v

.5".-.
g!“ u...

a

inn}.

 

 

v

A

:3.

 

 

W LIBRARY
Michigan State
University

f:
.(
Loo /

 

This is to certify that the
dissertation entitled

DEER AND SEDGE EFFECTS ON SEEDLING
REGENERATION DYNAMICS IN NORTHERN TEMPERATE
FORESTS

presented by

Jesse Allen Randall

has been accepted towards fulfillment
of the requirements for the

 

 

Doctoral degree in Forestm

_-"')

    
 
 

L"fillii‘ajot’ii’rofésfi’s SigW
/’ 4/ , 07/

Date

MSU is an Affinnative Action/Equal Opportunity Institution

 

~ .l-I--v-O-I-t-O-I-o----n-t----I-I-I-o-I-I--t---'--l-o-‘-'-I-o-. .—

 

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

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/07 p:lClRC/DateDue.indd-p.1

 

DEER AND SEDGE EFFECTS ON TREE SEEDLING DYNAMICS IN
NORTHERN TEMPERATE F ORESTS

By

Jesse Allen Randall

A DISSERTATION
Submitted to
Michigan State University
In partial fulfillment of the requirements
For the degree of

DOCTOR OF PHILOSOPHY
Department of Forestry

2007

ABSTRACT

DEER AND SEDGE EFFECTS ON TREE SEEDLING DYNAMICS IN
NORTHERN TEMPERATE F ORESTS

By

Jesse Allen Randall
Since the mid-20th century, a combination of forest and wildlife management practices in
northern Michigan have to led to white-tailed deer (Odocoileus virginianus) populations
that are perhaps unprecedented historically and approximately four-fold higher than pre-
European settlement estimates. Selective browsing of vegetation by deer at high densities
may alter vegetation composition and structure, and impact ecological and economic
values including forest tree regeneration. In addition to the proximal effects of
browsing, deer- altered vegetation communities, have been hypothesized to persist if deer
are reduced, by competitively excluding the reinvasion of deer sensitive species. For
aspen and sugar maple dominated systems, I used a series of natural and manipulative
experiments to 1) quantify the impact of deer browsing on forest understory composition
and structure 2) compare the relative contribution of deer browsing and competition from
deer impacted vegetation on tree seedling and forb dynamics, 3) investigate possible
physiological mechanisms underlying competition between tree seedlings and deer
altered vegetation, and 4) investigate techniques that could be used to decrease tree
seedling competition with deer impacted vegetation. In 61 Populus spp. stands
distributed among two relative deer density classes (~11 deer / km2 and 6.8deer / kmz
deer densities), and a range of site productivities (Site index 12-25m @ SOyears) and

stand ages (12—44 years) deer strongly affected understory vegetation composition and

 

structure. Changes caused by higher deer density include greater fern and sedge mass,
and, especially in more productive environments, decreased species diversity & richness
Deer also reduced tree stem density and species richness (0.6m -4m tall) suggesting
longer-term consequences for forest succession.

In managed northern hardwood forest with high long-term deer density (3 l/km2)
removal of the Carex pensylvanica dominated understory vegetation had modest positive
effects on tree seedling growth and survival and the magnitude of these effects increased
with canopy Openness (i.e. higher light levels). However, these effects, for larger tree
seedlings were only apparent if deer were removed. These results indicate that deer
effects on tree seedling dynamics supersede vegetation competition effects.

In the field, Carex—dominated understory vegetation lowered soil moisture
availability to tree seedlings, potentially explaining lower growth and survival in the
presence of vegetation. In a potted plant experiment with Acer saccharum, and Carex
pensylvanica monocultures and mixtures, drought treatments reduced Acer seedling but
not Carex survival. Surviving seedlings had shorter overall root and stems lengths
following drought, while sedge root length or mass were not reduced by drought.

In one acre field plots, herbicide application in late autumn greatly reduced Carex density
with little effect on other vegetation. However, given the dominant effects of deer on

vegetation, only in areas with reduced deer could a seasonally timed herbicide application

be effective in promoting tree seedlings.

 

Dedicated

To my wife Natalie & my entire family

iv

ACKNOWLEDGEMENTS

My family - for continued help, encouragement, and guidance with my project and along
the road of life.

My wife, Natalie - for the countless hours in the field, the lab, and the greenhouse. You
have supported me at every turn, making sure I headed in the right direction.

The Nelson’s - you first welcomed me into your home as a poor and hungry grad student,
and then into your family as a son. Thanks for all you have given me.

My major advisor, Dr. Mike Walters - a mentor for seven years and a friend for life.

My committee members - Ric Campa, Don Dickmann, Shawn Riley, and Dave Rothstein
- all provided quality feedback over the course of the project making it that much
stronger.

My lab mates - John Gerlach, Justin Kunkle, Joe LeBouton, Laura Marx, and Bhavesh
Shah, for their friendship and help in the field and the lab.

To all those who worked in the sun, rain, snow, and tick infested woods as a member of
the field or lab crew that collected, weighed and ground hundreds, if not
thousands, of samples over the years (Curt Mykut, Ross Colman, Andy Klein,
Tony Fox, Chris Gladstone, Melissa McDermitt, Lynn Holcomb, Dan Schillinger,
and Gail Simmonds).

The Rothstein lab - for the technical support, equipment, and time that you gave to my
project. The friendship and moral support wasn’t bad either.

Andy Klein - for his help in building deer exclosures in the middle of nowhere when it
seemed that the ticks outnumbered the black flies.

Tony F ox - for sticking with me when we got lost in the middle of the hunt club on a
summer day as the temperature topped out at 100.

Special thanks to

-Mid Forest Lodge members and staff for giving me access to the club lands and
for the deer and forest management records.

-Intemational Paper researchers and foresters for their technical and financial
assistance over the years.

-The personnel at MSU-UPTIC facility for their assistance, suggestions, and open
door policy when I needed to borrow equipment.

-Randy and Paul at MSU’s-TRC for your guidance, knowledge, and patience
when I took over the lathe house with sugar maple seedlings and sedge.

-Craig Albright and the Gladstone DNR personnel who provided me with an
office and expertise in both forest and wildlife areas.

Multiple Funding sources made this project possible
- International Paper
- Michigan Department of Natural Resources
- Mid Forest Lodge
- MSU - Department of Forestry
- MSU - Dissertation Completion Fellowship
- MSU - Land Policy Program - Graduate Research Scholar
- MSU - Plant Science Fellowship
- Safari Club International
- Sand County Foundation’s Bradley Fund for the Environment
- The Michigan Botanical Club — Hanes Fund

vi

INTRODUCTION

General Overview

I begin this dissertation with a general introductory overview highlighting past
forest and wildlife management practices that have resulted in an overpopulated white-
tailed deer herd in parts of Michigan. I present a review of the direct and indirect effects
of an overabundant deer herd on forested ecosystems from a global to an upper Midwest
perspective. I go on to discuss the current state of knowledge surrounding
restoration/remediation treatments used in forests to control competing vegetation, and
discuss the need for new management techniques to selectively control vegetation in the
understory of northern hardwood forests. Finally, I present the major questions addressed
in each research chapter (1-4).
History of forest management and deer in Michigan

Historically, Michigan’s forest communities were shaped by disturbance regimes
(Palik and Pregitzer 1991) with return intervals varying over several orders of magnitude
(Frelich 2002, Cleland et al., 2004). Prior to European settlement, Native Americans
exerted direct and indirect pressure on Michigan’s forested landscape through their use of
fire and hunting. The combination of having forests that were much more contiguous,
predator populations that were large and self-sustaining (cougars, wolves), and
subsistence hunting kept deer numbers low over much of the Upper Midwest (2 - 4
deer/km2 - Blouch 1984 and McCabe and McCabe 1984). Westward expansion helped
bring about the era of big timber and fire (1800’s to early 1900’s), which cleared almost
92 % of Michigan’s forestlands (Dickmann and Leefers 2003) and helped to drive deer

numbers even lower as suitable habitat declined and market hunting increased.

Efforts to suppress wildfire allowed for early seral species (aspen) to regenerate,
while those sites that did not burn regularly (northern hardwood, lowland conifers)
flourished due to increased light reaching the understory and nutrient enrichment from
decaying waste from past timber harvests. As forests were beginning to regrow, new laws
which restricted deer harvests led to an overall increase in herd size until the 1950’s. By
the 1950’s, the maturing second growth forest no longer provided favorable habitat
conditions and deer numbers declined. Even though a second round of timber harvesting
began in the mid-late 1950’s, deer numbers did not respond until mosaic-like conditions
(agricultural fields and forests at various successional stages), representing increased
levels of edge habitat, were extensively developed throughout the northern portion of the
state. As the deer population began to increase again (early to mid 1970’s), land
managers began to notice changes in forest understory compositions (increase in sedge)
and seedling regeneration levels (decreases in desired species and increases in
undesirables). By the late 1980’s and into the early 1990’s the deer herd approached two
million animals in the state, and in certain core deer areas (those areas with extreme
concentrations of deer- e. g. deer over-wintering yards) tree regeneration was no longer
assured. (Mike Young, personal communication of International Paper’s land
management records, Michigan DNR).

Today the deer herd is estimated at roughly 1.8 million animals, well above the
MDNR’s stated herd size goal of 1.3 million animals (MDNR). Because deer
management in Michigan is such a controversial topic, resource managers are often at
odds with multiple stakeholders (hunters, loggers, farmers, conservation groups, auto

insurance groups, etc.) as to how the herd should be managed. Resource managers must

often defend their decisions concerning deer and forest management using studies that
highlight deer impacts as justification for decreasing herd sizes or altering harvesting
regimes.

The impacts of elevated ungulatmopulations

Deer impacts in forested systems are not just a Michigan phenomenon. Research
work worldwide has quantified ungulate impacts on vegetation composition and structure
(McNaughton 1985, Alverson et al., 1988, Pastor and Naiman 1992, Rooney and Dress
1997, Kielland and Bryant 1998, Hester et al., 2000, Crete etal., 2001, Russell et al.,
2001, Motta 1995, Trembley et al. 2005, Weisberg et al., 2005, Danell et al., 2006).
Although deer can affect certain vegetation which are particularly sensitive to deer
browsing (6. g. hemlock seedlings) at low deer densities (Alverson et al., 1988), their
impacts are greater in areas with high deer densities (Alverson and Waller 1997, Waller
and Alverson 1997, Horsley et al., 2003, Coté et al., 2004, and Trembley et al., 2006),
and in systems with greater site productivity characteristics (Milchunas and Lauenroth
1993, Hester et al., 2000).

When compared to unselective disturbance agents such as wildfire, landslides,
windthrows, and ice storms, large scale herbivory may be a unique form of disturbance
(Hulme 1996). Ungulates (e. g. deer, elk, moose) are highly selective towards the plants
they consume, often causing understory forb layers to shifts away from being
compositionally diverse (Marquis 1974 & 1981, Anderson etal., 2001), especially in
areas with overabundant ungulate populations (Cornett et al. 2000, Crawley 1990,
Coughenour 1991, McInnes et al., 1992, and Jefferies et al., 1994, VanDeelan et al.,

1996, DeCalesta 1997, Horsley et al., 2003, Danell et al., 2006). Deer impacts on

vegetation are difficult to generalize however because, 1) browsing affects several
different vegetation characteristics and each of these may vary uniquely with browsing
pressure, 2) for each of these characteristics, density interacts with other factors
including, among others, characteristics of the plant community and resource availability
(Weisberg et al. 2005, Wisdom et al. 2006) and 3) there are both temporal and spatial
elements of deer browsing that can affect responses (Hester et al 2004, Danell et al.,
1994). Horsley et al’s, (2003) and Trembley et al’s, (2006) work in productive maple-
cherry forests in Pennsylvania and boreal balsam fir-birch forest of northeastern Canada,
respectively, provide the best quantitative data to date concerning intermediate levels of
deer browsing pressure for systems in the northeastern United States, but concrete
threshold values to explain deer damage still eludes researchers (Horsley et al., 2003).
Overpopulated ungulates can have direct and indirect impacts on entire systems
and their processes (Rooney and Waller 2003). Through browsing, the direct loss of
overall plant biomass (Manseau et al., 1996) and tree form (multi-stemmed shrub form
vs. a single stem -— Gill 1992) in areas with high ungulate densities can have long term
impacts to forest management (decreased timber quantity and quality resulting in
decreased economic returns and overall forest sustainability) and system functioning due
to changes in resource availabilities. Furthermore, as forbs never grow beyond the reach
of deer, they are under constant browsing pressure, with sensitive species (indicators-
Balgooyen and Waller 1995, and Rooney 1997 & 2000) being the first to be replaced
with more browse tolerant /non-browsed species. Indirectly, deer, via shifts in vegetation
compositions, can create a cascading effect on entire ecosystem and its functioning,

which can persist under certain conditions, even if deer are removed (Stromayer and

 

Warren 1997, Augustine et al., 1998, de la Cretaz and Kelty 1999, George and Bazzaz
1999). Specifically, ungulate induced vegetation shifis can lead to decreased nutrient
cycling rates when higher densities of low palatability, low nitrogen, decomposition
resistant species dominate (Pastor et al. 1993, Pastor and Cohen 1997). This alone can
change species compositions as species realign to reflect the change in nutrient
availability. Compositional shifts can also increase plant competition levels as
understories become saturated with the non-browsed or browse resilient species such as
grasses, ferns, and sedges. These compositional shifts can also create feedback
mechanisms as increased levels of plant growth (overcompensation) (McNaughton 1983,
Belsky 1986) can lead to increased secondary ungulate browsing further altering
composition. Overall, declines in forest productivity and ecosystem fertility (Risenhoover
and Maass 1987) can be caused directly and/or indirectly by deer via deer induced shifts
in composition. With declines in productivity and shifts in composition and forest
structure, habitat quality for some forest birds (Hall and Root 1999), insects, amphibians,
and small mammals (Hodorffeta1., 1988) may decline. Eventually, long lasting,
extensive shifts in vegetation compositions can alter the carrying capacity of the habitat,
resulting in a reduction of ungulate health and declines in herd size (Davison and Doster
1997). To date, the indirect effects of overpOpulated deer herds are not well understood
and mainly exist as untested hypotheses because of the complex interactions between the
temporal and spatial factors affecting deer browsing and a systems succession induced

changes in form and processes.

 

Restoring deer influenced degraded forests

Reversing compositional and structural shifts through direct remediation activities
is increasingly common, as more and more forested lands are being negatively impacted
by elevated ungulate populations and increased populations of ungulate induced plant
competitors. The most comprehensive work to date targeting the reduction of plant
competitors has taken place in conifer systems (Cogliastro et al., 1990, Lautenschlager
1995, Bell et al., 1997, Vreeland et al., 1998, and Wagner et al., 2004). In these systems,
emphasis is normally placed on increasing young seedling survivorship during the
establishment phase to maximize potential crop tree production and not the overall
ecological integrity and processes that maintain systems (Wagner et al., 2004).
Applicable work in northern hardwood stands is much more scarce, as most treatments
(i.e. herbicides, mowing, controlled burning) that control hardwood competitors (i.e.
grasses, sedges, and ferns) also impact the northern hardwood crop trees. Working to
promote northern hardwoods in areas with high deer densities by controlling competing
vegetation, has to date, resulted in a better understanding of deer impacts and the
effectiveness of various herbicides to control competing vegetation (Horsley 1981, &
1990), but no concrete evidence has been forthcoming that treatments can be used to
overwhelm and thus overcome overabundant deer herd browsing. The use of strategically
timed vegetation manipulation treatments which do not interfere with previously
established northern hardwood crop trees and which, at the very least, maintains
herbaceous layer diversity have been tried only sparingly (but see Horsley 1981,

Willoughby 2006)

Deer effects have been studied by several researchers in various northern
temperate forests (Aspen - Campa 1989 & Raymer 1996, Cedar - VanDeelan et al., 1996
& VanDeelan 1999, Hemlock — F relich and Lorimer 1985, Mladenoff and Stearns 1993,
Northern hardwoods — Buckley et a1. 1998, Rooney and Waller 2003), often times with
an emphasis on individual components (understory composition, overstory structure, deer
available forage quality and quantity) at a specific site or time since disturbance (but see
Raymer 2000). In Michigan, it is now of greater importance to managers and
practitioners for researchers to identify if linkages exist between ungulate browsing and
other potential successional drivers in forested systems. A recently conducted survey
spanning all major forest types in Michigan revealed that large shifts in structure and
composition were evident in high deer areas, but as important, high sedge densities often
accompanied elevated deer densities (Randall and Walters unpublished data). In these
areas, the declines in forb and seedling composition and structural complexity cannot be
ascribed solely to increased deer browsing as high sedge or fern cover, resulting from
preferential deer herbivory may directly impact seedling dynamics via competition.
Furthermore, sedge could increase due to factors other than the direct or indirect effects
of deer browsing. For example, a long history of selection harvesting may favor sedge, as
relatively high light levels are maintained for vegetation near the forest floor (Metzger
and Schultz 1984, Horsley 1990, Wiegmann and Waller 2006) but see Jenkins and Parker
1999). In addition, populations of invasive earthworms have even been found to cause
declines in forb and seedling communities (Hale et al., 2006, Frelich et al., 2006), and

increased levels of sedge cover compared to areas without worms.

As forest composition and structure can be affected by several variables other
than deer, accounting for those variables (6. g. site and stand variation - Pinno et al., 2001)
while attempting to understand the mechanisms at play between deer, sedge, and deer X
sedge should enable managers to make better resource based decisions in the deer and
sedge dominated systems. This increase in knowledge surrounding the mechanisms and
the direct and indirect effects of deer should improve not only forest health and
sustainability, but also the health and sustainability of the deer herd.

The overarching goal of this dissertation was to use natural and manipulative field
based studies, along with a controlled environment study, to tease apart potential drivers
of the regeneration failure currently ascribed almost exclusively to overabundant deer. In
addition to providing resource managers with a better understanding of the sedge/deer
system, I also hoped to find restoration treatments that could be used to control sedge in a
manner that would not harm existing seedling and forb populations.

Specifically, in the field portion of the dissertation (Chapters 1-3), I asked:

1) i). How does deer density affect the cover, mass, species richness, diversity,
nitrogen content and structure of forest understory herbaceous and woody
vegetation?

ii) Do site index and/or stand age have interacting effects with deer on vegetation
characteristics?

iii) What are the potential successional legacies of changes in tree sapling
composition and densities, and what are their management implications?

2. i) What are the relative impacts of main effects and interactions of deer and of a

sedge-dominated understory on tree seedling establishment, survival and

growth, and forb diversity in highly productive northern hardwood stands

dominated by sugar maple?

_ ii) If there are effects of deer and sedge on seedling growth in these
productive stands, can this be associated with changes in soil nitrogen and
water availability?

iii) If there are effects of sedge on growth and survival, how do these effects vary
with light availability?

. i) If high sedge cover is partially responsible for inhibiting the reestablishment
of tree seedlings and forbs, is it possible to reduce sedge covers with
management interventions (timed herbicide applications) such that tree,
shrub, and forb establishment would increase?

ii) What are the effects of increased canopy openness conditions on seedling and
forb establishment and survival, and do responses change based on

vegetation manipulation treatment?

. Under controlled environment conditions, I asked:

i) Can Carex draw soil water to lower levels than Acer saccharum seedlings?

ii) Does Carex have greater survivorship than A. saccharum seedlings at low
water availability?

iii) Does a drought event negatively affect A. saccharum growth?

iv) Does Carex negatively affect soil mineral nitrogen availability?

This dissertation represents a compilation of publication manuscripts developed from
work which took place in aspen dominated stands (chapter 1), productive northern
hardwood sugar maple stands (chapters 2&3), and under controlled environmental
conditions targeting sedge-maple interactions during and after drought (chapter 4). The
common linkage throughout this dissertation is elevated levels of plant competitors
(sedge) and/or the effects of overabundant white-tailed deer on vegetation dynamics.
Results along these lines further both basic and applied knowledge in forest systems with
overpopulated deer herds and an almost uniform cover of sedge in the understory.
Briefly:

In Chapter 1, to observe vegetation dynamics and address questions li-iii from
above, I designed an experiment that accounted for stand age (12-44 years) and site
productivity (~12-25 m @ 50 years) influences in each of two long-term deer densities
categories (11 vs. 6.8 deer/kmz). Aspen monocultures were used to minimize the
influence of overstory compositions on resources levels in the understory.

Research work which formed the basis for chapter two (questions 2i-iii) was
driven by the desire to specifically address the influence of both sedge and deer
individually and in combination with each other on understory vegetation composition
and structure, as well as system functioning from a belowground resource availability
standpoint. To do this, a series of deer exclosures and vegetation removal treatments
manipulated deer and understory vegetation in sugar maple dominated forests.

In response to the work and results from chapter 2, and driven by conversations
with forest managers at local, state, federal, and industrial levels it became evident that

silvicultural techniques utilizing cost effective herbicide treatments beneath selectively

10

harvested overstories which control increased populations of sedge, while at the same
time minimized the impacts to established hardwood seedlings and understory forbs,
were lacking. I addressed questions 3i-ii using seasonally timed spray treatments, 0.2 ha
in size, under the canopy of selectively harvested sugar maple forests. In addition to
direct tests of treatments, levels of harvesting intensity varied throughout the stand and
provided a gradient of canopy openness conditions to test the response variables against.
Finally, chapter four is a mechanisms paper. This manuscript presents results
quantifying the impact of drought on sedge and sugar maple seedling survival and
growth. It furthers the understanding of sedge/maple interactions witnessed in the field-

based portions of the dissertation.

11

LITERATURE CITED IN THE INTRODUCTION

Alverson, W.S., D.M. Waller, and S.L. Solheim, 1988. Forest too deer: Edge effects in
northern Wisconsin. Conservation Biology 2(4):348-358.

Alverson, W.S. and D.M. Waller, 1997. Deer populations and the widespread failure of
hemlock regeneration in northern forests. pp280-297 in W. McShea and J.
Rappole, eds., The science of overabundance: Deer ecology and population
management, Smithsonian Institute Press, Washington, DC.

Anderson, R.C., E.A. Corbett, M.R. Anderson, G.A. Corbett, and T.M. Kelley, 2001.
High white-tailed deer density has negative impact on tallgrass prairie forbs.
Journal of the Torrey Botanical Society 128(4):381-392.

Augustine, D. J ., L.E. Frelich, and PA. Jordan. 1998. Evidence for two alternate stable
states in an ungulate grazing system. Ecological Applications 8(4): 1260-1269.

Balgooyen, CR and D.M. Waller 1995. The use of Clintonia borealis and other
indicators to gauge impacts of white-tailed deer on plant communities in northern
Wisconsin, USA. Nat. Areas J. 15(4):308-318.

Bell, F .W., R.A. Lautenschlager, R.G. Wagner, D.G. Pitt, J .W. Hawkins, and KR. Ride.
1997. Motor-manual, mechanical, and herbicide release affect early successional
vegetation in northwestern Ontario. The Forestry Chronicle 73(1):61-68.

Blouch, RI. 1984. Northern great lakes states and Ontario forests. pp 391-410. in L.K.
Halls (ed.). White-tailed deer ecology and management. Stackpole books,
Harrisburg, Pa.

Belsky, A.J. 1986. Does herbivory benefit plants? A review of the evidence. The
American Naturalist 127(6):870-892.

Buckley, D.S., T.L. Sharik, and J .G. Isebrands. 1998. Regeneration of northern red oak:
positive and negative effects of competitor removal. Ecology 79(1):65-78.

Campa 111, H. 1989. Effects of deer and elk browsing on aspen regeneration and
nutritional qualities in Michigan. Pd.D Dissertation. Michigan State University.
East Lansing. 122pp.

Cleland, D.T., T.R. Crow, S.C. Saunders, D.I. Dickmann, A.L. Maclean, J .K. Jordan,
R.L. Watson, A.M. Sloan, and K.D. Brosofske. 2004. Characterizing historical
and modern fire regimes in Michigan (USA): A landscape ecosystem approach.
Landscape Ecology

12

Comett, M.W., L.E. Frelich, K.J. Puettmann, and PB. Reich. 2000. Conservation
implications of browsing by Odocoileus virgnianus in remnant upland Thuja
occidentalis forests Bio. Con. 93: 359-369.

Cote, S. D. T. P. Rooney, J. P. Tremblay, C. Dussault, and D. M. Waller. 2004.
Ecologicalimpacts of deer overabundance. Annual review of ecology and
evolutionary systems 35: 113-147.

Crete, M. J. P. Ouellet, and L. Lesage. 2001. Comparative effects on plants of
caribou/reindeer, moose and white-tailed deer herbivory. Arctic 54(4):407-
417.Comett et a1 2000

Cogliastro, A., D.Gagnon, D. Coderre, and P. Bhereur. 1990. Response of seven
hardwood tree species to herbicide, rototilling, and legume cover at two southern
Quebec plantation sites. Canadian Journal of Forest Research 20:1172-1182.

Coughenour, MB. 1991. Biomass and nitrogen responses to grazing of upland steppe on
Yellowstone’s northern winter range. J. of App. Eco. 28:71-82.

Crawley, M.J. 1990. Rabbit grazing, plant competition and seedling recruitment in acid
grassland. J. App. Eco. 27:803-820.

Danell, K., R. Bergstrom, P. Duncan, and J. Pastor. 2006. Large herbivory ecology,
ecosystem dynamics and conservation. Cambridge university press

Danell, K. R. Bergstrom, and L. Edenius. 1994. Effects of large mammalian browsers on
architecture, biomass, and nutrients of woody plants.

Davidson, W. R. and G.L. Doster Health characteristics and white-tailed deer population
density in the southeastern United States. Pg164-184 in W. McShea and J.
Rappole, eds., The science of overabundance: Deer ecology and population
management, Smithsonian Institute Press, Washington, DC.

DeCalesta, DS. 1997. “Deer, Ecosystem Damage, and Sustaining Forest Resources,”
online proceedings of Deer Management in Maryland Conference, October 27,
1997.

de la Cretaz, A. and M.J. Kelty 2002 Development of tree regeneration in fem-dominated
forest understories after reduction of deer browsing. Restoration Ecology
10(2):416-426.

Dickmann, DI. and LA. Leefers. 2003. The Forests of Michigan. The university of
Michigan Press. Ann Arbor, MI. USA. pp 297.

Frelich, LE. 2002. Forest dynamics and disturbance regimes: studies from temperate
evergreen-deciduous forests. pp. 276. Cambridge University Press

Frelich, LE. and CG. Lorimer. 1985. Current and predicted long-term effects of deer

13

browsing in hemlock forests in Michigan, USA. Bio. Con. 34(2):99-120.

Frelich, L.E., C.M. Hale, S. Scheu, A.R. Holdsworth, L. Heneghan, P.J. Bohlen, and PB.
Reich. 2006. Earthworm invasion into previously earthworm-free temperate and
boreal forests. Biological invasions 8:1235-1245.

Gill, R.M.A. 1992. A review of damage by mammals in north temperate forestsz3. Impact
on trees and forests. Forestry 65(4):363-3 88.

George, L0. and FA. Bazzaz. (1999) The fern understory as an ecological filter:
growth and survival of canopy-tree seedlings. Ecology 80 (3):846-856.

Hale, C.M., L.E. Frelich, and RB. Reich. 2006. Changes in hardwood forest understory
plant communities in response to European earthworm invasions. Ecology
87(7):1611-1874.

Hall, K.R., and TL. Root. 1999. Influence of browsing by white-tailed deer on migratory
bird population: is the black-throated blue warbler at risk. Progress report to the
Mich. Chap. TNC.

Hester, A.J., L. Edenius, R.M. Buttenschon and AT Kuitters. 2000. Interactions between
forests and herbivores: the role of controlled grazing experiments. Forestry
73(4):381-391.

Hester, A.J., P. Millard, G.J. Baillie, and R. Wendler. 2004. How does timing of
browsing affect above- and below-ground growth of Betula pendula, Pinus
sylvestris, and Sorbus aucuparia? Oikos 105-536-550.

Hodorff, R.A., C.H. Sieg, R.L. Linder. 1988. Wildlife response to stand structure of
deciduous woodlands. Journal of Wildlife Management 52(4):667-673.

Horsley, SB. 1981. Control of herbaceous weeds in Allegheny hardwood forests with
herbicides. Weed Science 29:655-662.

Horsley, SB. 1990. Control of grass and sedge in Allegheny hardwood stands with
Roundup-residual herbicide tank mixes. Northern Journal of Applied Forestry
7: 124-129.

Horsley, S.B., S.L. Stout, and D.S. deCalesta. 2003. White-tailed deer impact on the
vegetation dynamics of a northern hardwood forest. Ecological Applications
13:90118.

Hulme, PE. 1996. Herbivory, plant regeneration, and species coexistence, Journal of
Ecology 84(4):609-615.

14

Jeffries, R.L., D.R. Klein, and GR. Shaver.1994 Vertebrate herbivores and northern plant
communities: reciprocal influences and responses. Oikos 71 :193-206.

Jenkins, MA. and GP. Parker. 1999. Composition and diversity of ground-layer
vegetation in silvicultural openings of southern Indiana forests. The American
Midland Naturalist 142(1): 16.

Kielland, K. and J .P. Bryant. 1998. Moose herbivory in taiga: effects on biogeochemistry
and vegetation dynamics in primary succession. Oikos 82:377-383.

Manseau, M., J. Huot, and M. Crete.1996. Effects of summer grazing by caribou on
composition and productivity of vegetation: community and landscape level.
Journal of Ecology 84:503-513.

Marquis, DA. 1974. The impact of deer browsing on Allegheny hardwood regeneration.
US. Department of Agriculture Forest Service Res. Rep. NE-47

----- 1981. Effect of deer browsing on timber in Allegheny hardwood forests of
northwestern Pennsylvania. The impact of deer browsing on Allegheny hardwood
regeneration. US. Department of Agriculture Forest Service Res. Rep. NE-57

McCabe, RE. and TR. McCabe. 1984. Of slings and arros: an historical retrospective. in
L.K. Halls (ed.). White-tailed deer ecology and management. Stackpole books,
Harrisburg, Pa.

McNaughton, SJ. 1983. Compensatory growth as a response to herbivory. Oikos 40:329-
336.

McNaughton, S.J. 1985. Ecology of a grazing ecosystem: the Serengetti. Ecological
Monographs 55(3):259-294.

McInnes, P.F., R.J. Naiman, J. Pastor, and Y. Cohen. 1992. Effects of moose browsing on
vegetation and litter of the boreal forest, Isle Royale, Michigan, USA. Ecology
73:2059-2075.

