Can microbial functional traits predict
the response and resilience of
decomposition to global change?
Steve Allison
UC Irvine
Ecology and Evolutionary Biology
Earth System Science
allisons@uci.edu
Project goals
• Determine how microbial taxa respond to reduced
precipitation and increased N
• Determine the distribution of enzyme genes among
taxa
• Predict enzyme function and litter decomp based on
first two goals
• Test if microbial communities are resilient to
environmental change
Project design
Plot
Litter origin
A
Nitrogen
experiment
A
A
N
N
A
A
Ambient
N
Nitrogen enriched
N
Precip reduced
Mic. comm. origin
Precip
experiment
A
P
A
P
A
Ambient
A
P
A
P
N
Nitrogen enriched
P
Precip reduced
B
inoculation
2012
2011
Dec Feb
June
Dec Feb
June
2013
Dec Feb
composition samples
additional
samples
Allison lab responsibilities
• Litter mass remaining
• Fungal and bacterial counts
• Microscopy (fungi), flow cytometer (bacteria)
• Extracellular enzyme activities
• Litterbag and plot-level
• Litter chemistry
• nIR, C/N analysis
• Decomposition model
Litter mass remaining: Drought
• Microbes from reduced water leave more mass
remaining (6-12 months)
• Less mass loss in reduced water plots (6 months)
Microbe Origin (P=0.013)
Plot Effect (P=0.005)
100
Percent Mass Remaining
Percent Mass Remaining
100
90
80
70
60
90
80
70
60
X
R
X
R
Litter mass remaining: N addition
• Significant plot by litter interactions that differ at 6
vs. 12 months
Plot By Litter Interaction (P=0.008)
Plot By Litter Interaction (P=0.034)
100
100
Litter Origin
Litter Origin
Plot Effect
Percent Mass Remaining
Percent Mass Remaining
Plot Effect
90
80
70
60
90
80
70
60
XX
XN
NX
NN
XX
XN
NX
NN
Fungal counts: Drought
• More fungi in reduced water plots (3-6 months)
• Significant and contradictory microbial origin effects
Plot Effect (P=0.032)
10
Fungi/mg Litter
8
6
4
2
0
X
R
Bacterial counts: Drought
• Strong negative effects of reduced water; microbial
origin effect disappears by 6 months
Litter Origin (P=0.000)
2.0
2.0
1.5
1.5
Bacteria/g Litter x 10^9
Bacteria/g Litter x 10^9
Plot Effect (P=0.000)
1.0
0.5
0.0
1.0
0.5
0.0
X
R
X
R
Bacterial counts: N addition
• Positive effect of N in litter origin at 6 months
Litter Origin (P=0.000)
Bacteria/g Litter x 10^9
2.0
1.5
1.0
0.5
0.0
X
N
Enzymes: Drought
• Higher activities of all hydrolytic enzymes except LAP
Plot Effect (P=0.000)
10
2.5
8
2.0
Leucine aminopeptidase
Cellobiohydrolase
Plot Effect (P=0.000)
6
4
1.5
1.0
2
0.5
0
0.0
X
R
X
R
Enzymes: N addition
• Higher LAP in fertilized litter; other effects are weak
Litter Origin (P=0.000)
Leucine aminopeptidase
2.5
2.0
1.5
1.0
0.5
0.0
X
N
Initial litter chemistry
• Similar for litter from control and added N plots
• Litter from reduced water plots has more lignin,
protein, labile compounds; less cellulose and
hemicellulose
• Some differences are maintained after 3 months:
Litter Origin (P=0.000)
Litter Origin (P=0.000)
5
14
12
4
Sugars
Lignin
10
8
6
3
2
4
1
2
0
0
X
R
X
R
Litter chemistry: Drought
• 3-6 months: relatively more labile constituents
remaining in reduced water plots
Plot Effect (P=0.000)
Plot Effect (P=0.016)
8
14
12
10
Lignin
Crude protein
6
4
8
6
4
2
2
0
0
X
R
X
R
Litter chemistry: N addition
• Greater lignin loss in litter from N plots (6 months)
Litter Origin (P=0.000)
14
12
Lignin
10
8
6
4
2
0
X
N
Data summary
• Reduced water effects generally stronger than N
effects
• Direct effects of plot on decomposition generally
stronger than indirect effects on plants and microbes
• Reduced water favors fungi over bacteria, slows
decomposition, and allows enzymes and labile
substrates to accumulate
Project goal: model integration
• Incorporate disturbance responses and gene
distributions into a model
• Predict response of litter decomposition to
treatments
• Validate model with reciprocal transplant results
Approaches to modeling decomposition
Exponential decay (Olson 1963)
Schimel and Weintraub (2003)
Moorhead and Sinsabaugh (2006)
“Guild decomposition model”
(functional groups)
What is a “trait-based” model?
