View That Tree-Ring Stable Isotopes Can Supplement Ring-Width Analysis of Plant Productivity by Offering Long-Term Records on the Underlying Physiological Mechanisms

Home , Canopy research

Prescribed Fire and Thinning Effects on Tree Growth and Carbon Sequestration in

Mixed-Oak Forests, Ohio, U.S.A.

A dissertation presented to

the faculty of

the College of Arts and Sciences of Ohio University

In partial fulfillment

of the requirements for the degree

Doctor of Philosophy

Alexander K. Anning

December 2013

Prescribed Fire and Thinning Effects on Tree Growth and Carbon Sequestration in

Mixed-Oak Forests, Ohio, U.S.A.

ALEXANDER K. ANNING

has been approved for

the Department of Environmental and Plant Biology

and the College of Arts and Sciences by

Brian C. McCarthy

Professor of Environmental and Plant Biology

Robert Frank

Dean, College of Arts and Sciences 3

ABSTRACT

ANNING, ALEXANDER K., Ph.D., December 2013, Plant Biology

Prescribed Fire and Thinning Effects on Tree Growth and Carbon Sequestration in

Mixed-Oak Forests, Ohio, U.S.A.

Director of Dissertation: Brian C. McCarthy

Since the mid-1990s, the use of prescribed fire and thinning as part of an integrated forest management strategy has increased dramatically across the United

States, spurring numerous studies into how these treatments influence forest ecosystems.

However, despite a burgeoning literature on this topic the response of residual trees has not been thoroughly investigated. In this study, the effects of prescribed fire and thinning on residual tree growth and carbon sequestration and their underlying mechanisms were examined to better understand the impacts of the treatments on forest ecosystems. The study involved tree-ring analysis of 348 trees (DBH) obtained from 80 0.1-ha permanent plots in two sites, each with four experimental units (control, thin-only, burn- only and thin+burn). Treatments were applied in 2000/2001. Trees were selected from five common overstory species or taxa in the study area—white oak (), chestnut oak (), black oak (), hickories ( spp.) and yellow- poplar (). In addition to elucidating the long-term (1991-2010) variability of tree growth in relation to the treatments, the study examined the modulating effects of tree age, size, competition, and soil moisture gradient. Furthermore, the variability of carbon isotope ratios (13C) of white oak was assessed to gain more insight into the physiological response of trees to the treatments. Key findings of the study 4 include: (1) treatments caused considerable increase in tree growth, though the mechanical thinning treatments were more effective than the prescribed fire treatment at eliciting growth changes; (2) tree growth exhibited a strong temporal trend characterized by a sharp increase in BAI during the first 5-year post-treatment period with a slight attenuation thereafter; (3) competition was the most important determinant of residual tree growth, exhibiting the greatest effect in the thin-only stands; (4) variations in topographically-controlled soil moisture stress/demand strongly influenced tree growth, but this effect was more pronounced in the control stand than in the managed stands, where treatment effects became the main drivers of growth; (5) interspecific variations in tree growth response were evident; and (6) white oak 13C declined over time, suggesting

13 12 an increased discrimination against CO2 relative to CO2 and a reduction in water-use efficiency of the species, most likely related to changes in regional climate than to the treatments or microclimate. The results demonstrate that prescribed fire and thinning influence tree growth and forest productivity by creating heterogeneity (i.e., by altering the competitive status of trees) within and among stands, though responses may vary depending upon the species, soil moisture status, size, and topographic position of trees, among other factors. The findings of this study have important implications for forest ecology and management—they show the need for long-term, spatially explicit, tree- based analysis of residual tree growth to fully understand prescribed fire and thinning impacts on forest ecosystems. 5

DEDICATION

ACKNOWLEDGEMENTS

I thank God for bringing me this far in my education. I would like to express my profound gratitude to my advisor, Dr. Brian McCarthy, for his expertise, support, mentorship and guidance, which made my graduate education in the US worthwhile. To my dissertation committee members, Drs. Glenn Matlack, James Dyer, Donald Miles, and Todd Hutchinson, your keen interest, invaluable suggestions, and constructive criticisms made this dissertation possible, and I am grateful. I also thank Dr. Daniel

Johnson, who stepped in at a crucial moment during my Comprehensive Examination, and provided the essential dimension of Plant Physiological Ecology. To the faculty and staff of the Department of Environmental and Plant Biology, I appreciate your support.

I thank the staff of the USDA Forest Service, Northern Research Station,

Delaware, OH, for site permission. Financial support for the study came via a Donald

Clippinger Fellowship, a Student Enhancement Award, a Graduate Student Senate

Original Work Grant, a Hiram Roy Fund, and funding from the Ohio Center for Ecology and Evolutionary Study—all of Ohio University, and I am grateful to them all.

My appreciation also goes to all my lab mates, especially Keith Gilland, Ryan

Homsher, Chase Rosenberg, Joseph Moosbrugger, Nathan Daniel, Stephen Murphy,

Lauren Bizzari, Corey Kapolka, William Rucker, Kathleen Gabler and Bailey Hunter for supporting me in diverse ways. I acknowledge with thanks field assistance from Joseph

Moosbrugger, Ryan Lima, Lauren Bizzari and Aaron Cranford. Many other people helped me in this study. To these and the many friends who have been with me through thick and thin, I say may God richly bless you all. 7

TABLE OF CONTENTS

Page Abstract ...... 3 Dedication ...... 5 Acknowledgements ...... 6 List of Tables ...... 10 List of Figures ...... 11 Chapter 1: Introduction ...... 14 Literature Cited ...... 20 Chapter 2: Long-Term Effects of Prescribed Fire and Thinning on Residual Tree Growth in Mixed-Oak Forests of Southern ohio...... 27 Abstract ...... 27 Introduction ...... 28 Methods ...... 31 Description of Study Sites and Experimental Design ...... 31 Increment Core Sampling, Preparation and Measurement ...... 32 Growth Trend Analysis ...... 34 Data Analysis...... 35 Results ...... 37 Species Chronologies ...... 37 Treatment Effects on Tree Growth Over Time ...... 38 Variation in Tree Growth with Fire Severity ...... 42 Discussion ...... 43 Treatment Effects on Tree Growth ...... 43 Temporal Patterns of Tree Growth Response to the Treatments ...... 45 Interspecific Variations in Tree Growth Response to the Treatments ...... 46 Conclusions and Management Implications ...... 48 Literature Cited ...... 50 Chapter 3: Effects of Competition, Size and Age on Tree Growth Response to Prescribed Fire and Thinning Treatments in Mixed-Oak Forests of Ohio ...... 66 Abstract ...... 66 Introduction ...... 67 Methods ...... 71 Description of Sites and Experimental Treatments ...... 71 Increment Core Sampling and Measurements ...... 72 Growth Analysis ...... 74 Neighborhood Analysis ...... 74 Assessing Size, Age and Competition Effects on Tree Growth ...... 76 Results ...... 78 Characteristics of Trees Sampled ...... 78 Size and Age Effects on Tree Growth ...... 78 Competition Effects on Tree Growth ...... 79 Relative Importance of Competition, Size and Age for Tree Growth ...... 81 8

Discussion ...... 82 Conclusions and Management Implications ...... 86 Literature Cited ...... 88 Chapter 4: Tree Growth Response to Prescribed Fire and Thinning Treatments Along a Topographic Moisture Gradient in Mixed-Oak Forests, Ohio, USA ...... 105 Abstract ...... 105 Introduction ...... 106 Methods ...... 109 Site Description and Experimental Design...... 109 Increment Core Sampling and Measurement ...... 110 Growth Analysis ...... 112 Modeling Soil Moisture Demand and Availability ...... 112 Data Analysis...... 114 Results ...... 116 Variations in Moisture Demand/Stress Across the Landscape...... 116 Effects of Soil Moisture Gradients on Tree Growth Response to the Treatments ...... 116 Variations in Tree Growth Response of Different Taxa to the Moisture Gradient Across the Treatments ...... 117 Discussion ...... 119 Literature Cited ...... 123 Chapter 5: Stable-Carbon Isotope Composition of White Oak Trees in Relation to Soil Moisture Stress and Restoration Management in Central Hardwood Forests ...... 139 Abstract ...... 139 Introduction ...... 140 Methods ...... 143 Study Sites and Experimental Design ...... 143 Field and Laboratory Works ...... 145 Stable Isotope Analysis ...... 147 Statistical Analysis ...... 148 Results ...... 149 Climate Characteristics of the Study Sites ...... 149 13C and % Carbon of White Oak Between Treatments and Over Time ...... 149 Effects of Soil Moisture Stress and Slope-13C Values ..... 150 Discussion ...... 152 Literature Cited ...... 156 Chapter 6: Conclusion...... 168 Management Implications ...... 171 Literature Cited ...... 173 Appendix 1A: The Complete Set of A Priori Models Explaining the Variations in BAI ...... 174 Appendix 1B: The Complete Set of A Priori Candidate Models Explaining the Percent Changes in BAI ...... 175 9

Appendix 2A: Stand-Level Basal Area of the Two Replicate Blocks Before and After the Treatments...... 176 Appendix 2B: The Five Set of Competition Indices Compared in Current study ...... 177

LIST OF TABLES

Page Table 2.1: Summary statistics of ring-width series of five mixed-oak species analyzed in this study ...... 58

Table 2.2: Multi-model analysis: best models indicating the relative effects of fuel reduction methods and time since initiation of treatments (two 5-yr period of growth; 2001–2005, 2006–2010) on post-treatment BAI and percent change in BAI (% GC) responses of five mixed-oak forest tree species. A priori set of candidate models were derived from combinations of site, species, treatments and time, and evaluated using mixed models and Akaike information-theoretic approach (Appendices 1A and 1B). Akaike weight (w) indicates the probability of a selected model being the best among the set of competing models. % GC was log-transformed to stabilize its variance before the analysis ...... 59

Table 3.1: Mean (± standard error) basal area increment (BAI), size (i.e., diameter at breast height, DBH) and age of residual trees sampled from the prescribed fire and thinning treatment units. BAIb and BAIa indicate annual BAIs before (1991– 2000) and after (2001–2010) treatment, respectively ...... 95

Table 3.2: Summary of analysis of covariance showing the effects of size (DBH) and age on basal area increment (BAI) responses of tree species to prescribed fire and thinning treatments in southeastern Ohio ...... 96

Table 3.3: Analysis of covariance (ANCOVA) results ( value and 2) showing the influence of competition on BAI responses of some common overstory species to prescribed fire and thinning treatments in southeastern Ohio ...... 97

Table 3.4: Effect of site quality on growth-competition relations of species following prescribed fire and thinning management in mixed-oak forests of southeastern Ohio. Site quality classification was based on the integrated moisture index (IMI; Iverson et al. 1997) ...... 98

LIST OF FIGURES

Page Figure 2.1:Vegetation characteristics and fuel reduction treatment operations in the study sites: (A) control site with a downed tree; (B) prescribed burning; (C) thinning followed by prescribed burning; (D-F) vegetation showing some residual tree species after the treatments ………………………………… ...... 60

Figure 2.2:Annual rings (19962005) compared for the five tree species analyzed for their radial growth responses to prescribed fire and thinning treatments. Note the increase in ring widths across species (particularly yellow-poplar) following the treatments in 2000……………………………………………………………... 61

Figure 2.3: Composite raw and standardized ring-width chronologies of the five mixed- oak forest species analyzed. Dotted vertical line indicates the period of growth analyzed and the arrows indicate when the treatments were applied...... 62

Figure 2.4: Periodic annual basal area increment responses of residual trees to prescribed fire and thinning treatments over time across species in mixed-oak forests of southern Ohio. Error bars represent one standard error. Note differences in y-axis. The dotted vertical line indicates the time the treatments were applied...... 63

Figure 2.5: Changes in basal area increment of residual tree species before and after prescribed fire and thinning treatments in mixed-oak forests of southern Ohio. Error bars represent one standard error. Note differences in y-axis. The dotted vertical line indicates the time the treatments were applied...... 64

Figure 2.6: Basal area increment responses of residual tree species to prescribed fire severity in the study area. Different letters in each panel indicate significant ...... 65

Figure 3.1: Relationships of tree size (A) and age (B) with basal area increment (BAI; 2001-2010) responses of residual trees to prescribed fire and thinning treatments in southeastern Ohio. For clarity, age axis has been scaled such that two oldest trees are not shown...... 99

Figure 3.2: Variation in competitive status of individual trees across prescribed fire and thinning treatments in southeastern Ohio. Different letters indicate statistical difference at = 0.001...... 100

Figure 3.3: Effect of competition on basal area increment of residual trees after prescribed fire and thinning treatments in mixed-oak forests of southeastern Ohio ...... 101

Figure 3.4: BAI-competition relations compared for five species in control and active treatment (prescribed fire and thinning) stands in mixed-oak forests of southeastern Ohio. *** indicates significant difference at = 0.001, ** at = 0.01, * at = 0.05, and NS is non-significant ...... 102

Figure 3.5: Size-dependent competition effect on basal area increment following prescribed fire and thinning treatments in forests of southeastern Ohio. Panel (a) medium-size trees (25 to < 40 cm DBH); panel ...... 103

Figure 3.6: Relative importance of competition, size, and age for basal area growth of residual tree species following prescribed fire and thinning treatments (data from control plots excluded from analysis) in southeastern Ohio. Bars represent means ± 95 % bootstrap confidence intervals with 1000 replications (Gromping 2006). All multiple regressions were significant at = 0.001...... 104

Figure 4.1: Location of Vinton County, Ohio where the study was conducted...... 131

Figure 4.2: Variations in precipitation and temperature at Raccoon Ecological Management Area (REMA) in 2003 and 2010 compared to normal climate (1981–2010). Normal climate conditions were obtained from the NOAA’s National Climatic Data Center (NCDC, 2012) at Carpenter 2S, OH, USA...... 132

Figure 4.3: Variations in annual PET and deficit (estimated for 2010) across parts of the landscape at the Raccoon Ecological Management Area (REMA) and the Zaleski State Forest, both in Vinton County, Ohio. Also shown are the topography and the locations of some of the sampled trees...... 133

Figure 4.4: Variations in seasonal potential evapotranspiration (PET), seasonal deficit and integrated moisture index (IMI) across physiographic positions in mixed-oak forests of Ohio. Bars represent mean values ± standard errors for individual trees. Note that the y-axis scale for panel (A) does not start from the origin. Different letters denote significant difference 134

Figure 4.5: Effects of seasonal potential evapotranspiration (PET), seasonal deficit (DEF), and the integrated moisture index (IMI) on basal area increment (BAI) responses of trees to prescribed fire and thinning treatment manipulations. BAI was computed for the 2006–2010 growth period (6–10 years post-treatment) while PET and DEF were modeled for 2010. Bars represent mean values ± standard errors for individual trees. Different letters on bars indicate significant = 0.05) difference in BAI among the moisture classes ...... 135

Figure 4.6: Effects of seasonal potential evapotranspiration (PET), seasonal deficit (DEF), and the integrated moisture index (IMI) on basal area increment (BAI; 13

2006–2010) responses of oaks (white oak, black oak and chestnut oak) and non- oaks (yellow-poplar and hickories) to prescribed fire and thinning treatment manipulations. PET and DEF were modeled for 2010. Bars represent mean values ± standard errors for individual trees. Different letters on bars indicate significant difference in BAI among the moisture classes...... 136

Figure 4.7: Effects of seasonal moisture deficit (DEF) and seasonal potential evapotranspiration (PET) on basal area increment (BAI; computed for 2006– 2010) of yellow-poplar in control and manipulated stands within the mixed-oak forests of southern Ohio. DEF and PET were modeled for the year 2010...... 137

Figure 4.8: Effects of seasonal moisture deficit (DEF) and seasonal potential evapotranspiration (PET) on basal area increment (BAI; computed for 2006– 2010) of white oak in control and manipulated stands within the mixed-oak forests of southern Ohio. DEF and PET were modeled for the year 2010 ...... 138

Figure 5.1: Topography of the study sites and location of some trees (dots) used for analysis...... 163

Figure 5.2: Mean seasonal (April-September) precipitation and temperature variability from 1991–2010 for southeastern Ohio...... 164

Figure 5.3: Temporal changes in (A) percent carbon and (B) stable carbon isotope composition (13C) in tree rings of white oak compared for control and thin+burn stands. Error bar represents one standard error. Treatments were applied in 2000...... 165

Figure 5.4: Effect of soil moisture gradient (indexed by the integrated moisture gradient, IMI) on the temporal variability of carbon isotope composition (13C) of white oak trees growing in (A) control and (B) thin+burn stands in mixed-oak forests. Error bar represents one standard error. Treatments were applied in 2000. Analyses were performed at = 0.05...... 166

Figure 5.5:Effect of topographic-aspect position (SAP) on the temporal variability of carbon isotope composition (13C) of white oak trees growing in (A) control and (B) thin+burn stands in mixed-oak forests. Error bar represents on standard error. Treatments were applied in 2000. Analyses were performed at = 0.05...... 167

CHAPTER 1: INTRODUCTION

Fire has long been an integral part of the ecology of North American mixed-oak forests. Many dendrochronological and paleoecological studies have established that periodic, low-intensity surface fires were a regular and natural component of Native

American forests, and played a dominant role in ecosystem structure and function (Watts

1980, Van Lear and Waldrop 1989, Abrams 1992, Brose et al. 2001, Hutchinson et al.

2008, Ryan et al. 2013). Research has also indicated that many forest ecosystems throughout the region evolved with fire, with several important species (e.g., white oak) showing strong adaptation to it (Abrams 2003). Furthermore, the periodic, low-intensity fire regimes reduced the forest litter and undergrowth, opened up the stands, and thus forestalled wildfire occurrence or reduced its severity (Murphy et al. 2007).

However, following European settlement (ca. mid-18th to early 19th century), fire intensity, size and frequency changed dramatically with stand-replacing fires occurring more frequently (Abrams 2003). This change engendered a nationwide campaign aimed at detecting, suppressing and preventing fire (Brose et al. 2001). Many researchers posit that this fire suppression was an important underlying cause of the current alteration in species composition of many regional forests, though the roles of increased abundance of browsers and smaller canopy gaps have also been recognized (Nuttle et al. 2013). This compositional shift is characterized by increased density of trees (mainly fire-sensitive and shade-tolerant species), increased fuel load, increased vulnerability to insect and pest attacks, fewer and smaller canopy gaps, as well as increased potential wildfire severity

(Brose et al. 2001, Nowacki and Abrams 2008, Schwilk et al. 2009, Ryan et al. 2013). 15

These widespread community modifications necessitated the re-introduction of prescribed fire back into forest ecosystems.

Consequently, over the past two decades, prescribed fire and thinning treatments have seen an increased application across the coterminous United States as part of an integrated forest management strategy. This increase has been driven largely by recognition of the need to mitigate wildfire effects and to restore regional forests to historical structure and function (Brose et al. 2001). The surge in the use of these treatments has stimulated numerous studies to understand the responses of different ecosystem components, including flora, fauna, soils and fuels (Yaussy 2001). These studies, including a nationwide Fire and Fire Surrogate (FFS) study that spanned a five- year period from 2000 to 2005 and comprised a network of 13 sites, have led to a general consensus among researchers and managers that the treatments may be effective for achieving sustainable forestry (Phillips and Waldrop 2008, Schwilk et al. 2009, Stephens et al. 2009, Brose et al. 2013, McIver et al. 2013). Despite all of these studies, the effects of the treatments on residual tree growth, particularly at the level of the individual tree, and the concomitant sequestration of carbon, remain unclear and poorly documented.

Tree growth is an important process that reflects the health, productivity and overall response of forest ecosystems to environmental change (Fritts 1976, Dobbertin

2005). Tree growth is also critical from the perspectives of wildlife conservation, sawtimber supply and forest productivity. In particular, survival and growth of large residual trees following management disturbance is essential for maintaining short-term overstory conditions in forests (Thorpe et al. 2007, Brudvig et al. 2011). Moreover, tree 16 growth, together with regeneration and survival, drives plant community dynamics and ecosystem function (Fiedler et al. 2010, Gómez-Aparicio et al. 2011). Thus, analysis of tree growth response to prescribed fire and thinning treatments may be critical to increasing our understanding of the broad consequences of these management techniques on forest systems.

Tree growth is governed by multiple biotic and abiotic factors (Fritts 1976,

Kozlowski et al. 1991). One of the most important biotic factors is competition, which continues to receive research attention because of its strong modulating effects on tree growth, and other structural and functional attributes of forest ecosystems (McDonald et al. 2002, Weber et al. 2008). The strong relations between tree growth and tree size and age are also well documented (Macfarlane and Kobe 2006, Coomes and Allen 2007).

Tree growth varies greatly among species due to differences in their structural and physiological adaptations (Johnson et al. 2002). Of the abiotic factors, climate is widely acknowledged to be the most important driver of tree and forest growth (Rubino and

McCarthy 2000, Toledo et al. 2011). Iverson et al. (1997) and Dyer (2009), for example, observed that forest productivity is driven primarily by soil moisture, which is largely determined by the prevailing climate interacting with the biophysical characteristics of the local environment. Understanding how these myriad factors mediate tree growth response to prescribed and thinning manipulations may provide vital information to inform management decisions concerning the applicability and viability of the treatments.

Recent studies suggest that prescribed fire and thinning treatments may influence carbon sequestration through the removal of above-ground biomass of plants and by 17 increasing ambient CO2 concentration (Boerner et al. 2008, Chiang et al. 2008, Chertov et al. 2009, Davis et al. 2009, Hurteau and North 2010). Fire and harvesting disturbances also limit forest carbon storage through long-term changes in site quality (Gough et al.

2007). Hurteau and North (2010), for instance, observed an initial reduction in carbon stocks following fuel reduction treatment but also indicated that recovery is possible over a reasonable time if treatments do not remove large, fire-resistant overstory species.

Davis et al. (2009) found significant responses only in thin and thin+burn treatments. In spite of these findings, the effect of prescribed fire and thinning treatments on carbon sequestration rates in the temperate forest is far from clear.

The variability of stable carbon isotope composition (13C) in tree rings has received increased attention among dendroecologists in recent times as a powerful means to elucidate plant response to environmental change (McCarrol and Loader 2004,

13 McDowell et al. 2006). Tree-ring C reflects the internal CO2 concentration of the leaf at the time of assimilation, and can be used as a reliable proxy for carbon balance or intrinsic water use-efficiency of trees (Farquhar et al. 1982). Like radial growth, stable- carbon isotope composition in tree rings is influenced by several environmental factors including changes in precipitation, temperature, solar radiation, and atmospheric CO2 concentration (Leavitt and Long 1982). This sensitivity to environmental changes has led to the view that tree-ring stable isotopes can supplement ring-width analysis of plant productivity by offering long-term records on the underlying physiological mechanisms

(Leavitt and Long 1982, West et al. 2006, Brooks and Mitchell 2011). Thus, analysis of 18 tree-ring 13C may lend itself to understanding the long-term physiological response of trees to forest management disturbance or climate fluctuations.

Dendrochronology, the technique that uses tree-rings dated to their exact years of formation to analyze spatial and temporal patterns, is one of several methods available for analyzing the long-term variability of tree growth in relation to environmental change

(Fritts 1976). Radial growth analysis has been used extensively to study the influence of disturbance regimes on forest productivity because of its sensitivity to changes in the environment (Biondi 1999, Rubino and McCarthy 2004). This analytical approach is important given the multiple application of prescribed fire, and the short-term nature of the few studies conducted so far on residual tree growth response. Radial growth can provide fine-to-coarse temporal resolution of tree growth and carbon sequestration responses to the prescribed fire and thinning treatments (Skomarkova et al.

2006).

This study was designed to elucidate the long-term growth responses of residual trees to prescribed fire and thinning in mixed-oak forests of the central Appalachians,

Ohio, USA. This study used dendrochronology, GIS modeling, stable isotope analysis, and statistical modeling to address two broad questions: a) How do tree growth and carbon sequestration respond to prescribed fire and thinning treatments over time? and, b)

What environmental or biophysical factors mediate these responses? The overarching hypothesis was that treatment manipulations would cause considerable structural changes in forest stands, leading to increases in tree growth and carbon sequestration rates of 19 residual trees, with responses varying among treatments, species, and over space and time.

Though the study focused mainly on growth responses at the scale of individual trees and populations, stand-level responses to the management disturbances were also inferred by pooling data for all species. Besides the need to include the dominant/codominant and most important overstory species, tree selection was carefully done to represent the major ecological functional diversity, including shade-tolerance, relative growth rates, resistance to fire, and drought/soil moisture stress tolerance. This study does not provide a direct assessment of forest productivity, which is often measured per unit area basis. However, it is important to note that the temporal variability of tree growth provides a good indication of forest productivity, and thus carbon sequestration changes (Chiang et al. 2008, Campioli et al. 2011).

This dissertation is organized into six chapters. Chapter one (this chapter) provides a brief background to the study, including identification of the existing gaps in the ecological literature on the consequences of prescribed fire and thinning management in relation to residual tree growth, justification of the study, questions/hypotheses addressed, and the scope. In chapter two, the long-term patterns of tree growth responses to the treatments are analyzed. Further, the effects of fire severity and species on the tree growth responses have been addressed. Having established the growth responses to the treatments over time and across species, chapter three focuses on how these responses are modulated by tree age, size and competitive status. The landscape in the present study area is topographically complex, and is characterized by a strong gradient of soil moisture 20 demand/stress that exerts profound influence on many ecological processes (Iverson et al.

1997, Dyer 2009). Thus, chapter four examines this topographically-controlled moisture gradient, and its relation to residual tree growth in the managed stands or among treatments. To gain more insight into the physiological mechanisms of tree response to the treatments at the time of carbon assimilation (e.g., water use efficiency), stable- carbon isotope composition of white oak (L.) was analyzed. These results are presented in chapter five. Finally, a synthesis of all the major findings, along with their ecological and management implications, is given in chapter six.

Literature Cited

Abrams, M. D. 1992. Fire and the development of oak forests. BioScience 42: 346–353.

Abrams, M. D. 2003. Where has all the white oak gone? BioScience 53:927–939.

Biondi, F. 1999. Comparing tree-ring chronologies and repeated timber inventories as

forest monitoring tools. Ecological Applications 9:216–227.

Boerner, R. E. J., J. Huang, and S. C. Hart. 2008. Fire, thinning, and the carbon economy:

Effects of fire and fire surrogate treatments on estimated carbon storage and

sequestration rate. Forest Ecology and Management 255:3081–3097.

Brooks, J. R., and A. K. Mitchell. 2011. Interpreting tree responses to thinning and

fertilization using tree-ring stable isotopes. New Phytologist 190:770–782.

Brose, P. H., D. C. Dey, R. J. Phillips, and T. A. Waldrop. 2013. A meta-analysis of the

fire-oak hypothesis: does prescribed burning promote oak reproduction in eastern

North America? Forest Science 59:322–334. 21

Brose, P. H., T. Schuler, D. Van Lear, and J. Berst. 2001. Bringing fire back: the

changing regimes of the Appalachian mixed-oak forests. Journal of Forestry

99:30–35.

Brudvig, L. A., H. M. Blunck, H. Asbjornsen, V. S., Mateos-Remigio, S. A. Wagner, and

J. A. Randall. 2011. Influences of woody encroachment and restoration thinning

on overstory savanna oak tree growth rates. Forest Ecology and Management

262:1409–1416.

