on the line, but the corner of 13,14,23,24 is noted as: "the intersection in Prairie with regions (?) of timber through it"] First half Branch ridge poor land residue stoney hills thinly timbered post oak + 49860 4 11 E N 23,24 80 80 80 Black Jack [no p marked on the line] branch ridges…sandy soil Black Jack bushy with 49862 4 11 E N? 13,14 0 0 80 some groves of timber 147 through it

First half poor barrons Black Jack bushes residu undulating 2nd rate land oak and 49864 4 11 E N 11,12 15 54 80 Hickory 4 12 E NA 4 12 E Not in fdn viewer 21898 4 13 E No P, lots of swamp NO P swampy, pine, 22667 4 14 E marsh 24477 4 1 W N 3,4 0 160 160 thinly timbere post oak black jack and some 49861 4 11 E W 23,24 0 80 80 small pine This mile through timber but open barrens at a short distance on either side switch cane and bushes on the branch small timber post oak black jack and some 49863 4 11 E W 12,13 0 80 80 pine impossible to determine the direction traveled here. 2-5-09: changing to east direction, 2nd mile NBL of T4R1E per 8263 5 1 E E 5 mi SBL 57 88 180 point location scheme impossible to determine the direction traveled 8263 5 1 E E 6 mi SBL 50 70 180 here 8259- 8259-8278 are 8444, boundary lines. 8279 17938- begins T6R1E…17938 18045, is a typed 1938 survey, 49123- 49123 is just junk - a 49135 5 1 E dead end. 1st 1/2 level oak 29182 5 1 W N 13,14 80 160 160 hickory ** prairie land principally prairie of 2nd 29183 5 1 W E 12,13 0 50 160 quality rolling poor land 29183 5 1 W N 11,12 0 160 160 principally prairie [no 1/2 mile bearing 29183 5 1 W E 1,12 0 160 160 tree] prin open prairie expt br poor E all prairie soil 2nd [rate?] 1/2/sp [I think here 29187 5 1 W E 11,14 0 160 160 means half section 148 post, marks two hickories]

29188 5 1 W N 2,3 0 160 160 rolling poor prairie 21971 5 2 E N 13,18 103 155 160 there is a branch running through this, he crosses many times. At 103: "a heavy growth of cane briars and vines near the margins of the branch on each 21972 5 2 E N 12,7 11 160 160 side open prairie." 21972 5 2 E N 1,6 0 20 160 6th mile EBL open prairie wavy stiff 8333 5 2 E E 3,34 97 160 160 soil 8333 5 2 E E 2,35 0 34 160 8340 5 2 E N 17,18 13 46 160 8358 5 2 E W 4,9 87 138 160 8359 5 2 E N 3,4 92 105 160 8380 5 1 E W 30,31 118 133 160 8381 5 1 E W 19,30 137 160 160 8383 5 1 E N 17,18 62 85 160 8384 5 1 E W 7,18 110 160 160 29197 5 2 W NO P 48095- 48199 5 3 E NO P 5 4 E NA 5 4 E Not in fdn viewer 5 5 E NA 5 5 E Not in fdn viewer 7327- 7456 5 6 E read on 5-20-09: no p level land blackjack post oak + pine poor prairie soil 5-20-09: I am moving this to thinly timbered. It came out as 5th mile on the toby_pnts anyway, 7189 5 7 E N EBL mi 6 0 80 80 getting rid of it there. 8721 5 8 E W 22,27 35 76 80 land broken and good 8722 5 8 E E 15, 22 17 65 80 land level and good Land level and rich timber blackjack and 8723 5 8 E E 10,15 16 70 80 oak. 8724 5 8 E E 3,10 13 77 80 Land broken and rich land level and good, timber oak and black 8729 5 8 E S 26.27 1 62 80 jack

149 9575-9671: a 1940 survey, 15385-15933, 8702- 16926-16969, 63197- 8744 5 8 E 63261 1st half poor land gently undulating 46772- residue good land 46834 5 9 E E 29,32 54 80 80 principally blackjack. 1st ahlf mile level poor land blackjack. Residue gently undulating poor 5 9 E E 17,20 44 70 80 land. 1st half first rate land oak and hickory level. Residue gently undulating first rate 5 9 E E 8,17 32 55 80 land. first 1/2 2nd rate land gently undulating. Residue poor land gently undulating. 5 9 E S 8,9 9 40 Blackjack + oak. This line runs down or near the edge of the prairie and is poor land gently undulating blackjack and oak. 5- 20-09: put this in thinly 46792 5 9 E S 16,17 timbered poor land oak pine blackjack gently 5 9 E E 13,24 60 66 undulating. Blackjack prairie undulating timber 6650 5 10 E E 5,8 0 80 80 blackjack post oak +c Prairie soil undulating 6678 5 10 E S 22,23 0 80 80 timber oak blackjack clay soil gently undulating. Blackjack barren timber post oak 6694 5 10 E N 1,2, 0 80 80 blackjack +c except in prairie, land poor and level, timber 12138 5 10 E E 4,33 of 6 0 8 80 chiefly post oak + pine same 2nd rate land in prairie. Timber out of prairie post oak and 12139 5 10 E E 1,36 of 6 10 52 80 black jack chiefly. appears to be there, identified on 42383. Negative image scan, 42384- readable.42384-42412: 42412, all pine-oak. 45842:not 45942 5 11 E T5R11; 49931 is 150 readable and has "barrens"

