:13" flunk.
, .

v. .l‘vhiv
4

Laura an...
3.11;. .
2 i1?
.. L u. 2
ii.
5.2? t. A.

, . I,»
Winn-”z
.‘tix! "L?

PL: y:-

‘ I

{-‘ 4p: 1- r

a 'i_ a ‘

3:, a

c.

:
1.33...
‘ . 3. . a hm». ‘9‘ .
1.... i .v r \9
V 3:... J) inexcuSIis
t . .. t . . . u
5-. f...
‘ . . 2...... A!
s a z... 2.35.. {yrkizfi
3‘. x. ch. an aura
1...... .l) x v .
, 32— 1 lt»\‘al.fl,l¢-‘ r: L ‘3.) t
3’ I {Db-K9 ‘ hi. 3‘
i . . . ;
3-. .3i :9 .0: ‘r I zl!‘
I J.»‘\!.1

on
71.

I: 5.15..
:53; :
11>...)

..
. c2. .5...
3.). .5. :L:i£1
\OJAO‘I‘ .3;:.Vi.
2.....xiaxzf. 1.... 2.9:.
I. 1.3.5.125.
21:132.?
V E...
. 15.... . . .
l ._ \
1..»
’5’. Q. ‘ J. .
. 9:...2....:.i..:..
i. a. .a .3

5....
.2 t
w h
«b
.3. 1:)
\1 3!. i $x\\}\..vll~.2.
u ‘53); 3.. 3)“ 1i.
‘ . «.1. 3.15..) .5 73
\n? :9:..... ,
5.3.9..

ti .9328: .2.‘
1;... this}. huh!»
. h?! , \\. u
L 1.? a .\ . ,
3.233;}: 3
3:11.251!»
.y a? sit. .4.

5.3:
4.2353153...
2.3! 1.1.15.5:
A .23}; 12....
2 f IF...
4... F2 \
it..- .35. ,
12a}??? ‘
a: 2.55.».

 

c3."
.1 in...
. .81.-..
\(w 1' :|\ Ahr
I ‘1‘ II:\» I.
\(G'fixl... ..
.i I r

9:13.»
, xfill
3.2.x
\~§f.~st‘ 3.... ‘
0‘31; .

|\ A). an)?
{Ifixtsauqh
, s.. I 3...! ‘5‘:

l! E). .t

. ‘ kn» «2%.!er
2. 1 in .21.»!
1371.3}, 42.3.?! ..
. x. ‘k6l.>\3 521;}: hhrnvio

\i‘illliiiiiiiiiiiii

312

T1

This is to certify that the

dissertation entitled

USING AN ECOLOGICAL CLASSIFICATION
AND WILDLIFE HABITAT MODELS
IN FOREST PLANNING

presented by

Gary John Roloff

has been accepted towards fulfillment
of the requirements for

PhD degreein Wildl'ife

Major profess 1'

Date NOV. 15’ 1994

MS U i: an Affirmative Action/Equal Opportunity Institution 0- 12771

     

 

LIBRARY l
Mlchigan Statel
University i

 

 

manna-um BOXmmmuhMMMymm.
TO AVOID FINES Mum on orbdmddoduo.

DATE DUE DATE DUE DATE DUE

    

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

I II I l

MSU IoAn Affirmative Min-l Opporhmlly Intuition
WM‘

 

 

USING AN ECOLOGICAL CLASSIFICATION SYSTEM AND
WILDLIFE HABITAT MODELS IN
FOREST PLANNING

By

Gary John Roloff

A DISSERTATION

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

DOCTOR OF PHILOSOPHY

Department of Fisheries and Wildlife

I994

ABSTRACT

USING AN ECOLOGICAL CLASSIFICATION SYSTEM AND WILDLIFE HABITAT
MODELS IN FOREST PLANNING

By

Gary J . Roloff

The usefulness of integrating an ecological classification system (ECS) and wildlife
habitat suitability (HSl) models into vegetation sampling and wildlife habitat assessments was
evaluated. Vegetation attributes (required for HSI models) were measured using a variety of
plot, line intercept, and subjective evaluation techniques from a stratified random sample
design. Sampling was stratified by USDA Forest Service cover type and underlying
ecological land type phase grouping (ELTPGXa sub-level of the ECS hierarchy). Replicates
were also categorized by: 1) Forest Service cover type, and 2) ELTPG. Relative efficiency
tests were used to evaluate sampling efficiency between the 3 stratification schemes.

Results of the relative efficiency tests between cover type-ELTPG and cover type
stratifications indicated (Ps0. 10) that sampling precision generally increased when stratified
by cover type-ELTPG. Overall, the ECS appears to be a promising tool for reducing the
variance associated with sampling vegetation attributes.

Wildlife population indices for H species. including ruffed grouse (Bonasa umbellus).
pileated woodpecker (Dryocopus pileatus), ovenbird (Seiurus aumcapillus). American redstart
(Setophagus ruticilla). black-throated green warbler (Dendroica virens), eastern meadowlark
(Sturnella magna), upland sandpiper (Bartramia longicauda), white-tailed deer (Odocoileus
virginianus), gray squirrel (Sciurus carolinensis), wood duck (Aix sponsa), and eastern
bluebird (Sialia sialis), were collected at systematically located points during I99! and 1992.

Cover type and ELTPG maps were digitized and HS] models were programmed into a

geographic information system (GlS); all of which were linked to the cover type-ELTPG
database thereby automating HSl computations. HSl values were calculated for each sample
point at home-range sized spatial scales. and Spearman’s Rank Correlation tests were
performed between population indices and associated HSl values. HSl values were also
computed at other spatial scales (03482.4 ha) to determine the appropriate spatial scale of
model applicability.

Correlation coefficients varied by species and ranged from -0.427 to 0.644 and -0.223
to 0.514 for 1991 and 1992, respectively. For HSl models combined across years (n=8), all
exhibited significant (PSO. 10), positive correlation coefficients, ranging from 0.131 to 0.307.
The spatial scale at which the HS] models should be applied corresponded to a biologically
meaningful scale (e.g. home range).

The integration and application of ECS and HS] models as coarse filter-fine filter
approaches to landscape level planning are demonstrated. Implications for forest management

are also discussed.

ACKNOWLEDGEMENTS

Typical dissertation acknowledgements begin with gratitude for the funding agencies
or faculty involved on the graduate committee. Although I am deeply appreciative of these
contributions, 1 choose to begin my acknowledgements by thanking my wife and children.
Thank you for 10 years of support, encouragement, understanding, and love. We were
warned it would be tough, and indeed it was. There are few women who could raise a family
through field seasons; solitary holidays, weekends, and evenings; graduate student salaries;
and university housing and continue to support their husband’s dreams. From the bottom of
my heart, I thank you Kimberly for understanding. To my children, Joshua, Cassondra, and
Jordan, the time has come for you to have a ”real” father.

Funding for this project was provided by the US. Forest Service and the Michigan
Agricultural Experiment Station. Special thanks to Rex Ennis, Huron-Manistee National
Forest, for his patience and support. Thanks to my major advisor, Dr. Jonathan Haufler, for
giving me advise, opportunity, and the flexibility to explore numerous aspects of this project.
Also, thanks to my other graduate committee members, Mr. Rex Ennis, Dr. Scott
Winterstein, Dr. Larry Leefers, and Dr. Peter Murphy. The experience of my oral exams
and graduate committee will live with me forever! Special thanks are due Dr. John Hart for
taking time (with short notice) to review the dissertation and participate on my graduate
committee. Thanks to Dr. Winterstein for his advice on data analysis and Dr. Murphy for

valuable ecological insight. Dr. Leefers provided an important forestry component to the

iv

project, and I also thank him for his friendship. I appreciated and enjoyed our discussions.
Although not part of my graduate committee, I would like to thank Dr. Henry Campa, III, for
his advice, support, and friendship. ln Dr. Haufler’s absence, Dr. Campa was always
available. I sincerely hope to continue collaborating and conversing with all of you, I am an
immensely better biologist and person because of our interactions.

The individuals involved with field work deserve recognition. Thanks to Mark
Parkinson, Ed Melling, Brian Kemohan, Bill Smith, Tom Kelley, Mike Reynolds, Doug
Weeks, and Tony Duffiney for their diligent work. I know we worked our tails off, but I
considered myself lucky to have had such devoted field crews. You are all part of this
document, and I appreciate your efforts. Good luck with your careers.

Thanks to Bill Enslin, Michigan State Center for Remote Sensing, for his cooperation
in obtaining geographic information system (GIS) coverages. Also, thanks to Ryan Simmons
from the Michigan State Computer Center for his 618 technical support. Thanks are also due
Peter O’Grady and Phil Huber, USDA Forest Service, Huron-Manistee National Forest, for
their assistance and advise in data collection, management, and analysis.

Numerous graduate and undergraduate students shaped my experience at Michigan
State. Lou Bender, Brian Kernohan, Cathy Cook, Rich and Donna Minnis, Paul Padding,
Greg Hancock, Steve Negri, Tim VanDeelen, and Kelly Millenbah were all good co-workers
and friends. Special thanks to Lou, Brian, Rich, Donna, Steve, and Cathy for the long
discussions about all aspects of life. Lou, l have met few people that have your intellect and
perspectives on the wildlife profession. Your influence on my thinking has been tremendous
(for the better I hope!), but most of all, thanks for your friendship. Special thanks are due to
Brian "Tex" Kemohan for his friendship, advice, and concern regarding this project. The

memories are many, and the good far outweigh the bad.

Heartfelt gratitude is due my father, James Roloff. and my grandfather, the late John
Golz. In addition to exposing me to the outdoors, they taught me to be honest, hard-working,
sincere, and forthright. At times I stray from their teachings, but their influence is
undoubtedly expressed in this document. Long before I read the ”Land Ethic" by Aldo
Leopold, these individuals taught me to respect the land, recognizing that we were just one
cog in the environmental wheel. They are truly good men, and I will continue to emulate
their values.

Lastly, I wish to thank my parents, James and Jane Roloff. Thanks for tolerating the
often bizarre behavior of a wildlife biologist in your home. I promise not to bring any more
dead animals home! Your encouragement and support were invaluable. This document is as

much a part of you as it is me, thank you.

vi

TABLE OF CONTENTS

LIST OF TABLES, ix
LIST OF FIGURES, xi
INTRODUCTION, 1
STUDY AREA, 4

CHAPTER 1: USING AN ECOLOGICAL CLASSIFICATION SYSTEM IN VEGETATION
SAMPLING, 7
Introduction, 7
Methods, 9
Ecological Classification System, 9
Vegetative Cover Typing, 11
Sampling and Analyses, I I
Results, I 6
Relative Efficiency Tests-Cover Type and ELTPG versus Cover Type, 20
Relative Efficiency Tests-Cover Type versus ELTPG, 30
Discussion, 39
Summary and Conclusions, 46

CHAPTER 2: WILDLIFE HABITAT MODEL VALIDATION, 47
Introduction, 47
Methods, 51
Results, 59
Discussion, 74
Habitat Model Validation, 74
Spatial Scale of HS] Model Applicability, 79
Summary and Conclusions, 84

CHAPTER 3: APPLYING AN ECOLOGICAL CLASSIFICATION SYSTEM AND HSI
MODELS TO LANDSCAPE PLANNING, 86

Introduction, 86

The Ecosystem Diversity Matrix, 87

Integrating HSI Models: The Fine Filter, 96

Summary and Conclusions, 101

vii

APPENDICES
Appendix A: Description of Ecological Classification System Components, 102
Appendix B: Habitat Suitability Index Model for the Ovenbird, 108

Appendix C: A Habitat Suitability Index (HSI) Model for the Black-throated Green
Warbler in the Huron-Manistee National Forest, Michigan, 110

Appendix D: A White-tailed Deer HSI for the Upper Great Lakes Region, 121

Appendix E: Habitat Suitability Index Model for the Eastern Bluebird in Northern
Michigan, 152

Appendix F: A Habitat Suitability Index Model for the Upland Sandpiper in the
Huron National Forest, Michigan, 161

Appendix G: Habitat Model-American Redstart (Setophaga ruticilla), I 77

LITERATURE CITED, I89

viii

Table 1.

Table 2.

Table 3.

Table 4.

Table 5.

Table 6.

Table 7.

Table 8.

Table 9.

Table 10.

LIST OF TABLES

Components (terminology and numeric coding following Cleland et al. (1993))
of ecological land type phase groupings (ELTPG’s) used in this study (also see
Appendix A for explicit definitions), 10

Summary of vegetation parameters, associated sampling techniques (see Hays
et al. 1981), and data transformations (Sokal and Rohlf 1981), 13

Number of replicates for cover type-ELTPG combinations, cover type, and
ELTPG used for sample precision analyses, I 8

Summary of relative efficiency tests between samples stratified by forest cover
type-ecological land type phase groupings (ELTPG’s) or forest cover type for ‘
overstory vegetation variables. Values are F-ratios (i.e. variance ratios), 21

Summary of relative efficiency tests between samples stratified by forest cover
type-ecological land type phase groupings (ELTPG’s) or forest cover type for
understory vegetation variables. Values are F-ratios (i.e. variance ratios), 23

Summary of relative efficiency tests between samples stratified by forest cover
type-ecological land type phase groupings (ELTPG’s) or forest cover type for
dead and downed vegetation variables. Values are F-ratios (i.e. variance
ratios), 26

Summary of precision gains for each vegetation variable stratified by forest
cover type-ecological land type phase grouping (ELTPG) compared to samples
stratified by forest cover type, 28

Summary of precision gains for each forest cover type stratified by forest
cover type-ecological land type phase grouping (ELTPG) compared to samples
stratified by forest cover type, 29

Summary of precision gains for each ecological land type phase grouping
(ELTPG) stratified by forest cover type-ELTPG compared to samples
stratified by forest cover type, 31

Summary of relative efficiency tests between samples stratified by forest cover
type or ecological land type phase grouping (ELTPG) for overstory vegetation
variables. Values are chi square statistics derived from combining the results
of 10 relative efficiency tests (see Fig. 4), 32

ix

 

Table l 1.

Table 12.

Table 13.

Table 14.

Table 15.

Table 16.

Table 17.

Table 18.

Table 19.

Table 20.

Table 21.

Table 22.

Table 23.

Table 24.

Summary of relative efficiency tests between samples stratified by forest cover
type or ecological land type phase grouping (ELTPG) for understory
vegetation variables. Values are chi square statistics derived from combining
the results of 10 relative efficiency tests (see Fig. 4), 34

Summary of relative efficiency tests between samples stratified by forest cover
type or ecological land type phase grouping (ELTPG) for dead and downed
vegetation variables. Values are chi square statistics derived from combining
the results of 10 relative efficiency tests (see Fig. 4), 37

Means and standard errors (parenthesis) for number of replicates required
(£0.10; accuracy 0.20 of the mean) according to Freese (1978) for vegetation
variables sampled by Forest Service cover type and ELTPG, 41

Vegetation variables and associated sampling techniques. Methodology for
each technique is described by Hays et al. 1981, 53

Wildlife species, home range estimates, and supporting research, 55

Spearman rank correlation coefficients, associated probability, and sample size
for tests between habitat suitability indices and wildlife population indices. ‘
Probability was one-tailed and tested if the correlation coefficient equaled 0
(Siegel 1956), 60

Spatial scale(s) of wildlife model applicability. Estimates are based on
subjective evaluations of histograms depicting the range of HSI values and
their evenness across the landscape for different spatial scales, 61

Spearman rank correlation coefficients, associated probability, and sample size
for randomly generated habitat suitability indices (HSI’s) and wildlife
population indices. Probability was two-tailed and tested if the correlation
coefficient differed from 0 by chance (Siegel 1956), 63

Studies validating habitat suitability index (HSI) models and the spatial scale at
which the validation was performed, 81

Density-habitat relationships for black-throated green warblers in various
vegetation types, 116

Territory size-vegetation relationships for the black-throated green warbler,
116

Definitions of habitat variables and associated measuring technique, 159

Summary of model variables, associated vegetation cover types, and suggested
sampling techniques for habitat suitability index determination, 1 73

Average home range sizes of American redstart over its breeding range, I 80

Figure 1.

Figure 2.

Figure 3.

Figure 4.

Figure 5.
Figure 6.

Figure 7.

Figure 8.

Figure 9.

Figure 10.

Figure 11.

Figure 12.

Figure 13.

Figure 14.

LIST OF FIGURES

Location of study sites, Alcona County, Michigan, 5

Ecological land type phase groupings, naturally associated overstory, and
corresponding Forest Service Corporate Database cover types sampled, 12

Example of resource planning map and associated relative efficiency analysis,
where s‘2 denotes sample variance, 15

Relative efficiency tests between cover type and ecological land type phase
grouping (ELTPG), where 5‘2 denotes sample variance, 17

The influence of spatial scale on HSI model output for ovenbird, 64
The influence of spatial scale on HSI model output for gray squirrel, 65

The influence of spatial scale on HSI model output for white-tailed deer fall-
winter foods, 66

The influence of spatial scale on HSI model output for white-tailed deer
security cover, 67

The influence of spatial scale on HSI model output for American redstart, 68

The influence of spatial scale on HSI model output for black-throated green
warbler, 69

The influence of spatial scale on HSI model output for ruffed grouse, 70

The influence of spatial scale on HSI model output for pileated woodpecker
cover and reproduction, 71

The influence of spatial scale on HSI model output for pileated woodpecker
food, 72

The influence of spatial scale on HSI model output for gray squirrel winter
food, 73

xi

Figure 15.

Figure 16.

Figure 17.

Figure 18.

Figure 19.

Figure 20.

Figure 21.

Figure 22.

Figure 23.

Figure 24.

Figure 25.
Figure 26.
Figure 27.
Figure 28.
Figure 29.
Figure 30.
Figure 31.

Figure 32.

The influence of spatial scale on HSI model output for upland sandpiper
openings, 75

The influence of spatial scale on HSI model output for white-tailed deer spring
foods, 76

Area weighted HSI values for selected wildlife species at 10 randomly selected
inventory points. HSI’s were calculated for different spatial scales at each

point, 82

Overlay of existing vegetation and underlying ecological classification system
to create a resource planning map, 88

The ecosystem diversity matrix for Hoist Lakes and vicinity, Reed Lake, and
Sandpiper Fields, Alcona County, MI. Matrix cells are hectares, 90

Cover type diversity graphs for pine, oak, and lowland conifer cover types.
The x-axis is ecological land type (BLT), 91

Cover type diversity graphs for lowland deciduous, northern hardwoods, and
aspen cover types. The x-axis is ecological land type (BLT), 92

Hypothetical habitat suitability assessment for pileated woodpeckers across the
”Hoist Lakes" landscapes, 98

Habitat suitability index (HSI) histogram depicting management objectives for
a hypothetical landscape, 100

Ecological classification system hierarchy. Based on the works of Cleland et
al. (1993), 103

Habitat suitability index graphs for the ovenbird, 109

Habitat suitability index graphs for the black-throated green warbler, 115
Woody browse variables for FALL/WINTER FOODS, 129

Mast variables for FALL/WINTER FOODS, 130

Agricultural field variables for FALL/WINTER FOODS, 132

Area function for FALL/WINTER FOODS, 134

Suitability curves for SPRING FOOD openings components, 136
Suitability curves for SPRING FOOD openings components. Suitability of

agricultural crops as deer SPRING FOOD, I37

xii

Figure 33.
Figure 34.

Figure 35.

Figure 36.

Figure 37.

Figure 38.
Figure 39.

Figure 40.

SPRING FOOD suitability of forested stands, 139
Suitability curves for deer SECURITY COVER variables, 141

Distance relationships for THERMAL COVER. Suitability curves for
THERMAL COVER variables, 144

Suitability index curves for THERMAL COVER, 146

Vegetative variables and habitat suitability index (HSI) relationships for the
eastern bluebird, 158

Suitability index graphs for the upland sandpiper, 172
Percent canopy cover variables for American redstart, 182

Forest stand description variables for American redstart, 184

xiii

INTRODUCTION

Since the mid-1970’s, legislation and social demands have necessitated dynamic,
ecosystem-based natural resource plans. As a result, the US. Forest Service was mandated to
manage for multiple-resource use with specific directives to maintain or improve biological
diversity at the genetic, species, and ecosystem levels (Hunter 1990). The feature of
"multiple-use" or ”biodiversity" management that set it apart from the traditional single-
resource. sustained yield paradigm of forestry is an ecosystem perspective (Behan 1989).
Despite the general acceptance of "ecosystem" management, the concept remains ambiguous.
and the slow development of ecological analysis techniques reflects the incomplete
understanding of ecosystem structure and function (Hawkes et al. 1983, Hunter 1990).

To employ ecosystem management, we must be able to: I) measure or estimate inputs
and outputs of ecosystem materials, 2) describe processes involved in regulating ecosystems.
3) simulate these processes and their results at various spatial and temporal scales, and 4)
respond to rapid changes in social, economic, and political forces (Hawkes et al. 1983, Boyce
1985, Kaufmann and Landsberg 1990). One of the major problems facing forest and wildlife
professionals in meeting the demands of ecosystem management is incorporating these
concepts into forest plans. Forest plans must (by law) address minimum legal wildlife
conditions (Mealey et al. 1982, Gokee and Joyce 1992), and thus, tools are needed that
evaluate the impacts of forest management policies on wildlife populations (Farmer et al.

1982, Mello and Haufler I983).

Maintaining wildlife populations as viable, natural units requires management, and
sometimes protection, of the entire support system. Yet, effective field applications of
ecosystem-based wildlife habitat management is still in its infancy, although numerous
techniques have been developed to estimate habitat conditions for resource planning (see
reviews in Hawkes et al. 1983, Morrison et al. 1992). Wildlife-habitat models and ecological
classification systems (ECS’s) have the potential to provide ecosystem-based planning tools
with predictive capabilities. Procedures to rapidly determine relative values of land as
wildlife habitat are required by decision-makers, and wildlife-habitat models provide
quantitative habitat values that relate to wildlife abundance or productivity and can also be
used for monitoring the implications of management alternatives. Since integrated ECS’s are
simplified ecosystem maps, they appear to be well suited as a component of ecosystem-based
wildlife habitat mapping (Kerr 1986, Hunter 1990). The ECS also provides predictive
capabilities of ecosystem dynamics (e.g., succession).

Recent advances in computer technology have made geographic information systems
(GlS’s) available to resource planners. A GIS is capable of displaying, managing, and
analyzing geo-referenced spatial data. With the increased need for location-specific.
analytical, multi-purpose information bases (Davis 1980), the use of GlS‘s for ecosystem
management is imminent.

The goal of this research was to evaluate the utility and precision of combining
wildlife-habitat models (US. Fish and Wildlife Service 1981; Schamberger and Krohn 1982)
with an ECS (Barnes et al. 1982, Pregitzer et al. 1983, Pregitzer and Barnes I984, and Spies
and Barnes 1985) for forest planning using a GIS. Justifiably, concerns regarding the
usefulness of incorporating ECS into wildlife habitat assessments and wildlife-habitat model

validity were expressed by management personnel. Therefore, the overall objectives of this

study were to:

1. evaluate the use of ECS in habitat evaluation and management,

2. evaluate the utility of wildlife-habitat models for forest planning, and

3. demonstrate landscape (forest-wide) applications of this approach.

Specific objectives are outlined in each chapter.

This document consists of 3 chapters. Chapter 1 evaluates the effects of incorporating
an ECS into measures of vegetation structure and composition. Specifically, sampling
precision was compared for 3 landscape classification schemes. In Chapter 2, the validity of
wildlife habitat models is evaluated. Also, the apprOpriate spatial scale(s) at which to apply
habitat models is documented. A conceptual approach of integrating ECS and wildlife habitat

models into forest planning across a landscape is presented in Chapter 3.

STUDY AREA

Study sites were denoted as "Hoist Lakes Recreation Area & vicinity” (Hoist Lakes),
"Reed Lake Recreation Area" (Reed Lake), and ”Sandpiper Fields" and totaled approximately
6,000 ha (15,000 ac) in southwest Alcona County, Michigan (Fig. l). Portions of Hoist
Lakes and Reed Lake are currently designated ”non-mOtorized, semi-primitive" and the last
timber harvest occurred in 1992. The USDA. Forest Service manages these sites and
emphasizes providing visitors with remote, recreational experiences (Lantz 1976).

The Sandpiper Fields are part of the Forest Service’s Kirtland’s warbler (Dendroica
kittlaudit) management areas. These areas are characteristically jack pine (Pinus banksiana)
plains subjected to frequent disturbance. The Forest Service emphasizes habitat management
for Kirtland’s Warbler which typically involves whole-tree harvests followed by jack pine
planting.

The climate of this region is temperate. Average winter and summer temperatures are
-6.6°C and 21°C, respectively (US. Department Commerce 1979). Annual precipitation
ranges from 71-81 cm during a normal year, and the greatest amounts of precipitation occur
during May, June, and September (US. Department Commerce 1979). Average annual
snowfall accumulation is approximately 152 cm (US. Department Commerce 1979).

Landforms of the areas are characteristic of changes caused by continental glaciation.
The study sites are located in the morainal outwash plains (Sommers 1984). Relatively steep,

hilly moraines predominate the northern latitudes, whereas flatter, outwash plains are found

 

Alcona County
0.25 cm=1.6 km

Hoist. Lakes Recreation Area (& vicinity)

CE] 4] Reed Lake Recreation Area
Cl

C] C] "Sandpiper Fields”

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 1. Location of study sites, Alcona County, Michigan.

6 .
progressively south (Lantz 1976). The underlying bedrock is composed of shale, limestone,
and sandstone (Sommers 1984).

Soils typically have high sand concentrations of the Grayling-Rubicon series (Lantz
1976, Sommers 1984). Some sites, particularly Reed Lake and the northern part of Hoist
Lakes, consist of a sandy-clay to loam-clay layer, causing variation in site fertility (Lantz
1976). Overall, the soils are classified as low in moisture content, low in humus, medium
acid, and low in fertility (Lantz 1976). Isolated areas of peat and muck are also distributed
throughout the areas.

Different vegetation associations are interspersed throughout the areas as results of
past forest management and soil conditions. Major vegetation associations (Forest Service
vegetation classification) include (in decreasing order of predominance) quaking aspen
(Populus tremuloides), northern red oak (Quercus rubra). jack pine (Pinus banksiana), mixed
oaks (Quercus spp.), sugar maple/beech (Acer saccharum/Fagus grandtfolia), big-tooth aspen
(P. grandidentata), and red pine (P. resinosa).

Understory vegetation consists of numerous species including dogwoods (Camus
spp.), viburnums (Viburnum spp.), cherries (Prunus spp.), witch hazel (Hamamelis
virginiana), beaked hazel (Corylus rostrata), and serviceberry (Amelanchier spp.). Ground
cover is similarly diverse with abundant species including blueberries (Vaccinium spp.),

brambles (Rubus spp.), bracken fern (Ptert'dium aquilina), and sedges (Carer spp.).

Chapter 1: Using an Ecological Classification System

in Vegetation Sampling

From 1900 to the late 1960’s, forest management in the United States was concerned
mostly with fire protection and the efficient harvesting of timber (Carmean 1975).
Throughout the 1970’s and 80’s, however, an increasing environmental awareness by the
public and government confirmed the need for a structured ecological database to meet
multiple use/biodiversity forest objectives. This "new" direction of forestry focused on the
idea that complex ecosystems could be described by structure and how the component parts
were connected (Boyce I985).

Ecosystem components include climate, landform, soils, biota, and associated
ecological processes (e.g., nutrient cycling, competition). The importance of individual
components to resource management has been documented; including relationships between
soils and forest productivity (see reviews in Carmean 1975, Harvey and Neuenschwander
1991), soils and wildlife (e.g., Albrecht 1944, Denny 1944, Crawford 1950, Hundley 1959,),
topography and vegetation (Pregitzer et al. 1983), and climate and vegetation (Moore 1976,
Brubaker and Cook 1983). These single-component classifications proved to be ineffective
multiple-use/biodiversity management tools (Lewis 1967, Van Lear and Hosner 1967,
Broadth 1969, Murphy and Porath 1969, Carmean 1975, Barnes et al. 1982, Jones 1989),

thus, the need for multi-factor. ecological approaches to resource management was imminent.

8

It has long been recognized that ecosystems represent a combination and integration of
all biotic and abiotic elements (Tansley 1935). As a means of addressing ecosystem
organization for resource management, ecological classification systems (ECS’s) have received
considerable support (Hirsch et a1. 1978, Nelson et al. 1978, Rowe 1978, Barnes et al. 1982.
Bailey 1985, Spies and Barnes 1985, Bazzaz and Sipe 1987, Swanson et al. 1988, Barnes
1989, Jones 1989). An ECS unites land elements; including vegetation, soils, water, climate,
and landform; to form a coordinated entity (Hirsch et al. 1978, Rowe 1978, Driscoll et al.
1984, Pregitzer and Barnes 1984, Bailey 1987) capable of predicting ecosystem processes
(e.g., succession; Host et al. 1987, Jones 1989, Johnson et a1. 1991).

Wildlife populations are subject to limits imposed by ecosystems (Farmer et a1. 1982),
yet ecosystem-based approaches to wildlife habitat management have been rare. Single-factor
classifications have historically been used by managers to quantify and display wildlife habitat
(e.g., ”cover” type maps), but due to previous silviculture activities, land use practices, and
disruptions in disturbance regimes, natural relationships between ecosystem components have
been disassociated. Also, existing forest inventories seldom contain the information necessary
to effectively manage wildlife habitat. For example, 53 input variables from 12 wildlife
habitat suitability models reviewed for this study were related to vegetation structure or
composition. The Forest Service’s Corporate Database System could only account for 36% of
these input variables.

The ECS provides a multi-factor, ecosystem-based tool that can be used in wildlife
habitat assessment. Numerous examples of ECS development and use in resource planning
have been published (e.g., Wiken 1982, Van Lear and Jones 1987, Mengel and Tew 1991.
Hornbeck and Swank 1992), but no studies have quantified the usefulness of ECS for wildlife

habitat management. The goal of this chapter is to evaluate the use of ECS in habitat

9

evaluation and management. Specific objectives are:
1. to compare sampling precision of wildlife habitat data stratified by: 1) Forest Service
cover typing and an underlying ECS, and 2) Forest Service cover typing.

2. to evaluate the use of ECS to stratify wildlife habitat sampling.

METHODS

Ecological Classification System

The ECS for this study was founded on the Baden-Wiirttemberg model used in
Germany and is based on the works of Barnes et al. (1982), Pregitzer et al. (1983), Pregitzer
and Barnes (1984), Spies and Barnes (1985), and Cleland et al. (1993). Early emphasis (early A
1970’s) of this ECS was to obtain relatively rapid, inexpensive inventories for broad level
planning purposes (Russell and Jordan 1991). Forest plan implementation, however, required
a finer level of land and ecosystem stratification, thus, ecological land type phases were
delineated for operational forest planning (Russell and Jordan 1991). For this study, similar
ecological land type phases (as delineated by ECS field manuals and Cleland et al. (1993))
were grouped into "ecological land type phase groupings" (ELTPG’s) (that typically
encompassed multiple vegetation types) (T able 1). ELTPG’s used in this study were
intermediate to the ecological land type phases and ecological land types delineated in Cleland
et al. (1993) (Table 1; Appendix A). The finest level of ecological stratification was the
ecological land type phase, however, field logistics did not permit the acquisition of suitable
sample sizes at this level. The intermediate ELTPG stratification was chosen as a spatial and
logistic compromise between using ecological land type phases or ecological land types. The
statistical procedures employed by Cleland et al. (1993) to delineate the ECS support the

contention that some ecological land type phases were similar and could be combined without

10

Table 1. Components (terminology and numeric coding following Cleland et al. (1993))
of ecological land type phase groupings (ELTPG’s) used in this study (also see
Appendix A for explicit definitions).

Cleland et al. (1993)

 

 

Ecological Land Type Ecological Land
Phase Grouping (ELTPG) Type Phase(s) Ecological Land Type

Outwash I 1 Outwash Plains
Outwash II 10,11 Outwash Plains
Outwash; Ice 13 Outwash Plains

Ice; Moraine I 20.21 Dry Ice-Contact and Sand Hills
Ice; Moraine 11 23,25 Dry Ice-Contact and Sand Hills
Moraine l 30,31,33 Mesic Ice-Contact Sand Hills
Moraine II 35 Mesic Ice-Contact Sand Hills
Moraine Ill 37,38 Mesic Ice-Contact Sand Hills

11

significant effects on the sample variance of vegetative parameters (pers. comm., David
Cleland, Lake States Area Ecologist, US. Forest Service, Rhinelander, WI). The relatively
smaller spatial scales obtained by grouping similar ecological land type phases (typically 1 to
10’s of contiguous hectares) as opposed to stratifying by ecological land types (10’s to 100’s
of hectares) was also more conducive to quantifying vegetative structures for operational level
assessments.

Eight ELTPG’s were sampled ranging from ”white oak (Quercus alba)/Deschampsia
on well-drained sands of outwash plains" (termed Outwash l) to ”red oak-red maple (Acer
rubrum)/Desmodium on moderately well-drained sandy loams" (Moraine III; Fig.2). Similar
ecological land type phase(s) within each ELTPG contained like plant associations and overall
soil characteristics. Often, primary differences between similar ecological land type phases
were soil texture and morphology in the substratum (Appendix A). Numerous Corporate

Database cover types were typically sampled in each ELTPG (Fig. 2).

Vegetative Cover Typing

The US. Forest Service’s Corporate Database vegetation classification was
used. The Corporate Database system classifies vegetation into ”stands” according to
dominant commercial tree species, stocking density, and age (US. Forest Service 1988).

Nine cover types at various stocking densities were sampled for this analysis (Fig. 2).

Sampling and Analyses

Vegetation sampling was conducted from 1989-1992 throughout the Alcona County
study sites (Fig. 1). Vegetation variables (as delineated by wildlife habitat suitability models:
Table 2) were measured using a variety of plot, line intercept, pacing, and subjective

evaluation techniques (Table 2; see Hays et al. 1981) from randomly located points within

60382 889 .38

838.5 223.50 33.3% “was... waves—mots Ea 52396 33.8%.“ 3.282 .mucaaoaw 82a 25 e5. Esme—cum .N 239m
mmeiono mas—ago
SSSEBEm 3:qu w goBELm 323% h
83:35»: 33qu SSSasoh 33cm
d3. torquamek U d3. be Shame ”
$35 i8. wanted H €895 d3. 82.85 U
8%: Sebav ES: 3885 .
323...; 5:85 . 3.5a»; 8.2% H 8335:»: “sack . H 3.5%; 5:35
98:38 35.x H c.8238 3E H 95»: .38an _ 5659: SEE H Eases wart
853:8 3:: H 9332 3E U 8:32 as: a 8338 SE ” Sag 3.:

Bap—am 89C. 850 838mm onLoBoU oomEom amp—om

:25 fi :82 G “so mm a: *

b:5a=a>< 0.5352 9523.

 

 

 

 

exam...

Gen

5 05822
a 05802 = 8:82 ”8—
_ 2:222 3:382 ”8. 8— 53350 = £3330 _ snags

ESP—30 woufioommes 320E800 a. $59.80 SSE 2.3. e53 gamma—com

(m) UOIIBA9|3

.232. .93... .a ESSEQH 1mm.

.28 9. m9 .5 «in as... .922... 8......5 9. .9 .5 9m
usage .5 3.9%. 9. m9 .5 «iv was. .953 =<u§§=£mo 9. n9 .5 sin .8: =<nb§§c _

 

13

 

8. 0am. 2.5953 5... .8 £9 E 8. x 9 bacon. 8.
53. 2...... a... .2. 8.9 a 8. x 9 :3 as...

53. Bed .9. .2. 8.9 a 8. x 9 Ema :5 9:5 .988 new
.358 e and

2.02 38.2... 2.... 3.. 3. E m... 3.8mm .82 com 62

63. 0.3.5-0585 2.8.2... 0...... c... 3. E 3 .260 8...). com 2.8.0..
.8... Bed .2. .8 5.9 e 2 x N 92.5 389.8

83. .85 a... .8 5.9 a n x N 925

.8. 2%... 8.35.3 .280 na... .8 8.9 e a x 9 9...... 32.989

83. 330-9582 =0..~E..mm .230 so... .8 3.8 E a x o. .3... E mA .960 9.8.0..
83. uaacm-o=.8.< cozaEzmm .230 so... .8 8.8 E mm x o. ..a... E mA .960 2.3.0..
.9533:

252 a... .3 8.9 a 2 x 9 been

252 .2. as 8.9 a n x 9 8.8.... 8: dz

252 .2089... a... na... .8 8.9 a fix 9 9...:

252 .95 £95.... ”a... .2. 8.9 a 2x 9 ammo

8oz .2. .8 8.9 a 2x 9 82.2.. .23

.EE

8.38.893... 25.9.8... mezeEam eaoEfiax. .8230»;

 

.23. .23. .05W
.889 22368.35... 82. E... .23. ..n .o 9...: com. 8.5.5.8. MERE... 83.88.. .m.u.oEn.§. Susan? .6 EEEam .N 0.2a...

14

cover type-ELTPG combinations. Each sample point was verified as to cover type and
ELTPG. Throughout the analyses, Forest Service cover types on ELTPG’s were the
replicates, and each replicate had a minimum of 3 samples. Vegetation variables were
thus described by means. Only non-zero measures of vegetation variables (excluding count
data and percentages) were used in the analyses, consistent with Forest Service data
management (pers. comm., Peter O’Grady, U.S.D.A. Forest Service, Systems Analyst,
Cadillac, Ml).

The parametric assumption of data normality was evaluated for each vegetation
variable using frequency distributions, probability plots (Sokal and Rohlf 1981), and
Shapiro-Wilk’sW test (£50.05) (SAS Institute Inc. 1985, D’Agostino et al. 1990).
Parameters that exhibited non-normal distributions were analyzed for outliers using Dixon’s
test statistic (nSZS) or Grubb’s test statistic (n>25; £50.05) (Sokal and Rohlf 1981).
Data transformations were performed as needed to normalize vegetation parameters (Sokal
and Rohlf 1981) (T able 2).

Parametric relative efficiency tests (Sokal and Rohlf 1981) were performed to
evaluate sampling precision. Relative efficiency tests evaluate ratios of sample variances
(Sokal and Rohlf 1981), with the premise that lower sample variances equate to more
precise estimates. Precision was evaluated between samples obtained by: 1) Forest Service
cover type and ELTPG,2) Forest Service cover type (Fig. 3), and 3) ELTPG. Two
aspects of the relative efficiency tests were used: 1) the magnitude of the variance ratio
("> 1" indicated a gain in precision), and 2) the significance level. Significance tests were
one-tailed (PSOJ; df=n-l.n(s-l); where n=replicates and s=samples within replicates)
and tested the null hypothesis that samples obtained by Forest Service cover type exhibited

precision equal to the precision of samples obtained by cover type and ELTPG

15

028...? 0.9.... 865.. ~<. 0.2.3 $.33. 65.050 3.3.0. 33.8... E... aaE MESS... 3:6... .6 2.83m. .m 0.29..

