Systems Ecology for Aquatic Resource Management
Transcript of Systems Ecology for Aquatic Resource Management
6/27/13
1
Systems Ecology for Aquatic Resource Management
Systems Ecology Beijing 2013
Systems Ecology Case Studies
Earth System
Earth is an open system (both energy and matter can cross the boundary), but we often treat it as a closed system
Hydrosphere
6/27/13
2
Water Quantity
USGS
All water
Liquid fresh water
Water in Lakes and Rivers
1,386,000,000 km3
Relative Spheres
10,633,450 km3
(99% is ground water)
http://ga.water.usgs.gov/edu/2010/gallery/global-water-volume.html
70% of earth surface is covered by water
Hydrosphere Oceans and freshwater bodies on Earth
Jacobson et al. 2000
World Ocean Currents Water Cycle
Fig 2.15
6/27/13
3
Importance of Water
http://www.lccdnet.org/wp-content/uploads/2011/01/importance-of-water-conservation.jpg
Importance for life, biogeochemistry, ecosystem services
Aquatic Ecosystem Services
Food Production
Waste water processing
Nutrient Recycling
Recreation
Transportation
Drinking Water
6/27/13
4
Case Studies
1. Shrimp Trawling in Core Sound, NC (USA) 2. Lake Sidney Lanier, GA (USA) 3. Neuse River Estuary, NC (USA) 4. Cape Fear River Estuary, NC (USA)
Case Studies
1. Shrimp Trawling in Core Sound, NC (USA) 2. Lake Sidney Lanier, GA (USA) 3. Neuse River Estuary, NC (USA) 4. Cape Fear River Estuary, NC (USA)
6/27/13
5
Ecosystem Impacts of Shrimp Trawling
American Fisheries Society 2012
Borrett et al. in prep.
Jeff Johnson
Joe Luckovich
Becky Deehr Stuart Borrett
Application of Throughflow Centrality
Systems Ecology and Ecoinformatics Laboratory
Shrimp – An important Fishery
FAO 2008
Shrimp images from fishwatch.gov
$425 mil yr-1
Estimated USA Economic Value
Deehr 2012 NCDMF
6/27/13
6
Ecosystem Impacts of Shrimp Trawling
Warner et al. 2004 http://coresound.com/
Direct impacts on shrimp populations
Bycatch (5:1)
Physical disruption of benthic habitats
http://www4.ncsu.edu/
Impacts
Skimmer Trawls and Otter Trawls
TEDs, BRDs
Regulate Access (space & time)
Solutions
Gear Type
Deehr 2012
Core Sound NC and Shrimp Nursery Areas
Google Earth Image from 2011
Cor
e So
und
Nursery Areas Closed to Shrimping
6/27/13
7
Objectives & Approach Assess the whole ecosystem impact of shrimp trawling on the Core Sound, NC ecosystem
• Direct and indirect effects (e.g., trophic cascades) • Focus: Relative functional importance of species (T)
• Ecosystem understanding à adaptive management
Approach • Construct food web based ecosystem models
• Open and Closed to trawling • Compare areas with Ecological Network Analysis
Systems Ecology and Ecoinformatics Laboratory
Food web ecosystem model • Ecopath model
• n = 65 compartments (g C m-2) • 2 non-living compartments: bycatch and
detritus
• L = 614 carbon flows (g C m-2 y-1) • e.g., feeding relationships
• Parameterized with primary data and literature values as needed • NCDMF Trip Ticket Program
• Ecopath estimated trophic level corroborated with stable isotopes
http://core.ecu.edu/BIOL/luczkovichj/core_sound/core_sound.html
Atl. Croaker Atl. Menhaden
6/27/13
8
Models for Comparisons
Spring (April, May, June)
Fall (Aug., Sept., Oct.)
