Assessing the Pre-modern Tree Cover of the Ackerman
Unit, Tombigbee National Forest, North Central Hills, MS,
Using GLO Survey Notes and Archaeological Data
Evan Peacock, John Rodgers, Kevin Bruce, and Jessica Gray
Southeastern Naturalist, Volume 7, Number 2 (2008): 245–266
Full-text pdf (Accessible only to subscribers.To subscribe click here.)
Site by Bennett Web & Design Co.
2008 SOUTHEASTERN NATURALIST 7(2):245–266
Assessing the Pre-modern Tree Cover of the Ackerman
Unit, Tombigbee National Forest, North Central Hills, MS,
Using GLO Survey Notes and Archaeological Data
Evan Peacock1,*, John Rodgers2, Kevin Bruce3, and Jessica Gray4
Abstract - General Land Office (GLO) survey data from the Ackerman Unit of
the Tombigbee National Forest, MS are used to characterize early 19th-century tree
cover in a part of the North Central Hills physiographic province. Archaeological
settlement-pattern data indicate that the area was abandoned ca. A.D. 1000 and that
early Historic-period settlement was minimal by the time the GLO surveys were
done. The GLO data therefore represent forest conditions as they developed in the
absence or near-absence of human infl uence. Tree cover consisted of oak-dominated
hardwoods with a non-clustered pine component. The distributions of some hardwood
species were related to geological controls. Based on previous archaeological
work, the argument concerning minimal human impact can be extended to the entire
North Central Hills province, with consequent implications for forest management
on federal lands.
General Land Office (GLO) survey notes have long been a standard data
source for scientists interested in landscape conditions as they existed prior
to extensive modern impacts. One criticism levied against studies employing
GLO data is that researchers were paying insufficient attention to the
residual effects of American Indian land-management practices on early
19th-century forest cover (Peacock 1998), a practice exemplified by the
use of the term “presettlement” (e.g., Anderson and Anderson 1975). This
practice continues today (e.g., Bragg 2003, Brewer 2001, Farley et al. 2002,
Perkins and Matlack 2002), despite a number of studies that combine GLO
data with ethnohistorical and/or archaeological data to clearly demonstrate
the infl uence of American Indian populations on the landscape. For example,
Foster et al. (2004) recently employed catchment analysis of GLO data to
show vegetation differences between early Historic-period Creek Indian
town locales and uninhabited control areas in east-central Alabama. GLO
data also have been used to document late-prehistoric human environmental
impact in Arkansas, where wood was intensively exploited as fuel for boiling
off salt brine (Williams 1993).
Importantly, however, archaeological data also can be used to argue that
American Indian land use was not a significant factor in early 19th-century
1Cobb Institute of Archaeology, PO Box AR, Mississippi State University, 39762.
2Department of Geosciences, PO Box 5548, Mississippi State University, 39762.
3Marcell/Deer River Ranger District, Chippewa National Forest, PO Box 308, 1037
Division Street, Deer River, MN 56636. 4109 Old Farm Road,Perry, GA 31069. *Corresponding
author - firstname.lastname@example.org.
246 Southeastern Naturalist Vol.7, No. 2
forest composition, if those data indicate a lack of human occupation in an
area for a substantial period of time prior to the GLO surveys. In this article,
we present evidence that land now comprising the Ackerman Unit of the
Tombigbee National Forest, north-central Mississippi, was abandoned by
American Indians centuries before the GLO surveys were undertaken, and
that Historic-period reoccupation of the area was minimal until the mid-19th
century. We argue that the GLO data therefore represent unmodified to minimally
modified forest conditions, the major control for which seems to have
The Ackerman Unit of the Tombigbee National Forest includes approximately
40,000 acres of federally controlled land lying almost entirely
within the North Central Hills physiographic province of central Mississippi
(Fig. 1), with the eastern tip extending into the Flatwoods. The province
encompasses a broad, highly dissected belt of uplands, with a dendritic
drainage pattern feeding generally small streams. The largest waterway
crossing the Ackerman Unit, the Noxubee River, has its headwaters in
the northwestern part of the unit and eventually flows into the Tombigbee
River at Gainesville, AL. Geologically, the North Central Hills consist of
Tertiary sands, sandstones, clays, shales, and silts. The sand members are
mostly non-marine in origin and were deposited during the Upper Paleocene
and Middle Eocene coincident with sea-level trangressions and consequent
landscape aggradation. A more detailed discussion of area geology
can be found in Peacock and Fant (2002; see also Mellen and McCutcheon
1939, Vestal and McCutcheon 1943).
The area is home to a rich prehistoric record, although, as discussed in
detail below, American Indian settlement varied in intensity through time
(Peacock 1997). Historic settlement in the area began in the 19th century,
aided in part by the construction in the 1820s of Robinson Road, a post
route that ran through the Ackerman Unit area (Phelps 1950, Rafferty 1979).
Access to the area was facilitated by the Treaty of Dancing Rabbit Creek
in 1830, in which the land was ceded by the Choctaw Indians. Subsequent
settlement consisted primarily of small farmsteads around which cotton and
corn agriculture was practiced, with pigs being the major livestock (Moore
1988). With the coming of railroads around the beginning of the 20th century,
logging increased in importance in the local economy (Adkins 1979, Carroll
1983, Coleman 1973) as Mississippi became a national leader in lumbering
(Fickle 2001). Lands acquired by the government in the 1930s and early
1940s were administered by the Soil Conservation Service until control was
passed to the US Department of Agriculture in 1959, when the Tombigbee
National Forest was created. Since that time, the land has been managed
primarily for timber and game, with pine being maintained as the species of
primary economic importance.
2008 E. Peacock, J. Rodgers, K. Bruce, and J. Gray 247
Archaeological site ages and distributions
Data on the distribution and age of archaeological sites on the Ackerman
Unit have been gathered over a number of years via intensive shovel-test and
open-ground surveys designed to locate all sites in proposed impact areas.
These surveys have largely focused on ridge tops prior to logging, but a
number have been undertaken on other landforms, such as first terraces and
bottomlands, with one such project being designed for the sole purpose of
evaluating the survey methods (Peacock 1995a, 1996). These surveys have
resulted in several hundred prehistoric and Historic-period sites being recorded
on the Ackerman Unit. Approximate ages were assigned to individual
Figure 1. Location
A c k e r m a n
248 Southeastern Naturalist Vol.7, No. 2
occupations based on the presence of diagnostic artifacts (see Peacock 1995a
for details). In the case of prehistoric sites, these included projectile point
and pottery types with known temporal spans. In the case of Historic-period
(post-1700 A.D.) occupations, artifacts such as decorated ceramics and nails
were used, along with bottle glass color and form.
