Ecosystem Modeling Presentation

Development of an Atlantis ecosystem model
to study food web impacts of DWHOS
Cameron Ainsworth
USF College of Marine Science
April 29, 2014. NOAA Southeast Regional Office Brownbag
Overview
• Atlantis model
• GOM implementation
– Diet
– Biomass
– Other data
• Estimating oil response from DWH
• Integrated Ecosystem Assessment
• Collaborators
Atlantis
Biogeochemical deterministic ecosystem model
Fisheries
Target, bycatch, habitat effects,
ports, costs, compliance
Effort models: CPUE / cost based
Fleet 1
2
3
4
5
6
Ecology sub-models
Consumption / production, waste
production, migration, predation,
reproduction & recruitment, habitat
dependence and mortality
Diet
matrix
Physical and
biogenic habitat
Nutrients (Si, N)
Biogeochemistry
Oceanography sub-model
Climate and Oceanography
Nutrient
& waste
cycling
Hydrodynamics
Fulton, E. A., A. D. M. Smith, and C. R. Johnson. 2004. Biogeochemical marine ecosystem models i: Igbem - a model of marine bay ecosystems. Ecological
Modelling 174:267-307.
Atlantis
GOM polygons
• Bacteria to apex predators
(“end-to-end”)
• Irregular polygons
• Fully age structured
• Larval transport
• Space limitation
• Biogenic & physical habitat
• Nutrient and waste cycling
• Detailed fisheries accounting
Diet
Migration
Biomass
Life history
Baseline data
SEAMAP cruises
Fisheries
statistics
Habitat
Oceanography
Zooplankton bongo tows (Daly, USF)
Database/Source
Data Type
FWRI Diet Data Database (FWRI)
Sealifebase.org
Fishbase.org
ICCAT Manual http://www.iccat.es/en/
Animal Diversity Database
http://animaldiversity.ummz.umich.edu
NOAA Technical Memoranda
Diet composition by weight
Diet composition by weight
Diet composition by weight
Annual pelagic fish migration
Sealifebase.org
Fishbase.org
Animal Diversity Database
http://animaldiversity.ummz.umich.edu
SEDAR http://www.sefsc.noaa.gov/sedar/
National Marine Fisheries Services
http://www.nmfs.noaa.gov/pr/species/mam
mals
SEAMAP (FWRI)
SEFSC (NOAA) (Miami)
University of South Alabama
Annual bird migration
Turtles & mammals
Life history- wet weight, growth,
size, etc.
Life history- wet weight, growth,
size, etc.
Life history- growth, size, max age,
etc.
Catch/biomass
Total biomass- GOM specific
Fish & bycatch CPUE
Fishery log books
Reef fish abundance surveys
NODC (NOAA)
Temperature, salinity, oxygen
NCDDC (NOAA)
Hydrodynamics (FCOM)
MODIS (NASA) via GIOVANNI portal
http://disc.sci.gsfc.nasa.gov/giovanni/overvi Chl A
ew/index.html
NOAA Gulf of Mexico Data Atlas
Sediment / habitat
http://gulfatlas.noaa.gov/
FWRI
NMFS statistical zones, jurisdictions,
bathymetry, coral/seagrass
distribution, sediment
NOAA Office of Science and Technology
http://www.st.nmfs.noaa.gov
Catch statistics
ICCAT Manual http://www.iccat.es/en/
Catch statistics
Food web analysis
C-IMAGE longline surveys
• 136 stomachs analyzed in lab
• 19 under-sampled species (bycatch/deep)
Gut content analysis
Other data
235 spp.
15 spp.
905 spp.
FWRI Diet database
Oil impacts
(B. McMichaels)
Morphometrics for gape-limited feeding
Statistical fitting of diet compositions
Likelihood profiles
Bootstrapped data w/ fitted β distributions
Prey 1
Prey 2
Prey 3
Prey 4
Prey 2
Ainsworth, C.H., Kaplan, I.C., Levin, P.S. and Mangel, M. (2010) A statistical approach for estimating fish diet compositions from multiple data
sources: Gulf of California case study. Ecological Applications, 20(8): 2188-2202.
