Presentation at 2009 EID Meeting in Park City, Utah

Agent-based Modeling, Functional Connectivity, and Disease Transmission in Felines

Jeff A. Tracey, PhDSue VandeWoude, PhDKevin R. Crooks, PhD

Colorado State University

Introduction

● Agent-based Modeling (ABM)

● Role of host behavior in the ecology of infectious disease

– Movement and contact– Habitat fragmentation due to urbanization

Our Study System

● Species:– Bobcats (Lynx rufus), mountain

lions (Puma concolor), domestic cats (Felis catus)

● Locations:– Southern California, Colorado,

Florida, ...● Diseases:

– FIV, plague, calicivirus, toxoplasmosis, others

Bobcats in Southern California

HABITAT

BOBCAT

NON-HABITAT

Questions

● How does movement behavior interact with habitat fragmentation to affect contact rates and spread of an SI disease?

● How does movement behavior and landscape structure affect the transmission of the disease within and between habitat patches?

Agent-based Models (ABMs)● a.k.a. Individual-based Models (IBMs)

● Agent-level:– Individual bobcats– Sensing, decisions, behavior– Interactions

● System-level: emergence; collect data, summarize

● “Computational laboratories”

Start Simple...

● No birth, death, or life history● No social interactions● Agents have same move rule● Simple SI, direct contact disease (FIV-like)

● Incrementally, we will add detail / specificity

ABM Components

● Landscape● Agents● Disease● Networks

– Contact– Disease– Later, will add social networks

Hig

hLo

w

High LowFragmentation

Hab

itat A

mou

nt

Movement Rules

● Parts of movement rule– Landscape response– Basic movement rule

● Basic movement rules:– Random– Home range-like– Dispersal-like

● More realistic rules...

Simulation Experiments

● Vary parameters– Disease

● Proportion of initially infective individuals● Probability of disease transmission

– Landscape● Amount of habitat● Fragmentation

– Movement● Non-habitat preference● Move rule and parameters

Agent Locations and State

Contact Network

Disease Network

Simulation Experiments

● Summarize system-level properties:– Proportion infective at end of simulation– Proportion of within-patch transmissions– Proportion of between-patch transmissions– Proportion of patches with infective individuals– many others...

Movement Behavior Matters

30 % Habitat

Variation Due to Movement Behavior Only

10 % Habitat

Variation Due to Movement Behavior Only

Effects of Movement Can Be Nonlinear

Example: Between patch transmission Highest at:● Intermediate non-

habitat permeability● Intermediate

directionality for dispersal-like movement

Landscape: highly fragmented, 30% habitat

Directionality

Fragmentation Does Not Always Slow Disease Spread

● Effects of landscape can have opposite effects on prevalence depending on movement

Dispersal-like Movement50 % Habitat

Landscape Change as a Dynamic Process

● Therefore:– Connectivity and

movement should change

– Disease dynamics should change

● More vulnerable at particular stages?

● Species-specific

Western Futures ModelsDr. Dave Theobald, CSU

Landscape Change as a Dynamic Process

● Therefore:– Connectivity and

movement should change

– Disease dynamics should change

● More vulnerable at particular stages?

● Species-specific

Western Futures ModelsDr. Dave Theobald, CSU

Landscape Change as a Dynamic Process

● Therefore:– Connectivity and

movement should change

– Disease dynamics should change

● More vulnerable at particular stages?

● Species-specific

Western Futures ModelsDr. Dave Theobald, CSU

Landscape Change as a Dynamic Process

● Therefore:– Connectivity and

movement should change

– Disease dynamics should change

● More vulnerable at particular stages?

● Species-specific

Western Futures ModelsDr. Dave Theobald, CSU

Important Points:

● Agent-based modeling approach– Computational laboratory

● Movement matters– Can be nonlinear, counter-intuitive

● Simulation software

Acknowledgments

Funding:● National Science

Foundation● Burroughs Wellcome

Travel Scholarship

Other CSU Team Members:● Mo Salman, Mike Lapin● Sarah Bevins● Justin Lee, Jesse Lewis,

Martha MacMillan, Linda Sweanor, Robert Alonso

Dr. Jun Zhu (UW-Madison/CSU)

Collaborators From:● UC Davis, University of Florida, San Diego

State University, US Geological Survey, NIH, CDC, NPS, USFWS, Colorado Division of Wildlife

Thank you to the conference organizers!

Thanks

Website: http://feline-eid.colostate.eduE-mail: [email protected]