THESIS

")1

MI ICHI IGAN STATEU

N l (Ll! IllUlllllllllllllllllllllllll

301691 8892

"'m

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

This is to certify that the

thesis entitled
HABITAT QUALITY AND UTILIZATION ANALYSIS IN A SPATIAL

CONTEXT: THE CASE OF LUPINUS PERENNIS L. AND LYCAEIDES
MELISSA SAMUELIS NABOKOV, (LEPIDOPTERA: LYCAENIDAE)

presented by

Lisa Michelle Greenfeld

has been accepted towards fulfillment
of the requirements for

Has ters Forestry

degree in

 

 

a”

Major professor

  
 

 

 

Date I? Sef’l'CWLQV ’qq‘,

0-7639 MS U is an Affirmative Action/Equal Opportunity Institution

 

 

LIBRARY

Michigan State
Unlverslty

 

 

PLACE IN RETURN BOX

to remove this checkout from your record.
To AVOID FINES return on or before date due.

 

DATE DUE

DATE DUE DATE DUE

4‘19?

 

”14% mnriraé J

 

mylflfirm

   

W7

 

I'EB 2 5 2005

@824”

 

INOV 0 9 2006

 

 

 

 

 

 

 

 

ma (:ICIHCJDmDmmS-p."

 

HABITAT QUALITY AND UTILIZATION ANALYSIS IN A SPATIAL CONTEXT:
THE CASE OF L UPINUS PERENNIS L. AND LYCAEIDES MELISSA SAMUELIS
NABOKOV, (LEPIDOPTERA: LYCAENIDAE)
By

Lisa Michelle Greenfeld

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements

for the degree of

MASTERS OF SCIENCE

Department of Forestry

1997

ABSTRACT
HABITAT QUALITY AND UTILIZATION ANALYSIS IN A SPATIAL CONTEXT:
THE CASE OF L UPINUS PERENNIS L. AND L YCA EIDES MELISSA SA MUELIS
NABOKOV, (LEPIDOPTERA: LYCAENIDAE)
By

Lisa Michelle Greenfeld

The first objective of this research was to evaluate tree-moderated environmental
variables that affect lupine growth and density: light and soil pH, via a field study and a
controlled greenhouse study. My results indicated that a pH range of 4.2 to 5.6 and 100%
light provide the best conditions for lupine growth. However, because of the potential for
weather extremes, land managers protecting Kamer blue butterfly (KBB) sites should aim
to provide lupine within a range of light levels by adjusting canopy cover.

The second objective of this study was to assess how KBB and lupine distribution
are affected by spatial parameters and landscape level forest characteristics via an
exploratory analysis of stand data, and lupine and KBB distribution. Patches where
lupine and KBB were present tended to be in medium density stands, closer to one
another, more complex, and larger, than lupine areas where KBB was absent. Patches
where lupine was present were more likely to be in the low stand density class and less

likely to be in the high density class than areas where lupine was absent.

To Chuck, with all of my love.

iii

Acknowledgments

This research was funded by the McIntire-Stennis competitive grant, MSU
MAWGAP fellowship and the MSU Department of Forestry.

I would like to thank the many people who helped to make this project a success.
I could not have done it without my committee: Dr. Jeremy Fried, Dr. Deb McCullough,
and Dr. Robert Haack. I appreciated their time, patience, helpfulness, suggestions, and
ideas.

Many others provided valuable assistance to this study. Dr. Phu Nguyen provided
lab space, ecology and chemistry knowledge; Dr. Don Dickmann provided flight time and
the first review of my first proposal; Dr. Carl Ramm provided hours of statistical advice
and expertise; Dr. Rich Kobe helped me with his ecological insights and statistical
suggestions; Dr. Dan Brown provide assistance with F RAGSTATS and GIS questions;
Mary Rabe and Cathy Papp Herms were very generous in sharing their Kamer blue data
with me; Dr. John Biembaurn, Dr. An Cameron, Randy Klevickas, Paul Bloese, Dr.
Royal Heins, Dr. Boyd Ellis, and Dr. Bruno Reissmam were very helpful with my
greenhouse study; Maria Albright and John Lerg at ASGA provided data, guidance,
assistance, and much friendliness during my field work; Dr. Sam Chan and Dr. Andy
Gray generously allowed me to use their photo analyzing equipment, both in the field and
in their lab; the secretarial staff was very helpful with the many details. I’d like to

express my sincere gratitude to all of these individuals.

I would also like to thank the many friends that were supportive in times of stress,
fun, and hard work and that went above and beyond the call of helpfulness during my
project (e. g. watering plants, standing out in the freezing cold with the GPS, helping me
make maps, re-teaching rne statistics, shoveling soil, sifting soil,...). You helped me to
learn a lot about forestry, relationships, and myself. Thanks to Sarah Synowiec, Sean
Canavan, Demetris Gatziolis, Mark Zweifler, Greg Winter, Dave Neumann, Georgia
Peterson, Gem Castillo, Andy David, Peggy Payne, Joe Zeleznik, and Steve and Jan
Michmerhuizen.

Finally, I would like to thank my family for all of their support through out my
education. I sincerely appreciate my mother’s encouragement, love, and wisdom and my
sisters’ patience and support, and Chuck’s love, understanding, help, and listening

abilities, without which I would not have made it through the last three years!

TABLE OF CONTENTS

List of Tables ................................................................................................................... viii
List of Figures ................................................................................................................... ix
Chapter 1
General Introduction ............................................................................................................ 1
Introduction and Literature Review .................................................................................. 2
Chapter 2
The Effect of Light and Soil pH on Lupinus Perennis Growth and Density in Michigan
Oak Savanna Forest Stands .................................................................................................. 6
Abstract ............................................................................................................................. 7
Introduction ...................................................................................................................... 8
Methods .......................................................................................................................... 10
Field study ................................................................................................................... 11
Study Area ................................................................................................................ 1 1
Sampling .................................................................................................................. 11
Data Analysis ........................................................................................................... 15
Greenhouse study ........................................................................................................ 16
Methods .................................................................................................................... 16
Data Analysis ........................................................................................................... 17
Results ............................................................................................................................ 18
Field Study .................................................................................................................. l8
Greenhouse Study ....................................................................................................... 23
Discussion ....................................................................................................................... 25
Conclusions .................................................................................................................... 28
Bibliography ................................................................................................................... 29

vi

Chapter 3
The Effect of Landscape Level Spatial Parameters and Forest Structure on the

Distribution of Lupinus Perennis and Lycaeides melissa samuelis Nabokov. .................. 32
Abstract ........................................................................................................................... 33
Introduction .................................................................................................................... 34
Methods .......................................................................................................................... 40

Data processing and analysis ....................................................................................... 40
Results ............................................................................................................................ 48
Discussion ....................................................................................................................... 52
Conclusions .................................................................................................................... 53
Bibliography ................................................................................................................... 55

Overall Conclusions ........................................................................................................... 59

Appendix A ........................................................................................................................ 61

Appendix B ........................................................................................................................ 62

Bibliography ...................................................................................................................... 63

vii

LIST OF TABLES

Table 1 - Area, perimeter, and number of plots at the six study sites ....................... 12

Table 2 - Regression model statistics for each iteration of the factor-ceiling
distributions partitioned regression for proportion of direct light vs.
lupine stem density (number/5 m2) ..................................................... 20

Table 3 - Effects of light on Lupinus perenm's biomass when grown in a
greenhouse for 115 days; N = 41 ...................................................... 24

Table 4 - Geographic information system (GIS) coverages and associated
attributes used to analyze landscape properties and their relationship
to KBB and lupine habitat at Allegan State Game Area, Allegan
Michigan, 1995 ............................................................................ 42

Table 5 - Class level pairwise comparisons of spatial statistical output
from FRAGSTATS for areas with lupine absent (LA), lupine
present (LP), lupine present and KBB absent (LPKA), and
lupine and KBB both present (LPKP) ................................................. 49

viii

LIST OF FIGURES

Figure l - Map of study sites at Allegan State Game Area (ASGA), Allegan,

Michigan, May 1995 ....................................................................... 13

Figure 2- Scatterplot of lupine density on soil pH at Allegan State Game Area,
Figure 3 - Scatterplot of stem density on proportion of direct light at Allegan

State Game Area (ASGA), June-July 1995; “predicted” refers to
the regression: stem density = 0.5007 (proportion of direct light)

...18

+ 5.4351 (N = 443; adjusted r squared = 0.102) ....................................... 21

Figure 4 - Scatterplot of stem density on proportion of direct light at Allegan
State Game Area (ASGA), June-July 1995; “predicted” refers to the
fitted frontier regression: stem density =1.6085 (proportion of direct light)

+ 96.538; (N = 20; adjusted r squared = 0.739) ........................................... 21

Figure 5 - Histograms of the number of plots by proportion of direct light
for four lupine foliar desiccation classes: a) green; b) mostly green;

c) mostly brown; d) brown ................................................................. 22

Figure 6 - Map of areas with lupine present, KBB present (LPKP); areas
with lupine present, KBB absent (LPKA), and areas with
lupine absent, KBB absent (LAKA) at Allegan State Game

Area (ASGA), Allegan, MI 1995 ......................................................... 23

Figure 7 - Frequency of stands by density class for the qualifiers: all
stands (AS), lupine absent (LA), lupine present (LP), lupine
present, KBB absent (LPKA), and lupine present, KBB present

(LPKP) at Allegan State Game Area, Allegan, MI, 1995 ............................. 51

Figure 8 - Frequency of stands by size class for the qualifiers all stands
(AS), lupine absent (LA), lupine present (LP), lupine present,
KBB absent (LPKA), and lupine present, KBB present (LPKP)

at Allegan State Game Area, Allegan, MI, 1995 ....................................... 51

Chapter 1

General Introduction

Ix.)

Introduction and Literature Review

The Karner blue butterfly (KBB), Lycaeides melissa samuelis Nabokov,
(Lepidoptera: Lycaenidae) was added to the federal endangered species list in December
1992 following a period of extensive habitat decline (Dirig, 1988; Haack, 1993). To
protect this species, it is imperative to know what site characteristics are commonly
linked with healthy KBB populations. The intent of this study was to develop knowledge
about KBB’s habitat needs so that managers can better protect this species.

The Kamer blue butterfly, first identified in 1861 by Edwards, was ultimately
named by Nabokov in 1943 after Kamer, New York, the collection site of the type
specimen (Haack, 1993). Kamer blue have historically been observed in a region
stretching from New Hampshire to Minnesota, and as far south as Ohio. Today, they are
known to occur in parts of New Hampshire, New York, Michigan, Indiana, Wisconsin,
and Minnesota (Haack, 1993). Adults are small (2.2-3.2 cm long), purplish-blue,
bivoltine (two generations per year) butterflies with orange and black spots. Eggs and
larvae are green and difficult to spot when attached to vegetation (Clough, 1992; Back,
1993; Opler and Krizek, 1984; Scudder, 1889). Larvae feed exclusively on the
leguminous perennial Lupinus perennis L. (Haack, 1993; Smallidge and Leopold, 1992;
Stuart and Ricci, 1988; WI DNR, 1993). Adults of the first generation (May-June) rely
on lupine flowers for some of their nectar needs, but adults of the second generation
(July-August) require other nectar sources because lupine plants are no longer in flower
(Haack, 1993). Ant tending of KBB larvae appears to play an important role in larval

survival (Haack, 1993). The predators and parasites which affect Kamer blue include

3

various spiders, a stink bug, robber flies, assassin bugs, 3 tachnid fly, a braconid wasp,
and two ichneumonid wasps (Haack, 1993 and references therein).

Scientists have shown that habitat preservation is the key to species preservation
(e. g., Loehle and Wein, 1994). This is particularly evident in the case of the Kamer blue
butterfly, a habitat specific species (Dirig, 1988). Recently, the KBB recovery team
(mandated by the Endangered Species Act) have been working on a description of
Kamer blue habitat. In Michigan, the habitat has been described as oak savannas with an
understory of lupine and other nectaring plants (Hermes, 1996; Pickering and Gebauer,
1993; Shuey, et. al, 1987; Welch, 1993). Because this description does not address such
specifics as the amount of lupine required, the size, shape, and distribution of suitable
areas, or the ecological requirements of lupine, managers still lack critical knowledge for
reversing the decline of Kamer blue.

To restore remnant populations of Kamer blue butterflies to sustainable levels,
land managers need to consider the impacts of their actions on lupine, the Kamer blue’s
sole larval food plant (Haack, 1993; Smallidge and Leopold, 1992; Stuart and Ricci,
1988; WI DNR, 1993). A better understanding of lupine’s ecological requirements is an
important first step towards managing the recovery of the Kamer blue.