• Organisms are represented explicitly (biomass,
physiology, etc.)
• Each taxon possesses a specific set of trait values
• Trait values can be randomly chosen and/or
empirically derived
• Community composition
is an emergent property
www.brooklyn.cuny.edu
Represented traits
• Extracellular enzymes and uptake proteins:
•
•
•
•
Gene presence/absence
Vmax, Km
Specificity
Production and maintenance costs
• Carbon use efficiency
• Cellular stoichiometry
• Dispersal distance
www-news.uchicago.edu
Model structure
Example question and application
• Under what conditions are generalist versus
specialist strategies favored?
• Generalist = broad range of enzymes produced
Specialist
Generalist
Model set-up
• 100 taxa, 100 x 100 grid
• Taxa may possess 0 to 20 enzymes
• 12 chemical substrates (approximates fresh litter)
• Each degraded by at least 1 enzyme
Enzymes
20
1 0
1
0
… 0
0
0
100 1
0
0
Taxa
…
12
2.5
0
… 0
0
1.2
20 1.7
0
0
1
Enzymes
…
1
Substrates
1 0
Vmax
values
Model set-up
• Equivalent uptake across taxa
• Could also implement uptake matrices
…
20
1 0
1
0
… 0
0
0
100 1
0
0
Taxa
1
Monomers
Transporters
Transporters
…
14
2.5
0
… 0
0
1.2
20 1.7
0
0
1
1 0
Vmax
values
Model experiments
•
•
•
•
Simulate leaf litter decomposition (no inputs)
Test effect of tradeoffs in enzyme traits
Increase litter N or lignin
Model validation with Hawaiian litter
Model results
3
Microbial density [ log10(mg cm )]
• Taxa vary in density over time (succession)
0.5
0.0
-0.5
-1.0
-1.5
-2.0
-2.5
0.0
0.5
1.0
1.5
log10(days)
2.0
2.5
Model results
• Should be selection to link uptake with enzymes
No correlation
Maximum density [ log10(mg cm )]
1.0
3
3
Maximum density [ log10(mg cm )]
Enzymes and uptake correlated
0.5
0.0
-0.5
-1.0
-1.5
-2.0
0
5
10
15
Number of enzyme genes
20
1.0
0.5
0.0
-0.5
-1.0
-1.5
-2.0
0
5
10
15
Number of enzyme genes
20
Model results
• Species interactions are present but vary by taxon
and model conditions
0.25
Average correlation
0.20
0.15
0.10
0.05
0.00
-0.05
-2.0
-1.5
-1.0
-0.5
0.0
0.5
3
Maximum density [log10(mg cm )]
Model validation
• Fits are better for decomposition than enzymes
12
Unfertilized
Fertilized
Empirical CBH activity
Empirical k-value (1/yr)
10
8
6
R2 = 0.81
P < 0.001
Slope = 1.7±0.2
4
2
Unfertilized
Fertilized
R2 = 0.35
P < 0.001
10
8
6
4
2
outliers
0
0
0
2
4
6
8
Model k-value (1/yr)
10
0.0
0.1
0.2
0.3
0.4
0.5
Model CBH activity
0.6
Model summary
• Enzyme genes and uptake proteins should be
correlated
• Species interactions may be important
• Empirical and genomic data can tell us about
tradeoffs, trait correlations, and trait distributions
Thank you!
NSF ATB, DOE BER, audience