A dynamic species modeling approach to assessing impacts from climate change on vegetation Lydia P Ries Biogeography Lab Bren School of Environmental Sciences.

A dynamic species modeling approach to assessing impacts

from climate change on vegetation

Lydia P RiesBiogeography Lab

Bren School of Environmental Sciences & ManagementUniversity of California, Santa Barbara

ICESS SeminarApril 29 2008

Post-doctoral Collaborators:Frank Davis, David Stoms Biogeography Lab, Bren

Lee Hannah Conservation International, Bren

International Collaborators:Guy Midgley Kirstenbosch Research Institute,

South African National Biodiversity Institute

Ian Davies Australian National UniversityResearch School of Biological Sciences

Wilfried Thuiller Université J. Fourier,

Laboratoire d'Ecologie Alpine

Changwan Seo Seoul National University

Climate change…

…is affecting ecosystems worldwide.

species distribution

phenology

community associations

Now what?

Euphydryas editha

-assess climate change impacts on CA biodiversity,

-develop a tool to estimate changes in potential and realized niches for individual species under climate change scenarios.

The PIER Ecosystem Modeling Project

Outline

Species distribution modeling

Dynamic species model development: BioMove

BioMove applications

Case study examples

Species distribution modeling

Lessons from the past: climate oscillates pollen records-species shift latitude and elevation shifts maintain biodiversity

Using same principles, model future response to climate change

Species Distribution Modeling (Thuiller et al. 2003)

• Building and validating individual species models

• Reducing uncertainty of models

• Predicting potential niches

Building & validating modelsBuilding & validating models

Selecting species & environmental dataSelecting species & environmental data

GLM/GAM/GBM/CART/ANN

Species Distribution Model Methods

Bayesian Approach

Climate data Soil data

Max Temp of Warmest Month

Min Temp of Coldest Month

Annual Temp Range

Mean Temp of Wettest Quarter

Mean Temp of Driest Quarter

Precipitation of Wettest Quarter

Precipitation of Driest Quarter

Available Water Capacity

Soil depth

Soil pH

Salinity

Depth of water table

+

Species Distribution Model Methods

Building & validating modelsBuilding & validating models

Selecting species & environmental dataSelecting species & environmental data

Projecting models to the future environmentProjecting models to the future environment

Estimating the change of species range Estimating the change of species range

GLM/GAM/GBM/CART/ANN

Optimizing model outputsOptimizing model outputsConsensus Model

Bayesian Approach

Species Distribution Model Methods

Quercus agrifoliaHadley A2 Scenario

Present

Quercus agrifoliaHadley A2 Scenario

20802080

loss

gain

Species Distribution Model Output To Date:

314 Species

13 Pinus spp.15 Quercus spp.13 Ceanothus spp.9 Arctostaphylous spp.

89 California Endemic

Mojave

Mixed conifer

Coastal sage scrub

Endemic RichnessCurrent

2050

high

low

Species distribution modeling

Lessons from the past:

climate oscillates

pollen records-species shift

On what time scale?

Rapid events

Reid’s paradox (observed > calculated)

Other mechanisms of movement?

What about…Dispersal?

Short and long distance events

Competition?

Plant functional type-age and size classes, fecundity, mortality, germination

Disturbance?

Fire, grazing

BioMove; a dynamic modeling approach

A spatially explicit model used to predict the movement of individual species

Simulates dispersal, competition and disturbance

Outline

Species distribution modeling

Dynamic species model development: BioMove

BioMove applications

Case study examples

Model Overview

Species demographic model

PFTssuccession model

Initial Conditions

Dispersal model

Disturbance model

Annual time-step

PFTs Species

BIOMOVE

Competition for light

Vegetation structure

Global Changemodel

Dispersal model

Disturbance model

Global Changemodel

Species demographic model

PFTssuccession model

Initial Conditions

Dispersal model

Disturbance model

Annual time-step

PFTs Species

BIOMOVE

Competition for light

Vegetation structure

Global Changemodel

Dispersal model

Disturbance model

Global Changemodel

Demographic model

Death

DispersalSub-Model: Dispersal

CompetitionSub-Model: Competition

Sub-Model:

Climate Change

DisturbanceSub-Model: Disturbance

Outline

Species distribution modeling

Dynamic species model development: BioMove

BioMove applications

Case study examples

Model Applications

1. Assess threat to species from climate change

• 2 % listed at risk from CC• based on bioclimatic envelopes• extinction risk study assumptions

Model Applications

2. Management coordination and decision support:

• Impact of fire on vegetation• Conservation area designation (leading, trailing)• Species of concern (threatened, small populations)• Land use planning

Zaca Fire, 3 August 2007

Model Applications

3. Invasive species assessment

• Studies on conservation areas, agriculture, forestry– Global (Ficetola et al. 2007, Richardson & Thuiller 2007)– Regional (Parker-Allie et al. 2007)

• Need to incorporate spatial and

temporal dynamics

Bromus madritensisCalflora.net

Model Applications

4. Advanced climate change research

• The climate is changing fast…and so are the models!– inter-disciplinary model development– GHG stabilization

BioMove: a tool for assessing GHG stabilization scenarios

GHG Stabilization Targets:

EU: 2° C global mean temperature change (~450 ppm, IPCC 2007)

U.S.: USCAP 450-550 ppm CO2 eq.

Targets set to avoid ‘dangerous interference’ in the climate system, e.g. avoiding mass extinctions

Current levels 420-480 ppm CO2 eq.

(Pew Center on Global Climate Change, 2007)

BioMove: a tool for assessing GHG stabilization scenarios

Time

CO

2 Em

issi

ons

(ppm

) 550 ppm

450 ppm

Outline

Species distribution modeling

Dynamic species model development: BioMove

BioMove applications

Case study examples

BioMove; A Case StudyPinus lambertiana

“This is the noblest pine yet discovered, surpassing all others not merely in size but also in kingly beauty and majesty. “ (John Muir, 1894)

http://www.lib.berkeley.edu/EART/tour/TahoeNationalForest.gif

Species distribution model results: Pinus lambertiana

A2 B2

C

H

2010

BioMove; A Case Study Pinus lambertiana 2010

BioMove; A Case Study Pinus lambertiana 2050

BioMove; A Case Study Pinus lambertiana 2080

Disturbance: Fire• CA Dept. of Forestry and Fire Protection, Fire and

Resources Assessment Program (FRAP)• 54 Year Average Fire Frequency• 4 size classes

BioMove; A Case StudyPinus lambertiana

Zaca Fire, 3 August 2007

Quercus douglasii (Blue Oak)

Habitat Prob.

High

Low

loss

no change

gain

outside range0 90 180 270 36045Kilometers

0 90 180 270 36045Kilometers

Predicted current distribution Predicted change in distribution

Mixed blue oak habitat

Blue oak under climate change

B2 A2

C

H

Bromus madritensis (red brome)Lower elevations distribution

Fire effects from pine

Moving upslope

In summary

314 species library for distribution model outputs Coupled outputs with spatially explicit demographic

model Refined dynamic species model Predict climate driven shifts CA species Continue incorporating competition and disturbance

models Predict differences between GHG stabilization

projections Implications for management strategies, conservation

allocation

Future work

Predict differences between GHG stabilization projections

Implications for management strategies, conservation allocation

Collaborations with:San Diego County 2050 Project

CA Air Resources Board

IUCN

thank you

thank you