FOREST COMPOSITION AND SOIL TEXTURE
RELATIONSHIPS ACROSS THE TENSION ZONE,
CENTRAL LOWER MICHIGAN, _

Thesis for the Degree Of M. A,
MICHIGAN STATE UNIVERSITY.
KIMBERLY ELLEN MEDLEY
1985

PLACE N RETURN BOX to mnovo ml- chockanfl'om your ncoml.
TO AVOID FINES Mum on or Moro duo duo.

 

DATE DUE DATE DUE DATE DUE

 

F1:

.

f‘fifii‘ 3 1’ "\I ‘

 

OIUV. " £1225

 

’____

New. J <' 'wfi

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSUIIAn.““

 

ABSTRACT

FOREST COMPOSITION AND SOIL TEXTURE RELATIONSHIPS
ACROSS THE MICHIGAN TENSION ZONE, CENTRAL LOWER MICHIGAN
BY

Kimberly Ellen Medley

Relationships between forest composition and soils across the
Michigan vegetation tension zone were examined.]1;was hypothesized
that patterns of tree species and soil texture are significantly
associated and that coarse-textured soils increase abruptly northward
across the tension zone. It was also hypothesized that needleleaf trees
increase significantly northward, they are more strongly associated to
coarse-textured soils than are deciduous species to fine-textured
soils, and the relationships between forest physiognomy and soil
texture are stronger north of a sharp transition toward coarse-textured
soils.

Data on presettlement forest composition and soils were recorded
fromIthe land office survey notes and the modern soil surveys along
three south-north belt transects. The results from several statistical
analyses demonstrate that distribution of tree species, forest
physiognomy, and soil texture are significantly related. Needleleaf
trees are most frequent on coarse-textured soils, and sugar maple on
fine-textured soils, but the nearly equal distribution of most

deciduous species, and the predominance of beech, on all soil textures

Kimberly Ellen Medley
lessened the strength of the association and dependent variable (tree
species) predictability. Coarse-textured soils and needleleaf tree
percentages increase signicantly north across the tension zone with
similar northward trends. Contrasts identified between the transects
reflect east-west variation in overall forest composition and the
distribution northward of coarse-textured soils. It is concluded that
the occurrence of needleleaf trees is closely related to coarse-
textured soils within this region such that the underlying soils

pattern helps define the tension zone.

FOREST COMPOSITION AND SOIL TEXTURE RELATIONSHIPS

ACROSS THE TENSION ZONE, CENTRAL LOWER MICHIGAN

BY

Kimberly Ellen Medley

A THESIS

Submitted to
Michigan State University
in partial fulfullment of the requirements
for the degree of

MASTER OF ARTS

Department of Geography

1985

To James A. Rinier
Yes, I dig the soil, plants, landscape... And
I thank you, Professor Rinier, for the many
years of enthusiastic instruction you have
given on these topics within the Department of

Geography at Kent State University, Ohio.

ii

ACKNOWLEDGEMENTS

I would like to extend a most special "Thank You" to Dr. Jay
Harman, my advisor, for the many hours of discussion he has given me on
my academic program in general, and on this thesis in particular. His
skills has a researcher and instructor, his appreciation of nothing
less than superior quality, and his perceptive advice have been
invaluable contributions. Sincere thanks go to my thesis committee,
Dr. Bruce Pigozzi and Dr. Robert Thomas, for their questions,
suggestions, and critical review of the thesis draft, and to Dr.
Delbert Mokma, Department of CrOp and Soil Science, for his assistance
in acquiring the necessary soil surveys and suggestions concerning the
soil analyses. I am grateful to all members of the Department of
Geography, faculty, staff, and<uilleagues for their instruction and
assistance, tolhu.Gary Manson, chairman, for the financial support
provided through a fellowship and graduate assistantship, and to Mike
Lipsey for the innumerable pleasant hours I have spent in the

cartographic center.

iii

TABLE OF CONTENTS

LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . .
LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . .
INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . .
Chapter

I. REVIEW OF LITERATURE AND STATEMENT OF THE PROBLEM .

Literature Review . . . . . . . . . . . . . . . . .
Statement of the Problem . . . . . . . . . . . . .
Statement of Hypotheses . . . . . . . . . . . . .
The Study Rationale . . . . . . . . . . . . . . .

II 0 METHODS O O O O O O O O O O O O O O O O O O O O O O 0

Definition of the Study Area . . . . . . . . . . . .
Data Retrieval (the Sampling Design) . . . . . . . .
The Belt Transects . . . . . . . . . . . . . . .
The Presettlement Forest . . . . . . . . . . . .
Soil Texture . . . . . . . . . . . . . . . . . .
Data Analysis . . . . . . . . . . . . . . . . . . .
Analysis of Forest Composition and Soil Texture
Relationships . . . . . . . . . . . . . . . .
Analysis of the Soil Texture Gradient . . . . . .
Analysis of Forest Physiognomy and Soil Texture
Relationships (Transects B and C) . . . . . .

III 0 RESULTS 0 O O O O I 0 O O O O O O O O O O O 0 O O O 0
Analysis of Forest Composition and Soil Texture

Relationships . . . . . . . . . . . . . . . . . .
Presettlement Forest and Soil Texture . . . . . .

Tree Species Categories-Soil Texture Class Frequencies.
StatiStical ReSUItS O O O O O O O O O O O O O l O O O 0

Analysis of the Soil Texture Gradient . . . . . . .

Location of a Sharp Gradient .. . .. . .. . ..

Student's T-Test Results . . . . . . . . . . . .
Regression Analyses . . . . . . . . . . . . . . .
Analysis of Forest Physiognomy and Soil Texture
Relationships (Transects B and C) . . . . . . . .
Needleleaf Tree Percentages . . . . . . . . . . .

iv

Page

vi

vii

1O
10
11

13

13
13
13
16
18
21

21
2M

26

29

29
29
3O
32
3U
3U

36

38
38

Chapter

Forest Physiognomy-Soil Texture Relationships . .
ForestPhysiognomy-SoilTexture Relationships

South andNorth . . .. . .. . . . . . . . .

Results Summary . . . . . . . . . . . . . . . . . .

Iv. DISCUSSION 0 O O O O O O O O O O O O 0 O O O O O O O O

The Hypotheses and Related Observations . . . . . .

Forest Composition and Soil Texture Relationships

The Northward Gradient in Coarse-textured Soils .

Forest Physiognomy and Soil Texture Relationships

Recommendations for Further Research . . . . . . . .

v0 SUWARY AND CONCLUSIONS 0 O O O O O O O O O O O O O 0

LIST OF REFERENCES 0 O 0 I 0 O O O O O O O O O O O O O O I O

APPENDICES O O O O O O O O O O O O O O O I O O 0 O O O O I 0
Appendix

1. The Sources of Recorded Data . . . . . . . . . . . . .

2. The Data Recorded . . . . . . . . . . . . . . . . . .

3. Tables Compiled in the Data Analysis . . . . . . . . .

Page

“0

Mo
113

AA
AA
ALI
A5
A?
51
55
58

62

62
66

7O

LIST OF TABLES

Table Page
1. Characteristics and final classes of soil management units
used in the analysis . . . . . . . . . . . . . . . . . . . 20

2. Tree species categories-soil texture class relationships,
the statistical results . . . . . . . . . . . . . . . . . 33

3o StUdent'S t-teSt PeSUltS o o o o o o o o o o o o o o o o o o 35

A. Forest physiognomy and soil texture relationships, the
statistical reSUltS O I 0 O O O O O O O O O O O O O 0 0 0 N1

5. Forest physiognomy and soil texture relationships south
and north, the statistical results . . . . . . . . . . . . N2

A2a. Presettlement forest species . . . . . . . . . . . . . . . . 66
A2b. Tree species categories . . . . . . . . . . . . . . . . . . 67
A20. Soil management units . . . . . . . . . . . . . . . . . . . 68
A2d. Soil texture classes . . . . . . . . . . . . . . . . . . . . 69

A3a. Tree species categories-soil texture classes--the
contingency tables .. .. .. .. .. .. .. .. .. .. 70

A36. Relative proportions of needleleaf trees to all trees
(calculated at 9.6 kilometers) .. . .. . .. . .. . .. 73

ABC. Forest physiognomy—soil texture classes--the contingency
tables 0 O O O O O O I O O O O O O O O O O O O O O O O O O 7”

A3d. South and north forest physiognomy—soil texture classes--
the contingency tables . . . . . . . . . . . . . . . . . . 75

vi

The study area .

LIST OF FIGURES

Location of the belt transects and the sampling design .

Frequency percentages averaged from the three transects
Graphic representations of the linear equations

Northward trends in needleleaf percentages and mean

soil texture .

vii

Page

14
15
31
37

39

INTRODUCTION

The "tension zone" in the central lower peninsula of Michigan is a
south-north change in vegetation from a predominantly deciduous forest
to one characterized by a mixture of deciduous and coniferous trees.
Like "tension zones" elsewhere, this one separates two vegetation
associations with species from either association attaining dominance
in response to local site factors. This sharp transition may be
described floristically through an analysis of the presence or absence
of particular species, or ecologically by the change in the relative
dominance of all plant species. More simply, it may be described
physiognomically} for example, the frequency of needleleaf trees in the
present (and presettlement) forest appears to increase abruptly
northward across the tension zone. The nature and possible locational
control of this physiognomic transition will be the focus of this
thesis.

In Michigan, the northward gradient from a deciduous forest to a
mixed coniferous forest appears to be controlled ultimately by climate,
but the sharpness of the change may be attributable to other factors.
These factors may be biotic (e. g. species competition), topographic
(e. g. specific to landform type) or edaphic (e. g. soil texture) in
nature. This study is concerned specifically with the possible
relationship between forest composition and the texture of the upper
soil profile across the tension zone. The underlying hypothesis is
that evergreen species, increasing in abundance northward, show a high

1

2
association with coarse-textured soils, and that these soils are more
prevalent north of a sharp transition across central Lower Michigan.
If this view is correct, the distribution of needleleaf trees is
edaphically controlled and the underlying soils pattern across the

central lower peninsula helps define the tension zone.

CHAPTER I

REVIEW OF LITERATURE AND STATEMENT OF THE PROBLEM

Literature Review

References to the Michigan tension zone occur in three broad
contexts in phytogeographical literature; studies of the major forest
associations of the lower peninsula, the tension zone itself, and the
relationship between soil and forest distribution near, or within, the
study area. This review will focus on a summary of the literature
within each context.

In an early description of Michigan plant geography, Real and
Wheeler (1892) maintained that the flora of the lower peninsula was
generally'common to all regions due to the "uniform surface of the
peninsula," and that the "pinelands" of the northern lower peninsula
were intermixed with hardwoods south to a latitude of #3 degrees north,
with the pinelands becomming less frequent to nonexistent south of that
latitude. Darlington (1945), reemphasized the importance of that
latitude of change in his review of the taxonomic literature on the
flora of Michigan, stating that the principal deciduous area was south
of’that line. Dice (1931) described the boundary between two biotic
provinces, Ohioan and Alleghenian, at that approximate latitude, and
noted a belt of intergradation along this boundary where components of
the Ohioan province, characterized by its deciduous trees, intermingle
with components of the Alleghenian, an extensive development of pines.