Metzger, F. and J. Schutz. 1984. understory response to 50 years of management of a
northern hardwood forest in Upper Michigan. American Midland Naturalist
1 12(2):209-223.

Milchunas, D.G. and WK. Lauenroth. 1993. Quantitative effects of grazing on vegetation
and soils over a global range of environments. Ecological Monographs 63(4):327-
366.

Mladenoff, DJ. and F. Stearns. 1993. Eastern hemlock regeneration and deer browsing in

the northern great lakes region: a re-examination and model simulation.
Conservation Biology 7(4):889-900.

15

Motto, R. 1995. Impacts if wild ungulates on forest regeneration and tree composition of
mountain forests in the Western Italian Alps. For. Eco. Manage. 88:93-98.

Palik, B.J. and KS. Pregitzer. 1991. The vertical developmanet of early successional
forests in northern Michigan, USA. Journal of Ecology. 81:271-285.

Pastor, J. and R.J. Naiman. 1992. Selective foraging and ecosystem processes in boreal
forests. The American Naturalist 139(4):690-705.

Pastor, J ., B. Dewey, R.J. Naiman, P.F. McInnes, and Y. Cohen. 1993. Moose browsing
and soil fertility in the boreal forests of Isle Royal National Park. Ecology
74(2):467-480.

Pastor, J. and Y. Cohen. 1997. Herbivores, the functional diversity of plants species, and
the cycling of Nutrient in Ecosystems. Theoretical Population biology 51 :165-
179.

Pinno, B.D., V.J. Leiffers, and K.Stadt. 2001. Measuring and modeling the crown and
light transmission characteristics of juvenile aspen. Canadian Journal of Forest
Research 31:1930-1939.

Raymer, D.F.N. 1996. Current and long-term effects of ungulate browsing on aspen stand
characteristics in northern lower Michigan. M.S. thesis. Michigan State
University. East Lansing. 206pp.

------ . 2000. Effects of Elk and white-tailed deer browsing on aspen communities and
wildlife habitiat quality in northern lower Michigan: an 18-year evaluation. Ph.D.
Dissertation. Michigan State University. East Lansing. 371pp.

Risenhoover, KL. and SA. Maass. 1987. The influence of moose on the composition and
structure of Isle Royale forests. Can. J. For. Res. 17:357-364.

Rooney, TR and W.J. Dress. 1997. Species loss over sixty-six years in the ground-layer
vegetation of Heart’s Content, an old growth forest in Pennsylvania, USA.
Natural Areas Journal 17(4):297-305.

Rooney, TR and D.M. Waller. 2003. Direct and indirect effects of white-tailed deer in
forest ecosystems. Forest Ecology and Management. 181:165-176.

Russell, F .L., D.B. Zippin, N.L. Fowler. 2001. Effects of White-tailed Deer (Odocoileus
virginianus) on Plants, Plant Populations and Communities: A Review. The
American Midland Naturalist. 146(1): 1-26.

Stromayer, K. A. K., and R. J. Warren. 1997. Are overabundant deer herds in the eastern

16

United States creating alternate stable states in forest plant communities. Wildlife
Society Bulletin 25(2):227-234.

Trembley, J .P., J. Huot, and F. Potvin. 2006. Divergent nonlinear responses of the boreal
forest field layer along an experimental gradient of deer densties. Oecologia
150:78-88.

Van Deelen, T.R. Deer-Cedar interactions during a period of mild winters: implications
for conservation of conifer swamp deeryards in the great lakes region. Natural
Areas Journal 19(3):263-274.

Van Deelan, T.R., K.S. Pregitzer, and J .B. Haufler. 1996. A comparison of presettlement
and present-day forests in two northern Michigan deer years. American Midland
Naturalist 135:181-194.

Vreeland, J .K., F.A. Servello, B. Griffith. 1998. Effects of conifer release with glyphosate
on summer forage abundance for deer in Maine. Canadian Journal of Forest
Research 28:1574-1578.

Wagner, R.G., M. Newton, E.C. Cole, J .H. Miller, and 8D. Shiver. 2004. The role of
herbicides for enhancing forest productivity and conserving land for biodiversity
in North America. Wildlife soicuety Bulletin 32(4): 1028-1041.

Waller D.M. and W.S. Alverson. 1997. The white-tailed deer: a keystone herbivore.
Wildlife Society Bulletin 25:217-226.

Weisberg, P.J., F. Noavia, and H. Bugmann. 2005 Modeling the interacting effects of
browsing and shading on mountain forest tree regeneration (Picea abies).
Ecological modeling 185:213-230.

Wiegmann, SM. and D.M. Waller. 2006. Fifty years of change in northern upland forest
understories: Identity and traits of “winner” and “loser” plant species. Biological
Conservation 129(1): 109-123.

Willoughby, 1., FL. Dixon, D.V. Clay. 2006. Dormant season vegetation management in
broadleaved transplants and direct sown ash (Fraxinus excelsior L.) seedlings.
Forest Ecology and Management 222-418-426.

Wisdom, M.J., M Vavra, J .M. Boyd, M.A. Hemstrom, A.A. Ager, and B.K.Johnson.
2006. Understanding ungulate herbivory — episodic disturbance effects on
vegetation dynamics: knowledge gaps and management. Wildlife Society Bulletin
34:283-292.

17

CHAPTER 1

DEER DENSITY EFFECTS ON VEGETATION IN ASPEN FOREST
UNDERSTORIES OVER SITE PRODUCTIVITY AND STAND AGE GRADIENTS

Executive summary

1 quantified main effects and interactions of deer density, site productivity (site
index), and stand age on forest understory vegetation structure and composition
characteristics in the understory of closed canopy aspen stands in Michigan’s lower
peninsula. Sites on state-owned lands (6.8deer km'z) and on a nearby private hunt club
(~11 deer km'z) comprised two long-term (>30 years) deer density categories. Deer
density, stand age and site index affected several vegetation characteristics, but
interactions were generally weak and deer effects dominated. Higher deer densities
resulted in approximately 3-fold greater bracken fern (415 kg/ha vs. 130 kg/ha) and sedge
biomass (220 kg/ha vs. 84 kg/ha), 90% lower herb biomass (1.1 kg/ha vs. 10.2 kg/ha),
3.4-fold greater woody stem (<O.25m tall) biomass (9.3 kg/ha vs. 2.7 kg/ha), and 90%
lower woody stern densities (0.6m — 4m tall) with only Prunus virginiana saplings at
high deer density. Aspen stand age and deer interacted strongly on sedge biomass with
high deer densities areas having much greater sedge in young than old aspen stands,
while in lower deer density areas sedge was found at low densities across all ages. Both
fern and sedge mass were high at high deer and high 81, but near zero at lower deer.
Bracken fern is typically abundant on poorer sites suggesting that higher deer may be

driving vegetation on richer sites toward species that are both less palatable and adapted

18

to poorer sites. Herb layer species richness and diversity were greater at lower than
higher deer density. Deer density did not affect vegetation N concentrations, but total
vegetation N content in the deer browse zone (i.e. < 1.5 m height) was higher (11 vs. 3.6
kg N /ha) in high deer density areas due to higher sedge and fern biomass. In summary,
controlling for overstory density and composition with closed canopy aspen stands, I
found that higher long-term deer densities had strong impacts on understory vegetation,
with changes in sapling composition and density potentially affecting successional
trajectories on long time frames. Relative effects on understory vegetation were generally
greater on more than less fertile sites, which is consistent with data for non-forested

systems.

19

Introduction

Interacting with a myriad of factors, ungulate browsing is a key driver of
vegetation dynamics in ecosystems worldwide (Russell et a1. 2001 , Pastor and Naiman
1992, Alverson et al., 1988, Hester et al., 2000, Crete et al., 2001, Trembley et al., 2005,
Weisberg et al., 2005, Danell et al., 2006). In the forested ecosystems of Eastern North
America, white tailed deer (Odocoileus virginianus, hereafter called deer) are often the
dominant ungulate herbivore. In the last several decades, land use practices and wildlife
management policies have resulted in deer densities that are thought to be historically
unprecedented in much of its range. There has been widespread concern that deer are
having strong impacts on forest vegetation composition and structure, and research to
date has demonstrated some of these effects (e. g. Rooney and Waller, Augustine,
Horsley, Tremblay). Deer impacts on vegetation are difficult to generalize however
because, 1) browsing affects several different vegetation characteristics and each of these
may vary uniquely with browsing pressure, 2) for each of these characteristics, density
interacts with other factors including, among others, characteristics of the plant
community and resource availability (Weisberg et al., 2005, Wisdom et al., 2006) and 3)
there are both temporal and spatial elements of deer browsing that can affect responses
(Danell et al., 1994, and Hester et al., 2004).

Examples of vegetation characteristics affected by deer in Eastern North America
include reduction or local extirpation of highly browse sensitive species (T axus
brevifolia, Parks et al., 1998, Maianthemum canadense, Balgooyen and Waller 1995,
Rooney 1997, T suga canadensis and Thuja occidentalis Rooney and Waller 2003 ),

decreased density of reproductive structures on browsed plants (Trillium spp., Anderson

20

 

1994; total forest floor community Tremblay et al., 2006 ), composition changes and
decreased diversity of tree regeneration (Horsley et al., 2003, Tremblay et al., 2006), and
increases in non-preferred browse species (e. g. ferns, Horsley et al., 2003). The
magnitude of deer effects may vary uniquely among response characteristics. For
example highly sensitive species, like T suga canadensis saplings growing the in zone of
deer herbivory (i.e. to approximately 1.5 m height) can be locally extirpated at deer
densities as low as 4/km2 (Alverson and Waller 1997), whereas density of black cherry
(Prunus virginiana), a non preferred browse species, does not increase in density in
harvest gaps until deer density exceeds 15 km2 (Horsley et al., 2003).

Obviously, vegetation dynamics depends on factor other than deer herbivory, and
these factors may interact with deer herbivory in complex ways. For example, the
impacts of herbivory may vary among vegetation communities that differ in their
evolutionary browse histories (Milchunas and Lauenroth 1993), or in their particular
composition and the differences in competitive interactions and relative preferences to
herbivores among species.

Resource availability could also influence plant responses to deer browse
pressure. For example a metanalysis of ungulate grazing effects on non-forested systems
indicates that richer sites (higher nutrient and/or water avaialability) are more negatively
impacted than poorer sites (Milchunus and Lauenroth 1993), perhaps because plants on
poor sites are more prone to disturbance and drought and thus allocate more biomass and
energy storage to roots, thus making them more grazing tolerant. Responses should vary
with light availability as higher light energy would enable plants to quickly compensate

for tissue loss, and for tree saplings, allow individuals to outgrow deer’s reach in height

21

quicker (Weisberg et al., 2005). Differences in vegetation responses to deer herbivory
between harvested and unharvested areas (e. g. Horsley et al., 2003, Tremblay et al.,
2006) are a clear reflection of the interaction of resource availability and deer density
affects on vegetation.

The forest understories of closed canopy forests may be especially sensitive to
deer browse pressure. Because of strong light limitations to growth in forest understories,
the consequences of a given amount of tissue removed for plant fitness as well as for
community structure and composition is greater and more long lasting than in less growth
limited environments. Furthermore, deer browse impacts on tree sapling in forest
understories could change the tempo and composition of forest succession for decades to
centuries because their browse could affect the composition and density of saplings that
recruit to canopy positions.

I designed a study to test the effects of differences in long-term deer density and
site quality on vegetation composition and structure in the understories of 61 closed
canopy aspen stands distributed over state-owned and private hunt club land that have
differed in deer density for more than 30 years (lower vs. higher respectively). I focused
on the understories of closed canopy mono-dominant stands because 1) these criteria
insured that light availability, a strong driver of vegetation dynamics, varied minimally
among my study plots, and 2) because deer effects could strong due to light limitation
and have long legacies. Locating stands across site quality gradients (assessed using site
index, (Lundgren and Dolid 1970) allowed me to examine the notion that deer herbivory
has greater impacts on more fertile/moist sites. In addition, I located stands among a

fairly limited range of ages and used age as a variable, reasoning that understory

22

vegetation characteristics could vary somewhat with age (Pinno et al., 2001, Leiffers et
al., 1996) despite the fact that all sites were in young, fully stocked, self thinning stands.
Given this design I was able to address the following questions:
2) How does deer density affect the cover, mass, species richness, diversity, nitrogen
content and structure of forest understory herbaceous and woody vegetation?
3) Do site index or stand age have interacting effects with deer on vegetation
characteristics?
4) What are the potential successional legacies of changes in tree sapling
composition and densities, and what are their management implications?
Methods
Study Area
My study sites were in aspen (Populus tremuloides Michx. and P. grandidentada)
stands located on state and private forestlands in Roscommon, Gladwin, and Clare
counties MI, USA. The study region’s long-term climate averages include: annual
precipitation, 730 mm; growing season (May-August) precipitation, 300 mm; annual
snow fall, 192 cm; growing season length (days where the minimum temp is > 0 °C), 126
days; and July mean temperature, 21 °C (NOAA National Virtual Data Systems). The
region contains level to moderately undulating hills (0-18% slopes) with post-glacial
geological features dominated by ice contact ridges and sandy outwash plains. Soils in
the region are a mixture of Rubicon — Menominee, Graycalm and Grayling sands (USDA
soil survey of Gladwin County 1972 and Roscommon County 1972).

Site Selection

23

In early summer 1999, I selected aspen-dominated sites in approximately equal
numbers on two ownerships: a) within Mid Forest Lodge (MFL), a 7,317 ha private
hunting club, characterized by high long-term deer densities; and b) on state lands,
located >1.6 km and < 10 km from MFL and characterized by lower deer densities than
MFL. Unlike state lands, MFL has hunter access restrictions, an aggressive winter-
feeding / food plot program, and, in the past, club rules limiting doe harvests.

Several decades of track count data collected in a uniform manner each year by
MF L indicated sustained high deer densities (unpublished MF L data), and although
similar data do not exist for state lands, larger-scale regional deer pellet count data
indicated much lower deer densities on these lands. To confirm these a priori
classifications, I conducted deer fecal pellet count estimates (described below).

From a broad list of candidate stands, I selected 32 stands with relatively higher
deer densities (i.e 11 deer/kmz, MFL property) and 29 with relatively lower deer densities
(i.e. 6.8 deer/kmz, state property), distributed among two target aspen stand age strata
(10-20 yr old and 30-40 yr old) and three relative site productivity strata (low, moderate,
and high). MFL and DNR harvest records were used to identify potential aspen stands
that fell within five years of the two age strata. Productivity strata were determined from
landform mapping and interpretations of vegetation characteristics collected on site
according to an ecological classification system (Cleland et al., 1993). Candidate sites
were discarded if management records did not match site criteria (6. g. not aspen
dominated), were not closed canopy, self-thinning stands, or if they came from over-
represented strata. Within a selected stand I randomly located a sampling area, and

randomly located a site plot center which I marked with a permanent metal stake.

24

Measurements
Deer density estimates

In early April, 2001, I counted deer fecal pellets just after snowmelt on a
randomly selected subset of sites within each deer density class, (29 of the 61 sites).
Starting at the permanent center stake, an observer walked four-150 m long transects
along the cardinal directions (N,S,E,W) and counted pellet groups within 2 m of either
side of the transect (2400 m2 sample area). Spring deer population densities were

estimated using the following formula modified fi'om Hill (2000):

 

 

pellet groups/m2 sample area X (10,000 m2- km'z)
Deer-km'2

 

 

13.37 (fecal groups ' deer'l - day'l) X (leaf off period in days)

 

Vegetation

A four-layer nested sampling plot design, centered over the permanent site center
stake, was used to measure forest vegetation composition and vertical structure during the
peak of the growing season (late July). Plot dimensions and the characteristics measured
were as follows: 1) in a 3 x 3 m grid, % cover by species of all vascular plants <25 cm
tall estimated occularly and simultaneously by two observers (to remove observer error,
the same two observers conducted all % cover estimate measurements at all sites); 2) in a
3.1 m radius circular plot, heights of all woody tree and shrub stems 0.25 m to 1.4 m in
height by species; 3) in a 5.64 m radius circular plot, the diameter (at DBH), total height
(hypsometer pole) of all tree and shrub stems by species > 1.4 m tall and <10 cm DBH;
and, 4) in a 12 m radius, the diameter at 1.4 m height and total height (Sunto® precision

clinometer) of all trees > 10 cm DBH by species.

25

 

To characterize forest floor layer vegetation composition, biomass and nutrient
content in July 2001, I established a 1.5 m2 plot 15 m north of the permanent stake. In
this plot, I described the forest floor vegetation as detailed above and then destructively
harvested all aboveground biomass of the forest floor vegetation layer including shrub
and tree seedlings <0.25m in height. Harvested vegetation was separated into 5 categories
(forbs, sedge & grass, bracken fern, and shrub and tree seedling (hereafter referred to as
woody seedling) leaves and stems) in the field. The sedge and grass category was
dominated (> 90%) by Carex pensylvanica Lam. and will hereafter be called sedge. I
dried the samples at 70 °C for 72 hours in the lab prior to measuring dry mass. All dried
samples were passed through a Wiley mill, pulverized in a hammer mill (KLECO —
Kinetic Laboratory Equipment Company, Visalia, CA) and a 7-10 mg subsample was
analyzed for N concentration by the Dumas combustion method with a CHN analyzer
(Carlo Erba, Milan, Italy). Vaccinium was excluded for analyses because of its absence in
higher deer density areas and its overall small and inconsistent mass in lower deer density
sites.

Site Productivity

To determine and index of site productivity, I measured age at 1.4 m height for
two co-dominant trees at each site. To age trees I used either two increment cores at 90°
angles from one another, or a single radial cross section taken after felling the tree. Air-
dried cross-sections (n=78) and cores (n=mounted on grooved wooden boards were
sanded with 220 grit sandpaper prior to counting and recording annual rings on scanned
digital images with WinDendro (Regent Instruments, Blain, Quebec). On trees which

were increment cored (n =20), I measured tree heights using a Sunto® precision

26

clinometer. Felled tree heights were determined with a tape measure. To check
clinometer measurement accuracy I compared clinometer and direct height measurements
for a subset of felled trees. I used age and height data to determine site index from
published site index curves for the Lake States region (Lundgren and Dolid 1970).
Analysis

Preliminary analyses of stand age and site index data revealed that my a priori age
and productivity target categories resulted in relatively continuous and even distributions
of both stand age and site index over both deer density categories. Thus, I chose to
consider these variables as continuous rather than categorical. Due to the fact that age
and site index ranges differed somewhat between higher and lower deer density strata, I
selected a subset of stands to achieve similar ranges for both stand age (12-24) and site
index (17-26 m at 50yrs). This subset was used for some analyses (structure and tree
species composition) with the objective of minimizing confounding deer density with SI
and/or stand age. I used JMP (5.1) statistical software (SAS institute, Cary, NC, USA) for
all statistical analyses.

Seedlings >0.25 m tall were placed into height classes to facilitate analyses of
stem density and vertical structure. Vegetation was analyzed with mixed least squares
models using a full factorial treatment design with main effects deer density (higher vs.
lower), as a discrete, nominal variable, and site index and stand age as continuous
variables. If model results indicated that interaction terms were insignificant beyond the
threshold suggested for pooling variances (P > 0.25, Bancroft 1964), then the highest

order interaction term with the highest P value was removed and the model was re-run.

27

 

This process was repeated iteratively until a final model was constructed where all
interactions >0.25 were removed.

I examined species richness for vegetation <0.25 m tall by bootstrapping data
using R statistical software to obtain multiple estimates of the number of unique species
for each area (9 m2 plot) gradation. The mean of these estimates were used to develop
smoothed species area curves for each combination of deer density (lower, higher) and
site index (low, high) categories. Initially, I separated stands into young and old stand age
cohorts, but due to a non-uniforrn distribution of site productivities in older stands I chose
to only evaluate young stands. The methodology for bootstrapping dictated that I assign
productivity classes rather than using continuous data. Therefore I chose two classes that
included sites at lower and higher ends of the SI range and excluded intermediate sites to
clearly highlight differences.

Similar to the previously mentioned methods for analyzing forest structure and
composition, 1 tested Shannon-Weiner diversity indices (H’) on a subset of sites, which
had uniform stand age (15-38) and site index (17-26 m at 50yrs) ranges across the two
deer density strata.

ME
Deer density

Spring 2001 fecal pellet counts collected for a subset of sites showed that MF L
had 62% greater relative deer density than adjacent state owned lands (11 vs. 6.8
deer/kmz, respectively). Large variation among sites made these differences only
marginally significant (P=0.068). Corroborating my independent pellet counts, track

count data collected by MFL members since 1968 indicate an average herd size of 10

28

deer/km2 over the last 25 years prior to 2001. Additionally, MDNR deer pellet counts
collected in 1981, 1995, and 1999 (averaged across the three sampling years) estimated
21 deer/km2 in the MF L and 15.7 deer/km2 on adjacent state lands. Although values
differ among estimation methods and time frames they consistently indicate sustained
long-term relative differences in deer populations between my higher and lower deer
density strata. Vegetation browse indicators (Balgooyen and Waller 1995, Rooney 1997
& 2000) were also consistent with my higher and lower deer density catagories.
Mianthemum canadensis cover was ~1 l-fold greater at higher deer than lower deer
(P<0.0001), while Gaultheria procumbens and Aralia nudicaulis were 3.5-fold greater.
Vegetation biomass, structure, and composition

Several forest floor vegetation mass and structural characteristics varied with deer
density, site index, stand age, and/or their interactions, but deer effects generally
dominated (Appendix 1.1, 1.2). Total aboveground forest floor vegetation biomass (i.e.
ferns, sedges/ grasses, forbs, low shrubs, and woody seedlings <0.25 m tall) was greater in
higher than lower deer density areas (657 vs. 252 kg/ha, P< 0.0001). Although bracken
fern and sedge dominated forest floor biomass in higher and lower deer density
categories, higher deer density sites had, on average, three times greater bracken fern,
sedge, and woody plant mass, and one-tenth the forb mass (Figure 1.1). Bracken fern and
sedge mass decreased as site productivity increased and mass was greater at higher-than
lower deer density over the entire range of site index (Figure 1.2A & B). Total
aboveground vegetation decreased with increasing site productivity but only in areas with
high deer densities. Sedge mass decreased with increasing stand age such that sedge

density was similar at high and low deer densities in stands >35 years old (Figure 1.2C).

29

In contrast, woody plant density <0.25 m in height increased with aspen stand age but did
not interact with deer density (Appendix 1.1D, Figure 1.2D).

Higher deer density decreased forest floor layer plant species richness and the
magnitude of the decrease was greatest at higher productivities (45 % decline in richer
sites vs. 21 % decline in poorer sites vs. their respective lower deer density site
productivity controls). (Figure 1.3, Appendix 1.2D, & 1.3). I found no interactions
between deer density and stand age or site productivity levels for Shannon-Weiner
diversity values (H’). H’ increased with increasing site index for both deer densities and
was uniformly lower (24%) in higher than lower deer areas across the range of SI (Figure
1.4). Stand age showed no relationship with diversity (Appendix 1.2D).

Deer density affected total shrub and tree density, but effects varied among height
classes and with stand age and site quality (Appendix 1.2A-C). In low deer density areas,
total stem density in all height strata generally increased with stand age and site index,
but deer sharply reduced woody stem density in 0.6-1.5 m and 1.5 —4.0 m strata over the
range of SI and stand age (Figure 1.5). For a data set restricted to a common range of SI
and age, higher deer density had 84.4% lower total stem density than lower deer density
(P=0.0003, Figure 1.1E), and only black cherry stems remained. For the 1.5 m to 3.9 m
tall height class, a zone affected by past deer browsing, results were similar to the 0.6-1.5
m tall height class with higher deer resulting in 91% lower stem densities (P=0.0092,
Figure 1.1F) with witch hazel, oaks and red maple nearly eliminated and mostly black
cherry stems remaining (Figure 1.5, Appendix 1.4).

Vegetation nitrogen

30

Greater total N was found in forest floor vegetation at higher than lower deer
density (Figure 1.6A.). This difference was not driven by higher tissue concentrations
(Appendix 1.5C-G), but by greater total vegetation mass due to increased sedge, bracken
fem, and woody seedling < 0.25 m tall (Figure 1.1A, B, & D). Tissue N concentrations
for bracken fern, sedge, and forbs increased with SI and decreased for tree seedling leaf
N and stem N concentrations with stand age (Appendix 1.5). In contrast to total forest
floor vegetation N content, non-bracken and sedge (i.e. tree/shrub seedling leaf (< 0.25 m
tall) + forb) N content was roughly twice as high in lower deer density areas (Figure
1.68) and deer density, stand age, and site index model effects were of approximately
equal magnitude (Appendix 1.53). Forest floor seedling + forb N increased with site
index and stand age (data not shown) and decreased with deer density (Figure 1.63). The
lack of interactions with deer density indicates that deer reduced N content similarly
across stand ages and site indices (Appendix 1.5).

Discussion

All lines of evidence used in this study to assess the deer population status (site
specific fecal pellet counts, historical track counts, and MDNR estimates) showed that
my relative deer density categories across the two ownerships were in fact different.
These differences in relative deer density levels strongly influenced aspen forest
understory plant composition, mass, structure and N content, although some of these
responses were dependent on site index and stand age. Deer affected forest floor
vegetation in younger more than older stands (range 12-44 year old), where higher deer
densities areas had much greater bracken fern and sedge mass than lower deer density

areas. The strong stand age x deer interaction on sedge mass could possibly be explained

31

 

by sustained low light availability beneath closed aspen canopies (Pinno et al., 2001).
Over time, the low light availability beneath aspen stands may diminish shade intolerant
sedge whose populations were initially elevated by the combination of high light
following harvesting and lack of browsing by deer. In aspen stands with elevated
ungulate densities, browsing did in fact cause canopies to be less dense and allowed for
more “old field” higher light demanding forb species to establish and grow while more
shade tolerant forbs were found in exclosures (Raymer 2000). Although all of my sites
were closed-canopy aspen stands, higher sapling density in lower deer density areas may
have intercepted more of the light transmitting through the aspen crowns further lowering
light levels at the forest floor. I did not measure light directly, but young stands at low
site index had few stems >0.6 m tall in either higher or lower deer density areas, and it
was in these stands that some of the largest differences in sedge and fern density between
higher and lower deer density occurred. Thus, I did not find indirect support for my
speculation that differences in light between higher and lower deer density areas affected
sedge and fern density.

The increase in bracken fern at higher site index suggests that less palatable
vegetation common to lower fertility-more disturbance prone areas (Berger and Kotar
2003) may become more common in higher fertility areas when deer browse pressure is
high. The reduced light associated with increased fern cover may have long-term
implications for seedling establishment (George and Bazzaz 1999) as well as
consequences to below- ground nutrient cycling (for vegetation induced shifts in
belowground cycling see Pastor et al., 1993). Increases in non-browsed species (fern and

sedge) may also alter disturbance regimes on richer more productive sites (more frequent

32

surface fires etc) such that in forests that are not evolutionarily adapted to fire (northern
hardwoods) we may see drastic changes in all vertical layers following a ground layer fire
carried by bracken and sedge litter.

Changes in composition associated with high deer density might be partially
explained by species characteristics. Bracken fern and sedge are both rhizomotous,
clonally growing species, which lack permanent aboveground stems. Additionally
sedge’s intercalary meristems may reduce their sensitivity and susceptibility to deer
browsing. Although I found no differences in N concentrations, an index of protein
concentrations, between sedge, fern, woody seedling and forbs, sedge and fern may be
less desirable as browse because of accumulated silica in foliage (Prychid et al., 2003).
Thus, over time, as deer selectively remove more nutritionally desirable and non-
rhizomous forb and tree species, fern and sedge, via clonal growth, are able to quickly
sequester the growing space Opened by browsing. Ultimately this may lead to their
dominance, and in turn may be perpetuated by resource sequestration (e. g. light
interception by bracken fern (de la Cretaz and Kelty 1999, George and Bazzaz 1999),
nutrients (Knoop and Walker 1985), moisture (Dodd et al., 1998), and germination sites
(Horsley 1993a & b)).

Forest floor vegetation N content was elevated in higher deer density areas, and
these effects were due to increased amounts of vegetation (mostly sedge and bracken
fern), and not by increases in tissue N concentrations. It is important to note, however,
that forest floor vegetation samples did not include shrub and tree stems >025 m tall.
Stem density was lower in the 0-0.6 m tall class but greater in the 0.6 - 1.5 class at lower

deer densities. If mass 0.25 — 1.5 m had been included, absolute differences in total N

33

between higher and lower deer densities would probably be diminished somewhat, but
rank differences would likely be preserved because there is only one additional 0.6 - 1.5
m tall stem per 5 m2. The increase in forest floor vegetation caused by deer should not be
misinterpreted as an increase in available N to deer, as the increase occurs in sedge and
fern.

Greater forest floor seedling biomass (0-25 cm, Fig 1D) and stem density in the 0-
0.6 m tall class in higher deer density areas is a reflection of high browse pressure
lowering the overall stature of shrubs and tree seedlings, and a snow pack which provides
minimal browse protection to seedlings throughout the winter months. These older,
stockier individuals, in shorter height classes, fail to grow into the higher classes and
result in a drastically altered vertical structure. Although my study was not designed to
test hypotheses concerning fem-tree seedling resource competition, bracken fern’s
potential impacts on seedling resources, including light, could be inhibiting seedlings
from growing into taller height classes (George and Bazzaz 1999).

The near complete loss of seedlings in the1.5m — 4m height class indicates both
the legacy of past deer densities and it portends a shift in tree species composition in the
future, if the stands are not maintained for aspen by clearcutting. Based on my data, areas
with high deer density will likely have black cherry and little else, as oak, red maple and
witch hazel were essentially extirpated by deer. Others, (Tilgham 1989, Yahner 1995,
Raymer 1996, and Horsley et al., 2003) have documented increasing stem densities of
black cherry in higher ungulate areas along with red maple. It is possible that the red

maple component, or lack there of, in the understory was no more than an artifact of past

34

stand harvesting or cleaning practices, which could have promoted or reduced seed
availability or sprout origin stems of these species.