Closed Least impact
Open Most impact
Systems Ecology and Ecoinformatics Laboratory
ENA Statistics (response variables) Example Network
~T = (Tj) =nX
i=1
fij + yj
FCI =Cycled
TST
APL =TSTP
zi
TST =X
Tj Like GDP
Like multiplier effect
Recycling
6/27/13
9
Centralization characterizes the concentration or dispersion of the centrality (throughflow activity)
FC =
Pni (Tmax
� Tj)
n� 1
Where Tmax is the maximum value f Tj
Lower FC à More dispersed Higher FC à More centralized
Interpretation
Systems Ecology and Ecoinformatics Laboratory
TST, FCI, and APL
Fall > Spring Open > Closed in Spring Closed > Open in Fall
There is more TST per unit input in the Spring-Closed model APL is less with Trawling
Cycling is less in the Fall Cycling is less with Trawling
Tota
l Sys
tem
Th
rou
gh
flow
Systems Ecology and Ecoinformatics Laboratory
6/27/13
10
Throughflow Centrality - Spring Closed to Trawling
Open to Trawling
Units: gC m-2 yr-1
The two sites have a difference in the magnitude and distribution of throughflow in the ecosystem
N90% TST (n = 3) Detritus, Meiofauna, Phytoplankton
N90% TST (n = 8) Detritus, Phytoplankton, Meiofauna, Atlantic Menhaden, Benthic Bacteria, Benthic Macroalgae, Sea Cucumbers, Benthic Microalgae
Impact (Δt) = TOpen - TClosed
Systems Ecology and Ecoinformatics Laboratory
52+, 11- Net difference: -217 Fall
Impact of Trawling on T
Showing differences greater than 0.5 gC m-2 yr-1
More species have an increase in throughflow than decline Net impact is negative in Fall
45+, 16-
Impact differs by season; bigger in Fall
Net difference: +25 Spring
Systems Ecology and Ecoinformatics Laboratory
6/27/13
11
Centralization
Ecosystems open to trawling are less centralized.
The ecosystems are more
centralized in the Fall
Units of dispersion are the same as T
The disturbance spreads the activity across more species.
Systems Ecology and Ecoinformatics Laboratory
Summary & Discussion • Detritus and Phytoplankton are consistently two of the
most central nodes – Meiofauna and Atlantic Menhaden
• Trawling appears to stimulate throughflow activity in more compartments than are reduced
• Magnitude of negative impacts were greater with trawling in Fall, less in Spring – Suggests closing nursery areas may be appropriate
• Shrimp trawling tends to de-centralize the ecosystem activity – Unexpected consequence
• Network analysis shows whole ecosystem impacts (direct and indirect) of Shrimp trawling
Systems Ecology and Ecoinformatics Laboratory
6/27/13
12
Case Studies
1. Shrimp Trawling in Core Sound, NC (USA) 2. Lake Sidney Lanier, GA (USA) 3. Neuse River Estuary, NC (USA) 4. Cape Fear River Estuary, NC (USA)
Lake Sidney Lanier, GA, USA
http://www.mylakelanier.com/xsites/Agents/dianathomsen/content/uploadedFiles/lakelanier1.jpg
Metropolitan Population 2011 5,457,831 (Wikipedia)
Atlanta
6/27/13
13
Lake Sidney Lanier, GA, USA • 150 km2 Reservoir, 1958 • Uses
– Drinking Water Supply – Waste Water Discharge – Recreation – Navigation
• Challenges – Competing demands, stakeholders – Small watershed, sensitive to environmental
variation – Eutrophication, excess Phosphorus – Data limitations
Project Goals • Characterize Lake Lanier ecosystem • Construct a phosphorus model to evaluate
system state – Consider data uncertainty
• Apply Network Environ Analysis – Whole system indicators – Characterize sensitivity of indicators to model
uncertainty
Borrett & Osidele. 2007. Ecological Modelling
Kaufman & Borrett. 2010. Ecological Modelling
6/27/13
14
Conceptual Model: P in Reservoir
Modified from Osidele & Beck (2004) Consistent currency: Phosphorus Units: Flow: mg P m-2 d-1 Storage: mg P m-2
Model Construction • Constant Structure
(conceptual Model) à – 11 compartments – 26 within system connections – 5 boundary inputs, 4 boundary losses
• Identify Plausible Model Parameterizations – Used Regionalized Sensitivity Analysis to identify
parameterizations that fit known empirical data for SRP, phytoplankton, fish, and detritus.
– Donor controlled transfer functions – Monte Carlo Simulations (500, 1000, 5000)
• Randomly select parameters from uniform distribution • Range based on biological knowledge • Compare model behavior to observed system behavior
Variable Flow and Storage à Plausible parameterizations
6/27/13
15
Examples of Plausible Model Simulations
Range defined by empirical observations
Epilimnion SRP Phytoplankton
Larval-juvenile Fish
Plausible models generated behavior within known constraints.
