Transcript Stochastic colonization and extinction of microbial

Stochastic colonization and
extinction of microbial species on
marine aggregates
Andrew Kramer
Odum School of
Ecology
University of Georgia
Collaborators:
John Drake
Maille Lyons
Fred Dobbs
Photo by Maille Lyons
Dynamics of small populations
• Extinction
• Invasion
• Outbreaks
Woodland caribou
Gypsy moth caterpillar
biology.mcgill.ca
Important characteristics:
- stochastic fluctuations
- positive density dependence
(Allee effects)
Tools
• Experiments: zooplankton, bacteria (planned)
• Computer models
– Stochasticity crucial
– Simulation approaches
• Programmed in R and Matlab
• Parallelization to speed computation time
– Computing time remains substantial
• No experience with individual-based approaches
– Want to relax assumptions, such as no inter-individual
variation
Bacteria on marine aggregates
• Lifespan: days to weeks (Alldredge and Silver 1988, Kiorboe 2001)
– Carry material out of water column
www-modeling.marsci.uga.edu
• Variable size, shape, porosity
• Microbial community on aggregate:
–
–
–
–
bacteria
phytoplankton
flagellates
ciliates
Aggregates and disease
• Enriched in bacteria
– Active colonization
– Higher replication (e.g. 6x higher
(Grossart et al. 2003)
• Favorable microhabitat for waterborne,
human pathogens
– Vibrio sp., E. Coli, Enterococcus,textbookofbacteriology.net
Shigella, and others (Lyons et al 2007)
)
Pathogen presence and
dynamics
• When will pathogenic bacteria be present?
– Source of bacteria
– Aggregate characteristics
– Extinction?
• How many pathogenic bacteria?
– Predation
– Competition
– Colonization/Detachment
Pathogen dynamics model
(Non-linear stochastic birth-death process)
Permanent
Detachment
Birth
Predation attachment
dPU
F
  B PD   PU   B PU 
FPU   PU
dt
1   F F BT
Pathogen
dPA
F
  PU   PA 
FPA
dt
1   F F BT
dBU
F
  B BD   BU   B BU 
FBU   BU
dt
1   F F BT
Bacterial
community dBA
F
  BU   BA 
FBA
dt
1   F F BT
C
F
dF
Flagellate
  F FD  YF
FBT   F F 
CF
dt
1



B
1



F
consumer
F F T
C C
C
dC
Ciliate
 C CD  YC
FC   C C
top predator dt
1   C C F
Colonization
•
Gillespie’s direct
method:
1. Random time step
2. Single event occurs
3. Length of step and
identity of event
depend on probability
of each event
•
Assumptions:
1. Well-mixed
2. No variation among
species
3. No variation within
species
(modified from Kiorboe 2003)
Representative trajectories for 0.01 cm radius aggregate
Higher density (1000/ml)
Extinctions
Low density (10/ml)
Motivations and challenges
• Increased understanding of importance of
individual variation in bacteria
• Computational techniques
– Scaling up
– Model validation, model-data comparison
• Unpracticed with individual-based and
spatially explicit modeling techniques
Possible further application:
• Aggregate as mechanical vector
– Extend pathogen lifespan
– Transport
– Facilitate accumulation
in shellfish (Kach and Ward 2008)
www.toptenz.net
• Shellfish uptake, agent-based model
– What scale? Shellfish bed or individual
animal?
Discussion
Knowledge gaps
• Pathogens are average?
– Density
– Colonization, extinction
• Does extinction occur?
– Yes
• On what time scale?
– Is it longer than aggregate persistence?
Testing the models
• Experimental tests
– Isolate mechanisms
– Measure parameters for prediction
• Use new techniques to parameterize
stochastic models with data
– Particle filtering method to estimate maximum
likelihood
Hypotheses
• Are species-specific traits important?
– Detachment
• Are aggregates a source of new pathogen?
– Mortality
– Competition (Grossart et al 2004a,b)
– Predation
• Do pathogens interact with aggregates in
distinct ways?
Implications
• Identify new environmental correlates for human
risk
• Quantification of human exposure and infection
risk
• Surveillance techniques for current and
emerging waterborne pathogens
• Improved control:
– hydrological connections between pollution source
and shellfish beds
– Aggregate formation and lifespan (e.g. mixing)