Outline: Part I. Potential redistribution of tree species following climate change

1
(No Transcript)
2
Outline Part I. Potential redistribution of tree
species following climate change

• Climate change some past and possible future
  trends
• Evaluation of current tree biodiversity with 2
  data sets
• Potential changes in suitable tree species
  habitat (DISTRIB)
• Potential changes in forest type habitat
• Migration through fragmented habitats (SHIFT)

3
GLOBAL ENVIRONMENTAL CHANGE
Vitousek, 1993
4
www.whitehouse.gov/Initiatives/Climate
5
www.whitehouse.gov/Initiatives/Climate
6
(No Transcript)
7
Proportion of USA Affected by Extreme Rainfall
Events
Percent of US affected by gt2 per day Rainfalls
8
(No Transcript)
9
Ross Iceberg off Antarctica
March 21, 2000
10
Potential Changes in Temp and Precip, 2000-2099
Temperature Change, C
Temperature Change, C
Canadian Climate Centre
Hadley
Precipitation Change,
Precipitation Change,
gt100
50
0
-50
-100
www.cgd.ucar.edu/naco/vemap/trends.html
11
NATIONAL ASSESSMENT Mid-Atlantic Regional
Assessment
12
Potential Impacts Mid-Atlantic
Parameter 2030 2095 Confidence
CO2 () 20 to 30 50 to 120 Very High
Sea level (cm) 10 to 30 37 to 100 High
Temperature (C) 1.0 to 1.5 2.7 to 5.3 High
Precipitation () -1 to 8 6 to 24 Medium
Runoff () -2 to 6 -4 to 27 Low
13
Summary of Impacts Mid-Atlantic Region
Most Certain Negative Positive
Agriculture production
Coastal zones
Temp-related health (heat)
Moderately Certain
Forestry production
Temp-related health (cold)
14
Summary of Impacts (cont)
Uncertain Negative Positive
Biodiversity
Fresh water quantity
Fresh water quality
Ecological functioning
Vector/water borne disease
Environmental effects from agriculture
15
Evaluation of current tree biodiversity

• Elbert Little (US Forest Service) range maps
• USFS Forest Inventory and Analysis data

16
Littles Range Maps

• Elbert Little spent career in USFS assessing
  tree ranges and published atlases of
• Major hardwoods and conifers (1971)
• Minor eastern hardwoods (1977)
• We digitized maps for 78 species and acquired 57
  species from USGS 135 species in eastern US
• Overlaid maps in GIS
• species richness

17
Examples of Littles range maps
18
(No Transcript)
19
(No Transcript)
20
(No Transcript)
21
Forest Inventory and Analysis

• FIA has been in operation for some 70 years
• Provides consistent, credible, and periodic
  forest data for all forest lands (public and
  private) within the United States 
• Approximately one sample location (FIA plot)
  every 2,400 ha  
• Under annualized system, FIA strives to sample
  15 of plots in the eastern US and 10 of the
  plots in the western US every year as a base
  federal program (enhanced by some states, but not
  Ohio)

22

• FOREST INVENTORY (USFS)
• 37 states east of 100th meridian
• 135 tree taxa
• 103,488 plots
• 2,938,518 tree records
• PROCESS
• Extract FIA by State
• Calculate IV based on number of stems basal
  area
• Create point coverage of IV values by plot
• Aggregate points to county or 20 x 20 km
  polygons
• OUTPUT
• Importance Value (IV) for 135 tree species, by
  county or 20 km cell

23
(No Transcript)
24
Other FIA examples
25
(No Transcript)
26
(No Transcript)
27
Potential changes in biodiversity following 2
scenarios of climate change

• Potential changes in suitable habitat (DISTRIB)
• Potential migration of species in 100 years
  (SHIFT)

28
(No Transcript)
29
Regression Tree Analysis (DISTRIB)

• Data are recursively partitioned into subsets
  based on a single (best) predictor
• Output is a tree-diagram with the branches
  determined by splitting rules and a series of
  terminal nodes which contains
• the mean response

Advantages of RTA
Compared to linear models, RTA better captures
non-additive and non-linear relationships.
Highly suited for distributional mapping where
different variables operate at different
geographic regions can map predictors driving
the distribution.
Can be used to screen out irrelevant predictors.
The variables that operate at large scales are
used for splitting criteria early, while
variables that influence the response locally
are used in rules near the terminal nodes.
30
(No Transcript)
31
(No Transcript)
32
(No Transcript)
33
Top 20 Gainers
Quercus phellos 118.9 Q. falcata var.
falcata 134.6 Pinus palustris 137.8 Maclura
pomifera 181.3 Carya tomentosa 208.9 Quercus
marilandica 212.2 Celtis laevigata 362.4
Ulmus alata 410.6 Taxodium distichum
var.nutans 425.1 Quercus stellata 445.6