Model Sample Size Plausible Models
500 14
1000 18
5000 90
Total 122
Next: Apply ENA to models
NEA Whole System Indicators
6/27/13
16
Indicator Results
14, 18, 90 models
1. 90 models is sufficient sample size of plausible models,
2. Considerable uncertainty remains
3. Network non-locality, aggradation, homogenization, and amplification are present
4. Average Phosphorus recycling 39%
Conclusions
Indicator Correlations
Borrett & Osidele 2007
Not all Indicators are independent!
Indicator redundancy? Significance?
6/27/13
17
Statistical Factor Analysis
2 key latent factors Explain 80% variation
Borrett & Osidele 2007
Conclusions (1) • Small variability in the ecosystem indicators lets us
– circumvent part of the modeling and data uncertainty – draw more robust conclusions regarding the condition of
the Lake Lanier ecosystem
• FCI, Indirect/Direct, IFI, AGG, and HMG are robust to to model uncertainty
• Internal processes heavily influence phosphorus flow and storage – Well developed ecosystem – P is well mixed (HMG) – Changing system dynamics would be difficult by simply
changing the nutrient inputs
6/27/13
18
Conclusions (2) • Common indicator
variation can be mapped onto two factors
• Tentative Interpretation – Factor 1
• system integration • Network growth • Growth Form II
– Factor 2 • Boundary growth • Growth Form 0
Case Studies
1. Shrimp Trawling in Core Sound, NC (USA) 2. Lake Sidney Lanier, GA (USA) 3. Neuse River Estuary, NC (USA) 4. Cape Fear River Estuary, NC (USA)
6/27/13
19
Temporal Variation of Indirect Effects
in the Neuse River Estuary
Neuse River Estuary, NC
http://www.coe.uncc.edu/~jdbowen/neem/
(Christian and Thomas 2003)
6/27/13
20
Objectives
Q: Should the 30% decrease in N loading rapidly alter the system’s trophic state?
H1: Indirect effects are dominant If indirect effects are dominant, then we would not expect rapid change
H2: Indirect effects vary seasonally; moderate inter-
annual variation discrete-time series analysis
Throughflow Decompositions & Indirect Flows
TST = Boundary + Direct + Indirect
Indirect/TST
Indirect/Direct
TST = non-Cycled + Cycled 2. (Finn 1976)
Finn Cycling Index = Cycled/TST
1.
Network Environ Analysis
6/27/13
21
TST and FCI
Sp S F W Sp S F W Sp S F W Sp S F W
Tota
l Sys
tem
Thr
ough
flow
(mm
ol N
m-2
sea
son-1
)
4000
8000
12000
16000
20000
24000
Finn
Cyc
ling
Inde
x
0.00
0.25
0.50
0.75
1.00
TST FCI
1985 1986 1987 1988 1989
Sp S F W Sp S F W Sp S F W Sp S F W
Indirect/Direct
0
25
50
75
100
125
150
175
1985 1986 1987 1988 1989
Indirect Effects
Indirect Flows Dominate
H1: Indirect flows are dominant
System Currency Indirect/Direct Sourcecold spring energy 1.02 Higashi & Patten 1989oyster reef, SC energy 1.58forest, Puerto Rico nitrogen 6.14Lake Lanier, GA phosphorus 7.45 Borrett & Osidele 2007
6/27/13
22
Temporal Variation
Little Temporal Variation
Spring Summer Fall Winter
Indirect/TST
0.0
0.5
1.0
Seasons
BA A A
yc1985 yc1986 yc1987 yc19880.0
0.5
1.0
Years A A A A
H2: Indirect flows vary seasonally; moderate inter-annual variation
Summary & Conclusions
• Indirect flow dominates direct
• Stable ecosystem organization
• N load reduction will not have rapid effect on trophic state of estuary
6/27/13
23
Case Studies
1. Shrimp Trawling in Core Sound, NC (USA) 2. Lake Sidney Lanier, GA (USA) 3. Neuse River Estuary, NC (USA) 4. Cape Fear River Estuary, NC (USA)
Hines et al. 2012, Hines et al. in review
Nutrients in Estuarine Waters
Eutrophication
Recycling and removal of nutrients
(Costanza et al., 1998)
usbr.gov oceanservice.noaa.gov
eutrophication.yolasite.com
Ecosystem Services
6/27/13
24
Nitrogen Cycling and Removal
N2
NH4 NO3
NH4
NO2
NO3 NO2
N2 Atmosphere
Water
Sediment
DNRA
Fixation
Nitrification
Nitrification
Anammox
De
nitrifica
tion
Simplified Nitrogen Cycle
Upstream
Coupled Nitrification-Denitrification Coupled DNRA-Anammox
Coupled processes à indirect pathways
Complications
Sea level rise
treehugger.com Average rise of 1.7mm year-1
(IPCC, 2007)
earthtrends.wri.org, IPCC 2007
Salt water intrusion (Giese et al. 1985)
6/27/13
25
Effects on Microbes
arizona.edu
Hypotheses H1: Coupled nitrification-denitrification will be higher in the oligohaline portion of the estuary compared to polyhaline sites
H2: Coupled DNRA-anammox will be lower in the oligohaline portion of the estuary compared to polyhaline sites
H3: Microbial nitrogen removal capacity will be higher in the oligohaline portion of the estuary compared to polyhaline sites
6/27/13
26
Study Sites
Wilmington
Wilmington
Conceptual Model
Hines et al. 2012
6/27/13
27
Wilmington
Conceptual Model
Low Network Model
W- NH4
W- NOx
W- M
W- ON
S- NH4
S- NOx
S- M
S- ON
3.1
5.2
136.
