Impacts of climate change on greater sage - grouse populations in the

advertisement
Impacts of climate change on greater
sage-grouse populations in the
Northern Rockies and Great Plains:
understanding the combined impacts
of climate, altered fire regimes and
emerging infectious disease
Anne M. Schrag*, Joy Ritter^, Steve C. Forrest*
*Northern Great Plains Program, World Wildlife Fund
^American Wildlands
Credit:Hargreaves
Greater sage grouse
€
€
€
€
€
Credit: Cox
Long-lived (3-6 yrs)
with low reproductive
rates (6-9 eggs/clutch)
and low overwinter
mortality (2-20%)
Regional declines
between 17-47%
Currently occupy 56%
of historic range
Very little sagebrush
formally protected
Petitioned for listing
numerous times under
ESA
Threats to sage-grouse habitat
€
€
€
Changing fire regimes
Exotic plant invasions
Land-use change
y Conversion to agriculture
y Urbanization
y Energy development
○ Current restrictions: 0.4km
○ Effective impact: ~3.2km
€
Infectious disease
y West Nile virus: 2.4-13.3% mortality
€
Climate change
Credit: sagebrushsea.org
Objectives
€
€
€
How will the spatial distribution of Wyoming
big sagebrush (Artemisia tridentate v.
wyomingensis) and silver sagebrush
(Artemisia cana) change under various
climate-change scenarios?
How will climate change impact fire
regimes, and what will be the resultant
impacts on the distribution of shrublands
versus grasslands?
How will climate change impact the
occurrence of West Nile virus?
Credit: Cox
Species distribution models
€
€
Sagebrush point
data (presence
only): LANDFIRE
Program
Current climate
data: PRISM
y seasonal and annual
avg temp and total
precip
€
Future climate data:
WCRP CMIP3
y A1B scenario
y 6 models at three
time steps: 2030,
2070, 2099
MAXENT
€
€
€
€
€
Estimate species distribution by assuming that it
must agree with all known constraints and without
imposing any unwarranted constraints
Find the distribution that is “closest to uniform”
Use background points to represent suite of
environmental conditions within the study area
Project this distribution using future climate
scenarios
Test against null model AUC to ensure that model
results are different from random
MAXENT
€
Parameters chosen
y 75:25 train-test split (want to know ability of model to predict
independent test data, e.g., future climate scenarios)
y 10,000 background pts
y Presences pts:
○ ARTCAN: 973
○ ARTTRI: 1160
y Used only linear, quadratic and hinge features due to evidence of
overfitting when using auto-features option with >79 samples
€
Null model testing
y Built null model of random pts
○ Sample size * 999 iterations (for each species)
y Developed a frequency histogram of AUC values from these 999
runs
y Compare AUC values from model to AUC value distribution from
null model
Model validation & conclusions
€
July precip contributed
most to model:
y ARTCAN: 49.4%
y ARTTRI: 50.1%
€
Model AUC fell outside
the 95% confidence
interval
y ARTCAN: .912 (train),
.903 (test)
y ARTTRI: .923 (train),
.895 (test)
y 95% CI: .592
€
Significantly different
from random model
Sagebrush species move in
heterogeneous ways across
landscape
€ Highly dependent on
precipitation regime
€ Possible shift to silver
sagebrush in some areas
€
Climate, fire and sagebrush
€
Shifts in sagebrush vs grassland likely with
changing fire regimes
y Increased fire frequencyÆfavors grasslands
y Increased cover of exotic grassesÆfavors increased fire
frequency
€
TELSA/VDDT landscape simulation model
y Vegetation succession, disturbance and management
€
Ignites and spreads fires annually based on
following parameters:
y Land cover
y Human use (Wildland-Urban Interface)
y Historic fire data (ignition locations, size)
€
Then builds vegetation based on potential
vegetation types
Model development
€
€
€
€
€
Inputs: LANDFIRE sagebrush map, 20-year
repeated fire regime, WUI-SILVIS lab (2000
Census data)
Trend developed between current climate and fire
occurrence and size
Then re-examined using future climate scenarios
and applying trends at three future time steps:
2030, 2070, 2099
Does not incorporate invasive species (lack of
consistent data across study area)
Used model outputs to adjust sagebrush map
based on successional structure (where early
successional structure=<5% sagebrush and is
likely grassland type)
Model results
102-148% change in
fire frequency and
extent with UKMO
scenario
€ Decrease of ~5
million hectares of
sagebrush by 2099
€ Likely due to
increase in
frequency of fires
€
UKMO Climate Model: Change in Ha of
Sagebrush Vegetation
Millions Ha
20
15
10
5
0
2008
2030
2070
Years
2100
Caveats & improvements
€
Pulses at future time
steps
y Model likely
underestimating
decrease
€
Resolution of
climate
data/availability of
variables
Time series of future
climate
€ Incorporate invasive
species data
(cheatgrass)
€ Incorporate changes
to future fire regimes
€ Incorporate growth
scenarios
€
West Nile virus modeling
€
Sage grouse move to water
earlier during drought
y Increased infection rates
y Increased standing water due
to CBM development
ArcGIS .NET degree-day
model
€ Estimates WNv potential
using spatially explicit
weather station data
€ Identifies completion of
Extrinsic Incubation Period
for the virus (=time it takes
for the virus to reach
infectivity)
€
West Nile virus modeling
Temperature threshold for Culex tarsalis=14.3°C
with EIP for transmission=109°C/day
€ Uses preceding 12 days of temp data to calculate
accumulated degree days and determine whether
the threshold has been exceeded
€ If labeled yes, then WNv transmission is supported
at that location
€ Data inputs
€
y Daily tmax and tmin data from 54 locations across MT and
WY for June-Sept
y Future data: Tavg calculated for current data, compared to
Tavg from future 12-km resolution data, difference added
to (or subtracted from) Tmax and Tmin for 2007 to show
spatial variation in changes
Conclusions
As magnitude of temperature change
increases, likely to see increased
occurrence of WNv, particularly in
northern Rockies
€ WNv is likely to spread to areas of best
sage-grouse habitat
€ Areas of best sage-grouse habitat
overlap with areas of intensive energy
development
€ Submit to new listing decision
€
Acknowledgements
HP and WWF Climate Adaptation
Program
€ Karen Short: LANDFIRE Program
€ Sarah Olimb: American Wildlands
€ Scott Miller and Sarah Konrad: Univ of
Wyoming Spatial Analysis of Watershed
and Landscape Systems Group
€
Download