Lupinus perennis is a member of the pea family and is the only lupine native to
the eastern US (Hultman, 1978). The name “lupine”, derived from lupus, was chosen
because the genus was thought to “wolf up” resources (Edsall, 1985); however, this belief
proved false. Lupine fixes nitrogen, which is important since it usually grows on
nutrient-poor, sandy soils (Cryan and Dirig, 1978) that are well to excessively well-

drained and prone to drought (Smallidge and Leopold 1992). Lupine have 7-15

4

oblanceolate to obovate leaflets with strong midribs and an entire margin per palmate leaf
(Hultman, 1978; Miller and Whiting, 1895; Smith, 1966). Flowers ranging from white to
blue in color have five petals that are approximately one cm long on short pedicles in
loose terminal racemes (Miller and Whiting, 1895; Smith, 1966). Seeds, formed in
brown pubescent seed pods in mid-summer, are thrown as much as 3 meters when ripe
(Edsall, 1985).

Several researchers have demonstrated links between spatial patterns of
understory vegetation and variations in microhabitat caused by the differential influence
of individual trees on soil properties and light intensity (e.g., Crozier and Boemer, 1984;
Parker and Muller, 1982; Zinke, 1962). Because L. perennis is rarely found in dense
forests, most researchers have assumed that the species cannot survive under deep shade
(Bess, 1989; Maxwell and Givnish, 1993; Pickering and Gebauer, 1993; Smallidge and
Leopold 1992). However, relationships between lupine characteristics (e.g. percent
cover, number of flowering stalks, number of stems) and indices of light such as percent
tree and shrub cover (Pickering and Gebauer, 1993) have not been quantified.

Some have suggested that pH may have a strong influence on lupine growth
(Pickering and Gebauer 1993). On alkaline soils, growth is highly variable among many
lupine species and genotypes, but lupine growth is typically poor relative to soils with
lower pH. The optimal pH range for Lupinus angustifolius growth has been estimated via
controlled experiments at 5.0 to 5.5 (Tang et al., 1992). No studies have yet addressed
the link between soil pH and plant growth for L. perennis.

A better understanding of the underlying ecological factors that determine the

quality of Kamer blue habitat is essential to facilitate recovery efforts for the species.

5

This research has two foci. The first was to evaluate the effects of two tree-moderated
environmental variables which affect lupine growth and density: light and soil pH. A
field study and a greenhouse study were used to analyze these relationships. The second
objective was to evaluate the spatial parameters that affect lupine and KBB distribution. I
used a GIS and F RAGSTATS (a spatial statistics program) to answer questions about the
relationships between KBB and lupine distribution and (1) landscape metrics (e.g., size,
shape, edge, proximity, patch density) and (2) forest characteristics (e.g., overstory size

class, and overstory density).

Chapter 2

The Effect of Light and Soil pH on Lupinus

Perennis Growth and Density in Michigan Oak Savanna Stands.

Abstract

The Kamer blue butterfly (KBB), Lycaeides melissa samuelis Nabokov,
(Lepidoptera: Lycaenidae) was added to the federal endangered species list in December
1992 following a period of extensive habitat decline. To restore remnant populations of
Kamer blue to sustainable levels, land managers need to consider the impacts of their
actions on lupine, the Kamer blue’s sole larval food source. This research focused on two
tree-moderated environmental variables that affect lupine growth and density: light and
soil pH. The relationships among these variables were examined in both a field study and
in a controlled greenhouse study. Light and soil pH appear to affect lupine grth and
density. Lupine were found almost exclusively on soils with a pH range of 4.2 to 5.6.
The highest available light (100%) provides the best conditions for lupine growth.
However, because of the potential for weather extremes such as early summer drought
and late spring frost, land managers should aim to promote lupine by adjusting canopy

cover to provide a range of light levels.

Introduction

Scientists have shown that habitat preservation and/or restoration is the key to
species preservation (e. g., Loehle and Wein, 1994). However, moving from a conceptual
to an operational definition of habitat for particular species is no small task. This is
especially true in the case of the Kamer blue butterfly, (KBB), Lycaeides melissa
samuelis Nabokov (Lepidoptera: Lycaenidae), a species dependent on one of the most
rare habitats in the northeastern US, oak savanna and pine barrens (Dirig, 1988). This
species was added to the federal endangered species list in December 1992 following a
period of extensive habitat decline (Dirig, 1988; Haack, 1993). Lack of an operational
habitat definition for KBB has slowed recovery efforts and led to intense efforts to
identify critical habitat for this species.

In Michigan, KBB populations occur primarily in oak savannas with an
understory of lupine and other nectaring plants (Pickering and Gebauer, 1993; Shuey, et.
a1; 1987; Welch, 1993). Because this description does not address such specifics as the
amount of lupine required, the size, shape, and distribution of suitable areas, or the
ecological requirements of lupine, managers lack information that is critical for reversing
the decline of Kamer blue.

To restore remnant populations of Kamer blue to sustainable levels, land
managers need to consider the impacts of their actions on lupine, the Kamer blue’s sole
larval food source (Haack, 1993; Smallidge and Leopold, 1992; Stuart and Ricci, 1988;
WI DNR, 1993). A better understanding of lupine’s ecological requirements is a critical

step towards managing the recovery of KBB. Lupinus perennis is a member of the pea

9

family and is the only lupine native to the eastern US. (Hultman, 1978). Lupine fixes
nitrogen and grows on nutrient-poor, sandy soils (Cryan and Dirig, 1978) that are well to
excessively well-drained and prone to drought (Smallidge and Leopold 1992).

Several researchers have demonstrated links between spatial patterns of
understory vegetation and variations in microhabitat caused by the differential influence
of individual trees on soil properties and light intensity (e. g., Crozier and Boemer, 1984;
Parker and Muller, 1982; Zinke, 1962). Such findings suggest that the response of
herbaceous plants to their environment depends on highly localized characteristics that
are rarely captured by landscape or stand level averages (Clark, 1991). Tree-induced
environmental variation is particularly noticeable in savannas and variation in soil
properties is a frequently cited example (Bray, 1955 cited in Curtis, 1959; Holland, 1973;
Parker, 1977). Trees can have major effects on many environmental factors that probably
influence lupine growth, density and distribution and subsequently, KBB populations.

Because L. perennis is rarely found in dense forests, most researchers have
assumed that it cannot survive under deep shade (Bess, 1989; Maxwell and Givnish,
1993; Pickering and Gebauer, 1993; Smallidge and Leopold 1992). Flowering has been
correlated to canopy cover, reportedly reaching maximum levels in stands of intermediate
canopy cover (40-60%) (Maxwell and Givnish, 1993; WI DNR, 1993). In the Albany
pine bush in New York, home of one of the largest single populations of Kamer blue, L.
perennis has the greatest stem density where canopy cover is between 10 and 40%, and
the greatest reproductive success under 10-30% canopy cover (Pickering and Gebauer,
1993). In another study, L. perennis density was observed to be low and its flowering

ability impaired in open areas (0-5% canopy cover) in southern Ontario (Boyonoski,

10
1992). Though not entirely consistent, these results highlight the importance of light in

determining the distribution of L. perennis (Maxwell and Givnish, 1993; Pickering and
Gebauer, 1993). Managers attempting to enhance lupine will need more precise
quantification of the relationships between lupine characteristics (e.g., number of
flowering stalks, number of stems) and indices of light such as prOportion of direct light
(PDL) (Pickering and Gebauer, 1993).

Soil pH may also influence lupine growth (Pickering and Gebauer 1993). On
alkaline soils, growth is highly variable among many lupine species and genotypes, but is
typically poor relative to soils with lower pH. Growth variability has been attributed to
variation in root growth (Tang et al., 1992), shoot growth in response to iron deficiency
(Cowling and Clements, 1993), and decreases in symbiotic nitrogen fixation (Tang and
Robson, 1993). No studies have yet addressed the link between soil pH and plant growth
for L. perennis. A better understanding of the underlying ecological factors that
determine the quality of Kamer blue habitat is essential to facilitate recovery of this
species. This research focuses on two tree-moderated environmental variables that affect
lupine growth and density: light and soil pH. The relationships among these variables
were examined both in a field study in areas known to be occupied by KBB and in a

controlled greenhouse study.

Methods
Two complementary approaches were taken to evaluate the relationships between
light and lupine growth. A field study was conducted to assess environmental conditions

where lupine is present. A controlled experiment in the greenhouse provided a

ll

consistency check and an opportunity to address the effects of light under controlled
conditions. In the field study, the dependent variable was stem density on fixed plots; in

the greenhouse study the dependent variable was growth.

Field study
Study Area

The Allegan State Game Area (ASGA) in southwest Michigan’s Allegan County
was selected as the location for field experiments because it is one of two major areas in
Michigan where Kamer blue is found today. Forests of black and white oak (Quercus
veluntia and Quercus alba ) cover most of the area. Soils are very well-drained and
sandy; most are in the Oakville fine sand series. Attempts to farm this land early in the
century failed because the sandy soils were too infertile to support crops. By the 1930’s,
much of this area reverted to the state when landowners failed to pay taxes (John Lerg,

Forester, ASGA Pers. com., 1995).

Sampling

In spring 1995, six temporary study sites were established at ASGA in Open areas
where Kamer blue population sampling had been conducted in 1993 and 1994 (Herrns,
1996). These grass/forb dominated openings within the oak forest matrix varied in shape,
size, and lupine density (Table 1). At each site, plot center-points were established on a
10 by 10 meter grid spanning open areas and forest edges. There were 42 to 100 plot
centers/site depending on the size of the site. Plots were positioned from a random
starting point using a measuring tape and compass, and plot centers were marked with

numbered flags.

 

Table 1. Area, perimeter, and number of plots at the six study sites.

 

 

 

Study sites Area (In!) ' Perimeter (m) Number of plots
1 28,223 263 100

2 9,995 l21 100

3 32,720 321 100

4 20,485 347 70

5 10.637 238 42

6 30,969 305 70

Total 133,029 1,595 482

 

 

' The area and perimeter of study site 2 were estimated.

l
l

 

Roads

[:1 Swamp Hardwood
- Water
Wet Wetland
Study Sites‘
,. Private Land

 

1 2 Miles
Land Cover 1., ‘ A.

Aspen
.3 . Grass
5;; Jack Pine

*ASGA i

i
*All study sites are on the grass cover type. I

Figure 1. Map of study sites at Allegan State Game Area (ASGA), Allegan,
Michigan, May 1995

14

A five m2, temporary circular plot was established at each plot center by rotating 3
1.262 m stick with one end held over the plot center. Lupine stems and flowers were
counted within the plots at peak flowering, in early June.

Canopy cover was determined using hemispherical photography, a method which
provides a consistent, objective index of available light (Chan et al., 1986). Black and
white negative film (Kodak T-max 100) was used and exposures were made with a
Cannon AE-l Program fitted with a 7.5 mm focal length proportional fisheye lens
mounted on a tripod 38-51 cm above the ground. Exposure goals were: F stops 2 5.6,
and shutter speeds of 1/ 125 second or faster. I used a light meter to bracket the F stop
and shutter speed at each plot center. Excluding the sun from the camera’s field of view
is a requirement of this method, so most exposures were made near dawn or dusk when
the sun was below or near the horizon. A total of 964 exposures were completed between
June 27 and July 20, 1995. Proportion of direct light (PDL) at each plot center was
estimated from the corresponding hemispherical photo using a software program called
CANOPY (Los Alamos National Laboratory, Los Alamos, New Mexico) which utilizes
an efficient image combination algorithm to categorize digitized images of the slides as
PDL (Rich, 1989) (see appendix A).

One soil core of the entire A horizon was collected at each plot center with a 2 cm
diameter soil corer. The samples from each plot were stored in paper bags and air dried.
Each sample was then evenly mixed, and hand-processed to remove roots. Ten grams of

soil were drawn from each homogenized sample and combined with 10 ml deionized

15

water on a shaker for 30 minutes. The pH of the resulting slurry was measured using a
Corning ion analyzer 250 (Coming, Corning, New York).

The early summer of 1995 was abnormally hot and dry, and lupine senescence
occurred far earlier than normal (Dirig, 1973). To document this phenomenon, lupine
desiccation on all plots was assessed using a subjective, ordinal scale (1 - entirely green,
2 - mostly green, 3 - mostly brown, and 4 - entirely brown).

In 1996, aerial photography was attempted with the intent of delineating all
recognizable lupine patches. Unfortunately a late May frost which killed virtually all of
the lupine flowers, and an extended period of cloudy weather doomed this effort. Casual
on-site observations revealed intact flowers only on lupine plants growing under the

protection of the forest canopy or individual trees.