5
1.7
2.1
1.9
9.8 7.4
119.
2
16.0
850.
0 5.5
39.0
146.7
150.0
14.1
144.0
119.2
212.8
109.0 82.0
853.9
257.1
172.
0 0.3
3.9
276.0
1008.6
3.9e-5
1159.6 10
80.0
6.0
104.
1
130.0
5.0
(a) (b)
(c)
(d) (e)
3.9e
-7
79.0
173.
2
1238
.2
1020.0
1160.0
3.9e-5
1 2 3 4
5 6 7 8
10.5 27.6 1.0e-6 55.9
682.6 4.2 1.0e-5 7110.0
nmol N cm-3 d-1 Summer
Hines et al. 2012
6/27/13
28
Low Network Model
W- NH4
W- NOx
W- M
W- ON
S- NH4
S- NOx
S- M
S- ON
3.1
5.2
136.
5
1.7
2.1
1.9
9.8 7.4
119.
2
16.0
850.
0 5.5
39.0
146.7
150.0
14.1
144.0
119.2
212.8
109.0 82.0
853.9
257.1
172.
0 0.3
3.9
276.0
1008.6
3.9e-5
1159.6 10
80.0
6.0
104.
1
130.0
5.0
(a) (b)
(c)
(d) (e)
3.9e
-7
79.0
173.
2
1238
.2
1020.0
1160.0
3.9e-5
1 2 3 4
5 6 7 8
10.5 27.6 1.0e-6 55.9
682.6 4.2 1.0e-5 7110.0
Nitrification DNRA Denitrification Anammox Boundary
nmol N cm-3 d-1
Coupled NOx à NH4 NH4 à NOx NOx à N2 NH4 à NOx NOx à N2
NOx + NH4 à N2
Summer
Coupled NOx + NH4 à N2
NOx à NH4
Hines et al. 2012
Low Network Model
W- NH4
W- NOx
W- M
W- ON
S- NH4
S- NOx
S- M
S- ON
3.1
5.2
136.
5
1.7
2.1
1.9
9.8 7.4
119.
2
16.0
850.
0 5.5
39.0
146.7
150.0
14.1
144.0
119.2
212.8
109.0 82.0
853.9
257.1
172.
0 0.3
3.9
276.0
1008.6
3.9e-5
1159.6 10
80.0
6.0
104.
1
130.0
5.0
(a) (b)
(c)
(d) (e)
3.9e
-7
79.0
173.
2
1238
.2
1020.0
1160.0
3.9e-5
1 2 3 4
5 6 7 8
10.5 27.6 1.0e-6 55.9
682.6 4.2 1.0e-5 7110.0
nmol N cm-3 d-1 Summer
Hines et al. 2012
6/27/13
29
High Network Model
W- NH4
W- NOx
W- M
W- ON
S- NH4
S- NOx
S- M
S- ON
3.1
5.4
55.3
0.7
7.3
1.7
0.5 1.4
119.
2
0.5
423.
8 1.5
104.4
146.7
100.0
7.7
77.5
119.2
186.2
53.2 159.5
425.8
253.2
136.
7 7.8e
-3
2.0
132.4
308.9
3.9e-5
1246.8 10
06.9
127.