Campioli, M., B. Gielen, M. Gockede, D. Papale, O. Bouriaud, and A. Granier. 2011.

Temporal variability of the NPP-GPP ratio at seasonal and interannual time scales

in a temperate beech forest. Biogeosciences 8:2481–2492.

Chertov, O., J. S. Bhatti, A. Komarov, A. Mikhailov, and S. Bykhaovets. 2009. Influence

of climate change, fire and harvest on the carbon dynamics of black spruce in

central Canada. Forest Ecology and Management 257:941–950.

Chiang, J.-M., R. W. McEwan, D. A. Yaussy, and K. J. Brown. 2008. The effects of

prescribed fire and silvicultural thinning on the aboveground carbon stocks and

net primary production of overstory trees in an oak-hickory ecosystem in southern

Ohio. Forest Ecology and Management 255:1584–1594.

Coomes, D. A., and R. B. Allen. 2007. Effects of size, competition and altitude on tree

growth. Journal of Ecology 95:1084–1097.

Davis, S. C., A. E. Hassl, C. J. Scott, M. B. Adams, and R. B. Thomas. 2009. Forest

carbon sequestration changes in response to timber harvest. Forest Ecology and

Management 258:2101–2109. 22

Dobbertin, M. 2005. Tree growth as indicator of tree vitality and of tree reaction to

environmental stress: a review. European Journal of Forest Research 124:319–

333.

Dyer, J. M. 2009. Assessing topographic patterns in moisture use and stress using a water

balance approach. Landscape Ecology 24:391–403.

Farquhar, G. D., M. H. O'Leary, and J. A. Berry. 1982. On the relationship between

carbon isotope discrimination and the intercellular carbon dioxide concentration

in leaves. Australian Journal of Plant Physiology 9:121–137.

Fiedler, C. E., K. L. Metlen, and E. K. Dodson. 2010. Restoration treatment effects on

stand structure, tree growth, and fire hazard in a ponderosa pine/douglas-fir forest

in Montana. Forest Science 56:18–31.

Fritts, H. C. 1976. Tree rings and climate. The Blackburn Press, Caldwell, NJ.

Gómez-Aparicio, L., R. García-Valdés, P. Ruíz-Benito, and M. A. Zavala. 2011.

Disentangling the relative importance of climate, size and competition on tree

growth in Iberian forests: implications for forest management under global

change. Global Change Biology 17:2400–2414.

Gough, C. M., C. S. Vogel, K. H. Harrold, K. George, and P. S. Curtis. 2007. The legacy

of harvest and fire on ecosystem carbon storage in a north temperate forest.

Global Change Biology 13:1935–1949.

Hurteau, M. D., and M. North. 2010. Carbon recovery rates following different wildfire

risk mitigation treatments. Forest Ecology and Management. 260:930–937. 23

Hutchinson, T. F., R. P. Long, R. D. Ford, and E. K. Sutherland. 2008. Fire history and

the establishment of oaks and maples in second-growth forests. Canadian Journal

of Forest Research 38:1184–1198.

Iverson, L. R., M. E. Dale, C. T. Scott, and A. Prasad. 1997. A GIS-derived integrated

moisture index to predict forest composition and productivity of Ohio forests

(USA). Landscape Ecology 12:331–348.

Johnson, P. S., S. R. Shifley, and R. Rogers. 2002. The Ecology and Silviculture of Oaks.

CABI Publishing, Wallingford, UK.

Kozlowski, T. T., P. J. Kramer, and S. G. Pallardy. 1991. The physiological ecology of

woody plants. Academic Press, Inc., San Diego, CA.

Leavitt, S. W., and A. Long. 1982. Stable carbon isotopes as a potential supplemental

tool in dendrochronology. Tree-Ring Bulletin 42:49–55.

Macfarlane, D. W., and R. K. Kobe. 2006. Selecting models for capturing tree-size

effects on growth – resource relationships. Canadian Journal of Forest Research

36:1695–1704.

McCarrol, D., and N. J. Loader. 2004. Stable isotopes in tree rings. Quaternary Science

Reviews 23:771–801.

McDonald, E. P., E. L. Kruger, D. E. Riemenschneider, and J. G. Isebrands. 2002.

Competitive status influences tree-growth responses to elevated CO2 and O3 in

aggrading aspen stands. Functional Ecology 16:792–801. 24

McDowell, N. G., H. D. Adams, J. D. Bailey, M. Hess, and T. E. Kolb. 2006.

Homeostatic maintenance of ponderosa pine gas exchange in response to stand

density changes. Ecological Applications 16:1164–1182.

McIver, J. D., S. L. Stephens, J. K. Agee, J. Barbour, R. E. J. Boerner, C. B. Edminster,

K. L. Erickson, K. L. Farris, C. J. Fettig, C. E. Fiedler, S. Haase, S. C. Hart, J. E.

Keeley, E. E. Knapp, J. F. Lehmkuhl, J. J. Moghaddas, W. Ostrosina, K. W.

Outcalt, D. W. Schwilk, C. N. Skinner, T. A. Waldrop, C. P. Weatherspoon, D. A.

Yaussy, A. Youngblood, and S. Zack. 2013. Ecological effects of alternative fuel-

reduction treatments: highlights of the national fire and fire surrogate study (FFS).

International Journal of Wildland Fire 22:63–82.

Murphy, A., J. Abrams, T. Daniel, and V. Yazzie. 2007. Living among frequent-fire

forests: human history and cultural perspectives. Ecology and Society 12:17–31.

Nowacki, G. J., and M. D. Abrams. 2008. The demise of fire and mesophication of

forests in the eastern United States. BioScience 58:123–138.

Nuttle, T., A. A. Royo, M. B. Adams, and W. P. Carson. 2013. Historic disturbance

regimes promote tree diversity only under low browsing regimes in eastern

deciduous forest. Ecological Monographs 83:3–17.

Phillips, R. L., and T. A. Waldrop. 2008. Changes in vegetation structure and

composition in response to fuel reduction treatments in the South Carolina

Piedmont. Forest Ecology and Management 255:3107–3116. 25

Rubino, D. L., and B. C. McCarthy. 2000. Dendroclimatological analysis of white oak

( L., Fagaceae) from an old-growth forest of southeastern Ohio,

USA. Journal of the Torrey Botanical Society127:240–250.

Rubino, D. L., and B. C. McCarthy. 2004. Comparative analysis of dendroecological

methods used to assess disturbance events. Dendrochronologia 21:97–115.

Ryan, K. C., E. E. Knapp, and J. M. Varner. 2013. Prescribed fire in North American

forests and woodlands: history, current practice and challenges. Frontiers in

Ecology and Environment 11:e15–e24.

Schwilk, D. W., J. E. Keeley, E. E. Knapp, J. McIver, J. D. Bailey, C. J. Fettig, C. E.

Fiedler, R. J. Harrod, J. J. Moghaddas, K. W. Outcalt, C. N. Skinner, S. L.

Stephens, T. A. Waldrop, D. A. Yaussy, and A. Youngblood. 2009. The national

fire and fire surrogate study: effects of fuel reduction methods on forest

vegetation structure and fuels. Ecological Applications 19:285–304.

Skomarkova, M. V., E. A. Vagavov, M. Mund, A. Knohl, P. Linke, A. Boerner, and E.-

D. Schulze. 2006. Inter-annual and seasonal variability of radial growth, wood

density and carbon isotope ratios in tree rings of beech () growing

in Germany and Italy. Trees 20:571–586.

Stephens, S. L., J. J. Moghaddas, C. Edminster, C. E. Fiedler, S. Haase, M. Harrington, J.

E. Keeley, E. E. Knapp, J. D. McIver, K. Metlen, C. N. Skinner, and A.

Youngblood. 2009. Fire treatment effects on vegetation structure, fuels, and

potential fire severity in western U.S. forests. Ecological Applications 19:305–

320. 26

Thorpe, H. C., S. C. Thomas, and J. P. Caspersen. 2007. Residual-tree growth responses

to partial stand harvest in the black spruce () boreal forest.

Canadian Journal of Forest Research 37:1563–1571.

Toledo, M., L. Poorter, M. Peña-Claros, A. Alarcón, J. Balcázar, C. Leaño, J. C. Licona,

O. Llanque, V. Vroomans, P. Zuidema, and F. Bongers. 2011. Climate is a

stronger driver of tree and forest growth rates than soil disturbance. Journal of

Ecology 99:254–264.

Van Lear, D. H., and T. A. Waldrop. 1989. History, uses and effects of fire in the

Appalachians. USDA Forest Service General Report SE-54, Ashville, NC.

Watts, W. A. 1980. Late Quaternary vegetation of the central Appalachian and the New

Jersey coastal plain. Ecological Monograph 49:427–469.

Weber, P., H. Bugmann, P. Fonti, and A., Rigling. 2008. Using a retrospective dynamic

competition index to reconstruct forest succession. Forest Ecology and

Management 254:96–106.

West, J. B., G. J. Bowen, T. E. Cerling, and J. R. Ehleringer. 2006. Stable isotopes as one

of nature’s ecological recorders. Trends in Ecology and Evolution 21:408– 414.

Yaussy, D. A. 2001. Study plan and establishment report: consequences of fire and fire

surrogate treatments; the Ohio Hills site.

Http://www.fs.fed.us/ne/delaware/4153/ffs/ohio_studypland.pdf

CHAPTER 2: LONG-TERM EFFECTS OF PRESCRIBED FIRE AND THINNING ON

RESIDUAL TREE GROWTH IN MIXED-OAK FORESTS OF SOUTHERN OHIO

Abstract

Long-term (10-yrs) growth responses of residual trees to prescribed fire and thinning were evaluated using standard dendrochronological protocols to understand the broader effects of the treatments on mixed-oak forest ecosystems in southern Ohio.

permanent plots distributed evenly across four treatments (control, thin, thin+burn, burn) indicated substantial increase in tree basal area increment (BAI) following the treatment.

Post-treatment mean BAI of trees from the three active treatments ranged from 20.52-

23.55 cm2 yr-1 compared with pre-treatment values of 16.86-17.07 cm2 yr-1. BAI rates in the control plots averaged 15.13 cm2 yr-1 and 16.33 cm2 yr-1 respectively for pre-and post- treatment periods. Mechanical treatments were more effective than prescribed fire at enhancing BAI of trees. However, basal area growth depended to some degree on the severity of prescribed fire. Analysis of percent BAI change revealed a temporal trend with moderate to major growth releases during the first 5-yr post-treatment period, and a slight attenuation thereafter, suggesting the need for periodic application of treatments to sustain growth over a longer timescale. Growth responses varied greatly among species, with yellow-poplar and hickories exhibiting the highest and lowest post-treatment BAI rates of 31.11 and 15.71 cm2 yr-1, respectively. Given their variable growth responses, integrating residual trees into current monitoring programs may help in elucidating the consequences of prescribed fire and thinning on forest dynamics and development. 28

Introduction

Prescribed fire and thinning treatments are commonly used as part of an integrated forest management strategy to mitigate potential wildfire effects and to restore regional forests to historical structure and function (Schwilk et al. 2009, Stephens et al.

2009). In the mixed-oak forests of eastern US, these management strategies are increasingly used to alter stand structure and promote oak regeneration (Abrams and

Downs 1990, Albrecht and McCarthy 2006, Hutchinson et al. 2012). Numerous ecological studies, spurred by the growing interest in and application of these restoration methods, have led to consideration of prescribed fire and mechanical thinning as sustainable forest management options (Boer et al. 2009, Brose et al. 2013). However, how these treatments, particularly prescribed burning and the combination of prescribed burning and thinning, affect residual tree growth over the long term remains unclear and poorly documented. Knowledge of the long-term responses of residual trees, at both population and stand levels, is critical not only to understand the broader consequences of the treatments on forest ecosystems, but also to refine management aims and practices.

Tree growth is an important component of plant community dynamics (Gómez-

Aparicio et al. 2011), and a variable process that reflects the function, health, energy flux, and overall response of forests to environmental change (Dobbertin 2005). In particular, growth of large trees provides a good indication of forest productivity, and plays a major role in global carbon dynamics (Xu et al. 2012). Large trees are also important for the purposes of sawtimber supply and wildlife conservation. Growth of residual trees following thinning is important for maintaining overstory conditions in the short term, 29 and for evaluating the viability of prescribed fire and thinning treatments (Brudvig et al.

2011, Thorpe et al. 2007).

It has long been established that thinning increases diameter growth of trees through reduction in stand density (Hilt 1979, Johnson et al. 2002). Forest fires are also known to influence tree growth, though this effect often varies with species (Brose et al.

2006, Hutchinson et al. 2005). Despite the progress made so far in understanding the ecological effects of these forest management strategies, many uncertainties remain regarding the long-term growth responses of residual trees. For example, while several studies have documented significant but transient increases in tree growth and productivity associated with these management operations (Boerner et al. 2008, Chiang et al. 2008, Hurteau and North 2010, Sala et al. 2005), reports of growth pulses lasting up to

20 years following the treatments are not uncommon (Thorpe et al. 2007). Furthermore, although different species respond differently to forest stand disturbances due to variations in their structural and physiological adaptations (Johnson et al. 2002, Plauborg

2004), nearly all previous studies investigating the effects of fire and thinning disturbances have considered only stand-level responses.

Analyzing growth responses of surviving trees over a longer temporal scale can be an effective means of assessing the sustainability of prescribed fire and thinning management. This is important given that tree growth responses to environmental perturbations may occur over a long time-scale. Indeed, research indicates that prescribed fire and thinning effects can take many years or multiple applications to be fully realized

(Laughlin et al. 2008, McEwan et al. 2011). However, previous works have mostly been 30 of short-term duration (typically 1–4 years). One approach to assessing the variability of tree growth in response to environmental disturbances is through dendrochronology, the use of tree-rings dated to their exact years of formation, to analyze spatial and temporal patterns (Fritts 1976). This analytical tool is useful to ecologists and managers interested in long-term growth and development of forest stands for several reasons. Tree-rings are sensitive to changes in the environment and can produce fine to coarse resolution of past growth variability of trees associated with various environmental stressors (Biondi 1999,

Rubino and McCarthy 2004). Radial growth analysis can also overcome the limitation of inadequate information on longer-term before-treatment conditions (Hutchinson et al.

2008).

The purpose of this study was to evaluate the long-term (10-year) effects of prescribed fire and thinning treatments on residual tree growth in the mixed-oak forest of southern Ohio using tree-ring analysis. Specific questions explored were: () how does growth of residual trees respond to prescribed fire and thinning treatments, and in combination? () do changes in tree growth vary with time since treatment? ()are there interspecific variations in tree growth responses to prescribed fire and thinning treatments? It was hypothesized that tree growth rates would increase considerably following the treatments, and would differ among the three active treatments (mechanical thinning, prescribed fire, and their combination) and with respect to the control due to the differential treatment intensities and subsequent tree mortality in the case of prescribed fire. Tree growth response was also predicted to vary with fire severity, time since initiation of treatments, and among individual species. 31

Methods

study areas are replicate blocks of the Ohio Hills site of the national Fire and Fire

Surrogate (FFS) study (Waldrop et al. 2008). The Ohio Hills site lies within the unglaciated Allegheny Plateau physiographic region and is characterized by an heterogeneous landscape with ridges, hills and valleys (Hutchinson et al. 2005). Elevation ranges from 200 to 300 m. Soils are acidic, and consist of silt and sandy loams derived primarily from sandstone, siltstone and shale (Boerner et al. 2003). Annual precipitation and temperature average 1024 mm and 11.3 oC respectively. The vegetation is classified as upland mixed-oak hardwood forests (Iverson et al. 2008).

At each replicate site, there are four treatment units of approximately 20 ha each, and ten 20 × 50 m (0.1 ha) permanent plots distributed within each treatment unit for a total of 80 plots. The four units are untreated control (control), mechanical thinning

(thin), prescribed fire (burn), and mechanical thinning followed by prescribed fire

(thin+burn; Figure 2.1). Mechanical thinning, conducted between September 2000 and

April 2001, was essentially harvesting of midstory trees 15-30 cm diameter at breast height (DBH) that favored retention of oaks and hickories (Iverson et al. 2008). This treatment reduced basal area from 27.40 to18.68 m2 ha-1 and 28.94 to 22.56 m2 ha-1 for the thin and thin+burn units respectively at REMA. At the Zaleski State Forest, basal area 32 was reduced from 27.66 to 19.17 m2 ha-1 for the thin units and 25.47 to 18.38 m2 ha-1 for the thin+burn unit. Prescribed burning was first carried out in the spring of 2001 and repeated in 2005 and 2010. The prescribed fire treatment in 2001 was generally of low intensity (1-2 m flame length) but treatment in 2005 was more intensified resulting in variable flame lengths (1-4 m) across the landscape (Iverson et al. 2008, Waldrop et al.

2008).

Between June and December, 2011, increment cores of trees representing five of the most common overstory species in the study area were collected using an increment borer. The tree species sampled are white oak (), chestnut oak (), black oak (), hickories (spp. primarily ,and

) and yellow-poplar () (Iverson et al. 2008). To reduce variation in growth due to factors other than the ones being tested, healthy-looking codominant trees ( 25 cm DBH) of less than 50 % branch or twig mortality or foliage discoloration, dieback and defoliation (Yaussy et al. 2004) were selected for sampling using pre-existing FFS overstory data as a guide. Two increment cores were taken from each tree at breast height, and at roughly 180o from each other to account for circumferential variations in ring-widths, and in line with standard dendrochronological practice. The initial sampling goal was to obtain at least two trees of each species from each of the 10 plots within a treatment unit. If a species was missing in a plot, a substitute tree was selected within 20 m buffer outside the plot. In spite of this sampling effort, 33 some species were still missing in some sampling areas. Cores were taken parallel to the slope contours to avoid reaction wood (Stokes and Smiley 1968).

Growth responses of trees to fire severity, a measure of prescribed fire effect on the forest ecosystem (Keeley 2009), were assessed by estimating the extent of bark scorching and/or scarring on trees across the burn and the thin+burn plots. Based on these estimates, trees were assigned to one of three fire severity classes: low (no visible chars or scars), moderate (charred but no visible scars), and high (charred and scarred).

In the laboratory, increment cores were glued to wooden core mounts and air- dried for at least 24 hour before sanding. Cores were sanded using a belt sander with increasingly finer grades of sandpaper (100 to 600 grit), and micro-finished with 30, 15 and 9 micron Microfilm Sheets to reveal the ring structure (Speer 2010) (Figure 2.2).

Cross-dating was done using skeleton-plotting and the list method (Stokes and Smiley

1968, Yamaguchi 1991).

Ring widths were measured to the nearest 0.01 mm using a Velmex sliding stage with an Acu-Rite encoder, a Metronics digital readout (Velmex, Boomfield, NY) and the

Measure J2X program (Voortech Consulting Holderness, NH). Cross-dating and measurement accuracies were verified using the quality control program, COFECHA

(Grissino-Mayer 2001). Measurement and cross-dating errors detected from the

COFECHA output were corrected with the help of the program EDRM (EDit Ring

Measurement; Holmes 1999). Composite chronologies (of both raw and detrended ring- width series) for each of the five studied species were developed using the “dplR”

(Dendrochronology Program Library) package in R (Bunn et al. 2012). Detrending was 34 done using a 100-year smoothing spline with 50% frequency response. A 100-year spline was used as it retains much of the signals in ring structure due to disturbance, competition and climate in these closed-canopy forests (Cook 1985). Standard dendrochronological statistics such as the number of dated series, mean ring width, mean length of series and series inter-correlation were obtained from the raw ring widths as part of the COFECHA output. In addition, first order autocorrelation and Gini coefficient were computed from the standardized (detrended) ring-width indices using the “dplR” function . First order autocorrelation represents the year-to-year variability within a ring-width series while the Gini coefficient is an all-possible-lags measure of variability in ring width

(Biondi and Qeadan 2008a).

Radial growth of individual trees was measured by converting ring-width measurement into basal area increment (BAI), the net increase in the total cross-sectional stem area of tree. BAI is the preferred metric of growth for many dendrochronologists because it has been found to provide a more accurate approximation of the annual tree growth than ring width or diameter increment (Visser 1995, Biondi and Qeadan 2008b).

BAIs were computed using the function in “dplR”, which calculates the ring area from the bark to the pith (Bunn et al. 2012). Computations were based on diameter inside bark of trees, derived by first estimating bark thickness of the species using available regression equations (Hilt et al. 1983). BAIs of the two cores extracted from a tree were averaged to obtain a single value per year for that tree. Thus, the individual tree constituted the sampling unit. To filter out high frequency variability in annual tree 35 growth typically associated with climate, 5-yr periodic annual BAIs were computed by averaging the increment values within the 5-yr growth interval. For consistency, analysis was focused on a 20-year growth interval (1991–2010), which spanned two pre-treatment

(1991–1995, 1996–2000) and two post-treatment (2001–2005, 2006–2010) periods.

Proportional change in BAI associated with the prescribed fire and thinning disturbances was computed using a modification of the running mean method (Nowacki and Abrams 1997), by comparing sequential 5-yr periodic annual BAIs instead of 10-year means of ring-widths:

% , % = × 100 where is the periodic annual BAI of period and is the periodic annual BAI of the preceding growth period. The running mean approach was employed because it allows for the detection of “abrupt and sustained radial growth increases” associated with canopy disturbances (Nowacki and Abrams 1997).

All analyses were performed using the R programming language (R Development

Core Team 2012). The information-theoretic approach (Burnham and Anderson 2002) was used to determine the relative effects of the treatment types on BAI and percentage change in BAI observed during the two post-treatment periods. This analysis required three two-level treatment factors, namely, MECH (presence or absence of mechanical treatments), BURN (presence or absence of prescribed burning) and FUEL (presence of any treatment other than the control), which were created following the descriptions of

Schwilk et al. (2009), and in addition to treating each treatment unit as a separate and independent effect. Different combinations of these three derived treatment factors and 36 time since treatment initiation (i.e., two 5-yr post-treatment periods) yielded a set of nine candidate models, including:

i) BAI/ % GC = (Site); this model tests the random effect of site on differences

in the response variables.

ii) BAI/ % GC = FUEL; this model tests the effects of prescribed fire and

mechanical thinning treatments on BAI or percent change in BAI, assuming

no appreciable difference in the effects of the two treatments.

iii) BAI/ % GC = MECH; this tests the effect of mechanical thinning on

response variable, and assumes that prescribed fire has little effect.

iv) BAI/ % GC = BURN; this evaluates the effect of prescribed fire on BAI or

percent change in BAI, with the assumption that mechanical thinning has

little effect.

v) BAI/ % GC = MECH + BURN; this two factor model tests the distinct but

additive effects of mechanical thinning and prescribed fire on the response

variable.

Other models included in our analysis test whether time since treatment initiation

(used either as an independent fixed term or as an interactive term with treatment) has any effect on BAI or percent change in BAI (Appendices 1A and 1B). All candidate models were fitted with the function in the package “lme4” (Bates 2005), and included site and individual trees as random nested effects. This linear mixed-effects

(Pinheiro and Bates 2000) analytical approach was considered appropriate because of the longitudinal and unbalanced (due to missing values) nature of the data. Percent changes 37 in BAI were log-transformed to stabilize the variance. Models were compared using the

Akaike information criteria corrected for small samples (AICc), the change in AICc

(deltaAICc) and the Akaike weight (AICc weight or ) to identify the most parsimonious among them (Appendices 1A and 1B). Akaike weight is a second-order information- theoretic metric, which represents the probability of a selected model being the best among a set of competing models. Parameters were estimated using restricted maximum likelihood (REML), and model validation was performed using plots of residuals against each of the explanatory variables (Zuur et al. 2009). Data were analyzed separately for individual species and also pooled for general effects across species.

Analysis of variance (ANOVA) was used to investigate the variations in post- treatment BAI responses across fire severity classes. Since both mechanical thinning and prescribed burning are likely to influence growth responses of trees from the thin+burn treatment, data from this treatment unit were analyzed separately from those of the burn- only unit. Within each treatment unit, data were analyzed by species, and also pooled for the general effects across treatment. Post-hoc multiple comparisons (Tukey Honestly

Significant Difference, HSD) tests were used to evaluate differences in BAI among fire severity classes.

Results

A total of 696 increment cores (348 trees, five species) were analyzed across the four treatment units in the two sites. The species distribution was as follows: white oak

(94 trees), black oak (82 trees), chestnut oak (81 trees), yellow-poplar (47 trees) and 38 hickories (44 trees) (Table 2.1). Of these, hickories had the longest chronology of 264 years (1744–2011), although only a few trees were 200 years or older. Yellow-poplar had the shortest chronology of 147 years (1865–2011), whilst chronologies for the oaks ranged from 155 (black oak) to 162 (white oak) years. Thus, on average, series length was greater for hickories and white oak (103 years in both) than the rest of the species.

Conversely, mean ring width was largest for yellow-poplar (1.99 mm) and smallest for hickories (1.20 mm). Series inter-correlation, a measure of how well trees pick up the common signal at a site (Speer 2010), was greater than 0.62 for all species. First-order autocorrelation showed greater year-to-year variability in ring width among the hickories and yellow-poplar, and least among the oaks. Similarly, the Gini-coefficient suggested greater diversity in ring widths for yellow-poplar (0.347) and hickories (0.201) than in the oaks. However, raw and standardized (detrended) ring-width chronologies for the individual species (Figure 2.3) revealed considerable similarity in pre-treatment growth patterns among the species with concurrent releases (e.g., 1980) and suppressions (e.g.,

1954, 1966, 1988).

Radial growth of residual trees (all species combined) was impacted positively by the prescribed fire and mechanical thinning treatments (Figure 2.4A). Post-treatment mean BAIs in the three active treatments (thin-only, burn-only and thin+burn) ranged from 20.52 to 23.55 cm yr-1 compared with pre-treatment values of 16.86–17.07 cm2 yr-1.

Contrarily, BAI in the control unit did not change much over time, while pre-treatment mean BAIs were comparable among all treatments. After the initial pulse, growth rates 39 attenuated slightly across all treatments, although the thin-only trees continued to exhibit accelerated growth during the second 5-yr post-treatment period (2006–2011). Clearly, the mechanical thinning enhanced BAI to a greater degree than the prescribed fire, and was more effective when combined with prescribed fire. Our information-theoretic analysis indicated strong interactive effects of FUEL, time and species on the overall variance in post-treatment BAI of trees (AICc weight = 1.00; Table 2.2; Appendix 1A).

Species varied greatly in their growth responses to the prescribed fire and mechanical thinning treatments (Figure 2.4 A–F). Among the oaks, black oak exhibited the highest post-treatment BAI rate of 17.27–26.0 cm2 yr-1 (Figure 2.4B). Black oaks exhibited incremental growth after the treatments, with the thin and the thin+burn treatments producing consistently higher growth rates than the prescribed burn and the control units. Consequently, comparison of the competing models resulted in the MECH

× Time (period of growth) as the most parsimonious model for the variations in black oak growth (AICc weight = 0.81; Table 2.2). Chestnut oak growth decreased between the first and second pre-treatment periods, and subsequently increased continually across the three active treatments (Figure 2.4C). In contrast, growth in the control trees generally did not deviate appreciably before and after the treatment application. Differences in BAI of chestnut oak was best explained by FUEL × Time, although this model was less consistent (AICc weight = 0.45; Table 2.2). Unlike the other species, white oak growth in the burn-only plot declined substantially between the second pre-treatment and the first post-treatment periods, but barely changed for the other treatments (Figure 2.4D).