Written records, beginning with the GLO notes of 1832–1833 and including
the federal government acquisition records dating to the 1930s and
early 1940s (Peacock and Patrick 1997), were accessed as complementary
sources for site location and age determinations. Although GLO data have
been employed in charting Historic-period landscape use (e.g., Silbernagel
et al. 1997), maps and written records alone are insufficient for accurate
assessments of Historic-period site frequencies. Data from written sources
should be combined with archaeological survey data, as we have done here,
to insure that the full range and a more accurate number of sites are represented
(e.g., Peacock and Patrick 1997).
All site locations were entered into a Geographic Information System (GIS)
for display at a scale appropriate for the purposes of this paper, but which does
not compromise site security. Information on all sites on the Tombigbee National
Forest is available to qualified researchers through the US Forest Service.
Copies of the original GLO notes on file at the Tombigbee Ranger District
office, Ackerman, MS were used to compile witness-tree data. The
notes date from 1832–1833. All data from within the legal boundaries of the
Ackerman Unit were used except in cases where the notes could not be read
with certainty. One to four witness trees were recorded by the surveyors at
each Township, section, and quarter-section corner, with two being usual.
Although most of the data represent trees thus recorded “near” (generally
within 50 feet of) points on a half-mile grid, following the instructions of
Fisk (1832), surveyors corrected for errors by adjusting the size of sections
along the standard parallels. In the case of the study area, this led to a tier of
sections of unusually small size along the eastern boundary of Range 12E.
A description of the GLO surveys in Mississippi, including the debates and
decisions concerning resolution of errors, is provided by Burt (1992).
Bias in GLO notes can occur from a number of idiosyncratic sources,
such as a desire by surveyors to avoid conifers with sticky sap (Bourdo
1956, Whitney and DeCant 2001). This factor was assessed by comparing
distances to witness trees (Bourdo 1956; cf. Delcourt 1975) for all species
for which n > 50 (Quercus rubra [red oak], Pinus sp. [pine], Quercus stellata
[post oak], Carya sp. [hickory], Quercus velutina [black oak], and Quercus
alba [white oak]). Distances to each species were calculated from the
notes and the results compared. Because sample sizes are generally small
(Table 1), t-tests were used in which unequal variance was assumed, with
a 0.95 confidence interval. Differences between the mean distances to the
target species were not significant; hence, intentional surveyor bias does not
appear to be a problem with this data set.
2008 E. Peacock, J. Rodgers, K. Bruce, and J. Gray 249
Species pattern analysis. The first step in analyzing spatial patterns of
individual species was to determine if species distributions were clustered or
random within the GLO sampling frame. In order to carry this out, the original
GLO data were imported into a GIS. The X- and Y- coordinates (UTM
Zone 16N; NAD 1927) of each survey point in the original GLO database
were determined from USGS Toposheets. The survey points were then plotted
within the GIS, and the attribute table of this GLO data layer had fields
consisting of an individual point ID, the corresponding X- and Y- coordinate
values, and the number of observed individuals within each species. Next,
an attribute query was used to identify only those points that contained a
particular species, e.g., red oak. The selected points were then extracted to
a new GIS data layer. This step was repeated for each of the most abundant
species (red oak, black oak, hickory, pine, white oak, and post oak).
Table 1. Witness trees recorded for the Ackerman Unit area in 1832–1833.
Species Number Percent
Quercus rubra L. (red oak) 161 18.59
Pinus sp. (pine) 150 17.32
Quercus stellata Wangenh. (post oak) 107 12.36
Carya sp. (hickory) 98 11.32
Quercus velutina Lam. (black oak) 79 9.12
Quercus alba L. (white oak) 79 9.12
Nyssa sylvatica Marsh. (black gum) 30 3.46
Quercus marilandica Muencch. (blackjack oak) 17 1.96
Castanea dentate Marsh. (chestnut) 16 1.85
Quercus sp. (spanish oak) 16 1.85
Liriodendron tulipifera L. (poplar) 16 1.85
Ilex sp. (holly) 13 1.50
Liquidambar styracifl ua L. (sweetgum) 11 1.27
Cornus fl orida L. (dogwood) 11 1.27
Fagus grandifolia Ehrh. (beech) 11 1.27
Acer sp. (maple) 9 1.04
Sassafras sp. (sassafras) 6 0.69
Carpinus caroliniana Walt. (hornbeam) 6 0.69
Fraxinus sp. (ash) 4 0.46
Ostrya virginiana K.Koch (Ironwood) 4 0.46
Ulmus sp. (elm) 2 0.23
Quercus bicolor (swamp oak) 2 0.23
Oxydendrum arboretum (L.) DC. (sourwood) 2 0.23
Platanus occidentalis L. (sycamore) 2 0.23
Quercus phellos L. (willow oak) 2 0.23
Nyssa sp. (gum) 2 0.23
Persea sp. (bay) 2 0.23
Ulmus rubra Muhl. (slippery elm) 2 0.23
Prunus persica (L.) Batsch. (peach) 1 0.12
Magnolia acuminate L. (cucumber tree) 1 0.12
Chrysopsis lanuginosa Small (lynn) 1 0.12
Magnolia sp. (magnolia) 1 0.12
Morus rubra L. (mulberry) 1 0.12
Salix nigra Marsh. (willow) 1 0.12
Total 866 100.00
250 Southeastern Naturalist Vol.7, No. 2
Nearest neighbor analysis was used to test for significant clustering of
important GLO species. The assumption of this analysis is that a clustered
point pattern will have a smaller average (Euclidian) distance among samespecies
observations than a point pattern that is random. The ratio of the
average distance among observed points to an average distance of a hypothetical
random point pattern should be less than 1 if the pattern is clustered
and nearly equal to one if the pattern is random. The average nearest neighbor
function within ArcGIS (ESRI 2005) Spatial Statistics Tools was used
to calculate the nearest neighbor ratio of observed and expected average
distances. This function also reports a standardized normal score (Z-value)
that can be used to determine if the degree of clustering is significantly
different than a hypothetical random pattern. A Z value less than -1.65 is
significant at alpha = 0.10 and a Z value less than -1.96 is significant at
the alpha = 0.05. Nearest neighbor calculations were performed on each
of the individual species data layers, and the area used for the calculations
was the extent of the Tombigbee National Forest Ackerman Unit proclamation
(legal) boundary (311 square km). For ease in calculation, distance
between individual species as used in nearest neighbor calculations refers
to the distance between section corners where those species were noted to
occur (i.e., as presence/absence data), rather than to the XY coordinates
of individual trees. The average distance between witness trees of the ten
most abundant taxa was 1.47 km; Table 2 shows the observed mean distances
between members of each of these species.
One potential structural control on vegetation is geological, as the
Ackerman Unit is underlain by two geological formations, the Wilcox and
the Porters Creek. The Wilcox formation is described as irregularly-bedded
fine to coarse sands, ferruginous sandstone, lignitic clays, and limited
beds of kaolinitic and bauxitic clays of Eocene age (Moore 1969); it occurs
within the western two-thirds of the study area. The Porters Creek, in
contrast, is described as primarily dark-gray silty clays with abundant sand
Table 2. Nearest neighbor analysis of important species. For significance (sign.): * indicates
significant at alpha = 0.10, ** indicates significant at alpha = 0.05, and ns indicates not significant.
