. 1;},
.o r
. Alli-I!

a .55..)
. 1‘ {I I 'I.

r
.

L... ‘ ..
.ﬁuq
HI. $5.1:
gum. g.
. , ‘ , a. 3....
5+.
‘n'. '4. 43 . ‘
.. r1 .. is :
f. 353...:

ii...)
. \r . 0

 

;.
)1

u...

.5...

3
.n ..
eat... .53..

o I ﬁg!!!
5.5.1:: '1 J37: .}
.x»|.?££

 

, I»! A 4 .
3. x
4. ﬁgs”... {km

4 ﬁr.

:. .. 33$): :x

kit-12:13

2_ M LJBRARY
ic igan State
W University

This is to certify that the
dissertation entitled

SUBSTRATE LIMITATIONS TO TSUGA CANADENSIS AND
BETULA ALLEGHENIENSIS SEEDLING ESTABLISHMENT

presented by

Laura Michelle Marx

has been accepted towards fulﬁllment
of the requirements for the
Doctoral degree in Forestry and
Ecology, Evolutionary Biology.
/7 and Behavior

    

Major Professor’é’ Signature ’

247%“?

Date

MSU is an Afﬁnnative Action/Equal Opportunity Institution

 

PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2/05 p:/CIRC/DaleDue.indd-p.1

 

SUBSTRATE LIMITATIONS TO T SUGA CANADENSIS AND BET ULA
ALLEGHENIENSIS SEEDLING ESTABLISHMENT

By

Laura Michelle Marx

A DISSERTATION

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

DOCTOR OF PHILOSOPHY

Department of Forestry
Program in Ecology, Evolutionary Biology, and Behavior

2005

ABSTRACT
SUBSTRATE LIMITATIONS TO T SUGA CANADENSIS AND BET ULA
ALLEGHENIENSIS SEEDLING ESTABLISHMENT
By
Laura Michelle Marx

In this dissertation, I provide evidence that the distribution of hemlock (T suga
canadensis), yellow birch (Betula allegheniensis), and sugar maple (Acer saccharum)
decaying wood maintains two patterns of tree distribution in Upper Michigan: the eastern
hemlock-northem hardwood patch structure and the hemlock/yellow birch spatial
association. Patches (3-30 ha) of hemlock with scattered yellow birch have remained
hemlock-dominated and the same size for over 3000 years, even when adjacent to patches
of northern hardwood forest usually dominated by sugar maple. Across both patch types,
hemlock are most closely spatially associated with yellow birch, an association that
makes little sense from a life history perspective, since yellow birch is a gap-phase
hardwood and hemlock is a Iate-successional often slow-growing conifer. However, both
hemlock and yellow birch seedlings are most abundant on wood and, I demonstrate here,
in particular on hemlock wood. I show that hemlock wood is the most favorable substrate
for hemlock and yellow birch seedling establishment (seedling density = 0.42 hemlocks
/m2, 0.60 birches /m2), followed by yellow birch wood (0.21, 0.15), and that sugar maple
wood (0.08, 0.10) and undisturbed soil (0.01, 0.01) are less suitable and support few to no
hemlock and yellow birch seedlings older than three years. Sugar maple seedlings, in
contrast, do not establish on any species of decaying wood (sugar maple seedling density
= 0.03 to 0.09 /m2 across wood species). Hemlock and yellow birch wood are rare

everywhere, but are most abundant in hemlock patches where they cover 2.8% of the

forest ﬂoor, reinforcing the hemlock-northem hardwood patch structure and the spatial
association between hemlock and yellow birch.

I combine ﬁeld studies of seedling demographics, wood distribution, seed rain,
and decaying wood properties in three ﬁeld sites in Upper Michigan, USA with
greenhouse studies of seedling growth, ectomycorrhizal colonization, and nutrient content
to determine why hemlock wood and to a lesser extent yellow birch wood support higher
densities of hemlock and yellow birch seedlings than either sugar maple wood or soil.
Hemlock logs are more favorable for hemlock and yellow birch seedling establishment
for several reasons, among them lower pH, sufﬁcient nitrogen and phosphorus supply, a
tendency to decay more slowly than hardwood logs and to be attacked by brown rot rather
than white rot decay fungi, and a tendency to lose bark cover and develop moss cover. A
greater ability to provide ectomycorrhizal inoculum to seedlings and the relative absence
of sugar maple seedlings on hemlock logs may also contribute to the higher survival rates
of hemlock and birch seedlings. The full text of this dissertation is available free of

charge until at least 20] 0 at www.1auramarx.net.

ACKNOWLEDGEMENTS

My parents say that they knew I was going into science when I became
inseparably devoted to a millipede as a toddler. Since then, and quite possibly before,
they have provided incredible support and advice, as well as a ﬁrst-rate college education,
for which I thank them. I also thank my brother Steve, who may prefer computers to trees
but helped me rebuild a cat skeleton and allowed himself to be dragged to such horrible
places as Acadia National Park, Yosemite, and the Everglades on family vacations. All
of the Marxes have been examples of how to live a fulﬁlling, kind, and honest life.

Emily Leachman, an awfully science-savvy children's librarian, gets genuinely
excited when I tell her about the results of my newest statistical analyses. This is not an
easy thing to do, though both Emily and my family do a very convincing job of it. Emily
has been a wonderful friend since my ﬁrst week of college, but after she moved to
Lansing she was an even more constant source of support, excitement, and graph
proofreading. Not to mention spinach balls and homemade bread.

Rob Edge has never known me to not be working full-time on writing my
dissertation. That he has stuck around this long in spite of that speaks to his patience and
perhaps his masochism. I teach him about trees, he teaches me about music and math,
and it all seems to work out. I owe him one, though.

My advisor, Mike Walters, has consistently provided me with just about anything
I needed in my graduate career. In addition to signiﬁcant ﬁnancial support for this
project, Mike gave me the freedom to make enough of my own mistakes to learn from
them, but not enough to get discouraged. He has always allowed me to re-arrange my
research and even graduation plans in order to learn how to teach, and is quite possibly

the only advisor I know who would volunteer to help pick roots from pots full of wood on

a Saturday. For the original idea behind this research, help at every step of the way, and
especially for teaching me how to write convincing grants and clear papers, thanks.

My labmates and might-as-well—be-labmates were a great support network and
always fun to work with. John Gerlach, Justin Kunkle, Joseph Lebouton, and Jesse
Randall made me feel welcome even though my name doesn't start with J. Jeff Eickwort,
Meera Iyers, Steve Leduc, Nick Lisuzzo, Shawna and Jason Meyer, Sarah Neumann,
Laura Schreeg, Sue Spaulding, Corine Vriesendorp, and Zhanna Yermakov were always
willing to have me bounce ideas off of them or take a much-needed trip to the Dairy
Store. Justin, Steve, and Sue also inspired me to run the Grand Rapids marathon in
October 2005.

I worked with three of the most competent and consistently cheerful ﬁeld
assistants out there during the course of this project. Scott Kissman, Marcie Wawzysko,
and Laurie Gilligan were absolutely essential, and without any one of them this project
would not have been completed. Clyde the ﬁeld truck worked as hard as any of us. In
addition, virtually every student who ever worked for Justin or Jesse (and sometimes even
Joseph and Sarah), provided lab and greenhouse help when I needed it, especially Melissa
McDermott who helped with my abbreviated ﬁeld and lab season in 2005 and Trent
Thompson who helped with tree coring in 2004. Lynn Holcomb, Nick Lisuzzo, and
Gerald Mileski in the Rothstein lab, and Justin Zyskowski in the Kamdem lab ran
analyses of wood and plant samples for me and helped me turn computer output into
coherent statements. Michelle Holland at the US Forest Service and Dean Wilson and
William Brondyke in the Michigan DNR helped greatly with ﬁeld site selection and ﬁeld

sampling. David Gosling, William and Ann Manniere, and Kerry Woods at the Huron

Mountain Wildlife Foundation gave me access to an incomparable ﬁeld site and then
helped me ﬁnd the hemlock seedlings in it.

My college mentors, Oscar Will III, Siobhan Fennessy, and Karen Hicks, taught
me how to design and carry out a research project long before I started this one. Along
with Mike, they are examples of how to be good people and notjust good teachers and
scientists.

My committee members, Mike Walters, Kay Gross, David Rothstein, Pascal
Kamdem, and my might-as-well-have-been-committee-members Craig Lorimer at the
University of Wisconsin-Madison and Kerry Woods at Bennington College provided
much needed scientiﬁc and life experience. They have all greatly improved this project.
Mike, Kay (my link to KBSers and greenhouse space), David, and Pascal gave me great
questions on my prelims and impressed upon me the importance of Wedin and Tilman.
Craig was my host advisor in Wisconsin and he and Kerry shared natural history
observations based on more years in the ﬁeld than I have spent on earth. Kerry also
provided valuable feedback on early drafts of several of my dissertation chapters.

And ﬁnally, I'm still amazed by how many people were genuinely interested in
this research and willing to trust me with large sums of money in order to carry it out.
Michigan State University got me started with a Plant Science Fellowship, and continued
their support with a Research Enhancement grant and a Hanover Scholarship. The
National Science Foundation supported me for three years with a Graduate Research
Fellowship and my project for two with a Doctoral Dissertation Improvement Grant.
Two private foundations, the Huron Mountain Wildlife Foundation and the Hanes Fund

of the Michigan Botanical Club, gave generous support for my ﬁeld seasons.

vi

Many researchers grow to dislike their ﬁeld sites after spending so much time
battling mosquitoes, heat, boredom, and outright futility in them. I still feel privileged to
have spent so much time in old growth hemlock-hardwood forests, and to have learned so
much about such large, old trees by studying seedlings on such a small part of the forest

floor beneath them.

vii

TABLE OF CONTENTS

List of Tables ............................................................................................................. ix
List of Figures ............................................................................................................ xi
Chapter 1
General introduction ...................................................................................... 1
Literature review and natural history ............................................................. 2
Maintenance of hemlock and birch distribution ............................................ 5
Chapter 2

Tsuga, Betula, and Acer seedling distributions across forest ﬂoor substrates
in Upper Michigan, USA I: Patterns of seedling distribution and survival... 13

Introduction .................................................................................................... 14

Methods .......................................................................................................... l 7

Results ............................................................................................................ 2 I

Discussion ...................................................................................................... 27
Chapter 3

Tsuga, Betula, and Acer seedling distributions across forest ﬂoor substrates
in Upper Michigan, USA II: Mechanisms underlying seedling distribution

........................................................................................................................ 44

Introduction .................................................................................................... 44

Methods .......................................................................................................... 50

Results ............................................................................................................ 63

Discussion ...................................................................................................... 72
Chapter 4

Additional methods ........................................................................................ 94

Identiﬁcation of wood to species ................................................................... 94

Wood decay fungi classiﬁcation .................................................................... 99
Chapter 5

Conclusions .................................................................................................... 101
Literature cited ........................................................................................................... 103

viii

LIST OF TABLES

Table 2.1. Locations and characteristics of ﬁeld sites. Weather data were obtained from
NOAA COOP weather stations and are the sum of 2003 monthly averages: Marquette
weather station (46°33'N / 87°23'W) for the Huron Mountains, Marquette Wso Airport
(46°32'N / 87°33'W) for Sand River, Ontonagon (46:50N/89zl2W) for the Porcupine
Mountains, and the Lac Vieux Desert, WI (46:07N/89z07W) station for Sylvania.
Ecological sections follow Albert 1995. ATM = Acer-Tsuga-Maianthemum, ATD =
Acer- Tsuga-Dryopteris, and TMC = T suga-Maianthemum-Coptis (Kotar et al. 1988).
35

Table 2.2. Density (seedlings/m2 with s.e. in parentheses) of hemlock and birch seedlings
on different forest ﬂoor substrates. Established seedlings are older than one year, ﬁrst-
years refer only to seedlings in their ﬁrst growing season. Hemlock densities are
averaged over 2002-2004, while birch and sugar maple densities exclude a mast year in
2003 (2003 is excluded for ﬁrst-year seedlings, 2004 for established seedlings). n = 302
logs and soils in 2002, 308 in 2003, and 312 in 2004 ................................................ 36

Table 2.3. Percent survival of birch and hemlock seedlings on different log species. 2003
represents survival from late summer 2002 to mid summer 2003. 2004 represents
survival to the third, and 2005 survival to the fourth, growing season. For each column
(not row), percent survivals with different letters were signiﬁcantly different (Fisher's
exact test, individual contrasts between species, p < 0.05). n = 96 hemlock and 85 birch
seedlings, sugar maple seedlings not shown .............................................................. 37

Table 3.1. Best-ﬁt explanatory models of numbers of seedlings occurring on each log at
the Porcupine Mountains. Chi-square and p-values are given for each signiﬁcant factor in
the model. Akaike information criteria (AICc) were used to select each best-ﬁt model.
Not all second-order interactions (and no higher-order interactions) were tested since
many interactions, especially those containing categorical variables such as rot type,
resulted in models that did not converge ................................................................... 81

Table 3.2. pH, seed rain, canopy openness, fungal rot type, and water content of hemlock,
birch, and sugar maple logs. Means are presented with l s.e. in parentheses. Means
followed by the same letter are not signiﬁcantly different (student's t-tests, all possible
contrasts, alpha = 0.05). Data from all sites were pooled for factors not signiﬁcantly
affected by ﬁeld site. N = 63 logs for pH; 138 logs with seed traps; 314 logs for canopy
openness (manual threshold); 271 logs for fungal rot type, and 318 logs for moss cover.
...................................................................................................................... 82

Table 3.3 Effects of log species and light on seedling mass and N content. Values are p-
values from two-way ANOVA effect tests. Data were logarithm-transformed before
analysis. In all cases, the interaction between logs species and canopy openness was not
signiﬁcant (p > 0.250) and the term was removed from the ANOVA ....................... 83

Table 3.4 Macronutrient percent of plant dry mass (g/g, Ca, N, P, K) and micronutrient
concentrations (ppm, B, Cu, Fe, Mg, Mn, Zn) in whole birch seedlings grown on three
species of wood in the greenhouse. Each value represents the mean nutrient
concentration of composite samples (4 entire birch seedlings from a single pot make up
each sample, N = 5 to 7 for each wood species), except for reference values, which are
the ratio of macronutrients to one another (N is set to 100) in Betula verrucosa seedlings
at optimum nutrition (Ingestad 1971), and the % dry mass that would be expected in our
data given these ratios and a %N (setpoint for the ratios) of 1.60. l s.e. is in parentheses.
* = Pearson's correlation between nutrient concentration and birch mass is signiﬁcant,
alpha = 0.05 ................................................................................................................ 84

Table 4.1. Characteristics of hardwood wood for species identiﬁcation ................... 96

Table 4.2. Characteristics of conifer wood for species identiﬁcation ........................ 97

LIST OF FIGURES

Figure 2.1. Distribution of surface area of logs. All sites are pooled, n = 318 logs total.
....................................................................................................................... 38

Figure 2.2. Hemlock and birch seedling densities across substrates at the Porcupine
Mountains. Densities are means + l s.e. of established (older than one year) seedlings,
averaged across all years for hemlock and excluding the year following a mast year
(2004) for birch. n = 101 logs and 101 soils. Other sites show different densities but
similar patterns across substrates, as indicated in Table 2.2 ...................................... 39

Figure 2.3. Probability of seedling presence (in one or more years, 2002-2004) on logs, by
log surface area and species. n = 296 logs present for all three years; Sand River
excluded, minimum of 6 logs of each species in each surface area class .................. 40

Figure 2.4. Age distribution of hemlock, yellow birch, and sugar maple seedlings on
different log species. Log sample sizes are listed after each Y axis label. When
comparing across log species, note the different scale for sugar maple logs due to
oversampling of this species. n = 117 hemlock, 83 birch, and 32 sugar maple seedlings.
........................................................................................................................ 41

Figure 2.5. Estimated probabilities of survival of seeds and seedlings on hemlock logs,
sugar maple logs, and soil at the Porcupine Mountains ﬁeld site. Probabilities of survival
for hemlock seeds or seedlings are listed ﬁrst, followed by the probability of birch
survival. All probabilities were generated by dividing the density of seedlings in an age
class by the density of seedlings/seeds in the previous age class, except for probabilities
of survival from ﬁrst-year to establishment on logs. These probabilities are based on
survival of tagged seedlings ....................................................................................... 42

Figure 2.6 Paired hemlock and birch canopy trees. The birch tree is labeled with a B, the
hemlock tree with an H. Photo was taken at Sylvania Wilderness Area, near Devil's Head
Lake. ........................................................................................................................ 43

Figure 3.1. Field water content of wood pieces in 2002 and 2003 and water content of
wood measured at the point at which birch seedlings planted on each wood piece wilted
in the greenhouse. Means are presented with error bars representing 1 s.e. Field 2002 n =
239 logs, ﬁeld 2003 n = 172 logs, greenhouse birch n = 30 pots .............................. 86

Figure 3.23 and b. Ages of hemlock, birch, and sugar maple logs in decay stages 11 and
IV. a. (left) Conservative objective criteria for determining release. Release dates are the
average of two trees per log. n = 5 hemlock, 8 birch, and 9 maple logs. b. (right)
Subjective criteria for determining release. Release dates are averaged when both trees
for a single log showed a release, but included even if from a single tree. n = 10 hemlock,
26 birch, and 28 maple logs. Outer limits of each box represent the 25th and 75"I
percentile, while whiskers extend to 1.5 x the interquartile range ............................. 87

xi

Figure 3.3. Median rates of N mineralization (Nmin) and inorganic N concentrations
([N]) in wood in 2002 and 2004. Nmin is plotted on the secondary Y axis. Bars represent
medians, error bars represent the 25"I and 75'“ percentiles. ANOVA tests performed on
logarithm-transformed data indicated that the [N] of hemlock logs was lower than that of
both birch and sugar maple logs in 2002, and lower than that of sugar maple logs in 2004.
n = 128 wood samples in 2002 and 129 in 2004. Note that this is a larger sample than the
subset from which 2-year-old seedlings were collected and used to test seedling/N supply
relationships. 88

Figure 3.4. 3. Median seedling mass (g), and b. Median N content (mg), of birch and
hemlock seedlings growing on three different wood species. 11 = Birch: 50 logs with
birch seedlings in the ﬁeld and 60 in the greenhouse; Hemlock: 48 and 59 ............. 89

Figure 3.5. Mass of mycorrhizal and non-mycorrhizal greenhouse seedings. n = 233
birch and 258 hemlock seedlings. Only unfertilized seedlings are shown ............... 90

Figure 3.6. Greenhouse seedlings showed great variability in size. a. Mycorrhizal birches
growing on a hemlock log. b. Non-mycorrhizal birches on a birch log. 0. F ertilized
birch. d. Non-mycorrhizal hemlocks on a single sugar maple log - -----.9l

 

Figure 3.7. Response of birch seedlings to fertilization with N when grown on different
wood species. Symbols represent mean values of birch seedling mass, pooled by pot.
Error bars represent 1 s.e. n = 18 (6 hemlock, 8 birch, and 4 sugar maple) fertilized and
60 unfertilized pots (20 of each species) .................................................................... 92

Figure 3.8. Relationship between birch seedling N:P and mass (pooled mass of four birch
seedlings grown on the same piece of wood). 11 = 16-- - ...... -- - 93

 

Figure 3.9. Conceptual diagram of factors inﬂuencing ﬁrst-year and established seedling
survival on decaying wood. Factors signiﬁcant in best-ﬁt models are indicated with solid
lines. The dashed line indicates a factor that could not be tested with our data ....... 94

Figure 3.10. F irst-year sugar maple seedlings growing on a log and on nearby soil. The
seedlings on the log (to the right of the dotted line) show signs of nutrient deﬁciency,
unlike the ﬁrst-year sugar maples on soil beside them - .............. 91

 

xii

CHAPTER ONE
INTRODUCTION
GsnmLintmslusliQn

Let me begin by explaining the title without forest ecology terms. This is a study
of substrate limitation to eastern hemlock and yellow birch establishment, or a study of
the "preference," if seedlings could choose, of hemlock and birch seedlings to grow on
only certain parts of the forest ﬂoor (substrates), parts that are so rare that their scarcity
limits the numbers of those seedlings that are able to survive more than one year
(establishment). The substrates 1 am particularly interested in are pieces of decaying
wood (logs, downed branches, and stumps).

D . . E I I. .

I begin with a literature review and my own observations concerning the natural
history of hemlock/northem hardwood forests in Upper Michigan. I'll cover where
hemlock and yellow birch grow and why this is an interesting pattern to research. Then I
will list and explain some possible explanations for the distribution of hemlock and
yellow birch.

Chapter 2 can be thought of as the patterns chapter. This chapter, and chapter 3 as
well, contains a draft of a manuscript that will be submitted for publication. In chapter 2,
1 show numerical evidence for the patterns of hemlock and yellow birch distribution
described in the introduction. I also describe the distribution of decaying hemlock, birch,
and sugar maple wood in my ﬁeld sites.

Chapter 3 is a mechanisms chapter. I cover the various factors that I tested to see
which factor(s) could explain why certain species of decaying wood support more

hemlock and yellow birch seedlings than do others. These factors include: light levels,

seed rain, pH, nitrogen and phosphorus content, potential for mycorrhizal inoculation of
seedlings, wood moisture content, decay "pattern", and residence time of wood. Some of
these factors were measured in the ﬁeld, while others were examined in the greenhouse
using planted hemlock and birch seedlings.

Chapter 4 is a list of detailed methods not included in other chapters, including a
key to identiﬁcation of decaying wood to the species level, and chapter 5 is a general
conclusion.

Il' . I

Forests containing eastern hemlock (T suga canadensis (L.) Carr.) covered most of
Upper Michigan and northern Wisconsin until the end of the 19th century (Comer et al.
1998). Although hemlock is now much rarer, covering an estimated 0.5% of the
landscape in 1980 (Eckstein 1980) and declining at least slightly since then (Woods
2000), it is still an ecological dominant in the forests in which it occurs. Hemlock is a
long-lived canopy tree and has strong effects on the light availability and soil chemistry
beneath the canopy (Finzi et al. I988b, Ferrari 1999). Hemlock seedlings, however, have
high mortality rates in the ﬁrst year (Potzger and Friesner 1932 in Friesner and Potzger
1944) and grow very slowly, causing potential for problems with hemlock regeneration.
This contrast between adult canopy trees that are relatively hardy and survive even when
perched on boulders, steep hillsides, or stumps, and seedlings that are killed by relatively
mild weather events and even by hardwood litter has brought more attention to hemlock
than might be expected of a tree that has never been a particularly valuable timber species
(Tubbs 1995). Frederick Clements (Clements in Rogers 1978), Henry David Thoreau

(1854), and Aldo Leopold (1938) are among the early observers of hemlock forest natural

history, while Sarah Harlow discussed the physiology and requirements of hemlock
seedlings as early as 1900.

Yellow birch (Betula allegheniensis Britton) is similar to hemlock in distribution
and seedling requirements, but is an economically valuable timber species and less of an
ecological dominant than hemlock. Much of the research concerning birch (I use “birch”
rather than “yellow birch” throughout -- other birch species are mentioned only in
Chapter 3) has been related to forest management, and there is abundant information
about its nutritional and site requirements (Tubbs 1969, USDA Forest Service 1969,
Canavera 1978, Peterson and F acelli 1992). Birch is ﬁrmly in the middle of many
gradients used to group tree species (Sutherland et al. 2000) and so studies of its
regeneration ecology under old-growth conditions, or its natural history in general, are
slightly more difﬁcult to ﬁnd (Steams 1951, Reif 1992, Peterson 2000, Woods 2000).
Yet many researchers who explicitly studied only hemlock, as even I did in the ﬁrst few
months of designing this study, have included observations about birch in their
introduction or discussion sections (Hough 1936, Rogers 1980, Frelich et al. 1993).

II I l . E I I I I II I . I

In Michigan’s Upper Peninsula, hemlock occurs either in hemlock-dominated
stands or hemlock-hardwood (often dominated by sugar maple, Acer saccharum Marsh.)
stands (Pastor and Broschart 1990). In old growth forests such as the Sylvania
Wilderness, small patches of these two stand types often border one another, and the
boundaries of each patch have changed little in the past 3,000 years (Davis et al. 1993).
There are several ways in which hemlock patches could have originated in the Upper
Peninsula. One way is by replacement of white pine with hemlock in those areas with

suitable soil conditions for pine growth, possibly after ﬁre (Davis et al. 1995, Davis et al.

1998). Davis (1995) also speculates that warmer, drier climatic conditions during
hemlock invasion of Upper Michigan (approximately 3200 years ago) allowed hemlock
to invade areas with adequate moisture, while other areas changed from oak dominance to
sugar maple dominance.

One tree species is commonly found in both hemlock and sugar maple patches:
yellow birch. Birch is a gap-phase generalist species, but shows a distinct association
with hemlock throughout the northern Great Lakes forests, and in both hemlock and
hemlock-hardwood patches (Steams 1951, Forest Service 1965). In an analysis of the
1996 Forest Inventory Analysis data from northern Wisconsin, Kotar et al. (1999) found
that hemlock and birch have the highest degree of association out of all 22 species found
in these forests. The same studies (Davis et al. 1993, 1995) that show the replacement of
pine pollen with hemlock pollen during the formation of hemlock patches show a
corresponding increase in birch pollen, and replacement of paper birch macrofossils with
yellow birch macrofossils (S. Finkelstein, University of Toronto post-doc, pers. comm.
2005).