As deer browse has favored dominance by sedge and fern, which are less
palatable and/or inaccessible to deer for portions of the year, browsing pressure on the
remaining palatable plants (i.e. woody seedlings and forbs) will most certainly increase.
Bergvall et al., (2006) termed this “neighbor contrast susceptibility”. This mechanism is
proposed to hold true as long as deer are unselective as to the area that they browse while
being highly selective as to what they browse. Therefore, any management regime that
helps to maintain elevated deer densities in a fixed locality for extended periods of time
(food plots, supplemental feeding, natural deer yarding areas, and agricultural/forest
matrices) will promote the neighbor contrast susceptibility mechanism. Hunt clubs like
MFL, that have maintained high deer densities for decades, are a prime example of the
conditions promoting neighbor susceptibility, and as such, effects of deer on vegetation
can be expected to be greatest here and in other areas with sustained high long term deer
densities.

Given that deer in northern lower Michigan have home ranges that are often
between ~220 ha (2.2 kmz) and 500ha (5 km2)(Sitar 1996, and Garner 2001), and winter-
summer migrations in the area can be as long as 10km (Garner 2001), the minimum
buffer used in this study between higher deer density areas and lower deer density areas
(1.6 km) was a potential source of error. Given the multiple lines of evidence used to
determine if deer density categories were in fact different (explained above), I feel that
overlapping use of the area by deer was minimal, the categorized deer densities

differences significant, and that the observed deer effects on the system were real.

35

Altered composition and structural attributes caused by chronic browsing is not
limited to the aspen stands I studied. Similar changes often accompanied by increased
density of vegetation that may further hinder the establishment of deer sensitive trees,
shrubs and herbs has been found in other forest communities (Ferns — Horsley et al.
1993b, George and Bazzaz 1999, Sedge -— Randall and Walters unpublished date,
Weigrnann 2006). Management interventions in addition to decreasing deer populations
may be needed to restore the composition and structure of affected stands and these
interventions could be expensive (direct planting or seeding), time consuming (decades,

not years, to restock stands), and/or socially unpopular (herbicides).

36

Literature Cited:

Alverson, W.S. and D.M. Waller, 1997. Deer populations and the widespread failure of
hemlock regeneration in northern forests. Pp280-297 in W. McShea and J.
Rappole, eds., The science of overabundance: Deer ecology and population
management, Smithsonian Institute Press, Washington, DC.

Alverson, W.S., D.M. Waller, and S.L. Solheim, 1988. Forest too deer: Edge effects in
northern Wisconsin. Conservation Biology 2(4):348-358.

Anderson, R.C., E.A. Corbett, M.R. Anderson, G.A. Corbett, and T.M. Kelley, 2001.
High white-tailed deer density has negative impact on tallgrass prairie forbs.
Journal of the Torrey Botanical Society 128(4):381-392.

Augustine, D. J ., L.E. Frelich, and PA. Jordan. 1998. Evidence for two alternate stable
states in an ungulate grazing system. Ecological Applications 8(4):1260-1269.

Balgooyen, GP and D.M. Waller 1995. The use of Clintonia borealis and other
indicators to gauge impacts of white-tailed deer on plant communities in northern
Wisconsin, USA. Nat. Areas J. 15(4):308-318.

Bancroft, T.A., 1964. Analyses and inferences for incompletely specified models
involving the use of preliminary tests of significance, Biometrics 20:427-442.

Berger, TL. and J. Kotar. 2003. A guide to forest communities and habitat types of
Michigan. Published by the Department of Forest Ecology and Management,
University of Wisconsin-Madison.

Bergvall, U.A. and O. Leimar. 2005. Plant secondary compounds and the frequency of
food types affect food choice by mammalian herbivores. Ecology 86 (9): 2450-
2460.

Bergvall, U. A., P. Rautio, K. Kesti, J. Tuomi, and O. Leimer. 2005. Associational effects
of plant defences in relation to within- and between-patch food choice by a
mammalian herbivore: neighbour contrast susceptibility and defence. Oecologia

l47(2):253-260.

Cleland, D.T., J .3. Hart, G.E. Host, K.S. Pregitzer, and CW. Ramm et al. 1994. Field
guide: ecological cloassification and inventory system of the Huron-Manistee
National Forests. US. Forest Service, Huron-Manistee National Forest, Cadillac,
Michigan, USA.

Cote, S.D., T.P. Rooney, J .P. Tremblay, C. Dussault, and D.M. Waller. 2004.

Ecologicalimpacts of deer overabundance. Annual review of ecology and
evolutionary systems 35:1 13-147.

37

Crete, M., J .P. Ouellet, and L. Lesage. 2001. Comparative effects on plants of
caribou/reindeer, moose and white-tailed deer herbivory. Arctic 54(4):407-417.

Danell, K., R. Bergstrom, P. Duncan, and J. Pastor. 2006. Large herbivory ecology,
ecosystem dynamics and conservation. Cambridge university press

DeCalesta, D.S. 1997. “Deer, Ecosystem Damage, and Sustaining Forest Resources,”
online proceedings of Deer Management in Maryland Conference, October 27,
1997.

de la Cretaz, A. and M.J. Kelty 2002 Development of tree regeneration in fem-dominated
forest understories after reduction of deer browsing. Restoration Ecology
10(2):416-426.

Dodd, M.B., W.K. Lauenroth, and J.M. Welker 1998. Differential water resource use by
herbaceous and woody plant life-forms in a shortgrass steppe community.
Oecologia 117:504-512.

Frelich, LE. and CG. Lorimer. 1985. Current and predicted long-term effects of deer
browsing in hemlock forests in Michigan, USA. Bio. Con. 34(2):99-120.

Garner, MS. 2001. Movement patterns and behavior at winter feeding and fall baiting
stations in a population of white-tailed deer infected with bovine tuberculosis in
the northeastern lower peninsula of Michigan. Ph.D. Dissertation. Michigan State
University East Lansing. 270pp.

George, LC. and FA. Bazzaz. (1999) The fern understory as an ecological filter:
growth and survival of canopy-tree seedlings. Ecology 80 (3):846-856.

Hester, A.J., L. Edenius, R.M. Buttenschon and AT Kuitters. 2000. Interactions between
forests and herbivores: the role of controlled grazing experiments. Forestry
73(4):381-391.

Hester, A.J., P. Millard, GJ. Baillie, and R. Wendler. 2004. How does timing of
browsing affect above- and below- ground growth of Betula pendula, Pinus
sylvestris, and Sorbus aucuparia? Oikos 105-536-550.

Hill, H. 2000. The 2000 deer pellet group surveys. Wildlife Report No. 3324. Michigan
Department of Natural Resources.

Horsley, SB. 1993 a. Role of allelopathy in hay-scented fern interference with black
cherry regeneration. Journal of Chemical Ecology 19(11):2727-2755.

----- 1993b. Mechanisms of interference between hayscented fern and black cherry.
Canadian Journal of Forestry Research 23:2059-2069.

38

 

Horsley, S.B., S.L. Stout, and D.S. deCalesta. 2003. White-tailed deer impact on the
vegetation dynamics of a northern hardwood forest. Ecological Applications
13 :901 1 8.

Knoop, W.T. and B. H. Walker. 1985. Interactions of woody and herbaceous
vegetation in a Southern African savanna. Journal of Ecology 73:235—253.

Leiffers, V.J., K.J. Stadt, S. Navratil. 1996 Age structure and grth of understory white
spruce under aspen. Canadian Journal of Forest Research 26: 1002-1007.

Lundgren, AL, and WA. Dolid. 1970. Biological grth functions describe published
site index curves for Lake States timber species. USDA Forest Service Research
Paper NC-36.

Manseau, M., J. Huot, and M. Crete.1996. Effects of summer grazing by caribou on
composition and productivity of vegetation: community and landscape level.
Journal of Ecology 84:503—513.

Marquis, DA. 1974. The impact of deer browsing on Allegheny hardwood regeneration.
US. Department of Agriculture Forest Service Res. Rep. NE-47

----- 1981. Effect of deer browsing on timber in Allegheny hardwood forests of
northwestern Pennsylvania. The impact of deer browsing on Allegheny hardwood
regeneration. US. Department of Agriculture Forest Service Res. Rep. NE-57

McInnes, P.F., R.J. Naiman, J. Pastor, and Y. Cohen. 1992. Effects of moose browsing on
vegetation and litter of the boreal forest, Isle Royale, Michigan, USA. Ecology
73:2059-2075.

McWilliams, W.H., T.W. Bowersox, P.H. Brose, D.A. Devlin, J .C. Finley, K.W.
Gottschalk, S. Horsley, S.L. King, B.M. LaPoint, T.W. Lister, L.H. McCormick,
G.W. Miller, C.T. Scott, H. Steele, KC. Steiner, S.L. Stout, J .A. Westfall, and
R.L. White. 2002. Measuring tree seedlings and associated understory vegetation
in Pennsylvania’s forests. Proceedings of the fourth annual Forest Inventory and
Analysis Symposiumpp 22-26.

Milchunas, D.G. and WK. Lauenroth. 1993. Quantitative effects of grazing on vegetation
and soils over a global range of environments. Ecological Monographs 63(4):327-
366.

Parks, C.G., L. Bednar, and AR. Tiedemann. 1998. Browsing ungulates: and important
consideration in dieback and mortality of pacific yew (taxus Brevifolia) in a

northeastern Oregon stand. Northwest Science 72(3):l90-197.

Pastor, J. and R.J. Naiman. 1992. Selective foraging and ecosystem processes in boreal
forests. The American Naturalist 139(4):690-705.

39

Pastor, J ., B. Dewey, R.J. Naiman, P.F. McInnes, and Y. Cohen. 1993. Moose browsing
and soil fertility in the boreal forests of Isle Royal National Park. Ecology
74(2):467-480.

Pastor, J., B.Dewey, R. Moen, D.J. Mladenoff, M. White, and Y. Cohen. 1998. Spatial
patterns in the moose-forest-soil ecosystem on Isle Royale, Michigan, USA.
Ecological Applications 8(2):411-424.

Pinno, B.D., V.J. Leiffers, and K.Stadt. 2001. Measuring and modeling the crown and
light transmission characteristics of juvenile aspen. Canadian Journal of Forest
Research 31 :1930-1939.

Prychid, C.J., P.J. Rudall, and M. Gregory. 2003. Systematics and biology of silica
bodies in monocotyledons. The Botanical Review 69(4):377-440.

Raymer, D.F.N. 1996. Current and long-term effects of ungulate browsing on aspen stand
characteristics in northern lower Michigan. M.S. thesis. Michigan State
University. East Lansing. 206pp.

------ . 2000. Effects of Elk and white-tailed deer browsing on aspen communities and
wildlife habitiat quality in northern lower Michigan: an 18-year evaluation. Ph.D.
Dissertation. Michigan State University. East Lansing. 371pp.

Rooney, T.P. 1997. Escaping herbivory: refuge effects on the morphology and shoot
demography of the clonal forest herb Maianthemum canadense. J. Torr. Bot. Soc.
124(4):280-285.

------ . 2000. Influence of light, soil fertility, and deer browsing on the spatial structure,
relative grth rates, and population demography of Trillium grandiflorum. Ph.D.
Dissertation. University of Wisconsin-Madison.

Rooney, T.P. and D.M. Waller. 2003. Direct and indirect effects of white-tailed deer in
forest ecosystems. Forest Ecology and Management. 181 : 165-176.

Russell, F.L., D.B. Zippin, N.L. Fowler. 2001. Effects of White-tailed Deer (Odocoileus
virginianus) on Plants, Plant Populations and Communities: A Review. The
American Midland Naturalist. 146(1):1-26.

Sitar, KL. 1996. Seasonal movements, habitat use patterns, and population dynamics of
white-tailed deer in an agricultural region of northern lower Michigan. MS.
Thesis Michigan State University. East Lansing. 130pp.

Swift, E. 1948. Deer select most nutritious forages. Journal of Wildlife Management
12:109-110.

40

Tilgham N.G. 1989. Impacts of white-tailed deer on forest regeneration in northeastern
Pennsylvania. Journal of Wildlife Management 53:524-532.

Tracy, BF. and DA. Frank. 1998. Herbivore influence on soil microbial biomass and
nitrogen mineralization in a northern grassland ecosystem: Yellowstone National
Park. Oecologia 114:556-562.

Trembley, J .P., J. Huot, and F. Potvin. 2006. Divergent nonlinear responses of the boreal
forest field layer along an experimental gradient of deer densties. Oecologia
150:7 8-88.

United States Department of Agriculture.1972. Soil Survey — Gladwin County, Michigan.

----- . 1972. Soil Survey - Roscommon County, Michigan.

Waller D.M. and W.S. Alverson. 1997. The white-tailed deer: a keystone herbivore.
Wildlife Society Bulletin 25:217-226.

Weigrnann, SM. 2006. Fifty years of change in northern forest understory plant
communities of the Upper Great Lakes. Ph.D. Dissertation. University of
Wisconsin-Madison. 206pp.

Weisberg, P.J., F. Noavia, and H. Bugmann. 2005 Modeling the interacting effects of
browsing and shading on mountain forest tree regeneration (Picea abies).
Ecological modeling 185:213-230.

Wisdom, M.J., M Vavra, J .M. Boyd, M.A. hemstrom, A.A. Ager, and B.K.Johnson.
2006. Understanding ungulate herbivory — episodic disturbance effects on

vegetation dynamics: knowledge gaps and management. Wildlife Society Bulletin
342283-292.

Yanher, RH. 1995. Eastern deciduous forest: ecology and wildlife conservation
University of Minnesota press, Minneapolis. 220pp.

41

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

500 A. I Bracken fern
’8 1 P < 0.0001
£
go 300
S
w T
E 100 . 1
A 300 B- Sedge
“5 l P = 0.0001
.1:
So 200 ‘ 1
'35:
(D
«‘6’ 100‘ T
2 r
15 C. Forbs
’3 P = 0.0001 l
.c:
a) 10‘
'3,
i, 1
co 5 w
2
{—3—7
15 D. Tree andsTrub seedlings
Q <0.25 m tall
g) 10 . f P=0.0003
=5 1
VJ
a 5 .
E
600 ETrees and shrubs '
a 0.6m - 1.5m tall L
P = 0.0092
“é, 400‘
E
8
m 200«
i—LL
600 -F.Trees and shrubs
a 1.5m - 4.0m tall
4: P = 0.0003
h 400‘ f
E
8
m 200‘
0 ‘ 1

 

 

 

higher lower

Deer density

Figure 1.1. Mean biomass (i 1 SE) for bracken fern (2A), sedge (ZB), herbs (2C), and
seedlings (2D) and mean (:t: ISE) # of seedlings stems (0.6m — 1.5m in
height) / ha (2E), and mean (1: ISE) # of seedlings stems (1 .5m — 4m in
height) / ha (2F), for lower and higher deer density areas.

42

 

 

 

 

    

 

 

 

 

 

 

 

 

—0- Higher deer
"-0-" Lower deer
1200 . 600
,3 1000 ,A. Bracken fern ,3 500 j C. ’ Sedge
o
E, 800* . .0 .3 3,400.. o
v 600 ‘ . 0.0. v
a O ' a 300‘
a 400 l . o . . .. <6 00 ‘
S 200 . (3.60.303 ""00 g 2 .0 o O. o
m 0 ‘ O O O 8%..‘13 m 100 #:50‘00000000000 . .
a 500 . ,3 60 in Tree and Shrub 0.
§ 400 4 a Seedling
:5 40+
g 300 ‘ a
E .
.2 200 g 20.
m 100 w m
12 14 16 18 20 22 24 26 10

 

Site Index
(m at 50years)

Stand age

Figure 1.2. Higher and lower deer density average bracken fern (2A) and sedge (2B)
biomass and their relationship to site index, as well as higher and lower deer
density average sedge biomass and its relationship to stand age (2C). Seedling
biomass (2D) vs. stand age.

43

 

 

 

 

0 higher deer, high site index
0 higher deer, low site index
v lower deer, high site index
m A lower deer, low site index
a)
'g 21
g- 18 «
“5 15
r...
g 12 ‘
g 9
z 5 .

 

 

 

3 6 9 12 15 18
Sample area (m2)

Figure 1.3. Species per unit area sampled by deer density (lower and higher and site
productivity (low and high) categories.

44

 

 

2.4 -—0— higher deer
"'0"- lower deer

 

 

 

 

29

.2

D 2.0‘

83

.E

g 1.6<

é:

o

E 1.21

<6

.1:

m 08 . 9'. .1. 4 . 4 .

12 14 16 18 20 22 24 26

SiteIndex
(matSOyears)

Figure 1.4. Shannon-Weiner diversity indices across site productivity gradients in Lower
vs. Higher deer density areas. Higher deer densities areas did not show a
relationship between SI and diversity while lower deer areas did. The overall
least squared means were significantly higher in lower deer areas (1.7) verses
higher deer areas (1.24).

45

 

 

 

 

 

 

A.
p I Black cherry
g lower Red maple
1: E Witch hazel
:6; _ E] Oaks
Q lugher \
1.5-4.0mtall Other
3 * ***
’3 lower *
5
-o
8
O

0.6 - 1.5m tall

I l

500 1000 1500 2000 2500

”i
o I-‘

  
   

    
 

 
      

1
l . r r I I

0 2000 4000 6000 8000 10000
Stems/ ha

Figure 1.5. Stems/ha for a select group of species (black cherry, oak, red maple, witch-
hazel, and a combined group “other”) by deer density (lower & higher). All
stands represent a uniform subset of stands by age and site index. Symbols (*
and **) at the end of the columns represent significantly different (P=0.0247
and P=0.0060 respectively) mean total stems/ha between lower vs. higher deer
density areas. Symbols *, **, & *** above column represent significant
differences (P=0.05, P=0.001, and P<0.0001 respectively) within an
individual species between higher and lower deer density.

46

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

A 12 4A, All Vegetation
g; 10‘ I P<0.0001
ii 8 .
s 6.
50
E 4* 1
Z 2 «
A 0'4 egetation B.
g) 0 3 excluding bracken I
.2 ' fern and sedge
Z 02 .P=0.0131 j
0 .
go I
.2 0.1 1
2

0.0 . .

Higher Lower
Deer density

Figure 1.6. Total vegetation nitrogen (kg / ha) (Fig. 6A) and vegetation nitrogen (kg / ha)
excluding bracken fern and sedge (Fig. 6B) in areas with lower and higher
deer densities. error bars represent i ISE.

47

Appendix 1.1. Results of a standard least squares mixed model for the effects of deer
density, site index, stand age and their interactions on bracken fern, sedge,
forb, seedling < 25 cm, and # of stems/ha (0.6m — 4m). Interactions with P
> 0.25 were pooled with the error term (Bancroft 1964) and the models
rerun.

 

 

 

 

Vegetation biomass Anova Effects SS F P

A- Bracken Fern Deer 12655046 28.162 <0.0001
Site index 2815249 6.2648 0.0155
Age 1421.7 0.0316 0.8595
DxSI 1021506 2.2732 0.1377
Adj. R2 0.3399

B- Sedge Deer 24.642 23.42 <0.0001
Site index 3.293 3.13 0.0831
Age 1.2804 1.22 0.2754
DxA 4.436 4.22 0.0454
DxSI 4.655 4.42 0.0406
SIxA 1.823 1.73 0.1942
Adj. R2 0.4242

C- Forb Deer 1275.25 32.26 <0.0001
Site index 11.013 0.2786 0.5999
Age 147.46 3.731 0.0589
DxSI 184.99 4.6799 0.0351
Adj. R2 0.3831

D- Seedling (<0. 25m) Deer 494.266 14.153 0.0004
Site index 27.897 0.799 0.3756
Age 611.221 17.502 0.0001
DxSI 198.096 5.672 0.0209
Adj. R2 0.4500

 

48

Appendix 1.2. Results of a standard least squares mixed model for the effects of deer
density, site index, stand age, and their interactions on seedling stem
density by height class and Shannon-Weiner diversity indices. Interactions
with P > 0.25 were pooled with the error term (Bancroft 1964) and the
models rerun.

 

 

 

 

Stem density Anova Effects SS F P
A- stems (0 to 0.6m) / ha Deer density 31446045 1.4722 0.2466
Site index 99692286 4.6672 0.0500
Age 56610151 2.6502 0.1275
AxSI 77165216 3.6125 0.0797
Adj. R2 0.3050
B- stems (0.6m to 1.5m) / ha Deer density 12212029 6.3286 0.0247
Site index 87107.4 0.4514 0.5126
Age 1364515 0.7071 0.4145
Adj. R2 . 0.1669
C- stems (1 .5m to 3.99m) / ha Deer density 317104.9 10.7297 0.0060
Site index 47923.3 1.6215 0.2252
Stand Age 39779.9 1.346 0.2668
Adj. R2 0.4344
D- Understory forb Shannon —
Weiner Diversity Index Deer density 2.793 30.65 <0.0001
Site index 1 .295 14.208 0.0005
Stand Age 0.0019 0.2086 0.6504
DxA 0.173 1.902 0.1757
AxSI 0.559 6.138 0.0177
Adj. R2 0.5075

49

Appendix 1.3. Results of a standard least squares mixed model for the effects of deer
density, site index, and their interactions on species richness at a defined
area (3m2, 6m2, 9m2, and 12m2).

 

 

 

 

Sampling Area Effect SS F P
3m2 Deer 655.4 64.5 <0.0001
Site index 49 4.8 0.0305
str 60.8 5.99 0.0162
6m2 Deer 936.4 70.9 <0.0001
Site index 9 0.6817 0.411
DxSI 139.2 10.547 0.0016
9m2 Deer 1 149.2 161.3 <0.0001
Site index 98 13.76 0.0003
mm 65.6 9.21 0.0031
12m2 Deer 1369 240.8 <0.0001
Site index 29.2 5.1 0.0258
st1 108.2 19.023 <0.0001

50

Appendix 1.4. Results of a standard least squares model for the effects of deer density,
stand age, and site productivity by species (Black cherry, Red maple, and
Witch-hazel) on seedling stem density by height class. Interactions with P
> 0.25 were pooled with the error term (Bancroft 1964) and the models
rerun.
Stem density P ratio P

Species AOVA effects SS

 

 

 

 

 

 

 

 

 

A- stems (0-0.6m) Black cherry Deer 1230322 0.0682 0.7984
Site Index 5080728 0.2815 0.6054
Stand Age 14596792 0.8088 0.3862
Adi. R2 01191
Red Maple Deer 60385 0.0047 0.9463
Site Index 42442731 3.3319 0.0952
Stand Age 5817354 0.4567 0.5131
Adi. R2 0.1097
Witch hazel Deer 47540168 5.6163 0.0354
Site Index 8725328 1.0308 0.3300
Stand Age 3103245 0.3666 0.5561
Adi. R2 0.2161
B- stems (0.6-1.5m) Black cherry Deer 17118.6 0.1951 0.6720
Site Index 3025518 3.4479 0.1057
Stand Age 10363284 11.8099 0.0109
DxA 7273174 8.2885 0.0237
DxSI 353571.7 4.0293 0.0847
AxSI 2795297 3.1855 0.1175
DxAxSI 3128044 3.5647 0.1010
Adj.R2 0.7455
Red Maple Deer No red maple stems found
Site Index in this class
Stand Age
Adi. R2
Witch hazel Deer 34245647 23.99 0.0005
Site Index 935.8 0.0066 0.9369
Stand Age 5661601 3.9665 0.0718
DxA 5829616 4.0842 0.0683
Adi. R2 0.6247
C- stems (1.5 - 3.99m) Black cherry Deer 2661852 4.885 0.0442
Site Index 33178.3 0.6089 0.4482
Stand Age 79.9 0.0015 0.9700
Adi. R2 0.1725
Red Maple Deer 153491 5.985 0.0308
Site Index 11172 0.4356 0.5217
Stand Age 13097.7 0.5107 0.4885
Adi. R2 0.1815
Witch hazel Deer 37452148 6.9746 0.0204
Site Index 6502349 1.2109 0.2911
Stand Age 3662197 0.6820 0.4238
Adi. R2 0.2229

 

51

Appendix 1.5. Results of a standard least squares mixed model for the effects of deer
density, site index, stand age and their interactions on total vegetation N,
forb and seedling N, bracken % N, sedge %N, herb % N, seedling leaf %
N, seedling stem %N, Interactions with P > 0.25 were pooled with the

error term (Bancroft 1964) and the models rerun.

 

 

 

 

 

 

 

Vegetation nitrogen Effect SS F P

A- Total vegetation N Deer 27.94 33.09 <0.0001
Kg N/ha Site index 1.108 1.312 0.2572

Age 0.7437 0.8808 0.3523

SIxA 1.3558 1.606 0.2107

Adj. R2 0.3916

B- Forb & Seedling Deer 9.233 6.255 0.0157
Kg N/ha Site index 10.093 6.838 0.0118

Age 11.783 7.983 0.0068

Adj. R2 0.2703

C- Bracken % N Deer 0.00003 0.0003 0.9859
Site index 0.8068 7.8219 0.0077

Age 0.3339 3.2372 0.079

Adj. R2 0.1252

D- Sedge % N Deer 0.0020 0.0406 0.8411
Site index 0.7842 15.818 0.0002

Age 0.0779 1.5724 0.2154

Adj. R2 0.1903

E— Forb %N Deer 0.0281 0.2524 0.6189
Site index 0.5815 5.2217 0.0293

Age 0.2057 1.847 0.1839

DxSI 0.0927 0.8323 0.3686

DxA 0.3165 2.842 0.1019

SIxA 0.7443 6.684 0.0147

DxSIxA 0.7046 6.327 0.0173

Adj. R2 0.3134

F - Seedling Leaf % N Deer 0.0367 0.7392 0.3944
Site index 0.0006 0.0119 0.9137

Age 0.4833 9.7359 0.0031

DxSI 0.2974 5.9912 0.0183

SIxA 0.3382 6.8130 0.0122

Adj. R2 0.1939

G- Seedling stem %N Deer 0.0106 0.3979 0.5319
Site index 0.0094 0.3534 0.5556

Age 0.1634 6.1151 0.0179

SIxA 0.0934 3.5001 0.0689

Adj. R2 0.1143

52

CHAPTER 2

SEPARATIN G DEER BROWSE AND SEDGE EFFECTS ON UNDERSTORY
VEGETATION DYNAMICS IN TEMPERATE
NORTHERN HARDWOOD FORESTS.

Executive summagy

High deer populations since the mid 1970’s in Michigan’s central Upper
Peninsula have coincided with increased abundance of non-browsed or browse resistant
species such as upland sedge (Carex pensylvanica Lam.) and ironwood (Ostrya
virginiana), and a decrease in browse preferred tree seedlings and saplings (e. g. sugar
maple). It has been shown in systems that non-browsed species may persist, even if deer
are removed, by outcompeting browse sensitive species for resources, yet little is known
if the vegetation pressures from non-browsed species are additive to the pressures
associated with deer or if the non-browsed species pressures have become the primary
driver of failed regeneration in the system. I attempted to answer if deer , sedge, or deer +
sedge pressures are driving the vegetation and structural shifts found in areas with
elevated deer densities and widespread sedge covers. I separated deer and understory
vegetation (predominantly sedge) effects on vegetation-resource dynamics in selection
harvested stands with high (> 20 kmz) long-term deer densities with experiments that
included vegetation manipulations (all understory vegetation removal, vegetation
removal plus scarification, sedge removal, control) and deer treatments (exclosures/open)

After four years of deer exclusion and/or vegetation manipulation it was clear that
deer not sedge populations were driving most of the responses that I measured. Areas

protected from deer had significantly higher coverages of deer sensitive species

53

(Mianthemum canadensis, Trillium grandiflorum etc) while areas open to deer had
greater coverages of “weedy” species (dandelion, goldenrod) and browse tolerant Carex.

Deer effects on forest structural characteristics were clearly demonstrated as well,
as no seedling of any species grew to a height that was above the reach of deer over the
course of this study 4 years. Naturally established sugar maple seedlings were hardest hit,
being eliminated above 0.25 min areas open to deer, while planted sugar maple seedling
growth was reduced by ~1/3. Deer even reduced or outright eliminated ironwood
seedlings above 100 cm.

Although deer were primarily responsible for declines in seedling growth via
browsing, vegetation (mostly sedge) also reduced seedling height and mass, especially at
higher light levels and this was possibly driven by differences in soil moisture resources
found between areas with and without sedge. I found no evidence that deer or treatments
altered extractable inorganic N or rates of nitrogen mineralization.

This study does show that current deer densities are too high for canopy
recruitment of northern hardwood regeneration. If or when deer numbers decline, the
time needed for seedlings to grow above the browse zone can be reduced (by 6-10 years)
to approximately 8-10 years if sedge competition is controlled. More work is needed to
economically control sedge in an ecologically fiiendlier manner to minimize effects to

forbs, desired seedlings, and fauna in the understory.

54

Introduction:

Compared to non-selective disturbance agents such as fire, landslides,
windthrows, and ice storms, large-scale ungulate herbivory may be a unique form of
disturbance (Hulme 1996). Ungulates are highly selective in the plants they consume
(Swift 1948, Crawley 1990, McInnes et al., 1992, Comett et al., 2000, and J efferies et al.,
1994) and as such, at high densities, large ungulates (e. g. deer, elk, moose) can alter plant
species composition by selectively foraging on highly palatable species while ignoring
species of low palatability (Coughenour 1991, DeCalesta 1997, Horsley et al., 2003,
Randall chapters 1). For plants browsed by ungulates, the direct losses associated with
the removal of biomass can be compounded in subsequent years as browsed plants alter
growth (increased shoot growth — McNaughton 1983, Danell et al., 1985, Bergstrom and
Danell 1987, Campa 1989, and Molvar et al., 1993), regrth nutrient status (higher
foliar N -— Du Toit et al., 1990) and secondary defense levels (lower tannin and/or lignin
levels - Du Toit et al., 1990) increasing the potential for future browsing events (Danell et
al., 1985) Conversely, some browsed species lower their aboveground mass and foliar N
concentrations in response to browsing while increasing root growth and secondary
defense levels (ether extracts — Campa 1989). As such, sustained (seasonally repeating)
high browsing pressure on seedlings can cause differing responses. In some instances
trees may survive the browsing events but be chronically deformed (shrub formation vs.
single dominant stem)(Gill 1992), while others that succumb to browsing result in
outright tree recruitment failure (Til gham 1989, Randall Chapter 1) which overtime, can
lead to decreased vertical structure complexity (Horsley et al., 2003, Russell et al., 2001,

Randall Chapter 1) and loss of intermediate forest canopy layers. Chronic browsing has

55

also been shown to alter nutrient cycling regimes (Pastor et al., 1993, Pastor and Cohen
1997), and vegetation species richness (increase - Raymer 2001 or decrease - Frelich and
Lorimer 1985, Chapter 1). Increased richness in high deer areas is usually synonymous
with an increase in “old field” species (Milchunas and Lauenroth 1993, Raymer 2001)
driven primarily by increased light resources reaching the understory caused by deer
induced thinning of the overstory. In closed canopy forests (lower light levels), deer
directly remove preferred understory species, which are usually replaced by a single
dominant browse resistant or non-browsed species that is tolerant of intermediate to
lower light levels.