 

 

 

 

 

 

 

 

 

52»
EB»: .3230
36.3". ........ 2. 5830...“. a. 2.... .0689. ........ 2. £83.20 a. 2.... 3689.. ........................................................ .26 0 .8250
a . .2698. .. 5856 £88 .958. .8... .8.
56$
.6 a on... .u 8 2.3.: .3250
30.3.... ........ A. _. “.0.me WZQW .> .mmi .......... A. 5.96:0 d 2.... 360K... ......................................................... .960 .3930
< 282.5 £88.58. .8... .8.
5...... d on»... .960 336m .86...
:20
Eat... .3230
Aéu h0>OOvN<m .......................................................... .O>OU ~AQO=NU
8.28.. .8? .998. .8... .8.
.25.. .960 8.25 .86...
8.9.5389... .28.-.. oocmta> 2.28m .35 88.7.3.
.. £8350 ”OE”... . H1.9.3.5 ”gum.
.87 Hm.58.. .8? ”.88..
magnum .35 «Sam “on...
8. .18. ”8.... .26.“. ‘ / 8.. .8. “25 .98
556m «A»... .260 ..
ACES...
Man—=20 35 g fig qua—”28m ........................................

.8? U99.8..
.83. “8..

2.... an... ”9...... .260 1

 

16
combinations.

Relative efficiency tests were also performed between samples stratified by
cover type 91 ELTPG. For this analysis, more replicates were typically obtained for
ELTPG’sas opposed to cover types (Fig. 4). To standardize the effects of sample size on
estimates of variance, replicates were randomly selected from the ELTPG database (Fig.
4). The number of replicates within each ELTPG sub-set corresponded to the number of
cover type replicates in the comparison (Fig.4). This process was conducted 10 times for
each vegetation variable to minimize the effects of outlier ELTPG replicates. Thus, within
each vegetation variable, 10 relative efficiency tests per cover type were performed,
resulting in 10 f—probabilities per cover type (Fig. 4). The f-probabilities were subjected to i
a statistical technique that combines numerous probabilities into one probability that is
distributed as a chi-square statistic (see "Miscellaneous Methods” in Sokal and Rohlf
1981)(Fig. 4). The magnitude of the chi-square probability was used to determine
significance, and gains in sampling precision could be determined by comparing the
calculated probability to a chi-square probability derived from 10 F-ratios equal to 1 (i.e.
if the F-ratio is s l, sampling precision was not gained).

Vegetation parameters were grouped according to overstory. understory, and
dead and downed. Overstory was defined as woody species 211.4 cm (4.5 in.)
diameter breast height (DBH) and understory was woody species < 11.4 cm DBH. Dead
and downed material included snags (standing, visibly dead overstory) and logs(2 17.8

cm (7 in.) DBH).

RESULTS
The number of replicates sampled ranged from 3-ll for cover type-ELTPG

combinations, 6-l7 for cover types, and from 3-49 for ELTPG’s(T able 3). Univariate

008...;
0.9:... 88:0: ~<. 0.0...3 AGED... w:...=o.w 0.8.... 0...: :8. .8.wo..000 :5. 09.. .030 c0030.. ...0. 8:06....0 02.8.3. .8 Saw...

26-0.55... .6.

 

 

 

 

 

 

 

 

 

  

 

 

 

 

 

or. A .88.. .288.
2o-0....n.m..~<. .N
.x. ‘ N. ‘ .88.. .83...
N
20-0...559. ..
E ‘ 8-8.. .8. N...
b.....~..o.n. 0.8.0....0 83:382..-. .23.-..
. V a m. o .8 .85
. co. m. c v.80 .085
- . E. ' .o m v.80 v.08...
m
.5 0.9.5.? o.. . a \ ' .o m .80 .08...
. 88.: 2 m m. m. o 8.. 8..
. .o ...=o.m o. 80.0. 3:883. 00 .5. c 0:. n. :3.
MSHOAHEWVM .m ‘I .88....3. 5-0..qu .. no .5. w 0:. n. .03.
s 0.5.. P _ a. :6. o 8... .8.
mm ' .o m 0:... V.08..
.. h ' .o m 0:... «.08...-
.05.. .08.. ~<. ‘l .88.. 3. v 0.60.2... 0.08. m f E. ' .o v 0:... .083
no Iv 6 v 8.. .8:
.. -Iv a v 8.. .8.
005....» 0.08m .060 36:80 0...... 0.50 2...... .260
.080 30:80 :82 05.88.09.

 

8.8....3.

18

 

 

 

o a m 8. a n. e 1.28:... 2.5...
. . n 5.5. .536...
. e e n . 6N . 8.5.. .5565
o v . . . . a o 882...: .z
. . . . . . 0. e .5 B52
. . . n . . 2 o .5 8.. .z
m . o . . . a m .5 .8...
. : . . . : e 05.. .8.
v . . . v . 2 m 05.. .8.
. . . . n n e e .5. .8.
hall—U:
e a. n a. m 2 c 128...... 0.50
. . . 8.2 .5585
. o n z n 8...... .5389...
e e . 2 e .838... .z
. m m 0. o .5 .85:
m . . m . . 2 o .5 8.. .z
m . e . . . o m .5 .8...
. : . n m t e as... .8.
v . . v .. 2 n 05.. .8.
. m n o e .5. .8.
58.0
.= «5.32 = a5...... . .5262 = 085.5282 _ 0526.28. 8.58380 = 8.3.5 8.8.8.. .85 25 .260
0...... .060
.5880 as... 25 83 3.58.. 25 .260 8 508.0 8.189.,

8.32... :0...00... 0.050. .0. 00.0 0...... 0:0 .00... .060 ..:0..0:...:.00 0......m.-0..b .060 .0. .28....0. .0 80:52 .m 030...

.0300... 8.8.30. 0 ZN

.38. 00_>.0m 30.0.— .mdv 0.03.0008. 00.00..
-=03 0:. 5.300.: .-.000 .003.-0_00u0.m.0 ”00.003003 .00N..-w::0..um 0.0:? .0000 33003.0... 0.2.8.0 0...00.00fl0000_

 

 

 

w
|
e a. m 3 n 2 e 18.8.... 0......

. . . . . m 5.2 .5355

m . . n . 2 n 8.2. .5355

e v m . . . o c 88308: .2

. n n . . . 2 . .5 85:

n m . m . . 2 e .5 .8. .z

n . e . . . . n .5 .8..

. : . n a 2 e .5. .8.

v . . . e . 2 m .5. .8.

. . . . n m e v .5. .8.

32.5 d .80

.= .55.: = .55.: _ .52.: = .553”... _ .553”... 8.5835 = .8350 8.8:... .85 a... 880

000,—. .060
90000.0 03:0 00C. 051— .860.08— 0.03. .060 d blah—‘0 in;

|
.C.:oUv .m 030,—.

20 -

normality tests indicated most vegetation parameters required data transformations to

satisfy the parametric assumptions of the relative efficiency tests (Table 2).

Relative BfficiencyTests ~Cover'l‘ype and ELTPGversusCover Type

Relative efficiency tests indicated that sampling precision of all vegetative
variables could be potentially increased by incorporating ELTPG’s (Tables 4-6). Typically,
within each vegetation variable, greater than 50% of the sample variance comparisons
suggested gains in sampling precision (i.e., the F-ratio was greater than 1.00; Table 7).
For example, relative efficiency tests were conducted for basal area on 9 different cover
types representing 7 ELTPG’s(Table 4). Gains in sampling precision were implied by 13
(65%) of the relative efficiency tests, however. only 5 (25%) of the tests were significant
(250.10)“ able 7). For all vegetation variables, 43-74% (mean=59%) of the tests
suggested a gain in sampling precision, with 5-39% (mean=24%) of the relative
efficiency tests exhibiting significant gains (Table 7). At a lower probability (a=0.20),
the number of gains per vegetation variable typically increased, with several variables
demonstrating significant gains for >40% of the tests (Table 7). Vegetation variables that
exhibited the most significant gains in precision were understory cover 25m tall and stem
density, and snag diameter, whereas the fewest gains in precision were associated with
overstory number of tree species. numbers of soft mast species. and snag density (Table
7).

Samples stratified by cover type-ELTPG were also evaluated for each cover
type by vegetative variable grouping (i.e., overstory, understory, and dead and downed).
Generally, gains in sampling precision were suggested for > 50% of the relative efficiency

tests, however, the number of significant gains was variable (Table 8). Pole-sized,

21

 

N... . 8..< .5353

. 0... n 8.2 .5355

:....2 +3.. .1. . e .8905: .2
. wed 2:ch c 0.00 000.2
+.... .2. . $8.... . .5 8. .z

.N... . :0... . . .5 .8...

. S... 5.0.. 2.. o .5. .8.

5.... .... 8.. . . .5. .8.

. . 3... +8.. v .5. .8.

5...:

o... . 8.5. .5353

. 5...... n 8.3. .533...

.2. .2. +8.. . o .8868: .2
. .2. a... e .5 85:
c... o... . .8.. e .5 8. .z

N... . 8.. . . .5 .8...

. .N. 2.. .2. o .5. .8.

a... .2. :83. . . .5. .8.

. . +mm.n uuuwcd. v 05.— 300—.

3

8... . 8.2 .538...

. 3.0 m 500... 90—080....

3... +N..N .2. . e 5835.: .2
. 2.. 0... o .5 85:
.N... N... . .N. o .5 80 .z

0N6 . «35¢ . m 3.00 x005

. :0 :2... .0. e .5. .8.

8.. :8... 3.. .. . .5. .8.

. . N... 3...... e .5. .8.

82180

.0 .55.: 0 .55.: . .55.: .. .55.:8. . .55.:8. 8.0.8380 .. .8350 .280 ..... .260

 

$0.000.» 0.2.0 00... 000. 30.00.000.00... .060 30.0. .3 005...... .200... 0003.00 ...0. 600.0000 03.0.0. 00 0.00.0.5

.5550 .8... .9; 53 35.580

 

8.. 858.; a. .35.

.30.... 0000...? .0... .002-.. 0.0 .00.m> .0330? 00.000»? 0.0...06 .0. 00... .060 50.0.. .0 2.00.50.

.0 0.0m...

22

.8... 20.05.55.

._0.0w§« 00. .m0.0w§ .0..0w* .0~.0w + 0.003 00300.0 00.5.0.0...
.0>00V00_._00.0 00... .960 a: ”00500.0 00.5900... .0>00u00_._00.0 00... .0>00 no: 00 5.0. 05 .00 30.0.0.0... 00 .0>0_ 8.0000N

.800. 00_>..0m 50.00 .m.:. 0.02.0008. 00.00..
0.3 00. 5.0.000. £000 .000..-0_00n0.m.v ”00.00....03 .000..-w0__0..um 0.003 .0000 05.09.05. 0.33.0 0...00.0uu0000.

 

 

 

.0. . 5.2 0.0.5...

2.. . .8.. . 8.. . 5...... .535...

:5... .... . +8.. . . e 5835.: .2

. 3.. . +...N . .5 85:

+v—.N +3; . :«mod— 0 300 v0»— .2

a... . :3... . . .5 .8..

. . N... . :.N.. +.... e .5. .8.

+3.. . .2 . .0. . . .5. .2.

. . . . 0... o... e .5. .8.

. 015 B...

.0. . 5.2 .5355...

. .0. . 8.2 .535...

5.. .0. . 0... . . 5835.: .2

. 0... . 5.. e .5 B5:

.... .N. . .5. e .5 .3. .z

.. .. . 8.. . . .5 .8..

. . .0. . a... .0. e .5. .3.

2.6 . 2; . «32.6— . m 050 so:

. . . . 8.. 3... v .5. .8.

on... B... 5.5::

0. .55.: .. .55.: . .55.: 0 .55.:3. _ 55.2.... 8.55350 = .5350 .80 a... .960
.5555 a... o... .5... .8....80 a... 8..s...> a 5.3....

5.000. .0 0.00.0

IIIV1I.III‘«f-

.i Q .1 1“

\Ini'iOI-II\III\.‘I‘

l\:.~§~n‘~\.h .‘n‘

>Ls.:::2.¢.

.W Stink.

23

 

0...... n 5%.. 053.5;

n... . 3.... n 5%... 0536...;

unohbé— «not—.NQ . 0 3003—03: .2
. a... +2.. 0 .5 8...:
.2. 8.. . +03 0 .5 8.. .z

.3... . 2... . m .5 0.8...

. NZ. 9.... 0 8... .8.

2.. 8... . m 8... x8.

. .n. 8.. .. 8... 0.8.

30.0.. 3.38

a... n 5.2 053.5;

mm... . 8.. n 3.2 «Essay

3.5.8 3.3.2 . 0 300363: .2
. 3.8.9. .0..." 0 .5 85.2
+NM.N pound—N . :an o .30 vom— .Z

nan—.03. . .wd . m .30 0.02m

. 2.. .2. 0 a... .8.

v... 8... . m a... .8.

. .0... .2. v 8.. x8.

.1... an N: 880

g... n 8.5. 0538.5

.3 . a... m 8.2 05365

«ooh—.NN— :unnd— . c 3839...: .Z
. .... 2... 0 .5 8...:
8.. 3.8.2 . 1&3. 0 .5 8.. .z

303.: . mod . m .30 soc—m

. +3.. .2. 0 8... .8.

9.. R. m n a... .8.

. S... 3... e on... so:

.1... an M. .260

E 8.5: = 8.95.2 . 8.5: = 8:52.00. . u....a£..x.o. 8.2.228 = 5:25 .38 0...... .960

 

359.8» 00...... 0.5 v5. 303280.09: .030 3.0.5. .3 BEE... 00.95... 503.2. 30.0. 35.050 03.20.. .0 .5885

$386 as... 25 0.3 3.6.8...
|

 

25 8.3%; a 0.3!,

.82.... 005...? .0... 8:2-.. 0... 002.; 0030...... 8:80»? b23095 8.. 0%. .080 3.0.8 .o 3.9:.qu

.n 2...:

Afi. Ihhi'l\\ .v. t.V\A\§..-r.~t

24

 

+8." m 8.3 05355

+53 8.. 8.. S... n 5.2 053.5;

3... 8.0.... 2.... . . 0 883.5: .2
. .02 m... 0 .5 02:2
0:8.N— Gd . 0:. e 0.00 no: .2

a. .c . an. . m .5 .8...

. a... 2... 3... 0 oz... 0.8.

+2.0 +2... +8.0 . m a... .8.

. . 8.. 2... 0 E... .8.

8.88 I: :8 .oz

0:. m 5.2 asses

E... 8.. 2.... +30 M 8.2 05380;

a... 2... 2.... . . 0 883.5: .2
. 3.... :0.“ 0 .5 02:2
2.... S... . .50 0 .5 8.. .z

2... . 3... . m .5 x8...

. 3.. .38.”. .00.... 0 on... .8.

3.. 2.... x... . m on... .8.

. . R... S. .. 2.... as.

U80 ‘2 com

02. m 8.2 05388..

9... a... 53.0 3.03.. m 5.2 05380;

X... n... +0.0 . . 0 883.5: .2
. 9... .38... 0 5 8...:
an... 3.8.2 . +9.5 0 5 3. .z

.36 . an. . m .5 .8...

. on. 3&4». 55...... 0 2.... .8.

3.. a... .8..” . n a... .8.

. . m0... 3.. v on... 0.0...

in .50

=. 8.5: .. 8.8a: . 8.5: .. 05082.8. . 8.52.8. 8.2.33.5 .. €225 85 2.? .95

 

059.80 o2.... 25 05... 3.0280

 

25 8.300; a 0.3!.

6.50. .m 030,—.

25

Saw... a... .33: 6...? 2.3.. 22.3 .528... ESQ»...
.o>oov:o_m_8.a 2:. .38 a: ”cease... 09592.... .o>8u=o.m.8... on». .38 no: .8 .3. o... .8 8:85am... .o .32 8850.

.20.. .5658...“

.83. 3.20m .88". .md. 33.80%... 63.8...
-=.3 E... 3533:. (.2... .uofim-u.onue.m.v ”Es—8......» 6.5-3.3..."0. 0.2.3 8.5.. w=_._8.m\00_m 820.00 v.0.oEouuuuoP

Ara—o“: .w ”EMF

l!!uluu¢n.ll- {111:1 ll-|¢ 114‘. .1---«II!--.u--- ..l.-< 11--..-. 1111..-! .‘l. --a-nvuuqfll1€ {inductii ‘11:}...qu ,uciauc :yhflcnq...\u.\.\ni QI.¢.--.~!‘L..\a\ \uLEEFh-~.V. -\ .\\A\~.NN.

26

 

.N. m 8%... 9.250....

3.3.2» . 006 n :85. 3:288...

av... 3.8.x . c .8320: .z
. 3... :3... c .30 80.2
on... .c. . 2.... e .30 .8... .z

1.86. mm... . m .30 .305

. 00.. e on... .30.

m0... 0... m on... .300

. 0:. 3... v «a... .300

.15 03

3.. m :83. «53.8....

. n 6.2 0535.;

00... 3.3.2 o 0.803.802 .Z
. on. m0... 0 .30 80:2
9.... «We . .30... e .30 .8.. .z

.m. .v . no... . m .30 .305

. o... cad e on... .30..

mm... .300... m 05.. .30..

. :wa nummé 0. 0:5 300—.

anal... 0!.

me... m 892 9:358...

. 0 02.2 053880

3.8.0 056 0 8833.0: .2
. 2.... S. 0 .30 80.2
00.. mad . +5.0 0 .30 8: .z

3... . 3.... . m .30 .305

. no. me... e on... .300

o... 00.2.: 0. m 0:5 aqua

. +0.20 mm. v 2.... .30..

.15 0.8

.= 3.0.02 = 8.0.02 _ 050.02 .. 8.0.52.8. . 8.0.52.8. 8.5.0350 = 50350 .080 RE... .960

 

”£9.80 80.... on»... 9.0.. .8.mo_8m

 

25 8.3m“; .0 .32.;

.32.... 8..0...0> .0... 02.0.-.. 0.0 8...0> .8...0..0> 02.0.82. .8038 .80 v0"... .8 on... .88 .88.. .o 3.0...58

30.95.» 8.0.... a... 9.0. 3032892.... .38 .85.. .3 85.0... 8.0500 083.2. 0.8. 88.qu 03.0.0. .5 00585

.0 030...

<£fih
.\ .
lxi-\Y\\

. \.

27

..c.cw§* 0:0 .nc.ow..i 63cm... 625... 0.2.3 00.0.00... 00.03.09:
.0>ouv=o...00... 0%. .960 0.: ”00.0.00... 0.0.5-0.... .0>00u=o...00... 0...: .260 no: .0 30. 0... .o. 8:00....cm... ..o .20. 00.2.00.H

.30.. 80.05005.

.3... 8E0. .85. .3: 2308.8. .028...
._.03 0:0 .-E=...0:. (.8.. .000..-0.o..uo.m.v ”08.02.-..03 .000..-m=....0.um 0.2.... 00.80 05.39.00... 0.0.3.00 0.0.0900fl080.

6.5.0. .0 030...

28
Table 7. Summary of precision gains for each vegetation variable stratified by forest
cover type-ecological land type phase grouping (ELTPG) compared to samples
stratified by forest cover type.

—
No. Relative Efficiency Tests (% of Total)

 

 

 

Significant Gains
Total Relative _ '

Veg . vmbk Efficiency Tests Steamed Guns P5010 P5020
Oven-y

BasalArea 20 13 (65%) 5 (25%) 6 (30%)
Diameter 20 12 (60%) 6 (30%) 8 (40%)
Height 20 13 (65%) 6 (30%) 9 (45%)
Number Tree Species 20 9 (45%) 1 ( 5%) 1 ( 5%)
Tree Density 2?. 14 (64%) S (25%) 11 (50%)
Under-nay

Cover23m Tall 18 12 (67%) 5 (28%) 6 (33%)
Coversz Tall 18 ll (61%) 7 (39%) 8 (44%)
Deciduous Height 18 10 (56%) 3 (17%) 5 (28%)
Stem Density 23 17 (74%) 8 (35%) 10 (43%)
Soft Mast Cover 23 ' 10 (43%) 4 (17%) 5 (22%)
No. Soft Mast Species 23 13 (57%) 3 (13%) 8 (35%)
Dell! and Dowmtl

Snag Density 16 9 (56%) 2 (13%) 4 (25%)
Snag Diameter 16 9 (56%) 6 (38%) 6 (38%)
Log Density 17 9 (53%) 4 (24%) 4 (24%)

|F-ratio greater than 1.00.

29

Table 8. Summary of precision gains for each forest cover type stratified by forest cover

type-ecological land type phase grouping (ELTPG) compared to samples

stratified by forest cover type.

_
No. Relative Efficiency Tests

 

 

 

(% of Total)
Significant Gains
Total Relative

Cover Type Code' Efficiency Tests Suggested Gains2 P5010 Ps0.20
Ovalay

Jack Pine 4 10 6 (60%) 2 (20%) 4 (40%)
Jack Pine 5 15 12 (30%) 5 (33%) 6 (40%)
Jack Pine 6 15 10 (67%) 3 (20%) 4 (40%)
Black out 5 10 5 (50%) 3 (30%) 3 (30%)
N. Red Oak 6 15 10 (67%) 4 (27%) 7 (47%)
Mixed Oak 6 10 s (50%) 2 (20%) 3 (30%)
N. Hardwoods 6 15 6 (40%) 2 (13%) 6 (40%)
Trembling Aspen 3 7 s (71%) 2 (29%) 2 (29%)
Trembling Aspen 5 s 2 (40%) 0 (0%) 0 ( 0%)
Unlanry

Jack Pine 4 12 4 (33%) 0 (0%) 0 ( 0%)
Jack Pine 5 15 11 (73%) 1 (7%) 4 (27%)
Jack Pine 6 15 s (53%) 4 (27%) 5 (33%)
Black Oak 5 12 7 (58%) 4 (33%) 4 (33%)
N. Red Oak 6 18 12 (67%) 7 (39%) 10 (56%)
Mixed Oak 6 12 s (67%) s (42%) 6 (50%)
N. Hardwoods 6 15 10 (67%) 7 (47%) s (53%)
Trembling Aspen 3 1s 12 (67%) 2 (11%) 4 (22%)
Trembling Aspen 5 6 1 (17%) 0 (0%) 1 (17%)
Dad Ill Donal

Jack Pine 4 6 4 (67%) 2 (33%) 3 (50%)
Jack Pine 5 6 5 (83%) 2 (33%) 2 (33%)
Jack Pine 6 5 3 (60%) 0 (0%) o ( 0%)
BlackOalt s 6 2 (33%) 2 (33%) 2 (33%)
N. Red Oak 6 9 4 (44%) 1 (11%) 2 (22%)
Mixed 08k 6 6 3 (50%) 1 (17%) 1 (17%)
N. Hardwoods 6 6 3 (50%) 3 (50%) 3 (50%)
Trembling Aspen 3 2 l (50%) l (50%) 1 (50%)
Trembling Aspen 5 3 2 (67%) 0 (0%) 0 (0%)

'Code=Corporate Database size/stocking codes where 3=sapling sized, well-stocked;
4,5,6=pole sized, poor-, medium-, and well-stocked, respectively (US Forest Service
1988).

2F-ratio greater than 1.00.

30
medium-stocked black oak exhibited the highest significant sampling gains across all major
vegetative groupings (30. 33, 33% for oVerstory, understory, and dead and downed
respectively; Table 8). Other cover types that exhibited relatively high numbers of
significant precision gains included pole-sized, well-stocked northern hardwoods (47%) and
mixed oak (42%) for understory; pole-sized well-stocked northern hardwoods (50%) and
sapling-sized,well-stocked trembling aspen (50%) for dead and downed variables (Table
8). Low numbers of precision gains were consistently demonstrated by pole-sized,
medium-stocked trembling aspen (Table 8).

Generally, the cover type-ELTPG stratification exhibited higher precision for
cover types not specifically defined by the Corporate Database classification. For example,
"northern hardwoods" was an inclusive label typically referring to sugar maple-beech
forests, however, these stands also included samples dominated by yellow birch (Benita
allenghensis), paper birch (B. papyrifera), and oak. Similarly, the cover type denoted as
"black oak” for this study included black oak, scarlet oak (Q. coccinea), and hickory (Carya
spp.)(U.S. Forest Service 1988).

No ELTPG exhibited consistently higher or lower precision for samples stratified
by cover type-ELTPG (T ables 4-6). Some ELTPG’s were poorly represented in the sample
(e.g. "Ice;Moraine 1" and ”Moraine 111"), and these ELTPG groups typically exhibited the
highest number of precision gains (Table 9).

Relative Efficiency Tests - Cover Type versus ELTPG

Relative efficiency tests comparing cover type and ELTPG stratification schemes
indicated that for certain vegetation variables, an ELTPG stratification could provide more
precise estimates of vegetative variables (Tables 10-12). Specifically, most vegetation

variables sampled for pole-sized jack pine at all designated stocking densities exhibited

31
Table 9. Summary of precision gains for each ecological land type phase grouping
(ELTPG) stratified by forest cover type-ELTPG compared to samples stratified
by forest cover type.

—
No. Relative Efficiency Tests

 

 

 

(% of Total)
Significant Gains

Ecological Land Type Total Relative
Phase Grouping Efficiency Tests suggested Gains2 P5010 Ps0.20
Gouda-y
Outwash 11 10 7 (70%) 2 (20%) 4 (40%)
Qawashzlce 20 16 (80%) 6 (30%) 7 (35%)
lee;Moraine 1 5 4 (80%) 3 (60%) 3 (60%)
lee;Moraine 11 31 18 (58%) 7 (23%) 10 (32%)
Moraine 1 5 3 (60%) 0 ( 0%) 1 (20%)
Moraine 11 26 11 (42%) 3 (11%) 8 (31%)
Moraine ill 5 2 (40%) 2 (40%) 2 (40%)
um
Outwash 11 1, 5 (42%) 2 (17%) 2 (17%)
Outwash;lce ,; 10 (48%) 4 (19%) 6 (29%)
lce;Moraine 1 “9 7 (78%) 4 (44%) 6 (67%)
lee;Moraine 11 36 20 (56%) 3 (8%) 8 (22%)
Moraine l 6 4 (67%) 3 (50%) 3 (50%)
Moraine 11 33 23 (70%) 11 (33%) 14 (42%)
Moraine 111 6 4 (67%) 3 (50%) 3 (50%)
Dad and Downed
0mm}, 11 5 2 (40%) 1 (20%) 1 (20%)
mace 7 5 (71%) 3 (43%) 4 (57%)
alumina 1 3 - (67%) 1 (33%) - (66%)
lee;Moraine 11 15 9 (60%) 1 (7%) ‘ (7%)
Mom 1 3 1 (33%) o ( 0%) 0 (0%)
Moraine 11 13 7 (54%) 5 (38%) 5 (38%)
Mom In 3 1 (33%) 1 (33%) 1 (33%)

‘Code=Corporate Database size/stocking codes where 3=sapling-sized, well-stocked;
4,5,6=pole-sized, poor-, medium-, and well-stocked. respectively (US. Forest Service
1988).

2F-ratio greater than 1.00.

 

 

32

 

I '1

0083.5: 05...: 02...).
1.0.00 2... :0 3.00 e .00 02...).
. 30M and . 1 . 0 x00 001
MO.’ GW.F 00% a .OJ—N 5' O Xflo XOQ—m
00.0 60.. . . t . 0 cc... .8..
8.. 00.. 00 0. 8.. R .. 0 2.... .8.
. .nd 0306.3 8.9— 305—.3 :03.“ o 050— 0.0:

. . 00.0 8.0. .800 0
0.0.0..
.0... 00... 00.0 as 0 0083.5: 05...: .80.:
. +3.00 8.00 . c 6.00 80.2
.00 8.. . s... 00.0 00.... 0 0.00 .60
. 00... . 2.... 0.... . 8.0 0 0.00 08.0
0.0 0... . . 0 0.00 0.8.0
00... .0... 8.. a... 00.0 0 on... .5.
. 00.0 :0 .0... . 3.2.00 0 cc... 0.8.
. . 0.00 00... 3.8.0 0 as... 0.8.
in
2... 3.0 80 00.0 0 0083.5: 05...: 02.2
. 2.... 00.0 . e 0.00 80.2
0.... .2. 00... 00.0 8.0 0 .00 .5.
. 8.0 00.0 .00 . 00.0. 0 0.00.8.0
00... an... . .. 0 000 s.020
00.... :0. 3.00.00 2.0. 3.00.... 0 on... .5.
. 00.00 3.000.. .3020 . 3.0.8. 0 cc... 0.6...
. . $0.00 00.0 8.2....“ 0 on... v.8.
8! 1....
E 8.86: .. 8.86:. . 8.86.2 .. 8.6.62.8. . 65862.8. 8.0.33.5 .. 033.8 . 02350 .680 on... .6560

 

0308.0 0..an... 25 no... 16.0280

 

60.0 8.30m; 0 can;

.00.". 000. 0.00. 300.050
0.50.0. 0. .0 0.300. 0... 00.03500 50.. 003.00 00.0.0.0 0.00.00 20 0.0 000.0.» .0030. .0 > 00.00.00? E0502. .0. ADA—ham.
«0.0.5.0 000.... 0%. 2.0. 00.00.80 00 0%. .0>00 .005. .3 00.0.0.0 00.00.00 0003.2. 0.00. 0000.050 03.0.0. ..0 E0850» .0. 030...

0.00 00.05.00...

.00..00.....0..0 00.5.0. .0. 00.0.00... 0.00.00 .00»... 0 A. 00...

0.00.0 0.03 02.0.-.. .0... 00.000000 0..0..0.0 0.00.00.00 0... .0... 00.00.00. 80.0.0: ..o.ow.*. 000 .nodws. .0..ow. .o~.ow +
0.0....» 00.0.00... 00......0.V00.0.00... 00... .0>00 w: 00.0.00... 00.....mu00.0.00... 0%. .0>00 ..: .00. 0... .0. 00000....030 .0 .26. 00.0000.

 

 

 

33

 

 

 

.600.
00.>.0m .008... .3: 0.03.8000. 00000.0-..03 000 £02000. .000.0-0.0..u0.m 0.003 00000 00.08.0000 00000.00 0.0.00.00u0009
i
8.0 8.0 . .00 8.0 0 .0832... 05...: 0...:
. . . . . 0.0 0.0.2
0.0 0. 00 0 0 0. 0 B0
0... . 00.0 . .0. .00 E... 0
. . . . . . .3 000 0003
8 0 00 . 0. 0 3 . 0
3.0. . .00 . 0 .Hmflum
000—066— . 0003.~h 000—n.~.. . 0008.8. 000N650. e 0...... .00..
. . 00066.8 001%: :06— .E. . 0003.0? 0 .
. . . . . 0....— 000..
8 0 00 . 0
3.60 .2...
0.0 00.0 . 00.. 0... 0 .832... 2...... 0...:
. 00.. . 00.0 . 0 0.0 02...).
00.0. .00. . 8.0. 00.0 +2.00 0 0.0 .30
. .0. .0 . 0.00 3.0. . . 3.00.00 0 0.0 08.0
00.0 . 8.0. . . . . 0 0.0 00....
0... . 00.0 00.0 . 00.00 0.0. 0 .0... 0.0
. . 00.0 00.0 00.0 . 00.0 0 .0... .0...
. . 00.0. .0000 00.0. 0 .0... 00..
3...... B... has:
.0 .5802 .. 8.05:. . 8.80.2 .. 8.2.2.8. . 8.30.2.8. 8.0.82.0 .. 0.5.00 . 0.3.5 .800 2...... .260
.05 3.0%.... 25 80.00... 0 is... >

 

5.000. .0. 0.00...

 

 

unuhvfin 3.2 o an v0.52

.3 +8.R a; 8.2 .33 a: o a5 8:

. 303.; . .33.: 3.3.8 . 3.3.3 o "10 «.83

2.. a... 1: $5 3.. o 85 Ba

. ~62 8.: «3. 2.» 3.2 e on: a:

5.5:. 8.8 . 3: 8.2 .38 m 85 so:

. . . a: a... N; v 8a x:—
it: mi

5.» 8.2 o :5 85.2

2.: 3.: can 3...: 3.2 e x5 8..

. 8._ 3a .3 . a.“ e x5 “.85

«3 a: 8.: a: o 8: 3.

4 . 8.2 .33.: 3.: 2.8% o SE a...
3 S... 2.2 . 8.“ v.2 n 8E x...
.3 a... N. .960

a; 2 a e “.5 8::

a: a... 35% Rd. 3.2 ..:..R a e “.5 Ba

. $8 . 3.. 2.. . 9.... o .10 x85

and N: 33 28 Sa 0 8E 3.

. 5.2 an: .58 3.2 +8.5 0 SE .6:

m3 . 9.... . 8.2 92 ed n 8: v.2:

. . . . .38 «.2: 13.8 v 8E x...

.1... an N .260

E .552 = 03.32 _ .552 = .seozg. _ useozg 8.5.225 = 522.6 _ $25 .38 25 .260

 

 

mafia as... 25. 23 .8m28m £5. 8.5.“; a 3.1.;

.3. .mE 00$ 88. 358$» 0352
2 ho 8.38 05 9:59.59 Eat 33.0.. 852m: 826m Eu 2n 82.3 .8335; 5:80»? b.9802: .8 6958
9.32:» 823 on». was. .3328“. a. on». .96”. 3.28 E 3:223. SEES :85»: 38. 3:3qu 02.22 .6 ban—Ea .: 038.

 

35

 

 

1. I F

«3N9: 3. _ _ 3.8.3 .38.!” o mug—Echo

8.” 8.: . . a... n .22 85835
. :2. 8.8 on... n 592 ”5353
«on no: 53 :3. 0 $832.: 233: 8...:
. 3.8.2 as . e “.5 can:
3.: m3. .2 as" 2.2 8.2 . e “.5 Ba
. unuadv . uoaQNdv +£th . «co—56m c «30 Jun—m
«38.8 3.05.: . . . m an 335

+33 2: 3.2 b; 39$ e 2E B:

. .33.: 9d .338 8.2 in...“ o 2.: :2:

SN 8.9 . N3 9; awn n a: so:

. . 3.2 n2. 9... c 2.: x8.

880 I: :8

. 8.8.3 o? :32 2.2 o $588
and v: . . 98 n 8.5 Essaa
. 8.» 3.... 2.. n .22 ”canes...
and 3... 8a 3o 0 383.5: 22.5 85.2
. «33.: again . . o g @052
3.. 3: 2a 8.: .3 3.2.8. o 55 Ba
. 8s . m3. 2: . :3 e “.5 :85
Se .3 . . . w x5 “.88

3d of 3c . 3... 3o 0 as 3.

. «no 3.. 38.8. 2.: 8.2 o 2.5 so:

a: 2... . goods 8.: 8.2 n 8: x8.

. . 3.8.: m_._ 3.. v 05; x02.
3.8: as»
_= 8.8a: = 8.5: _ £5: = 8.523. _ 05.323. 8.5825 = 5228 _ 55.6 .88 a»... .23

 

 

«5386 as: a? 23 .833 25 8.3%; a. 3.3.5

 

.C.:oUv . _ _ 03¢.

.82. 28353....

£232.58... 095m .8 .528:— oEEam Saw... a 2 =2.
.28.» 953 8.2..."— .o._. 3.8%.... 2.3.5 22.3.25 0... .2. 8.8%... 3-39..— ..odwfia E... .modw: .c..ow.. 62“ +
22.3 .538... mun—hammeefiooa on». .260 m: E238... Dmhqmunofifioa 25 .38 no: .8. o... .8 3505:»... .8 .32 88.89

.33. 8.20m .88". .md. b02838. 63.83433 2... 3833:. .252.
.uoflmbfiquefié 68.83.33 .uo~_m-w:__aamnm ”35.—8:. o5 mwficonouc 22.3 mecca mac—83.3... 8.58.5 0....oEouuouoU.

 

 

 

 

J.
%

8... 9.... 3.. 3... a $558

3.." 3&2... . . 3.2.3 n .22 8583...

. 3... . .3 a... n .23 «saith

N: 85 . .3. a... c 883...“: 23...: .35.:

. .88 . than . e .5 .35.:

.2." 3... 9% u... 3.2 RS 0 x5 8..

. 3&9 . 5&3 5.5% . . 3.8.8 o .5 .8...

n: . 5n . . . . n .5 .8...

as . a... 8... . m3 3.... o a... .5.

. . on... 3... 9.... S... +3.8 0 a... .8.

9.2 . 33.—n . 3.3.3 :36.” «ganged m 0:5 :2...

. . . . 83$ 88 3.8% v 2.... .8.

luau. is. :8 dz

.= 8.52 .. 8.88.. . 8.5: = 8.52.8. . 95.528. 8....m5.5 .. 5.3.5 . .325 .36 95 .38

3.86 as... 25 23 3.3.8.. x5. 8.5%; a 9.3.;

 

6.58 ._ _ 03:.

. 1 . . ullnu :.|4. I11u41-Ink‘ I‘IIIIHIIHIIIIII Illufl‘nu‘IQIIllul '\lll‘ullflV.v\ ’i‘lh .‘ltz‘fi Li‘l‘ A”Lll~l\u~h'-
wcaflnanwhu SKELL «or—ha Urn-w. :FiHCTYJU HQ OLA. hU>CU .mbhaz A... UUttnhum tU~Q==mm CUUBaUC tutu: \fiUCDmuvtknv ”sf-2.4.0.5 k3 KKQ~C~=3VJ .N~ 3~Cs~h

37

 

 

 

mm.» 0000.; . . o .00 00.32

nu...— +hNSN +3.0... 0003.00 55.34. 86a 0 X00 00..
. o. .c . 3.0 cc... . cod 0 A00 0.005
0003.5 3.: cud- vmd N..—N o 0...... 00..

8.0— 0003.: . 008.Nn cc... +nNSN m 0:... 0.00..

. . +3.8 2..» SAN v 0...... .00..

3.80 03

. and mud . 0 0.00 000.2
00.. 3.0 mm. 5.0. 8.0 . e 0.00 00..
. 0002.3 8.: 000N510. . 309%? 0 0.00 0.005
0008.3 3.8 0003.: . 36 000360 e 05.. 00¢

. 3.: 0008.3 000M950 3.0 0003.8 c 05.. £00..

3..” on... . mnd vmd as... n 0:... .00..

. . 3&wa 3N. 000—0.: .0 0:... 300..

an al.»

moéu cm... . . 0 .10 0052

+3.3 8.: 0.9% 000N560 .. 0 £00 001

no. .. pm... .06.. . m0... 0 M00 0.005

008.3 3.0— 0008.Nv . 86 3.0— 0 05m 00..

. mad. 00Nv.Nn +8.: .0.m B... 0 0:5 x03

0008.3 0008.3 . 0008.3. 3.: «0003.3 n 0:... 0.00..

