Transcript Slide 1

Motive
• Konza: understanding disease, since there is no
apparent reason to manage native pathogens of
native plants
• Also have background information in the event that a new
pathogen is introduced
• Also background comparison for disease severity compared
to severity for potential newly introduced host species
• Xoo: develop durable resistance in the short run
and developing mechanistic understanding of
what produces durable resistance to make its
development easier in the future in other
systems
• Root SIR: figure out infection risks and how to
manage disease
Motive in your system?
Hypotheses to be tested or
parameters to be estimated?
• Konza:
– Disease incidence is substantially lower in
dryer environments, such as upland sites
– Burning reduces disease incidence
– Pathogen reproduction is host frequency
dependent
• Xoo: The abundance of virulent isolates
will remain low if they experience a
substantial cost of virulence
• Root SIR: What is the transmission rate
between roots of different ages?
Hypotheses to be tested or
parameters to be estimated?
Need for long-term data
• Konza
– Surprising result that pathogen populations did not
bounce back quickly after drought
– Could have very different view of “importance of
disease” depending on when disease is sampled
– Also different ideas about how direct the effect of, for
example, moisture availability is for disease risk
– Background information for evaluating new diseases
• Xoo: by definition, durability of resistance must
be studied over long time periods
• Root SIR: Especially for perennial plants,
responses may change greatly over time; annual
environmental variability can be studied
Advantages of long-term data in
your system?
What is the inference space?
• Konza: All sampling within KPBS, some within
small experiments. We can suggest that the
Flint Hills will be similar to Konza… perhaps
even tallgrass prairie, in general?
• Xoo: All sampling at one experimental site in the
Philippines. We can suggest that it is
representative of at least the Philippines
• Root SIR: Experiment would need to be
performed in a controlled environment.
Inference outside that environment…?
What is the inference space in your system?
Experimental unit
• Konza: For some analyses, individual
plants; for other analyses, plots in which
treatments have been imposed
• Xoo: Individual bacterial isolates
• Root SIR: Individual roots? Individual
plants if treatments applied at that scale?
• In plants, the definition of individuals is
more flexible… leaves, genets, clones
Your experimental unit?
How do pathogens enter and leave
your study system?
What are the response variables?
• Konza: Disease incidence (per quadrat)
• Xoo: Lesion length on resistant and
susceptible host plants (per isolate)
• Root SIR: Number of roots susceptible,
number of roots infectious, and number of
roots resistant
What are your response variables?
What are the predictor variables?
• Konza: Topographic position, Precipitation
rate, Grazing (+/-), Burn return time,…
• Xoo: Host plant genotype,…
• Root SIR: Perhaps Plant age
What are your predictor variables?
What are appropriate statistical methods for
estimation of effects of predictor variables?
• Konza: Analysis of variance with repeated
measures
• Xoo: Analysis of variance
• Root SIR: Perhaps analysis of variance for
evaluation of environmental effects, etc.,
or other maximum likelihood methods
What are appropriate statistical methods for
estimation of effects of predictor variables?
What are potential sources of bias?
• Konza: some plant species were selected
because of observed disease levels – therefore,
questions about typical pathogen loads across
plant species could encounter bias
• Xoo: isolates would tend to be collected in the
field when lesions are readily visible – therefore,
questions about isolates may not be
approachable on a “per lesion” basis
• Root SIR: larger roots might maintain their
integrity during infection to a greater extent and
so be more readily sampled
Potential sources of bias in your system?
What desirable data are not available?
Are there widely accepted models
for these systems available?
• Konza: Can apply some models from
agricultural systems as initial hypotheses
• Xoo: Can modify Leonard’s
parameterization of the cost of virulence to
incorporate changes in plant resistance
with temperature
• Root SIR: use of SIR model allows
comparison to many systems
Are there widely accepted models
for your system available?
What form of sensitivity analysis
might be useful?
• Konza: Given disease incidence data, combine
with models of disease effects on plant
productivity to look at range of possible effects of
disease on plant community
• Xoo: Given pathogen responses to host
genotype, consider the possible effects of
pathogen bottlenecks during non-conducive
weather
• Root SIR: Determine potential importance of
plant age at beginning of epidemic, for example
What form of sensitivity analysis
might be useful?
What forms of model validation
might be useful?
• Konza: Use of replicate watersheds in a
formal statistical analyses functions as a
validation tool; with a longer time series,
could also see whether models based on
earlier years worked in later years
• Xoo: Perhaps not relevant here
• Root SIR: Validating predictions based on
controlled environment analyses in the
field would be useful
What forms of model validation
might be useful?