Observed mean Ratio of
mean distance if observed to Pattern
Species distance (m) random (m) expected Z-value Sign. type
Red oak 942.9 1039.84 0.90 -1.910 * Clustered
Black oak 1124.4 1350.41 0.83 -2.410 ** Clustered
Hickory 1129.8 1233.49 0.91 -1.140 Ns Random
Pine 1014.0 1008.40 1.01 0.116 Ns Random
White oak 1262.7 1336.28 0.94 -0.870 Ns Random
Post oak 1145.0 1223.80 0.93 -1.135 Ns Random
Black gum 1861.1 1753.50 1.06 0.630 Ns Random
Chestnut 2122.9 2126.90 0.998 -0.014 Ns Random
Spanish oak 1717.8 1921.00 0.89 -0.810 Ns Random
Blackjack oak 2439.9 2040.80 1.19 1.449 Ns Random
2008 E. Peacock, J. Rodgers, K. Bruce, and J. Gray 251
beds and lenses in the upper part; it also is of Eocene age (Mellen and Mc-
Cutcheon 1939, Moore 1969) and occurs within the eastern one-third of
the study area. The Poisson probability distribution was used to investigate
differences in the occurrence of species between the Wilcox and Porters
Creek geological formations. A 1-km grid was superimposed on top of the
GLO GIS data layer and the frequency of species occurrence within each
grid cell was tabulated. The frequency values were plotted, and the expected
mean frequency per 1-km cell was determined for each species for both
geological formations. Additionally, the Poisson probability function was
used to calculate the probability of at least one individual occurring within
each geological formation.
Species diversity across the study area. Differences in species diversity
were investigated between the two geological formations. The total GLO
GIS data layer was subdivided into two files each containing points for the
Wilcox and Porters Creek formations, respectively. To accomplish this, the
Mississippi surface geology data layer (MARIS 2007) was subset to the
study area. This data layer had polygons representing the different geological
formations, and these polygons were intersected with the GLO point file
to generate data layers for each formation. Attribute queries were used to
determine the number of individuals within each species for each geological
formation. Relative abundance of each important species was also calculated
from the attribute queries.
The Shannon-Wiener diversity index was used to compare the diversity
of species between the two geological formations. This index is a measure of
both the total number of species and the evenness of individuals within each
species per unit area (Barbour et al. 1999). Index values are relative and can
be used to compare diversity among different regions.
Species distribution and topographic variability. The distribution of
each important species was examined in relation to slope, elevation, and
aspect. USGS digital elevation models (30-m resolution) for Oktibbeha
County, Choctaw County, and Winston County were merged to produce a
continuous elevation grid for the study area. ArcGIS Spatial Analyst (ESRI
2005) was used to derive slope and aspect from the elevation grid. Each
topographic grid was classified into categories in order to facilitate comparisons
with species distribution data. The elevation grid was classified
into four elevation classes using Jenks natural breaks (Jenks and Caspall
1971). These elevation classes were low (90–103.9 m), middle (104–
131 m), upper middle (131.1–157 m), and high (157.1–211 m). The slope
grid was classified using Jenks natural breaks into three classes consisting
of low slope (<2%), medium slope (2.1–6%), and steeper slope (6.1–31%).
The majority (90%) of the values in the steeper slope class, however, were
less than 15%. The aspect grid was manually classified into eight categories
consisting of north (338.1–0.0° and 0.0–22.0°), northeast (22.1–67.0°),
east (67.1–112.0°), southeast (112.1–157.0°), south (157.1–202.0°), southwest
(202.1–247.0°), west (247.1°–292.0°), and northwest (292.1–338.0°).
252 Southeastern Naturalist Vol.7, No. 2
Hawth’s analysis tools for ArcGIS (Beyer 2004) was used to intersect the
species point files with the classified topographic grids. The attribute
table from this intersection was imported into a statistical software package
(SPSS), and a non-parametric chi-square analysis was used to test for
independence of species abundance and the topographic variable classes.
Lastly, cross-tabulation of the topographic variable classes and geological
formation was used to investigate further the relationship of these two
variables. A chi-square contingency table was not possible due to the low
frequency of values within each category.
Prehistoric and historic settlement patterns
The archaeological site distribution data show clear temporal patterns
(Fig. 2). Few pre-Woodland-period occupations (those with artifacts diagnostic
of any part of the time span between ca. 12,000 and 800 B.C.) have
been identified on the Ackerman Unit; those that have been recorded are
mostly located along the major streams. In contrast, Woodland-period occupations
(including sites with Gulf Formational period diagnostics—see
Peacock 1997) dating from ca. 800 B.C. to A.D. 1050 are abundant and
are found on all landforms (Blitz 1984, Parrish 2006, Peacock 1997). Unlike
some other areas of the Eastern Woodlands (Scarry 1993), there is no
Figure 2. Archaeological site distributions on the Ackerman Unit of the Tombigbee
National Forest. Most Mississippian-period sites likely date no later than ca. A.D.
2008 E. Peacock, J. Rodgers, K. Bruce, and J. Gray 253
evidence of Woodland-period plant domestication in the study area (Johannessen
1993); rather, a general hunting-gathering lifestyle was followed.
The high numbers of Woodland-period sites may be related to the advent of
sedentariness (permanent settlement) and consequent population growth, as
sedentariness seems to have begun in the area during the first millennium
B.C. (Rafferty 1980, 2002).
There is a dramatic decrease in site numbers at about the beginning of
the second millennium A.D., with the advent of what archaeologists call the
Mississippian period. During this time, maize-based agriculture was widely
adopted across much of eastern North America, including the study area
(Peacock 2003). Although the entire span of the Mississippian period is used
in Figure 2, artifactual and chronometric (radiocarbon) evidence gathered
from surveys and excavations suggests that the area was abandoned early in
the period, around A.D. 1000–1100 (Blitz 1984; Peacock 1997, 2003), and
was not reoccupied until the early 19th century. Reasons for this early Mississippian
abandonment of the hills are unknown at this time. There does not
seem to have been a general depopulation event, as areas to the west (the
Bluff Hills, the Mississippi Alluvial Valley) and to the east (the Black Prairie,
the Tombigbee River Valley) show no signs of abandonment (Lipo and
Dunnell, in press; Rafferty 2003, in press). Indeed, a recent compilation of
radiocarbon and thermoluminescence dates shows a marked increase in occupation
of the Black Prairie by ca. A.D. 1200 (cf. Johnson and Sparks 1986,
Johnson et al. 1991, Rafferty and Peacock 2007). Work currently underway
on the Ackerman Unit indicates that Mississippian period occupations are
quite small and relatively ephemeral, with only a few artifacts diagnostic
of the period being recovered from intensive shovel testing on close-interval
grids across entire sites (Andrew Triplett, Tombigbee National Forest,
Ackerman, MS, 2007; pers. comm.). This finding supports our contention
that abandonment took place early in the period, before sufficient time had
elapsed for substantial archaeological remains to accumulate.