Data Analysis

The relationship between soil pH and lupine density was assessed via scatterplots
and correlation analysis. The relationship between light and lupine density was assessed
via factor-ceiling distributions partitioned regression, an exploratory data analysis
technique described by Thomson et al. (1996). Associations between light and lupine

desiccation was examined via histograms and Spearman’s rank correlation.

Greenhouse study

Methods

A controlled experiment was conducted to quantify effects of three light
treatments, 100%, 65%, and 35% of ambient light.1 The 35% and 65% treatments were
created by stretching shade cloth with these ratings over frames constructed of electrical
conduit and positioned 1 meter above the potted plants. F ifty-four, 25.4 cm deep pots
were thoroughly cleaned, dried, placed in saucers, and distributed among the three
treatments.

Lupine seed was collected from ripe pods at ASGA in July, 1995. Seeds were
removed from pods and stored at 3 ° C in a sealed plastic bag until sowing. Seeds free of
obvious deformities (e.g., wrinkled, dehydrated or underdeveloped) were scarified with a
file, soaked in deionized water for 12 hours and coated with Rhizobium lupini. Five seeds
were sown in each pot at a depth of ‘/2 cm in soiless potting media (a mixture of peat,
pearlite, and vermiculite) in February 19972. After two weeks, plants were thinned to one
plant per pot. Attempts were made to transplant excess seedlings into pots in which no
seedlings had emerged, but these were largely unsuccessful.

Media moisture in the pots was maintained (40-70% saturated by weight) via

subirrigation using well water with a fertilizer treatment of 20-2-20 NPK. Plants were

 

' This experiment was conducted in a temperature controlled greenhouse at Michigan State University, East
Lansing Michigan from February - June, 1997 in which ambient light was augmented with 110 watt
fluorescent lamps daily between pm and 1:10 am)

2 A previous attempt (from April to September 1996) to conduct this experiment, which also included pH
as a manipulated variable, was aborted due to high mortality of lupine unrelated to specific light and pH
treatment. In the course of this experiment, 1 leamed that pH could not be manipulated without undesirable
impacts on plant nutrition.

17

treated with fungicides monthly and sprayed for insect and other fungal problems as
needed. Appendix B contains a complete log of all treatments.

All plants were harvested in mid-June 1997, 115 days after planting. Roots were
separated from above-ground plant material and stored in paper bags. Leaves were
weighed and assessed for leaf area using a leaf area meter (Decagon Devices, Inc. Delta-T
Area Measurement System. Pullman, WA, USA). Leaves and roots were dried at 21 ° C
for 24 hours, then weighed. Any remaining potting media was removed from roots prior

to weighing.

Data Analysis
The relationship between light and growth was tested via the Kruskal-Wallis t-sample
test, a nonparametric rank-based test designed to handle non-normally distributed data

which is appropriate for small sample sizes (less than 30).

Results

Field Study

Although no statistically significant (p=.05) linear relationship between soil pH
and lupine density was identified via regression analysis (adjusted r squared = .039), a
scatterplot of this relationship for all plots on all sites suggests that lupine at ASGA was
restricted to soils with pH between 4.2 and 5.63 (Figure 2). Results among the six
individual sites were similar. The mean soil pH for plots with lupine stems present was

4.85 (sd :t 0.30); for plots without lupine stems, the mean was 4.70 (sd i 0.43).

 

 

 

 

 

 

 

 

 

 

 

 

 

250
d. 200 °
. E
m
, a
E 150
2
N
i .5 100
.8
E
3 50 -.__-- |
m
o
o 1 2 3

Soil pH

 

Figure 2. Scatterplot of lupine density on soil pH at Allegan State Game Area,
MI, 1995.

 

3 Plots that were within 1 meter of a road were excluded from this analysis due to the potential of
confounding effects from material applied to and scattered from the road.

l9
Regression of the PDL (the index of available light) on lupine density produced

no statistically significant, simple linear relationship (adjusted r squared = 0.102) (Figure
3). This was also the case with flower4 counts (adjusted r squared = 0.012). However, if
the variables are truly unrelated, multiple regressions run on the resulting positive and
negative residuals should also generate low adjusted r squared values (Thomson, et al.,
1996). Given the potentially large number of factors which may affect the suitability of a
site for lupine, it would be unlikely to find a strong predictive relationship based on light
alone. Factor-ceiling distributions partitioned regression was employed to identify the
relationship between the upper limit of lupine density and light (Thomson, et al., 1996).
Using this iterative technique, density was repeatedly regressed on the positive residuals
from the previous regression model (Table 2). The third and final5 iteration produced a
regression model which estimates the frontier line (upper boundary of the lupine density /

PDL relationship) with an adjusted r squared of 0.739 (Figure 4).

 

‘ Because n was smaller for flowers than for stems and were highly correlated, stems were used for all
further analysis.
5 By this iteration. n equaled 20.

20

Table 2. Regression model statistics for each iteration of the factor-ceiling distributions

partitioned regression for proportion of direct light vs. lupine stem density (number/5 m2).

 

 

 

 

n Adjusted b m Actual Signif. p value Accept!

R2 F F for b Rejectl

initial 443 0.102 5.4 50.1 51.281 3.370 0.096 reject
regression

partition 1 110 0.238 43.7 96.2 34.989 3 .953 0.0002 reject

partition 2 48 0.481 70.5 138 .7 44 .617 2.760 0.0002 reject

partition 3 20 0.739 96.5 160.9 54 .873 7.200 0.0002 reject

 

 

IAccept/Reject null hypothesis Ho: y = b; where b= a constant. Reject because Actual F
is greater than significant F.

2Significant at p=0.05

The Spearman’s rho correlation coefficient, (rs = 0.572; p < 0.001), between the PDL and

the observed desiccation class was significant (Figure 5). Lupine plants remained greener

for longer into the summer at lower PDL than at high proportion of direct light. This

suggests that while a high PDL values may be associated with greater lupine stem

densities, they also predict premature senescence during unusually dry years.

21

 

300 _-

 

 

 

 

 

 

 

'
250 -- _— .2 2-.. .._ __-,,--__..

E T
re 0.
E 200 -__- 2...- .1... ’
g . e I» e
2 150 L.—
“ O O : 9 observed
.6 O. 't O .,
E 100.1% We. 9— _predicted
5 ' “o 0’ e o...’ """""
2 5° ' ”e 3 e

0 9 e

O

0
o 20 40 60 80 100

Proportion of direct light

 

Figure 3. Scatterplot of stem density on proportion of direct light at Allegan State

Game Area (ASGA), J une-J uly 1995; “predicted” refers to the regression: stem density =
0.5007 (proportion of direct light) + 5.4351 (N = 443; adjusted r squared = 0.102).

 

250

150

Number of stemsls 1112 area

200-

100.

 

 

 

 

 

 

 

 

 

40 60 80
Proportion of direct light

100

 

. obsened
predicted

 

 

Figure 4. Scatterplot of stem density on proportion of direct light at Allegan State
Game Area (ASGA), June-July 1995; “predicted” refers to the fitted frontier
regression: stem density =1.6085 (proportion of direct light) + 96.538; (N = 20;

adjusted r squared = 0.739).

22

.8595 € 2563 .388 G 800%

>588 3 Eucam Au “mamas—o Baguio—u 3:8 2:9: .58 C8 Em: 82% mo :oEonoa 3 $03 .3 L382: 05 Co 3.83me .m SawE

 

2.»: 39:36 55.3..ch

600L9svszul
0000000000

-u-
di-
..
..

 

o u c

S u 5:86
R u .66 a
Q. n 52:

Kauenbeag

 

 

 

001

 

vmoecsovmo

2.»: 66:36 SEES.— Eas 2.88 a

68L9§P£ZI
0000000000

 

A A ii. I O
N 1.
cm": v u
81...... w s m
o. "$ch - w w
c. .A

mun—SQ: N.

 

 

 

   
 

 

 

  

Eu: 395:: 32.339:— coEw .259: 3

8 I. 9 S .7 C» 7y .I.

0 0 0 0 0 0 0 0 0

- . _ o
N
.. w
I . o b
00 I C .... w W
3. u 560:. o. w
A

«N u>oc S m—

3. u SSE 3
Eu: «out—Cc EELS—o...— cooLw Am
00 L 9 S .V C... Z I.

O 0 0 0 O 0 O 0 0
. c

an": m u
a
on u 562: o .w
mm ">38 em a m
"50:. E b

em N.

3

 

  

 

 

   

 

 

 

 

 

23

Greenhouse Study

Wet and dry leaf weight, dry root weight, and leaf area were measured because of
their physiological significance for lupine and probable relationship to KBB. Dry leaf
weight, dry root weight, root/shoot ratio (dry), and wet leaf/dry leaf mass ratio were
significantly (p < 0.01) different among light treatments according to the Kruskal-Wallis
test (Table 3).

The ratio of wet to dry leaf weight was greatest in 35% light (mean = 6.24), and
smallest in 100% light (mean = 4.93). This implies that desiccation is more of a problem
in areas with full sun exposure, and supports the results of the field study, indicating that
shadier areas protect against desiccation. Additional supporting evidence is provided by
the pattern of root to shoot ratio, which was greatest in 100% light (mean = 2.48), and
smallest in 35% light (mean = 1.01). The fact that lupine plants growing in high light
environments have relatively high root to shoot ratios suggests that they have adapted by
proportionally reducing leaf surface area to decrease water loss, and increasing root
surface area to increase water absorption.

However, plants were significantly (p < .05) larger under 100% light than in the
shade treatments (Table 3). Leaf area follows a similar trend, but the relationship was not
significant, probably due to variability among plants within treatments. This implies that
the plants grow larger‘ 11 under high light, potentially resulting in more food for the
KBB, and possibly more competitive lupine plants. With larger roots, the plants would
be better able to compete for nutrients and water. With larger leaves, they would have an

advantage in competing for light and Space.

24

 

 

 

m .1. E83: ..o 083?
And.» HE Abb.v HE Aom.m av
mmo. 36.6 3.3 2.6 $6. 84. 22m 24. mm.m m.m~ SST: 323638.83
Ahm.o av Amm.o av AHH.HHV
moo. Smd $4. £6 34 Tim mmé «Tm «m6 mad 932mm enact—<63
Am~.H av Amm.H «C Amo.o «V
So. 2.0.: mod. 3.0 mm; 84 mv.m 34 84 9mm Tmmméa 33.85563.
.mm.mma av Lev.mm av .mm.mm av
m3. 3.61m mmémv mm.m 3.2: 363 Sim: 2.43 «mam Eva 9323 $588.38..
Amm.a av Amm.o He .vm.o “C
25. mmm.mm mm.m mad mm; «m; :6 mm...“ 3.0 22.. 58mg. @Ew_ofoo.bo
A>N.H «C Amm.o «L Amm.o «v
moo. «3.3 3.6 3.0 .84 an; 34 mm; 36 oém WANTS Ersmezengbo
9:53 Son c252:
uoeauEeufi -EU .32 £2 .3m :32 $2: $3 $3 $2: $3 {own 83.3.34»
.Uzmzaom «nah. fiocmn—EOU 3:0:5a9—h. mesoeuwuhh. ap—wmx— ac mflcmaagon— mufloE~NUL.—. «SM:-
m_=a>) Ezmshv— =< LO.— muzmmuuow Gnu—uzwam “Eu 9‘02 be «:3: Ewe—)—

 

 

 

 

 

 

3 n Z 6.96 m: 8m 8:0: 59% a E :3on 5:3 amen—63 ”ESSA SSSN :o Em: mo Soobm .m 2an

 

25

Discussion

Managing vegetation (and indirectly managing available light) to provide habitat
favorable to KBB appears to involve a balancing act. If KBB require some minimum
amount of accessible lupine (yet to be identified), managers may strive to achieve a target
lupine density to maximize KBB population size. If such a density target falls on the
lupine density frontier, it is easy to identify a minimum level of light that would be
required for lupine to have any chance of meeting this density criterion. However, it is
also necessary to allow for a range of canopy densities and consequent light intensities in
order to offer protection for the second generation larvae from complete lupine
desiccation due to drought. In short, managers should provide habitat with a mix of
optimal environmental conditions for lupine growth and conditions which provide
protection for lupine (and thus, the butterflies) during years with “unusual” weather.

Because lupine are found exclusively on acidic soils at ASGA, it is not surprising
that they are frequently found on soils enriched by the acidic litter of overstory canopies
of oaks (and in other parts of KBB’s range, pines). These tree species may serve to
maintain low soil pH, particularly in areas with frequent fire. The fire regime likely to
have been common in oak savannas in pre-settlement times would have tended to elevate
soil pH (Tester, 1989). Judging by the narrow pH range within which lupine were found
at ASGA, it appears that soil pH affects the distribution of lupine. Maintaining healthy
lupine populations may require a combination of acidic conditions facilitated by
overstory trees and periodic disturbance (primarily fire) that keeps competition levels

low. There may also be a balance between lupine’s light and soil pH requirements.