3

The importance of this vegetation boundary near A3 degrees north
latitude has also been summarized by other investigators in the
description of nmjor forest associations. For example, buchols (1935)
stated that the boundary between the "Hemlock-White Pine-Northern
Hardwood Region" and a deciduous forest to the south crosses the
central lower peninsula of Michigan near this latitude. He emphasized
the transitional nature of this mixed forest association with its
southern boundary'("a great tension zone") consisting of large areas of
either coniferous species, deciduous species, or a mixed composition of
northern and southern climax types. Nichols speculated that deciduous
species would succeed at localities where they were edaphically
favored. Braun (1950) established a similar line between a "Hemlock
Hardwood” andaa"Beech Maple forestJ' Braun.and Nichols emphasized
the mixture of climax forest types across the boundary, supporting an
earlier conclusion of Cowles (1901) and, in Michigan, Whitford (1901)
that edaphic or physiographic climaxes may persist in a region
dominated by another major plant association. Curtis (1950) extended
the forest association boundary into Wisconsin, separating a southwest
southern-hardwood and a northeast conifer-hardwood forest association.
The boundaries established by Braun, Nichols, Dice and Curtis are all
coincident across central Lower Michigan.

The existence of a transition between major plant associations
described by these authors was not always supported by other research.
Quick (1923) concluded that there was no persistent separation of
forest types in Lower Michigan and that the beech-maple association was

capable, if left undisturbed, of occupying all of the soils of the

5

region. Gleason (192A) supported Quickms conclusion, emphasizing the
successional nature of the evergreen species. Benninghoff (1960)
agreed with Quick's "concept of unity" (one climax forest association)
after examining the beech-maple stand phytosociology both north and
south of the vegetation boundary defined by Braun and Nichols.
Therefore, one possible objective of an examination of the Michigan
tension zone would be to establish the location of significant change
in forest composition, hence, define the occurrence of more than one
plant association, (successional or climax) across central Lower
Michigan.

Several people have presented information concerning possible
controls of plant distributions across the tension zone of Michigan.
Livingston (1903) referred to a "zone of tension" when he described the
upland plant communities of Kent County, Michigan. Although he was
recording vegetation distribution on a somewhat disturbed landscape, he
noted a significant change in the predominant vegetation cover (from
deciduous to coniferous tree species) along "a wavy east and west line
passing through Rockford" (page 51). He suggested that the change in
plant species could be attributed to a south to north trend toward
cooler climate. Livingston emphasized, however, the importance of the
distribution of contrasting surficial material and the consequent
variation in the ability of the surficial material to hold water
(dependent on particle size). He suggested that the lighter surficial
material of Northern Kent County is less suited to hardwood species.
This conclusion was supported in a later study by Livingston (1905) in
Roscommon and Crawford Counties.

D. Harvey (1919) referred to a "great tension zone" in Lake County

and adjacent areas where the northward extension of the deciduous
forest formation intermingled with the southern relicts of the
northeastern evergreen forest formation. He stressed, however, that
the evergreen forest was an edaphic climax which had maintained itself,
and may do so perpetually, against a successional conversion toward the
regional climax association (mixed hardwood). Although this may be so,
he suggested: "It is evident that ecological conditions permit in
certain widely distributed areas over the entire region the development
of the deciduous climax forest . .. either as maple-beech or mixed
hardwood" (page 213).

Potzger (19u8) examined temporal forest change in the tension zone
as shown by pollen data. He defined the zone as an area approximately
thirty miles north and south of a line extending from Muskegon to
Saginaw. Although Potzger found evidence that temporal fluctuations in
vegetation along the tension zone far exceeded those in locations north
and south of the zone, he, as well as Curtis (1959) in Wisconsin,
identified a persistence in the tension zone throughout glacial times.
Potzger concluded that "it is not a time factor influencing the
northward movement of plants," and that the latitudinal gradient in an
unspecified climatic control appeared sufficient to prevent such a
successional trend through Michigan vegetation history.

In examining the transition from prairie and southern hardwoods in
southwestern Wisconsin to a conifer-hardwood forest in the northeast,
Curtis and McIntosh (1951) found only a continuum of forest change
rather than a sharp transition. The lessened gradient, they suggested,

may be attributed to the simpler floristic composition of the southwest

7
vegetation and the lack of a sharp corresponding gradient in certain
environmenta1.factors. They hypothesized that a greater floristic
complexity, or a sharper environmental gradient, may result in a more
distinct change in vegetation across a tension zone.

J. C. Elliot (1952, 1953) worked within Missaukee County, assuming
it to be representative of the tension zone. He concluded that: (1)
the vegetation now (a disclimax) is very different from that
interpreted from the original survey; and, (2) the data support the
contentions of Quick (1923) that the whole lower peninsula would
support a maple-beech association given time for the vegetation to come
into equilibrium with the climate. Ekemaintained that disturbance
factors (lumbering and fires) have eliminated coniferous species as
dominants within the region, and that the normal course of succession
is toward a maple-beech association.

T. Hushen, et. al. (1966) found that the presettlement forest in
Montcalm County consisted of a complex and variable mosaic of forest
associations characterized by abundant white pine and beech. They
observed that white pine, beech, and red maple maintain a "broad
ecological tolerance" for different soils which may explain their
prominence in the transition zone. Later, Kapp (1978), in a study of
presettlement forests in the Pine River watershed of Montcalm and
adjacent counties, identified the botanical ranges of several forest
species that compete within this area of the tension zone.

Recently two research projects were completed which focused on the
distribution of flora across the tension zone. McCann (1979) located a
"floristic tension zone"Iwhich she defined as a concentration of the

northern range limits of southern plants. Initially she established

that the range limits of northern species are not significantly
concentrated and therefore are not indicative of a fflxudstic tension
zone. In an attempt to determine an "ecological" or "physiological"
explanation for this concentration of southern species ranges, she
identified a number of possible causes and examined the gradient
expressed by each factor across the zone. . She noted that the
latitudinal change in growing degee days corresponded closely with the
designated floristic tension zone and concluded that this factor
(described as insufficient heat) limited enough species to produce the
transition.

Brewer (1982) studied the ecological basis of the tension zone by
examining the range limits of tree species. He noted a correlation
between the northern limits of southern trees and the southern limits
of hemlock and white pine. From this observation, Brewer concluded
that the "northern conifers played a major role in causing the
presettlement tension zone for southern treesJ‘

Soils, as an environmental control within the Michigan tension
zone, have not been adequately studied. Livingston (1903, 1905)
observed the presence of pines on a lighter surficial material but made
these observations at only two locations in the lower peninsula. As
mentioned, Hushen, et. al. (1966) examined soils within only one
county. McCann (1979) argued that soils were not a causal factor in
the tension zone on the basis that the glacial drift boundaries failed
to correspond with the floristic boundary, and also that the present
soil differences (e. g. podzolization) were probably the result of

vegetation and climate rather than causing the vegetation difference.

9

That soils may be important to plant distribution is clearly
revealed in other works. Wilde (1933) emphasized the correlation
between pines and sandy soils, hardwoods and loamy soils, and conifers
and hardwoods and clayey soils in the Lake States. Transeau (1935)
suggested that "mature" and "immature" soils may be an influence along
tension zones. Eggler (1938), in Wisconsin, observed that maple-
basswood forests grow only on till soils whereas pines occur on
extensive areas of outwash. Nutter (1973) identified contrasting
trends in vegetation composition on soils of different textures in
Antrim and Montmorency County.

Veatch (1959) published a map of the Michigan presettlement forest
which clearly depicts a northward transition from hardwood to mixed
coniferous forest across the tension zone. He compiled the map using
the established correlation between soils and vegetation distribution
enhanced at a few localities with information from survey notes and/or
botanical literature. The location of the vegetation transition
corresponded closely with a sharp gradient northward toward a
predominance of coarse-textured soils depicted on Veatch's earlier
soils map (1953L

More recently, data concerning water holding capacities of various
levels in the soil profile were plotted on several maps (Schneider and
Erickson, 1967). lidecrease in soil water holding capacity, at all
levels of the soil profile, was depicted northward across the tension
zone.

The northward trends observed in forest composition (toward mixed
coniferous), soils (toward coarse-textured soils) and soil water

holding capacities (toward lowered capacities) coincide across central

10

Lower Michigan. Eyre (1968), in the book Vegetation and Soils,

 

emphasized that a coarse and siliceous soil will weather to a material
that permits rapid leaching. It is well documented (e. g. Eyre, 1968;
Daubenmire, 1962) that coarse soil texture is associated with a low
water holding capacity, low nutrient availability, and low pH, and in
sum may be an important controlling factor in plant distribution. The
association between this factor and forest composition across the

tension zone should be more closely examined.
State-eat of the Problem

There appear to be three areas of insufficient understanding that
concern the tension zone: (1) whether a distinct south-north
transition in the major forest associations occurs in central Lower
Michigan; (2) the identification of a physical factor or set of factors
most closely associated with forest composition across this transition;
and, (3) the latitudinal distribution of that physical factor(s).

Statement of Hypotheses. The main hypotheses of this study are

 

concerned specifically with the latitudinal distribution of coarse-
textured soils northward across the tension zone, and the relationship
between forest composition and underlying soil texture in this region.
The first hypothesis is that presettlement forest composition, as
defined by the number of individuals of representative tree species, is
significantly associated with soil texture. Secondly, it is
hypothesized that the extent of upland areas characterized by coarse-
textured soils increases significantly northward across the tension

zone. I anticipate that a test of these hypotheses will reveal a close

11
association between needleleaf evergreen treesanuicoarse-textured
soils, and that the occurrence of these soils increases sharply from
south to north.

To further clarify these patterns, I will test three
subhypotheses which evaluate the trend in forest physiognomy, as an
indicator of the forest transition, and the corresponding association
to soil texture. These hypotheses are: (L)the percentage of trees
that are needleleaf increases significantly northward across the
tension zone; (2) the needleleaf trees are more site specific to sands
than are the broadleaf deciduous species to finer-textured soils; and,
(3)a.stronger association occurs between tree physiognomic type and
soil texture north of a sharp transition toward coarse-textured soils
within the tension zone.

The Study Rationale. An identification of all aspects of forest

 

composition that change across the Michigan tension zone is beyond the
expectations of this thesis. The presence of a transition, however,
may be better determined through an examination of forest physiognomy.
I will focus on this criterion in the examination of vegetation across
the tension zone.

If the location of representative broadleaved or needleleaf trees
is significantly associated with a physical factor such as soil
texture, this factor may be an environmental determinant of local
forest distribution. Moreover, if soils of specific textures are
distributed along a nonuniform gradient northward across the tension
zone, the forest transition may be related to this gradient. The
significance and strength of the association between representative

tree species and various soil textures should indicate the degree of

12

correspondence and therefore define the relationships that exist
between forest composition and underlying soil texture across the

Michigan tension zone.

CHAPTER II

METHODS

Definition of the Study Area

I incorporated background literature and a knowledge of the
physical environment in the demarcation of the study area. J. E.
Potzger’(19u8), in his pollen research across the tension zone, defined
the area as extending approximately thirty miles either side of a line
from Saginaw to Muskegon. This zone coincides with a transition in
soil types depicted on Veatch's soils map (1953) and a decrease in the
soil profile water holding capacity (Schneider and Erickson, 1963).
The tension zone examined in this study was defined by these works

(Figure 1).
Data Retrieva1.(the Sampling Design)

The Belt Transects. The sampling of vegetation and soils was
completed along three belt transects which extend south to north across
the tension zone in an upland physiographic region (Figure 2). No
transects were located in the Saginaw Bay Lowlands where the low
relative relief and the predominance of poorly drained mineral soils
within the lowlands might obscure soil texture-forest relationships.
Furthermore, lowland sites are known to have a successional development
and climax forest which differs from that characteristic of a region.
The sampling area, secondly, was located east of the approximate

13

111

 

 

 

 

 

 

I H; Vegetation Tension Zone
(Potzger. 1948)

b

-_., Soils Transition
(mm. 1953)

Transition in
.-..... Soil Water Hoiding

Capacities
(Schneider a. Erickson, 1965)

 

o 150 km

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

L L L l

 

Figure 1. The study area.

