I.
:3
4 .K-«Io-a‘v
hr z.

.v
x..l.:
d3 .
‘ 2 .. .
z. .
x
:‘l V... .1

1...: 5.7.
..

z. .

.791... .35.... I:

4.. 5...: .v.......!. 2..”-
: .5-3125) ;

V 4.-

f :e ,

3:2... .98 1.

$5.19.“. 3.1 .3. .2.
1:... .

3.511..
11.3.}

 

 

 

 

 

 

LIBRARY
Michigan State
University

 

 

 

This is to certify that the
dissertation entitled

Modeling the Effects of Management Approaches on
Forest and Wildlife Resources in Northern
Hardwood Forests

presented by

Christine Hanaburgh

has been accepted towards fulfillment
of the requirements for

_ Ph.D.4 Fish. &-w11d1.

degree in

 

 

lbw M WT

M a j r professo

Date August 23, 2001

MSU is an Affirmative Action/Equal Opportunity Institution 0- 12771

 

PLACE IN RETURN Box to remove this checkout from your record.
To AVOID FINB return on or before date due.
MAY BE RECALLED with earlier due date if requested.
DATE DUE DATE DUE DATE DUE
MAY 2 6 2007 DEC 0 2 £83
'9 .6 1 .4 .0 .7

JAN

JJUN 1 7

 

 

 

   
  
    
  
  
 

 

I VSQF‘MANAGEMENT APPROACHES 0N FOREST

’5 .. 7 traces iN' NORTHERN HARDWOOD FORESTS

By

Christine Hanaburgh

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
‘ DOCTOR OF PHILOSOPHY
zpepartrnent of Fisheries and Wildlife

.2001

 

  

  
  
  
  
  
 
 
 
  
  
  
 

,q .. 7" EFFECTS OF MANAGEMENT APPROACHES ON FOREST
V LIFE RESOURCES IN NORTHERN HARDWOOD FORESTS

By

Christine Hanaburgh

AN ABSTRACT OF A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
_ for the degree of
’ - ‘_ "H" " DOCTOR OF PHILOSOPHY

Department of Fisheries and Wildlife

 

   

2001 '

     

‘52..

b3“ ' Professor Henry Campa, III

  

MDT

 

 

' -.
i
...,..
5
3‘":
.a
.

 

 

 

 

\5
1"
u
‘.
.

 

 

 

     
  

finch: 1'. 11.. i. '
By
tilts 2:21.}. . '
Christine Hanaburgh
tooth" .

'-‘«~t.ii‘ ,
1 EM the effects of forest management regimes associated with 4 land
- Mgniiu". ~
. in Michigan’s Upper Peninsula on forest and wildlife resources in northern
._ not (1111:: 1,7
' ‘ . .-. forests. Management ownerships investigated were the State of
“autumn ' ‘

Man/Michigan Department of Natural Resources (MDNR) (management primarily
(p \(i; i

it! game species and timber products), Federal/USDA Forest Service (management for
multiple use), private/industrial (management primarily for timber products), and
iirivateann-indusuial (minimal manipulations with the goal of preservation).

I Twelve sites, 7-23 km2 each, were evaluated in the Lake Superior State Forest, the
Hiawatha National Forest, the Mead Company and Shelter Bay Forests property, and at
the privately owned Huron Mountain Club. From 1996-1998, data were collected on 35

forest attributes, and on the relative abundance of red- backed salamanders, forest

mgbirds, barred owls, and fishers. Landscape characteristics were quantified from 1991
\\ ti ill "
clusifid thematic mapper satellite imagery. Habitat suitability index models were
mgc G l. i.
Winged for the red-backed salamander, yellow-rumped warbler, and northern flying

 

 

 

 

 

-

fiif I]

 

 

 

 

 

 

    
  
  
  
  
  
  

. 'hl other ownerships. Timber industry stands had a larger
. 1“! My (0.5-5 m) canopy cover (1150.10). and Forest '

. ,' tube more similar to each other in structure and

. Mournerships.

- m were positively correlated with the density of trees >10 cm
, lily correlated with canopy cover 0.5-5 m in height. Relative abundance
imitpOJO) among ownerships, but tended to be greatest at the Huron

maul; Relative abundance of American redstarts and veeries was greater

 

MN) on timber industry sites, and yellow-rumped warblers and pileated woodpeckers
mm abundant on Huron Mountain Club sites. Habitat suitability index values for
meted-backed salamander, yellow-rumped warbler, and pileated woodpecker were
outdated (psOJO) with species population indices, indicating that these models
performd well. The fisher and veery models did not predict the distribution and
W of these species.
landscape characteristics at the scales and resolution evaluated were primarily

ilflnencedby historical and physiographic factors, rather than management approaches
“in mnerships. Results indicated that the 4 land ownerships investigated represent a
“Midlife habitat conditions found in northern hardwood forests, reflecting the
WW goals across the eastern Upper Peninsula, as well as
: among ownerships.

'. Mil-{5.1mm a ;

  
   

 

 

 

 

  

. .tpomble-thmugh funding from Michigan State University.

   
  
  
 

‘ Wildlife Fund, the Huron Mountain Foundation, and the Federal
‘ 1 Act under Project W-127-R administered by the Michigan
‘m Resources, Wildlife Division. I would like to thank Terry Minzey

’ ,- the=Cusino Wildlife Research Area for being so accommodating to the
“Mg our stay at Cusino. I am grateful to David Gosling, Arthur Turner, and
Maid-members at the Huron Mountain Club for providing project support,

‘ W'wwmmodations, and the very rewarding experience of working on such an
m: md majestic piece of property. I would also like to acknowledge the Mead
Company and Shelter Bay Forests for permission to conduct research on their land, and
brgroviding information on my study sites.

The expertise of my graduate committee has been invaluable in this undertaking.
lam grateful to Dr. Doug Lantagne for his continued interest in this project, all the way
from. Vermont, and to Dr. Richard Groop for his time and helpful comments during the
0mm of the project. Dr. Dean Beyer was instrumental in developing this project and
presided a valuable management perspective. Special thanks are extended to Dr. Kelly

M for generously filling a last minute committee opening. I have especially

W working with Dr. Scott Winterstein, and Ihave learned he can always be counted

 

 

“,1 u
‘ .

"I-alr it

v”! 3‘

 

it",

111-4.:

 

 

 

   

   
  
   
  

 

 

    

flu. field assistants who worked with me and provided the

' . ., observations, and hard work that made it possible for me to

, for this project. Most sincere thanks to Sarah Converse, Sarah

1 , April Woodward, and especially Tammy Giroux for her extra

., '4 with this project.

mothers I would like to recognize for their support and contributions

r Mhmey through graduate school. Delia Raymer and Meg Clark have been my
mum teaching assistant colleagues, and graduate school comrades from the very
beguiling. and [thank them for sharing their experiences and friendship with me. I am
also grateful to Dr. Bill Taylor for taking an interest in my graduate experience and for
giving me advice and opportunities that have helped shape my professional and personal
goals.

Cutainly without my family this whole endeavor would not have the personal
minglthat it does. I was truly fortunate to have the continual support, praise, and
encouragement of my parents David and Johanna Hanaburgh, my grandparents, and my
pleats inv'l‘awt Finally, no words can express my gratitude to my wonderful husband

M. for never questioning my goals and for making this dissertation part 'of his life as

‘ ‘ ,’Hl‘r,' ,..-
- “Form: )P .~

 

TABLE OF CONTENTS
LIST OF TABLES ...................................................... ix
LIST OF FIGURES ................................................... xvii
INTRODUCTION ....................................................... l
OBJECTIVES .......................................................... 8
STUDY AREA ......................................................... 9

CHAPTER 1 - Assessment and Modeling of Wildlife Habitat

INTRODUCTION ................................................... 14
METHODS ........... . ............................................ 17
Experimental design ............................................. l7
Vegetation sampling ............................................. 20
Wildlife species surveys .......................................... 26
Red-backed salamanders ..................................... 26
Forest bird species .......................................... 30
Northern flying squirrels ..................................... 31
Barred owls ............................................... 31
Fisher .................................................... 32
Habitat modeling ................................................ 33
Data analysis ................................................... 35
RESULTS ......................................................... 38
Vegetation and structural attributes ................................. 38
Principal components analysis ................................ 41
Overstory tree species composition ............................. 45
Wildlife population survey results .................................. 49
Red-backed salamanders ..................................... 49
Comparison of ground transect searches and cover boards for surveying
salamanders ......................................... 58
Forest bird species .......................................... 62
Principal components analysis of forest birds ..................... 78
Forest bird communities ..................................... 81
Barred owls ............................................... 83
Fishers ................................................... 83
vi

.JL.‘

ii

Habitat Suitability Analysis ....................................... 85
Red-backed salamander ...................................... 85
Ovenbird ................................................. 90
American redstart .......................................... 92
Veery .................................................... 92
Yellow-rumped warbler ...................................... 95
Pileated woodpecker ....................................... 100
Northern flying squirrel ..................................... 102
Barred owl ............................................... 104
Fisher ................................................... 106
Relationships between population indices and HSI model output ..... 106

DISCUSSION 110

Vegetation and structural attributes ................................ 110
Wildlife habitat relationships ..................................... l 12
Red-backed salamanders .................................... 112

Comparison of ground transect searches and cover boards for surveying
salamanders ........................................ 1 14
Forest birds .............................................. l 16
Barred owls .............................................. 1 18
Fisher ................................................... 119
HSI model performance ......................................... 1 19

CHAPTER 2 - Landscape Scale Wildlife Habitat Characteristics and Relationships

INTRODUCTION .................................................. 126
METHODS ....................................................... 129
Landscape analyses ............................................. 129
Explanation of landscape metrics ............................. 135
Data Analysis ................................................. 136
RESULTS ........................................................ 139
Vegetation cover type distributions ................................ 139
Landscape characteristics of ownerships ............................ 143
Road density .................................................. 149
Landscape characteristics of northern hardwood forest patches ........... 149
Relationships between wildlife species relative abundance and landscape
characteristics ............................................. l 53
Red-backed salamander ..................................... 153
American redstart ......................................... 156
Ovenbird ................................................ 156
Veery ................................................... 160
vii

,,
\
i.

\ :‘PIJ
.tltal

‘..“\

 

 

 

WES ....................................................... 185

Mi am analysis .................................................. 185
11 x - a .
MULTS ........................................................ 186
Forest stand characteristics ....................................... 186
‘ , Landscape characteristics ........................................ 191
Wildlife species ................................................ 191
DECUSSION ..................................................... 200
CONCLUSIONS ...................................................... 205
MANAGEMENT AND RESEARCH IMPLICATIONS ....................... 209
APPENDICES ........................................................ 213
:uAPPENDIX A. Vegetation and wildlife sampling point coordinates. .......... 214
APPENDIX B. A habitat suitability index model for the red- backed salamander
(Piethodon cinereus) in northern Michigan. .......................... 218

APPENDIX C. A model of habitat suitability for the yellow-rumped warbler

n- '(Dendroica coronata) in the Upper Great Lakes region of the United States. 232
APPENDIX D. A habitat suitability index model for the northern flying squirrel
As1,u(Glaucomys sabrinus) in the Upper Great Lakes region of the United States. 242
APPENDIX E. Scientific names of songbirds surveyed. .................... 256

Memorial) ................................................. 258

 

   
   

' - Mculaxmu ‘- LISTOFTABLES

' = w : of locations for Michigan Department of Natural
: -. ‘ " ~ U. S. Forest Service (USPS), timber industry (TI), and Huron
' . (lib (INC) study sites in Michigan’s Upper Peninsula .............. ll

, mmmffu

home range estimates for species surveyed in Michigan’s Upper
Mania. 1996—1998 .................................................. 18
TI“ in ."- 1: ' ‘
Twat MW HSI model variables and values associated with relatively good
new quality. Dbh=diameter at breast height .............................. 22

Table 4. Forest stand variables and methods used to measure them in the Upper
Peninsula of Michigan, 1996-1998. ...................................... 25

Table 5. Understory vegetation variables (means and standard errors) measured on
Michigan Department of Natural Resources (MDNR), U. S. Forest Service
(USPS), timber industry (TI), and Huron Mountain Club (HMC) land in
Michigan’s Upper Peninsula, July and August, 1996, 1997, and 1998. .......... 39

Table 6. Overstory vegetation variables (means and standard errors) measured on
Michigan Department of Natural Resources (MDNR), U. S. Forest Service
(USPS), timber industry (TI), and Huron Mountain Club (HMC) land in
Michigan’s Upper Peninsula, July and August, 1996, 1997, and 1998. .......... 42

Table 7. Overstory tree species composition (stems/ha) on Michigan Department of
Natural Resources (MDNR), U. S. Forest Service (USFS), timber industry (TI),
and Huron Mountain Club (HMC) land in Michigan’s Upper Peninsula, July and
August, 1996, 1997, and 1998. ......................................... 46

‘ Table 8. Number of salamanders surveyed per hectare (means and standard errors)
with ground transect searches, time spent ground searching, number of
salamanders found per minute of ground searching, and number of salamanders
found under cover boards in forest stands on Michigan Department of Natural
Resources (MDNR), U. S. Forest Service (USFS), timber industry (TI), and Huron
Mountain Club (HMC) land in Michigan’s Upper Peninsula, June, July and
August, 1997. ....................................................... 50

 

v"

 

 

m 91‘. Nmbetof salamanders surveyed per hectare (means and standard errors)
Mmd transect searches, time spent ground searching, number of
islanders found per minute of ground searching, and number of salamanders
Minder cover boards in forest stands on Michigan Department of Natural
Resources (MDNR), U. S. Forest Service (USFS), timber industry (TI), and Huron
Mountain Club (HMC) land in Michigan’s Upper Peninsula, June, July and
August, 1998. ....................................................... 51

Table 10. Number of salamanders surveyed per hectare (means and standard errors)
with ground transect searches, time spent ground searching, number of
salamanders found per minute of ground searching, and number of salamanders
found under cover boards in forest stands on Michigan Department of Natural
Resources (MDNR), U. S. Forest Service (USFS), and timber industry (TI) land in
Michigan’s Upper Peninsula, September and October. 1997. No significant
differences (p>0.10) were detected ....................................... 53

Table 11. Number of salamanders found per hectare (means and standard errors)
under cover boards in forest stands on Michigan Department of Natural Resources
(MDNR), U. S. Forest Service (USFS), timber industry (TI), and Huron Mountain
Club (HMC) land in Michigan’s Upper Peninsula, late August, September, and
October, 1998. ...................................................... 54

Table 12. Mean values for soil and vegetation variables measured during red—backed
salamander surveys in forest stands on Michigan Department of Natural
Resources (MDNR), U. S. Forest Service (USFS), timber company (TI), and
Huron Mountain Club (HMC) land in Michigan’s Upper Peninsula, June, July and
August, 1997 and 1998 ................................................ 55

Table 13. Spearman rank correlations (rs) between the mean number of salamanders
found (ground transect searches and cover board surveys) and 35 forest stand
variables in Michigan’s Upper Peninsula, 1997 and 1998. .................... 59

Table 14. Spearman rank correlations (rs) and probability levels (p) for the number of
salamanders found per stand between artificial cover boards and natural cover
objects grouped by size class. .......................................... 62

Table 15. Mean absolute frequencies (percent of points at which species occurred)
and standard errors (S.E.) on the means over 3 years of data collection of bird
species surveyed on Michigan Department of Natural Resources (MDNR), U. S.
Forest Service (USFS), timber industry (TI), and Huron Mountain Club (HMC)
land in Michigan’s Upper Peninsula, May-July, 1996, 1997, and 1998. Indicator
species are in bold .................................................... 63

 

7' '
lnli"
, 1&1

 

 

M 33. Mean absolute frequencies (proportion of points at which species occurred)
NW errors (8.13.) of bird species surveyed on study sites on U. S. Forest
W=WSFS), Michigan Department of Natural Resources (MDNR), and Huron
Mn Club (HMC) land in Michigan’s Upper Peninsula, May-July, 1996.
Mentor species are in bold ............................................. 69

Table 17. Mean absolute frequencies (percent of points at which species occurred)
and standard errors (SE) of bird species surveyed on U. S. Forest Service (USPS)
and Michigan Department of Natural Resources (MDNR) sites, 1 Mead Co.
(MEAD) site and Huron Mountain Club (HMC) land in Michigan’s Upper
Peninsula, May-July, 1997. Indicator species are in bold ...................... 72

Table 18. Mean absolute frequencies (percent of points at which species occurred)
and standard errors (SE) of bird species surveyed on Michigan Department of
Natural Resources (MDNR), U. S. Forest Service (USFS), timber industry (TI),
and Huron Mountain Club (HMC) land in Michigan’s Upper Peninsula, May-July,
1998. Indicator species are in bold ....................................... 75

Table 19. Frequency of barred owl responses (% of points sampled) and standard
errors among years (S.E.) on study sites on Michigan Department of Natural
Resources (MDNR), U. S. Forest Service (USFS), timber industry (TI). and Huron
Mountain Club (HMC) land in Michigan’s Upper Peninsula, 1996, 1997, and
1998 ............................................................... 84

Table 20. Ranges and mean values for habitat variables in stands where s2
salamanders were found and in stands where >2 salamanders were recorded, and
probability of statistical difference in the means of the 2 groups based on
independent t-tests .................................................... 86

Table 21. Mean suitability index values for each HSI model variable, and means and
standard errors of final HSI values for the red-backed salamander on Michigan
Department of Natural Resources (MDN R), U. S. Forest Service (USFS), timber
industry (TI), and Huron Mountain Club (HMC) land in Michigan’s Upper
Peninsula. .......................................................... 89

Table 22. Mean suitability index values for each HSI model variable, and means and
standard errors of final HSI values for the ovenbird on Michigan Department of
Natural Resources (MDNR), U. S. Forest Service (USFS), timber industry (TI),
and Huron Mountain Club (HMC) land in Michigan’s Upper Peninsula .......... 91

xi

 

 

P» a.“ rm 4 ~ . . a a at. .u z,
mliw p1 Ht. .\.. \. 1U unfit. H a 1! lie. .
L |

 

mu

 

7“”. Mean suitability index values for each HSI mode] variable, and means and
and!!! wars of final HSI values for the American redstart on Michigan
Woof Natural Resources (MDNR), U. S. Forest Service (USFS), timber
infirm (TI), and Huron Mountain Club (HMC) land in Michigan’s Upper
Peninsula. .......................................................... 93

Table 24. Mean suitability index values for each HSI model variable, and means and
standard errors of final HSI values for the veery on Michigan Department of
Natural Resources (MDNR), U. S. Forest Service (USFS), timber industry (TD,
and Huron Mountain Club (HMC) land in Michigan’s Upper Peninsula .......... 94

Table 25. Ranges and mean values for habitat variables in stands where no yellow-
rumped warblers were recorded and in stands where at least 1 yellow-rumped
warbler was observed in 1996, 1997, or 1998, and probability of statistical
difference in the means of the 2 groups based on independent t-tests ............. 97

Table 26. Mean suitability index values for each HSI model variable, and means and
standard errors of final HSI values for the yellow—rumped warbler on Michigan
Department of Natural Resources (MDNR), U. S. Forest Service (USFS), timber
industry (TI), and Huron Mountain Club (HMC) land in Michigan’s Upper
Peninsula. .......................................................... 99

Table 27. Mean suitability index values for each HSI model variable, and means and
standard errors of final HSI values for the pileated woodpecker on Michigan
Department of Natural Resources (MDNR), U. S. Forest Service (USFS), timber
industry (TI), and Huron Mountain Club (HMC) land in Michigan’s Upper
Peninsula. ......................................................... 1 01

Table 28. Mean suitability index values for each HSI model variable, and means and
standard errors of final HSI values for the northern flying squirrel on Michigan
Department of Natural Resources (MDNR), U. S. Forest Service (USFS), timber
industry ('1‘ D, and Huron Mountain Club (HMC) land in Michigan’s Upper
Peninsula .......................................................... 103

Table 29. Mean suitability index values for each HSI model variable, and means and
standard errors of final HSI values for the barred owl on Michigan Department of
Natural Resources (MDNR), U. S. Forest Service (USFS), timber industry (TI),
and Huron Mountain Club (HMC) land in Michigan’s Upper Peninsula ......... 105

 

  
 
 
 
 
 
 
 
   

 

 

,— ,. , index values foreach HSI mode] variable, and means and
.' w ; m HSI values for the fisher on Michigan Department of
‘ — 'WNR), U. S. Forest Service (USFS), timber industry (TI),
1 ' ‘ '1 w: ' Club (HMC) land in Michigan’s Upper Peninsula.
" . levels for tests of differences among ownerships were calculated with
‘ lis one-way analysis of variance (Siegel and Castellan 1988).. 107

_ . ' ,. . . - rank correlations between mean relative abundances per study

, .' m HSI values for species surveyed on Michigan Department of

.j' , ‘lmes (MDNR), U. S. Forest Service (USFS), timber industry (TI),
mm Mountain Club (HMC) land ................................... 108

M31Cmegories used in the classification of 1991 Upper Peninsula Landsat
Thematic Mapper satellite imagery (MacLean Consultants, Ltd. no date) ........ 130

Table 33. Percent of total land area (means and standard errors) in each cover type
-identified by satellite imagery for study sites defined with 80 m buffers on
Wchigan Department of Natural Resources (MDNR), U. S. Forest Service
(USFS), timber industry (TI), and Huron Mountain Club (HMC) land in
Michigan’s Upper Peninsula, calculated from 1991 satellite imagery. No
significant differences (p>0. 10) were detected ............................. 139

Table 34. Percent of total land area (means and standard errors) in each cover type
identified by satellite imagery for study sites defined with 800 m buffers on
Michigan Department of Natural Resources (MDNR), U. S. Forest Service
(USFS), timber industry (TI), and Huron Mountain Club (HMC) land in
Michigan’s Upper Peninsula, calculated from 1991 satellite imagery ............ 141

Table 35. Values for landscape metrics (means and standard errors) for study sites on
Michigan Department of Natural Resources (MDNR), U. S. Forest Service
(USFS), timber industry (TD, and Huron Mountain Club (HMC) land in
Michigan’s Upper Peninsula, calculated from 1991 satellite imagery. Values are
for study sites created with an 80 m buffer ................................ 144

Table 36. Values for landscape metrics (means and standard errors) for study sites on
Michigan Department of Natural Resources (MDNR), U. S. Forest Service
(USFS), timber industry (TI), and Huron Mountain Club (HMC) land in
T'bMiehigan’s Upper Peninsula, calculated from 1991 satellite imagery. Values are
“Wainscreatedwithan 800mbuffer ............................... 146
like 111 N. ,

 

    

flWiucs by cover type (m/ha, means and standard errors) for study
human Department of Natural Resources (MDNR), U. 3. Forest
”$95), timber industry (TI), and Huron Mountain Club (HMC) land 1n
Michigan’ 8 Upper Peninsula. No significant differences (p>0. 10) were detected.
Twit 4.7. ‘;{( ‘1‘ r
Tablet”. Values for landscape metrics (means and standard errors) of northern
Wood forest patches for study sites on Michigan Department of Natural
Resources (MDNR), U. S. Forest Service (USFS), timber industry (TI), and Huron
I t‘rMumtain Club (HMC) land in Michigan’s Upper Peninsula, calculated from 1991
smilite imagery. Values are for study sites created with an 800 in buffer. No
significant differences (p>O.10) were detected ............................. 152

150

Table 39. Spearman rank correlations between proportion of vegetation cover types
in the landscape and red-backed salamander relative abundance (1997 and 1998)

on study sites in Michigan’s Upper Peninsula .............................. 154

Table 40. Spearman rank correlations between landscape metrics calculated from
1991 satellite imagery and red-backed salamander relative abundance (1997 and

1998) on study sites in Michigan’s Upper Peninsula ......................... 155

Table 41. Spearman rank correlations between proportion of vegetation cover types
in the landscape and American redstart relative abundance (1996, 1997, and 1998)

on study sites in Michigan’s Upper Peninsula. ............................ 157

Table 42. Spearman rank correlations between landscape metrics calculated from
1991 satellite imagery and American redstart relative abundance (1996, 1997, and

1998) on study sites in Michigan’s Upper Peninsula ......................... 158

Table 43. Spearman rank correlations between proportion of vegetation cover types
in the landscape and ovenbird relative abundance (1996, 1997, and 1998) on study

sites in Michigan’s Upper Peninsula ..................................... 159

Table 44. Spearman rank correlations between landscape metrics calculated from
1991 satellite imagery and ovenbird relative abundance (1996, 1997, and 1998) on
study sites in Michigan’s Upper Peninsula. Values in bold are significant at

1150.10 ............................................................ 161

Table 45. Spearman rank correlations between proportion of vegetation cover types
in the landscape and veery relative abundance (1996, 1997, and 1998) on study

sites in Michigan’s Upper Peninsula ..................................... 162

 

2w
1‘.“
I“!
.ul

,4

k .

:15.
A ‘

 

" :1

x

31
.41‘1-1‘
“i\

i:-

"l

 

 
  
  
   
   
  
  

-' correlations betweenlandscapemetricscalculatedfmm
- , a ndveerynelativeabundance (1996, l997,andl998)on
. y ’s‘UpperPeninsula. .............................. 163

7 Mmkeorrelations (r,) between proponion of vegetation cover
. fl “scape and yellow-rumped warbler relative abundance (1996,
~ -i- 1398,emnbined) on study sites in Michigan’s Upper Peninsula. ..... 164

' ' M rank correlations (r,) between landscape metrics calculated from
-. imagery and yellow-ramped warbler relative abundance (1996,
MJQQS) on study sites in Michigan’s Upper Peninsula. No significant
1 .ifierences (p>0. 10) were detected ...................................... 166
lelc E” .
m. Spearman rank correlations (rs) between proportion of vegetation cover
types in the landscape and pileated woodpecker relative abundance (1996, 1997,
and 1998) on study sites in Michigan’s Upper Peninsula. .................... 167
Table 50. Spearman rank correlations (rs) between landscape metrics calculated from
1991 satellite imagery and pileated woodpecker relative abundance (1996, 1997,
Md 1998, combined) on study sites in Michigan’s Upper Peninsula ............ 168

',--r

 

Table 51. Spearman rank correlations between proportion of vegetation cover types
in the landscape and barred owl relative abundance (1996, 1997, and 1998) on
Study sites in Michigan’s Upper Peninsula. ............................... 170

Table 52. Spearman rank correlations between landscape metrics calculated from
1991 satellite imagery and barer owl relative abundance (1996, 1997, and 1998)
on study sites in Michigan’s Upper Peninsula. ............................ 171

Table 53. Proportion of home ranges (%) in each cover type at sampling points where
barred owls were detected and points where they were not detected for 1996,
1997, and 1998 combined on study sites in the Upper Peninsula of Michigan. . . . 172

Table 54. Spearman rank correlations (rs) between proportion of vegetation cover
types in the landscape and fisher relative abundance (1996, 1997, and 1998
combined) or! study sites in Michigan’s Upper Peninsula. ................... 174

Title 55. Spearman rank correlations between landscape metrics calculated from
1991 satellite imagery and fisher relative abundance (1996, 1997, and 1998,
u' .. Whirled) on study sites in Michigan’s Upper Peninsula. ................... 175

 

 

  
 
 
    
  
  
   

‘ 1., ‘ ‘ ‘ " ' “ measuredforstandsinthe
__f 'areasof theHuron Mountain Club 1n Michigan’ sUpper
' gamma, 1996. 1997, and 1998. ........................ 187

local-o -

 

- . vegetation variables measured for stands in the
_ »< ms of the Huron Mountain Club 1n Michigan’ s Upper
“august, 1996. 1997, and 1998 .......................... 189

~ My tree species composition (stems/ha) for stands in the reserve and

-. ' ‘ runs of the Huron Mountain Club 1n Michigan’s Upper Peninsula,

hiymdAugust, 1996 1997, and 1998. ................................. 190
3

We Mm values for landscape metrics for stands in the reserve and nonreserve
M Of the Huron Mountain Club rn Michigan’s Upper Peninsula, calculated
M11991 satellite imagery ............................................ 192

Twin 60. Mean absolute frequencies (percent of points at which Species occurred),
pooled over 3 years of data collection, for bird species surveyed in reserve and
nonreserve areas of the Huron Mountain Club in Michigan’s Upper Peninsula,
May-July, 1996, 1997, and 1998. Indicator species are in bold ................ 197

Table A1. Global positioning system coordinates (taken from the approximate center
of the stand) for stands sampled on northern hardwood forest study sites in
Michigan’s Upper Peninsula, June-August, 1996, 1997, and 1998. Salamander
surveys were conducted at the shaded coordinates. ......................... 214

Table El. Common and scientific names of bird species recorded on nonhem
hardwood forest study sites in Michigan’s Upper Peninsula, June-August, 1996,
1997, and 1998. .................................................... 256

’rsgm-‘r .
WC Fl'h‘i ‘

 

 

..
n

1
r 1

ii1
-\1
It
i/,'
a
be”
-;.:

111
'-u
9
"s
'v 'V.
"-1”

éL.‘ s

«>03

  

‘10. Aster-sq 1‘ . LIST OF FIGURES
. 'W prom" 1 ~ '
IO m bun.

W locations of 1996-1998 study sites on state (MDNR), federal
mutual)“ industry (Mead Company and Shelter Bay Forests), and Huron
Mountain Club land in Alger, Chippewa, Luce, Mackinac, and Marquette counties

1 maths Upper Peninsula of Michigan. ..................................... 10
w- .

Figure 2. Arrangement of cover boards and ground transects used for salamander
' _~ surveys in Michigan's Upper Peninsula, 1997 and 1998 ....................... 28

Figure 3. Scores for the first 2 principal components (PC) for vegetation variables
measured on study sites on Michigan Department of Natural Resources (MDNR),
U. S. Forest Service (USFS), timber industry (TI), and Huron Mountain Club
(INC) land in Michigan's Upper Peninsula, 1996, 1997, and 1998. ............ 43

Figure 4. Proportion of natural cover objects and artificial cover boards with
salamanders found beneath them, summer and fall, 1997 and summer, 1998 in
Michigan’s Upper Peninsula ............................................ 57

Figure 5. Scores for the first 2 principal components (PC) for songbirds which
occurred at a frequency 220% on at least 1 study site on Michigan Department of
Natural Resources (MDNR), U. S. Forest Service (USFS), timber industry (TI),
and Huron Mountain Club (HMC) land in Michigan's Upper Peninsula, 1996,
1997, and 1998. ..................................................... 80

Figure 6. Mean number of cavity nesting and noncavity nesting birds per sampling
point on Michigan Department of Natural Resources (MDNR), U. S. Forest
Service (USFS), timber industry (TI), and Huron Mountain Club (HMC) land in
Michigan's Upper Peninsula, 1996, 1997, and 1998 .......................... 82

Figure 7. Mean number of birds per sampling point by migratory status on Michigan
Department of Natural Resources (MDNR), U. S. Forest Service (USFS), timber
industry (TI), and Huron Mountain Club (HMC) land in Michigan's Upper
Peninsula, 1996, 1997, and 1998. ....................................... 82

Figure 8. Illustration of sampling points and buffers used to define landscape
boundaries for study sites in Michigan’s Upper Peninsula. ................... 132

Figure 9. Examples of the interspersion and juxtaposition index (1.11). Calculations
were performed with the Patch Analyst extension to ArcView. ............... 137

 

  
  
  

”Men of total land cover represented by each cover type and ',, -
. . , , 7 Menu! total roads in each cover type for study sites created with an 1

, flaming in the Upper Peninsula of Michigan. Probability that the proportion

' Whit”! cover type differs from the availability of each cover type is 0.28

:WW). ................................................... 151
m 11‘. flee diameter distributions for the reserve and nonreserve areas of the
M’Mormtain Club in Michigan’s Upper Peninsula, 1996 - 1998. ........... 188

m 12'. Average proportion of total land area in each land cover class for study
sites in reserve and nonreserve areas of the Huron Mountain Club in Michigan’s
Upper Peninsula, calculated from 1991 satellite imagery ..................... 193

Figure 13. Abundance and size of woody debris used by salamanders for stands in the
reserve and nonreserve areas of the Huron Mountain Club in Michigan’s Upper
Peninsula June, July, and August, 1997 and 1998. ......................... 195

Figure 14. Abundance and size of woody debris for stands in the reserve and non
reserve areas of the Huron Mountain Club in Michigan’s Upper Peninsula June,
July, and August, 1997 and 1998. Note that the scale of the y axis is logarithmic. 196

Figure 15. Average response rate of barred owls for stands in 2 reserve areas and 1
nonreserve area of the Huron Mountain Club in Michigan’s Upper Peninsula,
July—August, 1996, 1997, and 1998. .................................... 199

Figure B1. Relationship between Variable l, the density of trees 2 10.2 cm dbh, and
red-backed salamander habitat quality. .................................. 228

Figure BZ. Relationship between Variable 2, the percent canopy cover of shrubs and
regenerating trees 0.5-5 m high, and red-backed salamander habitat quality ...... 228

Figure B3. Relationship between Variable 3, the density of woody debris 10-40
cm wide, and red-backed salamander habitat quality. ....................... 229

Figure C1. Relationship between Variable l, the percent conifer cover in the
overstory, and yellow-rumped warbler habitat quality ....................... 239

Figure C2. Relationship between Variable 2, the average height of overstory trees (25
m tall, 210.2 cm dbh) and yellow-rumped warbler habitat quality .............. 239

Figure C3. Relationship between Variable 3, the total percent overstory canopy
cover, and yellow-rumped warbler habitat quality ........................... 240

 

«r N
l-“»
n

       

   

" «i A 1m flying squirrel habitat quality. ........................ 252
bib: il’llL’. ma".

    
  

' " *- ' v ' - flying squirrel habitat quality ................................ 253
Illa dull \ I 1

mm. Relationship between Variable 4, the percent conifer cover in the
W and northern flying squirrel habitat quality ....................... 253
M Shir.i.'..‘

agenda-

2il.‘Jil'. Ii ' ‘

l . 9.
[Mint

i
WW'DL3VIg‘ , . I

“dMit'9‘Jga'i .. .‘ ‘1 ‘ .i

   

- 1 .lti.‘ HWY“:
\.

 

 

Q‘QI.
1"

“skin-

 

 

 

 

 

 

 

r.
‘a.
.1.

"-ii

  
   
 

 

   

w the or m u. -
AI;7,‘.,,,""‘I' facing natural resource managers implementing ecosystem
.-. fmajorrwr "
" _. '5“ its integration of differing goals and objectives among land ownerships
' ~ today has I ...
(Allen 1994, Sample 1995). Often, the demands placed on individual

  

      
     

V " ‘ merry drive the land use decisions made by public and private landowners.
ol ‘ .

m then consideration for the role of that property in the landscape. However, the
*gbdfiir'u: . .

W of ecosystem management as the cornerstone of natural resources management

balsam. .; - I
III stimulated changes in the goals and management approaches of natural resource

:eiicies. In managed forests, these changes have involved a greater emphasis on
maintaining ecological integrity and biodiversity, integrating social perspectives into
management plans, and providing nontimber forest products, such as recreational
opportunities and nongame wildlife. Nonetheless, the administrative boundaries that
separate land ownerships remain obstacles to coordinating natural resources management
at the large scale necessary to encompass forest ecosystems and conserve many wildlife
species (Allen 1994, Sample 1995).

In a landscape that spans political boundaries, such as the northern hardwood
forests of Upper Michigan, the combined effects of differing management goals among
land ownerships can impact the outcome of management in each area. Approximately
80%: of Michigan’s Upper Peninsula is forested, and forest resources have historically

ml in‘a'u‘
been one of the major social, economic, and environmental driving forces of the area ,
i‘ .

“Myer/mugs .t. .
' . -— I993, Schmidt 1993). This heavily forested region is divided among

“TOWN . . . 1' - "'w? L3 u‘h moi:

  

 

x‘. .

.10

till, i

“(14‘

It;
,1

w
..‘
&

 

  

, . we; - L refleettheiversity

 

 

G If

  

‘ yltemgandthe diversity ofproduets that-the system-
. factors that has shaped the forests that dominate Michigan’s

   
 

, been the extensive logging that occurred throughout Michigan

  

' --d”early 19005. Prior to this time period, Upper Michigan was covered
.‘Wifuelt dominated by sugar maple (Acer saccharum), yellow birch (Betula
W), and hemlock (Tsuga canadensis) in drier areas, and balsam fir (Abies
m, spruce (Picea spp.), and northern white cedar (Thuja occidentalis) in lowland
In. There were also areas of pine (Pinus spp.) dominated forests on the drier sandier
soils, and a few pockets of mesic deciduous forest, composed of sugar maple and beech
(Fugue grandifolia) (Frelich 1995).
Before 1870 there were over 4 million ha of forest in the Upper Peninsula; by
1941, 75% of the original forest had been subjected to logging or other types of
development (Cunningham and White 1941). The Upper Peninsula logging industry was
driven by white pine (Pinus strobus) and Norway pine (Pinus resinosa) harvest
(langbome 1988), despite the fact that pine was the dominant species in only 15% of
Upper Peninsula forests (Karamanski 1989). Pine was initially the most important tree in
the logging industry because it was valued for its light weight and strength as a building
Internal. It also occurred in relatively large, isolated stands (Langhorne 1988). Perhaps

“My, it was buoyant enough to be transported to sawmills by water, which

hum transportation medium for that era and region.
’ awww.mmdxliernlock..begmtobeexploitedaswell. There

 

.11 't‘
U an:

9 .I
9.1"
..e-.\;

£25

"‘V‘s'
>.Il4\i

l
u,,,

. i
«'5'...

2}}.
Jim“

t
J-u

a“.

   
   
   
  
   

. . ,, n ahavtofhudwoodforest in theUpper Peninsula.
andhemlock, with smalleramountsofyellow birch,
» ,and green ash (meinus pennsylvam‘ca) (Karamanski 1989,

" ; W, hardwood species received little industrial attention until the

 

mmu extensive railroad transportation system had been established and
Wm diminishing.

micely'lm. timber supplies were greatly reduced, and the logging industry declined
Minder. the effects of the Great Depression. Agriculture in the Upper Peninsula had
uninvolved with the logging industry, and it too, suffered a decline immediately after
World War 1. leading many farmers to abandon their farmland. Between 1921 and 1932,
“.000 ha of abandoned forest land reverted to the State of Michigan. In 1931, the
Naional Forest Reservation Commission approved land purchases in the Upper
Peninsula to create the Hiawatha National Forest. The lands returned to federal

ownership were among those most severely impacted by previous logging activities.

Much of the land had been logged and then burned over several times, and after clearing,
the land was often maintained as agricultural plots. In addition, lumber companies whose
land had been cut over and were no longer returning a profit offered large pieces of land
for sale to‘the federal government. The Civilian Conservation Corps, established in 1933,
My improved the condition of publicly owned land by aggressively planting trees and

cancelling fines on national and state forest land (Karamanski 1989). Thus, the Upper

mm that confirmed past 1920 represented a transition from an era of i ‘

 

‘\ ‘r
it“s“

CHE:

' I)
:1;

(v.

&

'L a“...

‘2, ‘1'
.131 -
‘

 

  

' warmth an Wehuedbymynamral
. .hdzmm, the preciseobjectivesandmanagement
. may be quite different. Current management of US. Forest
' ~ -~ lands seeks to meet multiple use objectives, which includes
recreation, range, timber, watershed, and wildlife and fish
m umdued by the 1960 Multiple Use-Sustained Yield Act. The Forest
Wanted that an ecosystem approach to management is central to achieving its
“objectives, and has recognized the necessity of landscape scale assessments
(Hrs. Department of Agriculture 1995). For example, in 1994, The Forest Service
med in the Northern Lower Michigan Ecosystem Management Project, in
collaboration with the Michigan Department of Natural Resources (MDNR), several other
federal agencies, and state and local conservation groups (Michigan Department of
Natural Resources 2000). The intention of the project was to evaluate and make
recommendations on how public land in northern lower Michigan can be managed in
oursideration of the interactions among the social, economic, and ecological components
of the region. Products of this project have included a Resource Conservation Guide,
which explains the interrelationships between human and natural communities for public
audprivate stakeholders in Michigan natural resources management, and a classification
map of all ecosystems, grouped by climate, topography, soils, and vegetation, in northern
W not bound by the same legislation as the Forest Service, the MDNR has

 

  
   
   
     
 

Fill:
Jim-5

 

”MM
,,\|~\
, .

*‘W‘F‘r
vary}
A

 

"if.

~..-i...'.

mt

Cid.

 

V
"\r\-‘

h, [-u

 

P“.
‘i,_ A
i. w.

 

 

7.21

h‘

   
 

' ' . ‘ sm'prodncing game wildlife and timber revenues. In addition to

1.- u. .a in the northern lower Michigan ecosystem management

~lexunple of how the MDNR is moving in this direction is its draft of a
WW management plan for the Escanaba River State Forest in
W‘sflpper Peninsula (Michigan Department of Natural Resources 1991). One
thhlch this plan is different from past management efforts is that the focus of
managemt is moved away from forest stands to larger scale ecological management
units, in which classification is based on climate, physiography, soils, and vegetation
types. The plan also provides criteria for designation of old-growth forests that allow
more integrated management of current and future old-growth with the rest of the
landscape (Begalle 1991, Michigan Department of Natural Resources 1991).

Though under much less public pressure to conform to the principles of ecosystem
management, wood products industries are also making changes in their forest
management practices. The timber industry has recognized that sustainably high timber
production is not possible on all industry lands and that other objectives are compatible
with intensive timber management (Wright 1991). Wood products companies are also
aware that the industry can ultimately benefit from a more balanced management
approach that integrates public values and generates a positive public perception of
timber industry management practices. One example of timber company efforts to

”attire ecosystem level is the Total Ecosystem Management Strategies (TEMS)

' - '_ yd “Ecosystem Management Strategies is a collaboration initiated in 1989

 

tilt?

:llt’:

I}! «‘3
.n; ll»

TTV‘

1m and

32 06C

r4
"MM

‘2"

"up“
5

'Q.

I .'
{Jim

  
   
  

. a wood products company, and White Water Associates.
' t; firm (Ticknor 1993). The project used White Water’s
. r evaluate the effects of Mead’s management practices on ecosystem
.. j g 7 ' . 10“ ha forested landscape owned primarily by Mead. Part of the
law was to assess songbird and mammal use of the area, to relate the results
”timber harvesting practices, and to improve those practices to maintain the
Wof native ecosystems where possible.

.. n - Objectives for management of nonindustrial private lands, owned by individual or
private organizations such as the Huron Mountain Club or private hunt clubs, are perhaps
the most diverse of all types of land ownership. Private land owners are a more numerous
group, their lands range in size from less than a hectare to several thousand hectares, and,
unlike institutionalized resource management organizations that are accountable to public
and scientific opinion, private landowners are only accountable to governmental laws and
regulations for their management decisions. Nonindustrial private land management
objectives may include hunting, wildlife viewing, economic return from timber
production, maintenance or restoration of ecological integrity, or any combination of
these objectives. An additional example of private nonindustrial forest management is
the Huron Mountain Club’s goal of preserving the condition and natural processes of
their forests, much of which have never been logged.

Managing to meet different ownership objectives can result in differences in the

Structure and composition of forests and landscapes. In addition to current management

'~ ». (Idler factors that may influence landscape characteristics include past

 

 

,- 1
T. i. ‘l
U-..“

y
.

iLTT;

33m

\

 

 
  
  
  
 
 
 
   
  
 
  
   
 

 

  
 

makeimportanttoconsider when planningmanagement
“m’ai Ws‘fl' 1‘

We and ecosystem management objectives. For example, white-
u commit uh ‘

1|.'

, 4 , minianus) in the Upper Peninsula travel an average of 9.7- 14.5
bulliUpif: 11 "

11';

W and winter ranges (Verme 1973), and the spatial arrangement of
timber pr: 1 ‘

.go (si'v 1.

Id Winter ranges has been found to influence the population structure of deer on

Tim range (Van Deelen 1995). Patterns of silvicultural practices among
’ {men 0.

law” can also determine the spatial diversity of the entire landscape (Mladenoff et
at wildh » ~
11993).

In the northern hardwood forests of Michigan’s Upper Peninsula, the ult1mate
iiitptiets of varying management approaches, past disturbances and abiotic factors on
idea and wildlife resources across ownership boundaries have not been rigorously
documented. Therefore, the focus of this project was to evaluate and compare the
impacts of management approaches guided by the goals and objectives of several
different land ownerships on northern hardwood ecosystem characteristics This

information may be used for coordinating forest management that will meet 1ndtv1dual

agency and land owner objectives, yet will maintain biodiversity and ecological integrity
across the landscape.

\a'i’it

'i¥(

 

 

 

17:}.

“ M n

:0
f x
r.-.

y.
/

 

 

OBJECTIVES

Specific objectives of this project were to:

1.) Model and compare the effects of 4 different management approaches
(management for multiple use [federal forest management agency], management for game
species and timber products [state natural resources agency], management primarily for
timber products [timber industry], and minimal management with the goal of preservation
[private forest land that has never been manipulated]) on ecosystem conditions and forest
and wildlife resources, at spatial scales ranging from the stand to the landscape level.

2.) Compare forest and landscape characteristics, and wildlife relative abundance
in mesic hardwood forests that have been subjected to minimal human disturbance with
areas last manipulated in the early 20th century.

3.) Develop recommendations for ecosystem management that will maintain
biodiversity and ecological integrity across a landscape that is managed with different

goals and objectives.

 

.q r in.
1:1 1'"

rib: JIAI

“a. l“ 7.
\a.‘ fifth

"II

‘p‘ Jr

'-~slrh.\‘
I

.-- '5'.

Z'M'

n._j

 

 

  

‘ of the 4 management categories evaluated in this study. Three replicate study
d . ’_‘ - win the Hiawatha National Forest, representing the Forest Service
_ . _. .7 category, and 3 replicate sites were selected in the Lake Superior State Forest
. Effluent the MDNR ownership category (Fig. l). The private/non-industrial

K" , category was represented by the Huron Mountain Club. Three additional
' were identified in 1997, l of which was on land owned by the Mead Company, and
Rhfwhich were on land owned by Shelter Bay Forests. These sites were added to the
study to fill the private industrial forest management category of study sites.

Of the 3 study sites located in the Hiawatha National Forest, 2 were in the western
part of the forest in Alger County, and 1 was in the eastern part in Chippewa County
(Table l). The Mead site was also in Chippewa County and the 2 Shelter Bay sites were
in Alger and west—central Luce Counties. In the Lake Superior State Forest, 1 site was
located in Alger County, 1 was in northeastern Luce County, and l was across the border
of Luce and Mackinac Counties. All study sites ranged from 12-19 km2 and contained
the largest proportion of contiguous northern hardwood forests available, based on GIS
coverages, within each land ownership. There was a mix of successional stages and

amusement intensities among the study sites, ranging from recently thinned mid-

mortal standsto unmanipulated old-growth forest.
(”3“). Limos: . ‘
I; my Aha-.141“. and Chippewa counties are in aregionthatia Mu] y
y Michigan

-

    
 

   

   

Michigan‘s Upper Peninsula

Scoolcraft ”$4

County

  
  

   

Alger County

  
  

    

Marquette County

   
 

Schoolcraft Chippewa County

County

      
   

   
 

Delta County ‘

 

Mackinac
County

 
   
 

I - MDNR study sites

0 - USFS study sites
A - Mead Co. study site
V - Shelter Bay study sites

+ — Huron Mountain Club study site

Figure 1. Approximate locations of 1996-1998 study sites on state (MDNR), federal
(USFS), timber industry (Mead Company and Shelter Bay Forests), and Huron Mountain
Club land in Alger, Chippewa, Luce, Mackinac, and Marquette counties in the Upper
Peninsula of Michigan.

{BPS

331C

1'

Table 1. Legal descriptions of locations for Michigan Department of Natural Resources
(MDNR), U. S. Forest Service (USFS), timber industry (TI), and Huron Mountain Club

(HMC) study sites in Michigan’s Upper Peninsula.

 

 

Ownership County Township and Sections
category Range
MDNR Alger T47N, R17W l7 18 19 20 21 28 29 30 32 33 34
MDNR Luce T48N, R8W 18 19 20 29 30 31 32
MDNR Luce/Mackinac T45N, R10W 3O 31
T45N,R11W 25 26 35 36
T44N, R11W 1 2
USFS Alger T46N, R2OW 22 23 24 25 26 27 34 35
USFS Alger T46N, R21W 4
T47N, R21W 20 27 28 29 32 33 34
USFS Chippewa T46N, R5W 1 2 3 4 5 9 10
T47N, R5W 34 35
TI Alger T48N, R13W 30 31 32
T48N, R14W 25 26 35 36
TI Luce T47N,R11W 5678918
T47N,R12W 1 121314
TI Chippewa T47N, R6W 8 17 19 20 21 22 26 27
HMC Marquette T51N, R29W 1 2
T52N, R29W 13 14 23 24 25 36
T52N, R28W 18 19 20 29 3O

 

 

 

 

 

‘1' Hyf

‘blt

 

 

 

 

 

 

  
  
  
 

M about 114 days. Elevations range from 183-378 in, and the
«poorly drained sand lakeplains and well drained sand end-moraine
1.; average 14.4 C from May through September, with an average of
Uf'ptecipitation falling annually (Albert et al. 1986).

(hesite in the Lake Superior State Forest lies across the border of Luce and
WCounties, where northern hardwood forest is the dominant vegetation type in
opt-1d areas, and stands of northern white-cedar, balsam fir, and spruce grow in poorly-
drained areas. Limestone bedrock and sand lake plain occur in this region. Soils
daughout the area are sandy, and topography ranges from 178-317 m. Average annual
precipitation is 810 mm, and temperatures average 14.9 C in summer (Albert et al. 1986).

The Huron Mountain Club property, located in northwestern Marquette County, is
the largest and potentially most representative example of an old—growth northern
hardwood ecosystem in Michigan. The land is privately owned and consists of
approximately 7200 ha. A portion of the land, approximately 3300 ha in size, received
only light or no cutting before 1900 and is designated as the “reserve area.” The
remainder of the Huron Mountain Club land (the “nonreserve area”) surrounds the
reserve area and was last cut in the early part of the 20th century (Simpson et al. 1990).
Both the reserve and nonreserve areas of the Huron Mountain Club were sampled for this

project. The physiographic region in which the Huron Mountain Club is located is

Mud by steeper topography than other areas of Michigan, with elevations ranging

 

17ml; [

, b J
“LL“:

“fig

.t' l

rim

3,“, .
3.1.4.1
..

“w. A.
2",“:

n . '
if)“. I

if: r,

 

  
   
 
   
      

,‘ bmsively well drained sands. Northern hardwood forests and

’ courier swanps are the most prevalent habitat types. The average
' ... H (1i\" '

' is 15.0 C and the growing season typically lasts 89 days. An
practu r:
arm of precipitation falls annually (Albert et al. 1986).

Wild: H

i9“ and 1997, the average annual temperature for the eastern half of the
l . jalbod N . . s ‘

y I
will .'

   
    
  

was below the 20 year average, and in 1996 the average temperature (4 4
s that

an lower than in any other year since 1980. Cumulative precipitation for the eastern
milk :1

w Peninsula was above average in 1996 and below average in 1997 and 1998
gait [Ragional Climate Center 1999).

