&EPA
            United States
            Environmental Protection
            Agency
             Policy, Planning,
             And Evaluation
             (PM-221)
EPA-230-05-89-054
June 1989
The Potential Effects
Of Global Climate Change
On The  United States
Appendix D
Forests

-------
THE POTENTIAL EFFECTS OF GLOBAL CLIMATE CHANGE
               ON THE UNITED STATES:

               APPENDIX D - FORESTS
             Editors: Joel B. Smith and Dennis A. Tirpak
          OFFICE OF POLICY, PLANNING AND EVALUATION
           U.S. ENVIRONMENTAL PROTECTION AGENCY
                  WASHINGTON, DC 20460

                       MAY 1989

-------
                            TABLE OF CONTENTS


                                                                       Page


APPENDIX D:  FORESTS
      PREFACE
      ASSESSING THE RESPONSE OF VEGETATION TO FUTURE CLIMATE
      CHANGE: ECOLOGICAL RESPONSE SURFACES AND PALEOECOLOGICAL
      MODEL VALIDATION ............................................ > .....  1-1
      Jonathan T. Overpeck and Patrick J. Bartlein

      EFFECTS OF CLIMATE CHANGE ON FORESTS OF THE GREAT LAKE
      STATES [[[ 2-1
      Dr. Daniel B. Botltin, Dr. Robert A. Nisbet, Tad E. Reynales

      FOREST RESPONSE TO CLIMATIC CHANGE: A SIMULATION STUDY FOR
      SOUTHEASTERN FORESTS ................................................. 3-1
      Dean L. Urban and Herman H. Shugart

      ANCIENT ANALOGS FOR GREENHOUSE WARMING OF CENTRAL
      CALIFORNIA [[[ 4-1
      Owen K. Davis

      HARD TIMES AHEAD FOR GREAT LAKES FORESTS: A CLIMATE
      THRESHOLD MODEL PREDICTS RESPONSES TO CO--INDUCED CLIMATE
      CHANGE .................................... r ........................... 5-1
      Catherine Zabinski and Margaret B. Davis

      POTENTIAL EFFECTS OF CLIMATE CHANGE  ON U.S. FORESTS:

-------
                                            PREFACE


The ecological and economic implications of the greenhouse effect have been the subject of discussion within
the scientific community for the past three decades. In recent years, members of Congress have held hearings
on the greenhouse effect and have begun to examine its  implications for public policy.  This interest  was
accentuated during a series of hearings held in June 1986 by the Subcommittee on Pollution of the Senate
Environment and Public Works Committee.  Following the hiring*, committee members sent a formal request
to the EPA Administrator, asking the Agency to undertake two studies on climate change due to the greenhouse
effect

        One of the studies we are requesting should examine the potential health and environmental
        effects of climate change. This study should include, but not be limited to, the potential impacts
        on agriculture, forests, wetlands, human health, rivers, lakes, and estuaries, as well as other
        ecosystems and societal impacts. This study should be designed to include original analyses, to
        identify and fill in where  important  research gaps exist and  to solicit the opinions of
        knowledgeable people throughout  the  country through a process of public hearings and
        meetings.

To meet this request, EPA produced the report entitled The Potential Effects of Global Climate Change on the
United States.  For that report, EPA commissioned fifty-five studies by academic and government scientists on
the potential effects of global climate change.  Each study was reviewed by at least two peer reviewers.  The
Effects Report summarizes the results of aU of those studies. The complete results of each study are contained
in Appendices A through J.


                              Appendix                        Subject

                                A                           Water Resources
                                B                            Sea Level Rise
                                C                            Agriculture
                                D                           Forests
                                E                            Aquatic  Resources
                                F                            Air Quality
                                G                           Health
                                H                           Infrastructure
                                I                            Variability
                                J                            Policy
GOAL

The goal of the Effects Report was to try to give a sense of the possible direction of changes from a global
warming as well as a sense of the magnitude. Specifically, we examined the following issues:


        o  sensitivities of systems to changes in climate (since we cannot predict regional climate change, we
           can only identify sensitivities to changes in climate factors)

        o  the range of effects under different warming scenarios

        o  regional differences among effects

        o  interactions among effects  on a regional level


                                                 iii

-------
        o   national effects

        o   uncertainties

        o   policy implications

        o   research needs

 The four regions chosen for the studies were California, the Great Lakes, the Southeast, and the Great Plains.
 Many studies focused on impacts in a single region, while others examined potential impacts on a national scale.


 SCENARIOS USED FOR THE EFFECTS REPORT STUDIES

 The Effects Report studies used several scenarios to examine the sensitivities of various systems to changes in
 climate. The scenarios used are plausible sets of circumstances although none of them should be considered to
 be predictions of regional climate change.  The most common scenario used was the doubled CO2 scenario
 (2XCO2X which examined the effects of climate under a doubling of atmospheric carbon dioxide concentrations.
 This doubling is estimated to raise average global temperatures by L5 to 4-5°C by the latter half of the 21st
 century. Transient scenarios, which estimate how climate may change over time in response to a steady increase
 in greenhouse gases, were also used. In addition, analog scenarios of past warm periods, such as the 1930s, were
 used.

 The scenarios combined average  monthly climate change estimates for regional grid boxes  from General
 Circulation Models (GCMs) with 1951-80 climate observations from sites in the respective grid boxes.  GCMs
 are dynamic models that simulate the physical processes of the atmosphere and oceans to estimate global climate
 under different conditions, such as inrrgamng concentrations of greenhouse gases (e.g^ 2XCO2).

 The scenarios and GCMs used in the studies have certain limitations. The scenarios used for the studies assume
 that temporal and spatial variability do not change from current conditions. The first of two major limitations
 related to the GCMs is their low spatial resolution.  GCMs use rather large grid boxes where climate is averaged
 for the whole grid box, while in fact climate may be quite variable within a grid box. The second limitation is
 the simplified way that GCMs treat physical factors such as clouds, oceans, albedo, and land surface hydrology.
 Because of these limitations, GCMs often disagree with each other on estimates of regional climate change (as
 well as the magnitude of global changes) and should not be considered to be predictions.

 To obtain a range of scenarios,  EPA asked the researchers to use output from  the following GCMs:

        o    Goddard Institute for Space Studies (GISS)

        o    Geophysical Fluid Dynamics Laboratory (GFDL)

        o    Oregon State University (OSU)

Figure 1 shows the  temperature change  from current  climate to a climate with a doubling of CO2 levels, as
modeled by the three GCMs.  The figure includes the GCM estimates for the four regions. Precipitation
changes are shown hi  Figure 2. Note the disagreement in  the GCM estimates concerning the direction of
change of regional and seasonal precipitation and the agreement concerning increasing temperatures.

Two transient scenarios from the GISS model were also used, and the average decadal temperature changes
are shown in Figure 3.
                                                IV

-------
                          FIGURE 1.  TEMPERATURE SCENARIOS

                      GCM Estimated Change in Temperature from 1xCO2 to 2xCO2
o
o

UJ
cc
OC
UJ
Q.

Ul
                                                                  8
                                                                                      SUMMER
                                                                                        1
     Great Southeast Groat
     Lakes        Plains
California United
       States*
Great  Southeast Great California  United
Lakes         Plains        States*
Great  Southeast Great California United
Lakes         Plains        States*
                                                                                  * Lower 48 States

-------
    FIGURE 2. PRECIPITATION SCENARIOS
                                            «

GCM Estimated Change in Precipitation from 1xCO2 to 2xCC>2
I.U
>. 0.8
Q 0.6
a o.4
LU
t °2
2 0.0
=J -0.2
5»
12 -0.4
O fi

ANNUAL

•
&LD 11 «*-, iL iy~i
EJj KjJI ^— *
i


I.U
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
.nfi

WINTER
™
i
In JL
f/
i
i I
I \

LHTI




1.U-]
0.8
0.6
0.4
0.2
o.o-
-0.2
-0.4
n a



No
Change
i




i
si

1
&
Es2

n
%
\
\




SUMMER

rra

$

i
|22|
n
v/
\



1
i
[g£]
1 l^ v/ r~\
\VA W \
lij|_J IX
I


Great Southeast Great California United Great Southeast Great California United Great Southeast Great California United
Lakes Plains States' Lakes Plains States' Lakes Plains States'
                         GISS



                         GFDL



                         OSU
                                                Lower 48 Slates

-------
    4
   3.5
O   3
UI  95
ee  "
I   *
ui
a.  ...
B
    1
   0.5
    0
oc
UI
UI
3.5
 3
2.5

1.5
 1
as
 0
                                              3.72
                                              2.99
                                        2.47
                                   1.72
                              1.36
                   0.70
                        0.88
              0.30
             7Z2L
             1980s 1990s 2000s 2010s 2020s 2030s 2040s 2050s
                         TRANSIENT SCENARIO A
                                                   1.26
                                          1.02
                                 0.59
                        0.35
                       777/
           1980s     1990s     2000s    2010s
                      TRANSIENT SCENARIO B
                                                  2020s
FIGURE 3.
             GISS TRANSIENTS  "A" AND "B"  AVERAGE
             TEMPERATURE CHANGE FOR LOWER 48  STATES
             GRID POINTS.
                               vu

-------
EPA specified that researchers were to use three doubled CO-scenarios, two transient scenarios, and an analog
scenario in their studies.  Many researchers, however, did nothave sufficient time or resources to use all of the
scenarios. EPA asked the researchers to run the scenarios in the following order, going as far through the list
as time and resources allowed:

        1. GISS doubled CO2

        2. GFDL doubled CO2

        3. GISS transient A

        4. OSU doubled CO2

        5. Analog (1930 to 1939)

        6. GISS transient B


ABOUT THESE APPENDICES

The studies contained in these appendices appear in the form that the researchers submitted them to EPA.
These reports do not necessarily reflect the official position of the U.S. Environmental Protection Agency.
Mention of trade names does not constitute an endorsement
                                              vui

-------
ASSESSING THE RESPONSE OF VEGETATION TO FUTURE CLIMATE
       CHANGE:  ECOLOGICAL RESPONSE SURFACES AND
            PALEOECOLOGICAL MODEL VALIDATION
                      Jonathan T. Overpeck
               Lamont-Doherty Geological Observatory
                       Palisades, NY 10964
                 Goddard Institute for Space Studies
                         2880 Broadway
                      New York, NY 10025

                             and

                        Patrick J. Bartlein
                     Department of Geography
                      University of Oregon
                        Eugene, OR 97403
              Interagency Agreement No. DW80932629-01

-------
                                CONTENTS

                                                                      Page

ACKNOWLEDGMENTS	  ui

FINDINGS 	  H

CHAPTER 1: INTRODUCTION	  1-2

CHAPTER 2: DATA AND STUDY AREA 	  1-4
      OBSERVATIONAL CLIMATE DATA	  1-4
      MODERN AND FOSSIL VEGETATION DATA 	  1-4
      SIMULATED CLIMATE DATA	  1-6

CHAPTER 3: MODELING METHODS	  1-10
      EMPIRICAL ECOLOGICAL RESPONSE SURFACES	  1-10
      FOREST STAND SIMULATION	  1-10

CHAPTER 4: RESULTS  	1-14
      SIMULATED VERSUS OBSERVED MODERN VEGETATION PATTERNS	1-14
      FUTURE VEGETATION CHANGE - 2xCO2 SCENARIOS	1-14
      PAST CLIMATES - MODEL VALIDATION^	1-18
           Response Surfaces - The NCAR CCM	1-18
           Stand Simulation Model - The GISS GCM	1-23

CHAPTER 5: DISCUSSION	1-24
      SIGNIFICANT RESULTS	1-24
      UNCERTAINTIES AND THE NEED FOR FUTURE WORK	1-25

CHAPTER 6: CONCLUSIONS	1-27

ABBREVIATIONS  	1-28

REFERENCES	1-29

-------
                                   ACKNOWLEDGMENTS


       We thank R. Goldberg for important computational and climatological assistance. A. Solomon and L.
Tharp provided the stand-simulation model (FORENA) and insights about its design and use. R. Jenne, D.
Joseph, and R. Ruedy helped with GCM output D. Rind, C Rosenzweig, A. Solomon, G-King, R. Nielsen, T.
Webb HI, J. Smith, D. Tirpak, E. Cowling, M. Harwell, F. Miller, B. Smith, and anonymous reviewers provided
valuable discussion and comments.
                                             ui

-------
                                                                                           Overpeck
                                            FINDINGS1


      Climate and vegetation models can be combined to assess how natural vegetation might respond to trace-
gas-induced climate change. We used paleoclimatological and paleoecologjcal data to test the ability of (1)
atmospheric general circulation models (GCMs), (2) ecological response surface models, and (3) a forest-stand
model to simulate observed oast and present vegetation.  Our results suggest that these models can be used to
describe the broad-scale (1(T to 10s km2) equilibrium response of vegetation to past and therefore future climate
change. Model simulations suggest that significant climate warming across this region is likely in all seasons.
Our results suggest that this warming alone is sufficient to cause large-scale geographic shifts in plant populations
and in total forest biomass. Even the most modest simulated wanning is sufficient to cause significant vegetation
change. Future precipitation changes are more difficult to assess, but could exacerbate future vegetation change.

      We used two independent vegetation models coupled with three independent GCM scenarios to assess
the response of natural vegetation to future climate change. The results of these different modeling efforts are
in general agreement  Both the spruce-rich boreal  forests and mixed conifer-hardwood  forests of the Great
Lakes and New England regions could be significantly altered by future climatic change.  Spruce, fir, and pine
populations could decline dramatically in these regions and could be replaced by larger populations of oak and
other deciduous trees. The simulated eastward expansion of prairie forb populations south of the Great Lakes
coincides with simulated decreases in total  forest biomass. Vegetation change in  the southeast U.S. could be
large, but the nature of this change is difficult to assess because the projected extreme warmth in that region has
no modern analogs. The range boundaries of  southern pine populations could extend northward.

      Our model validation efforts suggest that the current generation of climate and vegetation models can
help in assessing the broad-scale equilibrium response of vegetation  to climate change. Our validation of the
models also demonstrates, however, that new research is needed before we can hope to simulate many details
of future climate and vegetation change.
        'Although the information in this report has been wholly or partly funded by the U.S. Environmental
Protection Agency under Interagency Agreement No. DW80932629-OL, it does not necessarily reflect the Agency's
views, and no official endorsement «hnnlr1 tv» !nf»rr»H from it
views, and no official endorsement should be inferred from it

                                                 1-1

-------
 Overpeck

                                            CHAPTER 1

                                         INTRODUCTION


       Current projections suggest that the concentration of radiatively active trace gases in the atmosphere
 could  double their pre-industrial levels within 100-200 years.  This change in an important climate boundary
 condition, and the climatic response (warming) to this change, have no analog in the past 18,000 years. This lack
 of a historical or geological analog means that we must rely on sophisticated numerical climate and vegetation
 models to predict future climatic change and its impact on vegetation.  In order to trust the predictive accuracy
 of these models, we must test their ability to simulate realistic responses to large changes in climatic boundary
 conditions. Although the past does not contain an analog for doubled atmospheric trace gas concentrations, it
 does contain evidence of many large and complex changes in other boundary conditions (e.g^ glacial-ice volume,
 insolation, and sea-surface temperatures). The validation of climate and vegetation models can be accomplished
 via the comparison of simulated and observed changes in climate and vegetation patterns of the past. This report
 uses this approach to lend confidence to our model results that suggest future climate-induced vegetation change
 in eastern North America could be significant.

       In this report, we describe our efforts to simulate past and future vegetation change by coupling a number
 of General Circulation Models (GCMs) with two  different types of vegetation model:  (1) empirical response
 surfaces, and (2) a stand-simulation model We compare simulations of past vegetation change directly with the
 observed geological record  of vegetation change.  The use of this paleoenvironmental data is key because it
 expands our matrix of "observed" climate and vegetation to include observations reflecting the response of the
 climate system to the changing boundary conditions of the past  18,000 years. In light of the large boundary
 condition change (Le, atmospheric trace gases) that is likely to occur in the future, it is crucial that models be
 tested  in a framework of changing boundary conditions and not just against 20th century observations.  Our work
 thus constitutes an important step toward the systematic validation of climate and vegetation models that are
 being used for future impact assessment

      The  use of records of past climate and vegetation change to improve the performance of models  is an
 active  area of research.  Simulated climates are compared with climates reconstructed from  geological  data.
 Alternatively, simulated paleodimate can be input into vegetation models (both empirical and dynamic) to yield
 simulated paleovegetation that can then be compared with observed paleovegetation data (e.g^ fossil pollen data).
 Initital research has demonstrated that GCMs simulate some of the general patterns in the observed climate and
 vegetation record of the past 18,000 years.  Our work complements data-model comparisons done to date at
 GISS,  the 18,000-year BJ>. experiments done at GFDL, and the more extensive COHMAP effort involving a
 NCAR GCM (Bamosky et aL, 1987; Broccoli and Manabe, 1987; COHMAP, 1988; Hansen et aL, 1984; Kutzbach
 and Guetter, 1986; Kutzbach, 1981; Kutzbach and Wright, 1985; Manabe and Broccoli, 1977; Overpeck et aL,
 1988; Rind, 1986; Rind et aL, 1986; Rind and  Peteet, 1985; Schneider et aL, 1987; Webb et aL, 1987; Wright,
 1987).  More importantly, this data-model comparison has led to improvements in our understanding of the
 climate-vegetation system and in the way we model these systems (Webb et aL,  1987; Overpeck and Cook,  1988;
 Solomon and Shugart, 1984).

     The  record of climate and vegetation change over the past 18,000 years offers more than a unique
 opportunity for model validation.   This paleodimate/paleovegetation record also shows  how vegetation
 responded to climatic change as large as that expected in response to increased atmospheric trace gases. For
 example, the record of late Quaternary vegetation change indicates that:  (1) climate induces large spatially
 coherent patterns of change in the vegetation; (2) species respond indrviduaUsticalry to climate change; (3) the
 composition and structure of vegetation regions (e^, biomes) vary temporally; (4) vegetation regions commonly
 appear, change, and disappear through time;  and (5) climates unlike the present-day can occur and may result
in vegetation assemblages unlike those of today (Gaudreau and Webb, 1985; Huntiey and Webb, 1988; Jacobson
et aL, 1987; Overpeck et aL, 1985; Overpeck and Cook, 1988; Webb, 1987). In this report, we use ecological
response surfaces and a dynamic forest stand simulation model as two independent methods for modeling this
type of observed climate-induced vegetation change.


                                                1-2

-------
                                                                                           Overpeck


      Ecological response surfaces represent a method for quantifying the equilibrium relationship between
climate and vegetation (Austin et at, 1984; Bartlein et aL, 1986). In  contrast to climate-vegetation classification
schemes that relate the distribution of entire vegetation regions to climatic indices (e.g^ Emanuel et al., 1985;
Holdridge, 1947; Koppen, 1931), response surface methods help identify and model the relationships between
individual plant taxa and climate. We use response surfaces to model the relationship between plant abundances
and climate, but a similar approach may be used to model plant growth (Graumlich and Brubaker, 1986) or
other aspects of plant biogeography. A focus on plant abundances is particularly appropriate when modeling
vegetation at the regional to subcontinental (l(r-l(r km2) scale, which is dose to the lower resolving limit of
GCMs. After presenting a new set of response surfaces, we document that they can be used to simulate modern
vegetation patterns.  We then couple the response surfaces with GCM scenarios of 2xCO2 climate to obtain a
new assessment of how vegetation could change in eastern North America.  We obtain an independent
assessment of future change using a dynamic forest stand simulation model (Solomon, 1986). The forest stand
approach has been used with success before to simulate climate-induced vegetation change (e^, Shugart, 1984;
Davis and Botltin, 1985; Solomon et aL, 1984; Solomon, 1986). In our study, the results of the stand modeling
complement and support the results obtained using response surfaces.

      After  presenting the 2xCO2 modeling results  for the two independent vegetation models, we test the
accuracy of the models by simulating past vegetation change and then comparing the simulated results with maps
of observed fossil pollen abundance. We demonstrate that vegetation change of the past was complex, but that
the GCMs and response surfaces can be  used to simulate many aspects of this past  change. We also extend
this model validation step to test the performance of the stand simulation model Although these analyses reveal
that the models can simulate many of the aspects of observed vegetation change, they also highlight the need
for model improvement, additional paleoclimate/paleoecology research, and an expanded model validation effort
                                                1-3

-------
 Overpeck

                                            CHAPTER 2

                                     DATA AND STUDY AREA


       Four principal types of data were used in our study of vegetation-climate interactions:  (1) present-day
 observed climate data, (2) present-day observed vegetation data, (3) paleovegetation data for the past 18,000
 years, and (4) simulated climate data for the past 18,000 years and for the 2xCO2 world (ca. 2100 AD).  The
 present-day climate and vegetation data were used to generate ecological response surfaces which were then
 combined with simulated climate data to yield simulated vegetation maps for selected periods of the past 18,000
 years and for the 2xCO2 world. Simulated climate was also used as input to a stand simulation model to yield
 a second set of simulated vegetation maps.  We were  able to assess the performance of the climate and
 vegetation modeling by comparing the maps  of  simulated paleovegetatation data with maps of observed
 paleovegetation data. This model validation step is crucial because it represents the only way to check how well
 climate and vegetation models can simulate observed vegetation and climate change when forced by changes
 in climatic boundary conditions.

       Eastern North America (Figure 1) is ideal for the study of long-term climate-vegetation interactions for
 several reasons.  The  subcontinental  scale of eastern North America closely matches the relatively coarse
 resolution  of the models being used to simulate past and future climate.  Climate and vegetation data are
 available in greater number for this region than any other comparably sized region in the world. This is true
 for the recent observational period as well as for the past 18,000 years. The changes of climate and vegetation
 in this region have been studied extensively and are relatively well understood. The eastern North American
 region therefore makes an excellent study area for use in the development and testing of climate and vegetation
 models.


 OBSERVATIONAL CLIMATE DATA

      The period 1951-80  was  used for the calculation  of present-day climate normals. The eastern North
 American observational network for this period is quite dense, particularly in the U.S. portion of the region. We
 used an inverse-distance-weighting method to interpolate climate from 1328 U.S. observational stations to a 100-
 km equal-area grid.  In most cases, a 100-km search radius was used in this interpolation, guaranteeing the
 averaging of at least 3  stations at each of the 416 ILS. grid points.  In 51 cases, the lack of nearby stations
 necessitated a slightly larger search radius (maximum • 200 km, although most were close to 100 km). Gridded
 data were used because spatial data inhomogeneity and local outliers could bias ungridded data.  The gridding
 also facilitated machine map contouring.  Duplicate analyses with ungridded data were found to be more noisy,
 but essentially the same (Webb et aL, 1987).

      Time constraints required that we not use original station data from Canada. Instead, we  digitized data
 from the National Climatic Atlas of Canada (1951-80 normals). Comparison of our digitized data  with available
 station data for Canada indicated that digitizing errors were not significant  Further work will include the
 gridding of original Canadian station data as we did with the U.S. data. However, this task will be complicated
 by the low station coverage north of 50°N.


 MODERN AND FOSSIL  VEGETATION DATA

      Vegetation data for  eastern North America are extensive and available in several forms.  We chose to
work first with pollen data, but later we hope to expand our analyses to utilize complementary Continuous Forest
Inventory data (Olson et aL, 1980) and species range data (Little, 1971).  In this report, we concentrate on
pollen data for two principal reasons.  First, the pollen samples in our database (Figures  1 and 2; Avizinis and
Webb, 1983; Delcourt et aL, 1984; Jacobson et aL, 1987; Webb and McAndrews, 1976) provide representative
abundance data for the  entire eastern North American region. Over 800 samples of pollen from  the surface


                                                 1-4

-------
130
                                                                                             60'
            100*            -     90*                   «0*                    70*
 Figure 1. Locations of lakes and mires with modern samples of pollen (Avizinis and Webb, 1983).
                                           1-5

-------
 Overpeck


 sediments of lakes and mires are available from all the vegetation regions of eastern North America (Figure 1).
 Second, the use of pollen data enabled us to compare simulated taxon distributions directly with observed taxon
 distributions reconstructed using fossil pollen data from radiocarbon-dated lake and mire sediments.  The
 available fossil pollen database consists of over 10,000 samples from over 250 lakes and mires in eastern North
 America (Figure 2).  We used standard curve fitting and interpolation between the radiocarbon-dated sediment
 samples to generate maps of pollen abundances for selected time intervals (Overpeck and Fieri, 1982).

      We were able  to represent many of the broad-scale patterns of past and future vegetation change using
 the relative abundances of seven pollen types: sedge (Cyperaceae), spruce (Ficea), birch (Betula), the northern
 and southern pines (Finus), oak (QuercusX and prairie forbs (the sum of sage (Artemisia), Compositae, and
 pigweed (Chenopodiaceae-Amaranthaceae)). The northern pines were separated from southern pines by dividing
 the contemporary distribution of pine pollen at 40°N (Webb et aL, 1987). Following the convention of previous
 studies  (Bernabo and Webb, 1977; Webb et aL, 1987; Jacobson et aL, 1987), we used a pollen sum of all tree,
 shrub, and herb pollen to calculate pollen percentages. These previous studies have documented how well maps
 of pollen abundance  display the modern distribution of vegetation across eastern  North America (Figure 3).

      Although the biome-scale patterns in vegetation can be reconstructed using  pollen data, the relationships
 between plant and pollen abundances are not usually one-to-one.  The pollen production rate and dispersal
 ability varies significantly among different plant species.  Some species are under-represented by their pollen,
 whereas other species can be over-represented. Fortunately, pollen representation has been well studied (Webb
 et aL, 1981; Delcourt et aL, 1984; Bradshaw and Webb, 1985; Prentice et aL, 1987), and the relationships between
 tree and pollen abundances are approximately linear for the five tree pollen types used in this study. The plant-
 pollen relationships for sedge and prairie forbs are less well understood.  Our procedures for simulating pollen
 abundances via response surfaces and for reconstructing past vegetation can be extended to more than these
 seven pollen types (Bartlein and Webb, in prep.X but not all plants produce or disperse enough pollen to permit
 effective quantitative  analysis.


 SIMULATED CLIMATE DATA

      Output from three different GCMs  was used as equilibrium 2xCO2 climate for eastern North America:
 (1) the GISS Model U 8°xlO° GCM (Hansen et aL, 1983,1984); (2) the GFDL 4.40x7-5° GCM, and (3) the OSU
 4°x5° GCM (Ghan et aL, 1982; Schlesinger and Zhao, 1988). These model experiments are described elsewhere
 in this volume, and differ from each other in several respects other than resolution.  Because of these differences
 and  different sensitivities to a doubling of atmospheric COj,  these models provide a useful range of future
 climate  scenarios.

      No coordinated set of past, present,  and future experiments has been run with any single GCM.  For this
 reason, we used output from two additional GCMs for the purpose of climate/vegetation model validation: (1)
 the GISS Model U 4°x5° GCM (Hansen et aL, 1983) and (2) the NCAR 4.4°x7-5° Community Climate Model
 (CCM)  (Pitcher  et aL, 1983;  Ramanathan et  aL, 1983;  Kutzbach and Guetter,  1986).  We used these two
 additional GCMs to  simulate paleodimates for a number of past periods in order to explore intermodel
 differences and to see how well the models performed given a range of climatic boundary conditions unlike those
 of the present-day.  Systematic model testing of this type began with simulations of the 18,000-year B J*. Ice Age
 Maximum by the CLJMAP research group (Williams et aL, 1974; Gates, 1976a,b; Manabe and Hahn, 1977), and
 has been expanded significantly by the COHMAP and GISS groups (Webb et aL, 1985; Kutzbach and Guetter,
 1986; Webb et aL, 1987; Hansen et aL, 1984; Rind and Peteet, 1985; Rind, 1986).  The recent development of
response surfaces for pollen data was motivated by the need for a quantitative method that could be used to
compare observed paleodimate data directly with climate data simulated by GCMs (Bartlein et aL,  1986; Webb
et aL, 1987).
                                                1-6

-------
                                                                                                  45C
                                                                              POLLEN SITES
                                                                            • 6000 yr  B.P.
                                                                            012 000  yr B.P.
                                                                            D18 000  yr B.P.
                        too
60C
                                                                             TO1
Figure 2. Locations off lakes and mires with pollen data for 6,000-years B.P. (black dots), 12,000-years B.P. (open circles), and 18,000-   jp
         years B.P. (open squares). Many sites with pollen data at 6,000-years B.P. contributed data back to ca. 10,000-years B.P., and  -3
         most sites contributed data form the mapped dates to the present (from Jacobson et al., 1987).                           £

-------
           Present     Observed Pollen Data
           Present     Simulated by Observed Modern Climate
              Sedge
Spruce
Birch
N. Pines
Oak
S. Pines     Prairie Forbs
   Figure 3. Isopoll maps of observed and simulated modern pollen percentages for seven pollen taxa. The observed pollen data maps were
           constructed using more than 1,200 samples of modern pollen (Figure 1). Response surfaces were used to obtain simulated
           estimates of pollen abundance using modern climate values of mean July temperature, mean January temperature, and annual
*»          precipitation. These estimates were then machine contoured. Three levels of shading Indicate pollen percentages greater than
E-         1% (lightest), 5%, and 20% (darkest).  Portions of the grid without sufficient data were not mapped.

-------
                                                                                          Overpeck


      Global average temperature change  over  the  past  18,000-year  glacial  to  interglatial interval was
approximately 4°C (Kutzbach and Guetter, 1986), similar in magnitude to the change anticipated in response to
a doubling of atmospheric trace gas concentrations. Although this past period was obviously not a strict analog
for the future, it does provide the only opportunity to perform rigorous tests of how well climate and vegetation
models can simulate change as large as that expected in the future (Overpeck and Cook, 1988).  For this reason,
we used GCM simulations of the past 18,000 years.

      NCAR CCM experiments for four time slices  (18,000, 12,000, 9000, and 6000-years B.P) were used to
simulate how the climate of eastern North America changed in response to changing solar radiation, sea surface
temperatures, glacial ice extent and height, sea ice extent, and atmospheric trace-gas concentrations (Kutzbach
and Guetter, 1986;  Kutzbach, 1987). The CCM experiments used in this report are updated versions of those
described in Kutzbach and Guetter (1986), and include experiments with longer run  integration times and
improved boundary conditions.  The  18,000-year CCM experiment was rerun to incorporate a more realistic
glacial ice height (20% lower than in the 1986 runs) over North America. The CCM experiments were not run
with the full annual cycle.  Instead, these model runs were run for January and July only.  Kutzbach and Guetter
(1986) determined that these "perpetual" January/July experiments can adequately approximate annual-averaged
climate.

      Because the annual cycle was not simulated in these experiments, we could not use CCM output as input
to the stand simulation model The GISS experiment differed in that it was run only for 18,000-year B.P., but
with the full annual cycle.  The 4°x5° GISS GCM was run with standard CLIMAP Ice Age boundary conditions
(similar to those used in the NCAR CCM 18,000-year B J». run). Altogether, the paleodimate GCM experiments
provide a clear view of how well these models, when coupled with available vegetation models, can simulate
observed vegetation patterns of the past

      Modern (0-year B J».) "control" GCM experiments all exhibit systematic biases when compared to observed
modern climate (Schlesinger  and Mitchell, 1987; Rind,  1986; Webb et aL, 1987).  In order to  minimise  the
potential effects of  these biases in our analyses, we used simulated climate anomalies (2xCO2 or paleodimate
experiment minus control experiment) applied to the observed modern climate values. Climate values simulated
by the  GCMs therefore were  not used directly.
                                                1-9

-------
 Overpeck

                                            CHAPTER 3

                                      MODELING METHODS


 EMPIRICAL ECOLOGICAL RESPONSE SURFACES

      The distribution of vegetation on the landscape is  related to a number of endogenous and exogenous
 processes (White, 1979; Shugart, 1984). Over sufficiently long time scales (102 to 106 years), the dominant
 influence of vegetation is climate (Webb, 1986). It is for this reason that modern ecotones are broadly coincident
 with the borders of climatic regions and that the abundances of plant species appear to have tracked favorable
 environments throughout the late Quaternary (Webb, 1986). Ecological response surfaces represent an effective
 way to quantify the manner in which a plant taxon's expected abundance (the response variable) depends on the
 combined effects of several environmental predictor variables (Bartlein et aL, 1986).  Ecological response
 surfaces can thus describe the equilibrium relationship between climate and the abundances of a particular taxon.

      Bartlein et aL (1986) and Bartlein and Webb (in preparation) describe the theory and application of
 response surface methods to pollen data. Conceptually, the construction of a response surface represents the
 transformation of taxon abundances in geographic space to taxon abundances in climate space, where climate
 is defined as two or more climate predictor variables (Figure 4). For this study, we developed response surfaces
 for seven pollen types (spruce, northern pines, birch, southern pines, oak, sedge, and prairie forbs)  using three
 climate variables (mean July temperature, mean January temperature,  and mean  annual precipitation) as
 predictors. These three climate variables represent the combined general effects  of summer warmth, winter
 temperature stress, and moisture availability on plant abundances. We have begun to explore  the potential for
 other seasonal climate variables, but the results presented in this  report would probably not be changed by the
 incorporation of additional predictor variables that are correlated with the three we used.

      Ecological  response surfaces are powerful because  they are nonlinear functions describing how two or
 more climate variables jointly control the abundance of  a plant taxon.  We used various  nonlinear variable
 transformations to allow for flexibility in surface shape and then generated the surfaces using local weighted-
 average regression (Bartlein et aL, 1986; Webb et aL, 1987). Best fitting surfaces were identified, estimated, and
 diagnosed using standard statistical procedures (Bartlein et aL, 1986; Bartlein and Webb, in preparation). The
 resulting response surfaces for each of  the seven pollen taxa are unique.  The surfaces can be coupled with
 simulated values  of mean July temperature, mean January temperature, and  annual precipitation to yield
 estimates of simulated pollen abundances. This calculation was done using simulated climates from each GCM
 experiment at each of the locations with fossil data Machine contouring yielded the finished maps of simulated
 pollen abundance.


 FOREST STAND SIMULATION

      Forest  stand simulation  models have been developed and described  extensively (Botkin et aL, 1972;
Shugart and West, 1977; 1979; Shugart,  1984; Davis and Botkin,  1985). The FORENA model was developed
from these earlier models expressly to simulate the forests of eastern North America (Solomon et aL, 1984;
Solomon, 1986). We coupled FORENA with three sets of climate data: (1) modern observed climate (1951-
80) from our 100-km grid; (2) 2xCO2 equilibrium climate simulated using the GISS 8°xlO° GCM; and (3) 18,000-
year BJ*. climate simulated by the GISS 4*x5*  GCM.  The 2xCO2  experiment was run to obtain a second
independent assessment of how vegetation may change in the future, whereas the modern and  18,000-year B.P.
experiments were run to explore how well a stand simulation model could reproduce  observed vegetation
patterns. In these experiments, we used Solomon's (1986) model with his relatively moisture-insensitive soil and
with a modification that allowed the length of the growing season  to vary with the simulated input climate. We
                                               1-10

-------
                                                                                             Overpeck
      ANNUAL PRECIPITATION (mm)
                                   n
                                                  •,,-S?
                                               ••".-. i-""'" ''
                                                 ,.
                                              '. • .'*..• «».s I .   .

                                                '* *•'•'•'"'.'"' '  '
                                                           Sedg*
                                                               n
                                                                 t«
                                                             If H • If »

                                                             tt N • M Of

                                                             » M - «t ft
                                                                                 ANNUAL PRECIPITATION (mm)
                                         ^™^^~»000      '400      JOOO
                                          ANNUAL PREaPTTATION (mm)
Figure 4.  Response surfaces showing the relationship between the percentages of seven pollen types and three
          climate predictor variables:  mean July temperature, mean  January temperature, and annual
          precipitation. Two surfaces are plotted for each taxa for display purposes, but the two surfaces for
          each individual taxon were fit simultaneously using local weighted-average regression. The three-
          dimensional surfaces therefore extend over all the modern data.
                                                 1-11

-------
Overpeck
       20   -10    0    »IO
   MEAN JANUARY TEMPERATURE CO
                                                         Stdg*
I.- -pv-^b
 • •  \ »•%
                                                             M • IV

                                                           It M • »

                                                           If «• • 19

                                                           II M • M
                                  "4-rj
                                    J -JO    -iO   - .0     J    • :C

                                        MEAN JANUARY TEMPERATURE CO
                                      MEAN JANUARY TEMPERATURE CO
Figure 4.  (continued)
                                                1-12

-------
                                                                                             Overpeck


simulated 20 forest plots at each of up to 40 equally spaced ice-free grid points in eastern North America (Figure
1).  We found that the model correctly simulated no tree growth in areas of climatic extremes, including far
northern areas and the prairie. We ran each simulation for 400 years, averaging tree abundances over  the final
100 years to yield relativebiomass values at each of the grid points.  The relative tree abundances for six tree taxa
(spruce, pine, birch, oak, fir, and maple) were then mapped along with simulated total tree biomass.
                                                 1-13

-------
 Overpeck

                                            CHAPTER4

                                             RESULTS


 SIMULATED VERSUS OBSERVED MODERN VEGETATION PATTERNS

       The fitted response surfaces for the seven pollen taxa portray how the abundances of these taxa relate
 to winter temperature, summer temperature, and annual moisture availability (Figure 4). Each surface is unique
 and reveals the optimum climatic conditions for a taxon. Steep gradients on a surface reveal where a taxon is
 most sensitive to a change in climate.  In addition, the surfaces define the approximate locations in climate space
 of range boundaries for each of the seven taxa. The response surfaces clearly reveal patterns  in the vegetation-
 climate relationship that are obscure when viewed in geographic space.

       The major vegetational patterns of eastern North America are captured by the response surfaces.  Sedge
 abundances are high in the cool dry portions of Canada, whereas the southern pines are favored in the warm
 moist southeast U.S. The surface for prairie forbs exhibits a steep gradient corresponding to the prairie-forest
 ecotone, a region that could be very sensitive to future climatic change. The surfaces for spruce, northern pines,
 birch, and oak all portray well-defined optima surrounded by gradients of varying magnitude and direction. Each
 of these taxa is susceptible to different combinations of climatic change. Altogether, the surfaces reveal that
 even slight climatic change can have a significant effect on natural vegetation.

       When coupled with observed modern climate data, the response surfaces are able to simulate most of
 the patterns exhibited in maps of observed pollen data (Figure 3).  High abundances of spruce presently are
 confined to the boreal  forest of Canada, with sedge dominating the northernmost tundra areas. Two distinct
 birch populations appear on maps of simulated and observed pollen abundances.  Shrub birch populations occur
 in the northern forest-tundra and tundra regions, whereas tree birch populations give rise to high birch pollen
 percentages in the boreal and mixed conifer-hardwood forests. The high abundances of oak and prairie forb
 pollen delineate the deciduous forest and Great Plains grasslands, respectively. The mapped patterns for both
 northern and southern pines show clear agreement and define the pine-dominated biomes (southeast pine forests,
 mixed conifer-hardwoods, and the western boreal forest). Quantitative measures of map association confirm that
 the simulated pollen maps reproduce  most of the patterns in observed  pollen abundance over North America
 (Webb et aL, 1987).  The only significant exception is the poor match  of sedge abundances in the prairie, an
 anomaly probably induced by the high abundances  of local wetland sedge populations.

      Most of the same observed patterns in the modem vegetation of eastern  North America are also
 reproduced by the stand simulation model (Figure 5). In general, the model simulated little or no tree biomass
 in areas presently characterized by  treeless landscapes  (e-g, the prairie and tundra).  Simulated spruce
 abundances are high in the boreal forest along with moderate amounts of pine, birch, and fir. Too little fir is
 simulated in the northeastemmost part of the map area, and the amount of diploxyion  pine in the northern
 boreal forest  and southeast U.S. are  underestimated  by the stand  model  Both northern  hard maples and
 southern soft maples are simulated fairly well, as are oak populations. Whereas the stand model cannot simulate
 the abundances of nonarboreal taxa, such as prairie forbs or sedges, it does do a good job of simulating
 subcontinental scale patterns in the abundance of most tree taxa over eastern North America.


 FUTURE VEGETATION CHANGE - 2xCO2 SCENARIOS

      Future  equilibrium  vegetation  patterns were simulated using  the output from  three  2xCO-  GCM
 experiments coupled with (1) the seven response surfaces and (2) the stand simulation model The results from
 these two independent vegetation models are in agreement, and suggest that trace-gas-induced warming could
 cause significant change in natural vegetation over  most of eastern North America (Figures 5 and 6).  Large
warming over ail of this region in both summer and winter characterizes all three 2xCO2 climate scenarios,
                                                1-14

-------
                                                                                       Overpeck
Present   Vegetation Simulated by Stand Simulation Model
   Spruce
      Pine
Oak
Birch
Fir
Maple
Biomass
2xC
Vegetation Simulated by Stand Simulation Model
   Spruce
     Pine
Oak
Birch
Fir
Maple
Biomass
18,000 years ago     Vegetation Simulated by Stand Simulation Model
                                                         •*>.
   Spruce
     Pine
Oak
Birch
Fir
Maple
Biomass
     Figure 5.  Vegetation patterns simulated with the forest stand model using (a) modern climate values, (b) the
              GISS 2xCO, scenario, and (c) the GISS 18,000-year B.P. climate simulation.  Biomass values
              (contoured less than 50 (lightest shading), 50 to 100, and greater than  100 (darkest))  are in
              megagrams of biomass per hectare, whereas the relative tree abundances are in percent (same
              shading as in Figure 3). In each case, the biomass map outlines the area in which vegetation could
              be simulated using the forest stand model (refer to the text for the reasons for this).  The extent of
              the Laurentide Ice Sheet and associated glacial lakes is delimited by diagonal shading on the maps
              for 18,000-year BJ».
                                                1-15

-------
  Overpeck
 Present     Simulated by Observed Modem Climate
   Sedge
   Spruce
Birch
N. Pines
Oak
S. Pines    Prairie  Forbs
2xCO2     Simulated by GISS Model Output
2xC
Simulated by GFDL Model Output
2xCO2     Simulated by OSU Model Output
   Sedge
  Spruce
Birch
N. Pines
Oak
S. Pines    Prairie Forbs
 Figure 6.  Simulated modern pollen abundances and simulated future pollen abundances for each of the three
          2xCO, scenarios: (a) GISS, (b) GFDL, and (c) OSU.  Modern pollen abundances were simulated
          using instrumental climate data as input to the response surfaces shown in Figure 4, whereas the
          simulated 2xCO2 pollen abundances were generated using the same surfaces coupled with putput
          from the 2xCO2 climate scenarios. Shading is the same as in Figure 3.
                                           1-16

-------
                                                                                           Overpeck


whereas changes in simulated precipitation are more regional and less coherent among the models. Each of
these climate scenarios produced patterns of vegetation change that are generally similar, thus suggesting that
climate warming alone has the potential to cause significant future vegetation change.  Our results also suggest
that  even modest climate warming (e^,  the OSU scenario) could lead to substantial changes in the natural
vegetation of eastern North America.

      Results generated using the response surface models suggest that components of all the major vegetation
regions in eastern North America will move northward in response to increased summer and winter warmth
(Figure 6).  Model results suggest that spruce populations in the northeast U.S. will decline in abundance, as will
populations of northern pines (P. strobus and P. resinosa).  Areas now dominated by  mixed conifer-hardwood
forest could become i~T»a«ingiy deciduous in nature, particularly with large increases  in oak abundance in the
Great Lakes and New England regions. Response-surface results generated using the GISS GCM results suggest
that  the northward spread of oak could be further encouraged by drying at mid-latitudes.  A large eastward
expansion of prairie forb abundances across the northern U.S. is simulated by all three GCM scenarios, even in
areas where simulated annual precipitation is greater than today (Le., in the GFDL  and OSU scenarios).  Our
results suggest that forest biomass could decline even without significant decreases in  precipitation.  The large
simulated eastward expansion of prairie forb populations in the GISS scenario indicate, however, that regional
drying could exacerbate the spread of xeric vegetation at the expense of more mesic vegetation.

      In addition to the increase of dry/warm adapted taxa across the Great Lakes and northeast regions, the
response surface results indicate that the U.S. could also experience  a northward expansion of southern pine
populations.  Taxa that do poorly at the southern ends of their ranges (spruce and  northern pines) all expand
in abundance further north.  Birch populations are notable  because their areas of greatest abundance could
contract significantly, changing the structure of the Canadian boreal forest  Tundra  regions, marked by high
abundances of sedge pollen,  appear to suffer most west of Hudson Bay where simulated warming  favors the
expansion of tree populations. The broad agreement between vegetation patterns simulated in each of the three
GCM scenarios suggest that anticipated changes in mean winter temperature, mean summer temperature, and
mean annual  precipitation could all contribute to significant vegetation change.

      An independent assessment of possible future vegetation change was produced by coupling the FORENA
forest stand simulation model with the GISS 2xCO2 climate scenarios (Figure 5). The broad agreement between
the three climate scenarios and between the three response surface assessments (Figure 6) suggest that these
stand simulation results are probably representative of those that could be generated  using the other two GCMs.
In general, the same major patterns of change that were produced by the response surface modeling and earlier
forest stand modeling  (Solomon, 1986) were reproduced by our  forest stand simulations.  Efforts to simulate
forest growth at the southernmost grid points failed because the 2xCO2 climate was  too warm for the model
This failure of the stand model to simulate southeast forests does not imply treeless  vegetation.  Instead, this
failure suggests that future  climate  in the southeast could have no  modern analogs and that the current
generation of empirically based stand models is inappropriate for a.
-------
 Overpeck

 characterized by an expansion of prairie forbs in the 2xCO2 response surface simulations.  This result lends
 confidence to the assertion that trace-gas-induced warming and drying could contribute to significant decreases
 in total forest productivity as well as to important shifts in the species composition and structure of forests in
 eastern North America.

       It is important to note that our use of response surfaces and the stand model implicitly assumes that the
 vegetation is in equilibrium with climate. The comparison of simulated and observed past vegetation data (see
 below, and Webb et aL, 1987) indicates that vegetation has been able to track climate over the past 18,000 years.
 Future trace-gas-induced climatic change is expected, however, to be more rapid than the changes of the past
 18,000 years.  This  raises the possiblity that the future time-dependent patterns of vegetation change could be
 different from those mapped in our equilibrium assessments (Figures 5 and 6). Our assessments show how
 vegetation patterns may look after sufficient time has elapsed for individual  plant  taxa to adjust fully to new
 climatic patterns.


 PAST CLIMATES - MODEL VALIDATION

       A major difficulty in simulating climate and vegetation of the past stems from our inability to specify
 accurately how climate boundary conditions change.  This is particularly true for the deglatial interval 18,000 to
 10,000-year B.P., when several boundary conditions (e.&, glacial ice height and extent, sea surface temperatures,
 sea ice extent, and atmospheric aerosols; see COHMAP, 1988) were changing rapidly. After 10,000-years B.P.,
 the boundary conditions (including the seasonal distribution of solar radiation) were different from today, but
 the rate and complexity of change diminished, malting it easier to specify the boundary conditions for the GCM
 experiments. Climate simulations for the interval from 10,000 to 0-year BJ*. are thus likely to be more accurate
 than those for the preceding deglacial interval It is in part for this reason that we have chosen to use a number
 of GCM sensitivity experiments, each with slightly different boundary conditions, to represent 18,000-year B.P.
 In addition to providing model validation opportunities, the data-model comparison described in this section
 helps to improve our understanding of how the climate system works.

 Response Surfaces - The NCAR CCM

       When coupled with our response surfaces, the NCAR CCM-simulated climate for 18,000-year B.P. yields
 maps of simulated pollen abundances that show some resemblance to observed pollen percentages (Figure 7).
 The abundances of spruce pollen are simulated fairly well, as is the absence of southern pine populations. With
 the exception of birch, the approximate ranges of the other taxa are also simulated satisfactorily.  Observed
 abundance patterns within these ranges, however, are not simulated as  well as might be hoped.  These
 mismatches could be related  to (1) the lack  of dynamic equilibrium between vegetation and climate (Webb,
 1986); (2) shortcomings of the response surfaces; (3) inappropriate reseating of the coarse spatial resolution of
 the GCMs output to the scale of the individual fossil-pollen sites; (4) inaccuracies of the GCMs, due either to
 inadequate model design or poor specification of boundary conditions (Webb et aL, 1987); or (5) changes in the
 model biases that could  make the use of perturbation-minus-control model anomalies inappropriate.  The
 improved performance of the models in simulating more recent (e.g^ 9,000 to 0-year B.P.) vegetation argues that
 poor boundary condition specification may be the major reason for mismatch between earlier simulated and
 observed maps. The anomalously high abundances of simulated oak and prairie forb pollen at 18,000-year B.P.,
 and low abundances of simulated northern pine pollen, suggest that the CCM simulation for the southeast U.S.
 was too warm (by 2*C) for this time period (Webb et aL, 1987). The large observed abundances of prairie forbs
 at 18,000-year B J*. suggest that the climate simulated by the NCAR CCM may also be too dry for this period.
Similar analysis using a GCM 18,000-year B J*. experiment with lowered sea surface temperatures may eliminate
 much of the mismatch between maps of simulated and observed pollen abunda
      The simulations for 12,000,9,000, and 6,000-year B .P. show an increasing degree of match with maps of
observed pollen abundances (Figures 8-10). Numerical measures of association between these two sets of maps
                                                1-18

-------
        18,000 years ago    Observed Fossil-Pollen Data
        18,000 years ago    Simulated by NCAR CCM Output
                                                               •\
           Sedge
Spruce      Birch
N. Pines
Oak
S. Pines     Prairie Forbs
Figure 7. Observed and simulated pollen data for 18,000-year B.P. Responses surfaces (Figure 4) and climate data from the NCAR CCM
        were used to produce the simulated pollen data. The shadings are as in Figure 3, except that the extent of the Laurentide Ice
        Sheet and associated glacial lakes are delimited by diagonal shading. Portions of the grid without sufficient data were not
        contoured.

-------
12,000 years ago   Observed Fossil-Pollen Data
                                      :\
12,000 years ago   Simulated by NCAR CCM Output
   Sedge      Spruce      Birch        N. Pines       Oak       S. Pines    Prairie Forbs
           Figure 8. Observed and simulated pollen data for 12,000-year B.P. See Figure 7 for details.

-------
9,000 years ago      Observed Fossil-Pollen Data
9,000 years ago      Simulated by NCAR CCM Output
MMmviVM^MWv^BI''**>BV*^ •9BWWHB9W39WBWV™"'""^™™"*"^"™I i3B9BB8MBMB80Mli^^TKT^^^™^^^^^ •5BBBBS9!B3B5EBHIB)^^^™^™^™""™*"^HH!
                                                                         '\
   Sedge       Spruce      Birch        N. Pines       Oak        S. Pines     Prairie Forbs
          Figure 9. Observed and simulated pollen data for 9,000-year B.P. See Figure 7 for details.

-------
        6,000 years ago     Observed Fossil-Pollen Data
        6,000 years ago     Simulated by NCAR COM Output
           Sedge      Spruce      Birch        N. Pines       Oak       S. Pines    Prairie Forbs
•a
i
Figure 10. Observed and simulated pollen data for 6,000-year B.P. See Figure 7 for details.

-------
                                                                                           Overpeck

between simulated and observed pollen percentages is better than for earlier periods (Webb et aL, 1987), and
that the response surfaces coupled with GCM climate can do a good job of simulating observed patterns in the
paleovegetation of eastern North America. This is a strong test of model performance, because the climate
boundary conditions at 9,000-year BJP. were significantly different than the present-day (Kutzbach and Guetter,
1986). The simulated maps for spruce, birch, northern pines, oak, and southern pines all reproduce the observed
individualistic response of these types to climate change during the interval  from 12,000 to 0-year B.P.  The
simulations for sedge and prairie forbs fare less  well, perhaps because of the GCNTs inability to model
precipitation as well as temperature, or in the case of sedge,  because plants of these taxa are more likely to be
responding to local site factors south of tundra regions than to regional macroclimate. The overall match between
simulated and observed pollen percentages is encouraging, but these paleoclimatic results should also temper our
efforts to attach significance to changes in simulated pollen abundance.

Stand Simulation Model - The GISS GCM

      One  of the major advantages of using pollen data as a proxy for vegetation is the ease with which
simulated pollen data lends itself to comparison with observed paleovegetation (fossil pollen) data.  Validation
of the forest stand simulation model is more difficult than the response surface model because maps of relative
biomass cannot be compared directly to maps of  relative pollen abundances.  Eventually, the relationships
between pollen abundance and tree abundance may be quantified to the extent that these two can be compared
directly in model validation experiments. In the meantime,  however, it is still possible to obtain a qualitative
assessment of how well  the stand model works by comparing the broad patterns of simulated taxon ranges,
abundance gradients, and optima with those in the maps of fossil pollen abundance (Figure 7).

      Because the annual cycle was not simulated in the NCAR CCM experiments, we used the GISS GCM
with similar 18,000-year BJP. boundary conditions to examine how well past vegetation patterns could be
simulated using GCM output as input to the forest stand succession model (Figure 5).  Like the NCAR CCM,
the GISS model simulated a colder climate for 18,000-year BJ». In contrast, the NCAR CCM simulated a dry
southeastern U.S. at 18,000-year BJ?., whereas the GISS GCM simulated higher precipitation in this same region.
This result is not suprising and reinforces the fact that we must be careful when using GCM output to make
regional assessments. In any case, the stand simulation model appears to do less well than the response surfaces
in modeling the 18,000-year BJP. vegetation patterns.  The stand model is able to simulate high abundances of
spruce and pine south of the Laurentide Ice Sheet and no tree growth in areas the fossil evidence suggest there
were few trees.  Although the stand model simulates the correct amount of these taxa at 18,000-year BJ?., it
overestimates the past abundances of some other taxa such as birch, oak, beech, ash, and chestnut As we
emphasized before, there are several reasons why simulated and observed vegetation patterns may not match.
New GCM experiments with the annual cycle and improved specification of boundary  conditions clearly need
to be run in order to obtain a thorough validation of forest stand succession models.
                                                1-23

-------
 Oerpeck

                                            CHAPTERS

                                            DISCUSSION


 SIGNIFICANT RESULTS

       Most assessments of future climate and vegetation change will center on the development and use of
 imperfect quantitative models. One primary component of this assessment process therefore must be to develop
 systematic methods for measuring the uncertainty in the models and for improving the models. For example,
 each of the GCMs presently used to simulate the 2xCO2 world yields a somewhat different view of our future
 climate.  What parts of these simulated views can we trust? We already know that the differences among GCM
 simulations can be significant, particularly where hydrologic variables or finer geographic patterns are concerned
 (Schlesinger and Mitchell, 1987). The fact that we wish to model large climate or vegetation change implies that
 it is not sufficient to validate a model solely against modern observational data  An important result of our study
 is that paleoclimatic and paleovegetation data can form a central basis for  «Me«ing how well climate and
 vegetation models simulate change given boundary conditions unlike the present day (Schneider, 1986).

       Our direct comparisons between observed paleovegetation  (pollen)  data and paleovegetation data
 simulated with  GCMs  and  vegetation models suggest  that we can begin  to use these models to assess
 subcontinental to regional scale (105 to 10s km2) patterns of future climate and vegetation change. This is the
 approximate resolution limit of current GCMs and the scale at which we were able to simulate observed patterns
 in fossil pollen data. Even at this scale, however, it is dear that some uncertainty exists. We cannot simulate
 the response of all plant taxa to future climatic change, nor can we simulate many processes that could influence
 future vegetation change. New climate and vegetation research will have to be conducted before we will be able
 to make  highly reliable assessments of future change.