 

15

Michigan Lower Peninsula

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Belt Transect I 0 43 km

Physical regions adapted trom Veetch. 1953 -

Figure 2. Location of the belt transects and and the sampling design.

16

longitude of Grand Rapids, Michigan, where at least some of the
climatic influences of Lake Michigan diminish (N.D. Strommen, 1971i).
The acquisition of data from three separate transects (identified as A,
B, and C) was designed to optimize the reliability of the study
results, and their location in a physiographic region with only limited
wetlands should facilitate the study of the vegetation and soil
transition on upland sites.

Each belt transect is a linear plot one township wide 0&6 km. or
6 mi.) and ten townships long (96 km. or 60 mi.). The width of these
transects provided sufficient data points at close intervals to examine
soil texture-forest relationships from south to north. The total area,
576.6 square kilometers (360 sq. mi.), and the number of samples taken
allowed separate analyses to be maintained along each transect. The
total area from which data were obtained was 1728 square kilometers
(1080 sq. mi.L

The Presettlement Forest. Agricultural and urban land uses are

 

currently dominant in the tension zone, and the woodlands that remain
have suffered considerable alteration. Consequently, they are limited
in extent and the spatial variation in their composition may be more
related to succession and peculiarities of site history. The
composition of the presettlement forest was probably more closely
related to the physical environment and, hence, was examined in this
study.

Presettlement forest data were obtained from the General Land
Office Survey notes (seeAppendix 1a).This sourme has been used

extensively in vegetation studies elsewhere, e. g. Ohio (Sears, 1925),

17

Indiana (Potzger, Potzger and McCormick, 1956), Michigan (Kenoyer,
1930, 19311, 19110, 19113; Hushen, et. al., 1966; Brewer, et. al., 19811).
Crankshaw, et. al. (1965) used survey notes as a source of quantitative
data in a multiple regression analysis which tested the relationship
between vegetation type and various soil parameters, and Nutter (1973)
used presettlement vegetation data in a quantitative analysis of "soil
texture and forest change" in Antrim and Montmorency counties. Bourdo
(1956) outlined the use of the General Land Office Survey notes,
detailing the techniques used to identify bias and its application to
quantitative research.

The survey notes are arranged by township in large volumes located
in the Lands Division of the Department of Natural Resources, State of
Michigan. As the surveyors mapped the section lines within a township
they recorded witness or bearing trees (at section corners and at the
middle of a section line) and line trees (along a section line) by
common tree name, location, and diameter at breast height. I recorded
only tree name and location.

Data available from all sections within a township exceeded that
needed for effective statistical analyses. Therefore, II collected a
stratified random sample along each transect by using a computer-
generated random numbers table which listed numbers one through six (J.
M. Lipsey, 1983L. Using these numbers, I chose one or more sections
from each six-mile wide row, a technique that ensured at least one
sampling unit (15-18 samples) at 1.6 kilometer (1 mile) intervals from
south to north (see Figure 2). A stratified random sampling design was
advantageous because it allowed for an examination of the entire

transect as well as specific portions along it from south to north.

18

The sample of presettlement trees obtained by this method was not
tested for bias, but the presence of bias was estimated by checking for
consistency between the sampled trees and the forest descriptions in
the survey notes. The likelihood that the witness trees represent a
seriously biased sample of the presettlement forest is small unless the
surveyors repeatedly chose trees not representative of the forest. I
saw no evidence of such bias as I collected the data. Further, I
assumed that the presettlement forest was relatively undisturbed, and
hence, would be representative of natural forest distribution. Hushen,
et. al. (1966) tested the sample obtained from the survey notes on
Montcalm County and found that bias in the trees identified was not
present.

Soil Texture. All recorded trees, identified by location in the

 

survey notes, were mapped on detailed soils maps available within the
modern county soil survey reports (post 1960; see Appendix 1bL.Each
mapped tree was assigned to a soil mapping unit and a respective soil
series.

All soil series located in Michigan, irrespective of county, have
been classified into a soil management group (D. L. Mokma, 1982) on the
basis of the dominant profile texture (upper 5 feet), and further
classified by drainage characteristics. Crankshaw, et. a1. (1965)
suggested that probably enough time had elapsed between the land survey
and the soil survey (approximately 100 years) to alter surface soil
characteristics such that a correlation between presettlement forest
composition and soils might be invalid unless characteristics of the B

horizon also were included with the other soil information. The soil

19
management unit, because it represents an average of the A and B
horizon textures, incorporates this information.

The soil textures and appropriate class designations used in this
study are identified in Table 1. I considered only the textural class
of zonal mineral soils, omitting from the data set those trees located
on organic, recent alluvial, and bedrock soils which are localized and
maintain a vegetation composition not characteristic of the regional
pattern. In the original scheme of soil management units U&.L. Mokma,
1982), those mineral soils developed from uniform parent materials are
represented by one number, and soils developed from two parent
materials or with contrasting textures in their profiles are
represented by a fraction. I later designated an average value to the
soils classed between two units and aggregated soils of similar
textures ("Final Class" in Table 1) in order to maintain an adequate
sample within each textural type.

A separate data set was acquired from the county surveys for the
analysis of the soil textural gradient. I used a stratified systematic
sampling technique to ensure an adequate sample, at close intervals,
from south to north. Data were obtained from every second section by
starting in the most southwestern section of the transect and sampling
east then west respectively (see Figure 2). Within each section, the
soil mapping unit was determined from the county soil survey map at .A
kilometer (.25 mile) intervals from east to west and south to north.
Each soil mapping unit was assigned a soils series name and
corresponding soil management unit, which then were combined as labeled
in Table 1. Only the distribution of coarse-textured soils (textures A

and 5) was observed in this analysis.

Table

1.

20

units used in the analysis.

Characteristics and final classes of soil management

 

 

 

 

 

 

 

 

 

Soil Management Units Textural Characteristics Final Class
Texture 1 fine clay, over 60% clay
clay, ”0-601 clay
clay loam &
silty clay loam 2
Texture 2.5 loam and silt loam
Texture 3 sandy loam
Texture 3/2 sandy loam, 20-A0 inches
over loam to clay loam 3
Texture 3/5 sandy loam 20-90"
Texture 3.5 between texture 3 & 9
Texture A loamy sand
1;
Texture 9.5 between texture A & 5
Texture 5 sand with strong subsoil
development
Texture 5.3 sand with medium subsoil 5
development
Texture 5.7 sand with weak or no
subsoil development
Texture “/2 sand to loamy sand 20-90"
over loam to clay loam
Texture N/1 loamy sand, 1A-40 inches, c/f
over clay
Texture 5/2 sand to loamy sand, “0-60"

over loam to clay loam

 

Modified from the Extension Bulletin (El-1262) entitled; "Soil
Management Units and Land Use Planning" by ILL. Mokma; reprinted April
1982, Cooperative Extension Service, Michigan State University.

21

Data Analysis

Analysis of Forest Composition and Soil Texture Relationships.

 

The tree species frequencies on each soil texture class, recorded along
each transect, were arranged in three contingency tables. Overall
patterns were observed by averaging, from the three transects, tree
species frequencies on each soil texture and the frequencies ofzall
tree species on representative fine (texture 2) and coarse-textured
(textures A and 5) soils, and further clarified, by identifying
contrasts and similarities between the transects in tree species
distributions. The first hypothesis addressing the significance of the
association between tree species and soil textural classes was tested
using the Pearsonts chi square statistic. The null hypothesis, that no
association exists between tree species and soil texture, depicted
frequencies that could be expected, and the Pearsonhs chi square test
determined the significance of the divergence between these expected
frequencies and the observed values (Blalock, 1979L

The assumptions of the test are that the population can be divided
into discrete classes, the outcomes are independent of one another, and
the sample size is relatively large. Taylor (1977) suggests, with
respect to sample size, that no more than twenty percent of the
contingency table cells have an expected frequency of less than five.
Although the sample size was large for each transect, the frequencies
of several of the trees were very low. It was necessary to combine the
frequencies of trees with assumed similar habitats to maintain at least

eighty percent of the expected cell frequencies at five or greater.

22

The chi square statistic is approximated by the following

X2- 2 (observed - expected)2
"' expected

The calculation procedure is detailed elsewhere (e.gp Winkler and

equation:

 

Hays, 1975; Blalock, 1979L thice, however, that a greater difference
between the observed and expected frequencies results in a larger chi
square value, and, furthermore, in a contingency table, the computed
result equals the sum of all individual cell chi square calculations.
In this study the contingency table compiled for each transect was
complex (6 rows by 6 columns) and the respective recorded sample size
was large. With everything else equal, the greater the number of
cells, the greater the chi square value, with this increase offset only
partially by the calculation of the degrees of freedom:
df = hP-1)(c«- 1)

where r and c are the number of rows and columns, respectively.
Moreover, the chi square value is very sensitive to the size of a
sample. If the sample is large, as recommended for this test,
"virtually any 'degree' " of a statistical relationship will show up as
a significant result (Winkler and Hays, 1975).

The significance level in the chi square analysis, consequently,
was not only an indication of the relationships between tree species
and soil textures, but also the complexity of the contingency table and
the large data set. Therefore, to enhance the interpretibility of the
results, I calculated two indices which measure both the overall
strength of the association and the predictive power of the independent

variable (soil texture). The phi coefficient (¢) measures association

23

strength by adjusting the chi square values for a large sample (Nie,
et. al., 1975; Winkler and Hays, 1975). The Cramer's V (V) statis:l_c
is a "modified version of the phi which is suitable for larger
contingency tables" (Nie, et. al., 1975, page 2211). The calculation

and-relationship between these two indices is as follows:

' 2
: X2 =
o A: V /|. - 1
phi coefficient Cramer's 11

By dividing the phi coefficient by the smaller dimension (rows or
columns) of the contingency table (L), which has been subtracted by
one, the index is adjusted to vary between zercI(no association) and
one (absolute association). One undeSirable feature of the Cramerfis V
statistic is that the computed value is dependent on the rows or
columns for its interpretation; a larger number of rows or columns will
result in a greater index value (W. J. Conover, 1971).

As an index of predictive association, the asymmetric lambda
statistic was calculated by the following equation:

Ab = SI- of the maxim- values) - maxim: value of the row
in each column marginals (totals)

 

n - maximzm value of the row marginals (totals)

(Winkler and Hays, 1975, Nie, et. al. 1975). The index assessed the
predictive power of the independent variable (soil texture) in
identifying the dependent variable (the occurrence of a particular tree
species) and varied from 0 (no improvement in prediction or reduction
in the probability for error) to 1 (prediction can be made without

error).

2A

The chi square statistic, the Cramer's V, and the asymmetric
lambda indices, calculated for each transect, should indicate: (1) the
significance of the association between tree species and soil texture;
(2) the strength of the association; and, (3) the predictive power of
the independent variable (soil texture). The analysis of Transect A
was completed manually, whereas the stastistics for Transect B and C
were calculated using the Statistical Package for the Social Sciences,
(Nie, et. al., 1975) on the mainframe computer (Control Data Cyber 750)
at Michigan State University.

Analysis of the Soil Texture Gradient. The stratified systematic

 

sample of soil texture allowed the frequency of each textural unit to
be examined at .14 kilometer (.25 mile) intervals from south to north,
and the analysis focused on the frequency of coarse-textured (texture A
or texture 5) soils at each interval. Scattergrams (Nie, et. al.,
1975) were produced to observe the northward trend (frequency of
texture 4 or 5 with distance) and to identify the location of the most
distinct increase in the occurrence of each respective soil texture. I
separated the recorded data at that location, thereby forming two
samples which represented south and north, respectively.