49931-49994 difficult 49931- scan, seems to be no 49994 5 11 E P, on edge of belt 45940- 45978 5 12 E No P, map on 45978 7189 5 7 E N 1,6 0 80 80 This line runs down or near the edge of the prairie and is poor land gently undulating 46792 5 9 E S 16,17 0 80 80 blackjack and oak. 24735 6 1 W NO P first glance shows lots 6 1 E of prairie fdn finder says 49122, which is unreadable, and 67397, which is west of the Pearl, and 67404 is just a 6 1 E continuation of that 8279 begins EBL 8290 begins sections but very hard to read. It gets better, ends on 6 1 E 8329. 8279 6 1 E N 25,30 84 114 160 2nd mile EBL 1st 40 chains rich 1st rate prairie some undergrowth of hickory - -next 40 chains thin 8281 6 1 E N 13,18 0 82 160 open prairie at 57: cross woodland 8286 6 1 E E 3,34 57 64 160 to rich prairie 8287 6 1 E E 2,35 41 60 160 8287 6 1 E E 2,35 80 110 160 8293 6 1 E W 19,30 100 132 160 8295 6 1 E W 18,19 88 95 160 wavy mulatto stiff soil 8298 6 1 E W 7,18 130 160 160 8301 6 1 E N 20,21 122 148 160 8302 6 1 E N 16,17 95 160 160 8303 6 1 E W 8,17 0 10 160 8303 6 1 E W 8,17 110 146 160 rich wavy stiff soil 8304 6 1 E W 16,21 86 126 160 8306 6 1 E N 7,8 111 146 160 8308 6 1 E W 5,8 67 81 160 loamy black soil last 34 8308 6 1 E W 5,8 126 160 160 ch prairie

151 8309 6 1 E N 26,27 141 152 160 8312 6 1 E W 10,15 130 157 160 8313 6 1 E N 9,10 1 22 160 8319 6 1 E N 25,26 26 43 160 8322 6 1 E E 13,24 118 160 160 8323 6 1 E N 13,14 83 97 160 21975 6 2 E 12,7 85 100 160 5th mile EBL 24753 6 2 W E 11,14 prairie 4th mile EBL: soil black and covered with 21955 6 3 E N 13,18 110 160 160 rough grass 21955 6 3 E N 12,7 0 40 160 5th mile EBL 61082- 61082-61166 not in 61166 6 3 W files 65752- bad scan 90% 65775 6 4 E unreadable readable but BARELY. Some swamp. Some "level bushy land", but 65752- nothing worth 65807 6 4 E investigating gently rolling land soil thin all but the prairie which appears black and rich [p is mapped 7217 6 5 E N 1,6 34 70 80 on 7218] 7479 6 5 E N 1,2 128 153 160 [very washed out hard 6 5 E to read 7457-7480 Hilly poor pine oak + 7495 6 6 E W 14,23 45 60 80 black Undulating poor pine oak hickory and black 7504 6 6 E S 4,5 0 4 80 clay land (?) level + poor clay land oak black + pine prairie 7512 6 6 E S 5,6 15 35 80 land 3rd rate 26101 6 7 E E 29,32 70 80 80 Prairie 1st rate Poor clay land - pine oak + black. Corner of 28.29.32.33 Prairie 26107 6 7 E S 28,29 65 80 80 land 1st rate [at 10; out of preraria - land 1st rate] Gently undulating + poor clay 26108 6 7 E E 28,33 0 10 80 land pine oak + black Praria 2nd rate balance 26108 6 7 E S 32,33 53 80 80 hilly + poor 26117 6 7 E E 23,25 40 50 80 Hilly poor land Hilly 3rd rate pine oak 26121 6 7 E N 23,24 45 57 80 black hickory 26127 6 7 E MAP 152 Chiefly low cane swamp to the prairie thence open all poor and of no value for 7229 6 8 E E 3,34 52 80 80 cultivation 19684- 19811 6 8 E No P 11714- 11750 6 9 E 1947 11714- 11750; 46647- 707 6 9 E 1947 clay soil poor blackjack land somewhat hilly timber post oak 6724 6 10 E E 28,33 0 80 80 blackjack pine +c clay soil poor blackjack barrens timber post oak 6725 6 10 E N 33,34 0 80 80 blackjack pine +c 50059- No P, lots of reed-open 50122 6 11 E woods-pine. Soggy. 1st 1/2 mile thin 3 rate residue rich wavy interspersed with prairies undergrowth 8260 6 1 E N 24, 19 80 160 160 hickory 1st rate Hky and Bjack interspersed with prairies - Loamy black wavy principal growth 8307 6 1 E N 5,6 0 160 160 Hky half section post:large red oak in edge of prairie, this mile very open soil lively red oak hickory and some black jack - partly prairie from 24756 6 2 W N 14,15 0 160 160 the branch. this is parlty prairie and 24757 6 2 W E 11,14 0 160 160 the rest very open this mile is partly narrow prairies, res O, 24767 6 2 W N 4,5 0 160 160 BJ, PO and + H 29509-29585: these show up in the finder but are skipped in the disc, they are out of 26916 7 1 E sequence or not there. Level 2nd prairie land with hickory and 29301 7 1 W N 35,36 0 160 160 blackjack Level 2nd prairie land 29301 7 1 W E 25,36 0 160 160 with hickory and 153 blackjack