       Our model validation experiments suggest that we can make several strong inferences regarding future
 climate and vegetation change in eastern North America. Given a doubling of atmospheric CO2 concentrations,
 it appears likely that large-scale climatic wanning will characterize the next couple of centuries. This wanning
 will  probably occur in all seasons.  Significant changes in precipitation wfll also probably occur, but intermodel
 differences suggest  that we must remain less certain about the geographic and seasonal patterns of future
 precipitation changes. Our results suggest that  temperature change alone could result in significant vegetation
 change across eastern North America, and that even the most modest simulated temperature change for eastern
 North America could induce  significant vegetation change.  Precipitation change, particularly  where it tends
 toward greater aridity, is likely to increase the magnitude of future vegetation change.

      Response surfaces and forest  stand modeling suggest that plant taxa will respond individualistically to
 trace-gas-induced climatic change. Vegetation models that cannot accommodate the the response of individual
 taxa are therefore of limited utility. The most dominant and likely pattern of future change will be large-scale
 northward shifts in  plant populations. Our model validation experiments indicate that our projected range-
 boundary movements are more reliable than our assessments of future taxon abundance gradients. In any case,
 both taxon range-boundary and abundance change could be large over the next two centuries.  Our use of two
 independent vegetation models coupled with three independent 2xCO2 climate scenarios reinforces these
 conclusions. These  model experiments also suggest that significant systematic changes in total  forest biomass
 and  longitudinal  shifts in plant populations could occur.

      The eastward extension of prairie-forbs simulated in our 2xCO2 response-surface experiment coincides
with a biomass decline south of the Great Lakes in our forest stand  simulation results.  Forests in this region
might become more open or savannalike.  Both models suggest that the composition of the mixed conifer-
hardwood forests across the Great  Lakes and New England regions could experience dramatic  change,
particularly with  the significant replacement of pine populations by oak populations.  Spruce-rich forests of the
northeast  U.S. could be  in danger of serious decline.  Both vegetation models suggest that  southern pine
populations could extend northward  as climate  warms.  Populations of most  other  tree taxa are likely to be


                                               1-24

-------
                                                                                           Overpeck

affected by trace-gas-induced climate change. The results of our model validation experiments lend confidence
to the foregoing 2xCO2 assessment Caution is advised, however, against taking our results without consideration
of the uncertainties discussed below, and particularly against interpreting our results at a  scale finer than about
Hrkm2.  We are relatively secure in projecting that major vegetation change could occur over the next two
centuries, but have little basis for speculating on the exact pattern of change in any given county or township.


UNCERTAINTIES AND THE NEED FOR FUTURE WORK

      One uncertainty in our work is the lack of  information on the transient response of the vegetation to
future climatic change. Our results portray the equilibrium response of the vegetation to climatic change and
bow vegetation patterns in eastern North America  might look when vegetation change has fully caught up with
the  unprecedented fast climatic change that might occur in the future. Empirical data on important factors are
lacking- We do not know how fast plant populations can track changing climate. Although evidence of significant
disequilibrium between plants and climate is lacking in the record of the past 18,000 years, it is possible that rates
of seed dispersal and soil development could limit the rates of future vegetation change. The inability of species
to migrate to sites newly favored by climatic change could increase the likelihood of significant biomass declines
in some areas (Solomon et aL, 1984; Solomon,  1986; Solomon and West, 1987).  Our results highlight some of
the  plant taxa that might be sensitive to future climatic change and where they will be sensitive. The results of
Botltin et aL (this volume) support ours, and indicate that significant changes in the vegetation of eastern North
America might take place as early as the year 2010.

      Our response surface modeling used mean January temperature, mean July temperature, and mean annual
precipitation as predictor variables. These variables are correlated with other climate variables in eastern North
America and serve to represent the influence of winter stress, growing season warmth, and water stress on plants.
Future work must consider the effects of other variables, including climatic variance (Solomon and West, 1985;
Neilson, 1986). Our stand-simulation experiments included the influence of monthly climate data, and supported
the  results of our response surface modeling.

      The conclusion that future climate-induced vegetation change could be large is supported by the broad-
scale agreement among results generated using two independent vegetation models and three independent
climate models. More importantly, our ability to simulate aspects of future vegetation is reinforced by our ability
to simulate past vegetation. The crucial role of model validation is further highlighted, however, by the partial
lack of fit between simulated and observed pollen maps. Although it is valid to test the sensitivity of a specific
local region or site to hypothesized climatic forcing, our results clearly show that local-scale predictions are not
yet  possible.

      Our validation experiments reveal that we can only model certain aspects of vegetation with confidence.
The forest stand model simulates the establishment, growth, interaction, and death of 72 tree species, but our
results suggest that not all of these species are simulated in a realistic manner. This result supports earlier work
with forest stand models (Solomon et aL, 1980; Solomon and Shugart, 1984). Part of the problem undoubtedly
lies in the fact that our vegetation modeling did not include the influences of all climatic variables, simulated
climatic variance, realistic soil processes, forest disturbance, and other important processes (Solomon and West,
1985; Davis and Botltin, 1985; Solomon, 1986; Pastor and Post, 1986). The subtropical tree species of Florida
were also omitted from the model  These additional influences must be considered in the next generation of
vegetation models, and the ability of these models  must be tested using paleoclimate and paleovegetation data.
Our modeling results clearly demonstrate, however, that the current generation of simple models can yield useful
broad-scale assessments of vegetation change.

      The experimental design of our model validation framework is also just a beginning.  The high degree
of fit between simulated and observed vegetation  after 9,000-year B.P. argues that much of the mismatch for
earlier periods is due to poor specification of climatic boundary conditions. Paleoclimatic research is needed to
reduce this uncertainty.  GCM improvement must include (1) systematic programs to compare the results of
different GCMs and  (2) validation  of  paleoclimatic simulations  against the observed  paleoclimate  and


                                                1-25

-------
Overpeck

paleovegetation record. We also need more paleovegetation data, particularly for the period prior to 10,000-
year B J».  We have began to work on response surfaces for Continuous Forest Inventory (Olson et aL, 1980;
Delcourt et aL, 1984) and species range (Little, 1971) data.  The use of range data will allow us to extend our
empirical climate-vegetation modeling to all species for which we have range data. These models can be tested
but will require the collection and compilation of plant macrofossil data. The full power of paleovegetation data
for testing models was  only touched upon in this report

      One major uncertainty regarding future vegetation change is  the rate at which this change will occur.
Our results only illustrate how large this change could be over the next 100-200 years and the direction that this
change will probably follow.  Even though we have models capable of simulating time-dependent change (e^,
the stand-model), we are limited in our ability to judge their accuracy.  The same holds true for vegetation
models that simulate the direct effects of rising trace-gas concentrations (e-g, CO, "fertilization") on plants.  At
this time, there are not  even enough data to access the potential impact of these direct effects on natural forests
(Strain, 1985; Solomon, 1986). Careful paleodimatological and paleoecological research could eventually provide
a means for testing the ability of models to simulate time-dependent vegetation change.
                                                1-26

-------
                                                                                           Overpeck

                                            CHAPTER6

                                          CONCLUSIONS


      The results presented in this report suggest that state-of-the-art climate and vegetation models can be
used to assess broad-scale patterns of past and therefore future vegetation change.  Plant taxa will respond
indrvidualisticaQy to trace-gas-induced H""at* change and these responses could be «igniRrafit across eastern
North America.  We have demonstrated how paleodimatic and paleoecological data can be  used to identify
model uncertainties. Our results highlight the need for further ecological, paleoecological, paleoclimatological,
and modeling research.  All future assessments of potential trace-gas-induced climate and vegetation change
should include provisions for model validation.
                                                1-27

-------
Overpeck
                                    ABBREVIATIONS
       CCM                Community Climate Model (NCAR)



       GCM                General Circulation Model



       GISS                Goddard Institute for Space Studies



       NCAR               National Center for Atmospheric Research



       OSU                Oregon State University



       year BJ*.             years Before Present
                                           1-28

-------
                                                                                        Overpeck


                                         REFERENCES


Austin,  M.P., R.B.  Cunningham, and  P.M. Fleming.  "New approaches  to  direct gradient analysis using
environmental scalars and statistical curve-fitting procedures." Vegetado 55,11-27,1984.

Avizinis, J. and Webb, T. in (1983). The computer file of modern pollen and climatic data at Brown University.'
(unpublished manuscript).

Barnosky, C.W, P.M. Anderson, and PJ. Bartlein. The northwestern U.S. during deglaciation; vegetational
history and paleodimatic implications." In: North America and Adjacent Oceans During the Last Deglaciation,
WJF. Ruddiman and H.E. Wright, Jr, eoX Geological Society of America, Boulder, Colorado, 1987.
pp. 289-321.

Bartlein, P J., I.C Prentice, and T. Webb m. "Climatic response surfaces from pollen data for some eastern
North American taxa." Journal of Biogeograpby 13: 35-57,1986.

Bernabo, J.C and T. Webb m. "Changing patterns in the Holocene pollen record  from northeastern North
America" Quaternary Research, 8: 64-96,1977.

Botltin, DJJ., JJF. Janak, and JJL Wallis. "Some ecological consequences of a computer model of forest growth."
Journal  of Ecology, 60: 849-872,1972.

Bradshaw, RJLW. and T. Webb  m. "Relationship between contemporary pollen and vegetation data from
Wisconsin and Michigan." Ecology 66: 721-737,1985.

Broccoli, AJ and S. Manabe. The influence of continental ice, atmospheric COj, and land albedo on the climate
of the last glacial maximum." Climate Dynamics 40:1410-1425,1987.

COHMAP Members. "Climatic changes of the last 18,000 years: observations and model simulations." Science
241,1043-1052,1988.

Davis, M.B.  and DJBJBotltin. "Sensitivity of cool-temperate  forests and their fossil pollen record  to rapid
temperature change." Quaternary Research 23: 327-340,1985.

Delcourt, P A, HJR. Delcourt, and T. Webb m. "Atlas of mapped distributions of dominance and modern pollen
percentages  for important tree  taxa of eastern North America."  American Association of Stratigraphic
Paiynologists Contribution Series 14:1-130,1984.

Emanuel, W.R., H.H. Shugart,and MX. Stevenson. "Climate change and the broad-scale distribution of terrestrial
ecosystem complexes." Climatic Change 7: 29-43,1985.

Gates, W.L. "Modeling the ice-age climate." Science 191:1138-1144,1976a.

Gates, WX. The numerical simulation of ice-age climate with  a global general circulation model" Journal of
Atmospheric Sciences 33:  1844-1873,1976b.

Gaudreau, D.C. and T. Webb m. "Late-Quaternary pollen stratigraphy and isochrone maps for the northeastern
United States." In: Late-Quaternary North  American Sediments,  VJVt. Bryant Jr  and R.G. Holloway, eds.
American Association of Stratigraphic Patynologists, Dalas Texas, 1985, pp. 247-280.

Chan, S J. et aL "A documentation of the OSU two-level atmospheric GCM model" CRI Report 35,1982,395
pp.


                                               1-29

-------
 Overpeck

 Graumlich, LJ. and  L.B.  Bnibaker. "Reconstruction of annual   temperature  (1590-1979) for  Longmire,
 Washington, derived from tree rings." Quaternary Research 25: 223-234,1986.

 Hansen, J., A. Laos,  D. Rind,G. Russell, P. Stone, I. Fung,  R. Ruedy, and J. Lerner. "Climate sensitivity;
 analysis of feedback mechanisms." In: Climate processes and climate sensitivity, J.E. Hansen and T. Takahashi,
 eds. American Geophysical Union, Washington D.C, 1984, pp. 130-163.

 Hansen, J., G. Russell, D. Rind, P. Stone, A.  Laos,  S. Ledeff, R. Reudy, and L. Travis. "Efficient three-
 dimensional global models for climate studies: Models I and U' Monthly Weather Review 111: 609-662,1983.

 Huntley, B. and T. Webb, m. Vegetation Dynamics. Kluwer, Dordrecht, 1988. Holdridge, LJl. "Determination
 of world plant formations from  simple climatic data." Science 105: 367-368,1947.

 Jacobson, GI~, T. Webb, m and E.C. Grimm. "Patterns and rates of vegetation change during the deglariation
 of eastern North America." In: North America and Adjacent
 Oceans During the Last Deglaciation,  W.F. Ruddiman  and H.E. Wright,   Jr,  eds^ Geological Society of
 America, Boulder, Colorado, 1987. pp. 277-288.                                           '

 Koppen, W. "Grande der Klimakunde Berlin." Walter de Gruyter,  1931. Kutzbach, J.E. "Model simulations of
 the climatic patterns during the deglaciation of North America." In: North America and Adjacent Oceans During
 the Last Deglaciation, WJ. Ruddiman and HE. Wright, Jr, eds. Geological Society of America, Boulder,
 Colorado, 1987. pp. 425-446.

 Kutzbach, JJL "Monsoon climate of the early Holocene, climatic experiment using the earth's orbital parameters
 for 9,000 years ago." Science 214: 59-61,1981.

 Kutzbach, J.E. and P J. Guetter. The influence of changing orbital parameters and surface boundary conditions
 on climate simulations for the past 18,000 years." Journal  of the Atmospheric Sciences 43:1726-1759,1986.

 Kutzbach,  J.E. and H.E. Wright, Jr. "Simulation of  the climate of 18,000  yr  BP:  results for  the North
 American/North Atlantic/European sector." Quaternary Science Reviews 4:147-187,1985.

 Little, EI~, Jr. Atlas of United States trees, VoL L Conifers and important hardwoods. U.S. Department of
 Agriculture, 1971.

 Manabe, S. and AJ. Broccoli The influence of continental ice sheets on the climate of an ice age." Journal of
 Geophysical Research 90: 2167-2190,1985.

 Manabe, S. and D.G. Hahn. "Simulation of the tropical climate of an ice age." Journal of Geophysical Research
 82: 3889-3911,1977.

 Neilson, R .P. "High-resolution diamtic analysis and southwest biogeography." Science 232: 27-34,1986.

 Olson, RJ, CJ. Emerson, and M.K. Nungasser. "GEOECOLOGY: a county-level environmental database for
 the conterminous United States." ORNL/TM-7351, Oak Ridge National Laboratory, Oak Ridge, TN, 1980.

 Overpeck, J.T. and EJl.Cook. "A Quaternary perspective in assessing how future trace-gas-induced climate
 change might effect natural vegetation." Quaternary Science Reviews, (in preparation) Overpeck, J.T. and E.C.
 Fieri. The  development of age models  for Holocene sediment cores: northeast North American examples."
American Quaternary Association Abstracts 7:152,1982.

Overpeck, J.T, L.C.  Peterson, N. Kipp, J. Imbrie, and D. Rind. "Climatic change in the tircum-North Atlantic
region during the last deglariation."  Nature (submitted).
                                               1-30

-------
                                                                                          Overpeck

Overpeck, J.T., T. Webb HI, and LC Prentice. "Quantitative interpretation of fossil pollen spectra: dissimilarity
coefficients and the method of modern analogs." Quaternary Research 23: 87-108,1985.

Pastor, J. and W.M. Post "Influences of climate, soil moisture, and succession on forest carbon and nitrogen
cycles." Biogeochemistry 2: 3-27,1986.

Pitcher, EJ., R.C Malone, V. Ramanathan,  Mi. Blackmon, K.  Puri,  and W. Bourke. "January and July
simulations with a spectral general circulation model." Journal of the
Atmospheric Sciences 40: 580-604,1983.

Prentice, I.C, BJL Berghmd, and T. Olsson. "Quantitative forest-composition sensing characteristics of pollen
samples from Swedish lakes." Boreas 16: 43-54,1987.

Ramanathan, V, EJ. Pitcher, R.C. Malone, and MX. Blackman. "The response of a spectral general circulation
model to refinements in radiative processes." Journal of Atmospheric Sciences 40, 605-630,1983.

Rind, D. "The dynamics of warm and cold climates" Journal of the Atmospheric Sciences 43: 3-24,  1986.

Rind, D. and  D. Peteet  Terrestrial conditions  at the last  glacial maximum and CLJMAP sea-surface
temperature estimates: are they consistent" Quaternary Research 24: 1-22,1985.

Rind, D., D. Peteet, W.S. Broecker, A. Mclntyre,  and W. F. Ruddiman. "Impact of cold North Atlantic sea
surface temperatures on climate: Implications for the Younger Dryas cooling (ll-10k)." Climate Dynamics 1:
3-33,1986.

Schlesinger,  MJL and JJ?J). Mitchell*. "Climate model simulations of the  equilibrium climatic response to
increased carbon dioxide." Reviews of Geophysics 25: 760-798,1987.

Schlesinger, MJL  and Z. Zhao. "Seasonal climate changes induced by doubled CO2 as simulated by the OSU
atmospheric GCM/mixed-layer ocean model" CRI report, 1988, 84 pp.

Schneider, S Ji. "Can modeling of the ancient past  verify prediction of future climates? An editorial." Climatic
Change 8:117-119,1986.

Schneider, SJL, DM. Peteet, and GJL North. "A  climate model intercomparison for the Younger  Dryas and
implications for paleodimatic data collection." In: Abrupt Climatic Change, W Ji. Berger and LD. Labeyrie, eds.
D. Reidel Publishing Company, Dordrecht, 1987, pp. 399-417.

Shugart, HJf.  and D.C West "Development of an Appalachian deciduous forest succession model and its
application to assessment of the impact of the chestnut blight"  Journal of Environmental Management 5:
161-170,1977.

Shugart, H.H. and D.C. West "Size and pattern of simulated forest stands." Forest Science 25: 120-122,1979.

Shugart, H Ji. A theory of forest dynamics. Springer- Verlag, New York, 1984

Solomon, AM. Transient response of forests to CO^induced climate change: simulation modeling experiments
in eastern North America." Oecologia 68: 567-579, £986.

Solomon, AJM., HR. Delcourt, D.C West, and T J. Biasing. Testing a simulation model for reconstruction of
prehistoric forest stand  dynamics." Quaternary Research: 14, 275-293,1980.
                                                1-31

-------
 Overpeck

 Solomon, AJVI. and H.H. Shugart Integrating forest stand simulations with paleocological records to examine
 long-term forest dynamics." In:. State and change of forest ecosystems - indicators in current research, G.I.
 Agren, ed Swed Univ. Agric. Sd,  Uppsala, Sweden, 1984, pp. 333-356.

 Solomon, AJrt., MX. Tharp, D.C West, GJL Taylor, J.M. Webb, and J.C. Trimble. "Response of unmanaged
 forests to  CCyinduced  climate change:  available  information, initial tests and data  requirements."  US.
 Department otEnergy, Washington, DC, 1984.

 Solomon, A.M. and  D.C. West "Potential responses  of forests to CCyinduced climatic change." In:
 Characterization of information requirements for studies of CO-effects: water resources, agriculture, fisheries,
 forests, and human health, MJR. White, ed. DOE/ER-0236, U^. Department of Energy, Washington, DC, 1985.
 pp. 145-169.

 Solomon, A31. and West, D.C "Simulating forest ecosystem responses to expected climate change in eastern
 North America: applications to decision malting in the forest industry" In: The greenhouse effect, climate change,
 and U.S. Forests, W.E. Shands and J.S. Hoffman, eds. The Conservation Foundation, Washington, 1987, pp.
 189-217.

 Strain,  B.R.  "Physiological and ecological  controls on carbon  sequestering  in  terrestrial ecosystems."
 Biogeochemistry: 1, 219-232,1985.

 Webb, T. m "Is the vegetation in equilibrium with climate?  How to interpret  late-Quaternary pollen data."
 Vegetatio 67: 75-91,1986.

 Webb, T. m. "The appearance and disappearance of major vegetational assemblages: long-term vegetational
 dynamics in eastern North America." Vegetatio 69:177-187,1987

 Webb, T. m, B J. Bartleuvmd J.E. Kutzbach. "Climatic change in eastern North America during the past 18,000
 years; comparisons of pollen data with model results." In: North America and Adjacent Oceans During the Last
 Deglatiation, WF. Ruddiman and HE. Wright, Jr., eds. Geological Society of America, Boulder, Colorado,
 1987. pp. 447-462.

 Webb, T. m, S.E.  Howe,  RJE. Bradsbaw,  and ICM. Heide.  "Estimating  plant abundances from pollen
 percentages: the use of regression analysis." Review of Paleobotany and Palynology 34: 269-300,198L

 Webb, T. m, J. Kutzbach, and FA. Street-Perron. "20,000 years of global climatic change: paleodimatic research
 plan." In: Global Change, T.F. Malone and J.G. Roeder, eds. ICSU Press, 1985,  pp. 182-218.

 Webb, T. m and JM.  McAndrews. "Corresponding patterns of contermpory pollen and vegetation in central
 North America." Geological Society of America Memoir 145: 267-
 299,1976.

 White, P.S. "Pattern, process, and natural disturbance in vegetation." The Botanical Review 45: 229-299,1979.

Williams, J., R.G. Berry, and W.W. Washington. "Simulation of the atmospheric circulation using the NCAR
global circulation model with ice age boundary conditions." Journal of Applied Meteorology 13: 305-317,1974.

Wright, HJL, Jr. "Synthesis; the land south of the ice sheets." In: North America and Adjacent Oceans During
the Last Deglaciation,  W.F. Ruddiman and HE. Wright, Jr., eds. Geological Society of America, Boulder,
Colorado, 1987. pp. 479-488.
                                               1-32

-------
EFFECTS OF CLIMATE CHANGE ON FORESTS OF THE GREAT LAKE STATES
                           Dr. Daniel B. Botltin
                           Dr. Robert A. Nisbet
                            Tad E. Reynajes
                          University of California
                         Santa Barbara, CA 93106
                       Contract No. CR-814595-0-10

-------
                                  CONTENTS
ACKNOWLEDGMENTS	 iii

FINDINGS  	 2-1

CHAPTER 1: INTRODUCTION: BASIC ISSUES	 2-5

CHAPTER 2: METHODOLOGY	 2-6
      THE FOREST MODEL	,	 2-6

CHAPTER 3: CHOICE OF WEATHER STATIONS AND LOCATIONS	 2-8
      CREATION OF EXPERIMENTAL WEATHER RECORDS 	 2-8
      EXPERIMENTS	 2-9

CHAPTER 4: RESULTS  	2-11
            Climatic Changes	2-11
            Natural Forests: The Boundary Waters Canoe Area	2-11
      IMPACTS OF CLIMATIC CHANGE	2-17
            Southern Region of the Great Lake States	2-17
                  Transition from Current Conditions	2-17
                  Comparison with Steady-State Conditions 	2-20
      ECONOMIC FORESTRY	2-23
      LIMITS OF THE RESULTS AND RESEARCH NEEDS 	2-27
      POLICY IMPLICATIONS	2-27
      SUMMARY 	2-29

REFERENCES	2-30

-------
                                    ACKNOWLEDGMENTS

     This work was supported in part by the following: the U.S. Environmental Protection Agency grant CR-
814595-0-10; NSF Grant DEB80-12159,1980 - 83; and a grant from the Andrew J. Mellon Foundation to D.B.
Botkin. Results dp not necessarily reflect views of any of these, and no official endorsement should be inferred
from this publication. Mr. Jon Bergengren and Dr. Lloyd Simpson of UCSB also contributed to this work.
                                              in

-------
                                                                                             Botkin


                                            FINDINGS1


      There is widespread concern over the climate change that may be induced by increases in the carbon
dioxide concentration in the atmosphere, but there have been few ways to assess the impact of such changes
on natural ecosystems.  Recently developed global climatic models project that mid-latitudes would experience
pronounced wanning and drying out of soils, which suggests that there might be major changes in mid-latitude
forests.  But the Irind and degree of such changes in vegetation have not been subject to quantitative evaluation.
One of the few methods available for such an evaluation is computer simulation. Fortunately there is a well-
established model of forest growth which has been shown to be realistic and has been used to study the response
of forests to long-term climatic change. We  report  the application of this model to evaluate effects on forests
of the Great Lake States to changes in temperature and rainfall induced by changes in the CO2 concentration
of the atmosphere.

     Two representative sites were considered, one in the southern and one in the northern portions of the
Great Lake States.  In the southern part, weather records from Mt  Pleasant, MI, were  chosen to represent a
heavily settled area where commercial forests are still an important economic resource. Forests in this area are
transitional between northern hardwoods and oak-dominated forests. In the northern portion, weather records
from Virginia, MN, were chosen to represent heavily forested areas  in and near the Superior National Forest
(where commercial forestry has been important) and the Boundary Waters Canoe Area  (BWCA), a nationally
designated wilderness area (important for recreation and biological conservation).

     The climate change is projected to lead to major changes in the forest composition, that is, in the species
of trees which dominate the forests, as illustrated in Figure 1 and Figure 2. In the north, boreal forests may be
replaced during the next 90 years by northern hardwood forests, now characteristic of areas to the south. Effects
depend on soil type and soil water conditions.  In the Boundary Waters Canoe Area, areas where balsam fir
dominates and upland areas where white birch or quaking aspen are now dominant may be converted to forests
dominated by sugar maple; white cedar bogs may be converted to treeless bogs (Figure  1).

     In the south, hardwood forests that are transitional between sugar maple-dominated hardwoods and oak
forests may be converted to oak woodlands or savannahs, which occur to the south under current conditions, or
even to treeless prairies, which occur much farther to the west (Figure 2).

     Wood production and the accumulation of total biomass may be greatly affected, but the effect depends
heavily on soil and soil water conditions. Forests on dry sandy soils in the southwestern part of the region may
be converted to prairie and savannah in which no significant wood production takes place.  Soils with abundant
soil water may continue to support trees but with a somewhat lower wood production and biomass.

     In the north, cedar bog land may be converted to a treeless  bog, while upland areas, in contrast, may
undergo an increase in wood production and biomass accumulation  where saturated soils are made somewhat
drier and better as sites for tree growth.

     The climate  change would have major effects on the forest  industry in the Great  Lake States.  This
industry is currently adapted for a certain complement of species, primarily for softwoods used in the production
of paper pulp and construction materials.  The species  that would become most economically important under
        'Although the information hi this report has been funded partly by the U.S. Environmental
Protection Agency under contract no. CR-814595-0-10, it does not necessarily reflect the Agency's views, and
no official endorsement should be inferred from it
                                                2-1

-------
 Botkin
                    Before

                     White Birch     Balsam Fir

                    After
                           Sugar Maple

                     /{  r^A)  *£^4^\
Figure 1.   Diagram of predicted change in forests of northern Minnesota beginning with the current climate
           in 1980 ("Before" in the diagram) and ending after undergoing the climatic change as projected by
           the GISS model transient A, 90 years later in 2070  CAfter" in the diagram).
                                                2-2

-------
                                                        Before
                                                                                         Botkin
                                                        After
                                                                              •1002
                                                                                   0}
Figure 2.  Diagram of predicted change in forests of southern Michigan beginning with the current climate in
          1980 ("Before" in the diagram) and ending after undergoing the climatic change as projected by the
          GISS model transient A, 90 yean later in 2070 CAfter" in the diagram).
                                              2-3

-------
Botkin

twice CO2 steady-state climate would be hardwoods such as oaks and maples, useful for furniture and other
decorative purposes, which would have much longer rotation times (harvesting could be done less frequently)
than the softwoods. Thus there would be a major shift in the character of the forest industry whose costs should
be evaluated; the shift would require different equipment and markets.

      Results from tests run to date suggest that significant changes in the forests could occur as early as the
year 2010 and as late as 2040.

      The time and funds available for this study placed a limit on the work that could be done, and it would
be advisable to extend the work in the following ways:

      (1)  test the sensitivity of the results to the value of parameters in the forest growth model, such as the
maximum longevity of trees and the temperature limits of growth;

      (2) test the sensitivity of the results to certain aspects of the climatic projections, including the combination
of weather records used as a basis for the simulation, and the calculation method to determine the ratios used
to modify control climate to -««iiii«t«» modified «*n«?iat**t;

      (3) consider the interaction between the climatic effects considered here and other environmental factors
including fire, pathogens, insects, air pollution, UV-B radiation fluxes, acid rain and gaseous oxidants, human
management and human-induced disturbances, as wefl as the frequency, intensity, and  duration of extreme
weather events including wind, flooding, and drought. In particular, it is especially important to determine how
sensitive the projected timing of changes in forest composition under the transient climatic regimes is to specific
values of parameters in the forest models and to key assumptions in the climate models.
                                                 2-4

-------
                                                                                             Botltin

                                           CHAPTER 1

                                 INTRODUCTION:  BASIC ISSUES


      Growing concerns with the possibility of a major climatic change due to the increase in carbon dioxide
in the atmosphere as a result of human activities have led to a need to predict the possible effects of such a
climate change. An important issue is the response of forests and woodlands to such a change.  Forests and
woodlands  are important not only as commercial crops for lumber, and pulp, but also in urban areas  as
recreational urban parks, in rural areas for multiple uses inrJwfing- soil conservation and erosion control; water
supply; habitat for wildlife; maintenance of streams in a form and shape that can support populations of fish; as
recreation for camping, fishing, hunting, nature viewing; for aesthetic values; and for conservation of biological
diversity. Woodlands containing endangered species and forests designated as wilderness preserves are of special
legal concern.  Finally, forests play an important role at a global level in terms  of chemical cycling throughout
the biosphere and in climate dynamics. Forest trees and soils together are estimated to store more carbon than
is found in the atmosphere (WoodweU et aL, 1977).  Under support from EPA, we have begun to investigate
the effects of global warming on the forests of the Great Lake States. The results of the work are reported here.
                                                2-5

-------
 Botkin

                                           CHAPTER 2

                                         METHODOLOGY


      The TIMBER! forest model ((C) 1983,1988 by D.B.BoUdn and T.E. Reynales), based on the JABOWA
 forest model developed by D.B. Botkin, J. Janak, and J. Wallis (1972X  was used to investigate the possible
 effects of climate change on forests of the Great Lake States.

      Projected climatic changes made use of the following global climate dynamics models, whose output was
 modified especially for these  studies by Dr. R. Jenne of NCAR: GISS  "normal" climate; GISS "twice CO,"
 climate; GISS Transient A and Transient B; GFDL "twice CO2" climate; and the OSU "twice CO2" climate. The
 outputs from these models were used to modify real weather records, which were obtained in computer form
 for stations in the Great Lake States.3 Note that this report concerns the  response of forests to climate change
 induced by of an increase in CO2 concentration; direct physiological effects of atmospheric CO2 increase on tree
 growth is not simulated.


 THE FOREST MODEL

      In 1972 Botkin, Janak, and Wailis first reported the  development of a computer model of forest growth
 (Botkin et aL, 1972X which has since been shown accurate  and realistic and has been applied to forests around
 the world (West et aL, 1981; Prentice, 1986). Other forest models that are now often referred to as "gap-phase"
 models are derived from and essentially identical to earlier versions of JABOWA. For example and for purposes
 of comparison for the EPA study, the model discussed by Shugart (1984) is derived from our model and to the
 best of our knowledge contains most of the algorithms from the version of JABOWA in Botkin et aL (1972).
 One of these derivative models has been used to investigate some aspects of CO-- induced climate change
 (Solomon and West, 1983), but these use climate projections from Mitchell (1983).  The TIMBER! model
 incorporates a number of advances including (a) a more complete method of handling the relationships between
 water and tree growth, making possible a separation of floodplain communities from bog and other wetland
 communities;4 (b) relationship between nitrogen concentration in the soil in terms of and tree growth;5 and (c)
 provides a more realistic treatment of growth and reproductive rates among species. TIMBER! also incorporates
 40 species of trees, which are the major native trees found in the northern hardwoods and boreal forests of
 eastern and mid-western North America and all of the tree species found north of Connecticut since the end
    2 Botkin, D.B., J.F. Janak and JJL Wallis. 1973. Some ecological  consequences of a computer model of
forest growth. J. Ecology 60: 849 - 871


    3 The weather records were provided in computer format by R. Jemw of NCAR.

    4This follows methods developed by Botkin, D.B., and TLE. Levitan. 1977. Wolves, moose, and trees: an
age specific trophic-level model of Isle Royale National Park. IBM Research Report in Life-Sciences RC
6834, 64 pp.

    5 This follows methods reported in Aber, J.S., D.B. Botkin and J.M. Melillo. 1978. Predicting the  effects
of different harvesting regimes on forest floor dynamics in northern hardwoods. fanM Ji Forest Research 8:
306-315.

                                               2-6

-------
                                                                                             Botkin
of the last major period of gladation. (A few minor species that are never dominant in the Great Lake States
are not included, such as boxelder and bur oak.)0

     JABOWA-type models have been used to simulate forest growth in many parts of the world (Shugart,
1988), and for many purposes. The JABOWA model has been applied to consideration of the effects of aad
rain on forest growth (Aber et aL, 1978; Aber et at, 1979). A version of TIMBER! was used  to investigate the
effects of climate change since the end of the last ice age on forests of New England (Davis and Botlon, 1985).

     Pertinent to this EPA study, one of our earliest applications of JABOWA was to simulate the effects of
CO- fertilization of forests resulting from the atmospheric increase of CO2 due to the burning of fossil fuels and
decrease in forest  and soil organic matter (Botkin et aL, 1973).

     JABOWA-tvpe models simulate growth of individual trees on small plots; trees compete for light, water,
and soil nutrients^ their growth is affected by environmental conditions including mean monthly temperature
and precipitation, sofl depth, characteristic soil particle size, and soil fertility. An assumption of the model is that
each plot is small  enough so that a large tree shades every other tree on the plot  Each year the growth and
mortality of individual trees and reproduction of species are determined based on competitive and environmental
conditions. Aspects  of the model are stochastic, so that the model can be used to consider variability within a
forest and to generate statistical means,  variances,  and confidence  intervals.   The  model has  been well
documented and verified elsewhere (Botkin et aL, 1972,1973; Davis  and Botkin, 1985).
    8 The species considered in this model are sugar maple (Acer sacchamm): yellow birch (Betula
alleeheniensis): white ash (Fraxinus ffllTlJT?!!^ mountain maple (Acer soicatum): striped maple (Acer
pensvlvanicum): pin cherry (Prunus pensvtvanica): choke cherry fPrunus virginiaV balsam fir (Abies
balsamea): white birch (Betula panvrifera): mountain ash (Sorbus ajp^ficapa: red maple  (Acer rubrum):
scarlet oak (Ouercus coccinea): hornbeam (Carpinus spp.); green alder (AJous. crispa): speckled alder (Alnus
ruqosa): black ash (Fraxinus nigra): butternut qyylana cjn£££aj;wbite spruce (Picea ylauca): black spruce
(Picea mariana): jack pine (Pinus bankmanafe red pine (Pinna resinosa): white pine (Pinus strobus): trembling
aspen (Populus tremuloides): white oak (Quercus alba): northern red oak (Ouercus rubra): white cedar
(Thuia occidental^; hemlock (Tsuga ranjjy1
-------
 Botkin

                                            CHAPTERS

                       CHOICE OF WEATHER STATIONS AND LOCATIONS


      Two weather stations were selected for the studies reported here:  Mt Pleasant, MI, and Virginia, MN.
 At each station 30-year records (1951-1980) were used. It is important to understand that the results reported
 here represent the responses of forests in the general area of each station and not only forests exactly at those
 stations. This is because the output from the climate dynamic models is for large areas and because the forest
 model described above can consider a wide variety of forests in many different habitats, with different soils and
 water relations.

      Mt Pleasant was chosen to represent the impact of climatic change on woodlands  in settled areas, on
 hardwood forests of commercial use, and on woodlands useful for other multiple use purposes.  In addition,
 our results indirectly provide insights into the possible response of certain fruit trees.  Mt Pleasant lies in the
 southern portion of the Great Lake States, in a region which was forested  in presettlement times by northern
 hardwoods forests  (forests typically dominated by sugar maple on good, well-drained soils), but near to  a
 transition to oak forests to the south.  Today this area is heavily settled.

      Virginia, MN, as chosen to represent  the impact  of climatic change on forest growth in on coniferous
 woodlands of commercial use and on a wilderness area.  Forests in this area are of commercial value for pulp,
 paper, and construction. In addition, Virginia is near the Superior National Forest, managed for multiple uses
 including timber sales, and it is near the Boundary Waters Canoe Area, one mQlion acres of federally designated
 protected wilderness important for recreation and biological conservation. The Boundary Waters Canoe Area
 is especially useful  because its history since the end of  the last ice age has been reconstructed (Heinselman,
 1970), the frequency of natural forest  fires is well known (Heinselman,  1973X and  the distributions and
 abundances of major tree species are comparatively well understood. In addition, we have done extensive field
 work in portions of this area (Botkin et aL, 1984; Hall et aL, 1989 in press). Virginia, MN, lies in the northern
 part of the Great  Lake States in the transition between northern hardwoods  forests and the boreal forests
 (typically dominated by spruce and fir, with aspen, white birch, and pines common in younger forests and drier
 sites).


 CREATION OF EXPERIMENTAL WEATHER RECORDS

      Weather records were prepared for control (normal) conditions and treatment conditions using theoretical
 output from climatic dynamic models to modify existing  weather records. The assumption implicit in creation
 of treatment conditions was that the yearly and seasonal variation of treatment climates was adequately simulated
 by the variation observed in the actual weather record. This assumption may be in error since the period of the
 Little Ice Age between the  late 1400*s and the mid 1800's was characterized by greater than usual variation in
 temperature and precipitation (Bryson and Murray, 1977).

      During control runs (normal CO2) the TIMBER! model used the actual 30-year weather  record, with
 the record repeated for simulations of more than 30 years. To model forest growth in the Boundary Waters
 Canoe Area where there are no weather stations,  the Virginia,  MN, weather records were  modified slightly to
 represent an area to the north as we have done previously for other sites (Botkin et aL,  1972). In this case  it
 was assumed that the area to the north would experience weather equivalent to increasing the elevation above
 Virginia  and  using standard temperature and precipitation lapse  rates; this method has been used  by us
 previously.

      The model is stochastic, allowing consideration of the variations to be expected in a forest  Multiple
experiments with identical  initial conditions and  weather patterns over time can be used to project mean,
standard errors, and confidence intervals. In the work reported  here, 60 replicates were obtained for each trial
Each replicate began with 1951 data and followed the same weather sequence.


                                                2-8

-------
                                                                                             Botkin
      In each treatment, projected weather records were prepared as follows: the ratios between the model
"normal" steady-state output and the COU-ennanced climate were calculated by R. Jenne and provided to us in
computer format  Ratios were calculatedior mean monthly temperature and mean monthly rainfall.  The actual
mean monthly temperature and precipitation were then multiplied by  the  appropriate ratios to generate
"treatment" climates. It must be emphasized that the  model results represent projections of forest response
based on the climate projection input to the model  We do not assume that the model projections will actually
occur, but rather, the projections represent various scenarios that might occur if any of the climate simulations
occur.   The various climate models  represent the state-of-the-art in climate simulation;  greater certainty
associated with projected forest responses can be expected as the climate models improve.

     The "normal" output from a climate model represents steady-state conditions assuming that the present
concentration of carbon dioxide is maintained. Twice CO2" represents a steady-state condition in which carbon
dioxide is twice current concentrations in the atmosphere.  Transients" provide a simulation of weather in a
transition from the model projections of climate under current CO2 concentrations to future climatic conditions.
These results were used to generate ratios of change for real weather records.
EXPERIMENTS

     The following experiments are reported: the growth of forests from a clearing were compared for (1)
1951-1980 climate; (2) GISS steady-state; (3) GFDL steady-state; (4) OSU Steady-state; (5) GISS transient A;
and (C) GISS transient B. Each experiment was replicated 60 times and the results were averaged and the 95%
confidence intervals were calculated. The southern portion of the Great Lake States was represented by Mt
Pleasant weather records, and for these weather records growth of forests from f^aringy was considered. The
northern portion of the Great Lake States was represented by Virginia, MN. weather records, modified to
represent the Boundary Waters Canoe Area and nearby Superior National Forest; for these, growth both from
clearings and of old age (400-year) forest stands was considered.  Four soil types were used:

     l)deep comparatively dry soils;
             soil depth - LO meters
             depth to water table

                       White birch
                       Yellow birch
                       White spruce
       2)    deep comparatively wet soils;

             soil depth
             depth to water table

                       Balsam fir
                       Sugar maple
- L2 meters

- 3378 * 56 cm2/100m2 basal area
  12.6 ± 0.2 stems/lOOm2

- 93 i 29 on2/100m2 basal area
- 1.9 ± 0.6 stems/lOOm2

- 911 L3 onVlOOm2 basal area
- 4.910.01 stems/lOOm2
  LO meters
  0.8 meters

  2768 t S3 cm2/100m2 basal area
  8.2 ± 0.15 stems/lOOin2

  224 112 on2/100m2 basal area
  4.0 10.05 stems/lOOm2
                                                2-9

-------
 Botkin

         3)   shallow wetland soils;

              soil depth                     - 0.5 meters
              depth to water table           - 0.2 meters

                        White cedar         - 2693 t 93 cm2/100m2 basal area
                                            -  1.0 ± 0.03 stems/lOOm2

         4)   shallow dry upland soils;

              soil depth                     - 0.5 meters
              depth to water table           - 0.6 meters

                        Trembling aspen     - 2078 t 72 cm2/100m2 basal area
                                            -  LO10.03 stems/lOOm2

                        Balsam fir          -  11311 cm2/100m2 basal area
                                            -  231 0.13 stems/lOOm2

                        Sugar maple         -  27811 cm2/100m2 basal area
                                            -  LO t 0.01 stems/lOOm2


      These soil types support a broad range of forest conditions from old age cedar bogs and old age balsam
 fir stands to regrowth of aspen on thin sandy soils. Using the monthly temperature and precipitation estimates,
 the model generates moisture conditions using a modified Thornthwaite water balance method (Thomthwaite,
 1948; Sellers, 1965) as described in Botkin et aL  (1973).

      Note that the transients involve projections from the present into the 21st century, while the steady-state
 experiments represent conditions under which the CO2 concentrations were to remain constant at twice the
 present atmospheric concentrations, a condition that is not expected to occur (the CO2 concentration will
 continue to vary further into the future X The steady-state expectations are of interest, however, to compare
 expected growth under current conditions with the effects of a doubling in COg.  Under each set of weather
 conditions, trials were conducted for two major soil types: deep comparatively dry soils and deep relatively wet
 soils.8  These two soil conditions  represent generally good sites for  forest growth under current  climatic
 conditions; most soils would be shallower than  these and therefore have less water available for tree growth or
 else, in wetland situations, be more saturated  with water and subject tree roots to stress because of lack of
 oxygen.  These relatively good conditions were chosen on  purpose to weight the results toward the more
 optimistic side. The soil particle size chosen is on the sandy side, which is common in the Great Lake States.
 A soil with a finer average particle size (a loam soil) would be somewhat more fertile and would have a better
 water holding capacity. Clay soils tend  to lead to saturated conditions, leading to a decline in forest growth
 under the normal climatic regimes used  in this work. The fertility of the soil used in the  experiments is high,
 but could be made even higher; however, to our knowledge, it represents a quite fertile natural soil of the region.
    sln the parts of Michigan and Minnesota considered in these simulations, the following species do not
occur  grey birch, eastern red cedar, red spruce, pitch pine, and (for Minnesota) beech; blights on American
elm and chestnut will exclude these as well; therefore, seed sources would not be available. These species
were therefore excluded from the simulation experiments described here; 33 species could enter any plot if
the site conditions were appropriate.

                                                2-10

-------
                                                                                            Botkin

                                           CHAPTER 4

                                            RESULTS
Climatic Changes
     The climate change is strong. Projected steady-state alterations of the climate for one year are shown in
Figure 3.  The GISS, GFDL, and OSU doubled CO2 outputs convert ML Pleasant records from winters that
commonly have months with mean temperatures below freezing to series of years with winters in which no month
has an average temperature below freezing.  The GISS and GFDL models increase January temperatures
approximately 10*F, whereas the mean January temperature never exceeded the mid 20s CF). From 1950 to 1980,
the projected range for Mt Pleasant January temperature ranged from the high 30s CF) to the mid 20s CF).
In other words the 1951-1980 maximum January mean value becomes the minimum value  hi the projected
climatic warming. The OSU model projects January means that are only slightly colder.

     Similar increases occur for the July mean temperature.  For the GISS model, average temperatures range
from the mid 70s CF) to above 80*F, in contrast to the 1951-1980 range from the high 60s CF) to the  mid 70s
CF). These changes in temperature and rainfall would have great effects on forests and other vegetation. Such
summer increases would greatly increase evaporation of water from soils and trees, and lead to a much drier
climate. The effects on rainfall are much less pronounced, with a slight decrease in rainfall in some years.

     The GFDL gives  considerably more severe  projections for summer temperatures, with July means
exceeding 90°F and never descending below 82°F.  The OSU model projects somewhat cooler July mean
temperatures than the two  other models, with most values remaining in the low to mid 70s CF).

     The GISS model transient A increasingly deviates from the actual climate so that July mean temperatures
show a definite increase after the first decade (Figure 4A). Transient B, which has projections for only 60 years,
shows increases on the order  of several degrees Fahrenheit by the fifth decade (Figure 4B).  Such  summer
increases would greatly increase evaporation of water from soils and trees, and lead to a much drier  climate.
The effects on rainfall are much less pronounced, with a slight decrease in some years.

     Projected effects are given for three models, GISS, GFDL, and OSU. Values are the average  monthly
temperature for the year 1951 for the actual weather record CNbrmaTX  and for this temperature record as
modified by the steady-state twice CO2 climate for each of three climate models. The altered temperature was
calculated by multiplying the actual mean monthly temperate by the ratio of the treated to  normal steady-state
climates for each model

Natural Forests: The Boundary Waters Canoe Area

     The transient climatic regimes lead to surprisingly rapid changes which are very sensitive to soil moisture
conditions; using the Giss transient-A climate beginning with 1980 conditions, the model predicts that a significant
change in forest composition would occur in the Boundary Waters Canoe Area by year 2010 and transient B by
year 2040. By 2010 under transient-A conditions, a 400 year-old stand dominated by balsam fir on deep, fertile,
moist soil (Figure 5) would decline to one-third of the balsam fir basal area under the 1951-1980 weather regime.
Sugar maple replaces fir as the dominant species, and the total btomass nearly triples.  In a wetland, a 400 year-
old white cedar forest with a total basal area of 2685 cnr/lOOnr declines to a nearly treeless bog with total basal
area of only 134 cnr/lOOm2 (Figure 6). On a deep, drier but fertile sandy upland soil the white birch dominant
declines to about 10 percent of its starting level in about 40 years, and is replaced by a sugar maple forest (Figure
7\ with a decline hi total biomass accumulation to 16 kg/m2, about half the biomass of 28 kg/m2 under normal
conditions after 90 years (Figure 8).  These dramatic effects of the transient-A climate  change also suggest that
we should be able to find and document some boreal forests that are currently changing to sugar maple forests.
                                               2-11

-------
Botkin
                              MT.  PLEASANT,  Ml
  CJ

  Ul
  QC.

  \—
  <

  Ul
  Q_
                                           —- GISS
                                                GFDL
                                            ---- OSU
                                                           8
                                                              10    11     12
                                    MONTH  (1951)
FigureS.
The projected effects of a doubling of the CO- concentration in the atmosphere on temperature for
Mt Pleasant, ML Projected effects are given for three models, GISS, GFDL, and OSU. Values are
the average monthly temperature for the year 1951 for the actual weather record ("Normal"), and
for this temperature record as modified by the steady-state twice CO2 climate for each of three
climate models. The altered temperature was calculated by multiplying the actual mean monthly
temperature by the ratio of the treated to normal steady-state climates for each model
                                           2-12

-------
    VI
    LJ
    O
    (O
    UJ
    Ul
    K
    O
                           MOUNT PLEASANT

                       JULY MEAN TEMPERATURE

                       GISS MODEL TRANSIENT A
                                             Botltin
         30
         20-
         15
                          NORMAL CLIMATE
                          TRANSIENT A CLIMATE
          1980      2000      2020      2040

                                 YEAR

                       GISS MODEL TRANSIENT B
         30
         25--
         20--'
         15
          1980
                      	 NORMAL CLIMATE
                      	TRANSIENT B CLIMATE
2000
2020
2040
                                 YEAR
                               2060
                               2080
2060
2080
Figure 4.  Temperature change during the next decades. These temperature changes are for the July mean
        temperature for Mt Pleasant, MI ("Normal") and as modified by the GISS climatic model transient
        A and transient B.
                                  2-13

-------
 Botltin
                         8000
                    o*


                    E


                    2   eooo 4-
                    o*
                    N

                    E
                    u
                              BWCA

                    (MODIFIED VIRGINIA. UN DATA)

                           BALSAM FIR

          Soil Depth a  1 .Om        Woter Table D«pth = 0.8m
                        BASAL AREA


              O	O NORMAL CLIMATE

              •	• GISS TRANSIENT A CLIMATE

              A	A GISS TRANSIENT B CLIMATE
   4000
                            1980
                                                          2080
 Figure 5.   Changes in forest composition during the next century for a deep, wet sandy soil in the boundary

           waters canoe area.  The projections shown here are for a 400-year-old balsam fir stand that is

           characteristic of northern Minnesota and throughout the northern portion of the Great Lake States.
                                                   BWCA

                                          (MODIFIED VIRGINIA, MN DATA)

                                                WHITE CEDAR

                                Soil Depth = 0.5m        Water Table Depth = 0.2 m
                      a-
                      n
                     O
                     O
o-
M


E
u
                          5000
    4000
                          3000
                                                BASAL AREA
            O
        O NORMAL CLIMATE

        • TRANSIENT A CLIMATE

     — A TRANSIENT B CLIMATE
                     £   2000
                                            •o—..-,-- -o—o
                     V)



                     CD
                          1000
                             04
            (Note - Initial conditions:

             400 yr. old forest)
1980
2000
                            2020
                                                           2040
2060
2080
                                                      YEAR
Figure 6.   Changes in forest composition during the next century for a cedar bog in the boundary waters canoe

          area. The projections shown here are for a 400-year-old white cedar bog which is characteristic of

          certain water-saturated soils in northern Minnesota and throughout the northern portion of the Great

          Lake States.
                                             2-14

-------
                                                                                       Botkin
                o
                o
o-
M
E
o
                    8000
                    6000
                    4000
                          Soil Depth = 1.0 m
                             BWCA
                    (MODIFIED VIRGINIA, MN DATA)
                          WHITE BIRCH
                                   Water Table Depth = 1.2 m
                         ••
                m
                     O
                     •
                     A'
                  BASAL AREA
                     NORMAL CLIMATE
                  • TRANSIENT A CLIMATE
                  A TRANSIENT B CLIMATE
                                      (Note - Initial conditions:
                                              400 yr. old forest)
                                                   /
2000
n .
•v
">L
• — •
---A- 	 .
	 • 	 •
A °— o— °\c_xr
•— i^.— .^.
1980      2000      2020      2040

                       YEAR
                                                                2060
                                                         2080
                                              BWCA
                                    (MODIFIED VIRGINIA. MN DATA)
                                           SUGAR MAPLE
                           Soil Depth = 1.0 m	Water Table Depth = 1.2 m
                 o
                 o
                 E
                 o
                 LJ
                 or
                 
                 m
                     8000
                     6000
                     4000
                     2000
                           BASAL AREA

                   •	• TRANSIENT A CLIMATE
                   A —A TRANSIENT B CLIMATE
                       (Note: No Sugar Maple in
                             normal climate)

                       (Note - Initial conditions:
                              400 yr. old forest)
                                  2000
                    2020     2040

                        YEAR
                                                2060
2080
Figure 7.  Changes in forest composition during the next century for a deep, dry sandy soil in the boundary
          waters canoe area. The projections shown here are for a 400-year-old forest on a deep, dry sandy
          soil which, in this case, is dominated by white birch in 1980.  Such stands are typical in northern
          Minnesota and throughout the northern portion of the Great Lake States.
                                             2-15

-------
Botkin
(A)
















(B)

















BWCA
(MODIFIED VIRGINIA. MN DATA)
100-
^K
o- 80
M
E
» 60
<«^^
VI
< 40
a
Sell Death s 1.0 m Water Table Deoth = 0.8 m

BIOMASS

O 	 O NORMAL CLIMATE
• 	 • GISS TRANSIENT A CLIMATE
A 	 A GISS TRANSIENT B CLIMATE •
^x*

X
o •'
• 20f— S*=58^,A 	 |^-f— A 	 0^<
J
0 •











)


1980 2000 2020 2040 2060 2080
YEAR

BWCA
(MODIFIED VIRGINIA. MN DATA)
.__ Sell Death * O.S m Water Table Depth a 0.6 m
lOO -I
^+
cr 80 •
•
,6
• 60
VI
2 40
a
o
* 20-

BIOMASS

O 	 O NORMAL CLIMATE
• 	 • GISS TRANSIENT A CLIMATE
A 	 A GISS TRANSIENT B CLIMATE

o-^0 	 o-^.
O^^f^*^ ^>^
^^ A o— (

[^£*-~*~~r~*^r~*~'~









)


1980 2000 2020 2040 2060 208C
                                                YEAR
Figure 8.   Changes in biomass for (A) the balsam fir stand shown in Figure 5, and (B) a dearout stand on a
           thinner, drier soil representative of the boundary waters canoe area and northern Minnesota
                                                2-16

-------
                                                                                             Botkin

     While the three climate models differ considerably in their projections, all projected twice CO2 steady-
state climates lead to major changes in vegetation.  In the BWCA, natural fire frequency has been once a
century (Heinselman, 1970) and early successional stands are common on the landscape;" thus consideration
of growth from clearings is important On deep, relatively moist soil, wetlands that now develop into
larch-dominated bogs regrow to red maple-dominated wetlands characteristic of warmer areas. Thus both old-
age and regrowth forests change from boreal to northern hardwoods.

     These conclusions appear to contradict those proposed by some workers (Wigley et aL, 1980), who have
suggested that an increase in CO- concentration would  augment tree growth because increases in the COj
concentration in  area above a potted plant in a greenhouse lead to great increase in seedling growth. This
consideration alone has led some scientists to believe that changes in forest composition during the next 25 to
30 years  were expected  to  be  only  barely measurable.   However, greenhouse and controlled-enyironment
experiments usually fail to consider the considerable effect of CO2-induced climate change on site conditions that
is projected by our work.10 CO2 fertilization may indeed increase growth of greenhouse plants, but these results
are difficult to extend to  the natural environment, where many factors operate to control plant growth.

     As  a consequence of these major changes in vegetation, large alterations would occur  in the forest
ecosystems; chemical cycling, storage of organic matter,  and rates of decomposition differ  between conifer-
dominated boreal forests and the northern hardwood forests. The flux of chemical elements from forests to
streams could change. The habitat for wildlife would be  altered, and one would expect the  dominant species
of wildlife to change with the vegetation.  For example, areas suitable today to moose would become favorable
to white-tailed deer. The entire character of the BWCA  as a wilderness area would be altered.

     The response of the forests is very sensitive to soil  conditions. This is because the climatic changes  lead
to a great increase in evapotranspiration; although rainfall increases, the evaporative losses increase more and
water becomes limiting.  This explains why upland sites  can be converted to savannahs  while  wetlands can
maintain  substantial, if reduced, forest growth of species presently dominant further south.


IMPACTS OF CLIMATIC  CHANGE

Southern Region of the Great Lake States

Transition from Current  Conditions

     Under the 1951-1980 actual climate, following a simulated clearing in 1980, a normal  forest develops in
which red oak, sugar maple, white oak, and white pine are dominants on drier sites (Figure 9), and sugar maple,
red maple, white ash, and basswood are dominants on wetter sites (Figure 10) throughout the first 100 years'
growth. In transient A climate, which contains projections for 90 years, sugar maple does not  occur on the drier
sites which are dominated by pin cherry in the earliest stages, then by red and white  oak and red maple (Figure
9B). On the wetter sites sugar  maple develops during the first decades but declines almost  to the point of
           on analyses of existing forests stands by remote sensing, carried out by F. Hall, D.B. Botkin, D.
Strebel, and S. Goetz.