The Studentfist.statistic was computed to determine whether the
mean frequency of soil texture A or 5 increased significantly north of
the established break. The precise calculation of the Student's-t
difference of means test is well documented (e. g. Nie, et. al. 1975,
Blalock, 1979, Winkler and Hays, 1975L. The test assumes that the
samples are drawn from populations of equal variances. An F test,

which measured the similarity in the sample variances north and south,

25

was used additionally to determine whether the results of the actual t-
test computation (a pooled variance estimate), or, if the variances
were not equal, the approximated t-value (a separate variance
estimate) were interpreted (W. K. Klecka, et. al., 1975). The null
hypothesis, that there is no significant increase in the mean
frequencies of texture A or 5 north of the line, was rejected if the
alpha value of a one-tailed test exceeded .01.

Although the t-test revealed the significance of the different
mean frequencies in coarse-textured soils north and south of a
designated line, it can not depict the trend that occurs northward
along each transect. A regression analysis, employing dummy variables,
was used to clarify the northward trend in the occurrence of coarse-
textured soils across the tension zone. This analysis was designed to
examine the frequency of texture 5 (the most coarse soil texture) as it
was related both to distance (I), and to its position north or south of
the line established in the t-test. A designation of north or south
was incorporated into the equation through the use of an additive and
multiplicative dummy variable (D1) where D:1 if I is north of the break
and D=O if I is south of the break. The final regression equation was
as follows:

I = a 4- D11 + b2(D1) + b3D2x

where I is the frequency of soil texture 5, a is the y-intercept, X is
the distance northward along the transect, D1 is the additive dummy,
and D2 is the multiplicative dummy. If I is north of the break the
equation would be as calculated above, but if I is south of the line

the effects of the dummy variable are eliminated from the equation:

26

Y = a '0' b1! 4- b2(0) + b3(0)

I=a+b1x

The examination of the results focused on the significance of the a (y-
intercept) and b (slope coefficient) values in the equation. If Y is
north of the break, the a value would have the additive effect of b2,
and the slope of X (b1) would increase by b3. Significant a and b1
values indicate that the linear trend south (and north) has a slope and
y-intercept different from zero, and significant b2, and b3
coefficients indicate that the linear trend north of the break is
different from the trend south of the breaks The Rz'value, for the
overall equation, depicts the percentage of variance in the northward
distribution of soil texture five explained by the equation. Further
detail on the application and interpretation of linear regression using
dummy variables may be obtained elsewhere (Johnson, 1972; Nie, et. al.
1975).

Analysis pf Forest Physiognomy and Soil Texture Relationships
(Transects B app _1. The first subhypothesis addressed the percentage
increase in needleleaf trees northward across the tension zone. From
the original data recorded on the presettlement forest I was able to
calculate, by township, the percentage of all trees that were
needleleaf. As I was interested in a possible northward trend in the
relationship between the percentage of needleleaf trees and the extent
of coarse-textured soils, I separated the recorded data at the same
location used previously in the t-test examination of the soil texture

gradient. The data sets, south (n1) and north (n2) remained large so it

27
was pos31ble to assume an approximate normal population, and the
significance of the different frequencies of needleleaf trees south
(k1) and north (k2) was determined through a computation of the 2-
statistic, where p, and p2 were the sample proportions south and north
respectively (Yamane, 1973). The true proportions of needleleaf trees

to all trees south ( 7T1) and north ( #2) were examined as follows:

 

“01 “=1 #:k‘I": . 01' P:
Maren; "r "2 '3 'F_—11-fi'll'ln.+ m

I was specifically interested in determining whether there was a
notable increase in the percentage of needleleaf trees north of the
preestablished division across each transect and in the computed
significance of this increase. I rejected the null hypothesis, that
the north percentage is not greater, if theaalpha value exceeded J31
(one-tailed test).

The data concerning presettlement tree species frequencies on
different soil textures were reorganized into a subtable to facilitate
determination of which portions of the overall contingency table
contributed most to the significance of the tree species-soil texture
association (chi square), strength (Cramerus V), and dependent variable
(tree species) predictability (asymmetric lambda). I examined the
frequencies of the broadleaved deciduous species and the needleleaf
evergreen species on each soil texture class. The reduction in the
number of rows and columns (cell frequencies) enhanced the
interpretability of the chi square and Cramer"s V statistic because
their computed values were not increased as much by the complexity of

the contingency table as in the examination of individual tree species.

28
Changes observed in the computed statistics, as they compared to the
examination of individual tree species-soil texture relationships
(hypothesis one), clarify the site specificity of these two
physiognomic groups.

The frequencies of the two physiognomic groups on each soil
texture were separated south and north of the predetermined break in
coarse soil texture. A recalculation of the statistics for each table
was completed to assess the relative increase in association
significance and strength, as well as the predictability of the

dependent variable (tree species),:north of this break.

CHAPTER III

RESULTS

Analysis of Forest Composition and Soil Texture Relationships

Presettlement Forest and Soil Texture. The number of trees

 

 

recorded along each transect, including only those on mineral soils,
was 1089, 1161, and 1236 for transect A,IL and C respectively (see

Appendix 2a). Beech (Fagus grandifolia) was the most common tree

 

recorded, exceeded only by oak (Quercus sppJ along transect C. Sugar

 

maple (Acer saccharum or Acer nigrum) was a well-represented species

 

 

along transect A (highest frequency) and B, but was lower in frequency
along transect C. In contrast to the distribution of sugar maple, oak
was most common on transect C and had a very low occurrence along

transect A. Elm (Ulmus spp.), maple (Acer spp., probably Acer rubrum),

 

 

black ash (Fraxinus nigra), lynn (Tilia americana) and other species

 

 

(e. g. Ostrya virginiana, Prunus spp.) were infrequently recorded. I

 

assumed these species to be similar in their preferred habitats, hence,
they were combined into one group which I labeled "other mesic
deciduous species" (Harlow, Harrar, and White. 1979).

The needleleaf trees recorded were pine (Pinus strobus, and

 

unspecified pines), hemlock (Tsuga canadensis), cedar (Ipuja

 

occidentalis) and tamarack (Larix laricina). Tamarack and cedar were

 

 

combined into one category, which I regarded as representative of a
poorly drained habitat. The combined frequency of tamarack and cedar

29

30

was the lowest of all tree categories and,jJisome cases,]Zcombined
the frequencies of these two species with hemlock, forming a category
which I labeled "other needleleaf species".

The recorded occurrences of fine and coarse-textured soils varied
between the transects (see Appendix 2b). A.higher frequency of fine-
textured soils was recorded along transect A than transect B or C.
Transect C had the highest frequency of coarse-textured soils (textures
A and 5), soil texture class 3 was most frequent along transect B, and
texture class 15 (coarse over fine soil textures) was infrequent along
the three transects.

Tree-Species Categories-Soil Texture Class Frequencies. Several

 

 

overall patterns were observed in the distribution of a tree species
category on all soil textures (Figure 3a) and in the distribution of
all tree species categories on representative fine and coarse-textured
soils (Figure 36). Beech, oak, and other mesic deciduous species,
although most frequent on soil texture 2, were generally abundant on
all soil texture classes. Sugar maple appeared to be primarily site
specific to fine-textured soils (soil texture 2),Iwhereas pines and
hemlock were identified primarily on coarse-textured soils, being most
frequent on soil texture 5 (Figure 3a). Beech was the most frequent
tree on representative fine (texture 2) and coarse-textured (texture A
or 5) soils, exceeded only slightly by pine on soil texture five, and
the occurrence of needleleaf species, relative to the other deciduous
species, increased noticeably'on the coarser-textured soils (Figure
3b).

An examination of the relationships present along the individual

31

Sugar Oak
Mapm

 

 

41114111111

     

2345c" 2345:" 2345M 2345c/I

a. Distribution of tree species or species category on all soil
texture classes.

 

4AL4AAAAL111

 

:9“: EA;
3335:: gages; 3

Sugar
0“
Other
PIne
Hml

b. Distribution of all tree species or species categories on
representative fine and coarse-textured soils.

Figure 3. Frequency percentages averaged from the three transects.

32

transects, however, revealed several major contrasts to the general
pattern8»(see Appendix 3aL. First, the frequencies of beech and oak
were actually greatest on texture 2 on only transect A, but were
highest on texture 3 along transect B and on texture 5 along transect
C. These patterns correspond directly to the extent of each soil
texture, with texture 2 being most frequent along transect A, texture 3
along transect B, and texture 5 along transect C. Second, the
localized distribution of sugar maple on soil texture 2 was most
apparent on transect C (70.6%) and least apparent on transect B (21%),
and further, along transect B, pine appeared less site specific to soil
texture 5 (31%, as compared to 62% on transect C). Third, pine was
actually most frequent on soil texture A along transect A, in contrast
to transects B and C where it was most frequent on texture 5. Finallyg
beech was the most common tree on the representative fine and coarse-
textured soils along each transect, with the exception that it was
exceeded by pine on soil texture 5 along transect B, and exceeded by
oak on texture A and 5 and pine on texture 5 along transect C. These
contrasts indicate that there is some east-west variation in the
observed site-specificity or relative frequency of a tree species
category.

Statistical Results. The chi square value, computed for each
transect, exceeded a critical value of approximately 53 at an alpha
value less than .001 (Table 2), indicating that the summed divergence
between observed and expected frequencies was significant. The
Cramer's V index, varying from 0 to 1, was highest along transect C
(.21) and lowest along transect B (.16) (Table 2). The chi square

values computed for each transect indicated, in addition to the

33

Table 2. Tree species categories-soil texture class
relationships, the statistical results.

 

 

Transects Chi square df(r-1)(c-1) Cramer"s V Asymmetric

 

lambda
Transect A 185.73 2A .21 O
Transect B 122.30 2A .16 .007
Transect C 231.79 20 .22 .03

 

Critical chi square is approximately 53

at alpha ( .001

3A
significance of the tree species-soil texture association, that the
values computed for the Cramer‘s V are statistically accurate
estimations of the strength of the association (Norcliffe, 1977). The
asymmetric lambda index calculated for each transect, varying from 0 to
1, equaled approximately zero (Table 2). There was no decrease in the
probability for error, in choosing a tree species, if a soil texture is

known .

Analysis of the Soil Texture Gradient

Location pf 3 Sharp Gradient. Scattergrams, which showed the

 

frequency of textures A and 5 at .A kilometers from south to north
along each transect, depicted a sharp increase in soil texture 5 (the
most coarse soil) at 67.2 kilometers along transects B and C, and much
less distinctly along transect A. The data set was separated, forming
a south and north sample, at that location.

Among texture A soils, no sharp increase was recognizable along
either transects B or c. On transect A, however, I observed a
significant break in the distribution at 76.8 kilometers, and because
it was on these soils in transect A that needleleaf trees were most
frequently recorded, I examined additionally the significance of this
break.

Student's '_I‘_ Test Results. The mean frequency of soil texture 5 at

 

.A kilometer intervals along transect A was 1.0A for the sample south
of 67.2 kilometers and 1.A6 for the north sample (Table 3). This
slight increase north of the recognized break was not significant
(alpha = .05), and, furthermore, the overall occurrence of this soil

along transects B and C and in each transect the north mean exceeded

Table 3. Student's t-test results.

35

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

TRANSECT F—TEST T—TEST
DIVISION MEAN Value Sig. Value $19
TRANSECT A (texture 5)

3 th ' '
OU l
T9N-T15N 1'05 I l
1.07 I .722 -1.89 . .060
North 1 46 ' |
T16N-T18N ' L (pooled)
TRANSECT a (texture sI
SOUUT I I
T7N-T13N 93 I I
h 2.59 : .000 -9.80 : < .001
Man
T14N-T16N 3'74 1 (sepfrate)
TRANSECT C (texture 5)
1 1
South I I
T5N-T11N 2-84 I I
2.17 I .000‘ -9.96 I < .001
North 7 I I
T12N-T14N '03 ! (”pfmm
TRANSECT A (texture 4)
South 1 307 T :
T9N-T16N - I
N h 2.54 : .000 -8.84: <.OOI
on
T1 7N-T1 8N 4.604 : (sears re)

 

 

36

that in the south (Table 3). The negative t-values (a separate
variance estimation) were significant at an alpha value less than .001,
thereby indicating that the mean south frequency of texture 5 (u,) was
significantly less than the mean north frequency (uz) along these two
transects. The mean frequency of soil texture 14, south of 76.8
kilometers along transect A (1.307), was significantly less than north
(“.60“) at an alpha value less than .001.