26

Individual trees often generate a pattern of soil pH in which values increase with distance
from the base of individual trees, with the magnitude of this effect depending on tree
species (Zinke, 1962). Given that preliminary investigations found lower soil pH under
the forest canopy than in adjacent open areas (Mary Rabe, Michigan Department of
Natural Resources, pers. Comm, 1996), one might assume that lupine would grow best
near trees, unless limited by low light. The best sites for lupine appear to be sandy sites
where both light and soil pH are favorable.

Another possibility is that the trees in oak savannas (and elsewhere in KBB’s
range, pine barrens) affect other environmental variables that in turn affect lupine growth
and density. Because trees affect multiple environmental variables, (e.g., soil moisture,
pH, nutrient levels and light) (Anderson, 1991; Crozier and Boemer, 1984) it is likely that
the apparent correlation between light and lupine density may be the result of
confounding factors.

Nutrient availability could be a significant issue in the relatively infertile savannas
and barrens where lupine are found. Numerous studies have shown that soils under trees
have significantly higher concentrations of organic matter, total and available N, Ca, K,
P, Na, and S (Kellman, 1979; Belsky etal., 1989; Young, 1989; Vetaas, 1992). Because
lupine fix nitrogen from the atmosphere, they may not benefit from this elevated nutrient
availability, and may in fact, be at a competitive disadvantage relative to non-nitrogen
fixing vegetation.

As an early successional, nitrogen-fixing species, lupine may have developed an
evolutionary strategy that is tailored to survival on depauperate, moisture-deprived,

barren sites and improving such sites via nitrogen fixation. Lupine is probably at a

27

competitive disadvantage on better sites. This nuance complicates interpretation of the
positive association between lupine density and light. It could reflect a physiological
light requirement, decreased competition from other plants in the nutrient or moisture-
limited open areas, or a combination of these factors.

One artificial element of the greenhouse study design introduced some caveats
that must be considered when interpreting results. All pots were watered when weighing
indicated a soil moisture below 35%. Presumably due to greater evapotranspiration rates,
plants in the 100% light treatment were watered more often than those in the 65% or 35%
treatments. Under field conditions, lupine grow on sandy soil where available water may
be limited, and the evapotranspirative stress characteristics of open areas (i.e. high light)
would cause desiccation and perhaps premature senescence. Another complication
introduced by the watering regime concerns nutrient availability. Because plants were
watered with a fertilizer solution, plants which received more water also received more
nutrients. If one or more nutrients were limiting, this would have unintended
consequences by introducing an independent variable (nutrition) that was not part of the

experimental design.

28
Conclusions

Light and soil pH appear to affect lupine growth and density. Results indicate that
a pH range of 4.2 to 5.6 and the highest available light provide the best lupine habitat.
However, because of the potential for weather extremes such as early summer drought
and late spring frost, land managers protecting the KBB should aim to provide lupine
habitat that covers a range of light levels. The easiest way to accomplish this goal is via
management of stand density. The habitat diversity thus created is especially important
for an endangered species. Managers need to be very cautious and not only provide the
best habitat for lupine, but be certain to provide protection against unusual or extreme
conditions.

Further research is needed to determine how much lupine is required to support
KBB populations. This study provided information about two tree-dependent
environmental variables and documented their relationship with lupine. Ultimately, this
improved understanding of lupines requirements should lead to a more complete
description of KBB’s habitat requirements. Information on the effects of other
environmental variables such as nutrient and water availability on lupine density would

further improve a habitat model of lupine and the KBB.

29

BIBLIOGRAPHY

Andersson, T. 1991. Influence of stemflow and throughfall from common oak (Quercus
robur) on soil chemistry and vegetation patterns. Canadian Journal of Forest
Research 21 :917-924.

Bess, J .A. 1989. Status of the Karner blue butterfly, Lycaeides melissa samuelis
Nabokov, in the Manistee National Forest. Report to Michigan Natural Features
Inventory, Lansing, Michigan 20 p. (unpublished).

Boyonoski, A.M. 1992. Factors affecting the establishment and maintenance of Lupinus
perennis, (wild lupine). Unpublished MS. Thesis, University of Guelph, Ontario.

Bray, J .R. 1955. The savanna vegetation of Wisconsin and an application of the
concepts order and complexity to the field of ecology. Unpublished Ph.D.
Dissertation. The University of Wisconsin, Madison. 251 p.

Chan, S.S., R.V. McCreight, J .D. Walstad, and TA. Spies. 1986. Evaluating forest
vegetative cover with computerized analysis of fisheye photographs. Forest
Science 32:1085-1091.

Clark, J .S. 1991. Disturbance and population structure on the shifting mosaic landscape.
Ecology 72:1119-1137.

Cowling, W.A., J .C. Clements. 1993. Association between collection site soil pH and
chlorosis in Lupinus angustifolius induced by a fine-textured, alkaline soil.
Australian Journal of Agricultural Research 44: 1821-1836.

Crozier, Carl R., and E. J. Boemer. 1984. Correlations of understory herb distribution
patterns with microhabitats under different tree species in a mixed mesophytic
forest. Oecologia 62:337-343.

Cryan, J .F., and R. Dirig. 1978. A report on the status of the Kamer blue butterfly
(chaeides melissa sarnuelis Nabokov) in New York State. Report for the New
York State Department of Environmental Conservation. August 29, 1978. 18 p.

Curtis, B. 1959. The vegetation of Wisconsin. The university of Wisconsin press.
Madison, Wisconsin. 657 p.

Dirig, Robert. 1973. The endangered Kamer blue. The Conservationist 28:6,47.
Dirig, Robert. 1988. Nabokov's blue snowflakes. Natural History 97:68-69.

Haack, RA. 1993. The endangered Karner blue butterfly (Lepidoptera: Lycaenidae):
Biology, management considerations, and data gaps. Proceedings of the 9th

30
Central Hardwood Forest Conference. U.S.D.A.-Forest Service, North Central
Forest Experiment Station. St. Paul, Minnesota. p. 83-100.

Herms, Catherine P. 1996. The endangered Kamer blue butterfly (Lepidoptera:
Lycaenidae) in Michigan: Habitat suitability, potential impacts of gypsy moth
(Lepidoptera: Lymantriidae) suppression and laboratory rearing. Unpublished
Masters Thesis, Department of Entomology, Michigan State University.

Holland, V.L. 1973. A study of soil and vegetation under Quercus douglasii compared
to open grassland. Unpublished PhD. Dissertation, Department of Botany,
University of California, Berkeley. 359 p.

Hultman, G. 1978. Trees, Shrubs, and Flowers of the Midwest. Great Lakes Living
Press, Chicago. 77 p.

Loehle, Craig, and G. Wein. 1994. Landscape habitat diversity: A multiscale
information theory approach. Ecological Modeling 73:311-329.

Maxwell, J ., and T. Givnish. 1993. Research on the Kamer blue butterfly at Fort
McCoy, Wisconsin: Progress report for the 1993 field station. Report submitted
to the US Fish and Wildlife Service, and the Department of the Army. 22 p.

Parker, V.T., and CH. Muller. 1982. Vegetation and environmental changes beneath
isolated live oak trees (Quercus agrifolia) in a California Annual Grassland.
American Midland Naturalist 107269-81.

Pickering M.G., and SB. Gebauer. 1993. Status report: Assessment of associations
between lupine population characteristics and vegetation and soil characteristics -
1991-92. Eastern New York Chapter of the Nature Conservancy. (unpublished)

Rabe, M. 1995. Personal communication.

Rich, RM. 1989. A manual for analysis of hemispherical canopy photography. Los
Alamos National Laboratory. Los Alamos, New Mexico. 80 p.

Shuey, J.A., J .V. Calhoun, and DC. Iftner. 1987. Butterflies that are endangered,
threatened, and of special concem in Ohio. Ohio Journal of Science 87:98-106.

Smallidge, P.J., and DJ. Leopold. 1992. Environmental factors and critical habitat for

Lycaeides melissa samuelis along transmission line rights-of-way in east-central
New York, USA.(Draft).

Spies, T.A., W.S. Ripple, and GA. Bradshaw. 1994. Dynamics and pattern of a
managed coniferous forest landscape in Oregon. Ecological Applications 4:555-
568.

Stewart, M.M., and C. Ricci. 1988. Dearth of the blues. Journal of Natural History 92:
64-70.

Tang, Caixian, J. Kuo, N.E. Longnecker, C.J. Thompson, and AD. Robson. 1992. High
pH causes disintegration of the root surface in Lupinus angustifolius L. Annals of
Botany 71:201-207.

a‘

.31

Tang, Caixian, and AD. Robson. 1993. pH above 6.0 reduces nodulation in Lupinus
species. Plant and Soil 152:269-276.

Thomson, James, D.G. Weiblen, B.A. Thomas, S. Alfaro, and P. Legendre. 1996.
Untangling multiple factors in spatial distributions: Lilies, gophers, and rocks.
Ecology 77:1698-1715.

Welch, Robert J. 1993. Dispersal and colonization behavior in the Kamer blue butterfly
(Lycaeides melissa samuelis) in central Wisconsin. Final Report to US. Fish and
Wildlife Service. Green Bay, Wisconsin. 50 p.

Wisconsin Department of Natural Resources Kamer Blue Technical Team. 1993. 3“1
draft: Wildlife Management Guidelines for Kamer blue. (unpublished).

Zinke, PJ. 1962. The pattern of influence of individual forest trees on soil properties.
Ecology 43:130-133.

Chapter 3

The Effect of Landscape Level Spatial
Parameters and Forest structure on the distribution of

Lupinus Perennis and Lycaeides melissa samuelis Nabokov.

32

33

Abstract

The Kamer blue butterfly (KBB), Lycaeides melissa samuelis Nabokov,
(Lepidoptera: Lycaenidae) was added to the federal endangered species list in December
1992 following a period of extensive habitat decline. To restore remnant populations of
Kamer blue butterflies to sustainable levels, land managers need to consider the impacts
of their actions on lupine, the Kamer blue’s sole larval food plant. Because KBB fits in a
metapopulation model, I expected that the spatial parameters of the KBB’s habitat would
affect its prospects for survival. In forested landscapes, spatial metrics (e.g., amount of
edge, connectedness of habitat patches, distance between habitat areas, patch shape and
patch size) have been directly linked to flora and fauna species composition. In this
study, exploratory analysis of stand data and lupine and KBB distribution was conducted
to determine (1) if the landscape properties required by KBB could be identified and (2)
if characteristics of forest stands could be linked to KBB habitat requirements. To do
this, stand (e.g., species, age, and size class) and landscape (e.g. patch size, density, and
shape, and nearest neighbor) characteristics were contrasted for two comparisons: (l) for
stands where lupine was absent (LA) and stands where lupine was present (LP), and (2)
for stands where lupine was present and KBB was absent (LPKA) and stands where
lupine and KBB were present (LPKP). Overall, LPKP patches tended to be closer to one
another, more complex, and larger than LPKA. Stands of medium density contained
more LPKP sites than LPKA sites. LP areas were more likely to be in the low stand

density class and much less likely to be in the high stand density class than LA.

34
Introduction

The Kamer blue butterfly (Lycaedies melissa samulis) (KBB) was added to the
federal endangered species list in December 1992 following a period of extensive habitat
decline (Haack, 1993; Dirig, 1988). Scientists have long known that habitat preservation
is the key to species preservation (e.g., Loehle and Wein, 1994). This is particularly
evident in the case of the Kamer blue butterfly, one of the most habitat dependent species
in the northeastern US (Dirig, 1988). Recently, scientists have been working on a habitat
description for the Kamer blue. In Michigan, the habitat has been described as oak
savanna with an understory of lupine and other nectaring plants (Pickering and Gebauer,
1993; Shuey, et al., 1987; Welch, 1993). Because this description does not address such
specifics as the amount of lupine required, the size, shape, and distribution of suitable
areas, or the ecological requirements of lupine, managers lack a critical tool for restoring
KBB populations. The first goal of this research is a more precise description of the
spatial component of Kamer blue’s habitat requirements.