• Pinus elliottii 39.8
• Quercus muehlenbergii 41.6
• Quercus laurifolia 47.0
• Celtis occidentalis 54.1
• Pinus taeda 54.2
• Morus rubra 61.7
• Quercus nigra 64.2
• Gleditsia triacanthos 76.6
• Magnolia virginiana 101.8
• Diospyros virginiana 115.7

34
Bottom 20 Losers
Acer pensylvanicum -71.5 Crataegus sp. -65.5
Fraxinus nigra -65.3 Acer rubrum -62.9
Pinus virginiana -59.0 Prunus
serotina -53.0 Carya glabra -50.0 Ulmus
rubra -46.6 Fraxinus americana -46.3 Tsuga
canadensis -41.6

• Populus grandidentata -99.5
• Acer saccharum -98.5
• Thuja occidentalis -97.9
• Abies balsamea -96.0
• Populus tremuloides -94.5
• Pinus resinosa -93.4
• Betula papyrifera -90.0
• Fagus grandifolia -85.7
• Betula alleghaniensis -83.3
• Tilia americana -73.1

35
Tree species richness potential changes according
to DISTRIB
36
(No Transcript)
37
(No Transcript)
38
SHIFT Model (Spatially Explicit Cellular
Automata Model)initially developed by Mark
Schwartz

• Assumption
• SHIFT assumes that species can maintain
• an average migration rate of 50 km/century
• in fully forested regions (the upper end of the
  range of observed migration rates during
  the Holocene)

• Calculates the probability of colonization of
  currently unoccupied cells based on habitat
  availability, abundance in occupied cells and
  distance between occupied and and unoccupied
  cells.

39
Animation of colonization using presettlement
vegetation Virginia Pine in Ohio
40
A
B
ABUNDANCE
HABITAT AVAILABILITY
41
(No Transcript)
42
Distance, km Percent Colonization
10 32.8600
20 7.2777
30 2.0105
40 0.7489
50 0.3252
60 0.1548
70 0.0784
80 0.0399
90 0.0187
100 0.0140
150 0.0055
200 0.0020
300 0.0014
400 0.0009
500 0.0003
Southern Red Oak Concentric Rings
200 km
100 km
500 km
10 km
43
Current
DISTRIB
SHIFT
Constrained by DISTRIB and SHIFT
44
(No Transcript)
45
Percent of new suitable habitat colonized by 2100 (constrained by both DISTRIB and SHIFT) S. Red Oak Sourwood Sweetgum Persimmon Loblolly Percent of new suitable habitat colonized by 2100 (constrained by both DISTRIB and SHIFT) S. Red Oak Sourwood Sweetgum Persimmon Loblolly Percent of new suitable habitat colonized by 2100 (constrained by both DISTRIB and SHIFT) S. Red Oak Sourwood Sweetgum Persimmon Loblolly Percent of new suitable habitat colonized by 2100 (constrained by both DISTRIB and SHIFT) S. Red Oak Sourwood Sweetgum Persimmon Loblolly Percent of new suitable habitat colonized by 2100 (constrained by both DISTRIB and SHIFT) S. Red Oak Sourwood Sweetgum Persimmon Loblolly Percent of new suitable habitat colonized by 2100 (constrained by both DISTRIB and SHIFT) S. Red Oak Sourwood Sweetgum Persimmon Loblolly Percent of new suitable habitat colonized by 2100 (constrained by both DISTRIB and SHIFT) S. Red Oak Sourwood Sweetgum Persimmon Loblolly






SHIFT-CCC gt2 7.6 12.7 11.6 2.7 8.4
SHIFT-CCC gt20 2.0 2.4 2.2 0.8 1.5
SHIFT-CCC gt50 1.2 1.0 1.2 0.6 0.6
SHIFT-HAD gt2 11.5 8.2 14.7 3.8 9.9
SHIFT-HAD gt20 4.1 2.2 5.1 1.3 3.2
SHIFT-HAD gt50 2.5 0.9 3.0 0.9 1.6
46
Updates to the DISTRIB Model

• Improved and updated input data for climate,
  soils, and FIA data
• 20 km grid resolution (instead of county
  resolution) from 2121 counties to 10000 cells
• 135 species (as compared to 80)
• Extending analysis to Canada
• More robust models using combinations of trees
  using bootstrap resampling, with averaging of
  trees-Bagging and Random Forest