level 2nd rate oak 29304 7 1 W E 24,35 70 100 160 hickory black Level prairie land 29305 7 1 W N 14,15 0 160 160 blackjack hickory Level poor Land black 29306 7 1 W N 10,11 50 146 160 post oak +c Rolling prairie land 2nd 29307 7 1 W N 27,28 0 160 160 qulty Level prairie with Hky 29308 7 1 W E 15,22 0 160 160 Bjack 26916: one page thrown in with random single pages from other 26916 surveys. Rich wavy 29509- level land, NBL, no 85 7 1 E mention of p principally blackjack + 7 1 W N 3,4 0 160 160 prairie 21992 7 2 E E 6,31 115 133 160 21996 7 3 E NBL NO P 65790 7 4 E N 2,3 150 157 160 65794 7 4 E E 1,12 30 118 160 65798 7 4 E E 4th mi on NB 158 160 160 65798 7 4 E E 5th mi on NB 0 110 160 not clear where he enters p but leaves at 64230 7 5 E N 17,18 100 119 160 119 64230 7 5 E N 7,8 8 63 160 no bearing trees on 64234 7 5 E W 5,8 110 160 160 corner Land wavy timber post 64234 7 5 E N 4,5 0 32 160 oak and blackjack land level timber post oak black ash and 64237 7 5 E W 9,16 130 160 160 blackjack at 110 note: "enter peraria @ north" and at 160 "departed to the south 20 L, land wavy timber ?? Ash black 64238 7 5 E W 4,9 110 160 160 ash blackjack 3te" at 151: "enter peraria @ SW" Land wavy timb poste ash + 64240 7 5 E N 22,23 151 160 160 Blackjack 3rd" 64240 7 5 E N 15,22 52 94 160 first 92 chains wavy peraria residue ???? 64241 7 5 E N 10,11 0 92 160 Black ash poste oak…