Early 19th-century sites (including a very few that show evidence of Historic-
period aboriginal occupation, possibly Choctaw Indian homesteads)
also are relatively rare, while many mid-19th- to early 20th-century sites have
been recorded on the unit (Peacock and Patrick 1997). This archaeological
assessment is borne out by records research, with very few improvements
(the Robinson Road and a few fields) being shown on the early 19th-century
GLO maps for the Ackerman Unit.
Taken together, these settlement-pattern data present a very interesting
picture where the reconstruction of forest cover is concerned. By the time
of the GLO surveys in the early 1830s, the Ackerman Unit area had been
essentially devoid of American Indian settlement for approximately seven
to eight hundred years, and Historic-period resettlement of the area was still
quite limited. The witness-tree data therefore should provide a good representation
of what forest conditions in the North Central Hills were like in the
absence of intensive human occupation.
254 Southeastern Naturalist Vol.7, No. 2
Witness tree data
Data were compiled for 866 witness trees on the Ackerman Unit. Common
names, numbers, and proportions of the 37 species represented are
given in Table 1. Common names are used, as different surveyors were
involved and precise identification of some species from the notes is not always
possible. Despite this limiting factor, a good characterization of forest
cover is obtained. In general, the area supported oak-dominated hardwoods,
with pine being common at slightly more than 17% of the total.
Spatial pattern analysis. Red oak and black oak witness trees were significantly clustered within the Tombigbee National Forest Ackerman Unit
study area, but all the other important species had a more-or-less random
distribution (Table 2). Red oak was more clustered within the Wilcox geological
formation of the Ackerman Unit, while Black oak was more clustered
within the Porters Creek formation (Fig. 3). The point-distribution maps of
pine and hickory across the Ackerman Unit (Fig. 4) suggest a slight preference
for geological formation (Wilcox), but the nearest neighbor analysis
did not show this trend to be significant. Both white oak and post oak had a
random distribution across the study area (Fig. 5).
The importance of geological formation to species distribution was examined
further by calculating the Poisson probability distribution for each
important taxon (Fig. 6). Red oak was twice as likely to occur within the
Wilcox (45%) than in the Porters Creek (22%) formation. Black oak, on the
other hand, was nearly four times more likely to occur in the Porters Creek
(33%) than in the Wilcox formation (8.5%). Both results corroborate the
results of the nearest neighbor analysis. As noted in the point distributions,
Figure 3. Distribution
absence) of red
oak and black
trees from the
data on the Ackerman
Red oak is more
clustered in the
oak is more
clustered in the
2008 E. Peacock, J. Rodgers, K. Bruce, and J. Gray 255
both hickory and pine had higher probabilities of occurring in the Wilcox
formation. Both white oak and post oak had similar probabilities of occurring
on each geological formation.
Species diversity across the study area. Because geological formation
was determined to be an infl uential factor, comparisons between the Wilcox
Figure 4: Distribution
of pine and hickory
witness trees from the
General Land Office
survey data on the
Ackerman Unit of the
Forest. Although both
species are slightly
more abundant in the
that their distribution
across the two geological
Figure 5. Distribution
of white oak and post
oak witness trees from
the General Land Office survey data on the
Ackerman Unit of the
white oaks are slightly
more abundant on the
Wilcox formation and
the post oak are slightly
more prevalent on
the Porters Creek formation,
a random distribution
of these species across
the two geological
256 Southeastern Naturalist Vol.7, No. 2
Table 3. Species diversity on the Ackerman Unit of the Tombigbee National Forest compared
between the Wilcox and Porters Creek geological formations.
Diversity measure Porters Creek Wilcox
Number of species 28 30
Total number of individuals (all spp.) 324 341
Shannon-Wiener diversity index -2.53 -2.42
Number GLO points 141 255
Figure 6. The Poisson probability of at least one witness tree from the General
Land Office survey data occurring within the Wilcox and Porters Creek geological
Table 4. Absolute abundance of taxa of n > 5 divided between the Wilcox and Porters Creek
Species Porters Creek (# of individuals) Wilcox (# of individuals)
Red oak 38 123
Black oak 60 19
Hickory 22 76
Pine 59 91
White oak 29 50
Post oak 43 64
Black gum 5 25
Chestnut 6 10
Holly 7 6
Spanish oak 5 11
Sweet gum 5 6
Poplar 7 9
Dogwood 2 9
Blackjack oak 4 13
Beech 6 5
Maple 6 2
Hornbeam 4 2
2008 E. Peacock, J. Rodgers, K. Bruce, and J. Gray 257
and Porters Creek formation were investigated in more detail (Tables 3–4,
Fig. 7). It should be pointed out that the Wilcox represents a larger portion
of the study area (64%) than the Porter Creek formation (46%). Both geological
formations had similar number of taxa: 28 species occurred within the
Porters Creek formation, and 30 species occurred within the Wilcox formation
(Table 3). The Porters Creek formation had fewer numbers of individuals,
but this was most likely related to the smaller area unit. Moreover, the Porters
Creek had a slightly higher Shannon-Wiener index value. Because this diversity
index is heavily weighted by evenness of individuals, the large number of
red oaks in the Wilcox formation (123 individuals; Table 4) may have created
an uneven situation that defl ated the index value for this area.
Regarding the abundance of important taxa (those for which n > 5),
76% of all red oaks occurred within the Wilcox formation (Table 4). The
Wilcox also had higher numbers of hickory, Quercus marilandica Muencch.
(blackjack oak), Nyssa sylvatica Marsh. (black gum), Cornus fl orida L.
(dogwood), and Quercus falcata Michaux. (Spanish oak; Fig. 7). In contrast,
76% percent of all black oaks occurred within the Porters Creek formation.
The Porters Creek formation also had slightly higher numbers of pine and
Figure 7. Relative abundance of the most frequent witness tree species represented
in the General Land Office survey data from the Ackerman Unit of the Tombigbee
National Forest. Relative abundance data are plotted for both the Wilcox and Porters
Creek geological formations.
258 Southeastern Naturalist Vol.7, No. 2
post oak. The remaining species were either evenly distributed across the
two formations or had such low occurrence values that meaningful comparisons
could not be made (Table 4, Fig. 7).
Several species were found only in the Wilcox formation, and these
included Prunus persica (L.) Batsch. (peach), Magnolia acuminata L. (cucumber
tree), Oxydendrum arboreum (L.) DC (sourwood), Chrysopsis lanuginosa
Small (lynn), Persea sp. (bay), Ulmus rubra Muhl. (slippery elm),
and Magnolia sp. (magnolia). The species that were found only in the Porters
Creek formation included Quercus bicolor Willd. (swamp oak), Platanus
occidentalis L. (sycamore), Morus rubra L. (mulberry), Quercus phellos L.