Regression analyses. The linear equation determined for transect

 

A had a low 32 value (.12), indicating that very little of the variance
in the occurrence of soil texture 5 was explained by the equation.
This result, as well as the insignificant t-test result, prevented any
detailed observations on the linear trend of texture 5 north and south
of the designated break along this transect. The linear equation for
transect B, however, had a R2 value equal to .4” and the y-intercept
(a), as well as each of the slope coefficients, was significantly
different from zero. These results indicate that, along transect B,
the south (and north) gradient was significantly different from zero
and the linear trend to the south (I = .55 + .01!) was significantly
different from the trend to the north (I = -6.05 + .12!) (Figure lla).
The linear equation calculated for transect C yielded a high “2 value
(.57) and the y-intercept (a), as well as each of the slope
coefficients, was significantly different from zero. As in transect B,
the south (and north) gradient was significantly different from zero
and the linear trend to the south (I = 1.95 + .26!) was significantly
different from that north of the break (I = 32.75 + -.31'!X) (Figure

lIb). The south and north linear equations for transects B and C reveal

37

 

 

 

 

 

a.
'TH‘WMSECH'B
In 5*
2
a 5-
g .
a I-
a 4
Z 3*
é“
‘5 1? SOUTH
#1
978 13.2 28.8 38.4 38715715 8'2 T8 85.4 68 kilometers
[3.
\
3.. \ musecrc
. ‘\
In”: \
528» \
322: \
3 p \
3...
° b
14:
g.
u. 6'
b ——
2»

 

 

 

8T8 182 25.8 38.4 18 5?.8 8f2 78.8 88.4 8‘8 18mm

Figure 4. Graphic representations of the linear equations.

38
a similarity in the trends along both transects south of the break, but

distinct differences are observed north of the break.

Analysis of Forest Physiognomy and Soil Texture Relationships

(Transects B and C)

Needleleaf Tree Percentages. A northward increase in the

 

 

percentage of all trees that were needleleaf occurred along both
transects; this percentage reached a maximum of N7 per cent at 96
kilometers along transect B and M5 per cent at 86a" kilometers along
transect C (see Appendix 3b). Along transect B, the mean percentage of
needleleaf trees south of the line (29%) was less than north of the
line (N3%). .Along transect C, the differences were less, with averages
of 25 per cent south of the line and 37 per cent north of the line.
The computed z-statistic was high for transects B (161.79) and C
(1117.62), and, at an alpha value less than .001, the north percentages
were significantly greater.

A graph of the northward trends in the percentage of needleleaf
trees and mean soil texture (an average of the frequencies of all
textural classes) shows a correspondence between these two measures
(Figure 5). In addition to the overall similarity in the two
distributions, the decline in needleleaf trees along transect C at 28$)
kilometers and the sharp increase again at 38¢“ kilometers corresponds
to a finer and coarser mean soil texture, respectively, Just as the
decline in needleleaf trees along this transect at 86.“ kilometers

corresponds to a trend toward finer-textured soils.

39

°/o
100

Needleleaf Tree Percentages

 

 

 

 

 

 

i l L l 1 l i L l l
9.6 19.2 28.8 38.4 48 57.6 67.2‘ 76.8 86.4 96 km.
south —— north

Figure 5. Northward trends in needleleaf tree percentages and mean
soil texture.

NO

Forest Physiognomy-Soil Texture Relationships. The deciduous tree

 

species were the most frequent physiognomic group on all soil textures,
although they were exceeded by needleleaf speciescnisoil texture 5
along transect C (see Appendix 30). Deciduous tree species exhibited
little site specificity to a soil texture, having similar frequencies
on all soils (transect B), or having highest frequencies on fine
(texture 2) and coarse (texture 5) soil texture (transect C; see
Appendix 30). The needleleaf species were most frequent on the coarse-
textured soils, with 58 per cent of the trees occurring;on texture 5
along transect B and approximately 75 per cent of the trees occurring
on textures M and 5 along transect C.

In each transect the chi square value exceeded the critical value
(20) at an alpha value less than .001 (Table ll). In this analysis of
physiognomic groups, the Cramerks V and the asymmetric lambda indices
increased slightly from those computed in the examination of tree
species categories (Table H and Table 2L.The strongest association
(highest Cramer"s V value; .25) was obtained for the physiognomic
groups along transect C (Table 8).

Forest Physiognomy-Soil Texture Relationships South and North. No

 

difference was apparent in the computed statistics for the south and
north portions of transect B, but a noticeable difference was observed
in the chi square statistic and the Cramerhs V index computed for the
south and north portions of transect C (Table 5). In the analysis of
transect C, the chi square value south of the break (12.63) was below
the critical chi square value (approximately 13), or, in other words,
insignificant (alpha = .01), but north of the break, although the

sample size was less, the chi square value (25.03) was significant

“1

Table “. Forest physiognomy and soil texture relationships,
the statistical results.

 

 

Transects Chi square df(r-1)(c-1) Cramer's V Asymmetric
lambda

Transect B 6“.96 “ .2“ .02

Transect C 76.33 “ .25 0

 

Critical chi square is approximately 20

at “ df and alpha < .001

“2

Table 5. Forest physiognomy and soil texture relationships
south and north, the statistical results.

 

 

Transects Chi square df(r-1)(c-1) Cramer's V Asymmetric
lambda

 

Transect B

south 30.27 “ .210 .002
(0:688)
north 18.0“ “ .200 .063
(n=“73)

Transect C

south 12.62 “ .119 0
(n=898)
north 25.03 “ .268 0
(n=3“8)

 

Critical chi square is approximately 13

at “ df and alpha = .01

“3
(alpha < .001). Furthermore, the Cramer's V index computed for the two
samples indicated that the strength of the association north of the
break (.268) exceeded the computed association strength to the south
(.119). The extent of soil texture 5 (”70% of the sampled sites) was
much greater north of the break, with a resulting high occurrence of
deciduous (62.8% of all deciduous trees) and needleleaf (77.6% of all
needleleaf trees) species on that soil texture (see Appendix 3d). As
in all prior analyses of the predictability of the dependent variable
(tree species or physiognomic group), the calculated asymmetric lambda

value was approximately zero.
Results Sunary

Tree species categories and physiognomic groups showed a
significant association to soil texture classes along the examined belt
transects, with the exception of a statistically insignificant
association computed for the south sample along transect C. Needleleaf
tree species were distinctly more frequent on coarse-textured soils,
and sugar maple was identified primarily on fine-textured soils, but
beech, oak, and other mesic deciduous species showed little specificity
to a particular soil texture class. The mean frequency of soil texture
class 5 increased significantly north of 62.“ kilometers along
transects B and C, though differences occur in their trends north of
the identified break. The occurrence of soil texture class “ increased
significantly northward along transect A. A significant increase in
needleleaf trees (as a percentage of all trees) was measured northward
along the transects and their relative percentages from south to north

corresponded directly to the distribution of coarse-textured soils.

CHAPTER IV

DISCUSSION

The Hypotheses and Related Observations

Forest Composition and Soil Texture Relationships. The first
hypothesis states that presettlement forest composition, as defined by
the number of individuals of representative species, is significantly
associated with soil texture. Based on the results of the chi square
analysis completed for each of the three transects, I reject the null
hypothesis that there is no association and conclude that the
association is statistically significant.

The high chi square values (and corresponding low alpha values),
however, were partially the result of the large data set recorded for
each transect and the large number of rows and columns in the
contingency tables. The Cramer's V index, adjusting for the large
recorded samples, indicated that the strength of the overall
association was not great. The localized distribution of pines on
coarse-textured soils and sugar maple on fine-textured soils increased
the Cramer's V index, while the high occurrence of beech, oak, and
other mesic deciduous species on all soil texture classes decreased it.
The frequency of those species not site specific exceeded the combined
occurrence of pines and sugar maple and is possibly the explanation for
the low computed values. Transect C had the highest Cramer's V value,
the most pines, and the greatest extent of coarse-textured soils,

““

“5

whereas transect B had the greatest extent of soil texture 3, a texture
to which no tree species appeared to be site specific, and the lowest
Cramerhs V value. The predominance of beech on nearly all soil
textures and its high overall frequency was the primary reason for the
low asymmetric lambda index values. In nearly every case for the three
transects, beech would be the predicted tree when the soil texture
class is known.

The contrasts in the relative frequencies of tree species
categories between the three transects appear to depict an east to west
transition in forest composition. Beech and sugar maple declined in
frequency westward, while pine, and especially oak, increased westward.
Because pines appear to be noticeably specific to coarse-textured
soils, their increase may be an expression of the increase westward in
coarse-textured soils. "Oak upland forestfi occupies more land area
than any other natural community in southern Michigan (Gysel and Arend,
1953; as cited in Parmelee, 1953) and this community type has been
described by Braun (1950) as extending northeastward over the
Wisconsinan-aged drift, from the Prairie Peninsula into southern
Michigan, as large isolated tracts. The westward increase of oak, as
well as the decrease of beech and sugar maple, that I detected may be
an expression of the overall transition from beech-maple to oak
dominated forests.

The Northward Gradient in Coarse-textured Soils. I hypothesized,

 

secondly, that the extent of upland areas characterized by coarse-
textured soils increases significantly northward across the tension
zone. Based on the t-test results, I accept this hypothesis, at least

regarding soil texture class 5 (the most coarse soil) along transects B

“6

and C and texture “ along transect A. At an alpha value equal to .05,
the null.hypothesis, stating that there is no significant increase,
could not be rejected in the analysis of soil texture 5 along transect
A.

The distribution of soil texture 5, and its association to forest
compostion, differed along transect A from transect B and C. The
occurrence of this soil texture was low, it failed to show a sharp
increase northward or a definable northward trend, and pines were not
particularily'frequent on this texture class, occurring most frequently
on soil texture “. It appears that the distribution of this soil
texture along transect A was localized and exhibited contrasting
characteristics, such as poor soil drainage or a high water table,
which were not examined in this study and complicated general
relationships between soil texture and tree species distributions.

Although the mean frequency of soil texture 5 increased.
significantly north of 67.2 kilometers along transect B and C, the
northward trend from that location was distinctly different between the
two transects. There was a sharp increase at 67.2 kilometers along
transect B and a gradual positive trend north of that location in
contrast to transect C, where a much sharper increase was observed at
the south-north division, but the trend north of that location was
negative. The break determined along transect C is located in the
region of the Muskegon River watershed which is predominantly overlain
by glacial outwash that is characterized as a layering of sand and
gravel (Farrand, and Bell, 1982). North of this location the transect

is on coarse-textured till (sandy loam-loamy sand) and to the south the

“7

transect is on medium to fine-textured till, so that there appears to
be a distinct change in the texture of the drift (to very coarse) at
the break examined along this transect in contrast to transect B which
trends north across predominantly fine to coarse-textured ice-deposited
features. Although this distribution of glacial deposits may explain
the difference between the two transects, it was beyond the scope of
this thesis to examine thoroughly the reasons for the observed
northward trends in soil texture.