Prior to European settlement, most known disturbances in Michigan northern
adv/OMS occurred in the form of wind and ice storms, which removed <20% of the
canopy m a stand. More severe disturbance events, which removed >60% of the canopy,
were much less common, and occurred on a rotation of more than 1000 years in the

Upper Peninsula (Frelich and Lorimer 1991). The result of the historical disturbance

patterns was a mostly uneven aged forest structure, with as much as 90% of the

 

Moment northern hardwood forest in old growth condition (Frelich 1995)

 

 

  

NRODI
Dc

, .«
Nldfi"v
".b.-\

, .9 .
... :‘gh ..

:saLlle\-

twp

‘Fi‘wsi
.AL.

 

 

 

 

   
 
   
 
  
  
 

" v 'on‘olf‘ wildlife habitat quality and quantity in relation to forest

_ » ”ghoes is an important link between landscape-level habitat evaluation
“:Iwildlife management. This link can be established by implementing a fine—
km of evaluation (Noss 1987) to describe important habitat elements and
$3993 that occur on a relatively small spatial scale (Allen 1994, Roloff 1994),
mm with a coarse filter approach (The Nature Conservancy 1982) to integrate
gormafion about the large scale habitat characteristics of a landscape.

At the stand level, a fine-filter evaluation can be applied by measuring vegetation
characteristics and determining habitat conditions for individual wildlife species. Habitat
suitability index (HSI) models are a widely used method of evaluating habitat quality for
individual wildlife species on a local scale. The fundamental premise of HSI models is
that species’ abundances and distributions can be predicted from measurements of habitat
parameters (Marcot et al. 1983). While many research studies have quantified key habitat
attributes for particular wildlife species, HSI models offer the advantage of being a
Standardized, repeatable evaluation method. Assumptions associated with HSI models

are that observed wildlife abundance is a function of habitat quality and that HSI model

outpm is positively and linearly related to habitat quality for the species of interest
dbl! 11:1, .v ..
Merger and O’Neil 1986).
w anal Vat \ 1
-Hahitat suitability index models are useful 1n a management context for

 

3342le

on
.2
.\

g» ..

11.,

3?:
I mg

m :
--l iii“:

in; ‘1
3.215. P

 

a
F;._ .
g“ R as

v.
a“

 

 

  
   
  
  
 

a» quality though time and in response to specific

- iltiihoutdimctly measuring population density (O’Neil and

‘ . 1983). This is an advantage when managers are interested in evaluating

-, ._ Llama] species at once, or when a species is difficult to survey. However, use
‘1,“ ~ " : has been criticized based on their lack of validation (Cole and Smith 1983,
filial. 1984). As use and development of HSI models has been scrutinized and
m. other researchers have identified additional considerations for using HSI models,
'mluding sample size adequacy (Cole and Smith 1983), the spatial scale of model
Qplicability (Roloff 1994), and the amount of variance associated with model inputs
(Bender et al. 1996). Other studies have demonstrated the accuracy and usefulness of
HSI models for particular wildlife species (Thomasma et a1. 1991, Roloff 1994, Negri
1995). Although there has been no established technique for consistently validating HSI
models, Roloff and Kemohan (1999) recently proposed a protocol for validating HSI
models and improving their reliability. These researchers concluded that, if used
conscientiously, HSI models are a valuable and practical tool for habitat assessment.

This chapter addresses part of the first objective of this project, by comparing the
effects of 4 different management approaches on forest and wildlife resources at the level
0f the forest stand. The fine filter evaluation in this study consisted of stand level
Inalyses and habitat modeling to reveal habitat variables that may be driving species

Minions and relative abundance, yet would be overlooked in a coarser, landscape-

lepis. This information on stand-level habitat characteristics and relationships to

 

the: s:

“will-1
1 aflul‘tl

   
  
 
  
 
 
  
  
 
 
 

"'P emf-for?!“ (\'.A,!U.1fi r.‘ 1'. .r r:

'H .‘mmt “wry" a -'

 

; lanii "1‘3 1 .. 1'

Wncd t“.
MMCCM:
a.
’ Ninewilrlzzv-

Gill"; ‘.1

   
 
 
 
 
  
 

hlpfiu‘ '

'-‘ forth» 1 2 2‘ ,
-‘- reqmz. . .
IMIRIIK‘M ".' n:

80011,th ‘ ,

 

.“

 
 
 

Di‘ 1‘

dcr 1. {‘11

 
 
   

(5181. 15).":

    

 

)' )’ CHOW ”J ”Rd *3", “A, '1 'u' ,. :mg:..‘.,t;~;11.tl_\10(

      
    

   

  

Wrist“), balm) “WI 153-'1‘ . "rs/'21:. and trshcr (Mann {mm}.
K I

 

llElH

Experi

It} 0

if“)?
to ,ix

f' M
it “-11
M.“

“grit
If] .1‘.

.
""""

 

4* Mg late successional stage northern hardwoods and minimizing human disturbance.
E I Nine wildlife species associated with northern hardwood forests were selected as
Mrcators of the range of conditions and relative habitat quality of hardwood forests in
6911 ownership category. These species were selected based on their use of northern
hardwood forests to meet their life requisites, and on the range of spatial scales at which
their habitat requirements occur (Table 2). Representation of wildlife with a variety of
habitat requirements was important to understand the contributions each ownership is
providing to wildlife habitat quality in the landscape. Species chosen were the red-
hacked salamander (Plethodon cinereus), pileated woodpecker (Dryocopus pileatus),
wan redstart (Setophaga ruticilla), ovenbird (Seiurus aurocapillus), veery

-‘

11
, .
. (Glauc‘omys sabrinus), barred owl (Strix varia), and fisher (Manes permanrr').

#-

   

fuscescens), yellow—rumped warbler (Dendroica coronata), northern flying

‘. ',.L- .7 _.. 393mm were selectedbecausetheymeettheirlifetequisites

  

  

 

 
   

,noora'neebergsdema murmur“; __ '

  

p..v. —-«.Ja vi—

:‘UQI. DLUIIII l .II-lilI-uo-Iill aria-Al. I ll.-olo.l“-..i-I\r’- no. Del-\Itvtl alo1r. 1..irea-l new! vice-lunohllllri riv‘lhlal til-IIII‘ .vria\1\§e§~.I~ II 1I\A\O\.\.

 

 

AonEob a: 82

do mE
2:32 3% .3 do 55% Ame—«EV a: Q: m n
30 Dawn
Swan: £2 :85 e5 32m 2 $3: _ _ e
comm: .2: 08 N09 23 a: 0%-va
. Z aiming“ Era. coowwo USN $63 an m.N_-_ m 3:33 meta E0552
some Sim— 3282 use 55 a: mvméfl
«ESE .Ewcoom omE :5:sz a: on coxoomcooa @0323
ofiao 82 5E2 2 2. 73¢ b8> m
canoe Sam 02 $383 comEEio=o>
5&322 RE 5am 2 3-3
08853. 52 Ewanw Ea 5E5 as 5.0 95:30
for >62 92 :85am an med
020 9&— EcBm a: omd 55?: 580:2.
53:22 $2 5503 98 53389 NE w.v-o.m

Hoefifiafim coo—82-3%

 

.8583 oocsom emcee 080: 389mm

 

.wagroaafi .fisweEom Saab PEMEOME E cabin—m 3.5on do» 33:58 owned 050: gnaw—nut .N 031“.

 

 

 

21:31.0

ns‘:\\t‘-
9,‘ a)!“

'.-,. -.\.1.

3:771 t1

 

ft
KU

r1)

 

    
 

‘ y31 kmz(Arthuretal. 1989). Veeriesaregmund

 
  
 
   

. .. er mist forests with a dense understory (Winnett-Murray 1991).

’ ,. (”rally associated with mature hardwood forests with an open

., and Villard 2001), and the American redstart prefers early

. timests with high stem densities (Bond 1957). Pileated woodpeckers and

'mnflying squirrels are both cavity nesters and depend on older forests to provide
W a supply of large dead trees (Cowan 1936, Weigl and Osgood 1974, Bull and
Meslow 1977, Carey and Witt 1991). The yellow-rumped warbler was selected because,
flthough it frequently occurs in Michigan northern hardwood forests (Eastman 1991), its
association with northern hardwood forests is weaker than the other species chosen. The
yellow-rumped warbler occurs regularly in conifer forests, and less often in northern
hardwoods unless they have a conifer component (Hagan et al. 1997). Therefore, the
yellow-rumped warbler represents a species which occurs at one end of the range of
conditions that may occur in northern hardwood forests.

In 2 of the designated land ownerships (MDNR and Forest Service), 3 study areas

were identified in each ownership and were sampled during the 1996, 1997, and 1998
field seasons. One site owned by Mead Co. was identified at the end of 1996 and was
included in 1997 songbird, barred owl, and red-backed salamander data collection.
Permission to work on 2 sites owned by Shelter Bay Forests was granted by Martin Wilk

lithe end 01’ 1997, and these sites were sampled in 1998.

'mlmflglication of sites within the privately owned, low intensity management

 

 

 

ll.

m

 

‘\ l
..’v-
a,“ I1
..
U
0
7",“

gr.

   
    
  
  
   
 

represerrwd by the Huron Mountain Club, which was
provide information on habitat conditions in a forested landscape
M1111 disturbances rather than silvicultural activities. For sampling

' Mountain Club was divided into 3 separate areas, approximately 2.5-
. E ‘_ filed on ecological and geographic boundaries, such as the numerous lakes

- property and the borders of the reserve and nonreserve areas of the Club.

' III-pins

WM: Vegetation characteristics were measured in 104 stands. On MDNR, Forest
Mice, andtimber industry land, 1 survey point was established in each of 7-12
Wood stands throughout each site, based on the size of individual sites. Points were
placed approximately every 1.6 km along permanent transects located 1.6 km apart to
fare a rough grid pattern across each study site. Locations of sampling points were
recorded with a Global Positioning System (GPS) unit (Appendix A). This systematic
sampling pattern was chosen to obtain data that would represent the range of conditions
in the northern hardwood forest stands present on each site. The Huron Mountain Club
WIS Mpled more intensively to allow comparisons of areas logged early in this century
with actions that have never been logged. In each of the 3 areas chosen for sampling at
Illefluron Mountain Club, 6-8 points, located approximately 0.6 km apart, were

mmmWithin each stand selected for vegetation sampling, 3 sampling points were

  
   

a wrthr' 'n a 200 m radius of the birding point associated with that stand.

, mmsunpledoneefromsummertoearlyfall. Am.-

 

 

no! ,1‘

1"” Ha

I....
L w

a“; i
‘1‘. i
«.4 t
’3 l
"hp-gr.‘
.

. - .
‘ \
A. r. .
- b

 

r . ‘
l‘ I
"AM. 5

in.

w ,L‘i‘l

 

‘tl’ I}?

 

 

 
   
    
  
    

“_. . the: specified in published HSI models to: the fisher (Allen 1983),
1987): pileated woodpecker (Schroeder 1982), American redstart

amide: 1994), ovenbird (Roloff 1994), and veery (Sousa 1982). Variables

a,

in

~ if'lclude canopy cover of conifers, deciduous trees, shrubs, ground

. : my total canopy cover; density, height, and diameter of trees, shrubs, and
"'3 1
t: aidsnag, stump, and log density (Table 3).

ublished HSI models do not exist for the red-backed salamander, yellow-rumped

 

H
- idler, ilnd northern flying squirrel, so habitat sampling was based on important habitat

t‘f’

al.:m'butes reported in the literature for these species. Vegetation features linked to red-
baCked salamander abundance include leaf litter depth (Pough et al. 1987, DeGraaf and
Yamasaki 1992), soil pH and moisture, vegetation canopy cover (Heatwole 1962, Wyman
and Hawksley-Iescault 1987), and the availability of cover objects such as decaying logs
and rocks (Jaeger 1980, Pough et al. 1987, Mathis 1989).

Published information on the habitat requirements of the yellow-rumped warbler
is scarce. The Michigan Breeding Bird Atlas survey indicated that the warbler is
common throughout the Upper Peninsula, occurring most often in mesic forests
consisting of mixed northern hardwoods. The yellow-rumped warbler also occurs in
other forested areas, including dry coniferous forests and certain edge habitats (Eastman
@91). Therefore, the variables specified for other wildlife species in this study were
WE inclusive of the characteristics that would be associated with a range of

-. warbler habitat quality.

 

 

a. W5 feast of flying squirrel habitat identified from the literature neat

 

 

    

 

 

ANZ Eo G3 805
wmA Eu wmA AN>V EB mA @8290 as can :32
3% 2E new
as 3 \omA «:3. Eco: NA :8 EA moot mo bison—
AN>V a:
\ooovéooN base 8on mEEmmBEAM
sac cm? 33:55 22% mczmem
Am>v ommfimm fifimaoav .0656 02H.
bombs
E 530 3052099 do
550
$3 eeoaA 30:8 3803qu a»
bombs
ANN/v be v E 550 do :5 e
3: 9e . a. u s
. Go e
c. u so? :38 .558 935m as
A :3
c
:5 am? so? at so? :Z $8-8 dose 388 recess es
A32 9&2 A32
awe 38% 580509 to is bean sauce,
35 55$ so
.0 coxoomuooa £1me :3 cam—MMWMWV Aoocodowomv
00>
Meagan 33:95 55E 36 pecan 565:2 88on
.233

 

a i .
cusps Beading based debs; seen 33:22 5? caduceus 8%; ens 83%.? 338 in 3:235 .n 03%.

22

 

 

 

 

 

 

 

>hldliv> .vld-uwId-na- -U.h-P—-rdx’fl U turd-fi/n.‘ ‘azlflv .vtv-I-wuv. A-.U.J-aU-h’\\ fid...J:JA§.fl
. l . 11.1 - :1] l 1 .11... l||1 .u-lsnlmtdns sit-w-» ass 5.....t.lt.....V~t...u-..-..-.~..tzm.\..v w... s..\.A\A w.
.t’ICIIQIOIO III-Iall-IO Illvltv‘ )Iod\’lul1-ril Ito-ts III.I'I¢I dlvllitlu ltd-IIIItI III-II 'iitlhlvfillItl CIVIilvlai ‘1‘... \Irll‘lllfiliauti u llhl§lifi About! '1 I .l\a§le.‘

 

   

fim>y

 

 

   

             

 

    

   

. fleas m N the??? h u ..H
at Bases .
- 508 05%»: gamma:— mom
. at so 3.8: 25 3.60 a
ah i.
a: vdeA $2 28 3:56 we 56qu
§
A36 «:3: .A Eu wmA madam no 330.“
. nae Ea
m. GE Eu wmA wmA mwaem .«o gov 099.02%
m . .303
H. «GE gmA 38035: HO Emma...— .m>..<
m ._, .. 33‘
.n we Eo.m-n.~ 25:202. mo .5 own—3M
a\
m d «esp—SA ..
. Se 3.. an
. . . - w. ans: v 32 fine A32 some:
.. .5.. . .. . H. . .- , t. “.23.: 8:5 Ex: 8:3 use £556 8829.05 .
. - .. $9.06 3.23% ,. .30 gm 50:05 86on W

    

 

 

.: .i

..1...£

titsisrss

      

Ozumnm- . ,,, . an

 

3'1‘ ", '
‘.!'| 0‘:

“M

t‘---.
‘ A
‘4‘.“
a

..
t. a. .
-- 17.x

i
‘

 

   
 
  
  

 

.. m unimportant for nesting and foraging. respectively

Tim 1"l .. 1

and Gates 1985), and overstory tree species composition (Payne et

. g. d.“ ‘
may), corresponding to variables in the HSI models (Table 3) and the
3st

m vegetation layers in the forest. In addition, conifer cover, deciduous cover, and

6% V. ' cover of hard mast producing trees (e.g., red oak [Quercus rubra], beech) >51 cm

diameter at breast height (dbh) were measured in the overstory. Combined canopy cover

oflhrub species (e. g, serviceberry [Amelanchier spp. ], yew [Taxus spp. ]) was measured

in the midstory, and the percent cover of woody debris and herbaceous vegetation in the

ground layer was recorded (Table 4).

Nested plots were used to obtain densities of forest stand attributes. Density of

each tree species, saplings (defined as trees 2.5-10.2 cm dbh [Bond 1957]), and snags

were measured within 10x50 in plots (Table 4). A snag was defined as a dead or partly

dead tree that had a dbh 210.2 cm and a height 21.8 to (Thomas et al. 1979). The density

“diameter of logs >15 cm wide at their midpoint, and density, height, and diameter of

W >15 cm wide were also measured within 10x50 m plots.

Within a 10x25 m subplot, dbh of individual trees 210.2 cm in diameter were

ml i'l'l «
4. " ‘ g; with adbh tape, and heights of trees were measured with a Haga altimeter

 

  

listen
553.33
P. 3’ V.

w 'b..\t1:\r‘\
‘.
1:111

L
I

"--‘-..!'

s .
' b.
\31‘ 14.
..

,. duthodsusedtomeasarethemintheum

  

  

 

" 1996-1998.

 

  
 

‘l

vi-

lliarnlrllur .

, mfsize class of cover
" a \ lw

   
 
     

~ “It i(‘ll..ttc’- '

. tree species composition

Sampling method
10x50 m plots

10x25 m plots, dbh tape
10x25 m plots, Haga altimeter

f ', ; _ .. 10x50mplots
i? ll; ' 10x50 m plots
.4 10x50 m plots
W 10x50 m plots, Haga altimeter, meter stick
m diameter 10x50 m plots, dbh tape
SL03 density 10x50 m plots
15}; length and diameter 10x50 m plots, meter tape, meter stick
Illinpdensity 10x50 m plots
Sump height and diameter 10x50 m plots, measuring stick
Density of shrubs and seedlings Variable sized plots
Height of shrubs and seedlings Variable sized plots, meter stick
Maceous plant height 5 points along 3 50 m belt transects and at 15
salamander cover boards
litter depth 5 points along 3 50 m belt transects and at 15
salamander cover boards
Canopy cover 20 m line intercepts
Woody debris 20 m line intercepts
:“StiilpH Kelway soil tester at 6 points in each
1?... I .. : . salamander survey grid
Soil moisture Kelway soil tester at 6 points in each
salamander survey grid

3 1x50 m plots in each salamander survey grid

ls.-.. 2,5

 

..h I‘

L... la:

{-s-

I
-1

3’- .Z r
i~,_,t,‘bll.\

O! I ‘
. rt

¢._‘._
0

o-..
f”,

“ 1

a

£33: :1

L.'-» a!
In»?
w 1‘

   
  

 

. » tineasured in plots 10x25 m. Occasionally, slu'ub suns

. . within smaller (2x2 m) plots in areas of exceptionally high

- . 1983). During rainy nights, salamanders typically forage above ground
”1962). During this time, the ground and tree trunks within 1x25 m transects
"searched, and woody debris and rocks were overturned to locate salamanders.
Salamander surveys were completed in 3-6 hardwood stands, located at least 1.6 km
Qlt, in each study site. Although published literature had suggested that red-backed
salamanders are relatively abundant in Michigan hardwood forests (Test and Bingham
1948), no more than 3 salamanders were found among all stands sampled at a site during
rainy night surveys. Therefore, the nighttime search method was discontinued and 2 new
alrvey methods were implemented during the second and third years of data collection.
in 1997 and 1998, 2 salamander survey methods were used concurrently to
identify the most useful method for surveying red-backed salamanders. For 9 of the 12
Ill-dwell forest study areas sampled during the project (3 in the Hiawatha National
m 3in the Lake Superior State Forest, and 3 owned by private timber companies),

”surveys were conducted in 5 stands selected randomly from a larger set of
k -

  
    
   

. 1.. . -. ms sampledfor vegetation, forest birds, and barred owl.

 

. -.-- J Aimsmdy-M'msepamdbyathmlo

 

o "~,\J ‘
“1..-l“ .‘H

3):;3 GL

 

l

f‘ '_‘il Tl

5. ms‘a .
1 u

"r 5')"
\5 .svkti

-"--‘< .
4 :-

v
L ,‘w .J‘.
Ire

. .‘l ‘
"'37 ‘. “C
I.

7 _~« .
RV“ do

 

  
   
   
 
   

-,mdomly selected for salamander surveys at each of the 3 Huron
a ‘ 'areas and were separated by intervals of 0.6-1.2 km.

A '7 ’ r ' fancy methods used was daytime ground searches (Mathis 1989)

‘ . whit] transects. Three 100 m2 transects laid out parallel to each other and

' ' “ ‘20 meters were searched in each stand. Within each transect, all movable

: such as logs and bark were overturned, and the underlying litter was

i ’ .. for salamanders. When possible, ground cover and vegetation disturbed while

Wgere replaced after searching. The amount of time spent searching each

I:lmet was also recorded to standardize survey results to the amount of effort expended.
: The other method used to survey salamanders was the cover board method,

Mind by DeGraaf and Yamasaki (1992), in which wood boards were placed on the

he“ floor to simulate natural woody cover, and observations were made throughout the

field season to find salamanders. Cover boards were untreated pine and were 90 cm long

x20cm wide x 2.5 cm thick. In each stand, 3 rows of boards were placed parallel to the

2x50 in belt transects used for cover object searches. Rows were spaced 20 m apart, and

0th row had 5 boards placed at intervals of 10 m, resulting in a 3x5 grid of cover boards

tWWW/cred approximately the same area as the belt transects (Fig. 2). Boards were

mm on MDNR, Forest Service, Huron Mountain Club, and l timber industry site

hxuly and June 1997, and on the remaining 2 timber industry sites, boards were placed

‘\\P..

 

m 1997. Surveys began in mid-June when salamanders are thought to have

,-

 

. , ' ‘ (J aeger 1979). Salamanders were located by carefully turning over
. RCA—“VIN (t ‘
,r _. ww“mnnderlyingh‘tterforsalaniariders,andafierlookingfor

     

    

 

 

X' COX

.,
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I

 

 

 

 

 

X
X
X
X
X

 

 

 

 

I
I
—-——-———————'—--—-——__——____————_—--—_-—d

X - cover board

 

I - ground transect, 2x50 m

 

 

ys

Figure 2_ Arrangement of cover boards and ground transects used for salamander surve

in Michigan's Upper Peninsula, 1997 and 1998.

28

 

 

.
1
,.,*1!" "

:I '
Lh- at"

D:

3"; (1 1'
s... I.
o

 

r7.
LIES}

T ;_r

Cr

 

salamanders, boards were put back into place.

During spring and summer salamander surveys, litter depth and height of
vegetation below 50 cm were measured at every cover board and every 10 m along
salamander search transects, for a total of 30 litter and vegetation height measurements
per stand (Table 4). Soil moisture and pH were measured at 6 points in each stand with a
Kelway Soil Tester (Kel Instruments Co., Wyckoff, NJ). During ground transect
searches, the size and type of cover objects turned over were recorded, and the presence
or absence of salamanders beneath each object was noted. Size of cover objects was
recorded in width classes of 0-5 cm, 5-10 cm, 10-20 cm, 20-30 cm, 30-40 cm, and >40
cm to determine if particular sizes of cover objects were used more often by salamanders
on some sites than on others, and cover objects were classified as wood, bark, rock, or
other (Table 4). Observations of herptile species other than red-backed salamanders were
also recorded during surveys.

Cover board searches were conducted concurrently with ground transect searches.
Both salamander survey methods were conducted once in each of the 54 selected forest
stands during the period from June through early August. In September and October,
1997, before salamanders went below ground for the winter, ground transect searches and
cover board searches were repeated in 20 of the stands that had been searched that
summer. The purpose of these surveys was to assess seasonal population trends and
obtain a population index before salamanders went below ground for the winter. In fall,
1998, cover board searches were conducted in every stand that had been sampled during

summer. After the second cover board survey at the Huron Mountain Club in August,

29

q' " . I
‘1‘, .1: fix

I
,3
but

..

:r-w t
V

~ I. -
b

 

.I".

1998, the boards were picked up from the sites and put into storage. Cover boards on all
other sites were left in place after fall surveys so that researchers may start compiling a
long—term data set on the relative abundance of red-backed salamanders in relation to
habitat characteristics.

Forest bird species

Relative abundance of all forest birds, including the American redstart, ovenbird,
veery, yellow-rumped warbler, and pileated woodpecker (i.e., the indicator species), was
determined with point count surveys (Whitcomb et al. 1981) in 1996, 1997, and 1998 in
all stands where vegetation sampling occurred. Each sampling point was surveyed once
during May and June, during territory establishment and breeding, and once in July, after
fledging. Surveying began at dawn and continued for 3 hours after dawn (Robbins
l98la). Bird surveys consisted of a II-minute settling period followed by a lO—minute
sampling period. The length of time used for each point count was determined from
preliminary surveys, during which the number of species observed was plotted against
time, and the point at which the number of new species began to level off was used as the
count period. The species, gender, and relative location of all birds heard and seen from
the survey station was recorded. To maintain fairly standard sampling conditions,
surveys were not conducted in weather conditions such as rain, fog, winds over 20 kph
(Robbins 1981b), or when water was dripping from the trees due to a recent rain. If
possible, surveys conducted when there was >70% cloud cover were repeated when cloud
cover was <70%, to minimize the inhibitory effects that cloud cover can have on bird

vocalization (D. Beyer, Michigan Department of Natural Resources, pers. commun.)

3O

.I'or

In I

‘1”:ng

n-

u 07-.)
... .I u.\
V'r' ? s

I. if it

\-»..'t

\%.;~U.
Vt.“ 'r
JA'UII. I

If: 53. II]

'A
en” :

\ "f its.

BI

i", ‘
T
o “:16

'I-‘an ;,
“t.

Northern flying squirrels

In 1996, during vegetation sampling, all snags within vegetation sampling plots
were hit with a stick in an attempt to flush and observe flying squirrels that might have
been in the tree. Because flying squirrels are nocturnal, they are expected to stay within
their nest cavities during the day, and usually flee their nest cavity when their tree is
disturbed, for instance by banging on the trunk of the tree (Sonenshine et al. 1973).
However, no flying squirrels were observed with this method, so in 1997 larger transects
were set up specifically to locate snags and observe flying squirrels that might be
inhabiting the snags. Observers walked together through the forest stand, covering an
area approximately 30 m wide and 200 m long, and knocked on each snag in the transect
with a large stick to prompt flying squirrels inhabiting the snags to emerge. Again, no
flying squirrels were located, either during formal sampling or during other field
activities. Other potential methods for surveying northern flying squirrels, such as track
plates (Carey and Witt 1991) or live trapping (Rosenberg and Anthony 1993) were not
attempted because the amount of time they required would have prevented collection of
other data needed for this project.

Barred owls

Taped playback surveys (McGarigal and Fraser 1985) were used to determine the
relative abundance of barred owls throughout all study sites. Barred owl surveys were
conducted once during July or August in 1996, 1997, and 1998. Surveys began after dusk
and were completed by 0500. Barred owl survey stations were established at each point

used for songbird surveys, spaced 1.6 km apart on MDNR, Forest Service, and timber

31

f.
IT». I.
,_ .mlu
r
"."J
L.urb

:2: lif‘lc

 

 

industry land, and 0.6 km apart on Huron Mountain Club sites. At each survey point, a
taped barred owl call was played through an amplifier for 2 3-minute periods, with 1
minute between each period. During the tape and for 10 minutes afterwards the number
of barred owl responses to the taped call were recorded. The approximate distance and
direction of responses were also estimated to determine if the same barred owl had been
detected more than once. Stands in which barred owls were surveyed were not surveyed
for songbirds the following morning.

Fisher

Fishers were surveyed by counting the frequency of fisher tracks in the snow
(Powell 1994) during February and March of 1997 and 1998 on MDNR, Forest Service,
and timber industry study sites, and in 1999 on I timber industry site and 1 MDNR site.
Observers walked along transects through each study site and recorded the number and
location of fisher tracks encountered. Between 2 and 4 transects, 1.6 km in length and
Spaced approximately 1.6 km apart, were used at each study site. Exact placement of
transects depended on winter road accessibility in each study site.

Track count data were used to obtain an index of fisher use of study sites and to
describe the intensity of activity on each site. The number of tracks per transect was
divided by the total length of the transect to obtain an index of fisher activity in units of
tracks/km for each study site. This approach was based on the assumption that an
increase in the number of tracks per unit distance, even if left by one animal, corresponds

to an increase in the suitability of the surrounding habitat for meeting that animal’s life

requisites.

32

Bat

1|. .J'Lflb

,'
Si-grp
7.‘ in.»

film:

E775 SUI

‘..
b m¢~

‘EKI:

 

Because accessibility to the Huron Mountain Club is limited during winter, fisher
track counts were not performed. Instead, scent post stations (Linhart and Knowlton
1975) were used during the summers of 1997 and 1998 as an alternate method for
determining fisher relative abundance in stands at the Huron Mountain Club. Scent
stations consisted of a circle of ground 1 m in diameter, cleared of vegetation, with a
cotton swab with fermented egg and cod liver oil placed in the center. Placement of scent
post stations followed the pattern of bird survey point locations, with stations located at
approximately 600 m intervals in hardwood forest stands at the Huron Mountain Club.
Each scent post station was checked for 3 consecutive days each in June and July.
Habitat modeling

For species without an existing HSI model (red-backed salamander, northern
flying squirrel, and yellow-rumped warbler), preliminary models were developed using
published information on habitat requirements in Michigan or in comparable habitats and
geographic areas. The preliminary models guided field data collection, and analyses of
ecological attributes measured in this study were used to finalize the red-backed
salamander and yellow-rumped warbler models. For each variable included in the model,
values thought to be indicative of high and low quality habitat conditions were used as the
basis for determining an index of habitat quality for a sampled area.

For the red-backed salamander model, 35 vegetation and structural attributes were
analyzed as independent variables in a multiple regression analysis of 30 of the 54 stands
where salamanders were surveyed. The number of salamanders found per stand during

Summer cover board and ground transect surveys, averaged over all years of data

33

I
«209,1 '
. 42.» .

{3815 h

 

I.

I

.. 'l‘

I‘ ‘

[1.1
raJ

\ r

a

(F. ..
"'9‘ HI

u‘:, 5.5
it“.

is“
‘L.\\
. .

a. .. ‘
Ml

 

ti id

collection, was used as the dependent variable. Initially, the analysis was performed on
data from 30 stands, randomly selected from the 54 stands where salamander surveys
occurred, to preserve some data to use later for testing the validity of the model. The
analysis began with forward and stepwise variable selection (Sokal and Rohlf 1981), and
through this process, several variables that described little variability in the data (partial
R1<0.05) were eliminated. Based on the regression results using 30 data points, an
intermediate model was developed. This model was tested on the remaining 24 data
points, and the results were used to further evaluate and refine the salamander model.

Testing of the regression model against remaining data resulted in a poor model fit
(R2=0.20), suggesting that the model would not accurately predict salamander habitat
quality. The next approach to improving the initial model was to divide data from the 54
stands into 2 groups. Stands with an average of $2 salamanders found during the
summer, under boards and on transects combined, were considered to represent areas of
unsuitable or very poor quality habitat, while the remaining stands represented varying
degrees of better quality habitat, with the highest quality habitat occurring in stands where
the maximum number of salamanders had been recorded. An independent t-test (Ott
1988) was used to identify habitat variables that differed significantly between the 2
groups. Final model variables were selected based on statistical results from field data
(regression analysis and t-tests) and published research on red-backed salamander habitat
use (Appendix B).

Field data for the yellow-rumped warbler were more limited than for the red-

backed salamander, so model development differed slightly from that of the salamander.

34

Tim .7
rep. :'
I,. ,,
Ti i‘i'll';
T‘p‘l ".

Hub |~I-

al.;Ii U

I
‘." “1;: n
numb Il

"\b,

1". 't 7‘7
""Ir:l_ |
.

is in

 

 

 

Potential model variables were first identified from a thorough literature review. In the
next step, independent paired t-tests were used on field data to compare stands where
yellow-rumped warblers were observed during data collection with stands where yellow-
rumped warblers were not observed. These results were then used to corroborate and
associate quantitative values with an index of habitat quality for each variable identified
from the literature (Appendix C).
No northern flying squirrels were observed during field sampling, so population

data were not available to use in model development. Instead, the model was derived
from an intensive literature review (Appendix D).

Data analysis

The parametric assumption of normally distributed data was tested for each
vegetation and soil attribute with the Shapiro-Wilk test. Many of the variables tested
were nonnorrnal, so appropriate nonparametric tests were used for univariate data
analyses. A limitation for conducting data analyses was the was the lack of replication, or
pseudo-replication, of Huron Mountain Club study sites. While this may affect the
strength of some conclusions about wildlife populations and habitat conditions at the
Huron Mountain Club, the data that were collected provide an important reference point

for the range of conditions that may occur in northern hardwood forests.
Comparisons of vegetation variables and species relative abundances among land
ownerships and among sampling periods (years for all species, and seasons within a year
for bird data) were made with the Kruskal-Wallis one-way analysis of variance or the

Wilcoxon-Mann-Whitney test (Siegel and Castellan 1988). Significant differences

35

s a w
I7 .

a.- U
‘ I
Jorge
...'7 .sl‘r
fiat-y .-

L‘.’ "‘EI

 

AK

 

 

 

 

43.6?

 

 

 

(p_<_0. 10) detected with the Kruskal-Wallis test were further analyzed with the Kruskal-
Wallis multiple comparison statistic (Siegel and Castellan 1988) to determine which pairs
of ownerships were different. Songbird comparisons were based on the pr0portion of
sampling points at which a species was detected.

Songbird data were compared across the 3 years of this study for each ownership
category to determine if it would be appropriate to combine the data for further analysis.
In this analysis, species which never occurred more than once in any of 6 possible
sampling periods (spring and summer, 1996, 1997, and 1998) on any of the sites within
an ownership were omitted from comparisons among years. The overall management
approaches of each ownership remained constant during the study, so differences in
species relative abundances among years were assumed to be related to local
environmental conditions, rather than treatment effects. Therefore, data collected in
multiple years were combined by calculating yearly values for each site, then averaging
yearly values among sites within each ownership to obtain a value for each ownership by
year, and finally by averaging values for all years of data collection within each
ownership. Principal components analysis (PCA) (Morrison 1990) was also used to
obtain a graphical representation and quantitative description of the multivariate
relationships among vegetation variables in all 4 ownerships.

Patterns in bird species communities among ownerships were investigated by

conducting PCA on a subset of all bird species that occurred at an average relative
frequency >20% on any ownership. Additionally, forest birds with similar life history

Characteristics (e.g., cavity nesters, year round residents) were grouped together and a

36

 

 

KruskaI-Wallis one-way analysis of variance was used to test for differences among
ownerships in the relative abundance of each group.

At each study site, an index of relative habitat quality was calculated using

existing HSI models for the ovenbird, veery, American redstart, pileated woodpecker,
barred owl, and fisher, and recently developed models for the red-backed salamander,
northern flying squirrel, and yellow-rumped warbler. For the ovenbird, veery, American
redstart, pileated woodpecker, barred owl, red-backed salamander, yellow-ramped
warbler, and northern flying squirrel, model outputs between 0 and 1.0 were calculated
from data collected at each stand survey point. For comparisons among ownerships, HSI
values were averaged across stands to obtain a value for the entire study site. Based on
the typical home range size of adult fishers, it was assumed for each study site that the
entire area was included in the home range of any fishers whose tracks were observed on
the site. Therefore, fisher habitat suitability indices were calculated by first averaging
data for vegetation sampling locations across the whole study site, and using the average
values as the model input, resulting in a single HSI value for each site. Comparisons of
relative habitat quality (HSI values) among ownerships were made with the Kruskal-
Wallis one-way analysis of variance (Siegel and Castellan 1988).

Spearman-rank correlations (Siegel and Castellan 1988) between species average
relative abundances and HSI values for all study sites were analyzed to test the validity of
the models that are available for the ovenbird, American redstart, veery, pileated
Woodpecker, barred owl, and fisher, and for the red-backed salamander and yellow-

l'umped warbler models developed for this project.

37

Ir.h.bd“
r .“.

“-7 .,

1‘, ‘.‘*I

l
T;~,J';r
link,“ \
9

lane;

 

 

RESULTS
Vegetation and structural attributes

Vegetation variables were measured in 104 hardwood forest stands, and the
majority of vegetation sampling was done in July and August of 1996 and 1998. Tests of
univariate normality suggested a nonnorrnal data distribution for 12 of 27 vegetation
variables, so nonparametric tests were used for univariate analyses. On the forest floor,
differences among ownerships were found in the average height of ground vegetation
(p=0.034), the size of logs measured (length, p=0.075; width, p=0.052), and the average
diameter of tree stumps (p=0.053) (Table 5). Sites managed by the MDNR had taller
herbaceous plants than the other 3 land ownership categories sampled, which corresponds
to the significantly greater amount of vertical cover of herbaceous vegetation also
measured on MDNR sites. Logs on MDNR and Forest Service sites were significantly
smaller in width (p=0.052) than logs measured on other ownerships, and the largest
diameter logs were at the Huron Mountain Club. Sites on timber industry and Forest
Service land had the greatest density (p=0.041) of shrubs (defined as woody species 50
cm - 5 m tall with a dbh <10.2 cm), and on timber industry sites there was more vertical
cover in the shrub layer of the forest (50 cm - 5 m) (0.038), while forests at the Huron
Mountain Club have relatively few shrubs, resulting in a much more open understory.
Approximately 98% of all stems classified structurally as shrubs were tree species: the
remainder were shrub species.

The diameter and basal area of overstory trees was significantly different among

the 4 ownership categories, with stands at the Huron Mountain Club having the largest

38

 

.Ifivfivu uo-~.. .\\w\o~ .-V~.\v~ .~/~uu1-/\ hun~uw>--h ...~.-J....n..r‘\-
nuns-nu- . Irv-n... -A- I v 1’ -I,-u-.u-- sides-n... .A .’.,.§r.~ a v uv.d.-.I ovuTv. ~I..i..-A '3‘ .1]. x h .A UVHZK ih.<v
v .1...IIAII1‘ It...’ Inlet-Ivuair‘rvtl I .r Iv§o§ei ~

AIL-Lilli! Illicit-OI‘OAI IIIIIV

lute-FI— .- 1.--lamina- quZ I... avg-:4. A. Viz-I v scan-i V logo‘s-unuavliz
tr-.lOIIIo-uri'l IOIIOlUIn’II III I!|.IIIIIIOIII.UI I Illuu‘Iol 10"; our! OIIIIelIil-.l-II llltlIrIlIl .'7IIOIII.IIIIUIE\

owed no.0 .39.: mod Em... C. d 93.... No.0 aN—é *8...on £35 2...

 

 

 

«we... 2... a2 .2. .8... 2.... as; a... .3... 3082....
.8... 8..“ as... ....m .83 Se .83 e3 3:" .e and
N2... 3. «on. we. 3.... 3m 83 mom 3.8 e 3...
€er ..o>8 1.3.282
5.2. cm... 8.... 3o :3. .3 cm... 2 .0 can es. 5%.. as:
..8.o no. .2... B... .32 cod .5... 8.. .32 Lee Ewe; .8835:
Sc... 8.. an: we... .83 Ed .32 a... .32 Lee ones... .3
m8... moo .23 $5 .8... moo .8... mac .23 .25 525. we.
o8... mm... 8.2.. 2.; «3.3 3.2 .32 2.2.. 23.2 223
m8... .3 .25... 2% mama an. 2.3 9.... an: .25. sees... 95%
$3 9% 2.48 a... 3% cm... 23.. 2.. :3... as. Ewe; seam
32. Sam m3... 3. 2...: 2.2 was: .3. 8...... «£283
:3 2.2 3.3 8.. «3.3 8.: 3.2. 8... 8:8. as. Ems: seam
.3... Se? .38.... 3%.... 3.5.6. £33 3.25.2 3.8.. .32»... .383£
Sod 2...... .22... 8.: 233.. .32 games... 2...: 3.8.. .seasamm
.m.m .x. .m.m m. .m.m w .m.m M czar;
Lo>o_
$338.. an... ufld - - sue: 3.52.3 an... M75...
bowwwo 3:20:30 II

 

.32 as .32 .82 Jana... as b... .2355.
3&3 9525.2 5 28. 625 230 £8552 85: Ea db cases .85.. .Ammmav bosom neon. .m .3 .Ezeé
wooed—03M H.532 «0 Segmen— SwwEomE :o .3532: Quote Rocco; was 2825 mo~cn§> noufiumo> boemuoccb .n Each.

39

 

 

 

.xK-m- .u...-... u\\.\. m.|u..\~.\..m .~l.:..4:/\ u...-w .Nus:~......~..-ul.-..—\.~.u..§I.|~..l.vlaxfij
l:--«.-..nln-..J-F\Ih a... uvanau- A: VFz-lv A----. V unna-nncnanvvc .unviI-u- tau-aw .A..-.v ).usluuui-ui .u-VA~.--— .AY..‘MJ.. u v :v.v- I~-i.fv. ~I.d~A-..§ .11. x u .A inA
u

IIIO-Ill-z .III Ill:lnllilldn..1'l Illdu.§il- Il-t.‘ III! IOIIQIIII,I|IIIII AtlIDI.I.IOI .IIIIIII-IIUYI III-l! iL-IlilI-IIV ’..I\I~IQDIII.’ IIII‘IIVCIIV‘II.I

 

h.‘ V 7:i-.d,ih~Av/cvhr~

.iifilvhl..fivlii-n~ nh-hl 0. .IV #1 -.I§A\II.‘.

.waa .5.—230 E...

.umo_m_ 03.283 5.59:8 032:8 m___a3-_§mav5 8— .OAE 2.283. .0: o... .8...— oEmm on. 5.3 2.: 06mm on. so 325?.

.Awwa 5:830 E... 336. 852...; .8 22:28 2.3-2.0 £33-...me on. 5.2. 8.2323 803 “.353. 232 bzfimnoi .

 

 

 

 

 

 

 

 

.2. 213.... a... #49 4.3%
.306on
So... So .8... 8... .8... 2.. 2...: 8... .2... pa... 3828a
GE .38 Hunks
.m.m .w. .m.m .m .m.m M .m.m w 032.;
:96.
523305 an... 02: GHE: Gnu... mmmD M Q
’ homage $58550

 

.32 Ba .82 .32 .22.... .5.. :3 .2355.— 8&3

PEWMEQSZ E US: A023. 920 55.502 :83. US. .95 25.0.3.5 .38.. A953. 335m 520m .m .D .934sz 38.83%
H.532 .0 30.55300 Gamma—22 :o nouamsoE 3.8.8 2.3.5.8 US. 2325 moEa§> cargo»? .COBBUGD .950. m. 2an

40

 

 

A6-v

' 1 J
1L. .ul‘

. .
., .mu.

v.

.. r" .

.. .
Wadi”

hi

"\‘““IMQ'

N. ‘
~11"?! 1.
' w“:

r.

f;-
.

1. .
p"
‘K pll

"f: .
:‘wfijfl

 

 

dbh (p=0.057) and basal area (p=0.031) of the stands sampled (Table 6). The Huron
Mountain Club also had the greatest amount of vertical cover of conifer species (p=0.044)
and the least canopy cover of deciduous tree species (p=0.063), due to the large hemlock
component and near absence of beech in forests at the Huron Mountain Club. The
proportion of hard mast producing trees, primarily beech, >25.4 cm dbh was greatest
(p=0.029) on timber industry sites (Table 6).

Principal components analysis

Conifer cover was one variable that exhibited a very nonnorrnal, skewed
distribution, with especially high values in stands at the Huron Mountain Club (Table 6).
This variable was arcsine transformed to more closely meet the assumption of
multivariate normality associated with PCA, and the analysis was run on both
transformed and untransformed data. Results of PCA on the untransformed data
corresponded most closely with the ecological relationships (e.g., differences in canopy
cover) that were evident in nonparametric tests and univariate statistical comparisons, and
only these results are included here.

The first 3 principal components (PC’s) of the analysis for 27 overstory and
understory vegetation variables accounted for 66% of the variability in the data set, and
With the fourth principal component, 76% of the variability was described. The first
Principal component explained 37% of the variance and describes a contrast between
coniferous and deciduous overstory cover, and also between basal area and understory
Characteristics such as shrub stem densities, rnidstory canopy cover, and herbaceous

Vegetation height (Fig. 3). Specifically, stands at one end of the gradient had more

41

nIlil‘l I

n-uuwunu-nnv—x‘ .ugu-.-~ ua--.‘ .A~,~.» >..-/,-.:- uiJA~.--.
Iii-q-U-hlwui .U-ul.‘ wilt-«Id-nhv ’1J\I§Q.¢I-i’

.22.... . 1.312391?)— ... .13... A. we): _ . $2.. v 47.176. ~. 3-1)..-.7. .I.U.:..~ Hf. .. u .« v.7: -.<.
Vbtl Institutvdvl .liI-IIIIZ in! .II!!!IIIIC~I.IIU' I Illa-.‘nl- I...)- nun' I‘llnlovllildulu All-niuhtl nhnvi§o.o-vvul\ld.’ \’ in-‘l..ri..’~;~ .|\ lV‘A‘Ufi ~.

.232 02.0.80 0::
.035. 002.08 080000.00 0.02:0. 850328—333 8213 E00003. .0: 0:0 .0..0_ 0.0.8 0:. 5.3 0:: 00.0.. 00. :0 320?.

.83. 5.20.2.0 00.. .0w0mmv 00:00.; .0 22:00.. 83-000 2:03-...me 0:. 5.3 00.00.20... 0.03 00.0000. 20.301.50.30... .

 

 

 

 

.000 .8. v.8.
88.. S... 3...... 8.. .8... 8... 20m 2 .. .8." 8 2A .08. .88 8....
8.... 88 .88 8.0 28.2. 0.0 .25. SN .38 .88. 828.80
3...... 88 .83... Ed 2:... 8.. 8.8 80 .88 .88. .8800
2m... 0... 8.8 3.... 8.8 8.. 2.... 5.. 2...... 8 mA
88. .260 82.00.,
.8... 0. .m 3.8 8.. 8.: 8... 8.8 88 8.2 .8... 8.088.. 0.8m
8m... 9... 8.... 2 ... 0...: 8.. 0...... 8... 2.... .8. 80...... 08m
E... 8... 8.... 3% 2.8 8.. 8.8 8.0. 8.2. 8.888
m8... 8... 8.: 8.. 8.2 8... 8.8 8... 8.8 .8. 80...... 8.. 02.8.0
Re... 58 .38 8... 28.8 8.. .88 8.. 28.8 28.. 000 8.. 09830
08... 8.8 8.2.. 8.8 8.8.. 8.8 8.3% 8...» 8.28 .888. 02830
.8... 8.0 .88.. 0... ......~ 80 .38 and .8..." 2.8.8. 8.... 880
.85. .08 w .08 w .08 w .08 .... 088.03:
8.838.. G"... 080' an... .0 an... 83 an... :20:
808.00 8:80:30 ll

 

.2... .82 88 .82 .82 .8808... 08. 0.0 .8888...
0008. D $002000). 0. 0:...— AUEEV 020 08.0002 00...: 000 .95 b.0000. 0000... A953. 00_>..0m .m000m .m .D .22sz
000M 3.0.02 90 50.0.8009 SEEQS). 00 0000000.: $00000 0.000030 000 00000.. 330.00., 000.30%? b0.m00>0 .0 050%

42

 

V
I
-

 

 

 

 

Snag
density TI
4 .
TI
2 a
N
g HMC USFS TI
0i
v v HMC USFS
HMC USFS MDNR
Log -2 ‘ MDNR
density MDNR
-4 '— r fl I I 1 fi
.8 -6 -4 -2 0 2 4 6
PC 1

Conifer canopy cover
Basal area

Deciduous canopy cover
Midstory canopy cover
Herbaceous vegetation height

Figure 3. Scores for the first 2 principal components (PC) for vegetation variables

 

measured on study sites on Michigan Department of Natural Resources (MDNR), U. S.
Forest Service (USFS), timber industry (T1), and Huron Mountain Club (HMC) land in
Michigan's Upper Peninsula, 1996, 1997, and 1998.

43

M ,4 ,.
.t...'t .

Y
”,1

...
"w‘ m
v‘i“¥1

 

 

 

 

coniferous overstory cover, higher basal area, and relatively little midstory canopy cover,
while stands at the opposite end of the range had more deciduous overstory cover, lower
basal area, more midstory cover, and taller herbaceous vegetation. Sixteen percent of the
variance was explained by PC2, which represents a gradient between stands with
relatively larger numbers of fallen or cut logs and stands with higher snag densities. The
third principal component accounted for 13% of the variability and describes a contrast
between a combination of overstory canopy cover and overstory stem densities, and a
combination of litter depth, shrub stem density, and ground cover.

Scores for PC] were negative for all 3 Huron Mountain Club sites and positive for
all other sites, reflecting the relatively high proportion of conifer cover and low
pr0p0rtion of deciduous cover in the overstory, along with the greater tree volume and
Open midstory on Huron Mountain Club sites. These characteristics distinguished Huron
Mountain Club sites from forests managed by the MDNR, Forest Service, and private
timber companies, which had higher proportions of deciduous canopy cover, denser
midstories, and taller ground vegetation. Among sites on MDNR, Forest Service, and
timber industry land, scores for PC] varied substantially, and these 3 ownership
categories are not graphically distinguishable along the first principal component axis
(Fig. 3).

Michigan Department of Natural Resources sites and timber industry sites
occupied opposite positions along the gradient described by PC2. This relationship
Corresponds to the relatively high number of logs and low number of snags on timber

industry sites, and the high snag densities and somewhat lower log densities on MDNR

44

”A...“ a

 

sites (Table 6). Huron Mountain Club and Forest Service sites fell in the middle of this
gradient, indicating intermediate ratios of logs to snags on these sites (Fig. 3).

Overstory tree species composition

Eighteen overstory tree species occurred in vegetation sampling plots across the 4
ownerships (Table 7). Sugar maple was the most abundant species in the overstory of
MDNR and Forest Service sites and accounted for 72% and 71%, respectively, of trees
present on those sites. On timber industry sites, the dominant tree species was beech,
which accounted for 40% of the trees sampled. Sugar maple and red maple together
made up 45% of the overstory trees on these sites. Hemlock was the most prevalent tree
Species at the Huron Mountain Club (49%) and was significantly more abundant
(p=0.089) there than on other sites. Of the 18 tree species identified during vegetation
sampling, only 2, beech and balsam fir, did not occur in sampling plots at the Huron
Mountain Club. Several tree species seen at the Huron Mountain Club, including
bigtooth aspen (Populus grandidentata), red oak, striped maple (Acer pensylvanicum),

and white ash (Fraxinus americana), were not recorded from sampling plots in any of the

3 other ownership categories (Table 7).