. . 000Nn.Nm NNd nniu v 0a... 0.00..
in 0......
=. 050.02 = 050.02 _ 00.0.02 = 050.02.00— . 05.0.0260. 00.2.3330 .. 3003.30 . 3003.30 .0000 0%... .050

 

 

x5 80503., 0 33...,
|
Av .30. 000. 0.00. 500.050 03.0.0. 0.

.0 0.300. 0... 9.0.3800 :8... 003.00 8.500.... 0.0:... 0.0 0.0 00:.0> 0030.5, 02.0.0»? 00:30.. 0:0 000.. .8 ACES“...
00.0.5.» 000.... 0%. 0:0. .00.»...000 a 0%. .0300 .00.... .3 00.....0... 00.0800 0003.2. 0.00. 55.050 03.0.0. .0 3.055% .N. 030...

3.?0 9.0.... 0%... 0:0... .8.wo.00m

38

.83 80653::

dozaoccmbm 095m .8 =2£8£ 0.953 Saw:— a 2 SE.

.28.» 203 828i .9: 3.833 332.5 Ems—5-20 on. .2. 8.8%.: 85.38 .detz. can .mcdw: .E.ow* .o~.ow +
22.3 5:605 Omhqmwgafioa on». .260 w: 25383 Oahu—mucofifioa 099 .260 no: 38 05 he 8:3:in we .30. 8859

A32 8E8
.35; .03 303638.. 6383433 was (5285 £08 .o~_m-o_oauc.m.v 22.3 .880 mac—8505 332nm ouahofioouoccy

6.88 .N. 033.

39

higher precision when stratified by ELTPG’s(T ables 10-12). Many of the differences in
sampling precision for jack pine were significant (£50.10). Also, dead and downed
vegetation variables for pole-sized, well-stocked red pine and pole-sized, medium-stocked
black oak were more precisely sampled by ELTPG (Table 12). Some overstory and
understory vegetation variables for these cover types also exhibited higher precision when
sampled by ELTPG’s(T ables lO-ll). Vegetation variables for most other cover types did
not exhibit gains in precision when sampled by ELTPG (Tables 10-12).

Most of the cover types found on ELTPG’swith greater moisture availability
(e.g., red oak. mixed oaks, and mixed upland hardwoods on ”lce;Moraines" and
Moraines"; Fig. 2) exhibited fewer gains in precision relative to cover types found on the
”Outwash” ELTPG’s(Tables 10-12). Sampling conducted on ”Outwash l" and "Outwash
ll" proved to be the most efficient for overstory variables relative to other ELTPG's (Table
l0). Conversely, gains in sampling efficiency for understory and dead and downed

vegetation variables exhibited no consistent trends across ELTPG‘sCI‘ ables ll-lZ).

DISCUSSION

The objective of stratification (in a statistical sense) is to minimize within block
variation such that between block variation is detectable. When choosing a vegetation
sampling protocol from more than one stratification scheme, one tries to select the
stratification that provides accurate estimates of sample parameters in the most efiicient
manner. A primary concern regarding the compilation of vegetative databases for
resource planning is the sampling effort required. Resource agencies do not have the
personnel or finances to collect data inefficiently, and with the need for ecosystem-based
data inventories increasing (Davis 1980), the need for efficient data collection is

necessary. Vegetation databases for the US. Forest Service have historically been stratified

40
by dominant commercial cover types (U.S. Forest Service 1988). Further stratification,
for example, by cover type and ELTPG,should provide more precise estimates of
vegetative variables, consistent with stratification theory. Often, the primary question is
whether the stratification scheme(s) are legitimate. An inappropriately designed or applied
stratification scheme may in fact increase within block variability.

The cover type-ELTPG stratification scheme generally followed stratification
theory. The probability of increasing sampling precision when obtaining a vegetation
sample stratified by cover type-ELTPG was almost always >SO% (as suggested by relative
efficiency tests), however, these gains were only significant (£50.10) for about 25% of
the comparisons. lnsignificance may have been due to low sample sizes,and indeed, the
required sample sizes (as indicated by Frees’s (1978) sample size formula)(Table 13) for
most of the vegetation variables were above the number collected.

The relatively consistent precision gains for cover type-ELTPG stratification
regardless of vegetative structure or ELTPG was encouraging. Increasingly complex
stratification is best suited (both in terms of sampling practicality and logistics) for
operational level forest planning, however, planning must also occur at the strategic and
tactical levels (Roloff and Haufler 1993). At these higher broader planning levels, systems
demonstrating consistency across landscapes are extremely valuable. The cover type-
ELTPG stratification appeared to provide reasonable gains in sampling precision throughout
the study areas. However, caution should be exercised when interpreting these results
because 1) the assumption is made that a representative sample of ”ecological units" were
obtained (Driscoll et al. 1984), 2) there may be large-scale variability in cover type-
ELTPG units (Bailey et al. 1978) not detected by the stratification scheme. and 3) the

personnel and financial constraints on this type of data collection are of concern.

41 -

Table 13. Means and standard errors (parenthesis) for number of replicates required
(30,0; accuracy 0.20 of the mean) according to Freese (1978) for vegetation
variables sampled by Forest Service cover type and ELTPG.

 

Forest Service Cover Type-

 

 

ELTPG Stratification
Vegetation Variable N‘ Replicates
Overstory
Basal Area 19 17(4)
Diameter 19 3(1)
Height 19 3(1)
Number Tree Species 19 26(6)
Tree Density 19 26(8)
Understory
Cover23m Tall 18 35(10)
Cover 25m Tall 18 45(13)
Deciduous Height 19 13 (3)
Stem Density 23 35 (8)
Soft Mast Cover 23 40 (7)
No. Soft Mast Species 23 15 (6)
Dad and Downed
Snag Density 16 48(11)
Snag Diameter 16 3 (1)
Log Density 17 81(18)

 

'N=number of replicates sampled to compute the minimum number of replicates
required.

42

For rapid, landscape assessments of some vegetative variables, the ELTPG
stratification may provide a ”quick and dirty” alternative. For most overstory variables,
ELTPG was a more efficient stratifier than the jack pine cover type classification,
particularly on the ”Outwash" ELTPG’s. Jack and red pine dominate these ELTPG types
across the study areas, and thus, samples from ”Outwash" ELTPG’swere relatively
homogenous with respect to cover type. In contrast, jack pine and red pine were sampled
on numerous ELTPG’s,including morainal systems. Morainal ELTPG‘s had greater
structural diversity potentials than ”Outwash” ELTPG’s,and thus, this diversity is included
in the cover type samples.

Understory variables exhibited no consistent trends in precision gain when
sampled only by ELTPG. Some ELTPG’s yielded more precise estimates of dead and
downed vegetative variables, particularly ”Outwashzlce" and "lcezMoraine l" for snag
density and snag diameter (Table 8).

A11 stratification schemes used in this study were complex. The cover type-
ELTPG stratification had 6 components, whereas the cover type and ELTPG schemes each
had 3 major components. The cover type scheme was based on dominant, commercial
tree species; stocking density, and age. The ELTPG’swere delineated based on landform.
soils, and potential vegetation (climate was viewed as equally applicable to both
stratification schemes). The cover type-ELTPG scheme was a combination of cover type
and ELTPG. The major difference (besides cover type being commercially driven as
opposed to the ecosystem-based ELTPG’s) was the spatial extent of each scheme. Cover
type-ELTPG sample units were delineated at small spatial scales, cover types were
delineated at intermediate scales, and ELTPG’swere delineated at large scales. For

example, 148 cover type-ELTPG units, 70 different cover types, and 14 ELTPG‘s were

43

identified on the study sites. To measure overstory basal area with 90% confidence and
20% allowable error according to Freese’s ( 1978) sample size formula, 8 replicates from
each cover type-ELTPG, 15 replicates from each cover type, and 15 replicates per ELTPG
would need to be sampled. Total number of replicates required for each stratification
scheme would be 1,184 for cover type-ELTPG, 1,050 for cover type, and 210 for
ELTPG. At first glance, one would recommend sampling basal area by ELTPG,however,
one must also consider the stratification scheme that provides the ”best" estimate of the
variable. If less variation around the mean equates to a ”better” estimate, then the ELTPG
stratification is inferior to the cover type—ELTPG strata (as suggested by the relative
efficiency tests). The issue now becomes management oriented. 1f gross, landscape
descriptions of basal area are acceptable, then ELTPG should by considered as a
stratification scheme. In contrast, for situations requiring precise, defensible estimates
(e.g., legal mitigation), the cover type-ELTPG stratification is recommended.

Relative efficiency tests were used to evaluate trends in sample precision as they
related to cover type, ELTPG,and vegetative variables. Spatial scale appeared to be an
important consideration in delineating these relationships. For example, at smaller scales
(cover type-ELTPG units), sampling precision among ELTPG’swas generally consistent,
with all ELTPG‘sdemonstrating some reductions in sample variance. At larger spatial
scales, (e.g. ELTPG's), vegetation variables for cover types naturally associated with
extremes on the ELTPG scale (e.g., jack pine naturally occurs on Outwash 1; Fig. 2) were
more efficiently sampled when stratified by ELTPG as opposed to cover type. Conversely,
vegetation variables associated with cover types that grow on more fertile ELTPG's(e.g.,
morainal systems) exhibited higher sampling precision when stratified by cover type.

These observations may be related to the inherent variability of the ECS units.

44
At small spatial scales, the sample is stratified such that within block variation is minimized.
At larger scales, within-stand heterogeneity causes increases in sampling error. More fertile
ELTPG’s have a greater potential for complexity (Spies and Barnes 1985). Whereas
"Outwash” systems could only support species adapted to well-drained soils, "Moraines"
could support most of the ”Outwash" species in addition to species requiring more
moisture. The morainal systems sampled for this study exhibited more diverse vegetative
communities than outwash systems (Fig. 2).

The relatively poor ability of ”Morainal" ELTPG’sto increase sampling
precision at larger spatial scales may also be attributed to performing the vegetative
analysis at too coarse of scale on the ECS hierarchy. Since morainal systems are more
vegetatively heterogenous, grouping similar ecological land type phases (Cleland et al.
1993) may have obscured the discriminatory capabilities of the system. However, when
one examines ELTPG‘sbased on only a single ecological land type phase (e.g., ”Outwash
1", "Ice; Moraine l”, "Moraine ll”), no consistent gains in sampling precision were evident.

lntegrating an ecological classification system into a vegetative sampling strategy
is appealing in that the delineated boundaries have greater ecological significance (as
opposed to cover type maps). An ecological classification system permits managers to
more accurately predict successional pathways and the effects of their management
activities on vegetation structure. Also, a knowledge of the ecological relationships
provides insight to natural disturbance regimes at a large spatial scale.

An explanation for the variability in sampling precision from the cover type-
ELTPG stratification scheme may be that silviculturally induced disruptions of the natural
ecological processes have created a vegetative structure not conducive to ecological

quantification. The ages of vegetative cover types varied. but most stands on the study

45
sites recently (<70 years) experienced some sort of silvicultural prescription or
disturbance. The variable results from the relative efficiency analyses may be
representative of: 1) some past silvicultural prescriptions being more ecologically based
than others, 2) the plasticity of terrestrial plants; relying on several individual and
population-level characteristics that buffer them from environmental unpredictability
(Bazzaz and Sipe 1987), or 3) different cover type-ELTPG units exhibiting temporal
variation in returning to a pre-disturbance state (O’Neill et a1. 1986, Swanson et a1.
1988). This would partially explain gains in sampling precision for some understory
vegetative variables (e.g., the soft mast attributes), especially at larger spatial scales (i.e.
the ELTPG scheme). ln theory, understory and ground components should exhibit greater
resiliency to disturbance than overstory species because of shorter turn-around times
(Cleland et al. 1993), however, inconsistent gains in sampling precision for most
understory variables may suggest that micro-site variation in physiography and soils not -
detected by the ECS further influences forest composition and understory recruitment
(Host et al. 1987, McNab 1991). Also, certain understory units may be associated with
specific overstory units (McNab 1991) rather than ELTPG‘s.

Vegetative attributes exhibited varying degrees of sample precision dependent
on stratification scheme. Jones (1991) found that vegetation patterns within pine types in
Georgia were not a function of environmental conditions (i.e., an ECS) but rather
conditions of stand establishment or subsequent anthropogenic influences. These factors
are also important in northern Michigan and undoubtedly explain some of the vegetative

sampling variability exhibited in this study.

46
SUMMARY AND CONCLUSIONS

The relationships between the ECS and vegetation (and associated processes) is
almost always through spatial patterns and structures (Bailey et al. 1978). Although
processes are controlled by structure, processes emerge only at the integration of
composition, structures, and associated interactions (Bailey et al. 1978). Each level of the
ECS is defined by an integration of the dominant ecological factors affecting biological
systems at that scale, thus, spatial concerns and large-scale interactions are incorporated
into the ECS(C1eland et al. 1993). Also, the ECS fulfills legal and policy requirements
for coordinating and integrating resource inventories on National Forest lands (Cleland et
al. 1993). The utility of incorporating ECS into vegetative sampling is warranted, 9
however, as demonstrated, there is seldom an individual classification scheme that
addresses all of the needs and concerns of resource managers. For overstory vegetative
descriptions, the Forest Service cover type is an effective tool, however, it is important to
remember that the Forest Service classification provides measures of existing overstory
conditions, whereas the ECS identifies potential natural vegetation (Russell and Jordan
1991, Cleland et al. 1993). Also, contrary to some cover types, ECS‘s are expected to

respond to management alternatives similarly (Driscoll et a1. 1984).

Chapter 2: Wildlife Habitat Model Validation

Habitat has been theoretically defined as the location supporting wildlife
including space, food, cover, and other animals (Giles 1978); often characterized by
vegetation, landform, and hydrology (Odum 1971). Resource management influences the
physical features of habitat, the effects of which are often paralleled by differences in
animal abundance (Farmer et al. 1982). Resource managers are responsible for wildlife '
habitat and must justify and account for their management activities, however, they seldom
know what kinds, amounts, and distributions of habitats are necessary to maintain
population viability for all species (as required by law; Hurley et al. 1982). With top-level
management, legislators, and fiscal officers wanting plans, budgets, and project proposals
in quantitative terms (as opposed to documents created by intuition or expertise) (Davis
1980, Jarvinen 1985, Maurer 1986, Flather et al. 1989), the need for rapid and precise
estimates of habitat quality has escalated. In response, resource management agencies have
turned to modeling wildlife-habitat relationships.

Numerous studies have identified specific vegetation characteristics important
for determining wildlife abundances and distributions. Avian studies dominate the
literature, with most showing generalized relationships such as avian species richness,
avian community structure, or bird species abundance to habitat heterogeneity, vegetation
abundance, foliage height diversity, and/or habitat complexity (MacArthur and MacArthur

1961, Brewer 1963, Karr 1968, Karr and Roth 1971, Willson 1974, Pearson 1975,

47

48 .

Franzreb and Ohmart 1978, Lines and Perry 1978, Moss 1978, Asherin et al. 1979, Noon
et al. 1980, Ambuel and Temple 1982, Dobkin and Wilcox 1986, Freemark and Merriam
1986, Freemark 1988, Johnson and Brown 1990, O’Brien 1990). Others have documented
relationships between bird species and specific vegetation structures (Jackson 1970, James
1971, Holmes and Robbins 1981, James and Warner 1982, Balda et al. 1983, Grue et a1.
1983, Morrison 1983, Bock and Webb 1984, Swift et al. 1984, Rotenberry 1985, Maurer
1986, Askins and Philbrick 1987, Askins et al. 1987). Research on other wildlife species
is more limited. Studies have correlated (with varying degrees of specificity) vegetative
structures and small mammal abundances (Rosenzweig and Winakur 1969, Brown 1973,
Miller and Getz 1977, August 1983). Pianka (1967) and Marcot et al. (1983) 1
demonstrated similar relationships for lizards.

Since the beginning of the 20th century, theories regarding the relationships
between wildlife populations and their habitats have abounded (see review in Morrison et
al. 1992; Chapter 1), however, the first effective efforts to organize and integrate available
information into resource planning did not occur until the middle 1970’s (Thomas 1986).
Based on numerous studies and theories supporting the concept of wildlife-habitat
relationships, quantitative models that describe these phenomenon were developed. An
appealing characteristic of habitat models is the relative ease of inventorying habitats as
opposed to wildlife populations. Also, habitat models establish a link between wildlife
populations and the land altered by management practices (Flather et al. 1989). Habitat
models are also more applicable (as opposed to population-based models) to regional and
national analyses (Cooperrider 1986).

Numerous habitat models exist (see Verner et al. 1986, review in Morrison et

al. 1992; Chapter 6), all with several inherent assumptions; that 1) animal distribution and

49

abundance can be explained and predicted by environmental attributes (Marcot et al.
1983), 2) similar patterns or configurations of habitat will reflect similar patterns of animal
abundance (Flather and Hoekstra 1985), and 3) animal abundance is positively associated
with habitat quality (Van Home 1983, Best and Stauffer 1986). These assumptions make
the task of model validation difficult, and indeed, validation and robustness studies have
seldom been conducted (Lancia et a1. 1982).

In wildlife—habitat relationships modeling, the assumption that species density is
a direct measure of habitat quality has seldom been questioned (Van Home 1983). Van
Home (1983) summarized the potential problems with this assumption. The most accurate
estimates of wildlife density and habitat quality relationships are dependent on an 1
understanding of population demographics and the factors influencing survival and
reproduction (Maguire 1973, Van Home 1983). However, until databases link population
dynamics to specific habitat attributes and configurations across landscapes, the most
feasible means of estimating wildlife potentials across a landscape is through habitat
measurements.

The Habitat Evaluation Procedures were developed by the US Fish and
Wildlife Service (1981) to provide a habitat-based approach to assessing environmental
impacts of resource development projects (Schamberger and Krohn 1982, Cole and Smith
1983). Habitat suitability indices (HSl's) are integral components of the Habitat
Evaluation Procedures and are best viewed as hypotheses of species-habitat relationships
(Schamberger and Krohn 1982). The value of HSl’s lies in documenting a repeatable
assessment procedure and providing an index to particular environmental characteristics
that can be compared with alternative management plans (Schamberger and Krohn 1982,

Morrison et al. 1992). HSl's are based on the premise that habitat suitability can be linked

50
to habitat attributes with linear relationships (Morrison et a1. 1992). Subsequently, these
attributes are mathematically combined into an index to habitat quality (Blake and Karr
1984, Laymon and Barrett 1986).

The development and implementation of H81 models has been flawed due to
the lack of field validation (Cole and Smith 1983). Critics of H81 models argue that
management decisions should not be based on untested models (Lancia et a1. 1982, Cole
and Smith 1983, Marcot et al. 1983, Laymon and Barrett 1986). Validation is the act of
increasing, to an acceptable level, the confidence that an inference about a simulated
process is correct for the actual process (Van Horn 1971). Using a variety of validation
techniques, some researchers have found positive correlations between model predictions 1
and measures of wildlife distributions and abundance (Lancia et al. 1982, Cole and Smith
1983, Cook and lrwin 1985, Dedon et al. 1986, Hammill and Moran 1986, Latka and
Yanhke 1986, Laymon and Barrett 1986, Laymon and Reid 1986, Raphael and Marcot
1986, Stauffer and Best 1986), whereas others have found negative or no correlations
(Seitz et al. 1982, Clark and Lewis 1983, Bart et al. 1984, Johnson and Temple 1986,
Lancia et al. 1986, Larson and Bock 1986, Seng 1991, Robel et al. 1993). A variety of
explanations for these discrepancies have been offered including inadequate or
unrepresentative sampling (Cole and Smith 1983), model equations not representative of
the actual wildlife-habitat relationships (Farmer et al. 1982, Cole and Smith 1983,
Cooperrider and Behrend 1983, Warwick and Cade 1988, Van Horne and Wiens 1991),
misinterpretation of results (Capen et al. 1986, Brower and Zar 1984), or application of
the model to inappropriate spatial scales (W iens 1986).

Successful habitat models should be valid, general enough that a single model

can apply to a wide range of situations without major modifications, and usable by land

51

managers (Van Horne and Wiens 1991). In terms of U.S. Forest Service assessment
goals, a synthesis of ecological theory into a modeling framework is necessary to fulfill
legislative mandates. The goal of this chapter was to validate several HSI models for
Forest Service management indicator species and to evaluate their usefulness in land use
planning. Specific objectives are:
l. to determine if habitat quality as predicted from HSl models is correlated
with measures of animal abundance, and

2. to evaluate the appropriate spatial scale of model applicability.

METHODS

Twelve wildlife management indicator species were selected to represent a
broad range of habitat requirements. Most of these species were included in the Land
Resources Management Plan for the Huron—Manistee National Forest and have been
designated as species indicative of the health, status, and quality of habitat management on
the forest. The management indicator species included gray squirrel (Sciurus carolinensis),
eastern meadowlark (Sturnella magna). pileated woodpecker (Dryocopus pileatus), wood
duck (At'x sponsa), black-throated green warbler (Deudroica virens), ruffed grouse (Bonasa
umbellus), eastern bluebird (Sialia sialis), upland sandpiper (Barrramia longicauda), white-
tailed deer (Odocoileus virginianus), black bear (Ursus americana), American redstart
(Setophagus ruticilla), and ovenbird (Seiurus aurocapillus).

Habitat suitability models existed for all of the aforementioned species.
Models developed by researchers affiliated with the U.S. Fish and Wildlife Service
included those for the gray squirrel (Allen 1987), eastern meadowlark (Schroeder and

Sousa 1982), wood duck (Sousa and Farmer 1983), pileated woodpecker (Schroeder 1983),

52
and black bear (Rogers and Allen 1987). The model for ruffed grouse was developed
specifically for Michigan by Hammill and Moran (1986). A model for ovenbird was
derived from Capen et al. (1986) (Appendix B). The remaining models were developed at
Michigan State University through literature reviews and consultation with experts and
followed the protocol of the U.S. Fish and Wildlife Service (1981). These models
included black-throated green warbler (Bender et al. 1989; Appendix C), white-tailed deer
(Bender and Haufler 1990; Appendix D), eastern bluebird (Moses et al. 1990; Appendix
E), upland sandpiper (Roloff and Haufler 1990; Appendix F), and American redstart
(Minnis and Haufler 1991; Appendix G).

Vegetation data used to run HSl models were collected from a stratified
random sampling design using a variety of plot, line intercept, and ocular estimation
techniques (Table 14). Sampling was stratified according to Corporate Database cover
type (U.S. Forest Service 1988) and underlying ecological land type (Cleland et al. 1993)..
Replicates were the cover type-ecological land type stands, and each replicate contained a
minimum of 3 samples. The wildlife habitat database‘ consisted of 56 measures of
vegetation structure and composition derived from overstory, understory, dead and
downed, ground, and wetlands categories. Arithmetic means of vegetation attributes for
each cover type-ecological land type (Appendix A) combination were used for habitat
model input.

To facilitate the computation of HSI values, Corporate Database cover type and

ecological land type maps were digitized and habitat models were programmed into an

‘Detailed documentation of the wildlife habitat database is provided in Roloff and Haufler
(1994), U.S. Forest Service Huron-Manistee National Forest, Corporate Database and
Ecological Classification System: Vegetation Databases and Documentation, Unpublished
Report. Cadillac, M1.

53

Table 14. Vegetation variables and associated sampling techniques. Methodology for
each technique is described by Hays et a1. 1981.

 

Vegetation Variable Sampling Technique

Overstory‘

Basal Area (total and by species) 10 X 25 m (0.06 ac) Plot

No. Tree Species 10 X 25 m (0.06 ac) Plot

No. Trees 10 X 25 m (0.06 ac) Plot

Height Haga Altimeter

No. Cavities 10 X 25 m (0.06 ac) Plot

Coverage 25 rn (82 ft) Line Intercept

Diameter Diameter Tape
Understory

Height Lowest Live Conifer Branch Meter Stick

Coverage >5 m Tall 10 X 25 m (0.06 ac) Plot

Coverage >3 m Tall 10 X 25 m (0.06 ac) Plot

Coverage <3 m Tall 10 X 25 m (0.06 ac) Plot

No. Stems 2 X 25 m (0.01 ac) Plot

Coverage Soft Mast 25 m (82 ft) Line Intercept

No. Soft Mast Species 25 m (82 ft) Line Intercept

Dead and Downed

Snag Density (Total and by species) 10 X 100 m (0.25 ac) Plot
Snag DBH Diameter tape
No. Cavities 10 X 100 m (0.25 ac) Plot
No. Logs 10 X 100 m (0.25 ac) Plot
No. Stumps 10 X 100 m (0.25 ac) Plot
Ground
Distance to Passerine Perch Site Pacing
Coverage 10 m (33 ft) Line Intercept
Coverage of Grass 10 m (33 ft) Line Intercept
Height of Spring Growth Meter Stick
Wetlands

Coverage of Water by Emergents/Debris Ocular Estimate
Snag Availability Ocular Estimate

 

f'tfiersiorpzill trees 211.4 cm (45"? diameter breast height (55H); Understory=All
woody stems < 11.4 cm (4.5") DBH; Dead and Downed=Snags 211.4 cm (4.5”) DBH,
Logs 218 cm (7.1”) DBH, Stumps 230 cm (11.8”) tall and 11.4 cm (4.5”) DBH;
Ground=All species associated with the forest floor (typically s 1m tall).

54

Arc/Info geographic information system (GIS)‘. Both sets of maps were obtained from U.S.
Forest Service field offices. Base map coverages (e. g., roads, lakes, rivers) were obtained
from Michigan State University, Center for Remote Sensing. Cover type and ecological land
type maps were verified at each sample point, and discrepancies were updated. Habitat
models were programmed using the Arc/Info macro language.

Wildlife population indices were collected in 1991 and 1992. Songbird surveys
(Robbins 1981a,b, Skirvin 1981, Askins and Philbrick 1987, O’Brien 1990), ruffed grouse
drumming counts (Petraborg et al. 1953, Davis and Winstead 1980, Robbins 1981a,b,
Rodgers 1981, O‘Brien 1990), pileated woodpecker surveys, deer pellet counts (Neff 1968.
Collins 1981, Rowland et al. 1984), and squirrel time-area counts (Bouffard and Hein 1978)
were conducted using the point-count method (Ryder 1986; p. 304). All point surveys were
Standardized as to weather, time, and observer to minimize sampling bias (Ralph and Scott
1981. pp. 251-325).

North-south transects (1.6 km long) were delineated across township sections.
Each township section contained 3 transects spaced a minimum of 0.4 km apart. Eight points
were systematically located 2150 m apart along the transects (Freeme and Merriam 1986,
O’Brien 1990). Sample points from 1991 and 1992 were independently located.

Two different approaches were used to validate HSI’s, depending on the home
range of the wildlife species (Table 15). For species having home ranges s 10 ha, a home-

range-sized circle was centered over each sample point and an HSI was computed.

‘Detailed documentation of the modeling process is provided in Roloff and Haufler. (1994)
U.S. Forest Service Huron-Manistee National Forest, Wildlife Habitat Management
Procedure: Documentation, Unpublished Report. Cadillac, MI.

55

Table 15. Wildlife species, home range estimates, and supporting research.

 

 

Home Range Estimate

 

Wildlife Species (ha) Citation

American Redstart 1.5 Sturm 1945, Ficken 1962,
Samson and Lewis 1979,
Sherry 1979

Ovenbird 1.5

Eastern Bluebird 1.5 Pinkowski (1976)

Black-throated Green Warbler 1.5 Kendeigh (1945), Morse
(1976), Sabo (1980)

Upland Sandpiper 10‘ Bowen (1976), Ailes (1980)

Gray Squirrel 0.4 Nowak (1991)

Pileated Woodpecker 70 Renken (1988)

White-tailed Deer 173 Larson et al. (1978)

Ruffed Grouse 16 Cade and Sousa (1985)

Black Bear 10,000 Norton (1981), Manville

 

 

 

(I983), Garshelis (1986)

 

‘lncludes only the area required for nesting. The model calculations accounted for the amount
of contiguous habitat surrounding the nest site.

56
The HSI score was subsequently compared to the number of individuals censused at the point.
For species requiring > 10 ha, circles spatially corresponding to the home range were
distributed across study sites (regardless of sample point location) and HSl’s were computed
for each circle. The mean number of individuals censused per point located within each circle
was compared to the corresponding HSI score. HSI values were computed using the area-
weighting procedures as described by the U.S. Fish and Wildlife Service (1981).

Home-range circles were excluded from the analysis if vegetation data were
unavailable for <75% of the area. Although most cover type-ecological land type
combinations were inventoried, several combinations were not found in sufficient numbers or
area to warrant sampling. Also, the GIS overlay procedure develops "sliver polygons” that.
may not truly exist. It was determined that excluding areas without data rather than counting
them as unsuitable habitat provided a better representation of habitat quality.

Numerous researchers have cautioned against using singing bird surveys to index
bird populations (Willson 1974, Bart and Schoultz 1984. Sieg 1991), but the technique
remains the most practical for landscape assessments. Songbird surveys were conducted 3
times annually at all sample points from sunrise to 1000 (Skirvin I981, Askins and Philbrick
1987), 15 May to 15 June. Only bird species selected for this analysis were recorded at each
point to minimize the "saturation effect” (Bart and Schoultz 1984). Individual observers
would remain at each point for 5 minutes tallying all unique individuals seen or heard.

Point surveys for ruffed grouse and pileated woodpeckers were conducted from
sunrise to 1000, 25 April to 5 May, annually. Grouse and woodpecker surveys were
conducted at alternating points along the transects to minimize duplicate counts (Rodgers
1981). Observers remained at each point 5 minutes, recording all unique grouse and pileated

woodpeckers seen or heard. Random sightings of grouse were also recorded as to location on

57

field maps.

Deer pellet group counts were annually conducted prior to leaf-out in a circular 40
rn2 plot at each wildlife sampling point. Pellet groups were defined as consisting of 25
pellets (Rowland et al. 1984), and pellet groups lying on the edge of a plot were
systematically included or excluded in the tally.

Population indices of gray squirrels proved difficult to obtain. Several survey
techniques were used including a modified time-area count method (Bouffard and Hein 1978)
in 1991. The original method involved estimating distances to observed squirrels, however,
the modified method only documented the number of squirrels seen. Squirrels were censused
once at alternating sample points for 20 minutes from sunrise up to 2 hours later, 7 June to 7
July. The time area count yielded poor results, thus, random sightings and live-trapping were
used to index squirrel populations in 1992. Live traps were systematically placed in specific
cover type-ecological land type combinations and monitored daily for 6 trap nights. All
captured gray squirrels were marked with a temporary dye and released. Random sightings of
gray squirrels were recorded as to location on field maps. The total number of squirrels from
all census techniques per cover type-ecological land type combination was the population
index.

Wood duck population indices were obtained through brood counts (Lokemoen
1982) on 29 wetlands throughout the "Hoist Lakes" area. Counts were conducted from 0530
to 0900 hours, 10 to 17 July and 8 to 15 August, corresponding to the periods during which
wood duck broods are most visible (Bellrose 1976). Counts occurred from predetermined
points (maximizing wetland visibility) along wetland edges for 10-minute periods. The sum
of July and August counts was the population index for each wetland.

Association between HSI values and population indices were tested using

58

Spearman’s Rank Correlation Coefficient (Siegel 1956, O’Brien 1990). The Spearman’s test
provides the statistic Rho as a measure of association between 2 variables and a probability of
significance can also be computed (Siegel 1956). Correlations were performed for individual
years and for years pooled. To help discern chance significance, Spearman’s tests were
performed on randomly generated data for 4 different sample sizes; 20, 100, 300, and 700.
Randomly generated HSI values were between 0.00 and 1.00, and random population indices
ranged from 0-5 (typical of any one sample point). Random numbers were drawn from a
uniform data distribution.

To determine the proper spatial scale for habitat model applicability, HSI values
were calculated at several spatial scales (ranging from 0.1 to 182.4 ha) for wildlife models. 1
Spatial scales deemed biologically inappropriate because of their small size (e. g. deer habitat
use typically would not be assessed at small (0.1-7.3 ha) spatial scales) were excluded from
the analyses (Laymon and Reid 1986). The results were plotted on histograms with
"percentage of total area evaluated” on the ordinate, ”HSI values" (grouped by 0.105) on the
abscissa, and "spatial scale" on the z-axis. The spatial scale at which the most habitat quality
categories were delineated (i.e. the discriminatory capability of the model was maximized)
was deemed the proper spatial scale for model applicability. This analysis applied only to
final model output and assumed that vegetative structures (i.e. model components) were

consistently perceived and responded to by wildlife species regardless of spatial scale.

59

RESULTS

Population indices were obtained for 9 of the 12 management indicator species.
No eastern meadowlarks and too few eastern bluebirds were censused to warrant statistical
analysis. Population indices for black bears were not estimated due to the lack of practical
field techniques. Only one year of suitable data was obtained for some species including
pileated woodpecker, black-throated green warbler, and wood duck (Table 16). Population
indices per census point ranged from 0-6 for songbirds, 0-3 for pileated woodpeckers, 0-15
for deer, and 0-4 for ruffed grouse. Two wood duck population indices were obtained,
ranging from 0-2 for number of broods and from 0-9 for ducklings per wetland. The number
of squirrels censused in cover type-ecological land type combinations ranged from 0-11. -

HSI values ranged from 0.00-1.00 for the American redstart, ovenbird, deer,
squirrel, and wood duck models. HSl‘s ranged from 0.00—0.75 for upland sandpiper and
ruffed grouse and between 0.00-0.60 for pileated woodpecker and black-throated green
warbler. HSI values were calculated at spatial scales corresponding to the home range
requirements of each species (Table 17).

Spearman correlation tests for 1991 and 1992 produced variable results. The most
consistent (across years) performing model was the American redstart (Table 16). All other
songbird models exhibited either non-significant correlations or inverse correlation coefficients
(T able 16). Correlation coefficients for combined years were all significant (_PSO.10)(Table
16).

Although the deer models appeared to provide significant positive correlations

with population indices (Table 16), some of the significance tests may have been due to

 

63.3.5: 295 226365 Be: bogs—om.
458535 .:2M
.3826: 8:: :85: Eat—tamer

.38 o :58 u a sag: : 2: 5-2.: H m .256. 8: 5:225: :22 =25:on H St.

 

 

 

 

 

an .2. 3.. .: . . . $528: 2 83> a:
a .2. 8m: . . . .88: 9 33> a:
3.2.: :83
a 2.: an: :N a... 8:: : m8: 3:: 388-2: :8". mean
a 2.: 3:: :N 2.: 8n: : .2. 8:.9 386:3 88: :38:
a 8.: an: :N no: 3:: : :_ .: 55 385-2: :8". 8.53::
ea: 8.5-223
NS 38.: En: 385-3: 883285.086
NS 8:: 43.: 388-3: :8". .253
.953: :6
on :8: 5.: :2 .2. :8: :o :_ .: x: .: 386 8:2:
.8 .2. :2 .: . . . .0852: 882.2%:
.8 :_ .: m: .: . . . 385:8 .3838".
58:83 8.8::
.3 .2. 824 . . . 83:5... :83:
as .2 :8? . . .. 5:33 :86 8.8.582:
N: 88.: 22.: 3... 88.: Rm: .3” .2. 9:: 3:86
8: 88.: «R: an 88.: 2:: SN :8: 8.: use! 56:05
a m at a u at a m .2: 38:: 0:253
an; 83:95: «:2 a:
:3

.Gmfl .095: o 3.3:: .5658: 5:22.50 2: : :38. :5 3:23-28 83 3:239: .8035 coca—sue: “58:3
:5 8:65 5:323: .332 e835: 8:8 .8 :5... £953 :5 5:328: 838%: £23qu8 3:228: :5: 553on .2 :33.

61

Table 17. Spatial scale(s) of wildlife model applicability. Estimates are based on
subjective evaluations of histograms depicting the range of H81 values and their
evenness across the landscape for different spatial scales.

 

Scale of Applicability Home Range

 

Wildlife Model (ha) (ha)
American redstart 1.5 1.5
Ovenbird 1.5-7.3 1.5
Black-throated Green Warbler 1.5 1.5
Upland Sandpiper .' 10
Pileated Woodpecker

Food/Cover Sub-model 16.4 70

Reproduction Sub-model 16.4 70
Ruffed Grouse 16.4-29.2 16
Gray Squirrel

Winter Food Sub—model 0.1-29.2 0.4

Cover/Reproduction Sub-model 0.1-29.2 0.4
White-tailed Deer

Fall-winter Food Sub-model 16.4 173

Security Cover Sub-model 16.4 173

Spring Food Sub-model . 173

 

‘Scale of applicability could not be determined from histogram.

62 -

chance (T able 18). However, all of the significant (£50.05) deer correlation coefficients
were positive (Table 16), unlike the random analysis (Table 18). Also, all of the deer
models combined across years provided significant (1350.10), positive correlations.

The gray squirrel and ruffed grouse models for combined years exhibited
positive, higher probabilities of significance (BS0.025)(T able 16) than random analyses on
similar sample sizes (Table 18). The correlation coefficients for pileated woodpecker and
wood duck models were also positive, however, the association could not be discerned from
chance (Table 16). The black-throated green warbler and upland sandpiper models yielded
negative, non-significant correlation coefficients (Table 16).

The appropriate spatial scale for most habitat models corresponded to the
species home range (Table 17). For example, the ovenbird model discriminated habitats
equally (as indicated by the number of habitat classes) across a broad range of spatial scales
(0.3-29.2 ha)(Fig. 5). The area of model applicability, however, was deemed as 1.5-7.3 ha
(Table 17) because of the uneven distribution of area in each 1181 class for 0.3, 16.4, and
29.2 spatial scales (Fig. 5). The effects of area weighting on the ovenbird HSI model are
evident at scales larger than 70.0 ha as depicted by the convergence of H81 scores (Fig. 5).

Generally, the ability to discriminate habitats of different quality decreased as
spatial scale increased (Figs. 5-14). These models typically converged into 2 or 3 H81
classes at large spatial scales (Figs. 5-14), a property related to the area-weighting
averaging procedure used in HSI calculations (U.S. Fish and Wildlife Service 1981).
Habitat models for wildlife species for which habitat assessments were based on the
presence or absence of a specific cover type (e. g. openings) exhibited nominal output (i.e.

the model only discriminated ”good” and ”bad” habitats). Models for upland sandpiper

63

Table 18. Spearman rank correlation coefficients, associated probability, and sample
size for randomly generated habitat suitability indices (1181’s) and wildlife
population indices. Probability was one-tailed and tested if the correlation
coefficient differed from 0 by chance (Siegel 1956).

 

 

Simulation 1 Simulation 2 Simulation 3 Simulation 4
(n = 20) (n = 100) (n = 300) (n = 700)
rho‘ P. rho 13 rho E rho g

 

0.134 n.s.2 -0.054 n.s. 0.011 n.s. 0.023 n.s.
0.347 0.05 0.042 n.s. 0.010 n.s. -0.096 n.s.
0.087 n.s. 0.013 n.s. 0.003 n.s. ~0.020 n.s.
-0.029 n.s. 0.111 n.s. -0.018 n.s. -0.019 n.s.
-0.383 0.05 0.023 n.s. 0.041 n.s. 0.035 n.s.
-0.268 n.s. 0.085 n.s. 0.008 n.s. -0.005 n.s.
0.035 n.s. -0.129 0.10 —0.014 n.s. 0.046 n.s.
-0.412 0.05 -0.088 n.s. 0.104 n.s. 0.008 n.s.