From a shade tolerance and life history perspective, the hemlock/birch association
is counterintuitive (Crow 1995). Hemlock should consistently shade out birch (in low
light), or birch should overtop hemlock (in high light), yet the two species not only
coexist but often form pairs or triplets of similar-diameter adult trees (Figure 2.6). While
the association of adult hemlock and birch trees is difﬁcult to explain, the association at
the seedling stage makes more sense. Hemlock and birch are both small-seeded, require
the same germination temperature, and need consistently moist substrates in order to
reach the sapling stage (Houle and Payette 1990, Tubbs 1995). As many as 88% of

hemlock and 74% of birch seedlings that germinate die in the ﬁrst year (Potzger and

F riesner 1932 in F riesner and Potzger 1944, Linteau 1948). Most of these seedlings die
when the substrate on which they have germinated becomes dry. Hemlock roots, and
birch roots when under dense shade, grow approximately 13 mm (0.5 in) into the soil
during the ﬁrst full year of growth, which means that even mild droughts resulting in
drying of the top layer of soil can result in seedling death (Friesner and Potzger I944,
Linteau I948, Godman and Lancaster 1990, Tubbs 1995). Hemlock and birch seedlings,
however, are often able to start growing in the same places, using germination substrates
that are not utilized by most hardwood species (see Tubbs 1995 and Rogers 1980 for
anecdotal evidence), possibly sharing nutrients through shared mycorrhizal networks
(Booth 2004 and in progress), and coexisting on these substrates through the sapling and
canopy tree stages.
n k ' i

Long-tenn maintenance of both the hemlock patch structure and the
hemlock/birch association must depend on strong feedback mechanisms (F relich et al.
1993). Below, I introduce ﬁve of the mechanisms known to help maintain hemlock and
birch distribution along with evidence to support each. I end by proposing my own

hypothesis, substrate limitations to the establishment of hemlock and birch seedlings.

l. Slower nitrogen cycling under hemlock canopies (Campbell and Gower 2000)
One form of positive feedback that may maintain hemlock and mixed
hardwood patches is vegetation-caused differences in soil nutrient availability and
chemistry under hemlock and hardwood canopies. Under current soil moisture and
climatic conditions, the only signiﬁcant differences in cation availability and pH

beneath each forest type appear to be directly caused by the plants currently

occupying each patch of soil (Bockheim 1997). It is not the case that hemlock and
hardwood stands became established on inherently chemically different patches of
soil (Frelich et al. 1993, Bockheim 1997). Once hemlocks are established on a site,
though, nitrogen mineralization rates tend to be lower under hemlock canopies than
under sugar maple canopies, and the high lignin:N ratio in hemlock litter helps to
maintain this difference (Ferrari 1999, Mladenoff 1987). In addition, hemlock wood
has smaller extractable pools of inorganic N than either birch or sugar maple wood
(see Chapter 3), suggesting that both hemlock litter inputs and wood inputs contribute
to the slower cycling of N under hemlock canopies. Not all studies have determined
that nitrogen mineralization rates are lower under hemlocks, though. The conclusion
authors have come to appears to depend on whether nitrogen cycling is measured at
the individual tree scale or the forest stand scale.

Mladenoff (1987) measured differences on the scale of canopy trees,
comparing the forest ﬂoor under hemlock canopies, maple canopies, and gaps in each
forest type. Both N mineralization and nitriﬁcation rates were higher under sugar
maple canopies than under hemlocks. In gaps, results were more confusing, with N
mineralization actually higher under hemlock gaps than under sugar maple gaps, and
nitriﬁcation rates approximately equal. Mladenoffs study suggests that differences in
nutrient cycling must be measured on the level of individual trees, not stands, which
are too variable to allow detection of differences in nutrient cycling. Finzi et al.
(1998b) supported Mladenoff’s results. Their measurements of the forest floor
beneath individual hemlock and sugar maple trees (along with four other species)
suggested that net N mineralization and nitriﬁcation rates were lower, though not

signiﬁcantly so, under hemlock than under sugar maple trees. Both papers suggest

that sugar maple and hemlock litter have similar amounts of nitrogen, and so it is the
speed of decay of litter that results in the differences in nutrient cycling beneath each
canopy. Finally, Ferrari (I 999) showed a direct correlation between litterfall lignin:N
ratio and N mineralization and nitriﬁcation rates in small plots in Sylvania, supporting
the conclusions of Mladenoff, F inzi, and others that hemlock litter is responsible for
the slower nitrogen cycling under hemlock canopies. Conifer species such as
hemlock have higher nitrogen use efﬁciency than hardwood species such as sugar
maple, and so slower rates of nitrogen cycling could give hemlock seedlings a

competitive advantage over sugar maple seedlings.

2. Lower pH and soil moisture than hardwood stands (Tubbs 1995, F inzi et al. 1998b)
Climate differences under hemlock canopies as opposed to hardwood canopies
include a lower soil pH, drier soil, and cooler temperatures (Finzi et al. 1998b, Steams
1951, Tubbs 1995). Hemlocks tend to establish on very mesic soils (Godman and
Lancaster 1990, Tubbs 1995), yet the soil under hemlock canopies is drier than that under
nearby hardwoods. Daubenmire (1930) compared evaporation from the soil surface
beneath a hemlock and a beech-maple canopy in Indiana. He found no signiﬁcant
differences in evaporation under the two canopies, but did ﬁnd that the top layer of soil
was drier under hemlock than under beech-maple because: 1) the dense hemlock canopy
blocked some precipitation from reaching the forest ﬂoor, and 2) hemlocks took up all of
the available moisture from the soil surface, actually bringing soil moisture conditions
below the wilting point for hemlock at several points during the summer. However,
Pregitzer et al. (1983) found that soil moisture varied widely under hemlock stands across

a slope in Michigan. Hemlocks were found on both a relatively dry upland site, and also

at the most mesic site at the bottom of the slope. The differences in topography in this
study were more important than any species-caused differences in forest floor moisture.
In the ﬁeld sites used in this study, I have hemlock forest floor conditions that range from
very wet (perched water table with almost constant standing water) to very dry (virtually
no understory plants at all, hemlock litter dry to the touch).

Unlike soil moisture and perhaps rates of N cycling, pH under hemlock
canopies does appear to be universally low. Rogers (1980) surveyed hemlock stands
on a transect from Wisconsin to Nova Scotia, and found that the average soil pH
ranged from roughly 4.0 to 4.75 across this gradient (the lowest pH was in east central
Ontario, and the highest on the shore of Lake Superior). No hemlock stand surveyed
had a pH of higher than 5.45, and the minimum pH found was 3.25. Likewise,
Daubenmire (1930) found pH ranges from 3.6 to 4.7 in his Indiana hemlock stands.
This range did not overlap at all with beech-maple stands nearby, where the soil pH
ranged from 5.3 to 7. Finzi et al. (l998b) determined in a study of pH beneath
individual trees that this lowered pH was due to the effects of hemlock litter. The soil
beneath hemlocks was signiﬁcantly more acidic than that under sugar maple, white
ash, and red maple, and slightly more acidic than beech and red oak soils. Soil pHs
below 7.5 cm depth, however, did not signiﬁcantly differ, indicating that the pH
change was not an inherent difference in soil microsites, but rather caused by the
vegetation. pH of hemlock logs averages 4.5 whether logs are found beneath
hemlocks or beneath sugar maples (see Chapter 3), so hemlock wood, in addition to
litter, may help to maintain this low soil pH. It is important to note that all three
dominant seedling species in hemlock-hardwood forests (hemlock, yellow birch, and

sugar maple) can germinate at pHs as low as 3.0 (Raynal et al. 1982), so this

mechanism could affect seedling survival but most likely not germination. It is
unclear whether or not lowered pH negatively impacts sugar maple seedling survival

more than hemlock survival.

3. Dense shade cast by hemlock (Bourdeau and Laverick 1958)

In addition to altering the soil chemistry and climate beneath them, hemlock trees
form an extremely dense canopy that causes deep shade in the understory and prevents
even the most shade tolerant seedlings from surviving past the ﬁrst growing season
(Catovsky and Bazzaz 2000). Though species such as sugar maple readily germinate on
the forest ﬂoor of hemlock stands, the dense shade (year-round) and low nutrient
availability result in the death of virtually all seedlings before they can reach the canopy
(Rogers 1978 and Ferrari 1993 in Davis et al. 1993, Frelich and Graumlich 1994). When
disturbance increases the amount of light beneath a hemlock stand, however, reciprocal
replacement is the rule rather than the exception. In mixed stands, sugar maples and
other hardwoods are generally more likely to replace fallen hemlock trees than are
hemlocks (Barden I979, Frelich and Graumlich 1994). Sugar maples take over about one
third of the small and medium sized gaps that form in old-growth hemlock stands (Dahir
1994). This suggests that dense shade is a critical component of stable hemlock patches,

preventing the ascension of sugar maple to the canopy.

4. Clumped hemlock and birch seed dispersal near adult trees (Houle and Payette I990,
Rooney and Waller I998)
Hemlock and especially birch seeds travel far enough to reach most parts of any

stand with adult seed-producing trees (Mchen and Curran 2004), but seeds do not travel

into adjacent stands. Houle and Payette (1990) found that although seeds of birch can
travel considerable distances (mainly by blowing over the snow since these seeds are
winter-dispersed), distribution of seeds is clumped near adult birch trees. Rooney and
Waller (1998), Catovsky and Bazzaz (2000), and Mchen and Curran (2004) did not ﬁnd
evidence of clumping of hemlock seeds near adults or a positive correlation between
hemlock basal area and seed abundance, but do speculate that seed rain would be an
important limit of seedling distribution on the scale of individual hemlock and sugar
maple patches (which range in size from 3 to 30 ha, Davis et al. 1998). Also, with the
characteristically low viability (less than 25% for hemlock, Godman and Lancaster 1990)
and high ﬁrst-year mortality of hemlock and birch (Linteau 1948, Chapter 2), small
differences in seed rain as distance from parent trees increase may still affect the number
of established seedlings. This would result in the majority of hemlock and birch

seedlings being found near adult hemlock and birch trees.

5. Susceptibility of hemlock and yellow birch seedlings to smothering by hardwood litter
(Koroleff I954, Tubbs I978, Frelich et al. 1993)

Smothering of hemlock seedlings by litter is commonly assumed to be another
reason that hemlock seedlings do not invade sugar maple patches (and since birch
seedlings are similar in size if not more fragile, this mechanism also works for birch).
Koroleff (1954) suggests that leaves directly smother (block light to) seedlings, while
Frelich et al. (1993) propose a combination of smothering and drought, since leaf litter
dries more quickly than underlying soil. In my study, areas with thick hardwood leaf
litter almost never had established hemlock or birch seedlings, but it is possible that

mortality attributed to smothering or drought damage is actually due to some other

10

property of hardwood litter, or that the mechanism is different for hemlock and for birch
seedlings. For example, Peterson and F acelli (1992) have suggested that hardwood litter
blocks germination cues in birch. Regardless of the mechanism, sugar maple seedlings
are less affected by leaf litter, since their radicles can penetrate leaf litter to reach mineral
soil beneath and even ﬁrst-year seedlings are large and tall enough to avoid being covered

by hardwood litter.

SI 1" . 1 ll ll'l ll' ll'l

The possible mechanisms listed above help to explain why hardwood species are
unable to invade hemlock stands except in gaps, but they only partly explain why
hemlocks do not successfully invade hardwood stands. Hemlock is the eastern United
States’ most shade tolerant conifer species, and arguably our most shade tolerant tree
(Curtis 1959, Dahir 1994). Hemlocks can become established in as little as 5% of full
sunlight, survive in a suppressed state for decades, and respond to release (increased
growth with sudden increases in light availability) until they are at least 240 years old
(Tubbs I978, I995). Hemlock is one of, if not the, only eastern tree species able to
ascend to the canopy without the help of a treefall gap (Frelich and Graumlich 1994).
Why, then, haven't hemlocks shown signs of any meaningful expansion (F relich et al.
1993) into hardwood patches? And why are bitches so closely associated with hemlock
but not sugar maple? I hypothesize that the seedling germination and early establishment
requirements of hemlock and birch limit them to substrates that are more common in
hemlock/birch stands than in hardwood stands, preventing expansion into sugar maple-

dominated stands.

11

Many authors have documented one part of this substrate limitation: the
limitation of hemlock (Nelson 1997, Rooney and Waller I998, Tobin 2001) and to a
lesser extent birch (Steams I951, Coffman I978, Reif 1992) to decaying wood. Because
both hemlock and birch seedlings are drought intolerant and need consistently moist
substrates to establish, decaying wood is an ideal substrate, maintaining high moisture
contents even under drought conditions (Boddy 1983). However, while decaying wood is
more abundant under hemlock than under mixed hemlock-hardwood canopies due to
slower decay rates of conifers (Campbell and Gower 2000), decaying wood is still
available under sugar maple canopies. The substrates to which hemlock and birch
seedlings are limited must be available in hemlock stands but rare or unavailable in sugar
maple stands. The substrate limitation hypothesis, then, can be further deﬁned as the
limitation of established hemlock and birch seedlings to hemlock and/or birch decaying
wood. The ﬁrst step in testing this hypothesis was to measure where on the forest ﬂoor

hemlock and birch seedlings are found, in Chapter 2.

CHAPTER TWO
TSUGA, BE T ULA, AND ACER SEEDLING DISTRIBUTIONS ACROSS FOREST
FLOOR SUBSTRATES IN UPPER MICHIGAN, USA I: PATTERNS OF
SEEDLING DISTRIBUTION AND SURVIVAL

Abstract: We measured the abundance, survival, and age class distribution of tree
seedlings of T suga canadensis (eastern hemlock), Betula allegheniensis (yellow birch),
and Acer saccharum (sugar maple) on decaying wood of the same species and on soil at
four ﬁeld sites in Upper Michigan, USA to determine whether species of decaying wood
differ in their ability to support seedlings that depend on wood as an establishment
substrate. We also wondered, if wood species differ, what are the implications for forest
structure in the primary hemlock-hardwood forest of Michigan? We hypothesized that
hemlock wood supports higher hemlock and birch seedling abundances and survival rates
than those on maple wood and soil. Birch and hemlock seedlings were highly dependent
on decaying wood for seedling establishment whereas sugar maple showed the opposite
pattern. Independent of seed rain, light, and log size, hemlock logs generally supported
the highest abundances of ﬁrst-year and established (>l yr old) seedlings of birch and
hemlock. Averaged over three sites, densities (seedlings/m2) of established seedlings on
hemlock:birch:sugar maple woodzsoil were 0.42:0.21:0.08:0.01 for hemlock seedlings,
0.60:0.15:0.10:0.01 for birch seedlings and 0.09:0.03:0.04:0.98 for maple seedlings.
Long-term seedling survival was also greater on hemlock wood, such that hemlock wood
supported seedlings as old as 13 years while on maple wood seedlings > 3 years old were
very rare. Despite this general pattern among sites, site differences in seedling density
were highly signiﬁcant and may be related to variation in water availability. We conclude

that hemlock wood is the preferred substrate for hemlock and birch seedlings whereas

13

maple wood and soil are not. We also conclude that the limitation of hemlock and birch
seedlings to hemlock decaying wood combined with the distribution of hemlock wood
help explain I) the close hemlock-birch association, 2) the maintenance of distinct,
temporally stable hemlock and hardwood patches, and 3) the decline of hemlock in
managed forests where mature hemlock trees are removed and hemlock wood is
consequently scarce.
Introduction

Numerous studies have shown the importance of decaying wood as an
establishment substrate for the seedlings of some tree species (Gray and Spies I997,
Comett et al. 2001, Lee and Sturgess 2001 , McGee 2001, Mori et al. 2004, Casperson and
Saprunoff 2005). The association of these species with decaying wood, which is rare
under all circumstances and varies in abundance spatially and with time since disturbance
or stand establishment, may be a critical mechanism structuring forests. For example,
diminished tree mortality and consequently lower decaying wood inputs following
selective logging (Newberry 2001, Hura and Crow 2004) can affect future canopy
composition by decreasing seedling recruitment for those species that favor decaying
wood for establishment (Casperson and Saprunoff 2005). Despite the recognition of the
importance of decaying wood in forest dynamics, wood is usually treated as a single
category or categorized by decay class (Christy and Mack 1984, Takahashi et al. 2000,
Mori et al. 2004), but almost never by species (Comett et al. 2001). If there is species-
speciﬁc variation in the suitability of decaying wood as a seedling establishment
substrate, what are the implications for forest dynamics?

Comett et al. (2001) showed that in northern Minnesota T huja occidentalis wood

was twice as likely as Betula papyrifera wood to be colonized by new T huja germinants,

14

but given that only new germinant populations were reported it is unknown if these
patterns translate into longer term patterns for established seedlings. In the primary
hemlock-hardwood forests of Upper Michigan we observed that decaying logs of eastern
hemlock (Tsuga canadensis (L.) Carr.) seemed to support more tree seedlings than sugar
maple (Acer saccharum Marsh.) logs, that seedlings were predominantly yellow birch
(Betula alleghem'ensis Britton) and hemlock seedlings, and that these seedlings appeared
to be rare on the forest floor. Furthermore, sugar maple seedlings appeared to be rare on
logs. The hemlock-northem hardwood system may be ideal for testing for species
differences in suitability of decaying wood because canopy tree distribution in these
forests may depend on seedling establishment substrate preferences and the distribution
of substrates.

Hemlock-dominated forest intermixed with sugar maple-dominated forests
covered large areas of Upper Michigan and northern Wisconsin, USA until the late
18003. These forests have declined markedly in area since harvesting began in the late
18005. In Upper Michigan, more than 99% of mature hemlock-hardwood forest has been
converted to other cover types (N055 and Peters 1995), and by 1993 hemlock occupied
only 0.5% of the landscape (Mladenoff and Steams 1993). Yellow birch, which is
strongly spatially associated with hemlock (Frelich et al. 1993, Kotar et al. 1999), is also
in decline (Woods 2000, Schwartz et al. 2005). The strong hemlock-birch association is
puzzling, as hemlock is among the most shade tolerant trees in north America and birch is
mid-tolerant. However, both birch and hemlock are small-seeded, drought-intolerant
species (Eckstein 1980, Erdmann I990, Godman and Lancaster 1990), and established
seedlings of both have been found to be associated with decaying wood in primary forests

(Reif 1992, Corinth 1995). Relic primary forests stands show a pronounced patch

15

structure with hemlock-dominated patches (3- 30 ha, Davis et al. 1998) with high
admixtures of yellow birch adjacent to sugar maple-dominated patches (Frelich et al.
1993). Pollen core studies indicate that these patch boundaries have changed little since
their formation about 3,200 years ago (Davis et al. 1993), and several self-reinforcing
mechanisms have been proposed to explain the long-term maintenance of patch structure.
These mechanisms include the diminished light, water, and perhaps nitrogen levels
beneath hemlocks which allow hemlock but not sugar maple seedlings to survive (F inzi et
al. l998a, Campbell and Gower 2000, Catovsky and Bazzaz 2000), and the abundance of
hardwood leaf litter in sugar maple stands, which smothers young, small hemlocks but
not the larger sugar maple germinants and seedlings (Koroleff 1954, Tubbs I978, Frelich
et al. 1993). While proposed mechanisms offer a partial explanation for the maintenance
of patch structure, they do not explain why hemlock patches almost always contain a
large basal area component of yellow birch, or why the presence of decaying wood in
sugar maple-dominated stands does not result in the establishment of hemlock trees
within sugar maple patches.

In this study we ask: are there differences in the suitability of wood species for
seedling establishment, and if so, do these patterns help to explain the close spatial
association of hemlock and birch and the stability of hemlock-dominated and sugar
maple-dominated patches? We hypothesized that hemlock and birch seedlings are
restricted to hemlock and birch wood for seedling establishment. To address this
question, we identiﬁed decaying wood to species and measured seedling abundance,
survival rates, and age distributions across decaying wood and soil at four primary forest

sites in Upper Michigan. Our speciﬁc predictions were:

16

1. Independent of variation in light, seed rain, and log size, hemlock and birch
seedlings are more abundant on hemlock and birch wood than on sugar maple
wood and soil, and sugar maple seedlings are more abundant on soil than on
wood.

2. Survival of hemlock and birch seedlings is greatest on hemlock and birch wood.

3. The greater abundance and survival of both hemlock and birch seedlings
combined with the predicted greater quantity of decaying hemlock and birch wood
in hemlock-dominated than in sugar maple-dominated stands partially explain the
maintenance of the hemlock-hardwood patch structure and the hemlock-birch
spatial association.

Materials and Methods

Between 2002 and 2005, we studied four primary hemlock-hardwood forests in
Upper Michigan (Table 2.1). Three of these sites, the Porcupine Mountains Wilderness
State Park, Sylvania Wilderness Area (Ottawa National Forest), and the Huron Mountain
Club Reserve (private ownership), are characterized by a patchy distribution of forest
types, with areas of hemlock/birch bordering hardwood forests dominated by sugar maple
with basswood (T ilia americana) (Pastor and Broschart 1990). Stands contain minor
components of balsam ﬁr (Abies balsamea), red maple (Acer rubrum), striped maple
(Acer pensylvanicum), northern white cedar (Thuja occidentalis), and Ostrya virginiana.
Field plots are located in areas that have been selectively logged only for white pine
(Pinus strobus) in the late 18008 (Woods 1981, Simpson et al. 1990). The fourth site, in
the state-owned Sand River area near Skandia, MI, is a patch of hemlock-dominated
forest surrounded by areas that were selectively harvested for white pine, sugar maple,

and birch. Unlike the other three sites, Sand River is poorly-drained, has been mainly

l7

cleared of birch seed sources, and has lower deer browse pressure (D. Wilson, MDNR,
pers. comm.) This site is treated separately in the results section.

At each site, paired ﬁeld plots (0.1 ha) were placed on either side of distinct
hemlock/hardwood borders to allow comparison between hemlock (>55% basal area
hemlock) and hardwood (10 - 35% basal area hemlock) stands which could differ in seed
availability, environmental factors, and surface area and species composition of logs.
There were no obvious topographic differences between the members of each pair, and
paired plots were separated by 40 to 110 m. Sixteen ﬁeld plots were located: seven
within the Huron Mountain Club Reserve (one hemlock plot had no suitable mixed plot
nearby), four each within Sylvania and the Porcupine Mountains, and one in Sand River
(where no uncut areas of hardwoods existed).

Within each ﬁeld plot, every log, stump, or downed branch > 10 cm in diameter
was counted, and dimensions, decay stage, and wood species were recorded for each
wood piece. Wood pieces are collectively referred to as logs for simplicity, while those
representing main stems are referred to as boles. Logs in decay stage V (the most highly
decayed stage, where wood is almost fully incorporated into surrounding soil; Pyle and
Brown 1998, Graham and Cromack 1982) were not counted. Wood species of the 413
logs present in ﬁeld plots by 2004 was determined by microscopic examination of thin
slices of wood (40x to 200x, microscopes at the USDA Forest Products Laboratory,
Madison, WI, see Chapter 4 for details), and 47 logs of species other than hemlock, birch,
and sugar maple were excluded from analysis. Note that identiﬁcation of wood in the
ﬁeld is not reliable for well-decayed logs; comparison of microscopic features on several

wood cross-sections was required.

18

All tree seedlings (stems S 30 cm in height in order to be comparable to Rooney
and Waller 1998) growing on a log were counted, identiﬁed to species, and their height,
diameter, age class (ﬁrst-year seedlings or established), and substrate (bark, litter, moss,
etc.) were recorded. For each log, an identical seedling survey was conducted on an equal
surface area of soil 1 m away from and in the same orientation as the log. Soil plots were
randomly placed on either side of a log. Large seedlings/saplings (> 30 cm tall but not in
canopy) were noted when they occurred on either logs or soil, but individuals in this
larger size class were rare or absent in most ﬁeld plots and are not reported. We
assumed, in order to calculate soil surface area, that stumps were ﬂat and holes and
branches were half-cylinders. Given that most boles were intermediate between ﬂat
surfaces and half-cylinders, this resulted in a conservative measure of seedling density for
boles. In 2002 and 2003, sugar maple seedlings were often so abundant on soil that only
a subsample of seedlings (every 10th or every 20th seedling) was measured. In 2004 each
sugar maple seedling on soil was placed into an age class but other characteristics were
not measured.

In late August 2002, after the high-mortality period of June and July when many
ﬁrst-year seedlings die of drought, 190 seedlings growing on hemlock (n = 18 logs), birch
(17) and sugar maple (13) logs were marked with plastic toothpicks. Logs were chosen
via stratiﬁed (by log species) random sampling from a list of logs of each species in each
site. On each log, we started at one end and marked several seedlings of each available
species, up to 10 seedlings in total without regard to seedling age. Seedling survival was
recorded in summer 2003, 2004, and 2005.

We also collected established (older than one year) hemlock and birch seedlings

from 88 logs in 2002 (see Chapter 3 for more detailed methods). For seedlings Z 2 mm in

19

diameter immediately above the root collar, we counted growth rings using a 50x
dissecting microscope. Second- and third-year seedlings < 2 mm in diameter were aged
in the ﬁeld by counting bud scars (ﬁeld aging of hemlock is highly imprecise above three
years of age). Because sugar maple logs only rarely support seedlings, in 2004 we
collected and aged seedlings from 16 additional sugar maple logs with established
hemlock or birch seedlings to ensure that our estimates of maximum seedling age on
sugar maple logs were accurate. Sugar maple logs have therefore been considerably
oversampled and were sampled over a considerably larger survey area than the other
wood species.

Canopy photographs were taken approximately 30 cm directly above each log in
ﬁeld plots and all logs from which seedlings were sampled using a digital camera (Nikon
Coolpix 995, set to grayscale) with a ﬁsheye lens. All photos were analyzed using GLA
software (Version 2.0, 1999, Institute of Ecosystem Studies) by a single technician.
SideLook software (v. 1.1.0], 2005, M. Nobis, www.appleco.ch), which was developed
in 2004, was used to automatically threshold a subset of 22 canopy photos for
comparison. The gap light indices (canopy openness) of automatically and manually
thresholded canopy photos were similar (matched pairs mean difference = 0.62%, s.e.
0.16, n = 22 pairs, p = 0.001; Pearson's correlation = 0.75).