The savanna type has been described as a structurally simple, seasonally water-
stressed community with a sparse overstory of trees and shrubs that forms structural
discontinuities above the grassland (Belsky and Canham, 1994). Trees invade savannas
via seeds dispersed by wind and animals and by clonal spread from adjacent woodlots. In
pre-settlement times, the size and age structure, species diversity, and dynamics of
savanna vegetation were controlled by local and regional scale disturbances such as fire
(low intensity surface, as well as “stand-replacing” fire), weather (e.g., wind-throw,

tornado), and herbivory. Savannas can only be maintained by a relatively high frequency

35

of these succession-interrupting disturbances (Tester, 1989). The oak savanna forest type
favored by the Kamer blue has become increasingly rare due to fire exclusion (Welch,
1993; Shuey, er. al., 1987) and increasing urbanization (Givnish, et al., 1988; Smallidge
and Leopold 1992; Cryan and Dirig 1978). The Kamer blue is now commonly found
within small patches of L. perennis resulting from human disturbance such as along gas
pipeline rights-of-way, trails, roads, and abandoned farm fields (Haack, 1993).

To restore remnant populations of Kamer blue butterflies to sustainable levels,
land managers need to consider the impacts of their actions on lupine, the Kamer blue’s
sole larval food plant (Haack, 1993; Smallidge and Leopold, 1992; Stuart and Ricci,
1988; WI DNR, 1993). Lupinus perennis is a member of the pea family and is the only
lupine native to the eastern US (Hultman, 1978). A better understanding of lupine’s
ecological requirements is a critical first step towards managing the recovery of the
Kamer blue. Lupine habitat is likely affected by trees (e.g., light, soil moisture, soil
chemistry). It is therefore possible that lupine habitat can be described at a landscape
level as relationships among forest variables such as size class and age class. Another
goal of this research is to see whether these relationships exist.

Structuremand connectivity are two classes of landscape properties which control
the distribution of species over the landscape (F orman and Godron, 1986; LaGro, 1991).
Structural properties include the shape, size, and juxtaposition of habitat patches (Burgess
and Sharpe, 1981; Forrnan and Godron, 1986). Connectivity properties affect the

potential for movement of organisms within a landscape (Baudry and Merriam, 1988;

Janssens and Gulinck, 1988).

36

Landscape connectivity can be influenced by the proportion of the landscape in
so-called “edge” communities which differ from those found in “pure” cover types
(Malcolm, 1994). The proportion of a patch characterized as edge varies with patch
structure (e.g., shape and size of the patch) (LaGro, 1991). Small patches may consist
entirely of edge-modified habitat (Kapos, 1989). The effects of edge may extend far into
convoluted patches, ultimately changing landscape connectivity (Alverson, et al., 1988;
Janzen, 1983; Johnson, et al., 1981; Kapos, 1989; N055, 1983; Ramney, et al., 1981;
Temple and Cary, 1988). Edge may play an important role for KBB management by
providing effective connectivity between habitat patches not previously recognized by
scientists and land managers as connected or connectable.

In forested landscapes, amount of edge, connectedness of habitat patches, distance
between habitat areas, patch shape and patch size have been directly linked to flora and
fauna species composition (Franklin and Forrnan, 1987; Harris, 1984; Thomas, 1979).
For many species, the spatial extent of landscape properties that affect habitat quality may
be many times greater than the aggregate of the home ranges of the individuals within a
population (Gustafson, et al., 1994). Landscape properties can also influence the
suitability of individual patches (Bach, 1988). For example, diversity of wintering bird
populations found within individual patches was found to be a landscape level function
determined by distribution of habitat within the forest matrix (Pearson, 1993).

Habitat fragmentation produces isolated patches, with an increased likelihood of
deme extinction due to a lack of immigration possibilities, particularly for smaller patches
with low carrying capacities (Harrison, 1991; Miles, 1994). Agriculture, urbanization,

and forestry practices have substantially altered the pre-settlement land cover patterns

37

over the past century, resulting in a landscape consisting of a mixture of patches
generated by natural processes and human activities (Krummel et al., 1987). A better
understanding of how this mixture of patches affects species composition at the landscape
level is critical if preservation efforts are to be successful (Loehle and Wein, 1994). The
effect of landscape properties on metapopulation dynamics appears likely to be important
in the case of the Karner blue,

A metapopulation is a set of populations persisting over space and time via

waxing and waning, extinction and recolonization (Ham'son, 1991). Four broad types of
metapopulations have been theorized: source-sink, patchy, non-equilibrium, and classical.
In the source-sink model, source demes are considered critical to metapopulation
maintenance, while sink demes are inconsequential or a hindrance in terms of survival.
In patchy metapopulations, high rates of dispersal tend to demographically unite highly
fragmented landscapes. Non-equilibrium metapopulations tend to decline due to low
recolonization relative to extinction. Classical metapopulations are those where a
dynamic balance between colonization and extinction is maintained (Harrison, 1991).

Harrison (1991) suggests the non-equilibrium model as the best fit for the Kamer

m M“
“- ww.

blue on the grounds that the ongoing reduction of their habitat can only lead to extinction.(/
Others have suggested that small, isolated populations of Kamer blue are often at risk of
local extinction from random demographic events when there is no replenishing effect

due to immigration from nearby populations (e. g. Givnish et al., 1988). However, these
dynamics may be relatively recent. Though the Kamer blue has low motility (less than

1.5 km) (Haack, 1993), the species has long thrived on lupine patches that were

constantly shifting in both time and space due to fire, herbivory, and succession (Haack,

38

1993; Smallidge and Leopold, 1992). A continuous series of local extinctions from and
recolonizations of these shifting patches of habitat probably accounts for the ongoing
stability of Kamer blue metapopulations in such dynamic environments. These dynamics
would only have been possible with a high degree of patch connectivity (Harrison, 1991).
This implies that Kamer blue metapopulations may be better fit by the patchyrnodel. V;

Local population (also known as deme) extinction and habitat patch
recolonization are central to metapopulation models (Harrison, 1991). Because both
processes depend on the spatial distribution of demes within a metapopulation, landscape
structure and connectivity are critical variables. For example, larger islands produce
more migrants, the probability of immigration decreases exponentially with distance, and
the probability of extinction is inversely related to the carrying capacity of the patch
(Gilpin, 1987). Thus an understanding of the spatial parameters of KBB habitat is critical
for managing and sustaining KBB metapopulations.

Deme extinction depends on genetic, demographic, and environmental forces
acting within the patches (Gilpin, 1987). For example, genetic stochasticity (the loss of
heterozygosity through drift and inbreeding), a frequent result of patch isolation,
increases the risk of deme extinction (Harrison, 1991).

Movement of individuals among patches is critical for maintaining stable
populations via recolonization (Backs and Eisfelder, 1990; Ellis and Lewis, 1967) and is
dependent on landscape connectivity (Baudry and Merriam, 1988; Gilpin, 1987; Janssens
and Gulinck, 1988). By itself, this would imply that managers should strive for
connectivity. But, environmental stochasticity may have a greater influence on

metapopulations than genetic stochasticity because it can affect and synchronize many

39

demes simultaneously across the landscape (Comins and Noble, 1985; Harrison, 1991;
Hastings and Wolin, 1989; Pacala, 1989; Shugart, 1984; Turner et al., 1989; Warner and
Chesson, 1985). Therefore, it may also be very important to maintain unconnected
refuges which can act as population sources following metapopulation extinction brought
on by localized catastrophic events (Ham'son, 1991).

The Kamer blue exemplifies species requiring such a balancing act. Although
Kamer blue demes are spatially distinct, exchange of genetic material is believed to occur
among them as a result of dispersal (Givnish, et al., 1988; Maxwell and Givnish, 1993).
Because Kamer blue habitat can only be maintained through disturbance (e.g., fire, which
is lethal to Kamer blue individuals at all stages in their life cycle), a Kamer blue
metapopulation’s risk of total extinction may be reduced by a horizontally complex
distribution of habitat across the landscape (Maxwell and Givnish, 1993). Thus the best
landscape structure may require a balance between connectivity and patch isolation.

Assuming that KBB is well represented by a metapopulation model, the survival
of this species is likely to be affected by landscape properties such as structure and
connectivity. Land managers have lacked information about the linkage between KBB
metapopulation structure and the effect of their management on landscape properties. It
is particularly important to understand the relationship between landscape

structure/connectivity and metapopulations because the ultimate conservation goal for

‘h-O—

wh—J‘

w...-—— «a...»

/his species is a shift from the non-equilibrium to the patch metapopulation model via
management of landscape structure and habitat quality. Exploratory analysis of stand

data and lupine and KBB distribution was conducted to determine (1) if landscape

40
properties required by KBB could be identified and (2) if characteristics of forest stands

could be linked to KBB habitat requirements.

Methods

Data processing and analysis

Allegan State Game Area (ASGA) in southwest Michigan’s Allegan County was
selected as the location for this study because it is one of the two major areas in Michigan
where Kamer blue are found today, and is the only Michigan area for which spatially
referenced information was available at the initiation of this study. Stand boundary and
attribute data were provided by game area managers. Locations at which lupine were
observed in 1995 and locations where KBB were observed over the past decade were
digitized. While lupine has been observed throughout ASGA, KBB sightings have been
confined to lupine patches south of the Kalamazoo River (Figure 7), possibly due to the
river forming a recently insurmountable barrier to movement of KBB. For this analysis, I
considered (1) lupine patches south of the Kalamazoo river habitat as one metapopulation
of KBB and (2) lupine patches north of the river as unutilized potential habitat for a
second KBB metapopulation. A key question was whether or not there are landscape
properties associated with lupine patches north of the river which preclude utilization by
the KBB.

The ASGA stand coverage was provided in a proprietary GIS format (C-map).
This data was converted to an Arc/Info vector coverage. Because the stand information
provided in this coverage (e.g. size and age class, and predominant overstory species) was

lumped into one attribute, reclassification was required to generate stand attributes (Table

41

4) suitable for use as query criteria and subtotal categories. The approximate geographic
locations of KBB sightings from this decade and 1995 lupine observations were recorded
on and provided as a paper mapb. Approximate on-screen digitizing7 was employed
because the paper map was of “field quality” and subject to distortions which would have
severely limited the accuracy of coverages generated by tablet digitizing or scanning
methods. Once completed, these vector coverages (Table 4) were converted to a raster
format with a 15.23 111 cell size (deemed adequate given the limited accuracy of the
source coverages). I analyzed the lupine and KBB data at the stand level (e.g., if any
portion of a lupine stand was coded as having lupine present, it was coded as lupine
present (LP)) for two reasons : (1) more confidence in accuracy of stand boundaries than
patch boundaries for lupine or KBB; (2) stands are the level of resolution at which forest

vegetation is managed.

 

5 Provided by Mary Rabe at The Michigan Natural Features Inventory
7 Using the ASGA coverage as a backdrop, lupine and KBB patches were sketched on screen with a mouse.
These resulting polygons were attributed and built into vector coverages.

42

Table 4. Geographic information system (GIS) coverages and associated
attributes used to analyze landscape properties and their relationship to KBB and lupine

habitat at Allegan State Game Area, Allegan Michigan, 1995.

 

 

 

 

 

 

Variable Variable Type Data Type Classification Codes Source
Tree density independent ordinal O-non forest ASGK' stand map
l-low
2-medium
3-high
Tree size independent ordinal O-non-forest ASGA stand map
l-sapling
2-pole timber
3-saw timber
Lupine dependent! nominal 0-lupine absent on-screen digitizing
independent l-lupine present of MINFIz paper
map overlay on
ASGA stand
boundaries and
reclass of ASGA
stands
Kamer blue dependent nominal 1-lupine present, KBB on-screen digitizing
butterfly absent of MINFI paper
2- lupine present, KBB map overlay on
present ASGA stand
boundaries and
reclass of ASGA
stands

 

 

 

 

 

‘ ASGA - Allegan State Game Area
2 MINFI - Michigan Natural Features Inventory

 

43

Stand (e.g., species, age, and size class) and landscape (e.g., patch size, density,
and shape; and nearest neighbor) characteristics were contrasted for two comparisons:
(1) for stands where lupine was absent (LA) and stands where lupine was present (LP),
and (2) for stands where lupine was present and KBB was absent (LPKA) and stands
where lupine and KBB were present (LPKP) (Figure 7). Stand characteristics were also
summarized for all stands (AS) to provide a baseline for comparison. The landscape
characteristics were generated by processing GIS coverages with F RAGSTATS
software”. Two new coverages were generated by reclassifying the stands coverage
according to the values of the lupine and KBB attributes (Table 4, Figure 6). Two types
of spatial metrics were used: those calculated at the level of the individual patch (e.g.,
area, patch density, size and shape, and edge metrics) and those which quantify spatial
relationships among patches and the surrounding matrix (e.g., nearest neighbor metrics)

and are calculated at the landscape level (McGarigal and Marks, 1993).

 

' FRAGSTATS version 2, Unix raster version by Kevin McGarigal and Barbara J. Marks. 1995

 

 

; 0 3 6 9 7‘ 12 Kilometers (
‘ E—W LAKA ‘ ' l
I LPKA i

-LPKP .
I L, NotASGA ,

Figure 6. Map of areas with lupine present, KBB present (LPKP); areas with lupine
present, KBB absent (LPKA), and areas with lupine absent, KBB absent (LAKA) at
Allegan State Game Area (ASGA), Allegan, MI 1995.

