Climate Change Impacts at the Range
Margins of Rocky Mountain Tree Species:
Interactions with Disturbance
Katie Renwick
PhD Candidate
Colorado State
University
Climate-Disturbance
Trees
are slow to migrate
Interactions
Climate Change
All
species:
up 11 m
decade-1
Just
trees:
up 2 m
decade-1
Chen et al.
2011
Renwick &
Rocca 2014
• Change slow because trees long-lived
• Recruitment limited by competition
Upward Migration
Fire can initiate range shifts
Brubaker 1986, Overpeck et al. 1990, Johnstone & Chapin 2003
What about biological disturbances?
What about biological disturbances?
• Forest insects and pathogens
Mountain Pine Beetle
Soil Undisturbed
Host-Specific
Will MPB accelerate range shifts?
Methods
-2
-1
0
1
2
+1°
-3
Temperature Anomaly (°C)
3
Study Area
Rocky Mountain National Park
1992
10
1982
2002
2012
-5
0
5
-.1 cm
-10
Precipitation Anomaly (cm)
1972
1972
1982
1992
2002
Climate Trends
2012
Methods
Transect Sampling
20 m
20 m
Ecotone Types
Low Elevation:
• Ponderosa-Lodgepole
• Douglas-fir-Lodgepole
High Elevation:
• Lodgepole-Limber Pine
• Lodgepole-Spruce/Fir
Variables Measured
Biotic:
• Trees (species, DBH)
• Seedlings (density)
• Mountain Pine Beetle
Severity
Abiotic:
• Elevation
Disentangling the Drivers
• Model 2012 density as function of:
–
–
–
–
–
Limber Pine
1992 density
Ecotone type
Elevation
Disturbance (Mountain Pine Beetle)
Elevation : Disturbance interaction
Douglas-fir
Lodgepole
Pine
Engelmann
Spruce
Subalpine
Fir
Lower Elevation Ecotones (Trailing Edge)
Ponderosa –
Lodgepole
Douglas-fir –
Lodgepole
Evidence of shift:
1. Mortality of trees
2. Low seedling recruitment
MPB: accelerate canopy
death and recruitment of
new species
Analyzed together: type effect
Lodgepole trees: disturbance drives mortality
5
Model: 2012 Density ~
1992 Density + MPB
Disturbance
MPB
None
Log(2012 Density)
4
3
2
1
0
0
1
2
3
Log(Initial Density)
4
5
Lodgepole trees: disturbance drives mortality
5
Model: 2012 Density ~
1992 Density + MPB
Disturbance
MPB
No disturbance  slight decline
None
Log(2012 Density)
4
High disturbance  large loss of trees
3
2
1
0
0
1
2
3
Log(Initial Density)
4
5
Lodgepole Seedlings: Low Recruitment
Model: D12 ~
D92 + MPB + Elev. + MPB*Elev.
1000
High disturbance severity 
higher seedling density
Seedlings/ha
750
500
No recruitment at the lowest
elevations
Disturbance
MPB
None
250
0
low
Elevation
high
Lodgepole Seedlings: Low Recruitment
1000
Seedling layer dominated
by Douglas-fir
Seedlings/ha
750
500
250
0
low
Elevation
high
MPB: accelerate canopy
death and recruitment of
new species
Upper Elevation Ecotones (Leading Edge)
Lodgepole –
Limber
Lodgepole –
Spruce/Fir
Evidence of shift:
1. Increase in recruitment
2. New seedlings at higher
elevations
Analyzed separately –
No MPB on Limber transects
MPB: accelerate
lodgepole recruitment
at higher elevations
Lodgepole – Limber: Ecotone Remained Stable
1000
Model: 2012 Density ~
1992 Density
(Same for trees and seedlings)
MPB
None
750
Seedlings/ha
• No change in tree
or seedling density
• No elevation effect
Disturbance
500
250
0
low
Elevation
high
Lodgepole – Spruce/Fir:
High mortality of mature lodgepole
Recruitment not moving up
• Model: 2012 Density ~ 1992 Density + MPB
• No elevation effect
1250
40000
Disturbance
MPB
None
1000
Seedlings/ha
30000
750
20000
500
10000
250
0
0
low
Elevation
high
Relative
Density:
1.5 %
Will MPB accelerate range shifts?