Seed rain was measured near 15 randomly selected logs within each of 12 plots (3
plots from which the surrounding area had been logged for hardwoods were excluded).
Seed traps were constructed from 22 cm-diameter plastic pots (366 cm2 surface area),
lined with plastic weed cloth and with a piece of 1/2 inch wire mesh covering each trap
about 1 inch below the top. Plastic canvas (6 squares per inch) was used in the bottom of

traps to allow drainage but prevent entry of seed predators. Seed traps were placed

20

alongside the midpoint of each log and seeds were collected from August 2003 to late
May/early June 2004, with leaves cleared from trap surfaces in October 2003 at ﬁrst
snowfall. In 2004, contents of the remaining 137 undisturbed traps (of 180 placed) were
dried at 65°C, and seeds were counted and a subsample was cut open to determine
percentage of seed ﬁlled. Birch seeds, the most abundant type, were counted up to 200
seeds, with the remaining number of seeds estimated by weight. Note that seeds were
collected in the year following a mast seed year for birch and sugar maple.

Our data are a combination of seedling density and count data, with individual
logs treated as the experimental unit (except for analyses of survival and age). Factors
examined include site, log species, stand type, and wood area. The large number of zero
values obtained (as many as 97% of logs within a study site lacked seedlings), while
biologically meaningful, made the data difﬁcult to normalize by transformation. We used
negative binomial regression (SAS 9.1.3, proc GLM, dist=nb, SAS Institute, Cary, NC) to
analyze count data, but these data are presented as density (seedling count divided by log
surface area) values in ﬁgures and tables. Our Sand River site contained only a single
plot, so we were unable to perform statistics on this site. In addition, in certain years
sample sizes of seedlings at the Huron Mountain ﬁeld were too low to allow successful
convergence of the maximum likelihood algorithm used by SAS when more than one
variable was included in the regression model. When reliable statistical results could be
obtained for this site, they are reported, otherwise data are reported without test statistics.

Results
Decaying wood distribution
Logs covered a larger percentage of the forest ﬂoor in hemlock (5.3% i 0.71 s.e.,)

than in mixed stands (4.3% d: 0.52, paired t-test of percentage of forest floor p = 0.048).

21

There was no difference in the average decay stage of each wood piece in the two stand
types (mean = 2.9 and median = 3.0 for both, stages follow Graham and Cromack 1982).
Hemlock logs comprised 35% of the log surface area in hemlock stands, birch 18%, and
sugar maple 29%, with 18% of area made up by minor species (versus 27%, 18%, 48%,
and 8% in mixed stands). Hemlock plus birch logs covered on average 2.8% (27.5 i
7.3m2) of the forest ﬂoor in hemlock stands versus 1.9% (18.6 :t 5.2 m2) in mixed stands.
At the Porcupine Mountains, but not at Sylvania or the Huron Mountains, hemlock logs
had a greater average diameter than did sugar maple logs (one-way ANOVA, p = 0.015,
hemlock = 32.1 cm, sugar maple = 22.7 cm). Surface area distribution of individual
wood pieces did not vary signiﬁcantly with species at any site (Figure 2.1).
General seedling distributions

Hemlock and birch seedling density were greater in hemlock-dominated than in
sugar maple-dominated plots, but because this difference was due to a single mixed plot
in the Porcupine Mountains, we pooled stand types for analyses of seedling abundance.
Established hemlock and birch seedlings were several orders of magnitude more abundant
on logs than on soils across ﬁeld sites (Table 2.2). Virtually all soil plots (91%, 274
plots) lacked hemlock and birch seedlings in all three years, while 35%(111 logs) of logs
supported at least one hemlock or birch seedling in at least one of the three years
measured. Sugar maple seedling densities showed a pattern opposite that of hemlock and
birch seedlings, with signiﬁcantly greater seedling densities on soil than on logs. 42% of
soil plots in 2003 and 75% of soil plots in 2004 (after a mast year in 2003) had at least
one sugar maple seedling. In contrast, only 3% of logs in 2003 although 33% of logs in

2004 supported at least one sugar maple seedling.

22

At all sites, for ﬁrst-year and established hemlock and birch seedlings, the number
of seedlings on hemlock wood was greater than that on sugar maple wood, with the
exception of established birch seedlings at the Huron Mountain Club (Table 2.2, Figure
2.2). Averaged over all sites except Sand River (see explanation below), densities (m2) of
established seedlings on hemlock:birchzmaplezsoil substrates were 0.42:0.21:0.08:0.01 for
hemlock seedlings, 0.60:0.15:0.10:0.01 for birch seedlings and 0.09:0.03:0.04:0.98 for
maple seedlings. Thus hemlock and birch seedlings were at least ﬁve times more
abundant on hemlock wood than on sugar maple wood, and at least 42 times more
abundant than on soil. In addition, although hemlock logs made up 33% of the number of
logs, 34% of the logs in decay stages II through IV (classes most suitable for seedling
establishment), and 36% of total wood surface area, they supported disproportionately
large percentages of the total hemlock (50% to 67%) and birch (40% to 75%) seedlings
on logs, depending on year. In contrast, birch (21% of surface area) and sugar maple
(43%) logs supported proportions of seedlings equal to or below their proportion of log
area.

The difference between the number of established seedlings on hemlock logs and
the number of seedlings on sugar maple logs was signiﬁcant at the Porcupine Mountains
(seedling counts: log species effect test chi-square p < 0.045 in each year, for each
seedling species; seedling presence averaged over several years: nominal logistic ﬁt,
Wald test p = 0.0025, Table 2.2), and borderline signiﬁcant at Sylvania Wilderness (log
species effect chi-square p > 0.071 in each year, for each seedling species).

Seedling distributions by site
Collectively, our results suggest that general patterns in abundance of seedlings on

log species are similar across sites, although absolute seedling densities are affected by

23

site factors including substrate moisture. There were significant site effects on seedling
counts in all years (chi-square p < 0.0001 in each year for each seedling species). At the
Huron Mountain Club, our driest site (based on data from a nearby weather station, Table
2.1) there was an almost complete lack of established hemlock and birch seedlings
despite abundant ﬁrst-year hemlock seedlings, with established seedlings on between 3
and 10% of logs, depending on the year and species compared. At Sylvania and the
Porcupine Mountains, seedling numbers were higher than at the Huron Mountain Club
and greater proportions of logs (9 to 37%) supported seedlings. Sand River was our only
poorly drained site, and as such was also wettest, with greater soil moisture (79.1%, July
2003) than other sites (average 55.5%, see Chapter 3 for methods). Due to stand size and
composition, Sand River was represented by only a single hemlock ﬁeld plot, making
statistical comparisons impossible, but there were patterns at this site that warrant
reporting. Established hemlock seedling densities were by far the highest at this site, with
densities on hemlock logs 7.82/m2 and soil 1.03/m2 compared to less than 0.40/m2 and
0.020/m2 at other sites on logs and soil, respectively. Compared to hemlock seedlings
birch seedlings were relatively rare (e.g. on logs birch densities ranged from 1.16/m2 to
2.00/m2 depending on year), probably due to the removal of most nearby seed trees, but
birch densities were still high compared to other ﬁeld sites. High hemlock densities on
logs and especially on soils at this site are likely due to the presence of continually wet
soil, even in late summer, and illustrate that although logs are still important
establishment sites in wet areas, seedling are less restricted to logs at wet sites than at dry
sites. If we included Sand River as a site there were signiﬁcant site effects on seedling
counts in all years (chi-square p < 0.0001 for each seedling species in each year, n = 4).

When Sand River was removed, the effect of site was still signiﬁcant in some cases (chi-

24

square p = 0.0001 for birch seedlings in 2004, p = 0.042 for hemlocks in 2003) but not in
all (p = 0.620 for birches in 2002).

By directly comparing the number of seedlings on each log to the seedlings on its
paired equal-area, nearby soil plot, we could remove site and other environmental effects
and isolate the effect of log versus soil substrate on seedling abundance. Paired t-tests
demonstrated the same patterns seen in the averaged log and soil seedling densities
reported above. With three sites pooled, in 2002 the number of hemlock and birch
seedlings on logs exceeded the number of seedlings on the same area of corresponding
soil by 0.62 (hemlock) and 0.71 seedlings (birch, n = 302, both species paired t-test p <
0.01). Logs also had 9.29 fewer sugar maple seedlings than nearby soils in (n = 302, p <
0.0001). In 2004, differences between logs and soils were even larger due to a birch and
sugar maple mast year in 2003; logs had 3.52 more birch seedlings than soil (n = 311, p =
0.005), and 30.45 fewer sugar maple seedlings (n = 311, p < 0.0001). To separate
seedling presence/absence from abundance, we repeated these same paired t-tests using
only pairs where at least one seedling was present (on the log, soil, or both). We obtained
the same results in all cases (with p-values S 0.003).

Environmental factors potentially affecting seedling abundance

Light is a critical resource limiting survival in forest understories, but light levels
did not vary consistently among log species (Table 3.2, Chapter 3). Log surface area also
did not vary among log species, and in most size classes hemlock logs supported the
greatest number of hemlock and birch seedlings (Figure 2.3). Variation in seed rain is
another factor that if confounded with log species could result in a false conclusion that
log species are driving seedling abundance patterns. Although seed rain of hemlock and

birch was greater in hemlock-dominated plots than in mixed plots, seed rain did not vary

25

signiﬁcantly among log species, with the exception that more hemlock seeds fell onto
birch logs than sugar maple logs in mixed stands at the Porcupine Mountains (Table 3.2,
Chapter 3). Because seed rain was only sampled near a subset of logs, the addition of
seed rain to the tests of the effect of wood species on seedling abundance reported above
halved our sample size. This prevented us from testing seed rain in hemlock versus
mixed stands. With stand types pooled, wood species effects on seedling abundance were
no longer signiﬁcant with seed rain in the model for hemlock seedlings at the Porcupine
Mountains, and the seed rain effect was signiﬁcant (chi-square p < 0.05). However, when
we tested the strength of the wood species effect alone in hemlock versus mixed plots
(i.e. partially controlling for seed rain and wood abundance differences by controlling for
overstory/seed source composition), the effect of wood species on seedling abundance
was no less signiﬁcant in mixed plots (where the average p value for wood species effect
was 0.09, with two out of three cases < 0.05) than in hemlock plots (average p = 0.15,
two out of three cases < 0.05) . While seed rain almost certainly affected ﬁrst-year
seedling abundance, wood species was an important factor determining seedling
abundance even in mixed plots at the Porcupine Mountains, the only site with signiﬁcant
seed rain differences across wood species.
Seedling survival

Mortality across three years was high, with only 50 of the original 190 tagged
seedlings surviving to 2005. Hemlock seedlings had the highest survival rates when
growing on hemlock logs and the lowest rates when growing on sugar maple logs (Table
2.3). Birch seedling survival was signiﬁcantly greater on hemlock logs than on sugar
maple logs from 2002 to 2003, even when ﬁrst-year seedlings (which had higher

mortality rates than established seedlings) were excluded. After 2003, though, birch

26

survival did not vary across wood species. We were unable to ﬁnd an adequate sample
size of sugar maple seedlings growing on logs, but did tag nine sugar maple seedlings, all
but one of which were found growing on hemlock logs. Based on differences in sugar
maple abundance in the year following a mast year, early sugar maple survival on logs
appears to be low. After a mast seed year in 2003, sugar maple seedling density on logs
at the Porcupine Mountains averaged 1.34/m2i 0.31 (versus 0.045/m2i 0.026 in 2003,
paired t-test p < 0.0001), but only one third (median = 33%, n = 95; 17% of logs still
supported sugar maples) of the established seedlings in 2004 survived the winter to be
counted in a seedling census in 2005.

The age distribution of hemlock and birch seedlings on logs (Figure 2.4) is
consistent with our 2002-2005 survival data. Both hemlocks and birches older than three
years were rare on sugar maple logs, despite oversampling of this log species in 2004.
Hemlock seedlings were as old as 13 years on hemlock and 12 on birch logs. Birch
seedlings had a similar distribution. Sugar maple seedlings on all log species were very
rare (Figure 2.4).

Discussion
Seedling distribution and survival

At all ﬁeld sites, hemlock and birch seedlings were more abundant on logs than
on soils whereas sugar maples were more abundant on soils than on logs. Hemlock logs
support the highest seedling densities of both birch and hemlock at each ﬁeld site (with
the exception of established birches at the Huron Mountain Club). These results support
our hypothesis that differences in the suitability of wood operate on the species level,
with hemlock and birch logs more favorable for seedling establishment than sugar maple

logs. In addition to numerous hemlock, birch, and sugar maple logs, our ﬁeld plots also

27

contained one to eleven T huja occidentalis (cedar), Abies balsamea (balsam ﬁr), Ostrya
virginiana, Tilia americana (basswood), Acer rubrum (red maple), Pinus strobus (white
pine), and Quercus sp. (oak) logs. Sample sizes of these minor species are too small to
allow statistical comparisons, but in general hemlock and possibly cedar logs serve as
establishment sites for large numbers of hemlock and birch seedlings, while balsam ﬁr,
another conifer, does not appear to support high seedling densities. Among the
hardwoods, only birch commonly supports seedlings.

The restriction of hemlock and birch to logs is likely due, in part, to higher and
more constant water content in logs and their ability to shed leaf litter (Christy and Mack
1984, Comett et al. 2000), but the scarcity of sugar maple on logs (also noted by Tubbs
1995) is puzzling. Field observations suggest that seedlings germinating on logs become
chlorotic mossibly N deﬁcient, see Chapter 3) and die, often within the ﬁrst growing
season. The endomycorrhizal sugar maple seedlings (Klironomos 1995) are unable to
join existing ectomycorrhizal (ECM) networks of hemlock and birch (Booth 2004) and to
become colonized by ECM fungi already in wood (Kropp 1982b), which may result in the
observed nutrient deﬁciency. Logs are one of the few sites on which young hemlock and
birch can escape competition from the much larger and initially taller sugar maples,
which carpet the soil around logs in mast years, even in hemlock stands.

Survival of tagged birch seedlings from 2002 to 2005 was uniformly low across
log types, but hemlock survival was highest on hemlock logs overall and at each site
except Sand River. The differences in early survival rate of hemlock but not of birch are
unexpected since hemlock grows slowly in low light and can survive in the understory for
well over 100 years (Dahir 1994). Our age distribution results are generally consistent

with seedling survival results, but indicate that the oldest birch seedlings are found on

28

hemlock logs. Long—term survival of birch seedlings, then, is highest on hemlock logs
even though short-term survival (2002 to 2005) of young seedlings does not differ across
log species. The lack of hemlock and birch seedlings older than four years on sugar
maple logs is particularly striking. Despite oversampling, we found only a single 6-year-
old birch and a single 6-year-old hemlock seedling on sugar maple logs.

Age distribution and seedling survival rates show the continued narrowing of the
niche for successful hemlock and birch seedling establishment. First-year seedlings can
germinate on substrates where established seedlings are almost never found, but quickly
die, often due to drought (Linteau 1948, Friesner and Potzger 1932 in Friesner and
Potzger I944). First-year seedlings present in May and June were often dead by late
August when we returned to collect wood samples. For the next two years, seedlings may
survive on sugar maple logs. By the end of the third growing season, hemlock and birch
seedlings are found almost exclusively on hemlock and birch logs. Under low light
conditions, they will remain there for decades before ascension to the canopy is a
possibility (Tubbs 1978, Dahir 1994). One important piece of data that is missing from
this study is the distribution of saplings, which would conﬁrm the restriction of seedling
regeneration to hemlock and birch logs. Unfortunately, we found very few hemlock and
birch seedlings taller than 30 cm. At Sand River, the only site at which we found
hemlock saplings, all saplings were either on soil or on conifer (hemlock, balsam ﬁr,
white pine, or spruce) logs. Birch saplings were found on both hemlock and birch logs at
the Porcupine Mountains and Sand River sites.

Possible mechanisms explaining these species-level differences in seedling
abundance are beyond the scope of this paper, and are addressed in Chapter 3. However,

we have shown that light levels and seed rain rarely vary across wood species and are

29

therefore not likely to be the cause of variation in seedling abundance. Seed rain was
abundant and varied widely across and within our ﬁeld sites, only varied with wood
species in a single case whether stand types were pooled (Chapter 3) or as analyzed
separately here, and the effect of wood species was important even in the single site and
stand type at which seed rain differences across wood species were found. Given the
differences in short-term survival of seedlings across wood species, it seems unlikely that
the effects of light and seed rain on ﬁrst-year seedling abundance (Chapter 3) were
important in determining established seedling abundance beyond the second or third year.
Ecological and management implications

Despite variation across ﬁeld sites, collectively our results indicate that the
species-level differences in seedling density and survival on decaying wood may help
maintain the boundaries of adjacent hemlock and sugar maple-dominated patches (Figure
2.5). The greater abundance of hemlock wood in hemlock stands (which is not
surprising) may maintain patch boundaries, since hemlock wood, like hemlock litter,
favors hemlock seedlings over sugar maple. Even in hemlock stands, hemlock wood is
rare (covers < 3%) and likely limits seedling establishment. We have shown that
hemlock and birch seedlings are both more abundant and attain a greater maximum age
on hemlock and birch logs than on sugar maple logs. There is, then, a larger and older
seedling bank in areas that have numerous hemlock and birch logs (hemlock stands) than
in those with fewer logs of these species (mixed hemlock-hardwood stands). Recruits
from the sapling bank are quite possibly the only hemlocks that have a chance of reaching
the canopy in between catastrophic disturbances; hemlock sapling that ascend to the
canopy in single-treefall gaps can be quite old (average is 149 years (Dahir 1994)). The

differences in the survival rate of hemlocks on different log species (Figure 2.5) suggest

30

that even small changes in the availability of microsites on which hemlock seedlings can
survive their ﬁrst few decades have a direct bearing on the number of these older saplings
in a stand. Our ﬁeld plots contained at least 10% hemlock basal area and contained
hemlock logs, yet hemlock saplings were very rare. In a sugar maple stand without
hemlock logs, saplings would likely be absent, preventing hemlock and birch from
invading until a ﬁre or other disturbance removed hardwood litter and allowed
regeneration from seed.

Our seedling distribution results also suggest an explanation for the close spatial
association of hemlock and birch stems (Figure 2.6). Hemlock is almost always found in
stands containing a large basal area component of birch (Brown and Curtis 1952,
McIntosh 1972, Rogers 1978). In fact, the spatial association between individual
hemlock and birch trees is stronger than that of any other tree species association found in
forests of the Great Lakes region (Kotar et al. 1999, Rogers 1980). The apparent
restriction of both hemlock and birch seedlings greater than four years old to hemlock or
birch logs, and the positive correlation between hemlock and birch seedling densities on
each log (averaged across years, 2004 excluded for birch, r = 0.50, p < 0.0003 at the
Porcupine Mountains) explain this pattern. Seedlings of hemlock and birch often become
established on the same log, and can apparently coexist until it decays, as evidenced by
pairs of stilt-rooted hemlock and birch trees with tangled roots (Figure 2.6). Booth
(2004) has found that hemlock and birch share mycorrhizal networks, offering one
explanation for the ability of the two species to co-occur in such close proximity even
after reaching the canopy. Birch occurs in a number of forest types, but appears to form
this association only with hemlock, further suggesting that it is hemlock wood that

maintains the hemlock-birch association.

31

The focus of this study is on differences in the suitability of three wood species
for seedling establishment, but an equally interesting result is the almost complete lack of
established hemlock seedlings at the Huron Mountain and Sylvania ﬁeld sites. Rooney
and Waller (1998) also noted in their survey of hemlock stands that many stands had no
existing hemlock regeneration, and that the proportion of stands without seedlings varied
by year. Graham (1941) hypothesized that it was not rare, in Upper Michigan, to have
several decades of poor regeneration in which few or no hemlock seedlings become
established, followed by a mast seed year and wet conditions that allowed abundant
regeneration. Our results suggest that at the Huron Mountain site, not only did very few
hemlock seedlings establish between 2002 and 2005, but in all of our ﬁeld plots except
for the Sand River plot, few hemlock seedlings that established in the several decades
before we began our study survived long enough to grow between 30 cm and canopy
height. At the same time, at least (we only surveyed each ﬁeld site once during the
middle of the growing season, making this a conservative count) 44 hemlock seedlings
germinated on logs (655 m2 total area) in 2002, 230 seedlings in 2003, and 173 seedlings
in 2004, so seed or germination limitation alone are not entirely responsible for the lack
of seedling establishment. Instead, there appears to be high mortality of ﬁrst-year
seedlings before and/or during their ﬁrst overwinter period, whether due to frost,
pathogens (although O'Hanlon-Manners and Kotanen 2004 suggest that at least for seeds,
logs provide a refuge from fungal pathogens), lack of sufﬁcient storage reserves to
continue growth the next summer, or browsing by deer in the short period between
snowmelt and hardwood seedling leafout. This high mortality continues over the next
several years, as indicated in Table 2.3, and likely increases again as seedlings reach a

height where they are browsed by deer. Slight differences in the availability of hemlock

32

wood, given this high mortality, can translate into large differences in the number of
hemlock and birch seedlings and saplings.

The restriction of hemlock and birch seedlings to decaying wood, and speciﬁcally
to hemlock wood which covers less than 3% of the forest floor, is one reason that
established seedlings are usually rare or absent in second-growth forests. Tyrrell and
Crow (1994a) estimate that hemlock-hardwood forests take 400 years to reach peak
coverage of decaying wood, and in forests managed by selection silvicultural systems
decaying wood coverage likely never peaks as trees are harvested rather than left to die.
Even in unmanaged old-growth forests, wood of all species pooled covers at most 10% of
the forest ﬂoor in these eastern forests (Corinth 1995, this study), and only an average of
about 5% in our sites. Given the lack of decaying wood in young secondary and selection
harvested forests, management for natural regeneration of hemlock and birch (if planting
of seedlings and saplings is not practical) should include methods to increase the amount
of logs (even scattering of wood pieces as small as 0.029m2 (43x43in) supported
seedlings). Seeding of existing logs may also be helpful. We seeded logs with hemlock,
birch, and sugar maple seeds in 2003, but very few seeds germinated and thus further
trials are necessary. At least in moderately wet forests such as those found in the
Porcupine Mountains, our results indicate that survival and establishment of seedlings
will be better on hemlock or birch logs than on sugar maple logs. Forest management
guidelines in Michigan and Wisconsin include leaving some snags (dead standing trees)
or trees of low economic value standing to become decaying wood in the future and for
wildlife habitat (Martin and Lorimer 1996, Neumann and Peterson 2001), but do not yet

specify which species should be left. Sugar maple logs support few young hemlock and

33

birch seedlings and no older seedlings, and log species must be considered in efforts to
understand and increase hemlock and birch regeneration.
Acknowledgements
Scott Kissman, Marcie Tidd, and Laurie Gilligan assisted with ﬁeld and lab work.
Help in selecting ﬁeld sites was provided by Michelle Holland of the USPS and William
Brondyke and Dean Wilson of the MIDNR. This research was supported ﬁnancially by a
NSF Graduate Research Fellowship and Doctoral Dissertation Improvement Grant to
Laura Marx, by the Huron Mountain Wildlife Foundation, and by the Michigan Botanical

Club Hanes Fund. We thank Ken'y Woods for helpful comments on an earlier draft.

34

Table 2.1. Locations and characteristics of ﬁeld sites. Weather data were obtained from

NOAA COOP weather stations and are the sum of 2003 monthly averages: Marquette

weather station (46°33'N / 87°23'W) for the Huron Mountains, Marquette Wso Airport

(46°32'N / 87°33'W) for Sand River, Ontonagon (46:50N/89: 12W) for the Porcupine

Mountains, and the Lac Vieux Desert, WI (46:07N/89:07W) station for Sylvania.

Ecological sections follow Albert 1995. ATM = Acer-Tsuga-Maianthemum, ATD =

Acer- Tsuga-Dryopteris, and TMC = T suga-Maianthemum-Coptis (Kotar et al. 1988).

