Swedish Meteorological and Hydrological Institute
A simple application of a complex
ecosystem model
Sofia SARAIVA
[email protected]
[email protected]
The role of bivalves in the Balgzand: first steps on an integrated modelling approach.
S. Saraiva1, L. Fernandes2, J. van der Meer3,4, R. Neves5, S.A.L.M. Kooijman4
1Swedish
Meteorological and Hydrological Institute, Norrköping Sweden
Modulers, Estrada Principal, 29 r/c Paz, 2640-583 Mafra, Portugal
3 Royal Netherlands Institute for Sea Research (NIOZ), Texel, The Netherlands
4 Vrije Universiteit, Dept. of Theoretical Biology, Amsterdam, The Netherlands
5Instituto Superior Tecnico, Lisboa, Portugal
2Action
Funded by
<<<<<<<<<<<<<<
(SFRH/BD/44448/2008)
Bivalves
Mytilus edulis
Questions
. Can we predict the
optimal growth of bivalves
(space, time)?
. Can we predict the impact
of bivalves cultures in the
ecosystem dynamics?
…
Source: Food and Agriculture Organization of the United Nations, FAO,Fishery Statistics, 2006
April 2009
Sofia Saraiva
Bivalves
P N
C
C
N
N
P
C
C
N
N
P N
P
P
C
C
Questions
. Can we quantify the role of bivalves in the biogeochemical cycle of
nutrients?
. Can we predict the impact of environmental changes
(natural/anthropogenic) in the organisms/system performance?
…
April 2009
Sofia Saraiva
Bivalves
Disentangle the processes
Ecological Processes
Hydrodynamic
Food
Availability
Biogeochemical
processes
Residence Time
Particulate Matter
Erosion/Deposition
Tide, Wind
Fresh Water Discharges
Density driven currents
Bathymetry and morphology
Organisms
Light
Nutrients (N/P)
Temperature
Individual processes
Metabolism
Food
Temperature
Ecological processes
Population processes
Inter-species competition
Predation
Competition for food
Competition for space
How should we model it?
Sofia Saraiva
Bivalves
Modelling approach
Hydrodynamic and
biogeochemical
model
Couple
Size-structured
population model
Individual Based Population Model
www.mohid.com
Dynamic Energy Budgets theory
Set of coupled models
Hydrodynamic, eulerian and
lagrangian advection-diffusion,
sediment transport and
biogeochemical/ecological models
Sofia Saraiva
Why do we need a
size structure population model?
The bivalve processes depend on the organism size
Man
Predation
N
P
C
P N
C
Predation
Birds
Pelagic phase
Physical processes
P N
C
Settlement
Physical processes
Substrate
Hydrodynamic processes
Tide, Wind
Fresh Water Discharges
Density driven currents
April 2009
Predation
Adult bivalves
Shrimp
Fish
Crab
Cannibalism
Abiotic factors
Shrimp
Light
Fish
Temperature
Substract
Competition
Intra & Inter
species
Ecological processes
Primary production
Biogeochemical cycles
Predation/Competition
Sofia Saraiva
Individual
Bivalve, DEB
Model validation
FILTRATION
Zooplankton
Phytoplankton
Organic Matter
Sediments
PSEUDOFAECES
INGESTION
Feeding Processes Model
(Saraiva et al., 2011, Ecol.Mod.)