0.237 n.s. -0.044 n.s. -0.004 n.s. 0.012 n.s.

 

‘rho=Spearman rank correlation coefficient, £=one-tailed probability.

2Not significant.

.2355 :8 59:: 3:2: 5: :0 2:8 3:3: 3 853:3 of. .m :59”:

$30 51
ad ad No 0.0 md v.0 md Nd Pd o

patenlena 881V letol lo afieiuaoiad

 

25:96

65

 

._u.:_=cm raw 8.. 59.5 .32: .m: :0 28m 328... 3 3:255 2:.

$90 51
md No 06 md v.0 md N6 _..0 o

........
.....................

.............................

...................

...................

...................

...........................................................
..................................................
........................
......................

 

..............
..............

......
......
......

 

 

 

 

 

 

 

 

..............

..............

..............
........

.....

......

................................

......................

.....................
..............

......

......

..........

...........................

.......

......

......

......

......

......

........

.................................

..........................
...............................

,.. .amxxu»... ........

..............................
...............

...........

...........................

 

 

:ozoauoammtgoo .. .mEaam >20

 

.o :53:

md

md

petemeng 891V late; )0 afieiuamad

.88.. 3.53:3: .8.. 00:28.33 .8 39:: 3:2: 5: a: 28... 3.8:. :o 8:25.: 0:... 6 2:9“:

$90 51
0.0 0.0 5.0 0.0 0.0 V0 0.0 N0 _..0 0

o
no

0%. N: W.
reed... m.
m
to w.
m
:.: m
3
an
m: m
W

F

 

wuoom :95; :0“. .. .500 00:30:23

 

.560 5:53. .80 3:96.23 .50 39:0 .002: a: .8 23m 38% 0o 8:305 of. .w 2:»:—

mmflo .9...
0.0 0.0 5.0 0.0 0.0 V0 0.0 N0 '0 0

‘0. V. (‘4
0 0 0

w
0'
pamnle/xa 231v mo; 10 afiewaomd

 

.550 £58m .mmo 3.30-223

.2882 58:25. ..8 39:0 ESE .0: :0 28m 3:2: .8 85:05 2:. .0 3:3..—

mmma .01
0.0 «.0 0.0 N0 ...0

............................................................
.............................
...............

 

 

...............
...............
................

.................

.................

................

................
.....u .... ”....Hvuwmm . h.” H.“ ..mf ,. . . . ..............................

 

 

 

 

 

 

paxemena 991v .9101 ;o afiezuemad

 

 

 

 

:Emumm cwotoE<

 

69

00.0.03 50% 028050.003 :8 59:0 .002: .0... :o 0.000 .0080 ..0 00:0: 0:. 0:...

000.0 .0...
0.0 0.0 0.0 0.0 P0 0

 

 

 

 

 

 

 

 

 

 

 

 

 

 

00.0.0.5 :005 00.09.50.005

 

.o. 050.”.

V. ' N
O O
pammeAa 991v law; ;0 afieauemad

‘9.
0

°°.
0

 

.0380 005.2 5.. 59:0 .009: .0: :o 0.000 .0080 .8 00:80:. 0:... ... 050.“.

000.0 ~02
0.0 0.0 5.0 0.0 0.0 0.0 0.0 0.0 F0 0

panama/q 991V [2101, J0 afimuaomd

 

00:20 00.3".

71

.:o..0:0o:.0: 0:0 ..0>00 00.02.0003 00.00.... .00 59:0 .008. .0: :0 0.000 .0030 ..0 85:05 0: ... .m. 050.”.

000.0 .0...

0.0 0.0 0.0 0.0 0.0 0.0 0.0 0

        

  

 
 
 

 

 

 

 

 

panama/g 261V mo; ;0 afiezuamad

 

 

 

 

:000:0200m:0>oo .. 0000000025 00002.1

.0000 00.02.0003 00.00:: :00 0:900 .009: .0: :0 0.000 .0080 00 00:30:. 0: ... .0. 0:00....

000.0 .0...
0.0 0.0 5.0 0.0 0.0 0.0 0.0 N0 ...0 0

 

 

 

 

 

 

 

panama/g eaJv IBIOJ. ;o afiezuemad

 

 

 

 

 

000”. .. 09.000000; 00:00.5

73

.0000 03.3 0:33 3000 .00 39:0 .009: .0: :0 0.000 .0003 00 00:30:. 0:...
000.0 .0...

0.0 0.0 5.0 0.0 0.0 0.0 0.0 0.0 0.0 0

.2 3.0....

 

 

 

 

 

 

 

 

 

 

 

 

e00 2.0 .- \_\. .\
00.. .mffi \ ....-- \i F
o. 00% \ \ \ . liki
290.0000an \ KN \\ - \
03.2% \ N \ .\.- \ NR
00. N N \ \ N \

 

 

 

 

000“. 00.005 - .00.:00 >000

 

 

V. c\!
O O
panama/V55 team 0101 ;o afiewamad

‘0.
0

°°.
0

 

74

(Fig. 15) and white-tailed deer spring food sub-model (Fig. 16) provided nominal output.
The histograms also proved to be useful for evaluating the quantity of different
habitat categories. At the appropriate spatial scale the ovenbird model produced a relatively
even distribution of HSI categories (Fig. 5). Most of the other models were skewed either
high or low. Species for which more high quality habitat existed were gray squirrel
cover/reproduction sub-model (Fig. 6) and white-tailed deer fall/winter foods and security
cover sub-models (Figs. 7-8). Species for which low quality habitat predominated the
landscape included American redstart (Fig. 9), blackthroated green warbler (Fig. 10),
ruffed grouse (Fig. 11), pileated woodpecker cover/reproduction and food sub-models
(Figs. 12-13), gray squirrel food sub-model (Fig. 14). and white-tailed deer spring food

sub-model (Fig. 16).

DISCUSSION

Habitat Model Validation

Wildlife habitat modeling is an evolving process. As hypotheses about nature.
models cannot be proven correct, but merely corroborated by experiences and tentatively
accepted until proven false (Naylor and Finger 1967, Caswell 1976). Validation is a
critical step in estimating the reliability of models for forest planning. Habitat model
validation is difficult because there are no consistent standards relating to habitat quality
and because the models are founded on concepts (e.g.. limiting factors. carrying capacity)
often viewed with ambiguity (Schamberger and O’Neil 1986). The approach of HSI models
is generally valid in that habitat quality is likely to exhibit thresholds below which
habitat becomes unsuitable and above which further changes make little difference in quality

(Van Horne and Wiens 1991, Williamson and Lawton l99l).

75

00:.0000 00:30:00 05:0: :00 39:0 .002: .0: :0 0.000 .0002 ..0 00:30:. 0: ... .0. 030.0
000.0 .0...
0.0 0.0 5.0 0.0 0.0 0.0 0.0 0.0 0.0 0
%\ .\ .\ .\ .\ .\ ................
\ N N N \ N

 

    
 

 

.....
.....

....
....
....
....
....
....
..... .... O
..... ....
..... ....
..... ....
..... ....
....
....
....
....
....
....

.....
.....
.....
.....
.....
.....
.....
.....

.....
.....

 

-----
.....
.....
.....
.....
.....

..u.
....
....
....
....
V». c 0'.-
, ..... .... I
..... ....
..... ....
ttttt ...I I
..... ....
....
....
....
....
....
...

.....
.....
.....
.....
.....
.....
.....
.....
.....

 

patenIeAa BGJV moi to afietuamad

 

 

 

 

000.0000 - 000.0050 000.0:

.0008 w:.:% 0000002070003 00. .0900 .0000. .m: :0 0.000 .0009. .0 00:02.0. 0;... .0. 0009”.

000.0 ..w...
md v.0

   

 

 

 

 

N
o

pazenleng eaJv [2101 ;o afiexuamad

 

 

 

V.
o

76

‘Q
o

 

°°.
o

 

 

 

 

08“. 0:000 .80 8:00-222,

77

The HSI models evaluated for this study generally exhibited positive
associations with wildlife population indices. Most models that performed poorly were for
wildlife species censused in low numbers (e.g.. black-throated green warbler, upland
sandpiper, pileated woodpeckerXT able 16). The upland sandpiper and pileated woodpecker
use relatively large areas and can be detected over long distances, and Verner et al. (1986)
demonstrated the difficqu in detecting changes in population numbers and habitat use for
these types of species. The HSI approach does appear to be a valid and useful tool,
particularly for comparisons of large sample sizes and for landscape habitat quantification
(see "Chapter 3: Applying an Ecological Classification System and HSI Models to
Landscape Planning").

Correlation coefficients of low magnitude may be attributed to measurement
errors that accumulated throughout the validation process. For example, each model
component is based on measures of vegetative structure or habitat spatial configurations.
There is error inherent in sampling vegetation structure (both sampling error and error
caused by the inherent variability of the vegetation) and in 615 maps. The conversion of
vegetative strucrures into indices of habitat suitability is also a source of error in that the
relationships depicted in most models are untested. When all of these components are
mathematically combined into a final HSI score. the error is compounded. As a result,
confidence limits around HSI values are probably quite large. Also, error inherent in
sampling wildlife populations is common. Although human-induced sampling biases can be
minimized, it is difficult (if not impossible) to account for natural variation in population
detectability.

Cumulatively, the amount of error in HSI validation is large. The consistent

performance of HSl models in this study is especially encouraging in considering the

78

potential error involved. The significance of the correlations indicated HSI models have
good potential as comparative tools for forest planning or habitat mitigation. As sources of
error in modeling and population estimation are reduced, higher correlation coefficients
would be expected.

Pooling data across years tended to increase the consistency of population index
and HSI association, often masking a poor within-year association (Table 16).

Comparisons to correlations on random numbers suggest that this phenomena is not simply
a function of larger sample sizes, but rather, as sample size increases the affects of outliers
have less influence on the correlation analysis.

Several characteristics of HSI modeling have been identified as sources of
potential error: 1) the linear relationships between wildlife density and habitat parameters
that do not reflect the inherent variability in these relationships over geographic regions
(Bart et al. 1984, Maurer 1986, Johnson et al. 1991), 2) adequacy of observed density as
an indicator of habitat quality (Van Home 1983, Noon 1992), 3) inter- and intra-specific
population interactions (Lancia et al. 1982, Flather and Hoekstra I985, Schamberger and
O’Neil 1986, Smith and Shugart 1987), 4) restricted model breadth (Farmer et al. 1982,
Rotenberry 1985), 5) the method used to calculate final HSI values from model components
(Bart et al. 1984, Van Horne and Wiens 1991), and 6) using means to describe vegetative
attributes (Van Horne and Wiens 1991). Also, differences in individual animal behavior as
to habitat preferences may introduce error (Cody and Walter I976, Collins et al. 1982,
Verner 1984). Undoubtedly valid, these concerns require attention in future modeling
efforts, however, there is an immediate need for habitat evaluation tools in forest planning.
Managers must understand that HSI models will likely only capture a portion (typically half

or less) of the variation in population estimates (Morrison et al. 1992). Most managers are

79

willing to accept a limited degree of error, and past experiences suggest that the error
inherent in HSl validity and landuse conversion functions are usually tolerable for strategic
decision making (Lancia et al. 1986).

Although individual animals exhibit specific habitat selection behaviors (Maurer
1986), the basis for using animal abundance to assess habitat quality is that all members of
a population have the same limiting habitat components, and although individual variation
in habitat selection occurs, the average habitat selection pattern will be consistent with their
limiting needs. Thus, for large samples, the individual behavior is averaged out.
Nonetheless, caution should be exercised when using animal abundance to assess habitat
quality. A general consensus exists among researchers that better estimates of habitat
quality are obtained through direct measures of population fitness, such as fecundity and
survival (Van Home 1983, Van Horne and Wiens 1991, Martin and Nur 1992, Noon
1992). Data on population demographics is extremely difficult and expensive to obtain, and
currently, an impractical way to assess habitat conditions. Until more efficient population
assessment techniques are developed, animal abundance will continue to be used as an

indicator of habitat quality, not by choice, but by necessity.

Spatial Scale of HSI Model Applicability

The scale of space and time on which ecological systems are viewed influences
the patterns and/or processes that are detected (W iens 1986). Spatial and temporal
characteristics of landscape structure potentially important to wildlife species include forest
stand size, variability in habitat conditions among stands (heterogeneity), temporal dynamics
in habitat use, and the regional context (extent and configuration) of forest stands in a
landscape (Logan et al. 1983, Freemark 1988). To date, few model validation studies have

evaluated habitats across landscapes, even though the effects of scale on model output is of

80

concern (Logan et al. 1983, Maurer 1986, O’Neill et al. 1986, Wiens 1986). Explicit
consideration of multiple scales is essential for addressing the complexity of habitat
selection, and our inability to deal with the complexity of the spatio—temporal framework
may be one of the underlying causes of model failures (O’Neill et al. 1986). Model
validations have been applied to spatial scales often associated with the organism’s life
requisites (e.g., home range, muskrat marshes) (Table 19). The assumption is made that
the habitat model operates most appropriately at the chosen spatial scale, however, no
quantitative support for this assumption is provided. HSI models are extremely sensitive to
spatial scale (Van Horne and Wiens 1991), especially in heterogenous environments. For
example. figure 1? demonstrates the response of area-weighted HSI values for 4 wildlife
species at 10 sample points as size of the analysis area changes. The influence of scale on
model output is related to the habitat configuration surrounding each point, explaining why
some points (for individual models) exhibit lO-fold changes in HSI value whereas others -
remain relatively constant (Fig. 17).

Vegetative heterogeneity provides a potentially complex basis for resource
partitioning within and among coexisting species (Denslow I985, Kaufmann and Landsberg
1990). In a vegetatively heterogenous analysis area, as spatial scale increases, the number
of polygons included in the analysis increases. Scaling up output from a small—scale model
is not just a matter of altering parameter values to account for the range of habitat
conditions; there may also be additional feedbacks and interactions at higher organizational
levels (Kaufmann and Landsberg I990). The convergence of HSI values at large spatial
scales follows standard statistical theory in that as sample size increases (i.e.- the number of
polygons in the analysis), the variance around the mean will decrease and the mean HSl

becomes the "best" estimator of habitat quality for the landscape. At this scale,

81

Table 19. Studies validating habitat suitability index (HSI) models and the spatial scale at

which the validation was performed.

 

 

Area of

Validation
Wildlife Species (ha) Citation
Pronghorn Antelope (Antilocapra americana) 6,500—12,000 Cook and Irwin (1985)
Ruffed Grouse 66-151 Hammill and Moran (1986)
Clapper Rail (Rollus longirosm's) 40 Clark and Lewis (1983)
Muskrat (0ndatra zibethicus) 80-280 Bart et al. (1984)
Gray Squirrel 49 Seng (1991)
Fox Squirrel (Sciurus niger)
Beaver (Castor canadensis) 2.4 km‘ Robel et al. (1993)
Lynx (Lynx canadensis) 1.252 Lancia et al. (1982)
Eastern Phoebe (Sayornis phoebe) 2-ll Cole and Smith (1983)
Red-eyed Vireo (Vireo olivaceus)
Prairie Warbler (Dendroica discolor)
Field Sparrow (Spizella pusilla)
Meadow Vole (Microtus pennsylvanicus)
White-footed Mouse (Peromyscus leucopus)
Cottontail Rabbit (Sylvilagus floridanus)
Spotted Owl (Strix occidentalis) 162 Laymon and Barrett (1986)
Douglas’ Squirrel (Tamiasciurus douglasiz) 42
Spotted Owl 4-162 Laymon and Reid (I986)

 

‘HSl’s computed per stream length.

2l-lSl’s computed per map grid cell.

82

00.00 ..000 .0 00.000 .00000 00000....0 00.. 0000.00.00
0:03 0 .5: 0.0.00 000.0006. 00000.00 2.000000 0. .0 00.0000 050...: 00.00.00 00.. 000.0> .m: 00.00.03 000.... .0. 00%.”.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.05 0.00m .00000 .05 0.00m 3:08
0.00. 0.00 0.00 ...0. 0.. 0.. 0.0 ..o o 0.00. 0.00 0.00 «.0. 0.. 0.. 0.0 ..0 o
v w
0 / 0
...0 m. . 0.0 o.
w w
0.0 m . w
0M0 1. o G W
A
0.0 m. . 0.0 M
o . m
. .
09080 08022 08”. 3.0.2.00“. - .80 02.2.8.0;
.2. 0.80 .0080 . . . 05.280 0.... 00 . . .
0.00. 0.00 0.00 0.0. 00 0.. 0.0 ..0 0 000. .00 ~00 0.0. 0. 0.. 00 ..00
.y ill!\‘
.0
M / . M
. tom“ . 0.0m
. . M - 0.0
A
- 0.0 m . 0.0 a.
n m
9 /,
. . , .

 

 

 

 

 

8:20 00000 .0022 8.038.000.0030. 00008082. 0302.0

83 °
heterogeneity in model output declines and the landscape is viewed as relatively homogenous
(Welsh and Healey 1993). This approach, however, ignores the biological limits of habitat
perception for the organism. Individuals do not respond to average HSI conditions for entire
landscapes, but rather, they respond to habitat patches at a scale compatible with obtaining their
life requisites (Wiens 1974, Rice et al. 1984, Mackie and Dusek 1992). Of additional
importance is identifying areas providing "source" or ”sink” habitats (Morrison et al. 1992).
Average HSI scores across landscapes will disguise these important habitat considerations. lf
habitat models are developed to discriminate habitat quality best at biologically meaningful
scales, then the scale at which between patch variation is maximized provides the "best" estimate
of habitat quality.

The effects of spatial scale on model output may also be confounded by the
mathematics within each model. For example, the mathematics of the grouse model make it
difficult to obtain high quality ratings (Fig. 17), a characteristic often associated with models
based on geometric means. Also, several models (e.g., ruffed grouse, white-tailed deer)
contained spatial components (e. g. proximity of winter food to cover for grouse) that could
potentially distort model output. For example, the HSI value for excellent (HSI = 1.00) grouse
habitat within l6 ha may equal 0.00 if no winter food occurs within the 16 ha. To avoid these
biases, spatial analyses should be conducted independently from model computations, i.e., the
proximity analysis of winter food to cover within the 16 ha grouse area should be conducted
prior to delineating the 16 ha analysis unit.

Prior to implementing habitat models, the proper spatial scale of model applicability
should be determined. The geographic scale at which models are applied should reflect the size
of the animal’s home range, the degree of habitat specialization by the animal, the heterogeneity

of the habitat, and the intended use of the model (Flather and Hoekstra 1985, Laymon and

84

Barrett l986). The use of histograms to evaluate the behavior and consistency of models was
demonstrated by Van Horne and Wiens (1991), however, their analyses did not incorporate
varying spatial scales. Wildlife species have been documented to exhibit different habitat
selection patterns over small geographic areas (Stauffer and Best 1986) and to detect differences

in habitat quality, HSI models must be implemented at appropriate spatial scales.

SUMMARY AND CONCLUSIONS

The current regulatory environment effecting resource management decisions requires
wildlife habitat assessments that are repeatable, scientifically credible, and legally defensible.
The HSI approach conforms to these requirements, and the persistence of habitat evaluation
procedures attests to their usefulness. There has been a general divergence of opinion regarding
the reliability of HSI models; with many "researchers” demonstrating and questioning "poor"
model performance (usually as defined by some pre-determined level of significance), in contrast
to many managers who are excited about a tool capable of predicting animal distributions with
60% reliability. The results of this study support the contention that HSI models are a legitimate
tool for increasing the reliability of habitat assessments. Recognizing that HSI models are not
static and should evolve with the acquisition of new data (e.g. validation studies), the
performance of HSI models should continue to improve.

An appealing characteristic of using HSl‘s integrated with GIS is the ability to
incorporate the spatial requirements of an organism. Most animals require certain configurations
of their life requisites to successfully reproduce and survive (e.g. the interspersion of food and
cover for ruffed grouse). and these relationships can be quantified and displayed. Caution must
be exercised when using HSI models in that the size of the evaluation area will influence the
results. This study demonstrated that HSI models should be used at biologically meaningful

scales (e.g. home range), the size of which will be influenced by an organism's needs and

85

heterogeneity of the habitats. Habitat assessments conducted on a stand-by—stand basis are no
longer acceptable for many species, and the HSI-GIS tools can provide insights as to the effects

of land management activities on these spatial constraints.

Chapter 3: Applying an Ecological Classification System and

HSI Models to Landscape Planning

Ecosystem management has recently been adopted as the approach to land
management on federal lands. Encompassing multiple spatial, temporal, and tr0phic levels.
ecosystem management is a response to address social, economic, and biological issues
including concerns about biodiversity. habitat fragmentation, population viability, and genetic _
heterozygosity. These issues cannot be addressed using stand-based, species by species
assessments, but rather, a combination of coarse and fine filter approaches to resource
management must be used.

Coarse filter approaches involve assessing resources by broad biological groupings
such as "ecosystems” or ”communities" (Hunter 1990, Morrison et al. 1992). Coarse filter
approaches for resource management assume that a representative sample of biological
groupings appropriately spaced across a landscape will insure ecological integrity, where
ecological integrity is the interaction of the physical, chemical, and biological elements of a
system in a manner that ensures long-term health and sustainability of the system
(Environmental Protection Agency 1994, memo referencing the ”Edgewater Consensus“).

The problem with coarse filter approaches is that some ecosystem components are certain to
slip through the analysis (Hunter 1990). The fine filter approach is designed to catch
individual components known to be missed or inadequately represented by the coarse filter.

For example, threatened and endangered plant and animal species are appropriate fine filter

86

87
candidates in that land managers must meet their legal requirements in resource plans.

To implement a coarse filter-fine filter approach, managers need large amounts of
data to account for wildlife objectives, and thus, the complexity of forest planning has
increased. Databases created from an ecological classification system, used in conjunction
with existing vegetative cover types, provide a coarse filter resource planning tool capable of
integrating ecological components into a manageable entity (Fig. 18). Used in conjunction
with HSI models (a fine filter tool) and GIS, the resource planning tool allows managers to:
1) quantify existing vegetation, 2) evaluate the correspondence of existing vegetation to
ecological units (as identified by ECS), 3) develop management alternatives consistent with
land capabilities (both plant and animal), 4) assess current wildlife conditions, 5) simulate the
impacts of management alternatives, and 6) estimate short- and long-term dynamics of wildlife
habitat across landscapes (Roloff and Haufler 1993).

The goal of this chapter is to review the ideas and concepts for landscape
assessments of wildlife habitat that have resulted from this research project. This is not an
exhaustive review of all landscape assessment tools, but rather, a summary of a methodology
that incorporates the systems and tools evaluated for this project.

The Ecosystem Diversity Matrix

The foundation of an ecological, landscape-based approach to wildlife habitat
management involves maintaining adequate representation of the diverse ecosystems inherent
in our landscapes (Haufler 1994). Haufler (1994) proposed the use of an ecosystem diversity
matrix for landscape assessments. The matrix is a coarse filter approach that allows for
quantification of cover types and seral stage by ecological unit and provides a basis for

identifying appropriate management options (Haufler 1994).

88

.00.: 00.0:0.0 00.5000: 0 0.0000 00 80.000 0000050003 .00.00.000 00.00.0000: 000 00.00.00? 00.00.00 ..0 00.030 .0. 0000.“.

000000.. 00:00:00.0 .0030: @0000. .0
000000 .0:0.0000000 00.00000 .N

50.00000: 000 00000005 .00.00.000> 00.00.00 00.00009 ..

000.....00000 00$. 05000.0 00.50000

00$. 00.0005 00000000

2 "50 .3000 .802

0. "50 .5000 .8... 00m.

 

 

 

 

Omaha—m 60.0-0.9. .00.: 0.00.8.0

500000 0000050005 .00.00.00m

 

.....m

 

 

 

 

 

0000... 00000”. 0.0.0.0.".

 

0.0000

00.00.00 - 0.2 «00.8.0.

.0003.

 

 

 

 

89

Specifically, general vegetative types by ecological classification, age class (from
which existing vegetative structure and composition can be described), and quantity can be
identified from the matrix (Fig. 19). Integral components of the ecosystem diversity matrix
are the successional linkages within ecological land type (Haufier 1994), e.g., aspen on
ELT’s in the 20’s and 30’s succeeds into oak and pine if undisturbed, and oak on the high
30’s ELT succeeds into northern hardwoods (Fig. 19). The successional linkages depicted in
figure 19 (i.e. down an ELT column) are crude by nature of the classification schemes. For
example, oak on BLT 30 has multiple successional pathways (into an oak-pine, white pine.
or northern hardwoods community) depending on ELTPG. Matrix cells containing
vegetative cover types and age classes not typically supported by underlying ELT can also be
identified (Fig. 19).

Vegetation structure can be estimated from cover type, age class, and underlying
ELT; and more exact estimates can be obtained by incorporating existing vegetative
inventories (e.g. the cover type-ELT database discussed in Chapter 2). Also, knowledge of
disturbance regimes and successional pathways can be used to identify management options
(Haufler 1994). For example, ELT’s that historically supported mature cover types (e.g.
northern hardwoods on ELT’s in the 40‘s) can be identified and managed to support older
forests. Similarly, ecological land types that historically exhibited frequent disturbances (e. g.
jack pine on ELT’s in the 0’s and 10’s) should not be managed on long timber rotations
(Haufler 1994).

The ecosystem diversity matrix and associated cover type graphs for "Hoist
Lakes”. ”Reed Lake“, and the "Sandpiper Fields” are presented in figures 19-21. Eight
major vegetative groupings were delineated (Fig. 19). Forest cover type categories were

sub-divided into 3 age classes; sapling-, pole-, (11.4-25.4 cm), and saw-sized (>25.4 cm)

Ecosystem Diversity Matrix
Existing Forest Age Ecological Land Type
Cover Types Class 0 10 20 30 4o 50 60 7o 80 Total

40 6
White 375 75
Jack Pine-Black Oak l4 16

601 635
413 571
14 15

83 50
Black, Red. White Pole 595
Red. White Pine-Oak 24

Northern Hardwoods
Beech,
Birch
Deciduous

Black-ash

Conifers

 

0 6 55 88 O 6 2 3 88 247

 

 

[Total | [1625[464 [2243[3147[ 89 [ 61 [ 51 [ 5 [501[8185[
E Denotes cells typically not supported by corresponding ELT (i.e. gaps in
successional linkage).

Denotes the cover type where successional processes slow or are reset.

 

Figure 19. The ecosystem diversity matrix for Hoist Lakes and vicinity, Reed Lake, and
Sandpiper Fields, Alcona County, MI. Cell contents are hectares.

10

91

 

 

 

 

 

 

[Omnibus-Saw]

I A B l - l

 

 

0 2O 40 60 80

 

 

 

 

 

 

 

 

 

 

 

 

1O

 

 

 

 

 

 

 

_ . Lowland Conifer Cover Ty
Figure 20. Cover type diversity graphs for pine, oak, and lowl

x-axis is ecological land type (BLT).

20 40 60 80 0 20 4O 60 80
JackPine RedPine WhitePine
Pine Cover Type
[imam-aw]
[
0520 ”LBJO‘BLOA 410L610 L 0A2; AAGBLOIO 0 20410 60 80
Black Oak N.Red Oak” White Oak Mixed Oaks
Oak Cover Type
[DWzBPueIS-wl
[
oiz‘o‘io‘éo‘ab 6‘233‘0453‘0 32310 so {0
White Cedar Mixed Conifers Black Spruce

ancPcCOmfer cover types. The

 

92

 

 

 

L 1 l J 1 1 1

A1
01020304050070

Black Ash
Lowland Deciduous Cover Type

 

 

 

bamaomls-w]

 

 

 

 

3.0233370521072180 0 mama-$1.5m.)
Birch Sugar Maple-Beech
Northern Hardwoods Cover Type

1 l 1 1

 

 

 

[OS-MOPoleISu ]

 

 

 

 

 

 

 

 

 

2» L
F .
L \ m‘
01L 7 11‘411 IL .49111
aiomaowaoeoroaooowzosowsooomao

Trembling Aspen Big-tooth Aspen
Aspen Cover Type

 

 

Figure 21. Cover type diversity graphs for lowland deciduous, northern hardwoods, and aspen
cover types. The x-axis is ecological land type (ELT).

93

(Fig. 19); consistent with U.S. Forest Service Corporate Database classification (U.S. Forest
Service 1988). Matrix cells contain cumulative hectares of vegetative type and corresponding
age class per ELT (Fig. 19). Forest cover types representing the seral stage at which the
successional process typically slows or is reset according to underlying ELT are denoted by
shaded cells (Fig. 19).

The ecosystem diversity matrix was quickly constructed with the aid of GIS. The
overlay capabilities of GIS were used to create the resource planning map, and subsequent
polygon attribute database queries were conducted to determine hectares in each matrix
category.

The ecosystem diversity matrix is a coarse filter approach to landscape
assessments. The matrix is designed to delineate generalized, inclusive units; a necessary
generalization to transform a continuum of ecological units and successional stages across
landscapes into manageable entities. Data trends in the ecosystem diversity matrix may be
masked by numerical complexity, thus a graphical tool was developed (Figs. 20-21). Cover
type diversity graphs are derived from and linked to the matrix, providing a tabular and
graphical representation of the landscape. The graphs allow for a species-specific evaluation
of cover type-ecological unit representations in the landscape, however, numeric details and
successional linkages are sacrificed (Figs. 20-21). To depict the same species-specific
information in the ecosystem diversity matrix, the matrix would have increased in complexity
by an additional 24 rows.

The most prevalent cover types on the study areas were aspen, oak, and pine, in
decreasing order of abundance (Fig. 19). Lowland conifers and lowland deciduous cover
types were least prevalent. The aspen cover types were equally distributed between

trembling and big-tooth, consisting of sapling- (17% of the landscape) and pole-sized (13%)

94

on ELT’s 20 and 30 (Fig. 21). The oak cover type was predominantly pole-sized on ELT’s
20 and 30, dominated by northern red oak (15% of the landscape; Fig. 20). The pine cover
type accounted for 19% of the landscape and was dominated by sapling- and pole-sized jack
pine (Fig. 20). Jack pine inhabited the lower, less fertile ELT’s, while red pine was found
on ELT’s 20 and 30.

White pine was historically a major forest component on the sandy soils typical
of the Hoist Lakes area, however, white pine is poorly represented in the ecosystem diversity
matrix and cover type diversity graphs (Figs. 19-21). White pine grows best on the sandy
loams of BLT 30 (accounting for 45% of the landscape), and indeed, substantial white pine
regeneration was noted, often under an oak canopy. In contrast, oak stands on ELT 20 were
more likely to contain red pine in the understory. Both 20 and 30 ELT’s have the potential
to produce saw-sized stands of pine—oak mixes (Fig. 19), however, the pines require release
from the existing overstory.

The majority (4,937 ha) of the landscape was in pole-sized timber, followed by
sapling size (2,917) and saw-sized (98 ha). The abundance of pole-sized timber reflects: l)
second-grth timber from past, broad-scale silvicultural applications, and 2) the disruption
of natural disturbance regimes. Most of the saw-sized timber consists of residual pine and
oak, although some aspen also exists (Fig. 19).

The most prevalent ELT’s were the 20‘s and 30's, consisting of mesic ice-contact
sand hills and herb-poor moraines, respectively (Cleland et al. l993). All of the prevalent
vegetative types (pine, oak, northern hardwoods, and aspen) occur in the potential
successional pathways for these ELT’s (Cleland et al. 1993). Also prevalent were ELT’s in
the 0’s and 10’s, consisting of outwash plains (Cleland et al. 1993). These ELT‘s

historically exhibited high fire frequency and thus. seldom attained saw-sized timber.

95
Outwash plains support pine and oak (Cleland et al. 1993).

The remaining ELT’s, the 40’s, 50’s, 60’s, 70’s and 80’s, accounted for 7% of
the landscape and included mesic sites (BLT 40) characteristically supporting northern
hardwoods and aspen-birch associations, dry ELT complexes (50’s and 60’s)(Cleland et al.
1993) supporting aspen and oak, and poorly drained sites (ELT’s 70 and 80) that support
lowland species (Cleland et al. 1993).

Generally, forest types appear to be appropriately distributed in regards to ELT’s.
Small amounts of aberrant acreages (e. g. 3 ha of lowland conifers on ELT 10) are probably
due to GIS mapping error. The 43 ha of northern hardwoods on ELT 20 (Fig. 19) consists
of a forest stand that may have been more appropriately classified as mixed oak (as suggested
by ground reconnaissance).

Of concern among forest type age classes is the disproportionate amount of
saplingzpole-sizedzsaw-sized timber. This analysis, however, has the potential to be
deceiving. For example, acreages of mature oak that were selective harvested on high 30’s
ELT’s frequently had an abundance of northern hardwoods saplings in the understory. Most
of these stands retained their ”oak" classification at a lower stocking density when in reality
122111 the oak and sapling-sized northern hardwood classification could have been appropriate.
Similarly, a flaw in grouping tree species (in the ecosystem diversity matrix) is evident in
that saw-sized jack pine is naturally infrequent, however, the potential for saw-sized red and
white pines is much higher (warranting the use of cover type diversity graphs). The key to
successful use of the matrix is to track matrix changes over time (perhaps on lO-year cycles)
and develop the proper successional linkages. For example, aspen on 30’s ELT will succeed
into northern red oak-red maple cover types (Cleland et al. 1993). Having a knowledge of

the silvicultural applications during the lO—year period, and assuming that the vegetative GIS

96

layer is current, one can track these natural shifts between vegetation types. The ultimate
goal of a coarse filter analysis using the ecosystem diversity matrix is to achieve adequate
ecological representation across the landscape (Haufler 1994).

Although designed to be evaluated over larger spatial scales, the question for the
Hoist Lakes area becomes ”what is adequate ecological representation?" The ecosystem
diversity matrix indicates that virtually none of the ”late-successional" ecological units are
represented (Fig. 19). The greatest potential (as indicated by ELT potentials and areas) to
meet this deficiency lies in creating saw-sized oak or oak-red pine mixes on ELT 20; saw-
sized oak, oak-white pine mixes, or white pine on ELTPG’s in the low 30’s; or saw-sized
northern hardwoods on ELTPG’s in the high 30’s. Similarly, the absence of "late-
successional" lowland conifers on BLT 80 may warrant consideration. On the other extreme.
adequate ecological representation or openings requires an understanding of historic
disturbance patterns. For example, large openings on ELT’s 0 and 10 historically existed as
a result of fires inherent to these sites. Presently, no "openings” exist on ELT 0, although
large areas of jack pine seedlings do occur. Species associated with vast openings (e.g.
upland sandpiper) need large openings to successfully reproduce. and adequate ecological
representation may not be achieved without sandpiper populations. These types of issues

need to be addressed on a larger and more diverse landscape than the Hoist Lakes area.

lntegrating HS] Models: The Fine Filter

With adequate ecological representation of cover type-ELT combinations, one
assumes that all wildlife species inherent to the landscape will be accounted for. However,
to check the effectiveness of the coarse filter, fine filter approaches for selected species are
warranted (Haufier 1994). Also, management and legal requirements may mandate species-

specific assessments.

97 -

The forest potentials as indicated by ELT’s help delineate "adequate ecological
representation” across the landscape (Haufier 1994). For example, late successional Stages
for ELT’s 30 and 40 (e.g. northern hardwoods) are under-represented, with the majority of
the ELT’s in secondary successional species (aspen and oak; Fig. 19). Thus, one would
assume that wildlife species associated with this ecological unit are also under-represented.

Where appropriate, HSI models can be used as the fine filter analysis tool.
Assume pileated woodpeckers were identified as indicators of "mature forests” or "late
successional stages” (as identified by shaded cells in Fig. 19). The ecosystem diversity
matrix suggests that the study areas provide inadequate ecological representation of this
vegetation type (Figs. 19-21), and the woodpecker model can be used to verify this
prediction in addition to geo—referencing the habitat linked to "mature forest" potentials (Fig.
22). 1f the management objecrives were to increase ”mature forest” representation across the
landscape, one would expect pileated woodpecker habitat quantity and quality to increase
over time. Thus, HSI maps can be used to document the progress towards meeting
management objectives. Similarly, if management objectives are to provide for a diversity of
ecological representation and wildlife species, several wildlife species can be identified and
HSI maps created to evaluate management effectiveness. HSl maps for several species can
be overlaid, and where high quality habitat overlaps for several species, these areas can be
identified and managed accordingly. This type of approach also allows for consideration of
patch and metapopulation dynamics.

Stand by stand assessments of most wildlife habitats are inadequate (see Chapter
2), and therefore, models must account for relationships between multiple stands across
landscapes. Areas must be identified that provide both structural m spatial habitat.

Recognizing that each point in the landscape may provide a unique habitat configuration, HSI

 

 

a .
” _l
!
I l I
D 1
fi 172‘
2)} i.
_ ..i .
‘ C .
7
Q l
k l
1
Habitat Suitability Index
. 1.00030 , .
Habltat assessments performed at grld
. 0.80-0.60 intersections (1000 x 1000 ft grid).
0 0.60—0.40
. 0.40-0.20
0"“ 0.20-0.00
Figure 22. Hypothetical habitat suitability assessment for pileated woodpeckers across the

98

Pileated Woodpecker Habitat Quality

 

 

 

 

 

”Hoist Lakes” landscape.

99

models should assess habitats from a representative sample of these points. One such
approach is to overlay the landscape with a grid (Fig. 22), the dimensions of which would be
dictated by the habitat requisites of the species. For example, the grid size for pileated
woodpeckers was 1000x1000m (Fig. 22), roughly corresponding to the scale of model
applicability (see Chapter 2). A home-range-sized circular plot was subsequently overlaid on
each grid intersection, and HSI values were computed (Fig. 22). Subsequently, maps
delineating regions of similar HSI values can be created and integrated into management
decisions.

The question still remains as to the fulfillment of wildlife management goals
across the landscape. The HSl maps are useful for evaluating the spatial distribution and
quantity of habitats, but there is still a need to efficiently summarize habitat quality in terms
of management goals. Assume HSI values were calculated at 175 points across a landscape
(as was done to create the pileated woodpecker HSl map (Fig. 22)). A histogram depicting
HSI value (abscissa) and the number of points (ordinate) can be used to evaluate landscape
potentials (Fig. 23). Minimum levels of acceptable habitat quantity and quality can be
identified and related to existing habitat conditions (Fig. 23). Acceptable levels of habitat
may be driven by minimum viable population requirements (e.g. Fig. 23), metapopulation
theory (i.e. providing large amounts of low quality corridors between isolated, high quality
patches), and/or arbitrary management objectives (e.g. an even representation of all HSI
classes across the landscape (Fig. 23)). For example, the habitat evaluation depicted in
figure 23 indicates that both minimum viable population requirements and management
objective are not being met, however, there is an abundance (above the management
objective) of low quality habitat that can be enhanced. Although the histogram is a useful

display tool, it cannot be separated from the spatial HSI map and both should be used

100

 

 

.8083“. 30.005002 0 5.. 02000.00 0008000000. 0000.000 08000.0... :05 0000. 00:38.00 08.00:

     

   

0020 mm:
06 0.0 We wd nd v.0 md Nd ..o o
1 00
msfimmfiai 53.5: 8
03000.00 0008000002 m .
l 8
l 00
2.08.30 0:20 \‘
800300 2.05 55.52 .