 

 

 

 

 

 

 

 

 

 

Site Latitude and Summer Ecological section and Growing
longitude precipitation subsection season (days
(UTM (inches) (May /Habitat type with min.
coordinates) through temp. >=
September) 32F)
Huron 46:52 N 13.02 inches IX.2 Michigamme 209
Mountain 087:51W Highland/ATM and
ATD
Porcupine 46242-8 N 16.53 IX.8 Lake Superior 180
Mountains 089:4l-:58W Lake Plain and IX.6.1
Gogebic-Penokee Iron
Range/ATM and ATD
Sylvania 46:12-3 N 17.72 IX.3.2 Winegar Not
Wilderness 089:14W Moraine/ATM and recorded at
ATD station.
Sand River 46:26N 16.58 VII.3 Dickinson/T MC 178
087:11W

 

35

 

Table 2.2. Density (seedlings/m2 with s.e. in parentheses) of hemlock and birch seedlings

on different forest floor substrates. Established seedlings are older than one year, ﬁrst-

years refer only to seedlings in their ﬁrst growing season. Hemlock densities are

averaged over 2002-2004, while birch and sugar maple densities exclude a mast year in

2003 (2003 is excluded for ﬁrst-year seedlings, 2004 for established seedlings). n = 302

logs and soils in 2002, 308 in 2003, and 312 in 2004.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Site Substrate Hemlock Hemlock Birch Birch density Maple aple
density density density (established) density density
(ﬁrst- (established) (ﬁrst- (ﬁrst- (established)
car) car) ear)
Huron Hemlock 1.08 0.19 (0.08) 0.07 0.13 (0.08) 0.39 0.02 (0.02)
IMountain logs (0.41) L004) (0.28)
Birch 0.47 0.07 (0.04) 0.15 0.01 (0.01) 0.20 0.02 (0.02)
logs (0.17) (0.10) (0.09)
Maple 0.22 0.05 (0.04) 0.12 0.14 (0.14) 0.15 0(0)
logs (0.05) (0.09) (0.06)
IAII logs 0.58 0.11 (0.03) 0.11 0.11 (0.06) 0.22 0.01 (0.01)
(0.15) (0.05) (0.09)
All soil 0.14 0.01 (0.00) 0.04 0 (0) 1.89 2.44 (0.50)
(0.04) (0.02) 0.24)
IIP/prcupine Hemlock 0.50 0.70 (0.21) 5.33 1.22 (0.53) 0.15 0.09 (0.03)
ountains logs (0.18) (2.47) (0.08)
Birch 0.41 0.41 (0.18) 1.88 0.16 (0.09) 0 (0) 0.05 (0.05)
logs (0.17) (0.66)
Maple 0.06 0.02 (0.02) 0.33 0.02 (0.01) 0.01 0 (0)
logs (0.06) (0.24) (0.01)
All logs 0.35 0.43 (0.11) 3.06 0.62 (0.26) 0.07 0.05 (0.02)
(0.10) (1.19) (0.03)
All soil 0.12 0.01 (0.00) 0.19 0.01 (0.00) 0.18 2.63 (0.52)
(0.04) (0.06) (0.04)
Sylvania Hemlock 0.52 0.76 (0.68) 4.89 0.44 (0.35) 0.23 0.17 (0.17)
Wilderness logs (0.45) (3.24) (0.18)
Birch 0.29 0.27 (0.15) 0.45 0.28 (0.26) 0.12 0.03 (0.03)
logs (0.13) (0.16) (0.06)
Maple 0.08 0.36 (0.10) 0.75 0.14 (0.08) 0.37 0.11 (0.06)
logs (0.05) 0.31) (0.23)
All logs 0.22 0.41 (0.14) 1.41 0.24 (0.10) 0.24 0.10 (0.05)
(0.09) (0.62) 0.12)
All soil 0.01 0.02 (0.02) 0 0.01 (0.00) 0.88 4.74 (0.76)
(0.01) (0.02) (0.15)

 

 

 

 

 

 

 

 

 

36

 

Table 2.3. Percent survival of birch and hemlock seedlings on different log species. 2003
represents survival from late summer 2002 to mid summer 2003. 2004 represents
survival to the third, and 2005 survival to the fourth, growing season. For each column
(not row), percent survivals with different letters were signiﬁcantly different (F isher's
exact test, individual contrasts between species, p < 0.05). n = 96 hemlock and 85 birch

seedlings, sugar maple seedlings not shown.

 

Birch survival Hemlock survival
2003 2004 2005 2003 2004 2005
Hemlock 62.7% a 20.9% a 1 1.6% a 90.3% a 67.7% a 55.2% a
wood
Yellow 56.5% a 26.1% a 13.0% a 66.7% b 46.2% a 35.9% ab
birch wood
Sugar 26.3% b 15.8% a 10.5% a 65.4% b 46.2% a 28.0% b
maple
wood

 

 

 

 

 

 

 

 

 

 

 

 

37

 

Hemlock

 

 

 

 

 

 

 

 

 

 

,6 5 4 4

r .A ' Ir‘ ' T 1' 1r 0' ol—1' or 0' or 0' OI_1' Q1

4 6 8 1O 12 14 16 18 20
Birch

.6 5

[7345», [’34) 0' 0. o_g q 01 o. o, o. o' o. o.

4 6 8 1O 1 14 16 18 20
Sugarmaple

 

 

8 1012 14 16 18 20

Wood surface area (m2)

Figure 2.1. Distribution of surface area of logs. All sites are pooled, n = 318 logs total.

38

2.0

 

 

- Hemlock seedlings
1.6 - CE: Birch seedlings

 

 

 

1.4 -

1.2 -

1.0 -

Seedlings/m

0.6 ..

0.4 1

 

 

 

 

 

0.0 -
Hemlock logs Birch logs Sugar maple logs Soil

Substrate

Figure 2.2. Hemlock and birch seedling densities across substrates at the Porcupine
Mountains. Densities are means + 1 s.e. of established (older than one year) seedlings,
averaged across all years for hemlock and excluding the year following a mast year
(2004) for birch. n = 101 logs and 101 soils. Other sites show different densities but

similar patterns across substrates, as indicated in Table 2.2.

39

 

E Hemlock logs
Birch logs

m Sugar maple logs
0.8

 

 

Hemlock seedlings

0.6 «
0.4 - g

0.2 -

   

0.8 -
Yellow birch seedlings

     

0.6

0.4 -

0.2 -

Probability of having at least one seedling

   

  

 

 

 

 

j...

> 4.5

O-O.4 0.4 - 1 1-2 2 - 4.5

Surface area class for decayed wood piece (m2)
Figure 2.3. Probability of seedling presence (in one or more years, 2002-2004) on logs, by
log surface area and species. n = 296 logs present for all three years; Sand River

excluded, minimum of 6 logs of each species in each surface area class.

40

 

a. Hemlock seedlings b. Birch seedlings c. Sugar maple seedlings

. l

nnﬂ Hana H Dﬂﬂnna l HUDD a

Hemlock logs (42)

 

 

 

 

 

 

d—L—L .34.;

 

 

 

    

 

 

 

 

 

 

 

 

 

 

 

s i

m r"

or“ ;

1530 4.

El 2, [l

g 0‘» BBIIEIEIDI [:1 ['1 J21- 1:]

8, 201;;

2 161,.

9

s 12:

E 8‘

a 4. ,.

m 3

aoﬁanﬁﬂmme
2 4 6 81012142 4 6 81012142 4 6 8101214

Age(years)

Figure 2.4. Age distribution of hemlock, yellow birch, and sugar maple seedlings on
different log species. Log sample sizes are listed after each Y axis label. When
comparing across log species, note the different scale for sugar maple logs due to

oversampling of this species. n = 117 hemlock, 83 birch, and 32 sugar maple seedlings.

41

Hemlock logs

    
  
 

   

 

    

 

  

 

 

 

 
 
  
   

 

 

 

 

 

 

   

 

Potential sapling
0‘038/ /\/ numbers
0.095 relatively high for
both hemlock
. and birch.
Filled 23.1.33; 5:313?“ maintaining the
seeds . 0.50 l 5.33, 0.70 I 1.22 ESQIIglck/birch
13 lm2 8011 m2 I\ lm’ ssociation.
22:21:32“, 0009/ 012 10.19 0.083/ 0.01 mm No recru ts
birch on [m2 Q 952 [m2 on 5011
all 0 06 l o 33/ 0 02 | 0 02 giggle; foal/Big?
wbsmtes m2 ' [m2 both hemlock
and birch,
preventing
establishment of
\/\ hemlockin sugar
maple stands
0'0 6 'th no hemlock
. component
Sugar maple
logs

 

  

Figure 2.5. Estimated probabilities of survival of seeds and seedlings on hemlock logs,
sugar maple logs, and soil at the Porcupine Mountains ﬁeld site. Probabilities of survival
for hemlock seeds or seedlings are listed ﬁrst, followed by the probability of birch
survival. All probabilities were generated by dividing the density of seedlings in an age
class by the density of seedlings/seeds in the previous age class, except for probabilities
of survival from ﬁrst-year to establishment on logs. These probabilities are based on

survival of tagged seedlings.

42

 

Figure 2.6. Paired hemlock and birch canopy trees. The birch tree is labeled with a B, the
hemlock tree with an H. Photo was taken at Sylvania Wilderness Area, near Devil's Head

Lake.

43

CHAPTER THREE
TSUGA, BE T ULA, AND ACER SEEDLING DISTRIBUTIONS ACROSS FOREST
FLOOR SUBSTRATES IN UPPER MICHIGAN, USA 11: MECHANISMS
UNDERLYING SEEDLING DISTRIBUTION
Abstract: Eastern hemlock (T suga canadensis) and yellow birch (Betula alleghenl'ensis)
seedlings in primary Upper Michigan hemlock-hardwood forests are limited to decaying
wood for establishment. In an earlier study, we demonstrated that seedling densities and
survival are greater on hemlock wood than on birch or sugar maple wood. Here, using a
natural experiment at three ﬁeld sites and a greenhouse experiment, we quantiﬁed
intrinsic wood chemical, physical and decay characteristics among three species of
decayed wood and related them to seedling abundance, survival and growth. Water and
light availability did not vary among decayed wood species. Instead, higher hemlock and
birch seedling survival on hemlock than sugar maple wood may have been associated
with the lower pH, more balanced nitrogen and phosphorus supply, greater likelihood of
seedling mycorrhizal infection, greater moss coverage, tendency to decay by brown-rot
fungi, and longer residence time in decay classes favorable for seedling establishment of
hemlock wood. These endogenous factors acted independently of exogenous factors such
as light availability and stand type, and suggest that other species of wood with similar
characteristics may be important for seedling establishment in hemlock-hardwood forests
and in other systems.
Introduction
In an earlier study we demonstrated that the abundance and survival of eastern
hemlock (Tsuga canadensis (L.) Carr.) and yellow birch (Betula allegheniensis Britton)

seedlings in primary hemlock-northem hardwood forests differ among establishment

44

substrates in the following order; hemlock wood > birch wood > sugar maple wood and
soil (Chapter 2). We also found that these patterns were independent of the exogenous
factors seed rain, light availability, and decaying wood piece size, suggesting that
variation in endogenous species speciﬁc-characteristics were responsible.

There is little information on species differences in decaying wood characteristics
of possible relevance to tree seedling survival and growth. However, efforts directed at
explaining the greater abundance of hemlock and birch seedlings on wood, in general,
than on undisturbed soil might offer some insight on what some of these factors might be.
Compared to soil, wood has greater water content (Boddy 1983, Tubbs 1995), less leaf
litter (Harmon 1989, but see Simard et al. 2003), attainment of germination temperatures
earlier in the spring (Godman and Lancaster 1990), fewer fungal pathogens (Zhong and
van der Kamp I999, O'Hanlon-Manners and Kotanen 2004), and reduced densities of
sugar maple (Acer saccharum Marsh.) seedlings (Tubbs 1995, Chapter 2). Of these, the
high water content of wood and its relative lack of hardwood leaf litter (which can kill
seedlings by smothering them, blocking germination cues, or drying out quickly; Koroleff
1954, Peterson and Facelli 1992, Corinth 1995) are two of the best-studied explanations
in hemlock-hardwood forests (Godman and Lancaster 1990, Frelich et al. 1993).
Abundant hemlock and birch regeneration on soil after litter-clearing disturbances such as
ﬁre or mechanical scariﬁcation is indirect evidence for the importance of litter shedding
by logs to seedling establishment (Maissurow 1941, Peterson and Facelli 1992, Strong
1995, Simard et al. 2003). Log species could vary in litter shedding if surface
characteristics differ, and western conifer species do vary in these respects (Harmon
1989). Wood that is effective at shedding leaf litter, however, also tends to shed seeds

(Harmon 1989), so a thin layer of litter might be a more favorable surface for seedling

45

germination than bare wood or bark. Soil beneath hemlock canopies is usually dry and
often below the wilting point of most seedlings and herbs (Daubenmire 1930, but see
Pregitzer et al. 1983). Decaying wood, on the other hand, is usually the wettest substrate
on the forest ﬂoor (Comett et al. 1997) and dries out slowly even under full sun (Boddy
1983), factors which might be particularly important for small-seeded, drought-intolerant
seedling species such as hemlock and yellow birch (F riesner and Potzger 1936, Linteau
I948, Erdmann I990, Godman and Lancaster 1990). In the one study we are aware of
that compared decaying wood water content among species, water content did not differ
between Betula papyrifera and Thuja occidentalis wood in the ﬁeld (Comett et al. 1997).
Wood species are known to differ in wood chemistry (Arthur et al. 1993), decay
rates (Arthur et a1. 1993, Tyrrell and Crow l994a ), and predominant decay fungi (label
and Morrell 1992), all of which are likely interrelated and have implications for factors
associated with tree seedling establishment. Differences in decay rates could affect
nutrient cycling (see below) and the residence time that wood is available as a substrate
for seedling establishment. Shorter residence time would reduce the chance of seeds
encountering a log and becoming established, and also may result in logs decaying out
from underneath slow-growing, light-limited saplings before those saplings are able to
physically support themselves. Seedlings as old as 13 (hemlock) and 9 years (birch) can
be less than 30 cm tall when growing beneath a closed forest canopy, with
correspondingly slow root growth (Marx, unpublished data). In general, hardwood logs
are thought to decay more quickly than conifer logs (Arthur et al. 1993), but comparative
ﬁeld data are scarce (Chapter 2). Furthermore, most seedlings are found on decay stages
11 through [V (middle decay stages, Takahashi et al. 2000), so the residence time of these

decay stages is particularly important.

46

Log species could differ in their abilities to provide mineral nutrients to tree
seedlings. These differences could be associated with interrelated differences in wood
chemistry, decay organisms, decay rates, and associated biota (e.g. mycorrhizae), but
these linkages are largely speculative and unexplored. Of all the macro- and
micronutrients, nitrogen (N) is often regarded as the most limiting nutrient in northern
temperate forests, (Cole and Rapp 1981), though evidence of N-Iimited seedling growth
and/or survival in understories is scarce and equivocal (Walters and Reich 1997,
Catovsky and Bazzaz 2002), perhaps because light is very low and most limiting in these
habitats. F urtherrnore, it is not known if seedlings should be expected to be
predominantly N-limited on decaying wood as nutrient cycling in decaying wood is little
studied. Studies to date suggest that mineral N concentrations and mineralization rates in
(or under) decaying wood are similar to those for forest ﬂoor (Takahashi et al. 2000,
Spears et a1. 2003). Species comparisons of total N content have indicated that hardwood
wood initially contains more N than conifer wood (Arthur et al. 1993), but the amount of
N changes with decay stage (Holub et al. 2001) and is highly variable. Spears et al.
(2003) found few differences not only in mineral N concentrations but also for organic N
and a broad range of cations in solutions collected with lysimeters beneath the decaying
wood of four conifer species. Arthur et al. (1999) found that total calcium and
magnesium concentration on a mass basis were lower in birch than in sugar maple wood.
Given what little attention has been paid to species speciﬁc differences in wood chemistry
related to plant nutrition, generalizations cannot be made. Conceivably, speciﬁc limiting
nutrients in decaying wood could differ from forest ﬂoor layers, and also among species

of decaying wood.

47

If there are differences in nutrient availability to seedlings among log species, they
could be mediated, in part, by differences in pH. Phosphorus and zinc, for example, are
more available on organic substrates at pHs well below neutral (Smith 1989). Another
possibility is that species may vary in their biotic communities in ways that affect plant
nutrient availability but which can not be measured as solute concentrations or
mineralization rates. For example, uptake of N and especially of phosphorus is increased
by ectomycorrhizal infection (Kytoviita and Amebrant 2000, Perez-Moreno and Read
2000). Mycorrhizae are not necessary for, but do increase the rate of, uptake of amino
acids, an organic N source for tree seedlings (Perez-Moreno and Read 2000, Persson and
Nasholm 2001). Mycorrhizae are likely to be a requirement for long-term survival of
eastern hemlock seedlings given the advantage they provide for nutrient acquisition and
given the obligate mycorrhizal nature of the closely related western hemlock (Christy et
a1. 1982). Furthermore, the logs of western conifers serve as refuges for mycorrhizal
fungi and serve as inocula for western hemlock seedlings (Harvey et al. 1976, Kropp
1982c), although it is not known if western hardwood species (e.g. Alnus rubra Bong.)
can serve the same purpose. In our study system, it is possible that, like western conifers,
eastern hemlock wood provides mycorrhizal inocula for eastern hemlock and yellow
birch seedlings (both ectomycorrhizal species) and it is possible that it does this to a
greater degree than does sugar maple wood.

Wood decaying fungi fall into three general classes, brown rots, white rots and
soft rots. Soft rots are associated primarily with angiosperm wood, typically attack only
the outer layer of wood early in decay (Goodell et al. 2003) and are numerically relatively
unimportant as wood decomposers compared to brown and white rots, thus they are not

examined further here. White- and brown-rotted wood have different textures and

48

patterns of decay, and although a single log may have sections decayed by both types of
fungi, in general deciduous species are attacked by white rots and conifers by brown rots
(label and Morrell 1992 ). Brown rot fungi primarily digest cellulose and hemicellulose,
altering but not consuming most of the lignin in the wood and resulting in wood with
characteristic cubical breaks (Goodell et al. 2003). White rot fungi can digest lignin as
well, and cause decaying wood to break apart in soft strings. White rots depolymerize
wood components and then immediately digest the products, while brown rot fungi
initially depolymerize wood more rapidly than they are able to consume it (label and
Morrell 1992). To our knowledge nothing is known about the implications of these
differences for nutrient availability to tree seedlings, but brown rot fungal decay could
result in excess, unconsumed nutrients that are then potentially available to seedling
roots.

In summary, fragmentary evidence suggests that several decayed wood
characteristics could vary among species, and in ways that could inﬂuence their quality as
tree seedling establishment substrates. Studies speciﬁcally addressing this topic are
lacking. In an earlier study of hemlock-hardwood forests, we found large and parallel
differences in hemlock and yellow birch seedling abundance among hemlock, birch, and
sugar maple wood substrates. Moreover, differences in seedling populations among
decaying wood species were independent of exogenous factors including seed rain, light
and wood piece size, suggesting that species-speciﬁc variation in endogenous decaying
wood characteristics were responsible. Our objectives in this study were to: 1) quantify
and compare intrinsic characteristics of hemlock, birch, and sugar maple wood that are
potentially associated with hemlock and birch seedling establishment, and 2) where

possible, relate wood characteristics to differences in seedling abundance, growth, and

49

survival across wood species while controlling for exogenous factors. We accomplished
this with complementary ﬁeld and greenhouse experiments. From ﬁeld-collected data
and samples we measured the exogenous environmental factors canopy openness, seed
rain, habitat type, and hemlock basal area, and the intrinsic wood characteristics decaying
wood chemistry (pH, water content, and nutrient concentrations and mineralization rates),
wood decay characteristics (fungal rot type, wood residence time), wood surface physical
characteristics (bark, litter, and moss cover), and seedling mycorrhizal inoculation. A
companion greenhouse experiment examined differences in growth rates, nutrition,
survival and mycorrhizal infection for birch and hemlock seedlings growing on decaying
wood of hemlock, birch, and maple. For ﬁeld data, we examined the effect of intrinsic
wood characteristics on seedling abundance using regression models that included
exogenous factors in models as covariates with seedling abundance as a response
variable.
Methods
Field site descriptions

For this study we used sites in three primary hemlock-hardwood forests in Upper
Michigan (Huron Mountain Club, Porcupine Mountains, and Sylvania Wilderness)
between 2002 and 2005 (see Chapter 2 for locations and ﬁeld plot selection methods).
Field plots (0.1 ha, 1.5 ha total area) were located in these sites in hemlock-dominated
patches (> 55% basal area hemlock) and in adjacent mixed hardwood patches (10 - 35%
basal area hemlock).

Habitat types of each plot were determined using a habitat classiﬁcation system
developed for northern Wisconsin (Kotar et al. 1988) but considered appropriate for

Upper Michigan (J. Kotar, pers. comm., 2003). Soil moisture had been measured in each

50

plot prior to habitat typing, but differences in average soil moisture were small and did
not correspond to differences in indicator species presence or abundance. Plots were
either ATM (Acer-Tsuga—Maianthemum, a dry-mesic to mesic, moderately nutrient-rich
habitat type often dominated in our plots by Maianthemum and ferns), ATD (Acer-Tsuga-
Dryopteris, a more mesic and more nutrient-rich type than ATM with greater fern cover),

or ATM-ATD/ATD-ATM habitat types.

Field sampling

Censuses of decaying wood and tree seedlings were carried out in each ﬁeld plot.
All decaying wood > 10 cm in diameter in decay stages I-IV (stages follow Graham and
Cromack 1982) in each plot was counted, identiﬁed to species, and measured. We
identiﬁed wood to species by microscopic examination (70x to 400x, microscopes at the
USDA Forest Products Laboratory in Madison, WI) of thin wood slices of each sample
mounted on slides. Wood was categorized as being attacked by brown rot (solid cubical
or rectangular chunks of dark brown wood on most of the surface) or white rot
(predominantly wet, soft, stringy ﬁbers of wood) fungi (Goodell et al. 2003). Coverage
of bark, moss-covered bark, moss, litter, and bare wood was recorded by categorizing the
substrate found under clear 2 inch by 2 inch square quadrats placed across the diameter of
each log every 1.5 m. Due to our method of scoring moss-covered bark (moss with
visible ridges, patchy moss cover with bark visible in bare spots) this category is most
likely equivalent to a thin moss layer, and will be referred to as thin moss. Every tree
seedling (deﬁned as less than 30 cm in height, which encompassed nearly all individuals
and which was comparable to Rooney et al. (2000)) growing on wood was also counted

and identiﬁed to species, and seedling counts on each wood piece were analyzed with

51

negative binomial regression models described below. More detailed seedling survey
methods can be found in Chapter 2.

In August 2002, we collected wood and 2-year-old hemlock and birch seedlings to
be destructively sampled for analyses of wood N supply rate and seedling mass and N
content. This necessitated sampling outside of our ﬁeld plots since plots were being used
to monitor seedling abundance and survival from 2002-2005. We walked around the
outside of four ﬁeld plots from each ﬁeld site, described in Chapter 2, starting 10 m
outside of the plot border and continuing in additional circuits at 10 m increments. The
decay stage of every log, stump, or downed branch (with diameter > 10 cm) visible in
each circuit was checked until reaching a sample size of 10 conifer and 10 hardwood
logs, all decay stage 111 (soft sapwood, usually little bark cover) or decay stage IV (soft
throughout, usually no bark cover, branch stubs easily pulled free, Graham and Cromack
1982) for each plot. We collected samples from a total of 260 logs (chosen such that half
of the logs had hemlock or birch seedlings) and collected the top 10 cm of soil from
beside 78 of these logs (6 near each plot, randomly selected) after removing intact leaves
from the surface. From each piece of log thathad hemlock or birch seedlings growing on
it in the ﬁeld, we collected one, or two if available, established seedling(s) of each species
for measurement of mass and N content. Entire seedlings were collected from as close to
the log midpoint (to avoid log edge effects) as possible. Only two-year—old seedlings
were used in this study.

In July 2004 we collected additional seedlings from Sylvania Wilderness and the
Porcupine Mountains for mass and N content measurements. Up to 4 two-year-old
hemlock and 4 two-year-old birch seedlings were collected from each of the 41 logs

sampled. Unlike in 2002, sugar maple logs were oversampled because seedlings on sugar

52

maple logs are so rare. This necessitated canvassing a large area around ﬁeld plots to
obtain an adequate sample size of seedlings on sugar maple logs.
Analysis of ﬁeld samples

N characteristics: Soil, wood, and seedling samples were transported on ice to
Michigan State University and stored in a 4°C cold room. In 2002, wood and soil
samples were stored for 1 to 10 days before extraction for inorganic N (N03' and NH4”).
In 2004, wood and soil samples were extracted within 48 hours. For each sample, we
determined gravimetric moisture content by drying an approximately 10 g sample at
105°C. In 2002, two subsamples (23 to 27g fresh weight) of each log or soil sample were
used for a laboratory potential N mineralization measurement (modiﬁed from Powers
1980). Subsamples were placed in a 125-mL plastic specimen cup, and the volume was
determined using the mL scale of the cup. An initial sample of each piece of decayed
wood or soil core was immediately extracted with 50 mL of 0.5M K2804 solution, while
ﬁnal samples were incubated at room temperature (23-24°C) for 28 days before
extraction. Extracts were analyzed for inorganic N colorimetrically with an autoanalyzer
(01 Flow Solution IV, 01 Analytical, College Station, TX), with N03'-N determined by
the cadmium reduction method, and NHi-N with the phenol-hypochlorite method (Page
et al. 1982). Refrigerated extracts that could not be analyzed within one month were
frozen until analysis. In 2004, wood samples were extracted with 50 mL of 2M KCl
instead of K2804. Net mineralization rates (Nmin) were calculated as the difference

between initial and ﬁnal inorganic N amounts. We chose to express both initial N

concentrations ([N]) and Nmin on a volume basis (ug N (mL wood)’1 or ug N (mL

53

wood)‘l (day)") because seedling roots exploit a given volume of wood or soil and
because soil is denser than wood, making comparisons on a mass basis misleading.

Field-collected seedlings were kept on ice until processed in the laboratory. 2002
ﬁeld seedlings were examined with a dissecting microscope (10x to 70x) for evidence of
ectomycorrhizae. 2004 ﬁeld seedlings were scanned immediately upon return to the
laboratory as high-resolution digital images which were later examined for evidence of
ectomycorrhizae. Colored felty mantles on roots or root tips that were orange or yellow
were considered evidence of mycorrhizae, and because seedlings were for the most part
only 2 years old, white, smooth, non-mycorrhizal bare root systems were easily
distinguishable. In both years, seedlings were then dried for at least 48 hours in a 65 °C
drying oven. Seedlings were ground to powder using a mortar and pestle, and analyzed
individually for N concentration by the Dumas combustion method on a CN analyzer
(Carlo Erba, Milan, Italy). Hemlock and birch seedlings were pooled by individual log
for all statistical analyses.