45

.ui-ai-a< nap-am .A—-a—i .-4—.ari-_g_.ua.— _.Jn—A—_. d f...-_-«....w_——.J.—../~ ~.- «4-_...— .. vF/L— — \ t-..—. \ .....c ..... Q I n u . I
.—.v ~:-:-......a—.JA.- C.....I~.....J.—\lk .uav A.«.—\l..-.J~I.v -av..~_~lnvf~..—A-.J I..-J.~-J.JA~I. .J.J.~— \fisa v.1...~.d>fiv .N Ho~a-flkt

ultJ-nf. .74 .—d .AV-ZA-sz I.-J.J-~.«I.JV- -«.---az

 

wwod

av

:35 azoxmzmu

 

 

 

 

 

0 .6 o a: o a: .38 won 505.82
nwuzmnusoe swans
owed mm amcm w _ Ann 0 sva a a: $3280:
3395. 233
R3 _ N _ H o c m m 2a 2:3 825
32.222. 352%
Red _v e a a 2 mm .1 a €23 seam
oBEeENEEw aéxmom
wwod _ _ .2 o a o a o 3 Evan soona
85343 83<
ccmd o c v a o c a. a E Sam—mm
eteotese $225
$06 _ _ em 0 a: N zN m cm *EB 528:2
cmobhtcgm gunk
wad o E M: m3 2 an.» v :5 283 cmotoE<
ensuing» ERR
acmd N. 3 o c o c o a 98333 cautoE<
. A25»: ombEflomv
L92 .m.m M .m.m M .m.m M . .m.m M 860% out.
55an05 GME USE SHE E, AmuE mmmD AmuE mZQE
bowofio 9:30:30

 

.3595. one b3 .Esmficom .595 MammEoF/H :
320m .m .D .22ng 38:80M 3532 «o E

.waa Ea flog dom—

_ 9:: 6ng £20 55552 :05: cam .95 >536“: 528: A958 835m
oEthoQ 53:22 so QEmEBmV 22:89:00 86on on: bogmno>0 N 035.

46

>._:—- .:_:l—.:..JA— ....::_ — J..............!¢ -

._ d .A V-ZA u: v lav.J.-.unvf-JV— _-.w.-4:~Z .uav

I'd-P.-

..unn—QAfi p..m.t.w...--J.-—\I~ p-AV A...:\I..:

.J-IV n:u..:lnvr—::V-J 7...»..-derul. Unfin- >.::J..as\va \

- IIII\V

.\\

KI.

9.53% 98$

 

 

ammo _ _ o 2 o e m m 82% as?
Embgzqum £38m

omno o c o e a. a o c :23 8:5
casings 225.3%

83 m m o e o c o c as 223
Eakfi~$9u§h kwqu

owed mm .3“ mm ea : at“ E 2:. base swam
Ezomceéagmm gmo<

m _ 9o N .e o a o a o a name easm
23.55 amo<

Eco me E. : 3 3.. an R am case 8m
8323:»: “3.30%

58 _ _ o a m m N a in SE8
385380 35$

Sod 2 a: v a. o s: o a £88 223 50552
- A253 ombEmwomv

L32 .m.m M .m.m M .m.m M .m.m M 360% out.

E5895 dulédfldl Jane E. u: : jg

 

iowofio 9:80:30

 

.wmg can Sag 6mg szwz< use

33. .Esmcmcom SASS Pcwwfiom—Z 5 6:2 AUSEV n20 cfiacnog :05: can .95 >539: .555: .AmmmDv 838m 880m .m
.D 5&2sz 88.53% ~38de mo €08than :meoaz :o AmfimEBmv :oEmomEoo 860% 00.: @9330 .95an h 2an

47

.A_:—- .3.../::..J.— 5.4:»: - 1..-.33....i-xt u... . ..... . .. \u a. .
livald.h~ ).hA Uh/th1.u..JA ~ Ann-lir\\ \I. It‘li‘.olt

.upv ...-¢p:-.:..p--JA — :.......«.:--4..-\/~ :n. A...-\.l,--.J~.fv :av..s.-I..XL::7J I.u...v.JA~l

pd .AV-Zh-FZV I.-J.J---Avrcndy- —-w-au—-az

.239 :2230 :5
Low2m: 0:233 523,288 2922: mEaBLEmemv 8— .OAB 2:22:20 5: 2:. 5:2 2:8 0:: ::3 0:: 05mm 0:: :o mos—Ex...

Ammo: 52.2me :5 .ow2mv 852:;
:0 29:2: $3.28 2:: 3-136% 0:: ::3 2022323 803 32:20:30 w:oEa 80:82:: Lo 28: :8 232 925305 a

 

 

 

 

mmcd 52 go 3 mwv ow «mm mm mm 3:588 86wa =<
. 3.2253323 35mm
Good n gem v ac m 3 m as .283 30:0?
$28: o5329mv
Lo>2 .m.m um .m.m M .m.m m. .m.m M 360% 08L.
5:338: dlsdfldlu sue :. Ialfinflflnlu jg]; :
homage 3:95:30

 

.32 as... .32 .32 semi e5
:3 .2355: can: mamas: E 22 62$ 95 59:82 8.5: as 5: bases 25 .Ammma 8Eom use: .0.

.D .322sz mop—sewed 2:532 00 EoEtwLoQ :awEoG: :o 352:0qu :oEmOLEOU 360% on: @9230 0:300 h 22$.

48

Wildlif

Mt

fem

. 1.
MRI

MICE:

 

Wildlife population survey results

Red-backed salamanders

During June, July, and August, 1997, an average of 3.8 salamanders per stand
were located through ground searches on MDNR sites, 2.8 salamanders per stand were
found on the Forest Service sites, an average of 1.6 were found in each of the 5 stands
sampled at the Mead site, and 4.2 salamanders were found per stand at the Huron
Mountain Club (Table 8). These numbers correspond to densities of 127 salamanders/ha
on MDNR sites, 93 on Forest Service sites, 53 on timber industry land, and 140
salamanders/ha at the Huron Mountain Club, based on the 300 m2 transect area used in
each stand. In addition to red-backed salamanders, 3 central newts (Notophthalmus
viridescens louisianensis) and 1 blue spotted salamander (Ambystoma laterale) were also
found during ground searches. These results were very different from 1996 results when

nighttime ground searches were used to survey salamanders, and salamanders were only

rarely detected.

In summer 1998, an average of 144 salamanders/ha were located through ground
searches on MDNR sites, 115/ha were found on Forest Service sites, 76/ha were found in

stands managed by Mead Co. and Shelter Bay Forests, and an average of 189

salamanders/ha were found in stands at the Huron Mountain Club (Table 9). Two central
newts (Notophthalmus viridescens louisianensis), 2 blue spotted salamanders

(Ambystoma laterale), and 1 spotted salamander (Ambystoma maculatum) were also

found during ground searches in 1998.

49

.h.4»..—..— .Af..._f._~. 3.1),...7. .J..La...- .J _- ..-.lI.I4. - .
.-cU—4P-d —v--Pv— fund—dauawa--m_-w’ uAU .uadfa-nn-AP- «dun-.0 .viw—nhp—HUh-wruvf.udflu.dn1.~flfl.uav .J----- .hLUAh ud-chyfi 1..-.v‘v-~w-—-w-w.l ,‘Av .~idn\--~—.~ --<-~n~.uv-unu.it §uF--I\v-Ii
ua-UP-ol. lulu-nu .IrJa-.i-wtvml a..v.Jr4.a.-<- nun-ndnviriw nun-.3 afflunvhtulv con-«nun-uwul .un~a~ rtn-aan-n-V IV,-.d~.drv-— nil-eh ‘va\f,..iU\.r.-~¢l I.hti§U-h-N~h~a.¢§-w’. .‘av .hilv?---Z .urrv tUNlfiuwfi

.2022 52005 08
0022 200% 0.80858 29:20 203-3835 2: .000 20506 6: ea 5:2 2:3 06 SE 0:: 2:: 2: 8 82;...

Awwfl 00:03.00 000 _0w0_mv 0000::
00 2220000 0903.000 2:03-00me 05 003 00020200 0003 8220030 mcoEm 0000000006 00 2000 000 m_0>0_ 5250005 2

0000.00.00 00 0300 00000 080030 00 00000005 ”002053 0:0 2 300..

LEE 838

 

 

 

080 E 3.8 .30 mm .3 3 .2 5,8 595 085003 % 0o :82
2.2 80 mi. 8... 5o 2:. 8.0 2.0 8508 230058 085508
«3.0 3 as 0.2. 3 0: a... as use. 08 3603 $352
09.0 0.2 2.: 0mm 08 «:8 0.8 $2 000 0828 8 0850.03 060
82 0o 0... 3. 0o 3. 3. .3 05 9x
83 mm 2 3. ed a... Qo 30 .5 210m
800 E E. S 8 M... mm 3 :5 8.8
was 2 an: .3. 3 ES 3 ER *5 8-2
88 03 E... a..." 02 Se 02 0.3 :5 20
ES 0 G 0...» SN E 0.8 02 EN =5 no

0N7; 50.30 00>OQ E m0£080w vcsouw mat—.6 6:509 «Emhuficmrcflam
02,2 .00 w w .00 w .00 w 2%me
005320 3| 0",: .F sue mmlmdl

 

 

 

>8ng 9020030

 

.39 Jm=w0< 000 33. .003 £30500; 00003 0.0020002 E 000— 635v 030 0030002 0000: 000 .95 .03005

00000: Ammma 000>00m 3000”“ .m .D .AMZQEV m0000000m 02302 00 0008:000Q 030022 00 00090 00008 E 80000 00>00
00000 00000 2000003000 00 000050 000 00200000 0.50% 00 0000.00 000 00000 0000000023 96 000.000 00.500000 05.0%
000% 0:5 £000.83 00098.: 0.50% :03 300000 00000000 000 0000:: 000000: 000 00.333 2000000003 00 0000.52 .w 030%

50

,—¢J—4n-d uvn-nunu— .Al..:uu:lua.llua.-a.,u .01. ul:

u—nan—I. uJ-u...~ .r..:.u-_‘J.-00.JZ ~.JaJl.-:4‘- uv-u-u .J....-.J.«.~ ,--JA~ nu-J.).J‘)..-:.f, I.-.v~v--w--w~:,f. ,hau ‘unuF---2 Av EJI-fpn‘

1... . . -
Avgnnw --.>) A1,:1LLU uv...m—v:.w~l —v:... l—.-w-a~..v

.200=00000 000 0000
30003 00.000000 0000009000 0.02000 0___03-_00_000MV fig .0A& 000000000 000 000 0000. 0.000 05 £03 00: 0800 05 00 000—00»...

$00— 00=00000 000 600000 00000000,
00 0003000 003-000 050308.000! 05 503 0000—00—00 0003 002000030 00080 00000000000 00 00000 000 0.05. 090500000 .0

 

 

 

 

gal

0.0— .0 0.0 0.0.0 0.0 0.0~ 0.00 0.00 0.0 0.: 000000 00>00 00000 00000000200

000.: 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00000000 00000500000 000000000000

000.: 0.0 0.00 0.: ~00 0.0 0.00 0.0 0.00 00000 000 00000000 0000002

00.0.: 0.00 h.00~ 0.00 Numb 0.00 0.0: 0.0.0 0.03 A0500 000000000 00 0000000000000 000,—.
000.0 0.0 0.: 0.0 0.: 0.0 0.0 0.0 0.0 80 00A
00m: 01.0 v.0 0.0 0.0 0.0 0.0 0.0 0.0 000 00.00
000.0 0.0 0.: 0.0 0.0 0.0 0.: S. 0.00 000 00-00
000.0 0.00 0.00 0.00 0.00 0.0 0.00 0.0 0.00 000 00-00
000.0 0.00 2.0.: 0.0 00.00 0.0 00.30 0.00 in? *000 00 -0
000.0 0.00 0.00 0.0 0.: 0.0 0.00 0.: 0.00 000 0-0

0000 000.30 00>00 >0 00000000 000000 000000 00000 05000000000000

EmMMwE 0.0 w .00 w .00 w .00 w 0000;

3| SUEZ. ”0H0: 0053 H0 20
000000000 000000030

 

.0000 .000w0< 000 0:00 .0000. 000000000 0000: 0.0000022 00 000— 625 0000 00000002 0000: 000 db 00000000

000000 .8003 003000 0000000 .0 .D .3029): 0000000000 .0030 Z 00 0000000020 00002002 00 000000 000000 00 000000 00>00
00000 00000 00000000200 .00 0000000 000 00000000 000000 .00 0000000 000 00000 000000000000 00 000800 00300000 000000
00000 00000 .00000000 00000000 000000 0003 900000 00000000 000 0000000 000000: 000 000003000 0000000000000 000 000.0002 .0 030.0

51

JED:

"998. F

m.

Mw

3408.

m

COW

D11:

1
“V;
451m

IN)

firm;

HOT.

h’ffl

fol

Although there were no significant differences among ownerships in the mean
number of red-backed salamanders found during ground transect searches in 1997 or
1998, stands at the Huron Mountain Club showed a consistent trend of having more
salamander observations than any of the other 3 ownership categories (Tables 8 and 9).
Across all stands surveyed for salamanders, the number of salamanders found through
ground transect searches remained fairly constant during both years of data collection. In
1997, an average of 110 salamanders/ha were found for all study sites combined, and in
1998, 123/hectare were recorded. However, the proportion of cover boards with
salamanders beneath them more than doubled, from 2.3% in 1997 to 5.2% in 1998.

In the summer of l997, the greatest numbers (p=0.076) of salamanders under
cover boards were found at the Mead site and at the Huron Mountain Club (Table 8).
During summer, 1998, the most salamanders, in terms of absolute numbers, were found
under boards on Forest Service sites and at the Huron Mountain Club (Table 9).

Results of surveys conducted in September and October, 1997 were similar to the
results of summer surveys. Again, there were no statistically significant differences
among ownerships in the number of salamanders found during ground searches.
Differences in numbers of salamanders located beneath cover boards were also
nonsignificant (Table 10). However, it should be noted that during fall 1997, ground
transect and cover board surveys were completed in only 20 of 54 stands surveyed during
summer, and none were included from the Huron Mountain Club.

The most notable aspect of the fall surveys is that the number of salamanders

found per ground transect search was slightly less than in the summer, but on MDNR and

52

Table 10.
grand tr
Tilt-IE 0
first. sic
Sffilte l

ad 0911.

Table 10. Number of salamanders surveyed per hectare (means and standard errors) with
ground transect searches, time spent ground searching, number of salamanders found per
minute of ground searching, and number of salamanders found under cover boards in
forest stands on Michigan Department of Natural Resources (MDNR), U. S. Forest
Service (USFS), and timber industry (TI) land in Michigan’s Upper Peninsula, September
and October, 1997. No significant differences (p>0. 10) were detected.

 

 

 

Ownership categog
MDNIL JSFS II?l Probability
Variable i S.E. Y S.E. Y level”

 

Salamanders found during ground searches
Cover object size

0-5 cm 41.7 24.0 13.3 l3.3 0.0 0.368
5-10 cm 33.3 16.7 41.0 21.7 0.0 0.513
10-20 cm 22.3 1 1.0 5.0 2.7 0.0 0.422
20—30 cm 0.0 0.0 2.3 2.3 0.0 0.513
30-40 cm 2.7 2.7 0.0 0.0 0.0 0.513
>40 cm 0.0 0.0 0.0 0.0 0.0 1.000

Mean of all salamanders on 100.0 38.7 61.7 36.7 0.0 0.361
transects

Minutes searched per stand 56.7 I 1.1 58.7 9.1 49.0 0.867
Salamanders/minute searched 0.1 0.0 0.0 0.0 0.0 0.260
Mean of all salamanders 14.0 0.4 20.0 10.3 0.0 0.676
Elder cover boards

3Only 1 site surveyed; therefore, no standard error could be calculated.

53

least 5
rare [l
agii.
nor.

their:

WM}
5.1....

Mill“.
:0 Atl‘
n,

{fixer

Thes

Forest Service sites, the proportion of cover boards with salamanders found beneath them
more than doubled from summer to fall. However, the increase was not statistically
significant (p=0.947) because of the relatively small sample size (7 sites sampled both 2
seasons). This trend was not observed on the 1 timber industry site sampled in 1997,
where no salamanders were found under cover boards (Table 10), most likely because
sampling was completed in only 1 of the 5 stands where boards had been placed.

During late August, September, and October, 1998, a total of 58 red-backed
salamanders (35.8 individuals/ha) were counted during fall cover board surveys,
compared with 36 salamanders (22.2 individuals/ha) found under the same number of
cover boards during the June through early August sampling period (Tables 9 and 11).
The seasonal increase was evident on all 4 ownerships.

In 1997 and 1998, there were several statistical differences among ownerships in

the number of cover objects examined while conducting salamander surveys (Table 12).

Table 11. Number of salamanders found per hectare (means and standard errors) under
cover boards in forest stands on Michigan Department of Natural Resources (MDNR), U.
S. Forest Service (USFS), timber industry (TI), and Huron Mountain Club (HMC) land in
Michigan’s Upper Peninsula, late August, September, and October, 1998.

k

 

 

 

 

 

~ Ownership cmory

W M: J1m=31 _HMC_m=31 Probability
__Y S.E. I S.E. X 3.13. X 3.13. levela
#44 11 4o 10 IL 2 60 60 0.351

 

‘ Probability levels for tests of differences among ownerships were calculated with the
Kruskal-Wallis one-way analysis of variance (Siegel and Castellan 1988).

54

 

\.R.U.\v - out-'m-d< “1.... Ant-.1 oaJ-hfid‘ ndwhnd’.h~.-.LUQ. ..-UFL‘F‘K U I... ~h~wnlWin\.J..\< -I ‘d-u-wN Av .Vh<§ h v h-u\. V
€1.11 '17...1.I.I.I.I..I .I.l. .- I .A ”1.1-! .II1.’..\ b 71.1 til.nI-Auvl.derI huh-idfiuiz In: ha§1.v-thI-hfl§uvA I -§I.In\..§-\.4.\I<

.IR..A¢- .423
I'll-Illllbvlv‘l Infill-III Oil... 1.. I v )III‘!-IICII l ItdI-aI-II In .1..0..I.Il.- Iv 14..I‘.II
.. .. . . . . . 1...-.-1. .. .111 1 .1.-. 1.1 1..- ...1n ....Iu-tu. nuov 1| I. 1O\1\n.

32$ 83% 88888 23:2: azgafimamv 2: .3: 32.56 8.. as 83 as“... 2: as, 2 .waE 5:236 as.
. = 683 65 co 32m
>*

.6 mix—«SN $3-25 $33.3me 65 £3, BEEQEU 89$ mQEmuocao wcoEm woocnvmwowmu HMHMMWWOWVMQ Bwommv 8:32;
. B>2 5330.5 ..

 

 

 

 

119131 2 $1 3. S a lfil laml
o: .o m we 0 QM 5 Na w: @N EH” AWVAJII
$3 mm a: .3. a: E 2 mm a :5 ommm
30.0 mm Gm NV 2m 5m RV am 3m :8 8.2
Red mo. :33 5m :33 E USNN cw 3mg .28 QTm
owed 0mm gamma cmm awmhv wmo ammfim 3% £35 ...Eo mé
0886: con $8.30 560 Mo 59:32
whmd NNd whd Omd HNdu mmd MAY? 0 _ .O mfidu AEUV Sumo—u “Bums.—
Leov £32
956 um .o :36 mod 32.8“ ow. _ 25.3 $6 ammfiw mzoooafiom
71.0 mad 3.: New... :6— wwd «Kama mm; mafia €er 83.9.68 mom
2: .o S .o ma.» bod vmé Nod 36 I ..o mmé In :om
L32 .m.m M .m.m M .m.m M .m.m 1x. oBaca>
525305 ”mug: USE SUNSHW Ilflnlaglwmddll 3%

 

bowouao B55230

 

.32 us. 32 .uwamfi. Ba .22 .05; .2355.” 5&3 @8383 5 use. 62$ £6
53552 :95: can .95 hEaEOQ .695: .AmmmDV oomtom “modem .m .D .AMZQSC mecca—03% 72:32 no EoEtmmoQ cam—com:
:0 3.53» awoken E whotsw. eonCMEn—um “60393-62 wctsn @2339: absurd; comanuuwu> tum. 20m .5.“ .0639, 53: .N~ QBMH.

55

 

HIE

Club

an 12'

36F»

36.“

8‘0'

0%

In the smallest category of cover objects measured, 0-5 cm, stands at the Huron Mountain
Club had more (p=0.086) cover objects than any of the other ownership categories
investigated. In the 5—10 cm size class, the fewest cover objects were counted in stands
on timber company sites and the most were counted during transect searches on Forest
Service sites (p=0.031). The number of cover objects in the 20-30 cm and 30-40 cm size
categories varied considerably among ownerships, with the fewest of each on Forest
Service sites and the most on Huron Mountain Club sites, but differences were not
significant (Table 12)

Despite differences in cover object abundance among ownerships, the number of
salamanders associated with each size class of cover objects varied little among the 4
ownership types in 1997 or 1998 (Tables 8 and 9). The total number of salamanders
found during ground searches was also divided by the amount of time spent ground
searching to standardize differences in the amount of effort spent searching on each site,
but differences in salamander abundance per unit effort among study sites were not
statistically significant.

There were no statistical differences in soil moisture or soil acidity among study
areas during the study (Table 12). Soil pH ranged from an average of 5.80 to 6.84 within
a stand, and soil moisture averaged between 2% and 50% within a stand.

Among all the salamanders detected during ground searches, the size of cover
objects was not related to the number of salamander observations. Relatively few
salamanders were associated with cover objects 35 cm in width, while cover objects 30-

40 cm wide had the highest proportion of salamanders found beneath them (Fig. 4). This

56

I..o4—aC-::::...l. 5...)? 74.4-1n... .7. up

n .79. -JA -

iects

I Natural cover ob

..,.,...a...-.r.r.....,.,,...
...........ev.u........y..

‘.

..t..‘.s

......) . . . . .. . .. 2.1.... I: 1.1....
ra-lrul .rl Jltnalrllrall'IIOI AIDO'CIIIIIII no: ll-I'ar'JOJIIIIIIIIII- It’ll-IIIIDOICI
3.12.2}... ...... $3221.15}:

.Iav.r¢plrr!0nrtovno’plo lll!‘)\l|"‘~\1‘ t:

6.7

....uuu -u~cs

...p'rr.llsur r11. rvarvvlrlrovrv'pro

.m..\...‘u.~...~....... t.uu.“..«...\...‘...\‘.. uneaan .‘n ua‘e«~eeu-
.. .. ..... .. .. .....y... .....,........¢ . ......r.....;......

.. ........,..........

...... , ............... .......................... .
......a..-lppp.-..... tortoi-

utnalvooo'rlooroi topple.
«-\I\-~\«e‘\.-\n.\\«~“
tulle-Iva: verorapir'1rnvln.
~.-‘..-‘\.\..u~
. \u.“u
....a..... ......oo rirrrlryporl.
. . ....,....... ......\~..\\‘.\.‘.\\‘. \«t~‘\\\.|\u§..
.......... ..... .Jr::9;r¢;t.a;: alr.aralppuf
................oaa.~.u..--u|\-‘euv‘~u\‘v \u. n~xeguue u‘w~
..........................a::.;-ar.-rp:p.au era-ppr-vurp.::::ra r.Irpur;nr.u-».
................a...-\.-..\uu. s...‘-u..u\..~..‘u ‘~\ -\-‘\-\uu“u“xu.‘\w

6

q.“\u..xs.e.u~l..uun.uxn

. ......yt...~‘u....t.

jects

o

.. ... ..t..~“......‘-~.
..v....:v-:nv.cc:ruuau

vuvp
captorpvrrr-rrppvravervur'
.-\u.\‘t-~‘.\\.-\.c.su\q.ueuu‘~\~uu\

..t......\...\..t.t‘...~\-.n‘uu.n«n v... ...»........,..

a] cover ob

lCl

{.4........-.....4...‘~‘s.-an.xa

.‘......s.......-u\‘.uu... .....u\‘.-I".u

“-‘\u\e-\-a\\g...x..-.‘.
I..:;u-.arrrvlvrrrpr' latrrutvarrclr

\....¢..

 

 

- dl - d 1 a 4‘ q

1
oo 7 6 5 4 3 2 1 0

mecca—«83am firs $8.30 mo amoeba

21-30 cm 31-40 cm >40 cm

-20 cm Boards

11

6-10 cm

0-5 cm

(20 cm)

|CCt width

Cover ob

Figure 4. Proportion of natural cover objects and artificial cover boards with salamanders
found beneath them, summer and fall, 1997 and summer, 1998 In Michigan 5 Upper

Peninsula.

57

sun:
58

W

W.

C'OU

mld

YES

rel

suggests that certain size classes of woody debris may provide better habitat for
salamanders than others.

Salamander relative abundance exhibited a strong (p30. 10) positive association
with the density of overstory trees 2.10.2 cm dbh (Table 13). Significant negative
correlations were found between the number of salamanders observed and several
midstory attributes, including midstory canopy cover (rs=-0.325), canopy cover of shrub
species (r,=-O.430), and the density of shrubs and regenerating trees 50-5 m tall
(rs=-O.297). No correlations with woody debris were statistically documented.

Comparison of ground transect searches and cover boards for surveying

salamanders

With ground transect searches, there were no statistical differences in salamander
relative abundance among study sites in any sampling period, while use of cover boards
resulted in differences (p=0.076) among ownerships in spring 1997 (Table 8). The
relative differences among ownerships indicated by each method were somewhat
different as well. For example, in summer, 1998 ground transect searches revealed more
salamanders on MDNR sites than on Forest Service or timber industry sites, but with
cover boards, fewer salamanders were found on MDNR sites than on sites in all other
ownership categories (Table 9). However, results of both methods suggested that the
relative abundance of salamanders was greatest in stands at the Huron Mountain Club.

In 1997, 14 out of 660 boards, or 2.1%, had salamanders under them during the
summer survey period. This proportion increased to 4.6% in the fall of 1997, and in

summer, 1998, 36 out of 810 cover boards, or 4.4% were observed with salamanders

58

Table 13. Spearman rank correlations (rs) between the mean number of salamanders
found (ground transect searches and cover board surveys) and 35 forest stand variables in

Michigan’s Upper Peninsula, 1997 and 1998.

 

 

Variable rs
Saplings/ha -0.07
Shrubs/ha -O.30*
Snags/ha -0.01
Stumps/ha -0. 12
Logs/ha 0.04
Shrub height (cm) -0. l8
Snag height (m) -0.07
Snag diameter (cm) -0.13
Stump height (cm) -0.03
Stump diameter (cm) 013
Log length (m) 0.19
Log width (cm) 0.06
Herbaceous height (cm) -0. l4
Litter depth (cm) 0.02
Basal area (m2/ha) 0.13
Overstory trees/ha 032*
Overstory tree DBH (cm) -0.01
Overstory tree height (m) 0.02
Vertical cover (%)
0-0.5 m -0.03
0.5-5 m -0.33*
Herbaceous -0.05
All shrub species -0.43*
Woody debris 0.10
>5 m 0.17
Conifer trees 0.10
Deciduous trees 0.07
Hard mast trees >10 in (25.4 cm) 0.01
DBH
0.02

Soil pH

59

Table 13 (Cont). Spearman rank correlations (rs) between the mean number
of salamanders found (ground transect searches and cover board surveys)
and 35 forest stand variables in Michigan’s Upper Peninsula, 1997 and
1998.

 

Variable rs
Soil moisture (%) 0.14
Number of cover objects per hectare
0-5 cm 0.16
5- 10 cm 0.10
10-20 cm 0.00
20-30 cm -0.1 1
30-40 cm -0.09
4420“! 0%

 

* Probability level 50.10.

60

SUD

rein

COT

ncg

pri

beneath them. In contrast, with ground transect searches 95 salamanders/ha were found
in spring, 1997, 83 salamanders/ha were observed in fall 1997, and 125 salamanders/ha
were found in spring, 1998. These results suggest that both seasonal increases and
increases specific to the sampling methods were observed.

Correlations between the relative abundance of salamanders observed with both
survey methods ranged from 0.15 to 0.23 among the 3 sampling periods (Table 14). The
relationship between the number of salamanders under cover boards and the number of
salamanders under natural cover objects 0—5 cm wide were consistently positive, but
correlations with other cover object sizes (>5cm) fluctuated between positive and
negative values for the 3 sampling periods (Table 14).

The proportion of artificial cover boards with salamanders beneath them and the
proportion of similar sized natural cover objects may be the most direct comparison
between the 2 methods. Although cover boards were 20 cm wide, they were used less
than natural cover objects 11-20 cm and 21-30 cm wide in proportion to their availability
(Fig. 4).

The time required to complete each type of survey is also an important
consideration when evaluating the usefulness of each survey method. An average of 70
minutes was required to search all 3 ground transects in each stand surveyed, in addition
to time used to set up the transects. Cover boards initially required a great deal of time
and labor to put in place for the surveys, but once established, the 15 boards in a grid

could be searched in approximately 8-10 minutes.

61

Table 14. Spearman rank correlations (rs) and probability levels (p) for the number of
salamanders found per stand between artificial cover boards and natural cover objects
grouped by size class.

 

 

 

Sampling period
Spring 1997 Fall 1997 Spring 1998
01%.) 01520) main

Cover object size L p L D L J)
0-5 cm 0.22 0.157 0.20 0.402 0.20 0.142
5-10 cm 0.10 0.500 0.24 0.307 0.19 0.168
10-20 cm 0.17 0.256 -0.31 0.181 0.20 0.147
2030 cm -0.04 0.788 0.35 0.127 -0.23 0.098
30-40 cm 0.22 0.145 -0.10 0.671 0.06 0.654
>40 cm -0.09 0.530 -0.20 0.295 NP NP
All sizes combined 9.2} 0.153 12.32 0.085 9.15 0.268

NP = Cover objects were not present.

Forest bird species

During the study, a total of 5] bird species were surveyed on study sites
(Appendix E). Forty-two species were recorded on MDNR sites, Forest Service sites had
44 bird species, 38 species were observed on timber industry sites, and species richness at
the Huron Mountain Club was 39 (Table 15). Twenty-seven of these species occurred on
stands in all 4 ownership categories, and 8 were detected on only 1 ownership category.
The brown-headed cowbird (Molothrus ater) and cerulean warbler (Dendroica cerulea)
occurred during sampling only on MDN R sites, the mourning warbler (Oporomis

philadelphia) was only observed on Forest Service sites, and the ruby throated

62

{Carl}: U:- Caa A.m.—..f..v I.=v.--.U ua...a—«:.:f.

—J--w A—vaJ.-.--.u.uni yummy-IIf-U

I.c\lf..ll

 

63

 

 

......o ...... .5. .....m ...... mam ...... 8.. ...... .83... .83 seam
...... NZ 8... ...... ...... .3 ....m ... .m m... 5.8.8.3 855..
m8... 8... .8... on. an... ...... .8... on. ...... .52.. .5658
8.... 8... 8... .. .5 Es ...... ...... 8... 8... .83.: SEER.
.... ... a... ...... ...... ...... 8.. 3.. m2. ...... 5...... 82.35.26
8.... 8... 8... 8.. 8... 8... 8... a... ...... 5...... 5.228
N8... 8... 8... 8... 8... 8... 8... 8... 8... 2.38 ......aaem
.8... 8.3 ...... 8... .8... ....N .....N a... ...... 12...... .56...
$2. .... 3.... .3 mm... m... a... 8... ...... ......m
5...... mm... ...... 5.... .....8 2.. .5.... mg .3... 1......» 52. 8.8.5.8...
.8... ....N .3... a... .38 Rd .38 8a 28.: .53.... 8.. 8.8.5-8....
8m... .2 .....N .5. 8.: $3 «a... a... S... 8.8.0.... .88.-..8...
3:. 22 Re a. .. a... 8.. S... ....N ....N 5...... 5.5.5.8...
.8... 8... 8... .3 ....N .....m 8.... mm... a... 8...... 8:58....
So... .. ... ......h. .....N .N..~ .... .8... mm. ...... tee... 2...; .5. .8...
8.... .... 8... .... .... .m. 8.. 8... a... ...... 50.2....
.8... t... ...... 3m .8... N. ... .8... .3 .2... .23.... 53.55...

2.... ...... a3 ... s ...... 8. .....N 8... ...n a... 5.58...
.mm_:m.o:3o .m.m M .m.m M .m.m M .m.m M 2...... 5.8800
ween... 8:23.... hm"... USE Am"... C. 8n... mmmD an"... MZQE
.6 525805 bong... 8:82.30

 

...... .. e. .28... .38.... .8. .5. .32

..8. 53-..: .2355. 5...... .835... ... .5. 62:. 86 53.5.2 .55.. .5. ....H. has... ...E. .88... 8E...
.88"— .m .D AMZQE. .8583. 3.2.2 ..o .coEtwmoQ camioaz co Riots. .20on ES .0 22.828 ...... ..o a...» m .o>o
.538 05 so 6mm. ...on ESE... ...... 32.580 8.8% 5033 ... 850.. ..0 38.53 38:033.... 0.203.. .802 .m. 2an

 

 

 

8.... 8... 8... 8... 8... ...... ...... a... S... 38......» 8.3%....
...... m . .m 8.... a... 5.... 8.. «...... m... um... 8.: 83-.....

0...... 8.. .....2 8... .8... ..m. ...... N2. .3... £2.52. 882.-....

.8... a... ...... 8... .8... 8... .8... 8... .8... Luz...» 2....

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

...... ...... ...... Sm ...... .... .....N 8... .8... 13.8883 .38....

5...... R. . ...... .m... ......N ....N. .8... . ... .8.... .2355

NR... .3 2... 8... 8... .... .....N a... a... 2...... 62......

...... 8... 8... .2 ...... ...... ...... m... a... ......a 05:82

8.... 8... 8... 8... 8... .... E... 8... 8... 3...... 8.5.5.).

m3... ...... 8.. .3 ....N ...... 8... 8... 8... 3...... 2.8....

an... ...... ......— a. .. ...... 8.... .2... .. ... .....na .5502... ......

.8... mm. .8.... .3 .3... a... .....N». .... .8... ....a... .20.:
...... 8... .8... a. .. ...... ... .. .5... 8... .8... ...... ...»...
...... ... .. 8.. a; ...... Nu. m... ..m. .3 .38....83 ......
8N... .... ...m 8... 5.... .... .... 8... a... .238... .28.. .8...
Rm... 8... 8... S. R... 9.... S... .2. ...... 0...... 8.320.528
....EwBEso .m.m M .m.m M .m.m M .m.m I... 0E... 008800

$.08. 0000.0..6 fidfii an"... E. ifimawmwdlu ii
..0 5:330... #8580 0.5.0030

 

.200 ... 0... 8.00% 8.00.05 .30. ...... .30. .000. .33

432 03.55.. .253 ......wEflE E ...... 62:. 020 0.0.5.02 00.5: 0:. .9... .930... .008: .Ammma 003.0m 80.0".
.m .D .2292. 30.83% 3.5... Z .0 .00Etwm0Q 5.3.6.2 :0 00x03... 8.00% 0.3 ..0 00:00:00 .3... ..0 .80» m .0>0 2.00:.
05 :0 Add. 80.0 0.30:... ...... 2.0.5000 .200... £033 ... 350.. ..0 .0080... «0.000008. 0.203. 5.02 .9008 n. 05....

MVP—u :nv A.m—..fnv {LAY-Luv

—v._...._u:..-I. .uC... A—7J._,_:-J.J.v 1.4...JUCI. ~_.J..:.$ ..u 1:... v: .... .....-nhlrux U;...;.a.a....2..-..‘ ..~u..-¢~..J.. {if-‘4.

tiilliu

LI

 

000.0

00.0

 

65

 

 

00.0 00 .0 ~00 00.. 00.N 00.0 00.0 5000.0... 00.0—00-3000?

.000 00.0 00.0 00.0 00.0 .0 .0 ~04 00.0 00.0 000.000: 000.00-30.00»

00 _ .0 00.0 00.0 00.0 3.0 .0. _ ~04 00.. 00.0 0.0.0. 0003
000.0 00.0 00.00 00.0 00.: 0 _ .0 00.00 00.0 00.0— 00.3 .30.?
000.0 3.. .00; 0. .0 .00.. 00.0 .00.: 00.0 .000 .3050. 00.00.00-000?
000.0 00.0 00.3 .020 3.0 ~00. «0.0 00.~ 00.0 000.500 000.00.000.03
:0... 0... ...... ...... ...... 0.... ...... .... .2...— ......>
000.0 3.. 00.0 00.0 00.0 00.0 00.0 00.0 00.0 5.00.3 00.8000...
000.0 00.0 00.00 00.0 00.0 00.0 00.00 0. .0 00.2 0.000 000.8030
000.0 00.0 00.0 00.0 00.0 .00 00.0 00.. 00.0 00.3 0.00000

0 0 0.0 00.0 00.0 0. .. 0—A 00. _ 3.0 00.. 00.0 .0000... 00.0.00
000.0 00.0 00.0 0. ._ 0_.— 00.. 0—.~ 00. 0 00.0 0080 E00000
000.0 00.0 00.0 00 .0 00.0 00.0 00.0 00.0 00.0 00000880: 000.000-0000
0.0.0 .00 RNA 00.0 000.0 00.0 .00... 0.0.0 .000. $80.80 000.00.00.30
...:0. .00 30 .m0 M .m.0 M .m.0 ..m .m.0 M 0800 008800

00080 0000.00.00 Ilflmfldal Gus E. 'fimEIwMWDI lawn—3|
00 000000000 0.000000 80.00030

 

.0_00 0. 0.0. .0600. 00.00.05 .0000 00.0 .0000 .000_ ...—0..

-002 ...—0.0800 .0003 $000.00.: 8 000— 0230 005 0030002 000000 000 .900 0.0.000. .0080 .A0m0a 003.00 0.00000
.0 .D .90an): 30000.30 080.02 .00 0008800000 000000.02 00 000030. .2000. 000 00 00:00:00 800 00 .8000 0 00>0 .0008
00... 00 Ant—.00 .0080 000000.. 000 0080000 .2000. 00.03 0. .0800 00 800.03 00.000000... 0.0.0.00 000—2 A8000 0_ 200.0

Up: 5». Air—ff; l._»....... 1-.—3.23.1 1: . ;~.a...t.-11-.

.2000. 02.2.00 000

.0020. 2.2.0.. 00.000800 0......08 ....03-.00..0.v.0 .0. 0A3 80.0.0.0 .00 0.0 .002 080. 05 5.3 00.. 080. 05 00 .00.0>*

..000. 0002.00 000 .0020. 0000..0>
.0 ......000 003-000 2:03.835 05 5.3 00.0.0200 0.03 3.080030 00080 .0000.0....0 .0 ...0. .0. 226. ......0000... .

.0...0|||||:0 0.0 8.0 00.0 8.0 80 00.0 4...... ...0.; .5....

 

 

 

 

 

 

000.0 00 .m .000 00.0 000.0 .00 ...0.0 00.0 .00.. 0.20.03 00088-30=0>
...0....0030 :00 w .m.0 M .m..0 I... :00 m. 0800 008800
00080 0000.0...0 an"... USE AMHdJH Anna. mum: Gun. MZQE
..0 ......0000... 5000.00 80.0030

 

0.00 0. 0.0 .2020. .9020... .000. 000 1000.600. 50..

-..0.). 0.0.0.00... .0002 2000.000). 0. 000. 6.23. 00.0 0.0.0002 00.0.. 000 ...0. .....000. .000... .0003. 00.>.00 ..0.0n.
.0 .D .2sz. 80.00.00 .8802 ..0 80800000. 000.020,. 00 000080. .2000. 020 ..0 00002.00 0.00 .0 ..000 0 .26 .0008
05 00 .000. ..0..0 0.000.... 000 .00....000 .2000. 00.03 .0 20.00 .0 800.08 .2000000... 0.0.0.00 000.). .0000. 0. 200...

66

hummingbird (Archilochus colubris) was only seen on 1 timber industry site. Species
that were identified only at the Huron Mountain Club were the pine siskin (Carduelis
pinus), pine warbler (Dendroica pinus), Tennessee warbler (Vermivora peregrina), and
yellow warbler (Dendroica petechia). On all 4 ownerships, the red-eyed vireo (Vireo
olivaceus) was one of the 3 most abundant species observed. The ovenbird and black-
throated green warbler (Dendroica virens) were among the 3 most abundant species on
MDNR, Forest Service, and Huron Mountain Club sites, while the hermit thrush
(Catharus guttatus) and American redstart were some of the most common species on
timber industry sites.

Of the 5 bird species identified at the beginning of the project as having
associations with particular aspects of northern hardwood forests, all exhibited
differences in relative abundance among ownerships when data were combined among
sampling periods (Table 15). Specifically, American redstaits were more abundant
(p=0.084) on timber industry sites than on other ownerships, and veeries were most
common on timber industry and least common on Huron Mountain Club sites (p=0.047).
The ovenbird occurred most frequently on MDNR sites, and least frequently on timber
industry sites (p=0.047). Pileated woodpeckers and yellow-rumped warblers were more
abundant (p=0.056, p=0.065) at the Huron Mountain Club than on all other ownerships
(Table 15).

Other species that were more common at the Huron Mountain Club than on other
ownerships were the black and white warbler (Mniotilta varia) (p=0.097), black-throated

green warbler (p=0.057), brown creeper (Certhia familiaris) (p=0.089), and pine warbler

67

(p=0.088). Relative abundances of the least flycatcher (Empidonax minimus) (p=0.052)
and rose-breasted grosbeak (Pheucticus ludovicianus) (p=0.017) were greater on MDNR
sites than on other ownerships, and the eastern wood pewee (Contopus virens) (p=0.060)
and white—throated sparrow (Zonotrichia leucophrys) (p=0.051) were most abundant on
MDNR and Forest Service sites. Finally, the black-throated blue warbler (Dendroica
caerulescens) was more common (p=0.075) on Forest Service sites than MDNR sites
(Table 15).

Several species varied in relative abundance among years, and many more did
not. The black and white warbler was the one species which showed the greatest
fluctuations in abundance during the study. At the Huron Mountain Club, the black and
white warbler was observed more often (p=0.074) in 1998 than in 1997. On MDNR and
Forest Service sites, abundance was greatest (p=0.022, p=0.035) in 1996. In general, this
species was relatively less abundant in 1997 than in the other 2 years on all 4 ownerships
(Tables 16, 17, and 18).

The least flycatcher also differed in abundance over time on Huron Mountain
Club (pr-0.061), MDNR (p=0.046), and Forest Service (p=0.022) sites. On each of these
3 ownership categories, this species was observed most frequently in 1996, and relatively
less frequently in 1997 or 1998 (Tables 16, 17, and 18). It is not known if the same
pattern occurred on Mead and Shelter Bay land, because songbirds were only surveyed in
1997 and 1998 on timber industry sites. On MDNR sites, differences in abundance over
time were detected for the hermit thrush (p=0.031), ovenbird (p=0.062), and white-

throated sparrow (p=0.034),

68

.200... 00038000200

69

 

 

 

.....0 00.0 8.0 8.0 8.0 .... 0...
0.0.0 0... .0... 00.. .00... 0... ...... .0032. 083 52.00
a... .00 .... .00 0... 0... a... 0.8.08; .560
0 00... co... .5... 00.0 00.0 00... 2.... 0030 008800
000... co... 8... 00.. 00.. cod 8... .020... 000.800
....0 8.0 00.0 0... 0... .00 .00 .0303 0000-35.20
.30 8.0 00.0 8.0 00.0 .... ...N 0.538 80800320
0.0.0 .0. 8.0. a... .... .00 .00 .0085 5580
0.0.0 00.0 .... ...... 0... .3. 0... 00.020
000... 00.0. 0...... 00... 00.00 00.0 00.0.. .2803 000.0 00.00.03.005
.000 o... 3.. ...... 05.0 $0. a... .0303 2.3 028.0020
$.00 .... 8.: a... ...: 0... 0.... 808.020 02.08.0020
.0. .0 3.. .... 0... 0... 8.0 8.0 .2003 0053.020
.000 8.0 8.0 5.. a... ...... ...0. 0808 0200.020
0.0.0 .00 an... 00.. 00.00 .. 0 ...: 03.03 20.3 0:0 .020
mmmd o. .0 00.0 00.0 00... 00.0 00... 0.00. 002.080.
0.0.0 8.0 .00... $0 .3... .0... .03. 0.5.02 08008..
.900 0... a... 00.0 8.0 .00 .00 380 30025.

..0.0..0030 .m.0 M .00 w :00 M 0800 008800

00080 0000.0...0 .wu0. 0.)...1 Aflnd. 0003 Ann... 02%:
.0 0.0.0000... 5000.00 0.080030

 

0.00 0. 0.0 .200... 8.00.00. 000. ...0.—.40.). 0.0.0.000 .000.. ..000.00..). 0. 000. .03.... 00.0 0.0.002).
00.0.. 000 .9292. 80.00.00 .8807. .0 80800000. 000.02.). ..0000. 00.>.00 ..0.0n. .0 .D 00 .0... 000.. 00 00.080.
.200... 0.... .0 00.0. ..080 000000.. 000 .0080000 .200... 00.03 .0 00.00 .0 000.0080. 8.000002. 0.0.0.00 000.). 0. 0.00...

.800.»

 

70

 

 

0.. 0 000 000 .0. .0... .... 0....
mam... mmd 2.0 co... 8... cod 8... .0383 03.0000...
...0 ..0 .0... .00 a... .0... .0... 0.25 .0850...
...0 .... .... 00.0 000 00.0 000 8..., 03.3.
.000 00.0 000 0... .5. ...... 00... 5...... .008.
.000 00.0 000 0... .3. .... 0.... 2.... 020......
0000 :1 ...... .... .80. .... .00... .030... 02.85-08.
..00 00.0 000 0... .... 00.0 000 380083 0.0.2.80
000... 00.“. 00.... .0... 9...... m . .N 00.00 u..00..> 0000.03.
..00 00.0 000 0... .... 000 000 b.8008; 00.8....
:00 00.0 ...0.... .00 .0... .0. ...... ...0....85
0..0 .... .... 0... 0... .... 03. .3... .5052
.000 00.0 000 0... .... .... ...~ .20.... 202.82
.....0 ...... a... 0... 0... .... .... .03.... «02...:
..00 .. 0 ...... .00 .3... a... :0... .058... ...0..
..00 N... ...... ...... ...... .... ...... ......e... ...0....
.....0 000 000 m. .. .... 00.0 000 .00. .3...
.000 000 000 .. .0 .3. 0.... 0.... 3.80.83 ......
.0... 00.0 000 .... .... .... .... .228... 02...... .8...

...0....5030 .m..m M .m..m .m .m.m 1M 0800 008800

00080 0000.0...0 an... 02.0% Am"... mama .mnlfi M753.
.0 5.00000... 0500.00 8.080030

 

0.00 0. 0.0 8.000. 8.00.00. 000.01.07.03. 0.0.0.000 8qu $0.082... 0. 000. 00.2.... 00.0 0.0.0002 00.0.. 000
.2752. 30.00.00 .8807. .0 808.5000. 0.00.0.3. ..mmmD. 00.>.00 00.0.. .m ... 00 .0... .00.. 00 00.030. .060... 0....
.0 ..md. 8000 0.000.... 000 300.8000 .060... 00.03 .0 0.0.0.. .0 000.000.... 3.000002. 0.0.0.00 000.). ...000. 0. 0.00...

.waa 5:8me ES
BmeH osmcfim car—3E8 29:2: m=_m>?-_mxm8v3 82KB thdtc 8: 2a 5:2 2:3 2: £3, 0:: 2:3 05 co mus—Er."

.Awwfl 5:8me can Ewofiv 8593
mo wax—«SW $3-25 333.?me 05 5:5 wean—5&3 803 mmEmhocao wcoEm 30:88:“. Mo 33“ H8 232 b25305 a

 

 

 

 

mm _ .o mod N3: cod 93.: 9am me.” 585 woo?
9: d aod 34m and «m6 mm.w Emma :83 82:3
2 _.o mmd Mad as mfivn awd cm..." 383% 3385-823
Vm _ .o 3N 8.3 No.0 mad— m _ .N 3.: 5035:: 3335-823
«EEEQEso .m.m M .m.m M .m.m .M 25: 58:80
98:8 858:6 AmuE DEE SHE mmmD amnE M752
.3 335305 Eowofio 358550

 

.28 E 2a 360% 5:36.: 63— 53.432 .flschom 5an PSNMEUGA E 2:2 625 920 58552 :85: was
.EZQSC much—80m .8332 mo Eofitmmon— :awanz .AmmmDv 83.5w 395m .m .D :o 8% 35% so 3.333 36on EE
“.0 A.m.mv E23 235% 98 80:38 380% £253 “a 350m «0 :Etomoav 885595 828% :32 .958 2 2an

71

 

72

 

88 03 3... a: 3 s a: 8m 03 228% cameo sec
2 _.o 8.0 8... a: um: $4 8.0 8... E32 8:35-528

0.2 .0 mod 8." a: a? 3...: 03 08a 8300-083 825
83 8.0 8... 8... 8.0 8.0 31V 3.... 38383 2260
«8.0 8.0 .8... 28.0 0 E 3.: cod .8... .522 85:50
30.0 8.0 .8... .8: 8o .8... 8.0 .8... £9.20 85:50
30.0 SN «3 8.0 a? an." 8.0 8... 5533 83-55.20
82 8.0 8... 8... 8.0 8... mod 23 3333 52800
an; 03 3... 8.0 8.0 8.0 8.0 8... 5320 5,80
£8 mg an." ..E 8.... «8 80 Rd 3.320
$8 8.0 8.3 3.: En 8.2. 2.: and. 5533 :88 83850.85
30.0 8.0 .8... 8.: c; .38 $0 8.3 .3533 2.3 385.0320
20.0 a} 5.2 a: Rm 0:: m 0 .v 3.0 8325 3&8;an
S00 : .... 2.0. 8.0 t .0. 3:. Sn mm.» 5533 583820
was 8.0 8.0 8.0 of an; «3 a3 8068 023-0620
82 8a 03 8.0 8.0 8.0 8.0 8.0 53;; 22; Ba 0.85
03.0 8.0 8... 8... 8.~ an... 5.0 5.0 :52 50:25.
023 m3 8.: 03.. a; an: Em 8.8 :58... =8t~E<
520 2.0 aha 3.: E4. and 3:. 3.... 380 505:2.
ammfiflgao .m.m M m. .m.m .M .m.m M 08: 88809

mecca 00880.50 IQMSJUZMII EH filmmadl lama
00 325305 .9888 3:80:30

 

.202 E 08 880% 080065 .32 53.432 .w_=m:_:0m 8&3 983532 5 0:2 025 92.0 53552 :2st 0:0 006
329.6 .00 08—2 _ .806 32nd): 80So8m 3080 Z 00 E0Etmm0O 582032 0:0 Gama 00_>._0m 3000a .m .D :0 00833
880% EB 00 dqmv £88 0805me 08 80.5500 880% 20E? .0 3500 .«o E00003 86830000 03—080 :82 .2 030%

 

73

 

2 no 03 03 v2. SN Rn one 5,: E3»
83 ohm on... 8... :0 an... 80 8... 525 pcacaam
was $0 03 8... 8.0 8.: SN 8." 85 92:8
80o 8.0 8... 8... a? an; cod 8... amass 358
$06 80 8... 8... mam an." 05. 05. 2:3 afiSm
Sod cod ‘8... ..Es cod .8... cod ._8... ...Eswgeeé B§§-E§
2: .o 80 8... 8... mom «3 s: R.» 0.838» 8,305-33.
Bed :2 «Sb 3.? E s cm...” ono 3.5 8% 833m
950 m: ...0.: .85 m2. ..mm... Nam 33 2323: 02305.3
83 80 an 8... 8d 8... 8o 8... 502... 2E
N: .o an 3.2 Es 0: 05. cod 8... .3838; 088.5
on; 8.2 02.0 Eda m3. 3.? is 3.3 2385
33 cod 8.: 3s a? an; 80 8." 5333 2:232
80o cod 8... 8... N? am.” cod 8... .5333 05.552
«a .o cod 8.: 3.» 3o 8... «3. mm.» .2285 $3
3 .o 93. 3.2 3% m3 5.: 20 3mm 52.: seam
9.3 80 8... 8... 80 can cod 8... 0% 0%:
98¢ 3m 5%— .3: 80 ‘8... 8o .8... 138383 be:

ammfifloaao .m.m M M .m.m M .m.m M 080: 58800

$880 08000006 lama—323' g lama—Ha 32%..