45
The patch indices were grouped by class (e.g., AS, LA, LP, LPKA, LPKP). Class

area (CA)9 (in ha) and percent of total landscape area (PCTLAND)'° were used to
describe the area of each class. Although area metrics do not quantify landscape pattern,
they do quantify landscape composition (McGarigal and Marks, 1993) which may prove
important to KBB. Patch density, size, and variability metrics include number of patches
(NP)”, patch density (PD)'2 (#/ 100 ha) and mean patch size (MPS)l3 (ha). Edge is
another metric quantified at the patch level. Edge density (ED)" has the advantage of
removing the issue of differences in class area. Because shape is difficult to quantify, I
focused on two indices which characterize shape-related properties. The first, edge to
area ratio, captures complexity but not morphology. The area weighted mean shape

index (AWMSI)'5 statistic calculated in F RAGSTATS was appropriate for this study

 

fl
9 CA = Z 0,3- (T000) where a”~ = area (m2) of patch j in class i; n = number of patches in the landscape
j=l .

n
.2 “:1
F1
A

 

'° PCTLAND= where A = total landscape area (m2)

” NP, =n,- where n, = number of patches in the landscape of patch type (class) i..

'2 PD,=n—A'(10,OOO)(100)

' ill—1 I
Hi 10,000

eik

" ED=k—=:r-(10,000) where m' = number of patch types (classes) present in the landscape; e“, = total

length (m) of edge in landscape between patch types (classes) i and k.

TM:
a

 

I3 MP5, =

M3.

-

 

 

'5 AWMSI= X where p,' = perimeter (m) of patch j of class 1.

=1 ’0“ n
1:1

 

 

L

46
because the areas of stands with LA, LP, LPKP, and LPKA are not equal. The other

2
shape metric used was fractal dimension. This involves the relationship A = k? D , where

P = perimeter; D = the fractal dimension; k = a constant; A = area. As shapes become
more complex, D approaches 2 (Burrough, 1986; McGarigal and Marks, 1993). The area
weighted statistic (area weighted mean patch fractal dimension, AWMPFD)“5 is the
recommended technique for use with small sample sizes, (in our case n < 20).

At the landscape level, two nearest neighbor statistics were calculated to test for
an effect from being near another like patch (e.g., does it matter to KBB if patches of
suitable habitat are near each other or not?) Mean nearest neighbor (MNN)I7 was used to
determine concentration levels of patch types and nearest neighbor standard deviation
(NNSD)ls was used to describe the regularity of patch distribution within classes, that

have nearest neighbors.

 

 

n 21n 25p-- 0..
'6 AWMPFD= z [ I ”I] '1
j=I

 

 

”I

2h,-

1=|
n'i
class i, based on edge-to-edge distance; n’i = number of patches in the landscape of patch type (class) i

'7 MNN=

 

where hi]. = distance (in meters) from patch j in class i to the nearest neighboring patch of

 

P f n' \72
n' :hlj
2 (hvl‘ 14'.

l

 

 

 

 

 

 

 

" NNSD = ‘ ‘

 

47

The raster coverages for lupine and KBB presence generated in Arc/Info were
converted to ERDAS image file format and processed by FRAGSTATS. Global
parameters used in FRAGSTATS included core area = 10 m, travel distance = 1 km, and
diagonal travel allowed. F RAGSTATS output was imported to a spreadsheet” format for
additional analysis and summarization of results. Two sample t-tests were used to
statistically compare the output of the F RAGSTATS analysis for the patch level statistics.

To determine if a relationship existed between forest properties and KBB and
lupine distribution, 1 plotted histograms in Are View for each pair of comparisons. The
amount of area corresponding to each category (e. g., LA, LP, LAKA, and LAKP) in each
class of each forest variable (e.g., size class, species, density class) was transformed to a

percentage of total area for plotting in bar charts.

 

'° Microsoft Excel version 7

48

Results

Stands where lupine and KBB are both present comprise a larger area (2274 ha)
than stands where lupine are present and KBB are absent (729 ha) (Table 5). It is
possible that the aggregate size of lupine patches north of the Kalamazoo river may be
too small to support a metapopulation. The mean patch size (MP8) for all continuous
lupine patches (across stand boundaries) (53.6 ha) is less than that for lupine patches with

KBB (88.3) but the difference is not statistically significant (p = 0.05).

49

.008 0.0.3330 29:00 95 3:8 95 05 08 God v 3 EEobE 3:80:in 20 a: 0:0 <4 me 0:32 N

.mvK 25 «..vK 0853 23 x: can <1— :0223 000E 0:033:80 _

 

 

 

 

 

 

 

 

 

 

 

0.3 0.00 34 8.0 00.0 4.0: 00.0 on 8.: 0000 ACE:
«.000 N mimom 34 00.0 004 0.0». 00.0 0m 3.0 03. <vr:
0.03 mama 34 00.4 3.0 4.00 0.0.0 00 NW: Son a:
Emma Emma 34 00.0 mm.m H.000 50.0 mm 04.00 $0: <4
AEV commcuEE x205
:o:£>ou AEV 300$ 30% Ct
23:30 Shims: :83 :02: :02: A935 A05 ONE A0: 2: 30 3:83 go A05
Shims: H0200: “08:33, “025303 56:20 :83 .3220 mo @00— 020
002002 0002 02¢. 02< owvm :82 seam 038:2 :5on 000—0 959
02.32: 3.50:. 02.50:. b=3010>
Lee—wt: .0232 3.32: 2.3m outm— cns Jim :82. £205.. :82. 823:. «0.5.

 

.EVEAV E805 Eon mm! was 2:9: Ea .Afimd E030 mmM 9.0 0:085 059: And 05on

0&9: .26 0:030 2:9: firs 820 Sm mH<HmO<¢m an 39:0 30:38.». 38% mo .mcatmafioo 80$qu :32 000—0 .m 030,—.

 

50

Patches where lupine and KBB are present have an edge density which is more
than two times as large as lupine patches with no KBB (2.4 m/ha vs.l.l m/ha), implying
that KBB may slightly favor a preponderance of edge. AWMSI for LPKP patches (4.95)
was not significantly different than for LPKP patches (2.99). Stands with LPKP were
more concentrated (p < 0.05) (mean nearest neighbor (MNN) = 69.9 m) and more regular
(nearest neighbor standard deviation (NN SD) = 57.8 m) than LPKA (MNN = 208.1 and
NNSD = 259.2).

I considered the frequency of stands by overstory density class for the
qualifications: 1) all stands, 2) lupine absent, 3) lupine present, 4) lupine present, KBB
absent, 5) lupine and KBB present (Figure 7). Although both lupine and KBB can be
found in every overstory density class, there appears to be a negative relationship
between overstory density and the likelihood that a stand will contain lupine and KBB
(Figure 7). However, both lupine and KBB are more likely to be present in low density
forest than in open grassland, implying that optimal conditions do not include light levels
associated with 100% sky. I also compared the frequency of stands by overstory size
class for the qualifications: 1) all stands, 2) lupine absent, 3) lupine present, 4) lupine
present, KBB absent, 5) lupine and KBB present. The relationships did not appear as
clear as they did with the density classes (Figure 8). It appears that LPKP areas are
relatively more prevalent under grass and sapling size classes than LPKA, while LPKA

have more area in the larger size classes than LPKP areas.

51

 

 

 

 

///////////////////////////////,

 

',//

 

///////

 

 

’//////////////

\

 

Stand qualifiers

 

 

 

Figure 7. Frequency of stands by density class for the qualifiers: all stands (AS),
lupine absent (LA), lupine present (LP), lupine present, KBB absent (LPKA), and
lupine present, KBB present (LPKP) at Allegan State Game Area, Allegan, MI,
1995.

 

 

 

 

 

90

80

70
I
'2 60 Home:
I
E 50 leaping
o
E 40 Ipole
0 law
2: 30

20

 

0

AS LA LP LPKA LPKP
Stand qualifiers

 

 

 

Figure 9. Frequency of stands by size class for the qualifiers all stands (AS), lupine
absent (LA), lupine present (LP), lupine present, KBB absent (LPKA), and lupine
present, KBB present (LPKP) at Allegan State Game Area, Allegan, MI, 1995.

52

Discussion

There appears to be a negative relationship between tree density and both lupine
and KBB, yet both are able to survive amongst large trees. I think the density relationship
is indicative of a positive correlation with light. However, both lupine and KBB appear
to prefer low density forest to open grassland, implying that the optimal light threshold is
less than 100 percent. I also think this is good news for forest managers with KBB on
their land. Through careful thinning, they can provide for the butterflies’ needs (lupine)
as well as continue producing timber. In fact, I think that although lupine and KBB
appear to be more common in low density stands than grassland and higher density forest
stands, it may be wise to manage for a mosaic of stands with differing densities. This
will provide ample habitat with high light levels and protected areas to serve as lupine
refugia during years with early summer drought or late spring frosts.

I think my results demonstrate that the spatial arrangement of these lupine patches
is important based on the nearest neighbor results which showed the area which sustains a
KBB metapopulation has patches that are statistically closer to one another. This means
that managers attempting to perpetuate a metapopulation of KBB will need to manage the
forest from the landscape level. This entails coordinating preservation efforts with
regards to size, edge, and connectivity.

Based on the two potential metapopulations studied, it appears that over 2000 ha
of habitat with a large proportion of edge is necessary. Connectivity also seems to be
important. It appears that individual patches of habitat which are intended to be an active

part of the metapopulation should be approximately 100 m apart from one another or

53

closer. However, in keeping with metapopulation theory and concerns about extinction,
some patches of KBB should be maintained at a distance greater than 70 m from the rest
of the metapopulation to provide some protection against localized catastrophic events.
This study only included two distinct areas for analysis. Many of the comparison
were not statistically significant (p > 0.1), however metapopulations of KBB are quite
rare so the opportunities to conduct replicated experiments are limited. This research
represents a starting point towards building a knowledge base about these ecological
relationships. To develop a metapopulation model, further research is suggested in the
following areas. First, KBB demographic information must be collected. Second, in this
study, I defined KBB habitat solely by the presence or absence of lupine, however, KBB
is known to have other requirements including ants and nectar sources (Haack, 1993).
These must be mapped and included in habitat model-building. Third, information about
KBB flight potential through different forest types is needed. For example, can KBB
travel as far in high density forests as in low density forests, how far can they travel
without relying on nectar sources? In the interim, studies similar to the one presented
here should be carried out for other KBB metapopulations to further test the hypotheses

advanced here.

Conclusions

I achieved my goals of developing a more precise description of the spatial
component of KBB habitat and exploring the relationships between lupine and forest
stand characteristics. Much work remains, but this study constitutes a good beginning

because it provides information about spatial parameters likely to affect metapopulation

54
dynamics, and forest attributes that affect KBB habitat quality. LPKP patches tend to be

closer to one another than LPKA patches. LPKP patches are more common in medium
density stands than are LPKA patches and LP patches were more likely to be in the low
forest density class and much less likely to be in the high density class than were LA
patches.

Although managing a forest for multiple use and for KBB protection may appear
challenging, a GIS containing vegetation attributes can facilitate this process. I do not
suggest that managers should attempt to achieve an exact match to the landscape
parameters for the KBB metapOpulation described in this paper; however, they may be a
good first approximation. I think the relationship among trees, lupine and KBB
distribution should be investigated fiirther and that the spatial parameters should be tested

for other KBB metapopulations.

55

BIBLIOGRAPHY

Alverson, W.S., Waller, D.M., and Solheim 8.8., 1988. Forest too deer: Edge effects in
northern Wisconsin. Conservation Biology 22348-358.

Bach, CE, 1988. Effects of host plant patch size on herbivore density: Underlying
mechanisms. Ecology 69:1 103-1 1 17.

Backs, SE, and Eisfelder, CH, 1990. Criteria and guidelines for wild turkey release

priorities in Indiana. Proceedings of the National Wild Turkey Symposium
6: 134-143.

Baudry, J ., Merriam, H.G., 1988. Connectivity and connectedness: Functional versus
structural patterns in landscapes. Connectivity in Landscape Ecology.
Proceedings of the 2nd International Seminar of the International Association of
Landscape Ecology p. 3-28.

Belsky, A.J., Amundson, R.G., Duxbury, J. M., Riha, S.J., Ali, A.R., and Mwonga, S.M.,
1989. The effects of trees on their physical, chemical, and biological

environments in a semi-arid savanna in Kenya. Journal of Applied Ecology
26:1005-1021.