Water content and pH: Water content of 239 logs in August 2002, 172 logs in late
June/early July 2003, and 127 logs in late May 2004 was measured gravimetrically as the
difference between ﬁeld weight (samples collected > 24 hrs after last rainfall) and oven-
dried (at least 3 days at 105°C) weight of wood samples, divided by ﬁeld weight. Logs
measured in 2002 were selected from outside ﬁeld plots as described above in Field
sampling, while logs in 2003 were randomly selected from within plots at all ﬁeld sites,
and logs in 2004 from within plots at the Porcupine Mountains and Sylvania Wilderness.
pH of 63 logs randomly selected from those collected in 2002, for which no data on
seedling abundance exists, was also measured. Unlike soil, once decayed wood is dried it

cannot easily be re-wetted, so we modiﬁed a standard soils procedure (Klute 1986) and

54

measured pH of fresh wood. For each sample a mass of wet wood equivalent to 3 g of
dried wood, calculated using the water content of each wood piece determined on a
subsample, was placed in a sample cup and deionized water (pH ~ 5.5) was added to
bring the total volume to 60 mLs. This was done to account for differences in density
among wood samples, and ensured a 20:1 ratio of water to dry wood for all samples,
leaving sufﬁcient water to immerse the pH probe. Samples were shaken for 1 hour and
allowed to settle. pH was read 6 minutes after immersing the probe.

Exogenous factors: light and seed rain

Canopy photographs were taken between May and August approximately 30 cm
directly above each log in ﬁeld plots and any log from which seedlings were collected in
2002. Photos were taken using a digital camera (Nikon Coolpix 995, set to grayscale)
with a ﬁsheye lens. All photos were analyzed using GLA software (Version 2.0, 1999,
Institute of Ecosystem Studies, Millbrook, NY) by a single technician. SideLook
software (v. 1.1.01, 2005, M. Nobis www.appleco.ch), which was developed in 2004, was
used to automatically threshold a subset of 22 canopy photos for comparison. The gap
light indices (canopy openness) of automatically and manually thresholded canopy photos
were similar (matched pairs mean difference = 0.62%, s.e. 0.16, n = 22 pairs, p = 0.001;
Pearson's correlation = 0.75).

Seed rain was measured near 15 randomly selected logs within each of 12
seedling census plots, four at each site (3 plots in the Huron Mountain Club from which
the surrounding area had been logged for hardwoods were excluded). Seed traps were
constructed from 22 cm-diameter (380 cm2 surface area) 2-gallon plastic pots, lined with
plastic weed cloth and with a piece of 1/2 inch wire mesh covering each trap about 1 inch

below the top. Plastic canvas (6 squares per inch) was used in the bottom of traps to

55

allow drainage but prevent entry of seed predators. Seed traps were placed alongside the
midpoint of each log and seeds were collected from August 2003 to late May/early June
2004, with leaves cleared from trap surfaces in October 2003 at ﬁrst snowfall. In 2004,
contents of the remaining 138 undisturbed traps were dried at 65°C, and seeds were
counted and a subsample was cut open to determine percentage of seed ﬁlled. Birch
seeds, the most abundant type, were counted up to 200 seeds, with the remaining number
of seeds estimated by weight. Note that seeds were collected in the year following a mast
seed year for birch and sugar maple.
Greenhouse experiment

We carried out a l 15-day(mid-July through early November) experiment at the
Michigan State University Plant Science Greenhouses to compare mass and N content of
hemlock and birch seedlings grown on hemlock, birch, and sugar maple wood. Both
birch and hemlock seedlings were planted on all three wood types in a full factorial
design. In late May 2004, stage III and IV wood was collected from 20 logs each of
yellow birch, hemlock, and maple logs in Sylvania Wilderness and the Porcupine
Mountains. [N] of each sample was measured in the laboratory as described in the
Analysis of Field Samples section above. From each wood sample, a 180 mL subsample
was placed into each of two pots (wood was embedded in moist perlite), one for hemlock
seeds and one for birch seeds. For 10 of the logs per species, two additional subsamples
(one for birch and one for hemlock seedlings) were used for a fertilizer treatment, and one
additional subsample was used for a seedling drought tolerance experiment using only
birch seedlings. Birch was used because birch leaves visibly wilt while in previous
experiments with spruce needles remained turgid even after seedlings had died, and

because birch roots grew quickly enough to have penetrated nearly the entire volume of

56

the wood pieces on which they were grown at harvest. The drought tolerance experiment
pots were set up using the same methods as the main experiment, except that after three
months birch seedlings were no longer watered and were instead monitored until they
wilted. When all birches within a pot had wilted, the wood on which they were growing
was collected and weighed for determination of water content.

Eastern hemlock seeds (Ontario source, Ontario Tree Seed, Angus, ON) and
yellow birch seeds (Michigan source, USFS Toumey Nursery, Watersmeet, M1) were
stratiﬁed in perlite and wet sand at 4°C for two months before being placed on trays of
wet perlite in the greenhouse in early July. One week later, after birch seeds had
germinated and most hemlock seeds had cracked seed coats but did not have visible
radicles, nine seeds of either birch or hemlock were pushed into the surface of the wood
in each pot. Wood was saturated every day to ensure that seeds in the top layer of wood
did not dry out. Birch seedlings quickly developed cotyledons, and were thinned to four
seedlings per pot. Hemlock seedlings grew vertically and lost their seed coats before
expanding their ﬁrst leaves almost three weeks after birch seedlings, and were not
thinned. Throughout the experiment, seedlings were shaded with aluminum (Aluminet,
Polysack USA, San Diego, CA) 70% shade cloth. Seedlings were grown under natural
light conditions in July and August and supplemented with standard greenhouse lighting
(16 hours per day) above the shade cloth for the remainder of the experiment.
Temperatures in the greenhouse ranged from 21 to 35°C.

For the ﬁrst month of the experiment, seedlings were watered to excess with
groundwater which contained some inorganic N. After this establishment period, ﬁlters

were added to remove inorganic N and seedlings were watered to excess with deionized

57

water containing negligible nitrate (< 0.1 mg/L NO3-N) and no measurable ammonium

every second day. Also after one month, Scotts Proturf 38-0-0 Poly—S was applied to pots
in the fertilizer treatment at 200 kg N per hectare (0.3188 g per pot). Surprisingly, this
application of N quickly killed approximately half of the fertilized seedlings (with no
detectable pathogen presence, MSU Diagnostics Lab). Seedlings that survived the ﬁrst
week after fertilization tended to survive to the end of the experiment.

After 1 15 days, all surviving seedlings were harvested, cleaned, and refrigerated.
Seedling root systems were systematically visually scanned under a dissecting scope (10x
- 70x) within 36 hours of harvest for evidence of mycorrhizal infection. Seedlings were
then dried at 65°C, weighed individually, pooled by pot, and ground by mortar and pestle
or with a steel ball pulverizer into a ﬁne powder. N content of ground seedlings was
analyzed as described above for ﬁeld seedlings. The remaining dried seedling powder
from a subset of birch seedlings (individual seedlings were pooled from each of 7
randomly selected hemlock, 6 birch, and 7 sugar maple wood samples) was analyzed for
boron, calcium, copper, iron, potassium, phosphorus, magnesium, manganese, and zinc
concentration (ppm). Samples were microwave digested at 205°C for 30 minutes (MARS
5, CEM Corporation, Matthews NC) with 1 part hydrogen peroxide, 1 part nitric acid, and
2 parts deionized water. Digested samples were diluted 1:1 with deionized water and
analyzed with a direct current plasma atomic emission spectrometer (SMI III,
Spectrametrics, 1nc., Andover, MA). Hemlock seedlings did not show variation in size in
the greenhouse and so were not analyzed for nutrients besides N.

Log residence time

58

Tyrrell and Crow (1994a) have already estimated residence time for hemlock logs
at sites similar to ours. To estimate residence time of birch and sugar maple logs, 38
birch and 46 sugar maple logs were located in the Porcupine Mountains and Huron
Mountain Club Reserve. We also sampled six hemlock logs to compare age estimates to
those determined by Tyrrell and Crow. Logs showing evidence of having spent time as
standing dead trees (bracket fungi growing perpendicular to the ground, extensive
woodpecker holes, early decay stage logs with no visible remnants of the crown) were not
selected. All logs used were clearly identiﬁable as decay stage 11 (only sapwood soft,
branches remain), 111 (most or all bark lost, except in the case of birch, branch stubs
remaining, and soft sapwood with a hard inner core), or IV (soft throughout, branches and
most branch stubs absent; stages follow Graham and Cromack 1982), could be identiﬁed
to species in the ﬁeld by using a hand lens to examine wood characteristics, and were
associated with at least two obvious release trees. Release trees were deﬁned as trees that
had responded to the gap created by the death of a canopy tree (now a sample log) with
increased growth. Release trees used were either small-diameter trees growing beside the
stump end of each log, or, more rarely, small trees whose branches had grown laterally
directly over the stump end of the log. Most trees that showed a clear release were
hemlocks (63%) or sugar maples (22%). Each release tree was cored at breast height (1 6-
inch 4.3 mm-diameter increment borer), and cores were dried in paper straws, mounted
on wood blocks and sanded (150 through 600 grit sandpaper). Date of release for each
tree was determined by measuring the width of each annual ring (Velmex Unislide with
QCI 100 measuring system, Velmex, Inc., Bloomﬁeld, NY) and comparing the uniformly
weighted moving average ring widths for both release trees of a given log in JMP (JMPIn

5.1, 2004, SAS Institute, Cary, NC) using a modiﬁcation of the Lorimer and Frelich

59

( I 989) method. If both trees showed a release (at least one 7-year average of ring widths
greater than 2 standard deviations above the mean 7-year ring width) within 10 years of
each other, the release dates of the most recent simultaneous release were averaged and
the log was assumed to have fallen in that averaged year. These were conservative
criteria and resulted in a low sample size, and may have slightly underestimated dates of
release since only logs where both trees had responded to release with a large increase in
ring width at the same time were included. A second set of release dates was determined
by subjectively scoring each tree separately using 5-year moving averages and counting
the most recent release even if it was just under 2 standard deviations above the mean for
trees with consistently thin growth rings but a clear peak in width. We included release
dates if one or both trees for a given log showed a release. Ages of stage III logs were not
included in ﬁgures or analyses since their age range spanned parts of decay stages 11 and
IV and yielded no additional information about the total stage II-IV residence time of
logs.
Data analysis

We tested the effects of various exogenous factors and wood characteristics on
seedling abundance and determined which of these factors vary with wood species.
Seedling abundance data from logs in ﬁeld sites were analyzed as seedling counts rather
than as densities based on the high number of logs with zero seedlings or extremely small
seedling and seed rain counts. A zero count is equal to a zero density, which is
problematic because a relatively high proportion of small logs have no seedlings at all,
and the remainder have very high variance in seedling densities due to multiplication of
few seedlings many times over to generate densities per square meter. Zero-inflated

count data are resistant to transformation and we ﬁrst tested various non-parametric

60

statistical methods (analyzing presence/absence and logs supporting seedlings separately,
negative binomial nonlinear regressions, placing logs into abundance classes before
analysis) as methods of screening the effects of environmental and wood characteristics.
We ultimately used parametric tests (t-tests, ANOVAs, and single linear regressions) to
screen individual factors measured on logs both inside and outside of ﬁeld plots, and
nonparametric tests to build explanatory models of seedling abundance on logs in ﬁeld
plots. Exogenous and endogenous factors compared across wood species were, with the
exception of nitrogen mineralization rates, normally distributed or could be transformed
and so ANOVA tests were appropriate.

Using single linear regression models, we related 12 factors that might affect
seedling abundance to seedling abundance measured in the ﬁeld: light, seed rain (for ﬁrst-
year seedlings), wood water content, decay stage, rot type, wood species, % bark, % thin
moss, % litter, log surface area, ﬁrst-year seedling history, and established seedling
history. Data were divided by site because, despite strong generality in trends among
sites, site effects were still signiﬁcant for most factors, including seedling abundance.
(Chapter 2). This exploratory analysis allowed us to generate a percent of the variation in
seedling abundance that was explained by each individual factor. In addition, we used
multiple linear regression to explore candidate best-ﬁt multiple regression models of
seedling abundance. Linear regression screening of factors gave easily interpretable and
potentially biologically relevant information, as well as our only measure of overall ﬁt of
candidate multiple regression models, but was used for exploratory purposes only.

Greenhouse data were analyzed using a combination of correlations, t-tests, and
linear regressions to determine the relationships between wood nutrient content,

fertilization, and mycorrhizal colonization and seedling growth and N content. With the

61

exception of nitrogen mineralization (Nmin) rates of wood, greenhouse data were
successfully logarithm-transformed. Because Nmin rates were resistant to
transformation, they were analyzed using non-parametric tests and Nmin and [N] data are
presented in Figure 3.3 as medians rather than as means.

In order to generate best-ﬁt multiple regression models of factors inﬂuencing
seedling abundance, we used negative binomial regression in SAS (proc genmod, dist=nb
option). Negative binomial modeling is the appropriate technique to use with the
overdispersed, zero-inﬂated seedling counts in our study. Each starting model included
seedling history (the number of established seedlings on the log in the previous year), and
then we added each additional factor in the order: seed rain (for ﬁrst-year seedlings only),
log size, light, log species, rot type, decay stage, % bark, % litter, % moss. For
established seedlings, we also added ﬁrst-year seedlings from the previous year and
determined which additional factors were signiﬁcant with both seedling history and ﬁrst-
year seedlings in the model, but we did not include ﬁrst-years in ﬁnal models because
often no other factors were important with both established and ﬁrst-year seedling history
in the model. The addition of second-order interactions that contained categorical
variables to models caused the model algorithm to fail to converge. This was not purely a
sample size problem, as in ﬁeld sites and years where the distribution of seedling
abundance data was not similar in shape to a negative binomial distribution (e.g. when
most data were zero values with several high outliers), even interactions containing only
numerical variables resulted in a lack of convergence. Consequently, not all second-order
interactions and no higher-order interaction terms are presented in our best-ﬁt models.

Akaike information criteria (AICc) were used to compare model ﬁts, and AICs

were corrected for our relatively small sample size. AICc corrects for overﬁtting of

62

models, and in several cases models containing only seedling history (our starting
models) had the lowest AlCc. We did not, however, include these seedling history-only
models as candidate best-ﬁt models but rather continued to add factors because models
with additional factors had similar AICc values and those additional factors were
signiﬁcant even with seedling history in the model. As we added each factor to the
candidate best-ﬁt model, factors remained if they were signiﬁcant (chi-square p-value <
0.05), and replaced existing factors when the new model AICc was > old model AICc +
2. Given the close agreement between negative binomial regression and standard
multiple regression, as well as between models selected by comparing log-likelihoods and
those comparing AICc, we suspect that all of the modeling techniques we used are
sensitive to the few logs with high abundances of seedlings. We are more conﬁdent that
the factors identiﬁed by these techniques are important in determining seedling
abundance than in the p-values assigned to each individual factor.
Results
Exogenous factors: habitat type, seed rain, and light

Percentage of ﬁlled seeds was low at all ﬁeld sites, ranging from 9% to 20% for
hemlock, 5% to 32% for birch, and 18 to 24% for sugar maple across sites, and did not
vary systematically across log species or stand type, thus viable seeds were considered to
scale with total seeds counted and will not be discussed further. Seeds of hemlock and
birch were well dispersed and abundant, with 131 of 138 seed traps (22-cm diameter)
containing at least one hemlock and at least one birch seed. Seed rain density did not
vary among log species with the exception of hemlock seeds in mixed plots at the
Porcupine Mountains, where traps alongside birch logs received signiﬁcantly greater

numbers of seeds than those alongside sugar maple logs (ANOVA, effect of log species

63

on hemlock seeds/m2, p = 0.034, Table 3.2). Stand type differences in seed rain are
discussed further in Chapter 2, and in this study although seed rain varied widely within
each log species it was positively related to the number of hemlock and birch germinants.
However, seed rain density explained little variation in ﬁrst-year birch or hemlock
seedling abundance in most cases (maximum R2 = 0.18), nor did the size of each log
(maximum R2 = 0.15). When seed rain falling into each trap was multiplied by the size of
each log to generate an estimate of total seed rain on each log, seed rain explained 8 to
58% of variation in ﬁrst-year hemlock seedling abundance (0 - 4% of birch seedling
abundance).

Basal area of hemlock affected seedling abundance (see Chapter 2), while neither
hemlock nor birch seedling numbers varied with habitat type in the Huron Mountain or
the Sylvania Wilderness sites. Hemlock abundances were signiﬁcantly greater on logs in
the less mesic ATM plots than in the more mesic ATD plots at the Porcupine Mountains
in 2002 and 2003 (t-test, p = 0.010 to 0.031, n = 100, data not shown), but not 2004, and
hemlock, birch, and sugar maple logs were equally likely to be found in each habitat type
at this site.

Canopy openness (a surrogate for light) varied from 2 to 17% across logs from all
sites, but 70% of the values were < 6%. Canopy photos cannot accurately detect small
differences in light availability below this point (Machado and Reich 1990), and thus the
small (~ 0.70%) differences in canopy openness above sugar maple logs as opposed to
birch and hemlock logs at the Porcupine Mountains and Sylvania are not biologically
meaningful. While sites varied in light availability, with values higher at the Huron
Mountain site, mean values were similar among log species at each site (Table 3.2).

Light was rarely a signiﬁcant predictor of ﬁrst-year or established hemlock and birch

64

seedling abundance, explaining at most 5% of the variation in seedling abundance in a
single-factor regression.
Endogenous factors: Water content, decay, and wood surface characteristics

Field water content did not vary with wood species at most sites (Figure 3.1) and
was only about 10% lower than ﬁeld capacity measured for 16 samples of hemlock and
sugar maple wood in the laboratory (data not shown). Sugar maple logs had greater water
content than both hemlock and birch logs (student's t-test, p < 0.03, mean difference = 9%
for both species) at the Huron Mountains in 2002, while in 2003 sugar maple logs were
drier than birch logs at Sylvania Wilderness (student's t-test, p = 0.03, mean difference =
7%), suggesting that sugar maple water contents may ﬂuctuate more than those of other
species. However, in a greenhouse drought tolerance study of birch seedlings planted on
hemlock, birch, and sugar maple wood, there was no difference across log species in the
number of days birches survived without water or the water contents at which they wilted.
Birches wilted at wood water contents between 16 and 50% (mean 24.5% :1: 1.4 s.e., n =
30, Figure 3.1), and we encountered only six out of 452 logs measured over 2002-2004 in
the ﬁeld that had less than 50% water. Water is always available in the lumen of wood
(and therefore available to seedlings) that is at greater than 30% water content (P.
Kamdem, pers. comm. 2005). Water content was not signiﬁcantly and positively related
to seedling abundance in most sites at most years, with the exception of ﬁrst-year
hemlock seedlings in 2003 at the Huron Mountain Club (R2 = 0.16, p < 0.001, n = 85).

We examined wood decay rate from the perspective of the length of time wood
can serve as a substrate for seedling establishment (usually decay stage II - IV, Takahashi
et al. 2000, Mori et al. 2004, Chapter 2) and hereafter call time spent in these decay stages

residence time. The residence time for hemlock logs has already been determined in two

65

of our study sites by Tyrrell and Crow (I994a) and their estimates matched our estimates
based on a small sample of six hemlock logs (Figure 3.2). The range of ages of logs in
decay stage II-IV (maximum possible residence time) was 2 to 57 years for hemlock, 6 to
45 years for birch, and 10 to 30 years for sugar maple, based on our conservative
estimates (Figure 3.2). Average surface area of the three log species does not vary in
ﬁeld plots (ANOVA, p = 0.48). However, average hemlock log diameter was greatest in
both ﬁeld plots (mean = 28.2 cm, birch = 24.5, maple = 23.9, includes boles, stumps, and
branches, ANOVA p = 0.054) and the boles aged by coring release trees (mean 57.5 cm,
birch = 44.0 cm, maple = 43.2 cm, ANOVA p = 0.008). These differences may partially
explain the slower decay rate of hemlock logs, since larger-diameter logs might be
expected to decay more slowly than smaller logs.

Hemlock logs are more likely than birch or sugar maple logs to be brown-rotted
rather than white rotted (Table 3.2; chi-square analysis, likelihood ratio p < 0.005 at each
site), and the Huron Mountains site has the most brown-rotted logs. Within each wood
species, brown-rotted logs did not have signiﬁcantly different numbers of seedlings than
white-rotted logs. Rot type explained signiﬁcant variation in seedling abundance at the
Huron Mountain Club when it was the sole factor in a regression (R2 = 0.05 to 0.14 across
seedling species and years, all p-values < 0.05) and appears in several of the negative
binomial multiple regression models detailed below (Table 3.1).

Wood species varied consistently across sites in the proportion of their surface
area that was covered by thick and thin moss cover, bark, and litter but differences were
signiﬁcant (one-way ANOVA p < 0.05) only at the Porcupine Mountains. Hemlock logs
had the least bark coverage, the most litter, and the most combined thick and thin moss

cover. Hemlocks appeared to accumulate moss cover over time, as decay stages 1- III

66

had the greatest percent cover of thin moss, while stage IV had the least percent cover of
thin moss and the greatest of thick moss. Bitches had the greatest percent bark cover at
all sites (p = 0.008 at Porcupine Mountains), due to their pattern of retaining an almost
intact ring of bark around a highly decayed bole, while sugar maple logs did not have any
differences in surface area cover that were consistent across all sites. Thin moss cover
explained signiﬁcant but small amounts of variation in abundance of ﬁrst-year birch and
hemlock seedlings at the Porcupine Mountains and Sylvania sites (R2 = 0.05 to 0.15
across species and years), and appears in one of the explanatory models below (Table
3.1).
Endogenous factors: Wood chemistry and mycorrhizae

pH of hemlock logs was low, and signiﬁcantly different from that of both birch
and sugar maple logs (ANOVA p < 0.0001, n=63, Table 3.2). pH was related to log
species and not the environment of each log, as evidenced by different pHs found on logs
and adjacent soils and the lack of site or stand type effects on pH (data not shown). Soil
pH, which ranged from 3.9 to 5.7, was affected by site (ANOVA p = 0.017, n= 31).

Mineral N characteristics of wood varied signiﬁcantly across species (N min 2002
Wilcoxon/Kruskal-Wallis chi-square p = 0.023, [N] 2002 ANOVA p = 0.004, [N] 2004
ANOVA p = 0.001), and was always lower in wood than in soil (Figure 3.3). [N] and
Nmin were inversely ranked across species, with the highest Nmin and lowest [N] in
hemlock logs, and vice versa for sugar maple logs (Figure 3.3). In all log species, NH,+
dominated [N], with N03‘ negligible in 2002 and approximately 30% of [N] in 2004,
regardless of log species (Marx, data not shown). Within each individual log, Nmin was
positively correlated with [N] measured at the start of the laboratory incubation (Pearson's

r = 0.346, p < 0.001). It is important to note that because 2002 and 2004 [N] were

67

measured at different times of the growing season (late May for 2004, late August for
2002), were extracted with two different salt solutions, and 2002 samples spent as long as
10 days in cold storage, these two measures are not directly comparable. 2002 values
were higher than 2004 values, perhaps due to these methodological differences, but
rankings across species were the same in 2002 and 2004.

Seedling mycorrhizal infection status of seedlings also varied with wood species.
Mycorrhizae, which may increase uptake of N and P, were rarely found on two-year-old
seedlings collected from the ﬁeld. Only 2 out of 45 (4%) 2-year-old seedlings in 2002
and 8 out of 186 (4%) seedlings in 2004 were infected with all log species pooled.
Mycorrhizae were more abundant on older seedlings collected from the ﬁeld in 2002,
with 12% of birch seedlings and 10% of hemlock seedlings on hemlock logs infected, and
0% of birches and 3% of hemlocks on birch logs infected (n = 50 birch and 88 hemlock
seedlings older than 2 years). There were almost no older seedlings on maple logs
(Chapter 2). In the greenhouse, rates of infection were higher than in the ﬁeld for ﬁrst-
year birch seedlings, with 20.3% of seedlings infected (n = 237 unfertilized seedlings) but
similar for hemlock seedlings (4.9%, n = 263). Hemlock and birch logs supported greater
percentages of mycorrhizal birch seedlings (25% on each wood species) than did sugar
maple logs (11%), while similar proportions of hemlock seedlings (3-7%, 13 out of 263
total) were infected across wood species.

Seedling nutrient concentrations and growth

In the ﬁeld and greenhouse we used seedling growth and nutrient concentrations
as bioassays of differences in nutrient availability among log species. In both the ﬁeld
and the greenhouse (unfertilized treatment) seedling birch seedling mass was greater on

hemlock logs than sugar maple logs (ﬁeld: t-test p = 0.003, greenhouse: t-test p = 0.092,

68

Figure 3.4). Hemlock seedling mass did not signiﬁcantly differ across wood species in
the greenhouse or the ﬁeld (ANOVA p-values > 0.1, Figure 3.4). Despite differences in
birch seedling mass and in wood N characteristics across species, there was no correlation
between either [N] or Nmin with birch seedling N content or mass, in either the ﬁeld or
the greenhouse (N content: Pearson's correlation always < |0.431|, p > 0.189; seedling
mass: < |0.519|, p > 0.124).

In the greenhouse, both N-fertilization and mycorrhizal colonization increased the
mass of seedlings under some circumstances. Fertilized birch seedlings growing on
hemlock and sugar maple logs in the greenhouse were three to four times as large as their
unfertilized counterparts growing on different pieces of the same log (paired t-test n = 10
pairs, p = 0.0004, Figure 3.6c, Figure 3.7). Fertilized seedlings on these two wood types
also had increased N contents (fertilized = 0.224 g, unfertilized = 0.065 g , n = 9 pairs, p
= 0.0001). However, birch seedlings growing on birch logs did not respond to
fertilization with increased mass or N content (masszp = 0.706; N content: p = 0.151, n =
8 pairs, Figure 3.7). Despite statistical signiﬁcance across wood species, hemlock
seedlings did not respond to fertilization with a biologically meaningful increase in mass
(median fertilized = 0.01 1 g, unfert. = 0.010 g, paired t-test with all species pooled, n =
14, p = 0.03), and differences for individual wood species were not signiﬁcant. Hemlock
seedlings had higher N contents when fertilized (fertilized = 0.025 g, unfert. = 0.017 g, n
= 14, p = 0.004). The median mass of mycorrhizal seedlings was greater than that of non-
mycorrhizal seedlings for both hemlock (t-test log-transfonned data p < 0.0001, Figure
3.5) and birch (p = 0.023, Figure 3.6a and b, Figure 3.5).