154 Unsure of where he left the p but it was before 65242 7 5 E N 35,36 0 19 160 the half mile. Land wavy timber poste oak black ash blackjack 65244 7 5 E E 14,23 0 40 160 3rd rate Land wavy timber poste 65244 7 5 E E 11,14 0 49 160 oak black oak 3rd rate 35197-35418 is actually the 1805 line after Doak's Stand. 64211- 64226 is the sliver of 35197- T7R5E that is in 35418 7 5 E Dancing Rabbit Creek. Peraria 3rd rate 7594 7 6 E W 28,33 42 80 80 balance level+ poor No timber near. Undulating 3rd rate 7594 7 6 E N 32,33 0 80 80 perarie (prrarie?) 7595 7 6 E W 29,32 0 80 80 3rd rate peraria level + poor pine oak 7595 7 6 E N 28,29 0 30 80 black Prairie 3rd rate note that at corner he enters a p that runs the line between 19 and 20. "Undulating + poor 7595 7 6 E W 20,29 80 80 80 Oak, Black, pine +c" wavering + poor pine 7599 7 6 E S 17,18 74 80 80 oak + Bjack Prairie 3rd rate balance 7600 7 6 E W 18,19 28 80 80 level pine oak etc. 7600 7 6 E S 19,20 0 5 80 wavering + poor pine 7600 7 6 E S 19,20 70 80 80 oak + Bjack prairie + poor pine oak 7600 7 6 E W 19,30 76 80 80 + Bjack 7601 7 6 E S 29,30 0 80 80 3rd rate prairie wavering + poor pine oak + Bjack } prairie 3rd 7601 7 6 E W 30,31 62 80 80 rate 7601 7 6 E S 31,32 50 80 80 Prairie 3rd rate 7603 7 6 E MAP of T7R6E 7526-7586 is a 1947 7526- survey. 1833 survey 7605 7 6 E starts at 7580. 18263- No p but a MAP on 18304 7 7 E 18300 19812- No p but a MAP on 19881 7 8 E 19879 46969- No p but a map on 47041 7 9 E 47040 9121- 9121 not in sequence 9186 7 10 E or missing 155 21993 7 2 E E 4,33 0 160 160 21995 7 2 E E 1,36 0 160 160 one page, no 29436 8 1 W information 29436 8 1 W one page, unreadable 8 1 E NA priarie woodland thinly timbered with oak and hickory soil 2nd quality yellow and clayey 65646 8 2 E N 1st mi on EB 0 160 160 surface wavy prairie surface wavy soil 2nd quality [I think this whole mile is p, the half mile post only took 65646 8 2 E N 2nd mi on EB 0 160 160 one tree, an oak.] land 2nd quality timber 65646 8 2 E N 3rd mi on EB 0 70 160 oak and hickory into prairie, thru prairie 65682 8 2 E N 7,8 60 105 160 to woodland to prairie, thru prairie to 65685 8 2 E N 16,17 100 140 160 woodland thru woodland to prairie, thru prairie to 65686 8 2 E W 8,17 20 131 160 woodland thru woodland to prairie, thru prairie to 65686 8 2 E N 8,9 10 55 160 woodland 65686 8 2 E W 5,8 65 80 160 to prairie 15 chs + surface wavy and interspersed with small 65687 8 2 E N 4,5 0 160 160 pieces of prairie at 60 chains to prairie 65688 8 2 E N 27,28 60 80 160 20 chains across 65689 8 2 E W 16,21 20 88 160 thru woodland to prairie at 7 chains to prairie 15 65690 8 2 E W 4,9 7 22 160 chains X at 66 to prairie 20 chanis X Land waving soil 2nd quality timbered thinly with scrubby oak and 65690 8 2 E W 4,9 66 86 160 hickory thinly timbered with oak 65691 8 2 E W 27,34 15 100 160 and hickory surface wavy soil 2nd quality the prairie interspersed with some scattering [trees?] of 65691 8 2 E N 26,27 30 145 160 scrubby oak land level soil poor 65694 8 2 E N 2,3 120 140 160 timber oak and hickory

156 surface wavy soil 2nd quality woodland thinly timbered with scrubby 65695 8 2 E W 35,26 20 60 160 oak and hickory surface wavy soil 2nd quality woodland thinly timbered with scrubby 65695 8 2 E W 35,26 82 150 160 oak and hickory surface wavy soil 2nd quality timber hickory + 65695 8 2 E E 25,36 20 60 160 oak surface wavy soil 2nd quality woodland thinly timbered with scrubby 65696 8 2 E W 23,26 140 148 160 oak and hickory thru prairie with some scattering scrubby oak 65696 8 2 E E 24,25 60 154 160 hickory 65700 8 2 E 1 0 MAP T8R2E prairie with some scattering groves of scrubby oak and 65653 8 3 E E? 30,31 70 160 160 hickory 15: to prairie thru woodland; 160: thru prairie interspersed w/ small groves of oak and 65654 8 3 E E? 19, 30 15 160 160 hickory 20: thru woodland to prairie; 160: surface wavy prairies interspersed w/ some groves of oak and 65658 8 3 E N 20,21 20 160 160 hickory thru prairie interspersed w/ small groves of oak and hickory to 65658 8 3 E W 17,20 0 140 160 woodland 65659 8 3 E N 16,17 0 74 160 thru prairie to woodland 75: to prairie thru woodland, 100: + 65661 8 3 E N 4,5 75 100 160 prairie to a pond thru woodland to prairie, thru prairie to 65662 8 3 E W 16,21 60 115 160 woodland …..prairie, at corner thru prairie surface 65663 8 3 E W 16,21 141 160 160 wavy prairie, at corner thru prairie w/ some scatterings of scrubby 65663 8 3 E N 15,16 6 160 160 oaks and hickory at 80 thru prairie to 65663 8 3 E W 9,16 0 80 160 woodland 157 ???? - at corner:thru prairie interspersed with groves of hickory 65664 8 3 E N 9,10 0 160 160 and oak at 133 enter woodland thru prairie interspersed w/ groves of oak and 65664 8 3 E N 3,4 0 133 160 hickory at 159.25 (corner) surface wavy 2nd quality of prairies interspersed w/ some groves of oak and 65667 8 3 E W 10,15 0 160 160 hickory 65667 8 3 E N 10,11 0 160 160 thru prairie to woodland 65673 8 3 E W 2,11 80 160 160 thru woodland to prairie surface wavy prairie interspersed with small groves of oak and 65673 8 3 E N 1,2 41 160 160 hickory 8 4 E NA 8 4 E Not in fdn viewer wavy 2nd rate land, 65886 8 5 E E NBL 2d mi 13 30 80 timber oak & hickory &c 65297- 65320 8 5 E 65883-65890 8 5 W NA 8 5 W Not in fdn viewer 7606 8 6 E NO P 50201- 50336 8 6 W No P 18874- 18998 8 7 E a 1947 survey 18874- 18998; 33688- 33691 8 7 E 1947 8 7 W NA 8 7 W Not in fdn viewer 19986- 20063 8 8 E No P but map on 20053 8 8 W NA 8 8 W Not in fdn viewer 1st half mile mostly open prairie - wavy - stiff soil. Residue wavy stiff - Post oak B Jack 939 9 1 W N EBL 1st mi 0 160 160 and Hky. Open woods Open wavy woods thin 939 9 1 W N EBL 2nd mi 0 160 160 stiff soil Post oak black jack hky 940 9 1 W N EBL 3rd mi 0 160 160 - wavy stiff poor open 158 woods