Gulf of Mexico food web
Partial food web; showing 20% minimum connectance
Michelle Masi, USF
Masi, M. and Ainsworth, C.H. (in press). Statistical Analysis to Provide a Probabilistic Representation of Fish Diet Compositions from Multiple Data
Sources: A Gulf of Mexico Case Study. Ecological Modelling.
Diet validation
Realized diet matrix is often not what you entered…
• Spatial co-occurrence constraint
• Changing concentrations of prey
• Changes in body size affect gape limitation
Diet proportions change dynamically
Error range
Atlantis vs. observed diet
Fulton et al. 2007
Fulton, E.A., Smith, A.D.M., and D.C. Smith. 2007. Alternative Management Strategies for Southeast Australian Commonwealth Fisheries: Stage 2: Quantitative
Management Strategy Evaluation. Australian Fisheries Management Authority, Fisheries Research and Development Corporation.
Diet continuation (MARFIN)
•
•
•
•
•
•
Additional stomach sampling by
FWC
Nitrogen isotope studies suggest
there are 2 competing pathways
for vertical energy transmission in
the Gulf
One based on benthic algae, one
based on phytoplankton
Dominance switches with water
clarity and nutrient loading
Links to sub-area food web study
(NRDA)
We are surveying about 30
species for which isotope ratios
have been studied (E. Peebles)
Enriched by TL
PATTERNS OF ENERGY FLOW
Planktonic source
Benthic source
Specimen collection
• Collaborating with FWC to make
use of their expertise, facilities
and inventory of samples
• Last year we worked with
SEAMAP trawl (Bob
McMichaels) and C-IMAGE
long-line (Steve Murawski) to
collect specimens but limited
resources for laboratory work
C-IMAGE longline surveys
• MARFIN: budgeted for a couple
road trips, will also bully Steve’s
students
SEAMAP cruises
Estimating biomass distributions
Other demersal fish
• Goal: to extrapolate species distributions
into unsampled areas for many functional
groups
• Generalized additive model (GAM)
describes abundance distribution of 39
fish & invert groups
• Abundance based on SEAMAP-NMFS
fisheries independent data (2005-2010)
Mike Drexler, USF
Predictor variables
Predictor variables Chl A, sediment type, dissolved O2, temperature & depth
Seasonal means due to data limitation
National Oceanographic Data Ceter (NODC), Gulf of Mexico Data Atlas, MODIS (NASA)
Bot. temp
Bot. oxy
Abundance
•
•
•
Depth
Sediment
e.g., abundance relationships fit using spline
Oxygen (%)
Chl A
Generalized additive model
g(η) ~ s(depth)+s(Chl a)+ s(temp)+ s(oxygen)+ factor(sediment type)+ offset(g(effort))
Shrimp abundance
•
η represents expected abundance
according to link function g
•
s is thin plate regression spline
•
Data is zero inflated
•
Negative binomial distribution with log link
•
2/3 of data used for model training
•
1/3 of data for model validation
•
Evaluates biomass at 1/12th degree &
averaged to polygon level
Averaged to polygon
Drexler M, Ainsworth CH (2013) Generalized Additive Models Used to Predict Species Abundance in the Gulf of Mexico: An Ecosystem Modeling Tool. PLoS ONE 8(5): e64458.
doi:10.1371/journal.pone.0064458
2nd application: Delta method
•
New paper Gruss et al. compares with Delta
method
•
WFS application
•
Uses binomial distribution with logit link to
model presence/absense
•
Uses Quasi-Poisson with log link to model
abundance when present
•
Multiplication of two distributions (delta method)
provides estimate of abundance
•
Delta method fits faster with a small loss in
predictive capability relative to neg. binomial
Binomial model
(presence/absence)
Logit link function
*
=
Quasi-Poisson
(abundance)
Log link function
Zero adjusted
estimate of
abundance
Gruss, A., Schirripa, M.J., Chagaris, D., Drexler, M., Simons, J., Verley, P., Shin, Y-J., Karnauskas, M., Olivoeros-Ramos, R., and Ainsworth, C. (in review). Evaluation of the
trophic structure of the West Florida Shelf in the 2000s using the ecosystem model OSMOSE. Journal of Marine Systems.