Mo bEnBoE E0308 9500550

 

.23 E 03 360%

08865 .32 5:732 £35.09“ 5&3 m.§w222 E was 625 9:0 £3552 5:5 98 0% PC bung:
.898: _ .003 2753: 80583. 12302 00 0:08:35 :meBE can Gama 0330.0, meow .m .D so Bag—3 360%
BE .«o 6mg 8800 230.8% 23 80:88 880% EB? “a 850m 00 30808 86:00:00.0 82030 502 .95an S 030,—.

 

.wae 5:230 :5
6w0_mfl ocmcfim :85:an 29:2: m=_m>?-_§m5v 2: .OAE 20:85: :0: 0:: 3:0. 08% 05 2:3 0:: 08% 05 :o 356?."

$3. 8:280 23 swam: 8§§>
.6 max—0:: 333:0 m5: 3-1:me 0:: £3 3:23—00 0:03 3:33:30 m:o:.8 30:88:: :0 $3: :8 203— 3:50:05 a

ABE—:23 0: 2:00 8:0 33:03 0: 6:80:05 60333 BB fl 3:0 .

 

 

 

 

 

 

 

 

as: as 3 So .c
v: _ .: 8.: 8.: 3H wm.m me 8.: 8.: ...:03238 3:62-326?
who: mud aha 8.: mm: mm; 8.: 8.: 3:83»: 3509326.:
:3: 8.: 8.: 3.5 m:.m m:.m 26 5.: :25: :83
v3 .: _ _.N_ «5.5 8.: mm: 8.: on: 3.: :83 3:53
an : .: 8.: 8.: 8.: 8: ::.n 8.: 8.: 38.8% 3385-263
m m m: :m.m :m.m 8.: Va: w~.m 8: ::.N .6355: 3335-863
am E5. :0: 3 o .m.m M M .m.m M .m.m M 0:5: :oEEoU

32:: 00:03am: lflmfiduflul QMSH amHE wme u 2
:0 3:60:05 E033 3:30:30

.20: E 03 360%

8:365 .33 53.4922 53:69: 3&3 m.::w_:0=2 E :5: 635V :20 5355—2 :05: :5 0am at @3335
3:8: _ .35. 92sz 30.5%3— 6:32 .6 E08333 536:2 :5 Ammma 00_>._0m :38& .m .D :o @0333 360%
BB .6 A.m.mv 30:0 20:5me :3 32.508 360% 623 3 3:6: :0 E0803 36:013.: 0:28pm :32 .388 C 030.:

74

 

75

 

 

000.0 8.0 00.0 00.0 00.0 00: 00.0 8.0 00.0 .000 0.30:
000.0 00.0 00.0 00.0 00.0 00.: 00.0 00.0 00.0 5080803 :0:
:00 00.0 03. 00.0 00.0 00.0 00.0 8.0 8.0 200%: 02020 020
000.0 8.0 8.0 8.0 00.0 00.: 00: 00.0 00.0 E000. 0003295200
000.0 8.0 .000 03. .000 00.0 .000— 000 1.0.0: .8300 083 82000
:00 00.0 00.0 00.0 00.0 00.: 00.0 00.0 03. 380083 05.6:
000.0 00.0 .000 00.0 .05: 00.0 .0000 2 .0 .2... .522 005800
00 _ .0 8.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 .2003 0003050000
000.0 8.0 00.0 8.0 8.0 00.. 00.0 8.0 8.0 608.0 :32:
000.0 00.0 00... 00._ _ 00.: 00.0 2.0 00.0 00.0 00.0.20
2 :0 00.0 R00 00.0 00.5. 0_ .0_ 00.00 00.2 0...: 5303 0020 020250.02:
0_ _.0 00.0 00.0 00.0 00.00 00.0 00.00 00.0 00.00 .2003 020 0288-002:
E .0 03. 00.00 00.2 00.00 00.0 00.00 00.0 00.0 0003020 3008-002:
000.0 00.0 00.00 000 00.0 00.0 00.0 8.0 8.0 55.03 E63002:
000.0 8.0 8.0 05. 05. 8.0 8.0 00.0 03. 0838 0200-002:
000.0 00.0 .0000 :0 .00... 00.0 300.0 8.0 .000 £0303 223 0:0 0020
000.0 00.2 3.00 00.0 00.00 00.0 00.: 00.0 00.0: 508 085:2
000.0 00.0 200.0 00.0 000.00 00.0 .0000 00.0 .000 .2832 08:00.4
000.0 00.0 00.0 8.0 00.0 8.0 8.0 00.0 03. 320 50:20...

.0&:0._0:3o .m.m M .m.m M .m.m .M .m.m M 080: :oEEoU
w:oE0 00:80:06 lflmgl Gum: E. GUS mmWD .IQMSIMHQZI.
.6 5:50:05 0:000:00 3:000:30

 

.20: E 000 360% 080065 :03 53.4002 0:00:60: .033 0:03.632 5 ::0_ 62m: QED E00552 :03E
::0 .95 3:035 002:: A953: 00_>00m 0300": .m .D .9952: 300.53% 6:802 :00 0:083:09 :0wE0M—z :o “00.00350
360% ES :0 Add: 0.5000 E03000 ::0 3000,5000 360% .633 00 8:6: :0 0:00:08 36:003.: 002030 :002 .m. 030:.

 

 

76

 

word 2 .v 5.9 mmd 3.9 mod 23” m _ .v mad 2825: 8535-223
Sod ”Z .2.” 8a .82 9% .38 m: :2: 38>
god a? .33 wow .32 2: ..Sa one and .535 pcomcai
£3 wZ aha cod 8... m2 «3 a: E... 8.; Exam
owed mom an." a? «an 9% Rs a: c5. 53.5 5.30m
mono 8d 8... w? «3 8d 8.: 8.0 8... 2:3 afiém
ammo 8a 8a cod 8... a? an." RN 3.... 338m Baseaom
33 8d 8... cod 8... 3o 8... mod ”3 Eggnog 89.26%
33 N3 2.; can 8.3 8.» 3.3. Ga anew 8.; 83-8”.
83 Sm as cod 8... 92 3.; SN 22 £225: 8.985-an
wwod mmN .3}. cod :8... cod .15.: cod as... 10333 8E
88 mg ”3 8o 8... 85 8... 8d 8.: :35 BE
£56 3 .v gems cod ‘8... cod :86 cod :8... ...—0393393 63525
mmod mod Lhmh 3.2 oveda mow ..hmém no; amedm ...—5:5.»6
a: .o one an... 3o 8... NE N3 cod 8... 22g 60552
83 cod 8... mom ”3 8.0 8.: cod 8... 53.2, 2:282
Ed mg aha oi. 3... a? 93 8o 8... .0333 23:32
Ed 0:. one o: 2.... 8o 8.: 8a ”3 2288 .33
Etc 5 _ 8.3 2 .0 3.: N2: «3.. NE 3...: 525 “Eco:
«8395550 .m.m 1M .m.m M. .m.m um .m.m w 0:8: 58:80
macaw 855:6 Anna HEM Anna: E. ANNE mmWD 3%
Ho 523305 Eowoau 955230

 

.23 5 8a 860% 82865 .wafl 53-x“: flamficom 5&5 MammEom—z E us: 623 £35 58552 :oSm can
.95 .939: “BE: .Ammma oomtom 220m .m .D .EZQEV 882.8% .85“ Z 90 E08539 sawing—2 so 333:... 860%
ES .6 0mg 95:0 23:83 23 68.508 880% £033 a 850m .8 E883 865st“ 838% .802 .958 M: 223.

.83. 5:830 .28
.035. 03.3% acmtmmEoQ 07.3.2: m...a>2-.8.wav.v 8. .OAE E23...” 8: 8w 8:0. 083 2.. 5.3 2:. 08mm 2.. :o 825?,

3.3. 5.2.6 as. swam. 85:?
.o Ewing $3-25 flak/.365 o... 5.3 33.3.3 803 3.5555 macaw 3820...... .o 83. 8.. £96. 3:538; .

 

 

 

 

 

 

Imam? MK; 8... QED cod 8... 8o 8... $553 32.8.
mom... xv... 8.2 cod .5.: mm. um; me and 33.33 3:83-3c..e>
om... cod 8... mm.m ac... mod 3% Ema 3.: 5.2.2.3 to...un-3o..o>
3K... cod .5.: mm. mm; mm. um.— mON ”A; 5:28»... to...on-Bo..o>
mm. .o cod 8.: m. ... 9:. cod .5... mod mm... 55:... .503
and ...c 8.5m mm... 3.: mad end. 36 8.x :83 .853
Ev... cod .5... wm.m «flu ocd ...m mod £3 38.8% 3385-353

$35855 .m.m M .md m. .m.m .w .md m. 25: coEEoU
wcofim 8:28.... uldmmadzfll SHE .H IIAmdewMWDIu Aide/4|:
.0 5253.5. bawuwfio 35.2.30

 

.22. a. 8d 8.8% 5.36:. .woo. $3732 Esmchn. $9.5 m.§w...o.2 a. .23. 025 9:0 58:33. :05... .55
A...» .9265 .38.. ..mmmDv ooEom :20”. .m .D .9sz. 88:83. .8387. .o 208$qu samufiaz co coxotnm 860%
ES .0 ..m..mv $86 €3.55 .23 80:38 8.8% 5.23 a 352. .0 E885 8.80:3... 25.8% 88.). .958 x. 0.2m...

77

On both MDNR and Forest Service sites, the rose-breasted grosbeak and
Swainson’s thrush (Catharus ustulatus) were most abundant in 1996, and less abundant
in 1997 or 1998. The veery was also recorded more frequently (p=0.051) in 1998 than in
1996 on Forest Service sites. On timber industry sites, where data were collected on 1
site in 1997 and 3 sites in 1998, the American crow (Corvus brachyrhynchos) was the
only species for which a difference (p=0.083) was observed among years (Tables l6, l7,
and 18).

Other species which differed in abundance across years at the Huron Mountain
Club were the black-capped Chickadee (Paras atricapillus), which was detected most
frequently in 1998 and least frequently in 1996 (p=0.077); brown creeper, which was
significantly less abundant in 1998 than in the other 2 years (p=0.095); and the American
robin (Turdus migratorius), which was more abundant (p=0.100) in 1998 than in the
other 2 years (Tables 16, 17, and 18). The frequency of red-breasted nuthatch (Sitta
canadensis) observations at the Huron Mountain Club was also statistically different
among years, but it is likely that some red-breasted nuthatch vocalizations were confused
with those of white-breasted nuthatches (Sitta carolinensis) in the field. When the
abundance data for these 2 species were combined and tested again across years, the
difference was not significant.

Principal components analysis of forest birds

The absolute frequencies (% of points where species occurred) of the 17 forest
bird species that occurred at a frequency 220% on at least 1 of the study sites were

analyzed with principal components analysis to describe the dominant bird communities

78

on each ownership. The first 3 PCs of this analysis explained 71% of the variability in
the data set, and the first 4 PCS explained 84% of the variability. Principal component 1
accounted for 36% of the total variance and represents a contrast between a bird
community dominated by the black-throated green warbler and ovenbird and a
community dominated by the raven and veery (Fig. 5). The second PC, which accounted
for 18% of the variance in the data set, describes a gradient between a bird community
composed of robins, red—breasted nuthatches, black-capped chickadees, and 9 other
species, and another community consisting of black-throated blue warblers, least
flycatchers, and rose-breasted grosbeaks. Twelve species, including the robin, red-
breasted nuthatch, and black-capped Chickadee had a positive component loading for
PC2, but only five had negative loadings, suggesting a gradient between a very diverse
bird community, and a less diverse community dominated by black-throated blue
warblers, least flycatchers, and rose-breasted grosbeaks. The third principal component
was dominated by the red-eyed vireo and robin in one direction and the black and white
warbler and winter wren in another.

Huron Mountain Club and MDNR sites were grouped similarly along the first
principal component, while Forest Service sites fell in the middle, and timber industry
sites were grouped at the end opposite the Huron Mountain Club and MDNR sites (Fig.
5). Along PC2, Huron Mountain Club sites occupied the positive range of the gradient,
while Forest Service and MDNR sites were grouped together at the opposite end of the
gradient. Timber industry sites occupied an intermediate position along the axis,

suggesting that they cannot be characterized by either of the bird communities

79

 

 

3.0 7
Robin, red-breasted
nuthatch, black- HMC
capped chlckadee,
etc. 2.0 1
T1
1.0 -
N
U HMC
a.
J HMC
0-0 USFS USFS 'n
MDNR TI
tlillack-th‘rflmted-1.0 ‘1 USFS
“6 war er’ MDNR DNR
least flycatcher,
rose-breasted
grosbeak
‘2.0 fl I l U fl
-2.0 -l.0 0.0 1.0 2.0 3.0
PC 1
Black-throated green Raven, veery

warbler, ovenbird

Figure 5. Scores for the first 2 principal components (PC) for songbirds which occurred
at a frequency 220% on at least 1 study site on Michigan Department of Natural
Resources (MDNR), U. S. Forest Service (U SFS), timber industry (TI), and Huron
Mountain Club (HMC) land in Michigan's Upper Peninsula, 1996, 1997, and 1998.

80

represented by PC2. Principal component 3 was less useful for characterizing the bird
communities of the 4 ownership categories because sites in all ownerships ranged widely
in their scores for PC3 (Fig. 5).

Forest bird communities

Eight of the 5 1 species encountered on study sites were cavity nesters. As a
group, cavity nesting birds occurred more often on Huron Mountain Club sites and least
often on MDNR sites. These differences, however, were not statistically significant (Fig.
6). Black-capped chickadees were the most numerous species in the cavity nesting bird
community.

There were no differences (p>0.010) among ownerships in a comparison of bird
species grouped by migratory status (year-round resident, short distance migrant, and
neotropical migrant), but several trends were noticeable. Neotropical migrants tended to
be most abundant on MDNR sites and least abundant on Huron Mountain Club sites (Fig.
7). Short distance migrants, defined as species that winter south of the study area, but
north of the tropics (Blake et al. 1994), were slightly more common on timber industry
sites, and resident species were seen most frequently on Huron Mountain Club sites.

The proportion of each species group out of the total of all species observed on an
ownership followed a very similar pattern as the numbers of observations per sampling
point. On MDNR sites, a larger proportion of the total species observations were of
neotropical migrants, and Huron Mountain Club sites tended to have a smaller proportion
of neotropical migrant species and a larger proportion of cavity nesters than other

ownerships. Of the 27 neotropical migrant species observed, the 3 most abundant were

81

 

 

 

 

 

 

 

 

 

 

 

 

 

  

 

 

E El Noncavity nesters EECavity nesters
: 4.5 1
£1 4 1
D:
3.5 -
5» 3 J
g 2.5 4
g 2 i
E 1.5 7
.8 1 -I
3 0 5
g '01 ,
g MDNR USFS TI
é) Ownership category

Figure 6. Mean number of cavity nesting and noncavity nesting birds per sampling
point on Michigan Department of Natural Resources (MDNR), U. S. Forest Service
(USFS), timber industry (TI), and Huron Mountain Club (HMC) land in Michigan's
Upper Peninsula, 1996, 1997, and 1998.

El Neotropical migrants
E3 Short-distance migrants
I Year-round residents

A
J

 

 

 

 

 

 

 

 

 

 

 

 

 

   

0.5 1
0 JL____,_
MDNR USFS
Ownership category

 

HMC

Mean number observations/sampling point
N

Figure 7. Mean number of birds per sampling point by migratory status on Michigan
Department of Natural Resources (MDNR), U. S. Forest Service (USFS), timber
industry (TI), and Huron Mountain Club (HMC) land in Michigan's Upper Peninsula,
1996, 1997, and 1998.

82

‘..

...: . N 1 tin (I

Inviu.‘

the red-eyed vireo, black-throated green warbler, and ovenbird. The most abundant short
distance migrant was the hermit thrush, and the most common resident species was the
black-capped Chickadee.

Barred owls

Barred owl responses varied widely among the 3 summers during which data were
collected. When average response rates for each site within an ownership were compared
among 1996, 1997, and 1998, there were no significant differences within an ownership
category (Table 19). The highest average response rate for a site within a year was 45%
on MDNR sites in 1998, and the lowest, 0%, occurred on timber industry sites in 1997
when only one site in the timber industry category was surveyed. The second lowest
response rate for a year was 5%, which occurred in 1996 when barred owls were detected
on only 1 of 3 Huron Mountain Club sites sampled. Response rates were slightly higher
on MDNR sites than on the other 3 ownerships in 1997 and 1998, but differences were
not significant in any year (Table 19).

Fishers

Combined over the 2 years in which fisher track count data were collected on
Forest Service and MDNR sites, and for the one year of data collected on timber industry
sites, the average proportion of transects on which at least 1 set of fisher tracks was
observed ranged from 33-80%. The sample size of fisher track count transects was not
sufficient for statistical comparisons of fisher relative abundance; descriptively, however,
the data indicate that portions of all sites in the study were used by fishers in the winter.

Activity indices obtained were 0.55 tracks/km on MDNR sites, 0.63 tracks/km on Forest

83

.280. 8:889 ......

.85. 28% 8.5.58 2.22: 2.33.3.8: 8. .08. 280.6 a: as 5:2 sea 2.. 5.3 2... 2:3 2.. 8 32a?

.00.? 03.00.80 0. .12.... 00. 0N.m 0.08am ----

Ammo. 8:280 0.8 .885. 00§wa>
.0 mafia...“ 83-000 $33.38.! 0... ......» 83.00.80 0.03 2.5.8030 wcofie. 80080:... .0 88. .0. 0.08. 3:50.05.

 

«0.8.. macs...

 

 

 

 

 

 

on»... 0.3... mg... Nah... 0088...... .0 5.3305

00050.00 800..

mum... ..m Qua 0.0 v... v... m6». w... 0.5m ...N .0. 08008. 90 S8.).
mnm... NNN Qua ..o. Qua 5.0. Wan 0d. 0.3. mam.
m8... m. mm as». -..... a... m .0. «.3 m. .N ...NV 30.
mm... a... N... ---- ...: 5.0 «An ..0 Q3 000.

.....Eeocao .m... w .m.m w .m.m m .m... m
RMMNHW... US... .... mme M75...
.0 ......0305 E0830 02.20030

 

.wam. .0...w .30. .000. 6.08.00. 8%.: 9585.3. ... .05.. 6.2... 00.0
0.3.50.2 00.0... .05. ...... >080... .000... Ammo)... 00.>.0m .88”. .m .D .3552. 80.083. .8307. .0 308805 08.50....
:0 8... >00... 00 ...”...m. 88.. 9.0:... ...0.... 08.05.... .08 ...0.me.. 30.0.. .0 Q0. 82.008. .30 00...... .0 8000.02... .0. 0...“...

84

Service sites, and 0.63 tracks/km on timber industry sites. The statistical probability of a
difference in fisher activity among the 3 ownership categories was not significant
(p=0.640).

On Huron Mountain Club sites, where fishers were surveyed with summer scent
post stations, fisher visitations occurred at 4 out of 105 active station nights. The 4
visitations occurred on 2 of the 3 Huron Mountain Club study areas.
Habitat Suitability Analysis

Red-backed salamander

In developing the HSI model for the red-backed salamander, data on stand
structure and composition, and indices of salamander relative abundance collected on
study sites were used to identify the specific variables that are potentially important in
determining the quality of red-backed salamander habitat. The preliminary model that
was developed from literature information and a subset of field data included density of
trees 210.2 cm dbh, average log width, shrub density, and the number of cover objects
measured in width classes of 5-10 cm, 10-20 cm, and 20-30 cm present on the ground as
potential determinants of red-backed salamander habitat suitability. After a statistical
comparison of stands where s2 salamanders were found and stands where >2
salamanders were found during the course of field work (Table 20), overstory tree stem
density and percent canopy cover of shrubs and regenerating trees (0.5-5.0 m tall) were
chosen as components of the final HSI model (Appendix B).

Salamander HSI values ranged from an average of 0.33 on timber industry sites to

0.68 on Huron Mountain Club sites (Table 2]). Habitat suitability index values differed

85

wood

mad—

Snow-60..

 

86

 

5.0. ..m. - .

mwnd ~N.N 00.0-ac. m Sfi vowMonov .80. :50... 800005.05
@mfid anfin on. — mtm. .0. $.MN mm. . Woo... SW“... 00.: we...
20... and .m.w-o>.~ av... mus-mam .E. ....w0h3 W0...
..od NEE 8.:-3.x. 3.0m 2.0-ow... .80. .088”. M”: Q.
mp... 05.00 wade-5.0m 3.0m Wm. .-mm.m~ .60. Em...” 088w
mom... ..~..N omwmb . d. «WNN modvévd. . Em

. . . - . . 0:0. 8.0.5.0 025

N05 0 mm A: m n.— O v 0% A: O.w.uO.N AEV «cw—o... waam
3:. m... ...-on .3 8 . -8 ...... ...0.... pa...
:8 S...” 82...... 3.8 8.8%.... :5 ...0.-... 8.. 09.55
mood mafia w0.~m-0m.0. 3.3 adv-00.0. .80. 3mm. 00.. 3.0.8930
80.0 ew— moméo a: 000.0 058.0.

. .wd mm. cum-ow mm. 9%-? 0.5.0506
wed mm MG. .0 a? CV. nO mgwmcm
m8... .3... EDA-m. .8.... 8.8.8. «£2.88
.30... a: 8.95. max o~..~-oc 2:80:03
n8... 3m mac. 6% 50 own-50m «£80.. 3.05.0.5
hum... 00.5 0.09000 8.3 3.3.05.0 0.5.8. :0... 38m.
88:. 803.05 00....» 0:5. 8...? 00.? 0:5. 80.? 0.50..m>

008.0...0 own-.03. 0:3. .0 083. 08.0.5. 0:0...- .0 0.33.
.o 0:58.... ANN-v.5 mm . Nan-5v

 

05.03.038.28 .0 .0580: 5.0.).

.38... 80080005 :0 0085 09.0% m 0.... .0 88:. 0... ... 00:0.0...0 .8500... .0 5:550... 0:: 000.000. 0.03 0.00—8.5.8
NA 0.053 00:05 ... 0:: 0:00. 0.03 $005.80.... mm 82.3 .05... :. 852.? .859. .0. 80.0., :88 0:0 885. .ON 050..-

87

 

 

wood 5... 0-0 3... m-.. 0.0

one... and . .-o and N . -o 0.0 ”Mom
wwmd 8.0. 0m-m 3.: 3A .00 CNN”
0mm... 5.0m mmém whom .N . -0m 0. 0 o. -m
New... 0.. 0m. mvmfin ...m .m— 0hméw .00 m-..

. . 000.. .00 .8030 .0>00 .0 .00807.
3N... nmfi m.- o.-w~ . .0... Nod-h... ...000 .0003
can... 0...... 3.3-3.3 5.2. N. . .N0-mw.mv .00.. 3000.000.
.00... Emu. 8.00-8.0 3.0. $50.00... .00.. .0..000
80.0 00.0 ...-00.0 00.0 3.0.8.0 .0600. 00.0. =<
who... a.» mam-mm... he... mon-wmd .000000.0..
000.0 am.~w 3.00-3.8 m...» h. .N0-co.~0 0. mA
$66 when 8.3-: .w 8.3. 8.3.-.. .m 0. m-m.o
...»... "v.0. mmdv-wm. 3.: mw.om-wm.0 0. n.o-o

...-.0. .0>00 .00...0>

ma... «....N omdméoN 3.0m >0.-m... .N .00. 0.0...00. ..om

0mm... :0 3.0-30 =3 3.0-000 .... ..0m

wed-o :6 mNW-SN ma... wN.m-N. .m 3.0. 5.00 .03....

.0008 0003.00 00.0.» 0:5. .00.0> 00.0.» 0:3. .00.0> 0.00..0>

0000.0...0 000.03. 000.. .0 0w00~. 0u000>< 000.. .0 0w00y.
.o .0022. Gav-Saw . 00"... 0.

 

000.....0000800. .0 $00.00 000.).

 

....0.-. .00000g0. 00 00.00 300%
m 00. .0 .0020 00. 0. 0000.0...0 .00.....0.. .0 ......0000... 000 000.000. 0.03 2000000000. «A 0.003 .000.. 0. 000 0000.
0.03 ..000080.0. aw .0 0M0.0>0 000 0.003 .000.. 0. .0.00..0> .0..000 .0. .00.0> 0000. 000 .0m00y. .0000. cm 0.00..-

 

 

 

359:8
owed $.38 ow.mmm-oo.w3 mvduu oo.oom-oo.o3 $0qu $50 *0 3N6 =<
35.0 2: Sided «A: ooééod Eu OVA

93% Ea $00.30 S>Oo we 597852

333 a ‘ 05E;

oocouotE away—95‘ 938 mo omcam 03.55. 38m mo omcwm

 

o a: _ a 2
w ..b a n— »NmuE NA ANNHE Nw

 

wfiuflmbwcmfifiam go 858:: 582

 

.383 “succumowfi so woman 38%

N 2: we 288 05 E 8:80th 30:28: no 3.5305 28 63:82 295 flow—SEES NA 80:? $55 E can 958
203 Bovsmfimfim Nw Ho awake? can 20:3 35: E mofiatg 55m: Ho» 32m.» :38 ES mowcmm .389 cm 2an

88

Table 21. Mean suitability index values for each HSI mode] variable, and means and
standard errors of final HSI values for the red-backed salamander on Michigan
Department of Natural Resources (MDNR), U. S. Forest Service (USFS), timber industry
(TI), and Huron Mountain Club (HMC) land in Michigan’s Upper Peninsula.

 

 

 

Ownership category Probability
HSI variables MDNR USFS TI HMC level“
V1 (tree stem density) 0.52 0.48 0.44 0.62 0.433
V2 (rnidstory canopy cover)* 0.68ID 0.46c 0.28° 0.75bd 0.026
V3 (cover object abundance) 1.00 1.00 1.00 1.00 1.000
Final HSI value* 0.59” 0.46c 0.33‘I 0.68” 0.052

SE. of the mean of the HSI 0.056 0.062 0.030 0.061
values

“ Probability levels for tests of differences among ownerships were calculated with the
Kruskal-Wallis one-way analysis of variance (Siegel and Castellan 1988).

*Values on the same line with the same letter are not different (p>0.10) (Kruskal-Wallis
multiple comparison statistic [Siege] and Castellan 1988]).

89

significantly among sites, with the greatest values on Huron Mountain Club and MDNR
sites, and the lowest values for timber industry sites (Table 21). The overall lower scores
on timber industry sites were primarily influenced by the relatively greater amounts of
rnidstory canopy cover (V2), which resulted in low scores for the second model variable.

Huron Mountain Club sites received higher scores (p=0.026) for the second model
variable than timber industry sites. Habitat quality on MDNR and Forest Service sites
was limited by tree stem density (V1), for which values were lower than that considered
optimal, and the amount of rnidstory canopy cover, which was generally greater than the
optimal range. The third model variable, which refers to the abundance of ground objects
which may be used by salamanders for cover, was above the minimum value required for
high quality habitat on all study sites. Therefore, the amount of ground cover objects was
not a limiting factor for red-backed salamanders on any of the 4 ownerships.

Ovenbird

The 2 variables used to calculate the suitability of ovenbird habitat in a given
forest stand are mean basal area and the density of shrubs and saplings. Nearly all stands
had basal areas very near or much above the minimum values considered to provide
suitable habitat (Table 22). Optimum values for shrub and sapling density are defined in
a very narrow range by the model; stands with <2000 stems/ha are assigned a suitability
index (SI) value between 0.75 and 1.0, and above 4000 stems/ha, suitability index values
decline, reaching 0 for stands with >10,000 stems/ha. Timber industry and Forest Service
sites tended to have too many shrubs and saplings to be considered suitable habitat by the

model, while Huron Mountain Club and MDNR sites had more suitable shrub and sapling

90

Table 22. Mean suitability index values for each HSI mode] variable, and means and
standard errors of final HSI values for the ovenbird on Michigan Department of Natural
Resources (MDNR), U. S. Forest Service (U SFS), timber industry (TI), and Huron
Mountain Club (HMC) land in Michigan’s Upper Peninsula.

 

 

 

Ownership category Probability
level“
HSI variables MDNR USFS TI HMC
Vl (basal area)* 1.00b 1.00b 0.98c 1.00b 0.088

V2 (shrub/sapling stem density)* 0.70b 0.32c 0.12b 0.81d 0.048
Final HSI value'“ 0.75" 0.37c 0.18c 0.89" 0.057
SE. of the mean of the H81 values 0.123 0.141 0.069 0.014

a Probability levels for tests of differences among ownerships were calculated with the
Kruskal-Wallis one-way analysis of variance (Siegel and Castellan 1988).

*Values on the same line with the same letter are not different (p>0.10) (Kruskal-Wallis
multiple comparison statistic [Siege] and Castellan 1988]).

91

stem densities. Final HSI scores were higher (p=0.057) for the Huron Mountain Club
than for other ownerships (Table 22).

American redstart

For the American redstart HSI model, the variables related to the percent
overstory canopy cover (V1) and tree stem density (V3) were near 1.0 on most stands
sampled (Table 23). Habitat quality for some stands at the Huron Mountain Club was
limited by a relatively high proportion of coniferous canopy cover (V2) and by a
relatively low density of saplings (V4) compared to MDNR, Forest Service, and timber
industry sites. The resulting HSI values were significantly lower for stands at the Huron
Mountain Club than for stands on other sites, and higher for MDNR and timber industry
sites than Forest Service sites (p=0.041) (Table 23).

Veery

The first variable in the veery lHSI model, the percent of cover type flooded, is not
applicable to the HSI calculation for nonwetland cover types (Sousa 1982). Values for
the second variable, the soil moisture regime, were based on observations made during
data collection in the spring of 1998. Sites were classified as falling into 1 of 3 moisture
regime categories and assigned a corresponding suitability index value. Because all sites
were comprised primarily of upland forest, most stands had relatively dry soils and were
assigned to the lowest of the 3 suitability categories. However, several upland forest
stands had noticeably moister soils, particularly stands located near streams, and these
were subjectively assigned to the intermediate suitability category. No stands occurred in

floodplain forest, therefore none were assigned the highest suitability value for the soil

92

Table 23. Mean suitability index values for each HSI model variable, and means and
standard errors of final HSI values for the American redstart on Michigan Department of
Natural Resources (MDNR), U. S. Forest Service (USFS), timber industry (TI), and
Huron Mountain Club (HMC) land in Michigan’s Upper Peninsula.

 

 

 

Ownership category Probability

HSI variables MDNR USFS TI HMC level“
V] (percent tree canopy 0.90 0.95 0.92 0.95 0.228
closure)

V2 (percent coniferous 1.00” 0.99” 1.00b 0.64“ 0.025
canopy cover)*

V3 (tree stem density) 0.95 0.88 0.90 0.92 0.658
V4 (sapling density)* 0.74“ 0.70” 0.95c 0.30d 0.023
Final HSI value* 0.78” 0.64“ 0.84” 0.31“I 0.041

SE. of the mean of the HSI 0.081 0.041 0.084 0.154
values

3 Probability levels for tests of differences among ownerships were calculated with the
Kruskal-Wallis one-way analysis of variance (Siegel and Castellan 1988).

*Values on the same line with the same letter are not different (p>0. 10) (Kruskal-Wallis
multiple comparison statistic [Siege] and Castellan 1988]).

93

Table 24. Mean suitability index values for each HSI mode] variable, and means and
standard errors of final HSI values for the veery on Michigan Department of Natural
Resources (MDNR), U. S. Forest Service (USFS), timber industry (T1), and Huron
Mountain Club (HMC) land in Michigan’s Upper Peninsula.

 

 

 

Ownership category Probability
level‘

HSI variables MDNR USFS Tl HMC
V1 (Percent of cover type NA NA NA NA NA
flooded)
V2 (soil moisture regime) 0.12 0.11 0.16 0.24 0.392
V3 (percent deciduous 0.18” 0.35” 0.60“ 0.08d 0.070
shrub cover)*
V4 (height of deciduous 0.52 0.58 0.58 0.39 0.516
shrubs)
V5 (percent herbaceous 0.00 0.00 0.00 0.00 0.392
canopy cover)
V6 (height of herbaceous 0.64” 0.49“ 0.49“ 0.23‘1 0.034
canOpy)*
Final HSI value 0.07 0.10 0.16 0.08 0.192

SE. of the mean of the HSI 0.003 0.021 0.032 0.054
values

NA - Variable is not applicable in the habitat evaluated.

’ Probability levels for tests of differences among ownerships were calculated with the
Kruskal-Wallis one-way analysis of variance (Siegel and Castellan 1988).

*Values on the same line with the same letter are not different (p>0.10) (Kruskal-Wallis
multiple comparison statistic [Siege] and Castellan 1988]).

94

moisture variable (Table 24).

The third veery HSI mode] variable (V3) considers the amount of cover provided
by deciduous shrubs and regenerating deciduous trees. Values for this variable were
indicative of moderately suitable habitat on timber industry sites and Forest Service sites,
which had intermediate amounts of rnidstory canOpy cover (Table 5). Huron Mountain
Club and MDNR sites received somewhat lower values for this variable because they had
average shrub cover values below those considered suitable by the model. Values for
variable 4, (the average height of deciduous shrubs) and variable 6 (average height of
herbaceous vegetation), were moderately suitable for most stands in all 4 ownership
categories, although stands at the Huron Mountain Club had slightly lower SI values than
stands at other sites.

The final HSI values for all stands and the average HSI’s for all ownerships were
limited more by the percent of herbaceous canopy cover than by any other variables.
Nearly all stands had too little herbaceous canopy cover to provide suitable habitat, as
defined by the mode]. Final HSI scores averaged near 0 for all ownerships, and
differences among ownerships were not significant.

Yellow-ramped warbler

Although data on the relative abundance of songbirds were collected from 104
forest stands, yellow-rumped warblers were positively identified in only 8% of those
stands, so model development was based on a combination of field data and published
literature. As with the red-backed salamander, the species data was divided into 2 groups.

One group consisted of stands in which there was at least 1 yellow-rumped warbler

95

observation during the 3 years of data collection, and the other was made up of stands in
which no yellow-ramped warblers were recorded. Independent t-tests of these 2 groups
indicated significant differences in shrub densities (p=0.040), height of herbaceous
vegetation (p=0.071), average log width (p=0.039), and log densities (p=0.049) (Table
25).

An initial review of the literature suggested that the percent overstory conifer
cover and the amount of shrub cover might be important habitat variables to consider in
developing an HSI model for the yellow-rumped warbler. Although not statistically
significant in this data set, the proportion of conifer cover also differed between the 2
groups with a probability of 0.11 (Table 25), giving moderate support to the previous
hypothesis that conifer cover may be an important habitat variable. The final yellow-
rumped warbler model based on these data analyses consists of 4 variables: percent
overstory (25 m) conifer cover, average height of mature trees (defined as trees 25 m tall
and 210.2 cm dbh), percent overstory (conifer and deciduous) canOpy cover, and shrub
stem density (Appendix C).

Suitability index values for the first mode] variable (overstory conifer cover) were
greater (p=0.042) on Huron Mountain Club sites than on MDNR sites, due to the high
proportion of hemlock in the overstory at the Huron Mountain Club (Table 26). The
percent overstory conifer cover was the most limiting variable on MDNR and Forest
Service, and was also very low on timber industry sites, all of which had much less
conifer cover than optimal model values. The most limiting variable on timber industry

sites was the density of shrubs and saplings <5 m tall (V4) because understory stem

96

 

97

 

 

E3 8.: :eES 3.: 8.8.82 Es is: 8803ch
33 «SN 8.2.8.: :3 __.$-~_.£ 2.3 523 ms
83 gm saw-$5.. 3m .23..” as swag m3
came 3.: 3.8-Ki 3.8 3.3 -3; Es 55:35 mean
9: .o 8.3 3.3%.: 3.8 one 738 Ase Ems; 95%
32. 3S 2.8-3.2 8.3 2.8-2.2 flee 53:36 9.8
33 m2 3.2-23 «3: «SN-mi 95 Ems; mam
5.8 8.»: coca-8% 3.2: mm 2 -9. Ass 232 seam
33 3.3 8.5-2.8 8.8 Ram-8.: as 232. 8e 8226
93 3.3 gem-8.2 3.3 once-3.2 Ea :mo 8: 8220
Sod 8.92 we -9. $.33 82 2&3
83 3.2— SN-S 222 8...-2 seesaw
53 2.3. 5-8 8.3 87o «imam
036 2.25 Sale 8.35: Sade-o «£32m
«So 38» Sax-S. ~38 83-9. «€338
23 858 98-84 36% 82-2: «£328 8385
$3 3.8 Sen-8.x 2.3 388% 8; Ram
mgofi 5223 2:?» 936 83? 2...; 153m was?» 293:;
oozes—emcee ou§o>< use; we ewcmm owEo>< 28m .6 owcwm
mo 325305 AwflE _N Adana: o

 

285 com men—n33 woman-730:0» «0 59:5: 532

 

.SmB-e “conceaoni 5 woman 38% N 95 no 2on 2:
E cacao-Eu Roumzfim mo QEBEQ Ea .wmfi Ho .82 .82 E 83030 33 8333 wean—2-30:0; _ ammo. a 20:3 "6.8:
E 28 8288 803 wee—Ema ESE-Bozo» o: been? messa- 5 853.2? 2332 e8 82? .82: new mowswm .nm 2an

 

 

 

 

 

98

 

 

:86 and- 86786 8.8-:._ 333'.
EE 68
686 m8 2 .3-86 8... 8. 3.86 6.86 5 2A 8...: ...-«E 63m
«$6 8.8 8.8-8.8 8.8 8.8-8.8 .85 3828a
2 _ _6 2.8 $66.86 3.2 8.886 .8: 3:56
5.66 26 $686 86 $886 36.8... pea .8865
866 26 $686 86 $886 860% gen... .2
886 mm.» n: 35 mm.» 8.8-86 388.com
8.6.6 8.8 8.8-8.8 8.8 8.8-3.8 8 mA
686 8.8 8.8-8.6 8.8 3w: .8 E 36
~86 8.: 8.8-88 86— 89.8; e 36
32 ~98 Bush;
866 86 8.?88 36 8.33 83 £86 .25
6538 5953 2:5» ...—Sm 62:? was.» .336 82? 0362.6;
858:8 away-3a. 256 go owcmm away}. @57- we owcmm
8 $525 due 8 8&3

 

288 5d 6.5333 vogue-30:0» ...o 858:: :32

 

.383 Eugene—65 no woman 38% N 05 ..o 862: 05 5 8:28.26
3036686 mo 559305 23 .32 8 .82 .82 E cotowno 83 .5383 unmask-323% _ 682 8 80:3 $6:de E was
@0988 203 80333 “Vaughan-Bozo» 0: 80:3 26:68 E 8385.» 6352 SM 82g :62: Ba mowcmm .0208 mm 2an

Table 26. Mean suitability index values for each HSI model variable, and means and
standard errors of final HSI values for the yellow-rumped warbler on Michigan
Department of Natural Resources (MDNR), U. S. Forest Service (USFS), timber industry
(TI), and Huron Mountain Club (HMC) land in Michigan’s Upper Peninsula.

 

Ownership category Probability
level“

 

HSI variables MDNR USFS TI HMC

 

V1 (percent overstory conifer 0.1 lb 0.30c 0.2lbc 0.95d 0.042
cover)*

V2 (height of mature trees) 0.98 1.00 0.96 0.99 1.000
V3 (% overstory canopy cover) 0.99 0.98 1.00 1.00 0.392

V4 (density of shrubs/saplings 0.63b 0.31c 0.12C 0.95d 0.028
<5 m tall)*

Final HSI value“ 0.10b 0.25c 0.15c 0.93“I 0.041
SE. of the mean of the HSI 0.058 0.037 0.055 0.037
values

 

“ Probability levels for tests of differences among ownerships were calculated with the
Kruskal-Wallis one-way analysis of variance (Siegel and Castellan 1988).

*Values on the same line with the same letter are not different (p>0.10) (Kruskal-Wallis
multiple comparison statistic [Siege] and Castellan 1988]).

99

densities on these sites were generally much higher than the HSI model’s optimal range.
Values for the second and third model variables (height of mature trees and percent
overstory canopy cover) were consistently high on all study sites, indicating that the
yellow rumped warbler’s minimum requirements for these features are being provided
within all ownerships.

Final HSI values strongly reflect the SI values for the conifer cover variable
because it is given more weight than the other 3 variables in the final HSI equation.
Consequently, habitat suitability indices were highest at Huron Mountain Club sites, and
lowest on MDNR sites (p=0.041). Forest Service and timber industry sites also had
relatively low values for yellow-rumped warbler habitat suitability (Table 26).

Pileated woodpecker

The first variable, tree canopy cover, considered in the habitat model for the
pileated woodpecker was optimal or near optimal for all stands sampled (Table 27). The
density of trees 251 cm dbh and snags 238 cm dbh (V2 and V4) were the most limiting
variables on MDNR, Forest Service, and timber industry sites. Stands at the Huron
Mountain Club had more suitably sized trees and snags, and therefore had higher values
for these model components. Variable 3, the density of suitably sized logs and stumps,
was not limiting to pileated woodpecker habitat quality on any of the sites sampled.
Values for this variable in each stand were far above the minimum values required for
suitable habitat, with the exception of 1 stand. This stand occurred on a Forest Service
site, and had relatively few stumps and logs, although all other stands sampled at the site

had sufficient number of stumps and logs to provide suitable foraging habitat for the

100

Table 27. Mean suitability index values for each HSI mode] variable, and means and
standard errors of final HSI values for the pileated woodpecker on Michigan Department
of Natural Resources (MDNR), U. S. Forest Service (USFS), timber industry (TD, and
Huron Mountain Club (HMC) land in Michigan’s Upper Peninsula.

 

 

 

Ownership category Probability

levela
HSI variables MDNR USFS TI HMC
V1 (percent tree canopy 0.98 0.96 0.99 0.99 0.663
closure)
V2 (density of trees >51 cm 0.12b 0.06b 0.00C 0.56d 0.021
dbh)*
V3 (density of tree stumps 1.00 0.99 1.00 1.00 0.392
and logs)
V4 (density of snags >38 cm 0.04bc 0.03b 0.17c 0.56d 0.052
dbh)*
V5 (mean dbh of snags >38 0.03b 0.01b 0.08b 0.49C 0.060
cm in diameter)*
Final HSI value* 0.01” 0.10” 0.01” 0.36c 0.067
SE. of the mean of the HSI 0.012 0.009 0.013 0.102

values

3 Probability levels for tests of differences among ownerships were calculated with the
Kruskal-Wallis one-way analysis of variance (Siegel and Castellan 1988).

*Values on the same line with the same letter are not different (p>0. 10) (Kruskal-Wallis
multiple comparison statistic [Siegel and Castellan 1988]).

101

woodpecker.

Although some stands met the model requirements for the minimum number of
large snags (V3) needed by woodpeckers for foraging and nesting, habitat quality was
limited in a subset of these stands because the average diameter of large snags was below
values associated with high quality pileated woodpecker habitat. Again, differences in
this model component were most evident between the Huron Mountain Club and the
other 3 ownership categories. Final HSI values were near 0 for MDNR, Forest Service,
and timber industry sites, primarily due to the lack of very large trees and snags. Habitat
suitability index values for stands at the Huron Mountain Club averaged 0.31 (Table 27),
although the number and size of large trees and snags were still the primary factors that
drove HSI values below 1.0.

Northern flying squirrel

The most limiting habitat variable for northern flying squirrel habitat quality for
all 4 ownership categories was the density of snags 230 cm dbh (V3). Even though this
variable was the most limiting factor on Huron Mountain Club sites, it was higher
(p=0.067) on Huron Mountain Club sites than on other ownerships (Table 28). Variable
l, the density of overstory trees, was moderately suitable (0.58-0.68) for all 4 ownerships,
with no significant differences observed. Values for the second mode] variable, the
density of trees 230 cm dbh, were higher on Huron Mountain Club sites than on all other
sites and lowest on Forest Service and timber industry sites (p=0.024). Although
differences were not statistically significant, overstory canopy cover on MDNR sites was

generally much lower than optimal model variables, slightly higher on Forest Service and

102

Table 28. Mean suitability index values for each HSI model variable, and means and
standard errors of final HSI values for the northern flying squirrel on Michigan
Department of Natural Resources (MDNR), U. S. Forest Service (USFS), timber industry
(TI), and Huron Mountain Club (HMC) land in Michigan’s Upper Peninsula.

 

 

 

Ownership category Probability

l vela
HSI variables MDNR USFS TI HMC 6
V1 (density of overstory 0.68 0.64 0.58 0.69 0.376
trees)
V2 (density of trees >30 cm 0.50b 0.37c 0.32c 0.85d 0.024
dbh)*
V3 (density of snags >30 cm 0.11”c 0.06” 0.21c 0.70d 0.067
dbh)*
V4 (overstory conifer cover)* 0.28” 0.47” 0.41” 0.99c 0.093
Final HSI value“ 0.39” 0.39” 0.38” 0.81c 0.099
SE. of the mean of the HSI 0.058 0.016 0.048 0.029
values

 

” Probability levels for tests of differences among ownerships were calculated with the
Kruskal-Wallis one-way analysis of variance (Siegel and Castellan 1988).

*Values on the same line with the same letter are not different (p>0. 10) (Kruskal-Wallis
multiple comparison statistic [Siegel and Castellan 1988]).

103

timber industry sites, and near Optimal at the Huron Mountain Club (Table 28). Final HSI
values were determined to be different among sites (p=0.099), and a Kruskal-Wallis
multiple comparison test showed that calculated habitat suitability was highest at the
Huron Mountain Club (Table 28).

Barred owl

The first variable considered in the barred owl HSI model is the density of trees
with a dbh 251 cm. Stands in which no such trees are present receive an SI value of 0.10,
stands with an average of 25 trees/ha 251 cm dbh receive an SI of 1, and stands with >0
but <5 trees/ha are assigned an intermediate SI value. Tree diameter data in this study
was collected from 3 250 m2 plots per stand, or a total of 750 m2 per stand. Any stand in
which 1 tree 251 cm in diameter was recorded would be estimated to have a density of 13
such trees per hectare. Therefore, because of the plot sizes and sampling methods used,
any stand in which at least 1 51 cm tree was recorded received an S1 of 1.0 for that model
component, and all others received an SI of 0.1 (Table 29).

Another of the barred owl model variables is the average diameter of all overstory
trees. This parameter varied among study sites, with SI values ranging from 0.14-0.50.
Most stands on Forest Service, MDNR, and timber industry sites were at the lower end of
the range and several Huron Mountain Club stands were at the upper end (Table 29).

The third component of the barred owl HSI model is the percent overstory cover,
with stands having 260% assigned a value of 1.0. All stands sampled had 260%
overstory canopy cover, with the exception of one recently thinned stand on a Forest

Service site in which the overstory canopy cover averaged 57%. As a result, this HSI

104

Table 29. Mean suitability index values for each HSI model variable, and means and
standard errors of final HSI values for the barred owl on Michigan Department of Natural
Resources (MDNR), U. S. Forest Service (USFS), timber industry (TI), and Huron
Mountain Club (HMC) land in Michigan’s Upper Peninsula.

 

Ownership category Probability
1 la
HSI variables MDNR USFS TI HMC eve

 

V1 (density of trees >51 cm 0.44” 0.35” 0. 14° 0.90d 0.023
dbh)*

V2 (dbh of overstory trees)* 0.28”° 0.33” 0.26c 0.50d 0.057

V3 (percent canopy cover of 1.00 1.00 1.00 1.00 0.392
overstory trees)

Final HSI value“ 0.31” 0.29” 0.l8° 0.65d 0.024
SE. of the mean of the HSI 0.048 0.029 0.016 0.096

values

 

“ Probability levels for tests of differences among ownerships were calculated with the
Kruskal-Wallis one-way analysis of variance (Siegel and Castellan 1988).

*Values on the same line with the same letter are not different (p>0. 10) (Kruskal-Wallis
multiple comparison statistic [Siegel and Castellan 1988]).

105

model variable was optimal or very nearly optimal in all stands surveyed. Final HSI
values ranged from an average of 0.18 on timber industry sites to 0.65 at the Huron
Mountain Club, with significant differences (p=0.024) among study sites (Table 29).

Fisher

For each ownership category, the lowest SI values determined by the fisher HSI
model occurred for variable 4, the proportion of the overstory canopy cover comprised of
deciduous trees. Suitability index values on MDNR, Forest Service, and timber industry
sites ranged between 0.22 and 0.24 (Table 30), because they had more deciduous canopy
cover than that considered to occur in high quality fisher habitat. In addition, the average
size of overstory trees contributed to the lower HSI values on MDNR, Forest Service, and
timber industry sites than at the Huron Mountain Club.

Relationships between population indices and HSI model output

Spearman rank correlations between population indices for 8 selected indicator
species ranged from -0.559 to 0.594 (Table 31). The fisher was the only species with a
negative correlation between the population index (# tracks/km) and HSI values.
Correlations between relative abundance and HSI scores were rather low for the barred
owl and for the American redstart, a moderately positive correlation was found for the
veery. For the pileated woodpecker, there was a significantly positive correlation
between relative abundance and HSI model output.

The 2 species for which HSI models were developed from project field data had
the highest correlations, indicating that the models are fairly good predictors of habitat

quality on the sites examined in this study. In the 6 of the 8 stands where yellow-rumped

106

Table 30. Mean suitability index values for each HSI mode] variable, and means and
standard errors of final HSI values for the fisher on Michigan Department of Natural
Resources (MDNR), U. S. Forest Service (USFS), timber industry (TI), and Huron

Mountain Club (HMC) land in Michigan’s Upper Peninsula. Probability levels for tests

of differences among ownerships were calculated with the Kruskal—Wallis one-way
analysis of variance (Siegel and Castellan 1988).

 

 

 

Ownership category Probability

level"
HSI variables MDNR USFS TI HMC
V1 (percent tree canopy 1.00 0.98 0.96 0.99 0.734
closure)
V2 (average dbh of overstory 0.53”° 0.59” 0.50° 0.80d 0.057
trees)*
V3 (tree canopy diversity) 1.00 0.83 0.92 0.83 0.326
V4 (percent of overstory with 0.22” 0.23” 0.24” 053" 0.075
deciduous trees)*
Final HSI value* 0.18” 0.18” 0.18” 0.46° 0.099
SE. of the mean of the HSI 0.013 0.018 0.015 0.141
values

 

a Probability levels for tests of differences among ownerships were calculated with the

Kruskal-Wallis one-way analysis of variance (Siegel and Castellan 1988).

*Values on the same line with the same letter are not different (p>0.10) (Kruskal-Wallis

multiple comparison statistic [Siegel and Castellan 1988]).