(willow oak), Nyssa sp. (gum), and Salix nigra Marsh. (willow). Swamp oak,
willow oak, gum, and willow are more hydric species (USFS 2007a, 2007b),
and their presence only within the clay-rich and more acidic soils of the Porter
Creek formation further illustrates the geological infl uence on the vegetation.
Species distribution and topographic variability. Slope, elevation, and
aspect showed a complex relationship to species distribution (Tables 5–7). It
should be pointed out that there was not much topographic variability within
the study area. Slope value ranges from 0% to 31%; however, the vast majority
of values (over 90%) had a slope less than 15%. Similarly, elevation only
ranged from 90 m to 211 m.
Black oak and pine showed a significant preference for the middle slope
(Table 5). Post oaks, however, showed a preference for low and middle
slope values. This result was somewhat counterintuitive because post oaks
are described as being more prevalent on dry ridges (Chester et al. 1987).
Red oak, hickory, and white oak showed no significant preference for slope
Table 5. Results of chi square analysis for slope classes among the six most abundant witness
tree species. Note: results are for observed frequency of witness-tree species; * = significant at
p < 0.05; ns = not significant.
Slope class Red oak Black oak Hickory Pine Post oak White oak
1 (<2%) 42 13 28 29 34 23
2 (2.1–6%) 32 36 22 57 34 22
3 (7.1–31%) 41 8 28 31 17 23
Significance 0.45 ns 0.00* 0.02* 0.63 ns 0.03* 0.48 ns
Table 6. Results of chi-square analysis for elevation classes among each of the six most common
witness-tree species. Note: results are for observed frequency of witness-tree species; * =
significant at p < 0.05; ns = not significant.
Elevation class Red oak Black oak Hickory Pine Post oak White oak
Low (90–104 m) 11 5 6 11 11 11
Middle (104.1–131 m) 41 24 28 29 27 25
Upper middle (131.1–157 m) 39 20 23 51 23 25
High (157.1–211 m) 24 8 21 26 24 7
Significance 0.00* 0.00* 0.00* 0.003* 0.07 ns 0.01*
2008 E. Peacock, J. Rodgers, K. Bruce, and J. Gray 259
class. The results of the cross-tabulation of slope and geological formation
indicate that red oaks were more abundant on higher slope classes within
the Wilcox formation and were more abundant on lower slope classes within
the Porters Creek formation (Table 8). The interaction between slope and
geological formation was not as clear for the other important taxa.
With the exception of post oak, all important taxa showed a significant
preference for elevation class (Table 6). Red oak, black oak, hickory, and
white oak were more abundant in the middle elevation class while pine was
more abundant in the upper middle elevation class. The cross-tabulation of
elevation class and geological formation indicated that none of the important
taxa occurred within the lowest elevations of the Wilcox formation (Table 9).
Table 7. Results of chi-square analysis for aspect classes among each of the six most common
witness-tree species. Note * = significant at P < 0.05; ns = not significant.
Aspect Red oak Black oak Hickory Pine White oak Post oak
North 16 8 10 13 9 13
Northeast 14 7 11 14 11 9
East 23 3 13 21 13 20
Southeast 9 7 4 11 7 10
South 15 7 6 15 7 5
Southwest 18 6 10 20 5 9
West 6 6 12 10 8 6
Northwest 14 13 12 13 8 13
Significance 0.07 ns 0.36 ns 0.42 ns 0.378 ns 0.64 ns 0.04*
Table 8. Cross-tabulation of slope class with geological formation. W = Wilcox and PC =
Low slope (<2%) Medium slope (2.1%–6%) slope (6.1%–31%)
W PC W PC W PC
Red oak 26 16 27 5 36 5
Black oak 4 9 9 27 1 7
Hickory 19 9 18 4 25 5
Pine 13 16 29 28 26 5
White oak 11 12 12 10 20 3
Post oak 20 14 18 16 13 4
Table 9. Cross-tabulation of elevation class with geological formation. W = Wilcox and PC =
Low elevation Middle elevation Upper Middle elevation High elevation
(90–104 m) (104.1–131 m) (131.1–157 m) (157.1–211 m)
W PC W PC W PC W PC
Red oak 0 11 27 14 38 1 24 0
Black oak 0 5 5 19 4 16 5 3
Hickory 0 6 18 10 21 2 21 0
Pine 0 11 11 18 35 16 22 4
White oak 0 11 13 12 23 2 7 0
Post oak 0 11 12 15 16 7 23 1
260 Southeastern Naturalist Vol.7, No. 2
In contrast, only a few individuals were present within the upper elevation of
the Porters Creek. It is not clear at this time if the absence of species within the
lower Wilcox or upper Porters Creek is an artifact of the GLO sampling frame.
Aspect was the least important topographic variable, and only post oak
had a significant result (Table 7). Post oak appears to prefer the eastern and
northern aspects within the study area.
Accounts of 18th- and 19th-century landscape conditions in north-central
Mississippi range from traveler’s accounts and other historical records
(e.g., Nutt 1805) to broad-scale, scientific descriptions (Hilgard 1860). Interpreting
the former is something of an art form, but previous studies have
indicated that such historic descriptions are, in fact, reasonably accurate
(e.g., Brewer 2001, Peacock 1992a, Peacock and Miller 1990). Hilgard’s
(1860:301) description of the hill country of northern Winston County as
“characterized by a growth of White, Black and Red Oak, stout Post Oak, a
good deal of Hickory, and sometimes Short-leaf Pine” is in general accord
with the GLO data from the Ackerman Unit.
Historical records, including GLO notes, may be interpreted in different
ways, however (e.g., Johnson 1990, Peacock 1992a, Peacock and Miller
1990). For example, Johnson (1988:59) suggested that “the acidic sands of
the North Central Hills were covered primarily in pine.” This statement contrasts
with the findings of the current study and with an earlier study of GLO
notes and a botanical survey of old growth forests in the North Central Hills
by Brewer (2001). Brewer compiled witness-tree data from two townships
in Lafayette County and found a hardwood-dominated early 19th-century forest,
with blackjack oak, black oak, and post oak being especially common;
red oak also was common, although it varied in abundance spatially (Brewer
2001). His results differ from ours mainly in the higher proportion of blackjack
oak reported; such variability within the province (see also Hilgard
1860:300–301) is worthy of further study.
The results of the nearest neighbor analysis, the point-distribution
maps, and the Poisson probability distributions clearly show that red
oak and black oak distribution patterns within the Ackerman Unit of the
Tombigbee National Forest correspond with geological formation. Black
oaks prefer rich, moist, acidic, and well-drained soils. The Porters Creek
formation may be more suited to black oak because it is very clay-rich
and acidic, especially where the North Central Hills transition to the Flatwoods
along the eastern margin of the Unit. Red oaks, in contrast, tolerate
a wide range of soil types, but generally prefer loam, silt, sandy loam, and
sandy soil textures. The preference of red oak for the coarser soil textures
may explain its higher abundance on the Wilcox formation. The distinction
between the Wilcox and Porters Creek formations were also apparent
within the pine and hickory data; however, this separation was not nearly
2008 E. Peacock, J. Rodgers, K. Bruce, and J. Gray 261
as sharp as with the red and black oaks. The other taxa did not show a
clear preference for a particular geological formation.