Subsequently, BAI responded more positively to the treatments. Interestingly, the 40 model which tests the distinct but additive effects of prescribed burning, mechanical thinning and time (i.e., MECH + BURN + Time) had the best explanatory power for the variance in white oak BAI (AICc weight = 0.63; Table 2.2). There was, however, a strong competition from the FUEL × Time model (AICc weight = 0.23; Appendix 1A).

Like the oaks, hickory and yellow-poplar exhibited strong temporal growth patterns across the treatments. BAI of hickory did not change much between the first and second pre-treatment periods, although it decreased slightly across all treatments (Figure

2.4E). However, BAI increased appreciably during the first post-treatment period (2001–

2005) among trees growing in the three active treatments relative to the control trees.

Subsequently, growth rates either weakened or decline slightly across treatments. Despite the increase in growth rate following the treatments, their effects appeared to be statistically indistinguishable; BAI of hickory was best explained by FUEL (AICc weight

= 0.40; Table 2.2) but there was strong competition from other models (Appendix 1A).

The temporal trend of growth exhibited by yellow-poplar differed markedly from those of the other species (Figure 2.4F). Its rate of growth increased rapidly from 1991 to

2000, peaked between 2001 and 2005, and declined sharply (particularly in the thin+burn treatment) in the subsequent years. Obviously, post-treatment BAI of yellow-poplar trees responded more positively to the mechanical treatments than the prescribed burning treatments. As a result, MECH × Time offered the best explanation for the differences in

BAI of yellow-poplar (AICc weight = 0.92; Table 2.2). Altogether, post-treatment mean

BAI (excluding the controls) was highest for yellow-poplar (31.11cm2 yr-1) and lowest for hickories (15.71 cm2 yr-1). 41

Comparison of the percent growth changes over time indicated moderate to major growth releases during the first 5-yr post treatment period (Figure 2.5A). Changes in BAI of thin-only (~ 40%) and thin+burn (~ 55%) trees during this period exceeded the minimum threshold of 25% commonly used by dendroecologists to identify growth responses to canopy disturbances (Nowacki and Abrams 1997). Growth rate continued to increase 6–10 years after the treatment, but with less rapidity than was observed in the preceding interval. Growth change in the control, although positive, only averaged

12.67% during this period. Variations in the percent BAI change were mostly explained by FUEL × Time × Species (AICc weight = 1.00; Table 2.2). Percent change in BAI also differed among species, with black oak and white oak recording less than 20% change across treatments and over time (Figures 2.5B and D). Proportional growth changes in black oak were best predicted by the differences between the two sites (AICc weight =

0.53), whilst the interactive effects of mechanical treatments and time mostly explained changes in growth of white oak (AICc weight = 0.89). Chestnut oak trees growing in the thin-only and the thin+burn treatments, respectively, had 44.70% and 47.58% growth changes between 2001 and 2005 (Figure 2.5C). The burn-only and the control trees did not result in any major changes in BAI. As expected, growth changes in chestnut oak were explained by MECH, albeit with less support (AICc weight = 0.39). Changes in BAI of hickories and yellow-poplar were best explained by MECH × Time (with AICc weight of 0.52 and 0.99, respectively). Growth changes in hickories from the thin-only (76%) and the thin+burn (86%) treatments were quite dramatic during the 2001–2005 growth period (Figure 2.5E), considering the negative trend prior to this period. With percentage 42 mean growth change of 163% and 213%, respectively in the thin-only and thin+burn treatments, yellow-poplar was by far the most responsive to the treatments (Figure 2.5F).

Interestingly, this species also exhibited the most rapid growth decline during the second

5-yr post-treatment interval (2006–2010), with BAI rates decreasing by ~ 30–50% than in preceding growth interval across all treatments.

As predicted, prescribed fire severity had a considerable effect on tree growth, especially in the burn-only treatment (Figure 2.6). With all species combined, trees from the burn-only treatment that were most severely impacted by prescribed fire exhibited significantly greater BAI response compared to those that received low to moderate prescribed burning ( = 11.25, < 0.001). However, this effect was not discernible in the thin+burn treatment ( = 1.38, = 0.256). The effect of fire severity on tree growth was also clearly visible at the species level, particularly in black oak. BAI rates of black oak trees increased with increasing severity of prescribed burning in both the burn- only ( = 8.48, < 0.001) and the thin+burn ( = 5.35, = 0.009) plots. Growth response of white oak to prescribed fire severity was less consistent. White oak trees from the burn-only treatment, which were moderately impacted by prescribed burning, exhibited much lower ( = 3.56, = 0.038) BAI rates compared to those that experienced low- and high-severity burns. However, white oak trees growing in the thin+burn treatments did not respond significantly to fire severity ( = 0.963, >

0.389), although those which experienced less severe burning recorded slightly higher 43

BAI rates. No significant relations were found between basal area growth and fire severity for the three remaining species (data not shown).

Discussion

Tree growth, along with regeneration and survival, drives forest ecosystem composition and dynamics, and is useful for understanding the impacts of prescribed fire and mechanical thinning management on forest ecosystems (Fiedler et al. 2010, Gómez-

Aparicio et al. 2011). The purpose of this study was to understand how this vital process varies over the long-term following prescribed fire and thinning disturbances in the mixed-oak forests. Results generally showed substantial increase in basal area growth of residual trees following the prescribed fire and thinning treatments. The enhanced growth is consistent with the observed accelerated growth of trees associated with reduction in overstory tree density (Brudvig et al. 2011, Johnson et al. 2002). Removal of neighboring trees through prescribed burning and/or its surrogates increases the available growing space, allowing the remaining trees access to more resources (e.g., light, water, nutrients), and subsequently elevating their photosynthetic and diameter growth rates. Comparing growth patterns before and after the treatments, and between the three active treatments and the control (the so-called before-after, control-treatment design) allowed a better understanding of the effects of the treatments on growth of surviving trees.

The view that mechanical treatments are more effective than prescribed burning at promoting desirable changes in forest composition and structure (Schwilk et al. 2009) is supported by results of this study, which indicated a generally high response rate of tree 44 growth to mechanical thinning among the five species analyzed. Results of the multimodel analysis further demonstrate the relatively high predictive power of mechanical treatments for BAI and proportional change in BAI across individual tree species compared to prescribed burning. For example, in three of the five species analyzed (i.e., black oak, white oak and yellow-poplar), mechanical treatments, either acting independently or interacting with time since treatment initiation, best explained the variance in BAI. For the two remaining species (chestnut oak and hickory), the effects of prescribed burning and thinning on differences in BAI were indistinguishable. Analysis of the percentage growth change even depicted a much stronger effect of mechanical treatment. Mechanical thinning reduced stand basal area at the studied sites to a greater extent than was achieved with prescribed fire (Chiang et al. 2008), which probably explains the stronger positive impact on basal area growth of surviving trees. Notably, this effect was maximized by the thin+burn treatment, indicating that the response varied with the extent of thinning disturbance, either through mechanical means or fire-caused mortality.

Contrary to the generally uniform application of mechanical treatment across the landscape, prescribed burning at the Ohio Hills sites was patchy, with a significant portion of the site experiencing low-intensity burns (Brose et al. 2006). Previous local and regional studies suggest that prescribed burning, typically conducted during the dormant period in spring when the risk of extreme fire behavior is minimal, could barely cause any injury to the large trees (Boerner 2005, Brose et al. 2006, Waldrop et al. 2008).

These factors may account for the moderate effect of prescribed fire on basal area growth 45 found in this study. However, as demonstrated here, more severe fires may lead to positive growth responses in some species (e.g., black oak, white oak). It is equally important to note that prescribed fire is better at reducing surface fuel load and risk of wildfires, and in controlling regeneration of fire-sensitive species (Boer et al. 2009;

Hutchinson et al. 2005), though its necessity for fuel reduction in moist eastern deciduous forests is still questionable. Whether the repeated burns had any significant effect on residual tree growth is unclear from the current results. However, as noted by several researchers (e.g., Brose et al. 2006, Iverson et al. 2008, Waldrop et al. 2008), multiple burns can cause additional tree mortality, leading to greater canopy opening.

Despite the slight attenuation in BAI following the initial pulse between 2001 and

2005, trees continued to show incremental growth 6–10 year after the initiation of the treatments, implying that maximum response had yet to be attained, at least in some species. This trend, coupled with the higher growth rates in the active treatment units, suggests continuing recovery of the initial loss of basal area or carbon as a result of the treatments (Boerner et al. 2008, Chiang et al. 2008). Thorpe et al. (2007) investigated the response of black spruce ( (Mill.) BSP) to different silvicultural treatments in northeastern Ontario, and observed a similar temporal growth trend which peaked 8–9 years after harvesting, and argued that the time-scale of the response depends upon species and treatment. It is hypothesized that this growth attenuation 6–10 year after the treatments may be a direct consequence of increased competition for limiting resources.

Over time, canopy openings created by mechanical thinning and/or prescribed burning 46 disturbances gradually close, limiting supply of vital resources (e.g., light) and increasing competition among overstory plants. However, with repeated burns, canopy openings may remain or new ones may be created, depending on fire intensity, to sustain residual tree growth.

Interspecific variations in tree growth response to prescribed fire and thinning were noticeable in this study, reflecting differences in the inherent ability of species to respond to disturbance (Johnson et al. 2002, Plauborg 2004). Similar interspecific variations have been reported in several previous studies that investigated the effects of prescribed fire and thinning management on forest ecosystems (e.g., Waldrop et al.

2008). Yellow-poplar, for instance, is long established to be an aggressive gap-phase species that has the capacity to outcompete oaks when light is not limiting (Trimble

1967). Thus, the strong positive growth response of yellow-poplar within the first five years after the treatment seems ecologically reasonable; this species simply took advantage of the gap created by the mechanical operations to grow, and consequently contributed immensely to the initial growth surge following the treatment as depicted by the general growth curves. The subsequent sharp decline is likely related to reduction in competitiveness of species as light and other essential resources become more limiting over time (Tilman 1985). By comparison, hickories grow somewhat slowly, while oaks are reported to be quite conservative in their growth responses to disturbances despite differences within the group (Johnson et al. 2002, Nowacki and Abrams 1997).

Coincidence of prescribed burning with initiation of oaks’ stem growth, as postulated by 47

Chiang et al. (2008), may also have explanatory power for their intermediate growth position relative to yellow-poplar and hickories.

Differences in growth responses of oaks to the treatments is particularly interesting, in light of the tendency of researchers to combine the oaks in analyses (e.g.,

Chiang et al. 2008, Waldrop et al. 2008). Black oak is more light-demanding, and on average, grows faster than chestnut oak and white oak (Johnson et al. 2002). Thus, the relatively higher growth rate of black oak is biologically meaningful, although its unresponsiveness to the treatment is surprising. Nonetheless, the strong positive relation between black oak growth and fire severity suggests that growth of this species may be significantly enhanced by more aggressive fire prescriptions. Interestingly, the percent change in BAI responded strongly to the random effect of site, probably driven by variations in prescribed fire intensity across the landscape. Conversely, the negative growth trend of white oak might be related to the interactive effects of fire and insect attack. In 2002, a widespread defoliation of white oak and chestnut oak by several insects occurred at the study areas killing most of the white oaks (T. Hutchinson, personal communication). Delayed response to sublethal fires (Schwilk et al. 2009) coupled with insect attack might be responsible for the white oak growth decline observed during the first post-treatment growth period. Chestnut oak growth represents a slight departure from the trends observed for white oak and black oak, because it responded well to both fire and mechanical thinning; information-theoretic analysis indicated equally strong support for models comprising these two independent variables, either acting individually or in concert, and with time. Chestnut oak, like the other oak species, is shade intolerant 48

(Abrams and McCay 1996) and would react to canopy disturbances caused by mechanical thinning. This characteristic, combined with its ability to grow optimally on more xeric sites where prescribed fire tends to be more severe (cf. Albrecht and

McCarthy 2006), may explain the fairly even response to both mechanical thinning and prescribed fire.

Conclusions and Management Implications

A major objective of prescribed fire and thinning management in oak-dominated forests of eastern North America is to alter stand structure and promote oak regeneration

(Albrecht and McCarthy 2006, Iverson et al. 2008, Hutchinson et al. 2012). While pursuing this objective is vital in the long-term for maintaining oak dominance in the mixed-mesophytic hardwood forests, growth of residual trees is critical for maintaining overstory conditions at least in the short-term, as argued by Brudvig et al. (2011). Thorpe et al. (2007) further emphasized the importance of residual trees by stating that prescribed fire and thinning treatments can only be considered viable if they lead to enhanced growth of such large trees.

The present results have demonstrated that prescribed fire and thinning treatments can influence forest dynamics by enhancing growth of surviving trees. The combination of prescribed fire and thinning was particularly effective due to its ability to create more canopy openings. Thus, if the goal of management is to quickly increase growth and productivity, mechanical thinning may provide the best option. Tree growth exhibited strong temporal trend with the greatest proportional change occurring during the first 5-yr post-treatment period, and a slight dip thereafter. The continued positive growth of trees 49 has significant implication for carbon sequestration, which is important given current and predicted trends in global carbon dioxide emissions. This growth trend is also likely to enhance oak regeneration and establishment, given the often positive relation between growth and reproduction (Gómez-Aparicio et al. 2011). Unfortunately, testing this hypothesis in periodically masting species has proven difficult in retrieving a strong treatment signal (Lombardo and McCarthy 2008). The attenuation in growth rates suggests the need for repeated application of the treatments (particularly the mechanical treatments) to sustain growth and productivity of surviving trees.

Interspecific differences in tree growth response to the treatments suggest that inferences about residual tree responses to the restoration methods based only on stand- level results may obscure some important patterns. The higher relative growth rate of yellow-poplar compared with the more desirable oaks also calls for careful consideration of candidate species for removal during thinning operations. As a composite measure, the observed tree growth responses may reflect several other factors including size, age, competitive status, moisture demand/stress, and water-use efficiency of tree species. In general, the current results demonstrate that residual tree growth is responsive to forest disturbances and ought to be regularly monitored together with regeneration and survival to better understand the full scale of prescribed fire and mechanical thinning effects on forest ecosystems. 50

Literature Cited

Abrams, M. D., and J. A. Downs. 1990. Successional replacement of old-growth white

oak by mixed mesophytic hardwoods in southwestern Pennsylvania. Canadian

Journal of Forest Research 20:1864–1970.

Abrams, M. D., and D. M. McCay. 1996. Vegetation-site relationships of witness trees

(1780-1856) in the presettlement forests of eastern West Virginia. Canadian

Journal of Forest Research 26:217–224.

Albrecht, M. A., and B. C. McCarthy. 2006. Effects of prescribed fire and thinning on

tree recruitment patterns in central hardwood forests. Forest Ecology and

Management 226:88–103.

Bates, D. M. 2005. Fitting linear mixed models in R. R News 5:27–30.

Biondi, F. 1999. Comparing tree-ring chronologies and repeated timber inventories as

forest monitoring tools. Ecological Applications 9:216–227.

Biondi, F., and F. Qeadan. 2008a. Inequality in paleorecords. Ecology 89:1056–1067.

Biondi, F., and F. Qeadan. 2008b. A theory-driven approach to tree-ring standardization:

defining the biological trend from expected basal area increment. Tree-ring

Research 64:81–96.

Boer, M. M., R. J. Sadler, R. S. Wittkuhn, L. McCaw, and P.F. Grierson. 2009. Long-

term impacts of prescribed burning on regional extent and incidence of

wildfires—Evidence from 50 years of active fire management in SW Australian

forests. Forest Ecology and Management 259:132–142. 51

Boerner, R. E. J., J. Huang, and S. C. Hart. 2008. Fire, thinning, and the carbon economy:

Effects of fire and fire surrogate treatments on estimated carbon storage and

sequestration rate. Forest Ecology and Management 255:3081–3097.

Boerner, R. E. J., S. J. Morris, K. L. Decker, and T. F. Hutchinson. 2003. Soil and forest

floor characteristics. Pages 47–56 E. K. Sutherland, and T.F. Hutchinson,

editors. Characteristics of mixed-oak forest ecosystems in southern Ohio prior to

the reintroduction of fire. General Technical Report NE-299. U.S. Department of

Agriculture, Forest Service, Northeastern Research Station.

Boerner, R. E. J. 2005. Soil, fire, water, and wind: how the elements conspire in the forest

context. Pages 104–122 Fire in Eastern Oak Forests: Delivering Science to

Land Managers. Proceedings of a Conference, November 15-17, 2005. Fawcett

Center, The Ohio State University, Columbus Ohio.

Brose, P. H., D. C. Dey, R. J. Phillips, and T. A. Waldrop. 2013. A meta-analysis of the

fire-oak hypothesis: does prescribed burning promote oak reproduction in eastern

North America? Forest Science 59:322–334.

Brose, P. H., T. M. Schuler, and J. S. Ward. 2006. Responses of oak and other hardwood

regeneration to prescribed fire: what we know as of 2005. Pages 123–135 M.

B. Dickinson, editor. Fire in Eastern Oak Forests: Delivering Science to Land

Managers. U.S. Forest Service General Technical Report NRS-P-1.

Brudvig, L. A., H. M. Blunck, H. Asbjornsen, V. S. Mateos-Remigio, S. A. Wagner, and

J. A. Randall. 2011. Influences of woody encroachment and restoration thinning 52

on overstory savanna oak tree growth rates. Forest Ecology and Management

262:1409–1416.

Bunn, A. G., M. Korpela, F. Biondi, F. Campelo, P. Mérian, and C. Zang. 2012. dplR:

Dendrochronology Program Library in R. R Package Version 1.5.6.

Http://CRAN.R-project.org/package=dplR

Burnham, K. P., and D. R. Anderson. 2002. Model Selection and Multimodel Inference:

A Practical Information-Theoretical Approach. 2nd ed. New York, USA:

Springer-Verlag.

Chiang, J.-M., R. W. McEwan, D. A. Yaussy, and K. J. Brown. 2008. The effects of

prescribed fire and silvicultural thinning on the aboveground carbon stocks and

net primary production of overstory trees in an oak-hickory ecosystem in southern

Ohio. Forest Ecology and Management 255:1584–1594.

Cook, E. R. 1985. A Time Series Approach to Tree-ring Standardization. Dissertation.

Dobbertin, M. 2005. Tree growth as indicator of tree vitality and of tree reaction to

environmental stress: a review. European Journal of Forest Research 124:319–

333.

Fiedler, C. E., K. L. Metlen, and E. K. Dodson. 2010. Restoration treatment effects on

stand structure, tree growth, and fire hazard in a ponderosa pine/douglas-fir forest

in Montana. Forest Science 56:18–31.

Fritts, H. C. 1976. Tree rings and climate. New Jersey, The Blackburn Press. Page 567.

Grissino-Mayer, H. D. 2001. Evaluating crossdating accuracy: a manual and tutorial for

the computer program COFECHA. Tree-Ring Research 57:205–221. 53

Gómez-Aparicio, L., R. García-Valdés, P. Ruíz-Benito, and M.A. Zavala. 2011.

Disentangling the relative importance of climate, size and competition on tree

growth in Iberian forests: implications for forest management under global

change. Global Change Biology 17:2400–2414.

Hilt, D. E. 1979. Diameter growth of upland oaks after thinning. Broomall, PA, USDA

Forest Service Research Paper NE-437.

Hilt, D. E., E. D. Rast, and H. J. Bailey. 1983. Predicting diameters inside bark for 10

important hardwood species. Broomall, PA. USDA Forest Service Research Paper

NE-531.

Holmes, R. L. 1999. User’s manual for program EDRM v6.00P. Laboratory of Tree-Ring

Research. The University of Arizona, Tucson.

Hurteau, M. D., and M. North. 2010. Carbon recovery rates following different wildfire

risk mitigation treatments. Forest Ecology and Management 260:930–937.

Hutchinson, T. F., R. P. Long, R. D. Ford, and E. K. Sutherland. 2008. Fire history and

the establishment of oaks and maples in second-growth forests. Canadian Journal

of Forest Research 38:1184–1198.

Hutchinson, T. F., R. P. Long, J. Rebbeck, E. K. Sutherland, and D. A. Yaussy. 2012.

Repeated prescribed fires alter gap-phase regeneration in mixed-oak forests.

Canadian Journal of Forest Research 314:303–314.

Hutchinson, T. F., E. K. Sutherland, and D. A. Yaussy. 2005. Effects of repeated

prescribed fires on the structure, composition, and regeneration of mixed-oak

forests in Ohio. Forest Ecology and Management 218:210–228. 54

Iverson, L. R., T. F. Hutchinson, A. M. Prasad, and M. P. Peter. 2008. Thinning, fire, and

oak regeneration across a heterogeneous landscape in the eastern U.S.: 7-year

results. Forest Ecology and Management 255:3035–3050.

Johnson, P. S., S. R. Shifley, and R. Rogers. 2002. The Ecology and Silviculture of Oaks.

CABI Publishing. Wallingford, UK.

Keeley, J. E. 2009. Fire intensity, fire severity and burn severity: a brief review and

suggested usage. International Journal of Wildland Fire 18:116–126.

Laughlin, D. C., J. D. Bakker, M. L. Daniels, M. M. Moore, C. A. Casey, and J. D.

Springer. 2008. Restoring plant species diversity and community composition in a

ponderosa pine-bunchgrass ecosystem. Plant Ecology 197:139–151.

Lombardo, J. A., and B. C. McCarthy. 2008. Silvicultural treatment effects on oak seed

production and predation by acorn weevils in southeastern Ohio. Forest Ecology

and Management 255:2566–2576.

McEwan, R. W., J. M. Dyer, and N. Pederson. 2011. Multiple interacting ecosystem

drivers: toward an encompassing hypothesis of oak forest dynamics across eastern

North America. Ecography 34:244–256.

Nowacki, G. J., and M. D. Abrams. 1997. Radial-growth averaging criteria for

reconstructing disturbance histories from presetlement-origin oaks. Ecological

Monograph 67:225–249.

Pinheiro, J. C., and D. M. Bates. 2000. Mixed-Effects Models in S and S-PLUS.

Springer-Verlag. New York, USA. 55

Plauborg, K. U. 2004. Analysis of radial growth responses to changes in stand density for

four tree species. Forest Ecology and Management 188:65–75.

R Development Core Team. 2012. R: A language and environment for statistical

computing. R Foundation for Statistical Computing. Vienna, Austria.

Http://www.R-project.org/

Rubino, D. L., and B. C. McCarthy. 2004. Comparative analysis of dendroecological

methods used to assess disturbance events. Dendrochronologia 21:97–115.

Sala, A., G. D. Peters, L. R. McIntyre, and M. G. Harrington. 2005. Physiological

responses of ponderosa pine in western Montana to thinning, prescribed fire and

burning season. Tree physiology 25:339–48.

Schwilk, D. W., J. E. Keeley, E. E. Knapp, J. McIver, J. D. Bailey, C. J. Fettig, C. E.

Fiedler, R. J. Harrod, J. J. Moghaddas. K. W. Outcalt, C. N. Skinner, S. L.

Stephens, T. A. Waldrop, D. A. Yaussy, and A. Youngblood. 2009. The national

fire and fire surrogate study: effects of fuel reduction methods on forest

vegetation structure and fuels. Ecological applications 19:285–304.

Speer, J. H. 2010. Fundamentals of Tree-ring Research. The University of Arizona Press.

Tucson, AZ.

Stephens, S. L., J. J. Moghaddas, C. Edminster, C. E. Fiedler, S. Haase, M. Harrington, J.

E. Keeley, E. E. Knapp, J. D. McIver, K. Metlen, C. N. Skinner, and A.

Youngblood. 2009. Fire treatment effects on vegetation structure, fuels, and

potential fire severity in western U.S. forests. Ecological applications 19:305–20. 56

Stokes, M. A., and T. L. Smiley. 1968. An Introduction to Tree-ring Dating. University

of Chicago Press. Chicago, IL.

Thorpe, H. C., S. C. Thomas, and J. P. Caspersen. 2007. Residual-tree growth responses

to partial stand harvest in the black spruce () boreal forest.

Canadian Journal of Forest Research 37:1563–1571.

Tilman, D. 1985. The resource-ratio hypothesis of plant succession. The American

Naturalist 125:827–852.

Trimble, G. R. 1967. Diameter growth in second-growth Appalachian hardwood stands –

a comparison of species. 5 pages. US Department of Agriculture, Forest Service

Research Paper NE-75. Parsons, W. Va.

Visser, H. 1995. A note on the relation between ring-widths and basal area increment.

Forest Science 41:297–304.

Waldrop, T.A., D. A. Yaussy, R. J. Phillips, T. A. Hutchinson, L. Brudnak, and R. E. J.

Boerner. 2008. Fuel reduction treatments affect stand structure of hardwood

forests in Western North Carolina and Southern Ohio, USA. Forest Ecology and

Management 255:3117–3129.

Xu, C.-Y., M. H. Turnbull, D. T. Tissue, J. D. Lewis, R. Carson, W. S. F. Schuster, D.

Whitehead, A. S. Walcroft, J. Li, and K. L. Griffin. 2012. Age-related decline of

stand biomass accumulation is primarily due to mortality and not to reduction in

NPP associated with individual tree physiology, tree growth or stand structure in a

Quercus-dominated forest. Journal of Ecology 100:428–440. 57

Yamaguchi, D. K. 1991. A simple method for cross-dating increment cores from living

trees. Canadian Journal of Forest Research 21:414–416.

Yaussy. D. A., M. B. Dickinson, and A.S. Bova. 2004. Prescribed surface-fire tree

mortality in southern Ohio: equations based on thermocouple probe temperatures.

Pages 67–75 D. A. Yaussy, D. M. Hix, R. P. Long, and P. C. Goebel, editors.

14th Central Hardwood Forest Conference, 2004 March 16–19. Wooster, OH:

U.S. Department of Agriculture, Forest Service, Northeastern Research Station.

Zuur, A. F., E. N. Ieno, N. J. Walker, A. A. Saveliev, and G. M. Smith. 2009. Mixed

effects models and extensions in ecology with R. Springer. New York, USA.