The presence/absence of deer browse sensitive herbaceous plants (Clintonia
borealis (Ait.) Raf, Aralia nudicaulis L. [Balgooyen and Waller 1995], Maianthemum
canadensis Desf. [Balgooyen and Waller 1995, and Rooney 1997, 2001], Osmorhiza
claytonia (Michx), Arisaema triphyllum (L) Schott., and Actaea pachypoda Ell. [Webster
and Parker 2000], Smilacina racemosa L. and Uvularia spp. L. [Fletcher, et al., 2001]
and Trillium spp. L. [Augustine and Frelich 1998]) are often used as indicators of deer-
browsing pressure.

In Michigan’s central Upper Peninsula (Menominee County averages for ’57-’75
& ’76-’99 were 15.8 and 31.3 deer/kmz, respectively, from Michigan DNR fecal pellet
count data obtained from Bob Doepker) increased second growth timber harvesting
resulted in a sharp increase in deer population numbers in the mid 1970’s as deer
ntunbers responded to increased preferred browse. These high deer population areas are
increasingly found to be in close association with highly managed northern hardwood

stands that have understories dominated by browse resistant, disturbed- canopy loving

56

Carex pensylvanica Lam. (Pennsylvania sedge). Although the observed high sedge
densities have been linked to high deer densities areas (International Paper management
records, Chapter 1, Horsley 1993, Weigmann 2006), other factors such as invasive
earthworms (Hale et al., 2006), and selection harvesting practices (Metzger and Schultz
1984, Horsley 1990, but see Jenkins and Parker 1999) may also be important in the
establishment of sedge and the reduction in regenerating seedlings. Regardless of the
cause of high sedge densities, Carex spp. have been found in other systems to be strong
competitors with vegetation, including tree seedlings. Thus it is possible that both deer
and sedge affect understory vegetation dynamics, including those of juvenile trees

In northeastern hardwood systems, vegetation changes caused by deer have been
hypothesized to persist (i.e. are stable) even if deer are removed. In these systems,
established non-browse preferred species may have self-reinforcing mechanisms that
suppress reinvasion of browse-preferred tree and shrub seedlings and forbs independent
of the direct deer effects (Stromayer and Warren 1997, Augustine et al., 1998, de la
Cretaz and Kelty 2002, George and Bazzaz 1999). The thick sedge documented in
managed northern hardwood stands could represent an alternate stable vegetation state,
maintaining understory dominance perhaps, by it being a strong competitor for nutrients,
water and/or space.

In this study, my goal was to separate deer and understory vegetation
(predominantly sedge) effects on vegetation and resource dynamics in selection-
harvested stands with high (> 20 kmz) long-term deer densities. My experiments focused
on the effects vegetation manipulations (all understory vegetation removal, vegetation

removal plus scarification, sedge removal, control), deer treatments (exclosures/open)

57

and their interaction on understory vegetation dynamics. Specifically, I asked, 1) what are
the relative impacts of sedge-dominated understories and deer on tree seedling
establishment, survival and growth, and forb diversity?, 2) If deer and understory
vegetation affect growth, are these effects associated with soil nitrogen and water
availability, and 3) if there are effects of vegetation on growth and survival, how do these
effects vary with light availability?

Methods:

Field Experiments

Location: I have two field experiments, both of which are located on land formerly
owned and managed by International Paper (IP) in Menominee County, Michigan. Stands
are classified as upland northern hardwoods dominated by sugar maple (Acer saccharum
Marsh), with white ash (F raxinus Americana L.), basswood (T ilia americana), bitternut
hickory (Carjya cordiformis (Wang) K. Koch), black cherry (Prunus serotina Ehrh.), and
hop hombeam (Ostrya virginiana (Mill.) K. Koch) also being found on the sites. These
forests have been selectively harvested at 8-12 year intervals for approximately 50 years,
Typically each selective harvest entry reduced basal area from approximately 25.3-27.6
mz/ha to approximately 17 mz/ha. (For stand characteristics and histories per replicate
study site see Appendix 1). Despite post-harvest conditions that should favor it, there has
been little recruitment of Acer saccharum saplings from seedling classes for over 30
years (Mike Young, IP, personal communication). Sedge (mostly Carex pennsylvanica)
comprises over 80% of herb layer biomass in these stands (Randall and Walters,
unpublished data, field observation over last five years). The five main study sites are on

five different drumlins (< 3 km from one another), which are part of the Northern Lake

58

Michigan (Hermansville) Till Plain, with soils being moderately to well-drained
spodisols and alfisols (Albert 1995) (For detailed belowground descriptions per replicate
site see Appendix 2). Drumlin parent material (dolomite and limestone) is generally
within 9.1 — 15.2 m of the surface (Albert 1995). The relatively mild winter climate,
along with the landscape vegetation structure (hardwood dominated drumlins used for
browse interspersed with lowland cedar swamps used as thermal protection in winter)
creates conditions favorable for supporting high winter deer populations.
Experimental designs

In summer 2000, with assistance from International Paper researchers and land
managers I filtered their stand inventory database and created a list of potential stands
that were characterized as 1) dominated by sugar maple with trees already in the sawlog
class (IP’s M6 designation), 2) was recently (within 2-3 years) harvested with single tree
or small group selection cuts (2-3 trees maximum) and were therefore not eligible for
reentry within an 8-10 year research window, 3) were harvested to reduce BA to 17m2/ha,
and lastly, were located on IP land holdings in Faithom Township, Michigan. Potential
stands were visually inspected for topographic consistency, as well as stand overstory
composition, structure, and basal area retention. 1 selected five replicate drumlin sites
(each site was a single stand) and established on each a 400 m2 fenced (to 2.6 m height)
deer exclosure paired at close proximity with a 400 m2 unfenced area, all five exclosures
and paired open areas were centered under harvest gaps (1-2 dominant/co dominant trees
removed). All 400 m2 fenced and unfenced areas were split into four 10 x 10 m treatment
plots that received one of four randomly assigned vegetation manipulations in mid-

August 2000. These were: all vegetation killed with glyphosate herbicide followed two

59

weeks later by a mechanical soil scarification treatment to expose mineral soil (Herbicide
+ Scarification), all vegetation killed with glyphosate with no scarification (Herbicide),
weeding of sedge only (to mimic a sedge-specific herbicide)(Selective), and a control.
Although effective in promoting seedling establishment in many regions, scarification
alone was not used as a treatment since an earlier industrial trial indicated that it was
ineffective at decreasing sedge and increasing tree seedling densities (Mike Young,
International Paper, personal communication). Prior to setup, subplots were randomly
assigned in the lab and tagged in the field to various measurements (10 subplots /
treatment / year (2000, 2002 and 2004) for % cover estimates and destructive biomass
harvests, 10 subplots / treatment for planted seedlings (five for sugar maple, five for
white ash), 10 subplots / treatment to quantify and track naturally establishing seedling
cohorts in 2001 and 2002). No subplot was ever reentered after a destructive harvest.

To ensure adequate numbers of experimental tree seedlings, I supplemented
naturally establishing seedlings by planting two-month-old greenhouse raised sugar
maple and white ash germinants into 10 subplots (five subplots received five sugar
maple, five subplots received five white ash) in each treatment replicate in 2001. Seed
was collected from a Western New York site (USDA hardiness zone 4b) for sugar maple,
and white ash was purchased from Sheffield seed company in NY (collected from USDA
Hardiness zone 4b). Seed was stratified in a cold room following the protocol in Young
and Young (1992). Seedlings where raised in the greenhouse at Michigan State
University’s Tree Research Station before being transported to the field sites where they

were immediately planted.

60

It is widely known that increased canopy openness conditions (a proxy for light
availability) can alter seedling survival, growth, (Walters and Reich 1997, Beaudat and
Messier 1998) and competition pressures for resources from surrounding vegetation
(Davies et al. 1999). Because all treatment areas (described above) were in a fairly
narrow range of light levels (~ 6.5-11 % of open sky measurements explained below)
when the larger Deer/No Deer [nested treatment] study began, I located 60 pairs of two-
year old naturally established sugar maple seedlings within the general study area in mid
September 2000. These 60 pairs were distributed over a broader range of light levels (1 -
22 % canopy openness measured with an LAI 2000 at a point 10 cm from the seedlings
apical bud) enabling me to quantify if sedge affects seedling growth and survival
disproportionately as light levels change. All seedlings had an exclosure constructed with
field fence (1m radius x 122 cm tall) to protect from deer and one seedling from each pair
had all vegetation (primarily sedge) removed from within a 1 m radius by hand. A
follow-up spray application of Glyphosate (rate of lquart/acre) the following growing
season removed all returning vegetation.

Measurements:
Vegetation cover, diversity, seedling densities, and vertical structure

Tree seedling and other vegetation measurements (destructive and non-
destructive) were first taken prior to treatment in summer 2000 and culminated with final
measurements taken in summer 2005. In 2000, 2002, and 2004 two observers
simultaneous estimated % cover occularly for each species in 10-1m2 subplots / treatment
area. Ocular estimates were then averaged, recorded and Shannon-Weiner diversity

estimates were derived using the equation {)3 Pi * LogPi}. Seedlings <25cm in height

61

were counted and estimated % cover was obtained by species. At the same time, woody
plant vertical structure (height distribution of woody stemmed plants) was obtained by
measuring all seedlings (woody stems > 250m but less than 1.45 m). Seedling heights
were recorded to the nearest 10cm height class (0.25m —035m, 0.35m-0.45m. . .) by
species. Because Herbicide and Herbicide + Scarification treatments killed all advanced
regeneration at the outset of this study, and because of the limited seed fall into the sites,
future vertical structure development was hypothesized to be severely impacted over the
course of this study. As such, 1 limited vertical structure development results to Control
and Selective treatments. Ironwood (Ostrya virginiana) was uniformly the dominant tree
species in the seedling layer across all sites at the beginning of the study, structural
results for this species represent data from all five replicate sites. Sugar maple is also
represented but only two of the 5 sites are used as the remaining 3 had little to no
advanced regeneration prior to the study. In 2004, concerned with the low numbers of
sugar maple seedlings actually measured in the 10 subplots / treatment, I sampled the
remaining 34 unharvested subplots on each site, which were set aside for longer-term
measurements. Individuals in the 34 extra subplots were remeasured in 2005 to obtain
growth from ’04-’05. I non-destructively monitored planted sugar maple and white ash
seedling survival on a monthly basis from May through October in 2001 and then semi
annually (spring and fall) from 2002-2004.
Vegetation biomass and N content

In 2000, 2002, and 2004 I destructively harvested aboveground vegetation in the
core area (0.5 x 0.5 m) from the 10 l-m2 sub-plots / treatment area used to obtain % cover

by species. Destructive measurements provided a more robust data set, in that, occularly

62

estimated % cover data collected in 2000, 2002, and 2004 may have been unintentionally
biased in Controls due to heavy sedge cover. Harvested vegetation was separated in the
field into one of eight categories (forbs, sedge & grass, tree seedling leaves and stems,
recruit seedling leaves and stems below 25 cm, and recruit seedlings leaves and stems>
25cm). For clarification, a recruit seedling is and individual that is greater than 25 cm tall
but less than 1.45m tall. The sedge and grass category was dominated (> 90%) by Carex
pensylvanica Lam. and will hereafter be called sedge. All samples were dried at 70°C for
72 hours in the lab prior to measuring dry mass. All dried samples were passed through a
Wiley mill, pulverized in a hammer mill (KLECO - Kinetic Laboratory Equipment
Company, Visalia, CA) and analyzed for N concentration by the Dumas combustion
method with a CHN analyzer (Carlo Erba, Milan, Italy).

All sugar maple individuals were harvested from the 60 pairs across an openness
gradient in early fall 2004. Seedlings were harvested by clipping at the root collar,
aboveground parts were stored in ziplock bags, transported to the lab at MSU, and were
evaluated for total height growth prior to drying at 70 C for 72 hours at which time dry
mass was then obtained.

Light and soil resource measurements

I measured canopy openness (a proxy for light) in 2001 at 16 evenly distributed
points (each sampling point touched the comers of 4 unique subplots ensuring all 64
subplots in the core area had an associated openness measurement) in each of the
treatment core areas (8 x 8 m) within the Deer/No[nested treatments] with a dual LAI
2000 (LiCor Inc., Lincoln NE) setup. Both instruments were placed side by side for

calibration of the optical sensor in an open field where there would be no interference

63

from trees (setup at least 3X the canopy height away from the forest / field edge. One
LAI sensor and data logger was left in the open to continuously measure and log total
potential canopy openness, while the other LAI unit was taken to measure openness
levels under the canopy in each treatment. In the lab, data was transferred to a computer,
and the C2000 (program provided by LiCor) was used to synchronize time stamps and
calculate canopy transmittance (openness) as differences in total potential open sky
verses measurements collect under the canopy which intercepts a certain level of the total
potential transmittance at each site. Canopy openness (same LAI 2000 protocol as above)
was also measured at 1m above each of the 60 pairs (n=120) of naturally established
sugar maple seedlings, where half had vegetation removed and half were vegetated
controls.

Nine, one month long in situ incubations were used to measure soil inorganic
nitrogen(for dates of incubations see Table 2.8). I randomly punched eight pairs of soil
cores in each treatment area and immediately pulled one from each pair. The remaining
eight cores were incubated in situ for 30 days and the following extraction procedure was
used for both initial and final incubation samples. All eight pulled cores per treatment
were sieved to pass a 4 mm screen in the field, thoroughly homogenized, subsarnples
were stored in ziplock plastic bags, refiigerated @ 1°C, and transported to the lab. Soils
were extracted within 48 hours with 2M KCl (20g field moist soil with 50ml KCl) on a
shaker table for one hour, and allowed to settle for 30 minutes before being filtered
through Whatman® #42 filter paper. Soil moisture was determined gravimetrically at the
time of each extraction to allow calculation of NO3'-N and NH4+—N on a per unit dry soil

basis. All extracts were refiigerated until being measured for NH4+-N and NO3'-N

64

(measured within 30 days of extraction) on an 01 Alpkem Autoanalyzer (01 Analytical,
College Station, TX) by phenol-hypochlorite and cadmium reduction methods,
respectively. I expressed inorganic extractable N (N O3'-N and NHf-N) on a gram N /
gram dry soil basis. Rates of N-mineralization were calculated as the change in inorganic
N (NO3' + NH?) on a dry soil basis from initial to final measurements {(Final NO;' +
NH4+) - (Initial NO3' + NH?) / incubation period (days)}

Soil moisture measurements (15 over the course of the four growing seasons see
Table 2.6) were a combination of the nine initial moisture measurements collected from
the initial in situ N measurements and six measurements determined gravimetrically from
punched soil cores randomly taken in each Deer/ No Deer treatment. Soil cores were also
punched to obtain gravimetrically determined soil moisture content around each of the 60
pairs of seedlings at 11 measurement dates (see table 2.7).

Analysis:

Prior to setup, the four treatments (Control, Herbicide, Herbicide and
Scarification, Selective) were randomly assigned to one of the 10 x 10m vegetation
manipulation treatment areas within each of the Deer / No Deer treatments in each of the
five replicates (sites). For analysis, vegetation manipulation treatments were nested
within Deer / No deer treatment and denoted as Deer / No Deer [treatment] throughout
the paper. To analyze vegetation mass, N content, diversity, and structural change over
time I included year as a factor in initial ANOVA model with main effects Deer / No
Deer and Deer/No Deer [nested treatment] effects. Subsequent models were rerun to test
Deer / No Deer and Deer/No Deer [nested treatment] effects within a given year. Tukey-

Kramer HSD was used to test for significant differences (P<0.05) among years and then

65

for treatments within Deer / No Deer. If nested treatments were not significantly different
(P>0.05) the model was reduced and Deer / No Deer effects were tested with Student’s T
tests. For soil resources (extractable N, N mineralization rates, and soil water) tests were
similar to above models (least squares models) with the only difference being Year was
not added to the initial model.

Because established seedlings influence, to a great degree, the advanced
regeneration layer in future years, 1 limited most of the vertical structure tests to Control
and Selective treatments (the two herbicide treatments killed all advanced regeneration at
the outset of the project). For both ironwood and sugar maple advanced regeneration stem
densities, initial ANOVA tested vertical height class (nominal), Deer/ No Deer [nested
treatment] effects. For each species, nested vegetation treatment effects were not
significant in the model and were reduced to just Deer / No Deer treatments. Deer / No
Deer effects on stem densities were tested by individual height class with Student’s T
tests.

Using Deer / No Deer and Deer / No Deer[vegetation treatments] as covariate
predictors, I model planted seedling survival with Cox’s proportional hazards methods
(Cox 1972) in J MP 5.1 statistical software (SAS Institute, Cary, NC, USA). Deer had
such an overwhelming effect on planted seedling survival and growth that I restricted
analysis of growth to only those seedlings in exclosures. Inside exclosures, growth was
analyzed with least squares regression in mixed models with main effects of canopy
openness (continuous), treatment (nominal) and their interaction. If model results

indicated that the interaction term was insignificant beyond the threshold suggested for

66

 

pooling variances (P > 0.25, Bancroft 1964), the interaction term was removed and the
model was re-run.

In the small companion study, treatment (vegetation/vegetation removed ——
Nominal variable), % canopy openness (continuous), and their interaction effects on
naturally established sugar maple seedling growth, stem mass and soil moisture resources

were tested using least squares regression in J MP.

Results

Vegetation Biomass

Two years after vegetation treatments (2002), Selective, Herbicide and Herbicide
+ Scarification treatments had much lower sedge biomass than controls in both Deer
(average, 99.7% reduction) and No Deer (94.1% reduction) treatments (Figure 2.1A & B,
Table 2.1A). In 2004, sedge biomass had started to rebound but was still, on average, 50-
55% lower in the three vegetation manipulation treatments than in the Control. There was
a trend in both 2002 and 2004 for greater sedge mass in Herbicide + Scarified than in the
Herbicide treatment.

F orb mass was unaffected by vegetation treatments (Figure 2.1, Table 2.1). In
contrast, forb mass was 50 % lower in Deer than No Deer treatments, but only in 2002, as
forb mass rebounded to Control levels by 2004 in these Deer areas. Although deer had
modest and temporary effects on forb mass in vegetation manipulation treatments, they
had greater and more lasting effects on forb composition. In 2004, compared to the Deer
treatment, the No Deer treatment had greater coverages of the following deer sensitive
forb species (control treatment only): Maianthemum canadensis (700%), Dentaria

(243%), Trillium grandiflorum (233%), Uvularia grandifolia (117%), and Ozmorhiza

67

claytonia (95%), and tree species: Acer saccharum (79%). In addition, flowering trillium
density was significantly greater in No Deer treatments (2965/ha vs. 45/ha). In contrast to
No Deer treatments, Deer treatments had greater forb coverages of Carex pedunculata
(447%), Ribes (200%), T araxz'cum oflicinale (120%), Solidago spp. (100%), Rubus
hispidus (100%) and Carex pensylvanica Lam. (32%), and seedlings coverages of Abies
balsamea (50%), Acer rubrum (100%), and Fraxinus americana (173%). In 2002 and
2004 the Selection treatment had the highest H’ diversity, the Controls the lowest, and the
Herbicide and Herbicide + Scarification treatments were intermediate (Figure 23A,
Table 2.1 E). Deer/No Deer treatments had no effect on H’ diversity.

In Herbicide and Herbicide + Scarification treatments seedling mass was 98%
lower than in Control and Selective treatments in 2002 (Figure 2.1E & F, Table 2.1C),
but by 2004, seedling mass had rebounded and differences among vegetation treatments
were marginal (Table 2.1, P=0.0584). In the ten l-m2 subplots I found no recruit
seedlings (seedlings > 25cm) in any of the replicate sites in 2000. Harvests in 2002 and
2004 resulted in so few recruit seedlings being harvested in Control and Selective

treatments (none were harvested from the Herbicide and Herbicide + Scarification
treatments) that we do not present them here.

Vegetation N content

As one might expect, vegetation N content (kg N/ha) tracked overall mass. Prior
to vegetation manipulation treatments, no differences were found for sedge, forb, or
seedlings <25 cm N content between Deer / No Deer treatments and their respective
nested vegetation manipulation treatments (Figure 2.2, Table 2.2A,B, & C). In 2002,

system-wide N content was 90-98% lower for sedge, but was again driven by the 94-977

68

% lower mass due to my vegetation manipulation treatments (Figure 2.2A & B). By

2004, sedge N content was 54-60% lower than controls, which again follows the roughly

50% lower sedge mass.
Forb N content in 2002 was highly variable among vegetation treatments in the

Deer treatment, but overall, N content was roughly 50% of that found in the No Deer
treatment. By 2004, forb N content levels were not significantly different, but did parallel
the large increase in forb mass (6.53X) from 2002 to 2004, increasing forb N content by
~5.85X (Figure 22C & D Table 2.28).

Two years after treatment, seedling N content was consistently higher in Selective
treatment areas than Herbicide and Herbicide + Scarification treatments. By 2004, large
variation in seedling N content caused the Deer / No Deer [vegetation manipulation
treatment] effect to not be significant (Table 2.2C). Like forb N content, seedling N
content increased from 2002 to 2004 by over 7X (Figure 2.2E & F). Selective treatments

had consistently higher N content in both Deer and No Deer treatments while Herbicide +

Scarification had the lowest N content.
N Concentrations were unaffected by Deer / No Deer [Vegetation manipulation

treatments] in 2000, 2002, and 2004 samples of sedge (2000 P=0.7799, 2002 P=0.8207,

2004 P=0.6093), forb (2000 P=0.4432, 2002 P=0.8619, 2004 P=0.7205), or seedlings <
25cm (2000 P= 0.5104, 2002 P: 0.4733, 2004 P=0.2366).

Planted seedling survival and growth
Averaged over all vegetation treatments, survival of planted sugar maple

seedlings was 63% lower in Deer, than No Deer treatments (Figure 2.4A, Table 2.3A).

Neither deer nor vegetation treatments affected white ash survival, and white ash had

69

much greater survival than sugar maple (73.6% vs. 46.4% averaged across all treatments
and Deer/N0 Deer areas) (Figure 2.43, Table 3B).

Planted sugar maple seedlings were significantly shorter in Deer than No Deer
treatments (Figure 2.5A, P<0.0001). In the No Deer treatment, seedlings had the greatest
height growth in the most severe vegetation manipulation treatment (Herbicide +
Scarification), and the least in the Control (P=0.0321). Stem biomass patterns were

similar to those for height, with seedlings having significantly more stern mass in No

Deer areas (Figure 2.58, P<0.0001).
In the No Deer treatments, with increasing canopy openness planted sugar maple

seedlings increased in height growth in Selective and Herbicide + Scarified treatments
(Figure 2.6A), and increased in stem mass in all but the Control treatment (Figure 268).
White ash height grth also increased with canopy openness in Control and Herbicide
treatments, while its stem mass increased with canopy openness in all treatments (Fig 6C
& D). Over a broad range of light (1-22%), naturally established sugar maple height and
mass showed similar results. Tree seedling growth (height and stem mass) increased as

canopy openness increased especially in high light areas where sedge was removed.

(Figure 2.7A & B).

Vertical structure
Vertical development of ironwood seedlings and saplings was not affected by

treatment (Control and Selective) but deer reduced stems > 100 cm (Figure 2.8, Table
2.4). Deer significantly reduced ‘04-‘05 growth across both Control and Selective

treatments but only within the four height classes between 26 cm-150 cm (Figure 2.9A,

Table 2.5). Interestingly, the loss in yearly growth attributed directly to deer was greater

70

as seedlings increased their initial 2004 size. For example, if a seedling was 26-50 cm tall
in 2004, the difference in ’04-’05 growth between a Deer and No Deer raised seedling

was 3.06 cm, while a 100-150 cm tall height class seedling would have a growth

difference of 22.45 cm.
At the outset of the study, only two sites had advanced sugar maple regeneration

in the seedling layer. Sugar maple was able to grow above 25cm but only in No Deer
treatments and stem densities were highly variable above 25cm (Figure 2.8.). In No Deer
treatments, sugar maple had lower yearly height grth and attained a maximum height
class of only 51-75 cm in Control treatments, whereas, in Selective treatments sugar

maple grew into the 151-200 cm height class (Figure 298).

Belowground resource availability (soil moisture, N pools, N mineralization rates)
Weather station data from Spalding, Michigan (1 7 km from my sites) showed that
2000, 2002, and 2004 had growing season rainfall totals that exceeded the 50 year
average, while 2001 and 2003 were the 6th and 2"d driest growing season totals,
respectively, over the same 50 year period (data not shown). Soil moisture responses to
Deer/No Deer and vegetation treatments were weak (Figure 2.10, Table 2.6). In contrast,
soil moisture was generally lower for vegetation intact, than vegetation removed
treatments for the naturally established sugar maple seedling experiment. (Figure 2.1 1,
Table 2.7), and soil water generally increased with canopy openness in this study.
Except for one of the nine measurement dates, KCl extractable soil NO3' -N was
not significantly affected by Deer / No Deer or vegetation treatments (Table 2.8). On this
date (July, 2003), both Deer and No Deer Selective treatments and No Deer Herbicide

treatments had similar and greater N03' -N values than all other treatment combinations

71

 

 

(data not shown). NH4+ -N differed only on July 2003 (Table 2.9) as well, with

significantly higher values in No Deer Selective treatments than Deer Selective and all

other treatments were intermediate
Extractable soil NO3' -N and NH4+ -N, generally increased over the course of the

experiment (Figure 2.12) with the only deviations fi'om this trend being lower values in
summer 2001 and 2003 which coincided with drought events. N- mineralization rates did
not differ among treatments for any of the dates (Table 2.10). N-Mineralization declined

49.82% as a whole over the course of the study (Figure 2.13) with most of the total

decline occurring during the 2001 growing season.

Discussion
After four years of deer exclusion, results were similar to Milchunas and Laenroth

(1993), Raymer (2000), and Weigmann (2006) in that areas open to browsing had altered
understory vegetation compositions as evidenced by the increased old-field component

(goldenrod, dandelion, Carex etc), while areas protected from deer had increased

populations of browse sensitive species (Mianthemum canadensis, Trillium grandifolia
etc). Deer induced compositional shifts did not result in changes to overall sedge, forb, or
seedlings biomass below 0.25m. Vegetation manipulation treatments initially reduced
biomass of sedge and seedlings but all rebounded to pretreatment levels within 4 years.
Control (vegetation, primarily sedge) treatment diversity remained uniformly lower, not
as a result of sedge overabundance altering the evenness factor in the equation (as I
excluded sedge from the calculated diversity), but rather, I believe sedge dominates the

understory at the expense of forbs and seedlings by its ability to survive periods of

intense stress (Chapter 4).

72

Results were mixed pertaining to the impacts of sedge on seedling growth and
survival. Planted white ash showed little response to both deer and vegetation, while
planted sugar maple altered growth in some treatments and not in others. The larger study
showed that both planted and naturally established sugar maple seedlings were capable of
surviving and growing with sedge under average light levels, showing no reduction in
growth when control (sedge dominated plots) and selective sedge removal (only sedge
removed plots) treatments were compared. Sugar maple successfully grew into small
sapling class stems but only where deer were excluded, as deer browsed all stems
>0.25m, and where I did not treat the vegetation with herbicide (removing advanced
seedling regeneration as well).

In contrast to the larger study site, results from the 60 pairs of seedlings
distributed across a broader range of available light showed increased negative effects of
sedge on seedling growth and mass as light increased. The fact that seedlings grew
significantly taller in areas without competing vegetation is not a novel result, as this has
been shown in many grassland and forested systems (Carter et al., 1984, Davis etal.,
1998, Dodd et al. 1998, Elliott and White 1987, Knoop and Walker 1985, Lbf 2000,

Peterson and Maxwell 1987, Sands and Narnbiar 1984). However, across a light gradient,
the increased negative impacts of sedge on seedling growth could mean that in these
selectively harvested stands sedge may, in fact, be increasing the successional period
needed to obtain a full stocked canopy. The 60 pairs of seedlings did have access to
greater soil moisture when vegetation was possibly explaining the increased seedling
height growth and mass in vegetation removal plots (higher soil water) vs. intact

vegetation plots (lower soil water)(Elliott and White 1987, Gordon and Rice 2000).

73

 

1 found similar vegetation manipulation and deer effects for ironwood as
ironwood stems grew equally well in both controls (sedge) and selective (no sedge)

treatments and deer reduced or eliminated all ironwood stems above 100 cm. For all

height classes where stems of ironwood were present, deer essentially standardized yearly
growth, causing small seedlings (<25 cm in height) to have the same yearly growth (~9-

11 cm/yr) as taller, presumably older seedlings (<150 cm). This resulted in a 100-150 cm
tall seedling in deer treatments needing over two years worth of growth to equal the
yearly growth of a seedling protected fi'om deer. This growth differential increased as the

size of the seedling increased until it reached approximately 2 m in height; above which

deer could no longer browse the terminal bud.

During the study, vegetation nitrogen content tracked treatment induced shifts in
vegetation mass, as N concentrations were not altered by my vegetation manipulations or

by deer. Although forb N content did not differ in 2004, rank orders of vegetation

manipulation treatment effects (Herbicide + Scarified < Herbicide< Selective< Controls)
indicated that deer might have trended to utilizing forbs in more open areas to a greater

extent (easier to see) than in areas dominated by sedge (controls).

I expected to find, but did not, that both deer and vegetation manipulation
treatments altered levels of extractable N and/or N- mineralization rates, as others have

found both can change due to shifting vegetation conditions induced by ungulate

browsers (Pastor et al., 1993 & 1998,and McInnes et al., 1992). I may not have witnessed

this shift because of the continued dominance of overstory litter inputs into the system

(85% of the total litter inputs in control area was from overstory trees).