FAECES
ASSIMILATION
RESERVES
SOMATIC
MAINTENANCE
MATURITY
MAINTENANCE
k
MOBILIZATION
GROWTH
STRUCTURE
Standard DEB Model
(Kooijman, 2010)
1-k
INORGANIC
COMPOUNDS
CO2
H2O
O2
NH3
PO4
REPRODUCTION
REPRODUCTION
BUFFER
MATURITY
SPAWNING
GAMETES
N
Parameter set for blue mussel (Saraiva et al., 2011, JSR)
P
C
Individual model validation (Saraiva et al., 2012, MEPS)
April 2009
Sofia Saraiva
Size-structured population model
DEB model
Spawning event generates a new cohort
Individual model for each cohort
Population is the sum of all the cohorts in the system
Add mortality by different causes
Aging
Initial Egg Mortality
Background Mortality
Extreme Starvation
Velocity
Settlement
Age limit
Viability of the gametes
Diseases, storms, local food depletion…
After long periods of starvation (not
enough food to cope with maintenance
costs)
> 0.5 m/s
Population
. Temperature
. Reproduction Buffer content (GSR)
Predators
. Predator identification
. Predator abundance
. Predator intake rate
. Predator size range
preference
Settlement probability
Sofia Saraiva
Population
Tide defines the population state
Theoretical setup
. one newborn mussel
. one food type, constant
. 12°C, 0.5 mgC/l
. Aging and background mortality (0.005 d-1)
1 – with
2 – without tide
. Tide changes the food availability:
during low tide the mussel bed is above
water an individuals are not able to feed
but still cope with maintenance
Frequent periods without food => frequent use
of reserves and reproduction buffer for
maintenance => less spawning events => lower
number of individuals
. With tide: smaller organisms, lower
number of individuals, lower biomass
April 2009
Sofia Saraiva
Population parameters
(more) Realistic setup
Population
Schematically simulate a mussel bed in the Balgzand
Higher uncertainty on the population model parameters, mainly
concerning mortality:
 Initial Egg mortality, adim [0,1]
viability of the gametes; exchange between mussel beds
 Background mortality, /d
diseases, storms, local food depletion
 Predators:
• Fraction of mussels in shrimp diet, adim [0,1]
• Fraction of mussels in crabs diet , adim [0,1]
• Fraction of mussels in birds diet , adim [0,1]
2004 data on mussel bed contours in the Balgzand (IMARES)
Bivalves: from individual to population modelling (Saraiva et al., 2014, JSR)
Sofia Saraiva
Population
Several scenarios were performed to cover
the range of mortality related parameters
60000 simulations
Population persistence
. Some parameters combination are not realistic because they lead to the population extinction
after 30 years
. Only in a small ‘region’ of the parameter space the predation by shrimps and crabs do not
determine persistence
. The effect of birds predation is not determinant for mussels population extinction
Sofia Saraiva
Population
Parameter selection
Realistic criteria
. Persistence
. The number of spawning events should be higher than
14 -> at least one spawning every two years
. The maximum size in the end of the simulation should
be higher than 1.2 cm (adult size), because there are
adults in the system
‘best scenario’:
Fraction of mussels in shrimp diet – 0.6
Fraction of mussels in crab diet – 0.2
Fraction of mussels in bird diet – 0.5/1
Background mortality
– 0.008 /d
Initial egg mortality
– 0.9
. Mussels abundance in the end of the simulation
should be less than 3000ind/m2, because that is the
maximum abundance found in the mussel bed
. Lowest error, compared with observation on total
number of mussels
Sofia Saraiva
Population
Model vs. Field Observations
(more realistic setup)
. Not all the larvae and spat
peaks are predicted by the
model (possible contribution
from other mussel beds?)
> 70 µm
[0.03-0.15] cm
. The timing of the peaks is
reasonable predicted by the
model
. The model is not able to
predict satisfactory the
number of individuals in the
population
de Vooys (1999)
April 2009
Rob Dekker (unpublished data)
Sofia Saraiva
Discussion on population model
stand alone results
Population
. The high number of individuals can be the result of an underestimation of the early life stage
mortality (larvae, spat)
. Possible causes for difference between model and field observations:
. No exchange with other mussel beds (food, predators and mussel larvae)
. No cannibalism (adult mussels feeding on mussel larvae)
. Differences in the food between mussel bed and data used
. Differences in the predators abundance and intake values (average values were used)
. Field data observations (used for model input and model comparison) were measured at
different sites
Couple with ecosystem model?
