30 years of forest conversion in the Northeast:
historical patterns and future projections
Alison Adams1,2, Jennifer Pontius1,3, Gillian Galford1,2,Scott Merrill4, David Gudex-Cross1
1UVM
Rubenstein School; 2 Gund Institute for Ecological Economics; 3 US Forest Service; 4 UVM Dept of Plant and Soil Science
The BIG
Picture
Why does modeling forests matter?
Understanding trends and patterns
Methods
Identify primary drivers of trends
Results &
Products
Estimate or simulate data that doesn’t
exist (temporally, spatially…)
Additional
analyses
Manage forests in changing conditions
Conclusions
Product use &
applications
The BIG
Picture
Forest pattern and extent
Methods
Results &
Products
Forests = 53% of
terrestrial carbon
Additional
analyses
Conclusions
Product use &
applications
Food and Agriculture Organization 2001
Photo by Ed Post
B. McRae, UCSB
The BIG
Picture
Objective
Methods
Analyze patterns of past changes in forest cover
Results &
Products
and
Additional
analyses
Conclusions
Product use &
applications
use these patterns to project future changes in forest cover
The BIG
Picture
Study area & land cover data
Methods
Results &
Products
Additional
analyses
Model calibration
Model
validation
Conclusions
Product use &
applications
Derived from Gudex-Cross et al. in progress
The BIG
Picture
Model flow
1985
2000
Methods
1. Calculate transition rates
Results &
Products
Additional
analyses
Conclusions
Product use &
applications
(About 61,000 ha)
2. Calculate correlations
(About 320,000 ha)
with other spatial
variables
(“weights of evidence”)
3. Refine and remove
insignificant and correlated
variables
2015
The BIG
Picture
Model flow
1985
2000
2015
Methods
1. Calculate transition rates
Results &
Products
Additional
analyses
Conclusions
Product use &
applications
2. Calculate correlations
with other spatial
variables
(“weights of evidence”)
3. Refine and remove
insignificant and correlated
variables
5. Compare
2015
4. Simulate
The BIG
Picture
Model flow
1985
2000
2015
Methods
1. Calculate transition rates
Results &
Products
2.
Constant
Calculate
climate
3.
Scenario B
(med.)
Refine and
Additional
analyses
Conclusions
Product use &
applications
2030
correlations
with other spatial
Scenario A
2030
variables
(slow)
(“weights of evidence”)
remove
correlated
Scenario
C variables
(fast)
5. Compare
2045
2060
2015
4. Simulate
2045
2060
2030
2045
2060
2030
2045
2060
The BIG
Picture
Methods
Results &
Products
Additional
analyses
Conclusions
Product use &
applications
Simulated maps
The BIG
Picture
Simulated maps – Burlington area
2015 (observed)
2030
2045
2060
Methods
Results &
Products
Additional
analyses
Conclusions
Product use &
applications
The BIG
Picture
Simulated maps – Maine coast
2015 (observed)
2030
2045
2060
Methods
Results &
Products
Additional
analyses
Conclusions
Product use &
applications
The BIG
Picture
Simulated maps – Adirondacks
2015 (observed)
2030
2045
2060
Methods
Results &
Products
Additional
analyses
Conclusions
Product use &
applications
The BIG
Picture
Observed & projected transitions
Methods
Results &
Products
Additional
analyses
Conclusions
Product use &
applications
Decreasing forest area over time
The BIG
Picture
Forest configuration
Methods
Results &
Products
Additional
analyses
Conclusions
Product use &
applications
All measures show increasing fragmentation over time
…even for 2000 - 2015, when forest area increased
The BIG
Picture
Methods
Results &
Products
Important explanatory variables
Forest regrowth likely…
• Low population
densities
• Far from roads
• High elevations
• Steeper slopes
Additional
analyses
•
Conclusions
Product use &
applications
•
•
•
•
Deforestation likely…
High population
densities
Near other non-forest
Low elevations
Flatter areas
Less protected
The BIG
Picture
Methods
Results &
Products
Additional
analyses
Conclusions
Product use &
applications
But what about that 2000 – 2015 forest
gain?
The BIG
Picture
But what about that 2000 – 2015 forest
gain?
Methods
Results &
Products
Housing Price
Index (HPI)
Additional
analyses
Conclusions
Product use &
applications
Federal Housing Finance Agency
The BIG
Picture
Conclusions
Methods
If 1985 – 2000 trends are consistent with what’s happening
today, and if those trends continue, forest area will decrease
over the next few decades
Results &
Products
Development appears to be a major factor in which locations
experience deforestation or reforestation
Additional
analyses
Climate doesn’t play a significant role at this temporal scale
Conclusions
Product use &
applications
The BIG
Picture
Methods
Results &
Products
How can these maps inform management?
Where forest is likely to disappear in the future, impact of
that for fragmentation, etc.
Basis for conservation for connectivity
Future estimates of forest-based ecosystem services
Additional
analyses
Many other applications…
Conclusions
Product use &
applications
In other words…
where to prioritize for conservation
The BIG
Picture
Methods
Questions
?
Results &
Products
Additional
analyses
Conclusions
Product use &
applications
Many thanks to…
The BIG
Picture
Product use considerations
Simulations of likely change
Methods
Results &
Products
Relies heavily on the state of the economy, development
pressures
Don’t differentiate between types of non-forest
Additional
analyses
Climate may play a role in longer-range simulations
Conclusions
Product use &
applications
The BIG
Picture
Methods
Results &
Products
Additional
analyses
Conclusions
Product use &
applications
Important explanatory variables
Non-forest to forest:
population density
Likely reforestation if
population density is
very low
The BIG
Picture
Important explanatory variables
Methods
Non-forest to forest:
distance to any road
Results &
Products
Likely reforestation far
from any road
Additional
analyses
Conclusions
Product use &
applications
The BIG
Picture
Important explanatory variables
Methods
Forest to non-forest:
conservation status
Results &
Products
Additional
analyses
Conclusions
Product use &
applications
Decreasing likelihood of
deforestation as
conservation strength
increases
The BIG
Picture
Important explanatory variables
Slope
Non-forest to forest
Methods
Positive likelihood on
steeper slopes
Results &
Products
Additional
analyses
Conclusions
Forest to non-forest
Positive likelihood in
flatter areas
Slope (degrees)
Product use &
applications
The BIG
Picture
Areas for future research
Improvement of stand age estimates
Methods
Results &
Products
Additional
analyses
Conclusions
Product use &
applications
Carbon storage estimates that don’t rely on stand age to
better isolate effect of tree species
Regional or local analyses of forest/non-forest transitions to
improve applicability to local management
More forest and non-forest classes in land cover change
models
Include or focus mostly on economic drivers for forest cover
change models