74

 

After four years it was evident that vegetation growing in my study areas
experienced seasonal periods of intense climate induced stress. In three of the four
growing seasons I observed soil moisture declines while a decline in growing season
rainfall was recorded at a local NOAA weather station (Spalding MI). Two of the four
growing seasons had droughts that were below the 50-year average. Although the main
study showed no soil moisture differences among vegetation manipulation or Deer

treatments perhaps because of larger variation and smaller sample sizes,

In general, Although the pressures of sedge on seedling growth has the potential
to be a significant driver of succession, this can only occur if/when deer numbers decline,

as deer are truly the key drivers of sugar maple regeneration in these northern hardwood

system (Waller and Alverson 1997).

Management Implications

If deer populations stay at current levels, the restocking of these highly managed
stands should be a concern to managers. As deer halt regeneration from below and trees
that are above the reach of deer continue to grow into higher height classes, significant
losses of intermediate forest canopy layers will continue and calls into question the long-

tenn sustainability of these managed systems. The fact that deer are eliminating

ironwood, a species many consider to be of lower preference to deer, highlights the

pressure that the forested systems are under and the suboptimal forage that deer are

having to ingest to survive.

If deer densities are lowered, removing sedge while opening the canopy will
result in greater seedling growth which reduces the time that deer densities will need to

remain lower. Using Figure 2.9 to give us a rough approximation of a sugar maple’s

75

 

 

potential growth under average light levels (6-11 % open sky). Here a seedling could

potentially reach 150+ cm in 7-8 years, if as this data shows, a seedling grows more per
year as it grows in stature. Conservatively, managers should use a 10-year window to

allow for unforeseen events (droughts, insect defoliations, etc).

The near complete removal of sedge two years after the vegetation manipulation
treatments was promising from a management standpoint, as was the 50-5 5% lower
sedge mass found after four years. Unfortunately, the near complete loss of seedlings and
recruit seedlings >25 cm, which were also susceptible to the two mid-summer herbicide
treatments (herbicide and herbicide + scarification), was a concern. Given the four year
“window” of reduced sedge densities achieved with the vegetation manipulation
treatments, the periodic mast seeding events common to northern hardwood tree species
(Houle 1999), and the potential for moderate to severe growing season drought events,
the loss of previously established desired tree seedling individuals should be minimized.

Although I found severe negative deer effects on ironwood growth, this species is

poised to dominate the next forest sapling stage on these sites if/when deer numbers
decline. Because little is known if high deer densities can be overwhelmed by spraying
large areas of competing vegetation, it is currently an economically unwise decision to
proceed with spraying before there is an immediate and clear determination on the part of
resource managers to reduce the deer herd in parts of Michigan. Further work is needed

in identifying methods of spraying which effectively controls sedge while leaving tree

seedlings and forbs intact.

76

Literature cited
Albert, DA. 1995. Regional landscape ecosystems of Michigan, Minnesota, and
Wisconsin: a working map and classification. US Forest Service North Central
Forest Experiment Station, St. Paul, MN

Augustine, DA. and LE. Frelich 1998. Effects of white-tailed deer on populations of an
understorey forb in fragmented deciduous forests. Con. Bio. 12(5):995-1004.

Augustine, D. J., L.E. Frelich, and RA. Jordan. 1998. Evidence for two alternate stable
states in an ungulate grazing system. Ecological Applications 8(4):1260-1269.

Balgooyen, CF. and D.M. Waller 1995. The use of Clintonia borealis and other
indicators to gauge impacts of white-tailed deer on plant communities in northern

Wisconsin, USA. Nat. Areas J. 15(4):308-318.

Bancroft, T.A., 1964. Analyses and inferences for incompletely specified models
involving the use of preliminary tests of significance, Biometrics 20:427-442.

Beaudat, M. and C. Messier. 1998. Growth and morphological responses of yellow birch,
sugar maple, and beech seedlings growing under a natural light gradient.
Canadian Journal of Forest Research 28:1007-1015.

Bergstrom, R. and K. Danell. 1987. Effects of simulated winter browsing by moose on
morphology and biomass of two birch species. Journal of Ecology 75:533-544.

Campa 111, H. 1989. Effects of deer and elk browsing on aspen regeneration and
nutritional qualities in Michigan. Pd.D Dissertation. Michigan State University.
East Lansing. 122pp.

Carter, G.A., J .H. Miller, DE. Davis, and RM. Patterson. 1984. effect of vegetation
competition on the moisture and nutrient status of loblolly pine. Canadian Journal

of Forest Research 14:1-9.

Comett, M.W., L.E. Frelich, K.J. Puettmann, and PB. Reich. 2000. Conservation
implications of browsing by Odocoileus virgnianus in remnant upland Thuja

occidentalis forests Bio. Con. 93: 359-369.

Coughenour, MB. 1991. Biomass and nitrogen responses to grazing of upland steppe on
Yellowstone’s northern winter range. J. of App. Eco. 28:71-82.

Cox, D.R. 1972. Regression models and life tables (with discussions). J oumal of the
Royal Statistical Society B. 34:187-220.

Crawley, M.J . 1990. Rabbit grazing, plant competition and seedling recruitment in acid
grassland. J. App. Eco. 27:803-820.

77

 

Danell, K., K. Huss-Danell, and R. Bergstrom. 1985. Interactions between browsing
mosse and two species of birch in Sweden. Ecology. 66(6): 1 867-1878.

Davis, M.A., K.J. Wrage, and PB. Reich. 1998. competition between tree seedlings and
herbaceous vegetation: support for a theory of resource supply and demand.
Journal of Ecology 86:652-661.

Davis, M.A., K.J. Wrage, P.B. Reich, M.G. Tjoelker, T. Schaeffer, and C. Muerrnann.
1999. Survival, growth, and photosynthesis of tree seedlings competing with
herbaceous vegetation along a water-light-nitrogen gradient. Plant Ecology
145:341-350.

 

DeCalesta, D.S. 1997. “Deer, Ecosystem Damage, and Sustaining Forest Resources,”
online proceedings of Deer Management in Maryland Conference, October 27,

1997.

de la Cretaz, A. and M.J . Kelty 2002 Development of tree regeneration in fem-dominated
forest understories after reduction of deer browsing. Restoration Ecology
10(2):416-426.

Dodd, M.B., W.K. Lauenroth, and J .M. Welker 1998. Differential water resource use by
herbaceous and woody plant life-forms in a shortgrass steppe community.
Oecologia 117:504-512.

Du Toit, J .T., J .P. Bryant, and K. Frisby. 1990. Regrowth and palatability of acacia
shoots following pruning by Afiican savanna browsers. Ecology 71(1): 149-1 54.

Elliott, K,J, and AS. White. 1987. Competitive effects of various grasses and forbs on
ponderosa pine seedlings. Forest Science 33(2):356-366.

Fletcher, J .D., W.J. McShea, L.A. Shipley, and D. Shumway. 2001. Use of common
forest forbs to measure browsing pressure by white-tailed deer (Odocoileus
virginianus Zimmerman) in Virginia, USA. Nat. Areas J. 21(2): 172-176.

F relich, LE. and CG. Lorimer. 1985. Current and predicted long-term effects of deer
browsing in hemlock forests in Michigan, USA. Bio. Con. 34(2):99-120.

Gill, R.M.A. 1992. A review of damage by mammals in north temperate forestsz3. Impact
on trees and forests. Forestry 65(4):363-388.

George, LC. and FA. Bazzaz. (1999) The fern understory as an ecological filter:
growth and survival of canopy-tree seedlings. Ecology 80 (3):846-856.

Gordon, D.R. and K.J. Rice. 2000. Competitive suppression of Quercus douglasii

78

(Fagaceae) seedling emergence and growth. American Journal of Botany
87(7):986-994.

Hale, C.M., L.E. Frelich, and PB. Reich. 2006. Changes in hardwood forest understory
plant communities in response to European earthworm invasions. Ecology
87(7):]611-1874.

Horsley, SB. 1990. Control of grass and sedge in Allegheny hardwood stands with
Roundup-residual herbicide tank mixes. Northern Journal of Applied Forestry
7:124-129.

------- 1993. Mechanisms of interference between hayscented fern and black cherry.
Canadian Journal of Forestry Research 23:2059-2069.

------- , S.L. Stout, and D.S. deCalesta. 2003. White-tailed deer impact on the
vegetation dynamics of a northern hardwood forest. Ecological Applications
13:90118.

Houle, G. 1999. Mast seeding in Abies balsamea, Acer saccharum and Betula
alleghaniensis in an old growth, cold temperate forest of north-eastem North
America. Journal of Ecology 87:413-422.

Hulme, PE. 1996. Herbivory, plant regeneration, and species coexistence, Journal of
Ecology 84(4):609-615.

J effiies, R.L., D.R. Klein, and GR. Shaver.1994 Vertebrate herbivores and northern plant
communities: reciprocal influences and responses. Oikos 71:193-206.

Jenkins, M.A. and GP. Parker. 1999. Composition and diversity of ground-layer
vegetation in silvicultural openings of southern Indiana forests. The American
Midland Naturalist 142(1):16.

Knoop, W.T. and B. H. Walker. 1985. Interactions of woody and herbaceous
vegetation in a Southern African savanna. Journal of Ecology 73:235—253.

Klein, D.R. 1970. Food selection by North American deer and their response to over
utilization of preferred plant species. Pages 25-46 in A. Watson, ed. Animal
populations in relation to their food resources. British Ecological Society
Symposium 10. Blackwell, Edinburgh. 47 7pp.

McNaughton, SJ. 1983. Compensatory plant growth as a response to herbivory. Oikos
40:329-336.

Metzger, F. and J. Schutz. 1984. understory response to 50 years of management of a

northern hardwood forest in Upper Michigan. American Midland Naturalist
1 12(2):209-223.

79

McInnes, P.F., R.J. Naiman, J. Pastor, and Y. Cohen. 1992. Effects of moose browsing on
vegetation and litter of the boreal forest, Isle Royale, Michigan, USA. Ecology
73 :2059-2075.

Milchunas, D.G. and WK. Lauenroth. 1993. Quantitative effects of grazing on vegetation
and soils over a global range of environments. Ecological Monographs 63(4):327-
366.

Molvar, E.M., R.T. Bowyer, and V. VanBallenberghe. 1993. Moose herbivory, browse
quality, and nutrient cycling in an Alaskan treeline community. Oecologia
94(4):472-479.

Pastor, J ., B. Dewey, R.J. Naiman, P.F. McInnes, and Y. Cohen. 1993. Moose browsing
and soil fertility in the boreal forests of Isle Royal National Park. Ecology
74(2):467-480.

Pastor, J. and Y. Cohen. 1997. Herbivores, the functional diversity of plants species, and
the cycling of Nutrient in Ecosystems. Theoretical Population biology 51 :165-
179.

Pastor, J., B.Dewey, R. Moen, DJ. Mladenoff, M. White, and Y. Cohen. 1998. Spatial

patterns in the moose-forest-soil ecosystem on Isle Royale, Michigan, USA.
Ecological Applications 8(2):411-424.

Peterson, TD. and B.D. Maxwell. 1987. Water stress of Pinus ponderosa in relation to
foliage density of neighboring plants. Canadian Journal of Forest Research
17:1620-1622.

Raymer. D.F.N. 2000. Effects of Elk and white-tailed deer browsing on aspen

communities and wildlife habitiat quality in northern lower Michigan: an 18-year
evaluation. Ph.D. Dissertation. Michigan State University. East Lansing. 371pp.

Rooney, T.P. 1997. Escaping herbivory: refuge effects on the morphology and shoot

demography of the clonal forest herb Maianthemum canadense. Journal of the
Torrey Botanical Society 124(4):280-285.

Rooney, T.P. 2001. Deer impacts on forest ecosystems: a North American perspective.
Forestry 74(3):201-208.

Russell, F.L., D.B. Zippin, N.L. Fowler. 2001. Effects of White-tailed Deer (Odocoileus

virginianus) on Plants, Plant Populations and Communities: A Review. The
American Midland Naturalist. 146(1):1-26.

Sands, R. and B.K.S. Nambiar. 1984. Water relations of Pinus radiate in competition
with weeds. Canadian Journal of Forest Research 14(2):233-237.

80

 

Swift, E. 1948. Deer select most nutritious forages. Journal of Wildlife Management
12:109-110.

Stromayer, K. A. K., and R. J. Warren. 1997. Are overabundant deer herds in the eastern
United States creating alternate stable states in forest plant communities. Wildlife
Society Bulletin 25(2):227-234.

Tilgham N.G. 1989. Impacts of white-tailed deer on forest regeneration in northeastern
Pennsylvania. Journal of Wildlife Management 53:524-532.

Waller, D.M. and W.S. Alverson. 1997. The white-tailed deer: a keystone herbivore.
Wildlife society bulletin 25:217-226

Walters, MB. and PB. Reich. 1997. Growth of Acer saccharum seedlings in deeply
shaded understories of northern Wisconsin: effects of nitrogen and water
availability. Canadian Journal of Forest Research. 27 :23 7-247.

Webster, CR. and GR. Parker 2000. Evaluation of Osmorhiza claytonia (Michx) C.B.
Clarke, Arisaema triphyllum (L.) Schott, and Actaea pachypoda Ell. as potential
indicataors of white-tailed deer overabundance. Nat. Areas J. 20(2):]76-188.

Weigmann, SM. 2006. Fifty years of change in northern forest understory plant
communities of the Upper Great Lakes. Ph.D. Dissertation. University of
Wisconsin-Madison. 206pp.

Young, J.A., and CG. Young. 1992. Seeds of woody plants of North America.
Dioscorides Press. Portland, Oregon. pp. 407.

81

Deer No Deer

 

+ control
—0- select
+ herbicide
—e—- herb +scar

 

 

 

 

Forb

 

Seedling <25cm

    

 

 

 

 

2000 2002 2004 2000 2002 2004
Figure 2.1. Harvested Sedge (A&B), Forb (C&D), and Seedling<25cm(E&F) biomass

across sampling years. Treatments nested within Deer and No Deer areas. :1: 1
standard error.

82

Deer No Deer

 

 

 

10
Sedge Sedge + Control
8 1 1 —°— Selective
£5" 6 1 + Herbicide
ED 4 ‘ 4 —°— Herb+ Scar
.M
2 l

 

 

kg N/ha

 

 

 

 

 

 

 

1'6 . Seedling <25cm , Seedling <25cm
g 1.2.
2 0.8
on
at 0.4
0.0 ‘ i
2000 2002 2004 2000 2002 2004

Date Date

Figure 2.2. Harvested Sedge (A&B), Forb (C&D), and Seedling<25cm(E&F), N content
across sampling years. Treatments nested within Deer and No Deer areas. :1: 1
standard error.

83

1.4

1.2 -
1.0 .
0.8 .
0.6 ~
0.4 .

HI

0.2

Deer

No deer

 

 

Shannon-Weiner Diversity

 

 

Shannon- Weiner Diversity

 

2000 2002

Year

2004

2000

 

 

Control
Selective
Herbicide
Herb + Scar

 

 

2002 2004

Year

Figure 2.3. Shannon-Weiner Diversity across sampling years by Deer/ No Deer areas
with nested treatments. i 1 standard error.

84

 

 

Planted Sugar Maple

No Deer

 

% survival

> Deer

 

—0— Deer controls
-6- Deer selective
.....vt Deer herbicide
---A---- Deer herb + scar
+ No Deer control
—9— No Deer selective
«HO-- No Deer herbicide
40 l ---<>- No Deerherb+scar

 

 

% survival

 

 

 

 

20 . . . . . . .
0 200 400 600 800 1000 1200

 

Days since planting

Figure 2.4. Planted sugar maple (4A) and white ash (4B) seedling survival after 1210
days by Deer/No Deer[nested treatments].

85

 

25‘

-} cf
201 .1.
15‘
10*

Seedling height (cm)

 

 

 

 

 

 

 

 

 

1.0 < _}

 

 

 

 

 

 

 

 

 

 

 

g 0.8 . r}
... f
:8 0.64 f
E
E, 0.41
m
0.2 r
0.0 \- . . .
, o , o
0960 009‘ .96 x90
0 \ 1°
6? ~69 03
‘9
Treatment

Figure 2.5. Planted sugar maple seedling growth (5A) and stem mass (5B) by treatment
and Deer (shaded bars) / No deer (open bars). :1: 1 standard error.

86

 

 
 

 

 

 

    

 

 

 

 

4.0
E 3.8 . Planted sugar maple Planted white ash
B 3 3.6 1 Treatment P=0.021 Treatment P=091 — Control
EEO 3 4 . Light P<0.0001 Light P=0.0005 fielrzctixée
.... - _ . = — e ici e
“g a 3.2 ‘TxL P—0.0077 TxL P 0.1846 ....... Herb.+Scar.
‘3 .S 3.0 ...................
‘6”"3‘ '7 .-
—' 8 2.8 ‘ ......
2.6 . "
E 1.0 1 Treatment P=0.002 ...o" Treatment P=0001
'8 ‘3,” 0'5 ‘ Lrght P<0.0001 ‘ Light P<0.0001
E a TxL P=0.1035 TxL P=0.2978
,9, E 0.0 .
U) o
5 a; -0.5 -
b on
E05 -l.0 .
f}, -15-
-2.0 - 1 1

2 4 6 8101214162 4 6 810121416
% canopy openness % canopy openness

Figure 2.6. Planted sugar maple seedling height (6A) & stem mass (63), and White ash
seedling height (6C) & stem mass (6D) by treatment across a gradient of canopy

openness in areas protected from deer.

87

Sugar maple

 

 

 

 

 

 

 

 

A —0— No sedge
E54 ".9... Sedge
e Eb
”’ .2
E 2° Treatment P=0.267
m, % Light P<0.0001
,3 § TxL P=0.0269
r 9
E in
93 g Treatment P=0.012
on 1 t < .
§ L'gh P 00001
:0 E TxL P=0.0408
'o
3 8
(I)
0 5 10 15 20
Canopy openness (%)

Figure 2.7. Naturally established sugar maple seedling height and dry mass across a
gradient of canopy openness in areas with vegetation and without.

88

coho p556 _ a 98.58.: coon oz 98 Doc 5 mam—o Ewe: 3 8383 8on 535303. amps 98 3553.: 99:6 .m.~ Semi

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

£58on
coon ooov oooM ooom cog 25m ooom cos 0
u .w r f4 omumm
..lllmH. i mm- fl w W
3
Tfls Ti COTE. met
coca 02 U W
HODQ I TH: " cf Tl CW~I~O~ GM:
0
E1 . Till.— OCNI .— W H m
Efieeoooq Dow 38333:» $936 - omonm
m .
r D—Qfla gwsm I UOOBGOHH TIIHH

 

 

 

 

89

 

50

 

 

 

    
 
  

Ironwood - Ostrya virginiana % ---- 0"" control - deer
40 . —e— control - no deer
u-v-o selective - deer
30 —fi— selective - no deer
20 r
................. i ......... i

 

fir U T

60 iSugar maple - Acer saccharum é

2004-2005 height growth (cm)

 

 

 

50-
40‘
sol
20‘

10 j * = Treatment differences

P=0.05

0 .,, C Ta 5: e o a

N V” 1‘ O W o In

' e —'« 'r "r “3 s.“

O N ‘n \O '— v— '—

l\ O V) O

'— v— N

Initial (2004) height class (cm)

Figure 2.9. 2004 — 2005 Ostrya virginiana and Acer saccharum height growth by
Deer/No Deer and Nested treatments control & selective) separated by initial
height class. i 1 standard error.

90

% Soil moisture

Figure 2.10. % soil moisture over the three growing seasons (2001-2003).

 

25i
20l
151

10‘

 

O

2001

l O

i

 

O

2002 i

<1

 

2003

 

Apr May Jun Jul Aug Sep Oct

Jul

Date

91

Jul

Sep

 

 

 

 

 

2° * ‘ 8 8‘
E 2001 2002 * 2003
"29515 J 8 o 1 1 0 O No Sedge
E 10 e 0 8 . * = dates where . 0 Sedge
g l 8 sedge/no sedge
,,\° 5 4 * are sign. P=0.05. 6 O 5

 

 

 

 

 

Jul Aug Sep Oct May Jun Jun Jul Jul Jun Jun Jul Jul Aug Aug
Date

Figure 2.11. % soil moisture from areas with and without vegetation surrounding two
year old naturally established sugar maple seedlings.

92

 

 

 

 

 

 

 

 

0.35
Soileools
- 0.30
:5; L 0.25 E
.E‘ - 0.20 .5
£0 £0
+ _ 'm
§¢ 0.15 CZ)
on - 0.10 no
- 0.05
0.00
O O '-‘ v-' N N M M g
Q Q 9 Q Q Q 9 9 \
6 E B E B E 3 E B
Date

Figure 2.12. Extractable NHf-N and N03' -N collected from 9 measurement points
throughout the study. :1: 1 standard error.

93

 

1

      

+ control
—0— selective
—v— herbicide
—°— herb + scar

 

 
  

 

 

 

 

S .1
e
t
a u
r I
n .
O T
In I
a
.n .
.m. ..
n .
Um 7'
N .
. n a .
8 6. 4. 2 0. 2
0 0 O O O 0

2&3 TE 3 :ocmmzfiofifi cowobi

EVER
32>»
3:2
Shh:
82:.
8:?
no):
NQCS
82R
NOE?
82:
52>:
52R
52)»
52:
25>:
OCER
ooh:V

Date

Figure 2.13. N-mineralization rates by treatment from 2000-2004. i 1 standard error.

94

Table 2.1. Results of a standard least squares mixed model for the effects of Deer/No
Deer, and Deer/No Deer[nested treatments] on sedge (A), forb (B), and seedling
(C) biomass, and Shannon-Weiner diversity (D) by years (2000,2002,2004).

Vegetation biomass ANOVA Effects 2000 2002 2004
A- Sedge Deer/No Deer P=0.6706 P=0.8244 P=0.3151
Deer/No Deer [Veg manip] P=0.4569 P=0.0003 P=0.0705

B- Forb Deer/No Deer P=0.4036 P=0.012] P=0.3297
Deer/No Deer [Veg manip] P=0.3256 P=O.1085 P=0.8801

C- Seedling Deer/No Deer P=0.7027 P=0.6398 P=0.4227
Deer/No Deer [Veg manip] P=O.9465 P<0.0001 P=0.0584

D- Shannon-Weiner Deer/No Deer P=0.0793 P=0.7207 P=0.6041
Diversity (H') Deer/No Deer [Veg manip] P=0.9614 P<0.0001 P=0.0987

95

 

Table 2.2. Results of a standard least squares mixed model for the effects of Deer/No
Deer, and Deer/No Deer[nested treatments] on sedge (A), forb (B), and seedling

(C) N content across years (2000,2002,2004) by Deer/No Deer, and Deer/No
Deer[nested treatments].

Vegetation N Content Anova Effects 2000 2002 2004
A- Sedge Deer/No Deer P=0.5642 P=O.1459 P=0.4854
Deer/No Deer [Veg manip] P=0.2686 P<0.0001 P<0.0001

 

B- Forb Deer/No Deer P=0.2377 P=0.0070 P=0.7366
Deer/No Deer [Veg manip] P=O.1476 P=O.1529 P=0.5701

C- Seedling Deer/No Deer P=0.6945 P=0.5687 P=0.7l49
Deer/No Deer [Veg manip] P=O.9596 P=0.0002 P=O.1152

96

Table 2.3. Chi Square results for planted sugar maple and white ash seedling survival
(1210 days after planting) by Deer/No Deer, and Deer/No Deer[nested treatments]

 

effects.
ANOVA Effect Chi Square Prob>Chi sq.
Sugar maple Deer/No Deer 158.9898 0.0000
Deer/No Deer [Veg manip] 17.5519 0.0075
White ash Deer/No Deer 8.68139 0.0032
Deer/No Deer [Xeg manip] 13.6935 0.0333

 

97

Table 2.4. Mixed model (Deer/No Deer, and Deer/No Deer[nested treatments] effects)
results for Ironwood density (stems/ha) by height class fi'om 34-1m2 sampling

 

 

 

 

 

plots/treatment area.

Height
class ANOVA Effect Sum of Squares F ratio Prob>F
25-50cm Deer/No Deer 5605536 2.462 0.1362
Deer/No Deer [Veg manip] 1107266 0.243 0.787
50-75cm Deer/No Deer 1401384 0.657 0.4294
Deer/No Deer [Veg manip] 640138 0.15 0.8618
75-100cm Deer/No Deer 276816 0.098 0.7587
Deer/No Deer [Veg manip] 1003460 0.177 0.8394
100-150cm Deer/No Deer 12149654 4.051 0.0613
Deer/No Deer [Veg manip] 2223183 0.371 0.6961
150-2000m Deer/No Deer 2491350 3.578 0.0768
Deer/No Deer [Veg manip] 311419 0.224 0.8021
200-300cm Deer/No Deer 1730104 3.738 0.0711
Deer/No Deer [Veg manip] 34602 0.037 0.9634

 

98

Table 2.5. Mixed model (Deer/No Deer, and Deer/No Deer[nested treatments] effects)
results for ironwood growth (2004-2005) by height class from 34-1m2 sampling

 

 

 

 

 

 

plots/treatment area.
IW Initial height
class ANOVA Effect
O-25cm Deer/No Deer
Deer/No Deer [Veg manip]
25-50cm Deer/No Deer
Deer/No Deer [Veg manip]
51-75cm Deer/No Deer
Deer/No Deer [Veg manip]
76-100cm Deer/No Deer
Deer/No Deer [Veg manip]
101-150cm Deer/No Deer
Deer/No Deer [Veg manip]
151-200cm Deer/No Deer
Deer/No Deer fleg. manip]
201-300cm Deer/No Deer

Deer/No Deer [Veg manip]

 

99

0.066957
0.9892297
2.0433812
1.0138625
1 1.590785
1.011 144
6.601605
0.0074903
5.7453133
0.0925699

Sum of Squares F ratio

0.2678
0.7912
5.2696
0.6537

Prob>F
0.61 19
0.5713
0.0237
0.6256

29.4843 <0.0001

1.2861

0.2809

21.1904 <0.0001

0.012

0.9881

26.4655 <0.0001

0.4264

0.5179
na
na
na
na

Table 2.6. Results of a standard least squares mixed model for the effects of Deer/No
Deer, and Deer/No Deer[nested treatments] on gravimetric soil moisture by

sampling date.

Date

ANOVA Effect

July 13 2000 Deer/No Deer

Deer/No Deer [Veg manip]

 

May 1 2001

Deer/No Deer
Deer/No Deer [Veg manip]

 

July 1 2001

Deer/No Deer
Deer/No Deer [Veg manip]

 

July 13 2001

Deer/No Deer
Deer/No Deer [Veg manip]

 

Aug 3 2001

Deer/No Deer
Deer/No Deer [Veg manip]

 

Aug 15 2001

Deer/No Deer
Deer/No Deer [Veg manip]

 

Aug 24 2001

Deer/No Deer
Deer/No Deer [Veg manip]

 

Oct 4 2001

Deer/No Deer
Deer/No Deer [Veg manip]

 

June 18 2002 Deer/No Deer

Deer/No Deer [Veg manip]

 

June 29 2002 Deer/No Deer

Deer/No Deer [Veg manip]

 

July 17 2002 Deer/No Deer

Deer/No Deer [Veg manip]

 

April 30 2003 Deer/No Deer

Deer/No Deer [Veg manip]

 

July 2 2003

Deer/No Deer
Deer/No Deer [Veg manip]

 

Sept 4 2003

Deer/No Deer
Deer/No Deer [Veg manip]

 

June 22 2004 Deer/No Deer

Deer/No Deer [Veg manip]

 

1 1.079983
86.519846
0.2341 15
46.728291
4.180169
85.971097
4.777724
23.353471
0.913556
30.125009
6.218704
15.48526
0.356394
23.4791
0.646442
10.176175
14.35204
20.00516
0.009661
32.859523
14.508202
18.472395
0.02025
31.422492
1.602817
31.998678
1.164153
22.955757
0.068048
68.031 137

100

0.5178
0.6739
0.0159
0.5301
0.396
1.3573
0.7208
0.5872
0.1405
0.7719
2.2825
0.9473
0.0689
0.756
0.096
0.2518
4.1912
0.9737
0.001
0.5492
3.0634
0.6501
0.0013
0.328
0.2678
0.8912
0.1258
0.4136
0.0042
0.6961

Sum of Squares F ratio Prob>F

0.477
0.6715
0.9003
0.7812
0.5336
0.2616
0.4022
0.7379
0.7103
0.5977
0.1407
0.4757
0.7947
0.6095
0.7587

0.955
0.0489
0.4588
0.9754
0.7668
0.0897
0.6897
0.9718
0.9173
0.6083
0.5129
0.7251
0.8645
0.9489
0.6545

Table 2.7. Results of a standard least squares mixed model for the effects of Treatment,
canopy openness, and their interaction on gravimetric soil moisture by sampling

 

 

 

 

 

 

 

 

 

 

date.
Date ANOVA effect Sum of Squares F ratio Prob>F
July 13 2001 Treatment 0.2709 5.3504 0.0225
% canopy openness 0.2796 5.5228 0.0205
TxCO 0.1750 3.4560 _ 0.0656:
Aug 2 2001 Treatment 0.0237 0.2960 0.5875
% canopy openness 0.0927 1.1583 0.2841
TxCO 0.0236 0.2952 0.5880
Aug 15 2001 Treatment 0.4352 5.4587 0.0212
% canopy openness 0.5584 7.0042 0.0093
TxCO 0.0107 0.1336 0.7154
Aug 24 2001 Treatment 19.1344 3.2503 20.0740:
% canopy openness 113.0280 19.2000 30.0001
TxCO 22.3763 3.8000 .00537?
Oct 5 2001 Treatment 1.0217 0.1398 0.7092
% canopy openness 44.9415 6.1484 0.0147
TxCO 6.3030 0.8623 0.3551
May 29 2002 Treatment 43.2607 1.9669 0.1635
% canopy openness 1 11.1630 5.0541 0.0265
TxCO 10.5850 0.4813 0.4892 _
July 17 2002 Treatment 65.6580 3.6681 . 0.0579,.
% canopy openness 254.2490 14.2039 0.0003
TxCO 2.2125 0.1236 0.7258
June 23 2003 Treatment 0.0470 0.6168 0.4339
% canopy openness 1.7962 23.5760 <0.0001
TxCO 0.0615 0.8075 0.3707
July 15 2003 Treatment 6.5255 1.7761 0.1853
% canopy openness 15.7887 4.2973 0.0404
TxCO 6.3237 1.7212 ‘ 0.1922 *
July 29 2003 Treatment 5.0340 3.1604 0.0781;.
% canopy Openness 15.8687 9.9626 0.0020
TxCO 2.7383 1.7191 01924
Aug 19 2003 Treatment 0.2403 3.2076 0.0782 I
% canopy openness 0.0312 0.4165 0.5211
TxCO 0.0806 1.0759 0.3036

101

Table 2.8. Results of a standard least squares mixed model for the effects of Deer/No
Deer, and Deer/No Deer[nested treatments] on NO3'-N by sampling date.