Larval model: connectivity
• Lagrangian passive drift model of larval transport
• Individual based model
• Driven by GoM HYCOM NCODA (Chassignet et al. 2007)
• Egg density calculated based on spawner biomass (from generalized
additive model, Drexler and Ainsworth 2011)
Mike Drexler, USF
• Source/destination matrix imported to Atlantis
• Provides connectivity for 46 spp.
Red grouper larvae
Provides connectivity
Larval model: oil recruitment impacts
Spawning schedule
Larval tracks
No oil mortality
Oil mortality
Oil simulations:
• Larval survivorship based on oil
interaction
• Oil location taken from far field
modelling within CIMAGE project
(Claire Paris, RSMAS)
• 100% mortality assumed
• Entered as recruitment anomaly in
Atlantis
Hydrodynamics
Contributed by NCDDC
Naval Research
Laboratory
1/25o GOM
HYCOM
Current,
temperature &
salinity fluxes
interpolated
across polygon
boundaries
• Russ Beard, Rost Parsons & Charles Carleton
• Developing an ‘ocean slicer’ tool that can be
used in all future GoM Atlantis models
• Interpolates current, temperature and salt
fluxes across polygon boundaries from a
physical model
• Uses Naval Oceanographic HYCOM data
• Currently looping 1 year of data, but could be
used for longer term forecasts
Identify oil concentrations
Oil trajectory modelling
• Far-field modelling by Clair Paris RSMAS
• Identify oil concentrations in 3D space
Hamburg
• 12 hour time steps
• Includes microbial degradation from IMS
Potentially benthic
exposure (Hollander)
• With and without dispersants
Satellite observations validate model
Hydrodynamic & oil behaviour modelling
Paris, C.B, Le Hénaff, M., Aman, Z.M., Subramaniam, A., Helgers, J., Wang, D., Kourafalou, V.H., Srinivasan, A. 2012. Evolution of the Macondo Well Blowout: Simulating the Effects of the
Circulation and Synthetic Dispersants on the Subsea Oil Transport. Environmental Science & Technology,; : 121203084426001 DOI: 10.1021/es303197h
Estimate functional response
Impacts on growth, recruitment and mortality (Mote)
• Fertility, immune function & DNA damage assays
• Linked to PAH concentration in tissue
• Life ate of interest based on known function within body
Impact on recruitment
Impacts on growth rate (USF)
• Growth rates before/after oiling
estimated from otoliths
Impacts on recruitment (USF)
• Oiled area compared with larval positions from individual
based model
Impact on
recruitment & growth
Combine curves to
estimate functional
response
Physiological impacts
New Atlantis code
• Developed by CSIRO (Nov 2013)
• Allows spatial forcing function for mortality, growth and recruitment
• Useful for oil, HABs and other spatial stressors
• Currently building input files
Lindsey
Dornberger, USF
Oiled area resolved by time step (12 hr) and depth
• Functional response applied within cells
• Affects mortality, growth or recruitment
rates
• Future work: uptake-decay model to
determine dynamic concentrations
Fishery closures
• Predator increase after fishing release
• Possible trophic cascades & top-down
effects
• Immediate closures and prolonged
reduced seafood demand
Model diagnostics
No fishing tuning scenario
Testing of the 1980 model
• Biomass of exploited groups increases
under no-fishing scenario
• Carrying capacity can be compared to
Bzero available from stock assessment
• Transient dynamics work themselves
out after 10-15 years
• Now in process of tuning historical
scenario
• Uses reconstructed fisheries catch,
spatial regulation and seasonal
hydrodynamics
Expected outputs
Recovery time (example data)
•
•
•
•
•
Variety of synoptic & species-specific indicators
Indicators developed for Gulf IEA with SEFSC-NOAA
Economic impacts for fisheries
Recovery time for species
Effect of dispersants
Ecosystem changes
Community structure changes
• Model uses prespill baseline
provided by ROV
surveys
• Validation
possible with
post-spill surveys
• W. Patterson
Next steps
1.) New hi-res Atlantis model for
Campeche Bay
• IXTOC an analog for DWHOS?