 

 

§

.2 2.00

admsplm'l u! Samoa [811 “010

101

together. Considerably more information is nwded to link habitat planning processes with
the needs for minimum viable populations, and in turn, adequate ecological representation for

the coarse filter approach.

SUMMARY AND CONCLUSIONS

The ecosystem diversity matrix and graphs provide static views of landscapes,
however, the dynamic characteristics inherent to all lands is also a critical component of
successful ecosystem management. Preserving biodiversity in temperate regions requires the
maintenance of all successional stages (Franklin 1988). With biodiversity at the forefront of
contemporary management issues, the dynamic nature of the landscape must be accounted _
for. Although succession is a continuous process rather than a discrete event (Flather and
Hoekstra 1985), techniques to incorporate successional change into forest planning are
required. The ecosystem diversity matrix, as a coarse filter, provides a basis for
incorporating the natural successional pathways of land into resource management plans.
With the addition of successional models, both ecosystem diversity matrices and HSI models

can provide quantitative pictures of landscape dynamics over time.

APPENDICES

APPENDIX A

Description of Ecological Classification System Components
The Ecological Classification System (ECS) of the Huron-Manistee National Forests is a
hierarchical classification (Fig. 24) developed in northern lower Michigan (Cleland et a1.
1993). Appendix A briefly summarizes the ecological land types (ELT’s), ecological land type
phase groupings (ELTPG’s). and ecological land type phases (ELTP’s) as described by Cleland
et al. (1993). Only ecological units used for analyses in Chapter 1 are described.
ELT: Outwash Plains
Characterized as black oak-white oak-Vaccinl'um plant association on excessively well drained
sands of outwash plains. Coarse textures and near siliceous minerology of these outwash
deposits cause few soil morpohological differences to exist within the ELT.
ELTPG: Outwash l
Characterized as pin oak (Quercus palustris)-white oak-Deschampsia plant
association on excessively well drained sands of glacial outwash plains. The
driest and most nutrient impoverished ELTPG. Differs from other outwash
ELTPG’s by having a more open canopy, no red maple in the understory and
low ground layer coverage.
ELTP: 01 - Same as ELTPG Outwash l.
ELTPG: Outwash ll
Characterized as black oak-white oak-Vaccinium plant association on

excessively well drained sands of outwash plains. May include thin coarse

102

103 ~

Ecological Classification System Hierarchy

 

 

 

 

 

 

 

 

 

 

 

 

 

ELT Outwash Plains Dry Ice-Contact and Mesic Ice-Contact
1 Sand Hills Sand Hills
Outwash l lce;Moraine l Moraine l
ELTPG lce;Moraine ll Moraine ll
Outwash 11
l Moraine lll
Outwash;lce
BLT? 01 10.11 13 20,21 23,25 30,31,33 35 37.38

Figure 24. Ecological classification system hierarchy. Based on the works of Cleland et al. (1993).

104

loamy bands throughout the soil strata. Typically a moderately closed canopy with
bracken fern (Preridium aquall'nium) and red maple in the understory.
ELTP: 10 - Excessively well drained sands of outwash plains;sandy
throughout the soil profile.
ELTP: 11 - Excessively well drained sands with thin coarse
loamy bands beneath sandy sola.
ELTPG: Outwash; Ice
Characterized as black oak-white oak-Vaccinium plant association on outwash
and ice contact sands and gravels. Gravel and cobble common throughout soil
profile; finer textured soils found below 150 cm.
ELTP: 13 - Same as ELTPG Outwash; lce.
ELT: Dry lee-Contact and Sand Hills
Mixed oak-red maple-Trientall's plant association on well to excessively well drained sands on
overwashed moraines, kame terraces, and glacial spillways. Includes red maple in the
overstory with a potential for aspen and paper birch (Betula papyrifera) seres.
ELTPG: lce;Moraine l
Characterized as mixed oak-red maple-Trientalis plant association on well to
excessively well drained sands or banded sands on ice-contact and overwashed
topography. Typically found in rolling topography with red maple and a
higher predominance of red oak in the overstory; also contains a spodic soil
horizon.
ELTP: 20 - Lack of sandy loam textural bands in the

substratum.

105

ELTP: 21 - Sandy loam textural bands in the substratum.
ELTPG: lce;Moraine ll (Note: Description based on limited data).
Characterized as mixed oak-red maple-Trientalis plant association on well to
excessively well drained sands on overwashed moraines, kame terraces, and
glacial spillways. Medium textured sands generally occur down to 50 cm; and
soils in the lower stratum have coarse loamy textural discontinuities.
ELTP: 23 - Unsorted sands and gravels. Some banding in the
solum.
ELTP: 25 - Unsorted sands and gravels. Presence of a fine
textured substrata.
ELT: Mesic Ice-Contact Sand Hills
Characterized as red oak-red maple- Viburnum plant association on moisture-enriched, well
drained sands on moraines and ice-contact topography. Subsections in this ELT are
distinguished primarily by differences in geologic parent material, soil morphology and
classification, and temperature gradient effects on forest succession. Soils are predominantly
well to moderately well drained sands with higher moisture holding and cation exchange
capabilities than the outwash ELT's.
ELTPG: Moraine 1 (Note: Description based on limited data).
Red oak-red maple- Viburnum plant association on well-drained sandy moraines
and ice contact topography. Typically occurs within glacial lakebeds and
adjacent to existing bodies of water. Underlain by perennial water tables with

sandy soil textures to depths generally greater than 1.5 m.

106

ELTP: 30 - Medium sands occur in substratum; coarse sand

and gravels are generally absent from substratum; E horizon 2-

7 cm thick are common.

ELTP: 3| - Generally has banding below 150 cm; gravelly

handed and highly stratified sand substratums of ELTP 230 are

included in this unit.

ELTP: 33 - The subirrigated unit of the Moraine 1 system with

perennial water tables lying 24.5 m deep.
ELTPG: Moraine ll
Characterized as red oak-red maple-Viburnum plant association on well-drained
sands with fine loamy substrata on moraines and ice-contact topography.
Sandy soils to depths generally -> 1.5 m with a substrata of sandy clay loam or
finer textures within depths of 1.5-4.5 m; usually more than 15 cm thick. This
ELTPG may be more nutrient rich than other ELTPG's in this ELT because of
higher bases available in loamy substrata, and possibly higher available
phosphorous due to higher soil pH.

ELTP: 35 - Same as ELTPG Moraine ll description.
ELTPG: Moraine lll
Characterized as red oak-red maple-Desmodium plant association on well to
moderately well drained sandy loams over loamy substrata on ground moraines
and fine-textured glacial lakebeds. High water availability; physiography is

level to gently undulating. This ELTPG occurs within or adjacent to

107

landforms characterized by frequent presettlement fires. May be viewed as a
fire climax unit, which will succeed to northern hardwoods without fire.
ELTP: 37 - Some pedons have loamy sand caps.

ELTP: 38 - Sandy or loamy cap is generally absent.

APPENDIX B

Habitat Suitability Index Model for the Ovenbird

Gary J. Roloff
Michigan State University, East Lansing, MI

The habitat suitability model for the ovenbird was based on the research conducted by
Capen et al. (1986). They described suitable ovenbird habitat as a mature deciduous forest
with ample regeneration in the understory.

Using univariate comparisons between means, Capen et al. (1986) identified 5 habitat
variables (out of 20) that differed between plots used and unused by ovenbirds. Discriminant
function analysis (DFA) was used to identify 2 habitat variables that were estimated to be
important in determining ovenbird habitat suitability (Fig. 25). The suitability graphs are not
empiricaly based, but rather, the graphs are designed to provide a high suitability rating to
stands that exhibit the structural and compositional characteristics of mature deciduous forest
with medium to dense regenration in the understory. The overall habitat suitability index is the

geometric mean of the suitability indices.

108

109

 

Suitability Index

 

 

 

J

 

00 A 5410A15‘20A25‘30
BasalArea(sq.m/ha)

 

 

 

 

 

0.8
6
3 0.6
5
g...
3
a)
0.2
o 1 J . 1 1 I l l l '
0 4.000 8.000 12.000
2,000 6.000 10.000 14,000
Shrub/Sapling Sterne (lbs)

0.5
HSI =(Basal Area Suitability Index * Shrub/Sapling Suitability index)

Figure 25. Habitat suitability index graphs for the ovenbird.

APPENDIX C

A Habitat Suitability index (HSI) Model for the Biack-throated Green Warbler in the
Huron-Manistee National Forest, Michigan.

LC. Bender, R. Moses, and LB. Haufler
Michigan State University, East Lansing, MI

The black-throated green warbler (Dendroica virens)(BTGW) is a common summer resident of
northern coniferous and mixed coniferous-deciduous forests of eastern North America.
Although occurring in a variety of vegetation types, the BTGW is most closely associated with
mature spruce (Picea spp.), hemlock (Tsuga canadensis), and pine (Pinus spp.) stands (Pitelka
1940, Bent 1963, DeGraaf and Rudis 1983). During the breeding season the BTGW is one of
the more frequent wood warblers in the conifer regions of northern lower Michigan, although
it is not common (Pitelka 1940). Despite being used as an indicator of mature conifer forests
(U.S. Forest Service 1989), the BTGW is less specific in habitat preferences than such
characteristic coniferous species as the pine warbler (D. pinus), and will occur in areas having
only a few scattered conifers for nesting (Pitelka 1940).
Habitat Use Information

EM

The BTGW is primarily insectivorous. although the species will eat a variety of plant
matter (DeGraaf and Rudis 1983). Feeding habits consist primarily of gleaning insects from
the limbs and foliage of conifers, usually between 3 and 15 m (IO-50 ft) above the ground.
BTGW’s will also forage in the upper canopies of tall shrubs or midstory trees (MacArthur

I958, Morse I967, Rabenold 1978, Sabo I980, Mauer and Whitmore 1981). Feeding

110

lll

behavior consists of hopping, peering, or hovering followed by gleaning; hawking is much less
common (DeGraaf and Rudis 1983).

Major foods of BTGW’s include leaf eating caterpillars, beetles, flies, gnats, plant lice,
spiders, moths, and ants (Reading and Hayes 1933, DeGraaf and Rudis 1983). Seeds and
fruits are taken to a lesser extent ; berries of poison ivy (Toxicodendron radicans) and juniper
(Juniperus spp.) are particularly preferred (Barrows 1912, DeGraaf and Rudis 1983).

BTGW’s forage in a variety of vegetation types, including swamps, orchards,
deciduous trees, and shrubs (Bent 1963). Morse (1973) felt BTGW’s were opportunistic
foragers and not food limited. BTGW’s can be selective in their foraging strategies, however,
Morse (1976a) noted preferences for foraging red spruce (P. rubens) in Maine, although food '
was more abundant in white spruce (P. glauca).

flats;

No information relating to water requirements was found in the literature. Due to the
nature of the BTGW’s insectivorous diet, it is unlikely that the species requires free water.

m;

BTGW’s utilize a variety of vegetation types; most commonly, however, some type of
conifer presence is included in habitats used (Pitelka 1940, Bent 1963, DeGraaf and Rudis
1983). Collins (1983) identified 5 plant communities commonly used by BTGW’s across their
geographic range. These communities were 1) pine forest, 2) spruce-arbor vitae (Ninja
occidentalis), 3) mixed spruce-fir (Abies spp.)-deciduous, 4) balsam fir (A. balsanufera), and
5) birch (Betula spp.)-maple (Acer spp.)-beech (Fagus grandifolia). DeGraaf and Rudis (1983)

felt BTGW’s to be primarily associated with hemlock in the northeastern states.

112

In Michigan, BTGW’s have been reported to occur primarily in upland forests of pure
coniferous (usually red (P. resinosa) or white (P. strobus) pine) or mixed coniferous-deciduous
growth; spruce bogs are also commonly utilized, while pure deciduous forest is used
infrequently (Pitelka 1940). BTGW’s also inhabit jack pine (P. banksiana) plains in northern
lower Michigan (Barrows 1912).

r i n

BTGW nest locations are almost exclusively in conifers (Pitelka 1940). However,
nests in maples, bitches, alders (Alnus spp.), and grape vine (Vitis spp.) have also been
reported (Bent 1963, Griscom and Sprunt 1979). Barrows (1912) listed common nest sites in
Michigan to be in balsam fir , spruces, hemlock, and tamarack (um larcinia). Pitelka (1940)
described nesting in red pine in northern lower Michigan.

Nest locations vary in the tree; nests have been recorded from 5 cm to 22 m (2 up to 7
ft) above the ground (Griscom and Sprunt 1979). Pitelka (1940) described the nest site
characteristics of 2 BTGW nests found in red pine in Michigan. The first nest was 7 m (23 ft)
above the ground, and 1.68 m (6 ft) horizontally from the trunk; the second nest was 3.66 m
(12 ft) high and 1.52 m (5 ft) from the trunk.

mam;

Home range sizes for BTGW’s averaged 0.64 ha (1.6 ac) in hemlock-beech vegetation
types in New York; home ranges varied from 0.2-1.0 ha (0.6-2.5 ac) in size (Kendeigh 1945).

Morse (1976a) found BTGW home ranges in spruce forests in Maine to range from 0.2-0.5 ha
(0.6-1.2 ac) in size. Home ranges in less preferred northern hardwoods vegetation types
averaged 1.2 ha (3 ac) in New Hampshire (Sabo 1980). BTGW territories are vigorously

defended during the breeding season (Pitelka 1940, Reading and Hayes 1940, Kendeigh 1945,

113
Bent 1963). Male BTGW’s seldom wander > 100 m (328 ft) outside their established

territories (Kendeigh 1945). Wandering is much less common in BTGW’s than in other wood
warbler species.

N 0 information was available on the minimum contiguous habitat necessary to support
viable populations of BTGW’s. As BTGW’s largely confine their activities and satisfy their
life requisites within their relatively small home ranges (Kendeigh 1945), it is likely that any
forested tract of acceptable habitat larger than the typical BTGW home range can support this
species.

i l i rati

Quantitative data regarding BTGW habitat requirements are extremely limited in the
literature. Collins (1983) hypothesized that BTGW’s take advantage of what local habitat
structure is available. As a consequence of this, regional analyses of BTGW habitat (even
within a single plant community type) may be of limited value for habitat management outside
of that particular region.

Habitat Suitability Index Model

The following model has been developed to objectively quantify habitat quality for the
black-throated green warbler in the Huron-Manistee National Forest (HMNF), Michigan. This
model assumes that the quality of BTGW habitat on the HMNF can be evaluated in terms of 3
habitat variables. These 3 variables are:

V1: Percent conifer canopy cover.
V2: Mean height of the overstory.

V3: Percent of the overstory with a deciduous or coniferous midstory 25 m in
height.

The unit of evaluation for this model is a forest stand.

Model Justification

114

Optimal cover and reproductive habitat for the BTGW is provided by dense stands of
mature conifers with a multi-layered deciduous or coniferous midstory from 30-70% the height
of the overstory trees. Species of conifer does not appear to be an important variable, although
highest BTGW densities are associated with spruce or spruce-fir communities (Morse 1967,
1976a, 1976b; Hamel et al. 1982). BTGW’s will occupy stands of deciduous species as long
as a few conifers are present for nesting (Pitelka 1940, Sabo 1980, Holmes and Robinson
1981, Robinson and Holmes 1982, Hamel et al. 1982). BTGW’s appear to be totally
dependent upon conifers for nesting (Pitelka 1940).

Highest breeding densities and smallest home range sizes for BTGW’s are associated
with dense pure conifer stands (Fables 20-21). Breeding densities of 40-80 pairs/ 100 acres
have been documented in these vegetation types (Morse 1976a, Hamel et al. 1982). Mixed
conifer-hardwood stands or open conifer stands are also suitable habitat, with documented
breeding densities of 10-40 pairs/ 100 acres (Kendeigh 1945, Holmes and Robinson 1981,
Robinson and Holmes 1982). Almost pure stands with only occassionally scattered conifers
typically support <10 pairs/100 acres (Sabo 1980, Hamel et al. 1982). Exact quantification of
vegetative variables associated with most of these studies is lacking, however. Collins et a1.
(1982) found optimal BTGW habitat by principal components analysis to consist of stands with
a mean canopy closure of 84.3% and an average overstory height of 20.2 m (66 ft). Collins
(1983) reviewed descriptions of BTGW habitat from a variety of geographical regions. Total
canopy closure ranged from 84.0% to 94.5% for areas occupied by BTGW’s; 23-76% of this
was conifer (Collins 1983). This model assumes that BTGW habitat will be optimal with a
conifer canopy coverage of 75-100% (Fig. 26). Habitat quality will deteriorate moderately as

conifer canopy cover drops from 75 % to 25%. Below 25% conifer canopy coverage, habitat

115

 

0.8

.0
o

.9
;

Suhbiily Index

0.2

 

 

 

 

 

 

 

 

0'
o 100
1
0.8 ~
goo ~
g 0.4 -
a
0.2 -
c J 1 L
0 5 10 15 20
”warm

 

0.8

 

0.2 '-

 

 

 

0 20 40 60 80 100

Figure 26. Habitat suitability variables for the black-throated green warbler.

116

 

 

 

 

 

Table 20. Density-habitat relationships for black-throated green warblers in various
vegetation types.
Density
(Pairs/ 100 ac) Vegetation Type Source
83 Mature white spruce Morse 1976
71 Mature red spruce Morse 1976
61 Mature white spruce Morse 1976
53 Mature red spruce Morse 1976
51 Mature white spruce Morse 1976
44 Sawtimber spruce-fir Hamel et al. 1982
43 Mature white spruce Morse 1976
36 Mature oak-maple Stewart and Robbins 1958
33 Mature n. hardwood scattered spruce Robinson and Holmes 1982
32 Mature n. hardwood with spruce and hemlock Holmes and Robinson 1981
26 Mature virgin spruce Rabenold 1978
12 Mature spruce-fir Rabenold 1978
9-17 Mature hemlock Kendeigh 1945
9 Mature n. hardwood Stewart and Robbins 1958
8 N. hardwood Sabo 1980
7 Sawtimber oak-hickory Hamel et a1. 1982
Table 21. Territory size-vegetation relationships for the black-throated green warbler.
Territory Size (ac) Vegetation Type Source
0.2-0.5 Mature spruce Morse 1976
0.64 Mature hemlock Kendeigh 1945

1.2 Mature n. hardwoods Sabo 1980

117 '

quality quickly deteriorates. If no conifers are present, habitat suitability is assumed to be
0.00, as BTGW’s are dependent upon the presence of conifers for nesting (Pitelka 1940).
BTGW'S prefer mature stands of large trees (Pitelka 1940, Bent 1963, Morse 1967 , 1976a,
1976b). Collins (1983) reported canopy heights of areas occupied by BTGW’s to range from
14 m (46 ft) to 20 m (66 ft) in height. Collins et al. (1982) reported optimal habitat in
Minnesota to have a mean canopy height of 20.2 m (66 ft). Robinson and Holmes (1982)
found BTGW habitat in the Hubbard Brook Experimental Forest in New Hampshire to be
characterized by canopy heights of 25-27 m (82-88 ft). Kendeigh (1945) similarly noted
preferences for large mature trees; specifically, 61 m (200 ft) eastern Hemlocks. Other
researchers have also noted preferences for large mature trees without quantifying exact canopy
heights (Morse 1967, 1976a, 1976b; Rabenold 1978, Sabo 1980, Holmes and Robinson 1981,
Hamel et al. 1982).

This model assumes that optimal BTGW habitat consists of forested stands of large

mature trees (Fig. 26). Optimal canopy height occurs in stands with a mean canopy height 220

m (66 ft). Suitability declines as mean canopy height drops from 20 to 7.5 m (66 to 25 ft).

Stands with a mean canopy height $7.5 m (25 ft) are assumed to be unsuitable BTGW habitat.
BTGW’s typically forage in an area between 30—70% of the mean overstory height
(MacArthur 1958, Morse 1967, Rabenold 1978, Sabo 1980, Mauer and Whitmore 1981).
BTGW’s glean insects from the inner, middle, and outer branches of all overstory trees in this
height strata, as well as from any tall shrubs or midstory that is present in this range. This
model assumes that the maximization of foliar biomass in this height strata will optimize
habitat by providing maximum foraging opportunities (Fig. 26). As the percentage of the stand

with a tall shrub layer or midstory greater than 5 m (16 ft) in height decreases, habitat

118

suitability for BTGW’s will decline. If a stand has no tall shrub or midstory in this strata, the
suitability index will be 0.20. This is because BTGW’s will still forage in the overstory in the
preferred height range, as well as occassionally foraging in the upper canopy (> 70% of the
mean canopy height) and low shrubs (< 30% mean overstory height)(MacArthur 1958, Morse
1967, Rabenold 1978, Mauer and Whitmore 1981).
Model Application/HSI Determination

The habitat quality of a forest stand for BTGW’s can be evaluated by determining the
individual suitability index values for the 3 habitat variables described above (Fig. 26). the
overall suiatbility of a stand for BTGW’s is determined by relating the individual suitability
values for the 3 model variables utilizing the following equation:

HSI = (V‘*V2*V3)°‘33

This model will thus generate an overall HSI score between 0.00 and 1.00, with 0.00
representing unsuitable habitat and 1.00 representing optimal habitat. Future verfication of this
model can be accomplished by comparing densities of BTGW’s with HSI scores from various
forest stands of differing BTGW habitat quality. A positive correlation should exist between
BTGW density and HSI score for this model to be considered valid.
Literature Cited

Barrows, W.B. 1912. Michigan bird life. Spec. Bull. Dept. 2001. and Phys., Mich. Agric.
College, East Lansing.

Bent, A.C. 1963. Life histories of North American wood warblers: Part 1. Dover
Publications, Inc. NY. 367pp.

Collins, S.L., F.C. James, and P.G. Risser. 1981. Habitat relationships of wood warblers
(Parulidae) in northern central Minnesota. Oikos 39:50-58.

. 1983. Geographic variation in habitat structure of the black-throated green warbler.
Auk 100:382-389.

119

DeGraaf, R.M., and DD. Rudis. 1983. New England wildlife: habitat, natural history, and
distribution. U.S.D.A. For. Serv. Gen. Tech. Rep. NE-108. 491pp.

Griscom, L., and A. Sprunt, Jr. 1979. The warblers of America. Doubleday and Co., NY.
302pp.

Hamel, P.B., HE. LeGrand, Jr., M.R. Lennartz, and S.A. Gauthreaux, Jr. 1982. Bird-
habitat relationships on southeastern forest lands. U.S.D.A. For. Serv. Gen. Tech.
Rep. SE-22. 417pp.

Holmes, R.T., and S.K. Robinson. 1981. Tree species preferences of foraging insectivorous
birds in a northern hardwoods forest. Oecologia 48:31-35.

Kendeigh, SC. 1945. Nesting behavior of wood warblers. Wilson Bull. 57:145-164.

MacArthur, RH. 1958. Population ecology of some warblers of northeastern coniferous
forests. Ecology 39:599-619.

Mauer, B.A., and RC. Whitmore. 1981. Foraging of five bird species in two forests with
different vegetation structure. Wilson Bull. 79:64-74.

Morse, DH. 1967. The contexts of songs in black-throated green and blackbumian warblers.
Wilson Bull. 79:64-74.

. 1973. The foraging of small populations of yellow warblers and American redstarts.
Ecology 54:346-355.

. 1976a. Variables affecting the density and territory size of breeding spruce-woods
warblers. Ecology 57:290-301.

. 1976b. Hostile encounters among spruce-woods warblers. Anim. Behav. 24:764-
771.

Pitelka, F.A. 1940. Breeding behavior of the black-throated green warbler. Wilson Bull.
5223-18.

Rabenold, K.N. 1978. Foraging strategies, diversity, and seasonality in bird communities of
Appalachian spruce-fir forests. Ecol. Monogr. 48:397-424.

Reading, D.K., and SP. Hayes, Jr. 1933. Notes on the nesting and feeding of a pair of
black-throated green warblers. Auk 50:403-407.

Robinson, S.K., and RT. Holmes. 1982. Foraging behavior of forest birds: the relationships
among search tactics, diet, and habitat structure. Ecology 63:1918-1931.

Sabo, S.R. 1980. Niche and habitat relations in subalpine bird communities of the White
Mountains of New Hampshire. Ecol. Monogr. 50:241-259.

120

U.S. Forest Service. 1989. Final environmental impact statement: land and resource
management plan, Huron-Manistee National Forest. U.S.D.A. For. Serv., Eastern
Region.

APPENDIX D

A White-tailed Deer HSI for the Upper Great Lakes Region
“DRAFT“I
Louis C. Bender and Jonathan B. Haufler.
Michigan State University, East Lansing, MI

A habitat suitability index (HSI) model was developed for white-tailed deer in the
Upper Great Lakes Region to assist land managers in assessing and providing for deer habitat
in land management practices. This model evaluates deer habitat via 4 distinct sub-models,
each dealing with a critical habitat component: Fall and Winter Food, Spring Food, Security
Cover. and Thermal Cover. Habitat suitability for each critical component is determined by V
mathematical equations relating variables that contribute to the habitat component. An overall
HSI score quantifying the quality of deer habitat in the evaluation area is then determined as
the lowest suitability score of the 4 component sub-models. Quantifying habitat quality for
key species such as deer greatly aids land managers in accounting for the habitat needs of
wildlife in land management decisions, and assessing the potential impacts of land
management decisions on wildlife.

The white-tailed deer are the most popular big game animal in North America
(Halls 1978). Highly prized as venison, as a trOphy, and for viewing, it is one of the most
valuable and important wildlife species in the Upper Great Lakes Region (UGLR).
Consideration of the habitat needs of white-tailed deer is therefore extremely important in land
management planning.

Although well quantified, the habitat needs of deer have not been extensively

modelled. Deer inhabit a broad range of climatic and vegetative conditions across their range

121

122
(Halls 1978). Due to this diversity, modelling of deer habitat must be localized. Models that
are too geographically broad will be so inclusive as to be trivial. Similarly, models too
narrow in focus will lack applicability to any area outside the narrow boundaries. This model
attempts to evaluate habitat quality for deer in the UGLR in general, and the Huron-Manistee
National Forest, Michigan, in particular. Modelling at this geographic level maximizes model
utility while maintaining model complexity within reasonable constraints.

The habitat suitability index model presented here involves the assessment of deer
habitat via 4 distinct and independent sub-models, each of which deals with a critical element
of deer habitat. These 4 sub-models are:

(1) FALL AND WINTER FOODS

(2) SPRING FOODS

(3) SECURITY COVER

(4) THERMAL COVER
Deer require all 4 of these habitat components for existence in most of the UGLR.
Evaluation of each critical element of deer habitat via a unique sub-model assures that each
life requisite will be given maximum consideration in the evaluation of deer habitat suitability
in the UGLR.
GENERAL INFORMATION

White-tailed deer foods follow similar seasonal patterns throughout the Upper
Great Lakes Region, despite local differences in vegetation (Rogers et al. 1981). Browse
constitutes the primary food source, mainly because it is the only source available year round
(Blouch 1984). Other categories of common deer foods include conifer needles, evergreen
forbs, deciduous leaves, non-evergreen forbs, grasses, fruit, and fungi (Rogers et al. 1981).

Grain residues left in fields after harvest are heavily utilized by deer from September to April

123
(Gladfelter 1984).
ri / mm r Food

Spring foods of deer consist principally of herbaceous vegetation (Healy 1971,
Rogers et al. 1981). The most common foods include grasses, sedges, basal rosettes of
perennial forbs, and emergent bracken fern (McNeill 1971). Grassy openings are especially
important at this time of year (McCafferty and Creed 1969), as are old fields, roadsides, and
powerline right-of-ways (Rogers et a1. 1981). As spring progresses, new green grasses.
emerging forbs, and new leaves of trees and shrubs eventually make up 90% of a deer’s diet
(Pierce 1975, Rogers et al. 1981). These foods are both nutritious and easily digestible,
providing the high quality diet necessary for both survival and reproduction in the face of
increased spring metabolic rates (Verme 1963, 1969; Verme and Ullrey 1972).

During late spring-early summer, deer feed heavily on aquatic emergents (Rogers
et al. 1981). These plants are nutritious and may provide a source of sodium and other
important nutrients for pregnant and lactating does (Jordan et al. 1974).

Summer deer foods consist of leaves of non-evergreen terrestrial plants.
mushrooms, and fruits (Kohn and Mooty 1971, McCafferty et al. 1974). Succulent new
leaves are the principal component of deer diets during this period (Blouch 1984). Leaves of
aspen seedlings < 1 year old are especially preferred.

WM

As summer progresses into fall, use of grasses increases while leaf and forb use
declines (Blouch 1984). During good mast years, acorns, beechnuts, and other hard mast are
highly preferred fall/winter foods. Deer shift their diets to woody browse with the first
frosts; the shift to browse is usually complete with the first snows (Blouch 1984). In the

George Reserve of Michigan, 7cm of snow changed the diet of deer from 63% to 0%

124 -
herbaceous (Coblentz 1970). Where available, agricultural crops, especially corn and
soybeans, can also make up a major portion of a deer’s fall and winter diet (Nixon et a1.
1970).

Woody browse forms the bulk of deer winter diets (Rogers et al. 1981). Northern
white-cedar, red maple, eastern hemlock, American mountain-ash, and altemate-leaved
dogwood are all highly preferred browse species (Blouch 1984). Eastern white pine, yellow
birch, mountain maple, serviceberry, and jack pine are slightly less preferred, followed by
aspen, northern red oak, beaked hazelnut, paper birch, balsam fir, and red pine. Last resort
foods include speckled alder, black and white spruce, and tamarack. Although cedar is highly
preferred, conifer needles are typically a poorer diet than most woody browse (Rogers et al. '
1981). Conifer needles comprise only a small proportion of the diets of healthy deer; needles
can comprise up to 50-60% of stomach contents of starved deer (Aldous and Smith 1938,
Dahlberg and Guettinger 1956).

A prolonged diet of woody browse causes malnutrition and starvation in deer
(Mautz 1978). Browse is typically poor nutritionally, and deer will lose weight on a steady
browse diet even if browse is available ad 11121.11!!! (Ullrey et al. 1964, 1967, 1968; Verme and
Ullrey 1972, Grigal et a1. 1979). As a result, at winters end deer fat reserves are usually
depleted and deer are nutritionally stressed (Verme 1969, Mautz 1978).

COVER
r' v r

Cover can be defined as any structural feature of the environment that is used as
protection from the environment (thermal cover) or from predators (security cover) (Boyd and
Cooperrider 1986). Security cover is used for protection from predators and/or humans.

White-tailed deer are hiders; they rely on suitable vegetative cover to conceal themselves from

125

predators, but they can also run effectively through dense vegetation to escape predators
(Boyd and Cooperrider 1986). Thus, security cover for deer consists of dense vegetation in
which the animal can conceal itself and flee, if necessary.

Thomas et al. (1979) noted that optimal habitat for deer requires hiding cover,
perhaps because it gives the animals a sense of security. Thomas et al. (1979) defined
security cover (hiding cover) as vegetation capable of hiding 90% of a standing adult deer at a
distance of s61m. Most use of security cover by deer tends to occur within 183m of the
edge between cover and forage areas (Reynolds 1966, Harper 1969, Thomas et al. 1979).

Thermal Cover

In the UGLR, snow depth, low temperatures. and lack of protection combine to '
limit the northern range of deer (Halls 1978, Biouch 1984). Snow limits deer movements and
covers food; cold temperatures and wind combine to drive deer energy reserves down. In
response to this environmental stress, deer- tend to concentrate (yard) in heavy coniferous
cover, where snow depth, wind, and radiant heat loss are minimized (Blouch 1984).

White-tailed deer movements begin to be restricted by snow when it reaches
depths of 36-43cm (Kelsall I969). Deer may travel considerable distances to reach suitable
thermal cover. Distances traveled in the UGLR range up to 10-48+km (Rongstad and Tester
1969, Dahlberg and Guettinger 1956, Verme 1973).

Many factors determine what constitutes adequate deer thermal cover. Weber et
al. (1983) found 4 variables to accurately predict deer use of potential thermal cover areas in
New Hampshire: (1) area site index, (2) stand basal area, (3) softwood crown closure, and (4)
stand size. These 4 variables were able to predict site utilization with a precision of 95%.
Additionally, species of conifer has been shown to be an important factor in deer yard quality

in the UGLR (Blouch 1984).

126

In New Hampshire, site index showed the greatest variation between conifer
stands utilized as deer yards and those not used (Weber et al. 1983). Site index is important
in deer yard quality, as the largest trees with the most fully developed and well shaped
crowns grow where site index is highest. This results in high softwood crown closure. For
northern white cedar, deer yards in New Hampshire averaged 17.4m in site index; non-deer
yards averaged 15.1m (Weber et a1. 1983). Site indices for northern white-cedar in the
UGLR are lower than in New Hampshire deer yards, typically 12m on the best sites, and 5m
on poorer sites (Johnston 1977).

Total basal area of a conifer stand was found to be inversely related to winter deer
utilization (Weber et al. 1983). Deer apparently seek out canopy openings to benefit from
incoming radiant solar energy (Aldous 1941). Openings also provide a potential source of
food; saplings and young sprouts often grow in areas receiving direct sunlight (Weber et al.
1983). Conifer stands utilized as deer yards in New Hampshire averaged 49.3m3/ha of total
basal area (34.5-67.5), while non-deer yards averaged a higher 57.7m2/ha (42.2-72.8) (Weber
et al. 1983).

Softwood crown closure has long been recognized as an important component of
winter deer habitat (Verme 1965, Nowosad 1967, Kramer 1970, Biouch 1984). Complete
conifer canopy closure, however, is not necessary and may be undesirable for deer yards
(Aldous 1941, Weber et al. 1983). Euler and Thurston (1980) found that deer use declined in
eastern hemlock stands when crown closure increased beyond 71%. Weber et a1. (1983)
found deer yards in New Hampshire to be characterized by softwood crown closures
averaging 66.7%. Non-deer yards averaged 54.3%. Thus. although high softwood crown
closure is important in snow depth and wind velocity reduction, total closure is undesirable as

it negates the patchiness associated with food production and solar hot spots previously noted.

127

Weber et al. (1983) also found stand size to be an important factor in deer yard
use. Deer yards in New Hampshire were larger than non-deer yards (mean of 63.6 ha versus
40.7 ha). Large areas are more likely to be located by migratory animals, and support
greater numbers of deer.

Conifer species is the final important factor in deer yard quality. Northern white
cedar provides the best cover and forage, and is heavily utilized (Blouch 1984). Eastern
hemlock, jack pine, balsam fir, and other dense stands of upland conifers are also utilized
where cedar or mixed conifer swamps are lacking.

MODEL JUSTIFICATION/APPLICATION
This white-tailed deer HSI model is composed of 4 distinct sub-models:

(l) FALL/WINTER FOODS

(2) SPRING FOODS

(3) SECURITY COVER

(4) THERMAL COVER
The model assesses the quality of the evaluation area for each sub-model habitat component
independently. The final HSI value for deer is based on a most critical limiting factor
theorem: whichever sub-model HSI value is the lowest (most limiting) will be the overall HSI
value for deer in the evaluation area. The reasoning behind this is that each critical habitat
component associated with a sub-model is considered equally important to the well being of
deer; thus, a high value for one component(s) cannot compensate for a low value of another.

White-tailed deer home ranges can vary significantly, based on geographic area,
climate, annual variation in weather, etc (Marchinton and Hirth 1984). Values reported in the
literature ranged from 559 ha to >520 ha (reviewed by Marchinton and Hirth 1984). Most

deer home ranges have radii of less than 1.6 km, however. Therefore, the unit of evaluation

128
for this white-tailed deer HSI model is an area of contiguous habitat of 22.56 km2 (256 ha).
MODEL 1: FALL/WINTER FOODS (FWF)

Fall and winter foods in the UGLR consist primarily of woody browse, mast, and
selected agricultural crops (Rogers et a1. 1981, Blouch 1984). This model assumes that
browse is a function of availability and the relative nutritional quality of the site (Fig. 27).
FALL/WINTER FOOD suitability should be optimal as available (21m tall, $6.3cm dbh)
woody stems per hectare reach 5000 (M,V,, Fig. 27). Additionally, the nutritional quality of
available browse can be related to site quality. This model assumes that higher quality sites
(as assessed by site index for red pine or ecological land type phase (ELTP) will produce
higher nutritional quality forage (M,V,, and alt M,V2, Fig. 27). The total suitability index
(81) for woody browse is a mean of the individual values for availability and nutritional
quality:

SIBROWSE = (MN, + M,V2)/2.

The main mast producing species in the UGLR are the oaks. Acorn production is
a product of the number of oak trees present. their size, and the diversity of species present.
as acorn production within a species can vary greatly annually (Rogers et al. 1981,
Armbruster et al. 1987). This model assumes that acorn production will be optimal when oak
basal area is 27.5m2/ha (M.V3, Fig. 28) (Armbruster et al. 1987). Similarly, larger oak
trees tend to produce the largest and most consistent mast crops. This model assumes that
acorn production will be maximized in trees 250cm dbh (M,V,, Fig. 28) (Armbruster et al.
1987). Production progressively declines as mean diameters drop below 50cm. Below 25cm,
oaks are typically too small to yield good acorn crops; hence, suitability drops to 0.0 (Fig.
28).

Three oak groups occur in the UGLR--the white oak group, the red oak group,

O

O

Suitability Index (WW)
:5

.°
N
I

a
0'1
r

Figure 27.

p
m
1

'01
f

Red Pine Site Index (alt M1V2)(m)’
8

 

 

O
0

41.000 2.

l

 

 

000 A 3.000 4.000 A5000

Woody Stems (Available Browse)(ha)

 

U'i
I

 

 

 

 

$00

220

240

260

#230

Ecological Land Type Phase (ELTP)

129

.0
co

.0
as

O

Suitability Index (M1V2)
1.

.°
N

 

 

J. L 1 g; A 1

 

 

O

5 10 15 20 2
Site Index (Red Pine)(m)

Woody browse variablesfor FALL/WINTER FOODS.