Forest Physiognomy and Soil Texture Relationships. The tension

 

zone is defined in this study by an abrupt increase in the occurrence
of needleleaf tree species. The z-test results, addressing the first
subhypothesis, show that the percentage of needleleaf trees is
significantly greater north of a significant increase in the occurrence
of soil texture 5. Furthermore, the northward trend in the frequency
of these species and this coarse soil texture, along both transects, is
positive, with any decrease in the frequencies of needleleaf tree
species corresponding to a decrease in the extent of coarse-textured
soils. The similarity between the northward trends in coarse scil
texture and needleleaf tree species and their statistically significant
increase northward across the tension zone suggest that these soils may
partially dictate the distribution of the needleleaf evergreen
physiognomic group.

The second subhypothesis states that the needleleaf trees are more
site specific to sands than are the broadleaf deciduous species to
finer-textured soils. The statistical examination focused on both
physiognomic groups (broadleaved deciduous and needleleaf evergreen)

together, where it is shown that they are significantly associated to

“8

soil texture classes. The greater frequency of deciduous species,
exhibiting little site specificity, resulted in a moderately low
observed association strength and low predictability of either
physiognomic group when given a soil texture class. The overall
results from this examination of physiognomic groups-soil texture
relationship differ little from the analysis of individual tree species
categories, but the site specificity of needleleaf species, over that
of the deciduous species, is further clarified.

Only along transect C was a stronger association (greater Cramerms
V value) observed between soil texture and forest physiognomy north of
a significant increase in coarse-textured soils (subhypothesis 3).
Moreover, in the analysis of this transect, the chi square value
computed for the north sample, with a smaller sample size (n), exceeded
that computed for the south sample. These higher values, however,
appear to be an expression of the higher frequency of soil texture 5
relative to the other soil textures north of the break and the
resulting high frequency of deciduous and needleleaf species on this
soil texture, rather than a northward trend toward greater site
specificity. The specificity of needleleaf trees to coarse—textured
soils was similar from south-north within the sampled region.

In sum, the subhypotheses served to clarify the existing
relationships between forest composition and soil textures by focusing
on the northward transition in forest physiognomy, forest physiognomy-
soil texture relationships, and the south-north contrasts between
observed relationships across the transition. It was beyond the limits

of this study to measure relative increases in each tree species

“9
category, at intervals northward, in order to determine the abruptness
of the transition in forest composition and its relationship to a soil
texture gradient. However, the significantly higher north percentage
of needleleaf trees suggests that the change in forest composition at
the tension zone is abrupt and supports a conclusion that more than one
plant association occurs across central Lower Michigan.

The correspondence that I found between the northward trend in
needleaf trees and coarse-textured soils suggests that theIaverall
physiognomic transition may vary in response to the soil texture
gradient. For example, a sharp gradient toward coarse-textured soils
is characterized by abrupt physiognomic change, a lessened soil texture
gradient by an obscured physiognomic transition, and a nonlinear
gradient in coarse soil texture by a nonlinear trend in the occurrence
of needleleaf trees. These patterns are consistent with a suggestion
by Curtis and McIntosh (1951) that the nature of a gradient in an
influencing environmental control may dictate the abruptness of a
vegetation tension zone. Although contrasts in the forest transition
and soil texture gradient were identified between transects B and C,
more'evidence would be needed to confirm the existence of significant
east-west differences in the abruptness of the forest transition as a
possible response to dissimilar northward gradients in the distribution
of coarse-textured soils.

The examination of forest physiognomic groups clarified the site
specificity of needleleaf trees to coarse-textured soils, but it failed
to support the relationship observed earlier between sugar maple and
fine soil texture. The tension zone, if defined indirectly by a

significant increase northward in coarse-textured soils, may be best

50

described by a sudden increase in the relative frequency of needleleaf
tree species, but one might expect this increase to coincide with a
decline in sugar maple, which has a low frequency on these soils.
Finally, the dissimilar results which concern the strength of the
association between forest physiognomic groups and soil texture classes
(transect B, transect C, and north and south) may be, as mentioned,
attributable to different trends in the distribution of coarse-textured
soils, or they may be a response to different relative frequencies of
all soil texture classes, but they do not appear to be a reflection of
a difference in the site specificity of a physiognomic group.
Combining the results from the examination of the subhypotheses
and the primary hypotheses, as they relate to the underlying research
problem of this thesis, I draw several conclusions. The significant
increase in needleleaf species northward indicates a noticeable change
in forest composition across the tension zone. These species, as a
physiognomic group, appear to be primarily site specific to coarse-
textured soils, occurring with deciduous species on these soils, and
increasing in frequency in direct correspondence to an increase in
coarse-textured soils. The extent of upland area overlain by coarse-
textured soils increases abruptly across the tension zone, even though
there is variation in the northward trend from east to west. The
abrupt increase in coarse-textured soils and needleleaf species
northward and the significant overall association identified between
the two suggest that the distribution of needleleaf tree species across

this region may be, to a degree, edaphically controlled.

51

The question arises in the interpretation of the overall results,
however, whether the statistical relationship observed between
needleleaf trees and coarse-textured soils is coincidence and the real
northward control is climate. Needleleaf trees were relatively more
frequent on these soils both north and south of the tension zone and no
consistent northward change in their site specificity, which would
indicate an influence by an additional factor, was observed. I infer
from this observation that the distribution of needleleaf trees within
the tension zone is not climatically determined. Of course, the
ultimate control of the transition toward a mixed-coniferous forest, as
characterized by the northern limits of southern broadleaf species and
also by the southern limits of northern needleleaf species, is
climatic. However, these results focus on the distribution of tree
species within this northward climatic transition, and it appears that
across this region the sharpness as well as the characteristics of the
northward trend toward needleleaf forest are at least partially
explained by the distribution of coarse-textured soils. By this view,
the tension zone, as defined by an abrupt increase in needleleaf trees,
may be best explained by the underlying soils pattern across central

Lower Michigan.

Reco-endations for Further Research

This study was limited to an examination of hypotheses which
address the relationships between one physical factor, soil texture,
and presettlement forest composition along three belt transects. In an
attempt to extend the focus of this research, I offer three

recommendations for further research, and one suggestion for a sampling

52

technique which may be used in subsequent tension zone or other
phytogeographical research.

The data recorded along three belt transects were examined
separately in order to improve the reliability of the results.
Realizing now the spatial variation observed in forest composition and
soils (east-west, and south-north), I recommend first that subsequent
study could focus on a comparative examination of the soil texture and
forest physiognomic transition along several south-north transects,
identifying contrasts in the northward trends in soils and their
relationships to forest physiognomy. Furthermore, additional transects
positioned east to west across central Lower Michigan may clarify
the possible longitudinal transition between oak and beech-maple
forests in this region.

It would be incorrect to conclude from the statistically
significant results compiled concerning needleleaf species and coarse
soil texture that their distribution is necessarily determined in every
case by the distribution of these soils. For example, Potzger (19“8)
emphasized the importance of climatic control when he identified former
north and south shifts in the location of the tension zone. McCann
(1982) observed an interesting correlation between the range lines of
flora and heating degree days, and Brewer (1982) presented evidence on
the possible influence of plant competition in determining the northern
limits of southern broadleaved species. I suggested earlier that soil
drainage may be the reason for the variation in the needleleaf trees-
soil texture relationships observed along transect A and, furthermore,

site history, a variable only partially controlled by the use of

53
presettlement vegetation records, may be a signicant factor determining
forest composition. Further investigation of the northward transition
in forest composition could address several possible controlling
factors, (e. g. soil texture, depth to water table, climatic variables,
topography) and attempt to assess the relative importance of each.

The similarities in the northward transition in presettlement
forest compostion and soil texture classes and the statistical
significance of their association suggest that similar relationships
may be observed between measures of present forest composition or
productivity, land use practices, and the northward trend toward
coarse-textured soils across the tension zone. This study was limited
to an examination of the presettlement forest composition, but a
subsequent project could examine the existing patterns of land use and
land productivity along the transects as they are related to soil
texture and presettlement forest composition. The study results may
depict the usefulness of soil texture or natural (presettlement)
vegetation as indirect measures of land productivity.

Each of the three suggested research areas demands an effective
sampling technique to obtain information on presettlement forest
composition, one or more physical factors, and current land cover/land
use. For instance, this study constitutes one of several studies in
Michigan based on data collected from presettlement forest records for
research analyses or vegetation mapping; the recorded data are
maintained in tables identifying location, tree species, and, in some
cases, tree diameter. I know of no program to compile this information

into a computer data base or a geographic information system.

5“

A computer data base consisting of point data compiled by
location could identify tree species as recorded from the survey notes
and a number of other factors that would be of interest to
phytographical research. If these data were available in such a
systematic form, several smaller projects could be combined into a more
thorough examination of the northward transition in vegetation across
central Lower Michigan.

A computer data base would be appropriate for detailed or site
specific analyses but would be more tedious to compile, whereas the
incorporation of data into a geographic information system would be
easier to compile and useful to broader research topics or applied to
current land use decisions. The Center for Remote Sensing at Michigan
State University (Lusch and Enslin, 198“) is currently developing a
"statewide data base of land-surface information" consisting of data
for areal units (grid cells) that are one square kilometer.
Incorporated into this geographic information system is an
identification of land cover, average available soil water capacity,
general topography and multiple vector windward distance to the Great
Lakes. Presettlement forest composition could be recorded at 1.6
square kilometer intervals (one section), using a classification system
satisfactory to most phytogeographical research and entered into this
statewide land-base system. These data could then be analyzed using
the other information available on the statewide data base and, most
importantly, data collected for one location could be easily used in

subsequent research.

CHAPTER V

SUMMARY AND CONCLUSIONS

The objective of this study was to identify possible relationships
between forest composition and the texture of the upper soil profile
across the vegetation tension zone of central Lower Michigan. It was
suspected that needleleaf species are site specific to coarse-textured
soils, which increase abruptly across this region. I hypothesized that
the measured association between tree species and soil textures is
statistically significant and, secondly, that the occurrence of coarse-
textured soils increases significantly across the tension zone. I
further examined these relationships through an analysis of forest
physiognomy, hypothesizing that, first, the percentage of all tree
species that are needleleaf increases significantly northward, second,
a statistically significant association exists between ferest
physiognomy (broadleaved-deciduous and needleleaf-coniferous) and soil
texture classes, and third, the measured association strength between
forest physiognomy and soil textures is greater north of a sharp
transition toward coarse-textured soils.

Data were recorded on the presettlement forest composition, its
relationship to specific soil texture classes, and the distribution of
soil texture classes along three belt transects which extend south to
north across central Lower Michigan. The association between tree

species categories or forest physiognomic groups and soil texture

55

56

classes was analyzed using the chi square statistic (statistical
significance), Cramer's V index (strength of the association) and the
asymmetric lambda index (predictability of the dependent variable; tree
species). Scattergrams were examined to identify a sharp increase in
the occurrence of coarse-textured soils (texture “ or 5) at .“
kilometer intervals and the Student's t-test measured the statistical
significance of that increase. Multiple regression analyses depicted
the linear trend in the occurrence of texture 5 northward along each
transect, and the differences in the trend north and south of the
identified sharp increase in that soil texture. The percentage of all
recorded trees that were needleleaf was calculated at 9.6 kilometer
intervals and compared to the northward trend in mean soil texture.
The significance of the increase in needleleaf tree percentages was
measured (z-statistic) at the break identified in the analysis of the
soil texture gradient.

Tree species categories are shown to be significantly associated
to soil texture classes, but the measured strength of that association
and the predictability of the dependent variable (tree species) is
lessened by the more frequent occurrence of trees that exhibit little
site specificity (e. g. beech, oak, and other mesic deciduous species),
and the predominance of beech on nearly all soil textures. Pines and
sugar maple appear to be predominantly site specific to coarse and
fine-textured soils, respectively. Pine and oak are more common in the
central and western transects, where sugar maple and beech are less
important.