107

Table 31. Spearman rank correlations between mean relative abundances per study site
and mean HSI values for species surveyed on Michigan Department of Natural Resources
(MDNR), U. S. Forest Service (USFS), timber industry (TI), and Huron Mountain Club
(HMC) land.

 

 

Species Correlation coefficient 2-tailed probability level
(r.)
Red-backed salamander* 0.594 0.042
Ovenbird 0.448 0. 144
American redstart 0.168 0.602
Veery 0.347 0.269
Yellow-rumped warbler* 0.569 0.053
Pileated woodpecker* 0.498 0.099
Barred owl 0.294 0.354
Fisher“ -0.559 0.1 17

 

* Relationship is statistically significant (p30. 10).

a Correlation calculated between fisher activity indices (tracks/km) on MDNR, USFS, and
TI stands and HSI values.

108

warblers were observed, the SI value for the conifer cover variable was >0.75, which
contributed strongly to the significant positive correlation (rs=.569, p=0.053) between

relative abundance and HSI values.

109

DISCUSSION
Vegetation and structural attributes

The divergent management goals of private timber companies and the Huron
Mountain Club are reflected in several forest stand characteristics that varied most widely
between Huron Mountain Club and timber industry sites. In contrast, attributes on
MDNR and Forest Service sites tended to fall in the middle of the range of values
measured. These relationships are evident in the statistical comparisons and in the
principal components representation of the 4 ownerships (Tables 5 and 6, Fig. 3). For
example, the significantly lower basal area on timber industry sites is an expected result
of the timber industry’s management goals of producing and harvesting timber.
Conversely, many of the differences that distinguish the Huron Mountain Club from other
ownerships, such as larger tree diameters and a sparse understory, are the result of the
relatively unmanipulated condition of the land and the unique management history of the
area. Logs at the Huron Mountain Club consist almost entirely of trees that have fallen
over naturally, and the larger width of logs there reflects the greater size to which trees
are able to grow in the absence of timber harvesting.

The proportion of hard mast producing trees in the overstory was another disparity
between timber industry and Huron Mountain Club sites. Beech was the dominant hard
mast producing tree observed, and it was significantly more abundant on timber industry
sites than on Huron Mountain Club sites (Table 7). As a tree species with low
commercial value, beech is generally not removed during sawlog thinning on timber

industry sites, and therefore beech trees continue growing until being clearcut. The

110

Huron Mountain Club is located beyond the western front of beech’s geographic range,
and therefore beech does not occur at the Huron Mountain Club (Davis et al. 1991). The
only hard mast producing trees observed at the Huron Mountain Club were oaks (Quercus
spp.), which occurred in some drier, higher elevation northern hardwood stands.

For other vegetation characteristics, Huron Mountain Club and timber industry
sites had similar values, and MDNR and Forest Service sites were at the other end of the
range of values observed. Examples are the larger stump sizes and log diameters on both
private ownerships (Table 5). Similarities between Forest Service and MDNR sites are
also evident in the principal components analysis, which shows the clustering of these
sites in terms of basal area, and standing and fallen dead material (Fig. 3). Another
interesting result was that bigtooth aspen was recorded only at the Huron Mountain Club
and not on any other ownerships. This was a result of treefall gaps at the Huron
Mountain Club, where the death of a large tree initiates a gap in the overstory, forming a
pocket in the forest that exists in an earlier successional stage than the surrounding forest.
On other ownerships, aspen was most likely eliminated through succession as forest
stands matured, and opportunities for aspen regeneration were rare.

Forest stand conditions in this study should not be viewed solely in terms of their
current management. Although land on MDNR and Forest Service study sites has
generally been under the same ownership since the 19203, 3 timber industry stands
sampled were purchased from a private owner by Shelter Bay Forests in the 19708.
Timber management activities on stands sampled on MDNR, Forest Service, and timber

industry sites have consisted primarily of intermediate thinnings during the last 60 years.

111

Several MDNR stands had not been out since the stand was established, and 4 stands on
Shelter Bay and Mead sites have not been cut in the last 40 years. In a forest type that is
generally managed as uneven-aged, the concept of “stand age” is generally less relevant
than in other forest stands (Frelich and Lorimer 1991), and most stands were of
indeterminate age beyond the fact that they were established after early 20th century
logging.
Wildlife habitat relationships

Red-backed salamanders

Although statistically significant differences in salamander relative abundance
among ownerships were not documented (Tables 8, 9, 10, and 11), the vegetation
conditions associated with each ownership can help explain the relative differences in
salamander relative abundance that were observed. The relationship between the density
of overstory trees (210.2 cm dbh) emerged as one of the strongest indicators of
salamander relative abundance on our study sites (Appendix B). This result has strong
support from red-backed salamander research conducted in other hardwood forest
ecosystems. Monti (1997) found a positive relationship between salamander relative
abundance and tree stem density in an oak-pine forest in Maine, and suggested that the
association may be related to the amount of litter that is produced, with stands that have a
high density of growing trees producing more litter than fewer large trees would produce
in a stand. Numbers of red-backed salamander were also positively associated with tree
stem densities in Pennsylvania hardwood forests (Rodewald and Yahner 1999) and in

New England oak forests (Brooks 1999). Interestingly, all 3 studies used artificial cover

112

objects to survey amphibian populations.

Despite relatively low tree stem densities at the Huron Mountain Club, slightly
greater numbers of salamanders were observed on those sites. Among all stands sampled,
understory canopy cover and shrub stem densities were negatively related to salamander
abundance, and the lack of understory canopy cover at the Huron Mountain Club may
partially explain the slightly greater relative abundance of salamanders observed there.
The negative relationship between salamander relative abundance and the amount of
rnidstory canopy cover may be because less moisture is retained by the litter and soil on
sites where there is an abundance of herbaceous and shrub cover, which may result in a
less favorable habitat for salamanders (Welsh and Droege 2001). Midstory canopy cover
may also be correlated with microhabitat characteristics that were not measured, such as
nutrient ratios in the soil or daily soil moisture fluctuations.

The positive relationship between numbers of salamanders detected and high tree
stem densities suggests that younger or regenerating forest stands may provide suitable
salamander habitat. However, several studies have found that terrestrial salamanders,
including Plethodon cinereus, are negatively impacted by intensive silvicultural
treatments such as clearcutting, and have attributed the results to the drier surface
conditions caused by the removal of canopy cover and litter (Pough et a1. 1987, Herbeck
and Larsen 1999, Howard and Caschetta 1999, Rodewald and Yahner 1999).

Because salamanders only used a very small proportion (1.6%) of all cover objects
turned over during surveys, it was assumed that cover objects were not limiting to

salamanders in the habitats examined. Although the importance of woody debris to

113

terrestrial salamanders has been documented and widely acknowledged (Test and
Heatwole 1962, J aeger 1980, Grialou et al. 2000), some studies (Rodewald and Yahner
1999, Aubry 2000) have failed to find an association between salamander numbers and
the amount of woody debris in their habitat. However, it is possible that certain forest
activities where large amounts of dead woody debris may be removed from a stand, such
as collection of dead and downed woody material for firewood, could reduce the amount
of woody debris below a critical threshold. Thus, it should not be assumed that habitat
quality is entirely independent of the amount of woody debris in a stand, but rather that
woody debris was not limiting to red-backed salamanders in the habitats sampled.

Comparison of ground transect searches and cover boards for surveying

salamanders

Ground transect searches and the artificial cover boards were useful methods for
measuring the relative abundance of red-backed salamanders in northern hardwood
forests in the Upper Peninsula of Michigan. Salamander relative abundance for both
methods combined ranged from 73 salamanders/ha to 229 salamanders/ha during summer
surveys on the 4 ownerships investigated. In comparison, Heatwole ( 1962) attempted a
complete count of red-backed salamanders and measured 8900 salamanders/ha in a
Michigan northern hardwood forest. Test and Bingham (1948) recorded 496
salamanders/ha during a surface census in a Michigan hardwood forest.

Although each method used in this study produced somewhat different results,
they were in agreement in that the Huron Mountain Club showed a trend of having more

salamanders than the other ownerships. The results of both methods should be viewed

114

with caution in 1997 when only 1 timber industry site was represented, and especially in
the fall, when only 20 stands were sampled before the first snowfall.

It appears that the boards may require an acclimatization period on a site before
they reach their full potential as a survey tool. When cover board searches were
conducted during summer, 1997, none of the boards had been in place more than 8
weeks, which may account for the comparatively low number of salamanders observed
under boards (Table 8). One reason for this may be that it takes many weeks for the
boards and the litter beneath them to decay to the point that they provide a suitable
microhabitat for red-backed salamanders. Some time may also be needed for
salamanders to locate the boards within their territories and to begin using them for cover.
Therefore, cover board survey results may have been biased during the first summer the
boards were out. By the time fall surveys began, cover boards had been in place for at
least 14 weeks, and boards placed on the Shelter Bay sites, which were added to the study
after fall data collection, had been in place for 7 months when the first surveys took place
on those sites in 1998.

The relative abundance of salamanders determined with each survey method
fluctuated among the spring 1997, fall 1997, and spring 1998 sampling periods (Tables 8,
9, and 10). Although it may appear that there was an overall increase in red-backed
salamander populations between 1997 and 1998, the increase is most likely due to the fact
that boards had only been in place for approximately 6 weeks when the first surveys took
place, and probably became much more attractive to salamanders as the litter beneath

them decayed during the first year the boards were on the ground.

115

There were generally low correlations between number of salamanders observed
with artificial cover board and ground transect search methods. One reason may be that
relatively few salamanders were observed under cover boards; thus, there was little
variation (14 salamanders/stand) among stands, resulting in relatively weak correlations.
In addition, salamander use of cover boards was generally lowest during the first year
boards were used and increased after the boards had been in place for a few months,
while ground transect search results were not subject to the same bias. Ground searches
of cover objects revealed more salamanders than the cover board method, however, the
time required to perform the ground searches according to the protocol used was also
substantially greater. For studies of red-backed salamander populations where data will
only be collected one time from a particular study site, ground transect searches like the
ones used in this project are a more efficient method than cover boards. However, when
population surveys will be repeated over time, the cover board method offers the
advantage of requiring less time than ground transect searches. A second advantage of
using artificial cover objects is that it allows researchers to standardize the amount of
effort used to survey each site since the number and size of objects searched does not vary
from site to site as it does when ground transect searches are used. Finally, cover boards
searches can be accomplished with minimal habitat disturbance compared to other
available survey methods.

Forest birds

Forest bird communities on Forest Service and MDNR sites were relatively

similar, while timber industry and Huron Mountain Club land were structured very

116

differently, based on principal components analysis and nonparametric comparisons
(Table 15, Figs. 5-7). Forest Service and MDNR sites were characterized by a bird
community that included the black-throated blue warbler, least flycatcher, rose-breasted
grosbeak, black-throated green warbler, and ovenbird, but was not strongly dominated by
any 1 or 2 species. The Huron Mountain Club bird community included 2 of the most
abundant birds surveyed, the ovenbird and the black-throated green warbler, and a
secondary assemblage of cavity nesting birds. On timber industry sites, the American
redstart and veery, species with preferences for younger stands, and the raven, a habitat
generalist, were important species.

Although black-throated blue warblers are expected to inhabit more mature
forests, they prefer sites dominated by sugar maple and with a dense shrub layer (Binford
199] , Bourque and Villard 2001), which explains the small role they play in the Huron
Mountain Club forest bird community. The blackbumian warbler is a species facing
expected declines in Michigan, due to harvesting of uneven aged mature coniferous
forests (Doepker et a1. 1992). Although blackbumian warblers (Dendroica fusca) were
encountered infrequently during field sampling, they were generally more common at the
Huron Mountain Club than other study sites, where they were presumably attracted to the
conifer component of the hardwood forests (Tables 6 and 7). This is a potentially
important finding, as it indicates the potential of unique areas such as the Huron
Mountain Club to provide a refuge for species whose primary habitat is threatened.

The distribution of birds grouped by migratory status among ownerships was

slightly skewed in favor of neotropical migrant bird species on MDNR and Forest Service

117

sites, and resident and short distance migrant species were relatively more important on
timber industry and Huron Mountain Club sites. This result is most likely due to the
influence of a few very abundant species, such as the black-throated green warbler,
ovenbird, and red-eyed vireo, that accounted for the majority of individuals in the
neotropical migrant group. In addition, 5 of the 8 birds classified as year round residents
are also cavity nesters, which tended to be more prevalent in the older forests found at the
Huron Mountain Club. Thus, neotropical migrants as a group may not exhibit
associations with forest characteristics that reflect overall management approaches.
However, Noon et al. (1979) determined that mature and undisturbed forested habitats
contained more regionally rare species, in terms of absolute numbers and proportion of
species present, that did not occur in early successional or disturbed forests.

Barred owls

The index of barred owl population used in this study was highly variable among
years and among locations within a year. Barred owl responses may have been more
influenced by weather conditions and time of night than had been expected, and sampling
may not have been intensive enough to accurately describe the population. Barred owl
responses to taped broadcast calls have been found to vary with weather, environmental
conditions, and time of night; they are particularly influenced by moon phase, occurring
most frequently when the moon is visible (i.e., a full moon with no cloud cover) (T akats
and Holroyd 1997). In this study, however, barred owl surveys were not controlled for
conditions other than precipitation.

No strong conclusions can be drawn about the effects of management approaches

118

on barred owl populations other than the fact that barred owls were distributed among all
study sites. These results might also be a function of the resolution of the data analyzed,
which may have been too fine to demonstrate the true variability in the population.

Fisher

There was not sufficient population data to determine if fisher habitat use differed
among ownerships. However, observations of fisher tracks necessarily indicate that the
fishers were moving, and because fishers are generally solitary in the winter (Powell
1982), fishers whose tracks were observed were either foraging on the study sites or
moving through the study sites en route to another part of their home range. As with the
barred owl, the scale of the analysis may have obscured some population characteristics
that would be apparent in a broader scale of analysis.
1118] model performance

The model developed for the red—backed salamander exhibited a good fit to the
data used to develop the model, based on the significant correlation with salamander
relative abundance. Possible sources of error in the red-backed salamander model include
population variation due to the effects of stochastic events (e.g., weather conditions)
during the 2 years of population monitoring, potential bias in population sampling
methods (e.g., differential use of cover objects in response to moisture conditions), and
unidentified multivariate relationships in the data. Also, HSI models are driven by
measures of habitat structure and composition, and do not consider the impact of
interspecific relationships or density-dependent factors on a population (Schamberger and

O’Neil 1986).

119

In New Brunswick, Canada, ovenbird densities and reproductive success were
lower in selection cuts than in uncut stands (Bourque and Villard 2001). In this study, the
lowest ovenbird numbers occurred on timber industry sites (Table 15) which presumably
have had more intensive cutting. Leaf litter depth and biomass of invertebrates in the
litter have also been positively associated with ovenbird territory establishment (Burke
and No] 1998). Although not significant, the correlation between ovenbird relative
abundance and HSI values was more positive than for other species, and as a model with
2 simply measured variables, it may be especially useful to forest managers.

As might be expected for an early successional species, calculated habitat quality
for the American redstart was highest for timber industry sites and lowest for the Huron
Mountain Club (Table 23). There was a positive but weak correlation between redstart
numbers and HSI values, although average HSI values followed the trend for redstart
relative abundance (Table 15) for all ownerships except Forest Service sites.
Discrepancies between model output and redstart abundance may be due to the fact that
there are contradictions in the published literature regarding redstart habitat requirements
(Minnis and Haufler 1994), and the model was not empirically based. Also, the model
was developed for deciduous, coniferous, and mixed forests, but the parameters that
define high quality habitat may not be the same in each forest type.

Although there was a small positive correlation between veery observations and
calculated HSI values, all ownerships had HSI values <0.2, indicating poor habitat quality
(Table 24). Nonetheless, the veery was not as uncommon on any of the ownerships as

might be expected from such low HSI values. Thus, the existing veery model may not be

120

a very useful tool for evaluating veery habitat quality in northern Michigan hardwood
forests. The one model variable that most strongly contributed to the low HSI values on
all study sites was the herbaceous canOpy cover component. The model specifies at least
90% herbaceous cover for a site to receive an overall HSI of 1.0, but none of the sites
sampled even approached that. The definition for this variable is perhaps too restrictive,
and might be expanded to include woody canopy cover on the forest floor (i.e., tree
seedlings) as a contributor to veery habitat quality.

Because the yellow-rumped warbler model results were based on the same data
used to develop the model, the significant positive correlation between yellow-rumped
warbler observations and HSI values only indicates how well the model fits the data. The
high HSI values for yellow-rumped warblers at the Huron Mountain Club were mainly
driven by the strong influence of conifer cover in the warbler model (Table 26).
Generally, mature hardwoods and immature hardwoods are avoided by the yellow-
rumped warbler (Howe et al. 1995), but yellow-rumped warblers in Michigan have also
been reported to use northern hardwoods more often than is commonly believed (Eastman
1991). Even though the Huron Mountain Club forests where most yellow-rumped
warbler observations occurred do have a significant conifer component, there may be
other factors unique to that area, understory vegetation species composition, that are
important to yellow-rumped warblers and should be built into the model.

Four species associated with mature forest (pileated woodpecker, northern flying
squirrel, barred owl, and fisher) had significantly higher HSI values on Huron Mountain

Club sites than on other ownerships (Tables 27, 28, 29, and 30). This, together with the

121

generally positive correlations between pileated woodpecker numbers and barred owl
responses, gives support to the utility of these 2 models for measuring habitat quality.
However, the small number of pileated woodpecker observations may mean that the
significant correlation was spurious and not reflective of a biological relationship. Negri
( 1995) reported that the pileated model accurately described pileated woodpecker
requirements in the Upper Peninsula, but validation was limited by the large scale
required to obtain multiple pileated woodpecker observations. A pileated woodpecker
model specific to Great Lakes region has also recently been developed (Felix et al. 1999),
and a logical next step in model validation would be to compare the performance of the 2
models on the same data set.

In attempting to evaluate the fisher HSI model in the Upper Peninsula, Thomasma
et al. (1991) found that habitat predicted to be high quality by the model was used more
by fishers than predicted low quality habitat, and they recommended using the existing
fisher model for habitat evaluations. In this study, a likely reason for the negative
relationship between the calculated HSI and the population index for the fisher is that
among the 3 ownerships tested (MDNR, Forest Service, and timber industry), there was
relatively little variability among suitability index values for the 4 model variables and
among the final HSI variables (Table 30). There was much more variability among
activity indices across study sites, and the HSI model may not have been sensitive enough
to reflect those differences if they were the result of differences in habitat quality. The
scale at which fisher surveys were conducted may also have contributed to the poor

relationship between HSI scores and activity indices. Fisher home ranges may encompass

122

16-31 km2 (Arthur et al. 1989), so even though HSI scores may indicate exceptional
habitat quality, the sampling area (12-19 kmz) may encompass only 1 or 2 fisher
territories, and population surveys may only indicate presence or absence of fishers.

Paragi et a1. (1996) reported that fisher natal dens and resting sites in Maine were
located almost exclusively in the cavities of large mature hardwood snags. The existing
fisher HSI model evaluates the diameter of overstory trees, but does not consider snags.
However, Paragi et al.’s recommendation to maintain a supply of hardwoods >40 cm dbh
corresponds well to the model’s requirement for an average overstory tree dbh of 38 cm
in high quality habitat.

Stand level variables may be important in determining fisher habitat quality if they
are averaged over a large enough area. Carroll et a1. (1999) found that measurements of 3
stand variables (overstory canopy closure, percent conifer cover, and hardwood tree
diameter) aggregated over a regional scale, rather at local scale (0.05 ha) were the best
predictors of fisher distribution in California. If the results are applicable to Michigan,
the Huron Mountain Club would be expected to have better fisher habitat than the other
ownerships, which all had smaller tree diameters and less conifer cover. Carroll et al.’s
findings also emphasize the importance of evaluating fisher habitat quality at the scale of
the fisher’s home range, as was done in this study.

Because the HSI models developed for the red-backed salamander, yellow-
rumped warbler, and northern flying squirrel have not yet been used in northern
hardwood forests outside of the study area, they should be validated using a framework

such as that prescribed by Roloff and Kemohan (1999). Their protocol outlines 7 criteria

123

for testing the reliability of an HSI model, including modeling at the scale of the animal’s
home range, evaluating a broad range of habitat quality, and using an appropriate measure
of species response to habitat quality. The process used to build the salamander model at
least partially met 6 of these criteria. However, the model could be strengthened by
testing it with a population data set that spans at least 3 years to reduce the impact of
stochastic events, and by evaluating stands representing a wider range of habitat
conditions, such as heavily thinned northern hardwood stands with slash and stands with
very little woody debris.

The models developed for the yellow-rumped warbler, and for the northern flying
squirrel in particular, are only initial attempts at modeling wildlife habitat relationships in
northern hardwood forests. The sample size of yellow-rumped warbler observations was
low, and the published research used to supplement the field data was sparse. Thus, these
models will need more extensive testing to be useful to natural resource managers.

Another factor that should be investigated before basing management decisions on
HSI results is the relationship between the population data used to develop the model and
other demographic characteristics, such as reproductive success and survival. Following
Van Home’s (1983) assertions that density is not necessarily an indicator of habitat
quality, researchers have become more cautious about testing the assumption that HSI
output corresponds to actual species fitness. For example, Breininger et al. (1998) tested
an HSI model that was based on relationships between species density and habitat
characteristics, and found that under certain habitat conditions, HSI scores were high,

Florida scrub-jay (Aphelocoma coerulescens) density was high, but mortality exceeded

124

reproductive success due to predation. Although this assumption may be very time
consuming and difficult to test, it should at least be borne in mind when conducting a

habitat evaluation or using the results.

125

CHAPTER 2 - Landscape Scale Wildlife Habitat Characteristics and Relationships

INTRODUCTION

Movements and habitats of forest wildlife species often extend beyond the
boundaries of a single forest stand. For many wildlife species, stand level habitat
evaluations may only partially describe a species’ response to its environment, and
broader scale landscape evaluations are needed to account for a larger set of habitat
variables. Fine filter evaluations that assess stand level habitat attributes may also be
time consuming to measure across a large management area such as a national or state
forest, and identification of larger scale landscape attributes that contribute to habitat
selection may facilitate wildlife habitat quality monitoring, especially when multiple
species are being considered. For example, a habitat assessment for a group of
ecologically similar species, such as edge sensitive forest songbirds, may include fairly
large scale habitat characteristics, such as the size of the forested area, the proportion of
different forest types in the management area, and the degree of forest fragmentation in
the landscape under management. This type of analysis which assesses the range of
habitat conditions for a group of wildlife species can be referred to as a coarse-scale
habitat assessment (The Nature Conservancy 1982).

Landscape characteristics arise through natural and human influenced processes.
In northern hardwood forests, natural disturbances, primarily in the form of windthrow,
can set back succession and create new patches with complex shapes. In forests where

human influence is absent, such processes may result in a landscape with a matrix of very

126

large patches and a much smaller proportion of small patches (Mladenoff et al. 1993).
Natural resource management activities such as timber harvesting, herbaceous plantings,
and road building may also directly affect the arrangement, size, shape, and number of
patches on the landscape. Depending on the management goals being pursued, the results
may range from a diverse but highly fragmented landscape to one that has a closer
resemblance to an undisturbed landscape. Although natural resource managers today may
recognize the potential for both direct and cumulative effects of landscape changes on
wildlife habitat quality, landscapes are a reflection of past activities that may have sought
different objectives and operated under different assumptions than exist today. For
example, guidelines for wildlife management in Michigan in the early 1970s specifically
stated that managers should promote shade intolerant forest types and prevent conversion
to mature forest (Michigan Department of Natural Resources 1973).

Furthermore, the tools for making decisions at a landscape scale and a strong
information base on wildlife responses to landscape characteristics are not yet available to
many managers. Despite the fact that many researchers have documented relationships
between landscape parameters and wildlife habitat quality, many tools for habitat
assessment were designed before landscape management principles were fimily in place.
Therefore, HSI models that have traditionally evaluated local habitat variables but do not
address the importance of landscape characteristics may be inadequate for conducting a
thorough evaluation of habitat quality. One of the biggest reasons that these stand-level
models are “incomplete” is that there still remains a very large need to identify landscape

level requirements of individual species. Although HSI models are valuable for

127

evaluating wildlife habitat at a small spatial scale, habitat suitability analysis can be
integrated over a range of spatial scales, depending on the typical home range size of the
species of interest, the extent of the area to be managed, and the total number of species
for which habitat will be evaluated.

Knowledge of the current condition of a planning landscape, including land
managed by other entities, may help natural resource managers evaluate the potential
effects of management activities and refine their management objectives to better meet
overall goals. In addition, identification of larger scale landscape attributes that
contribute to habitat selection may also facilitate wildlife habitat quality monitoring.
Using a coarse filter approach, this chapter will compare the characteristics of landscapes
within 4 different ownership categories, identify relationships among landscape variables
and wildlife population indices, and relate results to differences in wildlife relative

abundance in each area.

128

METHODS

Population survey methods for red-backed salamanders, forest birds, barred owls,
and fishers were described in Chapter 1. Relative abundance data for these species for
1996, 1997, and 1998 were used in landscape analyses.

Landscape analyses

Land cover data were obtained from the Michigan Department of Natural
Resources (MDNR) for each study county. These data consisted of 1991 Landsat
Thematic Mapper satellite imagery of the Upper Peninsula that was classified for a
project on deer winter habitat use (MacLean Consultants, Ltd., no date). The imagery
was recorded at a 30 m resolution, meaning that the smallest identifiable element is a 30
m x 30 m square cell. The land cover data came in ERDAS format, which is a raster
based GIS system, and was converted to Arc/Info format using the IMAGEGRID and
GRIDPOLY commands.

The classification system for the land cover data identified a total of 13
nonconiferous cover types and 17 coniferous categories (Table 32). The
agricultural/cropland category was combined with the herbaceous openland category for
this analysis because both are nonforested cover types that made up a small proportion of
the area evaluated. Because the satellite data had been classified for a winter deer habitat
study, the coniferous categories were emphasized in the classification system. Most of
the coniferous categories were subdivided based on the amount of canopy closure, but the
nonconiferous categories were not distinguished in this way. Therefore, the 2 mixed

conifer canopy closure categories (<70% canopy closure and >70% canopy closure) were

129

Table 32. Categories used in the classification of 1991 Upper Peninsula Landsat
Thematic Mapper satellite imagery (MacLean Consultants, Ltd. no date).

 

N on-coniferous cover types

Coniferous cover types

 

Urban

Nonvegetative
Agricultural/cropland
Herbaceous openland
Shrubland

Northern hardwoods

Oak

Aspen/birch

Lowland hardwoods

Dry hardwood/conifer mix
Wet hardwood/conifer mix
Wetlands

Water

Red pine
Jack pine
White pine

Other (mixed) pine

Tamarack
Hemlock

Black spruce

White spruce

Balsam fir

White cedar

Mixed conifer

130

<70% crown closure
>70% crown closure
<70% crown closure
>70% crown closure
<70% crown closure
>70% crown closure
<70% crown closure
>70% crown closure
<70% crown closure
>70% crown closure
<70% crown closure

>70% cr wn closur

also merged to make the level of discrimination within each category more uniform.

Road coverages were created from data on trails, public streets, county roads, and
highways in 1978 MIRIS base map files provided by the MDNR. The MIRIS base map
data for each county was obtained in Intergraph Design Format, and the roads data layer
was then extracted and converted to Arc/Info format with the IGDSARC command.

Each study site was defined as a landscape; however the boundaries of the study
area landscapes were defined separately for each species evaluated, depending on the
amount of area beyond survey points species were likely to use. For the Huron Mountain
Club study site, all 3 sampling areas were treated as a single study site landscape, unlike
the stand level habitat and species abundance analyses presented in Chapter 1, where the
3 sampling areas were analyzed as individual sites within the Huron Mountain Club
ownership category. Arc/Info was used to prepare digital representations of each site.
For each of the 10 study sites, landscape boundaries were first created in ArcInfo by
generating a polygon with the UTM coordinate points associated with red-backed
salamander, forest bird, and barred owl, and vegetation sampling points located on the
outer boundary of the study site as the vertices (Fig. 8). Survey points had been
established across study sites in a superficial grid pattern, approximately 1.6 km apart.
Generated coverages were projected from UTM coordinates to state plane coordinates,
which was the coordinate system of the satellite imagery and roads data.

A buffer was then added to the exterior of the polygon to capture area outside
survey points that was potentially used by the wildlife detected. The size of the buffer

depended on the species for which the landscape was being defined. For barred owls, the

131

800 m buffer

 

“““““
ooooo
1C a o

' O
P a n
.- °°°
I I O
-----

~6.4 km

.....

ooooooo

....
....

 

  
  

 

80 m buffer

 

O = vegetation sampling points

Figure 8. Illustration of sampling points and buffers used to define landscape boundaries
for study sites in Michigan’s Upper Peninsula.

132

buffer was 800 m, corresponding to the radius of a 200 ha home range sized circle
centered on the survey point (Fig. 8). The average barred owl home range is between 1 18
and 282 ha (Elody and Sloan 1985), so an intermediate value of 200 ha was chosen as the
basis for determining buffer size. The same landscape definition was used for the
pileated woodpecker, based on Bull and Meslow’s (1977) home range size estimates of
130-243 ha and Kilham’s (1959) estimate of 70 ha. Home ranges estimates for songbirds
chosen as focal species for this project range from 0.04 ha to 2.5 ha (Table 2). For
songbirds, an 80 m buffer was used, corresponding to the distance from a survey point to
the edge of a 2 ha home range circle placed over the survey point (Fig. 8). The landscape
boundaries for forest birds were also used for red-backed salamanders, and the barred owl
landscape definition was used for fisher landscape analysis. Buffered study site polygons
were clipped with the satellite imagery and the roads data for each county to obtain a land
cover and roads coverage for each study site.

For the barred owl, landscape analyses were performed at 2 spatial scales because
the barred owl home range is large enough (approximately 200 ha) to show variations in
the land cover data, yet the home range is smaller than the area across which population
data were collected (12-19 km”). The purpose of the home range scale of analysis was to
document landscape characteristics that may impact barred owl populations at a spatial
scale intermediate between the stand and landscape levels. A similar analysis of songbird
home range sized areas was discontinued because there was little variation evident in land
cover within 2 ha circular areas. Fisher home ranges are at least as large as each study

site, so the area defined in the study site scale of analysis was considered equivalent to a

133

home-range level analysis for fishers. For the red-backed salamander, landscape
characteristics were not analyzed at the home range level because of their relatively small
home range size (3—5 m2).

Separate coverages were created for each barred owl survey point and analyses
were performed on areas equivalent to a home range sized area centered on barred owl
sampling points. The GENERATE command in Arc/Info was used to create circular
coverages, representing hypothetical home ranges centered on each survey point.
Generated coverages were clipped with corresponding satellite imagery coverages to
isolate land cover polygons within the hypothetical home ranges.

The Patch Analyst software extension to ArcView 3.0a was used to calculate
landscape metrics for each study site and for hypothetical barred owl home ranges. Of
the 38 patch, class, and landscape metrics that Patch Analyst can calculate, a subset of
variables was chosen for analysis. A patch was defined as a polygon that had a cover type
classification different from adjacent polygons, and each land cover type designation was
considered a class. Selected metrics calculated for the study site level of analysis
included class area, patch richness, mean patch size, median patch size, edge density,
total patch edge, mean shape index, area-weighted shape index, fractal dimension, area-
weighted fractal dimension, interspersion and juxtaposition index, Shannon’s diversity
index, and Simpson’s diversity index. These metrics were chosen based on information
in published literature, to minimize redundancy among variables, and to quantify
relationships that might exist between spatial patterns and ecological functions (O’Neill

et al. 1997). Analyses of barred owl home ranges were focused on cover type

134

composition (class area) and total edge because of the relative homogeneity within circles
<200 ha.

Explanation of landscape metrics

Shape index is expressed as a unitless value 2 1.0, where 1.0 is the shape index of
a circle, and values >1 represent more complex shapes. Area weighted shape index has
the same properties, but patches are given weights proportional to their area in the
calculation. Thus, the shape properties of large patches will have a greater influence on
the final value than the shape properties of relatively smaller patches (McGarigal and
Marks 1995).

Patch density and mean patch size are redundant variables, but are both useful for
visualizing a landscape. Mean patch edge and edge density are not entirely redundant
variables because edge density is calculated by considering the edge between 2 patches
just once, while the calculation for mean patch edge uses the same shared edge for each
patch in the landscape (McGarigal and Marks 1995).

Patch fractal dimension can range in value from 1-2, with values near 1
representing relatively simple shapes, such as circles and squares, and values that
approach 2 indicating more convoluted perimeters. Area weighted patch fractal
dimension has the same properties, but patches are weighted according to their area in the
calculation (McGarigal and Marks 1995).

The interspersion and juxtaposition index, expressed as a percentage, is at a
maximum (100%) when all patches are equally adjacent to all other patches. As the

amount of edge shared between patches becomes more disparate, the index approaches

135

0% (McGarigal and Marks 1995) (Fig. 9).

Both Shannon’s diversity index and Simpson’s evenness index can be used to
quantify landscape compositional diversity. Of the 2 indices, Shannon’s diversity index
is more sensitive to patch richness and is more influenced by rare patch types. Shannon’s
diversity index is 0 in a landscape with only 1 patch type, and therefore no patch
diversity, and increases without limit in proportion to the number of patch types and the
equitability of the proportion of the landscape in each patch type (McGari gal and Marks
1995).

Simpson’s evenness index is based on the proportion of each patch type
represented in a landscape, and is more strongly influenced by common elements. The
index is expressed as a probability that 2 randomly selected patches will be different
types. The index is 0.0 when there is only one patch type and approaches 1.0 as the
number of patch types increases and the proportion of each patch type in the landscape
becomes more equal (McGarigal and Marks 1995).

Data Analysis

Landscape characteristics and cover type composition were compared among
ownerships with a Kruskal-Wallis one way analysis of variance, followed by the Kruskal-
Wallis multiple comparison statistic (Siegel and Castellan 1988) if significant differences
(p30. 10) were detected. A nonparametric test was used because of the small study site
sample sizes, due to the difficulty in obtaining additional replicate landscapes within each
ownership. For all statistical tests, the significance level was set at a = 0.10.

Comparisons of road densities by cover type were made with a Chi square test.

136

 

UI for northern hardwoods (solid gray matrix) = 88%
U I for all cover types combined = 69%

 

UI for northern hardwoods (solid gray matrix) = 47%
UI for all cover types combined = 63%

Figure 9. Examples of the interspersion and juxtaposition index. Calculations were
performed with the Patch Analyst extension to ArcView.

137

Spearman rank correlations were used to identify associations between relative
abundance for each selected species and landscape variables across the 10 study sites.
Species data were combined for all sampling periods within a year because landscape
attributes were assumed to have remained constant. Only 1 of each group of redundant
variables is presented in the correlation analysis

To identify relationships between landscape characteristics and barred owl
abundance, landscape variables within the 200 ha surrounding points where barred owls
were present were compared with the area around sampling points where owls were not
present. Presence/absence data were used because only a small proportion of barred owl
survey points each year had a positive owl response recorded, and during the study, very
few survey points accumulated more than 1 positive response. Landscape variables for
each of the 2 groups of sampling points were averaged across each study site, and the
nonparametric Wilcoxon Mann-Whitney test (Siegel and Castellan 1988) was used to

compare landscape characteristics of points where barred owls were either present or

absent.

138

RESULTS

Vegetation cover type distributions

The average proportion of land area covered by northern hardwoods ranged from
70% on Huron Mountain Club land to 89% on Forest Service land for study sites created
with 80 m and 800 m buffers (Tables 33 and 34). Each remaining cover type comprised
<8% of the landscape for all ownerships except the Huron Mountain Club, where the wet
hardwood/conifer mix cover type made up 21% for the 80 m buffered study site and 18%
for the 800 m buffered definition. Based on analyses of the smaller landscape definition
(80 m buffer), the second most common land cover type on MDNR sites was wetlands,
dry hardwood/conifer mix on Forest Service sites, and wet hardwood/conifer mix on
timber industry sites. For the landscapes defined by the larger buffer size, a dry
hardwood/conifer mix was the second most common cover type on MDNR and Forest
Service land, and the mixed conifer cover type was the second most common in the
composition of timber industry sites. A combined total of 16 land cover types (i.e patch
richness = 16) were represented among the ownerships, ranging from 9 (timber industry)
to 14 (Forest Service) types in an ownership (Tables 33 and 34).

Jack pine made up a greater (p=0.083) proportion of the land cover on Forest
Service and Huron Mountain Club study sites than on the other 2 ownerships (Table 34).
The jack pine stands at the Huron Mountain Club are approximately 80 years old and
have only been thinned in the last 5 years because of the fire hazard they pose.

No statistical differences occurred among ownerships for all other vegetation

cover types. However, at the most expansive scale analyzed, lowland hardwoods were

139

awe 5:230 Ea swag 85:?
.«o mix—«SN $3-28 £33-?me 05 £3, noon—sofio Bo? 838236 macaw 36:88:: :0 mac: 8: 20>»: b=3mn€m a

flown—=28 on 2:8 Hobo teased: o: .2882: 6382.2 : 8 courses 825 magnum—a2 m =< .

 

 

easel—2.

140

 

 

 

who: ::.: ::.: ::.: 3.: 3.: ::.: ::.: Eco... 22>?
mmc: ::.: ::.: ::.: 8.: m:.: m: .: mu: xoflmfieh
m2 .0 we... and an: 2 .o mg on. 8a 35a 352
:3 a»... 8d 8... 2 .o 2... cod 8... 8a as
S m: an: ::.: ::.: mod n:.: ::.: ::.: 2:: com
:3: =4 3.: an: S .: am: m _ .5 8.: $5303
mom: SAN ::.v :m.m mm: ::A 5.: an; 3:8 HotsooBooaean 33
am»: mean 8d :3” 91 had 3.: :58 me cotcofleooavan .09
a3 a: 8m can 3o 3— ado as 655%...
amm: ma.:b mus 24: mm; mad: and ~ ::.:. £60025; Eontoz
mmm: 3.: ::.: ::.: an: 3.: w _ .: ::.: cam—985m
Nam: 3...: 5.: 3.: and find am: Nb; .83_=otwm\m=oooe€om
3N: ::.: ::.: ::.: 5.: —:.: 5.: 8.: 5&3

amflnflocao x .m.m x .m.m. x .m.m x 399 850 :5...—
wcoEe. | l l I

855:6 mo «SHE U“ a: GUNS C. SHE mung Anna mam:

335305 Eowoamo Esau—.30

 

.8880: 203 R: .OAB 30:28:: 33$:me oz
Down—E 858mm 5:— 58.: boa—=23 .2355: 595 Fameozz E :5: 625 :20 5.8552 .5.—:3 use AFC .939:

5:8: .Ammma 836m :38": .m .D .2292: 38:83— 333 Z we 2:08:35 535:2 :o meet—.5 E :w 515 cacao:
82m 33m 8: bowed: 258% .3 BEEoE 25 .566 some 5 68b... Becca: use 2825 ~on :5: :88 no Beacon .mm 2an

 

 

who: ::.: ::.: ::.: mm: mm: 8.: m:.: .8on BE?

 

 

 

 

83 8... 8d 8... 89¢ 85v cod 8... Baa... 233
89o 8... 3o 3... 3o 8... 5o :5 cases.
ammo a: 8d a... mad 3... m3 «3 85a 852
Sod .mm... cod .8... z .o .8... 8o .8... .83 6.8.
a S em... 86 8... mod 3... 8o 8... 8a 8m
memo a: «no a: 95 as... a: chm €5.33
33 an: Nam 3m m3 «3 8.2 «an 5:. 5388.833; 33
£3 :3 3.». am... 8a 8.... E .m e: ea EESBSBEE to
83 8.: cod 8... 8o 8... cod 8... 88%;: 9333
”So 8.: 42 E." :3 m5 So e3 555%..
24d «5: $8 8.“: a2 2.3 mg FE $8322 8282
3.3 a... 3o S... Rd 3... _ S a... Bazaam
£3 9.... $5 8; :2 85 m _ .o a; .aaéacweaaoaoaeo:
63 3.... So :3 85 8... cod 8... 535

ammEanao x .m.m x .m.m x .m.m x 399 35o :53
macaw I I l l

gaseous co . u an: F 8.1.5 mam: u a

55:3on bowofio £23830

 

..CowaEm 8528 33 88: 33323 532:5: 5&3 Mcawfiozz E US: 02:: :20 58552 .553 28 .95 .9265
“BE: .Ammma 83.3 “meow .m .D .9292: 38:83. 35ng .8 EoEtwmoQ 535:2 so £8.35 8 0:: 5:5 :2on 3%
33m 8: >5me 3583. .3 “co—33:02 25 B>8 some 5 Quote Banana :5... 33.5 was v52 :38 mo Eoocom .vm 2an

141

 

 

.wag 5:830 can
Bwomfl 03285 83358 23:58 msfigéxmavc 2: .OAE 820ch Ho: 8a 532 2:3 05 5?» on: 253 05 :o mes—~35n

A32 5:3me can Ewflmv uncut?
ho $9295 $3-28 £33-?me 05 53, 3830.3 803 mmfimuocBo wcoEm weep—Boga Mo 33“ “8 £05— bzfimnoi A.

622528 3 2:8 .55 Banana 0: .8820...“ .oawomncfi _ ma uni—mam 203 323292 m :< a

 

 

 

 

 

~26 3.5 mod ”N.“ _~.o mvfl EA Mwnn Sacco @822
nmmancao x .m.m x .m.m x .m.m x 8%: B>oo 6:3
mecca l l l l
oocouumtn ho . n GUS E. SHE wwwd GUS Hafiz
$33on .60ng 9:80:30

 

 

.bome_ 3583
52 soc 32820 .2355 5&2 @3222 a 22 625 £6 €582 85: 23 .95 ESE Ea: .Ammma

83.6w 380"“ .m .D .EZQEV woesomom E532 “6 625.839 53:32 so acts: E cow ::3 cos—Lou 33m 38m
é bowQE. 353mm .3 35:52 RE 560 some 5 Quote 32:83 93 £325 we“ 98. .S9 mo 888m .95an X“ 2an

142

 

present on only 1 MDNR site, and white spruce was only identified on 1 Forest Service
site (Table 34), both represented by a single polygon. The relatively greater proportion of
the urban cover type at the Huron Mountain Club represents the Club “compound”, where
members and employees are housed and where many of the services of a small “town” are
located. The herbaceous/Openland cover type was also slightly more common on MDN R
and Forest Service sites, and probably represented openings maintained for wildlife.

The fact that the hemlock cover type was not represented in the satellite imagery
of the Huron Mountain Club (Tables 33 and 34) is misleading because mature hemlock
stands were regularly observed during field work. It is likely that the wet hardwood
conifer mix that makes up approximately 20% of the satellite imagery includes the
hemlock stands that are prevalent at the Huron Mountain Club, yet were observed much
less frequently on all other study sites.

Landscape characteristics of ownerships

Many landscape characteristics of study sites did not differ significantly among
ownerships. For both the 80 m and 800 m buffered landscape, timber industry sites
tended to have larger and fewer patches, as well as lower patch richness (i.e., number of
cover types) (Tables 35 and 36). For the 80 m buffer, the average perimeter associated
with each patch was also longer (p=0.089) in correspondence with the generally larger
patch sizes. The Huron Mountain Club, however, was made up of a relatively larger
number of smaller sized patches, and MDNR and Forest Service sites had intermediate
mean patch sizes. In the 800 m buffered landscape, both median patch size and average

patch edge were greater (p=0.075, p=0.045) on MDNR and timber industry sites than

143

owe o .3...“ .2 .o ash.— Sd .meé 2 .0 mm;

.3865 b§o>€ Paoccazm
€ch 5?:

  

 

G _ .o afiwm ow; . . .
£3 «.3 . 3 mm 2 m S 8 om: 3.8 8282332 98 cofiuneo
: _.o 34 cod vmé . . cm u coup—2:6 afloat 635mm?» no.2“
ammo a?“ . . oo.o em 2 mod «3 85588 38c 55
. I. o S m 3 o and E .o N¢.N 39: 0927. Beam—03 «84
”Mme «3 mod un— 86 32 5o «m2 Ems 23%
when im 2 2 .o .22 :.o :82 Ed awn.— 15:30 53..
we 8.82 8.2 8.2 8.: 3.5 8.8 3.3 afie bacon 03m
OVH o Naoc ~m.o ”fie" ~ —.0 Sad OV.O mnofi ANSV DNmm SUN—Nam gun”:
a.» o 8.9 R2 2.... :2 3% m3 :6 page .338” am“
. . _
acme R E 2.: 2.3. c5. 2.8 in 3.2 a: komvcwukfiwaz
5.2. 8.2 mad was $5 8.: m3 m2: 8 2M ”as Ba 33
m2 9 8.2 on.“ 3.2 m3 3.: m3 3.2 N
Dmgmnmhocao Utaoe QQNUWUE
95:8 M .m.m x .m.m 1m .m.m M.

00:80th N I ll Algal“... - Ana
HO a AMI-H: HR. I. M
0&3

lbzfianoi >8ng Esau—~30 - ,.

.bomag BEE

. E .
DHWHMMWMQEV 832: omaomncfl

8 2a mo=E>
db 9265 a
ooh Quote 32:3

m 5% 3:3 E S 5 5; @838 85 bosom a
. £22 5 22 625 26 5355: =85 9a
H.532 .8 EoEwamoQ $2222 :0 82m 38m

144

E
Ewfifl oumsflm Sargasso 23:2: m_=m>>-_§mavc AS .OAE Bogota 8: 2a 8:2 2:3 05 .23 o .

803
.932 5:830 new Bwommv 8.8.9? mo max—28 33.25 333.3me 05 5:5 33323

 

dos—:28 3 2:8 Stu 22:83 0: .8822:

  

 

 

 

 

 

l mm: .o
smmEanao . . . .m x .m.m
was“ w m m w m .. «33%
85th a u Anni E. l! mu lillllhil\
mo
E525 homes aeaocao 32:28
l -, _ ace .
5&3 S cm 5 5E 8320 8%. boa... c8 8“ 82; .5262 Away mwtom meow m aw.
E 22 625 36 @382 85: Es db 9265 .38: m €03 8E2: 098%

SN
Ho Buchanan— :wwEB—z so 8% 55:. c8 Amoeba 32:83 v

53035: _ mm nouxfics So

.232 5:296 Ea
_ 08.3 05 co mos—«>3.

notomou 292 559305 a

3 330:9“.— m =< .

          

145

 

 

 

  

 

 

wood .Nwé 56 #6.— mod awbé mod .mwé *xoUE bzmuofiv $55525

GE 53

83 3% 8.~ 8% 86 2.3 cam 3.3 838%:9. 23 886325

Sod mm; 86 .3 So «2 :3 SA 8:856 38¢ 85%; 32

:8 3... 86 R4 So e2 :3 >3 83556 BMW MUM

. . — >9

3.2 9.... So can 8o «5” E o a n SE 25: BM? 23%

:3 :2 So a: mod 3: memo 5.” have own.» 65

WVOAV AMNA w©.O %.fi m—.O aWfi w 0w. 0 «NM. Awe—\EV Aime—0% Uwfim

ammd 3.3— oo.w_ 3dr cad emfie ow 3 3.3 LE: 36 :83 5602

33 am... #8 a: 85 am; 2 .o .31 meiv 28% :28

$3 mm.» «2 ”me 33 Gm B _ 2w.“ 35 9% 53m

33 3: a2 5.3 N3” «2: S m «a 8% :23 he £632

88 8.2 who 8.3 on; 2.2 £8 an: $5103 .8th

and 8.2 a: 2.3 on.» 3.8 :.~....% 2%.: amomnqfi

gnaw—38:30 . . . . x .m.m M
muons M m m .m m m ... I
”820% g u sue F 5% a mag H
ho can:
b25305 homage 92.04230 oh wan—:28 .Samcwmwwuwomom
I E .
8:3» .bomaé 2:33 fiww meow .m D “WW/mg .3 295
3&3 E 8m 5w 2:3 @838 3% ~33» mewwmhzmsue “BE: AmmmbwmomoEoE amomncfi u
:d . o .

”Swag: 5 Ba. 625 26 59:82 85m n
1&5ng mo EDEEQQQ 5&5va so 83 >9:

m Sb. Echo Emu:

Sm Ea mam

146

w .wae Sw=8mmU can
.0 2E ozmcfim cowgafioo 29:2: ERBAEEEMV Q: .OAE “scumbag 8: 08 8:2 2:3. 05 :23 on: 083 05 :0 mo:R>*

Amwa~ 58:3me fig Emommv eon—Mtg, mo mmmh—acd xmzruoco mSEBI—QXwP—VH 05 53> tango—NU 0.53 Gov—0&2 m_o>o_ bmfinabokm a

.omaomncfl _ mm Bax—«ca 803 823292 m =< a

.35328 on 2:8 hobo Banana 0: .2822:

   

 

 

 

 

 

5 o S o E a! S o
nmmEEocao
$25 m .5 m 4,;
855:6 a u .. lama
mo 815?
3.3805 homage 3:30:30 \\

 

.bowSE 8:33AM

5 .558: A953 .
385m; 93 2325 8.52:

E cow an 5:5 @238 8% 33¢, how 8a 825»
E 95 625 :20 58:32 825 can .95 Disc
Mo EoEtaoQ fiwfioaz co 8% 3:: H8 383

147

Forest Service (p=0.075) and Huron Mountain Club sites (p=0.045), indicating that
landscape patches on MDNR and timber industry sites were skewed towards larger sized
patches, while the other ownerships tended to have a higher frequency of relatively small
sized patches.

Interestingly, both mean shape index and mean fractal dimension varied little
among ownerships (Tables 35 and 36), but the area weighted version of each parameter
tended to be greater for the Huron Mountain Club than for the other ownerships.
Although the difference was not statistically significant, the data suggest that the largest
patches at the Huron Mountain Club are also some of the most elaborately shaped
patches.

The interspersion and juxtaposition index is another landscape variable in which a
trend among ownerships was evident, although not statistically documented. Michigan
Department of Natural Resources sites exhibited the greatest degree of interspersion and
juxtaposition among patches for both landscape buffer sizes analyzed, indicating each
patch is in contact with a relatively greater number of other patches. As a partial
consequence of the lower patch richness on timber industry sites, cover type diversity,
expressed by Shannon’s diversity index, was lower (80 m buffer, p=0.089; 800 m buffer,

=0.098) on timber industry sites than elsewhere. Simpson’s evenness index followed a
similar trend as Shannon’s diversity index for each ownership, but the trend was not

statistically significant (T ables 35 and 36).

148

Road density

Density of trails, county roads, public streets, and highways ranged from 5.3 m/ha
on MDNR sites to 11.7 m/ha on timber industry sites (Table 37). The relatively low road
densities calculated (on the order of several m/ha) are due to the fact that the digital data
set for roads is based mainly on highways and county roads which occurred infrequently
on study sites. Logging roads were under represented in the MIRIS database, compared
to what was observed on study sites.

Although timber industry sites tended to have higher road densities than other
ownerships, Kruskal-Wallis comparisons of mean road density among ownerships were
not significant. In proportion to the land available in each cover type, roads occurred
slightly more often in northern hardwoods, upland pine, tamarack, white cedar, and
mixed conifers (Fig. 10), but not significantly so (p=0.28). Roads were slightly less
prevalent in the dry hardwood/conifer, wet hardwood/conifer, and wetlands cover classes,
compared to the proportion of these cover types in the landscape.