Table 2.1

Variable Black oak Chestnut oak White oak Hickory Yellow-poplar Number of series 164 162 188 88 94 Master series 1857-2011 (155 yrs) 1856-2011 (156 yrs) 1850-2011 (162 yrs) 1744-2011 (264 yrs) 1865-2011 (147 yrs) Mean length 96 (25) 95 (26) 103 (25) 103 (33) 78 (22) of seriesa (years) Mean ring widtha 1.819 (0.447) 1.639 (0.551) 1.602 (0.377) 1.205 (0.259) 1.999 (0.668) (mm) Series inter- 0.666 (0.140) 0.627 (0.164) 0.655 (0.145) 0.662 (0.135) 0.689 (0.108) correlationa 1st order 0.500 (0.16) 0.475 (0.16) 0.489 (0.15) 0.511 (0.14) 0.612 (0.15) autocorrelationa Gini co-efficienta 0.163 (0.04) 0.169 (0.04) 0.176 (0.04) 0.201 (0.05) 0.347 (0.42) 59

Table 2.2 – – Response AIC Species Best model c variable weight () BAI All species BAI = FUEL × Time × Species 1.00 Black oak BAI = MECH × Time 0.81 Chestnut oak BAI = FUEL × Time 0.45 White oak BAI = MECH + BURN + Time 0.63 Hickory BAI = FUEL 0.40 Yellow-poplar BAI = MECH × Time 0.92 Change in BAI All species % GC = FUEL × Time × Species 1.00 growth (GC) Black oak % GC = Site 0.53 Chestnut oak % GC = MECH 0.39 White oak % GC = MECH × Time 0.89 Hickory % GC = MECH × Time 0.52 Yellow-poplar % GC = MECH × Time 0.99

Vegetation characteristics and fuel reduction treatment operations in the study sites: (A) control site with a downed tree; (B) prescribed burning; (C) thinning followed by prescribed burning; (D-F) vegetation showing some residual tree species after the treatments. 61

Annual rings (1996-2005) compared for the five tree species analyzed for their radial growth responses to prescribed fire and thinning treatments. Note the increase in ring widths across species (particularly yellow-poplar) following the treatments in 2000. 62

Composite raw and standardized ring-width chronologies of the five mixed- oak forest species analyzed. Dotted vertical line indicates the period of growth analyzed and the arrows indicate when the treatments were applied.

Periodic annual basal area increment responses of residual trees to prescribed fire and thinning treatments over time across species in mixed-oak forests of southern Ohio. Error bars represent one standard error. The dotted vertical line indicates the time the treatments were applied. Note differences in y-axis.

Changes in basal area increment of residual tree species before and after prescribed fire and thinning treatments in mixed-oak forests of southern Ohio. Error bars represent one standard error. The dotted vertical line indicates the time the treatments were applied. Note differences in y-axis.

Basal area increment responses of residual tree species to prescribed fire severity in the study area. Different letters in each panel indicate significant difference at

CHAPTER 3: EFFECTS OF COMPETITION, SIZE AND AGE ON TREE

GROWTH RESPONSE TO PRESCRIBED FIRE AND THINNING TREATMENTS IN

MIXED-OAK FORESTS OF OHIO

Abstract

Prescribed fire and thinning treatments are increasingly used to mitigate potential wildfire hazards, alter stand structure and restore forest functions. In this study, the effects of competition, tree size and age on tree growth following prescribed and thinning were examined to better understand the consequences of these management tools on forest systems. Tree-ring data from 348 trees, comprising five species (white oak, black oak, chestnut oak, hickories, and yellow-poplar) were analyzed following standard dendrochronological protocols. Data were collected from 80 0.1-ha plots in two study sites, each with four treatment units (control, thin-only, burn-only, thin+burn) in mixed- oak forests of Ohio. A neighborhood analysis was used to assess the effect of competition on tree growth. Basal area increment (BAI) and tree size were positively related, with the strongest correlation found in the burn-only treatment. Age was negatively related to

BAI, though weakly. Competition was inversely correlated with BAI, with trees from the thin-only unit showing the strongest correlation. BAI was greater for larger trees when competition was low and declined at a steeper rate as competition increased. Smaller trees grew less in general but decreases in BAI were not as steep as competition increased. Overall, tree size, age, and competition explained ~ 40% of total variance in

BAI across all species. Values for individual tree species ranged from 30–57%, indicating considerable variation in the responses of species to these factors. Yellow-poplar 67 exhibited greater sensitivity to competition than the other species analyzed. Altogether, competition was more important than size and age for tree growth in these managed stands. Variation in competitive status of trees within treatments supports the view that prescribed fire and thinning influence forest growth by creating heterogeneity among stands, and thus demonstrates the need for individual tree-based analysis to fully understand prescribed fire and thinning impacts on forest ecosystems.

Introduction

Prescribed fire and thinning treatments are widely used to mitigate wildfire effects, alter stand structure and species composition, and restore forest functions

(Hutchinson et al. 2005, Schwilk et al. 2009). Over the past two decades, many studies have examined the effects of these treatments on forest ecosystems; however, the response of large residual trees has received considerably less attention. The few studies focused on residual tree growth following these management disturbances have demonstrated interesting response patterns over space, time, and across species (Boerner et al. 2008, Chiang et al. 2008, Lutz et al. 2012, Anning and McCarthy 2013). While these patterns are becoming apparent, the underlying factors have not been investigated thoroughly, particularly at the individual tree scale. Understanding these factors is essential for predicting productivity and development of forest stands following prescribed burning and thinning (Johnson et al. 2002). More fundamentally, knowledge of the relative growth rates of different species in relation to competition, size and age is useful to forest managers interested in determining which species to favor in partial harvesting (Trimble 1967). 68

Tree growth is regulated by many biotic and abiotic factors (Fritts 1976,

Kozlowski et al. 1991). Among these, competition between individual trees continues to receive greater research attention because of its strong controlling effects on stand structure and development (Kozlowski et al. 1991, McDonald et al. 2002, Weber et al.

2008, Thorpe et al. 2010). Competition affects the availability and acquisition of resources such as light, water, nutrients and physical space, and thus has profound implications for forest ecology and management. In the closed-canopy forest of eastern

North America, for example, competition has been identified as a major determinant of plant growth and productivity (Phipps 1982). Therefore, it is not surprising that release of certain desirable tree species from competition has become synonymous with many prescribed fire and thinning management efforts in these temperate hardwood forests.

Prescribed fire and thinning treatments may create structurally complex forest stands, with trees varying in age, size, species and spatial arrangement, which require spatially-explicit, individual tree-based models to understand (Lorimer 1983, Canham et al. 2004, Thorpe et al. 2010). Evidence from several studies suggests considerable differences in the nature and effects of these management operations, with prescribed fire being more variable but removing less basal area than mechanical treatments (e.g.,

Waldrop et al. 2008). Canham et al. (2004) noted that partial harvesting could lead to considerable changes in the physical and competitive environments of stands. Despite these differences or changes, and the fact that resources are generally spatially heterogeneous across landscapes (Coomes and Allen 2007), most researchers continue to rely on coarser descriptors (e.g., stand basal area or density) when analyzing tree growth 69 patterns within these complex systems. Consequently, how prescribed fire and thinning manipulations influence the competitive status of individual trees, and how this, in turn, mediates residual tree growth remain unclear.

Traditionally, the effect of competition on tree growth has been assessed via neighborhood analysis (Lorimer 1983, Canham et al. 2006). This approach typically requires the demarcation of the spatial extent of a tree’s competitive environment (i.e., its neighborhood), within which tree growth is assumed to be a function of the number, size, species and spatial configuration of neighboring trees. Thus, integrating these neighborhood characteristics, ecologists have developed a variety of competition indices with which to measure the extent of resource limitation by a plant’s growing environment

(Shi and Zhang 2003). The most popular of these are the distance-dependent and the distance-independent models (Wimberly and Bare 1996, Canham et al. 2006). Weiner

(1990) also distinguished between asymmetric competition models, which involve only individuals larger than the target tree and reflect competition for light or effect of shading, and symmetric competition models, which incorporate all neighbors irrespective of size and represent competition for nutrients. These indices are of great value in accounting for the spatial structure in community data (Shi and Zhang 2003), although the choice of a particular competition index is rather subjective.

Tree growth is strongly related to size and age (Wyckoff and Clark 2005,

Macfarlane and Kobe 2006, Coomes and Allen 2007). Studies have shown that larger trees usually produce more wood than smaller trees (Kozlowski et al. 1991). For example, McDonald et al. (2002) state that the size of an individual tree has a direct 70 effect on its future growth, while size in relation to competitors indirectly affects growth through competitive effects. The size of an individual tree relative to its neighbors also influences resource supply and tree growth. Smaller trees are often shaded and suppressed by their larger neighbors (Coomes and Allen 2007), although some trees naturally lack the capacity to grow into the canopy. Thus, failure to account for initial tree size in tree growth models may lead to residual size bias (Macfarlane and Kobe

2006), which in turn may obscure the effects of environmental stresses on tree growth.

The present study examined the influences of competition, size and age on the growth response of residual trees to fuel reduction treatments. Three specific questions were addressed: (a) Does the size of an individual tree and its age affect its growth response to prescribed fire and thinning treatments? (b) Does competition among individual trees mediate tree growth response to these treatments? and (c) What are the relative contributions of competition, size and age to tree growth following prescribed fire and thinning treatments? These questions were evaluated using tree-ring data from five common tree species in the mixed-oak forests of southeastern Ohio. It was predicted that basal area growth would increase with size and decrease with age of trees following the treatment, with different treatments having varying effects on different-sized trees. If treatments conferred differential competitive advantages on trees or structural changes among stands due to differences in intensity (Waldrop et al. 2008), growth responses of the individual trees would be expected to differ among the treatments. Finally, it was expected that competition would influence tree growth responses to the treatment more than size and age. 71

Methods

Two replicate blocks within the Ohio Hills site of the national Fire and Fire

Surrogate (FFS) study were used for this study. The Raccoon Ecological Management

Area (REMA) block is located within the Vinton Furnace State Experimental Forest

(39º12W), and the Zaleski block is within the Zaleski State Forest

(39º2W); both are in Vinton County, Ohio. The Ohio Hills site lies within the unglaciated Allegheny Plateau physiographic region. The landscape is dissected into ridges, hills and valleys (Hutchinson et al. 2005), with elevation ranging from 200 to 300 m (Waldrop et al. 2008). Soils are mainly acidic and are derived primarily from sandstone, siltstone and shale (Boerner et al. 2003). Annual precipitation and temperature average 1024 mm and 11.3 oC, respectively (Sutherland et al. 2003). The vegetation is classified as upland mixed-oak (spp.) hardwood forests (Iverson et al. 2008). Prior to the start of the treatments in 2000, the even-aged stands within both

2 -1 blocks were fully stocked with tree basal area ranging from 25.5 to 29.4 m ha

(Appendix 2A; Waldrop et al. 2008).

At each site, there are four experimental units, each ~ 50 ha in extent and containing ten 20 × 50 m (0.1 ha) permanent plots (i.e., 2 sites × 4 experimental units ×

10 plots = 80 plots total). The four treatments consist of an un-manipulated control, a mechanical thinning (thin-only), a prescribed burning (burn-only), and a combination of the two (thin+burn). These plots are distributed across the landscape from ridgetops to lower slopes based on the integrated moisture index (IMI) developed by Iverson et al. 72

(1997). The IMI combines four topographic and soil factors (hillshade, curvature, flow accumulation and water holding capacity of the soil) in GIS (geographic information systems) to derive relative moisture ratings for sites. The model has been used successfully to predict site productivity and species composition in the oak-dominated forests of eastern North America (Iverson et al. 1997).

Mechanical thinning was conducted in the fall and winter of 2000–2001. This operation was primarily thinning from below, which focused on removing midstory trees

(15–30 cm diameter at breast height, DBH), and resulted in ~ 30 % reduction in stand basal area, although variations in treatment intensity were discernible (Appendix 2A). For example, at REMA, the thin-only and the thin+burn treatments reduced stand basal areas by 31.4 and 18.9%, respectively, from their initial values of 27.4 and 27.9 m2 ha-1.

Prescribed fires were conducted in the spring of 2001, and repeated in 2005 and 2010.

The intensity of prescribed fire varied greatly over the years and across landscapes. In

2001, for example, fire intensity was generally low with flame lengths getting to about

1m. However, higher intensity fires (i.e., flame lengths reaching 4–5m) were deliberately created in 2005 and 2010, resulting in significant overstory mortality (Iverson et al. 2008,

Waldrop et al. 2008).

In the fall and winter of 2011–2012, 696 increment cores were extracted from white oak (), chestnut oak (), black oak (), hickories

(spp. primarily ) and yellow-poplar (). These tree species were selected for analysis because they are the most common overstory species in 73 the study area. To minimize noise in the data, healthy-looking co-dominant trees ( 25 cm DBH) having no more than 50% branch or twig mortality or foliage discoloration, dieback and defoliation (Yaussy et al. 2004) were sampled using pre-existing FFS overstory data as a guide. Two increment cores were taken from each tree at breast height, and at roughly 180o from each other to account for variation in ring-width around the circumference of the tree. To obtain at least one tree per species per plot, sampling was extended into a 20 m buffer outside the plot, if necessary. In spite of this sampling effort, some species were still absent in some sampling areas. For an estimate of tree age, one of the two cores from each tree was taken to the pith (Speer 2010). Diameter at breast height (DBH) was measured at the time of coring. Cores were extracted from trees using an increment borer and transported to the laboratory in paper straws kept in a metallic case to prevent mechanical damage.

In the laboratory, increment cores were glued to wooden core mounts and air- dried for at least 24 hrs. The cores were then sanded using a belt sander with increasingly finer grades of sandpaper (100 to 600 grit), and micro-finished with 30, 15 and 9 micron

Microfilm Sheets to clearly reveal the ring structure (Speer 2010). Calendar dates were assigned to each ring with the help of skeleton-plots (Stokes and Smiley 1968). Ring widths were measured to the nearest 0.01 mm using a Velmex measuring system

(Velmex, Boomfield, NY) and the standard Measure J2X program (Voortech Consulting

Holderness, NH). The quality control program, COFECHA (Grissino-Mayer 2001), was used to verify cross-dating and measurement accuracies. The age of each tree was estimated from a ring count in the core that contained the pith. If the pith was absent in a 74 core, a pith indicator (a series of concentric rings) was used to estimate the number of missing rings (Speer 2010).

To quantify the radial growth of individual trees, each ring was converted into a basal area increment (BAI), the net increase in the total cross-sectional stem area of a tree. BAI is the preferred metric of growth for many dendrochronologists because it has been found to provide a better approximation of annual tree growth than simple ring width or stem diameter increment (Biondi and Qeadan 2008). BAIs were computed using the function in “dplR”, which calculates the ring area from the bark to the pith

(Bunn et al. 2012). Computations were based on diameter inside bark of trees, derived by first estimating bark thickness of the species using available regression equations (Hilt et al. 1983). The BAIs of the two cores extracted from a tree were averaged to obtain a single value per year for that tree; thus, the individual tree constituted the sampling unit.

To accomplish the goal of understanding how tree size, age and competitive status influence post-treatment growth patterns of trees, analysis was restricted to only the post- treatment period (2001–2010). Thus, for each tree, mean periodic annual BAI was computed by averaging the annual BAIs from 2001 to 2010.

A neighborhood analysis (Lorimer 1983, Wimberly and Bare 1996) was used to quantify the effect of competition on tree growth response to the treatments. Considering the large number of competition indices that exist in the ecological literature, it was decided to empirically determine the best model for the tree growth analysis in this study. 75

This was accomplished by evaluating a set of five candidate models within three neighborhood radii—10 m, 15 m, and a variable radius (Appendix 2B). The variable radius was a function of the size of the target tree and was obtained by multiplying the diameter of the target tree by 40 (Lorimer 1983). All but one of the models were based on the traditional distance-dependent competition model (denoted here as the “basic” model;

Appendix B), which is derived by summing the ratios of the diameters of a target tree and its neighbors, weighted by the distance from the target tree:

= (1) where: denotes the competition index, and are the diameters of the target tree and the neighbor, respectively. All trees with DBH equal to 10 cm or larger were considered as neighbors. The distance between a target tree and its neighbor was measured using an Optic-Logic InSightTM 800X L Laser Rangefinder (Optic-Logic,

Tullahoma, TN). A tape measure was used to measure distances shorter than the minimum range of the rangefinder (3.6 m). Measurement of the neighbor DBH and distance from the target tree were done at the time of coring.

Using the simple linear regression model, the five models were fitted to the pre- treatment BAI (i.e., BAI for the period 1996–2000), and their predictive capacities for

BAI compared using the resulting values. Model parameters were estimated using the restricted maximum likelihood method (REML), whilst model validation was performed by plotting the residuals against each of the explanatory variables (Zuur et al. 2009). The 76 selected model (the model used in subsequent analysis) included only neighbors larger than the target tree with the neighborhood radius of 15 m (Appendix 2B).

All analyses were performed using R (R Development Core Team 2012). The effects of size, age and competition on BAI responses of trees to the fuel reduction treatments were evaluated using analysis of covariance (ANCOVA). ANCOVA is a general linear model commonly used to evaluate the covariance of a dependent variable

(e.g., growth metric) and a categorical predictor (main effect) while “controlling” for the effects of other continuous predictors, known as covariates. Treatment was specified as the main effect, and tree size, age, and competition status as covariates. The goal of this analysis was to determine whether significant interactions existed between the levels of the treatment variable and the covariates. Data were analyzed separately for each species and were also pooled together for the general effect across treatments. The value was used to assess the goodness of model fit. Linear regression models were used to assess the effect of species on BAI-competition relations, whilst ANOVA was used to evaluate variation in competitive status of trees among the four treatments. Further, BAI- competition relations for different species were analyzed by IMI (which stratified the landscape into mesic, intermediate, and xeric sites). USDA staff at the Northern Research

Station, Delaware, OH, supplied the IMI scores.

To investigate whether competition effect was dependent on size of target tree, all cored trees were sorted into three size classes according to the definition of Schwilk et al.

(2009). The three classes were medium (25–39.9 cm DBH), large (40–54.9 cm DBH), 77 and very large (55 cm or greater). However, considering the small number of trees in the very large size class, trees belonging to the last two size classes were lumped together as a single “large” category. A simple linear regression analysis was then used to evaluate the relationship between BAI and competition for each of the two size classes (medium and large). Further, trees were divided into two competition classes: those that experienced lesser competition—competitively advantaged (index value < 0.4313, mean

BAI = 28.81 cm2 yr-1), and those that received greater competition—competitively

2 yr-1) using the ‘rpart’ function in R (Therneau et al. 2012). In this analysis, BAI was modeled as a function of competition using the “anova” method for continuous endpoint. The splitting criterion was chosen to maximize the between group sum of squares as in simple ANOVA.

Variations in BAI between these two competition classes across treatments, and the different size classes were then compared with ANOVA.

Finally, the relative importances of size, age and competition for BAI following the treatments were quantified by decomposing the total variance explained () in a multiple linear regression by averaging sequential sum of squares over all orderings of the explanatory variables. The results were then normalized to sum to 100%. This analysis was implemented using the ‘relaimpo’ package in R (Gromping 2006). In addition to the relative importance values, the package contains functions for estimating

95% bootstrap confidence intervals using 1000 replications generated from the original data. This analytical approach overcomes the common problem of correlation among regressors, and thus has advantage over the use of from univariate regressions 78

(Gromping 2006). To simplify the analysis and better understand the effects of competition, size and age on tree growth in the managed stands, data from the control stands were excluded from this analysis.

Results

Characteristics of the trees sampled from the four treatment units, including the number of trees cored, the mean DBH, age, and periodic annual BAI before and after the treatment are summarized in Table 3.1. Tree size and age did not differ among the four treatments ( > 0.05). Average size of trees ranged from 42.12 cm in the control to 44.24 cm in the thin+burn stand, whilst mean age ranged between 109.49 (control) and 110.50

(for both thin-only and burn-only) years. Pretreatment BAI was quite comparable among all treatments ( = 0.084), averaging between 15.13 and 17.15 cm2 yr-1. However, post- treatment BAI was higher ( < 0.05) for trees growing in the thin-only (23.14 cm2 yr-1) and the thin+burn (23.55 cm2 yr-1) treatments than for those from the control (16.33 cm2 yr-1) and the burn-only (20.52 cm2 yr-1) stands.

When data for all species were pooled, tree growth responded strongly to tree size and age ( < 0.001; Figure 3.1, Table 3.2). As expected, BAI showed a positive correlation with tree size, and also responded strongly to the treatments ( < 0.001).

Trees from the thin-only and thin+burn treatments grew more relative to those from the control. No significant treatment × size effect on BAI was found ( > 0.05). However, growth response to prescribed fire appeared to vary among trees of different sizes. 79

Prescribed burning generated much greater BAI response among smaller trees compared to larger trees (Figure 3.1a). Together, treatment and size explained 25.5% of the variations in BAI. In contrast with size, tree age was only weakly associated with BAI

(Figure 3.1b), with younger trees showing slightly higher growth rates than older ones.

When combined in regression analysis, tree age and treatment only explained 7.51% of the variations in BAI (Table 3.2). No interactive effect of age and treatment on BAI was found ( > 0.05).

Analysis of BAI by species indicated a strong positive effect of size across all five tree species ( < 0.001; Table 3.2). After accounting for the effect of size, treatment effect on BAI was significant only for white oak, black oak, and hickory ( < 0.05; Table

3.2). There was no significant treatment × size effect on BAI for any of the species. The importance of tree size and treatment for BAI (as indicated by ) varied greatly among species. These factors were more strongly associated with growth of white oak (2 =

43.15%), hickory (35.8%) and yellow-poplar (35.2%) than with those of black oak

(23.17%) and chestnut oak (15.73%). With the exception of chestnut oak, tree age had no significant effect on BAI of trees ( > 0.05; Table 3.2). As with size, no significant treatment × age effect on BAI ( > 0.05) was found.

As predicted, the competitive status of individual trees differed among the four treatments ( < 0.001; Figure 3.2). Trees from the control plots experienced the most intense competition while those from the thin-only and the thin+burn units had the least competition. Competition had a strong negative effect on BAI across all treatments ( < 80

0.001, Table 3.3, Figure 3.3). A strong competition × treatment effect on BAI was also noticeable when results for all species were pooled ( = 0.012). At low competition intensity, trees from all three active treatments (particularly thin-only and thin+burn) exhibited higher BAI rates than those from the control. Conversely, only trees growing in the thin+burn plots exhibited higher BAI rates than the control trees at high competition intensity. Interestingly, tree growth in the thin-only stands was more sensitive to competition than in the other treatments. Together, competition and treatment explained

31.37% of the variance in BAI.

The strong negative effect of competition on BAI was also evident at the species level (Table 3.3, Figure 3.4), with growth rates declining rapidly as competition intensity increased across species. After “controlling” for the effect of competition, treatment was only significantly related to BAI of white oak ( = 0.020) and hickory ( = 0.014). The five species studied responded differently to competition and treatment, as indicated by variation in the (26.86-47.86%; Table 3.3). Yellow-poplar was more sensitive to competition than the other species (Figure 3.4). Hickory was the least responsive, whilst oaks were intermediate in their reaction to competition. Among the oaks, black oak and chestnut oak were more sensitive to competition than white oak.

Besides their independent effects, competitive and species also showed strong interactive effects on basal area growth in the manipulated stands ( < 0.05; Figure 3.4).

The two variables accounted for 53.34% and 40.53% of the variance in BAI for the control and the active treatments, respectively. Site quality index (represented by the IMI categories), somewhat mediated the BAI-competition relation (Table 3.4). This 81 interactive effect was exemplified by white oak, chestnut oak and yellow-poplar, which showed concordance between competition importance (indicated by values) and the optimal “physiographic” range of the species across the landscape (indicated by the intercepts estimated from the regression analysis). Competition effect on BAI following the treatment was size-dependent (Figure 3.5). Although larger trees generally grew more

(intercept = 32.95 cm yr-1) than medium-sized trees (intercept = 18.44cm yr-1), the latter were less sensitive (slope = -4.8 cm yr-1, = 12.72%) to changes in competition compared to the former (slope = -17.55 cm yr-1, = 24.96%).

Competition, size and age together explained ~ 40% of the total variance in BAI when data for all five species were combined (Figure 3.6). Of this value, competition alone accounted for an average of 46.44%, whilst tree size and age comprised 39.17 and

14.40%, respectively. The magnitudes of overall variance in BAI explained by competition, size and age varied among species. The values were generally greater for the individual species (except for black oak) than for the pooled data. Chestnut oak BAI was most strongly correlated with competition, size and age ( = 56.52%), whereas black oak BAI was least responsive to them ( = 30.33%). Decomposition of the total explained variance indicated that competition ( = 42–74%) was generally more important than tree size ( = 23–42%) and age (= 3–29%) for predicting BAI of the oaks. On the other hand, among the non-oaks, a greater proportion of the variance in BAI

(47–49%) was attributable to tree size. Interestingly, age ( = 29%) predicted chestnut 82 oak growth better than size, though it was only weakly ( < 9%) associated with BAI of the other species.

Discussion

The importance of tree size for predicting future tree growth and forest productivity has been emphasized in several previous studies (Lorimer 1983, McDonald et al. 2002, Macfarlane and Kobe 2006). In the current study, BAI correlated positively with tree size, indicating that larger trees indeed produced more wood than smaller ones.

The higher rate of wood production by larger trees likely reflects a greater capacity to capture light for photosynthesis due to their larger crown size (Kozlowski et al. 1991,

Wyckoff and Clark 2005). Mechanical treatments generated the greatest BAI response among larger trees as they reduced stand basal area to a greater extent than prescribed fire. These growth responses greatly depended on species, which tend to vary in their adaptation to fire. On the other hand, growth of suppressed trees with small crowns is reported to cease sooner in the growing season than in dominant trees with large crowns

(Fritts 1976). This difference in growth period partly explains the observed disparity in growth rates of different sized trees.

Contrary to the mechanical thinning, the prescribed fire treatment seemed to have benefitted the smaller trees more, judging from their relatively greater growth in the burn- only stand. Fire intensity in our study was generally low to moderate (Waldrop et al.

2008), which induced significant positive growth responses among smaller trees. This positive growth response may seem unusual due to the potential loss of vigor reported for some species after fire. However, enhanced growth of an individual tree following 83 prescribed fire is conceivable if the treatment removes some neighboring trees, and frees up more resources for the surviving trees. Indeed, earlier analysis of 5-yr periodic annual

BAI of trees in relation to fire severity in our sites showed a positive response for some species (e.g., black oak; Anning and McCarthy 2013). Other studies have demonstrated the variable effect of fire on forest communities (which may be positive, neutral, or negative; e.g., Hutchinson et al. 2005, Brose et al. 2006). Resistance to low-intensity fire due to increased bark thickness and reduced vigor might partially explain the slower growth response of larger trees to the prescribed burning.

The results of this study also revealed higher BAI rates among younger trees compared to older trees, agreeing with the well-known phenomenon of tree growth decline with age (Kozlowski et al. 1991, Yoder et al. 1994, Ryan et al. 2006, Fiedler et al.

2010). Potential explanations for this growth decline are found in several hypotheses including reduction in photosynthetic rates due to hydraulic limitation (Ryan et al. 2006), decrease in leaf area to sapwood ratio (McDowell et al. 2002) and increase in sapwood respiration (Ryan and Waring 1992). Age had a significant negative effect on growth rates of chestnut oak trees, but the exact physiological mechanisms are not immediately clear.

Differences in treatment intensity resulted in substantial variations in the competitive status of individual trees, largely explaining the variation in BAI among the treatments. Prescribed fire decreased competition between trees but to a much lesser degree than was achieved with thinning alone, or the combined treatment. The relatively higher sensitivity of trees to mechanical thinning suggests a more direct response to 84 release from competition in the thin-only stand. By contrast, the slow decline in growth with increasing competition in the burn-only and the thin+burn stands suggests alteration of the growth environment in some beneficial ways other than simply releasing trees from competition. Such benefits may be a function of increased nutrient availability (e.g., total inorganic nitrogen, calcium, phosphorus), and mineral soil exposure and permeability following fire as documented by previous investigators (Boerner et al.