A significant increase in soil texture 5 (the most coarse soil)

was identified toward the north along transects B and C and in soil

57

texture “ along transect A. Soil texture 5 was infrequent and
needleleaf species showed little preference to these soils along
transect A, suggesting that the distribution of this soil texture was
localized and exhibited characteristics, such as poor drainage, that
differed from transects BIand C and were not examined in this study.
Significant linear equations depicting the northward trend in soil
texture 5 along transects B and C indicated that the trend north and
south of the break along both transects was different from zero, and
that the linear trend north of the break was significantly different
from that south of the break. Additionally, the north trends were
noticeably different between the two transects.

The percentage of all trees that are needleleaf increases
significantly northward along transects B and C, and this trend
coincides with the relative distribution of coarse-textured soils.
Needleleaf tree species are noticeably site specific to sands, whereas
the deciduous species are equally frequent on all soil textures or are
most frequent on fine and coarse-textured soils. These relationships
identified between forest physiognomy (needleleaf and broadleaf trees)
and soil texture classes was similar south and north of a distinct
increase in the occurrence of soil texture 5. (Iverall, the results
demonstrate that needleleaf tree species and coarse-textured soils
increase abruptly across central Lower Michigan and that needleleaf
trees occur primarily on coarse-textured soils, such that the
distribution of these soils within this region.may help define the

vegetation tenSion zone.

LIST OF REFERENCES

LIST OF REFERENCES

Beal W..L and C.Efl Wheeler. 1892. Michigan Flora. Agricultural
College, Michigan.

 

Benninghoff, W. S. and A. I. Gebben. 1960. Phytosociological studies of
some beech-maple stands in Michiganks lower peninsula. Papers pf
the Michigan Academy pf Science, Arts, and Letters “5:83-91.

 

 

Blalock, H. M. 1960. Social Statistics. New York: McGraw-Hill Book
Company.

 

Bourdo, E. A. 1956. A review of the General Land Office Survey and of
its use in quantitative studies of former forests. Ecology

Braun, E. Lucy. 1950. Deciduous Forests pf Eastern North America. New
York: The Free Press.

 

 

Brewer, Lawrence. 1982.”A study of the vegetation tension zone in
Michigan using pre and post settlement tree surveys". Unpublished
research paper. Western Michigan University.

Brewer, Lawrence G., T. W. Hodler, and H. A. Raup. 198“. "The
presettlement vegetation of Southwestern Michigan". Map.
Kalamazoo: Western Michigan University.

Conover, W. J. 197“. Practical Nonparametric Statistics. New York:
John Wiley and Sons, Incorporated.

 

Cowles, Henry C. 1901. The physiographic ecology of Chicago and
vicinity. Botanical Gazette 31:31-73; 1“5-182.

 

Crankshaw, W. B., S. A. Quadir, and A. A. Lindsay. 1965. Edaphic

controls of tree species in pre-settlement Indiana. Ecology
“6:688-698.

Curtis, J. T. and R. P. McIntosh. 1951. An upland forest continuum in
the prairie-forest border of Wisconsin. Ecology 32:“76-“96.

Curtis, J. T. 1959. The Vegetation pf Wisconsin. Madison: University
of Wisconsin Press.

 

Darlington, H. T. 19“5. Taxonomic and ecological work on the higher
plants of Michigan. Michigan Agriculture Experiment Station
Techincal Bulletin 201.

 

 

58

59

Daubenmire, R. F. 1959. Plants and Environment. Second Edition. New
York: John Wiley and Sons, Incorporated.

 

Dice, L. R. 1931. A preliminary classification of the major
terrestrial communities of Michigan. Papers 22 the Michigan
Academy Sciencp, ArtsJ and Letters 1“:217-2“0.

 

 

Eggler, W. A. 1938. The maple-basswood forest type in Washburn County,
Wisconsin. Ecology 19:2“3-263.

Elliot, J. C. 1952. "The phytosociology of the upland second growth
hardwoods of Massauke County, Michigan". Ph. D. Dissertation.
Michigan State College.

Elliot, J. C. 1953. Composition of upland second growth hardwood
stands in the tension zone of Michigan as affected by soils and
man. Ecological Monographs 23:271-288.

 

Eyre, S. R. 1962. Vegetation and Soils. _A_ World Picture. London:
Edward Arnold, Ltd.

 

 

Farrand, W. R., and D. L. Bell. 1982. "Quaternary Geology of Southern
Michigan". Map. The Department of Geological Sciences, the
University of Michigan, and the Michigan Department of Natural
Resources, Geological Survey Division.

Gleason, Henry A. 192“. The structure of the maple-beech association
in northern Michigan. Papers pf the Michigan Academy ScienceJ
Arts, and Letters “:285-296.

 

 

Harlow, William M., Ellwood S. Harrar, and Fred M. White. 1979.
Textbook of Dendrology: Sixth Edition. New York: McGraw-Hill
Book Company.

Harvey, Leroy H. 1919. Some phytogeographic observations in Lake
County, Michigan. Michigan Acadexgy pf Science Report 21:213-217.

 

 

Hushen, Timothy W., et. al. 1966. Presettlement forest pattens in
Montcalm County, Michigan. Michigan Botanist 5:192-211.

 

Johnston, J. 1972. Econometric Methods. 2nd Ed. New York: McGraw-
Hill Book Company.

 

Kapp, Ronald G. 1978. Presettlement forests patterns of the Pine River
watershed (Central Michigan) based on original land survey
records. Michigan Botanist 17:3-15.

Kenoyer, L. A. 1930. Ecological notes on Kalamazoo County, Michigan,
based on the original land survey. Papers o_f_ the Michigan Academy
Science, Artsi and Letters 11:211-218.

6O

Kenoyer, L. A. 193“. Forest distribution in Southwestern Michigan, as
interpreted from the original survey (1826-1832). Papers 92 the
Michigan Academy pf Science, Arts, and Letters 19:107-111.

 

 

Kenoyer, L. A. 19“0. Plant associations in Barry, Calhoun and Branch
Counties, Michigan, as interpreted from the origingal survey.
Papers 2: the Michigan Academy Science, Arts, and Letters 25:75-
77.

Kenoyer, L. A. 19“3. Forest associations of Ottawa County, Michigan,
at the time of the original survey. Papers 33 the Michigan
Academy Science, Arts and Letters 28:“7-“9.

 

 

Klecka, W. R., N. H. Nie, and C. Hadlai Hull. 1975. SPSS Primer. New
York: McGraw—Hill Book Company.

 

Lipsey, J. M. 1983. Personal Communication. Michigan State
University, Department of Geography.

Livingston, B. E. 1903. The distribution of the upland plant
societies of Kent County, Michigan. Botanical Gazette 35:36-55.

 

Livingston B. E, 1905. The relation of soils to natural vegetation
in Roscommon and Crawford Counties, Michigan. Botanical Gazette
39:22-“1.

 

Lusch, David P., and William R. Enslin. 198“. "Microcomputer-based,
statewide, digital land-surface information". Center for Remote
Sensing. East Lansing: Michigan State University.

McCann, M. T. 1979. "The Plant Tension Zone". Master”s Thesis.
Western Michigan University.

Mokma, D. L. 1982. Soil management units and land use planning.
Michigan State University, Cooperative Extension Service Bulletin
E-12620

Nichols, G. E. 1935. The hemlock-white pine-northern hardwood region
of Eastern North America. Ecology 16:“03-“22.

Nie, N. H., C. H. Hull, J. G. Jenkins, K. Steinbrenner, and D. H. Bent.
1975. Statistical Package for the Social Sciences. 2nd ed. New
York: McGraw-Hill Book Company.

Norcliffe, G. B. 1977. Inferential Statistics for Geographers. New
York: John Wiley and Sons.

 

Nutter, M. D. 1973. "Soil Texture and Forest Change in Northern Lower
Michigan". Master's Thesis. Michigan State University.

Parmelee, G. W. 1953. "Oak Upland Continuum in Southern Michigan".
Ph. D. Dissertation. Michigan State College.

61

Potzger, J. E. 19“8. A pollen study in the tension zone of Lower
Michigan. Butler University Botanical Studies 8:161-177.

Potzger, J. E., M. E. Potzger, and Jack McCormick. 1956. The forest
primeval of Indiana as recorded in the original U. S. land surveys
and an evaluation of previous interpretations of Indiana
vegetation. Butler University Botanical Studies 13:95-111.

 

Quick, B. E. 192“. A comparitive study of the distribution of the
climax association in Southern Michigan. Papers pf the Michigan
Academy pf Science, Arts, and Letters 3:211-2““.

 

 

Schneider, I. F. and A. E. Erickson. 1967. "Water holding capacity and
infiltration rates of soils in Michigan". Maps. Data from
Michigan Agriculture Experiment Project “13. Soil Science
Department, Michigan State University in cooperation with State
Resources Planning Commission, Michigan State Department of
Commerce and Water Resources Commission, Michigan Department of
Conservation.

Sears, P. B. 1925. The natural vegetation of Ohio. Ohio Journal 9_f_‘
Science 25:139-1“9.

Strommen, N. D. 197“. Climate 22 Michigan py Stations. Michigan
Department of Agriculture, Michigan Weather Service, cooperating
with the NCAA-National Weather Service, United States Department
of Commerce.

Taylor, Peter J. 1977. Quantitative Methods _i_p Geography. Boston:
Houghton Mifflin Company.

 

 

Transeau, E. N. 1935. The prairie peninsula. Ecology 16:“23-“37.

Veatch, J. O. 1953. Soils and Land pf Michigan. East Lansing:
Michigan State College Press.

 

Veatch, .L. 0. 1959. "Presettlement forest 5J1 Michigan". Map.
Department of Resource Development. Michigan State University.

Whitford, H. N. 1901. The genetic development of the forests of
Northern Michigan. Botanical Gazette 17:91-99.

 

Wilde, S. A. 1933. The relation of soils and forest vegetation of the
lake states region. Ecology 1“:9“-105.

Winkler, Robert L. and William L. Hays. 1975. Statistics:
Probability, Inference, and Decision. New York: Holt, Rinehart
and Winston.

 

 

Yamane, Taro. 1973. Statistics: ep Introductory Analysis. New York:
Harper and Row, Publishers.

 

 

APPENDICES

APPENDIX 1

THE SOURCES OF RECORDED DATA

a.

62

The Presettlement Forest.

 

 

 

 

 

 

 

 

 

Transect A_ (R“w)

T18N
T17N
T16N
T15N
T1“N
T13N
T12N
T11N
T10N

T9N

Volume

Volume

Volume

Volume

Volume

Volume

Volume

Volume

Volume

Volume

“9

“9

“9

“9

“9

“9

“9

“9

33

33

1856

1850

1837

1838

1838

1838

1832

1832

1831

1831

 

 

 

 

 

 

IT

 

 

 

 

 

 

 

 

63

Transect B (R7W)
T16N Volume 52
T15N Volume 52
T1“N Volume 52
T13N Volume 52
T12N Volume 52
T11N Volume 52
T10N Volume 37
T9N Volume 37
T8N Volume 37
T7N Volume 37

1839

1839

1839

1839

1839

1839

1838

1838

1836

1831

6“

Transect Q (R10w)

 

- T1“N Volume 55 1838

 

 

 

T13N Volume 55 1837

 

T12N Volume 5“ 1837

 

 

 

"- T11N Volume “0 1837

 

’ T10N Volume no 1837

T9N Volume “0 1837

T8N Volume “0 1837

T7N Volume “0 1837

 

T6N Volume “0 1831

T5N Volume “0 1831

 

.All data were obtained from the General Land Office Survey Notes,
identified here by the volume number of the compiled notes and the
date the survey was completed. Sampled sections are black.