1st hafl mile mostly rich caney flat land. Ash Oak Hky +c - Residue wavy open Bjack, post Oak and Hky woods - 940 9 1 W N EBL 4th mi 68 160 160 3rd stiff soil- Post oak, Roak, Hky - wavy stiff. 2nd quality 940 9 1 W N EBL 5th mi 0 160 160 open woods [at 45 notes "+ wavy open to low thick juniper swamp" at 80 his bearing trees are juniper and willow. This is the only reference to juniper I 940 9 1 W N EBL 6th mi 0 45 160 have seen.] mostly flat stiff thin surface - redok, Hky + 947 9 1 E E NBL 4th mi 52 78 160 postok 1st half mile mostly 2nd rate wavy red ok hky +c - the creek bottom is cut by small ponds - last 1/2 mile wavy open woods thin soil - post 949 9 1 E E NBL 6th mi 80 160 160 ok, R ok and Hky 21614 9 1 E N EBL 1st mi 140 160 160 surface wavy soil poor timbered thinly with 21614 9 1 E N EBL 2nd mi 0 40 160 scrubby oak surface wavy soil 2nd quality timber oak and 21644 9 2 E N 28,29 15 150 160 hickory 21644 9 2 E W 20,29 10 30 160 21645 9 2 E N 20,21 60 75 160 surface wavy soil poor 21648 9 2 E N 27,28 125 155 160 timer oak and hickory land level soil poor and thinly timbered with oak 21651 9 2 E N 26,27 5 40 160 and hickory MAP 21681. Lots of p indications are in GE. Only one actual P at 21661 9 3 E W 19,30 115 155 160 21676. 21661-21861 21676 9 3 E E 25,26 70 145 160 65709 9 4 E W 20,29 45 80 80 53264- 53264-53273:WBL no 53352 9 5 W P 58816- 58823 9 5 E NBL only: no P 159 added this on 5-15-09 because it is close to the Jbelt - 5-19-09: not 9 6 E in fdn viewer Section Survey not available. 53274- 53279: WBL, NBL, no 9 6 W P 9 6 W Not in fdn viewer 50786 9 7 W 50786-50821 no prairie 50822 9 8 W 50822-50857 no prairie 51367- 51473 9 9 W 51367-51372 NBL no P no direct mention but "surface wavy and thinly timbered with 21642 9 2 E N 17,18 0 160 160 scrubby oak" "surface wavy and interspersed with small pieces of prairie, soil 2nd quality timber oak 21643 9 2 E N 32,33 0 160 160 hickory. surface wavy and interspersed with small pieces of prairie 1/2 soil 21645 9 2 E W 17,20 0 160 160 2nd quality 1/2 soil poor 21645 9 2 E N 17,16 0 160 160 surface wavy soil 2nd quality and intersperese with small 21647 9 2 E N 33,34 0 160 160 pieces of prairie land wavy soil poor and thinly timbered withh scrubby oak and 21647 9 2 E W 28,33 0 160 160 hickory surface wavy soil 2nd quality and interspersed with small 21651 9 2 E W 27,34 0 160 160 pieces of prairie land wavy soil 2nd quality and thinly timbered withh scrubby 21655 9 2 E W 26,35 0 160 160 oak and hickory surface wavy soil poor and thinly timbered with 21658 9 2 E E 12,13 0 160 160 oak and hickory 21662 9 3 E W 19,30 0 160 160 21663 9 3 E N 17,18 0 160 160 21663 9 3 E N 7,8 0 160 160 21663 9 3 E N 11,12 0 160 160 21663 9 3 E N 7,8 0 160 160 21667 9 3 E N 8,9 0 160 160