2009). Albrecht and McCarthy (2006, page 99) reported “sustained reductions in large- diameter (3–10 cm dbh) sapling densities” over a 4-year period following prescribed fire, whilst Hutchinson et al. (2005) reported significant fire × year effect on small tree (10–15 cm DBH) density. Thus, while our neighborhood analysis did not include these smaller plants, it is important to emphasize that their removal as a result of the prescribed fire treatment may stimulate growth of the larger residual trees. In general, the strong relation between BAI and competitive status of trees in these closed-canopy forests is ecologically meaningful and consistent with previous studies, which have highlighted strong negative effects of competition on tree growth and forest structure (Canham et al.

2004, Coomes and Allen 2007, Thorpe et al. 2010).

Competition effect on tree growth following the treatment varied with species.

Yellow-poplar, known for its vigorous growth in open canopies (Trimble 1967), benefited most from competition release, although a positive growth response was also evident in all other species. The higher sensitivity of yellow-poplar and, to some extent, chestnut and black oaks implies that intense and periodic thinning may be more beneficial to these species. In contrast, white oak and hickories were less reactive to 85 changes in competition, though they appeared more likely to respond to prescribed fire.

Site quality somewhat influenced the growth-competition relation of trees as indicated by the increase in competition importance for BAI within the optimum “physiographic” range (as stratified by the IMI) of species (Table 3.4). For example, competition effect on

BAI of white oak was strongest on xeric sites, yellow-poplar and hickories on mesic sites, and both black oak and chestnut oak on intermediate sites. This pattern is consistent with the prediction that “future stands dominated by oaks will be concentrated in low-site index areas (low IMI scores) where competition from more mesic and [shade-] tolerant species will be minimized …” (Iverson et al. 1997, page 345). However, it suggests more intense intraspecific competition within these optimal “physiographic” areas of the landscape. It also indicates that competition importance may vary in distinct ways along environmental gradients as noted by Kunstler et al. (2011).

The size dependence of competition observed in this study is in keeping with the findings of several previous studies (Lorimer 1983, McDonald et al. 2002, Easdale et al.

2012). As noted by McDonald et al. (2002), tree size affects resource acquisition, and larger trees are less susceptible to competition because they have more stored resources, which give them a greater competitive edge over smaller trees. Variation in the strength of the competition-growth relation between smaller and larger trees reflects differences in sensitivity to crowding and shading (Coomes and Allen 2007). As these authors argued, larger trees with well-developed and exposed canopy are more often limited by nutrient availability (as a result of crowding), while smaller trees are mainly limited by light availability (as a result of shading). Thus, while partial harvesting of stems may free up 86 nutrients, which can substantially enhance growth of larger residual trees, smaller trees may respond rather slowly due to persistent shading by their larger neighbors. This explanation seems plausible, though our competition model did not take into account the angular dispersion of neighbors around the target trees.

Tree growth in natural populations is a complex process and integrates multiple factors (Fritts 1976, Kozlowski et al. 1991). In this light, the ~ 40 % of variance in BAI explained by competition, size and age of trees after the treatments is quite striking, and demonstrates the importance of these factors within the managed stands. Despite the difficulty in disentangling their relative effects, the present results demonstrate with reasonable confidence that competition is a more important determinant of residual tree growth than size and age. Among the non-oaks, tree size exerted equal or even greater controls on BAI than competition, probably due to a slight size differential between them and the oaks; yellow-poplar and hickories likely experienced continued shading relative to their larger oak neighbors. The weak growth response to age could be due to the minimal variation in age of trees observed in these even-aged stands. Variance in BAI not explained by competition, size and age might be related to treatment manipulations, species, or site quality as earlier discussed, or they might be a function of climate and some biotic stresses (Fritts 1976).

Conclusions and Management Implications

The study revealed a substantial variation in competitive status of trees among the treatments—this supports the view that prescribed fire and thinning influence forest growth and development by creating heterogeneity among stands (Lorimer 1983, Thorpe 87 et al. 2010). Moreover, competition, size and treatment interacted in a complex fashion to influence BAI of residual trees. The higher sensitivity of trees to competition in the thin- only stands implies that mechanical treatments influence tree growth mainly through release from competition. However, prescribed fire may provide additional benefits via nutrient release, increased surface temperature and moisture availability (Kozlowski et al.

1991, Peterson et al. 1994, Boerner et al. 2009). These results also agree with the notion that mechanical treatment may not be a complete surrogate of prescribed fire (Schwilk et al. 2009). Overall, competition appeared to be a more important driver of residual tree growth than size and age, though its effect varied among species.

Prescribed fire and thinning treatments are frequently used to alter stand structure

(Hutchinson et al. 2005) to help promote tree growth and productivity. In this respect, mechanical treatments will be most effective as they generate greater BAI response. The current results, consistent with previous findings (e.g., Hutchinson et al. 2005, Waldrop et al. 2008), demonstrate that prescribed burning has minimal effect on large trees, probably due to its low intensity. Thus, in forests where mechanical thinning is not a management option or where there is the need to promote growth of fire-adapted species such as the oaks, repeated or more intense burns may be the best strategy to eliciting the desired responses among large trees. Variation in species response to competition in relation to site quality calls for careful consideration, not only of which species to favor during thinning operations, but also, of where and how the treatments are applied. For example, the high sensitivity of yellow-poplar to competition suggests the need for a higher treatment intensity to sustain the growth of this species. However, “wholesale” intense 88 fire and thinning will only give yellow-poplar and other fast-growing species a competitive advantage over the oaks. If the management goal is to enhance growth and productivity of a particular species, reduction in intraspecific competition within the optimal “physiographic” range of the species may be necessary.

Literature Cited

Albrecht, M. A., and B. C. McCarthy. 2006. Effects of prescribed and thinning on tree

recruitment patterns in central hardwood forests. Forest Ecology and Management

226:88–103.

Anning, A. K., and B. C. McCarthy. 2013. Long-term effects of prescribed fire and

thinning on residual tree growth in mixed-oak forests of southern Ohio.

Ecosystems 16(8):1473–1486.

Biondi, F., and F. Qeadan. 2008. A theory-driven approach to tree-ring standardization:

defining the biological trend from expected basal area increment. Tree-ring

Research 64:81–96.

Boerner, R. E. J., J. Huang, and S. C. Hart. 2008. Fire, thinning, and the carbon economy:

Effects of fire and fire surrogate treatments on estimated carbon storage and

sequestration rate. Forest Ecology and Management 255:3081–3097.

Boerner, R. E. J., J. Huang, and S. C. Hart. 2009. Impacts of fire and fire surrogate

treatments on forest soil properties: a meta-analytical approach. Ecological

Applications 19:338–358.

Boerner, R. E. J., S. J. Morris, K. L. Decker, and T. F. Hutchinson. 2003. Soil and forest

floor characteristics. Pages 47–56 E. K. Sutherland and T. F. Hutchinson, 89

editors. Characteristics of Mixed-oak Forest Ecosystems in Southern Ohio Prior

to the Reintroduction of Fire. Gen. Tech. Rep. NE-299. U.S. Department of

Agriculture, Forest Service, Northeastern Research Station.

Brose, P. H., T. M. Schuler, and J. S. Ward. 2006. Responses of oak and other hardwood

regeneration to prescribed fire: what we know as of 2005. Pages 123–135 M.

B. Dickinson, editor. Fire in Eastern Oak Forests: Delivering Science to Land

Managers. U.S. Forest Service General Technical Report NRS-P-1.

Bunn, A. G., M. Korpela, F. Biondi, F. Campelo, P. Mérian, and C. Zang. 2012. dplR:

Dendrochronology Program Library in R. R Package Version 1.5.6.

Http://CRAN.R-project.org/package=dplR

Canham, C. D., P. T. Lepage, and K. D. Coates. 2004. A neighborhood analysis of

canopy tree competition: effects of shading versus crowding. Canadian Journal of

Forest Research 787:778–787.

Canham, C. D., M. J. Papaik, M. Uriarte, W. H. McWilliams, J. C. Jenkins, and M. J.

Twery. 2006. Neighborhood analyses of canopy tree competition along

environmental gradients in New England forests. Ecological Applications

16:540–554.

Chiang, J.-M., R. W. McEwan, D. A. Yaussy, and K. J. Brown, K. 2008. The effects of

prescribed fire and silvicultural thinning on the aboveground carbon stocks and

net primary production of overstory trees in an oak-hickory ecosystem in southern

Ohio. Forest Ecology and Management 255:1584–1594. 90

Coomes, D. A., and R. B. Allen. 2007. Effects of size, competition and altitude on tree

growth. Journal of Ecology 95:1084–1097.

Easdale, T. A., R. B. Allen, D. A. Peltzer, and J. M. Hurst. 2012. Size-dependent growth

responses to competition and environment in . Forest

Ecology and Management 270:223–231.

Fiedler, C. E., K. L. Metlen, and E. K. Dodson. 2010. Restoration treatment effects on

stand structure, tree growth, and fire hazard in a ponderosa pine/douglas-fir forest

in Montana. Forest Science 56:18–31.

Fritts, H. C. 1976. Tree rings and climate. The Blackburn Press, Caldwell, New Jersey.

Grissino-Mayer, H. D. 2001. Evaluating crossdating accuracy: a manual and tutorial for

the computer program COFECHA. Tree-Ring Research 57:205–221.

Gromping, U. 2006. Relative Importance for Linear Regression in R: The Package

relaimpo. Journal of Statistical Software 17:1–26.

Hilt, D. E., E. D. Rast, and H. J. Bailey. 1983. Predicting diameters inside bark for 10

important hardwood species. USDA Forest Service, Research Paper NE-531,

Broomall, PA.

Hutchinson, T. F., E. K. Sutherland, and D. A. Yaussy. 2005. Effects of repeated

prescribed fires on the structure, composition, and regeneration of mixed-oak

forests in Ohio. Forest Ecology and Management 218:210–228.

Iverson, L. R., M. E. Dale, C. T. Scott, and A. Prasad. 1997. A GIS-derived integrated

moisture index to predict forest composition and productivity of Ohio forests

(USA). Landscape Ecology 12:331–348. 91

Iverson, L. R., T. F. Hutchinson, A. M. Prasad, and M. P. Peters. 2008. Thinning, fire,

and oak regeneration across a heterogeneous landscape in the eastern U.S.: 7-year

results. Forest Ecology and Management 255:3035–3050.

Johnson, P. S., S. R. Shifley, and R. Rogers. 2002. The Ecology and Silviculture of Oaks.

CABI Publishing, Wallingford, UK.

Kozlowski, T. T., P. J. Kramer, and S. G. Pallardy. 1991. The Physiological Ecology of

Woody Plants. Academic Press, Inc., San Diego, CA.

Kunstler, G., C. H. Albert, C. Courbaud, S. Lavergne, W. Thuiller, G. Vieilledent, N. E.

Zimmermann, and D. A. Coomes. 2011. Effects of competition on tree radial-

growth vary in importance but not in intensity along climatic gradients. Journal of

Ecology 99:300–312.

Lorimer, C. G. 1983. Tests of age-independent competition indices for individual trees in

natural hardwood stands. Forest Ecology and Management 6:343–360.

Lutz, J. A, A. J. Larson, M. E. Swanson, and J. A. Freund. 2012. Ecological importance

of large-diameter trees in a temperate mixed-conifer forest. PLoS ONE 7:e36131.

Macfarlane, D. W., and R. K. Kobe. 2006. Selecting models for capturing tree-size

effects on growth – resource relationships. Canadian Journal of Forest Research

36:1695–1704.

McDonald, E. P., E. L. Kruger, D. E. Riemenschneider, and J.G. Isebrands. 2002.

Competitive status influences tree-growth responses to elevated CO2 and O3 in

aggrading aspen stands. Functional Ecology 16:792–801. 92

McDowell, N., H. Barnard, B. J. Bond, T. Hinckley, R. M. Hubbard, H. Ishii, B. Kostner,

F. Magnani, J. D. Marshall, F. C. Meinzer, N. Phillips, M. G. Ryan, and D.

Whitehead. 2002. The relationship between tree height and leaf area: sapwood

ratio. Oecologia 132:12–20.

Peterson, D. L., S. S. Sackett, L. J. Robinson, and S.M. Haase. 1994. The effects of

repeated prescribed burning on growth. International Journal of

Wildland Fire 4:239–247.

Phipps, R. L. 1982. Comments on interpretation of climatic information from tree rings,

eastern North America. Tree-Ring Bulletin 42:11–22.

R Development Core Team. 2012. R: A language and environment for statistical

computing. R Foundation for Statistical Computing, Vienna, Austria.

Http://www.R-project.org/

Ryan, G., and R. H. Waring. 1992. Maintenance respiration and stand development in a

subalpine lodgepole pine forest. Ecology 73:2100–2108.

Ryan, M. G., N. Phillips, and B. J. Bond. 2006. The hydraulic limitation hypothesis

revisited. Plant, Cell and Environment 29:367–381.

Schwilk, D. W., J. E. Keeley, E. E. Knapp, J. McIver, J. D. Bailey, C. J. Fettig, C. E.

Fiedler, R. J. Harrod, J. J. Moghaddas, K. W. Outcalt, C. N. Skinner, S. L.

Stephens, T. A. Waldrop, D. A. Yaussy, and A. Youngblood. 2009. The national

fire and fire surrogate study: effects of fuel reduction methods on forest

vegetation structure and fuels. Ecological applications 19:285–304. 93

Shi, H., and L. Zhang. 2003. Local analysis of tree competition and growth. Forest

Science 49:938–955.

Speer, J. H. 2010. Fundamentals of Tree-ring Research. The University of Arizona Press,

Tucson, AZ.

Stokes, M. A., and T. L. Smiley. 1968. An Introduction to Tree-ring Dating. University

of Chicago Press, Chicago, IL.

Sutherland, E. K., T. F. Hutchinson, and D. A. Yaussy. 2003. Introduction, study area

description and experimental design E. K. Sutherland, and T. F. Hutchinson,

editors. Characteristics of Mixed-oak Forest Ecosystems in Southern Ohio Prior

to the Reintroduction of Fire. U.S. Department of Agriculture, Forest Service,

Northeastern Research Station General Technical Report NE-299.

Therneau, T. M., B. Atkinson, and B. D. Ripley. 2012. Rpart: Recursive Partitioning. R

package version 4.1-1.

Thorpe, H. C., R. Astrup, A. Trowbridge, and K. D. Coates. 2010. Competition and tree

crowns: A neighborhood analysis of three boreal tree species. Forest Ecology and

Management 259:1586–1596.

Trimble, G. R. 1967. Diameter increase in second-growth Appalachian hardwood stands:

A comparison of species. USDA Forest Service, Research Note, NE-75, Upper

Darby, PA.

Waldrop, T. A., D. A. Yaussy, R. J. Phillips, T. A. Hutchinson, L. Brudnak, and R. E. J.

Boerner. 2008. Fuel reduction treatments affect stand structure of hardwood 94

forests in Western North Carolina and Southern Ohio, USA. Forest Ecology and

Management 255:3117–3129.

Weber, P., H. Bugmann, P. Fonti, and A. Rigling. 2008. Using a retrospective dynamic

competition index to reconstruct forest succession. Forest Ecology and

Management 254:96–106.

Weiner, J. 1990. Asymmetric competition in plant populations. Trends in Ecology and

Evolution 5:360–364.

Wimberly, M. C., and B. B. Bare. 1996. Distance-dependent and distance-independent

models of Douglas-fir and western hemlock basal area growth following

silvicultural treatment. Forest Ecology and Management 89:1–11.

Wyckoff, P. H., and J. S. Clark. 2005. Tree growth prediction using size and exposed

crown area. Canadian Journal of Forest Research 20:13–20.

Yaussy, D. A., M. B. Dickinson, and A.S. Bova. 2004. Prescribed Surface-fire Tree

Mortality in Southern Ohio: Equations Based on Thermocouple Probe

Temperatures. Pages 67–75 D. A. Yaussy, D. M. Hix, R. P. Long, and P. C.

Goebel, editors. 14th Central Hardwood Forest Conference; 2004 March 16–19;

U.S. Department of Agriculture, Forest Service, Northeastern Research Station,

Wooster, OH.

Yoder, B. J., M. G. Ryan, R. H. Waring, A. W. Schoettle, and M. Kaufmann. 1994.

Evidence of reduced photosynthetic rates in old trees. Forest Science 40:513–527.

Zuur, A.F., E. N. Ieno, N. J. Walker, A. A. Saveliev, and G. M. Smith. 2009. Mixed

effects models and extensions in ecology with R. Springer, New York, NY. 95

Table 3.1

Variable Control Thin Thin+burn Burn value

Number of trees 84 87 94 89 - Age (yrs) 109.49 (2.53) 110.50 (3.22) 109.50 (1.66) 110.50 (2.01) 0.983 Size (cm) 42.12 (0.71) 43.27 (0.74) 44.24 (0.63) 42.45 (0.66) 0.119 2 -1 BAIb (cm yr ) 15.13 (0.58) 17.00 (0.62) 17.15 (0.87) 16.72 (0.90) 0.084 2 -1 BAIa (cm yr ) 16.33 (0.76) 23.14 (1.05) 23.55 (0.98) 20.52 (0.89) < 0.001

Table 3.2

Species/Variable (covariate) Treatment Covariate Treatment × Covariate All species Size 1, 340 < 0.001 < 0.001 0.143 25.50 Age 1, 315 0.001 0.023 0.172 7.51 White oak Size 1, 86 < 0.001 < 0.001 0.086 43.15 Age 1, 81 0.001 0.185 0.744 12.57 Chestnut oak Size 1, 73 0.377 < 0.001 0.944 15.73 Age 1, 63 0.108 < 0.001 0.367 14.05 Black oak Size 1, 74 0.014 < 0.001 0.843 23.17 Age 1, 67 0.016 0.276 0.504 8.18 Hickory Size 1, 36 0.035 < 0.001 0.788 35.80 Age 1, 35 0.082 0.178 0.423 16.60 Yellow poplar Size 1, 39 0.071 < 0.001 0.531 35.2 Age 1, 37 0.302 0.253 0.516 4.16 97

Table 3.3

Species Treatment × (%) Treatment Competition competition All species 1, 331 0.291 < 0.001 0.012 31.37 White oak 1, 84 0.020 < 0.001 0.016 44.42 Chestnut oak 1, 73 0.484 < 0.001 0.315 38.25 Black oak 1, 71 0.474 < 0.001 0.581 26.86 Hickory 1, 35 0.014 < 0.001 0.111 47.86 Yellow poplar 1, 36 0.660 < 0.001 0.148 40.79

Table 3.4

Species/Site quality class d.f. Intercept Adjusted (%) All species Mesic 1, 71 30.71 18.62 < 0.0001 Intermediate 1, 130 29.86 30.75 < 0.0001 Dry 1, 132 28.35 35.37 < 0.0001 Black oak Mesic 1, 10 33.36 7.58 0.1979 Intermediate 1, 30 31.75 40.79 < 0.0001 Dry 1, 33 29.07 34.84 0.0001 Chestnut oak Mesic 1, 10 31.90 28.94 0.0413 Intermediate 1, 33 32.03 42.89 < 0.0001 Dry 1, 32 30.33 37.38 < 0.0001 White oak Mesic 1, 16 23.02 35.30 0.0051 Intermediate 1, 36 23.68 27.96 0.0004 Dry 1, 34 26.66 44.28 < 0.0001 Hickory Mesic 1, 13 22.45 21.59 0.0462 Intermediate 1, 12 17.25 25.47 0.0379 Dry 1, 12 14.46 19.27 0.0656 Yellow -poplar Mesic 1, 14 50.18 46.44 0.0022 Intermediate 1, 11 44.95 20.48 0.0681 Dry 1, 13 37.42 40.38 0.0068

Relationships of tree size (a) and age (b) with basal area increment (BAI; 2001-2010) responses of residual trees to prescribed fire and thinning treatments in southeastern Ohio. For clarity, age axis has been scaled such that two oldest trees are not shown.

100

Variation in competitive status of individual trees across prescribed fire and thinning treatments in southeastern Ohio. Different letters indicate statistical difference at = 0.001.

101

Effect of competition on basal area increment of residual trees after prescribed fire and thinning treatments in mixed-oak forests of southeastern Ohio.

102

BAI-competition relations compared for five species in control and active treatment (prescribed fire and thinning) stands in mixed-oak forests of southeastern Ohio. *** indicates significant difference at = 0.001, ** at = 0.01, * at = 0.05, and NS is non-significant.

103

Size-dependent competition effect on basal area increment following prescribed fire and thinning treatments in forests of southeastern Ohio. Panel (a) medium-

104

Relative importance of competition, size, and age for basal area growth of residual tree species following prescribed fire and thinning treatments (data from control plots excluded from analysis) in southeastern Ohio. Bars represent means ± 95 % bootstrap confidence intervals with 1000 replications (Gromping 2006). All multiple regressions were significant at = 0.001.

105

CHAPTER 4: TREE GROWTH RESPONSE TO PRESCRIBED FIRE AND

THINNING TREATMENTS ALONG A TOPOGRAPHIC MOISTURE GRADIENT IN

MIXED-OAK FORESTS, OHIO, USA

Abstract

Soil moisture availability is an important factor that regulates forest ecosystem structure and function, yet little is known about how variations in soil moisture demand/stress influence tree growth in managed stands. This study examined the effects of microclimate and soil moisture variability on tree growth following prescribed fire and thinning in oak-dominated forests of Ohio. Increment cores from 348 trees, comprising five species, were collected from 80 0.1-ha plots distributed among four treatments

(control, thin, burn, thin+burn) in two sites. Ring widths were converted to basal area increments (BAIs). A GIS-based water balance (WB) approach was used to assess the potential evapotranspiration (PET) and moisture deficit for each tree. The integrated moisture index, also GIS-based, was further used to assess the long-term soil moisture availability across sites. Moisture demand/availability varied considerably across the landscape, with the highest PET and deficit on ridges and south-facing slopes. This variation had a considerable influence on BAI, but the effect was stronger in the control than in the managed stands, where treatment effects became the main drivers of growth.

Oaks and non-oaks showed contrasting patterns of growth in response to the moisture gradient, with the former exhibiting greater BAI on sites with intermediate moisture demand/stress whilst the latter had greater BAI on more mesic sites. Deficit and PET also interacted to influence BAI of yellow-poplar and white oak particularly in the control. 106

These results demonstrate the strong regulatory effect of the topographically-controlled soil moisture gradient on tree growth in mixed-oak forests, which can be explored to better understand community response to prescribed fire and thinning treatments.

Introduction

Soil moisture availability is a key attribute of forested landscapes that has long been recognized to influence ecosystem structure and function (Whittaker 1956, Fralish

1994, Yeakley et al. 1998). In many ecosystems, soil moisture demand and availability often vary along topographic and edaphic gradients, and constitute a key component of the forest microclimate (McCarthy et al. 1984, Iverson et al. 1997, Chen et al. 1999,

Boerner 2006). Such microclimatic changes in soil water content are driven in large part by variations in soil and air temperature, evaporation rates, solar radiation, wind speed, soil type and atmospheric moisture (Harmon et al. 1983). Researchers have also recognized the strong link between soil moisture gradient and many ecological processes or attributes including productivity (e.g., Fralish 1994, Iverson et al. 1997), species composition (Hutchinson et al. 1999, Pausas and Austin 2001), nutrient cycling and decomposition rates (Boerner et al. 2009), as well as photosynthetic capacity and tree growth (Wang et al. 2012, Anning et al. 2013). Given its pivotal role in many ecological processes and variability over space and time, knowledge of soil moisture gradient can be critical for assessing the viability of forest management regimes.

Several studies have acknowledged the interaction between soil moisture gradient and human-induced disturbances such as prescribed fire and thinning treatments (e.g.,

Iverson and Hutchinson 2002, Iverson et al. 2008), which are frequently used to reduce 107 potential wildfire hazards, and to restore historical forest structure and function (Schwilk et al. 2009, Stephens et al. 2009). As major canopy-changing disturbances, prescribed fire and thinning treatments may impose considerable influence on soil moisture demand and availability by altering the amount of precipitation intercepted by the canopy as well as plant evapotranspiration rates (Kozlowski et al. 1991, Peterson et al. 1994). Indeed,

Iverson and Hutchinson (2002) found a significant reduction in soil moisture on xeric sites following prescribed burning in oak forests of Ohio. Conversely, microclimatic changes in soil moisture conditions have been shown to influence the behavior and effects of prescribed fire in several forest ecosystems (Harmon et al. 1983, Boerner 2006,

Lafon and Quiring 2012). As a result, identifying the effects of the soil moisture gradient on residual tree growth patterns after prescribed fire and thinning is useful to understand not only the broad impacts of the treatments on forest ecosystems, but also how the system responds generally to environmental change.

Tree growth, together with regeneration and survival, is an important driver of plant community composition and dynamics, and often reflects the function, health, energy flux, productivity, and overall response of forests to environmental change (Fritts

1976, Dobbertin 2005, Fiedler et al. 2010). Growth of large residual trees following prescribed fire and thinning manipulations is important from the standpoints of wildlife conservation, carbon sequestration, and maintenance of overstory conditions in forest ecosystems (Thorpe et al. 2010, Xu et al. 2012). Given its ability to identify subtle environmental signals, dendrochronological reconstruction of tree growth can also provide useful insight into forest’s response to management disturbances and variations 108 in climate (Fritts 1976, Biondi 1999, Rubino and McCarthy 2004). Few recent studies, for example, have shown evidence of elevated tree growth and/or productivity in response to prescribed fire and thinning across stands and populations (Boerner et al.

2008, Chiang et al. 2008, Lutz et al. 2012, Anning and McCarthy 2013a). However, the influence of topographically-controlled changes in soil moisture conditions on residual tree growth in managed ecosystems, where these restoration treatments are routinely applied, remains an important research need.

Restoration is an important goal of prescribed fire and thinning management in the hardwood forests of eastern North America (Hutchinson et al. 2012). Despite their values as timber species, wildlife habitat and food, and their current dominance in the canopy, research has consistently shown that the oaks are gradually being replaced by less desirable, shade-tolerant and fire-sensitive mesophytic species such as maples

(Abrams 2003, Nowacki and Abrams 2008, Brose et al. 2013). This compositional shift has been attributed to reduced fire activity and moister climatic conditions, among other factors (Nuttle et al. 2013). In the present study, the effects of the soil moisture gradient on growth of residual trees following prescribed fire and thinning are examined to broaden understanding of the impacts of the treatments on mixed-oak forest ecosystems.

Specific objectives were: 1) to quantify the soil moisture demand and availability across the heterogeneous landscape studied; 2) to assess the effect of the soil moisture gradient on tree growth response to prescribed fire and thinning treatments; and 3) to investigate how the soil moisture gradient effect on post-treatment growth varies among species or taxonomic groups. It was hypothesized that the treatments would interact in distinct ways 109 with the soil moisture gradient to influence tree growth. As different species exhibit distinct moisture requirements (Kozlowski et al 1991), and respond differently to the treatments (Anning and McCarthy 2013a), their responses to the soil moisture gradient in the managed stands were expected to differ substantially.

Methods

This study was conducted within two of the three replicate blocks of the Ohio

Hills site used for the national Fire and Fire Surrogate (FFS) study (Schwilk et al. 2009).