Topographic variables appear to have had some infl uence on vegetation
patterns. Several species showed preference for slope and/or elevation
type, yet it is difficult to draw clear interpretations of this data. The multiple
significances with elevation, for example, are somewhat counterintuitive as
the slight elevation differences on the Unit should not pose any limitation
to species distributions. Similarly, slope class in relation to species distribution
was somewhat opaque. The preference of red oak for lower slopes in
the Porters Creek and steeper slopes in the Wilcox formations was the only
meaningful result. However the ecological significance of this is difficult to
determine because the lower slope and steeper slope classes only differed by
10 individuals. Thus, the topographic infl uence does not seem to be as instrumental
in shaping the distribution patterns of the important taxa within the
study area. The main conclusion borne out of the GLO data is that geological
formation had primary infl uence on vegetation patterns.
Based on the GLO notes, the Ackerman Unit area in the early 19th century
can be characterized as one of mixed hardwoods with a generally distributed
pine component. Based on archaeological data, this represents a picture of
conditions as they existed in the near-absence of human settlement over a
period of many centuries (cf. Brewer 2001). This is not to say that human actions
did not alter the forest; in particular, the role of burning by Indians in this
sparsely settled region following the Woodland period is not known at present,
and provides an interesting avenue for future research (cf. Vale 2002). Brewer
(2001), for example, has argued that the “presettlement” forests of the North
Central Hills in Lafayette County, north Mississippi, were more open than the
few remaining old-growth forests of today, something he attributes to modern
fire suppression. He points out that his study area was located some 70 km west
of the major area of Chickasaw Indian settlement in the Black Prairie physiographic
province (Johnson 2000), and that there are records of the Chickasaw
burning the landscape (Nairne 1988). A similar concentration of Protohistoric
to early Historic-period aboriginal sites exists in the vicinity of Starkville, approximately
24 km northeast of the Ackerman Unit (Atkinson 1979, Rafferty
2003). Thus, while there is little evidence for actual occupation of the study
area between ca. A.D. 1000 and the early 19th century, the landscape may have
been manipulated in ways that bear further investigation. At this point, however,
it can be argued that the GLO data for the Ackerman Unit represent as
accurate a picture of undisturbed woodland as is obtainable in the region. The
witness-tree data therefore represent one set of historically defensible baseline
conditions for forest restoration.
An equivalent amount of archaeological survey has been carried out
by the senior author on other lands in north-central Mississippi, including
the Trace Unit of the Tombigbee National Forest and the Holly Springs
262 Southeastern Naturalist Vol.7, No. 2
and Yalobusha Units of the Holly Springs National Forest (e.g., Peacock
1992b, 1992c, 1993a, 1993b, 1993c, 1993d, 1995b). The archaeological
settlement patterns in those areas are similar to what has been described
for the Ackerman Unit, with a relative lack of settlement prior to the Woodland
period, intensive settlement during the Woodland, and a dramatic decrease
in sites at the beginning of the Mississippian period (Peacock 1997).
These findings are mirrored in other studies (Ford 1980, 1981, 1989, 1990;
Johnson 1984, 1988). For example, Johnson (1988) compiled data from a
number of surveys and excavation projects in north Mississippi. Although
only limited data from the North Central Hills were available at that time,
no Mississippian sites are represented from that province in his study.
Morgan (1996) compiled data on diagnostic artifacts using site files at the
Mississippi Department of Archives and History. His research universe was
composed of 40 counties making up approximately one-third of the state,
from north- and east-central Mississippi west to the Mississippi River
and essentially encompassing the North Central Hills. One class of site he
looked for was “all sites falling within a twelfth- through seventeenth-century
[A.D.] timeframe, including Mississippi, Plaquemine, and Protohistoric
cultural manifestations” (Morgan 1996:2). His findings led him to state
that, “at least one fact is obvious: that occupation in the central portion of
the state [i.e., the North Central Hills] is virtually nonexistent throughout
the period,” and that “present evidence is probably ample for documenting
that this was an area of meager aboriginal occupation” (Morgan 1996:5).
Also similar to the Ackerman Unit, Historic-period settlement in the uplands
of north-central Mississippi was limited before the mid-19th century
(e.g., Peacock 1992a, 1992b, 1993a, 1993b, 1993c, 1993d, 1995b). Based
on this information, we would argue that GLO data can provide an accurate
baseline for minimally disturbed forest conditions throughout the hill belts of
It is questionable whether forest management practices can be structured
so as to reconstruct those baseline conditions, as there has been enormous soil
loss in the North Central Hills and other upland provinces in Mississippi since
the mid-19th century (Grissenger et al. 1982). It could be argued that a range of
target ecological conditions should be the goal of a balanced resource management
program, if maintaining biodiversity is the key goal (Minnis and Elisens
2000, Peacock 1998). Whatever path is chosen, the archaeological settlement
pattern and GLO data combine to provide an accurate picture of what mature
tree cover in the hills of Mississippi was like in the absence of significant human
settlement. This work adds to a growing number of studies that explore
the implications of historic forest conditions for resource management in the
state (e.g., Aquilani 2006, Brewer et al. 2000).
We would like to thank John Baswell, District Ranger on the Tombigbee National
Forest, and Joe Seger, Director of the Cobb Institute of Archaeology, Mississippi
State University, for their support for this project. We also would like to thank the
Mississippi Museum of Natural Science for providing copies of reports.
2008 E. Peacock, J. Rodgers, K. Bruce, and J. Gray 263
Adkins, H.G. 1979. The historical geography of extinct towns in Mississippi. Pp.
123–152, In P.W. Prenshaw and J.O. McKee (Eds.). Sense of Place: Mississippi.
University Press of Mississippi, Jackson, MS.
Anderson, R.C., and M.R. Anderson. 1975. The presettlement vegetation of Williamson
County, Illinois. Castanea 40:345–363.
Aquilani, S.M. 2006. Bird communities in silvicultural fragments of Holly Springs
National Forest, Mississippi. Southeastern Naturalist 5:135–148.
Atkinson, J.R. 1979. A historic contact Indian settlement in Oktibbeha County, Mississippi.
Journal of Alabama Archaeology 25(1):61–82.
Barbour, M.G., J.H. Burk, W.D. Pitts, F.S. Gilliam, and M.W. Schwartz. 1999. Terrestrial
Plant Ecology, Third Edition. Addison Wesley Longman, Inc., Menlo Park, CA.
Beyer H.L. 2004. Hawth’s analysis tools for ArcGIS. Available online at http:
//www.spatialecology.com. Accessed November 1, 2007.
Blitz, J.H. 1984. A cultural resources survey in the Tombigbee National Forest, Mississippi.