160 21673 9 3 E N 22,23 0 160 160 21674 9 3 E N 10,11 0 160 160 21676 9 3 E N 25,26 0 160 160 21676 9 3 E W 26,35 0 160 160 21677 9 3 E N 23,24 0 160 160 21677 9 3 E E 24,25 0 160 160 21677 9 3 E W 23,26 0 160 160 116: swamp to open woods 140: open woods to prairie, 157: prairie to open woods "last 44 chains wavy - redok, Hky and Bjack - 952 10 1 W N EBL 2nd mi 116 160 160 2nd rate stiff soil. 132: open woods to level cane land. 1st 132 chains open wavy stiff thin soil - post ok, Rok, Blackjack and Hky.- Residu rich surface growth white = redok ,Ash, elm, Dogwood, Hky +c. 952 10 1 W N EBL 3rd mi 0 132 160 mostly level 21: +level cane land to open level wood, 70: + open wood to prairie, 156: to thick briery and caney swamp Mostly 952 10 1 W N EBL 4th mi 21 156 160 level stiff soil …residue wavy open woods - thin soil - redok 954 10 1 W N EBL 5th mi 133 160 160 Hky + post ok, stiff soil. first 103 chains open woods - mostly flat stiff thin soil - Rok, postok, Hky +c - Residue mostly swampy switch 955 10 1 W N EBL 6th mi 0 103 160 cane…. 140: + bushy to flat open…….last 20 chain 957 10 1 E E NBL 1st mi 140 160 160 open prairie at 55: + prairie to open flat woods, at 83 + branch stand in holes. Final note: Bjack, post ok + Hky - mostly wavy 958 10 1 E E NBL 2nd mi 0 83 160 stiff thin soil …residu wavy thin Bjack, Hky and post ok 959 10 1 E E NBL 3rd mi 55 160 160 - stiff soil 959 10 1 E E NBL 4th mi 0 160 160 In open wavy wood

161 Bjack, Hky and post ok - wavy open wood - stiff 960 10 1 E E NBL 5th mi 0 160 160 soil - 3rd quality 0f the 1st half mile, the 1st 30 ch are flat post 961 10 1 E E NBL 6th mi 0 30 160 ok and Rok. …residue mostly thin postok, hky + Bjack - 961 10 1 E E NBL 6th mi 98 160 160 wavy stiff soil - at 9 chains to open prairie, at 25 chains + 63361 10 1 E N 10,11 9 25 80 (cross) prairie at 10.5 chains to open rich prairie, corner in 63368 10 1 E W 10,15 11 80 80 prairie 63369 10 1 E N 9,10 0 1 80 at 1.0 chains left prairie at 37 chains to open prairie, at 60.5 + (cross) 63371 10 1 E N 3,4 37 61 80 prairie corner in open prairie, at 1.0 chains + prairie 63376 10 1 E W 9,16 0 1 80 to open woods at 70 chains to prairie, at corner west margin 63377 10 1 E W 8,17 70 80 80 of prairie at 32.5 to an open prairie, at 50 to scrubby 63381 10 1 E N 17,18 33 50 80 blackjack at 76 to prairie, corner 63381 10 1 E N 17,18 76 76 80 in prairie (80.45) at 70 chains to open prairie, at corner (80.00) in prairie - moderately wavy w/ bjack, hky, post oak w/ heavy grass - last 10 63384 10 1 E W 5,8 70 70 80 open prairie at 22.5 to open rich prairie, at 30 + prairie to 63385 10 1 E N 7,8 23 30 80 open woods at 35 to a large prairie - " last 45 chains entirely 63385 10 1 E N 7,8 35 80 80 open prairie" 63386 10 1 E N 5,6 0 12 80 at 11.5 + (cross) prairie at 5 chains + prairie to 63387 10 1 E W 6,7 0 5 80 brushy woods at 39 to open level prairie poor, at 48.5 to 63389 10 1 E W 18,19 39 49 80 rich level woods 21639 10 2 E N 13,18 151 160 160 4th mile EBL 21639 10 2 E N 12,7 0 60 160 5th mile EBL 27709, 27709: in the middle of 67174- 10 2 E a clear but unidentified 162 67322 survey

the following two pages consist of a long note about how messed up 67184 10 2 E W 12,13 0 70 80 his measurements are. Level poor land, oak + 10267 10 3 E E 12,13 30 53 80 Black Jack principally level poor 10268 10 3 E N 11,12 17 75 80 prairie 10273 10 3 E MAP level poor post oak + black jack woods + 10226 10 4 E N 19,20 69 80 80 prairie first part prairie residue level poor post oak + 10227 10 4 E W 18,19 0 24 80 black jack first part prairie residue 10227 10 4 E N 17,18 0 19 80 low swamp 21 chains low swamp residue level oak+ black jack woods+ 10231 10 4 E W 17,20 70 80 80 prairie First 5 chains swamp, residue wavy oak + black jack woods + 10233 10 4 E N 8,9 39 80 80 prairie Level oak + black jack 10233 10 4 E W 5,8 1 20 80 + prairie Level poor Black Jack 10305 10 4 E E 5,32 (of T11) 77 80 80 land. Rolling poor Black Jack 10306 10 4 E E 4,33 (of T11) 0 5 8 land to prairie, corner in 64915 10 4 E S 5,6 77 80 80 prairie 64915 10 4 E S 4,5 0 3 80 at 3 chains X the prairie 64915 10 4 E E NB 3rd mile 0 5 80 at 5 chains X the prairie 53280-53282 WBL no 57811- P; 57811-57847 57847 10 5 W sections, no P 57818- 57847 10 5 W no P [at 23 notes "blackjack prairie soil middling good" the end note is "land middling good timber pine + 50339 10 6 W E 24,25 23 80 80 blackjack"] [at 60 notes "blackjack prairie soil middling" It is not clear if he is noting p at that point or 50340 10 6 W N 23,24 60 80 80 entering p that runs to 163 the end]

