Slides - Forest Ecosystem and Landscape Ecology Lab

Download Report

Transcript Slides - Forest Ecosystem and Landscape Ecology Lab

Lecture 14
Models II
Principles of Landscape
Ecology
March 31, 2005
Spatially dynamic
Raster models
X LANDIS-II
Patch
models
Gap models
Mechanistic detail
Individual
tree
models
Cellular Models
• A system of cell networks or grids
• Cells interact with neighborhood
• Each cell adopts one of m (m may be infinite) possible
states
• Transition rules for each state can be simple,
deterministic, or stochastic.
• Transition rules ~ f(abiotic constraints, biotic
interactions, disturbances)
What is a cellular landscape model?
Cellular landscape models simulate change through time in
response to endogenous processes (growth, competition)
and exogenous forcing (disturbance, climate change, etc).
Spatially explicit and spatially interactive
• GIS used to store/display data.
• Entities have map coordinates
• Include spatial processes: seed dispersal,
disturbance
Run iteratively over time
• The landscape has a memory of previous events
• ‘Equilibrium’ doesn’t apply in the sense of analytical models
Model Example: LANDIS-II
Spatially interactive cellular model
Each site exchanges information (seeds) and
energy (disturbances: chemical and mechanical)
with neighboring sites.
Species have unique life history attributes
-shade tolerance, fire tolerance
-longevity, maturity age
-seed dispersal capabilities
Species presence in age cohorts.
Example:
sugar maple 11-20, 41-50 +
basswood 61-70, 101-110
Model Example: LANDIS-II


Multiple disturbances and interactions
Scales tree growth up to the landscape scale
FORESTED LANDSCAPE
DISPERSAL
SPECIES
ESTABLISHMENT
INSECTS / DISEASE
HARVESTING
FIRE
WINDTHROW
CLIMATE
LIVING BIOMASS
DEAD BIOMASS
Model Application Example:
Climate change effects in
northern Wisconsin
Goal:
 Estimate the effects of climate change
and disturbance on forest composition
and biomass.
Current
Great Lakes
Forests
Forest tree
communities
Disturbance
Dispersal
Ecosystem
Processes
Why Species Composition and Biomass?
 Of concern to management
 Integrating variables
CLIMATE
CHANGE
Scenarios
Potential
Great Lakes
Forests
Spp composition
Aboveground Biomass
Simulation of climate change in
northern Wisconsin
Climate change effects modeled:
 tree species germination and establishment
 tree spp growth rates and competitive ability
Climate change effects not modeled:

changes in disturbance regimes

potential CO2 fertilization

changes in soil properties
Other processes not modeled:

herbivory

exotic species
Simulation of
climate change
in northern
Wisconsin
Climate Scenarios
3 climate scenarios:

Current Climate

Hadley Centre for Climate Prediction
+3.8°C and +38cm ppt

Canadian Centre for Climate Modelling
+5.8°C and +20cm ppt
Simulate forest change over 200 years

Years 1990 - 2090 includes climate change

Constant climate from 2090 - 2190

10 year climate averages
Disturbance Scenarios
Two Disturbance Scenarios:

No Disturbance

Wind + Harvesting

Wind equal to historic frequency

Clearcutting

Selective cutting

Heavy thinning
LANDIS-II Input Data
23 tree species with life history attributes
Probability of establishment calculated
from a forest gap model (0.1 ha)
Growth and decomposition rates
Establishment, growth, and
decomposition varied among
ecoregions due to climate and soils
Total Aboveground Live Biomass
Canadian GCM
No Disturbance
Wind and Harvesting
ANIMATION
Aboveground Live
Biomass
45 90 135 180 225 270 > 325 Mg/ha
Total Aboveground Live Biomass
Canadian GCM
No Disturbance
Wind and Harvesting
Aboveground Live
Biomass
45 90 135 180 225 270 > 325 Mg/ha
Results: Biomass change
Aboveground Live
Biomass (Mg/ha)
300
Current Climate
Hadley Climate
Canadian Climate
200
100
constant
climate begins
1990 2010 2030 2050 2070 2090 2110 2130 2150 2170 2190
Simulation Year
With Wind &
Harvesting
300
Aboveground Live
Biomass (Mg/ha)
Without Disturbance
200
100
1990 2010 2030 2050 2070 2090 2110 2130 2150 2170 2190
Simulation Year
Results: Change in community
composition
Without Disturbance:
 The landscape becomes dominated by sugar
maple
 Few opportunities for southern species to
migrate north
With Wind and Harvesting:
 Shift toward southern oak and hickory if climate
changes, although the shift is small
Why isn’t there a larger shift toward
southern species?
Interactions between climate change
and fragmentation
At the same time, many
species will be displaced.
Wisconsin
As climate changes, we
expect northward migration
of some tree species.
Illinois
Interactions between climate change
and fragmentation
However, species migration
limited by:


Wisconsin
distance-limited seed
dispersal
the priority effect occupancy by current
species
Other limits not considered:
 generational lags
 herbivory
Illinois
Interactions between climate change
and fragmentation
Wisconsin
Illinois
Fragmentation also
reduces migration:
 fewer available
colonization sites
 fewer seed sources
Interactions between climate change
and fragmentation
Consequences:
Spp richness reduced.
Wisconsin
Illinois
Decline in productivity and
aboveground live
biomass.
Why? Realized niche <>
fundamental niche. The
species best adapted to
new climate are not
widely dispersed.
Interactions between climate change
and fragmentation
Our Questions:
How will seed dispersal
limitations affect
aboveground live
biomass?
Wisconsin
Is seed dispersal limited by
fragmentation?
Is seed dispersal limited by
existing species?
Illinois
Estimation of the effects of
seed dispersal
How do we measure the effect that seed dispersal
has on aboveground live biomass (B)?
Estimate B from identical scenarios with distance
limited seed dispersal and without seed dispersal
distance limitations.
Calculate Difference
Total Bno distance limit Total Bdistance limited
= Total Bseed dispersal effect
Estimation of the effects of
seed dispersal
20
Aboveground Live
Biomass (Mg/ha)
No Disturbance
Hadley Climate
Canadian Climate
10
0
2000 2020 2040 2060 2080 2100 2120 2140 2160 2180
Simulation Year
20
Aboveground Live
Biomass (Mg/ha)
Wind and Harvesting
10
0
2000 2020 2040 2060 2080 2100 2120 2140 2160 2180
Simulation Year
Climate Change Conclusions
 Disturbance is a strong determinant of
future community composition under
climate change.
 But, five important tree species will be
extirpated and landscape diversity will be
reduced if the climate warms.
Climate Change Conclusions
 The northward migration of many
species is limited by seed dispersal.
 Aboveground live biomass is limited by
seed dispersal.
 Management will need to balance carbon
storage, maintenance of diversity, and
the reality of species loss.
Caveats:
 Only two possible climate change
scenarios out of dozens.
 We did not include many critical
processes, e.g. herbivores.
 Our results are unvalidated.
Question: What, if any, value does this
research have for management?
Modeling Conclusions
Modeling Conclusions
Preparing input data is the most arduous
task. Garbage in, garbage out (?).
Replication is not always helpful - depends
on the size of stochastic events.
You can never include everything.
Always focus on the questions first, tools
last.
Modeling Conclusions
Technical limitations remain
Increase in computer capability in past
decade is not a panacea.
Challenge of appropriate complexity in
spatial models remains
•Spatial data availability
•Spatial and temporal scale limitations
•Resolution—Extent tradeoff
Model caveats
http://www.env.duke.edu/landscape/classes/env214/le_mod0.html
Building model confidence:
data validation
Traditional validation: compare model data with
empirical data. However, there is rarely independent
landscape data collected at same scales. Data solutions
include:
Fine-scale data
Problem: wrong scale
Space-for-time
e.g. southern forests ~ climate change
Problem: different initial conditions, multiple changes
Reconstruct past responses
Problem: unknown starting conditions
lack of human behavior model
lack of climate data
Compare to other models
e.g. GCMs
Problem: few other FLSMs applied at regional scale
Both models wrong or right? Model autocorrelation.
Building model confidence:
alternatives to validation
Landscape validation is not always possible - need
to judge by different standards.
Process validation
Independent application, assessment, and review
Indirect Corroborating evidence
Development over time
Model transparency:
• open code
• generous comments
Building model confidence:
Summary
Model
acceptance
+
Confidence from:
• application
• review
• development
• mistakes!
Doubts from
increasing
complexity
-
Model development time