USDA Forest Service, Jackson, MS.
Bourdo, E.A. 1956. A review of the General Land Office survey and its use in quantitative
studies of former forests. Ecology 37:754–768.
Bragg, D.C. 2003. Natural presettlement features of the Ashley County, Arkansas
area. American Midland Naturalist 149:1–120.
Brewer, J.S. 2001. Current and presettlement tree species composition on some
upland forests in northern Mississippi. Journal of the Torrey Botanical Society
Brewer, J.S., S.M. Aquilani, and M.L. Warren. 2000. A review of research on upland
oak-pine communities in the Little Tallahatchie Experimental Forest and adjacent
Holly Springs National Forest. Report submitted to the Forest Hydrology
Laboratory, Southern Forest Experiment Station and the Holly Springs National
Forest, Oxford, MS.
Burt, R. 1992. The survey of Mississippi’s state, Indian, and township Boundaries.
Mississippi Association of Professional Surveyors, Mississippi State University,
Carroll, T.B. 1983. Historical Sketches of Oktibbeha County (Mississippi). Sullivan’s
Printing, West Point, MS. Originally printed in 1931 by the Dixie Press,
Chester, E.W, R.J. Jensen, L.J. Schibig, and S. Simoni. 1987. The Nut Trees of Land
Between the Lakes. The Center for Field Biology of LBL, Austin Peay State
University, Clarksville, TN.
Coleman, J.P. 1973. Choctaw County Chronicles: A History of Choctaw County,
Mississippi. Privately published, Ackerman, MS.
Delcourt, H.R. 1975. Reconstructing the forest primeval, West Feliciana Parish. Lousiana
State Univesity Museum of Geosiences. Mélanges Series No. 10. 13 pp.
Environmental Systems Resource Institute, Inc (ESRI). 2005. ArcGIS Version 9.1,
Farley, S.C., R.E. Masters, and D.M. Engle. 2002. Riparian landscape change in
central Oklahoma 1872–1991. Proceedings of the Oklahoma Academy of Science
Fickle, J.E. 2001. Mississippi Forests and Forestry. University Press of Mississippi,
Fisk, G. 1832. Instructions for Surveying in the State of Mississippi. Natchez: R.
Semple. Reprinted in 1992 by the Mississippi Association of Professional Surveyors,
Mississippi State University, Starkville, MS.
Ford, J.L. 1980. Alas, poor Womack! Mississippi Archaeology 25(2):26–31.
264 Southeastern Naturalist Vol.7, No. 2
Ford, J.L. 1981. Time and temper in the North Central Hills of Mississippi. Journal of
Alabama Archaeology 27(1):57–71.
Ford, J.L. 1989. Time and temper meets trend and tradition. Mississippi Archaeology
Ford, J.L. 1990. The Tchula connection: Early woodland culture and burial mounds in
north Mississippi. Southeastern Archaeology 9:103–115.
Foster III, H.T., B. Black, and M.D. Abrams. 2004. A witness-tree analysis of the effects
of Native American Indians on the pre-European settlement forests in eastcentral
Alabama. Human Ecology 32:27–47.
Grissenger, E.H., J.B. Murphy, and W.C. Little. 1982. Late-Quaternary valley fill deposits
in north-central Mississippi. Southeastern Geology 23:147–162.
Hilgard, E. 1860. Report on the geology and agriculture of the state of Mississippi. E.
Banksdale, State Printer, Jackson, MS.
Jenks, G.F., and F. C. Caspall. 1971. Error in choroplethic maps: Definition, measurement,
reduction. Annals of the Association of American Geographers 61:217–244.
Johannessen, S. 1993 Farmers of the Late Woodland. Pp. 57–77, In C.M. Scarry (Ed.).
Foraging and Farming in the Eastern Woodlands. University Press of Florida,
Johnson, J.K. 1984. Prehistoric settlement in the upper Yocona drainage. Mississippi
Johnson, J.K. 1988. Woodland settlement in northeastern Mississippi: The Miller
tradition. Pp. 49–59, In R.C. Mainfort, Jr. (Ed.). Middle Woodland Settlement and
Ceremonialism in the Mid-South and Lower Mississippi Valley. Mississippi Department
of Archives and History, Jackson, MS. Archaeological Report 22.
Johnson, J.K. 1990. Cedar glades and Protohistoric settlement: A reply to Peacock and
Miller. Mississippi Archaeology 25(2):58–62.
Johnson, J.K. 2000. The Chickasaws. Pp. 85–121, In B.G. McEwan (Ed.). Indians of
the Greater Southeast: Historical Archaeology and Ethnohistory. University Press
of Florida, Gainesville, FL.
Johnson, J.K., and J.T. Sparks. 1986. Protohistoric settlement patterns in northeastern
Mississippi. Pp. 64–87, In D.H. Dye and R.C. Brister (Eds.). The Protohistoric Period
in the Mid-South. Mississippi Department of Archives and History. Archaeological
Report 18. Jackson, MS.
Johnson, J.K., G.R. Lehmann, J.R. Atkinson, and A.B. Shea. 1991. Protohistoric
Chickasaw settlement patterns and the De Soto route in northeast Mississippi.
Report submitted to the National Endowment for the Humanities by the Center for
Archaeological Research, University of Mississippi. Oxford, MS.
Lipo, C.P., and R.C. Dunnell. In Press. Prehistoric settlement in the Lower Mississippi
Valley: A critical review. In J. Rafferty and E. Peacock (Eds.). Time’s River:
Archaeological Syntheses from the Lower Mississippi River Valley. University of
Alabama Press, Tuscaloosa, AL.
Mellen, F.F., and T.E. McCutcheon. 1939. Winston County mineral resources. Mississippi
State Geological Survey, Bulletin 38. University, MS.
Minnis, P.E., and W.J. Elisens. 2000. Biodiversity and Native America. University of
Oklahoma Press, Norman, OK.
Mississippi Automated Resource Information Systems (MARIS). 2007. MARIS
technical center website. Available online at www.maris.state.ms.us. Accessed
November 1, 2007.
Moore, J.H. 1988. The Emergence of the Cotton Kingdom in the Old Southwest: Mississippi
1770–1860. Louisiana State University Press, Baton Rouge, LA.
Moore, W.H. 1969. Geological map of Mississippi. Mercury Maps Inc., Jackson, MS.
Morgan, D. 1996. The Mississippi De Soto trail mapping project. Mississippi Department
of Archives and History, Jackson, MS.
2008 E. Peacock, J. Rodgers, K. Bruce, and J. Gray 265
Nairne, T. 1988. Nairne’s Muskhogean Journals: The 1708 Expedition to the Mississippi
River. (A. Moore [Ed.].) University Press of Mississippi, Jackson, MS.
Nutt, R. 1805. Diary of a tour through the western and southern parts of the United
States in America. (J.D. Jennings [Ed.].) Journal of Mississippi History 9:34–61.