[again not really clear where p begins or ends. Also this and the previous 2 mark pines and oaks as bearing trees. At 35 marks "unlevel prairie land timber pine oak" and at 43 marks "rich prairie 50340 10 6 W E 13,24 35 80 80 land" prairie, soil 2nd rate, 50343 10 6 W E 14,23 15 80 80 timber pine + blackjack prairie land middling 50344 10 6 W N 15,14 17 80 80 timber blackjack + pine [at 40 notes "middling prairie (or middling rich) land"] land poor timber 50347 10 6 W N 32,33 40 63 80 pine oak Prairie land rich best 50355 10 6 W N 7,8 60 80 80 2nd rate [ at 12.65 notes: "Chocktaw boundary at the margin of the 50356 10 6 W W 6,7 0 2 80 praire" [ at 3.40 notes "the chocktaw boundary, in an extensive prairie. No bearing tree in a reasonable distance" It is unclear where the p is on this mile, but it is 50356 10 6 W N 5,6 1 34 80 there. land of best 2nd quality [at 35 notes "margin of the great prairie 2nd rate land" there is no 50356 10 6 W W 5,8 35 80 80 tree at corner 6,8. [at 40 notes "land 2nd rate blackjack prairie tinmber oakpine blackjack" So the quarter section monument is in p, the bearing trees are blackjacks, but we don't know the size of the p. At the section corner the bearing trees are pine and the note is "land 3rd rate timber 50357 10 6 W N 9,10 40 40 80 oak pine"] 164 [surveyor is moving north from the old Choctaw boundary, that is, the Wayne County Clarke county line, not the section corner. Also this is a longer line because of the distortion of the townships just south of the new basis meridian. Also note made that this p gets wider toward 50380 10 6 W N 5,6 1 34 80 the east thus: s -A- n 2nd rate land, mostly prairie. Timber post 50383 10 6 W W 6,7 0 2 80 oak and blackjack land broken and poor pine and oak bushes 50403 10 7 W N 1,2 66 107 80 and grass [sections are generally 80 chains in this township. This township is a little distorted, doesn't line up with the one to the north. Only one bearing tree on corner] Land broken + poor 50507 10 7 W N 1,2 66 107 80 pine+ oak 10 8 W NA 10 8 W Not in fdn viewer 51391-51399: sections, 10 9 W no P 51420-51443: sections, 10 10 W piney woods, no P Level 2nd rate prairie and post oak + black 10304 11 4 E E 3,34 67 80 80 jack land 1st 113 chains mostly thin wavy stiff soil..post ok Bjack Hky and 962 11 1 E N WBL 1st mi 0 113 160 RedOk. Black, Post Ok Redok and Hky. Wavy stiff 963 11 1 E N WBL 2nd mi 0 160 160 soil. Open wood. at 55 + wavy ope to rich 963 11 1 E N WBL 3rd mi 0 55 160 level switch cane land