Date ANOVA Effect
Jul-00 Deer/No Deer
Deer/No Deer [Veg manip]

 

May-01 Deer/No Deer
Deer/No Deer [Veg manip]

 

Jul-01 Deer/No Deer
Deer/No Deer [Veg manip]

 

Sep-Ol Deer/No Deer
Deer/No Deer [Vegmanip]

 

Jul-02 Deer/No Deer
Deer/No Deer[Veg. manip]

 

May-03 Deer/No Deer
Deer/No Deer [Veg manip]

 

Jul-03 Deer/No Deer
Deer/No Deer [Veg manip]

 

Sep-03 Deer/No Deer
Deer/No Deer [Veg manip]

 

Jul-04 Deer/No Deer
Deer/No Deer [Veg manip]

 

0.285688
0.39012
0.00014

0.373826

0.0352817
0.795346
0.0015077
1.62414
0.1 149966
1.37408
0.4625708
0.863546
0.8157605
8.060257
0.0199885
0.13944
0.003929
0.084343

102

Sum of Squares F ratio

1.1606
0.2641
0.0005
0.2005
0.158
0.5936
0.0337
0.6051
0.2728
0.5433
1.968
0.6123
3.151
5.189
0.3484
0.4051
0.1073
0.3837

Prob>F
0.2894
0.9496
0.983
0.9742
0.6936
0.733
0.8555
0.7242
0.6051
0.7713
0.1703
0.7187
0.0857
0.0009
0.5592
0.8701
0.7455
0.8838

Table 2.9. Results of a standard least squares mixed model for the effects of Deer/No
Deer, and Deer/No Deer[nested treatments] on NH4+-N by sampling date.

Date ANOVA Effect
Jul-00 Deer/No Deer
Deer/No Deer [Veg manip]

 

May-01 Deer/No Deer
Deer/No Deer [Veg manip]

 

Jul-01 Deer/No Deer
Deer/No Deer [Veg manip]

 

Sep-01 Deer/No Deer
Deer/No Deer [Veg manip]

 

J ul-02 Deer/No Deer
Deer/No Deer [Veg manip]

 

May-03 Deer/No Deer
Deer/No Deer [Veg manip]

 

Jul-03 Deer/No Deer
Deer/No Deer [Veg manip]

 

Sep-03 Deer/No Deer
Deer/No Deer [Veg manip]

 

Jul-04 Deer/No Deer
Deer/No Deer [Veg manip]

 

0.00011933
0.02620413
0.010071 1
0.068348
0.1310189
0.78030672
0.0003377
0.36849644
0.0067181
0.5866698
0.05889284
0.53568514
0.6392784
1.4349168
0.16175376
0.4815391 1
0.08201 13
1.048238

103

Sum of Squares F ratio

0.0488
1.786
0.4389
0.4964
1.0922
1.0842
0.0064
1.1575
0.0988
1.4378
0.4038
0.6121
6.9242
2.5903
1.6591
0.8232
0.4395
0.9363

Prob>F
0.8266
0.1346
0.5124
0.8062
0.3038
0.3927
0.9369
0.3531
0.7553
0.231 1
0.5297
0.7188
0.0131
0.0376
0.207
0.5605
0.5121
0.4829

Table 2.10. Results of a standard least squares mixed model for the effects of Deer/No
Deer, and Deer/No Deer[nested treatments] on N-Mineralization rates by

sampling date.

Date ANOVA Effect
Jul-00 Deer/No Deer
Deer/No Deer [Veg manip]

 

May-01 Deer/No Deer
Deer/No Deer [Veg manip]

 

J ul-01 Deer/No Deer
Deer/No Deer [Veg manip]

 

Sep-Ol Deer/No Deer
Deer/No Deer [Veg manip]

 

Jul-02 Deer/No Deer
Deer/No Deer [Veg manip]

 

May-03 Deer/No Deer
Deer/No Deer [Veg manip]

 

Jul-03 Deer/No Deer
Deer/No Deer [Veg maniy]

 

Sep-03 Deer/No Deer
Deer/No Deer [Veg manip]

 

Jul-04 Deer/No Deer
Deer/No Deer [Veg manip]

 

0.00088

0.06162

0.10747
0.249227

0.0000027

0.146405
0.000294
0.237864
0.020889
0.155785
0.06849
4.724
1.186235
2.22805
0.00197
0.07134
0.00396
0.051089

104

Sum of Squares F ratio

0.1125
1.3155
4.0067
1.5486
0.0002
1.3828
0.0069
0.9349
0.5715
0.7104
0.1616
1.8575
1.4757
0.462
0.1166
0.702
0.3111
0.669

Prob>F
0.7395
0.2788
0.0541

0.1955
0.9902
0.2526
0.9341

0.4838
0.4552
0.5437
0.6904
0.1 191

0.2336
0.8309
0.7351

0.6501

0.5809
0.6752

Appendix 2.1. Stand history and structural attributes for the 5 replicate sites.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Stand structure Sites
1 2 3 4 5
habitat classification AVO AVO AVO AVO AVO
stand size (ha) 18.6 31.8 20.8 14.5 38.9
stand established 1934 1930 1924 1934 1934
cut year 1972 1975
'cut year 1987 1985 1977 1970 1974
last out year 1995 1995 1998 1995 1997
Total BA (mz/ha) 24.68 21.81 23.19 21.81 18.82
SM 6.03 10.90 10.04 12.34 13.77
WA 4.88 3.16 1.43 2.87 0.86
BW 4.59 7.75 9.47 6.03 2.87
Hick 8.32 0.00 0.00 0.00 1.43
8.27 7.74 4.27 6.12 6.92
2001 Light (%) (5.6-11.14) (574-10) (234-73) (3.95-7.26) (4.88-9.8)
5.7 5.73 3.23 3.87 4.75
2003 Light (%) (Ll-13.5) (1.4-11.9) (0.8-7.7) (1.2-8.3) (0.6-13.0)
2002 Litterfall (kg/ha) 3221 3367 3090 3186 2805
2003 Litterfall (kg/ha) 2816 2993 3620 3366 3725
2004 Litterfall (kg/ha) 3404 3233 3777 3556 3471
2000 SM seed rain/ha 10000 0 0 3333 11667
2001 SM seed rain/ha 0 0 0 1666 1667
2002 SM seed rain/ha 263499 880897 1686667 2439000 3596417
2003 SM seed rain/ha 1111 2222 0 555 1111

 

105

 

Appendix 2.2. Soil depth, Horizon Munsell color description, Soil texture (sand, silt,
clay, fine clay) by depth, Soil pH, and soil Bulk density measurements by

replicate site.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Site
Landform l 2 3 4 5
Horizon depth (cm)
Oi +6~+3 +5~+2 +4~+2 +3 ~+1 +6~+2
Oe +3 ~ +1
Oa +1~0 +2~0 +2~0 +1~0 +2~0
A O~-4 0~-4 0~-2.5 0~-3 0~-2
E -4 ~ -7 -4 ~ -6 -2.5 ~ -7.5 -3 ~ -7 -2 ~ -6.5
B -7~-32 -6~-31 -7.5 ~-41.5 -7~-34 -6.5 ~ -24
BC -32 -31 -41.5 -34 -24
Horizon color
A 10YR3/2 10YR3/1 2.5Y2.5/1 10YR3/1 10YR3/1
E 7.5YR5/4 7.5YR4/1 10YR5/2 7.5YR4/1 7.5YR4/l
B 7.5YR4/4 7.5YR5/4 7.5YR5/4 7.5YR5/4 10YR5/4
BC 5YR3/4 5YR3/4 5YR3/4 5/YRY3/4 5YR3/4
Texture (%) 0-20cm
Sand 42.56 42.51 45.56 42.71 42.76
Silt 49.33 49.40 46.34 49.16 51.13
Course Clay 5.07 6.07 6.08 6.10 5.09
Fine Clay 3.04 2.02 2.03 2.03 1.02
20-40cm
Sand 39.41 40.29 40.43 38.35 38.33
Silt 47.45 49.64 49.46 51.55 52.59
Clay 7.07 4.03 5.05 6.06 5.04
Fine Clay 6.06 6.04 5.05 4.04 4.04
40-600m
Sand 39.39 38.59 40.48 42.57 33.28
Silt 47.48 44.15 46.37 42.22 58.65
Clay 5.05 5.08 6.07 6.08 4.03
Fine Clay 8.08 12.19 7.08 9.12 4.03
Bulk density 0.79 0.82 0.73 0.81 0.61
pH 6.23 7.1 6.86 6.87 6.74

 

 

 

 

 

 

106

 

 

 

CHAPTER 3

CAN HERBICIDE APPLICATIONS BE TIMED TO CONTROL CAREX
PENSYLVANICA LAM. WHILE MIN IMIZING IMPACTS TO NON-TARGET
VEGETATION IN GREAT LAKES NORTHERN HARDWOOD FORESTS?

Executive summagy

Dense Carex pensylvanica Lam. (upland sedge) mats can be found in northern
temperate forests with higher deer densities and may strongly compete with regenerating
tree seedlings for resources. Because sedge is photosynthetically active in the autumn
when much of the deciduous vegetation is dormant, I hypothesized that sedge could be
controlled with a late fall herbicide application with minimal impact on other vegetation.
Here, I report the results of a glyphosate application timing (November 1, July 15,
control) experiment conducted in a managed northern hardwood understory. Two years
after treatment, November 1 herbicide application reduced sedge biomass 92%,
maintained herb biomass, and increased understory plant species richness. November 1
treatment also had the twice the sugar maple seedling germination, establishment, and
survival vs. control areas while seedling survival in July 15 treatments was higher than
controls, but still less than November treatments. Canopy openness conditions varied
within treatments and mixed statistical models showed that increased openness lead to
increased herb and seedling mass across all treatments, and increased sedge mass in the
controls. Questions still remain as to the November 1 treatment’s effectiveness at
establishing seedlings that grow into and through the zone of deer browsing in areas with

high deer densities. More work is needed to determine potential for large scale whole

107

stand level treatments

108

Introduction

Commercially productive selection-harvested northern hardwood stands in areas
of the southern Upper Peninsula of Michigan ofien have very high sedge (Carex
pensylvam‘ca Lam.) cover (i.e. > 85% of the total understory herb layer biomass (Chapter
1& 2), and in many of these areas, tree recruitment failure is prevalent (Chapter 1 & 2).
High sedge cover may be caused by the combination of a long history of selection
harvesting practices (frequent disturbances) and high long-term white-tailed deer
(Odocoileus virginanus) densities (~ 30 deer/km2 since the mid 1970’s). Sedge may
increase in areas with high deer densities as it is not preferentially browsed by deer,
possibly due to high foliar silica contents (Prychid et al., 2003), and/or lower foliar N
values due to stress induced retranslocation fi'om foliage to roots (Heckathom and
Delucia 1996). Lower foliar N concentrations when faced with elevated ungulate
pressures is not the rule as I found similar N concentrations between areas with
historically elevated deer browsing pressure and areas with lower browsing pressure
(chapter 1 & 2). Furthermore, sedge’s intercalary meristem growth from belowground
rhizomes, and not from a permanent aboveground stem with nutrient rich buds like tree
seedlings, protects the plants active growth point from deer. The protection is primarily
through avoidance, as the majority of sedge mass is inaccessible during the winter
months when deer are forced, due to snow cover, to consume forage of poorer quality,
often switching their diets to consume essentially only woody aboveground browse.
Sedge, once established, may maintain dominance for long periods of time, even if deer

are removed, perhaps because it competitively excludes other plants from colonizing (e. g.

109

sequesters all growing space clonally) or it creates lethal conditions, which increase
mortality of colonizing plants (i.e. decreases soil water, nitrogen, light etc.).

The use of herbicides to reduce competition with conifer crop trees is well
understood and is a common operational practice (Cogliastro et al., 1990, Bell et al.,
1997, Vreeland et al., 1998). In contrast, using herbicides to control competitors of
deciduous northern hardwood tree seedlings is less studied and applied (Horsley 2001,
but see Horsley 1981, and Willoughby et al., 2006). Using a non-selective foliar-contact
herbicide such as glyphosate to control competitors of broad-leaved species could have
some advantages, as it is inexpensive and highly effective. However, its obvious
disadvantage is that native, non-competitive plants and desired tree seedlings are equally
sensitive to the herbicide, thus potentially making broadcast applications difficult. A
possible solution is to apply non-selective herbicide when target vegetation is actively
growing (and thus herbicide sensitive) and non-target vegetation is dormant and
unsusceptible. In mid to late fall (late October- early November), young tree seedlings
and forbs are dormant while sedge leaves are green and presumably active. This late fall
period might provide a window to target sedge with a reduced risk of secondary damage
to the desired tree seedlings and forbs. If high sedge cover is partially responsible for
inhibiting the reestablishment of tree seedlings and forbs, then it seems logical that
reducing sedge covers with management interventions such as timed herbicide
applications would increase tree, shrub, and forb establishment.

In this study, I compared the effects of a summer spray (July 15), a late fall spray
(November 1) and an untreated control on sedge cover and mass, tree seedling

germination, grth and survival, as well as understory species richness and diversity of

110

non-target flora. Because light likely varied within and among treatments potentially
affecting covers and mass of understory flora I measured light availability at the forest
floor could by using canopy openness (proxy for light) as a covariate in the analysis of
treatment effects.
Methods
Site characteristics

This field experiment is located in a 195 ha managed northern hardwood stand on
International Paper (IP) lands in Menominee County, Michigan. The stand is dominated
by sugar maple (Acer saccharum Marsh) with white ash (F raxinus americana L.),
basswood (Tilia americana L.), black cherry (Prunus serotina Ehrh.), American elm
(Alnus americana L.), and hop hombeam (Ostrya virginiana Mill.) as minor components.
The stand is on a highly productive drumlin formed from dolomite and limestone parent
material that is generally within 9.1 — 15.2 m of the surface and is classified as being part
of the Northern Lake Michigan (Hermansville) Till Plain. Soils are moderately to well-
drained spodisols and alfisols (Albert 1995). The growing season averages 140 days and
355 mm of rain falls from May through August (48-year average, NOAA-Spalding MI).
Like most managed northern hardwood forests in the region, the stand has been
selectively harvested at 8-12 year intervals since the 1960s, typically to a residual basal
area of 17-18.3 mz/ha. The most recent selection harvest conducted three years prior to
the beginning of my study significantly reduced residual basal area to levels lower than
past harvests on the site. Residual basal area on the site was 11.5-13.8 mZ/ha in 2002
when the project began.

Experimental design

111

In June 2002 I located three sites within the same stand along a productive
drumlin and delineated three, 0.2 ha treatment areas per site centered under large
harvested canopy gaps. Each 0.2 ha area was randomly assigned to one of three replicates
of the three vegetation manipulation treatments: 1) July 15 Glyphosate application (4.7
liters/ha), 2) November 1 Glyphosate application (4.7 liters/ha), and 3) a control (no
herbicide). See experimental overview diagram (Appendix Figure 3.1.), which shows
layout of treatments, vegetation measurements (explained below) and seed trap
positioning (explained below) for one of the three replicates used in the study.

Vegetation Measurements

Within each 0.2 ha treatment area 1 established a grid (buffered by 8 m to
minimize edge effects including herbicide drifi), and marked 24-1.5 m x 1.5 m permanent
vegetation sample plots. In these sample plots 1, along with another observer, measured
vegetation percent cover with ocular estimates and counted tree seedlings <0.25m tall in
mid July of 2002 just prior to the July herbicide treatment, and again in mid July of 2003
and 2004. Spring ephemeral herb percent cover estimates were taken in May of 2003 and
2004. Given sedge’s dominance in the understory, which increases the potential
likelihood of biasing ocular estimates of vegetation growing within the sedge, two
observers simultaneously estimated cover and estimates were averaged. Tree seedlings
were censused for survival on a monthly basis from May to October of 2003, in late May
and late September of 2004 and 2005, and early June of 2006.

In 2004 I located a l m x l m sampling area directly to the north and west of the
non-destructive sample plots. In these plots 1 non-destructively estimated percent cover of

all forest floor vegetation, counted tree seedling stem density, and then destructively

112

sampled and pooled vegetation from the core area (0.25 m2) into sedge, forbs, and tree
seedlings < 0.25 m tall categories. For seedlings > 25 cm I placed biomass into vertical
strata (e. g. 0-25 cm tall, 25-50 cm, etc.). Vegetation was transported to Michigan State
University’s Tree Research Center, dried for 72 hours at 70°C, and weighed.

Tree seed fall, and canopy openness

I evenly dispersed 12-0.5 m x 0.5 m seed traps between the vegetation sampling
areas, elevated to 1 m above the ground (to reduce seed predation by rodents), in each of
the 0.2 ha treatment areas to measure yearly seed fall. Traps were collected after leaf drop
(prior to snow fall) and seed was sorted in the lab by species, counted for total seed, and
sugar maple seed viability was evaluated by opening a subset of samaras and observing if
the seed was filled or empty and if the embryo was alive (bright green and soft) or dead
(brown and dessicated).

Canopy openness, an estimate of light availability, was measured in mid July
2003 at 1 m above each non-destructive sample plot with a dual LAI 2000 (LiCor Inc.,
Lincoln NE) setup. The setup and calibration was exactly the same as in chapter 2 and the
open field instrument was placed in an open field < 1.6 km from where understory
canopy measurements were taken.

Analysis:

The study was originally designed to have three replicate sites with each site
having three treatment areas (control, November 1, and July 15), but due to vandalism I
was forced to abandon one of the three sites, leaving the study with two replicates of each
treatment. In each replicate I averaged data from vegetation plots (n=24) and seed traps

(n=12). I used J MP (5.1) statistical software (SAS institute, Cary, NC, USA) for all

113

statistical analyses. ANOVA was used to examine the effects of herbicide treatments on
sedge, seedling, and forb biomass, the density of viable sugar maple seed fall, and sugar
maple germinants/ha. For significant (P < 0.05) ANOVA effects, Tukey Kramer HSD
was used to test for significant differences among treatments.

Within treatment replicates, there was large variation in overstory canopy
openness, which could affect forest floor vegetation characteristics. Thus, in addition to
ANOVA analyses of treatment means, I analyzed vegetation measurements using sample
plots as experimental units (11 = 48, 24/ replicate x 2 replicates) with mixed models that
included treatment as a nominal variable, canopy openness as a continuous variable, and
their interaction.

I examined species richness for vegetation < 0.25 m tall by bootstrapping data to
obtain multiple estimates of the number of unique species for each 2.25 m2 plot area
gradation. By fitting the mean of these bootstrap generated estimates, I was able to
develop smoothed species area curves for each treatment. To facilitate testing of
species/area curves, 1 used ANOVA and tested only the largest sampling area unit (27
m2). The effect of herbicide application on seedling survival was analyzed with Cox’s
proportional hazards modeling (Cox 1972).

Results
In control plots, sedge dominated forest floor biomass, constituting over 85% of

the total. Two years after application, sedge mass was 98% lower in July 15 herbicide
areas and 92% lower in November 1 herbicide areas than in control areas (Table 3.1).
F orb mass did not differ among treatments (Table 3.1), although there was a weak trend
of approximately 30% greater forb mass in both herbicide treatments than in controls.

2002 seed fall, 2003 first year seedling density, and seedling germinant mass were not

114

significantly different among treatments despite large differences in means, perhaps due
to low replication (n=2) (Tables 3.1 & 3.2).

In models including both treatment and canopy openness, sedge mass increased
with canopy openness in control treatments and was negligible at all light levels in
herbicide treatments (Figure 3.1A.). Seedling stem density (only post-treatment
germinants), and mass (pre- and post-treatment germinants combined) both treatment and
canopy openness was significant, but their interaction was not (Figure 3.1C, D). Seedling
stem density and mass decreased with canopy openness and values were greater in
November 1 herbicide than in July 15 herbicide and control treatments. Increasing
canopy openness increased forb mass while individual treatments did not differ (Figure
3.1B). It is important to note that greater first-year seedling stem densities in the
November 1 herbicide treatment was not due to increased seed fall in the autumn of 2002,
as control and July 15 treatments had roughly 4 times more viable seed fall than the
November 1 treatment. By the start of the third growing season (1099 days), tree seedling
survivorship was significantly higher in the November 1 (59%) than the July 15 treatment
(44%), and both were greater than the control treatment (31%) (Figure 3.2).

Herbicide treatments affected mid-season (July 15) ground flora species richness
measured two years after herbicide treatments. The November 1 treatment had greater
species richness than the control treatment, and both November 1 and control treatments
had significantly greater species richness than the July 15 treatment (Figure 3.4A).
Species richness of spring (late May) censused forbs two years after treatments were
nearly identical for both November 1 and control treatments and both had significantly

more species than the July 15 treatment (Figure 3.4B). For mid-season species richness, a

115

significant treatment x canopy openness interaction (P<0.0001 Adj. R2 = 0.40) indicated
that the November 1 treatment responded positively to increased canopy Openness (Adj.
R2: 0.2921 , P< 0.0001), whereas both July 15 and control treatments did not (July 15, P
= 0.78 & Control P= 0.96 respectively). Spring ephemeral herb richness increased with
canopy openness in both November and July herbicide treatments, while control area
richness did not respond to increasing canopy openness conditions. Dividing mid-season
ground flora into “weedy” and “non-weedy” components did not yield any significant
treatment effects, but weedy species trended to increase in herbicide treatments
(Appendix 3.2.) and increase with canopy openness, but only in the November 1
herbicide treatment (Figure 3.1E).

Overall forest floor layer diversity (Shannon-Weiner) varied by year (Y,
P<0.0001), treatment, and year X treatment interaction (overall model Adj. R2 = 0.409,
P<0.0001). Within sampling years (2002, 2003, & 2004), November 1 treatment had
higher diversity than either the control or July 15 treatment. Although this order was
maintained from 2002-2004, the 2004 treatment effect was not significant (Table 3.3). In
mixed models of Shannon-Weiner diversity, treatment effects were significant in 2003
and 2004 and treatment X canopy openness interactions were significant in 2002 and
2004.

Discussion

A single application of a non-selective herbicide (glyphosate) during autumn
reduced Carex pensylvanica Lam. mass by 92% for at least two years, while having
fewer negative effects on non-Carex vegetation than a mid-summer herbicide application.

Forb mass did not show significant treatment effects but did trend towards increasing in

116

both herbicide treatments relative to controls, possibly as a result of increased availability
of rooting space when sedge was killed. Direct herbicide induced mortality of summer

and spring active forbs decreased overall species richness levels in the July 15 application
areas and those species that were present were more likely to be ruderal or weedy species.

Surprisingly, I found large gains in seedling mass (2.7-fold increase) and densities
(2.1-fold increase) in the November 1 treatment but not in the July 15 treatment
compared to the controls, despite greater seed fall in the July 15 treatment. The majority
of the sugar maple seed fell after the July 15th spray treatment but before the November
1St treatment (personal observation) and was therefore on top of the glyphosate killed
sedge in the July 15 treatment, potentially exposing the seed to increased levels of
predation and desiccation throughout the fall (DeStevens 1991, and Meiners and Stiles
1997). The vast majority of sugar maple seed fell prior to the November 1 treatment and
was then covered with herbicide killed sedge biomass potentially providing better
protection from seed predators and from desiccation.

For sugar maple seedlings that established post-treatment, survival rankings were
November 1 > July 15 > control treatment and was not likely due to deer herbivory as the
incidence of browse damage ranked differently; July 15 > November 1 > control. Instead,
lower survival in control areas was likely due to maple’s inability to survive drought
events(Chapter 4). Levels of drought severity may be increased by the additive effects of
sedge moisture withdrawal and use in the understory and the withdrawal and use of water
resources by overstory trees .

Greater browsing by deer in herbicide application areas may be due to greater

seedling visibility and/or a greater nutritional browse value (e. g. higher leaf nitrogen,

117

lower lignin or silica content) of the seedlings (Anderson et al., 2001) as a result of
decreased N competition when sedge populations decline. Interestingly, sedge foliar N
concentrations were found to be similar between control and herbicide treatments and
overall levels of foliar N were comparable to tree seedling foliage, which also did not
change with mid-summer spray or selective sedge removal treatments vs. controls
(Chapter 2).

Management Implications

Results suggest that managers, if they wish to control sedge using relatively
inexpensive non-selective herbicides (i.e. Glyphosate), should spray at the end of the
growing season (late October-early November) and not during the summer months when
both advanced tree seedling regeneration and forbs are actively growing.

The greater proportion of seedlings damaged by deer in herbicide treatments vs.
control areas may reflect deer being attracted locally to treatment units with higher
proportional forb and seedling mass. If true, then it is possible that the browsing patterns
I observed are an artifact of the 0.2 ha size of my treatment areas. Increasing the size of
the treatment area could overcome this effect, as deer may be more locally dispersed,
overwhelming the foraging capacity of the local deer herd and thus reducing browse
pressure (Buttrick 1923, Zillgitt 1950).

Managers need to consider other techniques in addition to treating the understory,
that will overwhelm a deer’s ability to browse all regeneration. This might involve
increasing the size and intensity of harvests, along with increasing the harvest interval.
Shifting from a single stand harvesting scale to larger landscape oriented practices to

reduce deer-preferred edge habitat (Andren and Angelstam 1993) and deer use

118

(browsing) in the core areas (Blymyer and Mosby 1977). Deer have been found to utilize
browse primarily within 300m of swamp conifer (cedar) stands in winter months where
snow depths range from 20-50cm (Morrison et al., 2002). Lebouton (personal
communication of unpublished data) found deer density decreased from 20 to 5 deer/kmz)
from lowland conifer stand edge to 800m into northern hardwood stands. A density of 5
deer/km2 (at 800m from conifer cover) is below the 7 deer/km2 threshold for timber
regeneration suggested by (Hester et al., 2004). As such, forest managers could reduce
the acreage of swamp conifer within a buffer zone (3 00-800m) around northern
hardwood stands prior to harvesting the hardwood stand. The reduction in thermal cover
and protection from predators, along with increased distance from remaining thermal
cover (increased energy expenditures — Moen 1976), and increased snowpack in large
cutblocks (decreased food availability) might alter a deer’s wintertime deeryard fidelity
(Van Deelan 1999). Taken together, this might relax over-winter browsing pressure long
enough for seedlings to grow above the reach of deer.

There are downsides to this approach as these lowland conifer stands are highly
diverse ecosystems, are preferred by deer for browse leading to regeneration problems,
and removal of cedar overstories can reduce water table depths altering revegetation
composition and structure. Socially, lowland swamp conifers are viewed by many as the
deer herds only chance of surviving Michigan’s harsh winters in the northern zones. The
state DNR has even placed an emphasis on acquiring strategic blocks of lowland conifers
with the sole intent of providing wintertime habitat for deer (MDNR-Deer Range
Improvement Program). In the past, during extremely harsh winters managers even

would strategically out small portions of cedar to provide deer with browse.

119

There are also several factors to be considered before deciding to use widespread
herbicide application on forest understories that have failed to regenerate. The first, is the
extent of sedge cover and its competitive effect on seedlings on the site in question. This
work (Chapters 2 & 4) has shown that sedge can decrease both growth and survival of
northern hardwood trees. However, high deer browsing pressure can have such strong
direct effects on regeneration that spraying alone would have little effect on seedling
recruitment into the sapling class (Chapter 2). Aside from the sedge and deer
components, the composition of the advanced tree seedling regeneration in the understory
is also important, as economically undesirable tree species may dominate (e. g. Ostrya
virginiana). Fall spraying may promote these unwanted species, causing further
complications to successful establishment of high valued timber species. If species such
as ironwood dominate, then a late summer/early fall spray could be used to control both
sedge and tree regeneration les desirable for management, but damage to summer active
forbs should be expected, while early spring ephemerals should be spared.

Once the extent of sedge is quantified, deer impacts are known, and advanced
regeneration has been quantified, the decision to spray becomes an economic one. With
the recent advances made in spray delivery systems, such as “boomless” Sprayers
mounted directly on or pulled behind ATV’s or tractors, and the increased availability of
generic, lower cost glyphosate, land managers may be able to effectively, and perhaps
economically spray entire forest understories. These new spray systems will also extend
the spray window, as it is not necessary to wait until overstory trees are leafless as is
required if herbicide is delivered with a helicopter or fixed wing aircraft. Finally,

although there may be many benefits associated with fall spraying of sedge, caution

120

should be used if its implementation is to work. Stand type, harvesting regime, site
differences (productivities) and spatial extent will yield different results. Managers
should plan cautiously and perhaps initiate their own trials to test the efficacy of this

work in their own stands.

121

Literature Cited

Albert, DA. 1995. Regional landscape ecosystems of Michigan, Minnesota, and
Wisconsinza working map and classification. US Forest Service North Central
Forest Experiment Station, St. Paul, MN.

Anderson, C.P., W.E. Hogsett, M. Plocher, K. Rodecap, and EH. Lee. 2001. Blue wild-
rye grass competition increases the effect of ozone on ponderosa pine seedlings.
Tree Physiology 21 (5):3 1 9-327.

Andren, H. and P. Angelstam. 1993. Moose browsing on Scots pine in relation to stand
size and distance to forest edge. Journal of Applied Ecology 30:133-142.

Bell, F.W., R.A. Lautenschlager, R.G. Wagner, D.G. Pitt, J .W. Hawkins, and KR. Ride.
1997. Motor-manual, mechanical, and herbicide release affect early successional
vegetation in northwestern Ontario. The Forestry Chronicle 73(1):61-68.

Blymyer, M.J. and HS. Mosby. 1977. Deer utilization of clearcuts in Southwestern
Virginia. Southern Journal of Applied Forestry 1(3):]0-13.

Buttrick, PK. 1923. Second growth hardwood forests in Michigan. Michigan
Agricultural Experiment Station special bulletin 123. In USDA Forest Service
Handbook 271. 1965. Silvics of forest trees of the United States. US. Department
of Agriculture, Washington DC. 20250.

