Transcript Treating populations and landscapes as signals old

Treating populations and
landscapes as signals.
A step towards
research collaborations?
(1) spread of disease,
(2) animal welfare vs transport,
(3) endangered species, oak landscape
(4) climate effects on lichens.
Associate Professor Uno Wennergren
IFM, Theory and Modelling,
Division Theoretical Biology
Who?
• PhD Annie Jonsson: University of Skövde.
• PhD Lars Westerberg: Linköping University
• PhD students:
– Frida Lögdberg
• Synchronization between patches
• Oak landscape
– Tom Lindström
• Dispersal, movements
• Spread of disease, Foot and Mouth disease
– Nina Håkansson
• Spatial patterns, temporal aspects also
• Animal transport, Networks, oak landscape
– Jenny Lennartssom
• Spatial patterns
• Spread of disease, Networks
• Collaborations:
– Kristin Palmqvist Anna Cabrajic Umeå Univ – lichens
– Mikael Rönnqvist Matthias Henningsson, Bergen and IEE – transport
– Susanne Lewerin Maria Nöremark, National Veterinary Institute (SVA)
Ecology 2008, Ecological modelling 2003, 2005, 2008. Oikos 1995, Nature 1995.
Outline
•
Conceptual framework
–
–
–
•
Complexity
Methodological questions
Relate to the scientific field Theoretical Biology
Examples by our projects
(1)
(2)
(3)
(4)
spread of disease,
animal welfare vs transport,
endangered species, oak landscape
climate effects on lichens.
Conceptual framework
In signal (time):
Population filter:
Out signal (time):
Temperature
Humidity
Other population densities
etc
Reproduction
Survival
Growth
Dispersal
Population density
• What characteristics of in signal relates to
specific characteristics of out signal (increase
risk of explosion or extinction)?
• What impact do the characteristics of the
population have on this relation on in and out
signal?
Conceptual framework
adding complexity
In signal
Population filter:
Out signal:
Temperature
Humidity
Other population densities
etc
Reproduction
Survival
Growth
Dispersal
Population density
Spatial domain:
Populations exist in a 2 dimensional heterogeneous
landscape (or even 3D). Hence the signals are in 2D.
Characteristics of 2D signals?
Predation and competition between populations:
Sets of interacting populations is the filter:
Characteristics of sets of out signals?
The effect of the characteristics of interactions, feedbacks?
Conceptual framework
methodological questions, part I
In signal
Population filter:
Out signal:
Temperature
Humidity
Other population densities
etc
Reproduction
Survival
Growth
Dispersal
Population density
Spatial domain and sets of population
What defines the characteristics of the signals?
What characteristics are important (extinction/explosion)?
variance
mean
autocorrelation/aggregation
synchronization
Conceptual framework
methodological questions, part II
In signal
Population filter:
Out signal:
Spatial domain and sets of populations
What defines the characteristics of the signals?
What characteristics are important (extinction/explosion)?
variance mean
autocorrelation-1/f noise-flicker noise , in time and space
synchronization between subpopulations
How to generate and analyze:
variance mean
In 1 dim,
autocorrelation
2 dim and…..
synchronization
FFT
FFT vs
Science in Theoretical Biology
• Analyzing time series to estimate 1/f noise of densities
• Testing different in signals and measuring impact on
probability of extinction
• Few studies on the relation between insignal and
outsignal measured by change of frequency spectrum
• Few studies (one or two) on resonance
– within system populations
– between system and insignal
• Few studies on how to generate or analyse time series
and landscapes by FFT with desired properties
• No studies made on landscape of resources (in signal)
and landscapes of densities (out signal) by FFT
– single populations
– Sets of populations
Our projects as examples
(1) spread of disease
(2) animal welfare vs transport
(3) endangered species, oak landscape
(4) climate effects on lichens
spread of disease
• Foot and Mouth disease (recall outbreak in Britain)
– Funded by Swedish Emergency Agency
• What matters to the spead of spread for a given
disease?