    10 The general trend of our results showing forest responses to double CO- over a 100 year period are
similar to those of another study of Michigan forests by Solomon and West, using a model derived from
JABOWA which does consider climate interaction and immediate doubling of COj.  However, Solomon and
West used only the GFDL model, projecting a more severe wanning and drying trend then the GISS models
and predicting future forests composed of even drier species like pine (Solomon and West, 1987). Simulating
forest ecosystem responses to expected climate change in Eastern North America:  Applications to decision
making in the forest industry. li The Greenhouse Effect, Climate Change, and U.S. Forests. W.E. Shands
and JJS. Hoffman, eds. The Conservation Foundation, WaslL, D.C. pp. 189-217).

                                                2-17

-------
 Botkin
        MOUNT PLEASANT, Ml
NORMAL CLIMATE. DEEP DRY SANDY SOIL
      (A)
                          Soil Depth =  1.0 m
                    Water Table Depth =  1.2 m
u-
vt
E
0
0
^
V)

u
U!
1
m

i ^\j\j •

1250

1000


750
500
250
n
O 	 O Sugar Maple O 	 0
A 	 A Red Oak V 	 T




4

AxA
/^°^X=^
.^i^J^?8^'5^
1980 2000 2020 2040
White Pine
White Oak


±
/
t

^o-o_,
^^^^




i




>


2060 2080
      (B)
                                               YEAR
                                       MOUNT PLEASANT, Ml
                                   GISS TRANSIENT - A CLIMATE
                                       DEEP DRY SANDY SOIL
              o-
              M
              o
              o
              I/I
              E
              u
                  1500
                  1250
                  1000
                         Soil Depth = 1 .0 m
                    Water Table Depth =  1 .2 m
              m
                                2000
2020      2040

    YEAR
                                 2060
2080
Figure 9.   Development of a forest from clearing in the southern portion of the Great Lake States. The forest
          represented here is near Mt Pleasant, NO, on a deep, dry, well-drained, sandy soil for (A) north
          (1951-1980 Mt Pleasant weather records) and (B) these records as modified by the GISS transient
          A climate, represent the transition from  current  conditions to climate under ingn»a.«mg CO2
          concentrations.
                                           2-18

-------
                                       MOUNT PLEASANT, Ml
                               NORMAL CLIMATE, DEEP WET SANDY SOIL
                    Botkin
    (A)
^5 4000 -
VI
E
o
2 3000
cr
VI
E
,0 2000
Ul
QC.
Soil Depth a t .0 m
O 	 O Sugar Maple
• — • Red Maple





s
Water Table Depth = 0.8 r£
+ 	 * Basswood
• 	 • White Ash
o/
s
Q
/
>°
o
X






              3
              03
                  1000
2000     2020      2040

               YEAR
2060
                                                                          2080
     (B)
                                       MOUNT PLEASANT. Ml
                                   GISS TRANSIENT - A CUMATE
                                        DEEP WET SANDY SOIL
                         Sj>il Depth = t.O m        Wafer Table Depth = 0.8 m
tr
M
£
o
o
X
o*
M
u
X^1
<
QC.
-J
v\
aa
•twv
3000

2000


1000-
0 J
19
0 	 O Sugar Maple A 	 A Red Oak
• 	 • Red Maple • 	 • White Ash



.CX

^^•—•—9 \
^§^1— »^« • A*
30 2000 2020 2040 ?nen ™
                                                                          2080
                                               YEAR
Figure 10.  Development of a forest from clearing in the southern portion of the Great Lake States. The forest
          represented here is near Mt Pleasant, MI, on a deep, wet, well-drained, sandy soil for (A) north
          (1951-1980 Mt Pleasant weather records) and (B) these records as modified by the GISS transient
          A climate, represent the transition from current  conditions to climate under increasing CO2
          concentrations.
                                            2-19

-------
 Botkin


 disappearance by the end (Figure 10B).  In transient B, which provides projections for 60 years, sugar maple
 follows a similar pattern as transient A on the wetter sites.

       On the drier site, forest productivity under transient A climate during the first SO years following clearing
 matches the normal productivity, while under transient B the productivity exceeds normal for the first 50 years.
 Afterwards, as the climatic warming effects increase, there is a decline by year 2040 under both transient A and
 transient B, with the biomass reaching a very low value by year 2070 (Figure 11).

       The forest represented here is near Mt Pleasant, MI on a deep, dry, well-drained, sandy soil for (A)
 north (1951-1980 Mt Pleasant weather records), and (B) these  records as modified by the GISS transient A
 climate, represent the transition from current conditions to climate under increasing CO2 concentrations.

 Comparison with Steady-State Conditions

       The transient results are consistent with the projections for twice CO2 steady-state climates under all
 three models. This is illustrated here for results from the Mt Pleasant weather records. The three steady-state
 models differ in the magnitude of their projections, with GFDL  giving the most severe changes and QSU the
 mildest changes.  For example, on the drier site for the Mt Pleasant region, total biomass accumulation  after
 100 years  of growth exceeds 20  kg/m ,  while  under the GISS and OSU twice CO2 climates the biomass
 accumulation after  100 years is less than  10, and under the  GFDL twice CO2 climate there is negligible tree
 biomass (Figure 12A).

       There is greater biomass accumulation on the wetter site under all steady-state climates, but the GFDL
 climate leads to a much smaller value, approximately 10 kg/m2 compared to the normal forest whose biomass
 is approximately 50 kg/m2 at year 100 (Figure 12B). Under dry soil conditions, the three climate models lead
 to predictions of a major change in dominant  species and therefore in forest type, from a northern hardwood-
 oak transition to an oak forest with red maple. With the GISS model twice CO, steady-state climate (Table 1),
 the forest model results in an open savannah forest of low biomass (dropping from  an average of 25 kg/m2 to
 less than 2 kg/m  of tree biomass) dominated by oaks which  are sparse and small even after 100 years (Figures
 9 and 10).  While the normal forest would have commercially useful hardwoods, the forest  under the steady-
 state altered climate would not produce commercially  useful hardwoods even  after 100 years.   From an
 ecological perspective, this is a very severe effect Under such a shift, wildlife species would change from those
 adapted to the more northern closed forests to those adapted to grassland- savannah. Sugar maple and other
 dominant species of the northern  hardwoods forests disappear.

      The GFDL model gives an  even more extreme effect; remaining at year 100 are  many very small trees
 which contribute almost no biomass to the plot; red and white oak and red maple contribute most of the small
 amount of basal area For example, even at year 100 there is only 13.6 cm  /100m2 red oak, equivalent to a single
 tree with a diameter of 4 centimeters at breast height (a sapling in other words).  Assuming grasses could survive
 under these conditions, then the GFDL predicts that the forest would be converted to a sparse savannah or a
 grassland with sparse, very small trees (Table  1). Eventually, it is also possible that tree species with an  even
 more southern distribution might migrate into  the area. This last possibility could be investigated by expanding
 the current species list available to the forest model

      Although the OSU model projects  the  mildest climatic change, its twice CO2 climate results with the
 forest model in a sparse open forest with  6 ± 3 kg/m2, a forest that would appear to vary from a savannah  to
 an open woodland. Consistent among the projections for the three steady-state climate is a considerable decline
 in biomass accumulation after the  third decade on the drier site (Figure  12A) and after the seventh decade on
 the wetter site (Figure 12B).  The GISS and GFDL climates result in negligible biomass accumulation on the
drier site for all time periods.  This suggests that upland areas would typically be converted to savannahs or open
woodlands.
                                                2-20

-------
                                                                                       Botltin
      (A)
                                       MOUNT PLEASANT. Ml
                                         DRY SANDY SOIL
                         Soil D.oth a 1.0 m        Wot«r TobU D«pth = 1.2 m
               E
               o»
               VI
               (SI
               <
               2
               o
      (B)
      40


      30


      20
                       10 +
                        1980
                     BIOMASS

             O	O NORMAL CLIMATE
             •	• GISS TRANSIENT A CLIMATE
             A	A GISS TRANSIENT B CLIMATE
2000      2020      2040

              YEAR
                                               2060
                                                  2080
cr
in
E
\
o>
.*

(/)
(/>
<
3
O
CD
70

60

50

40

.30

20

10
                                        MOUNT PLEASANT. Ml
                                          WET SANDY SOIL
                           Soil Depth = 1.0 m
                                   Woter Table Depth  = 0.8 m
                                O
                                •
                                A
          BIOMASS

   -O NORMAL CLIMATE
   •• GISS TRANSIENT A CLIMATE
   -A GISS TRANSIENT B CLIMATE
         1980      2000      2020      2040

                                YEAR
                                                                 2060
                                                   2080
Figure 1L  Biomass changes during the development of a forest from clearing in the southern portion of the
          Great Lake States.  The forest represented here is near ML Pleasant, MI, on a deep, dry, well-
          drained, sandy soil for (A) north (1951-1980 ML Pleasant weather records) and (B) these records
          as modified by the GISS transient A climate, represent the transition from current conditions to
          climate under increasing CO2 concentrations.
                                            2-21

-------
 Botkin
             Table  1.   Forest Conditions After  100 Years  of Growth
 Forest conditions after 100 years of growth  from  clearing on a deep dry sandy
 soil.   Soil depth 1.0  m;   depth  to water table 1.2 m (A) (dry soil) and (B) 0.8
 m (wet soil);  soil moisture holding capacity 150.00 mm/m of soil;
 soil nitrogen 150.

 Treatments are: Normal: growth using Mt. Pleasant, MI. 1951-1980 weather records;
 GISS,  GFDL, and OSU: steady-state twice CO, climates as  modified by each of these
 models;  TR-A,  a transient projected by the GISS model from current climate to
 future climate.
 (A)  Deep  Dry Sandy Soil
TREATMENT BIOMASS

NORMAL 25 ±3.7
GISS 1.9 ± 0.3
GFDL 0.1 ± 0.02
OSU 6.8 ± 3.1
TR-A1 4.9 ± 2.4
SUGAR MAPLE
BASAL AREA.
444 ± 53
0
0
0
0
RED OAK
BASAL AREA
1193 ± 152
118 ± 11
6 ± 1
232 ± 49
196 ± 53
RED MAPLE
BASAL AREA
136 ± 71
150 ± 24
10 ± 1
323 ± 122
164 ± 83
 (B) Deep Wet Sandy Soil
TREATMENT

NORMAL
GISS
GFDL
OSU
TR-A'
BIOMASS

52 ±
33 ±
12 ±
28 ±
12 ±

2.0
4.9
1.9
3.5
2.0
SUGAR MAPLE
BASAL AREA.
4008 ± 166
0
0
0
157 ± 33
RED OAK
BASAL AREA
108 ± 23
176 ± 34
53 ± 7
231 ± 39
239 ± 85
RED MAPLE
BASAL AREA
664 ± 199
2117 ± 389
1097 ± 157
1177 ± 292
458 ± 165
Units are: Biomass Kg/m2
Density: Number of trees/ 100 m3;
B.A.: basal area (cross sectional area)  cn2/100m2
(all values are mean ± S.E. for  60 replicates).

     ' Note that this  transient climate model output is available
          for only 90 years; year 90 is  shown here  for
          comparative purposes.
                                      2-22

-------
                                                                                             Botltin

        On deep, well-watered sites with sandy soils (soils with a saturated zone but a well aerated layer above),
the forest under normal conditions is transitional with sugar maple and red maple with other wetland and
floodplain species including white ash and hemlock, suggesting abundant water for tree growth (Table 2). The
GISS model twice CO2 climate results in a forest dominated heavily by red maple with some oaks present; none
of the northern hardwood forest species occur (Table 2). Red maple increases in basal area under the three
steady-state twice CO2 climate regimes.

     Under the 1951-1980 actual climate, the normal forest, sugar maple continues to increase throughout the
100 years, while no sugar maple grows in any of the steady-state twice CO2 climates.  Even in transient A
climate, which contains projections for 90 years, sugar maple declines almost to the point of disappearance by
the end.  Only in transient B, which provides projections for 60 years, does sugar maple remain (transient A
suggests little change in the abundance of sugar maple by the end of 60 years).

     Again the GFDL model shows a more severe  effect (Table 2). Red maple  and white and red oaks are
again dominant at  year 100. but the biomass  averages 12 t 4 kg/m2 (mean * 95% confidence interval) in
comparison to 33 110 kg/m2 under the GISS twice CO- steady-state climate, 271 7 kg/m2 under the GFDL
twice CO2 steady-state climate, and 52 ± 4 kg/m2 projected for the 1951-1980 climate.
     Under the GFDL twice CO2 steady-state climate, the basal area of the dominant species, red maple is
half that under the GISS climate.  Under the OSU twice CO2 steady-state, the forest has a little more than
half the biomass of the normal forest, but the basal area on red oak remains about the same as under the
GFDL climate.

     The results for the northern portion of the area are similarly consistent between the transients and the
steady-state climatic projections, and the steady-state climates maintain the same order of severity for the forests,
with GFDL given the largest and most severe change, GISS next, and OSU the least severe.


ECONOMIC FORESTRY

     The results described here concern harvests of forests allowed to undergo natural regeneration, a common
practice for  hardwood forests of the  Great Lake States. Even under the best of conditions,  the species
composition may change from a forest with the potential for growth of sugar maple, yellow birch, and white pine
to a forest with a potential to grow red and white oak.  On drier sites, forest productivity may  drop  below
economically useful levels.  For example, under the GISS twice CO2 steady-state climate, the biomass of 100-
year-old stands on sandy soils drops to l/10th that of the normal forest, to below 2 kg/m2 above ground biomass
(see Table 3).

     A 50 year rotation forest with natural regeneration produces an average of 13 kg/m2 in the normal forest
but less than 2 kg/m2 under the GISS twice CO2 steady-state climate.  Under the GFDL there is no harvestable
forest biomass at year 50 or at year 100 on the drier site (Table 3). On the wetter site most of the biomass is
in red maple, a species not of economic importance (Table 3). Thus the forest model projects strong  effects
under the modified climates for commercial forest productivity.

     The reduction  in number of days with hard frosts may have implications  for commercial fruit crops
including cherry and apple.  The increase in temperature and resulting increase in actual evapotranspiration
relative to potential evapotranspiration would also affect orchards, requiring a great increase in irrigation water
if the crops were to be continued. It is important to investigate the possible effects of the climate change on
orchards.
                                                2-23

-------
Botltin
(A)
\ J


V
P
X.
0»
J£

E
£
I/I
M
^
0
on




rigure 12. Biomas!

MOUNT PLEASANT, Ml
DRY SANDY SOIL
,« Soil Death a 1 .0 m Water Table Depth a 1 .2 m
60

SO

40

30
20


BIOMASS
O 	 O NORMAL CLIMATE
• 	 • GISS 2XC02 CLIMATE
A 	 A OSU 2XC02 CLIMATE

(Note: less than 0.6 kg/m sq.)
in GFOL climate.) Q 	 	 < )
^^.O""""1'
10-' /•s**"*"^"' A ~ i
1980 2000 2020 2040 2060 2080
YEAR

MOUNT PLEASANT. Ml
WET SANDY SOIL
,0 Soil Deoth * 1 .0 m Water Table Deoth * 0.8 m
/ W
60

50
40
30

20
10-
BIOMASS
O 	 O NORMAL CLIMATE
• 	 • GISS 2XC02 CLIMATE
A 	 A GFOL CLIMATE J>
A 	 A OSU CLIMATE o
°^°^ 1
A^"* T
^fc^^^^ ii ^^^"* *"^^™^ •• ^^" *™ •• ^"*11"* «W
•«k^5^^^ 1
^^r^y _
^§^,A^ ^-A— A— \A_,k
jjl^^^*2*^
1980 2000 2020 2040 2060 208C
YEAR
t accumulation during 100 years of growth under stead-state climates :
Mt Pleasant, MI, on (a) dry sandy soil, and (B) wet sandy soil
                                    2-24

-------
                                                                          Botltin
                           Table 2
            DOMINANT TREE SPECIES FOR MOUNT PLEASANT, MI
            Deep Relatively Wet Sandy Soil   YEAR:   100

MOUNT PLEASANT. MI
1951-1980 WEATHER DATA
Depth to the Water Table 0.8m; Soil Depth 1.0 m;  soil moisture holding capacity
150.00 mm/m of soil; soil nitrogen 150; 60 replicates.

DOMINANT TREE SPECIES   DENSITY (#/100mJ)   BASAL AREA (cm'/lOOm*)
                     AVERAGE   95%CI     AVERAGE    95%CI
  (1)  NORMAL C02
     1  SUGAR MAPLE       10.5      .77      4008.3     332.6
    13  RED MAPLE           .2      .12       664.1     397.3
     4  WHITE ASH           .2      .13       298.4     198.8
    36  BASSWOOD           2.7      .45       299.9     111.2
    30  HEMLOCK            2.7      .50       179.2      58.7
    28  NORTHERN RED OAK    .5      .22       107.8      46.3
    26  TREMBLING ASPEN     .1      .08        88.9      71.4

  (2)  TWICE CO, (GISS  MODEL)
    13  RED MAPLE          1.3     .64       2116.9     777.0
    39  BLACK CHERRY       1.7     .87        334.5     136.0
    27  WHITE OAK          2.2     .56        260.4      67.5
    28  NORTHERN RED OAK   1.6     .64        176.3      68.7

  (3)  TWICE CO, (GFDL MODEL)
    13  RED MAPLE          2.4     .55       1097.2     314.1
    27  WHITE OAK          3.8     .65         96.7      21.5
    28  NORTHERN RED OAK   2.4     .47         53.1      14.8

  (4)  TWICE CO, (OSU MODEL)
    13  RED MAPLE           .5     .26        1177.2     583.2
    39  BLACK CHERRY       1.7     .56        1001.4     271.4
    27  WHITE OAK          1.8     .42        267.1      65.8
    28  NORTHERN RED OAK   1.1     .46        230.7      77.1

  (5)  TRANSIENT MODEL A  (YEAR 90)*
13
39
28
1
27
RED MAPLE
BLACK CHERRY
NORTHERN RED OAK
SUGAR MAPLE
WHITE OAK
.3
.9
1.0
.6
1.1
.19
.33
.26
.20
.46
458.4
381.4
238.9
156.6
84.7
329.3
120.6
70.1
65.0
28.9
     '  Note that this transient climate  model output is available  for only 90
years; year 90  is  show  here for comparative purposes.
                                      2-25

-------
 Botlrin
                      Table 3 EFFECTS ON ECONOMIC FORESTRY

 Biomass  Production After 50 Years Growth Following Clearcut
       Based on weather records from Mount Pleasant,  Michigan

 (A)  Deep Dry Sandy Soil

       MOUNT PLEASANT,  MI
       1951-1980  WEATHER DATA
       Depth to the Water Table 1.2 m;  Soil Depth 1.0  m
         soil moisture holding capacity 150.00 mm/m of  soil;
         soil nitrogen 150;  60 replicates.

                                  BIOMASS (kg/m2)
                                  ABOVE GROUND
                                                        MEAN 95%CI
 NORMAL                                                 13.4  1.4
       (Dominant species aspen,  red oak and sugar maple)
 TWICE  CO, GISS                                          1.8  0.4
       (Dominant species red maple and red oak)
 TWICE  CO, GFDL                                         0.07 0.007
       (Dominant species:  pin cherry,  choke cherry)
 TWICE  CO, OSU                                           9.3  1.6
       (Dominant species:  red oak and red maple)

 (B)  Deep Relatively Vet Sandy Soil
       Depth to the Water Table 0.8 m;  Soil Depth 1.0  m
         soil moisture holding capacity 150.00 mm/m of  soil;
         soil nitrogen 150;  60 replicates.

                                  BIOMASS (kg/m2)
                                  ABOVE GROUND
                                                        MEAN 95%CI
NORMAL                                                  19.4   1.6
       (Dominant species sugar maple and red maple)
TWICE CO, GISS                                         26.0   3.6
       (Dominant species red maple and black cherry)
TWICE CO, GFDL                                         13.0   1.8
       (Dominant species:  red maple)
TWICE CO, OSU                                          21.    3.8
       (Dominant species:  red maple and black cherry)
                                      2-26

-------
                                                                                               Botkin
LIMITS OF THE RESULTS AND RESEARCH NEEDS
      The time and funds available for this study placed a limit on the work that could be done, and it is not
yet known how sensitive the results may be to specific factors. This work merely scratches the surface of the
effects of climate on forests of the Great Lake States, but points out the very strong and important changes
that result from the projected climatic effects of carbon dioxide increase.  Wise planning would require additional
research. There are six ways that the work should be extended:

      (1) Test the sensitivity of the results to the value of intrinsic parameters in the forest growth model, such
as the maximum longevity of trees, the temperature limits of growth.

      In particular, it is especially important to determine how sensitive the projected timing of changes under
the transient climatic regimes is  to specific values of species  parameters in the  forest models and to key
assumptions in the climate models.

      (2) Test the sensitivity of the results to certain extrinsic parameters of the model, such as the choice of
1951-1980 weather records as a basis for simulation of future climates, the use of ratios of control, and modified
climates to determine climatic change  parameters.

      (3) Also test the response of the forest model to the effects of different rates of seed dispersal

      (4) Extend the work to more soils with a wider variety of soils in regard to soil depth, texture, and fertility.
Choose additional sites for which good soil information exists. In areas where commercial forestry is important,
investigate the effects on forests with representative soils of those areas at a number of harvesting periods.

      (5) Extend  the model by adding more species representing those found even farther south to determine
the forest response to the altered weather conditions if these could migrate into the area.

      (6) Consider the interaction between the climatic affects considered here and other environmental factors
including fire, pathogens, insects,  air pollution, UV-B radiation fluxes,  acid rain and gaseous oxidants, human
management and human-induced disturbances, as well as the  frequency, intensity, and duration of extreme
weather events including wind, flooding, and drought

      (7) Consider the effects of various forestry methods  or new practices such as irrigation to increase forest
yield.


POLICY IMPLICATIONS

      Forestry and recreation in forested areas are important in the Great Lake States and generally considered
to provide an important future economic resource. For example, in 1980 the Governor of Michigan held a major
state conference on forest products; the conference concluded that the forests provided a "vast" resource covering
more than half of the state (more than 19 million acres X and that "current productivity could be doubled with
good forest management practices," and could be a basis for "considerable expansion" of the forest industry. In
1975, forest industry  in Michigan alone produced $12 billion  and provided 64,000 jobs,  and the conference
projected that this could be increased S3.4 billion, adding 21,000 new jobs (Milliken,  1980).

      The results of the simulations reported here suggest that it is necessary to  re-evaluate  potential forest
productivity in light of climatic change. These results indicate dramatic changes would occur even on good, well-
drained soils which are currently the  sites of most commercial forest production. On such sites, total wood
production under a 50-year rotation could drop under the twice CO2 steady-state climates in the southern part
of the area to essentially nothing, as  projected with GFDL steady-state climate, or to as little as 70% of the
current yields, as projected with the OSU model  In either case, the effect would be economically severe.  On
the positive side,  some lowland forests would become better drained and their wood production might increase,


                                                 2-27

-------
 Botkin

 as projected using the GISS and OSU models. (In the worst case these lowland forests would decrease by about
 30% as projected using the GFDL climate.) Whether the putative increase on lowland sites could compensate
 for the decrease in upland sites would have to be evaluated in future work.

       In all cases, there may be a major shift in species composition. Currently the forest industry in the Great
 Lake States is adapted for a certain complement of species, primarily for softwoods used in the production of
 paper pulp and construction materials. In the southern portion of the area, the species that would become most
 economically important under twice CO2 steady-state climate would be hardwoods such  as oaks, useful for
 furniture and other decorative purposes.  Moreover, the most economically valuable of these more southern
 species, red oak, would have a much longer rotation time with harvesting less frequent than for softwoods. Thus
 there would be  a major shift in the character of the forest industry whose costs should be  evaluated; the shift
 would require different equipment and markets.

       Michigan and Minnesota, like other Great Lake States, have a large tourist industry centered around
 their forests and the wildlife and waterways within the forests, which in turn depend on forest characteristics
 for their maintenance.  The  projected change in species  composition may affect all aspects of recreation,
 especially wildlife habitat, and the kind and quality of fisheries in lakes and streams.  While there can  be
 considerable recreation in the kinds of forests that would result, the character of the recreation would change.
 For example, in the southern portion of the region, people would hike in oak forests and oak savannahs rather
 than in northern hardwoods or boreal forests.  There would be different wildlife,  and one would expect less
 canoeing.  Whether this would increase, decrease, or leave recreational use at current levels should be examined.

      The above statements concern the projections for twice CO2 steady-state climates.  On the more positive
 side, the transition from current forests would take place comparatively slowly in human terms (although these
 are extremely rapid in terms of forest growth, more rapid than effects we have examined previously for periods
 during the last 10,000 years).  However, as the climatic stresses on  the trees increase  they will become more
 susceptible to insect outbreaks and diseases.  For example, balsam fir under stress are much more susceptible
 and likely to attract the spruce budworm, which kills the trees; outbreaks typically start in stands of stressed trees,
 triggering large outbreaks in otherwise healthy forests.  In addition,  forests with an increasing number of dead
 trees and  increasing percentage of stressed trees will provide greater fuel for fire. When combined with a
 generally drier soil, these factors could increase the probability of  fires, and lead to  additional decreases in
 economic yield and to more rapid changes in recreational attributes. The effects of such disturbances should be
 investigated.

      All of these implications are currently based on simulations of forest  growth in two regions on two soil
 types.  Until more areas can be considered with the forest  model, all implications must be treated with great
 caution.

      Our results contrast with previous qualitative conjectures that forest production would be uniformly
 increased by CO2 increases in the atmosphere.  These conjectures are based on laboratory experiments which
 show that  wen-fertilized and well-watered plants undergo great increases in production under CO--elevated
 atmospheres.  We do not expect this effect to be  significant for two reasons: (1) the water limitation wfll prevent
 any substantial CO-fertilization response; and (2) as we have shown elsewhere (Botkin et at, 1973), competition
 among trees  for light in a mixed species forest buffers the forest against CO- fertilization.  The  net growth
 increase in an entire forest is thus much less than that of a single plant in a laboratory. With competition for
 limited water and light, as well as for specific chemical elements in the soil such as nitrogen,  the direct response
 to CO- fertilization is likely to be insignificant; however, explicit consideration of the combined effects of changes
 in temperature, rainfall, and CO2 concentration should be investigated.

     These results are restricted to direct climatic change on two soil types.  No  other independent effects,
such as the effects of other air pollutants or the direct effect of CO-atmospheric concentration on tree growth,
were taken into account  It is important to consider the combinedimpact of these processes, including direct
fertilization effects of CO2 on tree growth and effects of acid rain and gaseous oxidants on tree growth, and to
consider these for more soil types. The potential for such projections now exists.


                                                 2-28

-------
                                                                                              Botkin
SUMMARY

      The general pattern that emerges from these findings is that the CO2*induced climatic change leads not
only to much warmer, but also to much drier conditions than occur at present; although rainfall increases, total
evapotranspiration increases more and as a result less soil moisture remains for tree growth. As a consequence,
wetter sites become wanner and somewhat better drained; sandy, wetter sites are able to support forests, but
these  are characteristic of areas to the south and there is a drop in total biomass. The dominant species shift
from those with commercial value to those of little commercial value. On drier sites, the climatic shift is severe
enough to convert substantial  forests to open woodlands, savannahs, or grasslands with small scattered trees.

      These results  are projected for good sites for forest growth. The soils are very deep and well drained,
and (although relatively sandy under the 1951-1980 climate) they are moderately fertile  and well to very well
watered; the wetter sites have  only a minimal amount of saturated soiL Similar, sandy, glacial-derived soils are
not uncommon in the region  around ML Pleasant, MI. Thus the projections represent the more optimistic
conditions than are to be expected. Additional trials with the models on a greater variety of soils would of course
be of considerable utility to broaden the scope and increase the accuracy of the projections.

      It is important to note  that previous speculations about the effects of CO2 increases on forest growth
have emphasized the direct "fertilization" effect of that increase: in laboratory  experiments, increasing CO2
increases the growth rates of  individual trees. Thus some authorities have speculated that the CO-increase
will simply increase  total forest productivity. However, we showed much earlier  (Botkin et aL, 1973) that
competition among  trees  in a mixed-species forest would result in a much smaller effect than predicted from
a laboratory trial with a single  tree. The results presented here suggest that the climatic changes are very strong;
our judgment based on past modeling experience is that the climatic effects would more than compensate  for
any CO, enhancement effect;  however, the combined effects of climate change and CO,  fertilization should be
studied.
                                                 2-29

-------
 Botkin


                                           REFERENCES


 Aber, J.S, DJJ. Botldn and JJVL Melillo.  Predicting the effects of different harvesting regimes on forest floor
 dynamics in northern hardwoods. Canad. J. Forest Research 8:306-315,1978.

 Aber, JDn D.B. Botkin and JJVL Melillo.  Predicting the effects of different harvesting regimes on productivity
 and  yield in northern hardwoods. rfrnTKlifffl J- Forest Research  9:10-14,1979.

 Botkin, DJJ, JJL Janak and JJL Wallis.  Rationale, limitations and assumptions of a northeast forest growth
 simulator. IBM J. of Research and Development 16:101-116,1972.

 Botkin, D.B, J.F. Janak and JJL Wallis.  Estimating the effects of carbon fertilization on forest composition
 by ecosystem simulation, pp. 328-344. Iff GJVl. Woodwefl and E.V. Pecan, edsn Carbon and the Biosphere.
 Brookhaven National Laboratory Symposium No. 24, Technical Information  Center, U.S JLE.G, Oak Ridge,
 TN, 1973.

 Botkin, D.B, J.E. Estes, R JVI. MacDonald, M.V. Wilson. Studying the Earth's Vegetation from Space. BioScience
 34:508-514,1984.

 Botltin, DJ3, J J7. Janak and JJL Wallis. Some ecological consequences of a computer model of forest growth.
 J. Ecology 60:849-872,1973.

 Bryson, RA. and TJ. Murray. Climates of Hunger. University of Wisconsin Press, Madison, WI, 1977.

 Davis, MIX and DJB. Botltin.  Sensitivity of the Cool-Temperate Forests and Their Fossil Pollen Record to
 Rapid Climatic Change. Ouarternarv Research 23:327-340,1985.

 Hall, Fn DJ3J3otkin, D. Strebel, and S.Goetz, Ten Year Change In Forest Succession Measured By Remote
 Sensing (unpub. manuscript).

 Heinselman, MX, Landscape Evolution, Peatland Types, and the Environment in the Lake Agassiz Peatlands
 Natural Area, Minnesota, Ecological Monographs:40 235-261,1970,

 Heinseunan, MX, Fire in the virgin forests of the Boundary Waters Canoe  Area, Minnesota, J. Quaternary
 Research 3:329-382,1973.

 Milliken, Governor W.G, Proceedings of the Governor William G. Milliken's Forestry Conference, Michigan
 Technological University, Houghton, MI, 90 pp. (citation on page 8X 1980.

 Mitchell, JJF.B. Q J.R. Meteorl. Soc. 109:113,1983.

 Prentice, LC, The design of a forest succession model, pp. 253 -256, Iff J. Fanta fed.) F?refl ^YlFm'cs Research
 in Western and Central Europe. Pudue, Wageningen. and Ionian*, R, 1986.

 Prentice, LC, Description and simulation of tree-layer composition and size distributions in a primaeval Picea-
 Pinus forest, Veeetatio 69:147-156,1987.

SeUers, WJD, Physical Climatology. Univ. of Chicago Press pp. 156-180.

Shugart, H JL,  A Theory of Forest Dynamics. Springer-Verlag, N.Y, 1984.
                                               2-30

-------
                                                                                            Botkin

Solomon A-M. and D.C.West, Simulating forest ecosystem responses to expected climate change in eastern
north America:  Applications to decision making in the forest industry.  Iff The Greenhouse Eff?di Q»mate
Chance, and U.S. Forests. WJLShands and XSJloffman Eds. The Conservation Foundation, Washington, D.C.,
pp. 189-217,1987.

Thornthwaite, C.W., An approach toward a ratignfli Classification of rlimate. Georgr. Rev. 38:55-94.

West, D., HJL Shugart,  D3. Botkin (eds.),  Forest Succession: Concents and Applications. Springer-Verlag,
NY, 198L

Wigley, TJvIJ^ P. D. Jones, and PJvL Kelly. Scenario of a warm, high-CO2 world. Nature 283:17-2,1980.

Woodwell, GJVIn R.H. Whittaker, WA. Reiners, G.E. Likens, CA-S. Hall, C.C. Delwiche, and D.B. Botkin.
The biota and the world  carbon budget Science 199:141-146,1977.
                                               2-31

-------
     FOREST RESPONSE TO CLIMATIC CHANGE:
A SIMULATION STUDY FOR SOUTHEASTERN FORESTS
                   Dean L. Urban
                 Herman H. Shugart
           Environmental Soences Department
               The University of Virginia
               Charlottesvffle, VA 22903
              Contract No. CR-814610-01-0

-------
                                  CONTENTS
FINDINGS [[[ 3-1

CHAPTER 1: INTRODUCTION  [[[  3-2
      FOREST TYPES AND STUDY SITES .........................................  3-2

CHAPTER 2: METHODS [[[  3-4
      FOREST SIMULATION MODEL .............................................  3-4
      SCENARIOS [[[  3-4
      SIMULATIONS [[[  3-5

CHAPTER 3: RESULTS  [[[  3-6
      CLIMATE-CHANGE SCENARIOS ...........................................  3-6
      FOREST DYNAMICS UNDER CLIMATE CHANGE .............................  3-6
            East Tennessee [[[  3-6
            Sooth Carolina [[[ 3-10
            Georgia [[[ 3-10
            Mississippi [[[ 3-10
      TRANSIENT RESPONSES TO CLIMATIC CHANGE ............................. 3-10
            East Tennessee [[[ 3-12
            South Carolina [[[ 3-12
            Georgia [[[ 3-12
            Mississippi [[[ 3-12


-------
                                                                                               Urban


                                            FINDINGS1


        We used a forest simulation model to explore the effects of climatic variability on southeastern forests.
All climate-change scenarios (GISS, OSU, and GFDL 2xCO2) predict a wanning and drying climate.  This in
turn predicts the northward migration of forests, such that areas now characterized as oak-pine types would be
replaced by more southern types characterized by loblolly pine.  Based on climatic correlates of present-day
forest distributions, the extreme southeastern forest  types could not be supported by projected climates; the
peculiar (confounded) biogeographic context of these sites does not allow us to confidently predict their fate.

        Transient climate scenarios suggest that the  short-term response to rapid climate change would be a
period of synchronous stress mortality, resulting in forest declines in all simulated cases. The magnitude of this
response would depend  on the magnitude of climate  change  relative to  species tolerances, and on  the
suddenness of the climate change. The timing of forest decline would likely depend on local weather patterns,
with episodic mortality triggered by a  period  of unusually stressful years.  Susceptibility  to climatic stress
depends on stand age and condition, with older  or already stressed  stands being especially vulnerable to
additional climatic stress.  These factors in combination make it unlikely that we could confidently detect or
accurately predict a short-term response to climate change for any particular case.

   This study illustrates our uncertainty about the proximate mechanisms that govern forest response to climatic
variability.  We  suggest research priorities for further  studies, toward a more fundamental understanding of the
mechanisms and consequences of environmental constraints on forest ecosystems. Perhaps the most  critical
uncertainty in this study concerns seed dispersal and species  migration rates; these are  critical to predicting
forest response  to transient climate, and we have little capability to estimate these effects.
        'Although the information in this report has been partly funded by the U.S. Environmental Protection
Agency under Contract No. CR-814610-01-0, it does not necessarily reflect the Agency's views, and no official
endorsement should be inferred from it

                                                 3-1

-------
 Urban


                                             CHAPTERl

                                           INTRODUCTION


        In this report we provide a preliminary assessment of the potential effects of climatic variability on
 forests of the southeastern United States.  We consider a spectrum of forest types throughout the southeast,
 ranging from oak-pine, through southern mixed  forest of the Piedmont,  to coastal-plain elements of the
 southeastern evergreen forest  Within each forest  type, we focus first on upland sites; for the coastal plain we
 examine bottomland forests as wefl.

    In the following sections we detail the rationale and methods we use to  relate forest dynamics to climatic
 variation, present .simulation results from a variety of climatic scenarios, and discuss the implications of these
 results. Our discussion underscores the uncertainties in our current understanding and suggests priorities for
 further studies of forest response to climatic variability.


 FOREST TYPES AND STUDY SITES

    This study is concerned primarily with upland forests of the Southeast In the southern Appalachians, these
 forests encompass oak-pine, oak-hickory, oak-chestnut, as well as other variants of southern mixed forest types
 as classified by Braun (1950X Kuchler (1964), and  Bailey (1976). In terms of potential vegetation cover, these
 forest types represent more land area in the continental United States than any other forest (Eyre, I960).  Key
 species include upland oaks (Quercus alba, Q. rubra, Q. velutinaX hickories  (Carya cordiformis, C. glabra, C.
 ovalis, C ovata, and C tomentosaX and shortleaf pine (Finns echinata). Through the Piedmont and onto the
 coastal plain, indicator species include sweetgum (Liquidambar styrarifluaX black gum (Nyssa sylvatica),
 southern oaks (Q.  falcata, Q.  shumardiiX and especially loblolly pine  (P.  taeda). The  southern  pinelands
 dominated by shortleaf and  loblolly pines  are by  far the most important commercial  forests in  the eastern
 United States.

    This spectrum of forest types was represented  in this study by four locations:  Knoxville,  Tennessee (35.8°
 latitude (latX 89.8° longitude  (lonX representing Appalachian oak-pine forests; Florence, South Carolina (342°
lat, 79.7° Ion) in the northern coastal plain; Macon,  Georgia (32.7* lat, 83.7* lonX on the southern margin of the
Piedmont; and Vicksburg, Mississippi (32.4* lat, 90.0" lonX  near the western margin of the coastal plain (Figure
 1).  The latter three sites represent a east-west gradient across the loblolly pine forest type, from mesic to more
xeric climatic  regimes.  It should be emphasized that the selection of study sites was larger/ constrained by the
availability of long-term weather records; we  do not intend to make specific  statements about these particular
locations, and the preliminary results of this study should not be overinterpreted in this respect.
                                                 3-2

-------
                                                                                                  Urban
                                                                                TOHCST VCGCTATON
                                                                                   lEASTVANl
Figure 1.    Forest vegetation of the eastern United States (adapted from Powells, 1965).  Study sites are
             indicated in east Tennessee, South Carolina, Georgia, and Mississippi
                                                   3-3

-------
 Urban

                                            CHAPTER 2

                                             METHODS


       This study follows the precedent set by a number of previous studies in using forest simulation models to
 explore the response of forests to climate-mediated environmental driven (Solomon 1986, Solomon et aL 1981,
 Solomon et aL 1984, Solomon and Tharp 1985, Solomon and Webb 1985, Davis and Botkin 1985, Pastor and
 Post 1988; see also Botkin et aL 1972, Shugart 1984). While these studies varied in purpose and detail, they
 share a common rationale and method; conveniently, they also have  used similar versions of the same basic
 forest model (Botkin et aL 1972, Shugart 1984).

       Modeling forest response to an environmental driver requires, in the simplest sense (1) a description of
 how trees respond to the environmental factor of interest,  and (2) a means to generate or provide reasonable
 scenarios of variability in the environmental driver.  While some aspects of this approach may seem rather
 complicated (perhaps  even realisticX  it should be emphasized that  the results of these  studies should  be
 interpreted at the level of these basic steps involved in their implementation. That is, model output should be
 interpreted as the  consequences  or implications of the assumptions in the model, rather  than as precise
 predictions of reality. We return to tins theme at a later point

 FOREST SIMULATION MODEL

        The model used  in this study was ZELIG, a versatile forest simulator developed for applications
 concerned with spatial patterns occurring at scales larger than the forest gap.  The model is an individual-based
 'gap model" (Shugart and West 1980)  derived from the FORET model of Shugart and West (1977).  Gap
 models simulate forest dynamics by modeling the demographics of each individual tree on a small model plot
 corresponding to the area of a forest gap. These models have been especially useful in simulating the  dynamics
 of mixed-age, mixed-species forest stands (Botkin et aL 1972; Shugart  and West 1980, Shugart 1984).

    Like its parent  and sibling models, ZELIG simulates forest dynamics by accounting the establishment,
 annual diameter growth, and mortality of each individual tree in the simulated forest The basic approach in
 modeling each of these demographic processes is to begin with a maximum potential behavior (e.&,  inseeding
 rate, diameter increment, survival probability) and subsequently modify this potential according to the status of
 the  individual tree  in the  context of the modeled gap.  The contextual contraints modeled  in ZELIG include
 available light, temperature, soil moisture, and sofl fertility. The ZELIG model is documented in more detail
 in the Appendix. The 45 tree species included in these simulations are summarized in Appendix Table A.L

 SCENARIOS

      Upland forests at each location were simulated as growing on the same soil type, in order to control this
 source of variation in modeled forest dynamics. Thus, the ZELIG model as implemented for this study is wholly
 driven by weather data provided as input  data.  Baseline  weather data comprised monthly precipitation and
 temperatures for the years 1951-1980.

      Climate-change scenarios were derived from GISS, GFDL, and OSU model output as provided by NCAR.
 Implementation of these scenarios was as dictated by EPA.  For the GISS scenario, data conversions were based
 on the grid point at  35.22* N lat, 80.00*  W Ion; GFDL conversions were based on grid point 3333" lat, 8150°
 Ion; the OSU grid point was 36* lat, 85* Ion. Because of the coarse resolution of the GCM-model grid cells, the
same conversion ratios were used for all four sites. A single transient scenario was considered, GISS Transient
A.  Relative to previous simulation studies of this sort, this study considers  alternative steady-state climates
under 2xCO- (represented by three GCMsX and emphasizes especially the transient climate change as predicted
by the  GISS model  This is in contrast to previous studies, which have assumed  a single steady-state climate
under altered CO, or have used a linearly interpolated transient climate (e.&, Solomon 1986, Pastor and Post
1988).


                                                3-4

-------
                                                                                               Urban


SIMULATIONS

    In simulations, the 30-year base weather was concatenated to yield climate scenarios that were of sufficient
length to elicit a meaningful forest response.  In each case, simulations were run for 200 years. The initial few
decades illustrate successional dynamics under each scenario, while the later years suggest  trends in more
mature forests. Results reported here are averages of 50 replicate model plots.

    The design of this study was to run four climate scenarios for an upland soil at each location (baseline,
GISS, GFDL, and OSUX as well as a bottomland site for the south-central location in Georgia. Modifications
to this design are explained more fully where applicable.
                                                  3-5

-------
 Urban

                                             CHAPTERS

                                              RESULTS


     Before presenting the results of the several forest simulations, it will be  instructive to  summarize the
 climate-change scenarios in terms of their computed drought-day and  growing degree-day indices.  Because
 these are  the effective environmental constraints incorporated in the forest  model, a knowledge of species
 tolerances to these indices can provide a helpful preview of model results. Following this preview, the model
 results will be presented in more detail.

 CLIMATE-CHANGE SCENARIOS

     Afl three climate models predict a warming and drying as compared to current conditions. The models
 differ hi that the GISS prediction is for moderate warming with a concomitant increase in precipitation, while
 GFDL predicts  a wanning  with a decrease in  growing-season  precipitation;  OSU  predicts GISS-like
 temperatures and GFDL-lOce precipitation.

     In terms of the constraints computed by the forest model, all three scenarios predict a large increase in
 annual growing-degree sums, while there is some variability hi predicted  drought-day indices  (Table 1).  For
 reference,  we should note that no tree species simulated in the model can tolerate a drought-day index greater
 than 0.6 (indexing the proportion of  the growing season under drought conditions); few species can tolerate a
 value greater than 0.5, and most species show significant growth reductions if the drought index is greater than
 03  (refer  again to  Appendix Table A.1).  A drought index of «0 J  (half the growing season under drought
 conditions) roughly corresponds  to the western margin of the eastern deciduous forest

     Similarly, few tree species of the southern mixed forest are found under temperature regimes of more than
 «6000 degree-days (a 5SOO-GDD isopleth roughly parallels the Gulf coast). We should note that our estimates
 of heat-sum  tolerances of some of these species are thus not very confident, because their distributions are
 limited by  geographic constraints (the Gulf) rather than climatological factors.  We return to this point later in
 this discussion, but note here that the model will grow only marginal forests under temperature regimes of more
 than «6000 degree-days.

 FOREST DYNAMICS UNDER CLIMATE CHANGE

    The 200-year simulations cannot indicate clearly which species will ultimately come to dominate southern
 uplands under the various climate scenarios, but the model results do indicate successional trajectories under
 these regimes. In general, the baseline simulations reproduce major serai trends for these southeastern forests:
 shortleaf pine in the more northern oak-pine forests of east Tennessee, and loblolly pine throughout the more
 southern forest types. While this is encouraging, for our purposes it is sufficient that the model provide a useful
 framework of reference for further simulations.  The following  sections  detail these  patterns for  each
 geographic location simulated.

 East Tennessee

    Simulated succession on upland oak-pine sites is strongly dominated by shortleaf pine, which reaches its
 greatest dominance  over the first hundred years and subsequently is replaced by more shade-tolerant hardwoods.
 These sites support »460 T/ha of above-ground woody biomass.  Under GISS and OSU  scenarios the same
 qualitative  pattern obtains, with only slightly less biomass (Figure  2).  An important difference is that  loblolly
 pine is the serai pine, virtually replacing shortleaf pine (Figure 3). This reflects the change in annual heat sums
 from «3600 to «5000 growing degree-days (refer again to Table 1). The more southern longleaf and slash pines
 are also common but never abundant (basal area of wl-4 m /ha).  Hardwoods that succeed the pines include
black gum,  elm, and Shumard oak, all of which are indicative of more southern forest types.
                                                 3-6

-------
                                                                          Urban
Table 1.    Median Growing Degree-day  and  Drought-day Indices  for Four Study
            Sites,  for Baseline and Three Climate Change Scenarios.  Scenarios
            Are GISS,  OSU, and GFDL 2xC02.   90th Percentiles Are in Parentheses.
                   Baseline
               GISS
               OSU
               GFDL
Knoxville, Tennessee

     Degree-days   3616 (3809)
     Drought-days  0.00 (0.23)
               4917 (5130)
               0.05 (0.33)
               4845 (5060)
               0.18 (0.39)
               5327 (5542)
               0.48 (0.52)
Florence, South Carolina
     Degree-days
     Drought-days
4328 (4651)
0.00 (0.32)
5753 (6079)
0.13 (0.47)
5698 (6027)
0.22 (0.51)
6162 (6492)
0.52 (0.55)
Macon, Georgia

     Degree-days   4682  (4866)
     Drought-days  0.09  (0.36)
               6075 (6294)
               0.33 (0.50)
               6023 (6240)
               0.43 (0.51)
               6488  (6707)
               0.54  (0.56)
Vicksburg, Mississippi

     Degree-days   4719  (4968)
     Drought-days  0.14  (0.39)
               6145 (6400)
               0.35 (0.48)
               6094 (6344)
               0.41 (0.50)
               6564  (6813)
               0.51  (0.54)
                                      3-7

-------
Urban
                                      EAST  TENNESSEE
                                         Weedy Btonun*
                                                 too
                                           Simulation Year
ISO
200
Figure 2.     Trends in above-ground woody biomass in upland forests in east Tennessee, as simulated under
            current and 2xCO2 climate scenarios.
                                            3-8

-------
                                                                                        Urban
                                      EAST  TENNESSEE

                                          Yellow Ptn««
                                                  100

                                            Simulation Y«or
190
200
Figure 3.     Basal area of dominant yellow pine in east Tennessee forests as simulated under current and GISS
            2xCO2 scenarios. Pine in baseline case is shortleaf; under GISS scenario, shortleaf is replaced by
            loblolly.
                                              3-9

-------
 Urban


    The GFDL scenario, which predicts the most severe warming and drying, resulted in no trees surviving in
 simulations, indicating that nonforest conditions (grassland or sparse  savannah) would be favored.

 South Carolina

    Under the baseline climatic regime for the northern coastal plain, loblolly pine strongly dominates succession
 on upland sites, reaching a maximum of «25 or/ha in basal area at «60 years.  Under the GISS scenario,
 loblolly pine is present but never common; longleaf pine is common but not dominant (maximum basal area «2
 m2/ha). Instead, mixed southern hardwoods occur even in the youngest serai stages; these include black gum,
 hackbeny, laurel oak, and elm, all of which are associated with the more southern and  western coastal plain.
 These forests as simulated under the GISS scenario support less than half the biomass  of the baseline forest
 («60 vs. «150 T/ha, Figure 4).

    Both the OSU and GFDL scenarios predict thermal regimes at or beyond the domain of species tolerances
 (Table 1 and Appendix Table A.1).  Marginal forests under the OSU scenario never reach more than 15 T/ha
 in biomass. No trees were supported under the GFDL scenario.

 Georgia

    Upland sites in Georgia, again, are strongly dominatd  by loblolly pine during simulated succession.  These
 forests support *150 T/ha of woody biomass.  Under the most moderate  of the climate-change scenarios
 (GISS), saplings of several southern coastal-plain species are established but none persist to develop a forest.
 The mix of species suggest that temperature was the predominant constraint on trees. Under the OSU regime,
 only xeric species are planted (post oak and blackjack oakX and these do not survive. These species suggest that
 drought became an operative constraint in this simulation, selecting only the most drought-tolerant trees from
 the southern species pool The extreme GFDL scenario predicts drought and GDD indices beyond the domain
 of modeled species tolerances; no trees are planted.

    Bottomland sites in  Georgia were  simulated  with  a moderate regime  of saturated-soil conditions,
 corresponding  to a spring flood duration of 1 month (mean FD-0.17).   This  regime results in a  forest
 dominated by black gum and sweetgum, with elms of nearly equal importance; these are key species of southern
 floodplain forests as described by Braun (1950). The climate-change scenarios predict responses very similar to
 those for upland sites, indicating  that the temperature effects and late-summer drought are sufficiently extreme
 to override the effects of flood duration.

 Mississippi

    Like the other southern sites, upland succession in Mississippi is strongly dominated by loblolly pine in
 baseline simulations. This site is sufficiently droughty that even under  present-day climate, more xeric soils can
 result in post oak savannahs instead of pinelands (simulations not included here). This  reflects this site's
 position at the western margin of the deciduous forest biome. Under all climate-change scenarios, the  site
 becomes too hot and dry to support trees as parameterized in the model (Table 1).

TRANSIENT RESPONSES TO  CLIMATIC CHANGE

    Transient scenarios were simulated by concatenating the 30-year weather data bases and applying the GISS
Transient A conversions to yield 90-year transients. Two cases were simulated for each  study site considered:
a 90-year successional sequence from bare ground, and a 90-year projection of a 100-year-old stand (these were
output from year  100 of the baseline simulations). In each case, the baseline simulations  provide a control
against which the transients can be compared (years 0-90 for the serai case; years 100-190 for the latter case).
                                                3-10

-------
                                                                                     Urban
                                      SOUTH  CAROLINA
                                         Woody BlomoM
                                                 100
                                           Simulation Yoar
200
Figure 4.     Trends in above-ground woody biomass in upland forests in South Carolina, as simulated under
            current and GISS 2xCO2 climate scenarios.  OSU and GFDL scenarios did not support forest
            vegetation.
                                            3-11

-------
 Urban

 East Tennessee

     The transient climate change for east Tennessee is evident in growing degree-days in «30 years, but becomes
 pronounced only after «60 years (Figure 5).  Forest succession as simulated under this  transient  climate
 proceeds as with the control for «60 years, followed by a decrease in woody biomass (Figure 6a). This reflects
 a decline in the dominant species, shortleaf pine (Figure 6b).  While these declines are appreciable after 90
 years, the variability in both biomass and species importance is considerable (standard deviations are consistently
 50-100% of means). This variability cautions against overinterpreting minor differences observed earlier in the
 simulations.
     More mature forests subjected to the transient climate show a pronounced decline in biomass during years
 60-70 of the simulation (Figure 7a). This decline corresponds to the sudden warming trend in that decade (refer
 again to Figure 5). The gap-phase mosaic nature of these more mature forests, reflecting the break-up and
 regeneration of the canopy, results the large variance in stand biomass observed in Figure 7a. Interestingly, the
 decline  in biomass is not  due simply to a parallel decline in the dominant species, shortleaf pine, but rather
 reflects  a decline in several codominant and minor species. Shortleaf pine shows a general  decline in the
 control  as well as transient  simulations, and the transient does not deviate appreciably from the control until
 after «80 years (Figure 7b). Loblolly pine, the species favored under the GISS 2xCO2 scenario, does not appear
 in the transient forest until after 70 years, and at 90 years is still a relatively minor component of the forest
 (dashed line, Figure 7b).

 South Carolina

     The South Carolina transient climate is somewhat erratic, with several wanning pulses recurring over the
 simulated years (Figure 8).  Responses of serai forests reflect these pulses, especially in the later years of the
 simulation.  This is evident at years 55 and 85 for woody biomass (Figure 9a) as well as the dominant loblolly
 pine (Figure 9b). In both cases, the first decline (at «S5 years) is rather minor, while the later decline (at «80
 years) is much more dramatic  These same two declines are also evident in simulations of more mature forest
 (Figures lOa&b). In these forests, the warming pulses at about years 55 and 80 are reflected in the baseline as
 well as the  transient simulations; the transient cases do not deviate substantially from the baselines until after
 year 80.

 Georgia

     As with the east Tennessee case, the transient climate for the Georgia site shows a dramatic warming decade
 at »60 years into the simulation (Figure 11). This warming would roughly correspond to a 4-5°  C. increase in
 mean annual temperature, which would be equivalent to the entire temperature change as projected over 90
 years being telescoped into a single decade.  Simulated responses of successional forests to this regime are
 obvious  and intuitive:   total biomass declines abruptly at year 60 (Figure 12a); the dominant species, loblolly
 pine, shows a similarly dramatic decline (Figure 12b).

     More mature upland forests show the same  qualitative pattern: a  dramatic  decrease in woody biomass
 (Figure  13a) and in the dominant pine (Figure 13b) at «60 years into the simulation.  Recall that the GISS
 2xCO2 simulation for this site predicted marginal forest or nonforest vegetation, and the transient approaches
 this within the  90 simulated years. This reflects the dramatic warming of years 60-70 in the simulation.