In order to determine why fertilization affected seedlings on some wood types but

not others, we compared micro- and macronutrient concentrations in a subset of our

69

between either [N] or Nmin with birch seedling N content or mass, in either the ﬁeld or
the greenhouse (N content: Pearson's correlation always < |0.43 II, p > 0.189; seedling
mass: < |0.519|, p > 0.124).

In the greenhouse, both N-fertilization and mycorrhizal colonization increased the
mass of seedlings under some circumstances. Fertilized birch seedlings growing on
hemlock and sugar maple logs in the greenhouse were three to four times as large as their
unfertilized counterparts growing on different pieces of the same log (paired t-test n = 10
pairs, p = 0.0004, Figure 3.6c, Figure 3.7). Fertilized seedlings on these two wood types
also had increased N contents (fertilized = 0.00224 g, unfertilized = 0.00065 g , n = 9
pairs, p = 0.0001). However, birch seedlings growing on birch logs did not respond to
fertilization with increased mass or N content (masszp = 0.706; N content: p = 0.151, n =
8 pairs, Figure 3.7). Despite statistical signiﬁcance across wood species, hemlock
seedlings did not respond to fertilization with a biologically meaningful increase in mass
(median fertilized = 0.011 g, unfert. = 0.010 g, paired t-test with all species pooled, n =
14, p = 0.03), and differences for individual wood species were not signiﬁcant. Hemlock
seedlings had higher N contents when fertilized (fertilized = 0.00025 g, unfert. = 0.00017
g, n = 14, p = 0.004). The median mass of mycorrhizal seedlings was greater than that of
non-mycorrhizal seedlings for both hemlock (t-test log-transformed data p < 0.0001,
Figure 3.5) and birch (p = 0.023, Figure 3.6a and b, Figure 3.5).

In order to determine why fertilization affected seedlings on some wood types but
not others, we compared micro- and macronutrient concentrations in a subset of our
greenhouse-grown birch seedlings (Table 3.4). For several elements (copper and iron), 3
strong negative correlation between seedling concentration of each element and seedling

mass combined with very high seedling concentrations indicated that seedlings had not

70

on a given log (Table 3.1, Figure 3.9). For established seedling models of both hemlock
and birch, most of the variation (R2 = 0.77 to 97) was explained by the number of
established hemlock seedlings in the previous year (seedling history), which is not
surprising (Table 3.1). However, even with these factors in the model, several additional
factors explained part of the remaining variation. For hemlock and birch seedlings in
2003, log size and rot type explained signiﬁcant additional variation in hemlock seedling
abundance, and rot type effects were especially strong (Table 3.1). Furthermore, when
log species was added to models, overall model ﬁt was lower and species effects were
weaker than those of rot type. 2004 models for established seedlings differed in some
ways. Like 2003 models, log size was signiﬁcant in many cases, but it often added
signiﬁcant explanatory power through its interactions with seedling history. In addition,
rot type was not as important as wood species for birch seedlings and neither factor was
important for hemlock. For birch, 2003 was a mast seedling germination year such that
established seedlings in 2004 were dominated by this young cohort. The lack of a strong
association between birch seedling abundance and rot type in this year could occur if rot
type were more important for long than for short term survival.

Not all factors inﬂuencing hemlock and birch establishment could be tested. For
example, the effect of pathogens and the date at which wood species reach the optimal
germination temperature of hemlock and birch are not examined here. Factors such as
these may be responsible for the lack of seedlings on the majority of logs (69 to 88% of
logs at the Porcupine Mountains), since none of the factors measured here could be used
to correctly predict which logs had either ﬁrst-year or established birch or hemlock
seedlings present. Our most accurate discriminant analysis, for ﬁrst-year hemlock

seedling presence, still misclassiﬁed half of the logs with ﬁrst-year seedlings as logs that

71

did not have seedlings. However, on those logs that do support seedlings, we found that
for both hemlock and birch seedlings, rot type, log size, seed rain and moss cover (for
ﬁrst-year seedlings), and survival of the previous years' established seedlings explain the
majority of variation in seedling abundance.
Discussion

In a previous study (Chapter 2), we identiﬁed hemlock wood as a better substrate
for hemlock and birch seedling establishment than birch wood or sugar maple wood, and
increased survival rate as one of the mechanisms explaining this difference. In this study,
over half of the variation in established hemlock seedling abundance was explained by
the established seedlings present in the previous year (seedling history), a ﬁnding similar
to that of Rooney et al. (2000). Established birch seedling distributions were also largely
driven by long-term seedling survival. However, even with seedling history as well as
light, log size, and seed rain in negative binomial regression models, several endogenous
wood characteristics explained signiﬁcant additional variation. Of the variables we could
test in ﬁeld wood samples, greater thin moss cover and brown rot fungi, which are both
more abundant on hemlock logs, were associated with greater numbers of seedlings. For
birch seedlings in the greenhouse, seedling mass and N content were both greatest on
hemlock logs, indicating that the endogenous factors identiﬁed here were associated not
only with hemlock and birch seedling abundance and survival but also with short-term
growth and nutrient uptake increases. Collectively, our results indicate that several
characteristics of hemlock wood may have contributed to increased seedling abundance,
survival rate, and growth on this wood type: 1) Greater nutrient availability (lowest pH,

greatest seedling growth rate and N content), 2) a greater percent of wood surface area

72

covered by moss, and 3) a more favorable pattern of decay (greater tendency to be brown-
rotted and longer residence time on the forest floor).
Wood and seedling nutrient supply

Hemlock logs have a much lower pH (~ 4) than either birch or sugar maple logs.
The pH of the litter layer in hemlock stands in northern Michigan and Wisconsin is
similar to that found in hemlock wood (Rogers 1980). However, the basal area of
hemlock, birch, and sugar maple in the area surrounding each log had no effect on pH,
and pH of wood and nearby soil were different, suggesting that unlike soil (F inzi et al.
1998b), hemlock logs maintain their internal pH regardless of environmental conditions
around them. The near-neutral pHs found on sugar maple logs are less favorable for
hemlock than the lower pHs found on hemlock logs, as several nutrients are less available
at these higher pHs including phosphorus (which drops off sharply at pH 7.5 in organic
soils), nitrogen, potassium, manganese, and zinc (Smith 1989).

Because nutrients were measured in a subset of seedlings, logs, and soil, and
could therefore not be included in best-ﬁt models, we can only speculate as to the
importance of nutrient supply in determining seedling abundance in comparison to the
other endogenous factors tested with regression models. Our greenhouse results suggest
that nutrient differences are important even within the ﬁrst one or two growing seasons,
and the response of birch seedlings growing on hemlock or sugar maple logs to N-only
fertilizer (Figure 3.7) indicates that N must be at least partially limiting on these wood
types. %N in unfertilized seedlings was lower than that found in Betula verrucosa
seedlings at optimal relative growth rates (Ingestad 1971) and was increased by

fertilization. This would suggest that N is important for seedling growth and therefore

73

plays a role in determining seedling abundance. However, taken as a whole, our results
also strongly suggest that N is not the sole limiting nutrient for seedling growth on wood.
Although growth, survival, and abundance of seedlings were all greatest on
hemlock logs (Figure 3.4, Chapter 2), neither N concentration ([N]) nor N mineralization
rate (Nmin) were signiﬁcantly correlated with birch or hemlock seedling N content or
biomass, whether wood species were pooled or tested separately. Several other studies
have also found a lack of correlation between N supply and hemlock or birch seedling
growth (Crabtree and Bazzaz 1993, F inzi and Canham 2000), and some have found that
N supply inﬂuences survival though not growth (Sivaramakrishnan and Berlyn 1999,
Catovsky and Bazzaz 2002). In our study, [N] was lowest on hemlock logs while Nmin
was highest. Both measures are potential indicators of N available to plants, and both
measures also ignore some parts of the N cycle (Binkley and Hart 1989). [N] is affected
by vegetation density and uptake, and thus likely a more accurate measure of N
availability for individual seedlings (Walters et al., in press., but see Grenon et al. 2005),
which is the scale in which we were interested in this study. Decaying hemlock logs
contain more roots than sugar maple logs (Marx, pers. obs.), and greater root uptake
could account for lower [N] despite higher Nmin in hemlock logs (Figure 3.3). Note that
differences in N supply were only signiﬁcant when we used the full set of wood samples
collected from 2002 or 2004 (as in Figure 3.3), and not signiﬁcant when we examined
only wood samples from logs that supported 2-year-old seedlings (as in Figure 3.4). This
was due to high variability in wood N supply in the smaller subset, and may have
contributed to the lack of correlation between wood N supply and seedling characteristics.

We did, however, ﬁnd strong evidence for co-limitation of seedling growth on wood by N

74

and phosphorus (P) supply. Three lines of evidence suggest that hemlock logs supply an
optimal balance of nutrients to seedlings:

First, birch seedlings grown on hemlock logs for just four months were
signiﬁcantly larger than those grown on sugar maple logs. The measurable differences in
growth in such a short period of time with exogenous factors held constant suggest that
nutrients may be very important for short-term growth and survival, and given the length
of time both hemlock and birch seedling roots remain in logs before the seedlings are able
to support their own weight or explore other substrate, likely important for long-term
growth and survival as well.

Second, mycorrhizal bitches and hemlocks in the greenhouse were signiﬁcantly
larger than their non-mycorrhizal counterparts (Figure 3.6a and b, Figure 3.5). Access to
both N and P is improved by mycorrhizae (Perez-Moreno and Read 2000), and both our
greenhouse and ﬁeld results suggest that seedlings growing on sugar maple logs (sugar
maple is endomycorrhizal; Klironomos 1995) may have a lower chance of becoming
colonized by ectomycorrhizae, possibly due to differences in ectomycorrhizal ﬁlngal
communities across wood species that parallel those in wood fungal communities (Kropp
1982b), or to a lack of tree roots that would provide a source of mycorrhizal inoculum
(Dickie and Reich 2005) in sugar maple logs. In the ﬁeld, older seedlings were more
likely to be mycorrhizal than 2-year-old seedlings, suggesting a survival advantage
conferred by mycorrhizal inoculation. One caveat in interpreting these results is that
seedlings may be infected with mycorrhizae because they are larger and have a greater
rooting volume, rather than larger because they are infected with mycorrhizae.

A third reason that bitch seedlings, if not both birch and hemlock, grow largest

and survive best on hemlock logs may be that hemlock logs supply low but sufﬁcient

75

levels of both N and P, resulting in seedlings with the most balanced N:P ratio (11.4),
similar to the ratio of 12.5 suggested for optimal growth with minimal nutrition (Ericsson
and Ingestad 1988). On birch logs, birch seedlings had %P values lower than those
required for optimal relative grth rate in Betula pendula (Ericsson and Ingestad 1988)
and high N:P ratios. Seedlings on this wood type also did not respond to N-fertilization,
suggesting that they were most limited by P. Thus, N:P explained only 13% of variation
in seedling mass on this wood type. On hemlock and sugar maple logs, N:P explained 54
and 42%, respectively, of variation in seedling mass, suggesting that in addition to N, P
limits growth on these species. Sufﬁcient nitrogen and phosphorus availability in
hemlock wood, whether mediated by pH, type of decay fungi, or mycorrhizal
colonization, increased the growth (and, in the ﬁeld, potentially the survival) of hemlock
and birch seedlings under greenhouse conditions.
Wood surface characteristics

The importance of percent cover of thin moss in explaining both ﬁrst-year
hemlock and birch seedling abundance is not surprising, but its association with hemlock
wood is novel. Several authors have reported a correlation of moss with seedling
densities or germination in the past. Goder (1961) reported that moss cover and hemlock
seedling density were positively correlated on decay stage II and III wood. In Oregon,
moss-covered wood surfaces retain more western hemlock seeds than some bark-covered
or bate surfaces (Harmon 1989), and this is likely true for eastern hemlock seeds in
Michigan. Although for the most part thin moss and total moss cover were
interchangeable, thin moss was a more reliable predictor of seedling abundance than the
two moss types combined. This may be because very thick layers of moss reduce

germination of seeds (Harmon and Franklin 1989) or prevent seedling roots from

76

accessing nutrients and water below the moss mat. When thin and thick moss cover were
combined, hemlock logs had a greater average percent cover of moss than did hardwood
logs (Table 3.2), which may again contribute to their greater numbers of ﬁrst-year
hemlock seedlings and higher birch and hemlock seedling survival rates. Litter cover was
also greater on hemlock logs, and litter is apparently not as harmful to seedlings on logs
as it is to those on soil. 21% of established hemlock and birch seedlings in 2003 were
growing on litter, which covered 20% of log surface area. As with moss, depth of the
litter layer may be important (Peterson and F acelli 1992). Most logs had a thin surface
covering of hemlock needles and small pieces of hardwood leaves, rather than the thick
litter layer found on nearby areas of the forest ﬂoor.
Wood decay

Existing studies have estimated the decay rates of hemlock (Tyrrell and Crow
1994a), birch, and sugar maple (yellow birch and maple spp.: Arthur et al. 1993), but this
is the ﬁrst study to measure residence time of all three species using the same method and
in the same ﬁeld sites. The longer residence time of hemlock logs may beneﬁt saplings
by providing a sturdy substrate that supports sapling weight for at least a median of 54
rather than 21-24 years. Also, if well-decayed wood provides moisture and nutrients to
saplings even after it no longer supports their weight, which seems likely given the
horizontal habit of sapling and tree roots along and within logs, then hemlock logs should
support a greater proportion of hemlock and birch saplings than do hardwood species.
They may also support more of the hemlock saplings that capture gaps to ascend to the
canopy, since such saplings are an average of 149 years old (Dahir 1994).

In addition to spending more years on the ground, hemlock logs tend to be

decayed by brown-rot fungi rather than by white-rot fungi or both types. Rot type was

77

one of the most consistently important factors in our models, sometimes remaining
signiﬁcant even when both seedling history and ﬁrst-year seedlings were in the model.
When log species and then rot type were entered into negative binomial regression
models, log species explained additional variation only in a single model. There are
several reasons that brown-rotted (hemlock) wood may be so favorable for seedlings.
One possibility is that brown-rotted wood remains structurally sound and provides a good
rooting medium for seedlings. A second possibility is that because brown-rot fungus
decays wood at a faster rate than it can consume the products, at least initially, seedlings
on brown-rotted wood have greater access to nutrients than seedling grown on white-
rotted wood where decay products are immediately consumed (Goodell et al. 2003). A
third possibility is that both decay fungi and seedlings respond to some other aspect of
hemlock wood, such as availability of micronutrients or wood pH, and that brown-rot
fungi and seedlings are most abundant under the same conditions.
Exogenous factors

Exogenous factors varied across ﬁeld sites and may have explained large-scale
variation in seedling abundance, but they rarely varied across species and were not
usually important in determining seedling abundance (Chapter 2) or mass. Although
water is important for the survival of both hemlock and birch seedlings, and may have
explained why average seedling density was greatest at our wettest ﬁeld site, wood water
content does not vary across wood species (Figure 3.1). In addition, water content is far
higher, even in summer, than that which causes birch seedlings to wilt in the greenhouse
(Figure 3.1). Attempts to obtain a water loss curve from wood by using a pressure plate
apparatus, which would be a more conclusive test of differences in water availability than

water content, failed. This is likely because wood has at least three separate curves

78

(water loss from vessels, pores, and between intact wood pieces, and possibly also
sapwood versus hardwood) that would be treated as a single curve by this technique (D.
Fredlund, pers. comm., 2004). Light and seed rain were important in determining the
numbers of ﬁrst-year seedlings, but like water content, they rarely varied across wood
species and most likely do not drive patterns of established seedling density (Chapter 2,
Table 3.3).
Ecological and management implications

The importance of hemlock wood for hemlock and birch seedling establishment
(Chapter 2) partly explains the stability of hemlock- and sugar maple-dominated patches
in primary hemlock-hardwood forests (Davis et a1. 1995) and also why hemlock forests
have not recovered after most were logged in the 18005 (Noss and Peters 1995). Our
results here show that, compared to sugar maple wood, the greater quality of decaying
hemlock wood as a seedling establishment substrate is associated with features of the
wood itself and not the microenvironment in which it is found. Likewise, we've shown
that sugar maple logs do not have the characteristics favored by hemlock and birch
seedlings regardless of where such logs are found. Thus hemlock and birch are unlikely
to establish in closed-canopy sugar maple forests unless decaying hemlock wood is
available as a substrate, and tend to either remain in hemlock-dominated stands where
decaying hemlock wood is more abundant or enter stands by germinating on mineral soil
after severe disturbance (Tubbs 1969, Erdmann 1990, Peterson 2000). Sugar maple
seedlings, on the other hand, are unable to use hemlock logs as seedling establishment
sites, and seedlings that germinate on logs become chlorotic and often die within the ﬁrst
several months (Figure 3.10). Sugar maple seedlings may be N-limited on wood; 2-year-

old sugar maples growing on logs have lower N contents than sugar maples growing on

79

adjacent soil (paired t-test, n = 9 pairs, p = 0.065, Figure 3.10). Hemlock wood therefore
has the additional advantage for hemlock and birch seedlings of being relatively clear of
sugar maple seedlings.

For areas where increasing or maintaining hemlock basal area is a management
goal, retention of hemlock decaying wood in particular is important. An attempt to seed
logs in the ﬁeld in 2003 failed due to either lack of seed germination or early mortality of
germinants, but even ﬁrst-year hemlock and birch seedlings are more abundant on
hemlock logs in our ﬁeld sites (Chapter 2). It seems likely, given our identiﬁcation of
brown-rot fungi, moss, pH, nutrient concentrations, and long residence time as wood
characteristics that inﬂuence seedling establishment, that other conifer wood species such
as cedar and white pine can also support hemlock and birch seedlings. Both white pine
logs and cedar logs were observed with seedlings in our ﬁeld sites (Marx, pets. obs.),
though they were too rare to examine statistically, and like hemlock, cedar also tends to
decay via brown-rot fungi, develop moss cover, and remain in the later decay stages for
long periods of time.

Acknowledgements

Thanks to Scott Kissman, Marcie Tidd, Laurie Gilligan, Melissa McDermott, and
Trent Thompson for their help in the ﬁeld and the lab. Gerald Mileski and Lynn
Holcomb gave advice and help with nutrient analysis methods. This work was supported
through a NSF Graduate Research Fellowship and Doctoral Dissertation Improvement
Grant to Laura Marx and by Michigan State University, the Huron Mountain Wildlife
Foundation, and the Michigan Botanical Club. We thank Kerry Woods for comments on

an earlier draft.

80

Table 3.1. Best-ﬁt explanatory models of numbers of seedlings occurring on each log at
the Porcupine Mountains. Chi-square and p—values are given for each signiﬁcant factor in
the model. Akaike information criteria (AICc) were used to select each best-ﬁt model.
Not all second-order interactions (and no higher-order interactions) were tested since
many interactions, especially those containing categorical variables such as rot type,
resulted in models that did not converge.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Year Response Model
2003 Hemlock seedling log size rot type
established history
seedlings
Chi-square 31.25 5.38 21.60
p-value (pr > x’) <0.0001 0.0203 <0.0001
seedling rot type ﬁrst-
history year
history
Chi-square 47.32 19.99 7.03
p-value (pt > x2) <0.0001 0.0002 0.008
2003 Birch seedling log size rot type light
established history
seedlings
Chi-square 21.38 8.13 19.65 6.44
p-value (pr > )8) <0.0001 0.004 0.0002 0.01 l 1
2004 Hemlock seedling log size seedling history x log
established history size
seedlings
Chi-square 21.60 17.64 9.21
p-value (pr > 76) <0.0001 <0.0001 0.0024
2004 Birch seedling log size wood Seedling history x log
established history species size
seedlings
Chi-square 14.39 17.33 9.75 8.16
p-value (pr > )8) 0.0001 <0.0001 0.0076 0.0043
2004 Hemlock ﬁrst- seed light % thin
year seedlings rain‘siz moss
e
Chi-square 12.12 8.58 8.02
p-value (pr > )8) 0.0005 0.0034 0.0046
2004 Birch ﬁrst-year seed % thin Seed rain x thin moss
seedlings rain per moss
trap
Chi-square 1.02 4.06 11.14
p-value (pr > 1’) 0.312 0.044 0.0008

 

 

 

 

 

 

 

 

 

 

81

Table 3.2. pH, seed rain, canopy openness, ﬁlngal rot type, and water content of hemlock,
birch, and sugar maple logs. Means are presented with 1 s.e. in parentheses. Means
followed by the same letter are not signiﬁcantly different (student's t-tests. all possible
contrasts, alpha = 0.05). Data from all sites were pooled for factors not signiﬁcantly
affected by ﬁeld site. N = 63 logs for pH; 138 logs with seed traps; 314 logs for canopy
openness (manual threshold); 271 logs for fungal rot type, and 318 logs for moss cover.

 

 

 

Porcupine Mts.

Hemlock Birch Sugar maple
pH 4.2 (0.08) a 5.2 (0.30) b 6&129) c
Hemlock seeds/m2
Hemlock |
Mixed stands
Huron Mts. 1,494(395)a l I,622(549)a | 2,366(373)a|
1,592(690)a 349(143)a 155(34)a

 

 

 

2,068(495)a | 3,382(582)a | I,283(336)a[
383(140)ab 2,375(l,446)a 230(48)b
Sylvania 421(88)a I 583(323)a | 658(242)a |
71 1(399)a 662(479)a 452(1 19)a
Birch seeds/m2
Huron Mts. 8,1 10(l,462)a I 8,246(1,424)a| 3,963(849)a!
963(526)a 1,289 (579)a l,386(310)a
Porcul’m Mts' 2,555(697)a I 2,720(603)a | 395(78)a |
. 8,215(3,099)a 11,743 (5,033)a 2,797(2,670)a
Sylvanla
3,963(849)a l 1,404(387)a I l ,451(490)a|
l,386(3 10):: 1,013 (203)a 470(71)a
Sugar maple 1 18(26)a | 55(13)a | l71(30)a |
seeds/m2 316(65)a 255(61)a 162(28)a
Canopy openness
Huron Mts. 7.05% (0.50)a 7.10% (0.57)a 8.38% (0.42)a
Porcupine Mts. 4.05% (0.09)a 4.18% (0.16)a 3.48% (0.14)b
Sylvania 3.22% (0.17)a 3.16% (0.10)a 3.84% (0.18)b

 

% of logs brown-
rotted

Huron Mts. (1 18)
Porcupine Mts. (82)
Sylvania (71)

74.4%
58.5%
23.1%

20.0%
16.6%
4.5%

10.2%
4.3%
0%

 

% total moss

 

 

 

 

 

cover

Huron Mts. 23.7% (4.46)a 15.9% (4.01)a 15.7% (2.90)a
Porcupine Mts. 48.9% (4.84)a 28.3% (6.96)b 25.6% (5.49)b
Sylvania 35.8% (8.07)a 20.7% (5.67)a 27.3% (4.1 1)a

 

82

Table 3.3. Effects of log species and light on seedling mass and N content. Values are p-
values from two—way ANOVA effect tests. Data were logarithm-transformed before
analysis. In all cases, the interaction between logs species and canopy openness was not
signiﬁcant (p > 0.250) and the term was removed from the ANOVA.

 

 

 

 

 

 

 

 

 

 

 

 

Seedling species Response Effect of log Effect of canopy
species openness

Betula, ﬁeld 2002 Mass 0.204 0.162

n = 17 logs N content 0.035 0.749
Betula, ﬁeld 2004 Mass 0.013 0.364

n = 33 N content 0.011 0.734
Betula, greenhouse Mass 0.064 --------

n = 60 N content 0.817 --------
Tsuga, ﬁeld 2002 Mass 0.921 0.049

n = 22 N content 0.735 0.210
Tsuga, ﬁeld 2004 Mass 0.734 0.081

n = 26 N content 0.673 0.027
Tsuga, greenhouse Mass 0.104 --------

n = 59 N content 0.220 --------

 

 

83

Table 3.4. Macronutrient percent of plant dry mass (g/g, Ca, N, P, K) and micronutrient
concentrations (ppm, B, Cu, Fe, Mg, Mn, Zn) in whole birch seedlings grown on three
species of wood in the greenhouse. Each value represents the mean nutrient
concentration of composite samples (4 entire birch seedlings from a single pot make up
each sample, N = 5 to 7 for each wood species), except for reference values, which are
the ratio of macronutrients to one another (N is set to 100) in Betula verrucosa seedlings
at optimum nutrition (Ingestad 1971), and the % dry mass that would be expected in our
data given these ratios and a %N (setpoint for the ratios) of 1.60. 1 s.e. is in parentheses.
* = Pearson's correlation between nutrient concentration and birch mass is signiﬁcant,
alpha = 0.05.

 

 

 

 

 

 

 

 

 

 

 

 

 

Hemlock Birch wood Sugar Betula Correlation
wood maple wood verrucosa with birch
ratios / mass
Expected % (species
dry mass pooled)
% dry mass of:
Calcium 1.35 (0.04)a 1.45 (0.11)a 1.55 (0.10)a 7 / 0.1 1 -0.254
Nitrogen 1.44 (0.12)a 1.60 (0.16)a 1.72 (0.16)a 100 = 1.60 -0.327
Phosphorus 0.12 (0.01)a 0.09 (0.01)a 0.13 (0.02)a I3 / 0.21 0.539“
Potassium 0.47 (0.04)a 0.53 (0.10)a 0.64 (0.10)a 65 / 1.04 -0.551*
ppm of:
Boron 0.12 (0.01)a 0.13 (0.01)a 0.12 (0.01)a 0.190
Copper 0.06 (0.01)a 0.06 (0.01)a 0.05 (0.01)a -0.723*
Iron 0.78 (0.11)a 1.00 (0.24)a 0.80 (0.16)a -0.922*
Magnesium 13.23 (0.79)a 13.47 (0.8l)a 12.38 (0.43)a -0.350
Manganese 1.84 (0.65)a 1.1 1 (0.28)a 0.78 (0.35)a 0.280
Zinc 0.47 (0.08)a 0.89 (0.11)b 0.39 (0.06)a -0.058

 

84

 

100

 

 

 

- Hemlock
- Biron
I: Sugarmaple

1;; O ____

tn 0 —

m __

E

E o

w- ‘0—

O

39

v

‘E

O

2 v“

C

O

O

h

2 o

2 ~—

0..