– Connection between farms (Dispersal)
– Location of farms (landscape)
since assuming that connection is distance dependent
– What location pattern is risky? (increase the spread)
• Describe the actual pattern
• Test different patterns, virtual farm locations
Farms in Skåne (region southern
Sweden)
• Spread of disease
• Foot and Mouth
disease
• Blue Tongue
• Etc
• 1/fγ noise
• Slope
gamma= 1.2023
• What spread of disease
to expect? (replicates)
• What actions are most
effective?
• For what dispersal/agent?
• Animal transport
strategies and
slaughterhouses.
Log amplitude
Farms in Skåne (region southern
Sweden)
Log frequency
Generating Coordinates
Generate by starting with random (white noise)
tilt the line in the frequency plane
By inverse Fourier Transform go back to landscape
Example on generating
• Different slopes in the
frequency plane
• Continous or ’binary’
landscapes
• Different amount of
primary habitat
Gamma=0
Continuous
landscapes
viewed
from the
side
Continuous
landscapes
viewed
from above
Digitalized
landscapes
with 10%
preferred
habitat
Digitalized
landscapes
with 40%
preferred
habitat
Gamma=1
Gamma=2
Open questions
• Improve the generating algorithm
– Aggregations in different directions, keep the
1/f possible?
– Improve how aggregation and density of
locations can be generated.
• Use the same technique for analyzing
invasion of new species in new areas (to
be expected by climate change)
• ?
Animal welfare vs transport
funded by Swedish Animal Welfare agency and Swedish Board of
Agriculture
• Animal transport between farms during routes to
slaughterhouses is an animal welfare issue.
• Can these routes be improved?
– From an animal welfare perspective?
– Is there a conflict with profit?
• What is the effect by the production system:
– Locations and specificity of slaughterhouses
– Locations and specificity of farms
– Trends today?
Generate production system - landscapes
Locations of farms and slaughterhouses
Endangered species living on old
oaks, the oak landscape
• What landscape increase the risk of extinction?
• What environmental noise (growth rates)
increase the risk of extinction.
– Degree of synchrony between subpopulations
• Too many oaks of same age
– Autocorrelation in time of environmental noise
• Known variation
– Dispersal strategy of organism (not included in this
presentation)
– Location of subpopulations (oaks)
Oaks in Östergötland (region close
to Linköping)
• 30 000 oaks
• What aggregation –
landscape?
• Oaks
• Species living on
oaks
• Management plans
Oaks in Östergötland (region close
to Linköping)
• Can the pattern be
improved?
• For what kind of
dispersal/species?
• What temporal pattern to
expect given age
structure of todays oaks?
Log amplitude
• 1/fγ noise
• the slope gamma=1.0125
Log frequency
Autocorrelation in time
Synchronization between subpopulations
• 1/f noise and variance. Different variance
in time vs subpopulations.
• Initial study: no geographic locations of
subpopulations.
• Different models of single population
growth. Density dependence.
Generated in signal, time series of
growth rates for subpopulations
Growth rate
insignal
time
How we generate – filtering a white
noise
Variance filter
amplitude
White noise
Filtered white noise
Noise filter
Questions
• How do synchronization effect the frequency
spectra of single and total population
• How do synchronization effect the extinction
risk of single and total population
• What is the combined effect of
synchronization and autocorrelation (time)?
subpopulation growth
and dispersal
Densitydep growth
(neg feedback)
?
Coming up..
• Time and landscape – 3D.
• Sets of populations. Resonance.
Lichens
• Platismatia glauca (Näverlav)
• Both fungi and green algae (and/or
’blue-green algae’)
• Needs specific combination of
– Humidity
– Light
– Temperature
to photosynthesis
A day of a lichen in June
Northern Sweden
Humidity
Can we describe the different
correlation patterns between
Humidity, Light and temperature?
For different kind of habitats
as edge and core?
Light,
edge of forest
Get rid of 24 h pattern/signal.
Temperature
Light,
Core of forest
Core habitat
Photosythesis
Light
Watercontent
Summing up
• Analyze, manipulate and generate time series or
landscapes or spatiotemporal patterns.
• Analyze time series and landscape and
spatiotemporal pattern from inherent population
level properties (feedback etc)
• Combine the two to seek for overall system
properties (resonances etc)