 

65

b. The Soil Survey Reports

 

 

Clare County Modern soil survey was published in 1979.
Soil survey maps are on an aerial photograph
background at a scale of 1:15.8“0.

Gratiot County Modern soil survey' was published in 1979.
Soil survey maps are on an aerial photograph
background at a scale of 1:15,8“0.

Ionia County Modern soil survey was published in 1967.
Soil survey maps are on an aerial photograph
background at a scale of 1:15,8“0.

Isabella County The field work for the modern soil survey is
completed but not yet published. Maps were
available at the Department of Crop and
Soil Science, Michigan State University.

Kent County The field work for the modern soil survey
was completed by May 198“. Maps were
available from the Department of Crop and
Soil Science, Michigan State University, and
the Soil Conservation Service Office in
Grand Rapids, Michigan.

Mecosta County Modern soil survey was completed in 1981. I
acquired maps from the Department of Crop
and Soil Science, Michigan State University.
The soil survey was published in May 198“.

Montcalm County Modern soil survey was published in 1960.
Lithographed (colored) maps are at an
approximate scale of 1:20,000.

 

APPENDIX 2

THE DATA RECORDED

66

 

 

 

 

Table A2a. Presettlement foresptree species.
Tree Type Transect A Transect B Transect C
Freq. 5 Freq. 5 Freq. 1
Beech 367 33.7 351 30.2 269 21.6
Sugar Maple 1“9 13.7 12“ 10.7 75 6.0
Elm 86 7.9 31 2.7 22 1.8
Lynn 5“ “.9 1“ 1.2 2“ 1.9
Maple 5“ 5.0 * * 62 5.0
Black Ash 70 6.“ 22 1.9 9 1.7
Oak “5 “.1 21“ 18.“ “3“ 3“.8
Pine 125 11.5 206 17.7 250 20.1
Hemlock 111 10.2 122 10.5 11 .9
Cedar ' ' 10 .9 1 .1
Tamarack 28 2.6 21 1.8 7 .6
Other * * “6 “.0 82 6.6
TOTAL 1089 100% 1161 100% 12“6 100%

 

'no data recorded

 

67

Table A2b. Tree species categories.

 

 

 

 

Tree Type Transect A Transect B Transect C
Freq. i Freq. S Freq. 1
Beech 367 33-7 351 30.2 269 21.6
Sugar Maple 1“9 13.7 12“ 10.7 75 6.0
Oak “5 “.1 21“ 18.“ “3“ 3“.8
Pine 125 11.5 206 17.7 250 20.1
Hemlock 111 10.2 122 10.5 19 1.5
Tamarack 28 2.6 31 2.7 * 8
Other 26“ 2“.2 113 9.7 199 16.0
TOTAL 1089 100% 1161 100% 12“6 100$

 

'no data recorded

 

68

Table A2c. Soil management units.

Texture Class

Transect A

Transect B

Transect C

 

 

Freq. % Freq. % Freq. %
TEX1 2““ 22.“ 25 2.2 167 13.“
TEX2 32“ 29.7 107 9.2 1“1 11.3
TEX3 61 5.6 1“0 12.1 9“ 7.5
TEX“ 199 18.3 26“ 22.7 232 18.6
TEX5 1“7 13.5 237 20.“ “80 38.5
TEX3/5 ' 8 30 2.6 16 1.3
TEX“/2 66 6 1 6“ 5.5 51 “.1
TEX3/2 “8 “.“ 29 2.5 21 1.7
TEX5/2 * * 12 1.0 15 1.2
TEX3.5 9 * 1“2 12.2 3 .2
TEX2.5 * * 93 8.0 2 .2
TEX“.5 * 9 18 1.6 2 .2
TOTAL 1089 100% 1161 100% 12“6 100%

 

.no data recorded

 

Table A2d. Soil texture classes.

69

 

Texture Class

Transect A

Transect B

Transect C

 

 

Freq. 1 Freq. 1 Freq. 1 .
TEX2 568 52.1 225 19.“ 310 2“.9
TEX3 109 10.0 3“1 29.“ 13“ 10.8
TEX“ 199 18.3 282 2“.3 23“ 18.8
TEX5 1“7 13.5 237 20.“ “80 38.5
TEXC/f 66 6.1 76 6.5 88 7.1
TOTAL 1089 1001 1161 100% 12“6 100%

 

“If-.3. m.“—
- ‘ t l-
r

 

 

APPENDIX 3

TABLES COMPILED IN THE DATA ANALYSIS

‘Wu Lena-m-
_ I Q P
l.
L

70

ITable A3a. Tree species categories-soil texture classes-- the
contingency tables.

 

 

Transect A

 

frequency
row 1
column 1 TEX2 TEX3 TEX“ TEXS TEX15 row total
210 36 61 “u 16 367
BEECH 57 1o 17 12 on
37 33 31 30 2“
101 16 8 9 15 1“9
SUGAR MAPLE 68 11 05 06 1o
18 15 o“ 06 23
26 9 5 2 3 “5
OAK 58 02 11 o“ 07
on 08 02 01 05
25 13 59 2“ 0“ 125
PINE 02 10 “7 19 03
o“ 12 3o 17 O6
38 6 3“ 23 10 111
HEMLOCK 3“ 05 31 21 09
07 05 17 16 15
11 3 1 11 2 28
TAMARACK1 39 11 o“ 39 07
02 03 o 07 03
157 26 31 3“ 16 26“
OTHER2 59 01 12 13 06
28 2“ 16 23 2“

 

 

column total 568 109 199 1“? 66 1089

 

'Table A3a continued.

71

 

 

Transect B

 

 

 

frequency
row 1
column 1 TEXZ TEX3 TEX“ TEXS TEX15 row total
6“ 116 89 58 2“ 351
BEECH 18.2 33.0 25.“ 16.5 6.8
28.“ 39.0 31.6 2u.5 31.6
26 37 33 17 11 12“
SUGAR MAPLE 21.0 29.8 26.6 13.7 8.9
11.6 10.9 11.7 7.2 14.5
58 81 “9 21 5 21“
OAK 27.1 37.9 22.9 9.8 2.3
25.8 23.8 17.“ 8.9 6.6
28 52 5“ 6“ 8 206
PINE 13.6 25.2 26.2 31.1 3.9
12.“ 15.2 19.1 27.0 10.5
9 22 28 50 13 122
HEMLOCK 7.“ 18.0 23.0 “1.0 10.7
“.0 6.5 9.9 21.1 17.1
12 6 5 8 u 31
TAMARACK1 38.7 19.“ 16.1 25.8 8
5.3 1.8 1.8 3.“ s
28 27 2“ 19 15 113
OTHER2 2“.8 23.9 21.2 16.8 13.3
12.“ 7.9 8.5 8.0 19.7
Column total 225 3“1 282 237 76 1161

 

'no data recorded

Table A3a continued.

72

Transect C

 

 

 

frequency
row 1
column 1 TEX2 TEX3 TEX“ TEXS TEX15 row total
88 2“ “0 91 26 269
BEECH 32.7 8.9 18.9 33.8 9.7
28.8 17.9 17.1 19.0 29.5
53 5 8 7 6 75
SUGAR MAPLE 70.7 6.7 5.3 9.3 8.0
17.1 3.7 1.7 1.5 6.8
77 75 108 161 13 838
OAK 17.7 17.3 28.9 37.1 3.0
28.8 56.0 86.2 33.5 18.8
27 11 37 156 19 250
PINE 10.8 “.“ 18.8 62.8 7.6
807 802 1508 3205 2106
5 1 7 8 2 19
REML00K3 26.3 5.3 36.8 21.1 10.5
1.6 .7 3.0 .8 2.3
60 18 38 61 22 199
OTHER2 30.2 9.0 19.1 30.7 11.1
19.8 13.“ 16.2 12.7 25.0
column total 310 138 238 880 88 1286

 

1

2Other mesic deciduous species

Tamarack and cedar are combined in this category

3Other needleleaf species, combining hemlock, tamarack and cedar

Table A3b. Relative proportions of needleleaf trees to all trees

73

(calculated at 9.6 kilometers).

DISTANCE
9.6 km
19.2
28.8
38.“

“8
57.6
67.2'
76.8
86.“

96 km

TRANSECT B
.009
.21
.17
.39
.29
.33
.32
.35
.“5

.“7

TRANSECT C
.01
.2“
.19
.39
.28
.33
.31
.32
.“5

.“1

 

 

 

' A break was established in the recorded proportions at this
location forming south and north samples, respectively; The
proportion measured at 67.2 kilometers was included in the

south sample.

 

 

Win. mm. m. an:
«Eden—I: .

7“

Table A3c.
tables.

W
m

Forest physiognomy-soil texture classes-- the contingency

Transect B

 

 

 

 

 

 

 

frequency
row 1
column 1 TEX2 TEX3 TEX“ TEXS TEX15 row total
176 261 195 115 55 802
BROADLEAVED 21.9 32.5 2“.3 1“.3 6.9
78.2 76.5 69.1 “8.5 72.“
“9 80 87 122 21 359
NEEDLELEAF 13.6 22.3 2“.2 3“.0 5.8
21.8 23.5 30.9 51.5 27.6
column total 225 3“1 282 237 76 1161
Transect C
frequency
row 1
column 1 TEX2 TEX3 TEX“ TEX5 TEX15 row total
279 122 190 320 67 977
BROADLEAVED 28.5 12.5 19.“ 32.8 6.9
89.7 91.0 81.2 66.7 76.1
32 12 ““ 160 21 269
NEEDLELEAF 11.9 “.5 16.“ 59.5 7.8
2.6 1.0 3.5 12.8 1.7
column total 310 13“ 23“ “80 88 12“6

 

75

Table A3d. South and north forest physiognomy-soil texture classes--
.the contingency tables. '

 

 

Transect B (south)

 

 

 

 

 

 

frequency
row 1
column 1 TEX2 TEX3 TEX“ TEXS TEX15 row total
1“2 191 127 “3 17 520
BROADLEAVED 27.3 36.7 2“.“ 8.3 3.3
81.6 80.3 72.6 53.1 85.0
32 “7 “8 38 3 168
NEEDLELEAF 19.0 28.0 28.6 22.6 1.8
18.“ 19.7 27.“ “6.9 15.0
column total 17“ 238 175 81 20 688
Transect B (north)
frequency
row 1
column 1 TEX2 TEX3 TEX“ TEXS TEX15 row total
38 70 68 72 38 282
BROADLEAVED 12.1 2“.8 2“.1 25.5 13.5
66.7 68.0 63.6 “6.2 67.9
17 33 39 8“ 18 191
NEEDLELEAF 8.9 17.3 20.“ ““.0 9.“
33.3 32.0 36.“ 53.8 32.1

 

 

column total 51 103 107 156 56 “73

 

,Table A3d continued.

 

76

 

 

Transect C (south)

 

 

 

 

 

 

 

frequency
row 1
column 1 TEX2 TEX3 TEX“ TEX5 TEX1S row total
252 116 165 197 51 781
BROADLEAVED 32.3 1“.9 21.1 25.2 6.5
89.“ 90.6 89.7 82.“ 78.5
30 12 19 “2 1“ 117
NEEDLELEAF 25.6 10.3 16.2 35.9 12.0
10.6 9.“ 10.3 17.6 21.5
column total 282 128 18“ 239 65 898
Transect C (north)
frequency
row 1
column 1 TEX2 TEX3 TEX“ TEX5 TEX15 row total
26 25 123 16 196
BROADLEAVED 13.3 3.1 12.8 62.8 8.2
92.9 100 50.0 51.0 69.6
2 0 25 118 7 152
NEEDLELEAF 1.3 0 16.“ 77.6 “.6
7.1 0 50.0 “9.0 30.“
column total 28 6 50 2“1 23 3“8

 

 

 

”‘11111181181111111“