• Evidence of lower range margin shift:
1. Mortality of lodgepole trees – YES
2. Low seedling recruitment – YES
MPB: accelerate canopy death and recruitment of new species – YES
• Evidence of upper range margin shift:
1. Increase in seedling recruitment – with MPB
2. New recruitment at higher elevations – NO
MPB: accelerate lodgepole recruitment at higher elevations – NO
Conclusions
• Location of disturbance matters
– Trailing edge of target species- accelerates range retreat
– Leading edge of target species- reduces seed source
• Competition can limit range expansion
• Overall: range contraction for lodgepole
Implications for Future Forests
Current Species Richness
Projected 2080s
# species
Questions?
Thanks to: Tom Stohlgren, Greg
Newman, Josh Hansen, Max
Barlerin, Jeff Connor, Judy Visty
Funded by:
McIntire-Stennis
appropriations to Colorado
State University
Disturbance facilitates upward migration
Disturbance Intensity
0
50
100
None
Medium
High
-50
• No disturbance 
recruitment declines
with elevation
• Upward shift depends
on disturbance
Predicted Change in Density
(seedlings/ha)
mean
elevation:
9.7m
150
Best model: 2012 density ~ MPB + elevation +
MPB:elevation
low
Elevation
high
Climate Change
Upward shift in
elevation:
•
•
•
•
•
Mountains forests
particularly vulnerable
Enquist & Enquist
2001
Beckage et al.
2008
Lenoir et al. 2008
Feeley et al. 2011
Felde & Kapfer
2011
Altered Species Distributions
High rate of warming
Species with limited extent
Barriers to migration
Forecasting Potential Range Shifts
• Problems:
– Fine-scale climatic
variability
– Biotic interactions
important at the
landscape scale
Research Objectives:
1. Incorporate local-scale climate variability and biotic
interactions into species distribution model
2. Examine potential climate change impacts on tree
species at the landscape scale
Methods: Study Area and Data
Rocky Mountain
National Park:
• Forest Data:
– Presence/Absence for 7
tree species (2002)
• Climate Data:
– Downscaled PRISM
– 90m resolution
– Future: 2080s, A1B
Data Sets
• Full Model:
– Entire park (n = 899)
– Goal: Forecast Change
• Forest Model:
– Plots with trees (n = 562)
– Goal: Local-scale species
dynamics
Rocky Mountain National Park
Modelling Framework
Climate
Predictor Variables:
•
•
•
•
•
Σ
Topography
Mean Annual Temp.
(Mean Annual Temp.)2
Mean Annual Precip.
Temp. Differential
Topographic Convergence
Response:
• 7 common tree species
Latimer et al. 2009, Ovaskainen et al. 2011, Pollock et al. 2014
Error:
• Correlation
Matrix
Model Performance and Comparison
• Forest Model:
• Full Model:
– AUC range .75 - .95
– May over-predict
diversity
# species
– AUC range .76 - .96
– Predicts trees in the
alpine tundra
Residual Correlation Differs Between
Models
• Full Model:
• Forest Model:
– All are positive
– Some negative
correlations
Low elevation
tundra
low Ponderosa
Douglas-fir
Limber Pine
forest
Aspen
Lodgepole Pine
Engelmann Spruce
shortgrass
steppe
high Subalpine
Fir
High elevation
Douglas- Limber
Lodgepole Engelmann
Ponderosa
fir
Pine Aspen
Pine
Spruce
1
-0.01
1
-0.03
0.15
1
-0.33
0.18 -0.11
1
-0.43
0.26
0.38 0.29
1
-0.28
-0.28 -0.01 0.21
0.16
1
-0.35
-0.16 -0.14 0.32
0.21
0.81
Change in Suitable Habitat
Full Model
increase
decrease
Forest Model
decrease
increase
loss
loss
Current
2080s
Decline in Overstory Diversity
Full Model
Forest Model
Current
2080s
Models may over-estimate future
ranges
Current Species Richness
Projected 2080s
# species
Models may over-estimate future
ranges
Current Species Richness
Projected 2080s
# species
Ponderosa Pine Distribution
Current
Projected 2080s
Conclusions
• Modelling Distribution Shifts:
– Joint distribution incorporates species interactions
– Extent of data matters
• Future Forests:
– Decline in high-elevation species
– Less diverse overstory
Questions?
Contact Katie:
katie.renwick@colostate.edu