Landscape characteristics of northern hardwood forest patches
Although there were no significant differences among ownerships for any attributes
of northern hardwood forest patches, Forest Service sites were distinguishable from
others by the relatively large average size (531 ha) of their northern hardwood patches
(Table 38). The very high value (67.25) for median patch size on timber industry sites is
due to one timber industry site that had a few very large blocks of unbroken northern
hardwoods, while other sites generally exhibited a range of hardwood patch sizes.

Patch edge represents the average perimeter length associated with each patch, and

149

Table 37. Road densities by cover type (m/ha, means and standard errors) for study sites
on Michigan Department of Natural Resources (MDNR), U. S. Forest Service (USFS),
timber industry (T1), and Huron Mountain Club (HMC) land in Michigan’s Upper
Peninsula. No significant differences (p>0.10) were detected.

 

 

 

Ownership category Probability
of
Landscape MDNR USE-S I1 HMS; difference
buffer size _flEll _flE-n _m53l (111:1)- among
x S.E. x S.E. x S.E. x ownershipsb
80m 5.27 3.39 6.13 1.37 11.73 3.76 6.90 0.433
800m 7.21 3.06 7.80 0.86 9.45 2.72 6.84 0.921

 

“ All 3 replicates were analyzed as l landscape, therefore, no standard error could be
calculated.

b Probability levels reported were calculated with the Kruskal-Wallis one-way analysis of
variance (Siegel and Castellan 1988).

150

.32 03:3 28 wad mm 25 eo>oo :03 we b23282“ 05 E8» £859 25 uo>8 some 5 meme.—
uo somtoaoa 65 35 523305 .SwwEB—Z mo «Ema—Eon— ..omaD 65 E Stan E Ow 5w 5;» 6338 32m .3on co» 099 .550
some 5 £38 :38 he somtoaoa owfioaw was 09$ :38 some .3 358652 560 US: 138 he .8an8.“ owmco>< .o_ beam—m

on? “950 can;

 

tog

-om

..om

 

waned U 850 mead I

 

 

 

.8

too

(%) [moi JO uotuodmd

.2.

 

..ow

 

..oo

151

.Awwa 5:830 98 331% ocean? .8 £93.25 33.25 £33-3wa 65 £3, cos—=28 803 8:38 20>»: bzmnanoi a

.vofiso—eo 3 2:8 hobo 935% o: .8822: 63823 _ ma conch—«SN 963 33238 m =< .

 

152

 

 

 

3.3%...
mwfid 3.3 Maw 5.3 3d 3&5 36 3.3 528933“ can 58535::
owed 5..— mod gm; 5.0 an; 56 an." 56.88% 325 BEE?» 8.2.
gad 3..— de em; mod 34 mod me; 36586 38c 53¢
83 gm m3 3... E .o 2... t .o can 62: a2... 83%; 82
wand 3A 93 22 v _ .c «A: ood FA 52: 092m
3nd chm mi. 3.3 mad mm.m~ $6 and c5: omen scam
Sod can m6 flan ad ”.3 OH as $5.5 bmmcoc ownm
wwwd 3.: 3.8 mafia 5d 36 3.0 3A 35 36 53mm 5622
ommd we... ood and ood 3.: Rd ah: ANEéE 93% :23
mmnd an 5 awn mmm Em C: awn $5 36 no:
$06 8.5 min 3.3 $3 8.5 Sum 3.3 Q5 38 .38.
5338830 x .m.m x .m.m x .m.m x 252: 258984
waoEm I l l l .
85.0% . u an: : Jamaal Ifiamfizl
mo
$3305 bowofio 956550

 

68083 803 8:13 mooeeotmu “505%? oZ cubes 8 cow as .23
@838 8% >95 .8.“ 2a 838/ .bowaEm 258mm 32 Set Bum—=33 .2355; 5&3 Pameosz 5 “ES 62“.: £20
58552 :85: can .95 .6332: 89:: AmmmDV 855m 3.20m .m. .D .AMZQEV moocsomom 1253 Z we E25330 cmwfiosz
:o moxm beam HE 3283 $88 “68322 565.8: .8 9.8.5 3353... ES 338v 8508 omaomn§_ c8 338/ .mm 2an

is larger for northern hardwood patches than other classes in correspondence with their
larger area. The noticeably high value for patch edge for the Forest Service ownership
was greatly influenced by a study site which had 1 extremely large patch that was
essentially functioning as a matrix in which many other patches, including additional
northern hardwood patches, were embedded. Shape index, fractal dimension, and the
interspersion and juxtaposition index were also generally higher for northern hardwood
patches than for the collection of all patches in the landscape (Tables 36, 37, and 38).
Relationships between wildlife species relative abundance and landscape

characteristics

Red-backed salamander

Correlations between red—backed salamander relative abundance and the proportion
of vegetation types on each study site revealed very few strong relationships (Tables 39
and 40). Three of the 4 significant (p30. 10) correlations that were calculated indicated a
negative relationship between the proportion of various pine dominated cover types in the
landscape and the number of red-backed salamanders observed. The fourth significant
correlation occurred only in 1998 and showed a positive association between the
proportion of the mixed pine cover type and salamander relative abundance (Table 39).

Landscape characteristics revealed few consistently significant relationships with
red-backed salamander relative abundance (Table 39). Patch size was the only landscape
metric for which a relationship occurred in the same direction for each year of the study,
with a significant negative correlation in 1997 and in both years combined (Table 40).

Significant correlations also occurred for median patch size, patch fractal dimension,

153

Table 39. Spearman rank correlations between proportion of vegetation cover types in
the landscape and red-backed salamander relative abundance (1997 and 1998) on study

sites in Michigan’s Upper Peninsula.

 

 

 

 

* Probability level 50.10.

154

Sampling period
1997 1998 Both years
Land cover type (n=8) (n=10) (n=lO)
Urban -0.082 -0. 172 0.529
Herbaceous/agricultural 0.048 -0. 164 0.273
Shrubland 0.270 —0. 175 0.381
Northern hardwoods -0.575 0.164 -0.418
Aspen/birch -0.060 -0.394 0.224
Dry hardwood/conifer mix 0.455 -0.261 0.503
Wet hardwood/conifer mix 0156 0.152 -0.018
Wetlands 0.491 -0. 152 0.261
Red pine 0.157 -0.701* 0.311
Jack pine 0.094 -0.683* 0.208
Mixed pines 0.719* 0.000 0.51 l
Tamarack -0.070 0.356 -0.097
White cedar -0.368 0.294 0.164
Mixed conifer '(Mfl .188 -0. 8*

Table 40. Spearman rank correlations between landscape metrics calculated from 1991
satellite imagery and red-backed salamander relative abundance (1997 and 1998) on study

sites in Michigan’s Upper Peninsula.

 

 

 

Sampling period
1997 1998 Both years

Landscape metric (n=8) (n=101 (n=10L
Total area -0.515 -0.685* 0.018
Number of patch types -0.018 -0.207 0.512
Patch size 0802* -0200 0624*
Median patch size -0.434 0.176 -0.650*
Edge density (m/ha) 0.563 0.067 0.345
Patch edge -0.108 -0.l88 -0.539
Shape index 0.359 -0. 152 0006
Area weighted shape index -0.347 -0.309 0.055
Patch fractal dimension 0.431 —0.1 15 0636*
Area weighted fractal dimension 0.252 -0.115 0.033
Interspersion and juxtaposition index 0.359 0188 0.297
Shannon’s diversity index 0.060 -0.467 0.588*

s in CL .3 - .4 1

im 8 ’s ev

 

* Probability level $0.10.

155

and Shannon’s diversity index when both years (1997 and 1998) of data were combined,
but results for each individual year did not suggest such a trend.

American redstart

The significant negative correlation between American redstart relative abundance
and the proportion of mixed pines in the landscape was the strongest cover type
relationship observed (Table 41). In addition, there was a consistently positive
correlation between redstart numbers and the proportion of northern hardwoods in the
landscape that was significant in 1996, and a negative correlation with the urban cover
classification that was statistically significant in 1998.

Several landscape metrics were significantly negatively correlated with American
redstart relative abundance combined over 3 years, including patch richness, patch fractal
dimension, and Shannon’s diversity index (Table 42). A positive correlation with patch
size was significant in 1997, and apparent to a lesser degree in other years. Redstart
numbers were also positively related to mean patch edge for observations in 1998 and
when all 3 years were combined.

Ovenbird

Several conifer cover types were positively correlated with ovenbird relative
abundance for at least 1 year of data (Table 43). Results for mixed pine were consistently
positive, but the red pine, jack pine, and mixed conifer cover types showed contradictory
correlations among years, particularly between 1996 and 1998.

Patch fractal dimension, interspersion and juxtaposition index, and the 2 diversity

indices all had significant positive correlations, and patch edge had a negative correlation

156

 

..- ,‘_ l—ozw 7”

Table 41. Spearman rank correlations between proportion of vegetation cover types in
the landscape and American redstart relative abundance (1996, 1997, and 1998) on study
sites in Michigan’s Upper Peninsula.

 

 

 

 

* Probability level $0.10.

157

Sampling period
1996 1997 1998 All 3 years
Land cover type (n=7) (n=8) (n=10) (n=10)
Urban -0.201 -0. 191 -0.561* -0.529
Herbaceous/agricultural 0.346 0.071 -0.049 -0. l 15
Shrubland 0.165 0.073 -O.233 0213
Northern hardwoods 0764* 0.333 0.274 0.333
Aspen/birch 0.182 0.452 0.255 0.224
Dry hardwood/conifer mix -0.309 -0.143 -0. 140 -0.212
Wet hardwood/conifer mix 0.055 -0.095 0.006 0.006
Wetlands -0.600 -0.3 10 -0.4 19 0422
Red pine -0.1 13 0.094 -0.026 0026
Jack pine 0.1 13 0.156 0.026 0.026
Mixed pines -0.927* —0.548 -0.738* -0.729*
Tamarack 0.1 13 0.038 -0.272 0213
White cedar 0.318 -0.156 —0.551 0510
Mr W

Table 42. Spearman rank correlations between landscape metrics calculated from 1991
satellite imagery and American redstart relative abundance (1996, 1997, and 1998) on

study sites in Michigan’s Upper Peninsula.

 

 

 

Samplingperiod
1996 1997 1998 All 3 years
Landscape metric (n=7) (n=8L (n=10) (n=10)
Total area 0.29 1 0.452 0.06 l 0. 127
Number of patch types 0.] l 1 -0. 120 0648* -0.604*
Patch size 0.564 0.738* 0.419 0.527
Median patch size 0.092 0.216 0.393 0.395
Edge density (m/ha) -0.691 -0.476 -0.219 -0.297
Patch edge -0.491 0.286 0596* 0.600*
Shape index -0.691 -0. 167 -0. 103 -0. l 15
Area weighted shape index 0.218 0.310 0.219 0.188
Patch fractal dimension -0. 182 -0.476 -0.863* -0.867*
Area weighted fractal dimension -0.691 -0. 143 -0.036 -0.103
Interspersion and juxtaposition ~0.255 0.071 -0.304 -0.188
index
Shannon’s diversity index -0.182 0.143 -0.675* 0600*
i s n’ v nness i ex — 0 - 24 - 4 - .4

 

* Probability level $0.10.

158

Table 43. Spearman rank correlations between pr0portion of vegetation cover types in
the landscape and ovenbird relative abundance (1996, 1997, and 1998) on study sites in

Michigan’s Upper Peninsula.

 

 

 

 

* Probability level 3010.

159

Samfling period
1996 1997 1998 All 3 years

Land cover type (n=7) (n=8) (n=10) (n=10)
Urban —0.451 0.357 0.322 0.201
Herbaceous/agricultural 0.000 -0.55 1 0.401 0.370
Shrubland 0330 0.430 0.253 0.537
Northern hardwoods «0055 -0.180 -0.432 -0.248
Aspen/birch -0. 164 -0419 0.377 0.224
Dry hardwood/conifer mix 0.000 -0.299 0.571 0.422
Wet hardwood/conifer mix 0546 0.120 -0.261 -0.406
Wetlands 0. 164 0.395 0.419 0.394
Red pine -0.748* 0.243 0664* 0.164
Jack pine -0.680 0.133 0638* 0.182
Mixed pines 0.546 0814* 0.366 0626*
Tamarack 0.679 0.415 ~03 18 0.213
White cedar 0.136 0.031 —0 104 0.208
Mixed conifer 0764* 108 - .48 — .12

with ovenbird numbers combined for 1996-1998 (Table 44). These trends were
consistent for each individual year of the study, reducing the likelihood that the
relationships are spurious. Area-weighted mean shape index was significantly correlated
in 1996, but the relationship fluctuated among years and was the variable was probably
not biologically important by itself.

Veery

The number of veery observations in 1996 was significantly and negatively related
to the proportion of shrubland and tamarack in the 80 m buffered landscape (Table 45).
There was also a positive relationship with the mixed conifer cover type that was not
significant in any 1 year of the study, but was significant for the 3 years of pooled bird
survey data.

Landscape metrics that may have influenced veery populations on the study sites
sampled were patch fractal dimension, which was negatively correlated with veery
relative abundance in 1998 and for all data years combined, and the interspersion and
juxtaposition index which was positively correlated and statistically significant in 1996
and 1997 (Table 46). Also, there was a significant negative correlation with mean patch

edge in 1996, and a positive correlation with Simpson’s evenness index in 1997, but

correlations for the pooled data were opposite in sign.

Yellow-ramped warbler

The only compelling correlation between cover type proportions and yellow-rumped
warbler relative abundance (Table 47) was the significant positive relationship for the

urban cover type. The most likely reason for this is that the urban cover type occurred

160

Table 44. Spearman rank correlations between landscape metrics calculated from 1991
satellite imagery and ovenbird relative abundance (1996, 1997, and 1998) on study sites
in Michigan’s Upper Peninsula. Values in bold are significant at ps010.

 

 

 

Samplipg period
1996 1997 1998 All 3 years

Landscape metric m=7) Q=8) (n=10) (n=10)
Total area -0655 -0156 0.523 0.042
Number of patch types -0 167 0.539 0.434 0.549
Patch size -0055 -0180 -0255 0285
Median patch size 0.138 -0.544 -0.335 —0.456
Edge density (m/ha) -0.055 0.036 0.243 0.103
Patch edge -0.273 -0240 —0 176 -0.564*
Shape index 0.000 0.371 -0 140 0042
Area weighted shape index -0873* -0479 0.304 -0. 164
Patch fractal dimension 0.327 0.467 0.559 0806*
Area weighted fractal dimension -0.218 -0156 0.292 0.018
Interspersion and juxtaposition 0873* 0.551 0.267 0721*
index
Shannon’s diversity index 0.055 0814* 0644* 0758*

. l * . 35 . l .73 *

Simpson’s evenness index

 

* Probability level 3010

161

Table 45. Spearman rank correlations between proportion of vegetation cover types in
the landscape and veery relative abundance (1996, 1997, and 1998) on study sites in

Michigan’s Upper Peninsula.

 

 

 

Samplingperiod
1996 1997 1998 All 3 years
Land cover type (n=7) (n=8) (n=10) (n=10)
Urban -0374 -0.384 -0396 -0500
Herbaceous/agricultural 0.214 -0288 0.036 -0285
Shrubland 0.775 * 0.565 -0.097 -0 149
Northern hardwoods 0.357 0.072 0.134 0.261
Aspen/birch 0.214 0.275 0.061 -0 176
Dry hardwood/conifer mix 0214 0.060 -0231 -0455
Wet hardwood/conifer mix -0 143 -0323 0.231 0.127
Wetlands —0.321 -02 l 6 -0176 0200
Red pine —0535 -0486 -0373 -0545
Jack pine -0.401 -0.517 -0321 -0493
Mixed pines -0107 0.275 -0.424 -0280
Tamarack 0852* 0.619 0.266 0.381
White cedar 0.223 -0220 -0.104 -0121

Mixed conifer

* Probability level 5010.

162

0.536 0.5 ZS 0.402 0i 221*

Table 46. Spearman rank correlations between landscape metrics calculated from 1991
satellite imagery and veery relative abundance (1996, 1997, and 1998) on study sites in

Michigan ’3 Upper Peninsula.

 

 

 

Sampling period

1996 1997 1998 All 3 years
Landscape metric Q1=7L 01:8) jn=10L (n=10)
Total area -0429 -0599 -0353 -0.467
Number of patch types 0.164 -0.224 -0355 -0.476
Patch size 0.250 0.036 0.128 0.309
Median patch size -0360 -0241 0.335 0.444
Edge density (m/ha) -0.429 -0132 -0.030 -0 176
Patch edge -0.821* -0 180 0.207 0.345
Shape index —0571 -0. 168 -0 152 0.018
Area weighted shape index —0429 -0.503 0.188 0212
Patch fractal dimension 0.036 -0.263 0602* 0600*
Area weighted fractal dimension -0464 -0.060 0.085 -0.164
Interspersion and juxtaposition 0786* 0826* -0.122 0.006
index
Shannon’s diversity index 0.250 0.168 —0523 -0.539

0.250 .6 * -024 - .224

Simpson’s evenness index

 

* Probability level $010.

163

Table 47. Spearman rank correlations (rs) between proportion of vegetation cover types
in the landscape and yellow-rumped warbler relative abundance (1996, 1997, and 1998,

combined) on study sites in Michigan’s Upper Peninsula.

 

 

Land cover type rs (n=10)
Urban 0615*
Herbaceous/agricultural -0350
Shrubland 0.167
Northern hardwoods 0.097
Aspen/birch -0052
Dry hardwood/conifer mix -0142
Wet hardwood/conifer mix 0.291
Wetlands -02 l 6
Red pine 0.383
Jack pine 0.255
Mixed pines 0.082
Tamarack -0008
White cedar 0.149
- 46

Mixed cog; fer
* Probability level $0.10

 

164

primarily at the Huron Mountain Club where most of the yellow—rumped warblers were
observed. Thus, it is more likely that the correlation occurred because of a biological
association between yellow-rumped warbler abundance and another of the unique
characteristics of the Huron Mountain Club. No other significant correlations between
yellow-rumped warbler counts and cover type composition or landscape metrics were
found (Tables 47 and 48), perhaps because of the low variability in yellow-rumped
warbler observations among study sites.

Pileated woodpecker

During the 3 years of the study, pileated woodpeckers were recorded in only 7 of the
stands included in this analysis, with no more than 2 observations for any of the stands.
Therefore, only correlations with variables calculated for the 800 m buffered landscape
are reported.

There was a significant positive correlation between pileated woodpecker numbers
and the proportion of the urban cover type in the landscape, and a negative correlation

with the mixed conifer cover type (Table 49). The significance of the urban cover type
relationship may again be that land classified as urban occurred predominately at the
Huron Mountain Club, as did the majority of pileated woodpecker observations.
Therefore, as with the yellow-rumped warbler, there are probably other biological
relationships underlying the correlation result. Area-weighted mean shape index and
area-weighted fractal dimension corresponded strongly to the number of pileated
woodpecker observations (Table 50). Values for these landscape metrics also tended to

be greater at the Huron Mountain Club (T ables 36 and 37).

165

Table 48. Spearman rank correlations (r,) between landscape metrics calculated from
1991 satellite imagery and yellow-ramped warbler relative abundance (1996, 1997, and
1998) on study sites in Michigan’s Upper Peninsula. No significant differences (p>010)

were detected.

 

 

Landscape metric rL(n=10)
Total area 0.500
Number of patch types 0.450
Patch size 0.127
Median patch size -0434
Edge density (m/ha) 0127
Patch edge -0.2l6
Shape index -0. 186
Area weighted shape index 0.142
Patch fractal dimension 0.022
Area weighted fractal dimension —0037
Interspersion and juxtaposition index -0231
0.529

Shannon’s diversity index

Simpson’s evenness index -0.142

 

 

166

Table 49. Spearman rank correlations (rs) between proportion of vegetation cover types
in the landscape and pileated woodpecker relative abundance (1996, 1997, and 1998) on

study sites in Michigan’s Upper Peninsula.

 

 

 

Land cover type ri (n=10)
Urban 0642*
Herbaceous/agricultural O. 198
Shrubland 0.384
Northern hardwoods -0.287
Aspen/birch 0.300
Lowland hardwoods 0261
Dry hardwood/conifer mix 0.328
Wet hardwood/conifer mix 0.205
Wetlands -0485
Red pine 0.307
Jack pine 0386
Mixed pines -0239
Tamarack 0.007
White spruce -0261
White cedar -0015
Mixed conifer -0.655*

 

* Probability level 5010

167

Table 50 Spearman rank correlations (rs) between landscape metrics calculated from
1991 satellite imagery and pileated woodpecker relative abundance (1996, 1997, and
1998, combined) on study sites in Michigan’s Upper Peninsula.

 

 

 

Landscape metric r, (n=10)
Total area 0. 184
Number of patch types 0.366
Patch size 0437
Median patch size -0.403
Edge density (m/ha) 0.423
Patch edge -0437
Shape index 0068
Area weighted shape index 0847*
Patch fractal dimension -0137
Area weighted fractal dimension 0703*
Interspersion and juxtaposition index -0.280
Shannon’s diversity index 0.116
Simpson} gvgnness ingex L

* Probability level 3010.

168

The number of pileated woodpeckers observed was too small to perform either a
correlation analysis with compositional characteristics of home-range sized circles or a
comparison of potential home ranges where woodpeckers were present or absent.

Barred owl

Positive correlations with barred owl response frequency were found for the
proportion of herbaceous/agricultural and shrubland cover types when all 3 years of data
were combined (Table 51). Additionally, there was a negative correlation with the wet
hardwood/conifer mix type in 1996, and with the wetlands cover type in 1997, and there
was a significant positive association with tamarack in 1997. However, trends for these
relationships were not constant among years, making it difficult to assign meaning to
them. Shannon’s diversity index was also positively correlated with barred owl response
frequency for the 3 years of pooled data, but fluctuated among individual years (Table
52).

In a comparison of home range sized (200 ha) circular areas surrounding points
where barred owls were detected with points where they were not detected in any of the 3
sampling years, dry hardwood/conifer mix and northern white cedar were the only cover
types which differed significantly in the proportion of the area that they composed (Table
53). The dry hardwood/conifer mix made up a larger proportion of the home-range sized
areas where owls were detected, but cedar was not present in any of the 200 ha areas
where owls were heard. There was also a nonsignificant trend of the wet

hardwood/conifer mix being more prevalent in areas where owl responses were not

recorded.

169

Table 51. Spearman rank correlations between proportion of vegetation cover types in
the landscape and barred owl relative abundance (1996, 1997, and 1998) on study sites in
Michigan’s Upper Peninsula.

 

 

 

Sampling period
1996 1997 1998 A113
Land cover type (n=7) (n=8) (n=10) years
(n=10)
Urban -0493 -021 1 0.3 15 0.068
Herbaceous/agricultural 0.214 0.216 0.195 0564*
Shrubland -0607 0838* 0.262 0766*
Northern hardwoods 0.286 0.263 0.116 0.297
Aspen/birch 0.214 -0419 0.231 0. 103
Lowland hardwoods 0.408 -0498 0.000 0290
Dry hardwood/conifer mix 0.107 -0.371 0.158 0.164
Wet hardwood/conifer mix 0714* -0503 -0158 -0.067
Wetlands 0.536 -0.683* 0. 109 0467
Red pine -0 134 0.016 -0. 143 -0147
Jack pine -0020 -01 10 -0.045 0.172
Mixed pines 0.107 -0.048 0.219 0.018
Tamarack -0408 0740* -0 169 0.3 19
White spruce 0.612 -0083 0.000 0.290
White cedar 0.045 0.063 0.338 0.320
Mixed cnnifer 0.35 0.012 0.182 -0 00

 

* Probability level 3010

170

Table 52. Spearman rank correlations between landscape metrics calculated from 1991

satellite imagery and barred owl relative abundance

in Michigan’s Upper Peninsula.

(1996, 1997, and 1998) on study sites

 

 

 

Sampling period

1996 1997 1998 All 3 years
Landscape metric (n=7) (n=8) (n=10) (n=10)
Total area 0.250 -0.323 0.474 0.333
Number of patch types -0. 147 -0025 0.475 0.539
Patch size 0.429 -0.036 0.316 0.127
Median patch size 0.321 -0.323 0.419 -0. 164
Edge density (m/ha) -0464 -0 180 -0073 -0212
Patch edge 0.357 -0.371 0.097 -0515
Shape index 0.250 -0.539 -0.152 0515
Area weighted shape index -0607 -0.036 0.188 0.32]
Patch fractal dimension 0.071 0.000 -0.310 -0. 176
Area weighted fractal dimension -0.643 -0.204 -0067 -0 176
Interspersion and juxtaposition -0429 0695* 0.061 0.418
index
Shannon’s diversity index -0.357 0.431 0.511 0758*

 

 

* Probability level $0.10.

171

Table 53. Proportion of home ranges (%) in each cover type at sampling points where
barred owls were detected and points where they were not detected for 1996, 1997, and
1998 combined on study sites in the Upper Peninsula of Michigan.

 

Owls detected Owls not detected Probability

 

 

(n=33) (n=4 1 ) level
Land cover type >‘< S.E. >7 S.E.
Urban 0. 15 0.0 0.88 0.4 0.550
Herbaceous/agricultural 2.23 0.9 2.63 l .3 0. 148
Shrubland 0.38 0.1 1.33 0.7 0.763
Northern hardwoods 67.46 5.8 66.70 6.1 0.336
Aspen/birch 2.63 1.3 2.74 1.6 0.221
Lowland hardwoods 1.28 0.0 0.00 0.0 0.265
Dry hardwood/conifer mix 6.52 3.0 2.54 1.1 0.067
Wet hardwood/conifer mix 1.46 0.8 7.42 3.6 0.1 l l
Wetlands 5.05 3.5 1.02 0.7 0 160
Water 6.44 3.4 7.39 3.7 0.323
Red pine 0.1 l 0.0 0.58 0.1 0.319
Jack pine 2.69 0.6 1.17 0.4 0.613
Mixed pines 1.88 1.2 1.55 0.8 0.506
Tamarack 0.52 0.2 0.19 0 1 0.446
White cedar* 0.00 0.0 1.62 0.4 0.067
Mixed conifer 1.19 W 0L

 

 

 

“ Probability levels for were calculated with the Kruskal-Wallis one-way analysis of
variance (Siegel and Castellan 1988).

* Probability level $010.

172

Fisher

Indices of fisher activity calculated over 3 years for MDNR, Forest Service, and
timber industry sites (Table 54) exhibited significant positive correlations with the
proportion of urban and herbaceous/agricultural cover in the landscape. The next largest
correlation was also positive, though not significant, and occurred for the proportion of
shrubland. Area-weighted shape index was the only landscape attribute to have

statistically significant relationship with the fisher population index (Table 55).

173

Table 54. Spearman rank correlations (rs) between proportion of vegetation cover types
in the landscape and fisher relative abundance (1996, 1997, and 1998 combined) on study
sites in Michigan’s Upper Peninsula.

 

 

Land cover type rs (n=9)
Urban 0685*
Herbaceous/agricultural 0.661 *
Shrubland 0.562
Northern hardwoods -0.068
Aspen/birch 0.220
Lowland hardwoods -0279
Dry hardwood/conifer mix 0.475
Wet hardwood/conifer mix -0153
Wetlands -0288
Red pine 0.000
Jack pine 0.093
Mixed pines -0170
Tamarack 0.190
White spruce 0.000
White cedar 0.395
summer 4.441

 

* Probability level 5010

174

Table 55. Spearman rank correlations between landscape metrics calculated from 1991
satellite imagery and fisher relative abundance (1996, 1997, and 1998, combined) on
study sites in Michigan’s Upper Peninsula.

 

 

Landscape metric rfi (11:9)
Total area 0.322
Number of patch types 0.496
Patch size -0085
Median patch size -0068
Edge density (m/ha) 0.119
Patch edge -0.220
Shape index -0102
Area weighted shape index 0814*
Patch fractal dimension -0.068
Area weighted fractal dimension 0.458
Interspersion and juxtaposition index -0102
Shannon’s diversity index 0.305
Simn§on’§ gvenness index M1

 

* Probability level 3010.

175

DISCUSSION
Landscape composition and structure

The structure and composition of the Upper Peninsula landscape is a result of a
combination of historical and contemporary factors, both of which have natural and
anthropological components. Thus, one would not necessarily expect landscape patterns
to follow management boundaries exactly, and it is not surprising that few significant
differences in landscape structure were found among ownerships in this study at the scale
and resolution analyzed. The lack of statistical differences in landscape composition was
also not surprising, because study sites were deliberately chosen to be dominated by
northern hardwoods, leaving little room for large variations in other cover types.

Patch density tends to increase with increasing fragmentation (Reunanen et al.
2000), but is not necessarily indicative of fragmentation. The high patch density and
large amount of edge at the Huron Mountain Club are a result of the diversity of cover
types that occur along topographic gradients throughout the site. In contrast, timber
industry sites had a more homogeneous landscape structure, as well as a lower
representation of land cover types.

Despite the fact that road density calculations did not yield statistically significant
results, roads may still have a measurable influence on forested landscapes. Roads may
impact wildlife by fragmenting the habitat, providing an avenue for the spread of exotics,
and altering hydrology (Trombulak and Frissell 2000). Roads may also affect
microhabitat conditions such as soil and litter attributes. Haskell (2000) found a

relationship between the presence of a road and a reduction in leaf litter depth up to 100

176

m into the adjacent forest. Roads also impacted prey availability for forest birds and
salamanders by reducing the abundance and diversity of macroinvertebrates in soil up to
15 m from the road. The effects were theorized to be caused by increased exposure to
wind, resulting in drying and accelerated decomposition of the litter, as well as physical
displacement of leaf litter.

Relationships between wildlife species relative abundance and landscape
characteristics

In evaluating species responses to landscape characteristics, it is necessary to make

some arbitrary definitions of what constitutes a patch or an edge, when in fact, most
landscape features can be described along a continuum. Therefore, the landscape
variables calculated represent only 1 point on this continuum, and can only be interpreted
for the scale and resolution at which they were measured. The scale and resolution of the
available data may or may not be meaningful to a wildlife species, depending on the grain
size at which a species perceives heterogeneity. For example, satellite imagery may be
able to distinguish patches based on the dominant tree species in the overstory, but within
a patch of land with similar overstory composition, distinctions among forest stands of
different ages or with different understory composition may not be discernible.
Additionally, remote sensing may indicate where edges exist between vegetation types,
but it does not provide information on how much of a contrast the edge represents to a
species. Wildlife responses are not likely to be uniform for different points along the
continuum of possible landscape definitions, and it is important to interpret results of

landscape analyses based on the natural history of the species under investigation.

177

The red-backed salamander is one species which may not perceive landscape
attributes as coarse as those measured from satellite imagery in this study, and very few
land cover composition characteristics or metrics described the abundance of red-backed
salamanders (Tables 39 and 40). Rather, red-backed salamanders were most strongly
influenced by characteristics of their immediate environment, as discussed in Chapter 1.

Evaluations of the effects of landscape level characteristics on red-backed
salamanders are scarce and inconclusive. Rodewald and Yahner (1999) found that red-
backed salamander p0pulations in a relatively unfragmented forest were not influenced by
the proportion of forest cover in the landscape. It may be that salamanders do not
perceive landscape level effects because of the salamander’s limited home range size, or
that effects occur at scales other than those examined (Rodewald and Yahner 1999). One
type of landscape change that may impact salamanders is the creation of forest edges
through silvicultural treatments or road construction. DeMaynadier and Hunter (1998)
determined that red-backed salamanders were highly sensitive to the effects of clearcuts
on microclimate conditions up to 35 m into an adjacent mature forest.

Many landscape attributes are ecologically relevant because they indicate the degree
of habitat fragmentation, which is a concern for its potential impacts to wildlife,
particularly song birds. A related concept is that of connectivity, which describes the
degree to which animals can access suitable habitat through dispersal. Connectivity is
impacted by habitat degradation and fragmentation, and is relative to both an animal's life
history and to landscape structural characteristics (Reunanen et al. 2000). Connectivity

does not have a mathematical definition, but can be partially described by the degree of

178

dispersion and interspersion of landscape patches.

In this study, ovenbird relative abundance was negatively correlated with patch edge
and positively related to patch fractal dimension, interspersion and juxtaposition index,
and the 2 diversity indices (Table 44). Ovenbird numbers were also positively related to
the pr0portion of the landscape in a conifer cover type (Table 43). It may be that the
conifer component on these sites was responsible for the higher landscape diversity
indices that were also associated with ovenbirds, and it is difficult to determine which
variable ovenbirds may be responding to.

There are conflicting conclusions regarding the sensitivity of ovenbirds to habitat
fragmentation (King et al. 1996, Merrill et a1. 1998, Pomeluzi and Faaborg 1999). One
consistent finding has been that ovenbirds compensate for lower nest success along edges
by laying larger clutches and renesting (Flaspohler et al. 2001). In northern and mixed
hardwood forests in Wisconsin, clutch size was higher in edge habitats, yet nest success
was higher in interior habitats, resulting in comparable productivity in the 2 habitats
(Flaspohler et al. 2001). These differences were apparent at about 300 m from a clearcut.
However, because the sources of decreased nest success (i.e., cowbird parasitism, ground
predators) depend on local responses of nest predators and brown-headed cow birds to
increasing amounts of edge, the effect of landscape fragmentation on ovenbird fitness
may be specific to a particular landscape. The positive association with the interspersion
and juxtaposition index for the ovenbird and the veery was perhaps an indication of the
importance of connectivity to these species. The veery in particular often uses lowland

hardwood forests (Howe et al. 1995) that may be associated with riparian areas.

179

Hunt (1998) showed that American redstart numbers were positively associated
with the proportion of early successional habitat in the landscape. In this study, the
significantly negative correlation between the proportion of the urban cover type in the
landscape and the relative abundance of American redstarts (Table 38) may reflect the
finding that the Huron Mountain Club had the largest representation of the urban cover
class, but at the same time there is little early successional habitat that would favor
redstarts. American redstarts are also believed to prefer less conifer cover than is found
at the Huron Mountain Club (Minnis and Haufler 1994) (Table 5). Therefore, the
negative correlation reflects the lower quality habitat at the Huron Mountain Club based
on variables other than the prevalence of the urban cover type.

Time constraints of this project only permitted examination of 2 possibilities of
what might be an appropriate scale to analyze barred owl habitat use (i.e., 200 ha and 12-
19 kmz). However, in a similar study of sandhill crane habitat use reported by Baker et al.
(1995), sandhill cranes were found to select habitat based on features in a core area within
the total home range that they used, and few differences in habitat characteristics were
found by comparing areas the size of an entire home range. Likewise, an analysis of
barred owl habitat use at a scale slightly smaller than 200 ha, or a comparison of several
different scales of finer resolution data, might reveal more significant landscape
characteristics important to barred owl habitat selection than were evident in this analysis.
Additionally, as in Chapter 1, the quality of the barred owl population data may have
limited the analysis.

One landscape characteristic that has been linked to fisher habitat quality is the

180

proportion of conifer cover in the landscape. Conifer cover is thought to be important in
the winter for reducing ground snow depth and allowing fishers to forage more efficiently
(Krohn et al. 1995). If so, then the Huron Mountain Club may provide the best fisher
habitat because of the dominance of hemlock in some northern hardwood stands, the
stands of pure hemlock that have been mapped there, and the larger area covered by the
wet hardwood/conifer mix cover type. The relative importance of conifer cover may also
vary geographically with average yearly snowfall and with local marten densities (Krohn
et al. 1995).

This analysis examined the relationship of land ownership and wildlife populations
with landscape variables that were discernible at a minimum patch size of 900 m2.
Intensive activities, such as timber harvests, that remove the entire canopy may have been
interpreted as distinct patches on satellite imagery, but less intensive or finer scale land
management activities such as forest stand thinning and logging road construction were
not visible at this scale. Analysis of an additional spatial data layer, representing the scale
at which management prescriptions are applied (i.e., forest stands), would help fill the gap
between the very fine resolution of the vegetation field data collected and the coarser
satellite imagery.

Finally, the absence of strong relationships between landscape characteristics and
wildlife relative abundance, and the apparent similarity in landscape characteristics
among ownerships should not be taken as evidence that management approaches do not

affect the landscape. The effects of management activities that may initially only be

evident at a small scale may have the potential to accumulate over a much broader spatial

181

and temporal extent (Everett and Lehmkuhl 1999). For example, some forest birds may
not respond directly to a propagation of small forest openings across a landscape, but
those openings could alter local deer abundance, which in turn could impact forest

understory characteristics, and alter habitat quality for some species.

182

CHAPTER 3 - Old Growth Characteristics and Wildlife Habitat in the Reserve and
Nonreserve Areas of the Huron Mountain Club
INTRODUCTION

The Huron Mountain Club is ecologically and historically one of the most unique
places in Michigan. The club was established in 1889 with 2,800 ha by a group of
prominent families, who agreed at that time to leave the forest and lakes in their natural
state. Even in 1938 when people could still remember similar forests before they were
logged in other parts of the Upper Peninsula, the property was recognized by Aldo
Leopold as a site of tremendous ecological value. When Leopold was invited to the
Huron Mountain Club as a consultant to research the area and prepare a management plan
(Monthey 2000), he determined that the biggest threat to forest health there was deer
overpopulation. The forested property in the 1930s was not capable of supporting the
deer densities in the surrounding area, yet artificial feeding and logging encroachment
were attracting deer to the area (Monthey 2000).

Leopold’s written plan for the Huron Mountain Club recommended that the Club
reserve the central area, nearly 2,500 ha, from timber harvesting and use only light
selective cutting in a surrounding buffer area (Simpson et al. 1990). At the time the plan
was written (1938), selective cutting was a novel alternative to clearcutting (Monthey
2000). As a result of Leopold’s recommendations, the only cutting that has occurred in
the reserve area was the removal of eastern white pine from about 20% of the 2,500 ha
between 1885 and 1900. The surrounding nonreserve area of the Huron Mountain Club

was last out in the 19205 and 19405 (W estover 1971, Simpson et al. 1990). Today, much

183

of the Huron Mountain Club retains the character of a presettlement forest. The general
public is not allowed access to the land, and several of Michigan’s largest living trees are
located there (Wells and Thompson 1972).

Old growth forest, such as that at the Huron Mountain Club, is important not only
because of its rarity, but because of the many unique features associated with it. As forest
succession proceeds, stand structure becomes more heterogeneous and complex,
providing a suite of conditions that are not completely represented in younger forests
(Welsh and Droege 2001). An important issue in old growth management has been
identifying which individual elements of old—growth forests wildlife respond to, and how
those qualities can be preserved or replicated through forest management.

As a relatively unmanipulated ecosystem, the hardwood forests of the Huron
Mountain Club can serve as an ecological baseline or “control” against which the effects
of management activities can be measured (Arcese and Sinclair 1997). The Huron
Mountain Club can also be used as a reference point for evaluating the long-term effects
of succession and natural disturbances. Identification of specific properties that are
associated with old growth forests may then provide managers with the knowledge to
emulate old growth characteristics in a more disturbed setting. This knowledge would
allow natural resource managers to maintain a wider range of ecological processes and
biodiversity in a landscape. For these reasons, the objective of this chapter is to compare
forest and landscape structure and composition, and relative abundance of selected
wildlife species, in portions of the Huron Mountain Club that have not been out since the

1890s and in areas that were last harvested in the early part of the 20th century.

184

METHODS

Reserve and nonreserve area boundaries were determined from Simpson eta1.’s
(1989) map of cover types of the Huron Mountain Club and from (Westover 1971). In
1996, vegetation and wildlife surveys were conducted in 2 reserve areas of the Huron
Mountain Club, and in 1997 and 1998 data were collected from the 2 reserve areas and
from a portion of the nonreserve area. Data collection was focused on the 3 areas of the
Huron Mountain Club where the largest blocks of northern hardwood forest were located.

Wildlife population data collection was described in Chapter 1. Methods for
generating cover type maps and calculating landscape metrics were generally the same as
those outlined in Chapter 2. Sampling sites at the Huron Mountain Club were in
relatively close proximity to each other (>l.3 km) and usually separated by more water
area than land. Therefore, boundaries of study sites were created using only 80 m buffers
to avoid the overlap that would occur if 800 m buffers were used.
Data analysis

Values for stand structure and composition variables were calculated by averaging
data for stands within each sampling site, and then averaging the results of the 2 reserve
sites. Comparisons of wildlife relative abundance, stand structure and composition, and
landscape variables between reserve and nonreserve areas were not tested statistically
because of the small number of replicate sites. Instead, data are presented as means for

each group and described qualitatively in terms of the history of each site.

185

RESULTS
Forest stand characteristics

Of the 14 reserve stands sampled, 12 had been cut for eastern white pine in the
1890s, and the remaining 2 have never been logged (A. Turner, Huron Mountain Club,
personal commun.). All of the nonreserve area stands were cut for white pine in the same
time period as the reserve stands and then clearcut from 1919-1923 (W estover 1971).

As expected in forests that have been harvested, the nonreserve area tended to
have more tree stumps than the reserve area. However, stump diameter and height were
slightly larger in the reserve area (Table 56). Generally, there were higher shrub densities
and slightly more rnidstory canopy cover in the understories of reserve area stands than in
the nonreserve area.

Basal area was similar for reserve and nonreserve stands, but trees in the reserve
area tended to be larger in diameter and there were fewer of them (Fig 13). In contrast,
the nonreserve area had a larger number of smaller diameter trees. Snag occurrence
followed a similar pattern, with reserve area stands having about 13% fewer snags, but
snag diameters that were approximately 41% larger.

Reserve stands had 22% more conifer cover in the overstory than nonreserve
stands (Table 57), yet stem densities of hemlock were actually greater in the nonreserve
area (Table 58). Tree species recorded in the reserve area that did not appear in the
nonreserve area were American basswood, white spruce, and white ash (Table 58).
Quaking aspen, black cherry, and white birch were recorded in the nonreserve area, but

not in the old growth portion of the Huron Mountain Club. The nonreserve area also had

186

Table 56. Mean values for understory vegetation variables measured for stands in the
reserve and nonreserve areas of the Huron Mountain Club in Michigan’s Upper
Peninsula, July and August, 1996, 1997, and 1998.

 

 

Reserve Nonreserve

Variable (n=2)“ (n=l)b
Saplings/ha 288.2 616.7
Shrubs/ha 1523.9 413.3
Shrub height (cm) 82.0 119.5
Stumps/ha 68.9 170.0
Stump height (cm) 67.9 53.8
Stump diameter (cm) 41.4 33.1
Logs/ha 190.0 176.7
Log length (m) 5.9 5.3
Log width (cm) 29.9 24.8
Herbaceous height (cm) 7.9 6.6
Litter depth (cm) 3.4 2.8
Vertical cover (%)

00.5 m 15.0 9.6

0.5-5 m 17.7 13.6

Herbaceous 8.0 6. 1

All shrub species 0.08 0.01

Deciduous shrub species 0.08 0.01

Woody erris 14 5,2

 

“ Mean values were obtained by averaging the stand values for each reserve site to obtain
a mean for each site, and then averaging the means of the 2 sites.

b Mean values were obtained by averaging values for the 8 stands sampled on the l
nonreserve site.

187

60 Reserve area

 

 

10 20 30 40 50 60 70 80 90 100

Tree diameter class (cm)

Nonreserve area

  

 

 

 

 

 

 

Egg En man

0 II1 I I Ifi I I I I I ITI I I—I I I I I I fil I W I—I—I—I

10 20 30 40 50 60 70 80 90 100

Tree diameter class (cm)

 

Figure 11. Tree diameter distributions for the reserve and nonreserve areas of the Huron
Mountain Club in Michigan’s Upper Peninsula, 1996 - 1998.

188

Table 57. Mean values for overstory vegetation variables measured for stands in the
reserve and nonreserve areas of the Huron Mountain Club in Michigan’s Upper
Peninsula, July and August, 1996, 1997, and 1998.

 

 

Reserve Nonreserve

Variable (n=2)a (n-_-1)b
Basal area (m2/ha) 49.6 46.0
Overstory trees/ha 518 807
Overstory tree DBH (cm) 36.1 27.2
Overstory tree height (m) 21.6 20.1
Snags/ha 45.2 52.2
Snag height (m) 11.3 9.9
Snag diameter (cm) 37.5 22.0
Vertical cover (%)

>5 m 81.2 85.4

Conifer trees 52.6 39.9

Deciduous trees 54.6 62.9

Hard mast trees >10 in (25.4 cm) DBH 0.79 0.00

 

“ Mean values were obtained by averaging the stand values for each reserve site to obtain
a mean for each site, and then averaging the means of the 2 sites.

b Mean values were obtained by averaging values for the 8 stands sampled on the l
nonreserve site.

189

Table 58. Overstory tree species composition (stems/ha) for stands in the reserve and
nonreserve areas of the Huron Mountain Club in Michigan’s Upper Peninsula, July and
August, 1996, 1997, and 1998.

 

 

Reserve Nonreserve
Tree species (Scientific name) (n=2)’ (n=1)b
American basswood (Tilia americana) 20 0
American elm (Ulmus americana) 19 23
Bigtooth aspen (Populus grandidentata) 1 33
Black cherry (Prunus serotina) 0 1
Eastern white pine (Pinus strobus) 1 3
Hemlock (Tsuga canadensis) 299 343
Northern red oak Quercus rubra 4 3
Northern white cedar (Thuja occidentalis) 3 49
Quaking aspen (Populus tremuloides) 0 3
Red maple (Acer rubrum) l 221
Striped maple (Acer pensylvanicum) 7 4
Sugar maple (Acer saccharum) 127 93
White birch (Betula papyrifera) 0 1
White spruce (Picea glauca) l 0
White ash (F raxinus americana) 3 0
Yellow birch (Betula alleghaniensis) 29 45
All species combined 518 823

 

’ Mean values were obtained by averaging the stand values for each reserve site to obtain
a mean for each site, and then averaging the means of the 2 sites.

b Mean values were obtained by averaging values for the 8 stands sampled on the 1
nonreserve site.

190

a substantially higher density of red maple than the reserve area, as well as more bigtooth
aspen and northern white cedar.
Landscape characteristics

The 2 reserve sites were both smaller than the nonreserve site because they
bordered on and were surrounded by lakes (Table 59). Although they spanned several
township sections, much of the area was occupied by water rather than land. Despite
their smaller size, the reserve area sites contained more cover type classes than the
nonreserve area. The urban, agricultural/herbaceous opening, wetlands, and jack pine
cover classes were present in the old growth study sites but absent from the nonreserve
site (Fig. 12). Northern hardwood forest made up a slightly larger proportion of the
nonreserve study site than the old growth sites, and nearly 25% of the area of the
nonreserve site was in dry hardwood/conifer mix category. Conversely, there was very
little of the dry hardwood/conifer mix cover class on the old growth sites, but close to
30% of the area was covered by the wet hardwood/conifer mix cover type (Fig. 12).

Edge density, patch density, the interspersion and juxtaposition index, and
Shannon’s diversity index tended to be higher on the reserve sites than on the nonreserve
site (Table 59). Conversely, patch size was larger for the nonreserve site.
Wildlife species

In both years of salamander data collection, relatively few salamanders were
found in stands on I reserve site, an intermediate number were found at the second
reserve site, and the most were found in the nonreserve area. During summer surveys in

1997 and 1998, a total of 111 salamanders/ha were located in old growth stands through

191

Table 59. Mean values for landscape metrics for stands in the reserve and nonreserve
areas of the Huron Mountain Club in Michigan’s Upper Peninsula, calculated from 1991
satellite imagery.

 

 

 

192

Landscape metric Reserve Nonreserve
(n=2) (n=l)

Total area (kmz) 5.88 7.23
Number of patch types 9 5
Patch size (ha) 11.29 15.73
Patch density (#lkmz) 3.42 2.93
Median patch size (ha) 0.79 1.95
Edge density (m/ha) 120.04 110.33
Patch edge (km) 1.24 1.70
Shape index 1.59 1.55
Area weighted shape index 2.20 2.44
Patch fractal dimension 1.46 1.35
Area weighted fractal dimension 1.30 1.34
Interspersion and juxtaposition index (%) 62.53 49.54
Shannon’s diversity index 1.54 1.27
Simpson’s evenness inggx 0:25 All

 

70a

604

50 4
IReserve E] Nonreserve

8

U)
C
L

Proportion of total area (%)

 

 

 

 

 

Land cover type

Figure 12. Average proportion of total land area in each land cover class for study sites in
reserve and nonreserve areas of the Huron Mountain Club in Michigan’s Upper
Peninsula, calculated from 1991 satellite imagery.

193

ground searches, and more than double that number (272 salamanders/ha) were located in
nonreserve area stands during the same time period. Numbers of salamanders found
under cover boards followed a similar trend, with 25 individuals/ha found under boards
on the reserve sites and 44 salamanders/ha under boards on nonreserve sites. Also, a
larger proportion of salamanders located through ground transect searches on the
nonreserve site were found under the smallest cover objects, while salamanders in old
growth stands tended to be distributed more equitably under all available cover object
sizes (Fig. 13). All but the smallest (0—5 cm) size classes of woody debris recorded
during salamander searches were more abundant in old growth stands than in nonreserve
area stands (Fig. 14).

Several bird species, including the black-throated blue warbler, blackbumian
warbler, least flycatcher, and northern parula (Parula americana), were observed during
surveys in the reserve area, but were not recorded in stands in the nonreserve area (Table
60). Some species, such as the downy woodpecker (Picoides pubescens), veery, and
yellow warbler, were only recorded in the nonreserve area, and although the hermit thrush
was detected on all 3 sites, it was much more abundant in stands surveyed in the
nonreserve area. The American robin, American redstart, hairy woodpecker (Picoides
villosus), and pileated woodpecker were observed at least twice as frequently on the
nonreserve site as on reserve sites. The red-eyed vireo, ovenbird, and red-breasted
nuthatch also tended to be more common in the nonreserve area (Table 60).

Barred owl response rates to taped calls varied greatly during the 3 summers

surveys were conducted (Fig. 15), and there was not a discernible pattern to the results.

194

120.00 1

 

 

 

100.00 4
lReserve ElNonreserve
80.00 4
(‘6
E
45
g 60.00 J r—
E
'3
m
40.00 ‘

 

 

20.00 1

 

 

 

 

 

 

h. -

5- 10 10-20 20-30 30-40 > 40

Cover object width (cm)

0.00 1

 

Figure 13. Abundance and size of woody debris used by salamanders for stands in the
reserve and nonreserve areas of the Huron Mountain Club in Michigan’s Upper Peninsula
June, July, and August, 1997 and 1998.

195

Cl Nonreserve

I Reserve

 

,...;4;.4.....l..4

......~............s..e..axttt..t

...»..\.s;..t..n....
...;a...a..¢oyv.

 

    

              
              

.......\‘...‘..l.....‘v..............\...~...:»\.~...~...~.4..
..........I.........1..........o........ou..a......ala.aanl...

 

...,...........y .. . . .. ..
1.4....

........l..:.;rnv....c.....l.:41..4a.......u.4c‘44....-

 

., ..t.‘\...... ..