The Raccoon Ecological Management Area (REMA) block is located within the Vinton

Furnace State Experimental Forest (39º12W), and the Zaleski block is within the Zaleski State Forest (39º2W), both in Vinton county, Ohio

(Figure 4.1). The Ohio Hills site lies within the unglaciated Allegheny Plateau physiographic region, and consists of upland mixed-oak (spp.) hardwood forests. The landscape is dissected into ridges, hills and valleys, with elevation ranging from 200 to 300 m (Waldrop et al. 2008). Soils are acidic and are derived primarily from sandstone, siltstone and shale (Boerner et al. 2003). Climate patterns measured at the

REMA weather station in 2003 and 2010 were generally comparable to the climate normal (1981–2010) recorded at the NOAA’s National Climatic Data Center at Carpenter

2S, OH, USA (NCDC 2012; Figure 4.2). Normal temperature ranged from -0.6 in

January to 22.5 oC in July, with an annual value of 11.4 oC. These temperature values were lower than that measured at REMA for 2003 and 2010. However, precipitation for both 2003 and 2010 (1296 mm and 1125 mm respectively) was above normal (1052 110 mm). Prior to the start of the treatments in 2000, the even-aged stands within both blocks

2 -1 were fully stocked with tree basal area ranging from 25.5 to 29.4 m ha (Waldrop et al.

2008).

Four experimental units were installed at each of the two sites, with each unit ~ 50 ha in extent and containing ten 20 × 50 m permanent plots (i.e., 2 sites × 4 experimental units × 10 plots = 80 plots total). The four treatment units are un-manipulated control, mechanical thinning (thin), prescribed burning (burn), and mechanical thinning followed by prescribed fire (thin+burn). The plots are distributed across the landscape from ridgetops to lower slopes to represent a range of IMI values (Iverson et al. 1997). The mechanical treatment focused on removal of mid-story trees (i.e., 15–30 cm diameter at breast height, DBH), and was conducted in the fall and winter of 2000–2001. This operation reduced stand basal area by ~ 30 %. Prescribed fires were conducted in the spring of 2001, and repeated in 2005 and 2010. In contrast to the relatively uniform mechanical treatment, the intensity of prescribed fire varied greatly over the years and across landscapes. At REMA for example, several plots in the burn-only and thin+burn units experienced high-intensity fires (flame lengths reaching ~ 4-5 m) in 2005, resulting in increased overstory mortality (Iverson et al. 2008, Waldrop et al. 2008).

Between fall of 2011 and winter of 2012, I extracted 696 increment cores from

348 trees, comprising five common species in the study area: white oak (), chestnut oak (), black oak (), hickories (spp., mainly

)and yellow-poplar (). To minimize noise in the data, 111 healthy-looking co-dominant trees ( 25 cm DBH) having no more than 50% branch or twig mortality or foliage discoloration, dieback and defoliation (Yaussy et al. 2004) were sampled using pre-existing FFS overstory data as a guide. Using a standard increment borer, two cores were extracted from each tree at breast height, and at roughly 180o from each other to account for variations in ring width around the circumference of the tree. To obtain at least one tree per species per plot, sampling was extended into a 20 m buffer outside the plot if necessary. In spite of this sampling effort, some species were still absent in some sampling areas. DBH of the trees were measured at the time of coring.

Cores were transported to the laboratory in paper straws and kept in a metallic case to prevent mechanical damage. The geographic location of each cored tree was captured with a Garmin GPSMAP 60Cx GPS (Garmin International Inc., KS, USA).

In the laboratory, increment cores were laid in a wooden core holder and air-dried for at least 24 hours. The cores were then sanded using a belt sander with increasingly finer grades of sandpaper (100 to 600 grit), and micro-finished with 30, 15 and 9 micron

Microfilm Sheets to reveal the ring structure (Speer 2010). Calendar dates were assigned to each ring with the help of skeleton-plots (Stokes and Smiley 1968). Ring-widths were measured to the nearest 0.01 mm using a Velmex measuring system (Velmex, Boomfield,

NY) and the standard Measure J2X program (Voortech Consulting Holderness, NH). The quality control program, COFECHA (Grissino-Mayer 2001), was used to verify crossdating and measurement accuracies. 112

To quantify the radial growth of individual trees, each ring width was converted into a basal area increment (BAI), the net increase in the total cross-sectional stem area of a tree. BAI is the preferred metric of growth for many dendrochronologists because it has been found to provide a better approximation of annual tree growth than simple ring width or stem diameter increment (Biondi and Qeadan 2008). BAI was computed using the function in “dplR”, which calculates the ring area from the bark to the pith

(Bunn et al. 2012). Computations were based on diameter inside bark of trees, derived by first estimating bark thickness of the species using available regression equations (Hilt et al. 1983). BAIs of the two cores extracted from a tree were averaged to obtain a single value per tree; thus, the individual tree constituted the sampling unit. To understand the management impact on tree growth, two 5-year periodic annual BAIs were computed for each tree by averaging the values from 2001–2005, and 2006–2010.

Variations in the soil moisture availability and demand across the studied landscape were assessed using the water balance (WB) approach (Dyer 2009) and the integrated moisture index, IMI (Iverson et al. 1997), both based on GIS (geographic information systems). Components of the WB include potential evapotranspiration

(PET), actual evapotranspiration (AET) and moisture deficit. PET is a measure of the amount of water that can be evaporated and transpired from a vegetative surface provided moisture is not limiting. AET is the actual quantity of water evaporated and transpired by a vegetative surface given availability of water, and moisture deficit is the difference 113 between PET and AET. The WB model computes monthly time-step changes in soil moisture conditions by combining climate variables (temperature, daily solar radiation, and total monthly precipitation), digital elevation model (DEM), and soil available water capacity (AWC) in GIS. The model assumes that moisture availability is dependent on precipitation and soil moisture storage, while moisture demand (PET) is controlled by temperature and solar radiation. By computing monthly moisture conditions in absolute terms, the WB approach quantifies the moisture availability and demand in a biologically meaningful manner, thus permitting comparison of modeled values across sites (Dyer

2009). Precipitation, temperature and solar radiation data were obtained from the weather

W), whilst DEM (~ 0.7 m resolution) and AWC

(1:12,000 scale) were respectively obtained from the Ohio Geographically Referenced

Information Program (OGRIP 2012) and the National Resource Conservation Service

(NRCS 2012) databases.

In line with the objective of determining their effects on post-treatment tree growth, estimates of the moisture variables were restricted to two post-treatment years:

2003 and 2010. These years were selected for modeling due to availability of reliable radiation data from the weather station at REMA; they also corresponded with the two post-treatment periods over which BAIs were computed (i.e., 2001–2005, 2006–2010).

Further, given the strong association between tree growth and growing season climate in southeastern Ohio (Sutherland 1997, Dyer 2004, Anning et al. 2013), analysis was limited to only growing season (April-September) water balance. A 10-m buffer around each sampled tree was used to extract the WB variables. 114

The IMI uses a suite of topographic (hillshade, flow accumulation, curvature) and edaphic (water holding capacity) proxies to derive a long-term relative moisture rating for sites (Iverson et al. 1997). The IMI is dimensionless, with scores ranging from 0-100

(lower scores indicating lower moisture availability), and is often used to stratify landscapes into three soil moisture classes: mesic, intermediate, and xeric (Iverson et al.

1997). The IMI grids (10-m resolution) for the two sites were obtained from L. R. Iverson

(USDA Forest Service, Northern Research Station, Delaware, OH). Index values were estimated for every tree sampled using the 10-m buffer. Both the IMI and WB approach are commonly used in eastern hardwood forests to predict several ecological processes including productivity and species composition (Iverson et al. 1997, Dyer 2009), but they have never been related to tree growth in the same study. Their use in this study was to complement each other. In addition to modeling the moisture status, the physiographic position of each sampled tree was estimated based on its topographic position and aspect.

Consequently, four physiographic positions were designated for all trees: ridge, SE-S-SW

(south-facing) slope, NW-N-NE (north-facing or sheltered) slope, and valley.

Differences inPET and moisture deficit and IMI among the four physiographic positions were evaluated using analysis of variance (ANOVA). Multiple comparison tests

(Tukey’s Honestly Significant Difference; HSD) were then performed to evaluate the differences among the physiographic positions. A cursory examination of the water balance data showed virtually no moisture deficit in 2003 (data not shown). As a result, the moisture variables for this year were excluded from all subsequent analyses. To 115 assess the effect of the moisture demand/stress on BAI responses of trees to the treatments, PET, deficit and IMI were independently categorized into low, intermediate and high classes in GIS. Breakpoints for this classification were obtained using the respective means and standard deviations (i.e., mean ± 0.5 standard deviations) of the moisture variables computed over the landscape within each of the study sites. Variations in BAI among the three classes of each moisture variable were assessed using ANOVA, followed by post-hoc multiple test with Tukey’s HSD.

To assess the variations in tree growth responses of different taxa or species to the moisture gradients across treatments, tree data were collapsed into two groups (oaks and non-oaks) due to the small sample size for some species. ANOVA was then used to assess the effect of the moisture gradients on the variations in growth responses of the two taxonomic groups. Finally, the interactive effects of PET and deficit on post- treatment basal area growth of yellow-poplar and white oak within the control and the manipulated stands (data from the three active treatments combined) were evaluated using scatter plots. To accomplish this, trees from each of the two experimental units

(i.e., control vs. treatment) were rank ordered by BAI and divided into two groups at their respective means. White oak and yellow-poplar were considered because an earlier analysis indicated that they seemed to perform best at the opposite ends of the soil moisture gradient (Anning and McCarthy 2013b). All analyses were done using R (R

Development Core Team 2013), and at 5% significance level. 116

Results

The soil moisture demand and availability varied considerably across the studied sites (Figure 4.3). The range of annual moisture deficit estimated for Zaleski was 0–73 mm, whilst that of PET was 323–822 mm. Annual moisture deficit (0–87 mm) and PET

(132–821 mm) for REMA were generally comparable to the values observed at Zaleski.

Analysis of the data by physiographic position showed considerable variations ( <

0.001) in seasonal PET, with greater values on south-facing slopes than on north-facing slopes (Figure 4.4a). PET was significantly higher on south-facing slopes (mean = 655 mm) compared to the other physiographic positions (mean PET < 632 mm). No statistical difference in PET was found between the north-facing slopes and valleys ( > 0.05). As with PET, the seasonal moisture deficit was greater on south-facing slopes (mean = 53.5 mm) than on other topographic positions, which had mean deficits not exceeding 42 mm

(Figure 4.4b). IMI scores were highest for the sheltered slopes (mean = 50.2) and lowest for the south-facing slopes (mean = 21.1; Figure 4.4c). IMI for the south-facing slopes was much lower ( < 0.05) than that of the ridges (30.7).

Variations in the soil moisture demand and availability strongly influenced BAI in the control plots but less so in the managed stands (Figure 4.5). With data from all species combined, BAI decreased with increasing moisture deficit, and increased with

IMI in the control unit ( < 0.05; Figure 4.5a). The pattern of BAI response to PET was similar to that of moisture deficit but no statistical differences ( > 0.05) were found. 117

BAI averaged 13.6 cm2 yr-1 for high-PET sites, 13.2 cm2 yr-1 for high-deficit sites, and

12.2 cm2 yr-1 for low-IMI sites. Values for sites with low-to-intermediate PET and deficit ranged from 17.7 to 18.9 cm2 yr-1, whereas sites with intermediate-to-high IMI scores recorded mean BAI of 18.1 cm2 yr-1 or greater.

The pattern of tree growth response to the moisture demand/stress in the managed stands varied somewhat among the treatment types (Figure 4.5b–d). The soil moisture gradient had a marginal effect on tree growth in the thin-only stands, where BAI rates increased slightly on sites with low PET and deficit, or high IMI score (Figure 4.5b).

Apparently, BAI rates for the thin+burn sites were higher at the intermediate positions

(averaged 25.2, 27.3 and 29.0 cm2 yr-1 correspondingly for PET, deficit and IMI) than at the extremes (< 22.8 cm2 yr-1) of the soil moisture demand/stress continuum. However, this effect was only statistically significant ( < 0.05) for moisture deficit (Figure 4.5c).

Similarly, tree growth responded weakly to the soil moisture gradient in the burn-only unit (Figure 4.5d), with a small increase in BAI as soil moisture demand/stress increased.

Different taxonomic or species groups showed detectable variations in their growth responses to moisture demand/stress, particularly in the control plots (Figure 4.6).

Growth rates of the oaks in the control plots were consistently higher (< 0.05) on sites with intermediate deficit (mean = 19.6 cm2 yr-1) and IMI (mean = 22.2 cm2 yr-1) compared to the extreme ends (mean = 13.8–15.6 cm2 yr-1) of the moisture gradient

(Figure 4.6a). Tree growth response to PET in the control plots showed a similar pattern 118 as those of deficit and IMI, but the relationship was statistically insignificant (> 0.05).

Growth of oak trees in the three manipulated stands did not show any significant response to the moisture gradient (Figure 4.6b). Yellow-poplar and hickories responded in a similar fashion to the moisture gradients as the oaks in the managed stands but their responses in the control unit were strikingly different (Figures 4.6c and d). These species recorded significantly higher growth rates (mean BAI > 25 cm2 yr-1) on more mesic sites than on drier sites (mean BAI approximately 10 cm2 yr-1).

Seasonal moisture deficit and PET also showed a strong interactive effect on BAI of yellow-poplar and white oak trees particularly in the control plots (Figures 4.7 and

4.8). In the control stand, yellow-poplar trees growing on dry sites characterized by high

PET and deficit exhibited limited BAI (averaging 9.2 cm2 yr-1), whereas those found on sites with low moisture demand and deficit showed substantially higher growth rates

(32.8 cm2 yr-1). These two clusters of yellow-poplar trees were clearly separated along the soil moisture continuum (Figure 4.7a). This link, however, disappeared within the manipulated stands (Figure 4.7b), as no clear separation was found between trees with high growth rates (34.6 cm2 yr-1) and those with low growth rates (15.3 cm2 yr-1). White oak trees portrayed a growth-moisture relation that contrasted with yellow-poplar; growth rates of white oak were higher on sites with intermediate-to-high PET and deficit than on those with high moisture content or low moisture demand (Figure 4.8a). However, there was a considerable overlap between these two tree clusters along the moisture gradient.

As with yellow-poplar, the link between moisture gradient and BAI of white oak broke down completely in the manipulated stands (Figure 4.8b). 119

Discussion

Overall, the results suggest marked variations in soil moisture demand and availability across the landscape studied, despite the generally moist regional climate.

However, these local variations influenced tree growth to a greater extent in the control stands than in the managed stands, where the pattern of response varied slightly among treatments. Oaks and non-oaks exhibited contrasting growth responses to the soil moisture gradient. An interactive effect of moisture deficit and PET on tree growth was discernible in the control stands that delineated the sites based on how well they supported the growth of particular species. In general, the variations in PET, deficit and

IMI across these physiographic positions corroborate the view that topography and edaphic factors control soil moisture variability (Iverson et al. 1997), and thus contribute markedly to the microclimate that regulates most ecological processes in these mixed-oak forests.

Tree growth, like most physiological processes, is widely acknowledged to respond to variations in soil moisture stress/demand along with several other environmental factors. Under conditions of moisture stress, many physiological processes in the tree become impaired (e.g., stomatal conductance and photosynthetic rate decrease), thereby limiting growth (Kozlowski et al. 1991). Similarly, trees exposed to direct sunlight or solar radiation, such as those found on south-facing slopes, also require more soil water to meet their evaporative demands (Lambers et al. 2008). The general decline in growth rates with increasing moisture stress/demand in the control sites clearly reflects these responses. On the contrary, the weak growth response to the soil moisture 120 gradient in the managed stands suggests that treatment effects, including reduced competition among trees (Anning and McCarthy 2013b), increased availability of nutrients, and increased mineralization rates (Boerner et al. 2009), became the main drivers of growth in the manipulated units. With these major drivers operating concurrently, the influence of the soil moisture gradient on tree growth may be diminished, though its role in species recruitment/establishment in manipulated stands is not in doubt (Albrecht and McCarthy 2006, Iverson et al. 2008). A differential effect of moisture gradient on tree growth and species recruitment/establishment may be expected due to differences in sensitivity of these two processes to resource availability. Tree growth is less likely to be affected by soil moisture changes within the top few centimeters of the soil to which seedlings are most sensitive.

An increase in soil moisture content associated with the prescribed fire and thinning treatments (Kozlowski et al. 1991, Peterson et al. 1994) may explain in part the weak response of tree growth to the soil moisture gradient in the managed stands.

However, this effect conflicts somewhat with an earlier finding indicating significant reduction in soil moisture on xeric sites following a low-intensity prescribed burning at the present study site (Iverson and Hutchinson 2002). Moreover, the specific effect of canopy disturbance on the soil moisture availability and demand is difficult to infer from this study as neither of the two modeling approaches used integrates canopy disturbance.

The different patterns of tree growth response to the soil moisture gradient across the managed stands likely reflect the distinct behaviors or effects of the various treatments. Growth-moisture relation in the thin-only treatment somewhat mirrored that 121 of the control (i.e., higher growth rate on moist sites), presumably due to the relative uniformity of the mechanical thinning across sites, which clearly favored the non-oaks that grow on intermediate to mesic sites. By contrast, BAI rates of trees from the burn- only stands increased slightly with soil moisture demand/stress, probably due to the greater intensity of fire on the drier sites (Iverson et al. 2004). This increase may be attributed to the larger positive effect of fire-caused mortality on the residual oak species, which tend to dominate intermediate to xeric sites (Iverson et al. 1997). The relatively higher growth rate on sites with intermediate moisture levels in the thin+burn unit suggests the treatment confers equal benefits to the oaks and the non-oaks. These results demonstrate that mechanical treatments may not always be a suitable substitute for prescribed fire as noted by some previous investigators (e.g., Schwilk et al. 2009).

The contrasting growth responses exhibited by the oaks and the non-oaks (yellow- poplar and hickories) most likely reflects differences in the morphological and physiological adaptations of these two groups of species to soil moisture stress. Oak trees, unlike the yellow-poplar and hickories, possess deep root systems and high stomatal density among other eco-physiological adaptations, which allows them to thrive in more xeric sites (Abrams 2003). Further, a study by Querejeta et al. (2009) involving the California live oak () showed that oak trees, with their deep roots, have the ability to shift the relative dominance of ecto- and arbuscular-mycorrhizal symbionts depending on soil moisture levels, enabling these trees to perform well in water-limited sites. These contrasting growth responses provide an indication of how different species may react to future climate change. The strong limitation of yellow-poplar growth on 122 sites with high moisture stress/demand, and to a lesser extent, white oak growth on more mesic sites within the control stands supports this hypothesis. Clearly, yellow-poplar is more sensitive to the topographically-controlled moisture gradient than the other species studied.

Based on compositional analysis of a secondary forest in southern Ohio, Iverson et al. (1997) predicted a predominance of oak species on sites with intermediate moisture stress/demand, whilst yellow-poplar was expected to dominate more mesic sites in future forest stands. The current finding indicating consistently higher growth rates of white oak trees on sites with intermediate-high PET and deficit supports this prediction.

Additionally, the results show that tree growth may be maximized with the combination of prescribed fire and thinning on landscape positions with intermediate moisture (i.e., ridges, south-facing slopes). Since fire is commonly used to enhance the competitive status of oak regeneration, it is necessary to carefully balance mechanical thinning with prescribed fire on these intermediate moisture sites to prevent competitive advantage from switching to yellow-poplar and other non-oak species. Such a balance is critical to sustaining growth and productivity of oaks in the face of extreme climate and the ongoing compositional shift across much of eastern hardwood forests (Abrams 2003).

In conclusion, the present findings demonstrate that variations in soil moisture across heterogeneous landscapes can have an important regulatory effect on tree growth; however, management disturbances may trump this effect. Different species may exhibit contrasting growth responses to the soil moisture gradient due to differences in their morphological and physiological adaptations to moisture stress. Results also show that 123 soil moisture gradient and prescribed fire and thinning treatments may interact in complex ways to influence tree growth. Thus, analyzing tree growth response to the soil moisture gradient in relation to prescribed fire and thinning can enrich our understanding of the impacts of these management practices on forest ecosystems.

Literature Cited

Abrams, M. D. 2003. Where has all the white oak gone? BioScience 53:927–939.

Albrecht, M. A., and B. C. McCarthy. 2006. Effects of prescribed fire and thinning on

tree recruitment patterns in central hardwood forests. Forest Ecology and

Management 226:88–103.

Anning, A. K., and B. C. McCarthy. 2013a. Long-term effects of prescribed fire and

thinning on residual tree growth in mixed-oak forests of southern Ohio.

Ecosystems 16(8):1473–1486.

Anning, A. K., and B. C. McCarthy. 2013b. Competition, size and age affect tree growth

response to fuel reduction treatments in mixed-oak forests of Ohio. Forest

Ecology and Management 307:74–83.

Anning, A. K., D. L. Rubino, E. K. Sutherland, and B. C. McCarthy. 2013.

Dendrochronological analysis of white oak growth patterns across a topographic

moisture gradient in southern Ohio. Dendrochronologia 31:120–128.

Biondi, F. 1999. Comparing tree-ring chronologies and repeated timber inventories as

forest monitoring tools. Ecological Applications 9:216–227. 124

Biondi, F., and F. Qeadan. 2008. A theory-driven approach to tree-ring standardization:

defining the biological trend from expected basal area increment. Tree-Ring

Research 64:81–96.

Boerner, R. E. J. 2006. Soil, fire, water, and wind: how the elements conspire in the forest

context. Pages 104–122 M. B. Dickinson, editor. Fire in eastern oak forests:

delivering science to land managers. Proceedings of a Conference, November 15–

17, 2005. Fawcett Center, The Ohio State University, Columbus Ohio.

Boerner, R. E. J., J. Huang, and S. C. Hart. 2009. Impacts of fire and fire surrogate

treatments on forest soil properties: a meta-analytical approach. Ecological

applications 19:338–358.

Boerner, R. E. J., J. Huang, and S. C. Hart. 2008. Fire, thinning, and the carbon economy:

Effects of fire and fire surrogate treatments on estimated carbon storage and

sequestration rate. Forest Ecology and Management 255:3081–3097.

Boerner, R. E. J., S. J. Morris, K. L. Decker, and T. F. Hutchinson. 2003. Soil and forest

floor characteristics. Pages 47–56 E. K. Sutherland, and T. F. Hutchinson,

editors. Characteristics of mixed-oak forest ecosystems in southern Ohio prior to

the reintroduction of fire. Gen. Tech. Rep. NE-299. U.S. Department of

Agriculture, Forest Service, Northeastern Research Station.

Brose, P. H., D. C. Dey, R. J. Phillips, and T. A. Waldrop. 2013. A meta-analysis of the

fire-oak hypothesis: does prescribed burning promote oak reproduction in eastern

North America? Forest Science 59:322–334. 125

Bunn, A. G., M. Korpela, F. Biondi, F. Campelo, P. Mérian, and C. Zang. 2012. dplR:

Dendrochronology Program Library in R. R Package Version 1.5.6.

Http://CRAN.R-project.org/package=dplR

Chen, J., S. C. Saunders, T. R. Crow, R. J. Naiman, D. Kimberley, G. D. Mroz, B. L.

Brookshire, J. F. Franklin, and K. D. Brosofske. 1999. Microclimate in forest

ecosystem and landscape ecology. BioScience 49:288–297.

Chiang, J.-M., R. W. McEwan, D. A. Yaussy, and K. J. Brown. 2008. The effects of

prescribed fire and silvicultural thinning on the aboveground carbon stocks and

net primary production of overstory trees in an oak-hickory ecosystem in southern

Ohio. Forest Ecology and Management 255:1584–1594.

Dobbertin, M. 2005. Tree growth as indicator of tree vitality and of tree reaction to

environmental stress: a review. European Journal of Forest Research 124:319–

333.

Dyer, J. M. 2004. A water budget approach to predicting tree species growth and

abundance, utilizing paleoclimatology sources. Climate Research 28:1–10.

Dyer, J. M. 2009. Assessing topographic patterns in moisture use and stress using a water

balance approach. Landscape Ecology 24:391–403.

Fiedler, C. E., K. L. Metlen, and E. K. Dodson. 2010. Restoration treatment effects on

stand structure, tree growth, and fire hazard in a ponderosa pine/douglas-fir forest

in Montana. Forest Science 56:18–31.

Fralish, J. S. 1994. The effect of site environment on forest productivity in the Illinois

Shawnee Hills. Ecological Applications 4:134–143. 126

Fritts, H. C. 1976. Tree rings and climate. The Blackburn Press, Caldwell, New Jersey.

Grissino-Mayer, H. D. 2001. Evaluating crossdating accuracy: a manual and tutorial for

the computer program COFECHA. Tree-Ring Research 57:205–221.

Harmon, M. E., S. P. Bratton, and P. S. White. 1983. Disturbance and vegetation

response in relation to environmental gradients in the Great Smoky Mountains.

Vegetatio 55:129–139.

Hilt, D. E., E. D. Rast, and H. J. Bailey. 1983. Predicting diameters inside bark for 10

important hardwood species. USDA Forest Service, Research Paper NE-531,

Broomall, PA.

Hutchinson, T. F., R. E. J. Boerner, L. R. Iverson, S. Sutherland, and K. Sutherland.

1999. Landscape patterns of understory composition and richness across a

moisture and nitrogen forests and nitrogen mineralization gradient in Ohio. Plant

Ecology 144:177–189.

Hutchinson, T. F., R. P. Long, J. Rebbeck, E. K. Sutherland, and D. A. Yaussy. 2012.

Repeated prescribed fires alter gap-phase regeneration in mixed-oak forests.

Canadian Journal of Forest Research 314:303–314.

Iverson, L. R., M. E. Dale, C. T. Scott, and A. Prasad. 1997. A GIS-derived integrated

moisture index to predict forest composition and productivity of Ohio forests

(USA). Landscape Ecology 12:331–348.

Iverson, L. R., and T. F. Hutchinson. 2002. Soil temperature and moisture fluctuations

during and after prescribed fire in mixed-oak forests, USA. Natural Areas Journal

22:296–304. 127

Iverson, L. R., T. F. Hutchinson, A. M. Prasad, and M. P. Peters. 2008. Thinning, fire,

and oak regeneration across a heterogeneous landscape in the eastern U.S.: 7-year

results. Forest Ecology and Management 255:3035–3050.

Iverson, L. R., D. A. Yaussy, J. Rebbeck, T. F. Hutchinson, R. P. Long, and A. M.

Prasad. 2004. A comparison of thermocouples and temperature paints to monitor

spatial and temporal characteristics of landscape-scale prescribed fires.

International Journal of Wildland Fire 13:311–322.

Kozlowski, T. T., P. J. Kramer, and S. G. Pallardy. 1991. The physiological ecology of

woody plants. Page 657. Academic Press, Inc., San Diego, CA.

Lafon, C. W., and S. M. Quiring. 2012. Relationships of fire and precipitation regimes in

temperate forests of the eastern United States. Earth Interactions 16:1–15.

Lambers, H., F. S. Chapin III, and T. L. Pon. 2008. Plant Physiological Ecology. Pages

321–374, 2nd edition. Springer Science+Business Media, LLC, New York, USA.

Lutz, J. A., A. J. Larson, M. E. Swanson, and J. A. Freund. 2012. Ecological importance

of large-diameter trees in a temperate mixed-conifer forest. PLoS ONE 7:e36131.

