Post on 18-Jan-2018
description
Applying stochastic models of geographic evolution to
explain species-environment relationships of bats in the
New World
J. Sebastián Tello and Richard D. Stevens
Department of Biological SciencesLouisiana State University
Baton Rouge, LA 70803
Variation in species richness at broad geographic extents
Introduction
Bird richnessHawkins et al. 2007
e.g. NPP
Species Richness
Deterministic effects of environmentalconditions is a prominent hypotheses
Introduction
Strong and frequent species-environment correlations
Introduction
Field et al. 2009
Climate
/
Produc
tivity
Heterog
eneity
Nutrients
Prim
acy
adj.
R2
Correlative studies have used simple non-biological null hypotheses
Expected by chance?
Observed Relationship
Environmental variable Environmental variable
Ric
hnes
s
R2=0.513 R2=0.000
Introduction
Richness gradients are formed by the overlap of individual species distributions
Introduction
RichnessGradient
Species Distributions
Species distributions are the consequence of the geographic diversification of clades
Introduction
Species Distributions
Biogeographic Processes and Constraints
a. Confined geographic domain of distribution
b. Aggregated distributions
c. Geographic range movements
d. Cladogenesis:• Speciation• Extinction
Environment assumed to affects richness via fundamental biogeographic processes
Introduction
Distributions
Biogeographic Processes
Richness
Environment
Environment assumed to affects richness via fundamental biogeographic processes
Introduction
Distributions
Biogeographic Processes
Richness
Environment
Biogeographic processes not necessarily driven by environment
Introduction
Distributions
Biogeographic Processes
Richness
Stochasticity
Biogeographic processes not necessarily driven by environment
Introduction
Distributions
Biogeographic Processes
Richness
Environment
Stochasticity
Introduction
Biogeographic Processes
Richness
Environment
Stochasticity
How do species-environment relationships change when random biogeographic processes are considered?
? Null Model
Family Phyllostomidae
146 species
America divided in cells of 100 by 100 km
1. Species richness of bats by geographic range overlap
Methods
Correlation estimated with adjusted R2 values
2. Empirical richness-environment correlations
Methods
Correlation estimated with adjusted R2 values
Richness correlated against three variable sets:
a. Energyb. Heterogeneityc. Seasonality
2. Empirical richness-environment correlations
Methods
2. Empirical richness-environment correlations: Uncertainty estimation by bootstrapping
Methods
R2adj.
Freq
uenc
y
0 10.5
Methods
R2adj.
Freq
uenc
y
0 10.5
2. Empirical richness-environment correlations: Uncertainty estimation by bootstrapping
a. Computer simulations in R
b. Random biogeographic processes:
1. Range spread2. Range movement3. Speciation4. Extinction
c. Constrained domain: the New World (cells of 100 by 100 kms)
3. Create a null model of the geographic diversification of Phyllostomid bats
Methods
3. Create a null model
Methods Start
3. Create a null model
Methods Start
Domain colonization
Time = 1
3. Create a null model
Methods Start
Domain colonization
Ranges too small?Range growth
Time = 1
Yes
3. Create a null model
Methods Start
Domain colonization
Ranges too small?Range growth
Range movement
Time = 1
No
Yes
3. Create a null model
Methods Start
Domain colonization
Ranges too small?Range growth
Range movement
Speciations?Speciation
Time = 1
No
Yes
Yes
3. Create a null model
Methods Start
Domain colonization
Ranges too small?Range growth
Range movement
Speciations?
Extinctions?
Speciation
Extinction
Time = 1
No
Yes
Yes
YesNo
3. Create a null model
Methods Start
Domain colonization
Ranges too small?Range growth
Range movement
Speciations?
Extinctions?
Time limit reached?
Speciation
Extinction
Time = 1
No
Yes
Yes
YesNo
No
3. Create a null model
Methods Start
Domain colonization
Ranges too small?Range growth
Range movement
Speciations?
Extinctions?
Time limit reached?
Speciation
Extinction
Time + 1
Time = 1
No
Yes
Yes
YesNo
No
No
3. Create a null model
Methods Start
Domain colonization
Ranges too small?Range growth
Range movement
Speciations?
Extinctions?
Time limit reached?
Speciation
Extinction
Time + 1
Time = 1
End
No
Yes
Yes
Yes
Yes
No
No
No
Simulation Model Richness MapsSpecies Distributions
Methods3. Create a null model of the geographic
diversification of Phyllostomid bats
Methods3. Create a null model of the geographic
diversification of Phyllostomid bats
START
Methods
1000 null modelruns
3. Create a null model of the geographic diversification of Phyllostomid bats
START
END
Methods
R2adj.
Freq
uenc
y0 10.5
3. Create a null model of the geographic diversification of Phyllostomid bats
Methods4. Test for effects of environment using null model
R2adj.
Freq
uenc
y
0 10.5
R2adj.
Freq
uenc
y
0 10.5
Significant t-test Non-Significant t-testEnvironmental effect No environmental effect
Methods5. Calculate effect size using null model
R2adj.
Freq
uenc
y
0 10.5
Richness of Phyllostomid bats in the New World is strongly associated with the environment
adj.
R2
Energy
Heterog
eneity
Seasonal
ity
Results
All three environmental predictors have a significant effect
Results
R2adj. R2
adj. R2adj.
Freq
uenc
y
Energy Heterogeneity Seasonality
However, the relative importance changes significantly when using null model
Resultsad
j. R
2
Energy
Heterog
eneity
Seasonal
ityEner
gy
Heterog
eneity
Seasonal
ity
Hed
ges’
d
Naïve null hypotheses are not appropriate for testing species-environment relationships
Expected by chance?
Observed Relationship
R2=0.513
Environmental variable Environmental variable
Ric
hnes
s
R2=0.000
Conclusions
Conclusions
Geographical evolution null models produce much more appropriate null hypotheses
Conclusions
Geographical evolution null models can significantly modify results
Jim CroninBret ElderdKyle Harms
Eve McCullochMercedes GavilanezMaria SagotLori Patrick
?
Acknowledgements