 

 

 

 

 

Field 2002 Field 2003 Birch wilting
Water content measurement

Figure 3.1. Field water content of wood pieces in 2002 and 2003 and water content of
wood measured at the point at which birch seedlings planted on each wood piece wilted
in the greenhouse. Means are presented with error bars representing 1 s.e. Field 2002 n =
239 logs, ﬁeld 2003 n = 172 logs, greenhouse birch n = 30 pots

85

 

 

 

 

«wt

 

 

 

 

 

 

Hemlock decay stage
4
l
Hemlock decay stage
4
1

Years since release Years since release

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0 0)

O) U)

E E

U) (D

g .- 113—: a ._ '43:

. was: t ~—:—-—u

5 l l l r l I l T 5 l I I l I l l I

E 10 30 50 70 a.) 10 30 50 70
Years since release Years since release

(D d)

U) U)

Q 9

(D (D

n .- ll a .. n£l:l—I

a ~~l:l}I % ~+——l:l:—~

%_ l 1 I l T l I I g l I I I l I I l

‘5“ 10 so 50 7o ‘2“ 1o 30 50 70

Years since release Years since release

Figure 3.2a and b. Ages of hemlock, birch, and sugar maple logs in decay stages 11 and
IV. a. (left) Conservative objective criteria for determining release. Release dates are the
average of two trees per log. n = 5 hemlock, 8 birch, and 9 maple logs. b. (right)
Subjective criteria for determining release. Release dates are averaged when both trees
for a single log showed a release, but included even if from a single tree. n = 10 hemlock,
26 birch, and 28 maple logs. Outer limits of each box represent the 25'“ and 75'h
percentile, while whiskers extend to 1.5 x the interquartile range.

86

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

m .. — I Nnin_2002 P g
I. - N_2002
E21 N_2oo4
Q —i —— u- ; A
>~
A _'_ 8
_.I
E n -1 — 2 D
3’ S
3 3
E N —1 ’- g .s
_1__, E
j Z
A
v- . AIL; '- 3
1-44..

 

 

  

 

 

 

 

 

 

 

 

 

 

Hemlock Birch

Substrate

Figure 3.3. Median rates of N mineralization (Nmin) and inorganic N concentrations
([N]) in wood in 2002 and 2004. Nmin is plotted on the secondary Y axis. Bars
represent medians, error bars represent the 25"I and 75'h percentiles. ANOVA tests
performed on logarithm-transformed data indicated that the [N] of hemlock logs was
lower than that of both birch and sugar maple logs in 2002, and lower than that of sugar
maple logs in 2004. n = 128 wood samples in 2002 and 129 in 2004. Note that this is a
larger sample than the subset from which 2-year-old seedlings were collected and used
to test seedling/N supply relationships.

87

 

1

0.020

I

Bitch mass (9. ﬁeld)
0 010

 

l

 

 

 

0.20 0.000

I

0.10

 

Birch mass (9. greenhouse)

0.00

in

 

 

Hemlock Bitch Sugar‘maple

 

0.010 0.020

J

Birch N content (9, ﬁeld)

 

 

 

0.000

 

l

l

l

1

l

1

 

1

Q

 

 

Birch N content (9, greenhouse)
000 010 020 030

Hemlock Bifch Sugar'maple

 

J J l I

J

1

Hemlock mass (9, ﬁeld)

 

I.

 

 

J

l

I

I

l

l

L

 

Hemlock mass (9. greenhouse)

 

0.000 0.010 0.020 0.030 0.000 0.010 0.020 0.030

Hemlock Bitch Sugar'maple

 

0.020 0.035

I

 

Hemlock N content (9, ﬁeld)

 

0.005

 

0.015 0.030

I

 

Mn

 

Hemlock N content (9. greenhouse)
0.000

Hemlock Bifch Sugar'maple

Figure 3.4. a. Median seedling mass, and b. Median N content, of birch and hemlock
seedlings growing on three different wood species. n = Birch: 50 logs with birch
seedlings in the ﬁeld and 60 in the greenhouse; Hemlock: 48 and 59.

88

 

 

 

 

Bichmass(g,ﬁold)

Birch N content (mg. ﬁeld) Birch mass (9. greenhouse)

Bren N content (mg, greenhouse)

 

0.020

0.010

L

 

 

 

 

0.10

0.20 0.000

 

 

 

 

 

 

 

 

 

 

 

8.. — — -

0 # I I
Hemlock Birch Sugar maple

a .

o .—

S!

d .

8_ _

o I l I
Hemlock Birch Sugar maple

In

N .

°. . _

N

In. ‘ _

q . _

“2 T Q

0

O. _ — .—

O I l l
Hemlock Birch Swat mapb

Hemlock mass (9. Md)
0.000 0.010 0.020 0.030 0.000 0.010 0.020 0.030

Hemlock mass (g, greenhouse)

Hemlock N content (mg, ﬁeld)

Hemlock N content (mg, greenhouse)

0.25 0.35

0.15

0.20 0.30 0.40 0.05

0.10

 

l

1

l

l

l

l

 

l

 

 

l

l

l

l

1

l

 

l

 

Henllock

Bifch SugarTmapIa

 

1

1

l

l

 

%

é

 

%- lI:]]—l

Birch

Sugar maple

 

l l l l l

l

 

 

T

l'bmbck

Birch

Swat mapb

Figure 3.4. a. Median seedling mass (g), and b. Median N content (mg), of birch and
hemlock seedlings growing on three different wood species. n = Birch: 50 logs with

birch seedlings in the ﬁeld and 60 in the greenhouse; Hemlock: 48 and 59.

89

 

 

 

 

 

Figure 3.6. Greenhouse seedlings showed great variability in size. a. Mycorrhizal birches
growing on a hemlock log. b. Non-mycorrhizal birches on a birch log. c. Fertilized
birch. d. Non-mycorrhizal hemlocks on a single sugar maple log.

 

Figure 3.10. F irst-year sugar maple seedlings growing on a log and on nearby soil. The
seedlings on the log (to the right of the dotted line) show signs of nutrient deﬁciency,
unlike the ﬁrst-year sugar maples on soil beside them.

90

 

 

 

 

 

 

 

 

 

 

 

0.4
+ Fertilized
—O— Unfertilized
“F

A 0.3 -

8’

(I)

§

or T

.E 0.2 ~

6

an

a)

U) .l-

.c

0

.~_-

“3 0.1 -
a

0.0 I I I
Hemlock Birch Sugar maple

Wood species

Figure 3.7. Response of birch seedlings to fertilization with N when grown on different
wood species. Symbols represent mean values of birch seedling mass, pooled by pot.
Error bars represent 1 s.e. n = 18 (6 hemlock, 8 birch, and 4 sugar maple) fertilized and

60 unfertilized pots (20 of each species).

91

 

22.5 o Hemlock —

. Sugar maple
20~ A . Yellow birch -— —

17.5“

15~

N:P

12.5~

10.

 

 

 

7.5

 

—

l l l l

0 .1 .2 .3 .4 .5 .6
Birch seedling mass (9)

Figure 3.8. Relationship between birch seedling N:P and mass (pooled mass of four birch
seedlings grown on the same piece of wood). n = 16.

92

Germinati on

 

 

 

V First-year
Seeds falling on log I seedings

Overwinter Year 1 i i Summer Year 1

 

 

 

 

 

 

 

% coverage

Light of thin moss i

 

 

 

 

 

 

 

 

Established seedlingsi
Summer Year 1

\ /\

Surface area of log 12:10:: mestygre

Established seedlings
Summer Year 2

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 3.9. Conceptual diagram of factors inﬂuencing ﬁrst-year and established seedling
survival on decaying wood. Factors signiﬁcant in best-fit models are indicated with solid
lines. The dashed line indicates a factor that could not be tested with our data.

93

CHAPTER FOUR
ADDITIONAL METHODS

Wood samples from ﬁeld plots were in many cases highly decayed (late stage IV).
I was able to identify even these pieces to species using methods taught to me by Alex
Wiedenhoeft at the US Forest Service Forest Products Laboratory in Madison, W]. I
selected the least-decayed piece of wood I could ﬁnd on each log as an ID sample.
Samples were transported from the field to the lab in sealed ziploc bags, because it is
critically important that decaying wood samples never be allowed to dry out. A dried
sample, unless it can be re-wetted which is not always possible, will not yield usable
slices for species identification. In the lab, I used GEM surgical-grade razor blades to cut
wet samples of wood into the thinnest slices possible, being careful to get a good radial,
tangential, and transverse section from each piece. Sections were mounted in a 1:1
mixture of ethanol and glycerine by putting sections in a small amount of solution on a
microscope slide, adding a cover slip, and then boiling slides on a hot plate at medium to
high heat until all of the ethanol and the water inside each wood section had boiled off.
Too high heat caused sections to move out from under the cover slip when the solution
boils and bubbles vigorously, and too low heat may not get rid of all of the water and
replace all of the air spaces in the wood with glycerine solution. A well-prepared slide
looks clean under the microscope; leftover air bubbles remaining from not enough boiling
time were easily visible. I used clear nail polish painted around the outside edge of each
cover slip to seal the slides, after which they could be stored vertically in a slide box and
should last many years without drying out. I found that I usually needed magniﬁcation of

at least 200x to see the finer characteristics of wood slides. Abies and Tsuga are often

94

indistinguishable from one another even at that magniﬁcation, and I used a high quality
microscope at as high as 400x at the Forest Products Laboratory to separate these species.
The problem is not one of making the features large enough to see, it is more of
resolution and depth of field, which is why a well-maintained and perfectly aligned
microscope is essential for identiﬁcation of these two species.

Key to wood species found in Upper Michigan hemlock forests
This key covers the following species: Tsuga canadensis, Abies balsamea, Picea glauca,
Thuja occidentalis, Pinus strobus, A cer saccharum, Acer rubrum, Betula spp. (it is not
possible to distinguish between B. papyrifera and allegheniensis with wood alone), Tilia
americana, Ostrya virginiana. Note that many of the characteristics listed here were

those I found most useful in identifying wood, and this is not a complete list.
Not included in the key, but present in my ﬁeld sites, are:

F raxinus and Quercus, both of which are ring porous and so will be immediately

distinguishable from the hardwoods listed above.

1. Does the wood have pores (cross-section or transverse section)?
Yes: Acer, Betula, Tilia, or Ostrya SEE hardwood table.

No: Tsuga, Abies, Picea, Thuja, or Pinus SEE conifer table.

2. Can’t tell from cross-section? Are there visible vessels or perforation plates on the

radial section?
Yes: Hardwood.

No: Conifer.

95

Table 4.1. Characteristics of hardwood wood for species identiﬁcation

 

 

 

 

 

 

 

Species Perforation Spiral Intervessel pit Ray size
plate thickenings size
Acer rubrum simple + NA l-S seriate
Acer saccharum simple + NA 1-3 and 5-8
seriate
Betula scalariform - tiny l-S
(many bars)
Tilia simple + small tall, laterally
compressed,
3-4 seriate
Ostrya simple and +? small, l-3 seriate
scalariform alternate, some
(few bars — books say
reverse medium to
footprint) large in size

 

 

 

 

 

 

Acer saccharum: Usually obvious from the cross section alone, since it is often possible
to pick out very thick rays and thinner rays in the same ﬁeld of view. Rays in tangential
section are, again, clearly of two different sizes. Spiral thickenings usually visible even

in rather poorly done radial sections.

Acer rubrum: More often identiﬁed by default than by any particular distinguishing
characteristic. Very easy to confuse with Ostrya or sometimes even Tilia, so be sure you
have three good sections to work with and are able to see all of the identifying

characteristics.

Betula: Perforation plates give this one away since none of the other species have
scalariform plates with so many bars (a striped appearance rather than just a few bars here
and there). Also, the tiny inter-vessel pitting is usually easy to see in the radial section.
Cross section shows rays larger than pores, and cross section looks very different from
maple but somewhat similar to Tilia and Ostrya. Oﬁen having bark samples makes

actually doing microscope IDs unnecessary for birch.

96

T ilia: In a fairly intact samples, the rays give this one away. Rays look like they have
been compressed sideways, with ray cells taller than they are wide. At ﬁrst glance rays
just look “wrong”. Be sure to check other characteristics as well (such as looking for
very prominent spiral thickenings?), though, since stressed wood can sometimes look like
basswood, and every once in a while there is a basswood that does not have extremely

compressed rays. This is easy to confuse with red maple if you’re not careful.

Ostrya: Like red maple, usually identiﬁed by default. The perforation plates are a deﬁnite
give-away, but are often very difﬁcult to see. The scalariform plates with few bars will
sometimes only have a single bar, creating a shape that almost looks like a reverse
footprint — instead of a large ball of foot section and small (think men’s shoes) heel, keep
the footprint as the same shape but have a line going halfway across the ball of foot
section, making a small toe part and a large heel. The cross section sometimes has pores

in multiples, looking a little different than maple.

Fraxinus and Quercus are ring porous, and oaks have HUGE rays, easily visible with a

hand-lens or even with the unaided eye.

Table 4.2. Characteristics of conifer wood for species identiﬁcation.

 

Resin canals YES

 

Epithelium Cross-ﬁeld pitting Ray tracheids
Picea thick walled piceoid +

Pinus (whites) thin walled fenestriform +

(looks completely
different from
piceoid and
taxodeoid)

 

 

 

Resin canals NO

 

Ray parenchyma end walls Cross-ﬁeld pittin Ray tracheids

 

 

 

 

 

 

 

 

Tsuga nodular piceoid +
Abies nodular taxodioid -
T hzg'a smooth taxodioid -

 

97

Picea: Looks like Tsuga but with resin canals.

Pinus: This one is fairly easy. Resin canals are fairly large, and the fenestriform pitting
looks different than any of the other species. Rays will appear to have large squarish
windows in them. Pine will often also be different in its decay pattern, since I have found
most of my pine samples as charred stumps or as a hollow stump with one or two pieces

of virtually intact wood attached.

T huja: Sometimes this will be easy since you can smell cedar. Again, decay
characteristics may help, but be careful: some cedar no longer looks stringy and is
mushier than you would expect. Cedar has no resin canals, and has very clearly smooth
ray parenchyma end walls. It tends, in my samples, to have a “neater” appearance than
other woods, in that the rays will be very clear, and your sections will look suspiciously
good. I have mis-identiﬁed cedar as hemlock in the past. If you really, truly, can’t see

any nodular end-walls, don’t invent them.

Tsuga and Abies: These two are very difficult to tell apart when decayed. Think of the

hem-ﬁr wood grade. If it’s that hard to tell the intact woods apart...

Tsuga has piceoid pits rather than taxodioid. Without a good microscope, piceoid
pits often just appear smaller than taxodioid. With a better one you will notice that the
taxodioid pits have sort of football-shaped openings while piceoid pits are narrower.
Tsuga also has biseriate bordered pitting. This means that as you look at the ﬁbers, you
will see lines of two pits going down each ﬁber, as opposed to Abies with just one line of
pits. BUT be careful that you are not actually looking at two uniseriate, but overlapping,
ﬁbers. Tsuga does have ray tracheids. These are on the outside of each set of ray

parenchyma, and the walls between them look different than the ray parenchyma end

98

walls. The ray parenchyma end walls have a lot of invaginations on them, while tracheid
end walls will often look like a line of X5, one on top of the other. Hoadley (1990) has a
great picture of this. Ray tracheids in T suga can be transparent and “scrappy” looking,
with wavy edges. Finally, in the radial section, you’ll see that Tsuga has strap-like
extensions of the torus on the bordered pits. This is only apparent with a very high-

powered microscope.

Abies, meanwhile, does not have ray tracheids. Look carefully, at a large section,
to make sure you’re notjust missing them. Abies also often has nodular side walls to its
ray parenchyma cells, notjust nodular end walls. 80 all walls of the cell have
invaginations, notjust the end walls. Abies lacks biseriate bordered pitting and torus
extensions. The most important section by far for this and for Tsuga will be the radial
section, so put a lot of these on the slide so you can try to be sure to get a good section.
It’s also important that these sections be thin. Often I am only able to use the very edges
of a section to see some of the more difﬁcult characteristics. Again, if you can use bark
to help you, do so. Extremely thick, reddish bark is probably not Abies. And very thin,

ﬁbrous bark is probably not Tsuga unless it came off of a branch.

Ilfll ﬁ'l'ﬁ'

Wood may be decayed by a number of organisms, including white rot fungi,
brown rot fungi, and soft rots. In my initial survey of decaying wood, I listed whether
wood was white-rotted or brown—rotted. This was based on a visual identiﬁcation of
wood characteristics. Brown-rotted wood was light to dark brown in color, and decayed
as discrete chunks, usually cubes or rectangles of wood. Brown-rotted wood often

snapped cleanly along vertical or horizontal breaks when it was pulled from the rest of the

99

log. White-rotted wood was distinguished by two sets of characteristics. White-rotted
wood was sometimes saturated with water and a mass of long soft strings. It was
possible, in highly decayed wood, to reach into a log and pull out a handful of these
strings, which would compress when squeezed, unlike brown-rotted wood. White-rotted
wood also sometimes looked very dry and was solid to the touch. The top surface of the
wood was occasionally black, but unlike charcoal did not leave black streaks on hands or
other surfaces. In this “dry” form of white rot, the wood under this hard surface showed

the characteristic form of white rot with wet strings throughout.

Several logs were brown-rotted on one end and white-rotted on the other, and
early decay stage logs often did not yet show signs of either fungal type. Soﬁ rots, stains,
and bacterial decay were not recorded in the ﬁeld, although stains were occasionally
present on sugar maple and oak logs found in ﬁeld plots, turning the wood a blueish
green color. Soft rots may have been present in my ﬁeld sites. These fungi degrade the
outside layers of wood, and often decay wood also simultaneously attacked by white rot
fungi (Goodell et al. 2003). It is possible that some of the early decay stage logs I
classiﬁed as not yet being attacked by fungi were in fact being attacked by soﬁ rots, since

soft rots attack the outside layers of wood but leave it relatively hard and intact.

100

CHAPTER FIVE
CONCLUSIONS

In Chapter 2, I demonstrated the restriction of established hemlock and birch
seedlings to decaying wood and in particular to hemlock and birch wood. First-year
seedlings are found on a broad range of substrates, and though rare are even found on
soil. By the fourth growing season, however, hemlock and birch are found almost
exclusively on hemlock and birch wood. Hemlock wood is a better substrate than birch
wood in most cases, especially for survival of hemlock seedlings. This may help
maintain the distribution of adult hemlock, birch, and sugar maple seedlings and the close
spatial association between hemlock and birch seedlings, saplings, and canopy trees. The
tendency of hemlock and birch seedlings to grow on wood is not enough to explain the
species association at either the stand level or the individual tree level. Their restriction
to hemlock wood, however, makes this an almost complete explanation. The one
remaining question is why paired hemlock and birch trees are the same diameter. One
reason may be that they grew at the same rate, possibly sharing nutrients with adult trees
or each other through the same mycorrhizal network. This could be tested by coring tree
pairs and comparing their growth rings. Likewise, the restriction of hemlock and birch
seedlings to wood types more abundant in hemlock-dominated than in sugar maple-
dominated stands ﬁlls a gap in the existing explanations for maintenance of the
hemlock/hardwood patch structure.

Explanations for the suitability of hemlock wood for seedling establishment were
explored in Chapter 3. Although some results are preliminary and will need further
study, hemlock wood: 1) remains in the decay stages that support the greatest numbers of

seedlings for a longer period of time than birch or sugar maple wood, 2) has a more

101

favorable pattern of decay, usually developing a layer of moss over brown-rotted wood,
3) provides a balance of nitrogen and phosphorus, in addition to providing sufﬁcient
phosphorus for seedling growth, 4) has a low pH, which may be related to its tendency to
be attacked by brown rot and its ability to provide phosphorus to seedlings. Although the
seedlings I studied were rarely mycorrhizal, there are several results in these chapters that
suggest that hemlock may be better able to provide mycorrhizal inoculum to seedlings
than sugar maple wood, and possibly birch wood. Mycorrhizae may also be one of the
reasons that sugar maple seedlings have such poor survival on wood.

In the introduction to this dissertation, I contrasted how easily hemlock seedlings
could be killed with the stability of stands of adult hemlocks. Adelges tsugae, hemlock
woolly adelgid, has destroyed many of the hemlock forests in New England and along the
east coast, and seems likely to kill both adults and young seedlings and saplings of
hemlock since it prefers new growth. Barring a major advance in control of the adelgid,
it will likely reach Michigan's hemlock forests at some point in the next decade, and the
old-growth systems studied here will lose a large proportion of their hemlock trees. The
results presented here suggest that leaving dead and downed hemlock trees in place will
be helpful in eventual restoration efforts, assuming that they don't provide a refuge for
adelgid. Additional research into the use of yellow birch as “placeholders” for hemlock
would be of interest, since birch logs can support at least young hemlock seedlings, and
birch and hemlock can coexist in close proximity. Any approach will need to exclude
white-tail deer or greatly reduce deer densities. Hemlock restoration efforts would be an
interesting test of some of the ideas about hemlock and birch regeneration and the

interdependence of hemlock and birch presented here.

102

LITERATURE CITED

Albert, Dennis A. 1995. Regional landscape ecosystems of Michigan, Minnesota, and
Wisconsin: a working map and classiﬁcation. Gen. Tech. Rep. NC-178. St. Paul,
MN: USDA Forest Service, North Central Forest Experiment Station.
http://www.npwrc.usgs.gov/resource/ 1998/rlandscp/rlandscp.htm.

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

Anderson H. W. and Gordon A. G. 1994. Hemlock: Its ecology and management. pp. 62-
114. Min. Nat. Resourc., Ont. For. Res. Inst., For. Res. Info. Pap. No. 113.

Anderson R. C. and Loucks O. L. 1985. White-tail deer (Odocoileus virginianus)
inﬂuence on structure and composition of Tsuga canadensis forests. J. Appl.
Ecology 16: 855-861.

Arthur M.A., Siccama T.G. and Yanai RD. 1999. Calcium and magnesium in wood of
northern hardwood forest species: relations to site characteristics. Can. J. For. Res.
29: 339-346.

Arthur M. A., Tritton L. M., and Fahey T. J. 1993. Dead bole mass and nutrients
remaining 23 years after clear-felling of a northern hardwood forest. Can. J. For.
Res. 23: 1298-1305.

Baldocchi D. and Collineau S. 1994. The physical nature of solar radiation in
heterogeneous canopies: spatial and temporal attributes. In: Exploitation of
environmental heterogeneity by plants (eds M. M. Caldwell and R. W. Pearcy) pp.
21-72. Academic Press, San Diego, Calif.

Barden, LS. 1979. Tree replacement in small canopy gaps of a Tsuga canadensis forest
in the southern Appalachians, Tennessee. Oecologia 44: 141-142.

Binkley and Hart S. C. 1989. The components of nitrogen availability assessments in
forest soils. Advances in Soil Science 10: 57-112.

Bockheim J .G. 1997. Soils in a hemlock-hardwood ecosystem mosaic in the Southern
Lake Superior Uplands. Can. J. For. Res. 27: 1147-1153.

Boddy L. 1983. Microclimate and moisture dynamics of wood decomposing in terrestrial
ecosystems. Soil Biol. Biochem. 15: 149-157.

Booth M. G. 2004. Mycorrhizal networks mediate overstorey-understorey competition in
a temperate forest. Ecology Letters 7: 538-546.

Bourdeau P. F. and Laverick M. L. 1958. Tolerance and photosynthetic adaptability to

103

light intensity in white pine, red pine, hemlock, and Ailanthus seedlings. Forest
Science 4: 196-207.

Brown R. T. and Curtis J. T. 1952. Upland conifer-hardwood forests of northern
Wisconsin. Ecological Monographs 22: 217-234.

Campbell J. L. and Gower S. T. 2000. Detritus production and soil N transformations in
old-growth eastern hemlock and sugar maple stands. Ecosystems 3: 185-192.

Canavera, D. 1978. Effects of various growing media on container-grown yellow birch.
Tree Planters' Notes Winter: 12-14.

Canham C. D. and Loucks O. L. 1984. Catastrophic windthrow in presettlement forests of
Wisconsin. Ecology 65: 803-809.

Casperson, J .P. and Kobe, R.K. 2001. Interspeciﬁc variation in sapling mortality in
relation to growth and soil moisture. Oikos 92: 160-168.

Casperson, J .P. and Saprunoff, M. 2005. Seedling recruitment in a northern temperate
forest: the relative importance of supply and establishment limitation. Can. J. For.
Res. 35: 978-989.

Catovsky, S. and Bazzaz, RA. 2000. The role of resource interactions and seedling

regeneration in maintaining a positive feedback in hemlock stands. Journal of
Ecology 88: 100-112.

Catovsky S. and Bazzaz F. A. 2002. Nitrogen availability inﬂuences regeneration of
temperate tree species in the understory seedling bank. Ecological Applications
12: 1056-1070.

Christy E. J. and Mack R. N. 1984. Variation in demography of juvenile T suga
heterophylla across the substratum mosaic. Journal of Ecology 72: 75-91.

Christy E. J ., Sollins P., and Trappe J. M. 1982. First-year survival of Tsuga heterophylla
without mycorrhizae and subsequent ectomycorrhizal development on decaying
logs and mineral soil. Can. J. Bot. 60: 1601-1605.