Cogliastro, A., D.Gagnon, D. Coderre, and P. Bhereur. 1990. Response of seven
hardwood tree species to herbicide, rototilling, and legume cover at two southern
Quebec plantation sites. Canadian Journal of Forest Research 20:1172-1182.

Cox, D.R. 1972. Regression models and life tables (with discussions). Journal of the
Royal Statistical Society B. 34:187-220.

DeStevens, D. 1991. Experiments on mechanisms of tree establishment in old-field
succession: seedling emergence. Ecology 72(3):lO66-1075.

Gleason HA. and A. Cronquist. 1991. Manual of vascular plants of Northeastern United
States and adjacent Canada. D. Van Nostrand company, Inc. Princeton, New
Jersey.

Hester, A.J., P. Millard, G.J. Baillie, and R. Wendler. 2004. How does timing of
browsing affect above- and below-ground growth of Betula pendula, Pinus
sylvestris, and Sorbus aucuparia? Oikos 105-536-550.

Horsley, SB. 1981. Control of herbaceous weeds in Allegheny hardwood forests with
herbicides. Weed Science 29:655-662.

122

Horsley, SB. 2001. Effects of Fence, Herbicide, and Lime on regeneration of sugar
maple in northern Pennsylvania. Proceedings from ESA 86th annual symposium.

Meiners, S.J., and E.W. Stiles. 1997. Selective predation on the seeds of woody plants.
Journal of the Torrey Botanical Society 124(1 ):67-70.

Moen, A.N. 1976. Energy conservation by white-tailed deer in the winter. Ecology,
57:192-198.

Morrison, S.F., G.J. Forbes, S.J. Young. 2002. Browse occurrence, biomassm and use by
white-tailed deer in a northern New Brunswick deer yard. Canadian Journal of
Forest Research 32(9): 1 5 1 8-1 524.

Prychid, C.J., P.J. Rudall, and M. Gregory. 2003. Systematics and biology of silica
bodies in monocotyledons. The Botanical Review 69(4):377-440.

Van Deelan, TR. 1999. Deer-cedar imteractions during a period of mild winters:
implications for conservation of conifer swamp deeryards in the great lakes
region.

Vreeland, J .K., F.A. Servello, B. Griffith. 1998. Effects of conifer release with glyphosate
on summer forage abundance for deer in Maine. Canadian Journal of Forest
Research 28:1574-1578.

Willoughby, 1., FL. Dixon, and D.V. Clay. 2006. Dormant season vegetation
management in broadleaved transplants and direct sown ash (Fraxinus excelsior
L.) seedlings. Forest Ecology and Management 222:418-426.

Zillgitt, W.M. 1950. Does the partial cutting of northern hardwoods result in inferior
reproduction? US. Forest Service Lakes States Forest Experiment Station
technical note 340. In USDA Forest Service Handbook 271. 1965. Silvics of
forest trees of the United States. US. Department of Agriculture, Washington
DC. 20250.

123

 

1600

 

 

 

 

 

 

 

 

Sedge biomass MOW-November 1
g 1200‘ _D__July15
g A ' ' v Control
a #800 l 'V v ' __
go .3400 4 v 9'" Treatment P<0.0001
w v 'v " Openness P=0.0423
0 1 "1' """ " TxO P=0.0378
6 ‘ v Forb biomass
v O O
3% 4 l . 18,,“080 ..... ----- 2' Treatment P=0.8777
g g 2 1 0 ' . Openness P<0.0001
g 3" ,0,- g ' TxO P=0.5330
: V O 1 v .0 g V
V o 0
g; 60 '
E 50 o 80 2004 Seedling biomass
O
a 40 . g ‘0 Treatment P<0.0001
g: 30 , 0:,“7 ' ' Openness P=0.0081
'8‘ ED 20 , e- ..O ' TxO P=0.2907
0 v Q
m V . ".0... O
8 10 . -.. ........ .
:1:; 0. Wmoo ' o
v 13
a 2003 Sugar maple
a 12 ‘ 0‘9: 0' ngerminant density
,5 11 . Treatment P=0.0005
E Openness P=0.0132
:0 1° ‘ TxO P=0.2078
3 9 .
V V O O V
12 , 2004 Weed cover. .
a; 8 . O . Treatment P=0.5310
é ' v G (:3 .-.- ...... Openness P=0.0001
.\ 4 . v 3°, v ...... , . TxO P=0.2031
0 d 8 '

 

 

 

 

0 10 20 30 40 50
Canopy Openness

Figure 3.1. Sedge biomass (1A), Forb biomass (1B), Seedling biomass (1C), Seedling
density (1 D), and Weedy % cover (1E) across a % canopy openness gradient
by treatment (November 1, July 15 & control).

124

 

  
 
 
  

100
Sugar maple

80 ‘ seedling surv1va1

g (1099 days)
E 60 .
U) 40 d 0.000. Contro' ..o.. ..

 

July15th ' Hm...

— November1st

20 . . . . -
0 200 400 600 800 1000 1200

 

 

 

Days

Figure 3.2. Sugar maple seedling survival by treatment (November 1, July 15 & control).
(this graph tracks on the survival of seedlings which established following the November

and July herbicide applications.)

125

3.0

 

 

 

 

 

TxO P= 0.1169

2 5 ] _.- CODtI'Ol
"-0" Novemberlst
2-0 * —-— July 15th
1
1'5 .. Treatment P= 0.2868
1.0 . ' OpennessP=0.0967
€0.05 0' TxO P= 0.0931

0.5 1 '0 o

$-

0)

.S

o r."

3 5 Treatment P<0.0001

8 E‘ OpennessP=0.0936

g 8

.1: 2.

(D “U

 

Treatment P<0.0001
Openness P= 0.1231
TxO P= 0.0879

 

0- % sugar maple
80 ‘ v" browsed

Treatment P<0.0001
Openness P: 0.0123
TxO P= 0.0562

% browsed

 

 

 

Canopy Openness %

Figure 3.3. 2002, 2003, and 2004 Shannon-Weiner diversity by treatment across a %
canopy openness gradient.

126

Species/area curves

 

25 tMid-Growing season
July 15th

    
 
   

 

 

 

 

 

 

 

.9“:
8
81
5H .
o .-
4'1: F.
5 -' Treatment P<0.0001
25 . . . . .
Sprlng —0— November lst
20 . (late May) "on Control
8 + July 15th
8 15 ~
8*
‘8 10 -
:1:];
5 @ 27m2
Treatment P<0.0001
0 e .

4 8 12 16 20 24 28
Area (m2)

Figure 3.4. Species per unit area sampled by treatment (November 1, July 15 & control)
for Mid growing season (top graph) and Spring (bottom graph) census
periods.

127

Table 3.1. Forb, Seedling, and Sedge biomass (kg ha") by treatment (Control, November
1, & July 15).

Treatment forb kg/ha seedling kg/ha sedge kg/ha
Control 49.2 :1: 13.13 a

8.6 :l: 1.54 a 345.2 :1: 2.658 a
November 1 64.0 :1: 33.07 a

23.0 i 1.047 a 27.1 3: 10.147 b
July 15 62.3 :1: 38.67 a 9.19 :l: 5.49 a 5.77 :1: 3.327 b

Table 3.2. Viable seed rain/ha, and Sugar maple germinants/ha by treatment (Control,
November 1, & July 15).

2002 viable seed 2003 sugar maple
Treatment fall/ha

1St year seedlings/ha
Control 1996205 :1: 1587295 a 38958 :1: 2292 a
November 1 412435 :1: 92564 a

82083 3: 24167 a
July 15 1843333 :1: 968333 a 48958 :1: 17292 a

Table 3.3. Shannon-Weiner diversity indices for years 2002, 2003, and 2004, and %
ruderal and non-ruderal herbaceous cover by treatment (Control, November 1,

& July 15). Different letters represent significantly different (P=0.05) means
within a column (year).

Shannon-Weiner Diversity: Mid-July
Treatment 2002 2003 2004

Control 1.50i0.02a 0.69i0.12b 1.44i0.16a
November11-27i0-04b 1.52i0.05a 1.93:0.16a

July15 1.4140.02a O.81:l:0.l4b 1.57i0.08a

% herbaceous layer cover
Non-ruderal Ruderal
22.2 :t 5.87 a 2.97 :t 1.95 a
12.6i1.75a 3.57i1.7a
8.6i2.21 a 1.18i0.55a

128

Appendix 3.1.Site diagram detailing layout of 1 of 3 replicate sites used in the study.

Each treatment area (Control, Nov.1, and July 15) was 0.2ha (1/2 acre) in size.
Vegetation and seedling plot (0) was 1.5m x 1.5m while each seed trap(0)
was 0.5m x 0.5m. Canopy openness (light) was measured at 1m above the
ground at each vegetation plot (0).

 

 

 

 

 

   

 

 

 

 

 

000 000 000
00 00 00
000 000 000
00 00 00
0.0.0 0.0.0 0.0.0
000 000 000
000 000 000
00 00 00
000 000 000
00 00 00
0.0.0 0.0.0 0.0.0
000 000 000
Nov 1 J ulyl 5 Control

129

. Seed trap locations

0 Vegetation &
Seedling plot
locations

Appendix 3.2. % cover of species found by treatment (November 1, July 15 & control)

Scientific name
Gramineae

Carex pensylvanica
Rubus spp. *

Rubus hispidus *
Adiantum pedatum
Ribes spp.

F ragaria spp.

Carex pedunculata
Violet palmata
Mentha spp.
Dryopteris spinulosa
Plantago spp.
Hepatica acutiloba
Sambucus pubens
Aralia nudicaulis
Cirsium spp. *
Solidago spp. *
Aquilegia canadensis
Mitella spp.
Sambucus canadensis
Uvularia grandiflora
T araxacum officinale *
Laportea canadensis *
Acer saccharum
Lactuca canadensis
Galium boreale
Ranuculata sp.
Arisaema atrorubens
Asclepias spp. *
Oxalis montane
Botrychium virginianum
Osmorhiza claytoni
Trillium grandiflorum
Hieracium spp.
Allium triococcum

Common name
grass

sedge

raspberry
dewberry
maidenhair fern
gooseberry
strawberry
pedunculata sedge
violet

mint

shield fern
plaintain

sharp lobed hepatica
red elder

wild sarsaparilla
thisle

goldenrod
columbine
miterwort
elderberry
bellwort
dandelion

nettle

sugar maple
wild lettuce
bedstraw
buttercup

jack in the pulpit
milkweed

wood sorrel
rattlesnake fern
sweet cicely
trillium
hawkweed

leek

130

Control November

(% cover) (% cover) (% cover)

14.19 2.68
13.56 2.55
5.18 0.75
4.00 0.00
3.00 2.50
2.50 1.08
2.44 1.31
2.38 1.02
2.22 1.78
2.20 0.50
2.00 1.33
1.69 1.50
1.68 1.75
1.60 0.00
1.50 0.90
1.50 1.00
1.50 6.55
1.34 1.50
1.33 1.25
1.33 0.00
1.33 0.00
1.21 1.07
1.21 1.25
1.13 2.17
1.08 1.11
1.06 1.01
1.01 1.51
1.00 1.94
1.00 0.00
1.00 0.88
0.97 1.04
0.96 1.11
0.83 1.75
0.75 0.88
0.70 0.50

July

1.99
1.68
3.00
0.00
1.00
0.00
1.25
1.06
0.77
0.92
0.00
4.59
1.30
0.00
0.50
1.08
1.00
0.75
0.50
0.00
0.00
0.81
0.92
1.40
0.50
0.88
1.67
1.00
0.00
1.00
0.75
0.50
0.50
0.50
0.64

 

Prunus serotina
Maianthemum canadensis
Ostrya virginiana

Abies balsamea

Acer rubrum

Aster macrophyllus
Cerastium spp. *

F raxinus american
Lonicera canadensis
Mitchella repens

Picea glauca
Polyginatum pubescens
Smilacina racimosa

Thuja occidentalis

Tsuga canadensis
Anaphalis margaritacea *
Arctium minus *

Aster spp.

Botrychium dissectum

Caulophyllum thalictroides

Cephalanthus occidentalis
Convoulvulus arvensis *
Crucifer spp. *
Dennstaedtia punctilobula
Dentaria spp.

Equisetum spp.

Erigeron spp. *

Erodium spp. *

Linaria spp.

Myosotis spp. *

T iarella cordifolia

Tilia americana

Ulmus americana
Verbascum thapsus *
Total cover

black cherry

wild lily of the valley
ironwood

balsam fir

red maple

large leaf aster
mouse-cared chickweed
white ash
honeysuckle
partridge berry
white spruce

hairy solomon's seal
false solomon's seal
northern white cedar
hemlock

pearly everlasting
burdock

new england aster
grape fern

blue cohosh

button bush
bindweed

mustard

hay scented fern
toothwort

horsetail

daisy

stork's bill

toadflax
forget-me-not
foamflower
basswood

elm

mullen

131

0.67
0.58
0.57
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
90.21

0.67
0.84
1.04
0.83
0.00
0.00
8.00
1.00
0.00
0.50
0.50
0.66
0.00
0.00
0.50
0.00
0.00
0.63
0.50
0.50
2.00
1.00
3.00
0.50
0.58
2.75
2.25
0.50
0.50
0.00
1.00
0.67
0.56
4.50
82.13

0.75
0.75
0.56
0.57
0.50
0.00
2.25
0.00
0.50
2.75
0.50
0.79
0.00
0.00
0.00
0.50
0.50
0.00
0.00
0.00
0.00
2.00
1.00
0.00
1.00
0.00
0.00
0.00
0.00
3.00
2.00
0.63
1.00
1.70
55.70

Chapter 4

CAREXPENSYL VANICA HAS GREATER SURVIVORSHIP THAN ACER
SACCHARUM SEEDLINGS FOLLOWING COMPETITION INDUCED SOIL WATER
DEFICITS.

Executive summagy

Thick Carex pensylvanica mats can dominate managed temperate hardwood
forest understories. In these systems, Carex is thought to reduce the establishment and
growth of tree seedlings via competition for water and/or nutrients. Here, I quantify
Carex impacts on soil water, inorganic nitrogen and Acer saccharum seedling survival
and growth. Potted plants grown in monocultures and Carex-Acer mixtures were given
drought and well-watered treatments. For drought treatments, water was withheld until
complete stomatal closure for each pot. I found that plant water potentials were similar at
stomatal closure for Carex and Acer. However, Carex survival was nearly 100% in
drought and well-watered treatments, whereas Acer survival was 90% in well-watered
treatments but < 50% in drought treatments. Similarly, Carex mass was unaffected by
drought, but Acer seedling mass was 25% less and height growth 40% less in drought
treatments the year after the drought. Soil in Carex monoculture pots had 50% greater
N03-N than soil in Acer monoculture, whereas NH4+-N concentrations were four-fold
greater than N03-N, and did not vary among treatments. Collectively, data indicate that
Carex growth and survival is less impacted by a given level of drought than Acer

seedlings. Thus a well-established Carex understory could negatively impact tree

132

seedling establishment by, reducing soil moisture to levels that compromise tree seedling,

survival and growth.

133

Introduction

In the Great Lakes region, forests with a long history of selective harvesting and
high deer densities (selective browsing due to silica levels Prychid et al., 2003) are often
characterized by dense Carex pensylvanica Lam. (Pennsylvania sedge) understories with
little tree regeneration. Selective harvesting every 8-15 years does not allow for sustained
canopy closure to lower light levels and reduce Carex populations (lower light levels
measured in the understory had lower sedge cover — Randall Chapter 3). In a field
experiment that included vegetation removal treatments I found that while deer browsing
had the greatest negative effect on tree seedling growth and survival, Carex also
negatively impacted tree seedling growth and survival (Randall Chapter 2). This raises
the possibility that Carex may slow forest succession and the reestablishment of deer
sensitive species, even if deer numbers are reduced (Stromayer and Warren 1997,
Augustine et al., 1998, de la Cretaz and Kelty 1999, George and Bazzaz 1999).

Competition mechanisms altering tree seedling survival and growth are not well
studied for systems dominated by upland sedge (Carex). Because of to the similarities in
structure and function between grasses and upland Carex species, 1 utilized the existing
grass competition literature to provide a basis for this work with Carex. Moreover, in
studies using herbaceous plants of similar stature, competition for nutrients and moisture
had greater impacts to young plant survivorship and growth than did light (Kosola and
Gross 1999, Wilson and Tilman 1993). It is unlikely that Carex outcompetes seedlings of
species with large germinants, like Acer and Quercus, for light as young tree seedlings
and Carex grow in the same forest understory stratum (Chapter 1,2, & 3). It is more

likely that in this sedge/seedling system, much like grassland dominated areas, below-

134

ground resource competition for moisture is the primary mechanism limiting tree
seedling growth and survival (Davis etal., 1999 & 2005).

Competition studies (see Casper and Jackson’s review —Plant Competition
Underground) have been conducted in multiple systems (grassland, old-field, savanna,
conifer & deciduous forests) quantifying seedling growth and survival under varying
regimes of moisture, nutrient, and light availabilities. These differing regimes alter the
intensity and outcomes of plant-plant competition (Sims and Mueller-Dombois 1968,
Davis et al. 1999 & 2005, Wilson and Tilman 1993, Scholes and Archer 1997).

Carex, which allocates approximately 80% of its biomass below ground to roots
and rhizomes (chapter 4) has the ability to share nutrients and moisture between genet
and ramet, thus increasing young plant survivorship in times of stress (de Kroon et al.,
1998). In contrast, tree seedlings are not connected to their parents. Furthermore as tree
seedlings grow they rapidly extend roots into deeper zones than grasses and Carex to
obtain moisture and nutrients. However, young seedlings, and seedlings stressed by
factors such as browsing (Frank and Evans 1997) and low light (J .M. Kunkle, M.B.
Walters, R.K. Kobe, unpublished data) can have most of their roots in shallow surface
zones, potentially increasing their competitive interactions with Carex and other herbs
(Knoop and Walker 1985, Dodd etal., 1998). Increased survivorship, growth (Elliott and
White 1987, Davis et a1. 1999, Gordon and Rice 2000), and recruitment (Pseudotsuga
menziesii- Dunne and Parker 1999), have often been observed for tree seedlings when
competing vegetation was removed or when water was added (Harrington 1991, Davis et
al. 1999). Conversely, some studies have reported that soil water potential did not change

when surrounding vegetation cover was removed (Coates etal., 1991), and that soil

135

moisture did not negatively impact seedling density (Maguire and F orrnan 1983).
Berkowitz et al., (1995) even suggests that surrounding vegetation might buffer young
seedlings during periods of drought.

Several factors could contribute to the lack of generality in the results of these
studies: variation in vegetation density and physiology, climate (precipitation,
evaporative demand), and soils. Nonetheless, it is clear that competition for water can
limit tree seedling grth and survival in some systems. It is less clear how this
competition occurs. Superior competitors: 1) could maintain photosynthesis and
transpiration (and thus growth) at lower water potentials, potentially conferring a growth
advantage; 2), by maintaining photosynthesis/transpiration at lower plant water
potentials, could diminish soil water potentials to levels jeopardizing the survival of less
drought tolerant species (Tilman 1982); 3) could better tolerate (i.e. survive) the effects
of a given level of stress (drought) (Grime 1977).

In addition to water, sedge could out-compete tree seedlings for nutrients,
including nitrogen (N), the most frequently limiting nutrient in temperate terrestrial
systems, but to date, data are sparse. Possible mechanisms for Carex effects on N include
1) Carex being able to reduce mineral soil N to levels inadequate for tree seedling grth
and survival (Wedin and Tilman 1993), and/or 2) Carex altering available N (e. g. N03’,
NHX, organic N) to forms that tree seedlings can not use as effectively or by sequestering
substantial N resources in rhizomes and utilizing those stored resources to rapidly replace
tissue lost due to browsing (Bryant et al., 1983)

Can competition for water help explain Carex’s negative impacts on tree seedling

growth and survival found in the field? Specifically, 1) Can Carex draw soil water to

136

lower levels than Acer saccharum seedlings? 2) Does Carex have greater survivorship
than A. saccharum seedlings at low water availability? 3) Does a drought event
negatively affect A. saccharum growth? In addition, I asked, does Carex negatively affect
soil mineral nitrogen availability? I examined these questions with a three-year potted
plant experiment that included monocultures and combinations of Acer saccharum
seedlings and Carex pensylvanica culms, either well-watered or subject to a single dry-
down treatment during the second growing season. Water was withheld until stomatal
closure for either Carex or Acer in both combination pots and monoculture pots. Survival
and growth for seedlings were monitored through the third growing season.
Methods.
Design and Materials

In November of 2000, Acer seed was collected from Western New York which
was in the same USDA hardiness zone (4b) that sedge was collected from, mixed with
moist perlite, and cold stratified (2—4°C) in ziplock® bags. As I was interested in the
effects of drought, I chose to use a soil that would not hold excess soil water for extended
periods of time. In late March 2001, a relatively infertile field soil was collected from just
below the A horizon on a sandy outwash plain in southern Roscommon County, MI. Soils
in the region are a mixture of Rubicon — Menominee, and Graycalm and Grayling sands
(USDA soil survey of Gladwin County 1972 and Roscommon County 1972). Sites
typically support stands of aspen (mixed Populus tremuloides and P. grandedentada) and
Oaks (Quercus alba, Q. vellutina).

In early April 2001, Acer seeds with emerging radicles were planted into nursery

trays (12.7 cm depth X 6.4 cm diameter) filled with field soil, and allowed to establish for

137

three weeks in the MSU Tree Research Center greenhouse under 18 hr /6 hr (day length /
night length) conditions. Carex culm mats were collected in early April from the same
location that field soil was collected, transported to the Tree Research Center where they
were slowly warmed to 21°C allowing them to break winter dormancy, and were
maintained at field water holding capacity for three weeks under the same greenhouse
environment as germinating Acer seeds.

Individual Carex culms (consisting of a single rhizome and root not exceeding 5
cm in length) and three week old seedlings were planted as monocultures (n=4 plants) or
combinations (1 Acer and 3 Carex culms) in Poly-tainer ® 2- gallon pots filled with
homogenized infertile field soil. Pots were allowed to establish in the greenhouse until
June 15th 2001, at which time they were fertilized with a three-month slow release
fertilizer (Osmocote 15,15,15 at the rate of 200kg N/ha), and transferred to an outdoor
lathe house (50% shade) for a one-year establishment period. I randomly assigned
replicate pots from each vegetation treatment (see Table 4.1 for explanation) to a drought
treatment (well-watered/water withheld) and to two harvests (First harvest = the morning
after the designated stomatal closure point was reached, and Final harvest # after leaf
drop following the subsequent growing season). The 160 pots were grown under 50%
shade lathe with well-watered conditions until August 12th 2002, at which time I began
drought treatments by excluding both rain and supplemental watering on the appropriate
drought pots. At this time I harvested 2 or 3 pots each of Acer, Carex, and Acer-Carex
pots and found roots to be abundant and well distributed throughout the soil volume in all
treatments, suggesting that drought treatments should result in more or less homogeneous

soil water deficits within the soil volume. In addition, for the duration of the experiment

138

Acer leaf canopies were at or above the height of Carex in mixed pots indicating that
competition for light was likely minimal.

Due to concerns over a potentially serious experimental design flaw which were
raised during the review of this section I provide justification for the validity of the
results using a secondary study. The concern originated because I used seed fi'om a
western New York source (hardiness zone 4b-5a) and planted the developing young sugar
maple seedlings into sedge and soil collected from an ice contact/sandy outwash region in
the northern lower peninsula of Michigan (USDA hardiness zone 5a). Sugar maple
forests are not commonly associated with these ice contact/outwash sites and as such it is
possible that the sedge plants that I collected and used had adapted to their original soils,
thus conferring an advantage to sedge. Concerned with this possibility, I reran the entire
study the following year using soil, sugar maple seed, and intact sedge culms all collected
from a site in the Upper Peninsula of Michigan. In the follow-up study, sedge was found
to close their stomates at more negative internal plant water contents than sedge from the
ice contact /sandy outwash plain (-6.8 MPa vs. -5.1 MPa), respectively. If the sedge
plants used in the original study were adapted to the xeric soil conditions associated with
outwash and ice contact landforms, and were therefore functionally different than sedge
found growing in the understory of productive mesic sugar maple stands, I believe that
sedge stomatal closure points would have reflected these differences. Specifically, Sedge
closure should have occurred at more negative not less negative internal moisture
potentials when compared to sedge from mesic hardwood sites, and they did not. I
therefore feel confident that the results I obtained do highlight unbiased differences in

sedge and sugar maple responses to drought and competition.

139

Plant measurements pre/post harvest

During the dry-down period leaf transpiration was measured daily for all water
withheld treatment pots and every other day for well-watered pots with several Li-Cor
1600 steady state porometers (Li-Cor Inc., Lincoln NE). Drought treatments continued on
an individual pot basis until stomates closed for the target (i.e. assigned treatment)
species (Table 4.1). For each of the drought and well-watered control pots assigned to be
destructively harvested in the first round, I estimated plant water status for each
individual with predawn Scholander pressure bomb measurements of xylem water
potential the morning after stomatal closure. Drought pots not selected for the first
harvest were rehydrated when they reached stomatal closure. Following the drought
treatment, all remaining well-watered and drought pots were maintained under well-
watered conditions, overwintered in a hoophouse, and measured for spring and fall
seedling survival and growth during the following growing season. For both Acer and
Carex, survival was measured on an individual seedling or Carex culm basis. Final
harvests were conducted in late fall 2003 after Acer leaf drop.

At all harvests, I determined for each plant organ and species respectively, above
and below ground plant mass, N concentration, N content, as well as leaf area and total
root length). For Acer, I also measured annual height & diameter growth increment.
Nitrogen concentration was measured by the Dumas combustion method on a CN
analyzer (Carol-Erba, Milan, Italy).

Soil Resource Measurements
Immediately following the first harvest, soil was collected from each harvested

pot to determine KCl extractable soil nitrogen concentration. Soil from each pot was

140

thoroughly homogenized and sieved with a 4 mm screen, subsarnples were stored in
ziplock® bags at 1°C, extracted within 5 days with 2M KCl (20 g field moist soil with 50
ml KCl) on a shaker table for one hour, and allowed to settle for 30 minutes before being
filtered through Whatman® #42 filter paper. At the time of each extraction, soil moisture
was determined gravimetrically so that I could relate these measures directly to plant
water status measurements and to allow calculation of N03‘-N and NH4+-N on a per unit
dry soil basis. All extracts were refrigerated until being measured for NO3'-N and NH:-
N (within 30 days of extraction) on an 01 Alpkem Autoanalyzer(01 Analytical, College
Station, TX) by phenol-hypochlorite and cadmium reduction methods, respectively.
Analysis

Initially, 160 pots where divided into four groups of 40 pots and were randomly
assigned to one of four species treatments (Acer monocultures, Carex monocultures,
Acer-Carex combination pots dried to the stomatal closure point of Carex, and Acer-
Carex combination pots dried to the stomatal closure point of Acer). Within each of the
four species treatments (~40 pots/speCies treatment), pots were randomly assigned to
either a water or no water treatment. Prior to withholding water resources, pots were
visually evaluated for uniformity in Acer seedling and Carex culm numbers per
monoculture pot and pots that were non-uniform were discarded from the study. Of the
remaining 154 pots, six pots in each species treatment by water/no water drought
treatment were randomly assigned to the first harvest (total N = 48) and all remaining
pots in each species treatment by water/no water drought treatment (N = ~14 pots) were

designated as final harvest pots.

141

For analysis, species and drought treatments were considered discrete, nominal
variables. Main effects and interactions of species treatment and water/no water for plant
stomatal conductance, above-ground (total seedling and Carex mass, organ mass, leaf
area, and stem length for Acer) and below-ground growth characteristics (mass, total root
length) and soil (Extractable N, water content) were analyzed with least squares models
using JMP 5.1 statistical software (SAS Institute, Cary, NC, USA). Using water / no
water and species treatments as covariate predictors, I modeled seedling survival with
Cox’s proportional hazards methods (Cox 1972). Tukey’s HSD was used to test for
significant differences among water/no water and species treatments for plant and soil
characteristics. When interaction terms were insignificant beyond the threshold suggested
for pooling variances (P > 0.25, Bancroft 1964), the highest order interaction term with
the highest P value was removed and the model was re-run. This process was repeated
iteratively until a final model was constructed where all interactions >0.25 were removed.
ME
Stomatal conductance

Seasonal variation in Acer and Carex photosynthetic capacity was clearly evident
as conductance declined for both Carex and Acer in the well-watered controls over the
course of the study (Julian date 226-261, Figure 4.1A & B, Table 4.2). For water
withheld treatments, Acer, Carex and ricer-Carex mixtures reached their respective
significant conductance differentiation point from well-watered control treatments
between Julian dates 245 — 248 for Acer and Julian dates 247 — 250 for Carex following

the commencement of withholding water. All pots exposed to the induced drought event

142

reached their respective stomatal closure points within a three-day period (Julian days
258 - 260).
Vegetation and soil resource measurements corresponding to drought severity

At stomatal closure, droughted Carex and Acer monocultures had similar predawn
water potentials (approximately —5MPa, P=0.8091). Mean Acer monoculture predawn
moisture potential (MPa) was significantly lower than Acer seedlings in combination pots
at stomatal closure (Figure 4.2A, Table 4.3B), while Carex showed mixed results.
Specifically, pots that were harvested when Carex closed their stomates had similar
moisture potentials to Carex monocultures, while pots harvested when Acer stomates
closed had significantly less negative potentials than Carex monocultures. (Figure 4.28,
Table 4.3C) Gravimetric soil water was similar among all well-watered treatments and
likewise for all drought treatments (Figure 4.20, Table 4.3A.). Well-watered Acer
monocultures had significantly lower soil nitrate concentrations than well-watered Carex
monocultures and combination pots (Figure 4.2D., Table 4.3D.). However ammonium
concentrations were 6.2 fold greater than nitrate across all treatments and were unaffected
by water or species treatments (Figure 4.2E, Table 4.3E.).
Survival

Acer seedlings grown in monoculture or mixed with Carex had similar and high
survivorship (Figure 4.3A.). While there was a trend for droughted seedlings grown in
combination with Carex to have lower survival (30%) than those from seedling
monocultures (50%), differences were not significant. Interestingly, Carex survival was
unaffected by drought, surviving equally under well-watered and drought conditions

(Figure 4.3B).