Belsky, A.J., and Canham, CD, 1994. Forest gaps and isolated savanna trees: An
application of patch dynamics in two ecosystems. BioScience 44:77-84.

Burgess, Robert L., Sharpe, D. M., and Brunet, M.C., 1981. Forest island dynamics in
man-dominated landscapes. Eds. Burgess, R. L. and Sharpe, D. M. New York,
Springer-Verlag. 310 p.

Comins, H.N., and Noble, LR, 1985. Dispersal, variability, and transient niches:

Species coexistence in a uniformly variable environment. American Naturalist
126:706-723.

Crozier, Carl R., and Boemer, E.J., 1984. Correlations of understory herb distribution
patterns with microhabitats under different tree species in a mixed mesophytic
forest. Oecologia 62:337-343.

Cryan, J.F., and Dirig, R., 1978. A report on the status of the Kamer blue butterfly
(Lyceides melissa samuelis Nabokov) in New York State. Report to the US. Fish
and Wildlife Service.

Dirig, Robert, 1988. Nabokov's blue snowflakes. Natural History 97:68-69.

Ellis, J .E., Lewis, J .G., 1967. Mobility and annual range of wild turkeys in Missouri.
Journal of Wildlife Management 31 :568-581.

56

Forrnan, R.T.T., and Godron, M., 1986. Landscape Ecology John Wiley and Sons. New
York, New York. 619 p.

Franklin, J .F ., and Fonnan, R.T.T., 1987. Creating landscape patterns by forest cutting:
Ecological consequences and principles. Landscape Ecology 1:5-18

Gilpin, Michael E., 1987. Metapopulation structure and population vulnerability. In:
Viable populations for conservation. Cambridge University Press. New York,
New York. p 125-139.

Givnish, T.J., Mengis, ES, and Schweitzer, D.F., 1988. Minimum area requirements for
long-term conservation of the Albany Pine Bush and Kamer blue butterfly: An
assessment. (unpublished).

Gustafson, Eric J ., Parker, G.R., and Backs, SE, 1994. Evaluating spatial pattern of
wildlife habitat: A case study of the wild turkey (Meleagg's gallogavo). Midland
Naturalist 131 :24-33.

Haack, R.A., 1993. The endangered Kamer blue butterfly (Lepidoptera: Lycaenidae):
Biology, management considerations, and data gaps. Proceedings of the 9th
Central Hardwood Forest Conference. USDA. -Forest Service, North Central
Forest Experiment Station. St. Paul, Minnesota. p. 83-100.

Harris, L.D., 1984. The Fragmented Forest. University of Chicago Press. Chicago,
Illinois.

Harrison, Susan, 1991. Local extinction in a metapopulation context: An empirical
evaluation. Biological Journal of the Linnean Society 42:73-88.

Hastings, A., and Woling, CL, 1989. Within-patch dynamics in a metapopulation.
Ecology 70:1261-1266.

Hultman, G. Eric, 1978. Trees, Shrubs, and Flowers of the Midwest. Great Lakes Living
Press, Chicago. p.77.

Janssens, P., Gullinck, H., 1988. Connectivity, proximity, and contiguity in the
landscape interpretation of remote sensing data. Connectivity in Landscape
Ecology. Proceedings of the 2nd International Seminar of the International
Association of Landscape Ecology p 43-47.

Janzen, DH, 1983. No park is an island: Increased interference from outside as park
size decreases. Oikos 41:402-410.

Johnson, W.C., Sharpe, D.M., DeAngelis, D.L., Fields, BE, and Olson, R. J., 1981.
Modeling seed dispersal and forest island dynamics. p. 215-239.

Kapos, V., 1989. Effects of isolation on the water status of forest patches in the Brazilian
Amazon. Journal of Tropical Ecology 52173-185.

Kellman, M., 1979. Soil enrichment by neotropical savanna trees. Ecology 67:565-577.

Krummel, J.R., Gardner, R.H., Sugihara, G., O'Neill, R.V., and Coleman, P.R., 1987.
Landscape patterns in a disturbed environment. Oikos 48:321-324.

57

LaGro Jr., James, 1991. Assessing patch shape in landscape mosaics. Photogrammetric
Engineering and Remote Sensing 57:285-293.

Loehle, Craig, and Wein, G., 1994. Landscape habitat diversity: A multiscale
information theory approach. Ecological Modeling 73:311-329.

Malcolm, Jay R., 1994. Edge effects in central Amazonian forest fragments. Ecology
75:2438-2445.

McGarigal, Kevin and Marks, B., 1995. FRAGSTATS: Spatial pattern analysis program
for quantifying landscape structure. General Technical Report to Pacific North
West -GTR-351. Portland, Oregon: USDA Forest Service, Pacific Northwest
Research Station. 122p.

Miles, Donald B., 1994. Introduction to the symposium: Contribution of long-term
ecological research to current issues in the conservation of biological diversity.
American Zoology 34:367-370.

Noss, R.F., 1983. A regional landscape approach to maintain diversity. BioScience
33:700-706.

Pacala, S.W., 1989. Plant population dynamic theory. In: Perspectives in Ecological
Theory. J. Roughgarden, R.M. May, and SA. Levin eds. Princeton University
Press. Princeton, New Jersey p 54-67.

Pearson, Scott M., 1993. The spatial extent and relative influence of landscape-level
factors on wintering bird populations. Landscape Ecology 8:3-18.

Pickering M.G., and Gebauer, SB, 1993. Status report: Assessment of associations
between lupine population characteristics and vegetation and soil characteristics -
1991-92. Eastern New York Chapter of the Nature Conservancy. (unpublished).

Ranney, J .W., Bruner, M.C., and Levenson, J .B., 1981. The importance of edge in the
structure and dynamics of forest islands. In: Forest Island Dynamics in Man-
dominated Landscapes. R.L. Burgess and D.M. Sharpe eds. Springer-Verlag.
New York, New York p. 67-95.

Shuey, J.A., Calhoun, J.V., and Iftner, DC, 1987. Butterflies that are endangered,
threatened, and of special concern in Ohio. Ohio Journal of Science 87:98-106.

Shugart, H.H.A., 1984. Theory of forest dynamics: The ecological implications of forest
succession models. Springer-Verlag. New York, NY.

Stewart, M.M. and Ricci, C., 1988. Dearth of the blues. Journal of Natural History
92:64-70.

Temple, S.A., and Cary, J .R., 1988. Modeling dynamics of habitat-interior bird
populations in fragmented landscapes. Conservation Biology 22340-347.

Tester, John R., 1989. Effects of fire frequency on oak savanna in east-central
Minnesota. Bulletin of the Torrey Botanical Club 116:134-144.

58

Thomas, J .W., 1979. Wildlife habitats in managed forests - the Blue Mountains of
Oregon and Washington. USDA. Forest Service Agricultural Handbook. 53 p.

Turner, M.G., Dale, V.H., and Gardner, RH, 1989. Predicting across scales: Theory
development and testing. Landscape Ecology 31245-252.

Vetaas, OR, 1992. Micro-site effects of trees and shrubs in dry savannas. Journal of
Vegetation Science 3:337-344.

Warner, RR, and Chesson, PL, 1985. Coexistence mediated by recruitment
fluctuations: A field to the storage effect. American Naturalist 125:769-787.

Welch, Robert J ., 1993. Dispersal and colonization behavior in the Kamer blue butterfly
(Lycaeides melissa samuelis) in central Wisconsin. Final Report to US. Fish and
Wildlife Service 50 p.

Wisconsin Department of Natural Resources Kamer Blue Technical Team, 1993. 3'd
draft: Wildlife Management Guidelines for Kamer blue. (unpublished).

Young, A., 1989. Agroforestry for soil conservation. International Council for Research
in Agroforestry. Nairobi, Kenya.

Overall Conclusions

This research provides information about KBB habitat that may prove useful for
managers attempting to protect this species. Specifically, light and soil pH appear to
affect lupine grth and density. Lupine and KBB distribution, at a landscape level, are
affected by forest characteristics and spatial parameters. This provides managers with
habitat information at both the individual patch and landscape scales.

The results of our first study indicate that a pH range of 4.2 to 5.6 and the highest
available light are associated with highest lupine densities. However, because of the
potential for weather extremes such as early summer drought and late spring frost, land
managers protecting the KBB should aim to provide lupine habitat that covers a range of
canopy, and thus light levels. Such habitat diversity is especially important for an
endangered species. Managers need to be cautious and not only provide the best habitat
for lupine, but be certain to provide protection against disaster. Managers also need to be
concerned with providing these two types of habitat at a spatial scale that biologically
benefits KBB. The results of our second study indicate that mean nearest neighbor needs
to be less than 70 meters. Thus managers should try to provide both high light and
protected areas within 70 meters of each patch of KBB.

To achieve this parameter, managers need not (perhaps should not) focus solely
on providing open areas, but instead manage for low to medium density forests. This

would provide a continuum of light, litter to maintain a low soil pH, protection from

59

60

extreme weather conditions, and a timber source. I recommend future research to (1)
better define KBB’s ecological requirements, and (2) better define a metapopulation
model of the KBB.

The broad range of forest density levels under which KBB has been found at
ASGA points out the need for a more general definition of oak savanna. Initially, I
interpreted oak savanna as openings with very sparsely distributed trees. However, I was
surprised to find KBB, (a species that is said to only live in oak savannas, pine barrens,
and sand dunes) in high density oak stands. It seemed to me that at ASGA the definition
of oak savanna would need to include all stand and density classes, thus broadening the
term, which I believe would benefit the KBB by providing protected habitat in many

weather conditions.

APPENDICES

APPENDIX A

Proportion of direct light measurement techniques

The camera was aligned with N on a compass by centering the middle of the word
Canon on the camera with the N arrow on a compass and the camera was then leveled
with the visual aid of a bubble. I laid on the ground below the view of the camera and
took the two pictures of the sky. The film was then developed into negatives that I cut
and placed into gesepii finger pressure slide cases. I discarded the negative of poorer
quality from each plot. In December I traveled to Corvallis Oregon where I used Oregon
State University Department of Forestry’s equipment and Canopy software of to analyze
the slides. Percent sky was determined by taking the ratio of the readings from the actual

fisheye image to the clear negative (Chan, et. al, 1986).

61

 

 

 

as: 8.2 ..a a ..a n \ ..a o _8_-Nmm-:§. .. 352:5 :2”: $86 2... 8... 3.. NE.
as: 8.2 ..a n. ..a n . ..a _ 5..-ch m... asses... 2.0.2 2... 8... 3: NE.
320 sas: No 2 ..a \ No . N32 ..soasss .m .225 2.. n . ._ 3.2.2.
as: 8.2 ..a N. ..a m . ..a e .8.-NR...§. .. Essa .722 $86 2.. 2.. 3.2:...
as: 3.2 ..a N .a m \ ..a . $3.3 m... eases... 2.0.2 2.. 2.. 8.2:...
as? .25 was: 8. n c .a 8. .8... o. 32.... msass... 820x 2.. 2... 8.8.2.
823 3:0 mam—5... no 2 Em \ no _ Nmésm 353555 .m 3.225 2.. 2... 23.2.5
ago sac: No o. a» \ No _ N3: sagasss .m .8225 2.. 9.: 3.. Q8
as: 3.2 ..a N ..a m . ..a _ 3.72:: 526 .312 536 2.. on... 3.38
as: 8.2 ..a N ..a m . ..a . $3.8 m... ss..s=.< 23.2 2.. on... 8.38
s25... ..a a. ..a m \ ..a a. 29.5. 33.502 2.35 2.. 8... 3... :No
as: 8.2 No _ ..a n he . .8.-N2-_.§. .. Essa :62... p.986 2.. 8... 8.....8

£8.— .oueah tom: ..E< .35. 8a: a we: <3— .5529: 953. 25.2 .269..— oEE. 8.5

 

 

Esau ooze—Emma .2 been... 533.3% 30.530 .0 63m...

m X_Dzmn_n_<

62

BIBLIOGRAPHY

BIBLIOGRAPHY

Alverson, W.S., D.M. Waller, and SS. Solheim, 1988. Forest too deer: Edge effects in
northern Wisconsin. Conservation Biology 2:348-358.

Andersson, T., 1991. Influence of stemflow and throughfall from common oak (Quercus
robur) on soil chemistry and vegetation patterns. Canadian Journal of Forest
Research 21:917-924.

Bach, CE, 1988. Effects of host plant patch size on herbivore density: Underlying
mechanisms. Ecology 69:1103-1117.

Backs, SE, and CH. Eisfelder, 1990. Criteria and guidelines for wild turkey release
priorities in Indiana. Proceedings of the National Wild Turkey Symposium.

6:134-143.