A30

130

 

.° .0 .0
as a: on -
I I '

Suitability Index (M1V3)

.0
N
1

 

 

1 l 1

0 2 4 6 8 10
Basal Area of Oaks (sq. mlha)

 

O

—l

 

 

   

 

 

 

E 0.8
E
g 0.6 -
U
E
g 0.4
3
2
5 0.2
0 1 2 3
1 Species Diversity of Oaks

E? 0.8 —

E

g 0.6

U

E

g 0.4 -

.5

S

5‘, 0.2 -

0 a l

 

 

 

0 20 40 60 80 100
Mean Oak DBH (cm)

Figure 28. Mast variables for FALL/WINTER FOODS.

131 -
and northern pin oak. This model assumes that acorn production will be maximized in areas
having > 1 type of oak present (M,V., Fig 28). Areas with 1 type present are assigned a SI
of 0.5. If no oaks are present in an area, the area receives a SI of 0.0 for mast production
(Fig. 28). The SI for mast as a deer FALL/WINTER FOOD is then calculated as a reduction
function of the above 3 variables:
SIMAST = M,V3 * MN. * M,V,.

This model assumes that browse is at least 4X as important in supplying fall and
winter foods as is mast, due to the unpredictability of mast production and the continuous
availability of browse (Blouch 1984). Therefore, the FALL/W INTER FOOD SI for a forest
stand can be determined by a weighted mean of the individual 51’s for browse and mast:

SIFOR ,7 = (4"‘SIBROWSE + SIMASTVS.

Agricultural fields can also be an important source of FALL/WINTER FOODS
for deer in certain areas of the UGLR. Suitability of agricultural fields as a deer
FALL/WINTER FOOD source is a function of the size of the field, the crop species
cultivated, and tillage practices (Thomas et al. 1979, Armbruster et al. 1987). This model
assumes that agricultural fields with a maximum width of 5 180m are entirely available to
foraging deer and hence receive a 81 of 1.0 (M,V,, Fig. 29). As the maximum width of a
field exceeds 180m deer will not fully utilize the area (Thomas et al. 1979); suitability
therefore declines. Fields 2360m in width will be utilized along their margins only; hence
these fields receive a SI of only 0.25.

The crop cultivated also contributes to an agricultural field’s suitability as a
FALL/WINTER FOOD area. Corn is a highly persistent, highly palatable crop and is
assigned a SI of 1.0 (M,V,, Fig. 29). Other crops are less persistent and/or less palatable and

are assigned lower Sl’s (Fig. 29). Additionally, management practices affect the availability

Suitability Index (M1V6)

p
on

.0
a:

C

Suitability Index (M1V8)
In

Figure 29.

.9
00

.o .o
A O!

.0
N

P
m

0

 

 

 

1

4L

 

 

 

0 A 90
Maximum Width of Fields (m)

180

270

 

Fall Plowed 0.0

 

1 a I 1 l

; Stubble Left 1.0

1

 

 

Agricultural Management Practice

132

Suitability Index (M1V7)

.0
co

.0 .o
A O)

P
N

 

 

Corn 1.0
Beans 0.5
Wheat 0.5

Hay (0.3)

k 1 a 1 a I 1 I

 

Crop Species Cultivated

Agricultural field variables for FALL/WINTER FOODS.

 

133

of agricultural crops as winter foods. Agricultural fields that are left untilled or under other
conservation tillage practices retain maximum crop residues and are assigned a SI of 1.0
(MN., Fig. 29). Fall plowed fields have little available crop stubble for deer food, and are
assigned a SI of 0.0.

The 81 for agricultural crops is calculated as a reduction function in the following
manner:

SIAG = MN“ * MN, * MNs.
Overall suitability of an evaluation area for deer FALL/WINTER FOOD is determined by
summing the individual 81’s of all forest stands and agricultural fields (i.e. all potential
FALL/W INTER FOOD areas) multiplied by the proportional area the total FALL/W INTER '
FOOD areas that they represent; i.e. SI(FWF area I)*(area of FWF area I/total area of all
FWF types) + ..... + SI(FWF area n)*(proportional area of Stand 11), or:
SIM = SUM,., u, ,[(Sli)*(area of j/total area in FWF)].
This value is then modified by assessing the total percentage of the evaluation area in
FALL/WINTER FOOD types (MN,, Fig. 30). Sixty percent or more of the total evaluation
area in FALL/WINTER FOOD types is considered optimal (Thomas et al. 1979); hence, it is
assigned a SI of 1.0. Suitability declines as the percentage of the evaluation area in
FALL/WINTER FOOD types drops below 60%. The final HSI for FALL/WINTER FOOD
for white-tailed deer is then calculated using the following equation:
HSIM = SIM * M,V.,.

MODEL 2: SPRING FOOD (SF)

Spring foods of deer in the UGLR are chiefly a function of forest openings,
agricultural fields, and forest ground cover and understory shrubs (Rogers et al. 1981, Blouch

1984). Spring food production and deer utilization of openings is a function of the openings

134

 

.0
co

Suitability Index (M1V9)
O
'0:

 

 

 

4

0120L40 60180100
Area in FALLNVINTER FOOD Types (96)

Figure 30. Area function for FALL/WINTER FOODS.

135

size, the amount of openings in the evaluation area, and the productivity of those openings
(Fig. 31). Openings with a maximum width 2180m will not be fully utilized by deer
(Thomas et a1. 1979), therefore suitability declines as opening widths exceed 180m (MZVD
Fig. 31). Alternatively, size of openings can be substituted for width, with openings being
progressively less used as size exceeds 10 ac (alt M,V,, Fig. 31). Total herbaceous
productivity, although not quantifying deer foods specifically, provides a relative idea of the
amount of deer food potentially available (Crawford and Marchinton 1989). Openings with
21785 kg/ha likely will produce ample spring foods (MZVZ, Fig. 32). As total productivity
drops below 1785 kg/ha, suitability as SPRING FOOD declines (Fig. 32). The SI of
herbaceous productivity can also be estimated utilizing a function combining woody cover and
herbaceous ground cover (alt szz, Fig. 33).

Approximately 10-30% of the evaluation area in openings represents optimal deer
habitat (M,v,, Fig. 32) (Armbruster and Porath 1980). Below 10%, insufficient openings are.
present to allow adequate spring food production. If over 30% of an evaluation area is in
openings, cover factors are likely to be limiting (Thomas et al. 1979); however, limitations of
cover attributes are assessed using the SECURITY COVER sub-model.

The SI for SPRING FOOD associated with any specific openings is calculated
using a reduction function in the following manner:

SIOPEN = MN, * sz2 * M2V3.
The contribution of all openings toward SPRING FOOD is determined utilizing an area
weighing procedure:
SIOPEN = SUM,.l ,,,[(SI,)*(area of j/total area in openings)l.
Agricultural fields can also provide spring deer food in the UGLR. Food

production and deer utilization associated with agricultural fields can be assessed by

136

 

Suitability Index (M2V1)
as .o .o
«h C) on

.0
N

 

 

 

0 a A g
0 90 180 270 360

Maximum Width of Openings/Fields (m)

 

 

 

 

 

 

 

1
c i
a 0.8 >
2
’6
V 0.6
s
u 1
E
.523 0.4
.3
£5
'5 0.2 .
a)
00 A T 12 18 24
1 Area in FALL/WINTER FOOD Types (ac)
:1?
a 0.8
3
g 0.6
e
.5
£04
3
g
3 0.2

 

 

0 . ' 1 1 a a 1
0 400 800 1,200 1.600
200 600 1.000 1.400
Total Herbaceous Productivity (kg/ha)

Figure 31. Suitability curves for SPRING FOOD openings components.

100

V
(Ti

50

Shrub Cover (96)

25

 

 

 

 

 

 

 

 

 

0.25 0.25 . 0.25 0.25
0.25 0.50 0.50 0.50
—-Ji- -»—- “”71 ---1
0.25 0.50 0.75 0.75
0.25 0-50 0.75 1.00
J -
0 25 50 75 100

Herbaceous Ground Cover (96)

 

P
be

.0
at

.°

Suitability Index (M2V4)
b.

.0
N

 

Alfalfa 1.0

Winter Wheat 1.0

Hay 1.0

Others 0.0

1

1

 

 

Crop Species Cultivated

137

alt M2V2

.0
a:

.o .o
b O)

Suitability Index (M2V3)
0
ix)

 

 

 

 

 

5
Area in Openings/Fields (96)

10

Figure 32. Suitability curves for SPRING FOOD openings components. Suitability of agricultural
crops as deer SPRING FOOD.

138 -

considering the size of the fields, the percent of the evaluation area in agricultural fields, and
the cultivated species. Field size and percent of area in fields is evaluated identically to
opening size and percent of area in openings as discussed above (MN. and M2V3, Figs. 31-
32, respectively). Crop species is the third agricultural variable (M,V,, Fig. 32). Three
crops likely to be available to deer in the UGLR in spring include alfalfa, winter wheat, and
various hays. All are assigned a suitability value of 1.0. (Fig. 32), as all are likely to be
green and actively growing during the critical spring period. All other crops have a
suitability of 0.0.

The suitability of a field for SPRING FOOD associated with agricultural
production is also evaluated using a reduction function:

SIAG = MN. * MN. * M.V,.
The overall suitability of an evaluation area for SPRING FOOD associated with agricultural
production is determined by area weighing:
SIM = SUM... ....|(Sl.)*(area of j/total area in ag types)].

The final component of spring foods, forest ground layer production, can be evaluated by
modifying the Slums; determined for FALL/WINTER FOODS by MZV, (Fig. 33). A forest
stand with at least 50% ground cover is assumed to contribute to spring food production.
Below 50% ground cover too little is contributed, so the SPRING FOOD index value is set at
0.0. The forest floor (FF) contribution to SPRING FOOD is then calculated in the following
manner:

SI...- = SUM... ....l(SI....owsfii. * M,V.)(Proportional area of 1)].
Total SPRING FOOD HSI for the evaluation area is then calculated by summing
the 3 SPRING FOOD components in the following manner:

HSISF = SIOPEN + 81,... + (0.2 * SI”).

139

 

Suitability Index (M2V5)
.0 .0 .0
b O) on

.0
N

 

 

 

0 20 40 60 301100
Forest Grbund Cover (96)

Figure 33. SPRING FOOD suitability of forested stands.

140
Forest stands are assumed to contribute much less to SPRING FOOD suitability than do
openings or agricultural areas; hence, the 0.2 modifier. If the summation of the 3 SPRING
FOOD components totals > 1.0, the value is taken as 1.0.
MODEL 3: SECURITY COVER (SC)

Deer need security cover as protection from predators and man (Boyd and
Cooperrider 1986). Security cover for deer in the UGLR is principally associated with
forested and shrub/sapling stands.

Security cover attributes associated with shrub/sapling stands are primarily a
function of the density of woody stems. In the UGLR, shrub/sapling stands with woody
densities 23000 stems/ha will provide optimal security cover and have a suitability value of
1.0 (M3V., Fig. 34). As mean woody stem densities decrease below 3000/ha, the quality of
security cover declines. Patchiness in regeneration/stocking make some areas with a mean of
<3000 stems/ha suitable hiding areas, however, so suitability of a shrub/sapling stand as
SECURITY COVER remains greater than 0.0 until woody stem densities decline to mm.

The quality of security cover provided by forested stands (mean overstory dbh
210.2cm) is a property of stand density and tree size (Armbruster et al. 1987). Smaller trees
require higher densities to produce the same degree of concealment as larger trees. The
relationship between tree size (represented by total basal area) and tree density and associated
suitability values are shown in M3V2, (Fig. 34).

The third factor important in assessing the SECURITY COVER attributes of an
evaluation area deals with the total amount of SECURITY COVER present. Thomas et al.
(1979) felt that at least 40% of an evaluation area should be in security cover for optimal deer
habitat in the Blue Mountains of Oregon. This model likewise assumes that 240% of an

evaluation area should be in security cover types for deer habitat to be optimal (MN... Fig.

O

P
a:

Suitability Index (M3V1)

Suitability Index (M3V3)

be

P

.0
N

.0
cc

 

#1

 

.°
CD

9
A

.°
N

0

- J

500

Shrub/sapling Density (Stems/ha)

1

1,500

Stand Density/Size
Relationship (M3V2)

2.000

 

2.500

 

3.000

 

 

 

 

 

0

Area in Security Cover Types (96)
Suitability curves for deer SECURITY COVER variables.

Figure 34.

40

60

141

i

.5
Lb
N
01

Tree (>4" DBH) Density (he

.s
O
O
o

1.250 ‘

875

750 ’-

625 1

500 r

375
250
125

V i

 

 

 

0.70

0.30

1

 

 

23 A 46 69
Basal Area (sq. mlha)

92

142
34). As the amount of security cover in an evaluation area drops below 40%, habitat quality
declines, as deer lack adequate protection from predation, hunting, and disturbance.

‘ SECURITY COVER in this model is assumed to be an attribute of shrub/sapling
stands and forested stands only. Security cover quality is evaluated on a stand by stand basis
utilizing MN. for shrub/sapling stands and MN2 for forested stands. The overall 81 for
SECURITY COVER for the entire evaluation area is determined by summing the products of
each individual stands quality value times the proportion that particular stand contributes area-
wise to the area grand total of all SECURITY COVER areas:

Slsc =SUM... .0 I,l(SI_,,T,.....,i_)"‘(area of j/total area in security cover)].
The overall HSI for deer SECURITY COVER can then be determined by taking the 51“
value calculated above and modifying it by the percentage of the evaluation area that is in
SECURITY COVER types (MN3) (Fig 34):
HSISC = 815C * MN...
MODEL 4: THERMAL COVER (TC)

This model assumes that thermal cover is necessary and utilized every winter by
deer in the UGLR, and that it is the duration of use (and not use itself) which is determined
by winter severity. This assumption is undoubtedly valid in the northern parts of the UGLR.
It is possible that in southern areas, however, thermal cover may not be needed each year.

Habitat suitability in terms of thermal cover can be evaluated in 2 manners with
this model. The first method involves simply defining what constitutes adequate deer thermal
cover. The second methodology involves assessing the individual quality of each thermal

COVCI' area .

143

fiiithl

This model defines thermal cover as any conifer stand (pole size or larger) with
275% softwood crown closure that is 22 ha in size (Thomas et al. 1979). The
presence/absence of such stand(s) is determined on the evaluation area utilizing a distance
relationship for suitability assessment (M,V., Fig. 35). If an area(s) satisfying the above
definition is located on or within 3.2km of the evaluation unit, the HSI for THERMAL
COVER is 1.0. Habitat quality declines as the distance to suitable wintering areas exceeds
3.2 km. Deer in the UGLR will migrate extreme distances to wintering areas--often >48 km
(Dahlberg and Guettinger 1956, Rongstad and Tester 1969, Verme 1973). Such extreme
migrations are not indicative of optimal habitat, however. That such migrations do occur. ‘
however, accounts for the suitability value of 0.1 applied to all thermal cover areas > 8 km
distant.
Ii A sm M h 1
The second method of thermal cover evaluation involves quality assessment of
each individual thermal cover area. This model assumes that 5 variables can adequately
evaluate the quality of deer wintering areas:
(1) Percent conifer canopy closure
(2) Site index for northern white cedar
(3) Size of area
(4) Basal area
(5) Dominant overstory species
Conifer crown closure is assumed to be optimal at 75% (MN,, Fig. 35) (Thomas
et al. 1979). Below 25% conifer crown closure, a stand is assumed to be unsuitable as deer

thermal cover due to loss of snow reduction and wind reduction attributes (Fig. 35).

144

 

 

 

 

 

1
$0.8
V
.3.

go.6~
'D
5
£04
3
a
§
(00.2
1
00 A 2 4 A 6 A 8

Distance to Thermal Cover (km)

 

 

 

 

 

1
g 0.8
v
E
g 0.6
t
E
g 0.4
.3
£2
5‘, 0.2
o ‘ - a m 1
0 20 40 60 80 100
1 Conifer Canopy Cover (96)
’g‘ 0.8
v
E
g 0.6
'D
S
g 0.4
3
fl
§
a) 0.2

 

 

 

 

o0L2‘4L6‘13 10

Site Index (Northern White Cedar)(m)

Figure 35. Distance relationship for THERMAL COVER. Suitability curves for THERMAL
COVER variables.

145 °

Site index gives a relative assessment of the fertility of an area. This model
assumes that a site index of 210m for northern white-cedar is optimal, as a site of this
fertility should allow large full canopied trees to develop, maximizing snow interception and
wind reduction (MN3, Fig. 35). Below site index 10m, habitat suitability declines as trees
are apt to be smaller and less fully crowned.

Deer yards of 50 ha or larger were found to be optimal in New Hampshire
(Weber et al. 1983). This model assumes that areas of similar size will be optimal in the
UGLR (M,V., Fig. 36). Areas of this size will winter large numbers of deer and will be
easily locatable for migratory deer. Smaller areas are sub-optimal but still provide winter
cover (Fig. 36).

Deer benefit from high total basal area in a yard via enhanced wind blocking
attributes and forage availability. However, basal areas that are too high result in a lack of
small openings (glades), which are important in food production and as ”hot spots" where
deer can benefit from direct solar radiation (Weber et al. 1983). This model assumes that
basal area is optimal between 40 and 60tha (81 = 1.0: MN,, Fig. 36). Above 60m2/ha
lack of small openings results in less optimal conditions due to lack of the features noted
above. Below 40m2/ha, stands may be insufficiently dense to have optimal wind blockage
ability. Stands below 20m2/ha are too open to provide adequate deer THERMAL COVER
(Fig. 36).

The final variable in THERMAL COVER quality assessment is the dominant
overstory conifer species. Northern white-cedar is considered optimal for deer winter habitat
in the UGLR due to its excellent snow and wind limitation characteristics as well as its forage
value (SI = 1.0; MN.,, Fig. 36) (Blouch I984). Spruces, hemlock, and balsam fir are less

optimal as THERMAL COVER ($1 = 0.7); they do not provide the excellent forage

P
at

O

Suitability Index (M4V4)
2:.

.°
N

P
m

Suitability Index (M2V6)

P
co

.0
at

p
A

P
N

0

 

 

A

 

A

75

Stand Size (ha)

 

100

 

i

p—

1.

.Northern White Cedar 1.0

.Spruce,Fir,Hemlock 0.8

[Upland Pines 0.4

 

A

I

 

 

Dominant Overstory Conifer Species

Figure 36.

146

O

Suitability Index (_M4V5)
25

.0
N

.0
co

9
at

 

 

I

 

 

0

20

Suitability index curves for THERMAL COVER.

40

60

80

A 100
Total Basal Area (sq. mlha)

147
associated with cedar, although their snow and wind blockage characteristics are excellent.
Upland pines will be utilized as THERMAL COVER by deer if cedar and other swamp
conifers are absent. Upland pines lack the weather modification and/or forage values of the
other types, however, and are thus of lower suitability (0.4).

The quality of an area as THERMAL COVER for deer is calculated as a reduction
function of the above 5 variables, as each is considered equally important in determining the
quality of an area as deer THERMAL COVER. The quality suitability index is thus
determined by:

SI... = MN2 * MN, * MN, * MN. * MN.,.
The HSI for deer THERMAL COVER is then determined by modifying the 8ch calculated I
above by the distance function (MN., Fig. 35) discussed under DEFINITION
METHODOLOGY in the following manner:
HSITC = SI... * MN..
WHITE-TAILED DEER HABITAT SUITABILITY

Habitat suitability for white-tailed deer is determined using a minimum function
relationship among the 4 life requisite HSIs described above. The overall white-tailed deer
HSI for an evaluation area is the lowest HSI calculated for any of the 4 critical habitat
requisites, i.e.,

HSlm = min {HSIM, H815... HSI“, HSITC}.
Determination of deer habitat suitability in this manner makes the assumption that the 4
critical life requisites modelled are non-compensatory. Additionally, assessing deer habitat
suitability in this manner allows identification of which life requisite is most limiting in the
evaluation area. This can greatly aid in land management efforts by indicating habitat

shortcomings in the evaluation area. Future management efforts could then be tailored to

148
correct the limiting factors.
LITERATURE CITED

Aldous, SE. 1941. Deer management suggestions for northern white cedar types. J.
Wildl. Manage. 5:90-94.

Aldous, SE, and CF Smith. 1938. Food habits of Minnesota deer as determined by
stomach analysis. Trans. N. Am. Wildl. Conf. 3:756-767.

Armbruster, M.J., and W.R. Porath. 1980. White-tailed deer mum virginiaggs).
Pages 12-26 in T.S. Basket (ed). A handbook for terrestrial habitat
evaluation in Central Missouri. U.S.D.I. Res. Pub. No. 133.

, R.S. Lunetta. J.B. Haufler, RJ. Seppala, and RA. Steinbach. 1987. A white-
tailed deer habitat model for agricultural areas in the Saginaw River
basin. Draft Tech. Rep., Waterways Exp. Stat., Corps of Engineers.
Vicksburg. MS.

Blouch, RI. 1984. Northern Great Lakes States and Ontario forests. Pages 391-410 jg
L.l(. Halls (ed). White-tailed deer ecology and management.
Stackpole Books, Harrisburg, PA.

Boyd, RS, and A.Y. Cooperrider. 1986. Ungulates. Pages 519-564 in A.Y.
Cooperrider, R.S. Black, and H.R. Stuart (eds). Inventory and
monitoring of wildlife habitat. U.S. Dept. Int., Bur. Land Manage.
Service Center. Denver, CO.

Coblentz, BE. 1970. Food habits of George Reserve deer. J. Wildl. Manage. 34:535-
540.

Crawford. HS, and R.L. Marchinton. 1989. A habitat suitability index for white-tailed
deer in the Piedmont. S. J. Appl. For. 13:12-16.

Dahlberg, BL, and RC. Guettinger. 1956. The white-tailed deer in Wisconsin.
Wisconsin Con. Dept. Tech. Wildl. Bull. No. 14. 282pp.

Euler, D., and L. Thurston. 1980. Characteristics of hemlock stands related to deer use
in east-central Ontario. J. Appl. Ecol. 17:1-6.

Gladfelter, H.L. 1984. Midwest agricultural region. Pages 427-440 i_n_ L.K. Halls (ed).
White-tailed deer: ecology and management. Stackpole Books,
Harrisburg, PA.

Grigal, D.F., L.F. Ohmann. and NR. Moody. 1979. Nutrient content of some tall
shrubs from Northeastern Minnesota. USDA For. Serv. Res. Pap.
NC-168. 10pp.

149

Halls, L.l(. 1978. White-tailed deer. Pages 43-65 jg J.L. Schmidt and D.L. Gilbert
(eds). Big game of North America: ecology and management.
Stackpole Books, Harrisburg, PA.

Harper, J.A. 1969. Relationship of elk to reforestation in the Pacific Northwest. Pages
67-71 jg H.C. Black (ed). Wildlife and reforestation in the Pacific
Northwest. Oregon St. Univ. Press, Corvallis.

Healy, WM. 1971. Forage preferences of tame deer in a northwest Pennsylvania
clearcutting. J. Wildl. Manage. 35:717-723.

Johnston, W.F. 1977. Manager’s handbook for northern white-cedar in the north central
States. USDA For. Serv. Gen. Tech. Rep. NC-35. 18pp.

Jordan, P.A., D.B. Botkin, A.S. Dominski, H.S. Lowendorf. and GE. Belovsky. 1974.
Sodium as a critical nutrient for the moose of Isle Royale. Proc. N.
Am. Moose Conf. Workshop 9:13-42.

Kelsall, JP. 1969. Structural adaptation of moose and deer for snow. I. Mammal.
50:302-310.

Kohn, BE, and JJ. Mooty. 1971. Summer habitat of white-tailed deer in north central
Minnesota. J. Wildl. Manage. 35:476-487.

Kramer, A. 1970. Notes on the winter ecology of the mule and white-tailed deer in the
Cypress Hill, Alberta, Canada. Can. Field Nat. 85:141-145.

Marchinton, R.L., and DH. Hirth. 1984. Behavior. Pages 129-169 jg L.K. Halls (ed).
White-tailed deer: ecology and management. Stackpole Books,
Harrisburg, PA.

Mautz, W.W. I978. Sledding on a brushy hillside: the fat cycle in deer. Wildl. Soc. Bull.
6:88-90.

McCafferty, K.R., and W.A. Creed. 1969. Significance of forest openings to deer in
northern Wisconsin. Wisconsin Dept. Nat. Res. Tech. Bull. No. 44.

104pp.

, J. Tranetzke, and J. Piechura, Jr. 1974. Summer foods of deer in northern
Wisconsin. J. Wildl. Manage. 38:215-219.

McNeill, R.E., Jr. 1971. Interactions of deer and vegetation on the Mid-Forest Lodge
and Gladwin Game Refuge. PhD Thesis. Univ. Michigan, Ann
Arbor. 214pp.

Nixon, C.M., M.W. McClain, and KR. Russell. 1970. Deer food habits and range
characteristics in Ohio. J. Wildl. Manage. 34:870-886.

150

Nowosad. R.F. 1967. An ecological study of important deer wintering areas in Nova
Scotia. M.S. Thesis, Acadia Univ, Acadia, Nova Scotia. Canada.
l98pp.

Pierce, D.E., Jr. 1975. Spring ecology of white-tailed deer in north central Minnesota.
M.S. Thesis, Univ. Minnesota, St. Paul. 55pp.

Reynolds, H.G. 1966. Use of openings in spruce-fir forests of Arizona by elk, deer. and
cattle. USDA For. Serv. Res. Note RM-66. 4pp.

Rogers, LL, 1.]. Mooty, and D. Dawson. 1981. Foods of white-tailed deer in the
Upper Great Lakes Region--a review. USDA For. Serv. Gen. Tech.
Rep. NC-65. 24pp.

Rongstad, 0.1., and LR. Tester. I969. Movements and habitat use of white-tailed deer
in Minnesota. J. Wildl. Manage. 33:366-379.

Thomas, J.W., H. Black, Jr., R.J. Scherzingcr, and RJ. Pedersen. 1979. Deer and elk.
Pages 104-127 jg J.W. Thomas (ed). Wildlife habitats in managed
forests: the Blue Mountains of Oregon and Washington. USDA For.
Serv. Agric. Handb. No. 553.

Ullrey, D.E., W.G. Youatt, H.E. Johnson, P.K. Ku, and L.D. Fay. 1964. Digestibility of
cedar and aspen browse for the white-tailed deer. J. Wildl. Manage.
28:791-797.

, L.D. Fay, B.E. Brent, and KB Kemp. 1967. Digestibility of cedar and
jack pine browse for the white-tailed deer. J. Wildl. Manage. 31:448-
454.

 

. , and _. 1968. Digestibility of cedar and balsam fir
browse for the white-tailed deer. J. Wildl. Manage. 32: 162-171.

Verme, LJ. 1963. Effects of nutrition on growth of white-tailed deer fawns. Trans. N.
Am. Wildl. Conf. 28:431-443.

. 1965. Swamp conifer deeryards in northern Michigan: their ecology and
management. J. For. 63:523-529.

. 1969. Reproductive patterns of white-tailed deer related to nutritional plane. J.
Wildl. Manage. 33:881-887.

. 1973. Movements of white-tailed deer in upper Michigan. J. Wildl. Manage.
37:545-552.

. and DE. Ullrey. 1972. Feeding and nutrition of deer. Pages 275-291 jg D.C.
Church (ed). Digestive physiology and nutrition of ruminants. Vol.
111. Oregon St. Univ. Press, Corvallis.

151

Weber, S.J., W.M. Mautz, J.W. Lanier, and J .E. Wiley, Ill. 1983. Predictive equations
for deer yards in northern New Hampshire. Wildl. Soc. Bull. 11:331-
338.

APPENDIX E

Habitat Suitability Index Model for the Eastern Bluebird in Northern Michigan
Raymond G. Moses, Gary J. Roloff, and Jonathan B. Haufler
Michigan State University, East Lansing, MI

HABITAT USE INFORMATION
Metal

Eastern bluebirds (Sialia sialis) range from east of the Rocky Mountains north to
southern Canada, and south into El Salvador and Honduras in North America (Bent 1939). ‘
Wintering range is usually southward of New York and Ohio in the east, and into southern
Kansas and Texas in the west. Eastern bluebirds in Michigan inhabit 01d fields, and a variety
of lumbered and burned vegetation types (Pinkowski 1979).

Serious declines in bluebird populations have occurred in Michigan. as is the case
throughout its range. Wallace (1959) noted the rarity of bluebirds statewide, and that they
were largely restricted to jack pine (Pinus banksiana) areas of the northern counties and to
more remote and abandoned or uncultivated farmlands.

Habitat loss and competition with exotic introductions of house sparrows (Passer
domesticus) and European starlings (Sturnus vulgaris) associated with developed areas, have
often been cited as the major cause for decline (Bent 1939). Agricultural lands provide vast
open areas that are unsuitable habitat and abandoned farmland may revert to unusable, dense
second growth timber (Conner 1974). In view of public interest in bluebird conservation over
recent years, it is important that key habitat requirements for this species be identified and
systematically managed.

152

153

The eastern bluebird is primarily insectivorous in feeding habits but will also feed
on plant material. Over half of the adult diet is comprised of Orthoptera, Lepidoptera,
Arachnida, and Coleoptera (Pinkowski 1979). During the non-breeding season, eastern
bluebirds often feed on fruit of staghorn sumac (Rhus ryphina) and smooth sumac (Rhus
glabra)(Pinkowski 1979).

Beal (1915) found, from stomach content analysis, that approximately 68% and
32% of the diet consisted of animal and vegetable matter, respectively. Orthoptera accounted
for the majority of prey items in every month, comprising more than 50% of the diet in
August and September. Coleoptera were the second most important food source, accounting
for 22% of the diet. Other important prey items included Lepidoptera, Hymenoptera, and
Hemiptera.

Pinkowski (1978) studied feeding of nestling bluebirds in Michigan and found
Lepidoptera larvae (32 %). Orthoptera (26%), and Arachnida (11%). Early maturing
fruit,mainly apples (Malus spp.), brambles (Rubus spp.), dogwoods (Camus spp.), cherries
(Prunus spp.), and honeysuckle (Lonicera spp.) comprised a small portion of the nestling diet
for a seasonally short time.

As summer progressed and vegetation increased in height, ground dwelling prey
become increasingly more difficult to catch. In June. bluebirds switch to phytophilous prey
species by foraging from perches greater in height (Pinkowski 1979).
flats:

No information pertaining to water requirements of eastern bluebirds was found.
It is assumed that water requirements are met from food items and free water sources where

available.

154
W

The eastern bluebird is a secondary cavity nester, preferring to nest in cavities of
dead trees in park-like situations on dry sites producing sparse ground cover (Pinkowski
1979). Conner (1974) found bluebirds nesting in Virginia clearcuts in standing dead snags
averaging 23.6 cm diameter breast height (DBH) and 134 m from woodland edges. Openings
to nest cavities ranged from 4-12 cm in diameter. Most nests were constructed by common
flickers (Colaptes auratus). Conner (1974) observed that suitable nesting habitat can be
provided if dead snags are left standing in clearcuts to up to 12 years following harvest.

Pinkowski (1976). on the Huron National Forest of Michigan, found that 89% of
nests examined were in dead trees. Tree species used were in proportion to availability with
60% in oaks (Quercus spp.), 37% in pines, and one each in white birch (Berula papyrlfera)
and apple. Abandoned woodpecker holes accounted for 66% of the nests, 3% were produced
from natural decay and fire. Cavity height averaged 3.6 m, entrance diameter averaged 6.9
cm, cavity depth was 22 cm, and interior diameter averaged 10 cm. Most (80%) excavated
cavities were created by common flickers, 15% by hairy woodpeckers (Picoides villosus), and
5% by downy woodpeckers (Picoides pubescens).

Even-aged timber management eliminates nest sites and foraging perches required
by bluebirds (Pinkowski I976). Pinkowski (1976) recommended that trees > 15-20 cm DBH
provide optimum cavity potential, and that management benefitting woodpeckers should also
benefit bluebirds.

Iceman

Eastern bluebirds maintain a breeding territory for mating, nesting, and feeding,

and may decrease territory size during the breeding season (Pinkowski 1979). In the Huron

National Forest, breeding territories ranged from 0.8-2.9 ha, with a mean of 1.4 ha. Nest

155
densities averaged 7.9/100 ha. the highest nest densities (9.6/ha) were found in an area that
had burned IO-years earlier (Pinkowski 1976).
i r i n

Conner (1974) suggested that snags left in clearcuts provide suitable habitat for
nesting eastern bluebirds. Of equal importance may be the presence of sparse ground cover
and optimal perch distribution so that bluebirds can efficiently utilize drop foraging tactics
(Pinkowski 1979).

Studies by Pinkowski (1979) make reference to prey abundance. perch
distribution, and ground cover as important habitat variables for bluebirds. Additionally, a
study (Pinkowski 1974) demonstrated that management for woodpeckers may be of critical ‘
importance. Woodpecker management would apparently rely on basal area and snag
availability.

Land management practices involved in perpetuating bluebird habitat include
logging and burning. In jack pine stands in northern Michigan, eastern bluebirds have a
greater preference for burned areas over timbered tracts (Pinkowski 1977). Land use
practices that involve prescribed burning of thinned jack pine will be most beneficial to
bluebirds.

HABITAT SUITABILITY INDEX MODEL
i Ar

This model was developed for the central and northern lower peninsula of

Michigan.
Seams
This model was developed to evaluate breeding habitat for the eastern bluebird;

late-March through August.

156

Cat/SLIM

This model is restricted for use in the deciduous forest (DF), evergreen forest
(EF), deciduous shrubland (DS), and evergreen shrubland (ES) cover types of undeveloped
uplands (terminology follows that of U.S. Fish and Wildlife Service 1981). This model is not
applicable to wetland or agricultural cover types.
Mi 1 1

Minimum habitat area is defined as the minimum amount of contiguous habitat
required for a species to live and reproduce in an area. Minimum habitat area for bluebirds
was not specified in the literature, thus, it is assumed that an area of 0.8 ha must be present
or the habitat suitability index (HSI) will equal 0.00.
V ri l i vel

This model will be evaluated with data collected from the ”Ecological
Classification System-Wildlife Habitat Model" project on the Huron National Forest,
Michigan.
M ri i 11

mm

This model evaluates the ability of habitat to meet the cover/reproduction
requirements of the eastern bluebird and should be used as an overall indication of breeding
habitat suitability. Reproductive requirements are assumed to be the most limiting factor.
Therefore, food and cover requirements are assumed to be satisfied by cover types meeting
the reproductive requirements. The cover/reproduction component of the model evaluates
habitat suitability through measurements of vegetative structure including basal area. number
of snags 220 cm dbh, and percent ground cover.

The following sections provide documentation of the H81 for the eastern bluebird.

157

Following are sections describing life requisites, definitions of habitat variables, and
suitability levels and the relationships between variables.
v r/ r i

Eastern bluebirds are secondary cavity nesters that have been found to nest in
snags 220 cm dbh in low stocked, semi-open vegetation types in north-central Michigan.
Optimal habitats are assumed to contain at least 5 snags/ha and 0.5-3 milha basal area with a
relatively open, park-like understory. The requirement for basal area can never equal 0.00
due to the tendency of bluebirds to nest in open fields providing a cavity is available.
Bluebirds, during the breeding season are almost exclusively insectivorous, preferring to
forage in areas with sparse ground vegetation. Optimum habitat is assumed to contain
between 60-90% ground cover. Figure 37 illustrates the relationships between the habitat

variables and habitat suitability. The suitability index values are related in equation (1).

Equation 1:

 

[(srmxSImeerw +11ij
2

D in i n

Habitat quality is expressed as a result of equation I, with 0.00 and 1.00
representing low and high quality habitat, respectively. If any of the SI values equals 0.00.
then the HSI equals 0.00 and the habitat is considered to be unsuitable breeding habitat. The
HSI value from equation 1 represent the capability of habitat to support breeding pairs of
eastern bluebirds.

Willa
Listed in table 22 are definitions, applicable cover types, and recommended

sampling techniques for habitat variables used in the computation of bluebird HSl’s.

 

Suitability Index

0.2 *

 

 

o 2 I A l x I J A v

0 1 2 3 4 5
Total Basal Area (sq. mlha)

 

.0 .o
05 O

Suitability index
0
as

0.2 .

 

 

 

60 A 80 A 100
Ground Cover(96)

4A;

 

0 20 40

158

 

.0 .0
09 O

Suitability index
0
a

0.2

 

I

2

I

3
No. Snags >20 cm DBHIhe

 

i L

4

 

 

Suitability index
.0 .o .o
A 0’ O

.0
re

 

 

 

0

0

500

1 .000

1 .500

2.000

2.500

3.000

No. Woody Stems <11.4 cm DBHIha

Figure 37. Vegetative variables and habitat suitability index (HSI) relationships for the eastern

bluebird.

159

Table 22. Definitions of habitat variables and associated measuring technique.

 

HSI Variable & Definition

Cover Types‘

Sampling Technique

 

Basal Area

Area if exposed stems of
woody vegetation if cut
horizontally at 1.4 m (4.5 ft)
from the ground.

Snags 220 cm

Standing dead trees 220
cm dbh at 1.4 m (4.5 ft)
tall.

Saplings

The number of woody
stems < 11.4 cm (4.5 in)
dbh.

Ground Cover

Percent of the ground that
is covered by vegetation s 1
m tall.

 

EF,DF,ES,DS

EF,DF,ES,DS

EF,DF,ES,DS

EF,DF,ES,DS

10 x 25 m Plots: Basal
Area Prism or Tube

10 x 100 m Belt Transects

2 x 25 m Belt Transects

Line Intercept

 

160

Literature Cited

Beal, PE. 1915. Food of the robins and bluebirds of the United States. U.S. Dept. Agr.
Dept. Bull. 171.

Bent, A.C. 1939. Life histories of North American thrushes, kinglets, and their allies.
U.S. Natl. Mus. Bull. 196. 452pp.

Conner, R.N. 1974. Eastern bluebirds nesting in clearcuts. J. Wildl. Manage. 38:934-
935.

Pinkowski, BC. 1974. A comparative study of the behavioral and breeding ecology of
the eastern bluebird (Siali'a siall's). Ph.D. Thesis, Wayne State Univ.,
Detroit. 47lpp.

_. 1976. Use of tree cavities by nesting eastern bluebirds. J. Wildl. Manage. 40:556-
563.

1977. Notes on the effeCts of fire and logging on birds inhabiting jack pine stands. .
J. Pine Warbler 55:92-94.

1978. Feeding of nesting and fledgling eastern bluebirds. Auk 90:84-98.
. 1979. Foraging ecology and habitat utilization in the genus Sialia.

U.S. Fish and Wildlife Service. 1981. Standards for the development of habitat
suitability index models. U.S. Fish Wildl. Serv. Div. Ecol. Serv. 103
ESM. n.p.

Wallace, GJ. 1959. The plight of the bluebird in Michigan. Wilson Bull. 71:192-193.