McCarthy, B. C., T. I. Vierheller, and W. A. Wistendahl. 1984. Species ordination of

upper slope oak-hickory stands of southeastern Ohio. Journal of Torrey Botanical

Club 111:56–60.

National Climatic Data Center. 2012. Climate data online. National Oceanic and

Atmospheric Administration (NOAA), Asheville, NC, USA.

Http://www.ncdc.noaa.gov 128

National Resources Conservation Service. 2012. Soil Data Mart (SSURGO).

Http://soildatamart.nrcs.usda.gov

Nowacki, G. J., and M. D. Abrams. 2008. The demise of fire and mesophication of

forests in the eastern United States. Bioscience 58:123–138.

Nuttle, T., A. A. Royo, M. B. Adams, and W. P. Carson. 2013. Historic disturbance

regimes promote tree diversity only under low browsing regimes in eastern

deciduous forest. Ecological Monographs 83:3–17.

Ohio Geographically Referenced Information Program. Geodata Distribution.

Http://ogrip.oit.ohio.gov

Pausas, J. G., and M. P. Austin. 2001. Patterns of plant species richness in relation to

n appraisal. Journal of Vegetation Science 12:153–166.

Peterson, D. L., S. S. Sackett, L. J. Robinson, and S. M. Haase. 1994. The effects of

repeated prescribed burning, on pinus ponderosa growth. International Journal of

Wildland Fire 4:239–247

Querejeta, J. I., L. M. Egerton-Warburton, and M. F. Allen. 2009. Topographic position

modulates the mycorrhizal response of oak trees to interannual rainfall variability.

Ecology 90:649–662.

R Development Core Team. 2013. R: A language and environment for statistical

computing. R Foundation for Statistical Computing, Vienna, Austria.

Http://www.R-project.org/

Rubino, D. L., and B. C. McCarthy. 2004. Comparative analysis of dendroecological

methods used to assess disturbance events. Dendrochronologia 21:97–115. 129

Schwilk, D. W., J. E. Keeley, E. E. Knapp, J. McIver, J. D. Bailey, C. J. Fettig, C. E.

Fiedler, R. J. Harrod, J. J. Moghaddas, K. W. Outcalt, C. N. Skinner, S. L.

Stephens, T. A. Waldrop, D. A. Yaussy, and A. Youngblood. 2009. The national

fire and fire surrogate study: effects of fuel reduction methods on forest

vegetation structure and fuels. Ecological applications 19:285–304.

Speer, J. H. 2010. Fundamentals of Tree-ring Research. The University of Arizona Press,

Tucson, AZ.

Stephens, S. L., J. J. Moghaddas, C. Edminster, C. E. Fiedler, S. Haase, M. Harrington, J.

E. Keeley, E. E. Knapp, J. D. McIver, K. Metlen, C. N. Skinner, and A.

Youngblood. 2009. Fire treatment effects on vegetation structure, fuels, and

potential fire severity in western U.S. forests. Ecological applications 19:305–20.

Stokes, M. A., and T. L. Smiley. 1968. An Introduction to Tree-ring Dating. University

of Chicago Press, Chicago, IL.

Sutherland, E. K. 1997. History of fire in a southern Ohio second-growth mixed-oak

forest. Pages 172–183 S. G. Pallardy, R. A. Cecich, H. E. Garrett, and P. S.

Johnson, editors. Proceedings of the 11th Central Hardwood Forest Conference

23–26 March 1997. Columbia, Missouri. US Department of Agriculture Forest

Service, Gen. Tech. Rep. NE-188.

Thorpe, H. C., R. Astrup, A. Trowbridge, and K. D. Coates. 2010. Competition and tree

crowns: a neighborhood analysis of three boreal tree species. Forest Ecology and

Management 259:1586–1596. 130

Waldrop, T. A., D. A. Yaussy, R. J. Phillips, T. A. Hutchinson, L. Brudnak, and R. E. J.

Boerner. 2008. Fuel reduction treatments affect stand structure of hardwood

forests in Western North Carolina and Southern Ohio, USA. Forest Ecology and

Management 255:3117–3129.

Wang, M., S. Shi, F. Lin, Z. Hao, P. Jiang, and G. Dai. 2012. Effects of soil water and

nitrogen on growth and photosynthetic response of Manchurian ash (

) seedlings in northeastern China. PLoS ONE 7:e30754.

Whittaker, R. H. 1956. Vegetation of the Great Smoky Mountains. Ecological

Monographs 26:1–80.

Xu, C.-Y., M. H. Turnbull, D. T. Tissue, J. D. Lewis, R. Carson, W. S. F. Schuster, D.

Whitehead, A. S. Walcroft, J. Li, and K. L. Griffin. 2012. Age-related decline of

stand biomass accumulation is primarily due to mortality and not to reduction in

NPP associated with individual tree physiology, tree growth or stand structure in a

Quercus-dominated forest. Journal of Ecology 100:428–440.

Yaussy, D. A., M. B. Dickinson, and A. S. Bova. 2004. Prescribed surface-fire tree

mortality in Southern Ohio: equations based on thermocouple probe temperatures.

Pages 67–75 D. A. Yaussy, D. M. Hix, R. P. Long, and P. C. Goebel, editors.

14th Central Hardwood Forest Conference; 2004 March 16–19; U.S. Department

of Agriculture, Forest Service, Northeastern Research Station, Wooster, OH.

Yeakley, J. A., W. J. Swank, L. W. Swift, G. M. Hornberger, and H. H. Shugart. 1998.

Soil moisture gradients and controls on a southern Appalachian hillslope from

drought through recharge. Hydrology and Earth System Sciences 2:41–49. 131

Location of Vinton County, Ohio, where the study was conducted.

132

Variations in precipitation and temperature at Raccoon Ecological Management Area (REMA) in 2003 and 2010 compared to normal climate (1981-2010). Normal climate conditions were obtained from the NOAA’s National Climatic Data Center (NCDC, 2012) at Carpenter 2S, OH, USA.

133

Variations in annual PET and deficit (estimated for 2010) across parts of the landscape at the Raccoon Ecological Management Area (REMA) and the Zaleski State Forest, both in Vinton County, Ohio. Also shown are the topography and the locations of some of the sampled trees. 134

Variations in seasonal potential evapotranspiration (PET), seasonal deficit and integrated moisture index (IMI) across physiographic positions in mixed-oak forests of Ohio. Bars represent mean values ± standard errors for individual trees. Note that the y- axis scale for panel (A) does not start from the origin. Different letters denote significant difference .

135

Effects of seasonal potential evapotranspiration (PET), seasonal deficit (DEF), and the integrated moisture index (IMI) on basal area increment (BAI) responses of trees to prescribed fire and thinning treatment manipulations. BAI was computed for the 2006- 2010 growth period (6-10 years post-treatment) while PET and DEF were modeled for 2010. Bars represent mean values ± standard errors for individual trees. Different letters on bars indicate significant difference in BAI among the moisture classes.

136

Effects of seasonal potential evapotranspiration (PET), seasonal deficit (DEF), and the integrated moisture index (IMI) on basal area increment (BAI; 2006-2010) responses of oaks (white oak, black oak and chestnut oak) and non-oaks (yellow-poplar and hickories) to prescribed fire and thinning treatment manipulations. PET and DEF were modeled for 2010. Bars represent mean values ± standard errors for individual trees. Different letters on bars indicate significant difference in BAI among the moisture classes.

137

Effects of seasonal moisture deficit (DEF) and seasonal potential evapotranspiration (PET) on basal area increment (BAI; computed for 2006-2010) of yellow-poplar in control and manipulated stands within the mixed-oak forests of southern Ohio. DEF and PET were modeled for the year 2010.

138

Effects of seasonal moisture deficit (DEF) and seasonal potential evapotranspiration (PET) on basal area increment (BAI; computed for 2006-2010) of white oak in control and manipulated stands within the mixed-oak forests of southern Ohio. DEF and PET were modeled for the year 2010.

139

CHAPTER 5: STABLE-CARBON ISOTOPE COMPOSITION OF WHITE OAK

TREES IN RELATION TO SOIL MOISTURE STRESS AND RESTORATION

MANAGEMENT IN CENTRAL HARDWOOD FORESTS

Abstract

Stable-carbon isotope analysis is commonly used to elucidate plant response to past environmental disturbances. Here, the variability of stable-carbon isotope

13C) of white oak (L.) trees over time and space was explored to better understand the physiological mechanisms influencing the response of this species to prescribed fire and thinning management in the hardwood forests of southeastern Ohio. Tree-ring samples (160 5-yr segments spanning the period 1991–

2010; = 40 trees) were obtained from thin+burn and control plots representing two soil moisture classes and two topographic positions. Whole-wood samples were analyzed for

13C using an elemental analyzer linked to a dual gas isotope ratio mass spectrometer.

Results indicated an initial increase in percent carbon of the wood samples following the treatment but there was no significant time or treatment effect. 13C signature of

13 white oak (ca. -26.63‰) was within the typical range of most C3 plants. C values declined considerably with time, with tree-ring segments for the interval 2006–2010 being more depleted in 13C relative to the pre-treatment samples, and indicating a decline in water-use efficiency over the long term. Prescribed fire and thinning had little influence on the 13C values of white oak (mean values for the thin+burn and control, respectively, were -26.55‰ and -26.71‰), presumably due to the countervailing effects of increased moisture availability and photosynthetic capacity associated with the 140 treatments. Soil moisture stress/availability and topographic position did not strongly influence 13C values of white oak. In general, the results demonstrate that macroclimate has much greater influence on photosynthetic water-use efficiency of white oak than microclimate and prescribed fire and thinning treatments.

Introduction

Stable-carbon isotope in tree rings is a widely-recognized source of historical information on plant response to environmental change (Leavitt and Long 1982, West et al. 2006, Werner et al. 2012). The carbon isotope composition of a plant provides a good indication of the internal CO2 concentration of a leaf during assimilation, and is increasingly used by ecologists and ecophysiologists to infer plant carbon balance and water-use efficiency(Farquhar et al. 1989, Brooks and Mitchell 2011). It is also used commonly to reconstruct climate influence on forest ecosystems (Au and Tardif 2009).

The technique is considered to provide a more integrated measure of carbon balance and water-use efficiency over a longer time interval than the traditional instantaneous measurements of gas-exchange (Beerling and Woodward 1995, Dawson et al. 2002,

Lambers et al. 2008). Thus, assessing the carbon isotope ratios in tree rings may offer new insights into the physiological responses of forests to environmental stress.

Farquhar et al. (1982) were the first to describe the biochemical basis for the discrimination by plant materials against the heavier stable-carbon isotope (13C).

13 According to these authors, the carbon isotope ratio in C3 pl C) is lower than that

13 Catm (-8‰) due to fractionation associated with differential

13 12 diffusivities of CO2 versus CO2 in air and stomata (a = 4.4‰) and the isotopic effect 141

12 caused by the preference of ribulose-1,5-biphosphate carboxylase (Rubisco) for CO2

13 13 over CO2 (b = 27‰). Farquhar et al. (1982) also demonstrated that C is positively related to plant water-use efficiency (WUE), defined as the ratio of net photosynthesis to transpiration (A/E), as both are controlled by the intercellular CO2. These relations were expressed as follows:

13 13 Catm – a – (b-a)(Ci/Ca) [1]

WUE = A/E = (Ca – Ci)/1.6v [2] where Ci/Ca is the ratio of the intercellular (Ci) to atmospheric CO2 (Ca), 1.6 is the ratio of diffusivities of water vapor and CO2 in air, and v is vapor difference.

Consistent with the above theory, many studies have documented the response of

13C to several environmental factors, particularly precipitation, temperature, relative humidity, and atmospheric CO2 (McCarroll and Loader 2005, Marshall et al. 2007,

Gebrekirstos et al. 2009). Another factor that greatly influences 13C values is soil moisture stress. Often associated with decreased precipitation and increased temperature, soil moisture stress decreases stomatal conductance, leading to 13C enrichment in plant tissue (Leavitt and Long 1982). Furthermore, 13C in woody plant materials varies with altitude as a function of leaf mass per unit area (LMA), leaf nitrogen content, stomatal conductance, stomatal density, and soil moisture stress (Van de Water et al. 2002, Ma et al. 2005). Topographic influences, related to differences in solar radiation or heat loads, and moisture stress, have also been observed (Garten and Taylor 1992). The variation in

13C with altitude has also been attributed to the effect of temperature on gas exchange, 142 and the inhibitory effect of decreasing oxygen partial pressure on photorespiration leading to a reduction in intercellular CO2 concentration (see Marshall et al. 2007).

Besides macroclimate, researchers have noted strong correlations between past forest management practices and 13C values, although responses are less consistent across studies and appear to vary with species (e.g., Brooks and Mitchell 2011). For example, McDowell et al. (2003), observed about 0.89‰ increase in discrimination against 13C in tree rings of old ponderosa pine () trees following stand density reductions, and attributed this to increased soil moisture availability or increased stomatal conductance as a result of the thinning treatment. Brooks and Mitchell (2011) found a significant positive effect of the combined treatment of thinning and fertilization on 13C of Douglas-fir () trees. Similarly, Warren et al. (2001) reported an increased in 13C of conifers following heavy thinning and concluded that this could be due to increased light and nutrient availability. However, Martín-Benito et al. (2010) found no changes in black pine (). Clearly, tree-ring 13C variability has become a useful proxy for the study of forest carbon dynamics and water- use efficiency in relation to climate change and management disturbances.

In eastern North America, prescribed fire and thinning are commonly used as forest management tools to reduce potential wildfire effects, alter stand structure, and promote oak regeneration (Albrecht and McCarthy 2006, Hutchinson et al. 2012). In part, these treatments have become widespread as a means to curb the ongoing composition shift in the central hardwood forests, where the oaks are being supplanted by less desirable shade-tolerant, fire-sensitive mesophytic species (Abrams 2003). Moreover, 143 concerns about rising atmospheric CO2 levels have led to a growing interest in how these treatments influence forest carbon dynamics (Brooks and Mitchell 2011). Consequently, several recent studies have examined the impacts of the treatments on ecosystem attributes including tree growth (Thorpe et al. 2007, Anning and McCarthy 2013a), above-ground carbon stocks (Boerner et al. 2008, Chiang et al. 2008, Hurteau and North

2009), and site quality (Gough et al. 2007). The modulating roles of several biotic and abiotic factors such as competition (Anning and McCarthy 2013b, Fiedler et al. 2010,

Thorpe et al. 2010) and soil moisture gradient (Anning et al. ) have also been investigated. Nevertheless, little is known about the long-term physiological response of trees to the prescribed fire and thinning management. The goal of this study was to understand the physiological mechanisms underlying white oak response to prescribed fire and thinning disturbances over time and space in the mixed-oak forests of southeastern Ohio. The specific objectives of the study were: (a) to identify the 13C signature of white oak and investigate its variability over time in response to prescribed fire and thinning disturbances, and (b) to determine the effects of soil moisture stress/availability as well as slope-aspect position on the 13C values in these stands. It was expected that the 13C value of white oak trees would vary with time, altitude, slope- aspect positions, and between treatment and control stands.

Methods

This study was conducted at the Raccoon Ecological Management Area (REMA),

144

Ohio. These study areas are part of the three replicate blocks within the Ohio Hills study site used for the national Fire and Fire Surrogate (FFS) study. The Ohio Hills site lies within the unglaciated Allegheny Plateau physiographic region. The landscape is heterogeneous and characterized by ridges, hills and valleys (Hutchinson et al. 2005), with elevation ranging from 200 to 300 m (Waldrop et al. 2008). Soils are mainly acidic and are derived primarily from sandstone, siltstone and shale (Boerner et al. 2003).

Annual precipitation and temperature average 1024 mm and 11.3 oC, respectively

(NCDC 2012). The vegetation is classified as upland mixed-oak (spp.) hardwood forests (Iverson et al. 2008). White oak (), chestnut oak (), black oak (), and hickories ( spp.) dominate the canopy, while maples

( spp.) and yellow-popular () are the most abundant mid-story and understory species (Waldrop et al. 2008). Prior to the start of the treatments in 2000, the even-aged stands within both blocks were fully stocked with tree basal area ranging

2 -1 from 25.5 to 29.4 m ha (Waldrop et al. 2008).

Four treatments are installed in each of the two replicate block chosen for this study; each treatment unit is ~ 50 ha in extent and contains ten 20 × 50 m (0.1 ha) permanent plots. The four treatments consist of an un-manipulated control, a mechanical thinning (thin-only), a prescribed burning (burn-only), and a combination of the two

(thin+burn). These plots are distributed across the landscape from ridgetops to lower slopes based on the integrated moisture index (IMI) developed by Iverson et al. (1997).

Index values range from 0-100 with higher values indicating greater moisture availability. 145

The model has been used successfully to predict site productivity and species composition in the oak-dominated forests of eastern North America (Iverson et al. 1997).

Mechanical thinning was conducted in the fall and winter of 2000–2001. The thinning operation focused mainly on midstory trees (15–30 cm diameter at breast height, DBH), and reduced stand basal area by ~ 30 % (from 27.4 to about 18.7 m2/ha). Prescribed fires were conducted in the spring of 2001, and repeated in 2005 and 2010. The intensity of prescribed fire varied greatly over the years and across landscapes. In 2001, for example, fire intensity was generally low with flame length reaching about 1m. However, higher intensity fires (i.e., flame length reaching 4–5m) were deliberately created in 2005 and

2010, resulting in significant overstory mortality (Iverson et al. 2008, Waldrop et al.

2008).

In the spring (April 18 to May 22) of 2012, 80 increment cores were extracted from 40 white oak trees ( 25 cm DBH) from the control and thin+burn stands at both

REMA and Zaleski, using an increment borer. Samples were obtained from only these two treatments to reduce cost of isotopic analysis. White oak was selected for analysis because it is one of the most desirable species in the forests of central Appalachians

(Rogers 1990) and there is a considerable body of literature about its biology. It is intermediate in shade-tolerance and is distributed across a wide range of sites (Abrams

2003). White oak has also been observed to perform better in intermediate or subxeric sites (Iverson et al. 1997, Anning et al. ) via a suite of morphological and physiological adaptations including deep roots, thick bark, low relative water content at 146 zero turgor, high stomatal conductance, lower water potential threshold for stomatal closure, etc. (Abrams 2003). White oak trees also cross-date well (Rubino and McCarthy

2000, Anning et al. 2013). Two increment cores were obtained from each tree at breast height, and at roughly 180o from each other to account for variation in ring-width around the circumference of the tree.

Trees were sampled such that the two experimental units and the two IMI classes

(mesic vs. xeric) were evenly represented. Diameter at breast height (DBH) and the geographic location (measured with Garmin GPSMAP 60Cx; Garmin International

Inc., KS, USA) were recorded at the time of coring. Using GPS points, a layer of tree locations was created and draped over a digital elevation model (DEM; 0.7 m resolution), obtained from the Ohio Geographically Referenced Information Program (OGRIP 2012;

Figure 5.1). The physiographic position of each tree was estimated based on its topographic position and aspect; the trees were then classified into two slope-aspect classes: south-facing (105–285o) and north-facing (286–105o). This analysis was implemented in ArcGIS 10x (ESRI, Redlands, CA). Climate data for 1991–2010 for the study sites were obtained from the CLIMVIS database (NCDC, 2012) to characterize the climate change during this period, and to qualitatively assess how this change might influence the physiological response of trees to the treatments (Figure 5.2).

In the laboratory, increment cores were air-dried for at least 24 hours and cross- dated after exposing the ring surfaces with a sharp blade. Starting from the bark end, four sequential tree-ring segments, each comprising five consecutive annual rings (covering the period between 1991–2010) were isolated for isotopic analysis. These included two 147 pre-treatment (1991–1995, 1996–2000) and two post-treatment (2001–2005, 2006–2010) tree ring segments. Thus, a total of 160 samples were analyzed (2 sites x 2 experimental units x 2 IMI classes x 5 trees x 4 tree-ring segments). Cutting of ring segments was accomplished with the aid of a sharp scalpel and a dissecting microscope. To avoid cross- contamination, no lubricants or pencils were used during increment core extraction and sample preparation, respectively.

A dual inlet isotope ratio mass spectrometry technique (DI-IRMS) was used to measure the isotopic ratios of the wood samples (Sulzman 2007)13C were determined using an elemental analyzer (Carlo Erba CHN EA 1108, now Thermo Fisher

Scientific, Waltham, MA) coupled to an isotope ratio mass spectrometer (Finnigan

Conflo III Interface and a Themo Finnigan Delta V Advantage mass spectrometer,

Bremen, Germany). The samples were air-dried, crushed to pass a 53µm sieve, weighed into Sn capsules and combusted at 1000 ºC in the elemental analyzer under a stream of oxygen. The evolved CO2 was transferred to the Conflo III Interface and subsequently

13 C values were determined. The N2 and CO2 reference gases were used as gas standards and measured within every single sample run.

Every 20th sample was acetanilide, which served as a stable isotope ratio reference

13C values per mille [‰] relative to the international standard, Vienna Pee Dee belemnite (VPDB), as follows:

C= 1 × 1000‰ [3] 148

13 12 where and are the ratios of C/ C in the sample and the international standard, respectively (Farquhar et al. 1982). The output from this analysis also included the percent carbon in wood samples. The analysis was done using whole wood samples, which have been found to show a very high degree of coherence with specific wood components such as cellulose and lignin (Loader et al. 2003). The analysis was done at the Stable Isotope Laboratory at the School of Environment and Natural Resources, The

Ohio State University, Columbus, OH.

All analyses were performed using R (R Development Core Team 2013). A linear mixed effects model (Pinheiro and Bates 2000) was used to assess the effect of time and treatment on 13C. The analysis was conducted using the

function in the R package “lme4” (Bates 2005), and included time and treatment

(control vs. thin+burn) and their interactions as the main effects. Individual trees (i.e., tree ids) were also included as random variables. Parameters were estimated using restricted maximum likelihood (REML), and model validation was performed using plots of residuals against each of the explanatory variables (Zuur et al. 2009). Significance of the fixed terms in the model was tested using analysis of variance (ANOVA; Type II sum of square test). Post-hoc multiple comparisons (Tukey contrasts) were performed using the function in the package “multcomp” (Hothron et al. 2008)13C values in relation to IMI classes, and slope-aspect positions across the treatments and over time were analyzed in similar fashion. 149

Results

Climate for the study area fluctuated considerably over the period analyzed—

1991–2010 (Figure 5.2). During this period, total seasonal (April – September) precipitation ranged from 43.0 mm to 86.2 mm, with a mean of 60.8 mm. Seasonal precipitation was highest during the intervals 1996–1998 (68.5 mm) and 2003–2004

(86.0 mm), with the lowest amount of precipitation recorded in 1999. Seasonal temperature for 1991–2010 interval varied between 16.7 oC and 20.4 oC, with an average of 18.3 oC. Minimum and maximum seasonal temperatures were recorded in 1997 and

2010, respectively. A small rise in temperature over the period was also noticeable.

Analysis of the data indicated an increase in percent carbon (% C) from 1991 to

2005, and a decline thereafter (Figure 5.3A). Despite this trend, no significant ( > 0.05) time effect was found. Furthermore, the proportion of carbon in the tree-ring segments did not differ statistically ( > 0.05) between the thin+burn and control stands, although the magnitude of change following the prescribed burning and thinning manipulations appeared to be greater relative to the control. The carbon content of the wood samples analyzed averaged 44.2% (9.8–60.4%) for the control and 43.2% (23.3–63.8%) for the thin+burn stand.

13C values of white oak decreased with time in both the thin+burn stand and the control stand (Figure 5.3B). Mixed-effect analysis of variance showed a significant time effect ( = 0.002) and a marginal effect of time × treatment ( 150

= 0.065) on the stable-carbon isotope composition of the trees. Post-hoc pairwise comparisons revealed that the trees growing in the thin+burn plots were isotopically lighter ( < 0.05) in 2010 (-27.05‰) compared to the pre-treatment period (-26.34‰).

However, the 13C values for 2010 did not differ from that of 2005 (-26.49‰). The 13C signature of white oak trees, calculated over the 1991-2010 interval, averaged -26.55‰ (-

29.07 to -23.60‰) for the thin+burn units and -26.71‰ (-28.89 to -24.66‰) for the control unit, though treatment alone had no significant effect ( 13C values. Apparently, trees from the thin+burn stands exhibited greater and more variable

13C values prior to the treatment, and even within the first post-treatment period (2001–

2005) compared to those of their counterparts from the control stands. During this period,

13C value for trees from the thin+burn unit was -26.39‰ whereas those from the control unit recorded an average of -26.64‰, indicating a slightly greater degree of enrichment in 13C among trees from the thin+burn stands compared to those from the control stands.

Soil moisture availability or stress (as indexed by the IMI) did not have any strong

13C values ( > 0.05) in both the control and the thin+burn stands (Figure

5.13C values of trees from both mesic and xeric sites decreased during the 1991–2000 period, increased slightly between 2001 and 2005, and declined subsequently (Figure 5.13C values from the two IMI classes showed a high degree of coherence over time. Similarly, for white oak trees growing in the thin+burn stands, variation in moisture availability or 151

13C values (> 0.05), despite the slightly higher 13C values in the xeric sites compared to the mesic sites. As 13C values over time was substantial.

13C values for the two slope-aspect positions closely mimicked the patterns observed for the IMI classes, particularly among trees from the control unit (Figure 5.13C values for trees growing on both north-facing and south-facing slopes in the control plots decreased (became more negative) between 1991 and 2000, increased slightly in the next five years, and decreased again thereafter (Figure 5.5A). Not surprisingly, the mixed-model analysis of variance indicated a significant time effect ( 13C values. This

13C values for trees growing on the south-facing slopes, which were lower ( = 0.042) in 2000 (mean: -

26.86‰) than in 1995 (mean: -26.47‰). When data were considered over the entire period analyzed (1991-13C values between the north-facing (mean: -26.72‰) and south-facing (mean: -26.69‰) slopes in the control plots. Also, no significant time × slope-aspect position effect was evident.

Prior to the treatment in 2000, the 13C increased for trees on the sheltered sites in the thin+burn stands, but decreased for those on the south-facing sites (Figure 5.5B).

Thereafter, the trees generally exhibited a decreasing pattern of 13C across both topographic-aspect positions. Trees from these two physiographic positions did not differ in their 13C ( > 0.05); overall mean values for trees growing on the south-facing and north-facing slopes were -26.67‰ and -26.46, respectively. However, as with the control 152

13C values of trees from the thin+burn unit showed a strong temporal trend, with a much lower mean for 2010 (-27.05‰) relative to the preceding intervals (-

26.29‰). There was no significant time × slope-aspect effect on the composition of carbon isotopes of the thin+burn trees.

Discussion

Changes in % C of tree-rings from the thin+burn unit relative to the control after the treatment suggest a marginal increase in C allocation to wood formation, likely reflecting the enhanced growth elicited by this combination treatment as observed in an earlier analysis (Anning and McCarthy 2013a). The % C values observed for white oak trees in this study are comparable to the estimated value of 49% for most hardwood tree species in temperature forests (Chiang et al. 2008)—though mean % C values were lower, maxima for both control and thin+burn stands (44.17% and 43.23%, respectively) were much greater than this regional estimate. In general, the results indicate a somewhat stable rate of carbon sequestration in white oak within the studied forests.