Baudry, J ., and HG. Merriam, 1988. Connectivity and connectedness: Functional
versus structural patterns in landscapes. Connectivity in Landscape Ecology.
Proceedings of the 2"GI International Seminar of the International Association of
Landscape Ecology p. 23-28.

Belsky, A.J. R.G. Amundson, J. M. Duxbury, S.J. Riha, A.R. Ali, and S.M., Mwonga,
1989. The effects of trees on their physical, chemical, and biological

environments in a semi-arid savanna in Kenya. Journal of Applied Ecology
26:1005-1021.

Belsky, A.J., and CD. Canham, 1994. Forest gaps and isolated savanna trees: An
application of patch dynamics in two ecosystems. BioScience 44:77-84.

Bess, J .A., 1989. Status of the Kamer blue butterfly, Lycaeides melissa samuelis
Nabokov, in the Manistee National Forest. Report to Michigan Natural Features
Inventory, Lansing, Michigan 20 p. (unpublished).

Bray, J .R., 1955. The savanna vegetation of Wisconsin and an application of the
concepts order and complexity to the field of ecology. Unpublished Ph.D.
Dissertation. The University of Wisconsin, Madison. 251 p.

Boyonoski, A.M., 1992. Factors affecting the establishment and maintenance of Lupinus
perennis, (wild lupine). MS. Thesis, University of Guelph, Ontario.

Burgess, Robert L., D. M. Sharpe, and MC. Brunet, 1981. Forest island dynamics in
man-dominated landscapes. Eds. Burgess, R. L. and Sharpe, D. M. New York,
Springer-Verlag. 310 p.

Chan, S.S., R.V. McCreight, J .D. Walstad, and TA. Spies, 1986. Evaluating forest
vegetative cover with computerized analysis of fisheye photographs. Forest
Science 32:1085-1091.

63

64

Clark, J .S., 1991. Disturbance and population structure on the shifiing mosaic landscape.
Ecology 72:1119-1137.

Clough, M.W., 1992. Endangered and threatened wildlife and plants: Proposed
endangered species status for the Kamer blue butterfly. Federal Register
(January 21, 1991). 57: 2241-2246.

Comins, H.N., and LR. Noble, 1985. Dispersal, variability, and transient niches: Species

coexistence in a unifome variable environment. American Naturalist 126:706-
723.

Cowling, W.A., and J .C. Clements, 1993. Association between collection site soil pH
and chlorosis in Lupinus angustifolius industed by a fine-textured, alkaline soil.
Australian Journal of Agricultural Research 44: 1 821-1836.

Crozier, Carl R., and R. E. J. Boemer, 1984. Correlations of understory herb distribution
patterns with microhabitats under different tree species in a mixed mesophytic

forest. Oecologia 62:337-343.

Curtis, B., 1959. The vegetation of Wisconsin. The university of Wisconsin press.
Madison, Wisconsin. 657 pp.

Cryan, J .F., and R. Dirig, 1978. A report on the status of the Kamer blue butterfly
(Lyceides melissa samuelis Nabokov) in New York State. Report to the US. Fish
and Wildlife Service.

Dirig, Robert, 1988. Nabokov's blue snowflakes. Natural History 97:68-69.

Edsall, Marian S., 1985. Roadside plants and flowers: A traveler’s guide to the Midwest
and Great Lakes area: With a few familiar off-road wildflowers. University of
Wisconsin Press, Madison Wisconsin.

Ellis, J .E., and JG. Lewis, 1967. Mobility and annual range of wild turkeys in Missouri.
Journal of Wildlife Management 31:568-581.

F onnan, R.T.T., and M. Godron, 1986. Landscape Ecology. John Wiley and Sons. New
York, New York. 619 p.

Franklin, J .F., and R.T.T. Forman, 1987. Creating landxcape patterns by forest cutting:
Ecological consequences and principles. Landscape Ecology 1:5-18

Gilpin, Michael E., 1987. Metapopulation structure and population vulnerability. In:
Viable populations for conservation. Cambridge University Press. New York,
New York p. 125-139.

Givnish, T.J., E.S. Mengis, and DP. Schweitzer, 1988. Minimum area requirements for
long-terrn conservation of the Albany Pine Bush and Kamer blue butterfly: An
assessment. (unpublished).

Gustafson, Eric J ., G.R. Parker, and SE. Backs, 1994. Evaluating spatial pattern of
wildlife habitat: A case study of the wild turkey (Meleagris gallopavo). Midland
Naturalist 131:24-33.

65

Haack, R.A., 1993. The endangered Kamer blue butterfly (Lepidoptera: Lycaenidae):
Biology, management considerations, and data gaps. Proceedings of the 9th
Central Hardwood Forest Conference. USDA. -Forest Service, North Central
Forest Experiment Station. St. Paul, Minnesota. p. 83-100.

Harris, L.D., 1984. The Fragmented Forest. University of Chicago Press. Chicago,
Illinois.

Harrison, Susan, 1991. Local extinction in a metapopulation context: An empirical
evaluation. Journal of the Linnean Society 42:73-88.

Hastings, A., and CL. Woling, 1989. Within-patch dynamics in a metapopulation.
Ecology 70:1261-1266.

Herms, Catherine R, 1996. The endangered Kamer blue butterfly (Lepidoptera:
Lycaenidae) in Michigan: Habitat suitability, potential impacts of gyspy moth
(Lepidoptera: Lymantriidae) suppression and labortory rearing. Unpublished
Masters Thesis, Department of Entomology, Michigan State University.

Holland,V.L., 1973. A study of soil and vegetation under Quercus douglasii compared to
open grassland. Unpulbished PhD. Dissertation, Department of Botany,
University of California, Berkeley. 359 p.

Hultman, G. Eric, Trees, Shrubs, and Flowers of the Midwest. 1978. Great Lakes Living
Press, Chicago. p.77.

Janssens, P., and H. Gullinck, 1988. Connectivity, proximity, and contiguity in the
landscape interpretation of remote sensing data. Connectivity in Landscape
Ecology. Proceedings of the 2nd International Seminar of the International
Association of Landscape Ecology. p. 43-47.

Janzen, DH, 1983. No park is an island: increased interference from outside as park size
decreases. Oikos. 41 1402-410.

Johnson, W.C., D.M. Sharpe, D.L. DeAngelis, E.E. Fields, and R. J. Olson. 1981.
Modeling seed dispersal and forest island dynamics. p. 215-239.

Kapos, V., 1989. Effects of isolation on the water status of forest patches in the Brazilian
Amazon. Journal of Tropical Ecology 5:173-185.

Kellman, M., 1979. Soil enrichment by neotropical savanna trees. Ecology 67:565-577.

Krummel, J .R., R.H. Gardner, G. Sugihara, R.V. O'Neill, and PR. Coleman, 1987.
Landscape patterns in a disturbed environment. Oikos 48:321-324.

LaGro Jr., James, 1991. Assessing patch shape in landscape mosaics. Photogrammetric
Engineering and Remote Sensing 57:285-293.

Loehle, Craig, and G. Wein, 1994. Landscape habitat diversity: A multiscale
information theory approach. Ecological Modeling 73:311-329.

Malcolm, Jay R., 1994. Edge effects in central Amazonian forest fragments. Ecology
75:2438-2445.

66
Maxwell, J ., and T. Givnish, 1993. Research on the Kamer blue butterfly at Fort McCoy,
Wisconsin: Progress report for the 1993 field station. Report submitted to the US
Fish and Wildlife Service, and the Department of the Army. (unpublished).

McGarigal, Kevin, and B. Marks, 1995. FRAGSTATS: Spatial pattern analysis
program for quantifying landscape structure. General Technical Report Pacific
North West-GTR-3S 1. Potland, Oregon: U.S.D.A. - Forest Service, Pacific
Northwest Research Station. 122 p.

Miles, Donald B., 1994. Introduction to the symposium: Contribution of long-term
ecological research to current issues in the conservation of biological diversity.
American Zoology 34:367-370.

Miller, Ellen, and M. C. Whiting, 1895. Wildflowers of the North-Eastem states. G.P.
Putnam’s Sons. The Knickerbocker Press, New York.

Noss, R.F., 1983. A regional landscape approach to maintain diversity. BioScience.
33:700-706.

Opler, P.A., and GO. Krizek, 1984. Butterflies East of the Great Plains. John Hopkins
University Press, Baltimore, Maryland. p. 120-121.

Pacala, S.W., 1989. Plant population dynamic theory. In: Perspectives in ecological
theory. J. Roughgarden, R.M. May, and SA. Levin eds. Princeton University
Press. Princeton, New Jersey. p. 54-67.

Parker, V.T., and CH. Muller, 1982. Vegetation and environmental changes beneath
isolated live oak trees (Quercus Agg'folia) in a California Annual Grassland.
American Midland Naturalist 107269-81.

Pearson, Scott M., 1993. The spatial extent and relative influence of landscape-level
factors on wintering bird poulations. Landscape Ecology 8:3-18.

Pickering M.G., and SB. Gebauer, 1993. Status report: Assessment of associations
between lupine population characteristics and vegetation and soil characteristics -
1991-92. Eastern New York Chapter of the Nature Conservancy. (unpublished).

Ranney, J .W., M.C. Bruner, and J .B. Levenson, 1981. The importance of edge in the
structure and dynamics of forest islands. p. 67-95.

Rich, P.M., 1989. A manual for analysis of hemispherical canopy photography. Los
Alamos National Laboratory. Los Alamos, New Mexico. 80 p.

Scudder, SH, 1889. The Butterflies of Eastern United States and Canada with special
reference to new England. Cambridge, Massachucetes. p. 964-969.

Shuey, J .A., J .V. Calhoun, and DC. Iftner, 1987. Butterflies that are endangered,
threatened, and of special concern in Ohio. Ohio Journal of Science 87:98-106.

Shugart, H.H.A., 1984. Theory of forest dynamics: The ecological implications of forest
succession models. Springer-Verlag. New York, New York.

67

Smallidge, P.J., and DJ. Leopold, 1992. Environmental factors and critical habitat for

Lycaeides melissa samuelis along transmission line rights-of-way in east-central
New York, USA. (Draft).

Smith, Helen Vandervort, 1966. Michigan Wildflowers. Cranbrook Institute of Science,
Bloomfield Hills, Michigan.

Spies, T.A., W.S. Ripple, and GA. Bradshaw, 1994. Dynamics and pattern of a managed
coniferous forest landscape in Oregon. Ecological Applications 4:555-568.

Stewart, M.M., and C. Ricci, 1988. Dearth of the blues. Journal of Natural History 92:
64-70.

Tang, Caixian, J. Kuo, N.E. Longnecker, C.J. Thompson, and AD. Robson, 1992. High
pH causes disintegration of the root surface in Lupinus angustifolius L. Annals of
Botany 71:201-207.

Tang, Caixian, and AD. Robson, 1993. pH above 6.0 reduces nodulation in Lupinus
species. Plant and Soil. 1522269-276.

Temple, S.A., and J .R. Cary, 1988. Modeling dynamics of habitat-interior bird
populations in fragmented landscapes. Conservation Biology 2:340-347.

Tester, John R., 1989. Effects of fire frequency on oak savanna in east-central
Minnesota. Bulletin of the Torrey Botanical Club 116:134-144.

Thomas, J .W., 1979. Wildlife habitats in managed forests - the Blue Mountains of
Oregon and Washington. U.S.D.A. - Forest Service Agricultural Handbook. 53 p.

Thomson, James, D. G. Weiblen, B.A. Thomas, S. Alfaro, and P. Legendre, 1996.
Untangling multiple factors in spatial distributions: Lilies, gophers, and rocks.
Ecology 77:1698-1715.

Turner, M.G., V.H. Dale, and RH. Gardner, 1989. Predicting across scales: Theory
development and testing. Landscape Ecology 31245-252.

Vetaas, OR, 1992. Micro-site effects of trees and shrubs in dry savannas. Journal of
Vegetation Science 3:337-344.

Warner, R.R.,and P.L. Chesson, 1985. Coexistence mediated by recruitment fluctuations:
A field to the storage effect. American Naturalist 125:769-787.

Welch, Robert J ., 1993. Dispersal and colonization behavior in the Kamer blue butterfly
(Lycaeides melissa samuelis) in central Wisconsin. Final Report to US. Fish and
Wildlife Service 50 p.

Wisconsin Department of Natural Resources Karner Blue Technical Team., 1993. 3rd
draft: Wildlife Management Guidelines for Kamer blue. (unpublished).

Young, A., 1989. Agroforestry for soil conservation. International Council for Research
in Agroforestry. Nairobi, Kenya.

Zinke, P.J., 1962. The pattern of influence of individual forest trees on soil properties.
Ecology 43:130-133.