 Mississippi

    The  transient climate for the Mississippi site begins to deviate from the baseline within 20 years and shows
 a rather  steady wanning over the entire 90-year projection (Figure 14).  As with the other study sites, there is
 a single decade that shows a pronounced warming, in this case, the sixth decade. Serai forests undergo dynamics
 similar to the baseline  simulations for several decades, with a a slight and gradual reduction in woody biomass
 (Figure 15a) and the dominant species (Figure 15b). In each of these cases, a dramatic decline follows at about
year 60, at which point the transient climate passes 6000 growing degree-days. Recall that this value is a critical
value in the simulator, beyond which most modeled tree species succumb to heat stress (a similar point is passed


                                                 3-12

-------
                                                                                  Urban
                            EAST  TENNESSEE  TRANSIENT
                                   Growing  Oogrco-Ooy*
10
20
M
                                           40      SO
                                        Simulation Y«or
M     70     10
M
Figure 5.    Transient climate, as annual heat sum, projected for east Tennessee (GISS transient A).
                                           3-D

-------
  Urban
                              EAST  TENNESSEE TRANSIENT
                                 Succession from Bar* Ground
180-1

twJ

140 J
           g  t(*H
           o
           o   MH
           O
           5   40H
               20 ^

               0
                   Woody  Btomass
                       Control
                        10
                              2O
                                            «O
                                                   90
                                                          6O
                                                                 70      80      90
                                          Simulation Year
                             EAST TENNESSEE  TRANSIENT
                                Sueemslon fron Bar* Ground
                       to     20
                      30     40     90     «0
                           Simulation Year
70     M
90
Figure 6.    Simulated forest succession under transient climate in east Tennessee, as (a) woody biomass and
           (b) basal area of the dominant species, shortleaf pine.
                                          3-14

-------
                                                                                  Urban
180-

1SO-

140-


100-

 80-

 M-

 40-

 20-

  0
                            EAST  TENNESSEE TRANSIENT
                                Dynamic* of Mature Forest
                                                            Woody  Blomass
                                                            — Control
                      10     20     30     40     SO     SO
                                        Simulation  Year

                            EAST TENNESSEE  TRANSIENT
                                 Oynamtes  of Mature Formri
                                                   70     M
90
                      10     20
                         30      «0     90
                             Simulation Y
                                                                70     N     90
Figure 7.    Simulated response of 100-year-old east Tennessee forests subjected to transient climate change,
           in terms of (a) woody biomass and (b) basal area of the dominant species, shortleaf pine.
                                           3-15

-------
 Urban
                              SOUTH  CAROLINA  TRANSIENT

                                     Graving 0«gr«*-0ays
            7000
         o
         o
         I
         I
         o
            6OOO-
            3000-
            4000-
                                    30
  40     SO     «0

Simulation r«ar
70
•0
90
Figure 8.     Transient climate, as annual heat sum, projected for South Carolina (GISS transient A).
                                           3-16

-------
                                                                                 Urban
                           SOUTH  CAROLINA TRANSIENT
                              Succession from Bar* Ground
                     '0      20     30     40     SO     60     70
                           SOUTH  CAROLINA  TRANSIENT
                              Succession from Bar* Ground
                PInus  taoda
                — C«ntr«!
                            20      40      40      90     60
80     90
Figure 9.    Responses of succession^ South Carolina forests to transient climate change, as (a) woody
           biomass and (b) basal area of loblolly pine.
                                          3-17

-------
Urban
   180-

   180-
?
^ 140-
N>

7 '»-
M
| 100-

3  80-

    80-

    40-

    20-

     0
                        -7—
                         10
                               SOUTH  CAROLINA  TRANSIENT
                                    Dynamics of  Motur* ror«»*
                                                                Woody  Biomass
                                                                — Control
                                                                «• • Tranvtant
                    —T—
                    20
—r-
 30
   40     SO     80
Simulation Y«ar
        70
        80     90
               2S
               20-
            d- '»H
            N
            O
            5  ,0-1
            •3
                3-
                               SOUTH  CAROLINA TRANSIENT
                                    Dynamics of  Uofur* Forest
                        10     20
                          —r-
                           30
       —r-
       40
        •T"
         90
—I—
 80
                                           Simulation Tear
                                                                    Plnus
—r—
 70
                                                                          •0     90
Figure 10.    Responses of South Carolina forests to transient climate change, for 100-year-old forests, (a)
            woody biomass and (b) basal area of loblolly pine.
                                           3-18

-------
                                                                                 Urban
                                 GEORGIA TRANSIENT
                                   Growing Oogro*-0oys
                     10     20
90
  40     90
Simulation Y<
Figure 1L   Transient climate, as annual heat sum, as projected for Georgia (GBS transient A).
                                           3-19

-------
Urban
                                     GEORGIA  TRANSIENT
                                 Dynamic* of  Mature  Upland  rarest
                         10
        20
        30      40      90
            Simulation Y
                                                            M
                                     GEORGIA TRANSIENT
                                Dynamic* of Mature Upland rarest
10
70
                                       50      40      M     M
                                           Simulation Year
                                           70      CO
                                                         90
Figure 12.    Responses of succession^ Georgia forests to transient climate change, as (a) woody biomass and
            (b) basal area of loblolly pine.
                                            3-20

-------
                                                                                     Urban
                                   GEORGIA  TRANSIENT
                             Upland Succession from  Bar*  Ground

                        10      20
M     40     SO
     Simulation r«ar
6O
                                                                          80     90
                                    GEORGIA  TRANSIENT
                              Upland Succession from Bar*  Ground
                        10     20
 JO     40     30      8O
      Simulation Year
         70      80      90
Figure 13.   Responses of Georgia forests to transient climate change, for 100-year-old forests, (a) woody
            biomass and (b) basal area of loblolly pine.
                                            3-21

-------
 Urban
                                    MISSISSIPPI  TRANSIENT
                                        Growing  D«gr««-0ay«
              7000
           o
           o
           I

           9
           O
           9

           O
              6OOO -
3000-
              4OOO
20
                          JO
                                               40     30      SO
                                             Simulation Year
                                                               80
90
Figure 14.   Transient climate, as annual heat sum, as projected for Mississippi (GISS transient A).
                                             3-22

-------
                                                                                     Urban
                                  MISSISSIPPI  TRANSIENT
                                  Succession from  Bar*  Ground
           •o
           I
                                             40      30      SO
                                           Simulation Year
                                  MISSISSIPPI TRANSIENT
                                  Succession from  Bar* Ground
           5
            3
10     20
                                              40     SO      60
                                           Simulation Year
                                                   80
                                                                                   90
Figure IS.    Responses of saccessional Mississippi forests to transient climate change, (a) woody biomass and
            (b) basal area of loblolly pine.
                                            3-23

-------
Urban
in the South Carolina transient at year 80; refer again to Figures 8-10). In the case of more mature forests, both
biomass (Figure 16a) and loblolly pine (Figure 16b) show an appreciable decline before  this critical point is
reached; these declines are evident after as little as 40 years.  This partly reflects the wanning pulse at about
year 30, which elicits an appreciable effect for this already hot and xeric site.
                                                  3-24

-------
                                                                                              Urban

                                            CHAPTER 4

                              INTERPRETATION AND DISCUSSION


    In this discussion we interpret the simulation results, realistic as well as artifactual, to synthesize a general
overview of how forests might respond to transient climate change. We begin with what we feel are realistic and
robust model results, and then turn to aspects of the results that are less certain or compromised by lack of
information. The aspects of transient forest reponses that are of interest here are the magnitude and timing of
the response, because these attributes largely determine our ability to predict and detect such a response.


MAGNITUDE OF FOREST RESPONSE

    Simulated forest change in response to the projected climate scenarios roughly corresponds to a northward
migration of southeastern forest species. That is, the oak-pine forest types of east Tennessee are replaced by
more southern elements (as indicated by loblolly pine). This prediction is consistent with results of Solomon et
aL (1984), who used a similar model to project geographic implications of climate change. As previous studies
have not utilized transient climates  as predicted by GCMs, we emphasize these transients in this discussion.

    In every case, the  magnitude of predicted forest change is sufficiently dramatic as  to make statistical
comparisons of the long-term results superfluous.  For the southern forests (sites in South Carolina, Georgia,
and MississippiX the prediction is that these forests will degrade to marginal forest or nonforest vegetation. This
prediction represents the current state of our information base,  and similar studies have reached similar
conclusions (Solomon et aL 1984, Solomon 1986). But as we will argue, the basis for this prediction is somewhat
uncertain (see following section on Model Uncertainties).

    Of more immediate concern is whether forest change might be detectable in the near-term future, say 10-20
years.   If so, then we could partially verify the  model  predictions, and  would have more confidence in
longer-term predictions of forest responses.  In every case fimMla*Ht forest response to climate change is an
initial decline in biomass reflecting a period of heavy stress mortality. Based on the degree of variability in the
modeled forest dynamics, we must  argue that we would likely have little power to discriminate such a decline
from natural background variability in forest dynamics.  Indeed, it seems that the alternatives would be to risk
"crying wolf" by over-reacting to  short-term fluctuations in forests triggered by natural variability in weather, or
to be more conservative and ignore a real transient forest response until the change was well underway.
Neither alternative is particular^ appealing from a management standpoint


TIMING OF FOREST RESPONSE

    An important aspect of forest responses to transient climate change is evident in Figures 5-16, which should
be emphasized again at this point  In every case, the forest decline is triggered by a rapid warming, a few
stressful years in a row. The actual timing of these periods reflects the 30-year  base weather traces used to
generate the transient climate. In the South Carolina data, a wanning pulse is obvious at about year 25 in the
baseline (Le^ 1975); this pulse recurs at years 55 and 85 in the concatenated transient climate (Figure 8).  Not
surprisingly, the most dramatic forest declines correspond to these pulses.

    The South Carolina case is particularly obvious, but the same behavior is evident for the other study sites.
Indeed, much of this behavior can be attributed to the shape of the transient climate predicted by the GISS
model  The transient, which was provided to EPA as smoothed (time-averaged) conversion ratios comparing
the transient data to GKS-generated baseline conditions under current COj, shows a nonlinear trend with rather
abrupt changes around yean 30 and 60 (Figure 17).  These pulses coincide with the pulsed diebacks in the
modeled forests. Thus, the transient forest responses are partly artifactual, a consequence of the GISS-modeled
transient as modified by baseline weather records for each study site.


                                                3-25

-------
 Urban
                                   MISSISSIPPI  TRANSIENT
                                   Dynamic*  of  Mature  Forut
           Q

           e
180-

160-

140-

120-

100-

 M-

 •0-

 40-

 20-

  0-
                    Woody  Blomass
                    • — Tranctant
                        10     20     M
                               40     90
                             Simulation T<
                                                            M     70     SO     90
               29
               20
               19-
            O

           J
           "3
            a
               9-
                    Pfnus
                                  MISSISSIPPI  TRANSIENT
                                   Dynamics of MoHir* Foro*t
                        10      20      JO     40     90     SO      70

                                           Simulation  Year
                                                            80     90
Figure 16.    Responses of 100-year-old Mississippi forests to transient climate change, as (a) woody biomass
            and (b) basal area of loblolly pine.
                                            3-26

-------
                                                                                  Urban
                                  GISS  TRANSIENT  A
                                       Cast Tenn«ss«€
            3500
            3000
                1970   1980   1990   2000   2010   2020   2030   2040   2090

                                          Simulation  Year
Figure 17.
Growing degree-day transient as predicted by GISS transient A scenario. Plotted values are 30-
year mean monthly temperatures multiplied each year by conversion ratios for east Tennessee site,
yielding a transient unmodified by interannual variability in the baseline data.
                                           3-27

-------
  Urban


     Yet this pulsed behavior in the modeled forests is a realistic attribute of forests. In real forests, we would
  expect weather anomalies to modulate forest dynamics, with particularly stressful periods resulting in episodes
  of heavy mortality. In the context of predicting forest response to climate change, the implication is that we
  might expect an appreciable forest decline at some time over the next several decades, but the exact timing of
  such a decline, as well as the rate or abruptness of the decline, would depend on actual weather patterns. Thus,
  we might expect a forest response perhaps 40-70 years from now, but we cannot hope to predict the exact
  timing of the response.

  Confounding Factors

     Forest responses to climatic variability may be confounded by a number of other factors. Among these are
  stand age, and the ameliorating or exacerbating effects of other environmental factors. Forest sensitivity to
  environmental stress increases with stand age, because older trees are more vulnerable to stress. This reflects
  the high maintenance costs of large trees, which leaves little margin for a reduction in photosynthate production.
  Thus, we would expect very mature trees to show the effects of climate change sooner than young trees.

         Other environmental factors surely contribute to observed patterns of forest growth, mortality,  and
  regeneration.  Additional stresses would have a synergistic effect with climatic (heat and drought)  stress,
  resulting in increased local mortality at such sites.  Thus, there is a sense in which already stressful environments
  (very xeric sites; harsh microenvironments) present themselves as likely points to monitor for the appearance of
  episodic or unusually heavy mortality that might indicate environmental change.  Unfortunately, because "stress"
 is so difficult to diagnose specifically, it seems that real difficulties could arise in any effort to ascribe observed
 mortality to a single environmental factor (here, climate change).

     We should note that  not all environmental change is for the  worse. It has been suggested that one direct
 effect of CO, enrichment may be to increase the water-use efficiency of plants (Strain and Cure 1985).  If this
 is the case, then forest response to possible drying (through higher  temperatures, lower precipitation, or both)
 might be somewhat ameliorated as compared to simulations based  on our current estimation of trees' drought
 tolerances.

     Pastor and Post (1988) have noted that positive feedbacks between soil water content and available nitrogen
 may make forest response to climate change more complex than intuition might suggest  Their model illustrated
 two possible trajectories for northern forests under a warming climate: on mesic soils with sufficient soil water,
 increased decomposition rates led to higher nitrogen availability and greater forest productivity; on  drier soils,
 drought stress reduced forest productivity.  We do not know at what soil moisture level this bifurcation in system
 behavior might occur.

    Several natural disturbances may be mediated by climate, either directly or indirectly.  For example, fires
 would be more likely under a warming and drying climate, because  of greater fuel loads and higher chances of
 ignition. It is difficult to predict the degree to which this might affect southeastern forests, as fire regimes would
 likely be contolled through management (e-g, supression). Pest outbreaks could also be affected by climate
 change, for example, in cases where the geographic range of a pest organism is currently bounded by winter
 temperatures.  Again, predictions about these factors are beyond the  scope of this report

 MODEL UNCERTAINTIES AND CAVEATS

     There are two sorts of uncertainties in simulation studies such  as this. The first sort concerns conceptual
questions about operative mechanisms in forest response to environmental drivers. The second problem  arises
when the mechanisms are thought to be  understood, but  implementation is limited by a lack of data of
sufficient precision or resolution.  The former problem appears as underspecified (or incorrectly specified)
algorithms in the model, while the latter problem translates into problems in parameter estimation..  Both sorts
of problems are inherent in this  study.
                                                 3-28

-------
                                                                                              Urban


    One parametric problem to which we previously alluded concerns the estimation of maximum tolerable heat
sums for southern tree species. As noted, southeastern forest species do not occur under regimes of more than
«6000 growing degree-days, a value that occurs in north-central Florida (the Gulf coast has an annual heat sum
of wSSOO GDD).  In the simulator, most  southern tree species thus have their thermal maxima artificially
bounded.  These species do not occur under wanner regimes because (a) they are bounded by water, (b) they
are bounded by tropical or coastal evergreen forest elements, and/or (c) they are bounded to the southwest (in
Texas and Mexico) by drought In each case it is not possible to specify what the actual maximum tolerable heat
sum might be for these species. Thus, the  model wfll not simulate forests under regimes of more than *6000
GDD, but in fact this may be an artifactuaOy low boundary condition.  Paleo-ecological records  offer no
additional insights, as there is no record of forest (pollen) distributions under climates warmer than  those
projected over the next several decades (Delcourt and Delcourt 1987). We have no means of correcting this
source of uncertainty.

    A second source of uncertainly in the  model concerns the implementation of stress tolerance and  stress
mortality.  In the model, a tree subjected to an environmental condition near its specified tolerance suffers
reduced growth; more extreme conditions result in no growth. Stressed individuals (showing less than minimal
growth) are subjected to a fixed, elevated mortality rate (see details in Appendix). This implementation does
not incorporate myriad physiological or morphological  adaptations to environmental stress (see Turner and
Kramer 1980).  A more detailed implementation of stress responses in trees is beyond the scope of this study,
hence, represents another source of uncertainty in our results.

          Mechanisms      ecies Response to
    It seems obvious that a more fundamental understanding of the mechanism* that affect tree response to
climate will be  cruciaL  Solomon et  aL (1984)  discussed various proximate  mechanism* that might be
interpreted as forest response to "climate" in a general sense. These mechanisms include direct weather effects
as well as indirect effects mediated by climate (e-g, insect or pest outbreaks, or disturbances such as fires and
floods). Moreover, trees are more or less susceptible to these factors depending on their age or size: flowers
may be sensitive to spring frosts, while seedlings are vulnerable to late-summer drought, and mature trees may
succumb to  multiple droughts over a few years.  All of these factors may be intercorrelated and interpreted as
"climate." Our current model implementations reflect our current understanding, and simplify the richness and
complexity of forest response to these many proximate mechanisms. Thus, our model-based predictions about
forest  responses to climatic variability are limited  by  a lack of direct evidence detailing the  proximate
mechanisms of this response.

    We must emphasize  that our  current empirical understanding of forest response to climatic variability is
sufficiently lacking that grandiose predictions about climate-change scenarios are probably premature.  This
study underscores the need for a  great deal of basic research into the mechanisms by which trees and forests
respond  to  climatic variability in general, and specifically, to identify the  proximate climate factors and
biological mechanisms that govern such responses.

Seed Av?ilflfrilitv «md Species Migration

    Perhaps the most critical uncertainty in this study concerns seed availability and rates of species migration.
In the model, this translates into the implicit assumption of which species are available for establishment in any
given year.  This study follows Botltin (this volume) in generously assuming that any species in the regional pool
is available  throughout the simulation.  Thus, under a Changing climate  in east Tennessee,  more southern
species were allowed to enter the modeled forest This implementation ignores constraints on seed sources and
seed dispersal   This is an  extremely critical assumption in the context of this study, because this assumption
dictates very different forest responses  to climate change.
                                                 3-29

-------
 Urban

     If one assumes that any species in the regional (southern) species pool is available in forest simulations, then
 more southern species enter east Tennessee forests Mien the climate  becomes too warm for the local species.
 The result is that species replacement occurs (e.&, loblolly pine for shortleaf pine, Figure 3), which mitigates the
 decline in forests due to stress mortality.  Although we had no basis for including an alternative species pool
 for the more southern study sites, it is reasonable to suggest that a similar replacement might occur on these
 sites if species migration rates were sufficiently fast relative to climate change.  Note that species replacement
 does not imply that no forest decline will occur.  Declines  evident in  the east Tennessee transient simulations
 (Figures 5-7) reflect heavy mortality under stress. The magnitude of this mortality would likely depend on the
 magnitude and suddenness of climate change. The degree to which species replacement might mitigate this
 episodic mortality would depend on how well the replacement species might be adapted to the new climate, and
 how rapidly such replacement might occur.

     Alternatively, if southern species are assumed not to  be available, then local  stress  mortality would go
 unanswered by species replacement  A  dramatic forest decline would  result,  even in  areas for which
 well-adapted species exist, if these species were not locally available. Current estimates of tree species migration
 rates, inferred from Holocene pollen records (Davis 1981, Delcourt and Delcourt 1967), suggest maximum rates
 on the order of 50-100 m/yr. These rates are nearly an order of magnitude too slow to track a transient climate
 as  projected  for the next several  decade.   If climate change is  sufficiently  rapid and  extreme to initiate
 widespread and synchronous  stress mortality, then we  must conclude that  much of  this  mortality would be
 unmitigated by species replacement

    As a perhaps more realistic intermediate case, some individuals of species better adapted to the changing
 climate would probably be locally available, and these would provide at least a small source pool of propagules.
 In this case, the magnitude of local forest decline would be determined in part by the richness of the local
 species pool (the  availability of alternative, better adapted species).  The duration of the decline would be
 determined by the rate at which appropriate species could  migrate into the area or  by the rate at which local
 individuals could mature to provide an adequate seed source.

    In a  management context, many effects of seed  availability might be  mitigated by  planting species
 appropriate to the changing climate.  The  practice of planting seeds from carefully selected genetic stock or
 provenances is well established in silviculture.  This could perhaps be extended as necessary, such as to provide
 tropical or southwestern species for particular sites, as climate projections indicated.  Again, careful research
would be required to identify appropriate provenances.
                                                3-30

-------
                                                                                              Urban

                                            CHAPTERS

                                          IMPLICATIONS


    Results of this study are in agreement with previous studies in suggesting that projected climate changes will
initiate a northward migration of forest species, with current geographic distributions displaced distances on the
order of hundreds of kilometers. Simulations suggest that some southeastern sites would become too xeric to
support well-developed forest, and instead would  be replaced by some  other vegetation type   (savannah,
grassland, or scrub).  Because of the peculiar biogeographic context of these  areas (the Gulf influence, the
proximity of the southern hemisphere flora, and the xeric  boundary represented by south Texas and Mexico),
it is not possible to predict the fate of these southeastern sites.

    Transient responses of simulated forests are characterized by dieback episodes corresponding to pulses of
rapid climate change.  Actual forest response would depend on the  magnitude and timing of such warming
pulses. Both of these aspects of transient climate change are subject to some uncertainty in terms of prediction
or detection.

    This study is compromised by uncertainties about species responses to proximate environmental factors, and
about actual mechanisms and adaptations for stress tolerance. Perhaps the most critical uncertainties concern
seed availability and  rates of species  migration.  The extent to which  species replacement can mitigate
synchronous stress mortality due to climate change is a principal factor determining the magnitude and duration
of forest decline under a changing climate. The highest research priority should be placed on resolving these
uncertainties.
                                                 3-31

-------
  Urban


                                           REFERENCES


  Aber, JD., D.B. Botkin,  and JM. Mellilo.  1979.  Predicting the effects of different harvesting regimes
  onproductivity and yield in northern hardwoods.  Can. J. For. Res. 9:10-14.

  Bailey, R.G. 1976. Ecoregions of the United States. USDA Forest Service Intermontane Region. Ogden, Utah.

  Baker, F.S.  1949.  A revised tolerance table.  J. For. 47:179-18L

  Bassett,J.R.  1964. Tree growth as affected by sofl moisture availability. Soil So. Proc. 28:436-438.

  Botkin, D.B., JJ7. Janak, and JJL Wallis. 1972.  Some ecological consequences of a computer model of forest
  growth. J. EcoL 60:849-873.

  Braun, EX. 1950. Deciduous forests of eastern North America. Hafner Press, New York.

  Chapin, FSn AJ.  Bloom, CJB. Field, and RJI.  Waring.  1987. Plant responses  to multiple  environmental
  factors.  BioScience 37:49-57.

  Davis, M.B. 1981.  Quaternary history and the stability of forest communities. Pp 132-153 in West, D.C, HJH.
 Shugart, and D.B. Botkin (eds.X Forest succession.  Springer-Verlag, New York.

 Davis, MJ3., and D.B. Botkin.  1985. Sensitivity of cool- temperate forests and their fossil pollen record to rapid
 temperature change.  Quartemary Res. 23327-340.

 Delcourt, PJL, and H.R. Delcourt  1987. Long-term forest dynamics of the temperate zone. Springer-Verlag,
 New York.

 Eyre, FJL (ed).  1980.  Forest cover types  of  the United States and Canada.  Soc American Foresters,
 Washington, DC

 Powells, HA.  1965. Silvics of forest trees of the United States. USDA For. Serv. Handbook No. 271, Govt
 Printing Office, Washington, DC

 Harcombe, PA.  1987.  Tree  life tables.  BioScience 37:557-568.

 Harlow, W.M., and ES. Harrar.  1%9. Textbook of dendrology. McGraw-Hill, New York.

 Kuchler, A.W.   1964.  Potential natural vegetation of the conterminous United States. Amer. Geogr. Soc
 Special PubL No. 36.116 pp. + map.

 Mitchell, Hi, and  R.F. Chandler.  1939.  The nitrogen nutrition and growth of certain deciduous trees of the
 northern United States.  Black Rock For. BulL 1L

 Monteith, JX.  1973. Principles of environmental physics.  Arnold, London.

 Pastor, J.,  and  W.M. Post  1984.   Calculating Thomthwaite's  and Mather's AET using an approximating
 function. Can. J. For. Res. 14:466-467.

Pastor, J.,  and  W.M. Post  1988.  Response  of  northern forests to CO.-induced climate chance.  Nature
334:55-58.                                                          2                ^^
                                               3-32

-------
                                                                                          Urban

Pastor, J.,  and W.M.  Post   1985.   Development  of  a linked  forest  productivity-soil process model.
ORNL/TM-9519.  Oak Ridge National Laboratory, Oak Ridge, TN.

Shugart, HJL  1984. A theory of forest dynamics.  Springer- Verlag, New York.

Shugart, HJEL, and D.C. West 1977.  Development of an Appalachian deciduous forest succession model and
its application to assessment of the impact of the chestnut blight  J. Environ. Manage. 5:161-179.

Shugart, HJL, and D.C West 1979.  Size and pattern of simulated forest stands.  For. Sri. 25:120-121

Shugart, HJL, and D.C. West 1980.  Forest succession models.  BioScience 30:308-313.

Smith, T.M., and DX. Urban.  1988.  Scale and resolution of forest structural pattern. Vegetatio 74:143-150.

Sollins, P., DJS. Reichle, and J.S. Olson.  1973. Organic matter budget and model for a southern Appalachian
Liriodendron forest EDFB/ffiP-73/2.  Oak Ridge National Laboratory, Oak Ridge, TN.

Solomon, A-M. 1986.  Transient response of forests to CO,- induced climate change:  simulation modeling
experiments in eastern North America.  Oecologja 68:567-579.

Solomon, A^L, and MX. Tharp.  1985.  Simulation experiments with late quartemary carbon storage in
mid-latitude forest communities. Pp  235-250 in The carbon cycle and atmospheric CO^  natural variations
archean to present Geophysical Monograph 32, Amer. Geophys. Union.

Solomon, A^L, and T. Webb. 1985.  Computer-aided reconstruction of late-quaternary landscape dynamics.
Ann. Rev. EcoL Syst. 16: 63-84.
Solomon, AMn MX. Tharp, D.C. West, G.E. Taylor, J.W. Webb, and JX. Trimble.  1984.  Response of
unmanaged forests to CO2-induced climate change: available information, initial tests, and data requirements.
DOE/NBB-0053.  Office of Energy Research. U.S. Dept of Energy, Washington, DC.

Solomon, A^l, D.C West, and LA. Solomon.  1981.  Simulating the role of climate change and species
immigration in forest succession. Pages 154-177 in D.C West, H.H. Shugart, and D.B. Botltin (eds.), Forest
succession: concepts and applications.  Springer- Verlag, New York.

Strain, B.R., and JJD. Cure  (eds.).   1985.   Direct effects of increasing carbon dioxide on vegetation.
DOE/ER-0238. U.S. Dept of Energy, Washington.

Teskey, R.O., and T.M. Hinckley.  1977.  Impact of water level changes on woody riparian  and wetland
communities. VoL H: southern forest region. USDI FWS/OBS-77/59.

Thomthwaite, C.W., and J.R. Mather. 1957. Instructions and tables for computing potenital evapotranspiration
and the water balance.  Publications in Climatology 10:183-311.

Turner, N.C, and PJ. Kramer (eds.). 1980. Adaptation of plants to water and high temperature stress. Wiley,
New York.

Weinstein, DA, HJi. Shugart, and D.C  West  1981  The long- term nutrient retention properties of forest
ecosystems:  a simulation investigation. ORNL/TM-8472.  Oak Ridge National Laboratory, Oak Ridge, TN.
                                               3-33

-------
 Urban

                                             APPENDIX

                    DOCUMENTATION OF THE FOREST SIMULATION MODEL


     This appendix summarizes the implementation of the ZELJG  forest simulator. The emphasis is on those
 details of the model  that are pertinent to this particular application, especially the environmental constraints
 affected by climate.  This section also overviews model demographics, statistical concerns with replication, and
 notes on parameterization.

 Gap-scale Constraints

     The implementation of shading effects retains the original FORET approach of modeling light extinction
 through a forest canopy as a negative-exponential decay following Beer's Law (Shugart 1984):


                 Qh - Qoexpt-kUh')],                  (1)


 where Qh is the light available at height h, QQ is incident light, and L(h') is cumulative leaf area (m2/m2) above
 height h.  Here, k is a constant describing light extinction through the  canopy; this constant is related to leaf
 angle, branching geometry, and the absorption properties of leaf tissue, and for deciduous forest canopies takes
 on values on the order of 0.25- 0.50 (Monteith 1973). If Qg equals LO, this decay defines the proportion of full
 sunlight available at a given height of the foliage profile. Species response to available light was  originally
 described for 2 shade-tolerance classes (Botitin et aL 1972), using equations of the form:


             r(Qh) - c^LO-exptc^-c,)]),            (2)


 where Qh is as described above, a, is a scaling constant, Cj determines the rate of change in growth relative to
 change in sunlight (L&, steepness of the response curve X and c^ is the compensation point (where net growth
 is 0).  ZF.LTG uses 5 shade-tolerance classes (Baker 1949, Powells 1965); these functional  forms are evenly
 interpolated between the 2 shade-response functions originally implemented in FORET (Figure 1).

    Temperature effects on tree growth are  simulated as a   function of  an annual heal sum, with growing
 degree-days tallied from a 5-56°C base.  This implementation reflects empirical correlations between species
 distributions  and annual heat sums, as observed at  a coarse  (e^, continental) scale of resolution.  ZELJG
 retains the original convention of modeling species  response to growing degree-days as a parabolic function
 (Botitin  et aL 1972, Shugart 1984):
          r(GDO) -
where GDD is growing degree-days, subscripted to denote the maximum and minimum values observed over
the present-day distribution of a species (Figure 2).

    The soil moisture routine in  ZELJG calculates a monthly water balance  for each soil type, which is
summarized as the percentage of days during the growing season for which there is inadequate soil moisture
                                               3-34

-------
                                                       Urban
                AVAILABLE  LIGHT
1.4-
1.2-
 SPECIES  RESPONSE

— Shad* Tolerant
--- lnt»rm«dlat«
• -•• Shad* Intolerant
                      Available Light
Figure A.L Multiplier used to adjust tree growth rate in response to available light
                         3-35

-------
Urban
                           TEMPERATURE
         1.0-
     .«  0.8-

     o
     CfL
     I  °-6-l

     o


     =  0.4H


     as,


        0.2 H
        0.0
                 SPECIES
                  Chestnut Oak    /
                 1000    2000   3000   4000    5000   6000   7000


                          Growing  Degree-Days
 Figure AJ. Multiplier used to adjust tree growth rate in response to temperature as annual heat sum.
                                 3-36

-------
                                                                                               Urban

(soil moisture below wilting point). This routine is based on the method of Thorthwaite and Mather (1957) as
modified by Pastor and Post (1984,1985). Because basal area growth has been shown to decrease linearly with
moisture stress (Bassett 1964), and basal area is a square function of diameter, species response to drought
stress (as  relative diameter increment) is modeled as a square-root function.  This function  relates the
percentage of drought days in the year to a species-specific maximum sustainable percentage of drought days
(Pastor and Post 1985):

                 r(D) - >/(D*-D)/D*                 (4)


where D is the percentage of drought  days during the growing season and D  is the maximum tolerable
percentage for a given species (Figure 3).

    To simulate bottomland forests, a new constraint has been incorporated into ZELIG to account for high
water table or flood conditions.  This implementation is simplistic in assuming only  that (1) a flooding regime
or depth to water table is so dependent on local topography and drainage characteristics that  it is reasonable
to postulate a high-water regime that is larger/ independent of soil conditions and regional weather; and (2)
relative to the mean high-water regime, conditions in any given year are loosely correlated with springtime water
balance (years with dry springs will have a lower water table, and vice versa).  Thus, a bottomland site in
ZELIG has a declared mean value for a parameter flood duration (FD), which is defined to be the  proportion
of the growing  season  during which the soil is saturated. FD is linearly related to annual springtime water
deficit (deviation from mean PET for March-April-May). Because this relationship is rather weak in empirical
studies (t2 M 0.6,  D. Hains, pen. comm.), stochastic noise is added to the  predicted value for FD.  Species
response to FD is modeled similarly to the sofl moisture response,  as a square-root function of the maximum
tolerable FD for each species. Species  tolerances were parameterized as classified by Teskey and Hinckley
(1977).

    Differential species response to soil fertility is modeled by specifying a nutrient-response category for each
species.  Species tolerant to nutrient stress grow adequately on poor soils and do not respond substantially to
fertilization (or rich sitesX while intolerant species fare badly on poor sites but are very responsive to enriched
soils; a third class is intermediate in response (Mitchell and Chandler 1939).  Three previous gap models (Aber
et aL 1979, Weinstein  et aL  1982, Pastor and Post  1985)  have used the empirical results of Mitchell and
Chandler (1939) to derive polynomial functions relating relative tree growth to nutrient availability. In ZELIG,
these functions are  doubly relativized so that both growth rate and soil fertility  vary between 0.0  and  1.0:


                                                      (5)


where the c*s are regressed constants.  Here, soil fertility (F) is defined relative to the best possible site, and
tree growth is  retarded on soils  of lesser productivity (Figure 4).  Note that   ZELIG does not model the
dynamics of nutrient availability, relative sofl fertility is a parameter provided as input data.

    The functional forms of each of these constraints provide  dimensionless multipliers on the interval [0,1]
(they are  truncated to  0.0 or  LO if they go out of bounds). ZELIG uses an interaction between below-ground
and above-ground constraints to modify potential  regeneration and growth of trees.  The  below-   ground
constraint is chosen as the minimum of the soil fertility and sofl moisture multipliers, an approach that assumes
that water-use efficiency and nutrient uptake are so inter-related  that it would be inappropriate to treat
them independently.   The   above-ground  constraint of available  light is multiplied by the   below-ground
constraint, an interaction that reflects the empirical observation that photosynthetic efficiency decreases  with
moisture stress (Chapin et aL 1987).
                                                 3-37

-------
Urban
                            SOIL  MOISTURE
                                       SPECIES  RESPONSE
                                            Drought Intolerant
                                         -  Intermediate
                                          •  Drought Tolerant
          0.0
                            Relative  Drought  Days
       Figure A3. Multiplier used to adjust tree growth rate in response to moisture stress.
                                 3-38

-------
                                                               Urban
                      SOIL  FERTILITY
                                 SPECIES  RESPONSE
                                    Nutrient  Stress  Intolerant
                               - - - Intermediate
                                 ••.Nutrient  Stress  Tolerant
        0   0.1   0.2   0.3   0.4   0.5   0.6   0.7   0.8   0.9   1.0
   0.0
                        Relative Soil  Fertility
Figure A.4. Multiplier used to adjust tree growth rate in response to relative soil fertility.
                             3-39

-------
 Urban

 Demographics

     Regeneration is stochastic, and incorporates stump sprouting as well as seedling establishment Sprouting
 depends on the  number and sizes of newly dead trees on a plot, and on the tendency of a species to sprout.
 Seedling establishment is based  on maximum potential inseeding rates for each species as  constrained by
 ground-level shading, soil moisture, and soil fertility. Seedlings are filtered (sensu Harper 1977) through three
 years of environmental constraints before they are established as .saplings.  In ZELIG, the number of sprouts
 and the constrained inseeding rates are used to define a probability of establishment for each species, and new
 trees are established stochastically according to these probabilities. This algorithm,  while simplistic, biases
 regeneration to reflect the prevalence of stump-sprouting in eastern forests, and adjusts the number and species
 composition of seedlings according to local site   conditions.  As with other FORET-class models, ZELIG
 assumes that any species included in the simulation may be established if site conditions are appropriate; there
 are no constraints to reflect seed availability or dispersal

     Annual diameter growth is based on species-specific functions relating volumetric growth to the current size
 (height  and diameter) of an individual tree (Botltin et aL 1972):

             dTE^HJ/dt - rL(1.0-DH/DmaxHmaxX       (6)


 where D is diameter at breast height (onX H is  height  (cm), Dmax and H ^ are the maximum attainable
 diameter and height for a given species, r is relative growth rate, and L is leaf area. By making a number of
 simplifying assumptions  about tree allometries, Botltin  et aL  (1972) were  able to specify an  equation for
 diameter increment:

              dD/dt.    GD(1.0-DH/DmaxHmax)        (7)
                     (274.0+3.0b2D-4.0b3D2)


 where the b's are allometric constants and G is a growth rate parameter that can be expressed in terms of the
 other variables (but see section on parameterization). This equation is admittedly simplistic in that it does not
 attend the complexities of photosynthate allocation within a tree. It offers the advantage that it seems to capture
 the essential pattern of tree growth (Figure 5), and can be readily parameterized with comparable accuracy for
 a large number of species.

    Mortality is modeled as a stochastic event, and may arise from 2 sources: natural  (age-related) mortality,
 and stress or suppression.  Natural mortality related to aging is stochastic, based on the assumption that 1%
 of individuals survive to reach a species-specific maximum age (Botltin et aL 1972, Shugart 1984).  The  further
 assumption that mortality is constant dictates, for suitably long lifespans, that the annual mortality rate can be
 approximated as:

                   m » 4.605/A,^,                   (8)

 where 4.605 derives from the natural log of 0.01 (1%) and Amax is maximum age in years.  Stressed individuals
 (those failing to achieve 10% of their potential growth increment, or achieving an absolute increment of less
 than 0.1 cm, for more than 2 consecutive years) are subjected to a mortality rate that  assumes  that only 1%
 of stressed trees will survive 10 years; this yields a mortality rate for stressed trees of 0369.  The criterion that
 a tree must grow at least 0.1  cm in diameter to escape stress mortality injects an element of age-dependence
 into mortality: as  a tree approaches its maximum  diameter  (in old  age), its maximum possible  increment
approaches  0.0 and it enters the  domain of stress mortality.  While  this approach is motivated  largely by
computational convenience  (the model need not account  tree age), the resultant mortality patterns are  largely
consistent with data from tree life tables (Harcombe 1987).


                                                3-40

-------
                                                               Urban
         130-
                        DIAMETER  GROWTH
                     Liriodendron  tulipifera
                     r
                    50
100     150      200

  Simulation Ytar
250
300
Figure A-5.  Diameter growth as simulated by the forest model  Inset: relation of diameter increment to
         current diameter.
                                 3-41

-------
 Urban


 Simulations

    In gap models, a model plot is typically large enough to  contain the influence of a large tree's dominance
 and death, without overly diluting this influence (Shugart and West 1979). ZELIG is implemented as an array
 of 0.04-ha cells.  The array is underlain by a soils map (representing a topographic gradient), providing for
 variation hi soil moisture and site productivity. The array of plots is simulated simultaneously, so that each plot
 experiences the same temperature and precipitation regime.  Thus, temporal variation in weather constraints
 as well as spatial variation in edaphic constraints are incorporated into the simulation.

    Because  the simulator is stochastic, a -tingle plot depicts a possible trajectory of forest dynamics, but not
 necessarily the mean or expected trajectory.  Thus, the model typically is run for a large number of replicate
 plots, and these are aggregated to describe the average trajectory of forest dynamics over time. Shugart (1984)
 and Smith and Urban (in press) have used gap models to illustrate the relationship between gap-scale dynamics
 and larger-scale forest dynamics.  Preliminary simulations with the ZELIG model suggest that  at least 30
 replicate plots are necessary to stabilize the variance in basal area or biomass.

 Parameterization

    Implementing the ZELIG program for a particular study site requires parameter estimates for tree species
 silvics, soil characteristics, and a weather regime. These parameters are generally readily available from
 published sources, or are easily estimated without additional on-site field studies.

    Tree species parameters used in this study are annotated in Appendix Table A.1.  These parameters are
based on silvicultural records in Powells (1965) and Harlow and Harrar (1969), and preliminary estimates were
computed as described elsewhere (Botltin et aL 1972, Shugart  1984, Solomon 1984, Pastor  and Post 1985).
Because of the lack of site-specific data of sufficient scope and resolution to calibrate the model over the broad
spectrum of forests considered in this study, these parameter estimates have not been adjusted to reproduce any
particular target forest   Therefore, model results should  be  interpreted cautiously,  at a coarse level  of
resolution.
                                                3-42

-------
                                                                             Urban
Table A.I.    Parameter1  Estimates for 45 Tree  Species Used in  Simulations of
              Southeastern Forests
A, D, H_ G LDNF Sprouts, Seeds GDD.ln .
ACNE
ACRU
ACSI
CACO
CAGL
CAOL
CAOT
CATO
CELA
COFL
DIVI
FAGR
FRAM

FRPE
LIST
LITU
Acer negundo Boxelder
75 100 2500 215 2233 1 2.5
Acer rubrum Red maple
150 125 3500 175 2233 1 2.5
Acer saccharinum Silver maple
125 150 3500 230 2223 1 2.5
Carya cordif ormis Yellovbud hickory
250 125 5000 165 3321 1 2.5
Carya glabra Pignut hickory
300 100 4000 115 3321 1 2.5
Carya oval is Red hickory
300 125 4500 125 3421 1 2.5
Carya ovata Shagbark hickory
250 125 4000 135 3421 1 2.5
Carya tomentosa Mockernut hickory
300 100 3500 100 3421 1 2.5
Celtis laevigata Sugarberry
200 75 3000 125 2312 1 2.5
Cornus florida Dogwood
100 35 1250 105 1411 1 2.5
Diospyros virginiana Persimmon
150 50 2000 115 4323 1 2.5
Fagus grandifolia Beech
350 150 3700 90 1221 1 2.5
Fraxinus americana White ash
250 150 3500 100 3412 1 2.5
Fraxinus pensylvanica Green ash
150 125 3250 140 3113 1 2.5
Liquidambar styraciflua Sweet gum
250 125 3500 120 4423 1 2.5
Liriodendron tulipifera Tuliptree
300 300 6000 170 4311 1 2.5
100.
100.
100.
100.
100.
100.
100.
100.
100.
35.
50.
0
0
0
0
0
0
0
0
0
0
0
30.0

20,
20
100
100

,0
.0
.0
.0
40
40
40
20
20
20
20
20
30
30
30
20

30
30
30
40
900
1250
1300
1900
1900
1900
1650
1900
2650
1900
2650
1300

1400
1050
2650
1900
•X
5200
6600
4700
5000
4500
5500
5000
6000
7000
6000
6900
5500

6000
5500
6000
6000
                                       3-43

-------
 Table A.I.  continued
       A, D, EU,    G     LDNF   Sprouts,  Seeds    GDD.ln  .„
 MAGR  Magnolia grandiflora  Southern magnolia
       300 200 4000  115   3212   1 2.5 100.0  20   3500 6000

 NYAQ  Nyssa aquatica  Swamp tupelo
       300 150 3500  100   4125   1 2.5 100.0  20   3000 6000

 NYSY  Nyssa sylvatica  Black gum
       300 150 4000  115   3413   1 2.5 100.0  30   1900 7000

 PECH  Pinus echinata  Shortleaf pine
       200 125 4000  170   5531   1 2.5  15.0  40   2650 5100

 PELL  Pinus elliotti  Slash pine
       200 100 3500  150   4332   0 0.0   0.0  40   4000 6000

 PPAL  Pinus palustris  Longleaf pine
       250 125 4500  150   5332   0 0.0   0.0  40   4000 6500

 PSER  Pinus serotina  Pond pine
       200 100 3500  150   5223   0 0.0   0.0  40   3500 6000

 PTAE  Pinus taeda  Loblolly pine
       250 150 5500  185   4432   0 0.0   0.0  40   3150 6000

 PLOC  Platanus occidental is  Sycamore
       400 300 5200  110   4113   1 2.5  50.0  30   1900 5500

 PODE  Populus deltoides  Cottonwood
       200 200 5300  205   5124   1 2.5  50.0  40   1700 5300

 PRSE  Prunus  serotina  Black cherry
       200 125 4000  170   4311   1 2.5 100.0  30   2100 6000

 QALB  Quercus alba  White oak
       400 200 4500   95   3421   1 2.5 100.0  20   1700 5500

 QCOC  Quercus coccinea  Scarlet oak
       300 150 3100   90   4531   1 2.5 100.0  20   2000 4500

 QFAL  Quercus falcata  Southern red oak
       300 200 3750  110   3422   1 2.5 100.0  20   2650 6000

QLAU  Quercus  laurifolia   Laurel oak
       300 200  3000   90   3222   1 2.5 100.0  20   4000 6500

QLYR  Quercus  lyrata  Over cup oak
       300  150  3500  100    3123    1  2.5 100.0  20   2900 5300

-------
Table A.I.  continued
A, D. H., G L D N F Sprouts
QMAR
QMIC
QNIG
QPHE
QPRI
QRUB
QSHU
QSTE
QVEL
SANI
TADI
TIHE
ULAM
Quercus
300 50
Quercus
300 200
Quercus
300 100
Quercus
300 ISO
Quercus
300 125
Quercus
300 200
Quercus
300 200
Quercus
400 125
Quercus
300 125
marilandica Blackj ack
1500 45 3631
oak
1 2.
michauxii Swamp chestnut
4000 115 3221 12.
nigra Water oak
4000 110 3223
phellos Willow oak
4000 115 3323
prinus Chestnut oak
4500 125 3531
1
1
1
rubra Northern red oak
5000 130 3421 1
shumardii Shumard oak
5500 140 3422 1
stellata Post oak
3000 65 3531
velutina Black oak
4500 125 3531
Salix nigra Black willow
200 250 4300 185 5124
1
1
1
Taxodium distichum Baldcypress
500 300 4500 80 3225 0
2.
2.
2.
2.
2.
2.
2.
2.
0.
5
, Seeds
50.
oak
5 100.
5
5
5
5
5
5
5
5
0
Tilia heterophylla White basswood
150 125 4300 245 2212 1 2.5
Ulmus americana American elm
300 200 4500 115 3423
1
2.
5
100.
100.
100.
100.
100.
100.
100.
20.
0.
100.
30.
0
0
0
0
0
0
0
0
0
0
0
0
,0
20
20
20
20
20
20
20
20
20
40
20
30
30
GDD.ln. ,
2500
3000
3000
3500
1900
2000
2400
2650
1800
1700
3000
2650
1200
ax
6000
5500
6000
5500
4100
4600
6000
6000
5100
5500
7000
4600
7000
1  A,  D,  and  H..,,  are maximum  age  (yrs),  diameter  (cm),  and  height (cm)  per
species;   G  is growth rate constant; L, D,  N, F are light  (1-shade-tolerant).
drought  (/10-maximum tolerable drought- days), nutrient (1-stress  intolerant),
and  flood  tolerance  (1-flood intolerant)  classes,  respectively;  sprouting
parameters are number, and minimum and maximum sproutable stump sizes (cm),  Seeds
is relative inseeding rate (stems/plot); GDD.,n. m, are minimum and maximum growing
degree-days.

-------
ANCIENT ANALOGS FOR GREENHOUSE WARMING OF CENTRAL CALIFORNIA
                                 by
                            OwenK. Davis
                        Department of Geosdences
                          University of Arizona
                           Tucson, AZ 85721
                       Contract No. CR-814606-01-0

-------
                                 CONTENTS

                                                                        fage

FINDINGS 	4-1

CHAPTER L  INTRODUCTION 	4-2
      PAST WARM PERIODS AS ANALOGS FOR GREENHOUSE WARMING	4-2
      THE HOLOCENE CLIMATIC HISTORY OF CALIFORNIA  	4-2

CHAPTER 2: METHODOLOGY	4-4
      THE DATA 	                          	4-4
      DISSIMILARITY COEFFTciENTS.  ....!........'.....'.'.'.'.'.'.'.'.'.'.'.'.......'....! .. .4-4
      DETRENDED CORRESPONDENCE ANALYSIS	4-7

CHAPTER 3: RESULTS  	4-8
      THE POLLEN DATA	4-8
            The Surface Samples	4-8
            The Fossl Sites	4-8
      DISSIMILARITY INDICES 	4-8
            Comparison of Modern Samples 	4-8
            Dissimilarity of Fossil and Modern Samples	4-16
      DETRENDED CORRESPONDENCE ANALYSIS	4-16
            Comparison of Modem Samples 	4-16
            DCA of Fossfl Averages and Modern Analogs 	4-28

CHAPTER 4: INTERPRETATION OF RESULTS	4-29
      UNCERTAINTIES	4-33

CHAPTER 5: IMPLICATIONS OF RESULTS	4-36
      ENVIRONMENTAL IMPLICATIONS  	4-36
      SOCIOECONOMIC IMPLICATIONS	4-36

CHAPTER 6: POLICY IMPLICATIONS 	4-37

LIST OF ABBREVIATIONS	4-38

REFERENCES	4-39

-------
                                                                                              Davis


                                            FINDINGS1


     Using vegetation records from five sites in the Sierra Nevada, two periods are evaluated as historic analogs
for greenhouse wanning.  During the earty-Holocene, 9000 years ago, the vegetation of the western Sierra
Nevada resembled that currently found east of the crest.  Xeric pine forest occupied areas now covered by Sierra
montane forest, precipitation was less and temperatures may have been cooler. Concurrently, precipitation east
of the Sierra crest may have been higher than today.  By 6000 yean ago, the modern vegetation zones had
become established on both sides of the Sierra Nevada. Some sites indicate precipitation greater than today,
whereas others indicate less. Temperatures were lower at most sites.

     These results are based on two numerical techniques, dissimilarity analysis and detrended correspondence
analysis, that indicate modern analogs for fossil samples.  The best  modern analogs for each sample are
converted into precipitation and temperature estimates using modern lapse rates for the area.

     Tourism, water supply, and the  logging industry will be negatively affected if precipitation changes during
the next century are in the direction and magnitude of those of the 9000-yr B J». analog. Increased precipitation
in the eastern Sierra could offset some of the effects.
        'Although the information in tins report has been funded wholly or partly by the U.S. Environmental
Protection Agency under Contract No. CR-814606-01-0 under the Clean Air Act as amended 103, it does not
necessarily reflect the Agency's views, and no official endorsement should be inferred from it

                                                 4-1

-------
 Davis


                                            CHAPTER I

                                         INTRODUCTION


 PAST WARM PERIODS AS ANALOGS FOR GREENHOUSE WARMING

      There are two alternatives for evaluating the potential effects of greenhouse wanning on our ecosystems.
 The first is to use computer models that translate estimates of precipitation, temperature, and other factors into
 diversity, biomass, runoff,  or other ecosystem parameters.  The second  alternative, the  one  used in this
 investigation, is to study the responses of ecosystems during times in the past when the climate was substantially
 wanner than at present

      The success of the analog approach depends in part on how closely past environments duplicate those of
 interest A historical analog for greenhouse wanning should match the atmospheric, geographic, and climatic
 factors predicted for the next century. Possible analogs for warmer climate include the 1930*5, the medieval
 warm period of AD. 1100 to 1250, the postglacial warmth maximum, or the Cretaceous period.  However, the
 atmospheric concentration of CO, was probably lower than today during the first two periods; and although CO,
 concentrations may have been higher than at present during the Cretaceous, geographic and other environmental
 conditions were vastly different.

      During the maximum of post-glacial warmth, summer temperatures were 2-3°C warmer than today, and
 CO, concentrations may have been 30  ppm  greater than  the pre-industrial average  (Neftel et aL, 1982).
 Although traditional scenarios have postulated a thermal maximum about 6000 years ago (Deevey and Flint,
 1957), more  recent investigations in western North America indicate maximum summer temperatures 10,000 -
 8000 ya, coincident with maximum insolation (Davis et aL, 1985; Davis et aL, 1986; Davis and Moratto, 1988;
 Elias, 1985; Hebda and Mathewes, 1984; Kearney and Luckman, 1983; Ritchie et aL, 1983; Vance, 1985)  .

      Climate models  such  as  NCAR's CCM  (Kutzbach and Guetter, 1986; Kutzbach, 1987) indicate July
 temperatures 9000-yr BJ*. in California ca. 2S'C greater and no change in precipitation relative  to today,
 compared with increases of 4&C (GISS) and 2.4'C (OSU model) and precipitation decreases of 0.09 and 032
 mm day*1 predicted for 2xCO, increases (prepared for this study).  Six thousand years ago, the CCM indicates
 temperature ca. 1°C higher and precipitation much higher for a combined region of California and the American
 Southwest (Kutzbach, 1987). Precipitation and  temperature reconstructions for California alone have not  been
 published.

     The primary cause of increased summer temperature  in the early and mid-Holocene was greater solar
 radiation in the northern hemisphere, which was matched by decreased insolation in winter (Davis et aL, 1986).
 January  2xCOu  model  results  for California are +4.8°C  (GISS) and +17*C (OSU), whereas  January
 temperature 9000-yr B J*. in California was the same as today according to the CCM (Kutzbach and Guetter,
 1986). CCM reconstructions of January temperature 6000-yr B.P. are near  modern values for most of North
 America (Webb et aL, 1987). Because most the native plants in the Sierra Nevada are dormant during winter,
 the importance of the difference between the 9000 yr B J". analog and 2xCO2 conditions is lessened, but it  is an
 important consideration for other topics such as citrus production and winter crops.


 THE HOLOCENE CLIMATIC HISTORY OF CALIFORNIA

     California is presently a land of climatic contrasts, and its regions have  had  contrasting climatic  histories.
 Coastal California was wet and cold during the last gladation, whereas the western Sierra Nevada was dry (Davis
 et  aL, 1985).  During the early Holocene, the western Sierra Nevada was drier than today, but the desert east
of the Sierra Nevada was wetter (Spauldmg and Graumlich,  1986; Davis and Moratto, 1988). Thus, the 9000-
yr  B .P. analog is different for each area.  By 6000 years ago,  the modern vegetation had become established at


                                               4-2

-------
                                                                                             Davis

most sites in California. Both 6000 and 9000 yr B.P. may be considered analogs for greenhouse wanning because
the first period has traditionally been considered the time of maximum global warmth, and because the 9000-
yr B.P. period has been shown to be warmer than today in western North America. Because the vegetation of
the Sierra Nevada was different during these two periods, they provide a two alternative analogs for greenhouse
wanning.

     In the rest of this report,  I will describe the  numerical techniques used to reconstruct the vegetation of
the Sierra Nevada 9000 ya. The techniques of Dissimilarity Analysis and Detrended Correspondence Analysis
both indicate the character of past vegetation by inditing the most similar modern vegetation sample via pollen
analysis.  Both techniques provide a measure of uncertainty by indicating numerically the degree of similarity.
The reconstructions  are  further evaluated by  comparing  the  climatic implications  of  the vegetation
reconstructions with the outputs of the GISS and OSU climate models. Finally, I will address environmental and
socioeconomic implications of the results and recommend policy changes.
                                                 4-3

-------
 Davis
                                            CHAPTER 2

                                          METHODOLOGY
 THE DATA
      The vegetation of the central Sierra Nevada is reconstructed using modern and fossil pollen samples
 collected by R. S. Anderson (1987) and Owen K. Davis (Davis et aL, 1985; Davis and Moratto, 1988). Forty-
 four modern samples were collected from moss polsten in a west-east transect beginning in grassland at 400 m
 east of Fresno, California, reaching 3445 m above Granite Lakes north of Tioga Pass, and ending at 1280 m
 elevation at Fish Slough near Bishop, California (Anderson and Davis, 1988). The transect passes through all
 the major vegetation zones of the Sierra Nevada, but more samples were taken in forested vegetation where the
 fossil sites are located.  The samples were taken in three sections (Fig. 1) rather than in one straight line, but
 this presents no problems to interpretation since no major north-south vegetation differences exist between the
 sampling localities, and the results are fully analogous to the two transects studied by Adam (1967).

      The fossil samples come from five meadow and lake sites in the central Sierra Nevada (Anderson, 1987;
 Davis et aL, 1985; Davis and Moratto,  1988).  Balsam Meadow (2005 m), Exchequer Meadow (2219 mX and
 Starkweather (2438 m) are in Sierra montane forest, Tioga Pass Pond (3018 m) is in subalpine forest and Barren
 Lake (2816 m) western subalpine forest (Figure 1). The age of the samples is established by 23 radiocarbon
 dates (Table 1).