APPENDIX F

A Habitat Suitability Index Model for the Upland Sandpiper
in the Huron National Forest, Michigan

Gary J. Roloff and Jonathan B. Haufler
Michigan State University, East Lansing, Michigan

HABITAT USE INFORMATION
General

The upland sandpiper (Bartramia longicauda) is a migratory bird which originally
bred throughout the grasslands of the United States and Canada (Bent 1929, Snyder et al.
1987). Currently, upland sandpipers breed from the northeastern coast of the United States to
eastern Washington and Oregon and northward to Alaska and south into Oklahoma (Barrows
1912, Bent I929). The upland sandpiper over-winters in the pampas of South America (Bent
1929).

Formerly abundant across most of the United States and Canada in the mid- to
late-1800’s, populations declined rapidly after 1900 (Barrows 1912). Factors that caused the
population decline were the continuous cultivation of the native prairies (Graber and Graber
1963), market hunting in the United States and Canada (Cooke 1914), and unlimited harvest
in South America (Beck 1937). In 1916. the Migratory Bird Conservation Act ended legal
hunting of the upland sandpiper in the United States and Canada and populations increased
(Mitchell 1967, Kirsch and Higgins 1976). Additionally, upland sandpipers appeared to adapt

to changing habitat conditions and nested in cultivated fields when unbroken prairies were no

161

162
longer available (Bailey 1930). However, populations have never recovered to the pre-l900
levels. Factors that appear to be restricting post-1916 population growth include habitat
changes on the wintering grounds (Osborne and Peterson 1984) and habitat loss in North
America, specifically the maturation and replacement of old fields (Tate 1986). Ailes (1980)
also suggested that adverse weather conditions during the breeding season may account for
annual population fluctuations.

Current breeding populations of the upland sandpiper in the United States are
localized, and much of the breeding range has been reduced (Johnsgard 1981, Osborne and
Peterson 1984). As of 1977, no upland sandpiper populations were known to be increasing
and there were no known improvements in habitat of significant scale (Jurek and Leach 1977). .
In Michigan, as in other states, prairies have been eliminated (Stearns and Lindsey 1977) but
local populations of upland sandpipers appear stable (Wallace 1977). Powell (1981) reported
relatively good numbers of sandpipers in 13 Michigan counties.

Food 11 Feedin

The upland sandpiper forages extensively on arthropods (phylum Arthropoda) but
seeds are also an important food in some locales (Bates 1907, Forbush 1912, McAtee and
Beal 1912, Bent 1929). McAtee and Beal (1912) examined 163 stomachs from upland
sandpipers and found that 97% of the diet was animal matter. Grasshoppers (Locustidae),
crickets (Gryllidae), and weevils (Curculioninae) occurred in greatest quantities (McAtee and
Beal 1912).

Vegetable matter in the diet of upland sandpipers consists of the seeds of
buttonweed (Abutilon theophrasn’), foxtail grass (Seraria spp.), and sandburs (Cenchrus spp.;
McAtee and Beal 1912). Forbush (1912) reported that the upland sandpiper fed extensively

on crowberries (Empetrum nigrum) in Labrador and Bates (1907) reported that waste grain

163
was an important food in Nebraska.

The use of foraging sites by upland sandpipers is influenced by vegetation height.
Pastures that are continually grazed by cattle are preferred over sedge-grass meadows,
plowed-seeded fields, ungrazed pastures, and hayfields (Dorio and Grewe 1979, Ailes 1980).
Dorio and Grewe (1979) found that the sedge-grass meadow was used as a foraging site only
during May (corresponding to low vegetative heights) whereas the grazed pastures were used
in May and June. Ailes (1980) reported that feeding areas during late summer were similar to
those used throughout the summer. Upland sandpipers used foraging areas that had vegetation
heights from 2.5 cm to 15.0 em. but abandoned foraging areas when vegetation height was
over 30.0 cm (Dorio and Grewe 1979). Ailes (1976) had similar results, noting that upland
sandpipers foragcd in fields with vegetation 5 cm to 10 cm tall and abandoned the fields when
vegetation height reached 15 cm.

Upland sandpipers apparently used the grazed pasture-type fields because they
were more successful in finding insects in low vegetation (Buss and Hawkins 1939, Bowen
1976, Dorio and Grewe 1979). Additionally, added peripheral visibility in low vegetation
may account for the preference (Dorio and Grewe 1979). Fields that are frequently (more
than once per year) cropped or mowed provide good foraging areas because the availability of
the preferred habitat is extended (Dorio and Grewe 1979).
flag;

No information pertaining to the water requirements of upland sandpipers was
found in the literature. It is thus assumed that the water requirements of upland sandpipers
are satisfied by localized water sources or diet. Therefore, water was not considered as a

habitat limiting factor across the upland sandpiper’s breeding range.

164
Cover / Nesting and Inggbatign Requirement;

Nesting activity of the upland sandpiper typically begins about two weeks after the
first birds arrive from South America (typically early- to mid-April: Lindmeier 1960, Ailes
1980). Upland sandpipers separate into pairs and select their nesting areas after arrival to the
breeding ground (Buss and Hawkins 1939, Bowen 1976), and colonial nesting is not
uncommon (Buss and Hawkins 1939). Data suggest that upland sandpipers will re-nest if
their first nest is destroyed (Buss and Hawkins 1939, Ailes 1980). In most areas, the
principle cause of nest failure is mammalian predation (Kirsch and Higgins 1976,
Buhnerkempe and Westemeier 1988), although Ailes (1980) reported high losses from
livestock trampling. The typical clutch size is four, however clutches of three and five have
been documented (Buss and Hawkins 1939, Lindmeier 1960, Kaiser 1979, Ailes 1980,
Buhnerkempe and Westemeier 1988).

In general, upland sandpipers prefer an open, climax prairie-type habitat for
nesting that includes native prairie tracts, old fields, pastures, and airports (Buss and Hawkins
1939, Higgins et al. 1969, Kirsch and Higgins 1976. Dorio and Grewe 1979, Kaiser 1979,
Ailes 1980, White 1983. Osborne and Peterson 1984. Buhnerkempe and Westemeier 1988).
Several researchers have also reported that upland sandpipers nest in wet meadows or marshy
areas (Barrows 1912, Johnsgard 1981) and in clearings in spruce forests (800thill and Soothill
1982).

In order to initiate nesting, upland sandpipers require several specific habitat
components. The height of the vegetation is an important factor in nest site selection
(Lindmeier 1960, Kirsch and Higgins 1976. Dorio and Grewe 1979, Kaiser i979, Ailes 1980.
Snyder et al. 1987). Kirsch and Higgins (1976), in North Dakota, found 12% of the total

upland sandpiper nests in vegetative cover ranging from 0.0 cm to 15.4 cm tall; 62% of the

165

nests in cover from 15.5 cm to 30.8 cm tall; 25% of the nests in cover from 30.9 cm to 61.5
cm tall; and 1% of the nests in vegetative cover greater than 61.5 cm tall. Similarly, Kaiser
(1979) found that vegetation between 12.7 cm and 63.5 cm tall contained 97% of the upland
sandpiper nests in South Dakota. Forty-two percent of these nests were in vegetation between
12.7 cm and 33.0 cm tall (Kaiser 1979).

Ailes (1980) found 15% of the upland sandpiper nests in vegetative cover 0 cm to
15 cm tall; 31% in cover 16 cm to 25 cm tall; 54% in cover 26 cm to 40 cm tall; and no
nests in vegetative cover that was greater than 40 cm tall in central Wisconsin. Although
vegetation height is important at the time of nest initiation, Ailes (1980) found that at the time
of hatching, vegetation was 70 cm tall in some areas.

Several researchers have suggested that upland sandpipers require a minimum
amount of overhead— and side-concealment around the nests. Vegetation that provides 50% or
more overhead concealment appears to be selected by upland sandpipers for nesting (Kirsch
and Higgins 1976, Kaiser 1979). Only a small portion of upland sandpiper nests had no side
concealment (Buss and Hawkins 1939, Kirsch and Higgins 1976). Kaiser (1979), however.
suggested that overhead concealment is more important than side concealment. Although
vegetative coverage is important in nest site selection, Lindmeier (1960) demonstrated (via
light intensity readings) that coverage surrounding upland sandpiper nests was relatively poor
when compared to other ground nesting species.

Kirsch and Higgins (1976), in an attempt to quantify upland sandpiper nest
occurrence and vegetative coverage, found that 10% of the nests occurred in poor cover (e. g.
overgrazed pastures. feedlots. mowed areas with little regrowth); 76% occurred in fair cover
(e.g. standing stubble, moderately grazed pastures. mowed areas with heavy regrowth,

undisturbed vegetation on poor soil); 13% occurred in good cover (e.g. lightly grazed

166 -
pastures, undisturbed vegetation on good sites); and 1% occurred in excellent cover (e. g.
undisturbed tall grass, grass-forb mixtures on good sites).

Dorio and Grewe (1979) suggested that the selection of nesting habitat by upland
sandpipers indicated that appearance characteristics of the cover rather than species
composition was important. Kirsch and Higgins (1976) noted similar results, finding that
upland sandpipers nested in a variety of vegetation including grasses, forbs, and brush.
However, upland sandpipers appeared to avoid grass-legume mixtures, presumably because
they were too tall for nesting (Kirsch and Higgins 1976, Kaiser 1979, Buhnerkempe and
Westemeier 1988).

Buss and Hawkins (1939) suggested that the proximity of suitable loafing and
feeding sites to the nest is important. Suitable loafing sites include pastures with low
tussocks, prairie meadows with sparse 'vegetation not taller than the plovers back, or
hayfields containing open patches that allow freedom of vision (Buss and Hawkins 1939).
Buss and Hawkins (1939) also suggested that the loafing site must be adjacent to the nesting
area to make it acceptable, however, Ailes (I976) documented daily movements of
approximately 400 m by both male and female during the nesting and incubating period.

A variety of management scenarios provide suitable nesting habitat for the upland
sandpiper. Kirsch and Higgins (1976) found that average nest densities were highest on
grasslands managed by prescribed burning. Similarly, Buhnerkempe and Westemeier (1988)
demonstrated that upland sandpipers selected fields that had been rotary mowed or burned the
previous season, however, upland sandpipers did not selectively nest in burned fields past
their first year of growth. Bowen (1976) found more nests in unburned pastures than
expected based on habitat abundance.

Seeding and management of cool-season grasses appear to be beneficial to upland

167
sandpipers when given time to diversify (Buhnerkempe and Westemeier 1988). Buhnerkempe
and Westemeier (1988) suggested that management of such seeded fields include a three-year
rotation of rotary mowing, no disturbance, and prescribed burning.

Agricultural land use patterns and farming practices have important influences on
nest site selection and hatching success of upland sandpipers (Higgins et al. 1969). Higgins et
al. (1969) found 93% of the upland sandpiper nests in their study in grasses that were either
unused by cows or intermittently mowed for hay. Kaiser (1979) suggested that moderate
spring grazing in which 2040% of the current year’s growth is removed did not restrict
nesting of upland sandpipers. Buss and Hawkins (1939) noted a high percentage of pasture
nests were found in tufts of tall grass fertilized by the previous year’s cow dung. Typically.
few nests are located in annually tilled cropland or in heavily grazed pastures (Buss and
Hawkins 1939, Kirsch and Higgins 1976, Ailes 1980).

Covgr / Broming Requiremgnt§

Upland sandpipers typically prefer close-cropped, open fields for brood-rearing.
Buhnerkempe and Westemeier (1988) observed broods of upland sandpipers in fields of wheat
stubble, recently hayed legumes, and on grazed pastures. Dorio and Grewe (1979) and Ailes
(1980) observed broods in similar habitats.

Good brood rearing habitat is short in height (0 cm to 10 cm, perhaps up to 20
cm), open, and wwdy (Dorio and Grewe I979, Ailes 1980, Buhnerkempe and Westemeier
1988). Such habitat facilitates movements of the chicks and offers an abundance of insects for
food (Buhnerkempe and Westemeier 1988). Broods may also use wind eroded areas within
the nesting habitat for foraging. thus upland sandpiper broods will remain in the natal area if
suitable brood-rearing habitat is available (Dorio and Grewe 1979, Ailes 1980). However,

broods will move considerable distances to reach suitable brood rearing habitat (Dorio and

168
Grewe 1979, Ailes 1980, Buhnerkempe and Westemeier 1988). Ailes (I980) documented
movements by broods of 338 m and 48 m in Wisconsin.

Once the family finds an acceptable brood rearing habitat, daily movements
become short (< 100 m) and within 20 ha (Buhnerkempe and Westemeier 1988). Different
family units will share brood habitat (Ailes 1980).

11111162225100

No quantitative data were available on the minimum area requirements of upland
sandpipers. Bowen (1976) found that nesting pastures ranged in size from 7.3 ha to 40.5 ha.
Although not specified, Bowen’s findings apparently did not include areas for feeding and
loafing, which are important considerations in nest site selection. Ailes (1976) found that
home ranges of male and female upland sandpipers was 8.5 ha and 85.6 ha, respectively.
Presumably, upland sandpipers were obtaining their life requisites within 86 ha, thus this
model assumes that upland sandpiper habitat requirements can be obtained within 86 ha of
contiguous cover type. Typically, the vegetation components necessary during the breeding
season can be found within one cover type.

White (1983) suggested that nesting fields should be large and unbroken, with a
low vegetation ”edge” rating. Upland sandpipers will nest near (6.0 to 23.4 in) field edges.
however, large amounts of edge are not preferred (Dorio and Grewe 1979, White 1983).

Buss and Hawkins (1939) reported a nesting density of one nest per 0.60 ha.
Other researchers have reported densities of one nest per 6.23 ha; one nest per 12.9 ha: and
one nest per 21.3 ha (Kirsch and Higgins 1976, Ailes 1980, Snyder et al. 1987, respectively).
Thus. nest densities are variable and are probably related to nesting habitat quality.

Bowen (1976) found that upland sandpiper populations were spatially clumped

during the nesting season. Buss and Hawkins (1939) also suggested that upland sandpipers

169

were colonial nesters. The relatively high nesting success of upland sandpipers (often greater
than 50%) may be related to social behaviors of mobbing, vocalizations, and the tendency to
concentrate nests (Bowen 1976, Kaiser 1979).

i r 1

Several researchers have suggested the importance of perch sites for upland
sandpiper breeding and security (Bent i929, Buss and Hawkins 1939, White 1983, Snyder et
al. 1987). Upland sandpipers use a variety of perch sites, including fence posts, telephone
poles. stumps, and rocks (Bent 1929, White 1983), thus, it is unlikely that preference for a
particular type of perch limits breeding habitat suitability.

HABITAT SUITABILITY INDEX (HSI) MODEL

Qeograghig grgg. This model was developed for application within the Huron
National Forest located in northeastern Michigan.

5mg. This model was developed to evaluate the breeding season (typically mid-
April through August) habitat of the upland sandpiper.

W. This model was developed to evaluate the habitat quality required
by the upland sandpiper in the following cover types: Cropland (C) and Pasture and Hayland
(PH) cover types of agricultural and built-up lands; Grasslands (G) and Forbland (F) cover
types of undeveloped uplands: and Herbaceous Wetland (HW) and Shore, Bottom Wetland
(SW) cover types of wetlands (U.S. Fish and Wildlife Service 1981).

Minimgm hgbitgt 35;. Minimum habitat area is defined as the minimum amount
of contiguous habitat that is required before a species will live and reproduce in an area. A
quantitative assessment of the minimum amount of contiguous habitat necessary for upland
sandpiper breeding was unavailable. However, estimations could be made from home range

data. If the amount of contiguous habitat is less than 86 ha, the suitability index is assumed

170

to be 0.

Vgrification Igvgl. This model will be verified using the data from the Ecological
Classification System (ECS) developed in the Huron-Manistee National Forest in north-central
Michigan.

I r' i n

mgrvigw. This model will evaluate habitat suitability for the upland sandpiper
during the breeding season. Components of the model include foraging, reproductive, and
brood-rearing requirements, and it is assumed that measurements of these components can
predict the habitat suitability of an area. Foraging and brood-rearing components were
evaluated for habitat suitability through measurements of vegetation heights. The reproductive
components evaluate habitat suitability through measurements of vegetative heights and
vegetative coverage.

The following sections provide documentation of the HSI model for the upland .
sandpiper. Habitat variables of each life requisite and the interrelationships between habitat
variables are discussed.

Cgvgr / gating and inggbatign. Upland sandpipers prefer to nest in short, open grassland
vegetation types. Important habitat variables that influence the selection of nesting sites by
upland sandpipers include vegetation height and vegetation coverage.

Upland sandpipers prefer to initiate nesting in vegetation that is 15 cm to 30 cm
tall. Nest initiation occurs in vegetation that is 0 cm to 15 cm tall, but these heights are not
preferred. Similarly, nest initiation by upland sandpipers occurs in vegetation from 31 cm to
60 cm tall, but these sites are selected less frequently. Upland sandpipers will not
initiate nesting in vegetation that is taller than 60 cm.

Upland sandpipers require at least 50% overhead vegetation coverage of the nest,

171

however, complete coverage is not preferred. Optimal nesting cover contains a diversity of
grasses and forbs. Habitat containing brush clumps will be used for nesting by upland
sandpipers but brushy areas are typically not preferred. Side concealment is of minor
importance in nest site selection.

The minimum area of habitat required for upland sandpiper nesting is 86 ha.
Within this area, colonial nesting is not uncommon. The most concentrated nest density
reported in the literature was one nest per 0.607 ha. Nesting fields should be relatively open
and shaped to minimize edge.

Figure 38 illustrates the relationships between habitat variables and habitat
suitability. The relationship of the habitat suitability index values are depicted as the
geometric mean between nesting height and cover:

HSI=(Vegetation Height Nesting SI*Overhead Coverage 81)"-5
Ecological land type phases (ELTP’s) may be useful predictors of habitat suitability for upland
sandpiper nesting. For example, the lower ELTP‘S (e.g. the 01’s to the 13’s) are typically
less fertile sites and would therefore provide sparser vegetation than a field found on a higher
ELTP. Thus, the lower ELTP’s may not provide the necessary vertical concealment required
by nesting upland sandpipers. Further research is needed to correlate ELTP’s and habitat
suitability for the upland sandpiper.
Walled

All habitat variables are equally weighted in the H81 model. The model will
generate a habitat suitability index between 0 and 1, with 0 representing no suitable habitat
and 1 representing optimal breeding habitat for the upland sandpiper. A summary of the

vegetation variables and recommended sampling techniques is presented in Table 23.

172

 

Suitability Index (V1)

 

 

 

 

0 A10‘20‘30840760960
Ground Vegetation Height (cm)

 

\

Suitability index (V2)

 

 

 

0 A20‘40A60A808100
Ground Cover (96)

Figure 38. Suitability index graphs for the upland sandpiper.

173

Table 23. Summary of model variables, associated vegetation cover types, and suggested
sampling techniques for habitat suitability index determination.

 

Variable (Definition)

Cover Types

Suggested Techniques

 

V. Vegetation Height Nesting
-the average vertical distance
from the ground surface to the
dominant height stratum of the
herbaceous vegetative canopy
during spring.

V2 Percent Overhead Coverage
-the percent of the ground that is
shaded by a vertical projection
of non-woody

vegetation.

Ce PHe G, Fe
HW, SW

C, PH, G, F.
HW, SW

Graduated Rod

Line Intercept
Domin-Krajina Plot
Daubenmire Plot

 

 

 

174
thgtatutg Qitgj
Ailes, I. W. 1980. Breeding biology and habitat use of the upland sandpiper in central
Wisconsin. Passenger Pigeon 42:53-63.
Bailey, A. M. 1930. The upland plover. J. Amer. Mus. Nat. Hist. 30:177-181.

Barrows, W. B. 1912. Michigan bird life. Special Bull. Dept. Zool. and Physiol.
Michigan Agric. Coll. Wynkoop Hollenbeck Crawford Co., Lansing,
MI. 822 pp.

Bates, J. M. 1907. The Bartramian sandpiper. Bird-lore 9:84.
Beck, H. H. 1937. Status of the upland plover. Cardinal 42163-166.

Bent, A. C. 1929. Life histories of North American shorebirds (Part 2). Bull. U.S. Nat.
Mus. 14621-412.

Bowen, D. E. Jr. 1976. Coloniality. reproductive success. and habitat interactions in
upland sandpipers (gm lgngigggda). PhD Diss., Kansas State
Univ., Manhattan. KS. 127 pp.

Buhnerkempe, J. E., and R. L. Westemeier. 1988. Breeding biology and habitat of
upland sandpipers on prairie chicken sanctuaries in Illinois. Trans.
Illinois Acad. Sci. 81:153-162.

Buss, I . O., and A. S. Hawkins. 1939. The upland plover at Faville Grove, Wisconsin.
Wilson Bull. 51:202-220.

Cooke, W. W. 1914. Our shorebirds and their future. Pages 275-294 i_g Yearbook of
Agriculture. U.S. Govt. Printing Off. Washington, DC. 715 pp.

Dorio, J. C., and A. H. Grewe. 1979. Nesting and brood rearing habitat of the upland
sandpiper. Minnesota Acad. Sci. 45:8-11.

Forbush, E. H. 1912. A history of the game birds. wild-fowl and shore birds of
Massachusetts and adjacent states. Mass. State Board Agric., Boston.
p.347.

Graber, R. R., and J. W. Graber. 1963. A comparative study of bird populations in
Illinois, 1906-1909 and 1956-1958. 111. Nat. Hist. Surv. Bull. 28:383-
528.

Higgins, K. F., H. F. Duebbert, and R. B. Oetting. I969. Nesting of the upland plover
on the Missouri Coteau. Prairie Nat. 1:45-48.

Johnsgard, P. A. 1981. The plovers, sandpipers, and snipes of the world. Univ. Neb.
Press. Lincoln. 493 pp.

175

Jurek, R. M., and H. R. Leach. 1977. Shorebirds. Pages 301-320 jg G. C. Sanderson,
ed. Management of migratory shore and upland game birds in North
America. Inter. Assoc. Fish Wildl. Agencies, Washington, DC. 358

PP-

Kaiser, P. H. 1979. Upland sandpiper nesting in southeastern South Dakota. Proc. S.
Dakota Acad. Sci. 58:59-68.

Kirsch, L. M., and K. F. Higgins. 1976. Upland sandpiper nesting and management in
North Dakota. Wildl. Soc. Bull. 4:16-20.

Lindmeier, J. P. 1960. Plover, rail and godwit nesting on a study area in Mahnomen
County, Minnesota. Flicker 32:5-9.

McAtee, W. L., and F. E. L. Beal. 1912. Some common game, aquatic and rapacious
birds in relation to man. U.S. Dept. Agric., Farmer’s Bull. No. 497,
pp. 15-16.

Mitchell, G. J. 1967. The upland plover and its status in relation to environmental
conditions and situations, past and present. Blue Jay 25:58-63.

Osborne, D. R., and A. T. Peterson. 1984. Decline of the upland sandpiper (Bagrgmia
jgggjgaufig) in Ohio: an endangered species. Ohio J. Sci. 84:8-10.

Powell, D. J. 1981. Michigan bird survey, spring, 1981. Jack-Pine Warbler 59:105-
112.

Soothill, E., and R. Soothill. I982. Wading birds of the world. Blandford Press, Poole.
U.l(. 334 pp.

Stearns, F ., and D. Lindsey. 1977. Environmental status of the Lake Michigan region.
Vol. 11. Natural areas of the Lake Michigan drainage basin and
endangered or threatened plant and animal species. Argonne Nat.

Lab., Argonne, IL. p.62.

Snyder, D. L., V. Riemenschneider, and V. W. Inman. 1987. The upland sandpiper.
gm jggigjcggdg, breeding area in South Bend, Indiana. Indiana
Acad. Sci. 96:537-542.

Tate, .1. Jr. 1986. The blue list for 1986. Amer. Birds 40:227-236.

United States Fish and Wildlife Service. 1981. Standards for the development of Habitat
Suitability Index models. 103 ESM. U.S. Fish Wildl. Serv., Div.
Ecol. Serv. n.p.

Wallace, G. J. 1977. Environmental status of the Lake Michigan region. Vol. 14. Birds

of the Lake Michigan drainage basin. Argonne Nat. Lab., Argonne.
IL. p. 92.

176

White, R. P. 1983. Distribution and habitat preference of the upland sandpiper
(Bagtrmig lgggtcggda) in Wisconsin. Amer. Birds 37:16-22.

APPENDIX G
Habitat Model - American Redstart (Setophaga ruticilla)

R. Minnis and LB. Haufler
Michigan State University, East Lansing, MI

HABITAT USE INFORMATION
General

The American redstart inhabits a variety of woodland vegetation types, principally
second growth and mature deciduous. mixed and, occasionally, coniferous forests (Hickey
1940, Baker 1944, Bono 1957, Ficken and Ficken 1967, Webb et al. 1977, Sabo 1980). It is
a breeding bird of the Eastern United States and Canada but occurs in openings in the boreal
forest zone of Canada almost to the Pacific coast (Peterson 1980). Most American redstart-
over-winter in the West Indies (Bent 1963).

Whelan (1987) suggested that tree species may be an important determinant of
habitat selection by insectivorous birds. Accordingly, the American redstart has been
reported to favor certain tree species within individual stands for nesting and foraging
(Holmes and Robinson 1981, Morris and Lemon 1988a, 1988b). There is, however, no
consistency in selection of tree species over its entire range or even among stands in a single
forest (Purcell 1987).

EM

American redstarts are primarily insectivores that alter feeding strategies between
stands (Ficken 1962, Maurer and Whitmore 1981). Hovering, gleaning, flush-chasing, and
hawking are common tactics employed by American redstart for foraging (Sherry 1979.

177

178

Maurer and Whitmore 1981, Robinson and Holmes 1982). American redstart commonly
forage between 3 and 18 m (IO-60 ft) (Sherry 1979, Seidel and Whitmore 1982) with the 12
to 15 in (40-50 ft) range being most frequently utilized (Holmes et al. 1979). Invertebrates
within the orders Homoptera, Diptera, Hymenoptera, and Coleoptera are foraged upon more
often than those in Lepidoptera and Arachnida (Sherry 1979, Robinson and Holmes 1982).
Robinson and Holmes (1982) found that Homoptera comprised 35% of the American redstart
spring diet in New Hampshire.

Wiens (1974) found that summer food resources do not limit bird communities.
Morse (1973) noticed the mean abundance of food closely correlated with the food
requirements of warblers at different stages of the breeding cycle. This study also revealed
that, although food abundance fluctuated over years. bird nesting success remained high
(Morse 1973). Therefore, in this model, food is not considered a limiting factor for the
American redstart.

_W_ate_t

No information on water requirements was recorded in the literature. It is
assumed that the insectivorous feeding strategy minimizes the need for free water.
QLCI‘

The American redstart is primarily a bird of mature forest interiors (Ficken 1962).
though it often appears elsewhere (Ficken and Ficken 1967). Baker (1944) found the
preferred habitat of American redstart in Michigan to be second growth maple (Acer spp.).
Collins et al. (1982) noted that the American redstart occurred predominately in aspen
(Populus spp.) and birch (Betula spp.) forests or along edges of other deciduous forests in
Minnesota.

Pure coniferous stands are seldom used by American redstart (Capen 1979, Sabo

179
1980). Morse (1973) found that American redstart males on spruce-clad (Picea spp.) islands
usually are unsuccessful at nesting, failing in most cases to even attract a mate. Sabo (1980)
determined an upper threshold of conifer canopy coverage for American redstart to be 70%.
In Minnesota, American redstart were found on sites with very low amounts of coniferous
vegetation (Collins et al. 1982).
R r i n

American redstart enter their breeding range in late April and early May (Bent
1963). They construct a cupped nest in which the nest stands firmly upright, supported
mainly from below (Morris and Lemon 1988b). Clutch size averages near 3.5. and normally
only one brood per pair is raised each year (Whitcomb et al. 1981, Morris and Lemon
1988b).

Many forest-associated birds, such as the American redstart, typically nest in
small openings created by windthrow trees and other natural disturbances in the interior of
large wooded areas (Brittingham and Temple 1983). Morris and Lemon (1988b) found that

nest placement within each tree species seemed to be influenced by the requirements for the
structural support of the nest and reflected differences in branching geometry of the tree
species. The most common nest site of the American redstart is an upright, 3- or 4- pronged
crotch of a dead or live hardwood sapling, such as a maple, ash (Fraxinus spp.), or birch
(Bent 1963). Morris and Lemon (19880) found the average height of American redstart nests
to be 2-5 m (6.5-16.5 ft) above the ground, depending on the species of the tree.
Intgrsgersign

Home range size for American redstart average about 0.4 ha (1 acre) or smaller

(Table 24). American redstart aggressively defend their territories during the breeding

180 -

Table 24. Average home range sizes of the American redstart over its breeding range.

 

 

 

Size (ha) Location Study
1.85 New York Hickey 1940
0.59 Ohio Sturm 1945
0.40 Ohio 28mm 1945
1.06 New Hampshire Sherry 1979
2.45 New York Samson 1979
1.98 ‘ Ficken 1962

 

 

'Location not given.

season (Morse 1973. Sherry 1979), but on 2 occasions their boundaries were observed to
break down when young were being fed in the nest (Kendeigh 1945).
Numbers of American redstart seem to be positively correlated with forest area
(Stauffer and Best 1980) and require extensive tracts of forest to initially become established
(W hitcomb et al. 1981). Temple (1986) found that American redstart only occurred in forest
fragments > 100 ha (247 ac). Stauffer and Best (1980) noted that a reduction of wooded or
shrub habitat along a riparian zone to less than 200 m (656 ft) in width reduced or even
eliminated American redstart populations in Iowa.
HABITAT SUITABILITY INDEX (HSI) MODEL
The following model was developed for the American redstart as a method to
objectively assess habitat quality. The model is intended to measure habitat over their entire
breeding range during the breeding season (May - August)(Bent 1963). It is based on the
assumption that American redstart habitat can be accurately quantified using the following 5
variables:
V1: Percent Canopy Cover
V2: Percent of Canopy Cover that is Coniferous

V3: Number of Trees per Hectare
V4: Number of Saplings per Hectare

181

This model can be used to evaluate habitat in deciduous, coniferous and mixed forests.
Malia!

Optimal habitat for feeding and reproducing American redstart is characterized by
large tracts of second growth deciduous forest with numerous pole-sized trees and 60-85%
canopy closure. Sherry and Holmes (1989) cited 3 observations that indicate that American
redstart prefer deciduous stands: (1) older male American redstart, which are normally the
first returning migrants (Bent 1963), settle in such stands consistently year-after-year, (2)
yearling American redstart occur randomly in both hardwoods and mixed conifer hardwoods
depending on American redstart population densities, and (3) such stands are routinely the
first stands to be settled by American redstart each spring. Morse (1973) noted that American
redstart tended to utilize coniferous foliage for nesting when spruce-woods warblers (ie.
black-throated green (Dendroica virens). blackbumian (D. fusca), yellow-rumped (D.
coronata), and magnolia warblers (D. magnolia» were not present. It was also noted,
however, that American redstart foraged heavily within the limited deciduous foliage in the
woods, indicating the possibility of the spruce-woods warblers limiting the presence of
American redstart in coniferous forests (Morse 1973).

This model assumes the presence of spruce-woods warblers in any forest of high
conifer cover. Results from a New Hampshire study showed that American redstart were
located in regions predominated by northern hardwoods and were absent when conifer cover
reached or exceeded 70% (Sabo 1980). Therefore, any forest with > 70% conifer cover
receives a zero value for V2 (Fig. 39). Typical American redstart habitat in Minnesota
contained only 4.7% coniferous vegetation (Collins et al. 1982), whereas in New Hampshire
American redstart occupied habitats with 38% conifer vegetation (Sabo 1980). Optimal

habitat for the American redstart has 340% conifer cover, and receives a value of 1.0.

182

V1: Percent Canopy Cover

1.0 --- i i

 

 

V1 _.

 

 

0.0 . 1

V2: Percent of Canopy Cover that is Coniferous

 

 

1.0 i i 4' 1

V2

2_1_-_.___—___ 4t". .

0.0

 

 

 

Figure 39. Percent canopy cover variables for American redstart.

183

American redstart typically utilize small openings in the forest canopy for nesting
(Brittingham and Temple 1983) and feeding (Seidel and Whitmore 1982). Vegetation
structure at a particular site provides both opportunities and constraints on the ways that birds
can successfully search for and capture prey (Robinson and Holmes 1982). This may be one
explanation for the observed tree species preference (Whelan 1987); American redstart
concentrate their foraging where food resources are most abundant or available (Holmes and
Robinson 1981). Holmes and Robinson (1981) found that yellow birch (Betula alleghensis)
consistently had more potential prey per sample unit than sugar maple (Acer saccharum) and
American beech (Fagus grandifolia),which were less utilized than the yellow birch.

Similar studies indicate a clear preference of American redstart for areas with a
canopy cover between 60 and 85% (Bond 1957, Webb et al. 1977, Sabo 1980, Freedman et
al. 1981, Collins et al. 1982). Densities of American redstart dropped rapidly when canopy
cover either exceeded 85% or were below 60% (Bond 1957). Thus. any forest with a canopy
closure within the range of 60-85% receives a value of 1.0 for V1. The variable value
declines outside this range (Fig. 39).

Freedman ct al. (1981) found that plots thinned to 460 trees/ha held far fewer
American redstarts per unit area than did the 1400-1900 trees/ha uncut plots. Bond (1957)
noted that American redstarts in Wisconsin were predominantly found in stands with 350-375
stems/ha. The frequency of American redstarts decreased steadily from stands with 350
stems/ha to stands with 325 stems/ha (Bond 1957). Therefore, stands with 2350 stems/ha
receive a 1.0 for V3 and the value declines to 0.0 at 325 stems/ha (Fig. 40).

Purcell (1987) found the number of maple saplings and the basal area (BA) of
beech to be significant factors in determining American redstart habitat preference. Collins et

al. (1982) discovered the American redstart to be the most dependant species of its guild on

V3: Number of Trees per Hectalrge1

 

1.0 " T" ‘ 1

V3 -.

 

 

0.0

 

 

+ ~———J;

300 ' 350 400

V4: Number of Saplings per Hectare

 

 

 

 

 

 

1.0 r i i
V4
1
- i.
0.0 . . j
0 i 5110 . 1000+

Figure 40. Forest stand description variables for American redstart.

185
secondary forests with abundant low vegetation on numerous small deciduous trees. Bond
(1957) found that the highest densities of American redstart occurred in stands with 2650
saplings/ha (260/ac). Therefore, stands with 2650 saplings/ha (260/ac) receive a 1.0 value
for V4 (Fig. 40). Bond (1957) also found that stands with <300 saplings/ha (120/ac) did not
support American redstart, and therefore receive a 0.0 value for V4.

In the eastern and north central forest regions of the United States. many species
of birds are dependent on extensive forest systems (Robbins 1979), and consequently are
considered to be sensitive to habitat fragmentation (Whitcomb et al. 1981). These species
generally exhibit certain characteristics (Whitcomb et al 1981): they are long-distance
migrants that winter primarily in the New World tropics; they are obligate inhabitants of
forest interior; they build nests in the open rather than in the protection of cavities; they raise
only a single brood of young each year; and they have a comparatively small clutch size. The
American redstart is listed as such a species (Whitcomb et al. 1979).

MODEL APPLICATION / HSI DETERMINATION

The overall HSI value for any forest stand in the American redstart breeding range
can be calculated by determining the value for each of the 4 variables described above (Figs.
39-40), and then using the following equation to compute an index to habitat quality:

HSI = (V 1 *V2 "‘(V3"‘V4)°-’)°-25

Variables V3 and V4 are combined because together they sufficiently describe the
structure and density of the forest, while the canopy cover variables (V1 and V2) adequately
describe the overall canopy structure and composition. The averaging of the pair of combined
variables with the canopy (V3, V4) will generate a value between 0.0 and 1.0. where 0.0-
represents unsuitable habitat and 1.0 optimal habitat.

This model may be verified in the future by relating the densities of American

186

redstart to the HSI values for different forest stands. A positive relationship between
American redstart density, productivity and HSI values needs to exist for the model to be

valid.

Lit r r 1
Baker, B.W. 1944. Nesting of the American redstart. Wilson Bull. 56:212-216.

Bent, A.C. 1963. Life histories of North American wood warblers. Dover Publ.. Inc. New
York, N.Y. 743 pp.

Bond. RR. 1957. Ecological distribution of breeding birds in the upland forests of southern
Wisconsin. Ecol. Monog. 27:351-384.

Brittingham, M.C.. and S.A. Temple. 1983. Have cowbirds caused forest songbirds to
decline? Bioscience 33:31-35.

Capen, DE. 1979. Management of northeastern pine forests for nongame birds. Pages 90-109
i_n R.M. DeGraaf, and K.E. Evans, eds. Management of North Central and
Northeastern forests for nongame birds. USDA For. Serv. GTR-NC-Sl.

Collins, S.L., F.C. James, and P.G. Riser. 1982. Habitat relationships of wood warblers
(Parulidae) in northern central Minnesota. Oikos 39:50-58.

Ficken, M.S. 1962. Maintenance activities of the American redstart. Wilson Bull. 74:153-
165.

, and R.W. Ficken. I967. Age-specific differences in the breeding behavior and
ecology of the American redstart. Wilson Bull. 79: 188-199.

Freedman, B., C. Beauchamp, l.A. McLaren, and 8.1. Tingley. 1981. Forestry management
practices and populations of breeding birds in a hardwood forest in Nova Scotia.
Can. Field-Nat. 95:307-311.

Hickey, 1.1. 1940. Territorial aspects of the American redstart. Auk 48:512-522.

Holmes, R.T., R.E. Bonney, and SW. Pacala. 1979. Guild structure of the Hubbard Brook
bird community: a multivariate approach. Ecology 60:512-520.

. and S.K. Robinson. 1981. Tree species preferences of foraging insectivorous birds
in a northern hardwoods forest. Oecologia 48:31-35.

Howe, HF 1974. Age-specific differences in habitat selection by the American redstart. Auk
91:161-162.

187 -

Jones, RE. 1969. Breeding bird census: urban woodlots. Am. Birds 23:726-728.
Kendeigh. S.C. 1945. Nesting behavior of wood warblers. Wilson Bull. 57:145-164.

Maurer, B.A., and R.C. Whitmore. 1981. Foraging behavior of five bird species in two
forests with different vegetation structure. Wilson Bull. 93:478-490.

Morris, M.M.J., and RE. Lemon. 1988a. Mate choice in American redstarts: by territory
quality? Can. J. Zool. 66:2255-2261.

, and RE. Lemon. 1988b. American redstart nest placement in southwestern New
Brunswick. Can. J. 2001. 66:212-216.

Morse. DH. 1973. The foraging of small populations of yellow warblers and American
redstarts. Ecology 54:346-355.

Peterson, R.T. 1980. A field guide to the birds east of the rockies. Houghton Mifflin Co.,
Boston, Mass. 384pp.

Purcell, KL. 1987. Response of forest understory birds to habitat change. M.S. Thesis,
Humboldt State Univ., Arcata, Calif. 82pp.