Current Future Directions
47
Bird Distribution Atlas (with S. Matthews R.
OConnor, Univ. of Maine)

• The distribution of birds are based on BBS
• The model is based on RTA, similar to DISTRIB
• The output of DISTRIB, climate and other
  bird-relevant variables are used as predictors
• An atlas along the lines of our tree atlas is
  under production

48
Uncertainties

• Climate change scenarios range from warm to
  warmer
• Input data approximate on species ranges,
  climate, soils
• DISTRIB assumes the species are in equilibrium
  with their environment
• SHIFT assumes certain migration rates
• Errors magnified when models are stacked
• GCM, DISTRIB, SHIFT
• Human disturbances limited in models (e.g., other
  mechanisms of global change may swamp these
  results)
• How much will humans move trees in trucks or
  trunks?

49
Conclusions - 1

• With DISTRIB, we estimate deterministically
• Potential future habitat distribution and
  abundance by species
• With SHIFT, we estimate stochastically
• Probability of colonization into fragmented
  landscapes over 100 year period
• With both together, we project realistic future
  scenarios which consider
• Individual tree species and their current
  abundance
• Fragmented nature of forested habitats
• Possible edaphic/landscape/climatic barriers to
  migration

50
Conclusions - 2

• Climate change impacts on forests could be
  substantial, especially composition change
  although much uncertainty remains
• DISTRIB
• Species richness increased north and slightly
  decreased south
• Most species expand suitable habitat north, more
  so with CCC than Hadley scenario
• Expansion of oak-pine and oak-hickory, loss of
  several northern forest types
• SHIFT
• Very little migration through fragmented habitats
  within 100 years
• Remote possibility of long distance (up to 500
  km) dispersal
• DISTRIB/SHIFT
• Only small fraction of new suitable habitat in
  2100 is likely to be colonized

51

• Outline Part II. Prescribed fire on oak
  ecosystems in southern Ohio

Introduction to S. Ohio Forests and
Fire Integrated Moisture Index Fire
Characteristics Monitoring Grid Point Data
Collection and Analysis Transects Data Collection
and Analysis
52
Ohio Forests

• Forest cover has increased since 1940
• Now at 30 forested, mostly S and E

53
Forest trends in southeast Ohio
54
Oak Challenges

• Oaks are being replaced by mesophytic species
  maples, blackgum, tulip-poplar and black cherry
• Red maple is becoming the most abundant species
• Native Americans and early European settlers used
  fire frequently, which favored oaks
• Oaks compete better on drier sites with
  intermediate light levels and they tolerate
  fire well

• So, we need to add some fire to kill young maples
  and thin some larger trees to add light to the
  forest floor!

55
A National Study of the Effects of Fire and Fire
Surrogates on Forest Ecosystems
56
Thinning
57
Montana Fires, 2001
58
San Diego Fires The fire moves south down Oak
Canyon toward the 52 freeway. October 2003.
59
Ventura County firefighters look at a twister of
flame from a wildfire in Simi Valley
60
Prescribed Fires
61
(No Transcript)
62
Location of Study Areas
63
Integrated Moisture Index
Hillshade (40)
Flow Accumulation (30)
Elevation
Curvature (10)
Site Fertility
64
(No Transcript)
65
(No Transcript)
66
(No Transcript)
67
(No Transcript)
68
(No Transcript)
69
Canonical Correspondence Analysis
Moist Fertile
Dry Infertile
70
Integrated Moisture Index -2 m Sample points 50
m Soils 100 m
N
71
IMI uses

• Integrates topographic soil factors and aids in
  the study of landscape pattern and function
• Aids in predicting productivity and composition
  of future stands
• Can help in exploring species-habitat associations

• Can be used as landscape stratifier for various
  ecological studies including

72
(No Transcript)
73
(No Transcript)
74
Flying Zaleski
Thin
Control
Thin Burn
Burn
75
Fire Characteristics Monitoring

• What is magnitude, duration, and pattern of air
  temperature near the ground surface during the
  fires ?