      Prior to the analysis of Balsam and Exchequer Meadows (Davis et aL, 1985; Davis and Moratto, 1987), it
 was thought that the Sierran meadows came into existence in the late Holocene (Wood, 1975). The radiocarbon
 dates for Balsam  and Exchequer Meadow (Table 1) document continuous and relatively uniform deposition of
 sediment since the late Pleistocene. The persistence of seeds and pollen of sedges and other wet-ground plants
 throughout  the records indicate that these Sierran meadows have existed  since the late Pleistocene.

      Although pollen concentration and influx (accumulation rate) have  been calculated for the  sites (Davis,
 1984,1987; Anderson, 1987), pollen percentages have been used in the numerical analyses of the fossil data for
 two reasons. First, the surface sample data cannot be easily converted to concentrations or accumulation rates,
 and even if they are, the differences among samples could reflect site-specific processes rather than vegetation
 differences.   Second, the pollen concentrations  vary greatly among adjacent samples for the meadow sites,
 probably reflecting variable depositional processes rather than changes in plant abundances.


 DISSIMILARITY COEFFICIENTS

     Measures of dissimilarity permit the multivariate comparison of modern analogs with fossil samples. Low
 dissimilarity values among modern and fossil pollen samples indicate they originate in the same type of
 vegetation. Overpeck et aL (1985) tested eight dissimilarity indices by comparing 1618 different modern spectra
 from eastern North America with fossil samples from three sites.  They found that squared chord distance often
 showed differences that the other coefficients could not

     In this  study, calculations of squared chord distance are based on 15 pollen types that Anderson (1987)
found to vary with elevation in the Sierra Nevada. The statistic is calculated as:
                                                4-4

-------
                                                                                         Davis
    ISO'OO'
                                            STARK WATr/M POND
                                                   /BALSAM flEAOOV
           0
           o"
                                                                                         3d" Off
.                   mi.   A    /
 10   '   20   '  30  **   I  ''
                                                                                         •srscr
Figure L Map of the central Sierra Nevada, showing the location of sites studied in this report
                                              4-5

-------
Table 1.  Radiocarbon Dates for Central Sierra Nevada Pollen Sites
Site Sediment Depth (cm)
Balsam Meadow (2005 m)





Exchequer Meadow (2219 m)





(34-44)
(54-64)
(81-91)
115-125
140-150
213-238
55-65
90-200
172-180
205-230
284-291
326-351
Starkweather Pond (2438 m) 213
Tioga Pass Pond (3018 m)



Barrett Lake (2816 m)





48-58
149-157
234-243
291-305
25-30
44-49
74-80
85-90
126-134
179-197
Age (yr B.P. )
470 + 90
2350 ± 120
3160 ± 140
2920 + 120
6160 + 140
9420 ± 200
1870 ± 60
2980 + 80
4540 ± 90
7070 ± 70
11490 ± 270
10330 ± 380
10879 ± 150
1910 + 80
4060 + 160
6100 + 140
8760 ± 240
695 ± 85
1110 -I- 90
470 ± 120
3930 + 90
8020 + 190
11730 + 430
Laboratory No.
A-3688
A-3689
A-3690
A-3691
A-3692
A-3693
Beta-16111
Beta-17183 v
Beta-17184
Beta-16112
Beta-17185
Beta-16113
AA-1133
A-4448
A-4449

AA-4450
SI-6676
SI-6677
SI-6678
SI-6679
A-4455
A-4456

-------
                                                                                             Davis


Where cL is the dissimilarity between samples i and j, and p,k is the proportion of pollen type k in sample L The
results are displayed graphically by representing each value as a shaded square from light (l°w dissimilarity) to
dark (high dissimilarity) on a matrix with the data sets on the horizontal and vertical axes.  This technique was
proposed by L. J. Maher (1987). The utility of this approach is demonstrated by comparing the 44 modern
samples with the averages for each of the seven major vegetation zones of the transect.

     The sample with lowest dissimilarity is noted with an asterisk (*) if the value is less than 0.1.  This "critical
value" was chosen empirically. Lower values produce too many "misclassifications," Len too many samples are
similar to averages of other vegetation zones. Higher critical values produce too few matches. Critical values
of 0.15 and 0.12 are proposed for the squared chord distance by Overpeck et aL (1985), using data from eastern
North America.


DETRENDED CORRESPONDENCE ANALYSIS

     A second technique for finding analogs is to arrange samples along axes of variation of multivariate data,
a technique known as ordination. Similar  samples group together when plotted on these axes. Detrended
correspondence analysis was performed with the DECORANA program (HOI, 1979) using the same 15 pollen
types used in dissimilarity analyses. Both modern and fossil samples were included in the calculations to reduce
the probability that fossil samples might spuriously plot dose to some modern sample (Jacobsen and Grimm,
1986).  For the same reason, each fossil site was analyzed separately; that is, analyzing several sites together
could result hi patterns reflecting differences among the fossil sites, rather than differences between the fossil
and modem samples.  Modern samples are plotted as symbols  by vegetation type. Fossil samples are plotted
as thousand-year averages connected by a solid line.  Only the first two DCA axes are shown. The technique
is illustrated by comparing the modern data set with Adam's (1967) 21 samples.  For example, the first value (1)
is the average of all values from 0 to 1000 yr BJ», the second number (2) ranges from >1000 to 2000 yr B.P.,
and so on.
                                                4-7

-------
 Davis

                                            CHAPTERS

                                             RESULTS


 THE POLLEN DATA

 The Surface
      A major feature of the modern sample transect (Figure 2) is the positive correlation between elevation
 and Pinus pollen, which exceeds 60% of the pollen sum in samples from forest vegetation, and is less than 35%
 in grassland samples west of the Sierra Nevada, and less than 40-50% in the Great Basin samples east of the
 Sierra crest  Distinctive pollen assemblages occur for each of the  seven  major vegetation zones.  These
 vegetation types are used informally to facilitate discussion of the modern vegetation and the comparison with
 vegetation descriptions for the Sierra Nevada (see Anderson and Davis, 1988, for a more complete discussion).
 The  differences among vegetation and pollen samples within the zones, such as  those demonstrated by the
 numerical analyses of this study, do not detract from the utility of the zone concept

      Relatively few samples were collected at low elevation, so the four lowest samples from the western Sierra,
 grassland, oak woodland, and chaparral, are lumped into a "oak grassland" (GR) zone dominated by the pollen
 of Ouercus. Gramineae, and Other Compositae. Sierra montane (SM) samples are dominated by 10-30% Abies
 pollen, with Cupressaceae reaching 30% below 2000 m.  Upper montane (UM) and eastern subalpine  (ES)
 samples contain up to 30% Tsuea pollen. Subalpine (SA) samples are characterized by the dominance (>80%)
 of Pinus pollen derived from Pinus  murravana and £. flsxjfe which form treeline in this  part of the Sierra
 Nevada. Sample 29 (3445 m) was collected above treeline. It too has a high (87%) pine pollen percentage.  Pine
 forest (PF) samples contain high Artemisia and moderate AJass percentages. Low elevation samples east of the
 crest contain up to 50% Artemisia pollen and 5-10% Sarcobatus pollen.

 The Fossil Sites

      The four sites west of the Sierra crest have similar stratigraphies.  The upper portions of the diagrams are
 dominated  by high (70-90%)  Pinus percentages,  whereas Gramineae, Other Compositae, and Artemisia
 percentages are high near the base.  Abies percentages increase gradually from  the base to the surface.  The
 Balsam Meadow pollen diagram (Figure 3) shows this transition between 160  and 180 on.  At Exchequer
 Meadow (Figure 4) the transition takes place between 200 and 300 cm., and at Starkweather Pond (Figure 5),
 between 180 and 190 cm.  At Tioga Pass Pond it takes place at  the base of the diagram (Figure 6) below
 300cm.

     The  Barrett Lake diagram (Figure 7) is dominated by very high Pinus percentages.  Above 100 cm, Abies
 and Tsuga percentages increase, and Artemisia percentages decrease.


 DISSIMILARrrY INDICES

 Comarison of Moder
     The dissimilarity diagram (Figure 8) shows a general correspondence between individual samples and the
vegetation zone averages.  The oak grassland, upper montane, subalpine, eastern subalpine, and Great Basin
zones are homogeneous, but the Sierra montane and pine forest groups are not Within the oak grassland (GR)
zone, sample 3 has a minimum dissimilarity greater than O.L  Half of the Sierra montane (SM) samples are
dissimilar (>0.1 dissimilarity) to the zone averages, with two samples (16 & 17) more similar (<0.1 dissimilarity)
to the subalpine zone averages than to SM values.  Four  of the five upper montane (UM)  samples have
dissimilarity values less than O.L  Both the subalpine (SA) and eastern subalpine (ES) samples are internally
consistent, being most similar to their respective zone averages at less than 0.1 dissimilarity. None of the six pine


                                               4-8

-------
  SURFflCE SflMPLES
     B IB 2038 40506078 8890188 182838
   18
  O
  2
  a"
_/

"\

                                    18 28 38 48
                                            8 IB a ia
                                                                                                1
                                                                                                >
                                 182838485868 8 IBM BJB B 18 8JB B 1828984858 B 18 BJB 8 IB 28 38 48 BJ8 B     1722   3445
Figure 2. Percentage pollen diagram of 144 modern samples collected In a transect across the central Sierra Nevada. Abbreviations In right   °
         margin refer to different vegetation types. GR = oak grassland, SM = Sierra Montane, UM = Upper Montane, SA = Subalpine,   |-
         ES = Eastern Subalpine, PF = Pine Forest, and GB = Great Basin.

-------
     BRL5RM  MERDOW
          j£%$&titf£&^£*
       ,•.«*         H-*  «&&&>&$W j&fr*' ^**'°
       0 10 20 30 40 50 60 70 80 90 0 102030 0 10 0 10 0 10 0 10 0 10 0 10 0 1020 0 10 0 10 0 102030 0 10 0 10 0 10
     20
     40
     60
     80
  —  100
    5
  ^ 5
  Q_
  UJ
  Q
     120
     160
     180
     200
     220
    10
     240
               i
\
           7
            7

            )
Flgur* 3. Percentage pollen diagram for Balsam Meadow, Fresno Co., California. Dots Indicate less than 1%. Numbers next to depth
     column are ages In thousands of radiocarbon years.

-------
    EXCHEQUER  MERDOW
        0  0
          20
          40
        2 60

          100
          120
        5 140
    ~    160
£  J618B
    ""  7 200
    ^E    220
    UJ  8 240
    Qg260
          280
      10  300
          320
       n
       12
340
360
380
400
             B 10 20 30 40 SB 60 7B BB 9B 0 IB2B3IB 0 IB 0 IB2B BJB B IB B IB2B3fl4BSB BIB2B B IB B IB 0 IB2B 0 IB B IB B IB B IB
               "                   	      ^                  •*•   *T    r   i  "*"  *    *  i-j-j   ..fc.«j  *• --•-'  *- --•    —*  *--—*

   Figure 4. Percentage pollen diagram for Exchequer Meadow, Fresno Co., California. Dots Indicate less than 1%. Numbers next to the
            depth column are ages In thousands of radiocarbon years.

-------
             5TRRKWEHTHER  POND
                0 102030405060708090 0 10 0 10 0 10 20 30 40 0 10 0 10 0 10 0 10 0  10 0 IB 0 10 0 102030 0 IB 0 10 0 10 0 10




20


1 40
. 60
4
5
/> 80
_ 6
e 7
u 100
*-^ 8
m
H- 120
Q_
UJ


160

10180

200


I
X
/
s
\
1


1
/

/
\
/
f

(
X
7
\
/
_^ — -^'
\
/
\





j

















\
y

r— |





























1

i
^7

1
J
1










A
ft
^\
x1
























































(











































i







:

















\
/
(
1




(











































L
i














;









\

/
z\


t
























t






































i



j








    Figure 5. Percentage pollen diagram for Starkweather Pond, Madera Co., California. Dots Indicate leas than 1%. Number* next to the death
5            column are ages In thousands of radiocarbon years.                                                       ««pm
(0
Q

-------
        TIOGfl PR55  POND
           o   0




             40


             60


           2  B0

             100
                B 182030405060708090   B IB B IB B IB 2030 B IB B IB B IB B  IB B IB B IB B IB B IB 20 0 10 0 IB 0 10 0 10
 E   140
 U

~   160

1 4
I-   180
Q_

Q 5200

     220



   6™
     260


   7 280


     300
                                                                                     I
Figure 6.  Percentaga pollen diagram for Tloga Pass Pond, Mono County, California. Dots Indicate less than 1%. Numbers next to depth
         column are ages In thousands of radiocarbon years.

-------
                BRRRETT  LRKE
                    0 102030405060708090 0 10 0 10 0 10 0 10 0 10 0 10 0 10 0 10 0 10 0 10 0 10 20 0 10 0 10 0 1020 0 IB
                 20
               2  60




            _  3  BB

             5  -
            —  5  100

            m  6

            n~  7  120
            B 8
              10


              11
140





160




180
§     Figure 7. Percentage pollen diagram for Barrett Lake, Mono County, California. Dots Indicate leas than 1%. Numbers next to depth

-            column are agea In thousands of radiocarbon years.

-------
                                                                                       Davis
    abcdefghijklmnopqrstuvwxyzABCDEFGH'IJKLMNOPQR
9
10

12
13
14
15
16
17
18
19
20
21
1/1
24
A 0
26
27
28
      I->, |  --{
      •. ..i< ---------- SM ---------- >< --- UM --- >< — SA — ><-ES — >< --- PF --- >< — GB->

    Dissimilarity Scale
    0.0 -  0.02
    0.02-  0.1
    0.1 -  0.2
    0.2-0.5
    0.5-0.8   J
    0.8 -  0.9  II
         >  0.9  •
      Figure 9.  Dissimilarity diagram comparing Balsam Meadow pollen samples with 44 modem pollen samples.
               Asterisks mark the most-similar modern samples. Numbers on extreme left refer to ages of the
               samples (3,6, and 9,000 yr B.P.). GR = oak grassland, SM = Sierra Montane Forest, UM = Upper
               Montane Forest, SA = subalpine, ES = eastern subalpine, PF = pine forest, and GB = Great Basin
               Steppe.

      Horizontal Axis = Modern Pollen Samples Sierra Nevada
      Vertical Axis = BALSAM MEADOW FRESNO CO. CA
                                               4-17

-------
 Davis
 forest (PF) samples is similar to the zone averages.  Five of the six are most similar (<0.1 dissimilarity) to the
 averages of various other zones.  One of the five Great Basin samples is similar to the eastern subalpine
 averages, and one has high (>0.1) dissimilarity to the GB averages.

      Dissimilarity analysis fails to match one-third of the samples to their zones. Nine (20%) of the samples
 do not match any zone average (>0.1 dissimilarityX and 8 (18%) of the 44 samples are similar to the wrong zone
 averages (<0.1 dissimilarity). The fossil dissimilarity diagrams must be interpreted with these errors in mind.
 General trends should be apparent, but occasional mismatches are anticipated.

             of Fossil and Modern
      The dissimilarity diagrams for the five fossil sites (Figures 9-13) form U-shaped patterns of low-dissimilarity
 (<0.1) samples surrounded by high-dissimilarity values on the sides and bottoms of the figures.  The sides are
 dissimilar to fossil values because they are never similar to low-elevation vegetation, the bottoms are dissimilar
 because the early-Holocene vegetation was different from any of the modern analogs. The transition from the
 no-analog vegetation of the early Holocene occurs shortly after 9000 yr BJ>. at Balsam, Exchequer, and Tioga,
 and is earlier at Starkweather.

      The early-Holocene samples from  Balsam, Exchequer, and Starkweather are similar (<0.1 dissimilarity)
 to samples east of the Sierra crest.  At each of the three sites, the transition to similarity with western samples
 takes place ca. 9000 yr B J*. Tioga and Barrett also show this trans-Sierran similarity transition, but during a late
 Holocene  (Tioga) and early Holocene (Barrett).

      Prior to ca. 7000 yr B.P., the Balsam Meadow samples (Figure 9) are similar to modern analogs in the
 pine forest and eastern subalpine zones.  From 7000 to ca. 3000 yr B.P., the fossil samples match the upper SM
 and UM zones. After 3000 yr B.P., the fossil samples are generally most similar to sample j, collected at the site.
 The Exchequer Meadow samples (Figure 10) are similar to western Sierra analogs prior to ca. 8200 yr B.P.
 Most of the Holocene samples are match samples from the upper Sierra montane forest Although four of the
 samples (3, 13, 14, and 25) from Starkweather Pond (Figure 11) appear to be misdassified, the general pattern
 is similar to that for Balsam and Exchequer. The east-west similarity transition takes place ca. 9000 yr B.P., early
 Holocene samples are similar to upper montane and subalpine analogs, and little change is evident after ca. 6000
 yr B J>. Most of the Tioga Pass Pond samples (Figure 12) are similar to subalpine samples from west of the
 crest, but samples 14-16 (ca. 3500 yr BJ».) match samples from  east of the crest The Barrett  Lake  samples
 (Figure 13) are primarily similar to samples from the eastern subalpine forest, but samples 28-35 (7900-12,100
 yr B J1.) are similar to samples from west of the Sierra crest


 DETRENDED CORRESPONDENCE  ANALYSIS

 Comarison of Modern S
     Figure 14 compares the 44 modem samples with a comparable set of samples collected in a west-to-east
transect across Tioga Pass by David Adam (1967). A line is drawn around each cluster to facilitate discussion.
Samples from the two low-elevation zones - oak grassland and Great Basin steppe — form distinct groups at the
lower left and extreme right margins of the  plots. The forested types form overlapping groups toward the top
of the diagrams, with the Sierra montane and upper montane samples to the left, and the subalpine, eastern
subalpine, and pine forest samples to the right

     This general pattern also appears in the fossil diagrams (Figures 15-19): the first axis provides a west-east
ordination of the samples, the second  axis is primarily an elevational ordination.  Note that the polarity (left-
right, up-down) of the axes shifts from figure to figure. For example, western Sierra samples plot to the left in
Figures 14, 15, and 16, but to the right in Figures 17, 18, and 19. A possible climatic interpretation is that the
                                                4-16

-------
1
2
3
4
5
C
D
7
8
9

10
13
14
15
16
17
18
19
20
21
0-5
6 £
23
24
^
25
26
27
28
                                                                                       Davis
    abcdefghijklmnopqrstuvwxyzABCDEFGH'IJKLMNOPQR
     '1                  *                                                                         «
                        *
                        *
    J
     1 fM& vvCwi «&8t
    <-GR — >< ---------- SM ---------- >< --- UM --- >< — SA — ><-ES — >< --- PF --- >< — GB->

     Dissimilarity Scale
     0.0  -  0.02
     0.02-  0.1
     0.1  -  0.2
     0.2-0.5    1
     0.5  -  0.8   il
     0.8  -  0.9   H
          >  0.9   •
      Figure 9.  Dissimilarity diagram comparing Balsam Meadow pollen samples with 44 modern pollen samples.
               Asterisks mark the most-similar modern samples.  Numbers on extreme left refer to ages of the
               samples (3,6, and 9,000 yr B.P.). GR « oak grassland, SM =» Sierra Montane Forest, UM = Upper
               Montane Forest, SA » subalpine, ES = eastern subalpine, PF = pine forest, and GB » Great Basin
               Steppe.

      Horizontal Axis = Modern Pollen Samples Sierra Nevada
      Vertical Axis = BALSAM MEADOW FRESNO CO. CA
                                               4-17

-------
          Davis
     1
     2
     3
3   5
    6
    7
    8
    9
6
         abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQR
         to       '           *                                                                       .  . B
                                                                                  XJ,,
         "xsi "X;'
     10
     1 1
     13
     14
     15
 9   16
     17
     18
     19
     20
     21
12   22
     23
     24
*

*
         <-GR—><	SM	><	UM	><—SA—><-ES—><	PF	><—GB->

         Dissimilarity  Scale
         0.0  - 0.02
         0.02- 0.1
         0.1  - 0.2
         0.2  - 0.5
         0.5-0.8  3
         0.8  - 0.9  H
               > 0.9  •
          Figure 10. Dissimilarity diagram comparing Exchequer Meadow pollen samples with 44 modern pollen samples.
                   Asterisks mark the most-similar modern samples. Numbers on extreme left refer to ages of the
                   samples (3, 6, and 12,000 yr B.P.). GR = oak grassland, SM = Sierra Montane Forest, UM  =
                   Upper Montane Forest, SA = subalpine, ES « eastern subalpine, PF = pine forest, and GB = Great
                   Basin Steppe.

          Horizontal Axis =  Modern Pollen Samples Sierra Nevada
          Vertical Axis = Exchequer Meadow Fresno Co. CA
                                                    4-18

-------
                                                                   Davis
 1
 2
 3
 4
 5
 6
 7
 8
 9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
27
     abcdefghijklmnopqrstuvwxyzAB
     *2
     ^•X
     4^
w.*
                                               CDEFGHIJKLMNOPQR
                      *


                      *


                      *


                      *


                      *


                      *

                           '* X-.-x
     __               	  •*" •»•- >«w«M«««aud .,1,  . .  ' JVA   t
   <-GR-><	SM	»<—«-.>;-"-«-
   Dissimilarity Scale
   0.0 - 0.02
   0.02- 0.1
   0.1 - 0.2
   0.2 - 0.5   !
   0.5 - 0.8  ?I
   0.8-0.9  H
       > 0.9  •
                                               ><-ES — ><	pp	><	pn_-
          samples(3,6,and9,OOOyrBJ.).
                                  4-19

-------
 Davis
    1
    2
    3
    4
    5
    6
    7
    8
    9
    10
    11
    12
    13
    14
    15
    16
    17
    18

    20
    24
    25
6   26

    28
    29
    30
    31
    32
9   33
    34
        abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQR
    *
    *
    *
    *
    *
    *
*
*
        <-GR—^><	SM	><	UM-	>< — SA—><-ES—><	PF	><—GB->

        Dissimilarity -Scale
0
0
0
0
0
0

.0
-
.02-
.1
.2
.5
.8

-
-
-
-
>
0
0
0
0
0
0
0
.02
.1
.2
.5
.8
.9
.9
          Figure 12. Dissimilarity diagram comparing Tioga Pass Pond pollen samples with 44 modern pollen samples.
                   Asterisks mark the most-similar modern samples.  Numbers on extreme left refer to ages of the
                   samples (3,6, and 9,000 yr B.P.). GR = oak grassland, SM - Sierra Montane Forest, UM =* Upper
                   Montane Forest, SA = subalpine, ES = eastern subalpine, PF = pine forest, and GB = Great Basin
                   Steppe.

          Horizontal Axis » Modern Pollen Samples Sierra Nevada
          Vertical Axis = Tioga Pass Pond Mono Co. CA
                                                    4-20

-------
       abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQR
3
     1
     2
     3
     4
     5
     6
     7
     8
     9
     10
     11
     12
     13
     14
     15
     16
     17
     18
     19
     20
     21
     22
     23
     24
     25
     26
     27
     28
     29
 9   30
     31
     32
     33
     34
12   35
     36
     37
     38
6
                                                                            *
                                                                            *
                                                                            *
                                                                            *
                                                                            *
                                                                            *

                                                                            *
                                                                            *
                                                                            *
          -,,; ,,.J ^ •:...-.;>
I\LJI
                                                                          *

                                                                         '*
       ^j%^:1%%3
                                                                          *
                                                                          *
                                                             *
                                                             * "
         r
         i  17;
       <-GR—><	SM	><	UM	>< — SA—><-ES — ><	PF	><—GB->
       Dissimilarity  Scale
       0.0 - 0.02
       0.02- 0
       0.1  - 0,
       0.2  - 0,
       0.5-0.8  ^
       0.8  - 0.9  H
             > 0.9  •
          Figure 13. Dissimilarity diagram comparing Barrett Lake pollen samples with 44 modern pollen samples.
                   Asterisks mark the most-similar modem samples. Numbers on extreme left refer to ages of the
                   samples (3, 6, and 12,000 yr B.P.).  GR = oak grassland, SM = Sierra Montane Forest, UM =
                   Upper Montane Forest, SA = subalpine, ES = eastern subalpine, PF =  pine forest, and GB =
                   Great Basin Steppe.

          Horizontal Axis = Modern Pollen Samples Sierra Nevada
          Vertical Axis = Barrett Lake Mono Co. CA
                                                   4-21

-------
 Davis
            e GRRSSLflND

            OSUBflLPINE

            + GRERT BflSIN
     160



     140



     120



CN   100

in
H— I
X
      60
      40
     20-1
      0
             XSIERRfl MONTflNE     D UPPER MONTflNE

             AEflSTERN SUBRLPINE  ZPINE FOREST
        0
                         Ntt
                                                          .20
30
60         90
     RXIS  1
120
150
180
Figure 14. Detrended correspondence analysis of the modern set of 44 samples combined with the set of 21
         samples collected by Adam (1967).
                                      4-22

-------
                                                                      Davis
           ® GRRSSLRND


           OSUBRLPINE


           + GRERT BRSIN
(N
X
cz
     140
     120
     100-
     80
     60-
     40
     20
      0
XSIERRfl MONTRNE     D UPPER MONTRNE


AERSTERN SUBRLPINE  ZPINE FOREST
  BflLSRM
                             e
             e
        0      20     40     60     80     100    120    140    160    180

                                RXIS   1
Figure 15. Detrended correspondence analysis of Balsam Meadow and modern samples, 1,000-yr averages
        plotted for fossil samples.
                                    4-23

-------
 Davis
            ©GRRSSLRNO       X 5IERRR MONTRNE     D UPPER MONTRNE


            OSUBflLPINE       AERSTERN SUBRLPINE  XPINE FOREST
           + GRERT BRSIN
     180-
     160-
     140
     120-
 CN
in
i—*   100
X
     60-
     40-
     20-
      0
   EXCHEQUER
                         Z
                                e
        0     20    40
60     80     100

    flXIS 2
120    140    160    180
Figure 16. Detrended correspondence analysis of Exchequer Meadow and modern samples, 1,000-yr averages
        plotted for fossil samples.
                                  4-24

-------
                                                                       Davis
             ® GRRSSLRNO


             O SUBRLPINE


             + GRERT BRSIN
                          X SIERRfl MONTRNE     Q UPPER MONTRNE


                          A ERSTERN SUBRLPINE  2 PINE FOREST
    160



    140




    120



CN  100

in

X
cr   80



     60




     40




     20




      0
                               STRRKWERTHER
                                 Z
                                      a
                     2
                   2
                +
               +
e
           0      20      40      60      80     100     120
                                 RXI5 1
                                                        140    160
Figure 17. Detrended correspondence analysis of Starkweather
        plotted for fossil samples.
                                             and modern
       averages
                                    4-25

-------
 Davis
       180
      160
      140
      120
  CM
  in
  i—i  100
  X
  cr
       60
       40.
       20-
        0
             e GRRS5LRND
             OSUBflLPINE
             + GRERT BRSIN
XSIERRR MONTRNE     D UPPER MONTRNE

AERSTERN SUBRLPINE   SPINE FOREST




  TIOGfl   PflSS   POND
                  z
                                     "9

         0      20     40      60     80     100    120     140     160
                                RXIS  1
Figure 18. Detrended correspondence anafyas of Tioga Pass Pond and modem samples, 1,000-yr averages
       plotted for fossil samples.                                           ,*
                                  4-26

-------
                                                                     Davis
           eGRflSSLflND      XSIERRfl  MONTRNE     D UPPER MONTRNE


           OSUBRLPINE      AER5TERN SUBflLPINE  Z PINE FOREST
           + GRERT BflSIN
180



160



140



1201
(N
I—I   100 •
X
cn
     60
     40-
     20.
      0
                      X
                              BflRRETT
           H-S-
        0       20
                                                           x    x
                   40     60      80
                            flXIS  1
100     120     140     160
Figure 19. Detrended correspondence analysis of Barrett Lake and modern samples, 1,000-yr averages plotted
        for fossil samples.
                                   4-27

-------
 Davis

 first axis is a moisture ordination from wet (western Sierra) to dry (eastern Sierra) and the second axis is a
 temperature ordination from warm (low elevation) to cold (high elevation).  The boundary between SM + UM
 samples versus SA +• ES +• PF is taken as a boundary between eastern and western Sierra Nevada types.

      Most of Adam's
 set  The positions i
 Adam, 1967). Sample         .     __       _          _        _  .             ,_._,.,.,
 very high Castanopsis percentages. Samples 16, 20, and 21 have high Artemisia percentages combined with low
 Pinus percentages, the  distance between these three samples and the groups is less than for samples 1, 2, and
 4. This comparison is a guideline for evaluating the fossil sequences. Samples as far removed as the six outliers
 do not have close analogs in the modem vegetation.

 DCA of Fossi Avpes anf^ Modern Analogs
      The Balsam Meadow 1000-year averages of fossil samples form a diagonal line from the upper right to
 lower left of the diagram (Figure 15), implying progressively greater moisture (eastern Sierra analog to western
 Sierra) and temperature (high to low elevation analog).  The 8000- to 10,000-yr averages plot with the eastern
 Sierra modern analogs, and the 5000- to 1000-yr averages plot with western Sierra analogs. The convergent
 pattern formed by the 5000- to 1000-yr averages implies relatively small vegetation change. The 7000- and 6000-
 yr averages are  near the east-west  transition.  The modern samples form a pattern similar to that of Figure 14
 - axis 1 forming an west (left) to east (right) ordination, axis 2 an elevational ordination.

      The DCA diagram for Exchequer Meadow (Figure 16) shows a similar pattern, but the east-west transition
 is between 9000 and 10,000 yr B J>, with the millennium averages plotting closer to low-elevation, western Sierra
 vegetation from 12,000 to 5000 yr B J*. Thereafter, the trend reverses. The earliest vegetation recorded in the
 Exchequer Meadow diagram (Figure 12) is missing from the Balsam Meadow diagram (Figure 11).

      The Starkweather Pond  analysis (Figure 17) shows relatively little change in the millennium averages.
 The fossil samples plot along the east-west transition, so there is no evidence for an east-west transition. Higher
 values on axis 2 imply ingraa«dng similarity to high-elevation vegetation, Le^ progressive cooling from 9000 to
 3000 yr B J*.  The trends for Starkweather Pond  (Figure 17) are similar. There is weak evidence for an east-
 west transition on axis 1, and the upward movement on axis 2 implying cooling from 9000 to 3000 yr B.P.

      The polarity of axes 1 and 2 for Barrett Lake (Figure 18) are opposite those for Balsam and Exchequer
 Meadows (Figures 15 & 16).  Low-elevation samples plot at the  top of the figure, and western Sierra samples
 plot at the right. Relatively little change is apparent in the millennium averages after 12,000 yr B.P. The 12,000-
 and 13,000-yr B J*. averages plot far to the right,  with western Sierra modern analogs.

      The western Sierra sites (Figures 15-18) show similar chronologies of change but conflicting "temperature"
 histories. Early Holocene trends reverse after 5000 (Balsam and Exchequer) to 3000 yr B J». (Starkweather and
 Tioga Pass Pond), roughly contemporaneous with the onset of the Neogiatial (Wood, 1975).  The ambiguity
 probably results from my  interpretation of axis 2 as an elevational, and hence temperature gradient The
 relationship between elevation and axis-2 score is not perfect  (Figure 14), and elevation is a complex gradient
 involving both temperature and precipitation.  I have interpreted  the mathematical derivation of DECORANA
 axes to prevent correlation (HOI, 1979) as an indication that moisture is not important to axis 2, because it is
 dearly important to axis 1, the  west-east axis.  However, this interpretation may fail for Starkweather and Tioga
 Pass  Pond (Figures 17 &  18X which show little  change on axis 1, the "moisture  axis."  A parsimonious
 interpretation is that axis 2 of the Starkweather and Tioga DCAs shows the same environmental history as axis
 1 of Balsam and Exchequer DCAs; that is, increasingly mesic vegetation until 3000-5000 yr B .P. and more xeric
 vegetation thereafter.

     Given the uncertainty resulting from the conflicting interpretations of the DCA results for the western
Sierra sites,  the pattern of millennial averages from 12,000 to 9000 yr B.P. at Barrett Lake (Figure 19) may
represent a trend toward more xeric vegetation, cooler temperatures, or both.
                                                4-28

-------
                                                                                               Davis

                                            CHAPTER 4

                                 INTERPRETATION OF RESULTS


     Dissimilarity analysis and DCA provide complementary scenarios of vegetation change in the central Sierra
Nevada. Prior to ca 9000 yr B J», the vegetation of the western Sierra Nevada resembled that found east of the
Sierra crest today, by 6000 yr B J*. modern vegetation patterns had become established.  The trend  toward
decreasing aridity which began in the early Holocene continued until 5000 to 3000 yr B J*., when the direction
reversed. The two techniques differ in particulars, but the direction and magnitude of the vegetation change are
consistent  The 9000-yr B.P. analog betokens xeric vegetation in the  western Sierra; the 6000-yr B.P. analog
indicates vegetation similar to today.

     DCA analysis  is useful for showing overall trends of vegetation change, whereas dissimilarity analysis
provides specific modem analogs, albeit with occasional misdassification. Tables 2 and 3 compare the results
of the two techniques. In Table 3, the Euclidian distance between millennium averages was calculated as a
measure of "greatest change" for DCA. For dissimilarity analysis, the number of samples in the transect was
used as a measure of change. DCA indicates an earlier east-west transition than dissimilarity analysis for Balsam
Meadow, later at Exchequer Meadow, and inconclusive results at Starkweather, Tioga, and Barrett  The east-
west transition is the greatest change recorded at Balsam, Exchequer, and Tioga, but the end of the ice age
produced greater change at Starkweather and Barrett (Table 3).

     Since the goal of the analyses is to use an ancient analog to estimate the potential effects of greenhouse
warming, it appropriate to compare the results of this study with those produced by the GESS and OSU climate
models.  Climate estimates are calculated for  fossil samples with good modern analogs (<0.1 dissimilarity;
Figures 9-13). Annual precipitation and mean annual temperature is calculated from lapse rates calculated by
Rourke (1988). The data set includes 44 stations from central California ranging from 7 m to 3231 m elevation.
The lapse rates for the eastern Sierra Nevada are 181.6 mm"1"1 and -5J6"C km"T, and 683.7 mm*™ and -4.93°C
km  Five-level Gaussian smoothing was performed to remove effects of extreme values (Table 4). Ordinarily,
the elevational-analog technique produces negatively correlated temperature and precipitation reconstructions,
but the east and west slopes of the Sierra Nevada have different lapse. Because lowest dissimilarity can switch
from one slope to another, changes in temperature and precipitation are not necessarily correlated.

     The western Sierra samples are discussed separately from Barrett Lake due to their mutual similarity and
their differences from the Barrett Lake climatic reconstruction.  During the 6000-yr B J». analog, mean annual
temperature is cooler than modern in three of the four western Sierra sites, and unchanged in the third. Annual
precipitation is greater than modern in two sites but less in the other two. During the 9000-yr BJ*.  analog,
temperature is lower in three sites, but greater in the third.  Precipitation is greatly reduced  (by over 50% !) in
three of the four sites, and slightly reduced in the third (Table 4). The combined history is of relatively little
change 6000 yr B .P, but of much more xeric vegetation 9000 yr
     The xeric character of the early Holocene vegetation is supported by pollen accumulation rates (Davis,
1984,1987). In the Pinus stratigraphic Zone (after ca. 7000 vr BJ.i the nine pollen influx at Balsam and Dinkey
Meadows is over 15,000 grains cm  yr   and fir pollen influx is over 2000 grains cm"2 yr  .  Earlier, in the
Artemisia zone, the rate for pine is less than 9000 grains cm"2 yr" , and for fir it is less than 600 grains cm'2
yF-
     During the early Holocene and late glacial, the climate at Barrett Lake was markedly different from that
of the western Sierra sites. During the 6000-yr BJP. analog, annual temperature was slightly reduced and annual
precipitation was the same as modern.  During the 9000-ya analog, temperature was lower than during the 6000-
yr B J. analog, and precipitation was increased nearly threefold. An eariy-Holocene precipitation increase is
supported by reconstructions of climate  for the Mojave Desert east of the Sierra Nevada (Spaulding and
Graumlich, 1986). A trans-Sierra contrast of climatic history has been suggested by Davis and Sellers  (1987).
                                                4-29

-------
Davis
                   Table 2. Modern Analogs for Past Vegetation
   Site & Age
Dissim. Analysis
DCA
   Balsam Meadow
      OK
      3K
      6K
      9K

   Exchequer Meadow
      OK
      3K
      6K
      9K
      12K

   Starkweather Pond
      OK
      3K
      6K
      9K

   Tioga Pass Pond
      OK
      3K
      6K
      9K

   Barret Lake
      OK
      3K
      6K
      9K
      12K
     Sierra Montane
     Sierra Montane
     Upper Montane
     Pine Forest
     Sierra Montane
     Sierra Montane
     Sierra Montane
     (Pine Forest)
     (Pine Forest)
     Sierra Montane
     Sierra Montane
     Pine Forest
     Subalpine
     Subalpine
     Subalpine
     Subalpine
     Eastern Subalpine
     Eastern Subalpine
     Eastern Subalpine
     Eastern Subalpine
     Subalpine
     Subalpine
Sierra Montane
Sierra Montane
Eastern Subalpine
Pine Forest
Sierra Montane
Sierra Montane
Sierra Montane
Sierra Montane
Great Basin
Sierra Montane
Sierra Montane
Sierra Montane
Pine Forest
Subalpine
Eastern Subalpine
Eastern Subalpine
Subalpine
Pine Forest
Pine Forest
Pine Forest
Pine Forest
Sierra Montane
                               4-30

-------
                                                           Davis
                  TableS. Tuning of Vegetation Change
     Site                 Dissim.  Analysis         DCA
Balsam Meadow
     East-West Transition      9K                  8K
     Greatest Change           9K                  9K

Exchequer Meadow
     East-West Transition      9K                  10K
     Greatest Change           9K                  UK

Starkweather Pond
     East-West Transition      9K                  9K
     Greatest Change           (3K)                 3K

Tioga Pass  Pond
     East-West Transition
     Greatest Change           —                  7K

Barrett  Lake
     East-West Transition      —                  12K
     Greatest Change           —                  12K
                             4-31

-------
Davis
                           Table 4.  Temperature and Precipitation Estimates
S i te Temperature ( C )
Balsam Meadow
OK
3K
6K
9K
Exchequer Meadow
OK
3K
6K
9K
Starkweather Pond
OK
3K
6K
9K
Tioga Pass Pond
OK
3K
6K
9K
Barret Lake
OK
3K
6K
9K
12K

8
6
4
6

8
5
5
7

6
6
5
3

3
4
3
5

6
5
5
3
3
Precipitation (cm/yr)

119
128
138
45

119
136
138
44

137
137
114
135

163
94
155
50

48
48
48
135
127
                                               4-32

-------
                                                                                              Davis

     The GISS and OSU climatic models for 2xCO2 do not closely match either historic analog. They predict
increased temperature (2-4°C) and reduced precipitation (15-25 mm yr"1) for the grid points closest to the
western Sierra. Both historic analogs call for reduced temperatures in the western Sierra, and the 9000-yr B.P.
analog calls for a much greater reduction in precipitation (ave. 65 mm yr*1) than do the 2xCO2 models.

     The models provide conflicting precipitation reconstructions for the eastern Sierra. The GISS grid point
closest to the eastern Sierra (3L30N 35.22N x HOW) predicts an increase of 51 mm yr*1 for 2xCO~ but the
OSU model indicates a 77 mm yr'1 decrease for its 32-34N x 115W grid point  Because the 6000-yr BJP. analog
for the eastern Sierra indicates little change, it differs from both models. The 9000-yr B.P. analog agrees with
the GISS, but not with the OSU model Both models indicate temperature increases for 2xCO, (5°C for GISS,
3°C for OSUX but both analogs indicate lower temperature.

     The GISS model predicts reduced precipitation west of the Sierra  Nevada and increased precipitation to
the east, but the overall correspondence between the ancient analog and computer model is poor due to the
analog's indications of lower temperature 9000 yr BJ*.  Some  of the differences may reflect nonclimatic
differences between the early-Holocene environment and that of today. In particular, the soils would have shown
the effects of glacial and periglacial action, and would have contained less organic matter and lower nitrogen.
Soil moisture-holding capacity should have been less, reinforcing the effects of lower precipitation. In addition,
the coarseness of the grid points of the GCMs is simply too great to represent opposite sides of the Sierra crest


UNCERTAINTIES

     The techniques  used herein are subject to several constraints.  The vegetation reconstructions based on
pollen analysis are most reliable if analogs exist in the modem vegetation. The reconstructions are based only
on close  (<0.1  dissimilarity) matches  for dissimilarity analysis,  but the  technique produces frequent
misdassulcations ("Figure 8).  DCA does not provide an index of the precision of the match.  Despite these
uncertainties, there is good general agreement of the timing and direction of vegetation change among the four
western Sierra sites.

     The reconstructions for the eastern Sierra are less certain because only one site has been studied. However,
the environmental sequence for Barrett Lake matches that of the desert east of the Sierra Nevada (Spaulding
and Graumlich, 1986). Additional sites at high and mid-elevation should be studied on the eastern slope of the
Sierra Nevada to confirm the Barrett Lake record and provide better coverage (Figure 20).

     More disturbing is the disagreement between the temperature sequence resulting from numerical analyses
and estimates based on the elevational distribution of indicator taxa at Exchequer Meadow (Davis and Moratto,
1988). The presence of Sequoiadendron pollen Juniperus or Calocedrus macrofossils in early Holocene sediment
from Exchequer meadow indicates that temperatures were no colder than today. Both Sequoiadendron and
Calocedrus would be near their upper elevational limit at Exchequer Meadow (2219 mX and neither is present
there today.  For these species to have been present near their current upper-elevational limits, temperatures
could not have been much colder than today.

     Both numerical techniques indicate lower temperatures during the early Holocene.  The closest modem
analogs indicated by dissimilarity analysis are samples taken near timberline or in pine forest (Figures 9-12), and
the eariy-Holocene millennium averages plot near high-elevation samples on the DCA axes (Figures 15-18).
Since the numerical techniques  are based on the analog technique,  they  could fail if the eariy-Holocene
vegetations have no  modern analog.   At Barrett  Lake  and Starkweather Pond, the few fjoys. macrofossils
preserved in eariy-Holocene sediment are of £. monticola. £. murravana. and P.. flggjlis. (Anderson, 1987). These
are not the species prominent in the modern pine forest of the eastern Sierra (£. ieffrevi and P.. ponderosa are),
so a modem forest may be a poor analog for the early Holocene pine forest at mid- to high elevation. However,
£. murravana and P.. flgajjs. are timberline species, so  their presence supports an interpretation of lower
temperature during the Early Holocene.
                                                4-33

-------
  Davis
                 4000



                 3000



                 2000



                 1000



                     0
                              MODERN
(West)
             Tioga Pass Pond


            Starkweather q
             Excheqi
             Baisam
                              a Barrett
                                      (East)
                 4000



                 3000



                 2000



                 1000




                     0
(West)
                                  6K
             Tioga Pass Pond


            Starkweather
             Exchequer
             Baisam
                                Barrett
                                      (East)
                 4OOO




                 3000




                 2000




                 1000 -
                    0

                 Elev. (m)
                              (West)
                                 9K
                                         a Barrett
 Tioga Pass Pond

Starkweather a
 Exchequera  '
 Baisam
                                                          Great Basin (?)
                                                  (East)
Figure 20.  Vegetation zonation in the central Sierra Nevada, 6,000 ya (6k) and 9,000 ya (9k) reconstructions
          based on DCA and DA.
                                           4-34

-------
                                                                                              Davis


     Given the  conflict between the interpretations,  it seems  best  to  conclude  that  the  early-Holocene
temperature record is uncertain. However, the interpretation of eariy-Holocene aridity in the western Sierra is
supported by the numerical analysis and the macrofossU abundances (Davis et aL, 1985; Davis and Moratto, 1988,
Anderson, 1987). It has been suggested that the presence of Sequoiadendron pollen at Exchequer Meadow is
inconsistent with greater  aridity.  Rondel  (1972)  has demonstrated that soil moisture  is greater within
                groves than beyond the grove boundaries.  However, this variable affects the distribution of
                     community scale, and not at larger biogeographic scales. Factors other than sofl moisture
must determine to elevational and aerial distribution of Sequoiadendron groves, because sites with sufficient soil
moisture but without Seauoiadendron exist both within the plants' range and beyond it The Seauoiadendron
pollen at Exchequer meadow indicates that soil moisture was sufficient to support a grove, a conclusion made
obvious by the presence of sedge pollen in the sediment (Davis and Moratto, 1988)

     A  more general uncertainty concerns the time period chosen as an analog for greenhouse warming.
Typically, the middle Holocene (ca. 6000 yr BJ».) is used because it has been traditionally thought to be a period
of higher temperature (Deevey and Flint, 1957). The reasons for considering an eariy-Holocene analog (9000
yr B.P.) for California are outlined in the introduction to this paper.  However, the numerical methods indicate
that both 6000 and 9000 yr B f. were colder than today. This contradicts the nearly universal prediction for the
effects of greenhouse wanning. Although the area may become drier due to greenhouse processes, the analog
appears to have failed to show the effects of elevated temperature on the forests of the western Sierra Nevada.

     Finally, the early-Holocene analog may not simulate the effects of rapid climatic change on the vegetation
of the western Sierra Nevada.  The rapidity with which climate  may change between now and the next century
could be without precedent  The processes of migration and ecological interaction that operate during climatic
changes  (Cole, 1985) may be  too slow to produce the types of communities that existed during the ancient
climatic analogs for  greenhouse warming. The rapidity of greenhouse-induced climatic change may produce
biotic communities for which there is no ancient analog.
                                                 4-35

-------
 Davis


                                            CHAPTERS

                                    IMPLICATIONS OF RESULTS


 ENVIRONMENTAL IMPLICATIONS

      The implications of the results differ for the two analogs. The 6,000-yr BJP. analog implies relatively little
 change. However, during the 9,000-yr BJP. analog, the forests of the western Sierra resembled those currently
 found in the eastern Sierra (Figure 20). Xeric vegetation surrounded the fossil sites that now are in mesic Sierra
 montane forest, and the pollen of lumber-producing pines and firs was less plentiful Some trees endemic to the
 central Sierra were more widespread than today. Giant sequoia pollen was common during the early-Holocene
 at Exchequer Meadow (Davis and Moratto, 1988) and near Boyden Cave (Cole, 1983) where none is present
 today. Six thousand years ago the vegetation zones were similar to those of today, but 9,000-yr BJP. trans-Sierran
 biotic differences were less pronounced.

      We have no fossil sites at low elevation, but the abundance of Ericaceae pollen at the base of the Balsam
 and Exchequer diagrams (Figures 3 & 4) may indicate that chaparral vegetation was much more extensive 9,000
 ya.  Fire frequency could have been greater than today, but the abundance of charcoal in the sediments is less
 at low elevation (Davis et aL, 1985; Davis and Moratto, 1988).  However, the charcoal abundance probably
 reflects the  lower fuel load and, hence, charcoal production in the early-Holocene vegetation rather than lower
 fire frequency.


 SOCIOECONOMIC IMPLICATIONS

      Forest changes in the direction and magnitude of those seen during the 9000-yr BJP. analog would have
 dramatic effects on tourism and the lumber industry.  Quantitative estimates are not possible with the lands of
 data presented here, but the total area occupied by forest was perhaps half that of today. This figure is consistent
 with the pollen accumulation rates cited in the Interpretation of Results." Most of the important lumber species
 were probably present in lower  numbers, but groves of giant sequoia and incense cedar may have been more
 widespread.

     The  open character of the vegetation might be  less  scenic for tourists, but reduced tourism would be
 unlikely if temperatures  were higher. The relatively cool mountains would be very attractive.  Supplying water
 to the increased tourist population would be more difficult due to reduced precipitation.

     The negative effects of climate change on the western Sierra could be offset by increased precipitation
 and  forest density for the eastern Sierra, but the geography of the mountain range reduces the compensation.
 The gentle western slope occupies far greater area than the eastern slope. A doubling of the area occupied by
 forest on the east slope  produces far less change than corresponding reduction on the west slope.  Abo,  the
 eastern slope is farther from population centers and receives less tourist visitation.

     The goal of this research is to document the effects of elevated temperature on the forests of the western
Sierra Nevada. However, the east-west climatic contrast has important implications for water supply.  Increased
runoff for the east slope could offset the reductions in water supply to urban areas due to greater aridity for the
western slope. The implications are interesting, but the projected increased is based on just one fossil site.
                                               4-36

-------
                                                                                              Davis


                                           CHAPTER 6

                                     POLICY IMPLICATIONS


     Three issues resulting from climatic change in the central Sierra Nevada must be addressed.  Reduced
timber production will result in lost revenue. The reduction appears unavoidable if climate changes during the
next century are in the direction and magnitude of those during the 9000-yr B.P. analog. Decreased precipitation
could result in greater tourist  pressure at  time the ecosystem is less able to withstand it It may become
necessary to regulate the numbers of tourists throughout the Sierra Nevada.  Reduced precipitation in the
western Sierra has implications for reclamation programs. If precipitation increases in  the eastern Sierra, the
greater demand for water west of the Sierra Nevada would increase the value of aqueducts originating east of
the crest
                                                4-37

-------
Davis
                                  LIST OF ABBREVIATIONS
    CCM
    DCA
    Holocene
    NCAR
    ppm
    yrBJP.
    ya
   Community Climate Model
   Detrended Correspondence Analysis
   the last 10,000 years, » postglacial
   National Center for Atmospheric Research
   parts per million
   calculated radiocarbon age Before Present
   calendar years before present
                                       PLANT NAMES
    Abies
    Alnus
    Ambrosia
    Arceuthobium
    Artemisia
    Cercocarpys
    Cheno/Am
    Cupressaceae
    Ericaceae
    Gramii
    Other Compos.
    Pinus
    Ouercus
    Sarcobatuq
    Tsuya
 •fir
 •alder
 • ragweed
 • dwarf mistletoe
 • sagebrush and wormwood
 • chinquapin
 • mountain mahogany
 • chenopod family and amaranth genUS
 • (TCT) juniper, incense cedar, cypress, yew
 • heath family, indnrfing manzanita
 • grass
  undif. members of the sunflower family
  pine
  oak
-hemlock
                                           4-38

-------
                                                                                              Davis
                                          REFERENCES
Adam.DJ'.  1967. Late-Pleistocene and Recent paJynology in the central Sierra Nevada, California,  pp. 275-301
in (E J. Cashing and H.E. Wright, Jr. Eds.) "Quaternary Paleoecology." INQUA voL 7. New Haven, Yale Univ.
Press.

Anderson, R£.  1967.  Late-Quaternary environments of the Sierra Nevada, California.  PhJX dissertation,
University of Arizona, Tucson. 290 p.

Anderson, R.S. and Davis, OJL  1988. Contemporary Pollen Rain across the central Sierra Nevada, California:
RElationship to modern vegetation types.  Arctic and Alpine Research 20(4): in press.

Cole, KX. 1983. Late Pleistocene vegetation of Kings Canyon, Sierra Nevada, California.  Quaternary Research
19:117-129.

Cole, KX. 1985. Past Rates of change, species richness, and a model of vegetational inertia in the Grand Canyon,
Arizona. American Naturalist 125:289-303.

Davis, (XK.  1984. Appendix B: Pollen analysis of Balsam Meadow, Fresno County, California, pp. 325-352 in
(Goldberg, SJL and Moratto, MJ. eds.) Archeological investigations at Balsam Meadow,  Fresno County,
California. Report submitted to Environmental and Regulatory Affairs Division, Southern California Edison Co.,
P.O. Box 800,  Rosemead, California, 91770.

Davis, OJL  1987. Late-Quaternary Environments of the western Sierra: pollen and plant macrofossil analysis
of Dinkey and Exchequer Meadows.  Research Report distributed by Theodoratus Cultural Research Inc, Fair
Oaks, California 95628.

Davis, (XK.  and W.D. Sellers. 1987. Contrasting climatic histories for western North America during the late
glacial and early Holocene. Current Research in the Pleistocene  4:87-89.

Davis, O. IL, J. C Sheppard, and S. Robertson.  1986.  Contrasting climatic histories for the Snake River Plain
result from multiple thermal maxima.  Quaternary Research, 26321-339.

Davis, O. K, R. S. Anderson, P. L. Fall, M. K. OHourke, and R. S. Thompson.  1985.  Palynological evidence
for early Holocene aridity in the southern Sierra Nevada, California.  Quaternary Research 24-322-331

Davis,  O. K. and Moratto, M. J. 1988. Evidence for a warm-dry early Holocene in the western Sierra: pollen
and plant macrofossil analysis of Dinkey and Exchequer Meadows. Madrono 35:128-145.

Deevey, EJSn and Flint, R.F. 1957.  Postglacial hypsithennal interval Science 125:182-284.

Elias, S. A.  1985.  Pateoenvironmental interpretations of Holocene insect fossil assemblages for four high-
altitude sites in the Front Range, Colorado, U.S.A. Arctic Alpine Res. 1731-48.

Hebda, R. J. and R. W. Mathewes.  1984. Holocene history of cedar and native Indian Cultures of the North
American Pacific Coast  Science 255:711-711

H3LM.O.  1979. Decorana, a Fortran program for detrended correspondence analysis and reciprocal averaging.
Ecology and Systematic*, Cornell University, Ithaca, New York.
Jacoosen, G. I*, Jr., and Grimm, E. C 1986. A Numerical analysis of Holocene forest and prairie vegetation
in central Minnesota. Ecology 67358-966.
                                                4-39

-------
 Davis

 Kearney, M. S. and B. H. Luckman. 1983.  Holocene timberline fluctuations in Jasper National Park, Alberta.
 Science 221:261-263.

 Kutzbach, J.E. 1987. Chapter 19, Model simulations of the climatic patterns during the degladation of North
 Amearica. p. 425-446

 Kutzbach, J.E. and Guetter, P J.  1986.  The influence of changing orbital parameters and surface boundary
 conditions on climatic simulations for the past 18,000 years.  Journal of the Atmospheric Sciences 43:1726-1759.

 Maher, L. Jn r. 1987. Palynological zonation and correlation using dissimilarity coefficients - A cautionary tale
 and modest proposal Xn Intern. Congr. Internat Union Quaternary Res. Progr. & Abstr. p.218

 Neftel, An H. Oeschger, J. Schwander, B. Staufler, and R. Zumbnmn. 1982. Ice core sample measurements give
 atmospheric CO2 content during the past 40,000 yr.  Nature (London) 295:220-223.

 Overpeck, J. T., T. Webb ffl, and L C Prentice.  1985. Quantitative interpretations of fossil pollen spectra:
 dissimilarity coefficients and the method of modern analogs. Quaternary Research 23:87-108.

 Ritchie, J. C, L. C Cwynar, and R. W. Spear. 1983. Evidence from north-west Canada for an early Holocene
 Mflankovitch thermal maximum.  Nature (London) 305:126-128.

 Rourke, MJX 1988. The biogeography and ecology  of foxtail pine, Pinus balfouriana (Grev.  & Balf.), in the
 Sierra Nevada of California.  PhJX dissertation, University of Arizona, Tucson.

 Rundel,  P.W.  1972.  Habitat restriction in giant sequoia: The environmental control of grove boundaries.
 American Midland Naturalist 87:81-99.

 Spaukting, W. G. and L. J. Graumlich.  1986. The last pluvial climatic episodes in the deserts of southwestern
 North America.  Nature (London) 320:441-444.

 Vance, R. E. 1985. Pollen stratigraphy of Eaglenest Lake, northeastern Alberta.  Can. J. Earth Set 23:11-20.

 Webb, T. m, Bartlein, P J, and Kutzbach, J.E.  1987. Chapter 20, Climatic change in eastern North America
 during the past 18,000 years; comparisons of pollen data with model  results,  pp.  447-462 in VoL K-3 of The
Geology of North America. Geological  Society of America.