Tree-ring 13C signature of white oak trees recorded in this study falls within the typical range of -33‰ to – 23‰ for C3 plants (Leavitt and Long 1982). Garten and

Taylor (1992) reported a mean foliar 13C value of about -29‰ for three oak species (

, , and ) in a temperate deciduous forest of Tennessee. Given that the 13C of woody materials are often 2-4‰ higher than that of leaves (Leavitt and Long

1982), the isotopic compositions observed here compare well with the value documented by Garten and Taylor (1992). However, it should be emphasized that whole-wood 13C 153 values, such as reported in this study, are typically 1.5‰ lower relative to cellulose-based measurements (Leavitt and Long 1982).

The general decline in 13C values of white oak tree rings over the long term indicates an increasing discrimination against 13C in favor of 12C and a decrease in photosynthetic WUE (Farquhar et al. 1989) during 1991–2010 interval. This declining trend may be a function of the rising atmospheric CO2 levels associated with anthropogenic activities as reported by several previous investigators (e.g., Schulze et al.

2004, Gebrekirstos et al. 2009). Schulze et al. (2004), for example, noted that long-term trends in isotopic composition are often related to integration of CO2 signals from fossil fuel combustion or to shift in climate, which may increase the diffusive supply of CO2 relative to the assimilation rate. An increase in growing season precipitation often causes an increase in discrimination and makes woody plant materials isotopically lighter

(Marshall et al. 2007). Such an increase was apparent in the studied sites, and might have contributed to the observed declining pattern of 13C values over the long term. Farquhar

13 et al. (1982) reported that the C signature of trees might decrease when CO2 assimilation rate is reduced by factors that directly reduces leaf metabolism—this may also offer partial explanation for the decrease in 13C with time.

By reducing stem density, thinning may increase the soil moisture available to residual trees, leading to an increase in stomatal conductance, and thus a decrease in 13C of woody plant materials (Kozlowski et al. 1991, McDowell et al. 2003). This may partially explain the slightly more rapid decline in 13C values during the 2006–2010 interval in the thin+burn stands compared to the un-manipulated stands. On the other 154 hand, an increase in light intensity and in nutrient availability with thinning is likely to elevate photosynthetic capacity and contribute to 13C enrichment in tree-rings (Farquhar et al. 1989, Warren et al. 2001, Brooks and Mitchell 2011), though extreme light intensity may down-regulate photosynthesis through photoinhibition (Lambers et al. 2008).

Perhaps, the negligible change in 13C after the prescribed fire and thinning treatments is the result of these two countervailing effects (i.e., increases in soil moisture availability and photosynthetic capacity). This result indicates no major alteration in WUE of white oak trees as a consequence of the management disturbance. In addition, analyzing 5-year tree-ring segments instead of annual rings might have obfuscated some important patterns in white oak 13C response to the treatments. Further, because this study considered only the thin+burn vs. control design, it remains uncertain how mechanical thinning and prescribed fire may independently influence 13C signatures of white oak trees in this mesophytic oak-dominated forested landscape.

Contrary to expectation, soil moisture availability/stress had only minimal effect on white oak 13C in both the control and the thin+burn stands, with trees from xeric sites being slightly more enriched in 13C than those from the mesic sites. In a moist forest environment such as the ones studied, soil moisture stress may play a secondary role to other biotic and abiotic factors (Van de Water et al. 2002); this may account, to a limited extent, for the unresponsiveness of the whole-wood 13C to the soil moisture gradient.

Similarly, the lack of a strong topographic effect contradicts the expected differences in

13C values for trees occupying different physiographic positions, which tend to vary in temperature, oxygen partial pressure, light intensity, and soil moisture stress across those 155 sites (Garten and Taylor 1992, Marshall et al. 2007). Previous data from the same sites indicated no substantial difference in white growth between these topographic positions despite strong variations in potential evapotranspiration and moisture deficit/availability

(Anning et al. ). These results further support the suspicion that prescribed fire and thinning manipulations could not alter considerably the photosynthetic WUE of white oak trees.

It is important to emphasize that white oak, like most oak species, is one of the most deep-rooted species in the temperate deciduous forests of North America, and has leaves characterized by greater thickness, mass per unit area, stomatal density, higher nitrogen content, and higher osmotic adjustment compared to non-oaks (Abrams 2003).

These eco-physiological adaptations, as suggested by Abrams (2003), allow the species to maintain a relatively high predawn shoot water potential to support a high level of gas exchange. These traits presumably underlie the ability of white oak to thrive and outcompete other hardwood species on xeric sites in much of the regional forests (Anning and McCarthy 2013b), despite the macroclimatic-related decline in intrinsic WUE as suggested by the present findings.

In conclusion, the present results demonstrate considerable changes in 13C over time likely associated with increasing atmospheric CO2 levels or an increase in seasonal precipitation during the studied period. Since 13C signature of plant materials are positively related to photosynthetic WUE, the results indicate a declining pattern of white oak WUE over time. Prescribed fire and thinning caused only minor changes in the photosynthetic WUE of white oak, likely due to the countervailing effects of increased 156 soil moisture availability and increased photosynthetic capacity following reduction in stand density. Apparently, white oak has developed a suite of morphological and eco- physiological adaptations that enable it to thrive and perform across a broad range of environmental conditions created by the prescribed fire and thinning treatments.

Literature Cited

Abrams, M. D. 2003. Where has all the white oak gone? BioScience 53:927–939.

Albrecht, M. A., and B. C. McCarthy. 2006. Effects of prescribed fire and thinning on

tree recruitment patterns in central hardwood forests. Forest Ecology and

Management 226:88–103.

Anning, A. K., and B. C. McCarthy. 2013a. Long-term effects of prescribed pire and

thinning on residual tree growth in mixed-oak forests of southern Ohio.

Ecosystems 16:1473–1486.

Anning, A. K., and B. C. McCarthy. 2013b. Competition, size and age affect tree growth

response to fuel reduction treatments in mixed-oak forests of Ohio. Forest

Ecology and Management 307:74–83.

Anning, A. K., J. M. Dyer, and B. C. McCarthy. . Tree growth response to fuel

reduction treatments along a topographic moisture gradient in mixed-oak forests,

Ohio, USA. Canadian Journal of Forest Research.

Anning, A. K., D. L. Rubino, E. K. Sutherland, and B. C. McCarthy. 2013.

Dendrochronological analysis of white oak growth patterns across a topographic

moisture gradient in southern Ohio. Dendrochronologia 31:120–128. 157

Au, R., and J. C. Tardif. 2009. Chemical pretreatment of tree rings:

implications for dendroisotopic studies. Canadian Journal of Forest Research

39:1777–1784.

Bates, D. M. 2005. Fitting linear mixed models in R. R News 5:27–30.

Beerling, D. J., and F. I. Woodward. 1995. Leaf stable carbon isotope composition

records increased water-use efficiency of C3 plants in response to atmospheric

CO2 enrichment. Functional Ecology 9:394–401.

Boerner, R. E. J., J. Huang, and S. C. Hart. 2008. Fire, thinning, and the carbon economy:

Effects of fire and fire surrogate treatments on estimated carbon storage and

sequestration rate. Forest Ecology and Management 255:3081–3097.

Boerner, R. E. J., S. J. Morris, K. L. Decker, and T. F. Hutchinson. 2003. Soil and forest

floor characteristics. Pages 47–56 E. K. Sutherland and T. F. Hutchinson,

editors. Characteristics of mixed-oak forest ecosystems in southern Ohio prior to

the reintroduction of fire. Gen. Tech. Rep. NE-299. U.S. Department of

Agriculture, Forest Service, Northeastern Research Station.

Brooks, J. R., and A. K. Mitchell. 2011. Interpreting tree responses to thinning and

fertilization using tree-ring stable isotopes. New Phytologist 190:770–782.

Chiang, J.-M., R. W. McEwan, D. A. Yaussy, and K. J. Brown. 2008. The effects of

prescribed fire and silvicultural thinning on the aboveground carbon stocks and

net primary production of overstory trees in an oak-hickory ecosystem in southern

Ohio. Forest Ecology and Management 255:1584–1594. 158

Dawson, T. E., S. Mambelli, A. H. Plamboeck, P. H. Templer, and K. P. Tu. 2002. Stable

isotopes in plant ecology. Annual Review of Ecology and Systematics 33:507–

559.

Farquhar, G. D., J. R. Ehleringer, and K. T. Hubick. 1989. Carbon isotope discrimination

and photosynthesis. Annual Review of Plant Physiology and Plant Molecular

Biology 40:503–537.

Farquhar, G. D., M. H. O. Leary, and J. A. Berry. 1982. On the relationship between

carbon isotope discrimination and the intercellular carbon dioxide concentration

in leaves. Austrialian Journal Physiology 9:121–137.

Fiedler, C. E., K. L. Metlen, and E. K. Dodson. 2010. Restoration treatment effects on

stand structure, tree growth, and fire hazard in a ponderosa pine/douglas-fir forest

in Montana. Forest Science 56:18–31.

13C within a temperate deciduous forest:

spatial, temporal, and species sources of variation. Oecologia 90:1–7.

Gebrekirstos, A., M. Worbes, D. Teketay, M. Fetene, and R. Mitlöhner. 2009. Stable

carbon isotope ratios in tree rings of co-occurring species from semi-arid tropics

in Africa: Patterns and climatic signals. Global and Planetary Change 66:253–

260.

Gough, C. M., C. S. Vogel, K. H. Harrold, K. George, and P. S. Curtis. 2007. The legacy

of harvest and fire on ecosystem carbon storage in a north temperate forest.

Global Change Biology 13:1935–1949. 159

Hothron, T., F. Bretz, and P. Westfall. 2008. Simultaneous inference in general

parametric models. Biometrical Journal 50:346–363.

Hurteau, M., and M. North. 2009. Fuel carbon modeled treatment storage wildfire and

under emissions scenarios. Frontiers in Ecology and the Environment 7:409–414.

Hutchinson, T. F., R. P. Long, J. Rebbeck, E. K. Sutherland, and D. A. Yaussy. 2012.

Repeated prescribed fires alter gap-phase regeneration in mixed-oak forests.

Canadian Journal of Forest Research 314:303–314.

Hutchinson, T. F., E. K. Sutherland, and D. A. Yaussy. 2005. Effects of repeated

prescribed fires on the structure, composition, and regeneration of mixed-oak

forests in Ohio. Forest Ecology and Management 218:210–228.

Iverson, L. R., M. E. Dale, C. T. Scott, and A. Prasad. 1997. A GIS-derived integrated

moisture index to predict forest composition and productivity of Ohio forests

(USA). Landscape Ecology 12:331–348.

Iverson, L. R., T. F. Hutchinson, A. M. Prasad, and M. P. Peters. 2008. Thinning, fire,

and oak regeneration across a heterogeneous landscape in the eastern U.S.: 7-year

results. Forest Ecology and Management 255:3035–3050.

Kozlowski, T. T., P. J. Kramer, and S. G. Pallardy. 1991. The Physiological Ecology of

Woody Plants. Academic Press, Inc., San Diego, CA.

Lambers, H., F. S. Chapin III, and T. L. Pon. 2008. Plant Physiological Ecology, 2nd

edition. Springer, New York, NY.

Leavitt, S. W., and A. Long. 1982. Stable carbon isotopes as a potential supplemental

tool in dendrochronology. Tree-Ring Bulletin 42:49–55. 160

Loader, N. J., I. Robertson, and D. McCarroll. 2003. Comparison of stable carbon isotope

ratios in the whole wood, cellulose and lignin of oak tree-rings. Palaeogeography,

Palaeoclimatology, Palaeoecology 196:395–407.

Ma, J.-Y., T. Chen, W.-Y. Qiang, and G. Wang. 2005. Correlations between foliar stable

carbon isotope composition and environmental factors in desert plant

(Pall.) Maxim. Journal of Integrative Plant Biology 47:1065–1073.

Marshall, J. D., J. R. Brooks, and K. Lajtha. 2007. Sources of variation in the stable

isotope composition of plants. Pages 22–60 R. Michener and K. Lajtha, editors.

Stable Isotopes in Ecology and Environmental Science, 2nd edition. Blackwell

Publishing Ltd, Malden, MA.

Martín-Benito, D., M. Del Río, I. Heinrich, G. Helle, and I. Cañellas. 2010. Response of

climate-growth relationships and water use efficiency to thinning in a

afforestation. Forest Ecology and Management 259:967–975.

McCarroll, D., and N. J. Loader. 2005. Isotopes in tree rings. Pages 67–116 M. J.

Leng, editor. Isotopes in Palaeoenvironmental Research. Springer, Dordrecht, The

Netherlands.

McDowell, N., J. R. Brooks, S. A. Fitzgerald, and B. J. Bond. 2003. Carbon isotope

discrimination and growth response of old trees to stand density

reductions. Plant, Cell and Environment 26:631–644.

National Climatic Data Center. 2012. Climate data online. National Oceanic and

Atmospheric Administration (NOAA), Asheville, NC, USA.

Http://www.ncdc.noaa.gov 161

Ohio Geographically Referenced Information Program. Geodata Distribution.

Http://ogrip.oit.ohio.gov

Pinheiro, J. C., and D. M. Bates. 2000. Mixed-Effects Models in S and S-PLUS.

Springer-Verlag, New York, NY.

R Development Core Team. 2013. R: A language and environment for statistical

computing. R Foundation for Statistical Computing, Vienna, Austria.

Http://www.R-project.org/

Rogers, R. 1990. L. white oak. Pages 605–613 R. M. Burns and B. H.

Honkala, editors. Silvics of North America Vol. 2: Hardwoods. US Department of

Agriculture, Forest Service. Agricultural Handbook No. 654.

Rubino, D. L., and B. C. McCarthy. 2000. Dendroclimatological analysis of white oak

( L., Fagaceae) from an old-growth forest of southeastern Ohio,

USA. Journal of the Torrey Botanical Society 127:240–250.

Schulze, B., C. Wirth, P. Linke, W. A., Brand, V. Horna, and E. Schulze. 2004. Laser

ablation-combustion-GC-IRMS—a new method for online analysis of intra-

annual variation of delta 13C in tree rings. Tree Physiology 24:1193–1201.

Sulzman, E. W. 2007. Stable isotope chemistry and measurement: a primer. Pages 1–21

R. Michener and K. Lajtha, editors. Stable Isotopes in Ecology and

Environmental Science, 2nd edition. Blackwell Publishing Ltd, Oxford, UK.

Thorpe, H. C., R. Astrup, A. Trowbridge, and K. D. Coates. 2010. Competition and tree

crowns: A neighborhood analysis of three boreal tree species. Forest Ecology and

Management 259:1586–1596. 162

Thorpe, H. C., S. C. Thomas, and J. P. Caspersen. 2007. Residual-tree growth responses

to partial stand harvest in the black spruce () boreal forest.

Canadian Journal of Forest Research 37:1563–1571.

Waldrop, T. A., D. A. Yaussy, R. J. Phillips, T. A. Hutchinson, L. Brudnak, and R. E. J.

Boerner. 2008. Fuel reduction treatments affect stand structure of hardwood

forests in Western North Carolina and Southern Ohio, USA. Forest Ecology and

Management 255:3117–3129.

Warren, C. R., J. F. McGrath, and M. A. Adams. 2001. Water availability and carbon

isotope discrimination in conifers. Oecologia 127:476–486.

13C variability with

elevation, slope aspect, and precipitation in the southwest United States.

Oecologia 132:332–343.

Werner, C., H. Schnyder, M. Cuntz, C. Keitel, M. J. Zeeman, T. E. Dawson, F.-W.

Badeck, E. Brugnoli, J. Ghashghaie, T. E. E. Grams, Z. E. Kayler, M. Lakatos, X.

Lee, C. Máguas, J. Ogée, K. G. Rascher, R. T. W. Siegwolf, S. Unger, J. Welker,

L. Wingate, and A. Gessler. 2012. Progress and challenges in using stable

isotopes to trace plant carbon and water relations across scales. Biogeosciences

9:3083–3111.

West, J. B., G. J. Bowen, T. E. Cerling, and J. R. Ehleringer. 2006. Stable isotopes as one

of nature’s ecological recorders. Trends in Ecology and Evolution 21:408–14.

Zuur, A. F., E. N. Ieno, N. J. Walker, A. A. Saveliev, and G. M. Smith. 2009. Mixed

Effects Models and Extensions in Ecology with R. Springer, New York, NY. 163

Topography of the study sites and location of some trees (dots) used for analysis. 164

Mean seasonal (April-September) precipitation and temperature variability from 1991 – 2010 for southeastern Ohio.

165

Temporal changes in (A) percent carbon and (B) stable carbon isotope 13C) in tree rings of white oak compared for control and thin+burn stands. Error bar represents one standard error. Treatments were applied in 2000. 166

Effect of soil moisture gradient (indexed by the integrated moisture gradient, IMI) on the temporal variability of carbon isotope composition (13C) of white oak trees growing in (A) control and (B) thin+burn stands in mixed-oak forests. Error bar represents one standard error. Treatments were applied in 2000. Analyses were performed 167

Effect of topographic-aspect position (SAP) on the temporal variability of carbon isotope composition (13C) of white oak trees growing in (A) control and (B) thin+burn stands in mixed-oak forests. Error bar represents on standard error. Treatments were applied in 2000.

168

CHAPTER 6: CONCLUSION

The overarching goal of this study was to assess the response of residual tree growth to prescribed fire and thinning treatments to understand the broader consequences of these management strategies on mixed-oak forest ecosystems and to provide information that may help in fine-tuning their application in the central Appalachians,

Ohio. Using tree-ring data obtained from five common overstory tree species (white oak, chestnut oak, black oak, yellow-poplar and hickories), the study addressed two broad questions: 1) what is the long-term pattern of residual tree growth in response to the prescribed fire and thinning treatments, and 2) what are the underlying mechanisms of these growth response patterns or the biophysical factors that mediate them?

Prescribed fire and thinning caused substantial increase in tree growth mainly by reducing overstory tree density and thus increasing availability of resources. Tree growth exhibited a strong temporal pattern, with the largest proportional change occurring 1-5 years after the treatments. Subsequently, trees continued to exhibit elevated growth, albeit at a reduced rate. These changes in tree growth suggest an increasing pattern of productivity or carbon sequestration rates at least at the individual tree level. Mechanical treatment was most effective at enhancing tree growth. By contrast, prescribed fire had only a modest effect on tree growth, although for some species (e.g., black oak), responses varied with fire severity. Interspecific variation in tree growth response to the prescribed fire and thinning was evident, with yellow-poplar being the most responsive, oaks somewhat conservative, and hickories exhibiting slow growth. These differences 169 reflect the variations in the intrinsic ability of species to respond to the management disturbances (Johnson et al. 2002).

As expected, tree growth correlated positively with tree size, but the relation was strongest for the trees in the burn-only treatment. The larger trees produced more wood because of their larger crowns, which gave them greater capacity to capture light for photosynthesis (Kozlowski et al. 1991, Wyckoff and Clark 2005). Tree age was negatively but weakly related to tree growth. Prescribed fire had very little effect on the growth rates of larger and older trees because of their generally low intensity, which caused only minimal reduction in stand basal area.

The treatments induced considerable variations in the competitive status of trees, which explained much of the differences observed in growth rates of individual trees across experimental units. Trees from the thin-only stands were more sensitive to competition, i.e., their growth rates decreased more sharply with increasing competition, whilst those from the burn-only stands were least sensitive. These results suggest that mechanical thinning influences tree growth primarily via competitive release, whereas prescribed burning may confound the effect of competition on growth by altering other factors, e.g., increased soil nutrients and mineralization rates (Boerner et al. 2009). The importance of competition for tree growth was species-dependent and was somewhat mediated by site quality. The competition-growth relation also depended on the size of trees, likely due to differences in sensitivity of smaller and larger trees to crowding and shading. Overall, competition appeared to be a more important driver of residual tree 170 growth than size and age, and influenced forest growth and development by creating heterogeneity within and among stands.

Soil moisture demand and availability varied spatially within the forested landscape studied. Trees on sheltered parts of the landscape (i.e., those on north-facing slopes) experienced low PET, moisture deficit and high IMI compared to those on warmer, drier, south-facing slopes. In the control stands, tree growth generally decreased with increasing moisture demand/stress due to increasing physiological constraints including impairment of stomatal function and photosynthetic activities. However, this link broke down in the managed stands, as treatment effects (e.g., reduction in competition, increased nutrients and moisture availability) became the dominant drivers of tree growth. Nevertheless, the treatments interacted somewhat with the soil moisture gradient to influence tree growth. The oaks and non-oaks exhibited contrasting growth responses in the control stands, with the former growing best on intermediate to xeric sites, whilst the latter performed better on more mesic sites. In general, these results demonstrate the strong regulatory effects of a topographically-controlled soil moisture gradient on tree, though this effect may be trumped by management disturbances.

The stable-carbon isotope (13C) values of white oak trees declined with time,

13 12 indicating an increased discrimination against CO2 relative to CO2 and a declining trend in water-efficiency, most likely related to increased ambient CO2 levels and increased precipitation in the study area. Despite a slightly more rapid decline in the thin+burn trees compared to the control trees during the 20062010 interval, the treatment had very little effect on 13C signature of white oak trees within the study area. 171

Similarly, soil moisture gradient and topographic position had minimal effects on the

13C of white oak trees. These results demonstrate that prescribed fire and thinning manipulations do not alter significantly photosynthetic water-use efficiency of white oak or the balance between diffusive supply of intercellular CO2 or photosynthetic dependence on it.

This study provides the most comprehensive analysis, to date, of the growth response of individual trees to prescribed fire and thinning management in the closed- canopy forests of central Appalachians. Given the importance of tree growth for forest community composition and dynamics, and the little attention it has received so far, the long-term, species-specific responses and their underlying biophysical mechanisms observed in this study clearly add to existing knowledge on the consequences of prescribed fire and thinning treatments on forest ecosystems. The new insights can be useful for fine-tuning the treatment application to achieve management objectives, e.g., high-intensity prescribed fire and thinning treatment may be applied on intermediate- moisture sites to elicit substantial growth response of trees. The combination of tree-ring analysis, stable-carbon isotope analysis, GIS modeling, statistical modeling and spatially- explicit individual tree-based analysis provided an unique opportunity to elucidate the mechanisms/factors driving the growth responses of these residual trees to the treatments.

Management Implications

The results obtained from this study have important implications for forest ecology and management. Whilst research efforts in the last two decades have mainly focused on oak regeneration response to prescribed fire and thinning for the obvious 172 objective of maintaining oak dominance in eastern North America, the strong response of residual tree growth to the treatment posits the need to incorporate these large trees in current assessment programs to fully understand the impacts of the treatment on forest ecosystems. The attenuation in growth rates observed 610 years after the treatments suggests the need for repeated or periodic application, particularly of mechanical thinning, if increased growth rates of larger trees is a management objective.

Differences in the effects of mechanical thinning and prescribed fire on tree growth, including their effects on different-size trees, and their distinct interactions with competition and to some extent soil moisture gradient, show that these two techniques may not be complete surrogates as noted by previous investigators (Schwilk et al. 2009).

If the management goal, for instance, is to quickly increase growth and productivity of individual trees, then mechanical treatment may provide the best option. On the other hand, recalling the minimal effect of prescribed fire on large tree growth, more intense fire prescriptions may be required to enhance growth of large residual trees in forests where thinning is not a management option or where managers desire to promote growth of fire-adapted species.

From their responses to the different types of treatments, competition and soil moisture demand/stress, the inherent differences among the five species studied were clearly evident. Thus, inferences about residual tree response to the restoration methods based solely on stand-level results may obscure important variations. For example, yellow-poplar is highly responsive to both canopy openings and the moisture gradient, and may benefit most from heavy thinning on moist sites than the oaks. This calls for 173 careful selection of candidate species for removal during thinning operations. In addition, this study demonstrates that the responses of different taxonomic groups to prescribed fire and thinning may be mediated by site quality. Consequently, there is need for careful consideration of where and how the treatments are applied, as indiscriminate application of the treatments may end up promoting growth of less desirable species.

Literature Cited

Boerner, R. E. J., J. Huang, and S. C. Hart. 2009. Impacts of fire and fire surrogate

treatments on forest soil properties: a meta-analytical approach. Ecological

Applications 19:338–358.

Johnson, P. S., S. R. Shifley, and R. Rogers. 2002. The Ecology and Silviculture of Oaks.

CABI Publishing, Wallingford, UK.

Kozlowski, T. T., P. J. Kramer, and S. G. Pallardy. 1991. The Physiological Ecology of

Woody Plants. Academic Press, Inc., San Diego, CA.

Schwilk, D. W., J. E. Keeley, E. E. Knapp, J. McIver, J. D. Bailey, C. J. Fettig, C. E.

Fiedler, R. J. Harrod, J. J. Moghaddas, K. W. Outcalt, C. N. Skinner, S. L.

Stephens, T. A. Waldrop, D. A. Yaussy, and A. Youngblood. 2009. The national

fire and fire surrogate study: effects of fuel reduction methods on forest

vegetation structure and fuels. Ecological Applications 19:285–304.

Wyckoff, P. H., and J. S. Clark. 2005. Tree growth prediction using size and exposed

crown area. Canadian Journal of Forest Research 20:13–20.

174

APPENDIX 1A: THE COMPLETE SET OF A PRIORI MODELS EXPLAINING THE VARIATIONS IN BAI

175

APPENDIX 1B: THE COMPLETE SET OF A PRIORI CANDIDATE MODELS EXPLAINING THE PERCENT

CHANGES IN BAI

a Priori

176

APPENDIX 2A: STAND-LEVEL BASAL AREA OF THE TWO REPLICATE

BLOCKS BEFORE AND AFTER THE TREATMENTS

BA (m2 ha-1) Change in BA (%) Site/Treatm 2000- 2000- 2000- 2000- ent unit 2000 2001 2004 2007 2009 2001 2004 2007 2009 REMA Control 27.6 28.2 29.2 28.7 28.7 2.2 5.9 4.1 4.3 Thin 27.4 18.7 19.4 19.5 21.3 -31.4 -28.7 -28.8 -22.5 Thin+burn 27.9 22.6 22.0 19.5 20.6 -18.9 -21.6 -30.6 -26.5 Burn 25.6 26.1 24.4 17.7 17.7 1.6 -6.4 -35.3 -35.1 Zaleski Control 29.4 29.8 30.4 30.4 32.0 1.4 3.3 2.8 8.5 Thin 27.7 19.1 19.6 19.4 21.2 -29.6 -28.0 -28.9 -22.1 Thin+burn 25.5 18.4 18.4 18.1 19.3 -28.3 -28.0 -29.4 -25.2 Burn 25.8 26.2 24.5 22.4 23.4 1.6 -3.9 -11.6 -7.5

177

APPENDIX 2B: THE FIVE SET OF COMPETITION INDICES COMPARED IN CURRENT STUDY

10 m radius 15 m radius Variable radius Model Equation Slope (%) Slope (%) Slope (%) Basic model -5.575 25.83 -4.902 26.78 -4.097 24.72 Basic with basal -4.607 28.03 -4.247 28.49 -3.805 27.23 area -8.269 29.70 -7.292 29.59 Asymmetric -0.729 27.95 -0.622 27.56 -0.581 27.12 Distance -0.810 26.33 -0.415 25.06 -0.158 22.08 independent /

Thesis and Dissertation Services