Parrish, J.L. 2006. An archaeological investigation of four Woodland-period sites
in the North Central Hills Physiographic region of Mississippi: 22CH653,
22WI536, 22WI588, and 22WI670. M.A. Thesis. Mississippi State University,
Peacock, E. 1992a. Some additional notes on forest reconstruction in the Black Belt.
Mississippi Archaeology 27(1):1–18.
Peacock, E. 1992b. A cultural resources survey of selected timber tracts in the Holly
Springs National Forest, Marshall County, Mississippi. USDA Forest Service
report on file, Mississippi Department of Archives and History, Jackson, MS.
Peacock, E. 1992c. A cultural resources survey of selected timber stands in Compartment
33, Holly Springs National Forest, Marshall County, Mississippi. USDA
Forest Service report on file, Mississippi Department of Archives and History,
Peacock, E. 1993a. A cultural resources survey of selected timber stands in Compartment
130, Holly Springs National Forest, Yalobusha County, Mississippi. USDA
Forest Service report on file, Mississippi Department of Archives and History,
Peacock, E. 1993b. A cultural resources survey of selected timber stands in Compartment
37, Holly Springs National Forest, Benton and Marshall Counties, Mississippi.
USDA Forest Service report on file, Mississippi Department of Archives
and History, Jackson, MS.
Peacock, E. 1993c. A cultural resources survey of a proposed land exchange in Compartment
24, Holly Springs National Forest, Benton County, Mississippi. USDA
Forest Service report on file, Mississippi Department of Archives and History,
Peacock, E. 1993d. A cultural resources survey of selected timber stands in Compartment
43, Holly Springs National Forest, Benton and Marshall Counties, Mississippi.
USDA Forest Service report on file, Mississippi Department of Archives
and History, Jackson, MS.
Peacock, E. 1995a. Shovel-test screening and survey methodology: A field study
from the Tombigbee National Forest, North Mississippi. USDA Forest Service,
Peacock, E. 1995b. A final report on heritage resource surveys of ice-damaged timber
tracts on the Holly Springs and Tombigbee National Forests, Yalobusha and
Chickasaw counties, Mississippi. USDA Forest Service report on file, Mississippi
Department of Archives and History, Jackson, MS.
Peacock, E. 1996. Archaeological site survey in wooded environments: A field study
from the Tombigbee National Forest, north-central Mississippi. North American
Peacock, E. 1997. Woodland ceramic affiliations and settlement pattern change in
the North Central Hills of Mississippi. Midcontinental Journal of Archaeology
Peacock, E. 1998. Historical and applied perspectives on prehistoric land use in
eastern North America. Environment and History 4:1–29.
Peacock, E. 2003. Excavations at Stinking Water (22WI515/516), a prehistoric
habitation site in the North Central Hills physiographic province of Mississippi.
Mississippi Archaeology 38(1):3–91.
266 Southeastern Naturalist Vol.7, No. 2
Peacock, E., and D. Fant. 2002. Biomantle formation and artifact translocation in
upland sandy soils: An example from the Holly Springs National Forest, northcentral
Mississippi. Geoarchaeology: An International Journal 17:91–114.
Peacock, E., and W.F. Miller. 1990. Protohistoric settlement patterns in northeast Mississippi
and the cedar-glade hypothesis. Mississippi Archaeology 25(2):45–57.
Peacock, E., and A.J. Patrick. 1997. Site survey and land-records research: A comparison
of two methods for locating and characterizing Historic-period sites on
the Tombigbee National Forest, Mississippi. Mississippi Archaeology 32:1–26.
Perkins, T.E., and G.R. Matlack. 2002. Human-generated pattern in commercial
forests of southern Mississippi and consequences for the spread of pests and
pathogens. Forest Ecology and Management 157:143–154.
Phelps, D.A. 1950. The Robinson Road. Journal of Mississippi History 3:155–226.
Rafferty, J.E. 1979. Cultural resource reconnaissance and project-oriented survey,
Noxubee National Wildlife Refuge, Mississippi. Mississippi State University,
Department of Anthropology. Report submitted to the National Park Service,
Interagency Archeological Services, Atlanta, GA.
Rafferty, J.E. 1980. Surface collections and settlement patterns in the Central Tombigbee
Valley. Southeastern Archaeological Conference Bulletin 22:90–94.
Rafferty, J.E. 2002. Woodland period settlement patterning in the northern Gulf
Coastal Plain of Alabama, Mississippi, and Tennessee. Pp. 204–227, In D.G.
Anderson and R.C. Mainfort, Jr. (Eds.). The Woodland Southeast. University of
Alabama Press, Tuscaloosa, AL.
Rafferty, J.E. 2003. Prehistoric settlement patterning on the Mississippi Black Prairie.
Pp. 167–193, In E. Peacock and T. Schauwecker (Eds.). Blackland Prairies
of the Gulf Coastal Plain: Nature, Culture, and Sustainability. University of Alabama
Press, Tuscaloosa, AL.
Rafferty, J.E. In Press. Settlement patterns, occupations, and field methods. In J.
Rafferty and E. Peacock (Eds.). Time’s River: Archaeological Syntheses from the
Lower Mississippi River Valley. University of Alabama Press, Tuacaloosa, AL.
Rafferty, J.E., and E. Peacock. 2007. The spread of shell tempering in the Mississippi
Black Prairie. Paper presented at the 72nd Annual Meeting of the Society for
American Archaeology, Austin, TX.
Scarry, C.M. (Ed.). 1993. Foraging and Farming in the Eastern Woodlands. University
Press of Florida, Gainesville, FL.
Silbernagel, J., S.R. Martin, M.R. Gale, and J. Chen. 1997. Prehistoric, historic, and
present settlement patterns related to ecological hierarchy in the eastern Upper
Peninsula of Michigan, USA. Landscape Ecology 12:233–240.
US Forest Service (USFS). 2007a. Silvics of North America, volume 2: Hardwoods.
Available online at http://www.na.fs.fed.us/spfo/pubs/silvics_manual/table_of_
contents.htm). Accessed November 1, 2007.
US Forest Service (USFS). 2007b. UCONN Plant Data base. Available online at http:
//www.hort.uconn. edu/plants/. Accessed November 1, 2007.
Vale, T.R. (Ed.). 2002. Fire, Native Peoples, and the Natural Landscape. Island Press,
Vestal, F.E., and T.E. McCutcheon. 1943. Choctaw county mineral resources. Mississippi
State Geological Survey. University, MS. Bulletin 52.
Whitney, G.G., and J.P. DeCant. 2001. Government Land Office surveys and other
early land surveys. Pp. 147–172, In D. Egan and E.A. Howell (Eds.). The Historical
Ecology Handbook. Island Press, Washington, DC.
Williams, G.I., Jr. 1993. Environmental setting. Pp. 15–28, In A.M. Early (Ed.).
Caddoan Saltmakers in the Ouachita Valley: The Hardaman Site. Arkansas Archeological
Research Series No. 43. Fayetteville, AR.