Coffman M. S. 1978. Eastern hemlock germination inﬂuenced by light, germination
media, and moisture content. Mich. Bot. 17: 99-103.

Cole, D.W. and Rapp, M. 1981. Elemental cycling in forest ecosystems. In Dynamic
properties of forest ecosystems. (ed D.E. Reichle) Cambridge University Press,
Cambridge. pp. 341-409.

Comer, P. J., D. A. Albert, and M. B. Austin. 1998. Vegetation of Michigan circa 1800:
an interpretation of the general Land Ofﬁce Surveys. Natural Heritage Program,
Wildlife Division, Michigan Department of Natural Resources, Lansing, MI.

104

Corinth R. L. 1995. Coarse woody debris and regeneration of eastern hemlock. (eds G. D.
Mroz and J. Martin) pp. 193-194. University of Wisconsin-Madison, Madison.

Comett M. W., Reich P. B., and Puettmann K. J. 1997. Canopy feedbacks and
microtopography regulate conifer seedling distribution in two Minnesota conifer-
deciduous forests. Ecoscience 4(3): 353-364.

Comett M.W., Reich P.B., Puettmann K.J. & Frelich LE. 2000. Seedbed and moisture
availability determine safe sites for early Thuja occidentalis (Cupressaceae)
regeneration. American Journal of Botany 87: 1807-1814.

Comett, M.W., Puettmann, K.J., F relich, LE, and Reich, RB. 2001. Comparing the
importance of seedbed and canopy type in the restoration of upland T huja
occidentalis forests of northeastern Minnesota. Restoration Ecology 9: 386-396.

Crabtree R. C. and Bazzaz F. A. 1993. Seedling response of four birch species to
simulated nitrogen deposition: ammonium versus nitrate. Ecological Applications
3: 315-321.

Crow, T.R. 1995. The social, economic, and ecological signiﬁcance of hemlock in the
Lake States. In: Hemlock Ecology and Management: Conference Proceedings,
September 27-28, 1995, Iron Mountain, MI (eds Mroz G.D. & Martin J .),
University of Wisconsin-Madison, Madison. pp. 11-17.

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

Dahir S. E. 1994. Tree mortality and gap formation in old-growth hemlock-hardwood
forests of the Great Lakes Region. MS Thesis. Madison, WI, University of
Wisconsin-Madison.

Daubenmire, RP. 1930. The relation of certain ecological factors to the inhibition of
forest ﬂoor herbs under hemlock. Butler Univ. Bot. Stud. 1: 61-76.

Davis M. B., Calcote R. R., Sugita S., and Takahara H. 1998. Patchy invasion and the
origin of a hemlock-hardwoods forest mosaic. Ecology 79: 2641-2659.

Davis M. B., Sugita S., Calcote R. R., Ferrari J. B., and Frelich L. E. 1993. Historical
development of alternate communities in a hemlock-hardwood forest in northern
Michigan, USA. (eds P. J. Edwards, R. M. May, and N. R. Webb) pp. 19-39.
Blackwell Scientiﬁc Publications, Oxford.

Davis M. B., Sugita S., Calcote R. R., Parshall T. E., and Ferrari J. B. 1995. 3000 years of

abundant hemlock in northern hardwood forests. (eds G. D. Mroz and J. Martin)
pp. 19-28. University of Wisconsin, Madison, Madison.

105

Dehayes, D.H., Waite, CE, and Hannah, PR. 1980. Growth of yellow birch seedlings
from eight provenances in varying growth media. Tree Planters' Notes Fall: 12-16.

Dickie, LA. and Reich, RB. 2005. Ectomycorrhizal fungal communities at forest edges.
Journal of Ecology 93: 244-255.

Dirzo R, Horvitz C.C., Quevdo, H., and Lopez, MA. 1992. The effects of gap size and
age on the understory herb community of a tropical mexican rain-forest. Journal
of Ecology 80: 809-822.

Eckstein R. G. 1980. Eastern hemlock (Tsuga canadensis) in north central Wisconsin.
Research Report 104. Wisconsin Dept. of Natural Resources, Madison, WI
http://digital. library. wisc.edu/1 71 I.dl/EcoNatRes. DNRRep104.

Erdmann G. G. 1990. Betula allegham'ensis Britton Yellow Birch. In Silvics of North
America: 2. Hardwoods (eds R. M. Burns and B. H. Honkala) US For. Serv.
Agric. Handbk. 654.

Ericsson T. and Ingestad T. 1988. Nutrition and growth of birch seedlings at varied
relative phosphorus addition rates. Physiol. Plant. 72: 227-235.

Eyres EH. 1980. Forest Cover Types of the United States and Canada. Society of
American Foresters, Washington, DC.

Ferrari, J .B. 1999. Fine-scale patterns of leaf litterfall and nitrogen cycling in an old-
growth forest. Can. J. For. Res. 29:291-302.

Finkelstein, S. 2005. Postdoctoral researcher, University of Toronto.

F inzi A. C. and Canham C. D. 2000. Sapling growth in response to light and nitrogen
availability in a southern New England forest. Forest Ecology and Management
131: 153-165.

Finzi, A.C., Breemen, N.V., and Canham, C.D. 1998a. Canopy tree-soil interactions
within temperate forests: species effects on soil carbon and nitrogen. Ecological
Applications 8: 440-446.

Finzi A. C., Canham C. D., and Breemen N. V. 1998b. Canopy tree-soil interactions
within temperate forests: species effects on pH and cations. Ecological
Applications 8: 447-454.

Forcier, L.K. 1975. Reproductive strategies and the co-occurrence of climax tree species.
Science 189: 808-810.

Forest Service U. S. D. A. 1965. Silvics of forest trees of the United States. (comp. H.A.
Fowells) USDA, Agriculture Handbook 271. Washington, DC.

106

Fredlund, D. February 10, 2004. pers. comm. Physicist and engineer, University of
Saskatchewan.

Frelich L. E. and Graumlich L. J. 1994. Age-Class Distribution and Spatial Patterns in an
Old-Growth Hemlock Hardwood Forest. Can. J. For. Res. 24: 1939-1947.

Frelich L. E., Calcote R. R., Davis M. B., and Pastor J. 1993. Patch formation and
maintenance in an old-growth hemlock-hardwood forest. Ecology 74(2): 513-527.

F riesner R. C. and Potzger J. E. 1936. Soil moisture and the nature of the Tsuga and
Tsuga-Pinus forest associations in Indiana. Butler Univ. Bot. Stud. 3:207-209.

Friesner, RC. and Potzger, J .E. 1944. Survival of hemlock seedlings in a relict colony
under forest conditions. Butler Univ. Bot. Stud. 6: 102-115.

Goder H. A. 1961. Hemlock reproduction and survival on its border in Wisconsin.
Transactions of the Wisconsin Academy of Sciences, Arts, and Letters 50: 175-
181.

Godman R. M. and Lancaster K. 1990. Tsuga canadensis (L.) Carr. Eastern hemlock.
(eds R. M. Burns and B. H. Honkala) pp. 604-612. US For. Serv. Agric. Handbk.
654.

Godman, R.M., Yawney, H.W., and Tubbs, CH. 1990. Acer saccharum Marsh. Sugar
maple. (eds R. M. Burns and B. H. Honkala) US For. Serv. Agric. Handbk. 654.

Goodell 3., Nicholas DD, and Schultz, T.P. (eds) 2003. Wood Deterioration and
Preservation: Advances in Our Changing World. ACS Symposium Series 845.
American Chemical Society, Washington, DC.

Graham R. L. and Cromack K. Jr. 1982. Mass, nutrient content, and decay rate of dead
boles in rain forests of Olympic National Park. Can. J. For. Res. 12: 511-521.

Graham S. A. 1941. The question of hemlock establishment. Journal of Forestry 39: 567-
569.

Gray, AN. and Spies, T.A. 1997. Microsite controls on tree seedling establishment in
conifer forest canopy gaps. Ecology 78(8): 2458-2473.

Grenon F., Bradley R.L., Jones M.D., Shipley B. and Peat H. 2005. Soil factors
controlling mineral N uptake by Picea engelmannii seedlings: the importance of
gross NH4+ production rates. New Phytologist 165: 791- 800.

Grier C. C. 1978. A T suga heterophylla-Picea sitchensis ecosystem of coastal Oregon:
decomposition and nutrient balances of fallen logs. Can. J. For. Res. 8: 198-206.

Hadley J. L. 2000. Understory microclimate and photosynthetic response of saplings in an
old-growth eastern hemlock (T suga canadensis L.) forest. Ecoscience 7: 66-72.

107

Harlow S. 1900. Roots of the hemlock. Journal of the New York Botanical Garden 1:
100-101.

Harmon M. E. 1989. Retention of needles and seeds on logs in Picea sitchensis-Tsuga
heterophylla forests of coastal Oregon and Washington. Can. J. Bot. 67: 1833-
1837.

Harmon M. E., Franklin J.F., Swanson F J., Sollins P., Gregory S.V., Lattin J.D.,
Anderson N.H., Cline S.P., Aumen N.G., Sedell J.R., Lienkaemper G.W.,
Cromack K. Jr., and Cummins K.W. 1986. Ecology of coarse woody debris in
temperate ecosystems. Adv. Ecol. Res. 15: 133-302.

Harmon M. E. and Franklin J. F. 1989. Tree seedlings on logs in Picea-Tsuga forests of
Oregon and Washington. Ecology 70: 48-59.

Hart S. C. 1999. Nitrogen transformations in fallen tree boles and mineral soil of an old-
growth forest. Ecology 80: 1385-1394.

Hoadley, RB. 1990. Identifying Wood: Accurate Results With Simple Tools. The
Taunton Press, Newton, CT.

Holub S. M., Spears J. D., and Lajtha K. 2001. A reanalysis of nutrient dynamics in
coniferous coarse woody debris. Can. J. For. Res. 31: 1894-1902.

Houle G. and Payette S. 1990. Seed dynamics of Betula alleghaniensis in a deciduous
forest of north-eastem North America. Journal of Ecology 78: 677-690.

Hough AF. 1936. A climax forest community on East Tionesta Creek in Northwestern
Pennsylvania. Ecology 17: 9-28.

Hura, CE. and Crow, T.R. 2004. Woody debris as a component of ecological diversity in
thinned and unthinned northern hardwood forests. Natural Areas Journal 24: 57-
64.

lngestad, T. 1971. A deﬁnition of optimum nutrient requirements in birch seedlings, II.
Physiol. Plant. 24:118-125.

Klironomos J. N. 1995. Arbuscular mycorrhizae of Acer saccharum in different soil
types. Can. J. Bot. 73: 1824-1830.

Klute, A.ed. 1986. Methods of soil analysis: Part 1 - Physical and mineralogical methods.
2nd edn. American Society of Agronomy and Soil Science Society of America,
Madison, WI.

Kobe R.K., Pacala, S.W., Silander, J .A., and Canham CD. 1995. Juvenile tree
survivorship as a component of shade tolerance. Ecological Applications 5: 517-
532.

Koroleff 1954. Leaf litter as a killer. Journal of Forestry 52: 178-182.

108

Kotar, J. 2003. University of Wisconsin-Madison, pers. comm.

Kotar J ., Kovach J. A., and Brand G. 1999. Analysis of the 1996 Wisconsin forest
statistics by habitat type. Gen. Tech. Rep. NC-207. Saint Paul, MN: US
Department of Agriculture, Forest Service, North Central Research Station. 166

PP-

Kotar, J ., Kovach, A., and Loecy, CT. 1988. Field guide to forest habitat types of
northern Wisconsin. Department of Forestry, University of Wisconsin-Madison,
Madison, WI.

Kropp B. R. 1982a. Formation of mycorrhizae on nonmycorrhizal western hemlock
outplanted on rotten wood and mineral soil. Forest Science 28: 706-710.

Kropp B. R. 1982b. Fungi from decayed wood as ectomycorrhizal symbionts of western
hemlock. Can. J. For. Res. 12: 36-39.

Kropp B.R. 1982c. Rotten wood as mycorrhizal inoculum for containerized western
hemlock. Can. J. For. Res. 12: 428-431.

Kytoviita M. M. and Amebrant K. 2000. Effects of inorganic nitrogen and phosphorus
supply on nitrogen aquisition from alanine by birch seedlings in symbiosis with
Paxillus involutus. Plant and Soil 219: 243-250.

Lee, P. and Sturgess, K. 2001. The effects of logs, stumps, and root throws on understory
communities within 28-year-old aspen-dominated boreal forests. Can. J. For. Res.
79: 905-916.

Leopold A. 1963. Report on Huron Mountain Club (1938). Huron Mountain Club, Big
Bay, MI.

Linteau A. 1948. Factors affecting germination and early survival of yellow birch (Betula
lutea Michx.) in Quebec. Forestry Chronicle 24: 27-86.

Lorimer, CG. and Frelich LE. 1989. A method for estimating canopy disturbance
frequency and intensity in dense temperate forests. Can. J. For. Res. 19: 651-663.

Machado, J .L. and Reich, PB. 1999. Evaluation of several measures of canopy openness
as predictors of photosynthetic photon ﬂux density in deeply shaded conifer-
dominated forest understory. Can. J. For. Res. 29: 1438-1444.

 

Maissurow D. K. 1941. The role of ﬁre in the perpetuation of virgin forests of northern
Wisconsin. Journal of Forestry 39: 201-207.

Margolis H. A. and Vezina L. P. 1988. Nitrate content, amino acid composition, and
growth of yellow birch seedlings in response to light and nitrogen source. Tree
Physiology 4: 245-253.

109

 

Martin, J. and Lorimer, CG. 1996. How to Manage Northern Hardwoods. University of
Wisconsin Forestry Extension Publication #81.

McClure S. J. 1992. Effects of implanted and injected pesticides and fertilizers on the
survival of Adelges tsugae (Homoptera: Adelgidae) and on the growth of Tsuga
canadensis. Journal of Economic Entomology 85: 468-472.

Mchen A. B. and Curran L. M. 2004. Seed dispersal and recruitment limitation across
spatial scales in temperate forest fragments . Ecology 85: 507-518.

McGee, G.G. 2001. Stand-level effects on the role of decaying logs as vascular plant
habitat in Adirondack northern hardwood forests. J. Torrey Bot. Soc. 128: 370-
380.

Mchen A. B. 2002. Seed dispersal and distributions of woody plants across temperate
forest fragments, PhD dissertation. University of Michigan.

McIntosh R. P. 1972. Forests of the Catskill Mountains, New York. Ecological
Monographs 42: 143-161.

Metzger F .T. 1977. Sugar maple and yellow birch (Acer saccharum, Betula
alleghaniensis) seedling growth after simulated browsing (damage). USDA Forest
Service Research Paper 140, North Central Forest Experiment Station.

Mladenoff, DJ. 1987. Dynamics of nitrogen mineralization and nitriﬁcation in hemlock
and hardwood treefall gaps. Ecology 68: 1171-1180.

Mladenoff D. J. and Steams F. 1993. Eastern hemlock regeneration and deer browsing in
the northern Great Lakes Region: A re-examination and model simulation.
Conservation Biology 7: 889-900.

Mori, A., Mizumachi, E., Osono, T., and Doi, Y. 2004. Substrate-associated seedling
recruitment and establishment of major conifer species in an old-growth subalpine
forest in central Japan. Forest Ecology and Management 196: 287-297.

Nelson C. R. 1997. Hemlock regeneration on the Nicolet National Forest, Wisconsin.
MS. Thesis. University of Wisconsin-Madison.

Neumann, D. and Peterson G. 2001. Northern Hardwood Forest Management. Michigan
State University Extension Bulletin E2769. East Lansing, MI.

Newberry, J.E. 2001. Small scale disturbance and stand dynamics in Inonotus tomentosus
infected and uninfected old-growth and partial cut wet, sub-boreal forests in
British Columbia. MS. Thesis. University of Northern British Columbia.

Noss, R.F. and Peters, R.L. 1995. Endangered Ecosystems: A Status Report on America's
Vanishing Habitat and Wildlife. Defenders of Wildlife, Washington, DC, pg. 82.

110

O'Hanlon-Manners D. L. and Kotanen P. M. 2004. Logs as refuges from fungal pathogens
for seeds of eastern hemlock (Tsuga canadensis). Ecology 85: 284-289.

Pacala, S.W., Canham, C.D., Silander, J .A., and Kobe R.K. 1993. Sapling growth as a
function of resources in a north temperate forest. Can. J. For. Res. 24: 2172-2183.

Page, A.L., Miller, RH, and Keeney, D.R. (Eds.) 1982. Methods of soil analysis. Part 2.
Chemical and microbiological properties. American Society of Agronomy,
Madison, WI.

Pastor J. and Broschart M. 1990. The spatial pattern of a northern conifer-hardwood
landscape. Landscape Ecology 4: 55-68.

Perez-Moreno J. and Read D. J. 2000. Mobilization and transfer of nutrients from litter to
tree seedlings via the vegetative mycelium of ectomycorrhizal plants. New Phytol.
145: 301-309.

Persson J. & Nasholm T. 2001. Amino acid uptake: a widespread ability among boreal
forest plants. Ecology Letters 4: 434-438

Peterson, C.J. 2000. Damage and recovery of tree species after two different tornadoes in
the same old growth forest: a comparison of infrequent wind disturbances. Forest
Ecology and Management 135: 237-252.

Peterson C. J. and Facelli J. M. 1992. Contrasting germination and seedling growth of
Betula alleghaniensis and Rhus typhina subjected to various amounts and types of
plant litter. American Journal of Botany 79: 1209-1216.

Powers, R.F. 1980. Mineralizable soil nitrogen as an index of nitrogen availability to
forest trees. Soil Sci. Soc. Am. J. 44: 1314-1320.

Pregitzer, K.S., Barnes, B.V., and Lemme, GD. 1983. Relationship of topography to soils
and vegetation in an Upper Michigan ecosystem. Soil Sci. Soc. Am. J. 47: 117-
123.

Pyle and Brown 1998. Hardwood log decay classiﬁcation. Journal of the Torrey Botanical
Society 125: 237-245.

Raynal D. J., Roman J. R., and Eichenlaub W. M. 1982. Response of tree seedlings to
acid precipitation. I. Effect of substrate acidity on seed germination. Environ. Exp.
Bot. 22: 377-383.

Reif C. B. 1992. Specimens of Betula alleghaniensis and Betula Ienta growing from
century-old stumps. Journal of the Pennsylvania Academy of Science 66: 116-
122.

Robertson G.P. and Vitousek RM. 1981. Nitriﬁcation potentials in primary and
secondary succession. Ecology 62: 376-386.

111

Rogers R. S. 1980. Hemlock stands from Wisconsin to Nova Scotia: Transitions in
understory composition along a ﬂoristic gradient. Ecology 61: 178-193.

Rogers, RS. 1978. Forest dominated by hemlock (Tsuga canadensis): distribution as
related to site and post-settlement history. Can. J. Bot. 56: 843-854.

Rogers, L.L., Mooty, J.J., and Dawson, D. 1981. Foods of white-tailed deer in the Upper

Great Lakes Region -- A review. USDA Forest Service General Technical Report
NC-65.

Rooney T. P. and Waller D. M. 1998. Local and regional variation in hemlock seedling
establishment in forests of the upper Great Lakes region, USA. Forest Ecology
and Management 111: 211-224.

Rooney, T.P., McCormick, R.J., Solheim, S.L., and Waller, D.M. 2000. Regional
variation in recruitment of hemlock seedlings and saplings in the upper Great
Lakes, USA. Ecological Applications 10: 1 119-1132.

Scherbatskoy T., Klein R. M., and Badger G. J. 1987. Germination responses of forest
tree seed to acidity and metal ions. Environmental and Experimental Botany. 27:
157—164.

Schwartz J.W., Nagel L.M. and Webster CR. 2005. Effects of uneven-aged management
on diameter distribution and species composition of northern hardwoods in Upper
Michigan. Forest Ecology and Management 21 1: 356-370.

Simard M. J., Bergeron Y., and Sirois L. 1998. Conifer seedling recruitment in a
southeastern Canadian boreal forest: the importance of substrate. Journal of
Vegetation Science 9: 575-582.

Simard M.-J., Bergeron Y., and Sirois L. 2003. Substrate and litterfall effects on conifer
seedling survivorship in southern boreal stands of Canada. Can. J. For. Res. 33:
672-681.

Simpson T.B., Stuart PE and Barnes B.V. 1990. Landscape ecosystems and cover types
of the reserve area and adjacent lands of the Huron Mountain Club. Big Bay, Ml,
Huron Mountain Wildlife Foundation. Occasional Papers of the Huron Mountain
Wildlife Foundation, No. 4.

Sivaramakrishnan S. and Berlyn GP. 1999. The role of site conditions in survival of
hemlocks infested with hemlock woolly adelgid: Amelioration through the use of
organic biostimulants. In Proceedings: Symposium on Sustainable Management
of Hemlock Ecosystems in Eastern North America, June 22-24, 1999. Durham,
NH. USDA Forest Service Northeastern Research Station General Technical
Report NE-267. (eds McManus K.A., Shields KS. and Souto D.R.).

Smith, EM. 1989. Fertilizing Landscape and Field Grown Nursery Crops. Extension
Bulletin 650. The Ohio State University, Columbus, OH.

112

Spears, J.D.H., Holub, S.M., Harmon, M.E., and Lajtha, K. 2003. The inﬂuence of
decomposing logs on soil biology and nutrient cycling in an old-growth mixed
coniferous forest in Oregon, USA. Can. J. For. Res. 33: 2193-2201.

Steams F. 1951. The composition of the sugar maple-hemlock-yellow birch association in
northern Wisconsin. Ecology 32: 245-265.

Strong, T.F. 1995. Hemlock regeneration by the shelterwood method: 20-year results. In
Hemlock Ecology and Management: Conference Proceedings, September 27-28,
1995, Iron Mountain, Ml. University of Wisconsin-Madison, Madison. pp. 187-
192.

Sutherland E. K., Hale B. J., and Hix D. M. 2000. Deﬁning species guilds in the Central
Hardwood Forest, USA. Plant Ecology 147: 1-19.

Takahashi M., Sakai Y., Ootomo R., and Shiozaki M. 2000. Establishment of tree
seedlings and water-soluble nutrients in coarse woody debris in an old-growth
Picea-Abies forest in Hokkaido, northern Japan. Can. J. For. Res. 30: 1148-1155.

Templer P. H. and Dawson T. E. 2004. Nitrogen uptake by four tree species of the
Catskill Mountains, New York: Implications for forest N dynamics. Plant and Soil
262: 251-261.

Thoreau H.D. 1854. Walden; or Life in the Woods. Ticknor and Fields, Boston, MA.

Tobin M. F. 2001. Mechanisms of patch maintenance in old-growth hemlock-hardwood
forests. PhD Dissertation. MN. University of Minnesota.

Tripler C. E., Canham C. D., Inouye R. S., and Schnurr J. L. 2002. Soil nitrogen
availability, plant luxury consumption, and herbivory by white-tailed deer.
Oecologia 133: 517-524.

Tubbs C. H. 1969. Natural Regeneration of Yellow Birch in the Lake States. pp. 74-78.
Forest Service, US. Department of Agriculture, Upper Darby, PA.

Tubbs C. H. 1995. Aspects of eastern hemlock silvics important in silviculture: An
overview. (eds G. D. Mroz and J. Martin) pp. 5-10. University of Wisconsin-
Madison, Madison.

Tubbs C. H. 1978. How to manage eastern hemlock in the Lake States. NC Forest
Extension Station, US Forest Service.

Tubbs C. H. 1969. The inﬂuence of light, moisture, and seedbed on yellow birch
regeneration. St. Paul, MN, North Central Forest Experiment Station, USDA
Forest Service.

113

 

Tyrrell L. E. and Crow T. R. 1994a. Dynamics of dead wood in old-growth hemlock

hardwood forests of northern Wisconsin and northern Michigan. Can. J. For. Res.
24: 1672-1683.

Tyrrell L. E. and Crow T. R. 1994b. Structural characteristics of old-growth hemlock-
hardwood forests in relation to age. Ecology 75: 370-3 86.

USDA, Forest Service. 1969. Northeastern Forest Experiment Station Birch Symposium
Proceedings. Forest Service, US. Department of Agriculture, Upper Darby, PA.

Walters MB. and Reich PB. 1997. Growth of Acer saccharum seedlings in deeply

shaded understories of northern Wisconsin: Effects of nitrogen and water
availability. Can. J. For. Res. 27: 237-247.

Walters M.B., Lajzerowicz C. and Coates K.D. In press. Soil resources and the growth
and nutrition of tree seedlings near harvest gap-forest edges in interior cedar-
hemlock forests, British Columbia. Can. J. For. Res.

Wilson, D. 2001. Forester, Michigan Department of Natural Resources, pers. comm.

Woods K.D. 1984. Patterns of tree replacement: canopy effects on understory pattern in
hemlock-northem hardwood forests. Vegetatio 56: 87-107.

Woods K. D. 2000. Dynamics in late-successional hemlock-hardwood forests over three
decades. Ecology 81: 110-126.

Woods, K.D. Interstand and intrastand pattern in hemlock-northem hardwood forests.
Ph.D. Thesis. 1981. Cornell University.

Yavitt J .B. and Fahey T.J. 1985. Chemical composition of interstitial water in decaying
lodgepole pine bole wood. Can. J. For. Res. 15: 1149-1153.

Zabel, RA. and Morrell, J .J. 1992. Chapter 5. Fungal metabolism in relation to wood
decay. Academic Press, Inc., San Diego, CA. pp. 116-135.

Zhong, J. and van der Kamp, B.J. 1999. Pathology of conifer seed and timing of
germination in high-elevation subalpine ﬁr and Engelmann spruce forests of the
southern interior of British Columbia. Canadian Journal of Forest Research 29:
187-193.

114

   

Itljl'lj‘lllljllgﬂljlll![111131