....t.....-......n..I...l...l~........u.n....u...........«

    

 

r14.l.-.o.,1p....p..lv.c».;,.1p.....c....v..
v:.u.v.~‘tuaxvvussvsuu“‘\"~‘uu..vc......-.v

\f‘..u.v.‘y‘stu~\‘\~‘vvuvuss‘\~‘.

     

                             
                         

 

....... .....i.............s......
~a<~.1‘....w.‘.v..~u1~.-..a...w:..aa.nila

 

..a.....a..a...aa.-...-‘q~.1v.o~.«......4.‘J‘..a;ac..

......»vav.v».......y.,..'.v..o...v.o.....Ilun-.a.l.

 

                    

 

x...
a)..4;.a’o.:¢.¢:.nu4a.1pr

 

..‘......q...‘.....u..-.-..

  

.:.«.'¢;-.at....vpv.'olpr.....
...-auunc.......c..‘a...‘.......\....
;.A1....aauo4'o1;.11.a;-............-4.44.‘..u...4...v
rs....\.-..-~.\.\.s~s........c»..k~........~\...x..........t.\.............s...s...vs..
....a........a.......at..4........4o....4.......4;..4..aa.aa.4.cca..44......

..;........l....l.p...:...¢.o.....oa..ca.....1......4111»;a;.444o.ao...41..aa.aa.a..aa.s:.:44....

     

 

. ..... ....

 

.. .c.................. ..... .
~.a........-..‘-v.s........n....».-....-.......
..4c...n44.raa....a..—........4...a.ano..4....

 

 

 

 

   

................‘..-...s..........a-‘.sn

.. ..,................................a

...................v‘........~t..~...
..................

       

 

..........................av.t

   

 
 

,l.lc....a..¢;........4.....4..4.:.....c4

muss-x..A.s.k..s.s...snssaslas..‘e..nssea-x...u..~.~‘.ys».ssk~..--“a\.-uus\~aa...
....4....c...u.....4....a..44...c.....a....4.4a‘.a—.1¢.aa....‘naaau.a...a...o....4.
....‘x....-...x.........-...».........-.ust....‘»..s....\.........\~...»\I.s......

...i..........».........~....v...v...s.»......t.k......n...»..».................\‘~...‘..

.l..ll,ll

1a...

 

                 
            

., ...,r. ,%.,..rl..,........,'4...41(...4... .I.....al.z1.......

 

....1.1.....ta....41.41...aa-n.olua..4.non-....npooosolntna...a.

 

.,/.........l.l......lv.i..l.42.......

.. .t.r.......~‘. ....~.,...‘...

., . I . , .1... v .1 i .. .. _. ... , .. . ...

...V“..........‘;...‘.,.\....~..,.........v...c.......t....\.~..:...v.....t.

.... ......l..........v....u......u...........o:.........-...Ia..\nl............-.....u.»......n.........i..-\.....

 

.rvaa.....na....:1........Ia44..1.....;..........r..............4......4.........

 

 
 
 
 
 
 
 

 

 

 

      

....c........v..\~........\.....‘....‘.......s.....\..........l.

  

 

.44........a‘.....a.4t..............a..:......n....4.......

 

 

1000a
1

«53er

30-40

2030
Width (cm)

10-20

5-1

Figure 14. Abundance and size of woody debris for stands in the reserve and non reserve

areas of the Huron Mountain Club in Michigan’s Upper Peninsula June, July, and August,

1997 and 1998. Note that the scale of the y axis is logarithmic.

196

Table 60. Mean absolute frequencies (percent of points at which species occurred),
pooled over 3 years of data collection, for bird species surveyed in reserve and nonreserve
areas of the Huron Mountain Club in Michigan’s Upper Peninsula, May-July, 1996, 1997,
and 1998. Indicator species are in bold.

 

 

Reserve Nonreserve

Common name (n=2)a Qi=1)b
American crow 2,94 (100
American redstart 6.86 16.67
American robin 4.90 20.83
Black and white warbler 20.59 8.33
Blackburnian warbler 9.80 0.00
Black-capped chickadee 17.65 29.17
Black-throated blue warbler 5.88 0.00
Black-throated green warbler 77.45 100.00
Blue Jay 3.92 8.33
Brown creeper 5.88 4.17
Chestnut-sided warbler 0.98 0.00
Common raven 0.00 4.17
Downy woodpecker 2.94 8.33
Eastern wood-pewee 1.96 0.00
Great crested flycatcher 2.94 4.17
Hairy woodpecker 2.94 12.50
Hermit thrush 17.65 4.17
Least flycatcher 18.63 0.00
Magnolia warbler 0.98 4.17
Northern parula 4.90 0.00
Ovenbird 52.94 75.00
Pileated woodpecker 4.90 16157
Pine siskin 1.96 0.00
Pine warbler 1.96 0-00
Red-breasted nuthatch 9.80 25-00
Red-eyed vireo 69-61 91-67
Rose-breasted grosbeak 1-96 0'00
Scarlet tanager 0-98 0°00

[.96 4.17

Solitary vireo

197

Table 60 (Cont). Mean absolute frequencies (percent of points at which species
occurred), pooled over 3 years of data collection, for bird species surveyed in reserve and
nonreserve areas of the Huron Mountain Club in Michigan’s Upper Peninsula, May-July,

1996Ll997¥and 1998. Indicator species are in bold.

 

 

Reserve Nonreserve

Common name (n=2)a jn=ljb
Swainson's thrush 15.69 8.33
Tennessee warbler 0.98 0.00
Veery 0.00 12.50
White-breasted nuthatch 16.67 0.00
White-throated sparrow 0.98 0.00
Winter wren 36.27 16.67
Wood thrush 4.90 0.00
Yellow-bellied flycatcher 0.98 0.00
Yellow-ramped warbler 5.88 8.33

0.00 4.17

Yellow warbler

 

a Mean values were obtained by averaging the stand values for each reserve site to obtain
a mean for each site, and then averaging the means of the 2 sites.

b Mean values were obtained by averaging values for the 8 stands sampled on the l
nonreserve site.

198

—"' 'Reserve *Reserve *Nonreserve
n=8 n=6 n=8

\0
O
1

33‘6838
\
\
\
\
\>
/
/
/
/

Response frequency (%)

 

 

1996 1997 1998

Sample year

Figure 15. Average response rate of barred owls for stands in 2 reserve areas and l
nonreserve area of the Huron Mountain Club in Michigan’s Upper Peninsula, July-
August, 1996, 1997, and 1998.

199

DISCUSSION

Simpson (1990) found a correlation between the amount of hemlock in the
overstory and the structure of the tree seedling layer. In stands where the overstory is
dominated by hemlock, there may be <l % canopy coverage in the seedling and sapling
layers. As the proportion of hardwoods in the overstory increases, understory canopy
coverage may increase to 30—80% (Simpson 1990). This is consistent with data collected
from the old growth and logged portions of the Huron Mountain Club. Average
deciduous canopy cover was slightly greater in the reserve area, as was the average
density of shrub stems. It was also apparent from observations in the field that individual
stands followed the same pattern, with nearly pure hemlock stands generally having a
completely open forest floor and understory.

Although eastern white pine is known to have been a component of early northern
hardwood forests, it was actually slightly more prevalent in the nonreserve portion of the
Huron Mountain Club (Table 57). However, the stem densities observed (1-3 trees/ha)
are even lower than Price’s (1994) estimate of 18 white pine trees/ha in presettlement
Michigan northern hardwoods, reflecting the stands’ history of pine removal.

A result of logging in the nonreserve area may be an increased density of red
maple in the overstory. As a species that establishes opportunistically after disturbance,
red maple has increased in prevalence since European settlement. It can also respond as
both an early and late successional species (Abrams 1998), although it was nearly absent
from the reserve area.

One explanation for the lower stem densities of hemlock, and the higher shrub

200

densities in the reserve area is the presence of treefall gaps in the reserve areas. Tyrell
and Crow (1994) determined that old growth hemlock-hardwood forests are composed of
at least 10% canopy gaps, with an average gap area of 50 m2. Treefall gaps are in an
earlier successional state than the surrounding stand, and typical tree species include
sugar maple, basswood, yellow birch, and striped maple (Wells and Thompson 1972).
These young trees occur at relatively high stem densities, but do not provide much canopy
cover, so measurements of the deciduous overstory canopy were still low.

The size and abundance of dead and down woody material was one of the most
consistent differences between the old growth and previously harvested study sites. The
larger stump diameters on the old growth sites probably resulted from the natural falling
of mature trees. Hale et al. (1999) similarly found that an abundance of large coarse
woody debris distinguished old growth stands from mature managed forest stands in
Minnesota. Although the forests in their study were structurally different, plant species
diversity was similar, providing managers with the possibility of mimicking old growth
characteristics by selectively leaving a few large trees during a timber harvest, and felling
cull trees.

Nearly half of the old growth area of the Huron Mountain Club is described by
Simpson (1990) as belonging to the hemlock-northern hardwood forest type. Hemlock
dominated hardwood stands were occasionally established through fire, and the structure
and composition of an old growth hemlock-hardwood stand provides subtle clues about
the stand’s history. For example, Simpson ( 1990) determined that stands with >75 %

hemlock in the overstory were established by crown fires that destroyed the previous

201

forest, whereas hemlock growing in the understory of these stands is an indicator of past
ground tires which removed the ground vegetation and litter layer. In contrast, sugar
maple—hardwood stands indicate that there has been no fire establishment for several tree
generations, and this type of stand is thought to be one of the most constant and
unchanging types of forest stands at the Huron Mountain Club (Simpson 1990).

Landscape features tended to be more complex in the old growth portions of the
Huron Mountain Club than in the clearcut areas. One possibility for this is that the
landscape complexity is a result of natural disturbances that have acted on the landscape
and propelled ecosystem processes. For example, large areas of the reserve area forest
were established through fires in the last ZOO-300 years (Simpson 1990), and Frelich and
Lorimer (1991) estimated that up to 15% of the reserve area of the Huron Mountain Club
has been disturbed in any one decade since 1850.

A second explanation, not mutually exclusive to the first, is that some of the
landscape complexity of the old growth area inhibited logging before the Club was
established. The steep topography of the interior areas of the Huron Mountain Club
where no logging has occurred has resulted in a patchier distribution of vegetation types,
and it may also have discouraged serious attempts at logging. The nonreserve area that
was sampled is located near the shore of Lake Superior, and the topography is much
smoother, making it a much easier prospect for timber harvesting.

Unexpectedly, pileated woodpeckers were surveyed more often in the nonreserve
area than on the old growth sites. Snags, which are a component of the pileated

woodpecker HSI model, were more numerous on the nonreserve site, but had smaller

202

diameters than those in reserve sites. The HSI values calculated for the Huron Mountain
Club did not correspond to the number of pileated observations; instead, the lowest
pileated score (0.16) was associated with the nonreserve site. It may be that all 3 Huron
Mountain Club sites met the pileated woodpecker’s minimum requirements for snag
diameter, but the greater density of snags on the nonreserve site was more important in
determining habitat quality, despite the model predictions.

Northern parulas were heard singing near streams in the reserve area, but there
were no streams near sampling points in the nonreserve area, which may explain the
difference in abundance between the 2 areas. Blackburnian warblers are associated with
stands of hemlock, and measurements of coniferous canopy cover suggested that both
reserve areas sampled had a greater proportion of hemlock in the overstory than in the
nonreserve area of the Club. One explanation for the greater abundance of American
robins in the nonreserve area may be that robins are more tolerant of human disturbance
than other species, and the nonreserve area was close to a more heavily used road than the
2 reserve area sites.

The birds found in the nonreserve area represented a wide range of successional
associations. Some early successional species, such as the veery and American redstart,
were more common in the logged than the old growth area, yet the hairy and pileated
woodpeckers, which tend to be found in mature forests, were also more common in the
nonreserve area. Nonreserve area stands also tended to be dominated by a few species, as
shown by the high relative abundance values for the black-throated green warbler,

ovenbird, and red-eyed vireo (Table 60). Bird communities in the reserve area were

203

represented by a larger number of species, but relative abundances of individual species
tended to be lower than in nonreserve stands. This diversity may be a reflection of the
structural complexity that existed in some uncut stands, where treefall gaps were often
interspersed with enormous old trees. Therefore, implementing management activities
that mimic the large degree of stand complexity in old growth stands and landscapes may

be another approach enhancing regional biodiversity.

204

CONCLUSIONS

The diversity of forest management approaches among state, federal, timber
industry, and the privately owned forest land in the eastern Upper Peninsula has resulted
in a heterogeneous landscape, and a broad range of wildlife habitat conditions. Some the
most conspicuous differences in forest stand characteristics occurred between Huron
Mountain Club and timber industry sites, with MDNR and Forest Service sites falling in
the middle.

Landscape composition did not differ among ownerships, and compared to
wildlife and forest stand variables, there were few structural landscape differences among
ownerships. Available data on landscape structure and composition were relatively
coarse compared to the home ranges of most of the species evaluated, and compared to
the level at which forest manipulations take place. As a result, landscape structure and
composition were only minimally influenced by land ownership patterns at the scale and
resolution examined.

Species selected for population and habitat evaluation in this project were chosen
as a sample of species that occur under different conditions in northern hardwood forests.
Thus, their relative abundance and distribution were expected to vary across a landscape
with varying forest conditions. The relative abundance of red-backed salamanders varied
considerably at a relatively small spatial scale, with few significant differences in relative
abundance observed among the 4 ownerships. Additionally, 2 salamander survey

methods were compared, and it was determined that artificial cover board searches are

205

preferable to ground transect searches when surveys will be repeated at the same location.
Although cover board searches yielded fewer salamander observations than ground
searches, the cover board method offers the benefits of being a standardized method that
is less disruptive to the habitat.

Distinctive bird communities were also associated with each ownership category.
For example, timber industry sites had a greater relative abundance of early successional
species, such as the American redstart and the veery, and fewer ovenbirds than all other
ownerships. Cavity nesting birds, including the pileated woodpecker, were generally
most common at the Huron Mountain Club. In general, MDNR and Forest Service sites
had relatively similar bird communities, while timber industry and Huron Mountain Club
sites had more divergent forest bird associations. Relative abundances of pileated
woodpeckers and the 4 songbird species selected for their associations with northern
hardwood forests also differed among ownerships. No differences in barred owl or fisher
use of study sites were identified.

Data collected at the Huron Mountain Club provided an important reference point
for the range of wildlife habitat conditions that may occur in northern hardwood forests.
One of the most consistently documented attributes of old growth forests is the abundance
of coarse woody debris and standing dead wood (Tyrell and Crow 1994, Carey and
Johnson 1995). This was true for the old growth stands at the Huron Mountain Club, and
nearly all categories of dead wood were larger in size in the reserve area than in the
nonreserve area.

Habitat quality, as calculated from HSI models, varied substantially among

206

species and across ownerships. The lowest calculated habitat quality (0.01) occurred for
the pileated woodpecker on MDNR and timber industry sites, and the highest was for the
yellow—rumped warbler at the Huron Mountain Club. The performance of HSI models
also varied among the 7 models tested in this study, and at least 1 model, the veery, could
benefit from revision.

The lack of statistical differences between MDNR and Forest Service sites for
many forest stand and wildlife characteristics may be the result of parallels in public
agency management approaches. Management goals for state forest land include using
commercial timber harvests to enhance timber and wildlife production on a sustained
yield basis (Michigan Department of Natural Resources 1991). Management of Forest
Service lands has sought to meet multiple use objectives, including wilderness resources,
wildlife habitat, recreational opportunities, and economic development, as mandated by
the 1960 Multiple Use Sustained Yield Act. These general descriptions of management
approaches on state and federal forest land are then translated at a local scale based on
social, economic, historical, and ecological factors. State and federal management
agencies are also subject to public scrutiny, and therefore are unlikely to reach extremes
in their management practices. As a result, forest conditions on MDNR and Forest
Service sites represent the likely outcome of management practices that attempt to
represent a broad range of interests.

Landscapes are+ dynamic, responding to changes in land use, environmental
conditions, and social policies over time, and the results of this study may no longer hold

true 5 or 10 years from now. Furthermore, some long term effects of past management

207

approaches may only now be becoming evident in the landscape, and the cumulative
effects of current management programs may not be observed for several more years. For
example, the Forest Service’s management paradigm of focusing on ecosystem
management to meet multiple use objectives was only adopted in 1993 (US. Department
of Agriculture 1995). Therefore, it is likely that the results of this study are a reflection of

management over the past several decades as well as current management approaches.

208

MANAGEMENT AND RESEARCH IMPLICATIONS

While the management goals of the land owners in this study are known in a
general sense, the results of this study help describe the effects of those approaches and
the effects on forest and wildlife resources across the study landscape. This information
may be useful in building partnerships and coordinating regional management goals that
support the individual management objectives of each stakeholder. Ecosystem
management should include a goal of enhancing biodiversity at a large scale, and
managing for complementary regional habitat conditions is one way to achieve that goal,
without necessarily sacrificing specific objectives such as commodity production and
recreation.

As a partial result of current and past management activities, a broad array of
ecological conditions are represented in northern hardwood forests in the Upper
Peninsula. The Huron Mountain Club is a particularly unique component the eastern
Upper Peninsula because of its old growth characteristics. Unfortunately, the Huron
Mountain Club occupies a much smaller proportion of the landscape than state, federal,
or industrial forest land, and it has ecological properties that are not easily replicated
through management. Shifts in management approaches of other ownerships that require
more intensive management, such as increased timber production on public land, will
likely result in a more homogeneous landscape. This may be offset if concurrent efforts
are made to emulate more unique ecological features, such as connectivity of sensitive

habitat types. Such efforts could be part of a larger goal of maintaining stand conditions

209

within historical parameters, and might be accomplished by using silvicultural practices
to emulate some elements of historical stand structure and composition.

Collecting adequate population data on many wildlife species, such as amphibians
and songbirds is often difficult, expensive, and time consuming, and assessing habitat
quality may be more practical for managers than direct population monitoring. Habitat
suitability index models are therefore valuable as one component of a comprehensive
habitat evaluation, to intensively manage for a single species, or to ensure that habitat
requirements are met when a relatively small area is being managed. Additionally,
several variables that are included in HSI models, such as overstory canopy cover, species
composition, and basal area, are attributes that are already measured in the MDNR’s
routine forest inventories, and are likely to be in the databases of other organizations, and
therefore, can be assessed relatively easily. However, some existing models may not fully
describe the habitat requirements for the species surveyed in the Upper Peninsula, and
until they receive more extensive validation, they should only be viewed as working
hypotheses. Habitat suitability index models developed for this dissertation should also
be considered working hypotheses for defining species habitat relationships in Great
Lakes northern hardwood forests. It is vital that they be tested and validated in northern
hardwood forests in other parts of their applicable range, ideally before the models are
used to make habitat management decisions.

Habitat assessment with HSI models may also become labor intensive and
inefficient when a large number of species are being evaluated across a landscape. An

alternative approach may be to employ limited HSI modeling in combination with a

210

coarse filter, as described by Haufler et al. (1999). Wall (1999) has also outlined an
approach to large scale biodiversity conservation on industrial forest land that integrates
fine filter habitat assessments with adaptive management to continually evaluate whether
species and habitats are responding as predicted.

Another recommendation for managers is to continue or initiate monitoring of
red-backed salamander populations in northern hardwood forests. Populations in the
Upper Peninsula are abundant and variable enough that trends can be detected with
standard techniques, and because salamanders are influenced by microhabitat and stand
structure characteristics, management applied to individual forest stands will likely have
the most dramatic impacts on salamander populations. Red-backed salamanders are also
integral to several fundamental ecosystem processes. They mediate forest litter
decomposition through their consumption of invertebrates, and serve as a major food
source for forest snakes, birds, and mammals (Wyman 1998). Because they are relatively
sessile and physiologically sensitive to environmental changes such as soil acidification
(Wyman and Hawksley-Lescault 1987), forest disturbance, and succession (Pough et al.
1987, DeGraaf and Yamasaki 1992, deMaynadier and Hunter 1998), red- backed
salamanders may be ideal indicators of ecosystem integrity (Welsh and Droege 2001).

The results of this project suggest some areas on which to focus additional
research. Several wildlife species surveyed in this project (e.g. veery and pileated
woodpecker) showed distinct associations with land ownerships, raising the question of
their usefulness as indicator species for determining the effects of management on

northern hardwood forest ecosystems. For example, if the response of pileated

211

woodpeckers to forest stand thinning is representative of the response of other cavity
dependent species, managers might be able to make stronger predictions about the effects
of management activities on a forest wildlife community. One way to approach the
feasibility of using indicator species to predict the effect of management on other species
would be to see how well an HSI model for an indicator species predicts the abundance of
additional species in the same habitat. This approach would simplify monitoring efforts,
but would require careful selection of potential management indicator species and
thorough research to build a strong empirical foundation (Niemi et al.1997).

The question of forest and landscape influences on northern flying squirrel
populations is one that has been weakly addressed and needs more intensive study.
Northern flying squirrels are associated with late successional forest characteristics, yet
little is known about their responses to forest management activities. Survey methods for
determining population status, habitat requirements, habitat associations with pileated
woodpeckers, and the current population status in Michigan would all be worthy research
endeavors.

Finally, important information could be gained by analyzing an additional data
layer at the resolution of forest management prescriptions and planning. For example,
forest compartment maps may provide the level of detail needed for performing home
range scale analyses for songbirds and for examining finer scale landscape features.
Combined with satellite imagery, such information would be useful for determining the
spatial scale at which forest management effects may accumulate and begin to impact a

landscape.

212

APPENDICES

213

APPENDIX A. Vegetation and wildlife sampling point coordinates.

Table A1. Global positioning system coordinates (taken from the approximate center
of the stand) for stands sampled on northern hardwood forest study sites in Michigan’s
Upper Peninsula, June-August, 1996, 1997, and 1998. Salamander surveys were
conducted at the shaded coordinates.

 

 

Ownership Replicate Easting (m) Northing (m) Elevation (m)
MDNR 1 539,364 5,146,788 296
539,191 5,145,687 219
539,916 ' 5,144,403 323
541,795 5,144,868 265
542,060 5,143,684 152
540,976 5,142,527 ' 302
542,724 5,142,177 296
542,621 5,141,000 460
543,796 5,141,428 296
2 625,316 5,155,955 311
626,924 5,155,981 390
627,086 5,154,607 390
627,110 5,154,653 375
627,737 5,154,739 222
625,602 5,152,634 390
‘ 627,313 5,152,794 390
627,327 5,151,187 302
3 605,002 5,124,425 152
606,879 5,124,854 152
605,764 5,123,577 180
606,960 5,123,946 317
' 605,070 5,122,181 283
606,830 5,122,173 222
605,528 5,120,588 149
USFS 1 515,375 ‘ 5,135,122 460 '
516,640 5,135,627 390
’ 518,622 ‘ 5,135,235 ‘ 390

214

Table A1 (cont). Global positioning system coordinates (taken from the approximate
center of the stand) for stands sampled on northern hardwood forest study sites in
Michigan’s Upper Peninsula, June-August, 1996, 1997, and 1998. Salamander surveys
were conducted at the shaded coordinates.

 

 

Ownership Replicate Easting (m) Northing (m) Elevation (m)

520,065 5, 1 35,478 390

515,047 5,134,073 460

516,990 5,134,120 207

518,146 5,133,774 329

519,967 5,134,097 390

514,932 5,132,416 168

516,622 5,132,282 219

517,239 5,133,085 253

514,889 5,131,033 207

2 659,305 5,144,154 168
662,477 5,144,05 1 168

657,933 5,142,266 168

659,038 5,142,648 168

660,821 5,142,163 219

662,703 5,142,516 219

655,986 5,140,594 168

657,221 5, 140,997 152

658,827 5,141,100 168

660,776 5,139,193 219

659,207 5,139,344 152

661,001 5,140,877 219

3 502,352 5,144,109 152
503,801 5,143,578 186

505,460 5,143,441 152

502,442 5,142,951 152

503,454 5,141,289 152

503,814 5,142,349 256

505,479 5, 142,292 235

503,742 5, 140,497 332

505,388 5,140,955 302

TI 1 645,588 5,146,037 448

215

Table A1 (cont). Global positioning system coordinates (taken from the approximate
center of the stand) for stands sampled on northern hardwood forest study sites in
Michigan’s Upper Peninsula, June-August, 1996, 1997, and 1998. Salamander surveys
were conducted at the shaded coordinates.

 

 

Ownership Replicate Easting (m) Northing (m) Elevation (m)
647,256 5,147,561 360
647,190 5,146,413 366
648,874 5,146,156 509
650,601 5,146,158 302
651,945 5,144,748 308
647,277 5,149,632 183
2 579,831 5,151,447 238
576,688 5,152,707 363
575,023 5,152,832 256
578,292 5,152,586 256
578,508 5,151,423 393
575,1 19 5,150,976 268
576,755 5,151,027 241
3 596,056 5,150,173 -6
595,927 5,148,501 82
595,941 5,146,730 320
597,559 5,146,863 180
597,270 5,149,958 326
597,557 5,148,423 320
599,007 5,148,397 171
600,624 5,148,583 146
600,570 5,148,345 238
HMC 1 432,989 5,191,152 219
428,880 5,190,550 244
429,609 5,190,054 189
429,637 5,189,459 189
428,956 5,188,813 430
429,815 5,189,165 219
429,575 5,190,509 219
428,669 5,188,114 308
2 429,546 5,192,084 552

216

Table A1 (cont). Global positioning system coordinates (taken from the approximate
center of the stand) for stands sampled on northern hardwood forest study sites in
Michigan’s Upper Peninsula, June-August, 1996, 1997, and 1998. Salamander surveys

were conducted at the shaded coordinates.

 

 

Ownership Replicate Easting (m) Northing (m) Elevation (m)
430,347 5, 1 92,066 393
431,530 5,190,326 265
43 1 ,039 5,192,270 296
432,046 5,191,975 207
432,330 5,191,786 232
432,388 5,191,323 232
432,656 5,190,698 265
3 428,790 5,194,195 375
430,057 5,194,304 189
427,016 5,193,731 238
428,599 5,193,721 378
430,1 12 5,193,699 189
432,083 5,193,760 165

 

217

APPENDIX B. A habitat suitability index model for the red-backed salamander
(Plethodon cinereus) in northern Michigan.
Christine Hanaburgh and Henry Campa, III, Department of Fisheries and Wildlife,
Michigan State University, 13 Natural Resources Building, East Lansing, MI 48824
HABITAT USE INFORMATION
General

Red—backed salamanders are distributed throughout forests in the northeastern
United States and southeastern Canada, ranging from eastern Minnesota in the west,
across to Wisconsin, Ohio and Indiana, and down to northern North Carolina in the
southern part of their range. Vegetation types in which red-backed salamanders live
include oak, aspen, hemlock, pine, northern hardwoods, and conifer swamps (Test and
Bingham 1948, Test and Heatwole 1962, Deer and Rudis 1990). In northern New
England, DeGraaf and Rudis (1990) found that red-backed salamanders were more
abundant in northern hardwoods and red maple (Acer rubrum) stands than in balsam fir
(Abies balsamea) forests, and Test and Heatwole (1962) found that in Michigan, red-
backed salamanders prefer northern hardwood forests over oak-hickory stands.

Within these vegetation types, red-backed salamanders may be found under
stones, in and under logs, and throughout the leaf litter and other organic material on the
forest floor (Heatwole 1962, Burton and Likens 1975a). Although the species is most
easily observed in the litter and beneath cover objects, a substantial proportion of
individuals may occur below ground, at depths up to 30 cm (Taub 1961). In eastern

deciduous forests, the surface density of red-backed salamanders can exceed 4

individuals/m2 (J aeger 1980), and Kleeberger and Werner (1982) identified home range

218

values of 3.04.8 m2 for red-backed salamanders in Michigan forests. There is evidence
that red-backed salamanders are territorial and actively defend the portion of the forest
floor included in their home range (Gergits and Jaeger 1988a). Research has also shown
that they have homing abilities and will return to within several meters of their original
location after being manually displaced by as much as 30 m (Kleeberger and Werner
1982), and that they often return to the most recently used cover object after a foraging
trip (Gergits and J aeger 1988b).

Red-backed salamanders are a major component of the ecosystems in which they
occur, aiding in the decomposition processes and serving as a food source for larger
snakes, birds, and small mammals. They may be the dominant vertebrate in terms of
biomass in a typical northern hardwood New England forest (Burton and Likens 1975b),
and are perhaps the most abundant terrestrial vertebrate in general in the northeastern
U.S. (Wyman and Hawksley-Lescault 1987).

There is little data available in the literature on the impact of structural
characteristics of forest stands on red-backed salamander habitat quality. In evaluating
salamander use of artificial cover boards in even-aged northern hardwood forests in New
England, DeGraaf and Yamasaki ( 1992) found no salamanders under boards in seedling
stands, intermediate numbers in sapling and poletimber stands, and the most red-backed
salamanders in sawtimber or at the edges of sawtimber stands.

Foraging habitat
As much as 75% of the red-backed salamander’s diet (by volume) is made up of

insects such as beetles and flies. Important non-insect foods include earthworms, snails,

219

and slugs (Jameson 1944). Foraging occurs most often at night on the surfaces of the leaf
litter and plants when the ground is wet from rain or dew, or when the relative humidity
of the air is high (Heatwole 1962, Burton and Likens 1975b, Pough et al. 1987).
Salamanders have been observed climbing on leaves and plant stems up to 2.8 m high
(Burton and Likens 1975b).
Breeding habitat

Unlike most other salamanders, the red-backed salamander spends its entire life
cycle on land. Mating occurs during the summer, eggs are laid in early fall beneath a rock
or log, and the young hatch in late summer or fall (Taub 1961). Red—backed salamander
embryos emerge resembling small adults with no intervening larval stage.

Test and Heatwole (1962) found that in habitats where decaying conifer logs
(black spruce [Picea mariana], northern white cedar [Thuja occidentalis], white pine
[Pinus strobus], or hemlock [Tsuga canadensisD were available, the majority of red-
backed salamander egg clutches were found within the conifer logs. On sites where
relatively few decaying logs were present, female red-backed salamanders deposited their
eggs below ground, using small tunnels and burrows that had been made by other
organisms such as earthworms. These researchers hypothesized that in habitats where
suitable logs for nesting are present, red-backed salamanders still utilize burrows and
underground nest sites, and in situations where a suitable, but undetermined, number of
decayed logs are not available, salamanders are dependent on existing burrows in the soil

for egg laying.

220

Microhabitat requirements

Microhabitat variables that have been examined as potential determinants of red-
backed salamander habitat quality include soil moisture (Heatwole 1962, J aeger 1980),
soil pH (W yman and Hawksley-Lescault 1987), soil temperature (T aub 1961), and litter
characteristics (Test and Heatwole 1962, Jaeger 1980, Pough et al. 1987).

In some environments, soil acidity has been found to influence the distribution
and density of red-backed salamanders in an area. In northern hardwood forests in New
York where soil pH ranged from 2.7-5.8, red-backed salamander relative abundance
showed a steep decline as soil pH dropped below 3.9 (W yman and Hawksley—Lescault
1987). Soil pH values in another study where salamanders were observed ranged from
3.6-5.0 (Burton and Likens 1975b), and Vemberg (1955) reported the preferred soil pH
range of red-backed salamanders as 6.2-7.2. Test and Heatwole (1962) considered litter
characteristics other than depth, and observed that leaves in the litter of Michigan oak-
hickory forests are tough and tend to curl, resulting in a looser litter that dries faster and is
therefore less favorable to salamanders than litter in northern hardwood forests.

J aeger (1980) found that the proportion of the surface population of red-backed
salamanders in the litter of the forest floor generally increases and the proportion under
cover objects decreases with increasing amounts of rainfall. However, the total surface
population, compared to the population of red-backed salamanders in the soil, remained
relatively steady during seasonal fluctuations in rainfall and soil moisture. Taub (1961)
found that salamanders respond to changes in available moisture by moving deeper into

the soil under dry conditions and returning to the surface when surface moisture

221

increases. Similarly, Heatwole (1962) found soil moisture to be a factor that influences
the movement of salamanders within their home range, but he concluded that soil
moisture did not affect the observed frequency of salamander activity above ground.
Therefore, soil moisture and rainfall events may be factors that determine the local
distribution of a salamanders within an established home range, but there does not seem
to be corresponding evidence to suggest that soil moisture has a strong effect on overall
habitat quality for red-backed salamanders.

Studies of the importance of litter depth to red-backed salamanders have had
mixed results. Pough et al. (1987), working in New York northern hardwood forests with
different silvicultural treatments, identified leaf litter depth as the strongest indicator of
the surface density of red—backed salamanders. However, Jaeger (1980), working in
mixed deciduous forests in Virginia, and DeGraaf and Yamasaki (1992) in New
Hampshire northern hardwood forests concluded that surface densities of salamanders
were not dependent on litter depths across a wide range of stand ages.

The range of temperatures tolerated by red-backed salamanders with no apparent
negative effects is between 4 and 25 C. At temperatures below 4 C, red-backed
salamanders retreat underground (Taub 1961). Burton and Likens (1975b) identified 10-
15 C as the optimum temperature range for salamanders. Although in Virginia
salamanders are thought to have fully emerged from hibernation and establish territories
by mid-June (Jaeger 1979), Caldwell (1975) observed congregations of active red-backed

salamanders in breeding condition occupying ant mounds during January in Indiana.

222

HABITAT SUITABILITY INDEX (HSI) MODEL
Model development

This HSI model was developed through an analysis of data collected from the
Upper Peninsula of Michigan, during June, July, and August of 1996, 1997, and 1998.
Data on red-backed salamander relative abundance, forest stand characteristics, and soil
attributes were collected in 54 northern hardwood forest stands within 4 different land
ownerships. The number of salamanders observed during ground transect searches and
artificial cover board surveys ranged from 0 to 450 salamanders/ha. Based on the
distribution of salamander relative abundances recorded and the fact that 40% of stands
sampled had <67 salamanders/ha observed, stands in which <67 salamanders/ha were
found were considered to provide relatively poor quality red-backed salamander habitat.
Stands in which 267 salamanders/ha were recorded were identified as sites where red-
backed salamander habitat quality was relatively high. Using the assumption that there is
a positive linear relationship between red-backed salamander relative abundance and
habitat quality, multiple regression analysis and the Kruskal-Wallis one—way analysis of
variance were used to identify the habitat variables that are most important in determining
red-backed salamander habitat suitability.
Model applicability

The data used to develop this model were collected in the central and eastern
Upper Peninsula of Michigan, in the Luce, Mackinac, and Michigamme districts of
Michigan classified by Albert et al. (1986). Therefore, the model will have the strongest

application to Michigan’s Upper Peninsula, but it is also expected to retain applicability

223

in forested portions of nearby Great Lakes states, including Wisconsin, Minnesota, and
Indiana. The ecological units where data were collected are described as having a cool
lacustn'ne climate influenced by the Lake Superior, a mixture of well drained sandy soils
and poorly drained sand and clay soils, with elevations of 178-604 111 (Albert 1986). The
model is based on vegetation measurements and relative abundance data collected during
June-August and is therefore appropriate during that time period. The model was
developed for use in individual forest stands, within which vegetation conditions are
assumed to have minimal variation, and where the salamander population can be
expected to respond uniformly to conditions in the surrounding environment, given their
small home range size. Additionally, this model is only applicable in areas where soil
conditions are not strongly acidic (pH must be 23.7).
Model variables

Based on the analyses described above, the density of overstory trees 210.2 cm in
diameter (Variable 1), the percent canopy cover of shrubs and regenerating trees (50 cm -
5 m tall and <10.2 cm dbh) (Variable 2), and the density (#lha) of cover objects were
chosen as components of the final HSI model. It was found that red—backed salamander
relative abundance was positively related to tree stem densities and negatively related to
several measures of the amount of vegetation between 0.5 and 5.0 m tall.

Variable 1 (Tree stem density)

Stands identified as representing relatively high quality red-backed salamander
habitat had densities of tree stems 210.2 cm dbh ranging from 280-1,093 stems/ha, and a

mean of 599 stems/ha. In stands with relatively low quality habitat, stem densities ranged

224

from 267-720 stems/ha, with a mean of 481. For this model, it was determined that
stands with 0 trees would receive a suitability index (SI) value of 0 for Variable l, and as
the number of trees increased to 1,100 stems/ha, habitat suitability would increase
proportionally, reaching a value of 1.0 in stands with 21,100 stems/ha. The equation for
calculating a suitability index value for Variable l is 811 = (stems/ha)/l, 100 (Fig B 1).

One explanation for the biological importance of tree density to salamanders may
be that a higher density of tree stems might hold moisture in the soil around the tree roots,
and there may be a greater concentration of invertebrates surrounding tree trunks for
salamanders to feed on. The greater abundance of red-backed salamanders in stands with
more mature trees (210.2 cm dbh) may also be related to the amount or quality of litter
that is produced. Although litter depth was not identified as a determining variable of
salamander habitat quality, other characteristics of the litter, such as the volume, degree
of compaction, and moisture holding capacity, may be important for salamander foraging
activities or as hiding cover (Test and Heatwole 1962), and these variables may be
influenced by tree stem densities or a related factor.

Variable 2 (Shrub can0py cover)

In the final habitat model, salamander habitat quality is considered to be highest
when there is no vegetation in the 0.5-5.0 m stratum. Habitat quality then decreases
linearly with increasing vegetation cover in the midstory, reaching a value of 0 for stands
with 100% cover (Fig. B2). The negative relationship between salamander relative
abundance and the amount of shrub canopy cover may be because shrubs and

regenerating trees influence microhabitat characteristics that were not measured in this

225

study, such as nutrient ratios in the soil or daily soil moisture fluctuations. The equation
for calculating a suitability index value for Variable 2 is S12 = 1 — (% rnidstory canopy
cover/ 100).

Variable 3 (Abundance of cover objects)

The final variable included in the model is the abundance of cover objects that are
likely to be used by salamanders. All salamanders observed in this study were found
beneath cover objects, such as sticks >1 cm wide and logs. Although log area and cover
object abundance did not differ significantly (independent t-test, p>0.10) between sites
with relatively high and low habitat quality, woody debris has been noted as an important
element of red-backed salamander habitat (J aeger 1980, DeGraaf and Yamasaki 1992).
Of all the cover objects that we examined for salamanders, a greater percentage (5.5%) of
objects in size classes between 10 and 40 cm had salamanders beneath them than size
classes smaller than 10 cm diameter (1.1%), suggesting that salamanders prefer certain
cover object sizes. Based on our analyses, the abundance of cover objects was assumed
to be adequate in all stands where salamanders were surveyed, but woody debris 10-40
cm wide were considered optimal. The minimum density of cover objects, in the form of
woody debris 10—40 cm wide was 117 pieces/ha, and this value was chosen as the
minimum amount of woody debris that will provide high quality salamander habitat (Fig.
B3). Therefore, stands with 2117 pieces of woody debris/ha, receive an SI of 1.0, and
stands with 0 cover objects receive an S1 of 0.0. In stands with <117 pieces/ha, the SI

value is calculated with the equation $13 = woody debris pieces/ha *0.00855.

226

HSI determination

In the equation for the final HSI value, Variable 3 is used as modifying variable
for Variables 1 and 2 because salamanders are not expected to inhabit stands with no
cover objects, because cover objects allow salamanders to respond to precipitation and
changes in soil moisture Jaeger 1979). The equation for calculating the HSI value is the
geometric mean of the suitability index values for Variable l and Variable 2, multiplied

by 813: HSI = SI3((SIl + s12)l~0.5)

227

 

Figure B 1. Relationship between

Variable 1, the density of trees 210.2
cm dbh, and red-backed salamander ] '
habitat quality. 0.3 4

0.6 -1

Suitability inde:
p
A

 

.‘3
ON
m

 

I j T T— l

0 300 600 900 12001500

Tree stems/ha

 

 

 

 

 

 

Figure B2. Relationship between
Variable 2, the percent canopy cover of 1 1
shrubs and regenerating trees 0.5-5 m m
high, and red-backed salamander E 0.8 7
habitat quality. as 0.6 1

El

:3, 0.4 ~

5, 0.2 4

O I I 1 fi fi
0 20 40 60 80 100
Midstory canopy cover (%)

 

 

228

 

 

 

Figure B3. Relationship between
Variable 3, the density of woody debris 1
10—40 cm wide, and red-backed o I
salamander habitat quality. 2 0-8 ‘
g 0.6 a
g 0.4 ~
:3 0.2 —
0 7 7 I I
0 4O 80 120 160
Woody debris (pieces/ha)

 

 

229

LITERATURE CITED:

Albert, D. A., S. R. Denton, and B. V. Barnes. 1986. Regional landscape ecosystems of
Michigan. University of Michigan, Ann Arbor, Michigan, USA.

Burton, T. M., and G. E. Likens. 1975a. Energy flow and nutrient cycling in salamander
populations in the Hubbard Brook Experimental Forest, New Hampshire.
Ecology 56:1068-1080.

----- . 1975b. Salamander populations and biomass in Hubbard Brook Experimental
Forest, New Hampshire. Copeia 1975:541-546.

Caldwell, R. S. 1975. Observations on the winter activity of the red-backed salamander,
Plethodon cinereus, in Indiana. Herpetologica 31:21-22.

DeGraaf, R. M., and D. D. Rudis. 1990. Herpetofaunal species composition and relative
abundance among three New England forest types. Forest Ecology and
Management 32:155-165.

DeGraaf, R. M., and M. Yamasaki. 1992. A nondestructive technique to monitor the
relative abundance of terrestrial salamanders. Wildlife Society Bulletin 20:260-
264.

Gergits, W. F., and R. G. Jaeger. 1988a. Field observations of the behavior of the red—
backed salamander (Plethodon cinereus): Courtship and agonistic interactions.

Journal of Herpetology 24:93-95.

----- . 1988b. Site attachment by the red-backed salamander, Plethodon cinereus. Journal
of Herpetology 24:91-93.

Heatwole, H. 1962. Environmental factors influencing the local distribution and activity
of the salamander, Plethodon cinereus. Ecology 43:460-472.

Jaeger, R. G. 1979. Seasonal spatial distributions of the terrestrial salamander Plethodon
cinereus. Herpetologica 35:90-93.

----- . 1980. Microhabitats of a terrestrial forest salamander. C0peia 1980:265-
268.

Jameson, E. W., Jr. 1944. Food of the red-backed salamander. Copeia 1944:145-146.

Kleeberger, S. R. and J. K. Werner. 1982. Home range and homing behavior of
Plethodon cinereus in northern Michigan. Copeia 1982:409-415.

230

Pough, F. H., E. M. Smith, D. H. Rhodes, and A. Collazo. 1987. The abundance of
salamanders in forest stands with different histories of disturbance. Forest
Ecology and Management 20: 1-9.

Taub, F. B. 1961. The distribution of the red-backed salamander, Plethodon c. cinereus,
within the soil. Ecology 42:681—698.

Test, F. H., and B. A. Bingham. 1948. Census of a population of the red-backed
salamander (Plethodon cinereus). American Midland Naturalist 39:362-372.

----- , and H. Heatwole. 1962. Nesting sites of the red-backed salamander, Plethodon
cinereus, in Michigan. Copeia 1962:206-207.

Vemberg, F. G. 1955. Correlation of physiology and behavior indexes of activity in the
study of Plethodon cinereus (Green) and Plethodon glutinosus (Green). American
Midland Naturalist 54:382-393.

Wyman, R. L., and D. S. Hawksley-Iescault. 1987. Soil acidity affects distribution,

behavior, and physiology of the salamander, Plethodon cinereus. Ecology
68:1819-1827.

231

APPENDD( C. A model of habitat suitability for the yellow-rumped warbler (Dendroica
coronata) in the Upper Great Lakes region of the United States.

Christine Hanaburgh and Henry Campa, 111., Department of Fisheries and Wildlife,
Michigan State University, 13 Natural Resources Building, East Lansing, MI 48824

HABITAT USE INFORMATION
General

The geographic range of the yellow-rumped warbler during the breeding season
includes most of the northern United States (including Alaska), the Canadian provinces,
and the western United States (Eastman 1991, Grondahl 1999). Yellow-rumped warblers
are migratory, leaving the Midwest in October and returning during April and May
(Eastman 1991). They also winter farther north than any other warbler species, and may
remain in the southern parts of their summer range year round (Grondahl 1999).

Yellow-rumped warblers are typically associated with mature coniferous forests
(DeGraaf et al. 1991, Eastman 1991, Schulte and Niemi 1998), although they are also
frequently observed near the edges of forest openings, pine plantations, boreal bogs, and
mixed forests (DeGraaf et a1. 1991, Eastman 1991). Howe et a1. (1995) identified
lowland conifers, red pine (Pinus resinosa), and jack pine (Pinus banksiana) as forest
types preferred by yellow-nimped warblers, and very late successional and very early
successional hardwood forests as avoided forest types. Schulte and Niemi (1998) also
reported a very low abundance of yellow-rumped warblers in regenerating clearcuts and
burned forests, compared to other forest bird species observed.

In a study of industrial forests in Maine (Hagan et al. 1997), yellow-rumped

warblers were most abundant in medium aged (20-60 years) and mature (>60 years)

232

softwood forests dominated by balsam fir (Abies balsamea), white pine (Pinus strobus),
and spruces (Picea spp.) with “full or medium crown closure”, and they were less
abundant in clearcut or regenerating stands and in predominantly hardwood forests.
Thus, yellow-rumped warblers appear able to tolerate a fairly broad range of forest
conditions.
Foraging habitat

Yellow-rumped warblers are insectivorous during the breeding season, but they
can also survive on berries (their winter food) for short periods during the summer
(Grondahl 1999). Insects are obtained primarily by hawking, as well as by gleaning and
hovering (Franzreb 1983, DeGraaf et al. 1991). Franzreb (1983) documented that 80% of
male yellow-rumped warblers and 65% of females foraged from trees that were >18 m
tall. This study took place in a high-elevation mixed conifer forest in Arizona, and the
tree heights may not be directly comparable to those in Great Lakes forest types.
However, the data suggest that yellow-rumped warblers use relatively tall trees for
foraging, and therefore may choose habitats where taller trees are present.
Nesting habitat

The yellow-rumped warbler nests almost exclusively in conifers (DeGraaf et al.
1991, Grondahl 1999), and therefore requires an adequate number of conifer trees from
which to select a nest location. Materials used to build the bulky, open nest include
twigs, bark, feathers, and grass, and the nest is typically placed far out on a horizontal
branch (Eastman 1991, Grondahl 1999). Nests are generally located 3-6 m above the

ground, but nest heights may range from 1-15 m (Grondahl 1999).

233

An analysis of Hanaburgh’s (this volume, Chapter 1) data sets for yellow-rumped
warbler relative abundance and forest stand characteristics suggested that understory stem
density may be an important variable in yellow—rumped warbler habitat. In these data
sets, stands where yellow-rumped warblers were observed had a significantly lower
density (p=0.04) of shrubs and tree seedlings <5 m tall than stands where the warblers
were not encountered. In the one outside study consulted for this model where understory
stem density was evaluated (Schulte and Niemi 1998), the authors found that yellow-
rumped warblers in recently burned stands or in clearcut stands with residual trees were
associated more strongly with sites that had lower shrub densities. One way in which
stem densities may influence habitat quality is if a lack of understory allows more open
space for the birds to catch insects in flight, resulting in more favorable foraging

conditions.

HABITAT SUITABILITY INDEX (HSI) MODEL
Model development

In developing this model, a set of potential model variables was first identified
based on quantitative and qualitative reports of important habitat components and
characteristics in the published literature. This information was combined with empirical
data sets of bird species relative abundance and forest stand measurements collected in
northern hardwood forests in the Upper Peninsula of Michigan during 1996-1998. In
these empirical data sets, independent paired t-tests were used to compare stands where

yellow-rumped warblers were observed during data collection with stands where yellow-

234

rumped warblers were not observed. These results were used to associate quantitative
values with an index of habitat quality for each variable identified from the literature.
The empirical data were also used to identify shrub and seedling density as an additional
model variable which had significantly different (psOJO) values between stands where
yellow-rumped warblers were observed and stands where they were not observed,
although no published references to the importance of this variable were found.
Model applicability

This model is applicable for the Upper Great Lakes region, which includes the
northern parts of Wisconsin and Minnesota, the Upper Peninsula and northern Lower
Peninsula of Michigan, and southern Ontario. Although yellow-rumped warblers are
most commonly associated with coniferous forests (i.e., lowland conifers, red pine, and
jack pine), other forest types, such as northern hardwoods may also provide habitat for
yellow-rumped warblers if they contain a minor conifer component. Therefore, within
the specified geographic area, the model may be used in northern hardwoods forest,
mixed hardwoods/conifer forest, and boreal forest types. The model is applicable during
the breeding season only, from April through early October.
Model variables

The most important aspect of yellow-rumped warbler habitat quality is the
availability of mature coniferous trees to nest in, so the amount of overstory conifer cover
was chosen as a model variable. The average height of mature trees was identified as a
significant attribute of foraging habitat. Additional variables that determine yellow-

rumped warbler habitat suitability are the amount of overstory canopy cover in a stand

235

and the density of shrubs and seedlings in the understory.

Variable I (Overstory conifer cover)

In the data set used to create this model, overstory (>5 m) conifer cover values
averaged 31% (SE. = 7.8%) in stands where yellow-rumped warblers were recorded and
16% (SE. = 2.1%.) in stands where they were not observed. Based on this comparison,
30% was chosen to represent the minimum amount of conifer cover that will contribute to
relatively high quality habitat for yellow-rumped warblers (Fig. C1). Stands that meet or
exceed this minimum requirement are assigned an S1 value of 1.0, and SI values decrease
proportionally as the average amount of conifer cover decreases towards 0. For stands
with <30% conifer cover in the overstory, the SI value is calculated with the equation
S1] = [0.033*(% conifer cover)], where 0.033 is the slope of the portion of the graph (Fig.
C1) with conifer cover values between 0 and 30.

Variable 2 (Height of overstory trees)

In the empirical data set used to build this model, yellow-rumped warblers were
not found in stands where the average height of mature trees (defined as trees 25 m tall
and 210.2 cm dbh) was <20 m, although average tree heights ranged from 14-29 m
among all the stands sampled. Average tree height in stands used by yellow-rumped
warblers and stands not used by yellow-rumped warblers was not significantly different
(p=0.705), yet this variable has been indicated in the literature as an important habitat
characteristic for yellow-rumped warblers. Assuming that yellow-rumped warblers prefer
to forage in the taller trees in a stand, 20 m was used as the minimum average tree height

associated with high quality yellow-rumped warbler habitat. Stands with an average tree

236

height 220 m receive an SI value of l, and SI values in stands with an average tree height
<20 m are calculated with the equation SI2 = [0.05*(average tree height)], derived from
the linear equation for the slope of the portion of the graph with tree height values
between 0 and 20 m (Fig. C2).

Variable 3 (Overstory canopy cover)

Although published information indicates that yellow-rumped warblers tolerate a
broad range of overstory canopy cover conditions, they prefer forest stands with at least
moderate, if not complete canopy cover (Hagan et al. 1997). Overstory canopy cover was
not statistically different between stands where yellow-rumped warblers were observed
and stands where they were not observed, perhaps due to the small sample size. Yellow-
rumped warblers in Hanaburgh’s (this volume) Upper Peninsula study were found in
stands with at least 73% overstory cover. Rounding this value to 70%, stands with
overstory canopy cover values between 70% and 100% are considered necessary for
Optimal yellow-rumped habitat quality. An SI value of 1.0 is assigned to stands with 70-
100% cover, and for stands with <70% overstory canopy cover, the SI value is calculated
with the equation S13 = [0.014*(% canopy cover)] (Fig. C3).