Robbins, CS. 1979. Effect of forest fragmentation on bird populations. Pages 198-213 i_n_
R.M. DeGraaf, and K.E. Evans, eds. Management of North Central and
Northeastern forests for nongame birds. USDA For. Serv. GTR-NC-51.

Robinson, S.K., and R.T. Holmes. 1982. Foraging behavior of forest birds: the relationships
among search tactics, diet, and habitat structure. Ecology 63: 1918-1931.

Sabo. S.R. 1980. Niche and Habitat relations in subalpine bird communities of the white
mountains of New Hampshire. Ecol. Monog. 50:241-259.

Samson, PB. and SJ. Lewis. 1979. Experiments on population regulation in two North
American parids. Wilson Bull. 91:222-233.

Seidel. GE. and R.C. Whitmore. 1982. Effects of forest structure on American redstart
foraging behavior. Wilson Bull. 94:289-296.

Sherry, T.W. 1979. Competitive interaction and adaptive strategies of American redstarts and
least flycatchers in a northern hardwoods forest. Auk 96:265-283.

, and R.T. Holmes. 1989. Age-specific social dominance affects habitat use by
breeding American redstarts: a removal experiment. Behav. Ecol. Sociobiol.
25:327-333.

Stauffer, DE, and LB. Best. 1980. Habitat selection by birds of riparian communities:
evaluating effects of habitat alterations. J. Wildl. Manage. 44: 1-15.

188

Sturm. L. 1945. A study of the nesting activities of the American redstart. Auk 62: 189-206.

Temple, S.A. 1986. Predicting impacts of habitat fragmentation on forest birds: a comparison
of two models. Pages 301-304 i_n_ J. Verner, M.L. Morrison, and CJ. Ralph. eds.
Wildlife 2000: modeling habitat relationships of terrestrial vertebrates. Univ.
Wisconsin Press, Madison.

Webb, W.L., D.F. Behrend, and B. Saisorn. 1977. Effect of logging on songbird populations
in a northern hardwood forest. Wildl. Monog. 55: 35pp.

Whelan, CJ. 1987. Effects of foliage structure on the foraging behavior of insectivorous
birds. PhD Diss., Dartmouth College, Hanover, NH. 115pp.

Whitcomb, R.F., C.S. Robbins. J.F. Lynch, B.L. Whitcomb, M.K. Klimkiewicz. and D.
Bystrak. 1981. Pages 125-206 i_n R.L. Burgess, and D.M. Sharpe, eds. Forest
island dynamics in man-dominated landscapes. Springer-Verlay. New York, N.Y.

LITERATURE CITED

Ailes, I.W. 1980. Breeding biology and habitat use of the upland sandpiper in central
Wisconsin. Passenger Pigeon 42:53-63.

Albrecht, W.A. 1944. Soil fertility and wildlife-cause and effect. Trans. N. Amer. Wildl.
Conf. 9219-28.

Allen, A.W. 1987. Habitat suitability index models: gray squirrel. USDI Fish Wildl. Serv.
Biol. Rep. 82(10.35). l3pp.

Ambuel, B., and S.A. Temple. 1982. Songbird populations in southern Wisconsin forests:
1954 and 1979. J. Field Ornithol. 53:149-158.

Asherin, D.A., H.L. Short, and J.E. Roelle. 1979. Regional evaluation of wildlife habitat
quality using rapid assessment methodologies. Trans. N. Amer. Wildl. Nat.
Resour. Conf. 44:404-424.

Askins, R.A., and MJ. Philbrick. 1987. Effects of changes in regional forest abundance on
the decline and recovery of a forest bird community. Wilson Bull. 99:7-21.

_, _, and D.S. Sugeno. 1987. Relationship between the regional abundance of forest
and the composition of forest bird communities. Biol. Conserv. 39:129-152.

August, P.V. 1983. The role of habitat complexity and heterogeneity in structuring tropical
mammal communities. Ecology 64:1495-1507.

Bailey, R.G., R.D. Pfister, and J.A. Henderson. 1978. Nature of land and resource
classification-a review. J. For. 76:650-655.

_. 1987. Suggested hierarchy of criteria for multi-scale ecosystem mapping. Landscape
and Urban Planning 14:313-319.

Balda, R.P., W.S. Gaud, and JD. Brawn. 1983. Predictive models for snag nesting birds.
Pages 216-222 jg J.W. Davis, G.A. Goodwin, and R.A. Ockenfels, tech. coords.
Snag habitat management: proceedings of the symposium. USDA For. Serv. Gen.
Tech. Rep. RM-99.

Barnes, B.V., K.S. Pregitzer, T.A. Spies, and V.H. Spooner. 1982. Ecological forest site
classification. J. For. 80:493-498.

189

190

1989. Old-growth forests of the northern Lake States: a landscape ecosystem
perspective. Nat. Areas J. 9:45-57.

Bart, 1., DR. Petit, and G. Linscombe. 1984. Field evaluation of two models developed
following the Habitat Evaluations Procedures. Trans. N. Amer. Wildl. Nat.
Resour. Conf. 49:489-499.

_. and JD. Schoultz. 1984. Reliability of singing bird surveys: changes in observer
efficiency with avian density. Auk 101:307-318.

Bazzaz, F.A., and T.W. Sipe. 1987. E. Physiological ecology, disturbance, and ecosystem
recovery. Pages 203-227 jg E.-D. Schulze and H. Zwolfer, eds. Potentials and
limitations of ecosystem analysis. Springer-Verlag, New York.

Behan, R.W. 1989. Multiresource forest management with EZ-IMPACT simulation models.
Pages 182-195 i_n A. Tecle. W.W. Covington, and RH. Hamre, tech. coords.
Multiresource management of Ponderosa pine forests. USDA For. Serv. Gen.
Tech. Rep. RM-185. 282pp.

Bellrose, F.C. 1976. Ducks, geese, and swans of North America. Stackpole Books,
Harrisburg, PA. 540pp.

Bender. L.C., R.G. Moses, and 1.8. Haufler. 1989. Habitat suitability index models: black-
throated green warbler. Unpubl. Rep., Michigan State Univ., East Lansing. n.p.

_. and LB. Haufler. 1990. Habitat suitability index models: white-tailed deer, Great
Lakes region. Unpubl. Rep., Michigan State Univ., East Lansing. n.p.

Best. LB, and DP. Stauffer. 1986. Factors confounding evaluation of bird-habitat
relationships. Pages 209-216 jg J. Verner. M.L. Morrison, and CJ. Ralph, eds.
Wildlife 2000. Univ. Wisconsin Press, Madison.

Blake, 1.0.. and J.R. Karr. 1984. Species composition of bird communities and the
conservation benefit of large versus small forests. Biol. Conserv. 30:173-187.

Bock, CE. and B. Webb. 1984. Birds as grazing indicator species in southeastern Arizona.
J. Wildl. Manage. 48:1045-1049.

Bouffard, S.H., and D. Hein. 1978. Census methods for eastern gray squirrels. J. Wildl.
Manage. 42:550-557.

Bowen, D.E.. Jr. 1976. Coloniality. reproductive success. and habitat interactions in upland
sandpipers (Bartramia longicauda). PhD Diss., Kansas State Univ.. Manhattan.
127pp.

Boyce, S.C. 1985. Forestry decisions. USDA For. Serv. Gen. Tech. Rep. SE-35. 318pp.

191

Brewer. R. 1963. Ecological and reproductive relationships of black-capped and Carolina
chickadees. Auk 80:9-47.

Broadfoot, WM. 1969. Problems in relating soil to site index for southern hardwoods.
For. Sci. 15:354-364.

Brower. J.E., and J.H. Zar. 1984. Field and laboratory methods for general ecology.
William C. Brown Publ.. Dubuque, IA.

Brown. .I.H. 1973. Species diversity of seed-eating desert rodents in sand dune habitats.
Ecology 54:775-787.

Brubaker, LB, and ER. Cook. 1983. Tree-ring studies of Holocene environments. Pages
222-238 i_g H.E. Wright, Jr., ed. Late Quaternary environments of the United
States. Vol. II. The Holocene. Univ. Minnesota Press, Minneapolis.

Cade. 8.8.. and PJ. Sousa. I985. Habitat suitability index models: ruffed grouse. USDI
Fish Wildl. Serv. Biol. Rep. 82(10.86). 30pp.

Capen. D.E., J.W. Fenwick. D.B. Inkley. and A.C. Boyton. 1986. Multivariate models of
songbird habitat in New England studies. Pages 171-175 i_n J. Verner, M.L.
Morrison, and CJ. Ralph, eds. Wildlife 2000. Univ. Wisconsin Press, Madison.

Carmean, W.H. 1975. Forest site quality evaluation in the United States. Adv. Agronomy.
27:209-269.

Caswell, H. 1976. The validation problem. Pages 313-325 i_n B.C. Patten, ed. System
analysis and simulation in ecology. Vol. 4. Academic Press, NY.

Clark. .1 .D., and J.C. Lewis. 1983. A validity test of a habitat suitability index model for
clapper rail. Proc. Annu. Conf. Southeast. Assoc. Fish Wildl. Agencies 37:95-
102.

Cleland. D.T.. .1.B. Hart. G.E. Host, K.S. Pregitzer, and CW. Ramm. 1993. Field guide:
ecological classification and inventory of the Huron-Manistee National Forests.
USDA For. Serv., Huron-Manistee Nation. For., Cadillac, MI. varied pages.

Cody, M.L., and H. Walter. 1976. Habitat selection and interspecific interactions among
Mediterranean sylviid warblers. Oikos 27:210-238.

Cole, C.A., and R.L. Smith. 1983. Habitat suitability indices for monitoring wildlife
populations. Trans. N. Amer. Wildl. Nat. Resour. Conf. 48:367-375.

Collins, S.L., F.C. James. and P.G. Risser. 1982. Habitat relationships of wood warblers
(Parulidae) in northern central Minnesota. Oikos 39250-58.

Collins, W.B. 1981. Habitat preferences of mule deer as rated by pellet-group distributions.
J. Wildl. Manage. 45:969-972.

192

Cook, 1.0., and LL. Irwin. 1985. Validation and modification of a habitat suitability model
for pronghorns. Wildl. Soc. Bull. 13:440-448.

Cooperrider, A.Y., and DP. Behrend. 1983. Simulated experiments with a model of deer
populations and northern forest vegetation. Pages 313-321 i_n W.K. Lauenroth,
G.V. Skogerboe, and M. Flug. eds. Analysis of ecological systems: state-of-the-
art in ecological modelling. Elsevier Sci. Publ. Co., New York.

1986. Habitat evaluation systems. Pages 757-776 jg A.Y. Cooperrider, RJ. Boyd,
and H.R. Stuart, eds. Inventory and monitoring of wildlife habitat. U.S. Bur.
Land Manage. Serv. Center, Denver, CO.

Crawford, B.T. 1950. Some specific relationships between soils and wildlife. J. Wildl.
Manage. 14:115-123.

D’Agostino, R.B., A. Belanger, and RB. Agostino, Jr. 1990. A suggestion for using
powerful and informative tests of normality. Amer. Statistician 44:316-319.

Davis. DE, and R.L. Winstead. 1980. Estimating the numbers of wildlife populations.
Pages 221-245 jg S.D. Schemnitz, ed. Wildlife techniques manual. The Wildl.
Soc.. Washington, DC.

Davis, LS. 1980. Strategy for building a location-specific, multi-purpose information
system for wildland management. J. For. 78:402-406.

Dedon, M.F., S.A. Laymon, and R.H. Barrett. 1986. Evaluating models of wildlife-habitat
relationships of birds in black oak and mixed-conifer habitats. Pages 1 15-119 i_n,
J. Verner, M.L. Morrison, and CJ. Ralph, eds. Wildlife 2000. Univ. Wisconsin
Press, Madison.

Denny, AH. 1944. Wildlife relationships to soil types. Trans. N. Amer. Wildl. Conf.
92316-323.

Denslow. 1.8. 1985. Disturbance-mediated coexistence of species. Pages 307-323 jg S.T.A.
Pickett and PS. White, eds. The ecology of natural disturbance and patch
dynamics. Academic Press, Inc.. Orlando, FL.

Dobkin, D.S.. and BA. Wilcox. 1986. Analysis of natural forest fragments: riparian birds
in the Toiyabe Mountains, Nevada. Pages 293-299 jg J. Verner, M.L. Morrison.
and CJ. Ralph, eds. Wildlife 2000. Univ. Wisconsin Press, Madison.

Driscoll, R.S., D.L. Merkel, D.L. Radloff, D.E. Snyder, and 1.8. Hagihara. 1984. An
ecological land classification framework for the United States. USDA For. Serv.
Misc. Pub. No. 1439. 20pp.

Farmer, A.H., MJ. Armbruster, J.W. Terrell, and R.L. Schroeder. 1982. Habitat models
for land-use planning: assumptions and strategies for development. Trans. N.
Amer. Wildl. Nat. Resour. Conf. 47:47-56.

193

Ficken, M.S. 1962. Maintenance activities of the American redstart. Wilson Bull. 79:188-
199.

Flather, C.H., and T.W. Hoekstra. 1985. Evaluating population-habitat models using
ecological theory. Wildl. Soc. Bull. 13:121-130.

, _, D.E. Chalk, N.D. Cost, and V.A. Rudis. 1989. Recent historical and projected
regional trends of white-tailed deer and wild turkey in the southern United States.
U.S. For. Serv. Gen. Tech. Rep. RM-172. 22pp.

Franklin, J.F. 1988. Structural and functional diversity in temperate forests. Pages 166-174
jg E.O. Wilson and FM. Peter, eds. Biodiversity. Nation. Acad. Press.
Washington, DC.

Franzreb. K.E.. and RD. Ohmart. 1978. The effects of timber harvesting on breeding birds
in a mixed-coniferous forest. Condor 80:431-441.

Freemark, K.E., and H.C. Merriam. 1986. Importance of area and habitat heterogeneity to
bird assemblages in temperate forest fragments. Biol. Conserv. 36:115-141.

1988. Landscape ecology of forest birds in the northeast. Pages 7-12 i_g R.M.
DeGraaf and WM. Healey, compilers. ls forest fragmentation a management
issue in the northeast? USDA For. Serv. Gen. Tech. Rep. NE-l40. 32pp.

Freese, F. 1978. Elementary forest sampling. USDA. Agric. Handb. No. 232. 9lpp.

Garshelis, D.L. 1986. Ecology and population dynamics of black bears in north central
Minnesota. Minn. Dept. Nat. Resour. Res. Rep. 1986.

Giles. R.H., Jr. 1978. Wildlife management. W.H. Freeman & Co., San Francisco.
416pp.

Gokee, T.L., and LA. Joyce. 1992. Analysis of standards and guidelines in a geographic
information system using existing resource data. USDA For. Serv. Res. Pap.
RM-304. 12pp.

Grue, C .E., R.R. Reid, and H. Silvy. 1983. Correlation of habitat variables with mourning
dove call counts in Texas. J. Wildl. Manage. 47:186-195.

Hammill. J .H.. and RJ. Moran. 1986. A habitat model for ruffed grouse in Michigan.
Pages 15-18 i_n J. Verner, M.L. Morrison, and CI. Ralph, eds. Wildlife 2000.
Univ. Wisconsin Press. Madison.

Harvey, A.E.. and L.F. Neuenschwander. 1991. Proceedings-management and productivity
of western-montane forest soils. USDA For. Serv. Gen. Tech. Rep. INT-280.
254pp.

194 -

Haufler. JB. 1994. An ecological framework for planning for forest health. I. Sustain.
For. 2:307-316.

Hawkes, C.L., D.E. Chalk, T.W. Hoekstra, and CH. Flather. 1983. Prediction of wildlife
and fish resources for national assessments and appraisals. USDA For. Serv.
Gen. Tech. Rep. RM-IOO. 21pp.

Hays. R.L., C. Summers, and W. Seitz. 1981. Estimating wildlife habitat variables. USDI
Fish Wildl. Serv. FWS/OBS-8l/47. 106pp.

Hirsch, A., CT. Cushwa. K.W. Flach, and WE. Frayer. 1978. Land classification-where
do we go from here? I. For. 76:672-673.

Holmes. R.T.. and S.K. Robbins. 1981. Tree species preferences of foraging insectivorous
birds in northern hardwoods forest. Oecologia 48:31-35.

Hornbeck, J.W.. and W.T. Swank. 1992. Watershed ecosystem analysis as a basis for
multiple-use management of eastern forests. Ecol. Appl. 22238-247.

Host. G.E., K.S. Pregitzer. C.W. Ramm. J.B. Hart, and D.T. Cleland. 1987. Landform-
mediated differences in successional pathways among upland forest ecosystems in
northwestern lower Michigan. For. Sci. 33:445-457.

Hundley, LR. 1959. Available nutrients in selected deer-browse species growing on
different soils. J. Wildl. Manage. 23:81-90.

Hunter. M.L., Jr. 1990. Wildlife, forests. and forestry. Prentice-Hall. Inc.. Englewood
Cliffs. NJ. 370pp.

Hurley. J.F., H. Salwasser, and K. Shimamoto. 1982. Fish and wildlife habitat capability
models and special habitat criteria. Cal-Neva Wildl. Trans. 1982:40-48.

Jackson, J. 1970. A quantitative study of the foraging ecology of downy woodpeckers.
Ecol. 51:318-323.

James. F.C. i971. Ordinations of habitat relationShips among breeding birds. Wilson Bull.
83:215-236.

_. and ND. Warner. 1982. Relationships between temperate forest bird communities and
vegetation structure. Ecol. 63:159-171.

Jarvinen, O. 1985. Conservation indices in land use planning: dim prospects for a panacea.
Ornis Fennica 62:101-105.

Johnson, L.B., G.E. Host, J.K. Jordan, and LL. Rogers. 1991. Use of GIS for landscape
design in natural resource management: habitat assessment and management for
the female black bear. Pages 507-517 jg Proc. GlS/LIS ’91, Atlanta, GA.

195

Johnson, R.G., and S.A. Temple. 1986. Assessing habitat quality for birds nesting in
fragmented tallgrass prairies. Pages 245-249 i_n J. Verner, M.L. Morrison, and
C.J. Ralph. eds. Wildlife 2000. Univ. Wisconsin Press. Madison.

Johnson, W.N., Jr., and F.W. Brown. 1990. Avian use of a lakeshore buffer strip and an
undisturbed lakeshore in Maine. N. J. Appl. For. 7:1 14-117.

Jones, SM. 1989. Application of landscape ecosystem classification in identifying
productive potential of pine-hardwood stands. Pages 64-69 i_n Proc. pine-
hardwood mixture: a symposium on management and ecology of the type. USDA
For. Serv. Gen. Tech. Rep. SE-58.

1991. Landscape ecosystem classification for South Carolina. Pages 59-68 i_g D.L.
Mengel, D.T. Tew, eds. Ecological land classification: applications to identify
the productive potential of southern forests. USDA For. Serv. Gen. Tech. Rep.
SE-68.

Karr. J.R. 1968. Habitat and avian diversity on strip mined land in east-central Illinois.
Condor 70:348-357.

_, and R.R. Roth. 1971. Vegetation structure and avian diversity in several new world
areas. Am. Nat. 105:423-435.

Kaufrnann, M.R., and J .J . Landsberg. 1990. Advancing toward closed ecosystem models:
report on a workshop. USDA For. Serv. Gen. Tech. Rep. RM-201. 4pp.

Kendeigh, S.C. 1945. Nesting behavior of wood warblers. Wilson Bull. 57:145-164.

Kerr, R.M. I986. Habitat mapping. Pages 49-69 jg A.Y. Cooperrider. R.J. Boyd, and
H.R. Stuart. eds. Inventory and monitoring of wildlife habitat. USDI Bur. Land
Manage. Serv. Cent, Denver, CO.

Lancia. R.A.. SD. Miller, D.A. Adams, and D.W. Hazel. 1982. Validating habitat quality
assessment: an example. Trans. N. Amer. Wildl. Nat. Resour. Conf. 47:96-110.

_, D.A. Adams, and EM. Lunk. 1986. Temporal and spatial aspects of species-habitat
models. Pages 65-69 jg J. Verner, M.L. Morrison, and C.J. Ralph, eds.
Wildlife 2000. Univ. Wisconsin Press, Madison.

Lantz. G. 1976. Compartment prescription and environmental analysis report. USDA For.
Serv., 2410-1 (2/73), Harrisville, MI. 32pp.

Larson, D.L., and CE. Bock. 1986. Determining avian habitat preference by bird-centered
vegetation sampling. Pages 37-43 jg J. Verner, M.L. Morrison, and C.J. Ralph,
eds. Wildlife 2000. Univ. Wisconsin Press, Madison.

Larson, T.J., OJ. Rongstad. and F.W. Terbilcox. 1978. Movements and habitat use of
white-tailed deer in south central Wisconsin. J. Wildl. Manage. 42:113-117.

196

Latka. DC, and J.W. Yahnke. 1986. Simulating the roosting habitat of sandhill cranes and
validating suitability-of-use indices. Pages 19-22 jg J. Verner, M.L. Morrison,
and C.J. Ralph, eds. Wildlife 2000. Univ. Wisconsin Press, Madison.

Laymon, S.A., and R.H. Barrett. 1986. Developing and testing habitat-capability models:
pitfalls and recommendations. Pages 87-91 jg J. Verner, M.L. Morrison, and
C.J. Ralph, eds. Wildlife 2000. Univ. Wisconsin Press, Madison.

_. and J.A. Reid. 1986. Effects of grid-cell size on tests of a spotted owl model. Pages
93-96 jg J. Verner, M.L. Morrison, and C.J. Ralph. eds. Wildlife 2000. Univ.
Wisconsin Press, Madison.

Lewis, 1.8. 1967. Management of the eastern turkey in the Ozarks and bottomland
hardwoods. Pages 371-408 m CH. Hewitt. ed. The wild turkey and its
management. Valley Offset, Inc.. Doposit. NY.

Lines, I.L., Jr., and C.J. Perry. 1978. A numerical wildlife habitat evaluation procedure.
Trans. N. Amer. Wildl. Nat. Resour. Conf. 43:284-301.

Logan, J.A.. J .E. Heasley, and CB. Stalnaker. 1983. State-of-the-art and recommendations
for future efforts in modeling the functional role of animals in ecological systems.
Pages 983-985 i_n W.K. Lauenroth, G.V. Skogerboe, and M. Flug. eds. Analysis
of ecological systems: state-of-the-art in ecological modeling. Elsevier Sci. Publ.
Co., New York.

Lokemoen, J.T. 1982. Waterfowl production on stock-watering ponds in the northern plains.
Pages 201-211 i_n J.T. Ratti. L.D. Flake. and W.A. Wentz, compilers. Waterfowl
ecology and management: selected readings. The Wildl. Soc.. Inc. Bethesda.
MD.

MacArthur, R.H., and J .W. MacArthur. 1961. On bird species diversity. Ecology 42:594-
598.

Mackie, R.J.. and G.L. Dusek. 1992. Deer habitat relationships and management in the
northern Rocky Mountains and Great Plains. Western Wildlands 18:14-19.

Maguire, B. 1973. Niche response structure and the analytical potentials of its relationship
to habitat. Amer. Nat. 1072213-246.

Manville, A.M. 1983. Human impact on the black bear in Michigan‘s lower peninsula. Int.
Conf. Bear Res. Manage. 5:20-33.

Marcot. B.G., M.G. Raphael, and K.H. Berry. 1983. Monitoring wildlife habitat and
validation of wildlife-habitat relationship models. Trans. N. Amer. Wildl. Nat.
Resour. Conf. 48:315-329.

 

197

Martin, T.E., and N. Nur. 1992. Population dynamics and their implications for
vulnerability of species. Nation. Training Wkshp., Status and Manage.
Neotropical Migratory Birds. Executive Summ. Sept. 21-25, Este Park Center.
CO. pp. 9-10.

Maurer, BA. 1986. Predicting habitat quality for grassland birds using density-habitat
correlations. J. Wildl. Manage. 50:556-566.

McNab, W.H. 1991. Land classification in the Blue Ridge Province: State-of-the-science
report. Pages 37-47 i_n D.L. Mengel, D.T. Tew, eds. Ecological land
classification: applications to identify the productive potential of southern forests.
USDA For. Serv. Gen. Tech. Rep. SE-68.

Mealey, S.P.. J.P. Lipscomb, and K.J. Norman. 1982. Solving the habitat dispersion
problem in forest planning. Trans. N. Am. Wildl. Nat. Resour. Conf. 472142-
153.

Mello. P.J.. and J.B. Haufler. 1983. A computer simulation model for the study of deer- _
aspen forest interactions. Pages 259-265 i_n W.K. Lauenroth, G.V. Skogerboe,
and M. Flug. eds. Analysis of ecological systems: state-of—the-art in ecological
modeling. Elsevier Sci. Publ. Co., New York.

Mengel, D.L., and D.T. Tew. 1991. Ecological land classification: applications to identify
the productive potential of southern forests. USDA For. Serv. Gen. Tech. Rep.
SE-68. 149pp.

Miller, DH, and LL. Getz. 1977. Factors influencing local distribution and species
diversity of forest small mammals in New England. Can. J. 2001. 55:806-814.

Minnis, R., and J .B. Haufler. 1991. Habitat suitability index model for the American
redstart. Unpubl. Rep., Michigan State Univ., East Lansing. n.p.

Moore. PD. 1976. Effects of a long hot summer. Nature 263:278-279.

Morrison, M.L. 1983. Assessing changes and trends in wildlife habitat in a forest
management context. Pages 101-103 i_n J.F. Bell and T. Atterbury. eds.
Renewable resource inventories for monitoring changes and trends. Proc. int.
Conf., August 15-19, Corvallis, OR.

_, B.G. Marcot, and R.W. Mannan. 1992. Wildlife-habitat relationships: concepts and
applications. Univ. Wisconsin Press, Madison. 470pp.

Morse. DH. 1976. Variables affecting the density and territory size of breeding spruce-
woods warblers. Ecology 57:290-301.

Moses. R.G., G.J. Roloff, and J.B. Haufler. 1990. Habitat suitability index model for the
eastern bluebird in northern Michigan. Unpubl. Rep., Michigan State Univ.. East
Lansing. n.p.

198

Moss, D. 1978. Diversity of woodland song-bird populations. J. Anim. Ecol. 47:521-527.

Murphy, D.A., and W.R. Porath. 1969. Forest soils and game nutrition. Proc. Ann.
Conf. Southeast. Assoc. Game and Fish Comm. 23:18-25.

Naylor, T.H., and J .M. Finger. I967. Verification of computer simulation models.
Manage. Sci. 14:892-8106.

Neff, DJ. 1968. The pellet group count technique for big game trend, census, and
distribution: a review. J. Wildl. Manage. 32:597-614.

Nelson. 0., GA. Harris, and TE. Hamilton. 1978. Land and resource classification-who
cares? I. For. 76:644-646.

Noon. B.R., D.K. Dawson, D.B. Inkley. C.S. Robbins. and SH. Anderson. 1980.
Consistency in habitat preference of forest bird species. Trans. N. Amer. Wildl.
Nat. Resour. Conf. 45:226-244.

Noon. B. 1992. Thought of improving the biological value of bird-habitat models. Nation.
Training Wkshp., Status and Manage. of Neotropical Migratory Birds. Executive
Summaries. Sept. 21-25. Este Park Center. CO. 40pp.

Norton. N.C. 1981. Food habits, growth, and cover types used by northern Wisconsin
black bears. M.S. Thesis, Univ. Wisconsin-Stevens Point. 48pp.

Nowak. R.M. 1991. Walker‘s mammals of the world. Fifth edition, volume 1. The John
Hopkins Univ. Press. Baltimore, MD. 64pr.

O‘Brien, R. 1990. Assessment of nongame bird habitat using forest survey data. USDA
For. Serv. Res. Pap. INT-431. 8pp.

Odum, E.P. 1971. Fundamentals of ecology. 3rd ed. W.B. Saunders Co., Philadelphia.
574pp.

O’Neill, R.V., D.L. DeAngelis, J.B. Waide, and T.F.H. Allen. 1986. A hierarchial concept
of ecosystems. Princeton Univ. Press. Princeton. NJ. 253pp.

Pearson, D.L. 1975. The relation of foliage complexity to ecological diversity of three
Amazonian bird communities. Condor 77:453-466.

Petraborg, W.H., E.G. Wellein, and V.E. Gunvalson. 1953. Roadside drumming counts: a
spring census method for ruffed grouse. J. Wildl. Manage. 17:292-295.

Pianka, ER. 1967. On lizard species diversity: North American flatland deserts. Ecology
48:333-351.

Pinkowski, BC. 1976. Use of tree cavities by nesting eastern bluebirds. J. Wildl. Manage.
40:556-563.

199

Pregitzer, K.S., B.V. Barnes, and GD. Lemme. 1983. Relationship of topography to soils
and vegetation in an Upper Michigan ecosystem. Soil Sci. Soc. Amer. J. 47:117-
123.

, and . 1984. Classification and comparison of upland hardwood and
conifer ecosystems of the Cyrus H. McCormick experimental forest, upper
Michigan. Can J. For. Res. 14:362-375.

 

Ralph. C.J., and J .M. Scott. 1981. Estimating numbers of terrestrial birds. Studies in avian
biology No. 6. Allen Press, Inc.. Lawrence, KS. 630pp.

Raphael. M.G., and 8.6. Marcot. 1986. Validation of a wildlife-habitat-relationships model:
vertebrates in a Douglas fir sere. Pages 129-138 i_n_ J. Verner, M.L. Morrison,
and C.J. Ralph, eds. Wildlife 2000. Univ. Wisconsin Press, Madison.

Renken, RB. 1988. Habitat characteristics related to pileated woodpecker densities and
territory size in Missouri. PhD Diss., Univ. Missouri, Columbia. ll4pp.

Rice. 1., B.W. Anderson, and RD. Ohmart. 1984. Comparison of the importance of
different habitat attributes to avian community organization. J. Wildl. Manage.
48:895-911.

Robel, R.J., L.B. Fox, and K.E. Kemp. 1993. Relationship between habitat suitability index
values and ground counts of beaver colonies in Kansas. Wildl. Soc. Bull. 21:415-
421.

Robbins. C.S. l98la. Bird activity levels related to weather. Stud. in Avian Biol. No.
6:301-310.

1981b. Effect of time of day on bird activity. Stud. in Avian Biol. No. 62275-286.

Rodgers, RD. 1981. Factors affecting ruffed grouse drumming counts in southwestern
Wisconsin. J. Wildl. Manage. 45:409-418.

Rogers, LL, and A.W. Allen. 1987. Habitat suitability index models: black bear. upper
Great Lakes region. USDI Fish Wildl. Serv. Biol. Rep. 8200.144). 54pp.

Roloff, G.J., and J.B. Haufler. 1990. A habitat suitability index model for the upland
sandpiper in the Huron National Forest. Michigan. Unpubl. Rep., Michigan State
Univ., East Lansing. 20pp.

, and _. 1993. Forest planning in Michigan using a GIS approach. Int. Union Game
Biol. (In Press).

Rosenzweig, M.L., and J. Winakur. 1969. Population ecology of desert rodent
communities: habitats and environmental complexity. Ecology 50:558-572.

200

Rotenberry, J.T. 1985. The role of habitat in avian community composition: physiognomy
or floristics? Oecologia 67:213-217.

Rowe, J .S. 1978. The common denominator of land classification in Canada; an ecological
approach to mapping. Pages I95-198 i_n Ecological classification of forest land in
Canada and northwestern U.S.A. Centre for Continuing Educ., Univ. British
Columbia, Canada.

Rowland, M.M., G.C. White, and EM. Karlen. 1984. Use of pellet group plots to measure
trends in deer and elk populations. Wildl. Soc. Bull. 12:147-155.

Russell, W.B., and J .K. Jordan. 1991. Ecological classification system for classifying land
capability in midwestern and northeastern U.S. national forests. Pages 18-24 jg
D.L. Mengel, D.T. Tew, eds. Ecological land classification: applications to
identify the productive potential of southern forests. U.S. For. Serv. Gen. Tech.
Rep. SE-68.

Ryder, R.R. 1986. Songbirds. Pages 291-312 jg A.Y. Cooperrider. R.J. Boyd. and H.R.
Stuart, eds. Inventory and monitoring of wildlife habitat. USDI Dept. lnter., ‘
Bur. Land Manage. Service Center. Denver, CO.

SAS Institute Inc. 1985. SAS user’s guide: statistics, version 5 edition. SAS Institute Inc..
Cary, NC. 956pp.

Sabo, S.R. 1980. Niche and habitat relations in subalpine bird communities of the White
Mountains of New Hampshire. Ecol. Monogr. 50:241-259.

Samson, P.B., and SJ. Lewis. 1979. Experiments on population regulation in two North
American parids. Wilson Bull. 912222-233.

Schamberger. M., and W.B. Krohn. 1982. Status of the habitat evaluation procedures.
Trans. N. Amer. Fish Wildl. Conf. 1982:154-164.

_, and LJ. O’Neill. 1986. Concepts and constraints of habitat-model testing. Pages 5-10
jg J. Verner, M.L. Morrison, and C.J. Ralph, eds. Wildlife 2000. Univ.
Wisconsin Press. Madison.

Schroeder, R.L., and P.J. Sousa. 1982. Habitat suitability index models: eastern
meadowlark. USDI Fish Wildl. Serv.. FWS/OBS-82/ 10.29. 9pp.

1983. Habitat suitability index models: pileated woodpecker. USDI Fish Wildl. Serv..
FWS/OBS-82/ 10.39. 15pp.

Seitz. W.K., C.L. Kling, and AH. Farmer. I982. Habitat evaluation: a comparison of three
approaches on the northern Great Plains. Trans N. Amer. Wildl. Nat. Resour.
Conf. 47:82-95.

201 '

Seng. PT. 1991. Evaluation of techniques for determining tree squirrel abundance and
habitat suitability in central Missouri. M.S. Thesis, Univ. Missouri, Columbia.
128pp.

Sherry, T.W. 1979. Competitive interaction and adaptive strategies of American redstarts
and least flycatchers in a northern hardwoods forest. Auk 96:265-283.

Sieg, CH. 1991. Rocky Mountain juniper woodlands: year-round avian habitat. USDI For.
Serv. Res. Pap. RM-296. 7pp.

Siegel. S. 1956. Nonparametric statistics for the behavioral sciences. McGraw-Hill Book
Co., New York. 321pp.

Skirvin. A.A. 1981. Effect of time of day and time of season on the number of observations
and density estimates of breeding birds. Stud. in Avian Biol. No. 62271-274.

Smith, T.M., and H .H. Shugart. 1987. Territory size variation in the ovenbird: the role of
habitat structure. Ecology 68:695-704.

Sokal, R.R.. and El. Rohlf. 1981. Biometry. W.H. Freeman and Co.. New York. 859pp.
Sommers, L.M. 1984. Michigan. Westview Press, Inc. Boulder, CO. 254pp.

Sousa. P.J., and A.H. Farmer. 1983. Habitat suitability models: wood duck. USDI Fish
Wildl. Serv., FWS/OBS-82/ 10.43. 27pp.

Spies, T.A.. and B.V. Barnes. 1985. A multifactor ecological classification of the northern
hardwood and conifer ecosystems of Sylvania Recreation Area, Upper Peninsula,
Michigan. Can. J. For. Res. 15:949-960.

Stauffer, D.F., and LB. Best. 1986. Effects of habitat type and sample size on habitat
suitability index models. Pages 71-77 jg J. Verner, M.L. Morrison, and C.J.
Ralph, eds. Wildlife 2000. Univ. Wisconsin Press, Madison.

Sturm, L. 1945. A study of the nesting activities of the American redstart. Auk 62:189-
206.

Swanson, F.J., T.K. Krantz, N. Caine. and R.G. Woodmansee. 1988. Landform effects on
ecosystem patterns and processes. BioScience 38:92-98.

Swift. B.L., J .S. Larson, and R.M. DeGraaf. 1984. Relationship of breeding bird density
and diversity to habitat variables in forested wetlands. Wilson Bull. 96:48-59.

Tansley, A.C. 1935. The use and abuse of vegetational concepts and terms. Ecology
42:237-245.

202

Thomas, J.W. 1986. Wildlife-habitat modeling-cheers, fears, and introspection. Pages xix-
xxv i_n J. Verner, M.L. Morrison, and C.J. Ralph, eds. Wildlife 2000. Univ.
Wisconsin Press, Madison.

U.S. Department of Commerce. 1979. Climate atlas of the United States. Natl. Oceanic
Atmospheric Adm. 80pp.

U.S. Fish and Wildlife Service. 1981. Standards for the development of habitat suitability
index models. USDI Fish Wild. Serv. Release No. 1-81, 103 ESM. n.p.

U.S. Forest Service. 1988. Forest Service R-9 compartment prescription handbook. USDA
For. Serv. FSH 2409.21d. l43pp.

Van Horn, R.L. 1971. Validation of simulation results. Manage. Sci. 17:247-258.

Van Horne, B. 1983. Density as a misleading indicator of habitat quality. J. Wildl.
Manage. 47:893-901.

_. and J.A. Wiens. 1991. Forest bird habitat suitability models and the development of
general habitat models. USDI Fish Wildl. Serv. Res. 8. 3lpp.

Van Lear, D.H., and J.F. Hosner. 1967. Correlation of site index and soil mapping units.
J. For. 65:22-24.

Van Lear, D.H., and SM. Jones. 1987.- An example of site classification in the southeastern
coastal plain based on vegetation and land type. S. J. Appl. For. 11:23-28.

Verner, J. 1984. The guild concept applied to management of bird populations. Environ.
Manage. 821-14.

_, M.L. Morrison, and C.J. Ralph. 1986. Wildlife 2000: Modeling habitat relationships
of terrestrial vertebrates. Univ. Wisconsin Press. Madison. 470pp.

Warwick. J.J., and W.G. Cade. 1988. Estimating model reliability using data with
uncertainty. Ecol. Modelling 41:169-181.

Welsh, C.J.E., and W.M. Healey. 1993. Effects of even-aged timber management on bird
species diversity and composition in northern hardwoods of New Hampshire.
Wildl. Soc. Bull. 21:143-154.

Wiens. J.A. 1974. Habitat heterogeneity and avian community structure in North American
grasslands. Amer. Midi. Nat. 91:195-213.

Wiens. J.A. 1986. Spatial and temporal variation in studies of shrubsteppe birds. Pages
154-172 i_n J. Diamond. and TJ. Case. eds. Community ecology. Harper &
Row. Publ. New York.

203

Wiken. EB. 1982. Rationale and methods of ecological land surveys: an overview of
Canadian approaches. Pages 11-19 jg D.G. Taylor, ed. Land/wildlife
integration. Minister Supply Serv., Canada.

Williamson, M.H.. and J.H. Lawton. 1991. Fractal geometry of ecological habitats. Pages
69-86 jg S.S. Bell, E.D. McCoy, and H.R. Mushinsky, eds. Habitat structure:
The physical arrangement of objects in space. Chapman and Hall, New York.

Willson, M.F. 1974. Avian community organization and habitat structure. Ecology
55:1017-1027.