Wood, S Ji.  1975.   Holocene stratigraphy and chronology of mountain meadows, Sierra Nevada, California.
PhJX dissertation, California Institute of Technology, Pasadena, 180 pp.
                                               4-40

-------
    HARD TIMES AHEAD FOR GREAT LAKES FORESTS:
A CLIMATE THRESHOLD MODEL PREDICTS RESPONSES TO
           CO2-INDUCED CLIMATE CHANGE
                   Catherine
                   Margaret B. Davis
                 University of Minnesota
                 Minneapolis, MN 55455
               Contract No. CR-814607-01-0

-------
                                     CONTENTS
FINDINGS 	.	5-1

CHAPTER 1: INTRODUCTION	  5-2

CHAPTER 2: RESPONSES TO CLIMATE • A LrTERATIVE REVIEW	  5-3
      Eastern Hemlock	  5-3
      American Beech  	  5-4
      Yellow Birch	  5-4
      Sugar Maple  	  5-5
      Climatic Thresholds For Hemlock, Beech, Yellow Birch, and Sugar Maple	  5-5

CHAPTER 3: RESULTS OF THE CLIMATE THRESHOLD MODEL	  5-6
      Climatic Variables Used	  5-6
      Climate-Space Under GISS Scenario  	  5-6
      Climate-Space Under GFDL Scenario 	  5-7
      Rates of Dispersal and Colonization	  5-7
      I imitation*  	5-12

CHAPTER 4: IMPLICATIONS  	5-13
      Environmental Implications for the Great Lakes Region	5-D
      Sodoeconomk Implications  	5-D

CHAPTER 5: POLICY IMPLICATIONS 	5-15

REFERENCES	5-16
                                          n

-------
                                                                                           Zabinski


                                            FINDINGS1


        Geographical distributions of four timber trees important in the Great Lakes region have been predicted
assuming CO, doubling by 2090 AJ>n and using climate scenarios developed by the GISS and GFDL models.
The climate threshold model used for these predictions assumes that a tree species will only grow within a
climate-space identical to that within its present geographical range, and that species ran colonize new regions
at the rate of 100 km per century, a rate more than double the maximum ever recorded for temperate trees in
the Quaternary paleorecord. Under GISS-predicted climate, the ranges of sugar maple (freer saccharum). yellow
birch GBeJyja. alleghemensis). hemlock fTsuea canadensisX and beech (Fayus yrandifolia) contract markedly
within the Great Lakes region. The marginal nature of the climate makes it likely that hemlock, yellow birch,
and sugar maple wfll be much less abundant in those parts of Wisconsin and Michigan where they can still grow.
Beech wfll be eliminated completely from the lower peninsula of Michigan, where it is presently abundant. The
geographical ranges of all four species wfll be  much reduced under the GISS scenario, although potentially
suitable habitat wfll be created to the north in central Canada south of James Bay. In succeeding centuries, more
of this region may eventually be colonized.  Under GFDL-predicted climate, all four species wfll be eliminated
entirely from the Great Lakes region, persisting only in Nova Scotia and northern New England.  Larger
potential ranges wfll exist in eastern Canada at the latitude of James Bay.  This region might eventually be
colonized from the southeast if trees can adapt to pbotoperiods at that latitude. Although  the climate threshold
model  considers neither  competition from other species nor soils, it provides the first prediction of forest
response that takes geography and availability of seed source into account
        'Although the information in this report has been funded wholly or partly by the U.S. Environmental
 Protection Agency under Contract No. CR-814607-01-0, it does not necessarily reflect the Agency's views and
 no official endorsement should be inferred from it

                                                 5-1

-------
 Zabinslti


                                             CHAPTERl

                                          INTRODUCTION


         Dominant commercially valuable trees in the hardwood forests of the Great Lakes region are hemlock
 (Tsuga canadensis L.), American beech (Fagus grandifolia Ehrh.X yellow birch fBetuIa allegheniensis Britton),
 and sugar maple (Acer tftf^ttX™1* Marsh.). All four species reach range limits within the Great Lakes region,
 and therefore populations in this region might be  expected to react with sensitivity to climatic change.  The
 question is, are the  climatic changes expected with doubled CO2 of such a nature that they will  affect these
 species, and of such a magnitude that Great Lakes forests will be greatly altered?

         In this report we have assembled evidence from the literature that demonstrates a functional relationship
 between climate and the geographical limits for each species' range. Many of the observations are anecdotal,
 however, and even experimental results cannot readily be translated into quantitative parameters that describe
 the environment at the species limit  Nevertheless, the review was valuable in pointing to the climatic variables
 that are important in reproduction, seedling establishment, and growth of each tree species in various parts of
 its geographical range.  The  climatic  limits were  then quantified through correlation of range limits with
 geographical distributions of climatic variables. Although in several cases there is ample evidence that extreme
 events actually limit the survival of trees, we were forced to correlate with averages  because the General
 Circulation Model output does not include extreme  events. Monthly isotherms do coincide with range limits in
 many cases, however, suggesting that they are statistically correlated, at least regionally, with the extreme events
 that limit tree growth.

         The climatic conditions at a species' limit constitute threshold conditions for its survival. The thresholds
 that exist at the northern, western, and southern limits of yellow birch, hemlock, maple, and beech define a
 climate-space within which each species can grow,  and outside of which it cannot.  Once having defined this
 climate-space, we have inspected the predictions of General Circulation Models for climatic conditions under
 doubled CO2 to determine where similar climate-spaces will be located.  Because each species is limited by a
 different combination of climatic variables, the alteration of potential geographical habitat is different for each
 species.

        We have then taken an additional step to predict where each species will actually be found, given the
 constraints of seed dispersal from existing populations. To do this, it is necessary to specify a time dimension;
 we have  presumed that doubling of CO2 in the atmosphere will be achieved by about 2090 AD. Assuming a
 finite rate of range extension, based on rates of colonization of new regions recorded in pollen deposits during
the Quaternary, we have predicted the areas of potential climate-space that one might reasonably expect to be
occupied by each species by 2090 AJX
                                                5-2

-------
Eastern Hemlock
                                                                                          Zabinski

                                           CHAPTER 2

                      RESPONSES TO CLIMATE - A LrrERATTVE REVIEW
              canadensis (L.) Carr., commonly known as eastern hemlock, is a shallow-rooted species, which
readies reproductive maturity between 20 and 40 years of age when growing under good conditions, and may
live as long as 600 years (Powells, 1965). Hemlock reaches the edge of its westernmost distribution in the Great
Lakes region, and can be found growing in disjunct populations in Minnesota, Wisconsin, and Indiana, 50-70 km
away from the edge of the continuous range. The outlying populations provide clues to the factors important
at the species' range limit. Many of them grow in topographically unique areas, such as canyon walls, rims of
deep ravines (Daubenmire, 1931; Friesner and Potzger, 1931), within steep ravines (Cundiff, 1949; Calcote, 1986),
or on valley slopes (Adams and Loncks, 1971).

       Furthermore, hemlock throughout its entire range is found more commonly on north-facing slopes: in
North Carolina (Costing and Hess, 1956), in Georgia (Bormann and Platt, 1958), in Ohio (Black and Mack,
1976), in New York, and in Ontario (Kavanagh and KeUman, 1986).  North-facing slopes have characteristically
lower air temperature, soil temperature, and vapor pressure deficit than south-facing slopes (Cantlon, 1953).
North-facing slopes also have fewer temperature and moisture extremes, and considerably lower evaporation
(Cooper, 1961). Hemlock also grows in ravines in other parts of its range, such as New York, where hemlock's
shallow rooting system seems to be particularly adapted to steep, narrow ravines without much soil accumulation
(Lewin, 1974).

       In ravines with greater soil accumulation, hemlock grows on mid-slopes while oaks  and white pine
dominate the upper slopes (Lewin, 1974). Laboratory tests showed that photosynthetic efficiency of hemlock was
15% higher at temperatures characteristic of lower valley slopes than at temperatures characteristic of upper
slopes.  In the summer, when sofl moisture reaches  levels  low enough to severely restrict photosynthetic
capabilities, the cooler temperatures that result in higher photosynthetic efficiency become critical to the survival
and growth of hemlock (Adams and Loucks, 1971).

       In a comparison between populations of hemlock at the northern edge of the species' range  and the
center of the species' geographic range, Kavanagh and KeUman (1986) found that there  was little difference
between long-term growth at the northern edge versus the center of the range, but at the northern edge of the
range,  there seemed to be little recruitment after the initial establishment of the populations.  Hemlock made
up fewer than 5%  of saplings at northern sites.  An exception to this observation was at dry sites, where 30%
of the seedlings and saplings present were hemlock.

       Hemlock seeds from the northern edge of the species' range germinate optimally between 12* and 17°C
after 10 weeks of moist stratification (Stearns and Olson, 1958). In the field, seeds germinate and grow best on
moist mineral soil, moss beds, or rotting logs.  Seedlings are very sensitive to water stress; therefore, shaded
conditions reduce  the risk of desiccation (Olson, et ai, 1959).  The cooler, moister habitats of north-facing
slopes and steep ravines seem to be optimal habitat  for hemlock regeneration.

       Kavanagh and KeUman (1986) suggest that the limitation of successful establishment of hemlock to drier
sites at the northern edge of its range could be a result of competition with more mesk species, such as sugar
maple, striped maple, beech, and balsam fir. The cooler, drier sites which favor the establishment of eastern
hemlock could be  the sites where competing species exhibit slower growth rates.

        This same idea was advanced earlier by Daubenmire (1931) as an explanation for the  persistence of
hemlock in Indiana on soils with lower moisture content than the soils of the adjacent beech-maple stands. He
argued that hemlock was limited to the least favorable sites because  of an inability to grow with  superior
competitors on the more favorable sites.


                                                5-3

-------
 Zabinski
         Ability of hemlock seeds to germinate and grow on drier sites may be explained by the existence of two
 physiological races of hemlock in  Wisconsin, one growing in the northeastern corner of the state, which is
 characterized by cooler, moister conditions, and the other in southwestern Wisconsin. Seedings grown from
 seeds collected in southwestern Wisconsin had higher water-use efficiency, and higher absolute CO2 uptake,
 especially at low levels of irradiance (Eickmeier, et aL, 1975).
            eech
         Fapis grandjfolia. the American beech, is a common tree in the eastern United States, where it grows
 from Canada to Florida, and as far west as eastern Wisconsin and Oklahoma and Texas (Powells, 1965). Beech
 trees produce large, mammal- and bird-dispersed seeds. The seeds germinate well either on mineral soil or leaf
 litter.  Beech seedlings are slow-growing relative to other hardwood species, but are shade-tolerant and
 unpalatable to deer (Rushmore, 1961).  The persistence of beech in gaps in the Great Smokey Mountains was
 attributed to  the ability of beech to withstand wind (Russell,  1953).  In Pennsylvania, beech grew under all
 physiographic, soO, and site conditions, and was able to reproduce on dry, upper slopes underneath hemlock
 (Hough and Forbes, 1943).

         Beech is also  known for its ability to produce  root  sprouts,  a capability which contributes to the
 persistence of beech at a given site, but not to its spread to adjacent sites (Jones and Raynal, 1986).  Root
 sprouting appears to be more common in northern  sites than in sites in the central or southern parts of its
 geographic range, and more common at higher elevation sites in  the southern part of its range (Held, 1983).
 Both root injury and high density of superficial roots contribute to the occurrence of root sprouts.

         Studies near the western range limit of beech in Wisconsin suggest that moister and cooler conditions
 in northeastern Wisconsin increase seedling production relative to southeastern Wisconsin (Ward, 1961).  The
 author suggested (Ward, 1956) that climatic warming would result in a decline in beech, especially if warming
 was not accompanied by an increase in moisture. The width of annual rings on beech in Indiana correlates
 positively with total precipitation and  negatively  with temperature during June.  Ring width was  especially
 sensitive to low summer rainfall (DQler, 1535).

         Denton and Barnes (1987), however, found that beech distribution in Michigan was not limited by dry
 conditions. At this more northern site, beech is found where the ratio of July-August precipitation to potential
 evapotranspiration is low. They also found evidence that beech was limited by winter temperatures.  Thus, beech
 might be expected to expand northward given a climate warming.

 Yellow Birch

        The geographic distribution of Betula allegheniensis. yellow birch, parallels hemlock, possibly because
 of similar requirements for seedling germination (Godman and Mattson, 1980). Yellow birch seeds, like hemlock
 seeds, are very small, are shed during the late fall and winter, and germinate well on moist, mineral soil (Curtis,
 1971; Tubbs, 1969).  Establishment of seedlings is irregular and may depend on disturbance, as yellow birch is
 a shade-intolerant species (Stearns, 1951). Young  seedlings can tolerate hot, dry conditions once roots are well
 established, and growth of yellow birch seedlings increases with increasing light levels (Godman and Krefting,
 1960).

        There is also some evidence that growth of adult trees is decreased when more than adequate moisture
is available.  Radial growth of yellow birch on wet  sites and on well-drained sites during wet years was reduced
over that of drier sites or years, possibly as a result  of water-logging in the root system (Fraser, 1956). Seedlings
are also reported to be more vigorous and faster growing on well-drained soil
                                                 5-4

-------
                                                                                            Zabinski

      Maole
             sacchanim. sugar maple, is a shade-tolerant, long-lived species with a high reproductive rate.
Seeds are produced annually with periodic, large seed crops, are shed in the fall, and buried in the leaf litter
(Curtis, 1971). The seeds germinate at lower temperatures than other hardwood species (Godman and Mattson,
1980) so are able to monopolize the resources available in the early spring.  Sugar maple seedlings commonly
form a dense layer of growth on the forest floor (Stearns, 1951); Curtis, 1971) and then experience high mortality
as the cohort ages and resources become limiting

       Sugar maple  is a very competitive species because  of its ability  to survive  under densely shaded
conditions, to grow fast under high light regimes, and to survive heavy deer browsing.


Climatic Thresholds For Hemlock. Beech. Yellow Birch, and Sugar Maole

       Since hemlock seedlings are particularly sensitive to desiccation,  wanner and drier climate could
eliminate seedling recruitment. Adult trees could also be adversely affected as was seen in Menominee County,
Wisconsin, during the drought years of 1930-1937. Menominee County is 50 kilometers northeast of  the
southwestern edge of the continuous distribution of hemlock. Primary cause of death in a number of adult trees
was root death that resulted from a drop in the water table.  Both fungal and borer infection were found on dead
trees, but infection was a secondary cause of mortality, since only trees with a high proportion of dead roots were
infected.  Wind storms, the common form of disturbance in the Great Lakes forests, exacerbate the effects of
drought, as the removal of the overstory trees alters the microclimate of the understory, leaving seedlings and
saplings overexposed to high temperatures and high rates of evaporation (Secrest et aL, 1941).

       The response of beech to a climate warming is less easy to predict If beech is limited by cold winter
temperatures (Denton and Barnes, 1987), then climatic warming could result in northward expansion of beech's
range. Beech's sensitivity to a change in available moisture is less dear.  Although Denton and Barnes (1987)
found a correlation between the presence of beech in Michigan and low moisture availability, Diller (1935)
found that beech growth in Indiana, the western edge of beech's distribution, declines under warmer, drier
conditions. Although beech is reported to grow on dry sites in Pennsylvania (Hough and Forbes, 1943), the dry
sites in Pennsylvania may have higher available moisture than dry sites in the western Great Lakes region.  An
increase in temperature in the Great Lakes region would most likely result in moisture becoming limiting for
beech, as seems  to be the case in Indiana.

        Yellow birch's range limits parallel those of hemlock, providing support for the idea that the two species
have different thresholds for similar  climatic factors. Yellow birch seedling establishment, like that of hemlock,
is very sensitive  to moisture availability, so an increase in temperature without a corresponding increase in
precipitation could result in a contraction from a major portion of the species' present range.  Unlike hemlock,
however, yellow  birch grows as far west as northwestern Minnesota, which has lower annual precipitation.

        Sugar maple is likely to fare better than other species in the Great Lakes region after a climate warming.
Because of the high number of seedlings produced, and the high amounts of mortality necessary to reduce the
population as the cohort ages, there is potential for sugar maple to produce a cohort that is adapted to the
conditions at the time of establishment  There is some preliminary evidence to suggest that cohorts differ
significantly, but the reason for this can only be inferred (Mukahy, 1975). Although sugar maple ranges farther
west than the other species we have considered and has greater tolerance for lower moisture regimes, it too is
limited by moisture and does not grow at the edge of the prairies.
                                                 5-5

-------
 Zabinslti

                                            CHAPTERS

                        RESULTS OF THE CLIMATE THRESHOLD MODEL


 Climatic Variables Used

         The climate threshold model is a simple model which compares species' current geographic distributions
 to temperature and precipitation values to define the  species' climate-space. The assumption is made that
 conditions near a species' range limits represent threshold conditions; within the climate-space defined by these
 boundaries the species can grow, while outside it cannot

         The output generated by the GFDL and GISS General Circulation Models has been used to forecast
 a climate-space for each species under 2xCOj. Climate alone is considered, not soil fertility, dispersal rates, or
 any biotic factors.

         The climate  variables used were mean January temperature, mean July temperature, and  annual
 precipitation. Mean January temperature was used because of the correlation between northern distribution
 limit* for all four species and -15*C mean January temperature. Climate variables are often closely correlated
 with one another, and there may be one or several variables that are important in determining the threshold for
 growth for each species.  Minimum winter temperature is probably a  more  meaningful variable for the
 determination of a species' distribution than mean January temperature, and there is evidence that the killing
 point for sugar maple, beech, and yellow birch is between 40 and 45°C (Sakai and Weiser, 1973; George et aL,
 1974).  Since output of GISS and GFDL does not include temperature extremes, we have used mean January
 temperature instead; its coincidence with northern distribution limits suggest it is correlated with temperature
 extremes.

         Mean July temperature was chosen because of its close correlation with the southern distribution of
 yellow birch and hemlock.  Temperature  is an important factor in regulating many physiological processes, and
 has been correlated with initiation of growth in the spring (Fraser, 1956; Larsson, 1979; Hicks and Chabot, 1985),
 amount of shoot growth in the subsequent year (Kramer and Kozlowslti, 1979; Larsson,  1979),  changes in
 competitive dominance (Woodward, 1975), and failure  of sexual reproduction (Pigott and Huntley,  1978 and
 1981).

        Although this is a very simplistic approach to the interaction of climate and vegetation, we feel that it
 is appropriate for several reasons. Given the scale of the global climate models and the distance between grid
 points,  an analysis more sensitive to climate variation at a specific location would  be inappropriate. Second,
 the output of our model functions best to provide an estimate of potential range distributions for forest species.
 A more detailed analysis of the distribution, incorporating soil types, competitive interactions, and a finer analysis
 of climate variables, would be the next  appropriate  step. Interactions among variables are also important
 Higher  concentrations of CO, win increase the water-use efficiency of plants, possibly reducing sensitivity to
 drought (Morison and Gifford, 1984).

 Climate-Space Under  GISS Scenario

        The  responses of the four  species we have considered within the  Great Lakes region will depend
 significantly on the change  in precipitation that win accompany a temperature increase.  Because the global
 climate  models do not agree on the change in precipitation, we have given a range of possible outcomes. Both
 scenarios suggest, however, that the ranges of all four of these important forest trees win be reduced substantially
within the Great Lakes region.

        Given the output of the GISS model, the climate-spaces available for hemlock, beech, yellow birch, and
sugar maple are reduced in the Great Lakes region, but the species would not be extirpated from the region

-------
                                                                                            Zabinski

(Figures lb-4b). The less drastic changes in ranges, relative to the GFDL scenario, are the result of a smaller
increase in temperature, accompanied by a slight increase in precipitation.  The potential ranges of all  of the
species contract, as their southern limits move from central or southern United States to the Great Lakes region.
Beech, hemlock, and yellow birch would all be reduced in abundance in the Great Lakes region.

        Because of the generous spacing between grid points of the GISS model, predicted climate space is
largely based on extrapolation between points.  Both hemlock and beech in the Great Lakes region would be
limited by annual precipitation. Hemlock populations on wetter sites may survive, but the frequency of periods
with adequate moisture for seedling establishment would probably decrease.  At sites on the western edge of the
range, seedling establishment may be so infrequent as to eliminate hemlock from that area, but there will be a
time lag between climate change and the decrease of the geographic range because of the longevity  of the
species. It is to be expected that relict populations will persist in habitats with unusual microclimate

        Beech would decline in the Great Lakes region, limited by annual precipitation. Sugar maple and yellow
birch both have a more extensive potential range because of their current distribution on drier sites. Under the
GISS  scenario, the  northern limit  of the climate-space for all  four species is not  determined by winter
temperature, but by cool  summer temperatures and low annual precipitation in northeastern Canada.

nimale-Space Under GFDL Scenario

        Given the GFDL scenario, all four tree species will become extinct in the Great Lakes region (Figures
lc-4c).  The  predicted increase in summer temperatures in the Great Lakes region - 7-10 C - exceeds the
climatic tolerances of any of the species under any moisture regime. One would expect not only  failure of
seedling establishment  of the species, but mortality of adult trees also. The predicted climate-space for all of
the species (Figures  lc-4c) moves northeastward into Canada; the climate-spaces for sugar maple and yellow
birch extend  farther west than for hemlock or beech because  of their current distribution in drier areas. The
western and  southern  distributions of hemlock and beech are limited by summer temperature  and annual
precipitation. Within their current distribution, nowhere except Nova Scotia has a climate analogous to climates
predicted by  the GFDL model

Rates of Dispersal and, Cf?i9IlB?lti?n

        The  paleorecord provides  dear evidence  that  the  response of plants  to climate  change  is not
instantaneous. Trees grow slowly and take many years to reach maturity. Newly established seedlings of species
that may be well-adapted to changed climate may take many years to reach the canopy, owing to shading from
the resident canopy dominants (Davis and Botltin, 1985). When large geographic regions are considered,  the
limited distances of seed dispersal cause lags in the adjustments of species ranges to climate (Davis, 1986; Davis
et aL, 1986; MacDonald and Ritchie, 1986). In the past, tree species populations have expanded into climatically
favorable  regions at rates averaging 10-40  km per century (Davis,  1981).  In northwestern Canada  spruce
advanced northward at a record rate of 200 km per  century, a dispersal rate that is explained by winds blowing
northwestward along the  edge of the retreating ice sheet (MacDonald and Ritchie, 1986).

        Because  climatic changes resulting from doubled CO2 will occur with unprecedented speed, it is
important to consider  whether dispersal may limit biotic response. Doubling of CO2 or the equivalent in
radiativeh/ active gases is expected to occur by 2090 AJ>. at least We have therefore considered the possible
expansions of species ranges that could occur within 100 years. To display these expansions on maps, we have
drawn the predicted species range at 2xCO2 to include that part of  the present species range that falls within
the climate-space projected under a given scenario, plus an extension of 100 km into the unoccupied climate
space. This is a generous estimate of the possible range extension that might occur without human intervention,
as it is more than double the fastest known range extension for the species in question. (If the extension rate
has been doubled to 200 km, the resulting enlargement of range would be insignificant at the scale of the maps
shown in Figures 1-4).
                                                 5-7

-------
   Hemlock
                  Figure 1.     Present and future geographical range for hemlock. A:  Present range (modified from Powells,
                               1986). B: 2xCOj climate-space in 2090 A.D. under the GISS scenario. Black area is the predicted
                               species range; stippled area is potential range. C:  2xCO, climate-space in 2090 A.D. under the
                               GFDL scenario.  Black area is the predicted species range, stippled area is potential range.
13
.1
a

-------
Beech
                Figure 2.     Present and future geographical range for beech. A:  Present range (from Powells, 1965).  B:
                             2xCO2 climate-space in 2090 A.D. under the GISS scenario.  Black area is the predicted species
                             range; stippled area is potential range. C: 2xCO2 climate-space in 2090 A.D. under the GFDL
                             scenario.  Black area is the predicted species range, stippled area is potential range.

-------
Yellow Birch
                     Figure 3.     Present and future geographical range for yellow birch.  A: Present range (from Powells, 1965).
                                  B:  2xCO2 climate-space in 2090 A.D. under the GISS scenario.  Black area is the predicted
                                  species range; stippled area is potential range. O 2xCO2 climate-space in 2090 A.D. under the
                                  GFDL scenario.  Black area is the predicted species range, stippled area is potential range.

-------
Sugar Maple
                   Figure 4.     Present and future geographical range for sugar maple. A:  Present range (from Powells, 1965).
                                B:  2xCO2 climate-space in 2090 A.D. under  the GISS scenario.  Black area is the predicted
                                species range; stippled area is potential range.  C: 2xCO2 climate-space in 2090 A.D. under the
                                GFDL scenario.  Black area is the predicted species range, stippled area is potential range.

-------
 Zabinslti


        The differences between the predicted range and the predicted climate-space for each species (Figures
 1-4, b and c) is very large in all cases, indicating that dispersal of seeds and establishment of seedlings will be
 important factors limiting tree distributions in the coming century of rapid climatic change. Means for artificial
 seed dispersal into climatically suitable regions should be considered. It will be necessary, of course, to develop
 ecotypes that are appropriate for growth at the  high latitudes of the predicted climate-space regions indicated
 in Figures 1-4, b and c).
       There is a definite gap in the published literature on whole plant responses to climate and climate
change, and on factors determining the range limits of plant species.  We need more whole-plant studies along
the lines  of  those by Bordeau (1952) and Woodward  and Pigott  (1975).   Correlating climatic  isolines to
geographic distributions, and using statistical analysis to determine which factors may limit a species' distribution,
are good first steps for generating hypotheses. They in no way compensate, however, for the lack of experiments
that demonstrate how whole plants react singly, and in competition with other species, to a change in climate.
More needs to be known concerning the direct effects of CO2 on plants growing in natural ecosystems, and the
interactions of direct CO, effects and climate  on plant response (Bazzaz and Carbon, 1984; Canham and
McCavish, 1981; Conroy et aL, 1986; Funsch et aL, 1970; Higginbotham et aL, 1985; Williams et aL, 1986; Zangerl
and Bazzaz, 1984).

       We need to know more about physiological races within species. For example, there is a suggestion that
preference by some populations of yellow birch for wet sites in Wisconsin is the result of introgression of yellow
birch with bog birch Betula yladulifera (Curtis, 1971). The suggestion of two physiological races of hemlock in
Wisconsin, one that grows on dry sites and the other on moist sites, needs to be further investigated.  It might
be possible, for example, to manage forests for anticipated climate changes by moving ecotypes into areas where
they win have some time  to become established before a climate change occurs.
                                                5-12

-------
                                                                                           Zabinski
                                           CHAPTER 4

                                          IMPLICATIONS
                    ations for the Great Lak
       Major reductions in abundances are predicted using the GISS scenario, and loss of four major forest
species from the region is predicted given the GFDL scenario.  These changes will affect processes of soil
formation and will change the productivity of forested lands. Northern hardwood species will be replaced by
other trees characteristic of more southern latitudes, or by open prairie or scrubland.  Although the species in
question will not suffer extinction, ecotypes adapted to continental climate will be lost  Even under the GISS
scenario, northern hardwood forest communities will be much reduced in area.

       A major loss in biodiversity is to be expected under the GFDL scenario. Extinctions are to be expected
among spring ephemerals and other herbaceous species and fungi adapted to the moist, shaded conditions that
characterize the forest floor under northern hardwood and beech-maple forests.  These plants lack dispersal
capabilities sufficient to colonize the predicted hardwood region hundreds of kilometers to the northeast. Under
the GISS scenario, many species in the western and southern parts of the Great Lakes region will be endangered
because the hardwood forests in these areas wDl change greatly and contract in area. The marginal climate will
stress the trees,  resulting in increased mortality from insects and  pathogens.  Beech will be greatly reduced.  The
seeds of this species provide food for many birds and small mammal species.

       Woodland animals wfll also be endangered, especially under the  GFDL scenario.   However, deer
populations may be favored by increased early successional vegetation as weakened adult trees are knocked over
by wind or die as the result of attacks from insects or pathogens. Successional species such as trembling aspen
and paper birch provide fodder, and the deer population can be  expected to increase in the same manner it did
following logging around the turn of the century. Fires are likely  to be more common under the wanner climates
predicted by both models, although the heavy rainfall predicted by GISS may cut down the incidence somewhat.
Hydrology is unlikely to be greatly affected by the loss of these four tree species alone.

Socioeconomic  Implications

       Under the GISS scenario, forestry win be hard-hit Beech wfll be eliminated from lower Michigan, and
yellow birch and hemlock wfll contract from regions where they grew well prior to CO- doubling. Maple will
be marginal in the southern Great Lakes region, although it will still grow in northern Wisconsin and Minnesota,
and throughout Michigan. Given the GFDL scenario of climate change, the hardwood logging industry of the
Great Lakes region will be eliminated completely. Hardwood forests will exist only in Nova Scotia and northern
New England, regions with indigenous forestry industries.

       At present, most hardwood forests in the Great Lakes  region are not intensively managed. In future
decades, management will become necessary. Although sugar maple trees will still grow in the region, seeds will
probably have to be brought in from the south where varieties are adapted to warm climate.  Breeding will be
necessary to perfect varieties that withstand heat but have photoperiod sensitivities appropriate for the north.
Protection from wildfire, which may occur  more frequently with  wanner climate, will be necessary for these fire-
sensitive species.  As the climate changes and trees cease to grow rapidly and begin to show signs of stress,
salvage logging operations will be advised. Similar salvage operations will have become necessary some years
previously in states to the south; extensive salvage logging may depress market prices for hardwood, make  it
unlikely that salvaging wfll be highly profitable in the Great Lakes region.
                                                5-13

-------
       Methods will have to be developed for broadcasting seeds or seedlings ahead of the northward-moving
front for all four species in order to colonize the available climate-space more rapidly than will occur naturally.
Given  the sensitivity of seedlings of both hemlock and yellow birch, broadcasting seedlings may prove more
successful than seeds.  Two factors should be investigated before such procedures are widely applied:  1) the
suitability of soils in what is now a boreal forest region, and 2) the sensitivity of all four species to photoperiods
at the northern latitudes where seeding is to be attempted (Vaartaja, 1959).
                                                 5-14

-------
                                                                                            Zabinski
                                            CHAPTERS

                                      POUCYIMPUCATIONS


       Policies must be implemented to lessen the economic impact of CO, wanning on regions where forestry
is the main source of livelihood. Salvage logging win provide plentiful short-term jobs, but the long-term outlook
for a profitable industry is poor. Research is needed on the kinds of trees that can be grown in the Great Lakes
region under changed climate; development of varieties suited to northern latitudes should begin immediately.
Methods for introducing economically valuable species into regions where natural seeding is not occurring also
need development To preserve water quality, careful watch should be made on ecosystems where forest dieback
is occurring.

       The loss of biodiversity has large implications for social and esthetic values. Park managers and the
tourism industry have much at stake. Loss of biodiversity win have an economic impact as welL  Even for
species that are not completely lost, valuable ecotypes will become extinct Maps showing the response to
GFDL-predicted climate demonstrate that all the ecotypes of these four species of trees that are adapted to
continental, relatively xeric conditions could be lost from the species'  populations.  For colonization of the
climate-space in the mid-continent, entirely new genetic stock wfll have to evolve from die relict populations in
Nova Scotia and northern New England.  Evolutionary changes of this land are  slow within long-lived species;
for this reason colonization of the available climate-space might not occur given natural means for thousands of
years. In fact, natural dispersal into such a large region would take many centuries, if not several millennia.
                                                 5-15

-------
 Zabinslti

                                           REFERENCES


 Adams, MS., and (XL. Loucks. 197L Summer air temperatures as a factor affecting net photosynthesis and
 distribution of eastern hemlock (Tsuga canadensis L. (Carriere)) in southwestern Wisconsin. American Midland
 Naturalist 85:1-10.
 Bazzaz, FJL, and R.W. Carlson. 1984. The response of plants to elevated COy  L Competition among an
 assemblage of annuals at two levels of sofl moisture.  Oecologia. 62:196-198.

 Black,  RJL,  and RM.  Mack.   1976.  Tsuya <-anaHi»ngi«  in  Ohio:   Synecological and photogeographical
 relationships. Vegetatio.  32:11-19.

 Bordeau. P. 1954. Oak seedling ecology determining segregation of species in Piedmont oak-hickory forests.
 Ecology Monographs.  24:297-320.

 Bormann, FJL, and R.B. Platt. 1958. A disjunct stand of hemlock in the Georgia Piedmont  Ecology.  39:16-
 23.

 Bradford, KJ., and T.C Hsiao.  1983. Physiological responses to moderate water stress,  p. 263-324.  In: OX.
 Lange, PS. Nobel, GB. Osmond, and H. Ziegler. Encyclopedia of Plant Physiology, VoL 12B. Physiological
 Plant Ecology. IL Water Relations and Carbon Assimilation. Springer- Verlag, Berlin.

 Braun, EX. 1950. Deciduous forests of eastern North America. Blakiston Company, Philadelphia.

 Calcote, R.  1986. Hemlock  in Minnesota:  A rare species for 1,200 years.  Master's  thesis, University of
 Minnesota.

 Canham, A£n and W J. McCavish.  198L Some effects of CO«, daylength and nutrition on the growth of young
 forest tree plants. L  In the seedling stage. Forestry. 54:169-182.

 Cantlon, JJL  1953. Vegetation and microclimates on north and south slopes of Cushetunk Mountain, New
 Jersey.  Ecology Monographs.  23:241-270.

 Conroy, Jn E.WJL Barlow, and D.L Bevege.  1986.  Response of Pinus radiata seedlings to carbon dioxide
 enrichment at different levels  of water  and phosphorus: Growth, morphology,  and anatomy.  Ann. Botany.
 57:165-177.

 Cooper, A.W.   196L  Relationships between plant life-forms  and microclimate in southeastern Michigan.
 Ecology Monographs.  31:31-59.

 Cundiff, MJ7.  1949. A study in soil moisture, acidity, and evaporation in an upland woods at Turkey Run State
 Park. Butler University Botanical Studies. 9:108-123.

 Curtis, J.T.  197L The vegetation of Wisconsin.  University of Wisconsin Press,  Madison, Wisconsin.

 Daubenmire, RJ7. 193L  Factors favoring the persistence of a relic association of eastern hemlock in Tndiana.
 Butler University Botanical Studies.  229-32.

 Davis, M.B.  198L Quaternary history and the stability of forest communities,  pp. 132-153.  Iff "Forest
Succession," eo\, D.C. West, HJL Shugart, and D.B. Botkin, Springer- Verlag, New York.

Davis, M.B. 1986. Climatic instability, time lags, and community disequilibrium,  pp. 269-284. Iff J. Diamond
and T J. Case, eds^ Community Ecology.  Harper & Row, New York.


                                                5-16

-------
                                                                                          Zabinslti

Davis, MB., and DB. Botltin.  1985. Sensitivity of cool-temperate forests and their fossil pollen record to rapid
temperature change.  Quaternary Research. 23327-340.

Davis et aL  1986. Climate or dispersal as factors limiting the Holocene range extension of beech and hemlock
into the Great Lakes region. Vegetation. 67:65-74.

Denton,S.R. 1985.  Ecological climatic regions and tree distributions in Michigan. PhD. Thesis, University of
Michigan.

Denton, SJL,  and B.V. Barnes.  1987. Tree species distributions related to climatic patterns in Michigan.
Canadian Journal of Forestry Research. 17:613-629.

Diller, CXD. 1935. The relation of temperature and precipitation to the growth of beech in Northern Indiana.
Ecology. 16:72-81.

Eickmeier, W., M. Adams, and D. Lester.  1975. Two physiological races of Tsuea canadgmia in Wisconsin.
Canadian Journal of Botany. 53:940-95L

Powells, HA.  1965. Sflvics of forest trees of the United States.  U.S. Department of Agriculture. Agriculture
Handbook No. 271.

Fraser, DA 1956. Ecological studies of forest trees at Chalk River, Ontario, Canada.  H. Ecological conditions
and radial increments. Ecology.  37:777-789.

Friesner,R.G,andJ.E.Potzger.  193L  Studies in forest ecology. I. Factors concerned in hemlock reproduction
in Indiana. Butler University Botanical Studies.  2:29-32.

Funsch, R.W, RJL Mattson, and G-R. Mowry. 1970.  CO, supplemented atmosphere increases growth of Pjnus
strobus seedlings. Forestry Science.  16:459-460.

Gates,  DJVL  1983.  An overview, p.  7-20. ftp CO, and Plants: the response of plants to rising levels of
atmospheric carbon dioxide, EJL Lemon, ed. Westvww Press, Boulder, Colorado.

George, MJ7., MHJ. Burke, H.M. PeUett, and A.G. Johnson.  1974. Low temperature exotherms and wood
plant distribution. Horticultural Science. 9319-522.

Godman, R-M, and L.W. Krefting.  1960.  Factors important to yellow birch establishment in upper Michigan.
Ecology. 41:18-28.

Godman and Mattson. 1980.  Northern Hardwood Notes, 3.03. North Central Forest Experiment Station.

Grace, J. 1987.  Climatic tolerance and the distribution of plants.  New Phytologist  1065:113-130.

Held, M.E. 1983.  Pattern of beech regeneration in the east-central United States.  Bulletin of the Torrey
Botanical Club.  11O.55-62.

Hicks, DJn and B J7. Chabot 1985.  Deciduous forest  p. 257-277.  Iff BJ7. Chabot and HA. Mooney, eds.
Physiological ecology of North American plant communities.  Chapman and Hall, New York.

Higginbothani,K.OMJJVl. Ma)^S.LlIirodeUe,andDJCKrvstofiak. 1985. Physiological ecology of lodgepole
pine in an enriched CO, environment Canadian Journal of Forestry Research.  15:417-421.

Hough, AJ7., and RJ>. Forbes.  1943.  The ecology and sirvics of forests in the high plateaus of Pennsylvania.
Ecology Monographs. 13:299-320.


                                                5-17

-------
  Zabinslti

  Jones, RJi, and D Jl. RaynaL 1986.  Spatial distribution and development of root sprouts in FJBJS erandifolia
  (Fagaceae). American Journal of Botany.  73:1723-1731.

  Kavanagh, 1C, and M. Kellman.  1986.  Performance of Tsuea canadensfc (L.) Carriere at the  center and
  northern edge of its range: a comparison.  Journal of Biogeography.  13:145-157.

  Kramer, PJ.  1981.  Carbon dioxide concentration, photosynthesis, and dry matter production.  Bioscience.
  31:29-33.

  Kramer, P J.  1983. Water relations of plants. Academic Press, Inc. Orlando, Florida.

  Kramer, PJ., and T.T. KozlowskL 1979.  Physiology of woody plants. Academic Press, Inc.  Orlando, Florida

  Larsson, S.  1979.  Climate  as growth regulating factor in trees.  L A review of the literature.  Swedish
  Coniferous Forest Project, Technical Report 22.  Swedish University of Agricultural Sciences, Uppsala.

  Lewin, D.C  1974. The vegetation of the ravines of the southern Finger Lakes, New York region. American
  Midi Nat 91315-342.

  MacDonald, G.M., and J.L. Ritchie.  1986.  The patterns of post-glacial spread  of white spruce.  Journal of
  Biogeography. 13:527-540.

  Melancon, Sn and MJ. Lechowkz.  1987. Differences in the damage caused by glaze ice on codominant Acer
  sacrharum and Fayus erandifolia.  CamMsm Journal of Botany.  65:1157-1159.

  Morison, JH-, and RM. Gilford.  1984. Plant growth and water use with limited water supply in high CO2
  concentrations.   IL Plant dry weight, partitioning  and water-use efficiency.   Australian  Journal of Plant
  Physiology. 1L375-384.

 Mulcahy, Di.  1975.  Differential mortality among cohorts in a population of Acer saccharum (Aceraceae)
 seedlings.  American Journal of Botany. 62:422-426.

 Olson, JJS, F. W. Stearns, and H. Nienstaedt  1959. Eastern hemlock seeds  and seedlings:  response to
 photoperiod and temperature. Connecticut Agricultural Experiment Station. Bulletin 620.

 Costing, HJ., and D.W. Hess. 1956. Microclimate and a relic stand of Tsuga canadensis in the lower piedmont
 of North Carolina.  Ecology.  37:28-39.

 Pigott, CD., and JJP. Huntley. 1981.  Factors controlling the distribution of Tilia cordata at the northern limits
 of its geographical range.  New Phytologist. 87:817-839.

 Pigott, CD, and JJ». Huntley. 1978.  Factors controlling the distribution of Tilia cordata at the northern limits
 of its geographical range.  New Phytologist  81:429-441.

 Rogers, HJL, If. Thomas, and G.E. Bingham. 1983. Response of agronomic and forest species to elevated
 atmospheric carbon dioxide. Science.  220:428-429.

 Rushmore, RM. 1981. Sflvical characteristics of beech (Fagus eradifolia). USD A Forest Service, Northeastern
 Forest Experiment Station, Station Paper 161.

 Russell, NJi.  1953. The beech gaps of the Great Smokey Mountains. Ecology.  34-366-374.

Sakai, A^ and C J. Weiser.  1973. Freezing resistance of trees in North America with reference to tree regions.
Ecology. 54:118-126.


                                                5-18

-------
                                                                                         Zabinslti

Salisbury, F.B., and C.W. Ross. 1985. Plant Physiology. Wadsworth Publishing Company. Belmont, California.

Secrest, H.C, HJ. MacAloney, and R.C Lorenz. 1941.  Causes of decadence of hemlock at the Menominee
Indian Reservation, Wisconsin. Journal of Forestry. 39*3-12.

Stalter, R. 1981 Production of viable seeds by the American beech (Faeus eradifolia).  bulletin of the Torrey
Botanical Club. 109-342-544.                                  ^^ ***»***,

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

Steams, F.W, and J. Olson.  1958.  Interactions of photoperiod and temperature affecting seed germination in
Tsuea ffliiKteniTfc American Journal of Botany.  45:53-58.

Strain, BJL, and SIX Smith. 1985.  Response of Great Basin plants to atmospheric CO. enrichment. American
journal of Botany. 72:866.

Tinus,R.W.  1971 CO, enriched atmosphere speeds growth of ponderosa pine and blue spruce seedlings. Tree
Planter's Notes. 23:12-15.

ToUey, I*, and B.R. Strain.  1984a.  Effects of CO, enrichment on growth of Ui11tfain1?
-------
POTENTIAL EFFECTS OF CLIMATE CHANGE ON US. FORESTS:
    CASE STUDIES OF CALIFORNIA AND THE SOUTHEAST
                     James N. Woodman
                     Can Sasser Furiness
             Atmospheric Impacts Research Program
                 North Carolina State University
                     Raleigh, NC 27606
                    Contract No. 68-03-3439

-------
                                   CONTENTS


                                                                           Page

FINDINGS	6-1

CHAPTER 1: CLIMATE EFFECTS ON FOREST ECOSYSTEMS	  6-3
    INTRODUCTION	  6-3
    THE NATURE OF CLIMATE	                           6-3
    ESTIMATING CLIMATE CHANGE	  6-5
    EFFECTS OF HIGHER CO2 AND CLIMATE CHANGE ON FORESTS 	  6-5
       Physical Climate Effects	  6-6
       Carbon Dioxide Effects	  6-6
       Climate-Carbon Dioxide Climate
        Interactions	  6-7
       Ecosystem Effects	  6-7
       Forest Productivity and Species Composition  	  6-8
    OTHER EFFECTS ON FORESTS	  6-9
       Insects and Diseases  	  6-9
       Wildfires	  6-9
    SUMMARY AND CONCLUSIONS	  6-9

CHAPTERS FOREST-RELATED SOCIOECONOMIC SYSTEMS AND POLICIES 	6-11
    INSTITUTIONS AND POLICIES AFFECTING FOREST USE AND MANAGEMENT	6-11
       Non-Industrial Private Forestland (NIPF)	6-11
       Forest Industry-Owned Forestland	6-11
       Public Forestland	6-13
    POTENTIAL EFFECTS OF CLIMATE CHANGE ON FOREST USE AND MANAGEMENT . 6-13

CHAPTER 3: POTENTIAL CLIMATE CHANGE EFFECTS ON CALIFORNIA FORESTS 	6-15
    INTRODUCTION	6-15
       Geography and Climate	6-15
    FOREST RESOURCES	6-15
       Major Forest Types	6-15
       Timber Production 	6-16
       Water	6-16
       Recreation  	6-17
       Forage 	6-17
       Wfldlife	6-17
       Factors Affecting Forest Health and Productivity  	6-17
       Trends in Forest Management	6-18
    SOCIOECONOMIC TRENDS AFFECTING FORESTS	6-18
    PROJECTED CLIMATE-CHANGE SCENARIOS	6-19
    POTENTIAL IMPACTS ON SPECIES COMPOSITION AND FOREST PRODUCTrVTTY	6-21
       Species Composition  	6-21
       Forest Productivity	6-21
       Potential Impacts on Forest-Based Sodoeconomk Systems	6-22
    CONCLUSIONS	6-22
                                       u

-------
                               CONTENTS (continued)

                                                                            Page

CHAPTER 4: POTENTIAL CLIMATE CHANGE EFFECTS ON SOUTHEASTERN FORESTS ... 6-24
    INTRODUCTION	6-24
        Geography and Climate	6-24
    FOREST RESOURCES	6-24
        Major Forest Types	6-24
        Timber Production 	6-26
        Water	6-26
        Recreation 	6-26
        Wildlife	6-27
        Factors Affecting Forest Health and Productivity  	6-27
        Trends in Forest Management	6-28
    SOCIOECONOMIC TRENDS AFFECTING FORESTS	6-29
    PROJECTED CLIMATE CHANGE SCENARIOS	6-30
    POTENTIAL IMPACTS ON SPECIES COMPOSITION AND FOREST PRODUCTIVITY	6-30
        Species Composition	6-30
        Forest Productivity	6-31
        Associated Effects	6-31
        Potential Impacts On Sorioeconomic Systems and Policies 	6-32
    CONCLUSIONS	6-33

CHAPTER 5: POTENTIAL POLICY CONCERNS	6-34
    KEY ASSUMPTIONS  	6-34
    POLICY ISSUES	6-34
    RESEARCH NEEDS	6-35
    CONCLUSIONS	6-36

REFERENCES	6-37
                                       ill

-------
                                                                                           Woodman


                                            FINDINGS1


        The purpose of this review was to assess the potential effects of higher CO, concentrations and climate
change on forests and their uses in the United States.  Three hypothetical scenarios of potential future climate
were applied to published information to evaluate how a different climate and higher CO- concentrations might
affect forest productivity and species composition in California and the Southeast. These trends were compared
with socioeconomic trends in these states to assess potential economic effects and public policy issues.

        A fundamental question about effects of climate change on forests is what will happen to them when
the historic equilibrium climate conditions under which they have evolved greatly change in a very short period
of time.  Today's tree species in natural forest ecosystems evolved over thousands of years in response to
relatively stable climate conditions.  Historically, the greatest changes were in response to abnormal weather
events, e^, spring droughts, rather than to long-term averages. The climate change which is projected over the
next century win have more abnormal weather events. The ability of species to adapt to the new climate will
depend on many factors.  Some of the most important are the rate and magnitude of climate change, inherent
ability of tree species to grow  and compete with other species in  a  different  microclimate, and the degree to
which higher ambient CO2 concentrations can offset  negative effects of higher temperatures and droughts.

        Some of the tree species and populations most susceptible to climate change are peripheral populations
in ecotones near the edges of species ranges, species with limited genetic diversity, specialized species which can
only exist within a narrow range  of conditions, poor seed dlspersers, and some cold-climate species.

        Most of the  potential effects of climate  change will lag many years behind the temperature and
precipitation changes presumed in this analysis. Impacts will be difficult to detect, as will ability to ascertain
whether they were caused by a real change in climate or natural variability in the present climate.  In general,
most  low-elevation forests in the southern half of the  VS. will probably experience reductions in biomass and
species  diversity after 100+ years. Higher elevation and more northern forests could become more productive
and increase in species diversity.

        The risk of tree  damage and mortality due to insects, diseases, and  wildfires is expected to increase.
Warmer temperatures and wetter conditions will generally increase the severity of and potential losses from many
insects and diseases. Drier conditions would increase the  risk of wildfire losses.

        The potential impact of climate change on the policies employed by society to manage and regulate U.S.
forests will depend on many factors. The impact will depend in a large part on the economic and noneconomic
values that society places on the various commodities and uses of forests and to what extent future climate
change impacts those benefits. The dominant factor now  determining the uses and benefits are  the goals and
management policies of the forest owners.

        In California, climate change would most likely decrease the productivity and species diversity in many
low-elevation forests.  Many higher elevation forests could become more productive and increase in species
numbers. Higher winter temperatures would reduce mountain snow packs and  summer stream flows. Tree
mortality will probably increase  owing to more wildfires,  larger insect  populations,  and increases in diseases.
Foresters wfll place more emphasis on managing forests for water production, recreation, and reducing losses
from  forest pests and wildfires.   Anticipated decreases in forest  productivity will most likely intensify public
pressure to legislate more multiple-use  practices on private forestlands. A reduced presence of the forest
industry is also expected.
        'Although the information in this report has been funded wholly or partly by the U.S. Environmental
Protection Agency under Contract No. 68-03-3439,  it odes not necessarily reflect the Agency's views and no
official endorsement should be inferred from it

                                                 6-1

-------
Woodman


        Most southern forests will probably have some decrease in productivity and species diversity over many
centuries of climate change.  Some upland hardwood forests may become more productive.  Climate change will
probably favor hardwood forest types more  than types containing the commercially important southern pine
species. Anticipated trends in reduced softwood production and loss of productive forestland to agriculture will
probably accelerate. Forest  industries dependent on low-cost softwood will face increasing economic pressures
to modernize their manufacturing facilities, increase softwood supplies from their own lands, or move to other
regions with more favorable wood supplies and economic conditions.  A potentially major loss of forestland to
other uses could result in greater regulation of forest practices on private lands.

        Public policy implications will vary by region because potential impacts will not be the same across the
country.  Socioeconomic impacts will depend on forest ownership. Federal forestlands have a legislated policy
which can rapidly respond to socioeconomic needs on a regional basis.  Private forest owners will be most
responsive to impacts on financial benefits. State and local government policymakers are most likely to legislate
new policy or regulations concerning forest effects when forest-based tax revenues and socially important values
derived from forests need protection.

        Some of the most important research needs are 1) more precise regional-scale climate models which
can provide more biologically useful information for  trees and forests; 2) tree-,  forest-, and ecosystem-level
models for major species that are sensitive to changes in CO2 concentrations and microclimate variables; and
3) more information on effects of long-term elevated CO2 concentrations on seedlings and large trees in forests,
especially differences between species.
                                                6-2

-------
                                                                                         Woodman
                                           CHAPTER 1

                         CLIMATE EFFECTS ON FOREST ECOSYSTEMS
INTRODUCTION
     Atmospheric concentrations of carbon dioxide (CO2) have increased 20% since 1960 and are expected to
double pre-industrial levels by the third quarter of the next century. The increase has been attributed to the
burning of fossil fuels, worldwide deforestation, and changes in land use (NAS, 1983). CO, and other radiatively
active or "greenhouse" gases in the troposphere affect the global heat balance of the earih (NAS, 1975b, 1979,
1983; Hansen et aL, 1981). Elevated levels of these gases are expected to increase the average temperature of
the earth's surface by L5* to 4.5°C and alter patterns of precipitation, relative humidity, cloudiness, and related
climatic factors in different regions. Although there is no consensus among climatologists that the earth Is now
wanner because of an increase in "greenhouse gases," there is consensus that the earth should become wanner
in the future. There remains considerable uncertainty about when, where, by how much, and how soon climate
will change. Assuming that climate does change, there is equal uncertainty about  the overall effects of those
changes on many aspects of life.

     The purpose of this review is to assess the  potential effects of higher CO- concentrations and climate
change on forests and their uses in the United States. In order to detect some ofthe most likely trends, three
hypothetical scenarios of future climate were applied to what is known about climate and CO-influences on
forest productivity and forestry related sodoeconomic trends in California and the Southeast These results,
identified information needs, and some potential public policy concerns are described below.


THE NATURE OF CLIMATE

      "Climate" is defined as the sum total of all individual meteorological occurrences, or weather processes,
over a period of years in a given place (Geiger, 1965). Climate includes average weather conditions, regular
sequences of weather, and repeatedly observed special phenomena such as tornadoes and late frosts.

      One of the most important attributes of climate that influences all organisms is the variation of weather
conditions over time and space.  Light, temperature, and atmospheric moisture conditions are rarely the same
for the same dates or months in each year.  There are a number of climate anomalies within the context of
interannual variability.  For example, Figures 1A and IB show the average March and April temperatures and
precipitation values for North Carolina from 1887  to 1982 (Karl et aL, 1983).  The two horizontal lines labeled
5th percentfle and 95th percentfle delineate  the values that were in the lowest  and highest 5th percentiles of
values. Although the average temperature for this  96-year period was 12.6°C, actual values deviated from 10.1°
to 16.4".  Similarly, annual mean precipitation values varied 100 mm around the long-term mean of 1% mm.
Only three of the annual temperatures equaled the long-term average. Only five  annual precipitation values
were within 5 mm of the  long-term average.  The  years where values are below the 5th or above the 95th
percentfle lines are considered "extreme events."

      The variation in climate factors such as precipitation and temperature is more important to understanding
species composition, plant succession, and biomass productivity of plant ecosystems than long-term averages.
Ecosystems exist in a state of "dynamic equilibrium" with their climates, and plants are constantly striving to
respond  to changes in temperature, precipitation, etc They often respond to  normal annual cycles of these
values in a fairly predictable manner until extreme events, e^, unusual droughts or frosts in the spring, cause
unexpected departures from the normal equilibrium condition. One of the most critical questions about future
global climate change concerns how these dynamic aspects will change.  Will the frequency of abnormal events
that injure, weaken, or kill plants increase or decrease from the historical climate?
                                                6-3

-------
  Woodman
                  .   A
         Ul
         oc



         B
         Ul
         a

         UJ
18


17


16


15


14


13


12


11


10


 9


 8
 1880   1890   1900  1910   1920   1930   1940   1950   1960   1970   1980   1990   2000


                                     YEAR
                                                                                     95th%
Mean
                                                                                     5th%
             350
             300
         3  250
             200
         O
         m
         OC
             100
                                                                                     95lh%
                                                                       Mam
                                                                                      5lh%
              50 •—*•
               1880   1890   1900   1910    1920   1930  1940   1950   1960   1970  1980   1990   2000

                                                   YEAR
Figure L   Average temperature (A) and precipitation (B) for North Carolina for March and April from 1887
           to 1982 (Karl et aL, 1983).
                                                 6-4

-------
                                                                                         Woodman

     The factors of climate that affect plants are usually different from those depicted in historical climate data
(Geiger,  1965).  Most weather data used in climate averages are taken from stations located outside the
influences of plants. Data from all stations in a region are averaged to characterize that region's "macroclimate."
The meteorological conditions within 2 meters of the ground or within a plant canopy are  called microclimate
or plant climate (Geiger, 1965). Plants create their microclimate by altering solar energy flux, wind movement,
and precipitation reaching the soil surface. They modify air and soil temperatures by increasing humidity through
transpiration, shading, and lowering wind speeds. Summer temperatures are often 2° to 3° cooler and relative
humidities 20% to 30% higher within forests than in the surrounding macroclimate.