Variable 4 (Density of shrubs and seedlings)

In the data set used to build this model, the mean understory stem density in
stands with yellow-rumped warblers was only 2,111 stems/ha (SE. = 566 stems/ha),
compared with 11,067 stems/ha (SE. = 1,236 stems/ha) in stands where yellow—rumped
warblers were not recorded. Therefore, stands with 5 2,000 shrub or seedling stems/ha

are assigned an S1 value of 1.0. and stands with 2 11,000 understory stems/ha are

237

assigned an 81 value of 0.0 (Fig. C4). Suitability index values for stands with 2,000-
11,000 understory stems/ha can be calculated with the equation
814 = [—0.000112*(understory stem density)+(1.236)]. In this equation, 1.236 is the y-
intercept and -0.000112 is the slope of the portion of the graph with stem density values
between 2,000 and 11,000 (Fig. C4).
HSI determination

The most important element of the yellow-rumped warbler’s habitat is the
availability of mature conifer trees to nest in. This attribute, represented by Variable 1 in
the model, is likely to have the strongest and most direct impact on a yellow-rumped
warbler’s choice of habitat. The other variables identified in this model (overstory
canopy cover, tree height, and shrub and seedling density) also contribute to habitat
quality; however, because the yellow-rumped warbler tolerates a somewhat broader range
of other forest conditions, these variables are less important in determining overall habitat
quality for an area. The final HSI equation represents this relationship by using the
conifer cover SI value to modify the average of the other 3 SI values:

HSI = SIl*[(SIZ+SI3+SI4)/3]

238

Figure C 1. Relationship between
Variable l, the percent conifer cover in
the overstory, and yellow-rumped warbler
habitat quality.

Figure C2. Relationship between
Variable 2, the average height of
overstory trees (25 m tall, 210.2 cm dbh)
and yellow-rumped warbler habitat
quality.

 

 

 

0.8 4
0.6 4

9
J2
L

Suitability index

p
N
P

 

 

O

I j— l I— *1

O 20 40 6O 80 100

% Overstory conifer cover

 

 

 

1 1
0.8 1
0.6 -

0.4 a

Suitability index

0.2 4

 

 

O Irrrrrfi—
0 4 81216202428

Average overs tory tree height (m)

 

 

A

 

239

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure C3. Relationship between V
Variable 3, the total percent overstory
canopy cover, and yellow-rumped warbler I _
habitat quality. 5 0.8 —
E
E 0.4 ~
:2 0.2 ~
0 r 1 l I fl
0 20 40 6O 80 100
L % Overstory canopy cover
Figure C4. Relationship between
Variable 4, the density of shrubs and 1 ~
saplings <5 m tall, and yellow-rumped x 0 8 _
warbler habitat quality. 8 '
.S -
E“ 0.6
3 0.4 ~
5
V3 0.2 5
0 , T I
0 4000 8000 12000
Shrub/sapling density (stems/ha)

 

 

 

 

 

 

 

 

 

 

240

LITERATURE CITED:

DeGraaf, R. M., V. E. Scott, R. H. Hamre, L. Ernst, and S. H. Anderson. 1991. Forest
and rangeland birds of the United States natural history and habitat use. US.
Department of Agriculture, Forest Service, Agriculture Handbook 688. Northern
Prairie Wildlife Research Center Home Page. [Online] Available
http://www.npwrc.usgs.gov/resource/1998/forest/forest.htm (Version 03NOV98).

Eastman, J. 1991. Yellow-rumped warbler. Pages 404-405 in R. Brewer, G. A. McPeek,
and R. J. Adams, Jr. The atlas of breeding birds of Michigan. Michigan State
University Press, East Lansing, Michigan, USA.

Franzreb, K. E. 1983. Intersexual habitat partitioning in yellow-rumped warblers during
the breeding season. Wilson Bulletin 95:581-590.

Grondahl, C. 1999. A closer look: The yellow-rumped warbler. North Dakota Outdoors
61:15.

Hagan, J. M., P. S. McKinley, A. L. Meehan, and S. L. Grove. 1997. Diversity and
abundance of land birds in a northeastern industrial forest. Journal of Wildlife

Management 61 :718-735.

Howe, R. W., G. Niemi, and J. R. Probst. 1995. Management of western Great Lakes
forests for the conservation of neotropical migratory birds. Pages 144-167 in F.
R. Thompson, editor. Management of midwestem landscapes for the
conservation of neotropical migratory birds. U. S. Forest Service General

Technical Report NC-187.

Schulte, L. A., and G. J. N iemi. 1998. Bird communities of early-successional burned
and logged forest. Journal of Wildlife Management 62: 1418-1429.

241

APPENDD( D. A habitat suitability index model for the northern flying squirrel
(Glaucomys sabrinus) in the Upper Great Lakes region of the United States.

Christine Hanaburgh and Henry Campa, HI, Department of Fisheries and Wildlife,
Michigan State University, 13 Natural Resources Building, East Lansing, MI 48824

HABITAT USE INFORMATION
General

Northern flying squirrels are found in coniferous and hardwood forests across
most of the northern half of North America, from the Atlantic coast to the Pacific coast.
Their range extends south into the Appalachians of North Carolina, Virginia, and West
Virginia where the federally endangered subspecies Glaucomys sabrinusfuscus exists in
scattered populations at high elevations (Payne et al. 1989). In the Upper Great Lakes
region, northern flying squirrels are most commonly associated with boreal, mixed
hardwood/conifer, and northern hardwood forests (Weigl 1978, Wells-Gosling 1982), and
they are found in both relatively fragmented and unfragmented landscapes (Bayne and
Hobson 1998). In Michigan, the species has been reported in conifer swamps (Green
1925), mixed hardwoods and upland conifers (Laundre 1975), and jack pine (Pinus
banksiana) stands (Manville 1948), and a mature forest of mixed conifers and hardwoods
forest was identified as favorable habitat in southern Quebec (Wrigley 1969).

Home ranges of the northern flying squirrel have been estimated at 3.1-12.5 ha in
Pennsylvania and North Carolina (Weigl and Osgood 1974) and 3.4-4.9 ha in western
Oregon (Witt 1992). In New Brunswick, Canada, Gerrow et al. (1998) found that the
median home range of female northern flying squirrels was 2.8 ha. The median home

range for males was 12.5 ha, and male home ranges often overlapped with several female

242

home ranges.
Foraging habitat

Because northern flying squirrels do not hibernate, they must have access to a
food supply throughout the winter. In the moist forests of the Pacific Northwest, lichens
and fungi are the staples of the northern flying squirrel diet (Maser et al. 1985).
However, unlike northern flying squirrels in western North America which are dependent
on lichens and fungi as a food source, northern flying squirrels in the Upper Great Lakes
seem to have a more flexible diet. Northern flying squirrels in the Great Lakes area cache
conifer seeds, beechnuts, hazelnuts, and acorns for the winter, and in the spring and
summer they also consume insects, soft mast, and bird eggs and nestlings (Baker 1983,
Vander Haegen and DeGraaf 1996, Bayne and Hobson 1997). Baker (1983) suggested
that in Michigan, a stored supply of the autumn seed crop is one of the most important
factors in a northern flying squirrel’s survival. Northern flying squirrels generally meet
their water requirements through the food they eat and they are not thought to have any
habitat requirements related to water (W ells-Gosling and Heaney 1984).

Much of the southern part of the northern flying squirrel’s geographic range
coincides with the northern limits of the southern flying squirrel’s (Glaucomys volans)
geographic distribution, and it has been hypothesized that in regions where the 2 species
are sympatric, the northern flying squirrel’s habitat use is influenced by competitive
interactions with the southern flying squirrel (W eigl 1978). In a laboratory setting
consisting of a divided outdoor cage planted with sweetgum (Liquidambar styraciflua) on

one side and red spruce (Picea rubens) on the other, Weigl (1978) observed that northern

243

flying squirrels paired with other northern flying squirrels did not show a distinct
preference for either hardwood or conifer habitats. However, when paired with southern
flying squirrels, the northern flying squirrels spent nearly 100% of their time in the
conifer habitat, while the southern flying squirrels spent nearly 100% of their time in the
hardwood habitat. Thus, although pure hardwood forests may meet the northern flying
squirrel’s habitat requirements, they are better able to compete for their habitat
requirements in habitats that have a conifer component. One factor that may give
northern flying squirrels a competitive advantage in conifer habitats is their unique ability
to subsist on lichens and fungi in winter, which are more abundant in forests with a
coniferous component than in purely deciduous forests (Weigl 1978).

An additional aspect of northern flying squirrel foraging requirements may be the
density of overstory trees. Gerrow et al. (1998) reported an overstory tree density of 933
trees/ha 24 m tall in areas used by northern flying squirrels and a significantly greater tree
density of 1,167 trees/ha in areas outside of the squirrels’ home ranges. One reason for
the difference may be that the wider spacing of fewer trees may allow the squirrels to
glide farther and more efficiently in search of food than they could in a more densely
treed forest.

Nesting habitat

Adequate nesting sites are considered a critical component of the northern flying
squirrel’s habitat (Weigl 1978). Northern flying squirrels nest either in tree cavities or in
exterior nests built in tree branches, tree crotches, or on the ground (Cowan 1936, Weigl

1978, Gerrow et a1. 1998). An individual flying squirrel may use 1-4 nests in its home

244

range, and several flying squirrels may share a nest temporarily (W eigl and Osgood
1974). Carey et a1. (1997) observed that cavities were used year round, and cavity use by
females increased in late spring and summer, corresponding to the period when young are
born and raised. The authors suggested that cavities provide better shelter from predators
and weather than outside nests, and may be particularly important in harsh climates.
Cowan (1936) emphasized the importance of cavities as winter nest sites, but observed
more outside nests used by flying squirrels (males and females) than cavities in the
summer. In New Brunswick, outside nests in trees were used during all seasons except
winter, cavity nests were used year round, and ground nests were used only in winter,
especially during periods of heavy snow cover (Gerrow et al. 1998). These accounts
indicate that tree cavities may be preferred as nest sites under more critical conditions,
such as while giving birth by females and during inclement weather, while outside nests
may be sufficient in less stressful situations.

In New Brunswick, the majority of outside nests were located in live conifers,
usually red spruce or balsam fir (Abies balsamea). The average diameter of trees in
which outside nests were built was 29 cm and nest trees were an average of 14.5 m tall.
Cavity nests in the New Brunswick study were found in trees with an average diameter of
34.4 cm in live trees and 29.7 cm in snags. Average tree heights were 7.4-9.8 m, and
most of the cavities used were in dead trees. Gerrow et al. (1998) also reported that the
average diameter of trees within flying squirrel home ranges was larger than that of
random trees measured outside the home ranges. The average diameter of trees with

cavity nests reported by Witt (1992) in Oregon was 63.5 cm, and the average height of

245

cavity nest trees was 19.6 m. Cavity nests reported by Weigl and Osgood (1974) in
Pennsylvania were located in hardwood trees with an average diameter of 57 cm. The
large variation in the size of nest trees in these studies is most likely due to the wide
geographic separation and the ecological attributes of the forest types and dominant tree
species at each study site. Despite these differences, common factors in these studies are
the northern flying squirrel’s slight preference for conifer trees and the requirement of a
tree structure that can support a relatively large cavity.

In the Upper Great Lakes region, pileated woodpeckers (Dryocopus pileatus)
create more large sized tree cavities than any other forest animal, and the cavities they
create are used by many forest animals, including northern flying squirrels, long after the
woodpeckers are done with them (Bonar 2000). Therefore, a logical assumption is that
habitats where pileated woodpeckers have chosen to excavate tree cavities should have
characteristics that overlap with the nesting habitat requirements of northern flying
squirrels. A pileated woodpecker habitat suitability index model developed for the Upper
Great Lakes region (Felix et al. 1999) evaluates pileated woodpecker nesting habitat
based on the density of large overstory trees and snags and the average diameter of
mature trees and snags. This model determined that optimal pileated woodpecker nesting
conditions are present in forest stands with at least 95 trees/ha 230 cm in diameter, an
average diameter of at least 55 cm for all trees and snags at least 30 dbh, and a minimum

of 0.6 snags/ha 255 cm dbh (Felix et al. 1999).

246

HABITAT SUITABILITY INDEX (HSI) MODEL
Model development

The majority of published literature on northern flying squirrels is based on
research conducted in the northwestern region of the United States, and a habitat
suitability index model has been developed for northern flying squirrels in the Pacific
Northwest and Intermountain West (Felix and Campa 1999). Some information from
populations in the Appalachian Mountains has also been published, but much less data
exists for northern flying squirrels in the eastern and midwestem portions of their range.
In the absence of relevant information on northern flying squirrel habitat requirements in
the Upper Great Lakes region, information from other geographic areas was consulted to
develop this model.
Model applicability

The geographic region in which this model is applicable is the Upper Great Lakes
region, which includes northern parts of Wisconsin and Minnesota, the Upper Peninsula
and northern Lower Peninsula of Michigan, and southern Ontario. This region is roughly
equivalent to the area of the Laurentian mixed forest described by Bailey and Cushwa
(1981). Within this geographic area, the model may be used in northern hardwood forest,
mixed hardwood/conifer forest, and boreal forest types. Northern flying squirrels are not
migratory and do not hibernate, so the model has been developed to determine the
minimum habitat quality provided in an area throughout the year.
Model variables

The key variable identified as a determinant of northern flying squirrel foraging

247

habitat quality is the density of overstory trees. An additional variable that influences
foraging habitat quality is the proportion of coniferous canopy cover in the forest
overstory. Variables that describe nesting habitat requirements are the densities of large
(230 cm dbh) trees and snags.

Variable 1 (density of overstory trees)

Northern flying squirrel habitat is thought to be best when the density of overstory
trees (210.2 cm dbh) is 933 stems/ha, and it is likely that a range of stem densities values
surrounding 933 stems/ha also provide high quality habitat. In the absence of empirical
data that define the true range of suitable habitat conditions, stem densities within 10%
higher or lower than 933 stems/ha (840-l,026 stems/ha) are assigned a SI value of 1.0.
Habitat suitability decreases as the tree density increases to 1,167 stems/ha, beyond which
habitat is considered unsuitable and is assigned an S1 value of 0 (Fig. D1). Habitat
suitability also decreases linearly from the lower end of the range of optimal stem density
values (840 stems/ha) towards 0 stems/ha. The suitability index value in stands with less
than 840 stems/ha can be calculated with the equation S11 = [(#trees/ha)/840]. Suitability
index values in stands with l,026-1,137 stems/ha can be described by the equation S11
=[-0.009*(trees/ha) + 10.51], and stands with >1,137 trees/ha receive a SI value of 0.

Variable 2 (Density of trees 230 cm dbh)

Published measurements on the average size of trees in which northern flying
squirrels use cavity nests range from 27 cm to 64 cm in various habitats. Based on the
Northern Great Lakes pileated woodpecker HSI model (Felix et al. 1999), tree densities

of 295 stems/ha (for trees 230 cm dbh) are assumed to maximize the possibility that an

248

area will be inhabited by pileated woodpeckers that will create cavities for flying squirrels
to eventually nest in. Stands that meet this minimum requirement will be assigned an SI
value of 1.0 (Fig. D2). For stands that have <95 trees (230 cm dbh) per ha, the habitat
quality relationship is expressed by the equation, S12 = [0.010526*(# trees/ha 230 cm
dbh)].

Variable 3 (Snag density)

In the absence of pileated woodpecker activity, flying squirrels will have to rely
on natural cavities, primarily in snags, to meet their cavity nesting requirements. The
average diameter of snags containing natural cavities used by northern flying squirrels in
New Brunswick was 30 cm. Using a minimum home range estimate of 2.8 ha per female
and a requirement of up to 4 nests per individual (Gerrow et al. 1998), a minimum of 1.4
cavities/ha would be required. To be considered suitable for northern flying squirrels, a
habitat with 1.4 snags/ha 230 cm dbh would necessitate that each snag have an
unoccupied suitable cavity and that northern flying squirrels be able to locate every
suitable cavity. At the same time, a northern flying squirrel probably does not require that
all 4 of the nests be a cavity nest. It is therefore assumed in this model that forests with at
least 1.4 snags/ha 230 cm dbh will generally provide suitable flying squirrel habitat, and
habitat quality will decline as the snag density decreases towards 0 (Fig. D3). For stands
in which there are <1.4 snags/ha 230 cm dbh, the relationship is expressed by the
equation, S13 = [0.7*(#snagslha 230 cm dbh)].

Variable 4 (Percent coniferous canopy cover)

Some degree of conifer cover has been identified as an element of preferred

249

habitat for northern flying squirrels, although the amount of conifer cover that defines
optimal habitat has not been quantified. Northern flying squirrel nests have been located
more often, but not exclusively, in conifer trees, and some research has indicated that the
structural diversity present in a mixed conifer hardwood stand is important to northern
flying squirrels (Gerrow et al. 1998). A conservative interpretation of these data is that a
nominal percentage of at least 10% conifer cover and a maximum of 90% coniferous
cover will provide suitable northern flying squirrel habitat. Therefore, Variable 4 is
assigned a suitability index value of 1 if at least 10% and <90% conifer cover is present in
the overstory of a forest stand (Fig. D4). Suitability index values in forest stands with
conifer cover values outside of this range decrease linearly towards 0. The suitability
index value equation for stands with <10% conifer cover is S14 =[0.10*(% conifer
cover)], and the equation in stands with >90% conifer cover is 814 = [~0.1*(% conifer
cover)+10].
HSI determination

Each of the variables described in this HSI model contributes to the overall
determination of habitat quality for northern flying squirrels in the area being evaluated.
An individual SI value of 0 does not mean that the habitat is entirely unsuitable for
northern flying squirrels, because moderate quality habitat may be present if some of the
flying squirrel’s other habitat requirements are met. Similarly, a value of 1.0 for a single
variable is not enough to assign the whole habitat optimal status if conditions described
by the other variables are less than optimal. Instead, the final HSI value is calculated by

combining the suitability index values for all 4 variables for a given area, giving each

250

value equal weight in determining overall habitat suitability of an area. The equation for
calculating the HSI value is the mean of the suitability index values for each of the 4

variables: HSI = (SIl+SIZ+SI3+SI4)/4

251

Figure D1. Relationship between
Variable 1, the number of overstory
trees/ha 210.2 cm dbh, and northern
flying squirrel habitat quality.

Figure D2. Relationship between
Variable 2, the number of overstory
trees/ha 230 cm dbh, and northern flying
squirrel habitat quality.

 

1‘1

 

 

 

 

 

 

5 0.8
E
g, 0.61
E 0.4 -
53 0.2 -
O 1 —I I f
0 300 600 900 1200
# Overstory trees/ha >10.2 cm dbh
1
'1
33 0.8
1:
.E 06
.«E‘ '
E 0.4
a”: 0.2
O r T I T r I l
0 20 40 60 80 100 120
# Overstory trees/ha > 30 cm dbh

 

 

 

252

Figure D3. Relationship between
Variable 3, the number of snags/ha 230
cm dbh, and northern flying squirrel
habitat quality.

Figure D4. Relationship between
Variable 4, the percent conifer cover in
the overstory, and northern flying squirrel
habitat quality.

 

 

 

 

 

l
0.8
0.6

lity index

n

l

0.4

Suitab

0.2

 

0fi4

0 0.3 0.6 0.9 1.2
# Snags/ha >30 cm dbh

I I 1

I

1.5

fl

1.8

 

1.00
0.80
0.60
0.40
0.20

Suitability index

 

 

 

0.00 w

0

_—T

20

T I

40 60
% Conifer cover

I

80

Y

100

 

253

 

LITERATURE CITED:

Bailey, R. G., and C. T. Cushwa. 1981. Ecoregions of North America. U. S. Geological
Survey. Washington, DC, USA.

Baker, R. H. 1983. Michigan mammals. Michigan State University Press, East Lansing,
Michigan, USA.

Bayne, E. M., and K. A. Hobson. 1997. Temporal patterns of predation on artificial
nests in the southern boreal forest. Journal of Wildlife Management 61 : 1227-
1234.

Bayne, E. M., and K. A. Hobson. 1998. The effects of habitat fragmentation by forestry
and agriculture in the abundance of small mammals in the southern boreal
mixedwood forest. Canadian Journal of Zoology 76:62-69.

Bonar, R. L. 2000. Availability of pileated woodpecker nest cavities and use by other
species. Journal of Wildlife Management 64:52-59.

Carey, A. B., T. M. Wilson, C. C. Maguire, and B. L. Biswell. 1997. Dens of northern
flying squirrels in the Pacific Northwest. Journal of Wildlife Management
61:684-699.

Cowan, I. M. 1936. Nesting habits of the flying squirrel Glaucomys sabrinus. Journal of
Mammalogy 17:58-60.

Felix, A. B., and H. Campa, III. 1999. Habitat suitability model for the northern flying
squirrel in the Pacific Northwest and Intermountain regions of the United States.
On file with Boise Cascade, Inc. Boise, Idaho, USA.

Felix, A. B., H. Campa, 111., and G. J. Roloff. 1999. A pileated woodpecker habitat
suitability index model for the northern Great Lakes region. On file with Boise
Cascade, Inc. Boise, Idaho, USA.

Gerrow, S., S. Flemming, and T. Herman. 1998. Habitat characteristics and home ranges
of northern flying squirrels on Fundy National Park and the Greater Fundy
Ecosystem. [Online] Available
httpzllwww.unb.ca/web/forestry/centers/cwru/soe/squirrel.htm.

Green, M. M. 1925. Notes on some mammals of Montmorency County, Michigan.
Journal of Mammalogy 6: 173-178.

Laundre, J. 1975. An ecological survey of the mammals of the Huron Mountain Area.

254

Huron Mountain Wildlife Foundation. Occasional Papers of the Huron Mountain
Wildlife Foundation No. 2.

Manville, R. H. 1948. The vertebrate fauna of the Huron Mountains. American Midland
Naturalist 39:615-640.

Maser, 2., C. Maser, and J. M. Trappe. 1985. Food habits of the northern flying squirrel
(Glaucomys sabrinus) in Oregon. Canadian Journal of Zoolology 63: 1084-1088.

Payne, J. L., D. R. Young, and J. F. Pagels. 1989. Plant community characteristics
associated with the endangered northern flying squirrel, Glaucomys sabrinus, in
the southern Appalachians. American Midland Naturalist 121:285-292.

Vander Haegen, W. M., and R. M. DeGraaf. 1996. Predation on artificial nests in
forested riparian buffer strips. Journal of Wildlife Management 60:542-550.

Wei g1, P. D. 1978. Resource overlap, interspecific interactions, and the distribution of
the flying squirrels, Glaucomys volans and G. sabrinus. American Midland
Naturalist 100283-96.

Weigl, P. D., and D. W. Osgood. 1974. Study of the northern flying squirrel,

Glaucomys sabrinus, by temperature telemetry. American Midland Naturalist
92:482-486.

Wells-Gosling, N. 1982. Distribution of flying squirrels (Glaucomys) in Michigan.
Michigan Academician 142209-216.

Wells-Gosling, N., and L. R. Heaney. 1984. Glaucomys sabrinus. Mammalian Species
229:1-8.

Witt, Joseph W. 1992. Home range and density estimates for the northern flying
squirrel, Glaucomys sabrinus, in western Oregon. Journal of Mammalogy

73:921-929.

Wrigley, R. E. 1969. Ecological notes on the mammals of southern Quebec. Canadian
Field-Naturalist 83:201-21 1 .

255

APPENDIX E. Scientific names of songbirds surveyed.

Table E1. Common and scientific names of bird species recorded on northern hardwood
forest study sites in Michigan’s Upper Peninsula, June-August, 1996, 1997, and 1998.

 

Common name
American crow
American redstart
American robin

Black and white warbler
Black-billed cuckoo
Blackburnian warbler
Black-capped chickadee
Black-throated blue warbler
Black-throated green warbler
Blue jay

Brown creeper
Brown-headed cowbird
Cerulean warbler
Chestnut-sided warbler
Common flicker
Common raven

Downy woodpecker
Eastern wood pewee
Golden-crowned kinglet
Great crested flycatcher
Hairy woodpecker
Hawk spp.

Hermit thrush

Least flycatcher
Magnolia warbler
Mourning warbler
Nashville warbler
Northern parula
Ovenbird

Pileated woodpecker
Pine siskin

Scientific name

Corvus brachyrhynchos
Setophaga ruticilla
Turdus migratorius
Mniotilta varia
Coccyzus erythropthalmus
Dendroica fusca

Parus atricapillus
Dendroica caerulescens
Dendroica virens
Cyanocitta cristata
Certhia familiaris
Molothrus ater
Dendroica cerulea
Dendroica pensylvanica
Colaptes auratus
Corvus corax

Picoides pubescens
Contopus virens
Regulus satrapa
Myiarchus crinitis
Picoides villosus

Buteo spp.

C atharus guttatus
Empidonax minimus
Melanerpes erythrocephalus
Oporomis philadelphia
Vermivora ruficapilla
Parula americana
Seiurus aurocapillus
Drycopus pileatus
Carduelis pinus

256

Table El (cont).

 

Pine warbler
Red-breasted nuthatch
Red-eyed vireo
Red-headed woodpecker
Rose-breasted grosbeak
Ruby-throated hummingbird
Sandhill crane

Scarlet tanager

Solitary vireo

Swainson's thrush
Tennessee warbler

Veery

White-breasted nuthatch
White-throated sparrow
Winter wren

Wood thrush
Yellow-bellied flycatcher
Yellow-bellied sapsucker
Yellow-rumped warbler

Yellow warbler

Dendroica pinus

S itta canadensis

Vireo olivaceus
Melanerpes erythrocephalus
Pheucticus ludovicianus
Archilochus colubris
Grus canadensis
Piranga olivacea

Vireo solitarius
Catharus ustulatus
Vermivora peregrina
Catharus fuscescens
Sitta carolinensis
Zonotrichia leucophrys
Troglodytes troglodytes
Hylocichla mustelina
Empidonaxflaviventris
Sphyrapicus varius
Dendroica coronata
Dendroica getechia

257

LITERATURE CITED

Abrams, M. D. 1998. The red maple paradox. Bioscience 48:355-364.

Albert, D. A., S. R. Denton, and B. V. Barnes. 1986. Regional landscape ecosystems of
Michigan. University of Michigan, Ann Arbor, Michigan, USA.

Allen, A. W. 1983. Habitat suitability index models: Fisher. US. Fish and Wildlife
Service FWS/OBS-82/10.45.

Allen, A. W. 1987. Habitat suitability index models: Barred owl. US. Fish and Wildlife
Service FWS/OBS-82(10.143).

Allen, C. D. 1994. Ecological perspective: Linking ecology, GIS, and remote sensing to
ecosystem management. Pages 111-139 in V. Alaric Sample, editor. Remote
sensing and GIS in ecosystem management. Island Press, Washington, DC,
USA.

Arcese, P., and A. R. E. Sinclair. 1997. The role of protected areas as ecological
baselines. Journal of Wildlife Management 61:587-602.

Arthur, S. M., W. B. Krohn, and J. B. Gilbert. 1989. Home range characteristics of adult
fishers. Journal of Wildlife Management 53:674-688.

Aubry, K. B. 2000. Amphibians in managed, second-growth Douglas-fir forests.
Journal of Wildlife Management 64: 1041-1052.

Baker, B. W., B. C. Cade, W. L. Mangus, and J. L. McMillen. 1995. Spatial analysis of
sandhill crane nesting habitat. Journal of Wildlife Management 59:752-758.

Bart, J ., D. R. Petit, and G. Linscombe. 1984. Field evaluation of two models developed
following the Habitat Evaluation Procedures. Transactions of the North American
Wildlife and Natural Resources Conference 49:489-499.

Begalle, D. 1991. Ecosystem management: Is it possible on state lands? A forest
practitioner’s perspective. Pages 106-109 in D. C. LeMaster and G. R. Parker,
editors. Ecosystem management in a dynamic society. Proceedings of a
Conference in West Lafayette, Indiana. Purdue University, West Lafayette,
Indiana, USA.

Bender, L. C., G. J. Roloff, and J. B. Haufler. 1996. Evaluating confidence intervals for
habitat suitability models. Wildlife Society Bulletin 24:347-352.

258

Binford, L. C. 1991. Black-throated blue warbler. Pages 402-403 in R. Brewer, G. A.
McPeek, and R. J. Adams, Jr. The atlas of breeding birds of Michigan. Michigan
State University Press, East Lansing, Michigan, USA.

Blake, J. G., J. M. Hanowski, G. J. Niemi, and P. T. Collins. 1994. Annual variation in
bird pOpulations of mixed conifer-northem hardwood forests. Condor 96:381-
399.

Bond, R. R. 1957. Ecological distribution of breeding birds in upland forests of southern
Wisconsin. Ecological Monographs 27 :351-384.

Bourque, J ., and M. Villard. 2001. Effects of selection cutting and landscape-scale
harvesting on the reproductive success of two neotropical migrant bird species.
Conservation Biology 15: 184-195.

Breininger, D. R., V. L. Larson, B. W. Duncan, and R. B. Smith. 1998. Linking habitat
suitability to demographic success in Florida scrub-jays. Wildlife Society Bulletin
26:118-128.

Brooks, R. T. 1999. Residual effects of thinning and high white-tailed deer densities on
northern red-back salamanders in southern New England oak forests. Journal of
Wildlife Management 63: 1 172-1 180.

Bull, E. L., and E. C. Meslow. 1977. Habitat requirements of the pileated woodpecker in
northeastern Oregon. Journal of Forestry 75:335-337.

Burke, D. M, and E. N01. 1998. Influence of food abundance, nest site habitat, and forest
fragmentation on breeding ovenbirds. Auk 115:96-104.

Canfield, R. H. 1941. Application of the line intercept method in sampling range
vegetation. Journal of Forestry 39:388-394.

Carey, A. B., and J. W. Witt. 1991. Track counts as indices to abundances of arboreal
rodents. Journal of Mammalogy 72:192-194.

Carroll, C. W. J. Zielinski, and R. F. Noss. 1999. Using presence-absence data to build
and test spatial habitat models for the fisher in the Klamath region, U.S.A.
Conservation Biology. 13: 1344-1359.

Cole, C. A., and R. L. Smith. 1983. Habitat suitability indices for monitoring wildlife

populations - An evaluation. Transactions of the North American Wildlife and
Natural Resources Conference 48:367-375.

259

Cowan, I. M. 1936. Nesting habits of the flying squirrel Glaucomys sabrinus. Journal of
Mammalogy 17258-60.

Cunningham, R. N., and H. G. White. 1941. Forest resources of the Upper Peninsula of
Michigan. US. Department of Agriculture Miscellaneous Publication No. 429.

Davis, M. B., M. W. Schwartz, and K. Woods. 1991. Detecting a species limit from
pollen in sediments. Journal of Biogeography 18:653-688.

DeGraaf, R. M., and M. Yamasaki. 1992. A nondestructive technique to monitor the
relative abundance of terrestrial salamanders. Wildlife Society Bulletin 20:260-
264.

deMaynadier, P. G., and M. L. Hunter. 1998. Effects of silvicultural edges on the
distribution and abundance of amphibians in Maine. Conservation Biology
12:340-352.

Doepker, R. V., R. D. Earle, and J. J. Ozoga. 1992. Characteristics of blackbumian
warbler, Dendroica fusca, breeding habitat in Upper Michigan. Canadian Field-
Naturalist [06:366-371.

Eastman, J. 1991. Yellow-rumped warbler. Pages 404-405 in R. Brewer, G. A. McPeek,
and R. J. Adams, Jr. The atlas of breeding birds of Michigan. Michigan State
University Press, East Lansing, Michigan, USA.

Elody, B. J ., and N. F. Sloan. 1985. Movement and habitat use of barred owls in the
Huron Mountains. J ack-Pine Warbler 6323-8.

Everett, R. L., and J. F. Lehmkuhl. 1999. Restoring biodiversity on public forest lands
through disturbance and patch management irrespective of land use allocation.
Pages 87-106 in Baydack, R. K., H. Campa, 111, and J. B. Haufler, editors.

Practical approaches to conserving biodiversity. Island Press, Washington, DC,
USA.

Felix, A. B., H. Campa, III, and G. J. Roloff. 1999. A pileated woodpecker habitat
suitability index model for the northern Great Lakes region. On file with Boise
Cascade, Inc. Boise, Idaho, USA.

Flaspohler, D. J ., S. A. Temple, and R. N. Rosenfield. 2001. Effects of forest edges on

ovenbird demography in a managed forest landscape. Conservation Biology
15:173-183.

260

Frelich, L. E. 1995. Old forest in the Lake States today and before European settlement.
Natural Areas Journal 15:157-167.

----- , and C. G. Lorimer. 1985. Current and predicted long-term effects of deer browsing
in hemlock forests in Michigan. Biological Conservation 34:99-120.

----- , and C. G. Lorimer. 1991. Natural disturbance regimes in hemlock-hardwood
forests of the upper Great Lakes region. Ecological Monographs 61:145-164.

Gilmore, R. M., and J. E. Gates. 1985. Habitat use by the southern flying squirrel at a
hemlock-northern hardwood ecotone. Journal of Wildlife Management 49:703-
710.

Grialou, J. A., S. D. West, and R. N. Wilkins. 2000. The effects of forest clearcut
harvesting and thinning on terrestrial salamanders. Journal of Wildlife
Management 64: 105-1 13.

Hagan, J. M., P. S. McKinley, A. L. Meehan, and S. L. Grove. 1997. Diversity and
abundance of land birds in a northeastern industrial forest. Journal of Wildlife
Management 61:718-735.

Hale, C. M., J. Pastor, and K. A. Rusterholz. 1999. Canadian Journal of Forest Research
29: 1479-1489.

Hann, H. W. 1937. Life history of the ovenbird in southern Michigan. Wilson Bulletin
49: 145-237.

Haskell, D. G. 2000. Effects of roads on macroinvertebrate soil fauna of the southern
Appalachian Mountains. Conservation Biology 14:57-63.

Haufler, J. B., C. A. Mehl, and G. J. Roloff. 1999. Conserving biological diversity using
a coarse-filter approach with a species assessment. Pages 107-126 in Baydack, R.
K., H. Campa, III, and J. B. Haufler, editors. Practical approaches to conserving
biodiversity. Island Press, Washington, DC, USA.

Heatwole, H. 1962. Environmental factors influencing the local distribution and activity
of the red-backed salamander, Plethodon cinereus. Ecology 43:460-472.

Herbeck, L. A., and D. R. Larsen. 1999. Plethodontid salamander response to

silvicultural practices in Missouri Ozark forests. Conservation Biology 13:623-
632.

261

Howard, J. H., and A. R. Caschetta. 1999. Effects of clearcutting on amphibian
abundance in western Maryland. Northeast Wildlife 54:35-33-43.

Howe, R. W., G. Niemi, and J. R. Probst. 1995. Management of western Great Lakes

forests for the conservation of neotropical migratory birds. Pages 144-167 in F.
R. Thompson, editor. Management of midwestem landscapes for the

conservation of neotropical migratory birds. U. S. Forest Service General
Technical Report NC-187.

Hunt, P. D. 1998. Evidence from a landscape population model of the importance of

early successional habitat to the American redstart. Conservation Biology
12: 1377-1389.

J aeger, R. G. 1979. Seasonal spatial distributions of the terrestrial salamander Plethodon
cinereus. Herpetologica 35:90-93.

----- . 1980. Microhabitats of a terrestrial forest salamander. Copeia 22265-268.

Karamanski, T. J. 1989. Deep woods frontier: A history of logging in northern
Michigan. Wayne State University Press, Detroit, Michigan, USA.

Kilham, L. 1959. Winter foraging and associated behavior of pileated woodpeckers in
Georgia and Florida. Auk 83: 15-24.

King, D. 1., C. R. Griffin, and R. M. DeGraaf. 1996. Effects of clearcutting on habitat

use and reproductive success in forested landscapes. Conservation Biology
10:1380—1386.

Kleeberger, S. R., and J. K. Werner. 1982. Home range and homing behavior of
Plethodon cinereus in northern Michigan. Copeia 3:788-791.

Krohn, W. B., K. D. Elowe, and R. B. Boone. 1995. Relations among fishers, snow, and

martens: Development and evaluation of two hypotheses. Forestry Chronicle
71:97-105.

Langhome, W. T., Jr. 1988. An investigation of the spatial, temporal, and organizational
processes of the industrial frontier: A case study from the Upper Peninsula of

Michigan. Dissertation, Michigan State University, East Lansing, Michigan,
USA.

Leatherberry, E. C. 1993. Forest statistics for Michigan’s western Upper Peninsula unit,
1993. US. Forest Service Resource Bulletin NC-153.

262

Linhart, S. B., and F. F. Knowlton. 1975. Determining the relative abundance of coyotes
by scent station lines. Wildlife Society Bulletin 3: 1 19-124.

MacLean Consultants Ltd. No date. Inventory of deer habitat in Michigan’s Upper
Peninsula utilizing Landsat Thematic Mapper data. Unpubl. report, on file with
Michigan Department of Natural Resources.

Marcot, B. G., M. G. Raphael, and K. H. Berry. 1983. Monitoring wildlife habitat and
validation of wildlife-habitat relationships models. Transactions of the North
American Wildlife and Natural Resources Conference 48:315-329.

Martin, N. D. 1960. An analysis of bird populations in relation to forest succession in
Algonquin Provincial Park, Ontario. Ecology 41:126-140.

Mathis, A. 1989. Do seasonal spatial distributions in a terrestrial salamander reflect
reproductive behavior or territoriality? Copeia 3:788-791.

McGarigal, K., and J. D. Fraser. 1985. Barred owl responses to recorded vocalizations.
Condor 87:552-553.

----- , and B. J. Marks. 1995. FRAGSTATS: spatial pattern analysis program for
quantifying landscape structure. US. Forest Service General Technical Report
PNW-GTR-351.

Merrill, S. B., F. J. Cuthbert, and G. Oehlert. 1998. Residual patches and their
contribution to forest-bird diversity on northern Minnesota aspen clearcuts.
Conservation Biology 12: 190-199.

Michigan Department of Natural Resources. 1973. Forestry Division Policy Manual.
Lansing, Michigan, USA.

----- . 1991. Escanaba River State Forest comprehensive resource management plan.
Final Draft. Michigan Department of Natural Resources, Lansing, Michigan,
USA.

----- . 2000. Northern lower Michigan ecosystem management project. [Online]
Available http://www.dnr.state.mi.us/www/fmd/nlp/intro.htm.

Minnis, R., and J. B. Haufler. 1994. Habitat model - American redstart. Pages 177-188
in G. J. Roloff. Using an ecological classification system and wildlife habitat
models in forest planning. Dissertation, Michigan State University, East Lansing,
Michigan, USA.

263

Mladenoff, D. J ., M. A. White, J. Pastor, and T. R. Crow. 1993. Comparing spatial
pattern in unaltered old-growth and disturbed forest landscapes. Ecological
Applications 3:294-306.

Monthey, R. 2000. Aldo Leopold, selective cutting, and forest stewardship planning.
[Online] Available httpzllwww.fs.fed.us/na/durham/misc_docs/1eopold.htm.

Monti, L. 1997. Redback [sic] salamander (Plethodon cinereus) habitat preferences in a
Maine oak-pine forest. Thesis, University of Maine, Orono, Maine, USA.

Morrison, D. F. 1990. Multivariate statistical methods. Third edition. McGraw-Hill,
New York, New York, USA.

Negri, S. J. 1995. Analysis of habitat suitability models for primary cavity-nesting birds
in Michigan’s Upper Peninsula. Thesis, Michigan State University, East Lansing,
Michigan, USA.

Niemi, G. J ., J. M. Hanowski, A. R. Lima, T. Nicholls, and N. Weiland. 1997. A critical
analysis of the use of indicator species in management. Journal of Wildlife
Management 61 : 1240-1252.

Noon, B. R., V. P. Bingham, and J. P. Noon. 1979. The effects of changes in habitat on
northern hardwood forest bird communities. Pages 33-48 in R. M. DeGraaf,
technical coordinator. Proceedings of the workshop on management of north
central and northeastern forests for nongame birds. US. Forest Service General

Technical Report NC-51.

Noss, R. F. 1987. From plant communities to landscapes in conservative inventories: a
look at the Nature Conservancy (USA). Biological Conservation 41 :1 1-37.

O’Neil, L. J ., and M. Schamberger. 1983. Habitat models as a monitoring tool. Pages
424-427 in J. F. Bell and T. Atterbury, editors. Renewable resource inventories
for monitoring changes and trends: Proceedings of an International Conference,
August 15-19, 1983, Corvallis, Ore., U.S.A. Oregon State University, Corvallis,
Oregon, USA.

O’Neill, R. V., C. T. Hunsaker, K. B. Jones, K. H. Riitters, J. D. Wickham, P. M.
Schwartz, 1. A. Goodman, B. L. Jackson, and W. S. Baillargeon. 1997.
Monitoring environmental quality at the landscape scale. Bioscience 47:513-519.

Ott, L. 1988. An introduction to statistical methods and data analysis. Third edition.
PWS-Kent, Boston, Massachusetts, USA.

264

Paragi, T. F., S. M. Arthur, and W. B. Krohn. 1996. Importance of tree cavities as natal
dens for fishers. Northern Journal of Applied Forestry 13:79-83.

Payne, J. L ., D. R. Young, and J. F. Pagels. 1989. Plant community characteristics
associated with the endangered northern flying squirrel, Glaucomys sabrinus, in
the southern Appalachians. American Midland Naturalist 121:285-292.

Pomeluzi, P. A., and J. Faaborg. 1999. Season-long fecundity, survival, and viability, of
ovenbirds in fragmented and unfragmented landscapes. Conservation Biology

13:1151-1161.

Pough, F. H., E. M. Smith, D. H. Rhodes, and A. Collazo. 1987. The abundance of
salamanders in forest stands with different histories of disturbance. Forest

Ecology and Management 20: 1-9.

Powell, R. A. 1982. The fisher. Life history, ecology, and behavior. University of
Minnesota Press, Minneapolis, Minnesota, USA.

----- . 1994. Effects of scale on habitat selection and foraging behavior of fishers in
winter. Journal of Mammalogy 75:349-356.

Price, D. L. 1994. An ecological study of the composition, structure, and disturbance
regimes of the pre-European settlement forests of western Chippewa County,
Michigan. Thesis, Michigan State University, East Lansing, Michigan, USA.

Reunanen, P., M. Mbnkkonen, and A. Nikula. 2000. Managing boreal forest landscapes
for flying squirrels. Conservation Biology 14:218-226.

Robbins, C. S. 1981a. Effect of time of day on bird activity. Pages 275-286 in C. J.
Ralph and J. M. Scott, editors. Estimating numbers of terrestrial birds. Volume

6. Allen Press, Lawrence, Kansas, USA.

----- . 1981b. Bird activity levels related to weather. Pages 301-310 in C. J. Ralph and J.
M. Scott, editors. Estimating numbers of terrestrial birds. Volume 6. Allen

Press, Lawrence, Kansas, USA.

Rodewald, A. D., and R. H. Yahner. 1999. Effects of forest management and landscape
composition on woodland salamanders. Northeast Wildlife 54:45-54.

Roloff, G. J. 1994. Using an ecological classification system and wildlife habitat models
in forest planning. Dissertation, Michigan State University, East Lansing,

Michigan, USA.

265

Roloff, G. J., and B. J. Kemohan. 1999. Evaluating reliability of habitat suitability index
models. Wildlife Society Bulletin 27:973-985.

Rosenberg, D. K., and R. G. Anthony. 1993. Differences in trapping mortality rates of
northern flying squirrels. Canadian Journal of Zoology 71:660-663.

Sample, V. A. 1995. Partnerships for ecosystem management on mixed ownership
landscapes. Transactions of the North American Wildlife and Natural Resources

Conference 60:415-421.

Samson, F. B., and S. J. Lewis. 1979. Experiments and population regulation in two
North American parids. Wilson Bulletin 91:222-233.

Schamberger, M., and L. J. O’Neil. 1986. Concepts and constraints of habitat model
testing. Pages 5-10 in J. Vemer, M. L. Morrison, and C. J. Ralph, editors.
Wildlife 2000. University of Wisconsin Press, Madison, Wisconsin, USA.

Schmidt, LT. 1993. Forest statistics for Michigan’s eastern Upper Peninsula unit, 1993.
US. Forest Service Resource Bulletin NC-150.

Schroeder, R. L. 1982. Habitat suitability index models: Pileated woodpecker. US. Fish
and Wildlife Service. FWS/OBS-82/10. 146.

Siegel, S., and R. Castellan. 1988. Nonparametric statistics for the behavioral sciences.
McGraw Hill, New York, New York, USA.

Simpson, T. B. 1990. Landscape ecosystems and cover types of the reserve area and
adjacent lands of the Huron Mountain Club, Marquette County, Upper Michigan.

Dissertation, University of Michigan, Ann Arbor, Michigan, USA.

----- , P. E. Stuart, and B. V. Barnes. 1989. Cover types at the Huron Mountain Club,
Marquette County, Michigan. Huron Mountain Wildlife Foundation (color map).

----- , P. E. Stuart, and B. V. Barnes. 1990. Landscape ecosystems and cover types of the
reserve area and adjacent lands of the Huron Mountain Club. Occasional Papers

of the Huron Mountain Wildlife Foundation No. 4.

Smith, T. M., and H. H. Shugart. 1987. Territory size and variation in the ovenbird: The
role of habitat structure. Ecology 68:695-704.

Sokal, R. R., and F. J. Rohlf. 1981. Biometry. W. H. Freeman and Company, New
York, New York, USA.

266

Sonenshine, D. E., D. G. Cerretani, G. Enlow, and B. L. Elisberg. 1973. Improved
methods for capturing wild flying squirrels. Journal of Wildlife Management

37:588-590.

Sousa, P. J. 1982. Habitat suitability models: Veery. US. Fish and Wildlife Service.
FWS/OBS-82/ 10.22.

Sturm, L. 1945. A study of the nesting activities of the American redstart. Auk 62:189-
206.

Takats L. D., and G. L. Holroyd. 1997. Owl broadcast surveys in the Foothills Model
Forest, Alberta, Canada. Pages 421-431 in J. R. Duncan, D. H. Johnson, and T.
H. Nicholls, editors. Biology and conservation of owls of the northern
hemisphere, second international symposium. US. Forest Service General

Technical Report NC-190.

Test, F. H., and B. A. Bingham. 1948. Census of a population of the red-backed
salamander (Plethodon cinereus). American Midland Naturalist 39:362-372.

Test, F. H., and H. Heatwole. 1962. Nesting sites of the red-backed salamander,
Plethodon cinereus, in Michigan. Copeia 1962:206-207.

The Nature Conservancy. 1982. Natural heritage program operations manual. The
Nature Conservancy, Arlington, Virginia, USA.

Thomas, A. H. 1983. First year responses of wildlife and wildlife habitat to sewage
sludge application in northern hardwoods forest. Thesis, Michigan State

University, East Lansing, Michigan, USA.

Thomas, J. W., R. G. Anderson, C. Maser, and E. L. Bull. 1979. Snags. Pages 60-77 in
J. W. Thomas, technical editor. Wildlife habitats in managed forests: the Blue

Mountains of Oregon and Washington. US. Forest Service Agricultural
Handbook.

Thomasma, L. E., T. D. Drummer, and R. 0. Peterson. 1991. Testing the habitat
suitability model for the fisher. Wildlife Society Bulletin 19:291-297.

Ticknor, W. D. 1993. Ecosystem management: the TEMS approach. American Forests
99:34-36.

Trombulak, S. C., and C. A. Frissell. 2000. Review of ecological effects of roads on
terrestrial and aquatic communities. Conservation Biology 14:18-30.

267

Tyrell, L. E., and T. R. Crow. 1994. Structural characteristics of old-growth hemlock-
hardwood forests in relation to age. Ecology. 75:370-386.

US. Department of Agriculture. 1995. The Forest Service program for forest and
rangeland resources: A long-term strategic plan. US. Forest Service FS-6200—
28b(4/88).

Van Deelen, T. R. 1995. Seasonal migrations and mortality of white-tailed deer in
Michigan’s Upper Peninsula. Dissertation, Michigan State University, East
Lansing, Michigan, USA.

Van Horne, B. 1983. Density as a misleading indicator of habitat quality. Journal of
Wildlife Management 47:893-901.

Verme, L. J. 1973. Movements of white-tailed deer in Upper Michigan. Journal of
Wildlife Management 37:545-552.

Wall, W. H. 1999. Maintaining biodiversity in an intensively managed forest: A habitat-
based planning process linked with a fine-filter adaptive management process.
Pages 127-140 in Baydack, R. K., H. Campa, III, and J. B. Haufler, editors.
Practical approaches to conserving biodiversity. Island Press, Washington, DC,
USA.

Weigl, P. D., and D. W. Osgood. 1974. Study of the northern flying squirrel,

Glaucomys sabrinus, by temperature telemetry. American Midland Naturalist
92:482-486.

Wells, J. R., and P. W. Thompson. 1972. Vegetation and flora of the Huron Mountains.
Occasional Papers of the Huron Mountain Wildlife Foundation No. 3.

Welsh, H. H. Jr., and S. Droege. 2001. A case for using plethodontid salamanders for
monitoring biodiversity and ecosystem integrity of North American forests.
Conservation Biology 15:558-569.

Western Regional Climate Center. 1999. [Online] Available http://www.wrcc.dri.edu.

Westover, A. J. 1971. The use of a hemlock-hardwood winter yard by white-tailed deer

in northern Michigan. Occasional Papers of the Huron Mountain Wildlife
Foundation No. 1. ~

268

Whitcomb, R. F., C. 8. Robbins, J. F. Lynch, B. L. Whitcomb, M. K. Klimkiewicz, and
D. Bystrak. 1981. Effects of forest fragmentation on avifauna of the eastern
deciduous forest. Pages 125-205 in R. L. Burgess and D. M. Sharpe, editors.
Forest island dynamics in man-dominated landscapes. Springer-Verlag, New
York, New York, USA.

Winnett-Murray, K. 1991. Veery. Pages 350-351 in R. Brewer, G. A. McPeek, and R. J.
Adams, Jr. The atlas of breeding birds of Michigan. Michigan State University
Press, East Lansing, Michigan, USA.

Witt, Joseph W. 1992. Home range and density estimates for the northern flying
squirrel, Glaucomys sabrinus, in western Oregon. Journal of Mammalogy
73:921-929.

Wright, R. T. 1991. Adaptation in industrial forest management. Pages 61-64 in D. C.
LeMaster and G. R. Parker, editors. Ecosystem management in a dynamic
society. Proceedings of a Conference in West Lafayette, Indiana. Purdue
University, West Lafayette, Indiana, USA.

Wyman, R. L. 1998. Experimental assessment of salamanders as predators of detrital
food webs: effects on invertebrates, decomposition, and the carbon cycle.
Biodiversity and Conservation 72641-650.

Wyman, R. L., and D. S. Hawksley-Iescault. 1987. Soil acidity affects distribution,

behavior, and physiology of the salamander, Plethodon cinereus. Ecology
68: 1819-1827.

269

 

    

MM, HHJAN HAIL UUI‘AHII S

3 1293 02177 0684