165 1st 1/2 mile mostly stiff 964 11 1 E N WBL 4th mi 0 93 160 prairie wavy nd generally light soil 3rd quality - open 964 11 1 E N WBL 4th mi 100 160 160 woods at 110 + cain to level bushy woods; at 130 + level to wavy open woods; at end: Bjack, Hky, post ok and Redok - mostly wavy 970 11 1 E E NBL 5th mile 130 160 160 thin soil - stiff surface Bjack Hky Red Ok + Post Ok. 1st 140 chains mostly wavy stiff thin soil - residue flat 970 11 1 E E NBL 6th mile 140 160 160 open woods sw corner in priaire, at 13.50 + (cross) the 63295 11 1 E N 15,16 0 14 80 prairie corner in open prairie, 63305 11 1 E W 16,21 0 7 80 at 6.5 + the prairie at 34 to open prairie, at 50 + the prairie, cheifly moderately wavy open 63305 11 1 E W 16,21 34 50 80 blackjack at 1 chain X o. prairie to 63311 11 1 E N 31,32 0 1 80 flat poor wood at 55 chains to rich open prairie, at 64 to 63319 11 1 E W 19,30 55 64 80 rich woods 11 2 E NA Land red oak + Hickory 10281 11 3 E E 3,34 (of 12) 58 80 80 land level poor prairies 38 chains poor level prairie residue rolling 10282 11 3 E E 2,35 (of 12) 0 38 80 poor Blk Jack land Level poor Black Jack 10288 11 3 E W 16,21 28 45 80 land. South half level poor Black Jack + prairie N. half level good red oak 10289 11 3 E N 9,10 5 30 80 land. Level 2nd rate post oak + hickory and poor 10294 11 3 E N 2,3 42 79 79 prairie 10302 11 3 E MAP at 38 chains cross 64901 11 3 E E NB 5th mile 38 38 80 prairie NE and SW corner in prairie, level poor black jack and 64904 11 3 E E 18,19 56 80 80 post oak 64904 11 3 E N 19,20 63 80 80 enter prairie, corner in 166 prairie X the prairie, [scan too light to read, but some 64905 11 3 E E 16,21 28 45 80 p on this line there is] to prairie, corner in 64906 11 3 E N 2,3 42 80 80 prairie level poor Blk Jack + Rd Oak Land and 10324 11 4 E N 2,3 45 80 80 Prairie enter prairie, corner in 64913 11 4 E E NB 4th mile 67 80 80 prairie woodland at 40 chains set post in east margin of prairie - west 1/2 mile good 64914 11 4 E E NB 5th mile 0 40 80 prairie land 57811- small wedge shape, no 57817 11 5 W P 36006 12 1 W NO P 12 2 E NA Got kicked out because it is too far west of 12 2 W Yazoo River 12 2 E Not in fdn viewer at 19 chains X the 64962 12 3 E N 34,35 0 19 80 prairie Checked, nothing to 12 4 E 0 0 80 report

167

APPENDIX B

GROUND-TRUTHING NOTES

168

A = On the MNHP inventory. B = A recognizable relict prairie. C = Not an intact relict, but some indicator species are present. D = Intensively used or managed: houses, row crops, mowing, pine plantation, etc, to the degree that any natural prairie would be highly degraded. E = No management apparent, wild, weedy areas that exhibit no indicator species and are indistinguishable from the surrounding landscape matrix.

Site A B C D E Note County 1 x Hardwoods, pine plantation Madison 2 x Pine plantation, silphium, gaura Madison 3 x Hardwoods Madison 4 x Corn, soybeans, woodland Madison 5 x Pine plantation. Persimmon Madison 6 x cedar glade, pine plantation Madison 7 x cedar Madison 8 x hay field Madison 9 x hay field, Asclepius tuberosa Madison 10 x residential neighborhood Madison 11 x old field, soybeans, pine plantation Madison 12 x old field Madison 13 x Madison 14 x Pasture Madison 15 x residential / Ag land Madison 16 x Madison 17 x pasture Madison 18 x row crops, cul‐de‐sacs Madison 19 x McMansions Madison 20 x Hinds Natchez trace (new) Rhamnus, Desmanthus, 21 x Persimmon Hinds 22 x Pine plantation and lake Hinds 23 x pines Hinds 24 x x x Woodland, many indicators Smith 25 x Private land, no access Jasper 26 x Private land, no access Jasper 27 x Large pine plantation Jasper 28 x x HWY 15 ROW. Wet prairie? Jasper 29 x Power line ROW, exotics Jasper 30 x Pine, hayfields, no indicators Jasper

169 31 x Old field, many indicators Jasper 32 x Many indicators Jasper 33 x Hardwoods and weeds Jasper 34 x x Silphium, liatris, lots of cedar Jasper 35 x Private, no access, oil well, exotics Jasper 36 x Natural gas pipline. Good prairie. Jasper 37 x Church. Mixed Hardwood, vines, red clay Clarke 38 x Intensive ag use Clarke 39 x Mix hardwoods and common weeds Clarke 40 Pine, hardwoods, wildflowers Wayne 41 x Good relict prairie Wayne 42 x Cut over, red clay, common weeds Wayne 43 x Intensive ag use, indicators along road Rankin 44 x Weedy, no indicators Rankin 45 x Cornfield Rankin 46 x Corn and soybeans Rankin 47 x Ag land, some indicators, panicum Rankin 48 x turfgrass mowed to roadside Rankin 49 x soybeans Rankin 50 x post oak, elm, corn Rankin 51 x #1 in top sites, road ROW unmowed Scott 52 x old number 2 in top sites: ag land Scott 53 x #2 in top sites, corn field Scott 54 x x #3 in top sites, cell phone tower Scott #4 in top sites, USFS forest. The only high score in 55 x BNF that is not on MNHP inventory Scott 56 x #5 in top sites, USFS forest Scott 57 x x Found during veg survey Scott 58 x Found during veg survey: new discovery Scott TOTAL 5 9 8 29 12

Number of sites lost to development or intensive management: 29 (50%) Of the remaining 29 sites: 12 (41%) show no sign of prairie. 8 (28%) show indicator species. 3 are on unmowed road ROWs, 1 is in USFS forest land. 9 (31%) are recognizable prairie relicts: cedar, prairie grasses and forbs covering a large area.

170