     Because forests do modify their environments, there may not be a strong correlation between annual or
short-term weather events and the response of plants or ecosystems to these events (Geiger, 1965). The degree
of correlation will depend on each ecosystem, the type and duration of abnormal climate events, and the presence
of other  stress-inducing factors in the ecosystem.  Thus,  future climate change estimates  of mean monthly
temperatures, precipitation rates, etc,  have limited value in assessing  trends and impacts on  terrestrial
ecosystems.


ESTIMATING CLIMATE CHANGE

     One method of forecasting future climate changes is to simulate known and theoretical effects of higher
greenhouse gases in the earth's atmosphere using general circulation models (GCMs) that have been developed
by atmospheric scientists (Hansen et aL, 1981; Rind, 1987).  GCMs simulate average climate conditions on the
earth's surface using  specified levels of CO, and other greenhouse gases. All existing GCMs project  higher
average surface temperatures with a doubling of atmospheric CO2 over pre-industrial levels of 300 parts per
million (ppm). These projections  differ in the relative amount of temperature increase and direction and
magnitude of precipitation change.

     GCMs have several deficiencies which limit their present  usefulness in a«*«ing effects of climate change
on forest ecosystems. The most serious is  lack of adequate spatial resolution. They can only project mean
values for very large areas encompassing thousands of hectares.  They cannot make projections for areas as small
as states  or regions within states. They cannot predict interannual variation of micro- or macroclimate variables
or the probable frequency of extreme weather events. They assume that climate change will not alter spatial and
temporal variation.

     Given these  limitations, three hypothetical scenarios of climate change were developed to derive some
indication of how they might affect forest productivity and species diversity in California and the  Southeast.
These scenarios were developed by applying ratios of climate change from three GCM models to historical data
from the National Climate Center for the period 1931 to 1980 (Gibbs and Hoffman, 1987).  The GCMs were
developed by the Goddard Institute for Space Studies (GISSX Geophysical Fluid Dynamics Laboratory (GFDL),
and Oregon State University (OSU). All GCMs assumed a doubling of CO2 concentrations. Sample values from
each scenario are shown with the descriptions of each region. These are only hypothetical scenarios of climate
change and are not predictions of change.


EFFECTS OF HIGHER CO2 AND CLIMATE CHANGE ON FORESTS

     Today's natural forest ecosystems and tree  species have evolved in response to  long-term  climate
conditions.  The species diversity in these ecosystems is  the  product of a relatively slow  process of natural
selection which took place over thoropwk of years.  Thus, the fundamental biological question about future
climate change centers on how forests will respond to unprecedentedh/ large and rapid increases in temperature
and  ambient CO-, and changes in precipitation. How will North American forest types respond to an increased
number  of extreme hot days or major increases or  decreases in precipitation?  How long will it take for them
to adjust to a new stable climate? How will the productivity and species diversity of these forests be affected as
they adjust to a new climate?


                                                 6-5

-------
 Woodman

       The problem of predicting effects of climate change on the health and productivity of American forests
 is similar to the problem of detecting and quantifying impacts of air pollution on forests (Woodman and Cowling,
 1987). Trees and forests have the inherent ability to tolerate or adapt to many stresses such as competition,
 resource deficiency, herbivory, and environmental extremes (Chapin et aL, 1987).  Tree death is often the result
 of a combination of factors (Franklin et aL, 1987). We lack baseline information on "normal" patterns of forest
 development, succession, and mortality. Without it, it will not be possible to separate and quantify the direct
 effects of climate change on forest productivity or health.

       Paleobotarrical studies provide insight into changes in forest composition and rate of change of past climate
 conditions. Using pollen data, Bernabo and Webb (1977) concluded that over a temperature increase of 4° to
 8°C, four major North American forest types migrated at the rate of about 100 km per 1,000 years over the last
 11,000 years.  These forest regions stabilized in their present form about 6,000 years ago but have  undergone
 small  changes since then because of slower migrating species.  Gajewslti (1988) described variation in the
 composition of eastern North America vegetation on time scales of centuries or millennia.  Bernabo  (1981) was
 able to detect species changes on time scales of decades to centuries.

 Physical Climate Effects

       High temperatures generally increase the rate of transpiration and the potential for dehydration in plants.
 Because the temperature optimum of respiration is higher than that of photosynthesis, carbohydrates may be
 depleted (Kramer, 1980).  Very high or prolonged leaf temperatures can induce denaturation of proteins and
 membrane damage, causing cell injury (Levitt, 1980). Nonlethal heat injury ("sun scald") in mature trees can
 occur  on the south sides of thin-barked species. Heat injury is more common on recently germinated or planted
 seedlings on sites when the soil is exposed to direct insolation.  Bark  near the soil surface is killed by acute
 temperatures, often resulting in seedling death (Levitt, 1980).

       Water stress is the factor most often limiting tree growth (Kramer and Sionit, 1987). Combinations of
 temperature, low precipitation, and low humidities create internal stresses that reduce transpiration and uptake
 of nutrients because of complete or partial stomatal closure. Sustained drought conditions reduce photosynthesis
 rates and cause premature foliage loss, which result in depletion of carbohydrate reserves and reduced growth
 potential (Waring, 1987; Schulze et aL, 1987). The cumulative effect of many transitory temperature and water
 stresses over a growing season is reduction of tree growth (Levitt et aL, 1980).  The formation of low density
 "earlywood" and high density "latewood" in conifers is influenced  by changes  in seasonal temperature  and
 precipitation (Miller et aL, 1987).

 Carbon Dioxide Effects

      Considerable research has been done to elucidate the effects of increased CO2 concentrations on plants
 (Rose, Volume C). However, most of it has provided information on short-term responses of herbaceous species
 (Pearcy and Bjorkman, 1983). Research on woody plants has also been generally short-term and conducted on
         in controlled chambers rather than on large trees in forests. Although seedling studies have provided
insight into the physiological mechanism* through which microclimatic factors control individual plant growth,
there are no mature tree or general forest process models which link effects within a tree to effects at the forest
leveL Experimentation at the ecosystem level has been limited primarily to arctic tussock tundra (Oechel and
Strain, 1985).  Predictions of effects on forest ecosystems are  mostly based on the tundra  research and
extrapolations from seedling studies.

      Sionit and Kramer (1986) and Kramer and Sionit (1987) have summarized research findings on the direct
effects of increased CO2 on tree physiology and growth. Higher CO2 concentrations increase the height, stem
diameter,  and dry weight of most seedlings  through increases in photosynthesis  rates,  inhibition  of
photorespiration, decreases in stomatal conductance of CO2 and water vapor, and increases in foliage area. The
optimum  temperature for photosynthesis may increase at elevated CO- levels. Other effects of higher  CO2
include increased branching, leaf area, and leaf thickness. Flowering and seed production is increased in some
tree species.


                                                 6-6

-------
                                                                                          Woodman
Climate-Carbon Dioxide Climate Interaction*
      Fried et aL (1986), Houpis et aL (1986), and Surano et aL (1986) studied effects of two years of high CO,
(600+ ppm) concentrations on ponderosa pine seedlings and saplings in a normal forest microclimate.  Their
research suggested that higher CO2 could have detrimental effects on some trees.  They found that trees in an
ambient plus 300 ppm CO2 environment had greater stem growth than untreated trees in the first year but
grew  substantially less in the second. The trees in the higher concentration had reduced total foliage area owing
to injury that killed half of each needle. Surano et aL (1986) speculated that this injury was caused by acute
foliar heat stress resulting from  reduced  transpiration rates.  Surviving  needles had  less chlorophyll  and
carotenoid concentrations (Houpis et aL, in press).  The net effect was that trees in ambient CO, grew more
biomass after 2 yean than the trees in the high CO2 environment

      There is some experimental evidence that higher CO, could change the competitive abilities of various
species in the same microclimates (Sionit et aL, 1984; Tolley and Strain, 1984, 1985).  Tolley and Strain  (1984,
1985) found that the amount of CO2 fixed by photosynthesis per unit of water (water-use efficiency) in high CO,
environments was greater in seedlings of sweetgum (Liauidamber stvraciflua L.) than in loblolly pine (Tinus
laejja. L.).  In most cases, higher CO2 partially compensated for effects of water stress on tree growth. It is
unclear to what extent elevated  COg can completely overcome negative effects of moisture stress and if there
are differential effects on other speaes. If increased CO2 is in general less favorable for gymnospenns (conifers)
than angiosperms (hardwoodsX this could have significant implications on future species composition of natural
forests.

      Higher CO2 levels may also compensate for the negative effects of higher temperatures on photosynthesis
rates.  Leverenz  and Lev (1987) speculated  that doubled CO2 might raise the optimum temperatures for
photosynthesis by as much as 10*C  They also  suggested that a doubling  of CO, could offset growth losses
resulting from up to a 5*C increase in temperatures.  It is not known if higher CO2 would affect the chilling
requirements of some species (Leverenz and Lev, 1987).

Ecosystem Effects

      Peters and Darling (1985) identified the following types of species and communities which would be most
vulnerable to climate change:  1) peripheral populations in ecotones near  the edges of species ranges; 2)
geographically localized species  which have no populations in other areas of suitable habitat; 3) species which
do not have the genetic diversity to adapt to a different climate; 4) specialized species which require a narrow
range of environmental conditions during some phase of their life; 5) poor seed dispersers; 6) montane and
alpine species which would have no place to migrate  and avoid  competition with other species; 7) arctic
communities, if temperature increases are higher toward the poles; and 8) coastal and freshwater lowland
communities, which will be subjected to more flooding, higher water tables, and saltwater damage.

      No  direct experimental evidence is available to predict the individual and interactive effects of CO2 and
CO,-induced climate change on forest ecosystems. Extrapolation from experimentation on individual plants is
limited mainly by lack of information  on  physiological processes that would allow scaling up of data from
exposures of minutes or hours to monthly or yearly estimations, and lack of micrometeorological data concerning
interactions of vegetation with the atmosphere (Shugart et aL, 1986).  Differential species responses within an
ecosystem to altered climate would logically lead to changes in structure and function, but predictions are not
feasible with presently available data (Dahlman et aL, 1985).

      The effect of climate change will depend on the phase of forest development at the time of climate-induced
stresses. They are 1) establishment phase, 2) self-dunning phase, 3) transition phase, and 4) steady-state phase
(Peet and Christensen, 1987).
      The eflflblfchment flhjjft a characterized by establishment of tree seedlings and rapid increases in mass
of young vegetation (Peet and Christensen, 1987). This usually follows after a disturbance, e.g^ timber harvesting,


                                                 6-7

-------
 Woodman

 harvesting, wildfire, abandonment of cropland.  It may last from one year to many years depending on the
 availability of seed sources, severity of microclimate conditions, condition of soil surface, presence of competing
 plants, predation by animals and insects, and other factors. Soil moisture, light, and temperature conditions
 during the first 60 to 90 days after seed germination are very important to seedling survival and growth.  Most
 seedling mortality is related to environmental factors.  Conifer seedlings are especially vulnerable to high
 temperatures at the soil surface (Levitt, 1980). Hardwood species have a competitive advantage over conifers
 in some situations because most regenerate from sprouts which are more tolerant of extremes in microclimate.
 The microclimate in newly regenerated forests can be more severe than the surrounding macroclimate until the
 plants become dense or large enough to modify it

      The self-thinning phase begins when the forest canopy doses and competition between plants for resources
 becomes most intense (Feet and Christensen, 1987).  As some trees increase in size more rapidly than others,
 the slower-growing and less vigorous  individuals cannot obtain enough light, moisture, and other resources to
 survive (Waring, 1987). Smaller and less competitive trees are susceptible to any perturbations which  exacerbate
 their struggle for limited resources or give competitive advantage to other individuals.  Climate-induced stresses,
 especially droughts, reduce growth rates (Fritts, 1976; Cook et aL, 1987) and increase tree mortality. Competitive
 stresses predispose weaker trees to attacks from insects and diseases (Boyce, 1948; Mamon, 1981).
      Tree mortality in the traflyifjop ph.aje is more often a result of insects, disease, lightning, and windthrow
 than of competition (Peet and Christensen, 1987).  Death of a dominant or codominant canopy tree forms a gap
 which cannot be completely filled by lateral growth of adjacent  trees.  Consequently, new seedlings become
 established, and previously established seedlings and suppressed trees receive enough light and other
 nutrients to increase in size. Forest trees in this stage of development are generally less sensitive to most climate
 stresses except droughts.

      The steady-state phase is dominated by climax species of varying ages distributed as relatively even-aged
 patches within previous canopy gaps (Peet and Christensen, 1987). The forest exists as a mosaic of various tree
 sizes and ages, sometimes overlapping and sometimes distinct In reality, large-scale episodic perturbations such
 as insect outbreaks and windthrow occur often enough that most stands do not appear to reach steady-state.
 Drought-related stresses, wildfires, wind storms, etc, are the most common causes of the relatively low  tree
 mortality rate characteristic of this type of forest

      Forest management practices of artificial thinning and fertilization can compensate for some of the negative
 effects of higher temperatures and drier conditions (Smith, 1986).  In severe  and highly variable climates,
 artificial regeneration methods reduce the risk of excessive tree mortality (Daniel et aL, 1979; Tappeiner et aL,
 1986). Site preparation, weed control, and nutrient additions are also used to increase tree survival and growth
 during establishment Some of these methods are used to establish and maintain many tree species, e.g^ loblolly
 pine, outside their natural ranges.

 Forest Productivity and Species Composition

      Shugart et aL (1986) summarized results  of climate change  on forest succession using "gap-phase" forest
 simulation models.  They concluded  that intermediate  and long-term changes in the extent, location,  and
 composition of forests could take place over hundreds of years. High-latitude and boreal forests would be very
 sensitive to increased temperatures. Tropical and sub-tropical forests would be more responsive to precipitation
 changes than to temperature increases.

     A recent study on the potential effects of a GISS-based climate change scenario on a forest type in eastern
Tennessee indicated that these climate conditions would have to exist for hundreds of years before permanent
change in species  composition and structure would occur (Urban  and Shugart, this  volume).  Shorter-term
changes (over decades) would be very difficult to distinguish from the natural variability in forest dynamics.
                                                 6-8

-------
                                                                                          Woodman

OTHER EFFECTS ON FORESTS

Insects and Diseases

     Hedden (1987) summarized the possible effects of elevated CO2 and associated climatic changes on forest
insect pests. There are no direct effects of CO- on insect pests, but temperature and moisture changes will alter
the growth and survival of some species.  Higher temperatures could increase the number of some  insect
generations each year.  Changes in moisture may also affect the kind, abundance, and behavior of many insect
species.  Because insects have much higher reproductive rates than trees, they will most likely respond more
quickly to climatic change than forests.

     Tree-pest interactions will also be influenced by changes in the susceptibility of trees to attacks.  Trees
in suboptimal sites would be most vulnerable to increased attacks (Miller et at, 1987).  Future hardwoods are
expected to have greater carbon-to-nitrogen contents in their leaves which would require defoliating insects to
consume more foliage to satisfy their nitrogen requirements (Hedden, 1987).

     According to Hepting (1963), various combinations of climate factors act as constraints against outbreaks
of severe forest diseases.  If any of these normally restraining conditions are altered, major epidemics could
result   Relatively small increases  in temperature could increase the severity of a disease depending  on its
optimum temperature. He postulated that temperature increases would increase the northern range of many
pathogens and cause some tropical diseases to move into North America.

Wildfires

     Future temperature increases and decreases in precipitation could greatly increase wildfire activity in most
forest regions.  Historically, wildland fire activity is related to the relative flammability of "fuel groups"  (plant
species, vegetation type, size of material) and regional weather patterns (Simard and Main, 1987). The most
important weather factors affecting the incidence and severity of fires are temperature, barometric pressure,
wind velocity, and the moisture  content of fuel groups (Pyne, 1984).  The USDA Forest Service integrates
weather conditions such as 24-hour mean air temperature, relative  humidity, and precipitation with fuel moisture
estimates to project fire behavior (Deeming et aL, 1977).  Higher temperatures and lower precipitation reduce
fuel moisture and increase the chances for a fire to start and spread rapidly through a forest Botkin et aL
(this volume) suggested that climate  change-induced increases in tree mortality and loss of vigor may also
increase the probability for more wildfires.


SUMMARY AND CONCLUSIONS

     The tree species that exist in natural forest ecosystems have evolved in response to long-term climate
conditions that have inherent  associated  variability.  The potential climate change induced by  increased
concentrations of "greenhouse gases" will increase the frequency of abnormal events as well  as mean climate
values.  Forests have historically responded more to interamraal variation and extreme events than to  long-term
averages in climate values. A fundamental question is what will happen to forests as the  historic equilibrium
climate conditions to which they have  evolved change.  The answer is complex, and will depend in part  on the
amount and direction of change and the inherent ability of the tree species to adapt to the new conditions. Tree
species that have  large genetic diversity will show the greatest resiliency to climate change. Species with narrow
diversity are  most likely to be adversely impacted.

      Effects of climate change on U.S. forests will undoubtedly lag behind projected increases in temperature
and changes in precipitation.  Impacts will be difficult to Htapntw as an effect of climate change and not natural
variation in extreme weather events.  Long-term consequences will depend  on the nature of climate change,
effects of higher CO2 concentrations on plant processes, sensitivities of species to changes, forest development
                                                 6-9

-------
Woodman

stages, soil fertility,  changes in insect and disease  populations, and many other factors.   In general, most
low-elevation forests in the southern half of the country will   probably experience  reductions in biomass
productivity and species diversity after 100+ years.  Forests at higher elevations and more northern latitudes
could become more productive and increase in diversity.

      Climate change will increase  the risk of tree damage and mortality due to insects, diseases, and wildfires.
Wanner temperatures will favor more cycles of most bark beetles and harmful bisects. Warm temperatures and
moister conditions would favor the  spread  and severity of many pathogens.   These trends with  higher
climate-induced stresses could cause major losses in unmanaged, older, and more dense forests.

      Some of the most important research needs are 1) more precise regional-scale climate models which can
provide more relevant information for  trees and forests; 2) tree-,  forest-, and ecosystem-level models for major
species which are sensitive to inputs of higher CO2 concentrations and changes in microclimate; and 3) more
information on effects of long-term elevated CO2 concentrations on seedlings and large trees in forests, especially
differential species effects.
                                                6-10

-------
                                                                                        Woodman

                                           CHAPTER 2

                FOREST-RELATED SOCIOECONOMIC SYSTEMS AND POLICIES


     The potential impact of climate change on the policies employed by society to manage and regulate the
forests of the United States win depend on many factors besides the direct and indirect effects on plants, animals,
other organisms, and abiotic systems which are part of forest ecosystems. It will depend in a large part on the
economic and noneconomic values which society places on the various commodities and uses of forests and to
what extent future climate change impacts those benefits. Estimates of the effects of climate change on the
sodoeconomic values of forests are even more difficult to make than estimates of impacts on trees and forest
ecosystems.  Such estimates require  projections of future sodoeconomic values of forests with and without
climate change. They necessitate estimations of future management objectives of forest owners  and trends in
regional and state  forest policies and regulations.  One must also consider trends  in population growth,
urbanization, regional land use, resource-based economies, employment opportunities,  economic standards of
living, development of new technologies, environmental protection, and other factors.


INSTITUTIONS AND  POLICIES AFFECTING FOREST USE AND MANAGEMENT

     The dominant factor now determining the uses and benefits derived from United States forests are the
goals and management policies of the forest owners. These management policies are influenced by productivity
of the land for timber  and other forest-derived goods and values, the goals or reasons for ownership, and
constraints again** achieving these goals. Some of the most common constraints for intensive management on
privately owned forestiand include unprofitable timber prices, lack of capital for investment or development, and
forest practice regulations in some states. National, state, and local political and economic conditions determine
options for management of public forests. Thus, the  potential impact of climate change in a given area will
greatly depend upon regional forest ownership patterns and the regional opportunities for forest  owners to
achieve their goals.

  o-dustrial Private For   and fNTPF)
     Most of the 194.9 million hectares (ha) of commercial timberland in the United States is owned by private
individuals (Figure 2). The USDA Forest Service defines "commercial timberlaw&as land on.which 10% of the
area is now or was formerly occupied by forest trees capable of producing 1.4 m /ha (20 ft /ac) of industrial
roundwood per year (USDA, 1982). In 1977, 72% (1403 million ha) of the nation's commercial timberland was
owned by about 7.8 million private individuals and nonforest industry companies, mostly in parcels less than 200
ha in size (USDA,  1982; Birch, 1983). Farmers, who are the single largest group of owners in this class, own
33.4% (46.8 million ha) of the total private forestland

     Binldey (1983) estimated the economic value of timber from non-industrial private forestland (NIPF) to
be $600 billion in 1982. Besides the production of income from sales of timber, reasons for ownership of private
forestland include esthetics, recreation, land investment, and inclusion in farm or residential property.  Due in
part to the wide diversity of interests of owners, less than half of NIPF land is  deliberately managed in some
way for timber, wildlife, recreation, or other uses.   Decisions to plant  trees and actively manage  for timber
production are often determined by current government-funded incentive programs, the availability of free or
low-cost management advice, and state and regional regulations.

Forest Industry-Owned Forestland

     The forest industry owns or controls 143% (27.8 million ha) of the commercial timberland in the United
States (Figure 2). Although a few companies own forests only for the purpose of growing and selling wood to
others, most forest industry companies that own forestland use their wood in their own facilities. In almost all
                                                6-11

-------
Woodman
     I
     •a
     CO
60

50

40

30

20

10
                      North
                                                                    Q   Other Public
                                                                    0   Federal
                                                                    |   Forest Industry
                                                                    •   NIPF
                            South         Rocky Mountain      Pacific Coast
                                     Region
           Figure 2.  Area of commercial timberland by region and ownership (USDA, 1982).
                                            6-12

-------
                                                                                        Woodman

cases, their primary reason for owning forestland is to maximize profits from it This is usually accomplished
by producing low-cost wood for manufacturing facilities and ensuring that a long-term wood supply is available
for their mills.

     The relative intensity of industrial forest management, as measured by dollars invested in regeneration
and sttvicultural  activities, is also influenced by the  availability of cheaper timber  from NIPF  and public
forestlands. The current surplus of such wood in the Northeast and Lake States is a major factor in the low level
of management now practiced in these regions (Benzie et aL, 1986; Rose et aL, 1987). Likewise, the large supply
of low-cost timber available from national forests in the Northern Rockies and Pacific Coast states has reduced
management incentives on industrial land in these states (Tappeiner et aL,  1986).

     In  all regions, most forest products companies depend  on wood  harvested  from NIPF  and public
forestlands.  Only a few companies can satisfy all of their wood requirements from timber grown only on their
own lands. Thus, the potential impact of reduced forest productivity from climate change will vary by company
and by region. The greatest impacts win occur where current timber supplies now balance demand or where
shortages in certain types of wood are predicted by the year 2030. Using a GFDL climate model-based scenario,
Rose et aL (1987) projected a 30%  shortage of softwoods and a surplus of hardwoods in the Lake States and
North Central regions.  Some shortages in hardwoods  and softwoods have been projected in a few small areas
of the South, irrespective of climate change (Healy, 1985; Rose et aL, 1987).

Public Forestland

     Altogether, federal, state, and local governments own or control 27% (54.7 million ha) of  commercial
forestland (USDA, 1982).  Federal  forestlands are managed for various purposes by  the U.S. Department of
Agriculture (USDA) Forest Service  (39.9 million ha), U.S. Department of the Interior  (USDI) Bureau of Land
Management (BLM) (13 million haX USDI National  Park Service (NPSX  and US. Department of Defense.

     The management goals  for national forests are regulated by a complex set of legal directives and
administrative procedures. The most important are the Organic Act of 1897, the Multiple Use-Sustained Yield
Act of 1960, the Wilderness Act of 1964, and the National Environmental Policy Act of 1970.  The Forest and
Rangeland Renewable Resources Planning Act and the  National Forest Management  Act, passed in  the
mid-1970s, require that 15-year management plans for each national forest incorporate the public needs and
uses of the forests for the region in which they are located.

      The Federal Land Policy and Management Act of 1976 gave the USDI BLM statutory responsibility to
manage their lands under principles of multiple use and sustained yield. It also prescribed  that decisions be
based on inventory and land-use  planning involving broad public participation.

      The USDI NPS is responsible for managing  all national  parks, monuments, historic sites, national
recreation areas, and national wild and scenic rivers.  Its legislated goal is to  protect each park's natural  and
scenic attributes for the "unimpaired enjoyment of future generations and to manage  them for  the use  and
recreational benefits of the current generation." This is accomplished through a comprehensive long-range
planning process and system of management zoning. Some of the controversial management strategies utilized
on these lands include permitting wildfires to run uncontrolled in parks, eliminating exotic plants, and controlling
human access to sensitive wilderness areas.


POTENTIAL EFFECTS OF CLIMATE CHANGE ON FOREST USE AND MANAGEMENT

      No assessment of the effects  of climate change on non-industrial private forest owners or public forest
managers has been published to our knowledge. Many  NIPF owners do not have the financial resources,
flexibility, or incentives to dramatically alter their current management practices. Increased risk associated with
the stress of  climate  change may prove an added detriment  to long-term investments in intensive forest
management by these  owners.  Because forests which are unmanaged are likely to be at greatest risk from


                                               6-13

-------
 Woodman

 projected climate changes, this class of forest ownership will most likely show the greatest negative effect in the
 future.  The forest industry's reliance on NIPF wood in some areas win cause reverberations in that sector as
 well The future productivity of private forests may be forced to rely on more extensive use of intensive forest
 management techniques.  In order for this to come about, greater incentive measures may be necessary, or even
 increased regulation of forest practices.

      The current statutes and management policies of the USDA Forest Service and USDIBLM are sensitive
 and flexible enough to respond to climate change given all its present uncertainties. Mandated preparation of
 management plans based on environmental,  biological, and economic information should give federal managers
 a good framework for responding to climate changes. The public is likely to have greater need in the future for
 the products, both economic and unpriced, that are derived from state and local forests, and the ability of the
 forests to provide them will depend upon rigorous long-term planning and strict  management policies.

      Given the large uncertainty about climate change on the biology and economic values in major U.S. forest
 regions, the forest industry wfll probably take a "wait-and-see" attitude about global climate change before
 altering their management strategies.  Most companies do not make major decisions based on long-term plans
 for periods greater than 5 to 10 years in the future.  The response of the forest industry to climate change will
 depend on how quickly changes occur and can be measured. Malac (1987) suggested that adaptation of specific
 management practices or  defensive actions will always  lag behind climate change because of the high degree of
 uncertainty of the changes on a particular region.

      Sandenburgh et aL (1987) suggested that forest products companies could minimise the negative impacts
 of climate change by acquiring forestland in areas with little or no predicted impacts on forest productivity,
 placement of new manufacturing facilities in lowest impact regions, and ghn»«mg future markets where climate
 change may increase profits. Larger companies could move their manufacturing plants and forestland operations
 to regions where they can make  the highest  profit  Since profitability, rather than productivity, primarily
motivates industry decisions, climate change could reduce forest productivity but increase profits if higher prices
are paid  for wood products. Likewise,  negative impacts on  other forest regions in the world could improve
profitability through increased exports of wood products to other countries.
                                               6-14

-------
                                                                                         Woodman
                                           CHAPTERS

             POTENTIAL CLIMATE CHANGE EFFECTS ON CALIFORNIA FORESTS
INTRODUCTION
     California is the second largest state in the continental United States.  Its 33.6 million ha of forests and
rangelands produce a variety of goods and services of importance to the inhabitants of the state, the nation, and
the world. Nearly all annual water runoff in the state falls as precipitation on forested watersheds.  In addition,
forestlands provide habitat to more than 650 different wildlife species, provide 15% of the average annual U.S.
production of softwood, support the grazing of nearly 2 million head of range livestock, and offer  recreational
opportunities for millions of visitors each year (CDF, 1988).
Geography and
     California has some of the most diverse and striking landscape  and physical diversity in the country
(Britanmca, 1982). Its long and relatively mountainous coastline ranges in elevation from 610 to 2500 m, while
the eastern portions of the state are occupied by relatively flat, sparsely settled desert  The Sierra Nevada
Mountains, in the eastern central region of the state, contain 11 peaks which exceed 4250 m in altitude. The
Central Valley, a 140-km trough between the Coast Ranges to the west and the Sierra Nevada to the east,
constitutes the state's agricultural heartland (Dudek, Volume C). Most of the densely populated area known as
southern California is  located on a coastal plateau and in valleys within 65 to 100 km of the coast, separated
by the Transverse Mountains from the Central Valley.

     California has a Mediterranean-type climate characterized by hot, dry summers and frequent summer
droughts. Great temperature contrasts exist across the state because of the influences of mountains, deserts, and
the ocean.  Summer temperatures in the southeastern desert often reach 54°C.  Maximum daily temperatures
in the high-elevation eastern desert (1220 to 2300 m) range from 24° to 32°C.  Although most of the annual
precipitation occurs in the fall and winter months, mean monthly precipitation varies similarly to the example
for  North Carolina (Figure IB).  Annual rainfall ranges from the extreme annual mean of 4420 mm in  the
northwest to traces in  the southeastern desert


FOREST RESOURCES

Maior Forest Types

     The natural vegetation of the state has been grouped into five broad cover types:  conifer, hardwood,
shrublands, grasslands, and desert (CDF, 1988).  Only a small portion of the hardwood cover class meets the
definition of commercial forest  Approximately 21% of the commercial forests are of the conifer cover type.
The seven most  commercially important vegetation types,  as defined by the California Division of Forestry
(1988),  within the conifer type are as follows:

        Mixed Conifer (40%) - mainly  ponderosa pine (Finns ponderosa Law&X Douglas-fir (Pseudotsuga
        menriesii (Mirb.) Franco), white fir (Abies concolor (Gord. & Glend.) LindL (ex Hildebr.)), incense-cedar
        (Libocedrus decurrens Torr.), sugar pine OL lambertiana DougLX  and Jeffrey pine GC, jgfjigyj Grev. &
        Half.). California black oak (Ouercus kelloyyii Newb) is a major hardwood associate;
        Ponderosa Pine (11%) - mostly ponderosa pine. Associated species may include white fir, incense-cedar,
        Coulter pine (Pinus coulteri D. Don), Jeffrey pine, sugar pine, and  Douglas-fir,
                                                6-15

-------
 Woodman


         Red Fir (8%) - mostly red fir (Abies magnifies A. Murr.) with some lodgepole pine (Pinus contorta
         DougL) and noble fir (Abies procera Rehd);

         Douelas-Fir (8%) - mainly Douglas-fir, but may include sugar pine, ponderosa pine, Jeffrey pine, incense
         cedar, redwood (Sequoia semoervirens (D. Don.) EndLX tanoak, madrone (Arbutus menriesii Pursh). and
         canyon live oak (Ouercus agrifplia Nee);

         Redwood (7%) - redwood with Douglas-fir, ritka spruce (Picea sitchensis (Bong.) Carr.), grand fir (Abies
         grandis (DougL) LindL), western redcedar fThuia plicata Donn.X western hemlock (Tsuea heterophvlla
         (Raf.) Sarg.), red alder (Alnus rubra Bong.), tanoak (Lithocarpus densiflorus (Hook. & Am.) Rehd),
         madrone, and bigleaf maple (Acer macrophvilum Pursh);

         Lodyepole Pine (3%) • mostly lodgepole pine, but may include some red fir; and

         Jeffrey Pine (3%) - mostly Jeffrey pine with some ponderosa pine, Coulter pine, sugar pine, lodgepole
         pine, incense cedar, and red fir.  Black cottonwood (Populus trichocaroa Torr. & GrayX aspen (Populus
         tremuloides Michx.X  and California black oak are hardwood associates.

      The five remaining noncommercial conifer vegetation types are Juniper (Junioerus spp. Carr.) (6%), Pinyon
 Pine (Pinus monophvlla Torr.  & Fre.>Juniper (6%X Montane Hardwood Conifer (5%), Subalpine Conifer (1%),
 and Gosed-Cone Pine Cypress (0.2%).

 Timber Production

      Approximately 18% (7.5 million ha) of California's land is commercial forestland (USDA, 1982; CDF,
 1988). Of the forestland not reserved for parks or wilderness areas, 52% is administered by the Forest Service,
 23% is owned by approximately 120 forest industry companies, and 21% is owned by 60 to 100,000 NIPF owners.
 Timber from all private land accounts for almost 59% of the total harvested, 31% comes from national forests,
 and 10% from other public lands (CDF, 1988).

      Most of the timber harvesting and manufacturing facilities are  concentrated in the northern half of the
 state. In addition, 45% of the total harvest from industrial and NIPF lands comes from the North Coast region
 (CDF, 1988). Nearly all of the timber harvested in California remains in the state for processing.  Almost 61%
 of the total wood used in the state is currently imported

      Timber harvesting in California has decreased significantly from a peak of 29 million m /yr in 1955 (CDF,
 1988). Given current social, demographic, and economic trends, without global climate change, the CALPLAN
 simulation model (Davis et  aL, 1986)  projected an annual harvest of 15 million m /yr in the year 2020 to 2030.
 Some timber harvest forecasts  project slightly higher levels by assuming increased cuts from NIPF lands and more
 intensive management practices (CDF, 1988).

      Given continuation of present trends,  the main deterrents to the future productivity of forest resources are
 the conversion of forestland to other  uses, wildfire,  forest pests, management practices, and climate  change.
 Nearly 2.0 million ha of forest and rangeland cover types were lost to urbanization and development of agriculture
 from 1950 to 1980. Another 0.9 million ha  are expected to be converted over the next 30 years (CDF, 1988).

 Water

     Water is the most important and controversial product  derived from the state's forests  (CDF, 1988).
 Current needs for water now exceed the available supply. Although 70% of annual precipitation falls on forests
 in the northern third of the state, most of it is consumed by agriculture and urban populations hi the southern half
 (Dudek, Volume C; Lettenmaier, Volume A).  Precipitation is greatest in the winter and is lowest in the summer,
when water demands are at their  peak.  In  order to help overcome water imbalances, California has developed


                                               6-16

-------
                                                                                           Woodman

an extensive water storage and distribution system (Lettenmaier, Volume A; Dudek, Volume C).  In addition,
water from Oregon and the Colorado River is imported to meet total water needs.

      The driest year in California's  recorded history was 1977. That year was also the second successive dry
year of the worst drought of more than 100 years of record. Average stream and river runoff was only 47% and
22% of average in 1976 and 1977, respectively.  This drought is believed to have directly or indirectly caused the
death of 5 to 12 million trees (CDF,  1988).

Recreation

      The state parks, National Forests, and National Parks in California draw more annual visitors than those
of any other state (CDF, 1988). Recreational use of national forests in 1986 amounted to 55 million RVD's (one
RVD represents 12 hours of participation in any recreational activity by any individual); National Parks measured
19.9 million, state parks 153 million, BLM lands about 8.7 million, and  all other federal lands about 5.4 million
(CDF, 1988).  The San Bernardino National Forest, located near Los Angeles, has the highest recreational use
of any national forest Yosemite National Park is the third most visited National Park in the country. The USDA
Forest Service forecasts that national forests in California will have the most rapid increase in forest and rangeland
recreational use of any region into the next century (USDA, 1981).

Forage

      Livestock production is an important renewable resource of California's forests, ranges, and desert lands.
The quality and volume of forage produced in  different vegetation cover types depends greatly on precipitation
patterns, denary of overstory vegetation, and species competition. Approximately 35% of the forage consumed
by livestock in 1985 came from the hardwood and conifer cover types (CDF, 1988). Although forage from forests
is not as abundant or nutritious for livestock as rangelands, it is important to the livestock industry because it
provides forage in the summer when little  is available from grasslands.

      Over the next 30 years, amounts of forage consumed by cattle and sheep are projected to decline by about
7%  and 10%, respectively (CDF, 1988).  These  predictions are based on  expected losses of rangeland to
agricultural and urban uses, reductions in the nutritional quality of natural forage due to curtailments in spraying
and burning of rangelands, and prices for beef and lamb.

Wildlife

      Conifer forests provide breeding habitat  for 108 species of mammals, 148 species of birds, and 55 species
of reptiles and amphibians (CDF, 1988). Hardwood forests provide important habitat for other bird and mammal
species, such as turkeys and feral hogs, two introduced wildlife species. Fishing and deer hunting are the most
popular wildlife-related uses of forests.  Nearly 15 million fishing licenses and 315,000 deer tags were sold in 1984
and 1985 (CDF, 1988).  Most hunting and fishing takes place on federal and state lands, but an increasing number
of private landowners sell hunting rights to their forestlands.

Factors Affecting Forest Health and Productivity

      Wildfire.  On the average, wildfires have damaged  more than  133,000 ha of forests and rangelands in
California each year since 1980 (CDF, 1988). In southern California, a  variety of critical fire periods combine to
produce an essentially year-long fire season  (Pyne, 1984).  Fires tend to be large  and episodic, often with
urban-related causes. In northern California, there is a distinct summer  fire season as a result of low precipitation
during those months (Pyne, 1984). Almost 60% of the fires in the Pacific states from 1973 to 1978 were caused
by lightning (USDA, 1980).

       Insects  and Diseases.  The annual loss  of merchantable timber  to insects and  diseases is thought to be
substantial (CDF, 1988).  Bolsinger (1980) estimated insect losses to be 23% of the total tree mortality in
California's interior forests during the late 1960s and early 1970s.


                                                 6-17

-------
 Woodman


       The most harmful forest insects are bark-cambium feeders and defoliators such as the western pine beetle
 (Dendroctonus brevicomis LeConte), mountain pine beetle  (Dendroctonus ponderosae Hopkins), Douglas-fir
 beetle (Dendroctonus oseudotsueae Hopkins), red turpentine beetle (Dendroctonus valens LeConte), fir engraver
 (Ips spp.X. pine engraver dps spp.i flatheaded fir borer (Melanophila drummondi Kirby), and the California
 flatheaded borer (Melanoohila califomica Van Dyke)  (CDF, 1988).  The most important defoliators are the
 Douglas-fir tussock moth (Qrevia pseudotsuqata McDunnough), budworms, needle  miners, and the gypsy moth.
 Dwarf mistletoe (Phoradendron sop.) is the most common disease causing tree mortality, affecting as much as 15%
 to 25% of Douglas-fir and other pine and fir species (CDF, 1988).  Major diseases include annosus root rot
 (Heterobasidion annosus (Fr.) Bref.X Armfllaria root rot (Annillaria spp.X and black stain root diseases (Xvlaria
 spp.). White pine blister rust (Tfonprtfom. ribicola Fischer) affects many species of pines.
      Air Pollution. There is growing public concern about effects of air pollution on California's forests. Ozone
 is the only regionally dispersed pollutant which is presently known to occur in concentrations high enough to cause
 foliar injury to some sensitive trees (Woodman, 1986, 1987a).  Los Angeles has the highest measured ozone
 concentrations of any city in the country.  Miller (1983) and others have identified ozone symptoms in 12 conifer
 and hardwood species in southern California. Injured trees tend to  have a greater susceptibility to droughts,
 reduced growth rates, and fatal attacks from bark beetles and root  diseases than uninjured trees.

 Trends in Forest
      Most of the commercial forest types in California occur on relatively steep mountainous terrain which is
 not suitable for uses other than forestry (Tappeiner et aL, 1986). The more accessible virgin forests in the coastal
 areas and lower Cascades and Sierras were logged in the late 1800*3 and early 1900's.  Preferential harvesting of
 pines have resulted in Sierran forests dominated by shade-tolerant fir species. The interior forests, e^, eastside
 ponderosa pine and lodgepole pine, have been repeatedly logged over many years as merchantable  species and
 sizes changed. Most forests are now dominated by young-growth rather than old-growth stands.

      Over the last 30 years, timber management practices have included salvage logging of dead and dying trees,
 intensive site preparation and tree planting, control of animal damage (e.g, porcupines, rodents), release of conifer
 seedlings from competing shrubs and hardwoods, and thinning (Barrett, 1980; Tappeiner et aL, 1986). Relatively
 few trees are planted  due to cost and poor survival  Ponderosa  pine and Douglas-fir are the species  most
 commonly planted.

      California's "forest practice rules" are viewed as being more restrictive to management of private forestlands
 than similar regulations in other states (Tappeiner et aL, 1986).  These laws require that timber harvesting and
 regeneration plans for private lands be prepared by a licensed professional forester and approved by a state
 forest-practices official (Tappeiner et aL, 1986). The species, stocking rates, and time within which reforestation
 must be accomplished are prescribed.  In addition, some counties have ordinances which regulate stand densities
 after partial cutting and the silvicultural system to be employed  in some forest types.

     Tappeiner et aL (1986) and Woodman (1987b) forecast that most private forest owners will  adopt a
 graduated system of forest management  intensities  over the next  few decades.  These  zones of management
 intensity will be shaped by considerations such as forest site productivity, proximity to manufacturing facilities,
 harvesting  costs,  and availability  and cost  of transportation facilities.  They will also  consider topography,
 competition with other sources of timber, availability of labor, and local ordinances.  The most capital-intensive
 practices will be limited to those zones capable of providing the greatest financial return to owners.  In zones
 where esthetics, watershed management, and other uses are  dominant, timber management will be minimaL


SOCIOECONOMIC TRENDS AFFECTING FORESTS

     Ten percent of the population of the United States are California residents.  The state's $500 billion annual
output of goods and services constitutes 12% of the U.S. gross national product In 1985, service employment was


                                                6-18

-------
                                                                                        Woodman

55% of total employment, manufacturing was 17%, and agriculture and forestry only 3% (CDF, 1988). The state
is dominated by urban dwellers, whose attitudes and values play a major role in determining policy on the state's
forests.  Three southern counties contain 47% of the people, with 12% of the total (33 million) living in the city
of Los Angeles. Almost 20% of the population lives in the San Francisco Bay area.

     These demographic and socioeconomic conditions have resulted in a population which is wealthier and
more educated than average, willing to make  investments in public  service,  and more  concerned about
environmental issues than citizens in many other states (CDF, 1988). They have stimulated formation of a number
of public interest groups who  are politically active and able to exert influence on  public policies related  to
environmental quality and management of private and public forestlands. These trends are not likely to change
and may intensify with greater public information on global climate change and other environmental issues.

     The 1988 report of the Forest and Rangeland Resources Assessment and Policy Act Committee identified
a number of socioeconomic trends which may influence the effects of climate change on California's forests and
citizens  (CDF, 1988). These include the following:

        .   increasing needs for water and outdoor recreation;

        -   increasing demands for lumber, plywood, paper, and other wood products;

        -   a decrease in the ability of the state's forest industry to compete with other regions and maintain its
           present share of the nation's wood products markets;

        -   a reduction in the number of wood processing facilities and experienced forest labor;

        -   "*pr*a«">e demands for livestock products without price increases;

        -   increasing costs of owning and managing forests without compensating increases in wood prices;
        .-  more public pressure to manage public and private forestland for noneconomic values, e^, water and
           recreation; and

        -  increased state and county government regulations which limit opportunities for adequate financial
           returns to forest owners and managers.


PROJECTED CLIMATE-CHANGE SCENARIOS

      The three hypothetical scenarios of global climate change in California postulate statewide increases in
average monthly temperatures and slight decreases or modest increases in precipitation depending on the location.
On an annual average basis, the GISS model-based scenario estimated the greatest mean change in monthly
temperature (4.7*C) and precipitation (7.0 mm).  The OSU-based scenario postulated the smallest changes in
temperature (23*C) and monthly precipitation (0.5 mm).

      A representative example of changes in monthly temperature and precipitation is shown for California's
North Coast climate division in Figure 3A and 3B. The very productive Douglas-fir and coastal redwood forest
types are found in this division. The  GISS scenario had the highest individual monthly changes in temperature
(+8.4* in September) and precipitation (+68 mm in March). The OSU scenario projected the smallest maximum
monthly temperature increase (L2* in January) and the GFDL the greatest monthly decrease in precipitation (-28
mm in January).

      Figure 3  also illustrates some of the seasonal differences between scenarios and how they might influence
future forest productivity and composition.   For example, forest regeneration  and good seasonal growth is
dependent on having moist soils and low water stress conditions in the spring (March-April) and early summer


                                                6-19

-------
Woodman

     (A)
     (B)
                    Jan  Fab  Mar  Apr  May Juna  July  Aug  Sapt  Oct  Nov  Dec  Av«

                                                    Month
                                   Apr  May  Juna  July  Aug  Sapt  Oct  Nov  Dae  Ava

                                                 Month
Figure!   (A) Change in mean monthly temperature by scenario for North Coast Climate Division. (B) Change
           in mean monthly precipitation by scenario for North Coast Climate Division.
                                              6-20

-------
                                                                                          Woodman


(May-June). High temperatures and water stress have a potentially greater impact on forests if they occur during
these periods than they do after July.  Thus, the conditions of the GISS scenario would have less negative impact
on North Coast forests than the OSU or GFDL scenarios.   GISS projections of higher winter and spring
precipitation with wanner spring temperatures would provide  the best growing conditions  for these forests.
Assuming that postulated higher spring rainfall cannot offset potentially drier soils due to less than average winter
precipitation, the OSU scenario would have the greatest potential impact on forest survival and growth in this
division.

      Climate change scenarios for other state climate divisions  generally postulate 10 to 60% less precipitation
in May, June, and July. All climate scenarios for Southern California presume future growing season conditions
which are 15° to 7.8* wanner and 5% to 50% less than the historic average.  These conditions would primarily
impact the noncommercial forest types in this  part of the state.  Climate change estimates for commercial forest
types are similar to those depicted for the North Coast region.


POTENTIAL IMPACTS ON SPECIES COMPOSITION AND FOREST PRODUCTrvrTY

Species Composition

      No  major assessment of the range  of  potential effects of global climate change  on California's forest
resources has been published.  Davis (this volume) used fossil pollen  samples to   reconstruct the species
composition of western Sierra Nevada forests growing  at equilibrium conditions almost 9000 years ago when
temperatures were 1"C higher and monthly precipitation 5 mm less than today. He concluded that much of the
present forested area was sagebrush steppe and that pine and fir trees were less abundant than today.

      Using a GFDL-based scenario, Leverenz and Lev (1967)  concluded that the current natural  range of
Douglas-fir, western hemlock, and Englemann spruce (Picea *nglmannii Parry) in California would eventually
disappear because of a reduced ability to naturally regenerate. They hypothesized that the conditions in this
scenario would favor expansion of ponderosa and lodgepole pine. They assumed that  up to a 50% increase in
evaporative  demand  over available  precipitation would be  compensated for  by benefits from higher CO2
concentrations. If COj cannot compensate to that extent, they postulated that the natural range of most western
conifers would be diminished over an unspecified period of time.

      Assuming that future climate  change  does not exceed the bounds of the  three  scenarios, there is  no
compelling scientific information to suggest that California will experience significant large-scale reductions in the
distribution of major tree species within the next century. None of the projected climate changes appears severe
enough to directly kill trees. Shifts in tree numbers and distribution will likely take place over hundreds of years
through periodic extreme temperature or drought events and inabilities of some species to naturally regenerate
themselves after catastrophic mortality from wildfires and insects and disease epidemics. Although some species
may disappear from areas on the periphery of their natural ranges ("ecotones"), it seems unlikely that any of the
hypothetical climate change scenarios would eliminate all individuals of a species from  their present range.

      The major commercial tree species most likely to  be adversely impacted by climate change are those most
sensitive to drought and high temperatures.  These include Douglas-fir, western hemlock, red fir, and white fir.
Most of the major pine species have evolved under warm and dry conditions and will likely not be greatly impacted.
Changes hi climate will probably have the greatest effect on trees occupying marginal habitats or peripheral areas
of their natural ranges (Layser, 1980). These areas have  undergone the greatest contraction and expansion during
past cyclic periods of temperature and precipitation change.

Forest Productivity

      Higher temperatures (and concomitant, increased evapotranspiration demand), combined with small or no
increases in precipitation, will create drier summer conditions. The most probable short-term effects of such


                                                 6-21

-------
 Woodman

 conditions on low-elevation forests will be decreased annual growth.  Forests growing above 1*500 m elevation
 could become more productive as a result of warmer temperatures and longer growing seasons, although their
 natural ranges may be decreased.  The thresholds between no changes, decreases, or increases in forest growth
 cannot be estimated from available information.

       Lavender et aL (1986) raised concerns that Douglas-fir and other cool  climate species may experience
 reductions in growth and vigor solely from their inability to meet minimnm "chilling requirements" in a future
 warmer climate.  Douglas-fir seedlings need temperatures which do not exceed 5°C for 13 weeks (Lavender et aL,
 1986). In their assessment of climate change impacts on western conifers, Leverenz and Lev (1987) assumed that
 these requirements would be met if average temperatures did not exceed 9*C for one month. The OSU scenario
 is the only scenario which would meet  the Douglas-fir chilling requirements in the North Coast climate division
 where this species is primarily located.

       Any increase in temperature win likely increase  the potential for damage  by diseases and destructive
 populations of insects. Droughts are frequently followed by epidemics of bark beetles (Furniss and Carotin, 1977).
 One of the most  notable effects of the California drought of 1976 and 1977 was a significant increase in tree
 mortality as a result of bark beetle attacks (CDF, 1988).  Increased temperatures and relatively drier conditions
 projected by all scenarios will increase  the likelihood of wildfire occurrences and the danger posed by fires once
 they start

 Potential Impacts on Forest-Based Sodoeconomic Systems

      The most likely impact of climate change on current sodoeconomic trends are 1) increased disincentives
 to private owners and industry to manage forests for timber production, 2) additional need for production of water
 from forests, 3) increased demands for forest-oriented outdoor recreation experiences, and 4) increased need to
 salvage trees killed or injured by wildfires, insects, and diseases.

      The trends described in the  Forest and Rangeland Resources Assessment and Policy Act report (CDF,
 1988) raise the question of whether or not  private  and industrial forestlands will be able to  yield acceptable
 financial returns to their owners by the time of CO2 doubling.  The increased risks likely to be associated with
 future forest management may result in further reductions in the size and value  of industrial forests  in the state.
 Profit margins have decreased significantly for most timber companies due to higher harvesting costs, higher labor
 costs, old  and inefficient mills, and state and local  restrictions on timber harvesting and forest management.
 Without substantial increases in the value of timber, or  fees paid to forest owners for use of then- forests for
 recreation and generation of water,  the size and viability of the forest industry will undoubtedly decrease.

      Climate change will not likely alter the present trend in management policy on national  forests, national
 parks, or other federal lands. A.«mming that effects  of climate change are relatively subtle, the current trend in
 public demand for water and recreation will dominate policy considerations on these lands. The importance of
 water yield in national forest management plans is increasing and will continue  to influence the use of national
 forests without climate change.  If climate change reduces water yields, recreation and water supply fees could
 be established to help balance revenue reductions from timber sales. Future forest management practices will be
 dominated by needs to increase water from forested land, reduce risks of major insect epidemics, and salvage dead
 trees.  Utilization of salvaged trees may require some level of price support in order to maintain and encourage
 local wood utilization capabilities and a skilled labor force.


 CONCLUSIONS

      Projected warmer temperatures and possible shifts in precipitation would most likely decrease productivity
 and species diversity in many low-elevation California forests. Higher-elevation forests could experience greater
 productivity and an increase in the number of species. Higher winter temperatures will reduce mountain snow
 packs and summer stream flows.  Some of the potential negative effects of climate change could be ameliorated
by beneficial effects of COj, and the large number of tree species which have developed varying tolerances to


                                                6-22

-------
                                                                                            Woodman

extremes of temperature and water stress. Annual tree mortality will likely increase due to more wildfires, greater
insect populations, and increases in diseases.

     The state's distinctive sodoecooomic trends will most likely continue to greatly influence forest use and
management, regardless of climate change.  All forest managers will devote more attention to activities which
increase water production, provide more forest recreation opportunities, and lower the risk of catastrophic losses
to insects, diseases, and wildfires.  Potential decreases in forest productivity and losses of forestland to other uses,
will most likely intensify public pressure to legislate more multiple-use practices on  private forestlands. Most
timber harvesting from federal and state lands will come from tree salvage and forest sanitation operations which
remove dead, dying, and susceptible trees.
                                                  6-23

-------
 Woodman


                                             CHAPTER 4

             POTENTIAL CLIMATE CHANGE EFFECTS ON SOUTHEASTERN FORESTS


 INTRODUCTION

       The South, or Southeastern Region of the United States, includes the states of Virginia, North Carolina,
 South Carolina, Georgia, Florida, Tennessee, Alabama, Mississippi, Arkansas, Louisiana, Oklahoma, and Texas.
 It has been called the "nation's wood basket" because of its importance in the production of wood and the
 manufacture of wood  products.  Its mfld climate,  long growing season, and favorable precipitation  provide
 conditions conducive to the production of timber and many agricultural crops. These states contain approximately
 40% of the country's commercial forests, from which almost half of the nation's timber is harvested each year
 (USDA, 1982).

 Geography and Climate

       Compared to California, the topographic relief of the forested areas in the South is relatively flat (Britannica,
 1982). The largest and most productive of three  subregions is the Coastal Plain, encompassing 45% of the area
 (Figure 4).  It is a low-lying and flat 250- to 650-km-wide strip along the coast  It contains the Mississippi River
 Delta, Coastal Flatwoods, and the Florida Peninsula. The Piedmont is the subregion between the Coastal Plain
 and Appalachian mountains.  It has a mild topographic relief characterized by raiting hills and narrow streams at
 elevations of 100 to 250 m above sea level The Uplands subregion incorporates high plateaus, steep ridges, and
 mountain peaks with elevations ranging from 300 m to 2000+ m. The Uplands include the Blue Ridge Mountains,
 the Appalachian Plateau, the Ouachita Uplands, and the Ozark Plateau.

      The Southeast has a continental climate with maritime influences which is characterized by moderate to
 cool moist winters and moist warm summers (Britannica, 1982). The 1000 to 1500 mm of annual precipitation is
 generally well distributed over the year and droughts are more common in the spring and fall than in the summer.
 East Texas and Oklahoma have the lowest average precipitation (82  mm/month) and  greatest frequency of
 severe spring droughts.  Louisiana, Mississippi, Alabama, and Florida  have the highest rates  (110 to  118  mm
 /month). Monthly precipitation is highest between June and September in  the Coastal Plain and January through
 August in the Piedmont and Uplands.

      Mean annual air temperatures range between 164" to 20°C  (Britannica, 1982). The coldest temperatures
 (3.4° to 6.0°) occur in January and hottest (25° to 28°) in July. The highest temperatures and humidities occur
 in the south coastal areas. Historically,  the probability of clear sunny days are 60% to 80% in the summer and
 40 to 60% in the winter. Frost-free periods of 215+ days in the Coastal Plain and Piedmont permit forests to grow
 longer than in any other major forest region (USDA, 1969).