76
Fire Characteristics Monitoring
Maximum Temperature
Heat Index integral under curve
Duration
Time of Day
77
Zaleski Max. Temperature
78
Zaleski Heat Index
Zaleski
79
Zaleski Rate of Spread
Rate of Spread, m/min
Thin Burn
Burn
80
Burning at Zaleski State Forest April 4, 2001
81
Zaleski Temperature Slices April 4, 2001 Burn
82
(No Transcript)
83
Results, Fire Temperatures2001 fires

• ThinBurn treatment cooler (138oC) fires than
  Burn-Only (174oC)
• Skid road barriers
• Uncured slash barriers
• Burned earlier in day
• Xeric sites gt Intermediate gt Mesic sites on heat
  measurements
• Probe-data recorder system adds value over other
  sensors because we can calculate heat index,
  duration of heat, time of spike, as well as
  animate the fires

84
Grid Point Data Collection and Analysis

• What are longer-term, indirect effects caused by
  fire and thinning treatments on
• Soil Temperature
• Soil Moisture
• Forest Floor Light Environment
• Oak and Hickory Seedlings and Saplings

85
Soil Temperatures following burns
Control
Thin
Thin/burn
Burn
86
Soil temperature vs. fire/thinning April 2001
Zaleski
Control
Thin/Burn
Burn
Thin
87
Soil temperature vs. fire/thinning June 2001
Zaleski
Thin/Burn
Control
Thin
Burn
88
Control
Thin/Burn
89
(No Transcript)
90
Control
2000 to 2001
Thin/Burn
91
(No Transcript)
92
Tree sampling

• Seedlings, saplings, trees size and species
• Cover of litter, woody, grass, forb, bare
  ground, coarse woody debris

93
Control
ThinBurn
Many species did have vast differences from TB
94
Results, Grid Point Studies1st year after TB

• Soil temperatures
• Higher on TB vs. Control (as much as 4oC)
• Higher in drier IMI classes
• Moisture
• Higher on TB vs. Control
• Higher on mesic IMI classes
• Light
• Higher on TB (11.9) vs. Control (7.1)
• Higher on drier IMI classes

95
Results, Oak-Hickory Regeneration

• Oak-Hickory seedlings
• More seedlings on Xeric Intermed. vs. Mesic
• No first-year treatment effects
• Oak-Hickory saplings
• Very few only 40/ha in 2000 and 48/ha in 2001,
  compared to gt6400 seedlings/ha in both years
• Essentially no saplings on Mesic sites

96
Transects Data Collection and Preliminary
Wavelet Analysis

• Eight 400 m transects set up on two treatments
  (Control and Thin Burn) at two sites
• Each transect connected the dots along 9 grid
  points
• Sampling every 3.1 m 16 between each grid point
• Preliminary wavelet analysis to show pattern
  along transect at 5 scales 3.1, 6.2, 12.5, 25,
  and 50 m

97
Thin and Burn
Control
N
Zaleski State Forest
98
Z1A073 - Control
99
Z3B037 Thin Burn
100
Variables Evaluated on Transects(every 3.1 m x
1028 locations)
Light Slope Cover-cwd
Soil moisture Aspect Cover-bole
Litter-cover depth Coverwoodies Cover-rock
Tree basal area Cover-graminoid Cover-bare
Saplings-size class Cover-herb GIS-30m DEM
Seedlings-size class Cover-moss GIS-1m DEM
101
3.1 m
800 m
6.2 m
12.5 m
25 m
50 m
102
(No Transcript)
103
3.1 m
800 m
6.2 m
12.5 m
25 m
50 m
104
(No Transcript)
105
Litter Volume, cm3/m2
106
(No Transcript)
107
(No Transcript)
108
(No Transcript)
109
(No Transcript)
110
(No Transcript)
111
Results, Transect Studies2nd year after TB

• Vast amount of fine scale variation in most
  variables
• Litter lower on TB
• Overstory Basal Area lower on TB
• Light - higher on TB
• Moisture - higher on TB
• Percent Plant Cover higher on TB
• Seedlings
• Oaks and hickories more on TB
• Tulip poplar, sassafras, sourwood, black gum
  more on TB
• Small maple seedlings less on TB

112
Mop-Up Remarks

• Geospatial tools are important for several
  aspects of the study
• Integrated Moisture Index useful for scoring
  moisture and fertility gradients as well as
  stratification of study sites
• Thermocouples and buried data recorders provide
  valuable information on fire characteristics
• Data at three scales (plot, grid point,
  transects) important
• ThinBurn treatment tends to increase soil
  temperature, moisture, light, and oak
  seedlings/saplings
• Stay tuned to see if we get oak-hickory
  regeneration!

113
For Further Information

• Web site for most data presented
  today
• Littles boundaries
• FIA data grouped by 20x20 km cell
• Climate change atlas (DISTRIB)
• Several related papers (pdf)
• Fire animation
• Background on fire studies
• www.fs.fed.us/ne/delaware/4153.html
• For hard copy of atlas or reprints
• liverson_at_fs.fed.us
• Thanks to USDA FS Northern Global Change Program
  and the Joint Fire Science Program for support,
  and to you for the invitation to come to Toledo!