3
32.7
72.5
3.6
(a) (b)
(c)
(d) (e)
3.9e
-7
39.1
345.
5
975.
1
381.1
1255.1
3.9e-5
1 2 3 4
5 6 7 8
2.9 15.1 1.0e-6 49.28
107.9 14.3 1.0e-5 3724.4
nmol N cm-3 d-1 Summer
Polyhaline Model Parameterization Evaluation
77% of parameters were high or medium quality
6/27/13
30
Environs are the internal system environments - Input & Output for each node - sub-networks
ENA Environ Analysis
A
C
B
D
E
Input history of material in a given output
Analysis performed using NEA.m (Fath & Borrett, 2006) Patten 1978, Patten 1982
Input Environs
Denitrification Environ at Low
W- NH4
W- NOx
W- M
W- ON
S- NH4
S- NOx
S- M
S- ON
3.8e-3
6.4e-3
0.2
12.0e-3
16.0
e-3
43.0e-3
0.2 0.2
2.7
82.0e-3
4.4 0.3
2.0
7.7
7.9
7.5
77.1
3.1
5.5
2.8 2.1
10.3
3.1
172.
0
0.2
(b)
1.0
92.5
64.8
7.7
6.0
3.9e-5
1 2 3 4
5 6 7 8
10.5 27.6 1.0e-6 55.9
682.6 4.2 1.0e-5 7110.0
nmol N cm-3 d-1
Sub-Network
6/27/13
31
W- NH4
W- NOx
W- M
W- ON
S- NH4
S- NOx
S- M
S- ON
W- NH4
W- NOx
W- M
W- ON
S- NH4
S- NOx
S- M
S- ON
Hypothesis Testing H1: Coupled nitrification-denitrification will be higher in the oligohaline portion of the estuary compared to polyhaline sites
Hypothesis Testing H1: Coupled nitrification-denitrification will be higher in the oligohaline portion of the estuary compared to polyhaline sites
Uncoupled
Coupled to nitrification
De
nitr
ific
atio
n (
nm
ol N
cm
-3 d
-1)
% C
ou
ple
d N
itrifi
ca
tion
-De
nitr
ific
atio
n
6/27/13
32
Hypothesis Testing H2: Coupled DNRA-anammox will be lower in the oligohaline portion of the estuary compared to polyhaline sites
Uncoupled
Coupled to DNRA
An
am
mo
x (n
mo
l N c
m-3
d-1
)
% C
ou
ple
d D
NR
A-a
na
mm
ox
Findings H1: Coupled nitrification-denitrification will be higher in the oligohaline portion of the estuary compared to polyhaline sites
H2: Coupled DNRA-anammox will be lower in the oligohaline portion of the estuary compared to polyhaline sites
6/27/13
33
Hypothesis Testing H3: Microbial nitrogen removal capacity will be higher in the oligohaline portion of the estuary compared to polyhaline sites
Uncoupled denitrification
Denitrification coupled to nitrification
Uncoupled anammax
Anammox coupled to DNRA
N R
em
ova
l (n
mo
l N c
m-3
d-1
)
% N
Re
mo
ved
Wilmington
Implications
• Rising sea level may result in a decreased coupling of nitrification-denitrification
• DNRA coupled to anammox may become more important
• Overall N2 removal may
not be significantly altered
6/27/13
34
Additional Examples
Urban Water Metabolism Beijing, China Sarmato, Italy
Bodini and Bondavalli 1992 Zhang et al. 2010
Considered options to improve system sustainability Scenario Analysis
Evaluate changes from 2003 to 2007 Apply Utility analysis to characterize changing nature of relationships.
6/27/13
35
Hydrology Okefenokee Swamp, Georgia, USA
Uplands Lowlands
Surfa
ce
Su
b-S
urfa
ce
Patten & Matis 1982
Characterizes flow of water through the system Origin & Fate
Mao et al. 2013
Virtual Water Trade Virtual water = water embodied in food
Yang et al. 2012
Analyzed complex interdependencies among regions
Used control and utility analysis (ENA)
“Virtual water is useful as it globalizes perspectives on water scarcity, ecological sustainability, food production, and consumption”
6/27/13
36
Summary Examples of using systems ecology and ecological network analysis to study water resources
– Water quality – Water quantity – Aquatic ecosystem structure and function – Ecosystem services
• Model Construction & Evaluation • Systems Analysis • Applications of ENA are still a frontier of the
science