 FOREST RESOURCES

 Major Forest Types

      More than 100 commercial tree species grow in the South (Boyce et aL, 1986).  According to the USDA
Forest Service (1982), the major forest cover types in order of total forest  area are as follows:  oak (Ouercus
spp.Vhickorv (Carva SOP.) (31%); loblolly pine (Pinus taeda L.>shortleaf pine (L echinata Mill) (24%); oak-pine
(16%); oak-gum (Nvssa spp.Vcvpress  (Cupressus sop.) (14%); longleaf pine QL palustris MilL)-slash pine (P_j
elliottii Engelm. var. elliottii) (9%); elm (Ulmus spp.Vash (Fraxinus spp.Vcottonwood (Populus sop.) (2%); maple
(Acer spp.Vbeech-birch (Betula sopA white (£» strobus L.)-red (£,  resinosa AiL>jack pine (£» banksiana Lamb.),
and spruce (Picea rubens Sarg.)-fir (Abies fraseri (Pursh) Poir.) (0.4%).
                                               6-24

-------
                                                                                            Woodman
                                    THTitf Spring OUfTVTMf
                                QISS  -25   +3   +20
                                GFDL  4-17   4.15   -21
                                O6U   -10   -10    +«
QISS  -28   43  420
QFOL  +35  440  -32
CflU  .13   +•  +30
                                                                                         QISS   -9*32*33
                                                                                         GFDL  +17   +15  -21
                                                                                         OSU    -5    -9  +24
                                                                                              v Spring Sunwnv
                                                                                      QtSS   -8+22+32
                                                                                      GRX   -3   +12  -<•
                                                                                      OSU   ^    -9  +24
                                \
                                 Coastal Plain
                                                                          QISS   -8  +22  +32
                                                                          GRX  -3  +12  -49
                                                                          OSU  -10  -19   +•
                                                                                         prng
                                                                                    -24   -7  +12
                                                                              GFDL -13  -11  -32
                                                                              OSU   +5  -10
                                     QISS -25+3+20
                                     GFDL +12   +5   -37
                                     OSU  -16   -H  +22
          -ai   -10
    OFDL  4«  +16
    OSU  4-14   -21  481
Figure 4.   Percent change of «»g«niH precipitation projected by three climate-change scenarios for locations in
           three major physiographic regions of the South. Monthly values averaged for each season were:
           winter-November, December, and January; spring-February, March, and April; summer-June, July,
           and August.
                                                 6-25

-------
 Woodman

       Most of the pine types are found in the Coastal Plain and Piedmont (Boyce et aL, 1986). The spruce-fir
 type is  confined to  ridge and mountain tops in the Appalachian  Mountains.   The oak-gum-cypress and
 elm-ash-cottonwood types are mostly found on moist sites near rivers and streams. The highest quality sites for
 hardwood species are in bottomlands.  The oak-pine and oak-hickory cover types are especially important for
 production of wildlife and timber.  Loblolly pine is the most commercially valuable tree species in the region
 (Boyce et aL, 1986).  Other commercially valuable conifer species are slash, longleaf, and shortleaf pines.  The
 most valuable hardwoods are sweetgum ^iT'jdflnify'ff stvraciflua L.), yellow poplar (Liriodendron tuplioifera L.),
 white oak (Ouercus alba L.),  southern red oak (Q. falcata Michx.), and northern red oak (Q. rubra L.).

 Timber Production

      Approximately 90% (76 million ha) of southern forests is classified as commercial timberland (USDA,
 1982).  In 1976, approximately 44% of the 283 million m3 of softwoods and 50% of the 119 million m3 of
 hardwoods harvested in the U.S. came from  this region.  Most of this timber came  from Coastal  Plain and
 Piedmont forests (Boyce et aL, 1986).

      Southeastern forest ownership differs greatly from California and other western states (Figure 2).  Private
 individuals and nonforest industry companies own approximately  66% of the commercial forests (USDA, 1982).
 The forest indusry owns or leases 23%, and about 10% is in national forests or other public ownership (Alig et
 aL,1986).

      Approximately 59% of the annual harvest of softwood species (primarily pine) come from nonindustrial
 private forests (NIPF), 32% from the forest industry, 5% from national forests, and 4% from other sources (Boyce
 et aL, 1986).  Most of the  hardwood volume comes from NIPF and forest industry lands.

      A large amount of the current forestland in the South was once used for agriculture (Healy, 1985). It was
 abandoned as marginal cropland after the 1930s and was allowed to naturally regenerate or was planted to forests
 in the 1950s  through federal programs.  NIPF land  decreased 5% over the last 10  years; farmer ownership
 decreased 18%, while nonforest industry ownership increased 20%.  The area of forestland owned by public
 agencies has remained constant  Forest industry acquired nearly  half a million ha since 1977. Alig et aL (1986)
 estimated that forest industry acquisition of forestland may decline in the future and may be replaced with more
 intensive management on remaining lands and more long-term leases and cutting rights agreements with NIPF
 owners .

 Water

      Historically, abundant and well-distributed precipitation has adequately supplied public needs for water.
 Today, there  is growing  concern about the quality of future water supplies and how some forest practices, e^,
 harvesting, road building, and use of herbicides, might contribute to river sedimentation and water pollution (Boyce
 et aL, 1986).  Public perception that forests are important sources of public water supplies is nominal compared
 to California

 Recreation

      Although the economic importance of forest-based recreation in the South has not been fully assessed, it
is an important part of the lifestyle of most people. More than 40% of the public is involved in nonconsumptive
uses, e.g, wildlife observing, photography (Healy, 1985). More Southern adults hunt (29%) and fish (12%) than
the national average.  The Great Smoky Mountains National Park is the most visited national park in the country
(CDF, 1988). Recreational demand in the South is expected to increase 50 to 60% over the next two decades
(USDA, 1981).
                                                6-26

-------
                                                                                          Woodman

Wildlife

      Southern forests provide habitat for 170 species of amphibians, 720 species of birds, 204 species of mammals,
and 209 species  of reptiles (USDA, 1981).  Forest land conversion and harvesting of  old growth  forests has
decreased the area suitable for many species.  Conversion of bottomland hardwoods to agriculture and mixed
pine-hardwood forests to pine plantations are the most serious losses (Heaty, 1985).  Some endangered wildlife
species, e.&, the red-cockaded woodpecker, which nests hi old-growth pine forests, have been especially affected.

Factors ^ffectiny Forest Health and Productivity

      Wildfire. Although wildfires can occur during the entire year, most fires generally occur in the spring or
fall after long periods of dry air (Pyne, 1984). The frequency, severity, and social attitudes about fires and forests
are closely related to the cultural history of the South.  Fires were deliberately set in frontier days, and today in
some areas, for hunting, habitat and range improvement, and slash-and-burn subsistence  farming. From 1973 to
1978,55% of all fires were wfflfully set (incendiary) and 20% were caused by the burning of debris (USDA, 1980).

      Due to  unproved fire suppression technology, annual wildfire damage has decreased from 12 million ha
in the 1930s to 0.8 million in the 1980s  (Simard and Mam, 1987). Fire frequency influences the composition of
forestland:  frequent fires encourage shrub growth over pine;  less frequent  fire, pine over hardwoods; and
infrequent fire, hardwoods over pine.  Longleaf pine has been nearly completely replaced by loblolly and shortleaf,
which is favored by less frequent fire (Pyne, 1984).

      Insects and Diseases.  The most economically destructive forest insect is the Southern pine bark beetle
(Dendroctonus frjQDlaOs Zimmerman) (KaDcstein, 1981; Hedden, 1987). It attacks dense  pole- or sawtimber-size
pine trees which have usually been weakened by prolonged drought stress.

      The most serious diseases affecting pines are the following: fusiform rust (Cronartium ouercuum (Berk.)
Miyabe ex Shirai f. sp. fusiforme Burdsall & SnowX brown spot needle blight (Scirrhia acicola (Dearnb.) Siggers),
littleleaf Hi«»g!M» (Pnvtoohthora cjmjgQojni Rands), needle blights, pitch canker (Fusarium mnniliforme Sheld. var.
subghitinans Wollenw. & Reink.), and Annosus root rot (Heterobasidion aimosus (Fr.) Bref.).  Fusiform rust and
brown spot blight damage and/or loll trees under 10 yean old. Needle blight, pitch canker,  and Annosus root rot
generally weaken trees by injuring needles, branches, and reproductive structures. Warm, moist weather conditions
favor the spread of all of these diseases.

      The most serious insect and diseases impacting hardwood forests are hardwood borers (Stvloxus spp.) and
oak dieback or decline (Sphaeropsis auercina Cke. & EH).  Hardwood borers attack and kill twigs and branches
on oaks.  Unfavorable weather conditions predispose red oak, chestnut oak, and other oak species to oak decline,
which lolls twigs, branches, and leaves of infected trees.

      Soil Fertility. Soil fertility is a major factor in the productivity of southern forests (Allen and Morris, 1983).
Most soils tend  to be  relatively old and well developed, to be highly acidic,  and to contain relatively  few
weatherable minerals. They also tend to be low in calcium and magnesium, with most organic matter on the
surface (Gregory, 1983). Nitrogen and phosphorus are the nutrients which limit growth on most Southern soils
(Allen and Morris, 1983). Many  Piedmont and Upland soils have been severely eroded.

      Forest Productivity. One of the  main reasons that the South is the most important timber region in the
country is the high growth rates of southern pines.  Unmanaged, naturally regenerated, well-stocked stands of
loblolly, slash, and shortleaf pine on average soils produce 2000 to 2500 m3/ha of wood in a  30-year rotation (time
from regeneration to harvest) (Schumacher and Cofle, 1960). Unmanaged and managed plantations grow 40 to
125% more wood than comparable natural stands (Cofle and Schumacher, 1964; Gutter et  aL, 1983). Plantations
established on "old fields" (abandoned farmland) produce  20 to 50% more wood than forested sites which were
never farmed (Cofle and Schumacher, 1964). High growth rates, short rotations, and low harvesting costs offer
forest owners the highest financial returns on forest land of any region in the country.
                                                 6-27

-------
 Woodman

       The region's standing hardwood resource has been yvreiunnf* in all subregions except the Mississippi Delta
 (Boyce et aL, 1986).  Annual net growth of hardwood forests is 22 times  greater than the rate of harvest.
 Approximately two-thirds of the current total annual growth is in small trees less than 50 years old.

       Boyce et aL (1986) stated that many private landowners who invest in pine plantations do so on the basis
 of 'hope for high financial rewards rather than on historical certainty.  Based on the 1967 producer price index, the
 noninflated price paid for hardwood and pine pulpwood and sawtimber has declined or had no major appreciation
 since 1955 (Skog and Risbrudt, 1982).  This was attributed to an oversupply of low-cost  natural timber and
 advances  in manufacturing  technology which  permitted utilization of smaller trees  and wood waste  from
 manufacturing processes.

       Recent forecasts of shortages in future softwood supplies in parts of Virginia, North Carolina, and Georgia
 has been called the "pine regeneration" problem (Heaty, 1985). It has been attributed mainly to inadequate pine
 reforestation on NIPF lands.  In addition, although an increasing amount of land is being planted to pines, the total
 hectares in pine is declining (Knight, 1985). Many harvested pine forests are naturally succeeding to oak-pine
 and upland hardwood forest types because of inadequate regeneration.  Other causes are the conversion of pine
 forestland to  agriculture and other  uses, higher  than average mortality rates in natural pine forests,  and  an
 apparent slowdown in growth of some forests (Sheffield et aL, 1985).

       The determination of a growth slowdown was based on an analysis of sample trees in U.S. Forest  Service
 inventory plots in five Southeast states (Sheffield et aL, 1985). Some forests grew 15% to 25% less in radial growth
 over the last one or two  decades than in previous periods. Most of these sample plots were in natural pine and
 hardwood forests on NIPF land. Plots in industrial and public forests did not show the same decreases. Although
 many possible causes were suggested, analyses on some of the data indicated that 70% of the decrease could be
 attributed to older trees, higher tree densities, and more droughts than in the past (Zahner and Myers, 1986; 1987).
 Another explanation was that the earliest growth rates used in the comparison were biased by a high percentage
 of abnormally faster growing trees in "old field" plantations.

 Trends in Fors
       Expectations of high financial returns have induced many industrial and public owners to practice some of
 the most intensive forest management of any region in the world (Boyce et aL, 1986). Standard cultural practices
 include piling and/or burning logging debris after harvesting, controlling of sprouting hardwoods with cultivation
 or herbicides before planting, and planting pine seedlings which have genetically improved growth rates, disease
 resistance, and wood quality (Daniel et aL, 1979; Boyce et aL, 1986; Smith, 1986). Many 2- to 7-year plantations
 are "weeded" with herbicides or prescribed burned to reduce hardwood competition (Boyce et aL, 1986).  The
 current 9+ million ha of pine plantations is projected to double by the year 2030 (Alig et aL, 1986). Low-cost
 natural regeneration methods are  used when owners choose not  to invest  in plantations (Boyce et aL, 1986).
 Although longer time is  needed to produce  merchantable size  trees, natural regeneration and inexpensive
 prescribed burning provide a financially attractive alternative (Boyce et aL, 1986; Smith, 1986).

      Most hardwood forests are regenerated naturally because their slower growth rates and lower log values
 produce relatively small fliw\*\ returns. Almost all important hardwood species can regenerate well from seed
 or sprouts. Management of high-value hardwood species, e.g,  black walnut, white and red oak species, require
 a series of selective cuts which remove species of lower value (Smith, 1986).

      All Southern states provide some funding and cooperative assistance to private forest owners hi fire control,
insect and disease control, forest management planning, and sflvkultural research (Boyce et aL, 1986).  General
land-use planning and regulation have been limited due to public concerns against unnecessary infringement of
personal property rights (Healy, 1985). Only Mississippi and Virginia have laws which regulate reforestation of
private land after timber harvesting.  These statutes are less  comprehensive and  restrictive than the laws in
California and other western states.
                                                 6-28

-------
                                                                                         Woodman
SOCIOECONOMIC TRENDS AFFECTING FORESTS

      The 12 Southeastern states are home to just less than one-fourth of the total U.S. population (Healy, 1985).
The region  experienced more than a 40% increase  in population from 1960 to 1980, accompanied by an
approximate doubling of per capita real income (Alig et aL, 1986).  The population is expected to increase 31%
between 1980 to 2000, with the greatest increase in Florida (79%) and lowest in Alabama (13%) (Healy, 1985).

      The region has many social and economic distinctions.  Healy (1985) cited the South's low wage structure
and lack of unionization, low living costs, changing racial and social attitudes, and warm winter climate as reasons
for its population and economic growth of the last few decades. The average personal income per capita is below
the national  average in all Southern states except Virginia, and a greater percentage of Southerners than average
live below the national poverty level (U JS. Bureau of Census, 1986).

      The South has historically had a very large rural population compared  to the national average (Healy,
1985).  Although this is still the case, the rural population has shifted dramatically from fanners  to nonfann
residents.  Agriculture employed only 32% of the labor force in 1986 (USDA,  1987). The urban population is
distributed among many small cities or nonsuburban towns (with less than 500,000 people) and a few large cities,
such as Atlanta, Miami, Tampa, and New Orleans. Urban densities are very low, with a much greater than average
amount of land per urban resident

      Many Southerners are politically conservative, placing great importance on state and individual rights and
protection of those rights from infringement Compared to California, public interest groups concerned about
forest environmental issues have not been successful in influencing many public policy issues. This trend will most
likely reverse as environmental issues come ingraarin^y to the forefront and education levels foster a more
informed public

      The annual value of the forest resource in the South has been  approximated at $63 billion (Marx et aL,
1986).  The estimated value of forests in the four leading states are as follows:  Georgia ($8.6 billion), Arkansas
($83 billionX Florida ($7.4 billionX and Alabama ($6.5 billion).  Historically,  the area of forestland has been
related to the economic health of agriculture (Healy, 1985).  When actual or expected farm incomes were high,
forestland was cut and used for crops. When farming became marginal or unprofitable, the land was abandoned
and reverted to "old field" natural pine or pine-hardwood forests.  The amount of  land in forests decreased 1.8
million ha between 1977 to 1985 (Alig et aL, 1986). One-third of this area was in bottomland hardwoods (Boyce
etaL.1986).

      Sodoeconomic trends related to forestry over the next 50 years without  climate change includes the
following:

        increases in softwood prices (USDA, 1982);

        relatively small increases in the area of forestland owned by the forest industry (Alig et aL, 1986);

        increased  demands for lumber, plywood, paper, and other wood products (USDA, 1982);

        increased  opportunities for the export of wood products from the South (USDA, 1982);

        increased  hardwood biomass compared to softwood (Boyce et aL, 1986);

        increases in public pressure to manage public and private forestland for multiple uses rather than only
        timber (Boyce et aL, 1986);
                                                6-29

-------
 Woodman


         losses of total forestland to urbanization and agriculture (Alig et aL, 1986); and

         increased need for state and local incentive programs to increase management of most valuable tree
         species on private lands (Healy, 1985).


 PROJECTED CLIMATE CHANGE SCENARIOS

       The three  hypothetical  climate change scenarios developed for the South included average monthly
 temperature increases of 2.0*  to 7.7*C, depending on month and location.   Monthly temperatures from the
 OSU-based scenario were 0.5* to US' lower than the other scenarios. The GFDL scenario projected temperatures
 0.4* to 1.0° higher than the GISS scenario over most of the Coastal Plain and Upland regions. The GISS scenario
 was 0.4° warmer than the GFDL in the Piedmont

       The GFDL climate scenario was the only one with an overall average decrease in monthly rainfall.  The
 GISS scenario predicted average monthly increases of 2 mm, 12 mm, and 7 mm in the Coastal Plain, Piedmont,
 and Uplands regions, respectively. The OSU scenario projected average increases of 7 mm in the Coastal Plain,
 6 mm in the Piedmont,  and 1 mm in the  Uplands.  The drier GFDL scenario had reductions of 6 mm in the
 Coastal Plain, 13 mm  in the Piedmont, and remained unchanged in the Uplands.

       Monthly precipitation changes averaged over the year obscure the seasonal variability which  is most
 important to the growth and survival of Southern forests. Estimates of percent changes in winter, spring, and
 summer precipitation for eight representative climate divisions are shown in Figure 4. In all locations, the GISS
 scenario had less than average monthly precipitation in winter and  more than average in summer.   Smaller
 increases in spring precipitation were postulated for all but two Gulf Coastal Plain sites.  The  GFDL scenario
 presumed much drier summers in all locations except east Texas. GFDL winter and spring rainfall were higher
 than the historical average except in Florida.  In six divisions, the OSU scenario  projected less than  average
 precipitation in whiter and spring, and more than average summer rain in all locations.

      From a forest effects perspective, less than historic precipitation in winter and spring would probably have
 the greatest negative impacts on forest regeneration and seedling  growth.  Drier  summers would most likely
 reduce forest growth and predispose trees to attacks from insects and root diseases. None of these scenarios was
 consistently worse or better than others.  OSU*s drier winters  with drier springs make conditions harsher for
 forests in five of the eight locations.  Wetter winters and wetter springs in the GFDL scenario would be the least
 harmful in five divisions.  GISS results are the most variable among the sites, and potential impacts would depend
 on the specified conditions at each location.


 POTENTIAL IMPACTS ON SPECIES COMPOSITION AND FOREST PRODUCTIVITY

 Species Composition

      Climate change will decrease or increase species diversity depending on the ecosystem, the genetic diversity
 of the species, the migration rates of new species, and many other factors (Layser, 1980).  Using a relatively hot
 and dry climate scenario and results from a forest succession analysis by Solomon et aL (1984), Miller et aL (1987)
 concluded that the natural range of loblolly pine would move  northward and increase  in area by 16,000 ha.
Although no  time frame was emphasized, the process would require many centuries. The northern limits of
 loblolly are controlled by the frequency and severity of spring frosts  (Powells, 1965). The projected loss of area
 in East Texas, Oklahoma, Arkansas, and Coastal Plain was based on severe drought conditions projected by this
scenario. None of the  scenarios described above project conditions that are as severe. No compensating effects
of higher CO2 concentrations were considered.
                                               6-30

-------
                                                                                          Woodman

      Urban and Shugart (this volume) applied the GISS scenario to their mixed-species forest succession model
in order to evaluate long-term climate  change on an oak-pine forest type in eastern Tennessee. Assuming no
effects from higher CO2 or migration of new species into the ecosystem, they concluded that the warmer and drier
conditions would lead to the elimination of chestnut oak, black oak, and pine species from the forest type. Using
another climate change scenario, Field et aL (1988) evaluated potential long-term impacts on wetland ecosystems
in the Tennessee  River basin.  They concluded that the warm-and-wetter   scenario would eliminate less
water-tolerant species due to a higher frequency of spring flooding.

      One of the cover types most likely to be negatively affected by increased temperatures is the Fraser fir-red
spruce type found above 1400 m elevation in the Southern Appalachian Mountains. Delcourt and Delcourt (1984)
suggested that present climate warming has limited the re-expansion of this cover type since the late Holocene.
Higher temperatures would most likely severely limit or eliminate the natural range of Fraser fir, red spruce, and
other colder climate species in the southern mountains.

      It is apparent that changes hi forest species composition will be very region-specific  Changes in species
will be influenced by the actual scenario of climate change which comes into play in a particular region, especially
actual changes in precipitation,  and local topographic and  soil conditions.

Forest Productivity

      Most existing forests will probably experience some decrease in productivity due to higher temperatures,
if higher CO2 concentrations and/or more moist conditions in the  spring and summer do not compensate for
higher respiration  and evapotranspiration rates. The potential magnitude of losses cannot be estimated at this
time. P-«kHng growth and yield models are reliable only for forests grown in historical climate conditions.  None
of the forest succession simulation models were designed  for growth and yield estimations (Solomon and West,
1987). Unmanaged older forests will be most vulnerable to decreased productivity.

      Decreases in productivity and species diversity may be compensated for by planting tree species which are
better suited to wanner climate, e^, Caribbean pine, various Mexican pines, and  several eucalyptus species.
KeQison and Weir (1987) suggested that the genetic diversity of current species is large enough to  select and
propagate families and clones of trees which wfll grow well in a new climate.

Associated Effects

      Sea Level Rise - A large area of southern forests grow on the Coastal Plain on soils that require drainage
in order to be productive.  Rises in sea level would impact  an unknown portion of these forests through increased
flooding, saltwater intrusion, and ineffectiveness of drainage systems. The areas of highest risk are the northeast
North Carolina coast and the southern tip of Florida because of their low topography.

      Wildfires - Increased temperatures, if not compensated for by increased precipitation, will undoubtedly
increase losses of forests to wildfires. Assuming no change in the distribution of precipitation, Simard and Main
(1987) estimated than an average temperature increase of 2.4'C and a 4% decrease in annual precipitation would
increase  wildfire occurrence in the South by 8%.  Afl  of the current GCM climate models project higher
temperatures (Kellogg, 1987). As shown above, some locations in the region may become drier and others wetter.
The OSU scenario, which projects dry  springs, would severely increase fire danger.

      Insects and Diseases.   The potential effects of climate change  on southern pests will depend on  the
magnitude of temperature increases and changes in moisture conditions. Higher temperatures will generally favor
the northward expansion of the current range of most  insects and dfe»am.  Warmer temperatures will probably
increase the population levels and range of southern pine beetles (Kalkstein, 1981X and Nantucket pine tip moth
in some areas (Hedden, 1987).  Drier conditions would favor decreases in severity of brown spot and rust diseases
but increasd losses from littleleaf disease and annosus  root rot In general, forest pest problems will be greatest
in stands that contain trees which are stressed or of low vigor (Hedden, 1987).
                                                 6-31

-------
 Woodman

 Potential Impacts On SocJoeconomic Systems and Policies

       Almost any climate change will heighten future public concerns about land use.  Projected changes in the
 area and productivity of agricultural land in the United States and other areas in the world (Cooper, 1982) will
 provide a stimulus to convert the South's most productive forestland to agricultural production in areas where
 water supplies and soil conditions permit Since much of the most productive land is now used for pine production,
 the loss of this land will contribute to further shortages in the long-term supply of softwood in the region. If this
 occurs, federal and state policymakers will need more comprehensive land-use planning, state and regional land
 use zoning, and new  forest protection and incentive legislation.

       Private forest  owners  will face increasing pressure to use their land for agriculture or to establish and
 manage pine plantations to meet expected shortages in softwood. If production of food or wood declines in other
 areas, prices and the  potential profitability of land ownership will undoubtedly increase in the South. The needs
 of NIPF owners for short-term profits will likely influence them to sell or use their land for food crops. Nonforest
 industry owners, who  have purchased forestland for investment purposes, will likewise choose the alternative which
 gives them the highest economic return.

       Climate change will not likely impact the southern forest industry as quickly as other private owners because
 of long-term strategic planning, intensive management practices, and shorter rotations between regeneration and
 harvesting (Malac, 1987). Industry managers will  most likely sell land with the lowest potential financial returns
 and intensify management on their best  land. Those paper and solid wood products companies which depend on
 large supplies of softwood for their business will probably buy more land and intensify management practices on
 their existing land in  order to ensure their future profits.

       Given the  additional losses of pine forests because of climate change,  future softwood supplies are not
 likely to meet demands in many areas of the South. Many companies will be  forced to use available hardwood
 resources, move their manufacturing facilities to locations with favorable softwood supplies, or cease operations.
 Companies which are very dependent on wood from NIPF and public forests will be most vulnerable to decreased
 wood supplies and increased  prices.  These and other companies with timberland may decide to move out
 of the South when their mills become technologically inefficient  and greater profits can be made by investing in
 new facilities or land  in other forest regions.

       Cubbage et aL (1987)  outlined the possible economic impacts of global climate change on commercial
 forests and forest-based industries.  Using a 10%  decrease in loblolly pine  volume production, they estimated a
 $98 million per year loss in average cost  of standing timber sold to a buyer.  Further economic losses could occur
 from decreased quality of pulpwood (because of lower specific gravity of wood produced), shifts  in forest
 management  practices, and relocation of pulp and saw mills to  more economically favorable locations in other
 regions.   Southern forestry economic models assume that all  direct-value-generated   contributions  of forest
 products have an indirect multiplier effect of 13 on other economic sectors  (Cubbage et aL, 1987). Thus, effects
 on the forest industry would have significant  impacts on other aspects of southern economic life.

      Given the size, location, and policies controlling management of public forestlands, public forests would
 be impacted in the same ways  as described for public ownership in California, Climate change and current federal
 agency policies will emphasize more long-range planning, multiple-use management, and less emphasis on timber
 production. A shrinkage in private forestland wfll increase recreational uses, watershed protection, and wildlife
 management activities in public forests.

      The current downward  trend in forest-based employment opportunities will also be accelerated by climate
 change. Projected decreases in the area of commercial forests, reduction in the number of manufacturing facilities,
 and more automated mills will reduce future labor requirements.

      Climate change  will increase the need for highly trained foresters and forest managers. Increased complexity
and difficulty in meeting forestland owner needs will require greater knowledge about forest biology, new forest
management regimes, and a greater environmental consciousness in the future.


                                                 6-32

-------
                                                                                          Woodman


      Climate change  will likely accelerate concern about most of the  environmental issues described for
California.  These concerns include the adequacy of future water supplies, recreation, and wildlife habitat. Water
quality and quantity will become a major concern as ground and surface water supplies become used more in
irrigated farming. As the public becomes better informed about environmental issues, including the value of forests
in ameliorating potential climate change impacts and reducing atmospheric CO2 concentrations, policymakers will
face more pressure to regulate practices on all forestland.


CONCLUSIONS

      Higher temperatures, with or without increased precipitation, will most likely decrease the productivity of
most Southeastern forests. Some Uplands forests which are now limited by low temperatures and short growing
seasons would become more productive.  The timing and relative magnitude of changes in forest productivity
cannot be estimated  at this time.  Depending on the forest type and actual changes in climate, most impacts on
forests wfll probably be gradual and occur over several centuries. Given present knowledge on species and trends
in forest productivity, climate change wfll tend to improve hardwood productivity but decrease softwood production.
Current trends toward converting the more productive pine land to agriculture win be accelerated, which would
further decrease  softwood supplies in the region.  Forest  industries dependent on large supplies of low-cost
softwood will face increasing economic pressures to modernize their manufacturing facilities, increase the supply
of softwood from their lands, purchase more land, or move to other regions with more favorable wood supplies.
Potentially major losses of forestland to agriculture and urbanization would likely cause major price increases for
timber, more comprehensive land-use planning and zoning, and greater regulation of forest practices on all private
and public lands.
                                                 6-33

-------
 Woodman
                                              CHAPTERS

                                   POTENTIAL POLICY CONCERNS
 KEY ASSUMPTIONS
       The scientific basis for climate change rests on measured increases in greenhouse gases, projected emissions
 of these gases now and in the future, conceptual understanding of the global carbon cycle, and estimates from
 first-generation global circulation models (MAS, 1983). Although most knowledgeable scientists agree that global
 temperatures will rise regardless of actions taken now to control future emissions of these gases, significant
 uncertainties exist in all aspects of the issue. Likewise, even if climate does change according to one of the climate
 scenarios, there  is considerable uncertainty about the potential impacts on the biology of forest ecosystems and
 sotioeconomic trends in the United States and world.

       From the perspective of potential effects on forest ecosystems and the benefits or values which the American
 people receive from forests, the most pressing immediate questions about climate change are: when? where? how
 much? and what can we. da about it?  The answers to these critical questions cannot be provided now with any
 degree of certainty or acceptable risk.  We win  need new information on the  relationship between forest
 ecosystems and climate and weather systems, new and different forest management systems designed to reduce
 climate change impacts on the most important forest resources in a region, and development of regional, national
 and international sodoeconomic models that can simulate or project how various scenarios of climate change might
 impact the goods and services derived from forests.

       A major premise of this review is that public policy implications of global climate change on U.S. forests
 cannot be determined without assessing  the potential effects on a region-by-region basis. Each climate-change
 scenario varies by region, so that the relative effect is region-dependent  The forests in each region contain
 different tree species, different soil and physiographic conditions, differing intensities of forest management, and
 current limitations of climate which will greatly influence their short- and long-term response to dimate change.
 Potential changes in economic benefits from forests wifl also vary greatly between regions because of differences
 in forest ownership, the contribution of forest-derived values to regional economies, and opportunities for using
 forested land for agriculture.

      A second major premise is that climate change impacts on a region's sodoeconomic system will be a more
 dominant factor  in determining needs for forest policy changes than direct effects on spedes diversity or forest
 productivity. A  key factor in determining regional sodoeconomic impacts is related to forest ownership. The
 amount of forestland in private or public ownership will influence both the nature of impacts and the manner in
 which public policy may alter these impacts. Federal forestlands already have a legislated policy which can rapidly
 respond to sodoeconomic needs on a regional basis. Private forest owners are most likely to respond in proportion
 to impacts on the profits they derive from their forests. State and local government policymakers are most likely
 to legislate  new  policy or regulations concerning forest  effects  when forest-based tax revenues and socially
 important values derived from forests need protection.


 POLICY ISSUES

      The California and Southeast assessments suggest how different forest ownership patterns,  historic uses
of forests, and population  demographics might interact with climate change to influence future public policy in
those  regions.  Some limited extrapolations of these trends will apply to other forest regions:

        Public forestland managers will be more sensitive to public concerns and impacts on "noneconomic" values
        than private owners.  Private owners will be more responsive to changes in ffrv»n«ai effects, availability
                                                6-34

-------
                                                                                           Woodman
        of low-interest capital, and the risks associated with long-term capital investments in forests, which will
        increase in an unstable and less favorable climate.

        Federal, state, and local policy makers will respond to impacts of climate change in proportion to the
        expressed concerns of citizen's groups, forest industry, and local governments. In states like California,
        where forests contribute relatively little direct economic value  to the state's economy,  shortages in
        nonvalued forest resources, e^, recreation and water, will be more important factors in determining
        public policy than losses in timber supplies, forestry employment, or taxes.

        Potential impacts on future wood supplies and related sorioeconomic values will most  likely determine
        public policy in regions where forest-based resources are more  important to the regional  and state
        economics.

      One of the greatest uncertainties affecting  future public policy will be the need for converting suitable
forestland into land for agricultural production. This need will intensify as climate change*. Policymakers will have
to consider new or different land-use policies to encourage and protect forest management on private forestlands
that are not  suitable  for agriculture. This wfll require intensive planning and zoning of lands for these purposes.

      Another major factor which wfll influence future public policy is the dependence of the forest industry on
NTPF timber. Inadequate regeneration and lack of intensive management on these lands has already resulted in
projections of shortages in softwood in small areas in the South, Northeast, and Lakes States regions.  If this trend
is not reversed, the consequences  of tins  problem on the  industry and economies in  those  areas  will be
exacerbated further by climate change.  This suggests that new incentive programs and forest protection laws must
be considered to promote better regeneration practices and protection of private forests.

      Other potentially important public policy questions are related to climate change:

        To what extent should the federal government help protect private forestland owners from catastrophic
        losses due to climate change?

        To what extent should the public underwrite credit for long-term investments on private forestlands given
        the nature of higher risks and uncertainties of forests achieving merchantable sizes in an acceptable length
        of time?

        Who wfll pay for the costly long-term research needed to answer critical policy questions and provide new
        technologies to ameliorate impacts on sodoeconomk systems?


RESEARCH NEEDS

      One of the highest priority research needs is for reliable and precise estimates of future climate  change on
a regional or local basis.  Predictions  should include expected frequency and severity of departure from mean
monthly values, estimates of potential daily evapotranspiration, daily solar radiation, potential risk of  freezing
temperatures in early spring and fall, and the number of high temperature days in mid-winter.

      Of equal importance is an unproved understanding of how major forest ecosystems will respond to
concurrent  increases in  atmospheric CO* higher  temperatures,  and changes in seasonal distribution of
precipitation.  Most  of the research applicable to these issues has involved small plants over short time  periods.
Research is needed to be able to scale up from single plants to forest stands and ecosystems. It is essential that
we be able  to distinguish between effects on  forest  productivity that are related to climate change and those
effects caused by variability in plant competition, different tree stocking levels, and normal environmental stresses.
                                                 6-35

-------
Woodman


      Another very important research need is to determine the extent to which potential growth reductions and
impacts on species diversity due to climate change might be compensated for by enhanced levels of CO2. Such
studies will require long-term experiments with plants growing under continuous high CO2 exposures, along with
increased temperatures and different water regimes.  Information of differential effects on forest tree species,
especially hardwoods vs. conifers, is vital to the prediction of species composition of natural forest ecosystems and
potential effects on the forest industry.


CONCLUSIONS

      This review and assessment can only identify some of the emerging issues, trends, and questions about the
potential effects of global climate change on American forests.  The  need for more in-depth analyses  and
information from new research is obvious.  Undoubtedly,  the extent to which public policy makers fund new
research, pass new laws, and deal with forest-related issues will be determined by its importance within the context
of other problems related to climate change. Legislators wfll be faced with major funding needs related to effects
of sea level rise on coastal cities, needs for irrigation development to offset  decreased crop productivity, more
nonpolluting electric power generation facilities, etc  Although the overall effects of climate change on forest
ecosystems and associated goods and values may not be fully realized for a number of decades, we must begin now
to improve our present understanding of the biological and sodoeconomic factors that will determine those effects
and allow the assessment of their importance in relation to other public  policy issues.
                                                6-36

-------
                                                                                        Woodman


                                          REFERENCES


Alig, RJ., HA. Knight, and RA. Birdsey. 1986. Recent area  changes in southern forest ownerships and cover
types. USDA For. Ser. Res. Paper SE-260. Ashevffle, NC. 10 pp.

Allen, HJ~, and LA. Morris. 1983. Nutritional management of forest stands. In North Carolina State Forest
Fertilization Cooperative Forest Soils Shortcourse. School of Forest Resources, NCSU. pp 119-128.

Barrett, J.W. (ed). 1980. Regional Silviculture of the United  States, 2nd edition. Wiley & Sons, New York. 551
pp.

Benzie, J.W, AA. Aim, T.W. Curtin, and C Merritt 1986.  Silviculture - the past 30 years, the next 30 years: Part
V. the North Central Region. J. For. 84(8)35-41

Bemabo, J.C 198L Quantitative estimates of temperature changes over the last 2700 years in Michigan based on
pollen data. Quat Res. 15:143-159.

Bernabo, J.C. and T. Webb. 1977.  Changing patterns in the Holocene pollen record of northeastern North
America: a mapped  summary. Quat Res. 8:64-96.

Binkley, CS. 1983. Private forest land use: status, trends, and projections. In Nonindustrial Private Forests: A
Review of Economic and Policy Studies. Eds. JJ». Royer and CD. Risbrudt Sch. of Forestry and Environ.
Studies, Duke U., Durham, NC. pp. 51-70.

Birch, T.C 1983. Private forestiand owners in the U.S.: their numbers and characteristics. In Nonindustrial Private
Forests: A Review of Economic and  Policy Studies. Eds.  JJ*. Royer and CD. Risbrudt Sch. of Forestry and
Environ. Studies, Duke U., Durham, NC pp. 71-75.

Bolsinger, CL. 1980. California forests: trends, problems and opportunities. USDA For. Serv. Res. Bull PNW-89.
138pp.

Boyce, LS. 1948. Forest Pathology. McGraw-Hill Co, New York. 550 pp.

Boyce, S.G., E.C Burkhardt, R.C Kellison, and D.H. Van Lear.  1986. Silviculture - the past 30 years, the next
30 years:  Part  UL the South. J. For. 84(6):41-48.

Boyer, J.N. and D.B. South. 1984. Forest nursery practice in the South. South. J. AppL For. 8:67-75.

Britannica. 1982. The New Encyclopaedia Britannica. Encyclopaedia Britannica, Inc, Chicago.

CDF. 1988. California's Forest and Rangelands: Growing  Conflict Over Changing Uses. Calif. Div. of Forestry
and Fire Protection, Forest and Rangeland Resources Assessment and Policy Act  Committee, Sacramento, CA.

Chapin, F.S., m, AJ. Bloom, CB. Field, and RJi. Waring.  1987.  Plant responses to multiple environmental
factors. BioScience  37(l):49-57.

Gutter, J.L, J.C Fortson, L.V. Pienaar, G. H. Blister, and  Ri. Bailey. 1983. Timber Management:  A
Quantitative Approach.  John Wiley & Son, New York. 333 pp.

Coile, T5. and FJC Schumacher. 1964. Soil-Site Relations, Stand Structure, and Yields of Slash and Loblolly Pine
Plantations In The Southern United States. T.S. Coile, Inc. Durham, NC. 296 pp.
                                                6-37

-------
 Woodman

 Cook, E.R., AJL Johnson, and T J. Biasing. 1987. Forest decline: modeling the effect of climate in tree rings. Tree
 Physiology  3(1):27-40.

 Cooper, CF.1982. Food and fiber in a world of increasing carbon dioxide. In Carbon Dioxide Review 1982, Ed,
 W.C Clark. Oxford  Univ. Press, New York, pp 297-319.

 Cubbage, F.W., D.G. Hodges, and Ji. Regens. 1987. Economic  implications of climate change impacts on forestry
 in the South. In Proceedings of the Symposium on Climate Change in the Southern United States: Future Impacts
 and Present Policy Issues. May  28-29, New Orleans, La. pp.266-279.

 Dahlman, R.C, BJt Strain, and HJEL  Rogers. 1985. Research  on  the response of vegetation to elevated
 atmospheric carbon  dioxide. JJinviron. QuaL 14(l):l-8.

 Daniel, T.W., JA. Helms, and F.S. Baker. 1979. Principles of Silviculture. McGraw-Hill Book Co., NY. 500 pp.

 Davis, LSn R. Marose, and LJ. DeLain. 1986. CALPLAN: A model to simulate outputs from California's forests
 and rangelands under alternative futures. In Proc of The 1985 Sympos. On Systems Analysis In Forest Resources;
 Eds. PJL Dress and R.C Field. Georgia Center for Continuing Education, U. of Georgia, Athens, GA. pp. 29-42.

 Deeming, JJL, RJL Burgan, and J JX Cohen. 1977. The National Fire Danger Rating System-1978. USDA Forest
 Service Gen. Tech. Rep INT-39. Intel-mountain. For. and Range Expt Stn, Ogden, Utah. 63 pp.

 Delcourt, HJL and PA. Delcourt. 1984. Late-quaternary history  of the spruce-fir ecosystem in the southern
 Appalachian mountain  region. In The Southern Appalachian Spruce-Fir Ecosystem: Its Biology and Threats^ Ed,
 PJS. White. US Dept of Interior, National Park Service, Research/Resources Management Report SER-7L pp.
 22-35.

 Field, RJn NA. Nielsen, and R.T. Men. 1988. The future of freshwater wetlands in the southeast United States
 - potential  impacts of the greenhouse warming effect US EPA.

 Powells, HA. (compiler). 1965. Silvics of Forest Trees  of the  United States. USDA Agric. Handbook No. 271,
 USDA Forest Service,  Wash, DC 762 pp.

 Franklin, J.F., HJL Shugart, and ME. Harmon. 1987. Tree death  as an ecological process. BioScience 37(8):
 550-556.

 Fried, J.S., KA. Surano, P.F. Daley, JJL Shinn, and P.  Anderson. 1986. Biomass production and nutrient
 responses of ponderosa pine to long-term elevated CO2 concentrations. In Proc of Ninth North American Forest
 Biology Workshop, Agricultural  Conf. Oklahoma State Un StiUwater, OK. pp. 11-18.

 Fritts, H.C. 1976. Tree Rings and Climate. Academic Press, London. 567 pp.

 Furniss, RJL and VM. Carolin.  1977. Western Forest Insects.  USDA For. Serv. Misc. Pub. No. 1339. 654 pp.
 Gajewski, K. 1988. Late Holocene climate changes in eastern North America estimated from pollen data. QuaL
 Res. 29:255-262.

 Geiger, R. 1965. The  Climate Near The Ground Harvard Univ. Press, Cambridge, Mass. 611 pp.

 Gibbs, M J. and J.S. Hoffman. 1987. An approach for generating climate change hypotheticals given limitations
 in current climate  models.  In The Greenhouse Effect, Climate Change, and VS. Forests, Eds. W£. Shands and
J.S. Hoffman. The Conservation  Foundation, Wask, DC. pp. 91-112.

Gregory, J J). 1983. Southern forest soils. In North Carolina State Forest Fertilization Cooperative Forest Soils
Shortcourse. School of Forest Resources, NCSU. pp 41-60.


                                               6-38

-------
                                                                                       Woodman


Hansen, J, D. Johnson, A. Lads, S. Lebedeff, P. Lee, D. Rind, and G. Russell 1981. Climate impact of increasing
atmospheric carbon dioxide. Science 213:957-966.

Heaty, R.G. 1985. Competition for Land in the American South:  Agriculture, Human Settlement, and the
Environment The Conservation Foundation, WaslL, DC. 333 pp.

Hedden, R. 1987. Impact of climate change on forest insect pests.  In Proc. of Sympos. on Climate Change In The
Southern United States: Future Impacts and Present Policy Issues, EA, M. Meo. U.  of Oklahoma, May 22-29,
1987. pp. 309-328.

Hepting,  G M. 1963. Climate and forest diseases. Ann. Rev. Phytopath. 1:31-50.

Holdridge, LJR. 1947. Determination of world plant formations  from simple climatic data. Science 105:367-368.

Houpis, JXJn KA. Surano, PJF.Daley, and JJLShinn. 1986. Growth and morphology of Pinus ponderosa seedlings
exposed to long-term elevated atmospheric carbon dioxide concentrations. In Proc. of Ninth North American
Forest Biology Workshop,  Agricultural Conf. Oklahoma State U, Stillwater, OK. pp. 19-26.

Houpis, JXJ? KA. Surano, S. Cowles, and JJL Shinn. 1988. Chlorophyll and carotenoid concentrations in two
varieties of Pinus ponderosa seedlings subjected to long-term elevated carbon dioxide. Tree Physiology (In Press)

Kalkstein, LS. 198L Differential response of loblolly pines  to climatic stress. Prof. Geographer 33(1>.122-128.

Karl, TJR., UK. Metcalf, M.L. Nicodemus, and R.G. Quayte. 1983. Statewide Average Climatic History: North
Carolina, 1887-1982; Historical Climatology Series 6-1, NOAA, National Climatic Data Center, Asheville, NC.
35pp.

KeUison,  R.C and RJ. Weir. 1987. Breeding strategies in forest  tree populations to buffer against elevated
atmospheric carbon dioxide levels. In The Greenhouse Effect, Climate Change, and U.S. Forests, Eds^ WJE.
Shands and J.S. Hoffman. The  Conservation Foundation, WaslL, DC pp. 285-294.

Kellogg, W.W. 1987. Future climate scenarios for the southern  United States. In Proc. of Sympos. on Climate
Change In The Southern United States: Future Impacts and Present Policy Issues, Ed, M. Meo. U. of Oklahoma,
May 22-29,1987. pp. 1-21

Knight, HA. 1985. Southern U.S. timber supplies. Proc. 3rd North Amer. IIASA Network Meeting. USDA For.
Serv. SE  For. Exp. Sto, Asheville, NC, 11 pp.

Kramer,  P J. 1980. Drought, stress, and the origin of adaptations. In Adaptation of Plants to Water and High
Temperature Stress, Eds.  Turner, N.C and P J. Kramer. John WOey  & Sons, New York, pp.7-20.

Kramer, P J. and N. Sionit 1987. Effects of increasing carbon dioxide concentration on the physiology and growth
of forest  trees. In The Greenhouse Effect, Climate Change,  and U.S.  Forests, EoX W.E. Shands and J.S.
Hoffman. The Conservation Foundation, WaslL, DC pp. 219-246.

Lavender, D.P., RJK. Hermann, and D. McCreary. 1986. Chilling requirements of Douglas-fir. In Proc of Ninth
North American Forest Biology Workshop, Agricultural Conf. Oklahoma State  U, Stillwater, OK. pp.  85-93.

Layser, EJ7.1980. Forestry and climate change. J. Forestry pp. 678-681

Leverenz, J.W. and D J. Lev. 1987. Effects of carbon dioxide-induced climate changes  on the natural ranges of
six major commercial tree species in the western United States. In The Greenhouse  Effect, Climate Change,
and US. Forests, Eds. WJL Shands and J.S. Hoffman. The Conservation Foundation, WaslL, DC. pp. 123-156.


                                               6-39

-------
 Woodman


 Levitt, J. 1980. Responses of Plants to Environmental Stresses.  2nd Edition. Volume I. Chilling, Freezing, and
 High Temperature Stress. Academic Press, New York. 497 pp.

 Levitt, J, H.H. Wiebe, J.S. Boyer, JJL McWiffiam, J.T. Ritchie, A. Blum, and F. Bidinger. 1980. Adaptation of
 plants to water and high temperature stress: summary and synthesis. In Adaptation of Plants to Water and High
 Temperature Stress, Eds. Turner, N.C  and PJ. Kramer. John Wiley & Sons, New York, pp.437-456.

 Malac, BJ7.  1987. Climate change impacts on the southern U.S.:   forest products industry's concerns and
 perspectives. In Proc. of Sympos. on Climate Change In The Southern United States: Future Impacts and Present
 Policy Issues, Ed, M. Meo. U. of Oklahoma,  May 22-29,1987. pp. 329-333.

 Marion, PD. 1981. Tree Disease Concepts. Prentice-Hall, Inc, Englewood Cliffs, NJ. 399 pp.

 Marx, D.H, E.B. Cowling, and J.N. Woodman. 1985.  Effects of  airborne chemicals on southern commercial
 forests: a scientific research plan and budget for the Southern Commercial Forest Research Cooperative. USEPA,
 USDA-FS, NCASL and NAPAP (unpublished).

 Miller, P.R.  1983. Ozone effects in the San Bernardino National   Forest   In Proc Air Pollution and the
 Productivity of the Forests. Eds, D.D. Davis, AJL Millen, and L. Dochinger. Izaak Walton League, Arlington,
 VA. pp 161-197.

 Miller, W.F, PJVL Dougherty, and GX. Switzer. 1987. Effect of rising carbon dioxide  and potential climate change
 on loblolly pine distribution, growth, survival, and productivity. In The Greenhouse Effect, Climate Change, and
 U.S. Forests, Eds, W.E. Shands and IS. Hoffman. The Conservation Foundation,  Wash, DC  pp. 147-188.

 National Academy of Sciences. 1975. Understanding Climatic  Change. National Academy Press, WaslL, DC.

 National Academy of Sciences. 1979. Carbon Dioxide and Climate, a  Scientific Assessment National Academy
 Press, Wash, DC.

 National Academy of Sciences. 1983. Changing Climate. National Academy Press, WaslL, DC. 4% pp.

 Oechel,  W.C. and  B.R. Strain.  1985. Native species responses to  increased atmospheric  carbon dioxide
 concentration. In Direct Effects of Increasing Carbon Dioxide on Vegetation. U.S. Dept. of Energy, WaslL, DC.
 pp. 117-154.

 Pearcy, R.W. and O. Bjorkman. 1983. Physiological effects. In CO- and Plants. The Response of Plants to Rising
 Levels of Atmospheric Carbon Dioxide. Ed, E. Lemon. Westview Press, Boulder, CO. pp.65-106.

 Peet,  RJC, and NX. Christensen. 1987. Competition and tree  death. BioScience 37(8): 586-595.

 Peters, RX. and J.D.S. Darling. 1985. The greenhouse effect and  nature reserves. BioScience 35(1): 707-717.

 Pyne, S J. 1984. Introduction to WHdland Fire. Fire Management in the United States. John Wiley & Sons, New
 York. 455 pp.

 Rind, D. 1987. Predicting regional climate change:  improving the  models. In The Greenhouse Effect, Climate
 Change, and U.S.  Forests,  Eds, W.E. Shands and J.S. Hoffman. The Conservation Foundation, Wash, DC. pp.
 77-90.

Rose, D.W., AJl. Ek,  and KX. Belli 1987. A conceptual framework for assessing impacts of carbon dioxide
change on forest industries. In The Greenhouse Effect, Climate Change, and U.S. Forests. Eds. WJL Shands
and J.S. Hoffman. The Conservation  Foundation, Wash, DC. pp. 259-276.


                                              6-40

-------
                                                                                      Woodman


Sandenburgh, R., C Taylor, and IS. Hoffman. 1987. How forest products companies can respond to rising carbon
dioxide and climate change. In The Greenhouse Effect, Climate Change, and US. Forests. Eds., W.E. Shands
and J.S. Hoffman. The Conservation Foundation. Wash., DC. pp. 247-258.

Schuize, E.-D.,  RJL  Robichaux, J. Grace, P.W. Rundel, and J.R.  Ehleringer. 1987. Plant water balance.
BioScience 37(1): 30-37.

Schumacher, FJC and T.S. Cofle. 1960. Growth and Yields of  Natural Stands of the Southern Pines. IS. Coile,
Inc, Durham, NC. 115 pp.

Sheffield, TOMn  N.D.  Cost, WA. Bechtold, and J.P. McClure. 1985. Pine growth reductions in the Southeast.
USDA For. Serv. Res. Bull SE-83, Asheville, NC. 112 pp.

Shugart, HJL, M.Y. Antonovsky, P.G. Jams, and AJ>. Sandford.  1986. CO.,  Climatic  Change and Forest
Ecosystems. In The Greenhouse Effect, Climatic Change, and Ecosystems. Eds* B. Bolin,  B.R. Doos, J. Jager,
and R A. Warrick. Scope 29. Wiley & Sons, Chichester. pp. 475-522.

Simard, AJ. and WA. Main. 1987.  Global climate change:  the potential for changes in wildland fire activity in
the Southeast  In Proc. of Sympos. on Climate Change In The Southern United States:  Future Impacts and
Present Policy Issues, Ed, M. Meo. U.  of Oklahoma, May 22-29,1987. pp. 280-308.

Sionit, N. and PJ. Kramer. 1986. Woody plant reactions to CO.  enrichment In Carbon Dioxide Enrichment of
Greenhouse Crops. Volume n Physiology, Yield, and Economics. EdX H.Z. Enoch and  BA.  KimbalL CRC
Press, Boca Raton, FL. pp.69-86.

Sionit,  N., BJEL Strain, H. Hellmers, GIL Reichers, and CH.  Jaeger. 1985. Long-term atmospheric CO.
enrichment affects the growth and development of Liquidambar styraciflua and Pinus taeda seedlings. Can. f.
Forest Res. 15:468-471.

Skog, YL, and C. Risbrudt 1982. Trends in economic scarcity of U.S. timber commodities. USDA For. Serv. For.
Prod. Lab. Res. Bull FPL-1L 25 pp.

Smith, DM. 1986. The Practice of Silviculture. John Wiley &  Sons, New York. 527 pp.

Solomon, AM. and D.C West 1987. Simulating forest ecosystem responses to expected climate change in eastern
North America:  applications to decision making in the forest  industry. In The   Greenhouse Effect Climate
Change, and US. Forests, Eds. W.E. Shands and J.S. Hoffman. The Conservation Foundation, Wash-, DC.  pp.
189-218.

Solomon, A-M, Mi. Tharp, D.C West G.E. Taylor, J.W. Webb, and  JX. Trimble. 1984. Response of
Unmanaged Forests  to CO.-Induced   Climate  Change: Available  Information,  Initial Tests,  and Data
Requirements. DOE/NBB-0033. US. Dept of Energy, Wash, DC 93 pp.

Surano, IUL, P f. Daley, JUT. Houpis, J JL Shum, JA. Helms,  RJ. Palassou, and M.P. Costella. 1986. Growth
and physiological responses of Pinus ponderosa Dougl ex Laws, to long-term elevated CO. concentrations. Tree
Phys. 2^43-259.                                                                ^

Tappeiner U, J.C, WH. Knapp, CA. Wierman, WA. Atkinson, CJ>. Oliver, J£. King, and J.C Zasada. 1986.
Silviculture - the past  30 years, the next 30 years: Part H. the Pacific Coast J. For.  84(5)37-46.

Tolley, L.C. and B.R. Strain.  1984. Effects of atmospheric CO.   enrichment and water stress on growth of
Liquidambar styraciflua and Pinus taeda seedlings. Can. J. Botany 62:2135-2139.
                                               6-41

-------
 Woodman

 Tofley, L.C. and B.R. Strain. 1985. Effects of CO- enrichment and water stress on gas exchange of Liquidambar
 styraciflua and Pinus taeda seedlings grown under different irradiance levels. Oecologia 65:166-172.

 US Bureau of Census. 1986. State and Metropolitan Area Data Book,  1986. U.S. Govt. Print. Off., Wash., DC.
 697pp.

 USDA. 1987. Agricultural Statistics, 1987. US Gov. Print Off., Wash, DC. 541 pp.

 USDA Forest Service. 1969. A Forest Atlas of the South. Southern  For. Exp. Stn, New Orleans, LA, and
 Southeast. For.  Exp. Stn, Asheville, NC. 27 pp

 USDA Forest Service. 1980. 1978 Wildfire Statistics. US Gov.  Print Off., Wash, DC. 55 pp.

 USDA Forest Service. 198L An Assessment of the Forest and Range Land Situation in the United States. USDA
 For. Serv. For. Res. Rep. 22. U-S. Gov. Print Off, Wash, DC 352 pp.
                                                                                    •
 USDA Forest Service. 1982. An Analysis of the Timber Situation in the United States 1952-2030. USDA For. Serv.
 For. Res.  Rep. 23.  U.S. Gov. Print Off, Wash, DC. 499 p.

 Waring, RJL 1987. Characteristics of trees predisposed to die. BioScience 37(8): 569-574.

 Woodman, J.N. 1986. Pollution and a changing climate -  implications for world forests. In Proc, National
 Convention, Soc. of Am. Foresters, Birmingham,
Woodman, J.N. 1987a. Pollution-induced injury in North American forests: facts and suspicions. Tree Physiology
Woodman, J.N. 1987b. Potential impact of carbon dioxide-induced climate changes on management of Douglas-fir
and western hemlock. In The Greenhouse Effect, Climate Change, and U.S. Forests, Eds, W.E. Shands and J.S.
Hoffman. The Conservation Foundation, Wash,  DC. pp. 277-284.

Woodman, J.N, and E.B. Cowling. 1987 Airborne chemicals and forest health. Environ. So. TechnoL 21(2):120-126.

Zahner, R. and  RJC.  Myers.  1986. Assessing the impacts of drought on forest health. In Proc, National
Convention, Soc  of Am. Foresters, Birmingham, AL. pp. 227-234.

Zahner, R. and RJC Myers. 1987. Tree-ring model confirms undefinable growth decline in Piedmont loblolly pine
stands. Progress  report to the USDA Forest Serv. Southeastern Forest Exp. Sta, Asheville, NC,  Nov. 1,  1987.
20pp.
                                              6-42

-------