143

Size, morphology, and N concentrations

Drought did not affect Acer (p=0.3950) or Carex (p=0.5403) total root length one
year alter the induced drought. In contrast, one year after the drought Acer mass was 23.6
% lower (Figure 4.4, P=0.0025), the subsequent growing season stem extension grth
was 52% lower (P=0.0006), and drought had no effect on Carex mass (Figure 4.4,
P=0.1213). Furthermore, drought did not affect root mass ratio, leaf mass ratio, specific
root length, leaf area ratio, and specific leaf area, for either Acer or Carex except that
specific leaf area (mz/ g) declined in response to drought for Carex (SS=187.5, F Ratio:
10.9, P=0.0025). Although we measured N content in plant organs immediately following
the drought, no clear pattern was discernable. I did, however, find that seedlings grown in
species combinations had higher concentrations of N in leaves, stems, and roots
compared to monocultures, while Carex had lower foliar and root N concentrations in
combination pots.
Discussion

Acer and Carex had virtually identical predawn plant water levels when grown as
monocultures, but Acer survival was roughly 50 % lower than monocultures of droughted
Carex (~100% survival). For monocultures of Acer, seedlings closed their stomates at
predawn moisture contents that were near their lethal leaf water potential (-5.76 MPa -
Auge et al., 1998). Although not significantly different than droughted monocultures,
Acer survival was more variable and trended lower in droughted combination pots,
especially when rehydration occurred at Acer stomatal closure. Seedlings in combination
pots seem to “sense” the presence of Carex and regardless of the point of rehydration

(Carex (A c-) or Acer (a C-) stomatal closure), close their stomates at significantly less

144

negative predawn moisture contents (approximately 40% less negative than droughted
Acer monocultures). Adding to the inconclusiveness of the results, seedling survival
trended lower in pots where rehydration occurred at seedling not Carex stomatal closure.
The shift in lower survival in combination pots with less negative water potentials does
not correlate with the lethal leaf water potentials put forth earlier. Taken at face value,
one would be inclined to surmise that seedling survival would have increased in these
combination pots if moisture content was the driving mechanism. As I did not see
stomatal conductance measurements decline more rapidly or complete stomatal closure
occur earlier in the combination pots, I concluded that seedlings were exposed to
equivalent drought acclimation periods.

Immediately following the first harvest, soil samples revealed that extractable
N03' -N levels were significantly lower in well-watered Acer than well-watered Carex
monocultures or combination pots. This contradicts Templer and Dawson’s (2004) work,
which showed Acer as preferential using soil NH4+ -N more than N03' -N. This result
needs to be viewed carefirlly as the overall levels of soil NH4+ -N were six times as great
as NO3' -N and showed no effect of water/no water treatments or species treatments.
However, variable N03' -N levels in the pot microcosms are in contrast to results from a
field study which looked at areas with higher vs. lower deer densities and found that at
higher deer densities, which also had three times the Carex biomass, had soil N03' -N
levels that were constant, whereas NH4+ -N levels almost doubled (Chapter 1). One
concern in comparing the two studies is that field sites were aspen dominated
monocultures with very little, if any, Acer trees attaining intermediate canopy levels in

these stands and thus overstory demands for differing forms of N may have driven the

145

differences. If, as this greenhouse study shows, Carex does not utilize available N03' -N
to levels that are lower than those of Acer monocultures and has no effect on NHa+ -N,
one might conclude that overabundant Carex does not reduce nutrients to levels that are
lethal to Acer.

While most of the organ level N results were mixed and unclear, foliar N
concentrations increased in Acer combination pots by 20% vs. Acer monocultures, while
Carex N concentrations decline 11% in combination vs. monocultures pots. The differing
reaction to the presence or absence of competitors by Acer and Carex suggests that in
areas with high Carex densities and regenerating Acer, seedlings become more palatable
to deer while Carex, which is already quite unpalatable (high silica contents- Prychid et
al. 2003), becomes even less palatable. The negative drought effects on sedge leaf and
root mass in combination pots were short lived, failing to be significant one year after the
induced drought. For surviving Acer seedlings, drought had no effect on growth the year
of the drought, but grth of seedlings was reduced the year after drought by 27%.

Tilman’s R“ model was not supported by the Carex X Acer interactions in this
study. Carex (the surviving plant) did not reduce soil moisture or nutrient resources to
lower levels in combination pots. In fact, Acer monocultures reduced N03' - N levels
below Carex monocultures and combination treatments. Even thought Carex internal
water potentials were lower than Acer water potentials in both combination treatments at
stomatal closure, both Carex and Acer in combination had, for the most part, less
negative potentials than their respective droughted monocultures. Again, seeming to not

support Tilman’s R* theory.

146

In this study, Carex seems to be both a good competitor and a good stress
tolerator. Carex’s ability to 1. rapidly usurp belowground space (1000+ m of roots by
year three for Carex vs. 18m for Acer), 2. protect itself from browsing through structural
means (intercalary meristem) & high foliar silica contents- Prychid et al.,2003) and
differential partitioning of resources to protected areas (80%+ of biomass is
belowground) and 3. Possibly consume luxury amounts of nutrients while storing the
nutrients in rhizomes giving it more of an advantage in periods of resource limitation,
makes sedge a great competitor. But, the uniformity in which Acer and Carex respond to
soil resource availabilities (equivalent internal moisture points in monocultures at
stomatal closure), and the differing survival responses to drought also means that Carex
could be labeled as a stress tolerant (Grime 1977).

In short, Carex, either as monocultures or in combination with Acer seedlings can
competed for nutrients and moisture resources and survive drought events much better
than Acer. Even thought Acer closed their stomates at a less negative water potentials in
the presences of sedge, survival was still compromised. Over time, as Acer seedlings fail
to survive and are replaced in the system with the established, surviving, clonal growth of
Carex, it may become more difficult for seedlings to acquire the needed resources
(moisture, nutrients, light, space) to survive (initial advantage altering competition —
Wilson 198 8, Gurevitch et al., 1990). We can only postulate, but subsequent drought
events may have even greater impacts on seedling survival.

Summagy
Continued use of a short rotation (8-15 years) selection based harvesting in

northern hardwood forests with elevated deer populations virtually assures that high

147

coverage of Carex will remain, if not expand. This study shows that sedge has the
potential to greatly reduce the number of seedlings in the forest understory following a
single drought event. Given the lack of information concerning impacts to seedling
survival following subsequent drought events in these systems, and the potential that
drought has cumulative impacts on seedlings survival, it is possible that virtual all sugar
maple seedlings could be eliminated from the understory prior to attaining a tall enough
stature to be considered browse for deer. For systems at least in Michigan, it is not
uncommon for mid-summer drought events to occur yearly or every other year (N 0AA
NCDC weather data). Furthermore, for those seedlings that do survive single or multiple
drought events, Carex induced water deficits may decrease and, in some extreme cases,
halt the extension growth of young seedlings completely (chapter 2 & 4). Cessation of
growth could place seedlings in an almost perpetual state of herbivory, adding yet more
stress to young seedlings.

Contrary to work by Pastor et al., (1993 & 1998) I did not find large shifts in
belowground resources with the addition of a “lower” quality species. The fact that Carex
and Acer have 1) similar stomatal responses to drought at roughly equivalent internal
moisture levels in monocultures, 2) similar levels of soil water in monocultures and
combination pots at stomatal closure, and 3) the surviving species (Carex) does not
reduce soil nutrient resources to lower levels than Acer seems to more closely fit Grime’s
CSR theory than Tilman’s R* model. More work is needed detailing the interactive
effects of competing species (Carex vs. Acer) under varying light regimes as light has

previously been shown to alter oak and grass competitive responses to drought (Davis et

148

al. 1999) and, more specifically, under low light conditions Acer may better tolerate

drought than Carex, while under high light, the reverse may be true.

149

 

Literature Cited

Auge, R.M., X. Duan, J .L. Croker, W.T. Witte, and CD. Green. 1998. Foliar dehydration
tolerance of twelve deciduous tree species. Journal of Experimental Botany
49(321):753-759.

Augustine, D. J ., L.E. Frelich, and RA. Jordan. 1998. Evidence for two alternate stable
states in an ungulate grazing system. Ecological Applications 8(4):]260-1269.

Bancroft, T.A., 1964. Analyses and inferences for incompletely specified models
involving the use of preliminary tests of significance, Biometrics 20:427-442.

Berkowitz, A.R., C.D. Canham, and V.R. Kelly. 1995. Competition vs. facilitation of tree
seedling growth and survival in early successional communities. Ecology
76(4):1156-1168.

Bryant, J .P., F .S. Chapin, and D.R. Klein. 1983. Carbon/nutrient balance of boreale
plants in relation to vertebrate herbivory. Oikos

Coates, K.D., W.H. Emmingharn, and SR. Radosevich. 1991. Conifer seedling success
and microclimate at different levels of herb and shrub cover in a Rhododendron-
Vaccinium- Menziesia community of south central British Columbia. Canadian
Journal of Forest Research 21:858-866.

Cox, D.R. 1972. Regression models and life tables (with discussions). Journal of the
Royal Statistical Society B. 34:187-220.

Davis, M.A., K.J. Wrage, P.B. Reich, M.G. Tjoelker, T. Schaeffer, and C. Muermann.
1999. Survival, growth, and photosynthesis of tree seedlings competing with
herbaceous vegetation along a water-light-nitrogen gradient. Plant Ecology
145:341-350.

Davis, M.A., L. Bier. E. Bushelle, C. Diegel, A. Johnson, and B. Kujala. 2005. Non-
indigenous grasses impede woody succession. Plant Ecology 178:249-264.

de la Cretaz, A. and M.J . Kelty 2002 Development of tree regeneration in fem-dominated
forest understories after reduction of deer browsing. Restoration Ecology
10(2):416-426.

DeForest, J .L., D.R. Zak, K.S. Pregitzer, and A.J. Burton. 2004. Atmospheric nitrate
deposition, microbial community composition, and enzyme activity in northern
hardwood forests. Soil Science Society of America 68: 132-1 38.

De Kroon, H., E. van der Zalm, J .W. A. van Rheenen, A. van Dijk, and R. Kreulen 1998.

The interaction between water and nitrogen translocation in rhizomatus sedge
(Carex flacca). Oecologia 116:38-49.

150

Dodd, M.B., W.K. Lauenroth, J .M. Welker. 1998. Differential water resource use by
herbaceous and woody plant life-forms in a shortgrass steppe community.
Oecologia 117:504-512.

Dunne, J .A. and VT Parker. 1999. Species-mediated soil moisture availability and
patchy establishment of Pseudotsuga menziesii in Chaparral. Oecologia 119:36-
45.

Elliott, K,J, and AS. White. 1987. Competitive effects of various grasses and forbs on
ponderosa pine seedlings. Forest Science 33(2):356-366.

Frank, DA. and RD. Evans. 1997. Effects of native grazers on grassland N cycling in
Yellowstone National Park. Ecology, 78(7):2238-2248.

George, L0. and RA. Bazzaz. (1999) The fern understory as an ecological filter:
growth and survival of canopy-tree seedlings. Ecology 80 (3):846-856.

Gordon, D.R. and K.J. Rice. 2000. Competitive suppression of Quercus douglasii
(Fagaceae) seedling emergence and growth. American Journal of Botany
87(7):986-994.

Grime, J .P. 1977. Evidence for existence of 3 primaru strategies in plants and its
relevance to ecological and evolutionary theory. American Naturalist
111(982):1169-1194.

Gurevitch, J ., P. Wilson, J .L. Stone, P.Teese, and R.J. Stoutenburgh. 1990. Competition
among old field perennials at different levels of soil fertility and available space.
Journal of Ecology 78:727-744.

Harrington, G.N. 1991. Effects of soil moisture on shrub seedling survival in a semi-arid
grassland. Ecology 71(3):1138-1149.

Knoop, W.T. and B.H. Walker. 1985. Interactions of woody and herbaceous vegetation in
a southern African savanna. Journal of Ecology. 73:235-253.

Kosola, K.R. and KL. Gross. 1999. Resource competition and suppression of plants
colonizing early successional old fields. Oecologia 118:69-75.

Maguire, DA. and R.T.T. Forman. 1983. Herb cover effect on tree seedling patterns in a
mature hemlock-hardwood forest. Ecology 64(6): 1367-1380

Pastor, J ., B. Dewey, R.J. Naiman, P.F. McInnes, and Y. Cohen. 1993. Moose browsing

and soil fertility in the boreal forests of Isle Royal National Park. Ecology
74(2):467-480.

151

Pastor, J ., B.Dewey, R. Moen, D.J. Mladenoff, M. White, and Y. Cohen. 1998. Spatial
patterns in the moose-forest-soil ecosystem on Isle Royale, Michigan, USA.
Ecological Applications 8(2):41 1-424.

Prycid, C.J., P.J. Rudall, and M. Gregory. 2003. Systematics and biology of silica bodies
in monocotyledons. The Botanical Review 69(4):377-440.

Scholes, R.J. and SR. Archer. Tree-Grass interactions in savannas. Annual review of
ecology and systematics. 28:517-544.

Sims, HP. and D. Mueller-Dombois. 1968. Effects of grass competition and depth to
water table on height growth of coniferous tree seedlings. Ecology 49(4): 597-
603.

Stromayer, K. A. K., and R. J. Warren. 1997. Are overabundant deer herds in the eastern
United States creating alternate stable states in forest plant communities. Wildlife
Society Bulletin 25(2):227-234.

Templer, RH. and TE. Dawson. 2004. Nitrogen uptake by four tree species of the
Catskill Mountains, New York: implications for forest N dynamics. Plant and Soil
262:251-261

Tilman, D. 1982.Resource competition and community structure: Monographs in
population biology. Princeton University press. pp 296.

Wedin, D and D. Tilman, 1993. Competition among grasses along a nitrogen gradient —
initial conditions and mechanism of competition. Ecological Monographs
63(2): 199-229.

Wilson, J .B. 1988. The effect of initial advantage on the course of plant competition.
Oikos 51 :19-24.

Wilson, SD. and D. Tilman. 1993. Plant competition and resource availability in
response to disturbance and fertilization. Ecology 74(2):599-611.

152

 

1.5

 

 

 

 

 

 

 

A.
Carex
Mono-
1.0‘

AC- 3C.
0 N. _
o 0.5‘ H
g .., ............... Mom,
”’3? 6‘
:3\ 1.\ 000000000 Ac-
'8” ‘Oo\.~n
as ,, ....Tfft'r; — .o
"E . B —-- aC-
3" I Acer
°° Ac+
E
9
m

 

 

1.0 ‘

0.5 ‘

 

 

 

0.0

 

Julian date

Figure 4.1. Daily Acer and Carex stomatal conductance (mol/mz/s) by treatment over the
course of the study (Julian days 226-261 ). Lines represent point at which
water vs. water withheld treatments became significantly different (P=0.05).

153

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

A. A LIJ A LI—J A La—J
-1 .
-2 .
53 b b
2 '3 +
-4 l a
-5 4 Acer
0 MPa @ stomatal closure
B. A L+I B W B l—I—T
.1 l
-21
53 3
2 44 ab :4
a i-
-5- “I— Carex
6 MPa @ stomatal closure

o’ 0" 0
$0? 09 Y”

a

Y‘

Ix
6‘15;be

Species Treatment

% Soil Water

No,‘ -N

NH,+ -N

18

16-
14-
12-

0.20

0.16 -

0.12

0.08 1
0.04 '

1.2

1.0 -

0.8

0.6 r

0.4

0.2 1

0.0

H
N-fiOOOO

 

1

 

C.

% Soil water _}
@ stomatal closure—P

B B B B

A A A A
['1 1"] m ['1

 

 

 

 

 

 

 

 

 

 

A- A+ C- C+Ac—Ac+aC-aC+

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Species Treatment
'3' t Extractable No,'-N
l
i l j
A B B B
E' l Extractable NH4’-N
l I 1

A A A A

 

 

 

 

 

 

 

 

 

 

f I T I

X X 0X X
Y, O Y’ «>0
Species Treatment

Figure 4.2. Acer (2A) and Carex (2B) predawn moisture potentials (MPa) fiom seedling
and Carex culms reaching their respective stomatal closure points or from
well-watered control individuals. Gravimetric soil water by treatment at the
first harvest (2C). Extractable N03" (2D) and NHa+ (2E) immediately
following the first harvest. Error bars represent 3:1 St. Error. Large cap letters
in 2A&B represent differences between Acer and Carex within a given
species treatment, while small cap letters represent differences (P=0.05) in
species treatments within either Acer or Carex. In 2C, large cap letters
represent differences (P=0.05) between water and no water treatments. Large
cap letters represent differences in either extractable N03' -N (2D) or NH4+ - N

(2E) across well-watered treatments.

154

 

100, Acer __ __ ] Carex _ _

+
7680i -—L
> ABABAB
§601
m AAAAAA
°\°40<

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

20‘ m

 

 

X X
0

X 0’ Oi Oi X 0’ X cl
9&0 ‘19Q y- ? $0 to @090 4‘09 t?" y- e to

Species Treatment

Figure 4.3. Acer (3A) and Carex (3B) survival one year after an induced drought event.
Error bars represent :1:] St. Error. Letters represent differences (P=0.05) in

survival across species treatments within Acer or Carex.

155

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5 40
£9 4 ~ Acer 13".”; “35": 39 Carex
‘8 I m 30 ‘
to r 8
a 3 , 1 e l
1: '° 1
g 2 :, g 20 ‘
5" 1 '1 5’
g 1 < g 10 '
m m

0 v x"? ." .- .- ‘ 0 Y

no water water no water water

Figure 4.4. Acer and Carex belowground grth by water / water withheld treatments.

 

156

Table 4.1. Explanation of treatments used in the study.

Treatment ID

Mono- Monocultures of either Carex or Acer, dried to their respective stomatal closure points
Mono+ Monocultures of either Carex or Acer, well- watered controls

A c- 1 Acer 3 Carex, dried to the stomatal closure point of Carex

A 0+ 1 Acer 3 Carex, well- watered controls for A c-

a C- 1 Acer 3 Carex, dried to the stomatal closure point of Acer

a C+ 1 Acer 3 Carex, well- watered controls for a C-

 

157

Table 4.2. Results of a standard least squares mixed model for the effects of water/no

water, species treatment, date, and their interactions on Acer and Carex
stomatal conductance. Interactions with P>0.25 were pooled with the error
term (Bancroft 1964) and the models rerun.

 

Stomatal conductance Anova effects SS F P

Acer Water/No Water 0.3353 31.5112 <0.0001
Species Treatment 0.2377 11.171 <0.0001
WxST 0.0733 3.4469 0.0321
Date 8.0169 753.529 <0.0001
WxD 0.5572 52.3706 <0.0001
STxD 0.0384 1 .8062 0.1647
WxSTxD 0.1 171 5.5025 0.0042
Adj. R2 0.4344

Carex Water/No Water 0.231 1 46.6571 <0.0001
Species Treatment 0.0346 3.4891 0.0307
WxST 0.0472 4.7633 0.0086
Date 2.7134 547.926 <0.0001
WxD 0.3255 65.7348 <0.0001
STxD 0.0018 0.186 0.8303
WxSTxD 0.0078 0.7919 0.4531
Adj. R2 0.3371

 

158

 

 

Table 4.3. Results of a standard least squares mixed model for the effects of water/no
water, species treatment, and their interaction on gravimetric soil moisture
(3A), predawn seedling moisture potential (3 B), predawn Carex moisture
potential (3C), and standing pools of N03' (3D) and NHa+ (3E) at the point of

stomatal closure.

 

 

 

 

 

1.Stomatal closurt Source DF SS F Ratio P>F
A. Soil Water Water/No Water 1 1592.5 173.48 <0.0001
Species Treatment 3 56.055 2.0354 0.1252
WxST 3 61.523 2.234 0.1
Adj R2 0.7618
B. Acer water Water/No Water 1 9877.6 439.34 <0.0001
Potential Species Treatment 2 955.13 21.24 <0.0001
WxST 2 718.16 5.97 <0.0001
Adj R2 0.9376
C. Carex water Water/No Water 1 16120.5 800.47 <0.0001
Potential Species Treatment 2 295.63 7.34 0.0025
WxST 2 269.41 6.69 0.004
Adj R2 0.9592
D. N03- -N Water/No Water 1 0.2817 1.921 0.1738
Species Treatment 3 3.5773 8.085 0.0003
WxST 3 1.4281 3 .2456 0.0324
Adj R2 0.3902
E. NH4+ -N Water/No Water 1 0.0014 0.052 0.8209
Species Treatment 3 0.068 0.8332 0.484
WxST 3 0.1517 1.8601 0.1528
Adj R2 0.0143

 

159

 

Table 4.4. ANOVA results for plant water potential testing between Acer and Carex
within a given species treatment (monoculture, A c, a C). only for no water
treatment individuals.

Source DF SS F Ratio P>F
MonoculturesSpecies 1 0.021888 0.0617 0.8089
Error 10 3.54971
C. Total 11 3.5716
Ac Species 1 5.005208 9.1832 0.0127
Error 10 5.450417
C. Total 11 10.455625
aC Species 1 1.960208 7.1226 0.0235
Error 10 2.75208
C. Total 11 4.71229

160

 

Conclusions by chapter
1. Impacts to aspen understog structure and composition across site productivity and
stand age gradients in higher and lower deer density areas.

Deer have altered the understory composition of forbs and seedlings, and have
directly reduced the structural characteristics of future stands. Although stand age and
site productivity did explain some responses, their lack of interaction effects with deer
focused most of the discussion and management implications on elevated deer density
effects. Specifically this study found that:

- Levels of non-browsed (unpalatable) species (sedge and fern) have tripled on sites

with higher relative deer densities across the entire range of site productivities,

 

increasing aboveground (light) and belowground (soil water, nutrients, and space) i
resource competition between these species and establishing seedlings.

- While higher relative deer densities have reduced forb biomass in aspen stands to
1/10‘h that of lower, more moderate deer density areas, seedling stem densities (0-
0.6m tall) have remained constant.

- Compositionally, deer have simplified forb and seedling layers having greater
negative impacts to richness on more productive sites. Increased red maple stem
densities 0-0.6m tall, at the expense of oak and intermediate canopy species such
as witch hazel, indicate a shift in the future stand composition.

- Stand management goals, objectives and methods, along with deer populations and
inherent site factors, will play a part in determining stand stocking levels. As
such, it would be misguiding for managers to take the results of this work and

definitively state that future stands will most likely be only marginally stocked

161

(250-300 stems/ha) with black cherry (the only species represented in the 0.6-4m

tall measurement zones).
Aspen management implications

It would be advisable for resource managers to coordinate harvesting and wildlife
population management activities at a landscape level. This would allow forest managers
to concentrate traditional forest harvesting activities in areas with lower deer densities
and perhaps use more experimental harvesting or site treatment methods in areas with
higher deer populations in hopes of minimizing or ameliorating the negative effects of
deer. Resource managers need to realize that forest harvesting will, by its practice,
promote an increased quantity and better quality of available deer browse, and that deer
populations will respond accordingly. The potential for future deer impacts to the
herbaceous layer compositions should be evaluated with site productivity in mind, and be
incorporated into management plans in all stands, with preference and resources going to
stands in areas with high deer densities. Furthermore, the costs, both social and on the
ground, associated with stand and understory remediation will rise if steps are not taken

to control the deer herd in areas of active forest management.

11. Deer and sedge impacts to vegetation dynamics in northern hardwood systems of the
Upper Peninsula.
While sedge does play a minor role in altering seedling growth in areas protected
from deer, enough seedlings were able to establish, survive, and grow in dense sedge
areas that the I cannot target sedge as the primary cause of the regeneration failure

common in these high sedge - high deer environments. Deer effects, on the other hand,

162

 

dominated the system, altering understory forb and seedling compositions and outright
eliminating seedling growth into height class above the zone of deer browsing.
Specifically, results indicated that

- Mid summer (July) herbicide treatments are effective at reducing sedge mass (94-
97% after 2 years and 50% after 4 years), providing a window of reduced
vegetation competition to help tree seedlings establish. Unfortunately, tree
seedlings were also susceptible to foliar herbicide application, but through
seedling recruitment, biomass was not significantly different 4 years after
treatment.

- While vegetation treatments did not affect forb mass, the presence or absence of
deer greatly influenced herbaceous layer compositions after 4 years. Populations
of Carex, dandelion, goldenrod, raspberry as well as seedling covers of red maple,
balsam fir, and to a lesser extent, white ash increased in areas open to deer, while
trillium, lily-of-the-valley, toothwort, sweet cicely, and sugar maple all increased
in areas protected fi'om deer.

- Although seedling biomass levels rebounded by year 4 in spray areas, the practice of
mid summer spraying should be minimized. Summer spraying lengthens the time
that it takes to get a new cohort of trees beyond the reach of deer, primarily
because the new cohort must establish from seed as spraying kills advanced
regeneration.

- The exclusion of deer greatly improved tree seedling survival and overall growth
while herbicide treatments increase growth, especially as overstory conditions

improved (increased canopy openness), but had no impact on seedling survival.

163

- Measured soil properties (extractable N03- - N & NH4+ - N and N mineralization
rates) were not altered by deer or vegetation manipulation treatments in this study.
One might expect theses systems to be more resistant to belowground nutrient
changes as the vast majority of litter influx comes via the overstory, where
ungulate induced pressures are last to be felt in these systems.
Northern hardwood management implications
In selection harvested productive northern hardwood systems with elevated deer
densities, the sustainability of the current forest is in question. Deer have eliminated all
sugar maple regeneration in these areas and are even reducing ironwood. If forest and
wildlife management practices (selection harvesting, allowing deer herds to remain
elevated in areas where impacts to regeneration are already quantifiable) are not altered,
systems with high deer will continue to have chronic seedling recruitment failure, and
increasing loss of intermediate canopy trees and associated intermediate canopy wildlife
species. Eventually, the system may lose its overstory structure, compositional diversity,
and light reducing ability. The ability to reduce light through closed canopy conditions is
critical in these systems if sedge is to be reduced in a manner that does not rely on
chemicals. Along these same lines, forest managers should consider increasing the
reentry period from the standard 8-12 years, (which is the current practice on industrial
lands) to 30-35 years or more, as the longer timeframe between cuts should allow for
sustained low light conditions to promote low sedge mass levels in the understory.
Preferably, this would occur in tandem with a reduction in the deer herd to levels that

allow for tree regeneration (~7 deer/kmz).

164

III. Selectively removing sedge at larger scales with timed herbicide gpplications

Strategically timing herbicide applications can effectively reduce populations of
competing vegetation while minimizing the herbicide damage to desired tree seedlings
and forbs. At a 0.2ha scale, greater deer damage was found in treated areas. Perhaps if
treated areas were larger and deer feeding was not as concentrated, the positive effects of
the spraying would translate into greater numbers of seedlings growing into taller height
classes. Timed herbicide treatments resulted in:
- Sedge being as effectively controlled with a fall (November 1) application of
herbicide when compared to a July 15 spraying, and had the added benefits of not
reducing forb or seedling mass or spring ephemeral plant richness. Fall spraying
increasing tree seedling germination, establishment and survival over both julyl 5
spraying and controls.
- While levels of deer damage were higher than controls with November spraying,
July spray treatments received more than twice the deer damage at higher light
levels than comparable November spray areas.
Management implications to fall spraying in northern hardwood forests.

More work is needed to transfer the 0.2 ha results to a stand or landscape level.
Forest managers should be cautiously optimistic about fall spraying’s effectiveness at
controlling sedge, but questions remain as to the viability of this treatment at larger scales
when faced with elevated deer densities. Management of the landscape structure
surrounding northern hardwood stands (reduction in cedar swamps) in conjunction with
stand level fall applications of herbicides may be more effective than either treatment

individually. Managers need to retain larger diameter seed trees and attempt to plan

165

 

treatments (harvesting, spraying etc.) around good seed years. This is not always possible
but, to date, fall treatments give a 2 year (possibly up to a 4 year) window of reduced
plant competition before sedge reinvades and dominates the understory. Again, more
work is needed at greater spatial extents, controling for surrounding landscape features
(cedar swamps) but overall this work highlights the ability to use an inexpensive
herbicide in the fall to control seedling competitors while not harming the seedlings

themselves.

IV. Mechanisms driving the sedge - maple interaction
Although sugar maple and sedge monocultures respond similarly during drought
events by closing their stomates at equivalent moisture contents, few other similarities
exist between seedlings and sedge. In combination pots, sedge did not reduce soil
moisture or nutrients to levels that were lethal to seedlings and maintained its dominance
through its ability to survive drought events while not altering belowground growth. The
potential ramifications of the observed shift in foliar N concentrations where sugar maple
increased and sedge declined when sedge and maple were grown together could be a
factor in why seedlings are browsed more so than sedge. This controlled environment
study specifically showed that:
- Sedge and maple monocultures respond similarly to drought by closing their
stomates at extremely low yet equivalent internal moisture contents. For maple,
internal moisture contents at stomatal closure was close to known levels of lethal

leaf water content.

166

- Even with the readjustment of stomatal closure shown by maple (stomates closed at
less negative water potentials in combination vs. monoculture pots) and the
greater reduction in extractable N by maple, sedge post drought survival was
virtually 100% vs. maple seedlings, which survived 20-50% of the time.

- Droughted maples had less belowground biomass a year after the drought while
sedge was unchanged, possibly indicating that in subsequent drought years
seedling mortality might increase.

- Foliar N concentrations increased for sugar maple in combination pots while N
concentrations decline. Although interesting, the fact that my previous work
found greater browsing in areas where sedge was eliminated means that deer may
be focussing on seedlings that are visually easier to find vs. ones that have higher
N content. If the reverse was true, seedlings in the sedge controls should have had
the highest foliar N concentrations and the greatest amount of browsing damage,

which was not the case.

Management implications from sedge — maple interactions

Seedlings in the field are often exposed to multiple drought events during the

understory establishment phase. It is possible that after several years and several drought

events there may be very few, if any, seedlings reaching a stature that deer would then

utilize. Another possible explanation is that drought induced mortality lowers the

seedling population to such a low level that deer are then physically able to browse all

stems. In reality, the lack of regeneration in the highly managed forests in the Upper

Midwest may not be due to deer browsing alone, but rather through the combined effects

of direct deer browsing on the remaining seedlings whose populations have been reduced

167

 

through drought events and competition with sedge. There might also be a feedback
mechanisms in place where deer induced high sedge densities promote increased levels of
browsing on preferred seedlings (maple has elevated foliar N concentrations when grown

with sedge) and the increased seedling browsing decreases future levels of competition

(for light) that sedge will face if seedlings establish.

168

r

   

lliliilllllillililil