• Test hypotheses on shrimp failure
Campeche Bay
2.) Improved socioeconomics
• David Yoskowitz (Harte)
• Shore-based industry
impacts & indicators
o
o
o
o
o
o
o
o
o
o
Commercial harvesters
Primary dealers and processors
Seafood wholesalers and distributors
Grocers
Restaurants
Fuel service
Equipment retailers
Marinas
Hotels/motels/bed & breakfast
Boat building and repair
3.) MOSSFA (O2, benthic mortality)
4.) Coupling with Zoosim (Walsh/Lenes)
A nested suite of models
C-IMAGE / IEA
DEEP-C Atlantis model (NE GOM)
Felicia Coleman
Stephen Gosnell
•
Deep-C is funding Lindsey Dornberger to act as a liaison
with CIMAGE
•
Same oil forcing functions will be used in both models
•
Similar data and structure
•
Eventually, we may be able to feed boundary conditions
into Deep-C model
•
Focus: Mercury bioaccumulation
A nested suite of models
C-IMAGE / IEA
CSCOR - NCCOS
•
Doran Mason (GLERL) and post-doc Andrea Van Der
Woude developing a TX-MS-LA shelf model to look at
hypoxia effects (Dead zone)
•
Center for Sponsored Coastal Ocean Research (NCCOS)
sponsored research for reducing size of the hypoxic zone
by the Gulf of Mexico/Mississippi River Watershed Nutrient
Task Force.
•
Again, matches GOM Atlantis boundaries
Doran Mason
Andrea Vander Woude
Role of Atlantis in IEAs
IEA process includes 5 operational steps
1. Scoping
2. Indicator development
3. Risk analysis
4. Assessment
5. MSE
Ecosystem modelling supports steps 2-5
Other benefits:
• Expand Atlantis user base
• 6 students in training (USF & RSMAS)
• Integration of federal, state, regional &
academic data
Levin, P., Fogarty, M.J., Murawski, S., Fluharty, D. 2009. Integrated Ecosystem Assessments: Developing the Scientific Basis for Ecosystem-Based Management
of the Oceans. PLoS Biol 7(1): e1000014.
Management Strategy Evaluation (MSE)
Atlantis again considered “real world”
Simulates entire management process
•
•
•
•
•
Data gathering (with observation error)
Indicator calculation from pseudo-data
Assessment based on indicators
Enact harvest control rules (HCR)
Observe outcome & iterate to improve HCR
Observations
Ecology and
fishing
simulator
(Atlantis)
Requires models:
Population dynamics
Data collection
Analysis
Stock assessment
Integrated in Atlantis
Monitoring and
indicators
1 year
cycle
Assessments and
parameter estimation
Implementation
Management policies:
quotas, TAC, MPAs
Dealing with Uncertainty: A model ensemble
Atlantis
(USF)
OSMOSE
EwE
(NOAA)
(FWC)
• A model ensemble approach tests
model specification error
HAB mortality on gag grouper
• Contribution to SEDAR 33 (gag
grouper assessment) in 2013
Atlantis: biogeochemical inputs
OSMOSE: emergent diet matrix
EwE: trophic effects only
Atlantis: Ainsworth et al. in prep. NOAA Tech Memo. EWE: Chagaris 2014. OSMOSE: Grüss et al. in prep.
Collaborators
University of South Florida (Ainsworth, Murawski)
SEFSC-NOAA (Schirripa, Kelble, Zimmerman)
NWFSC-NOAA (Levin, Kaplan)
University of Miami (RSMAS) (Die, Babcock)
Florida State University (Coleman, Gosnell)
FWRI (Mahmoudi, Chagaris)
CSIRO (Fulton)
NCDDC (Beard, Parsons, Carleton)
& many others

Download Report