April 2009
Sofia Saraiva
ECOSYSTEM MODEL
www.mohid.com
HYDRODYNAMICS
TRANSPORT
BIOGEOCHEMICAL
PROCESSES
Nitrogen, Phosphorus
FOOD UPTAKE
PSEUDOFAECES
FAECES
INORGANIC FLUXES
POPULATION MODEL
Temperature
Salinity
Phytoplankton
Zooplankton
Particulate organic matter
Sediments
Nutrients (N and P)
(…)
FOOD
TEMPERATURE
CURRENTS
BIVALVE MODULE
Dynamic Energy Budgets
COHORTS
…
Each cohort
Size
Biomass
Condition
New Cohorts
April 2009
Ecosystem
INDIVIDUAL MODEL
Number Cohorts
Number Individuals
BIRTH
MORTALITY
PREDATION
3D model
FEEDING
MOBILIZATION
GROWTH
REPRODUCTION
MAINTENANCE
SPAWNING
Sofia Saraiva
Mussel density
Growth
Time
April 2009
Sofia Saraiva
Site Study
Study the role of bivalves in the ecosystem
Balgzand, Wadden Sea, The Netherlands
. Intertidal area, 50 km2, in the Wadden Sea
. Ecological relevant: stopover for migrating birds and nursery ground for North Sea fish
. Bivalves are a major component (more than 50% of the macrozoobenthos)
. Long term sampling program (since 1970)
. Many research projects
. Deltares OpenEarth website
Good study area for the implementation
of the integrated modelling tool
(Saraiva et al., in press, EcoMod)
Sofia Saraiva
The role of bivalves in the Balgzand
Site Study
Methodology
April 2009
Sofia Saraiva
Mussel Density
Results
. The spawning season starts exactly when temperature rises above the threshold (9.6 C)
. Spawning events are almost continuous during spring, summer and beginning of autumn
. Dispersion is important
. Only a few new born cohorts persist, most of the new cohorts die in the first month
. Intense larvae predation (adults and later on by shrimps) limits the number of new cohorts
April 2009
Sofia Saraiva
Balgzand
. Most of the new cohorts die in the first month
. Starvation is the main cause of biomass loss (98% ) total predation is about 0.1%
. But cannibalism has an extreme influence -> very high values of instantaneous mortality rate (105)
.The intense effect of cannibalism associated with the shrimps predation can result in the extinction of cohorts
April 2009
Sofia Saraiva
Balgzand
Mass fluxes exchange between areas
April 2009
Sofia Saraiva
Site Study
Mussel Beds effect on pelagic system
. Without mussels: output flux would be
15% more than the input flux
. The Balgzand is an area of intense
primary production, that would even
exports biomass without mussels
. Phosphorus: net consumption in both
scenarios (but more intense in the scenario
with mussels)
. Ammonia: net export in both scenarios
. Ammonia regeneration fuel primary
production and even to export about 40%
more that the input flux
. Suggests intense recycling of ammonia,
by mineralization of organic matter
April 2009
Sofia Saraiva
Site Study
Conclusions at the Balgzand
 There is no single mortality factor for the bivalve population
dynamics regulation:
• in the larvae stage, predation by adult mussels and shrimp (topdown) is very important and controls the persistence of the new
cohorts
• starvation (bottom-up) is the main responsible for bivalve
biomass loss over the year.
 By using a scenario without mussel beds the study quantifies their
effect over local biogeochemical processes. In the Balgzand:
-sink of phytoplankton (would be a source without mussels)
-source of ammonia (mussel intensify the export)
Quantification only possible with the complex model
Sofia Saraiva
Overview
It’s simple because it can be more complex!
 More species: oysters, macoma, other organisms, predators
 More food types: phytoplakton groups
 Larger domain, longer runs
Why would we do it?
 To be able to compare with data
 Describe the main processes
 To make projections on the system dynamics
 Short: next years, test management strategies
 Long time: decades, management and climate changes
Sofia Saraiva
Overview
Projections
Processes
description
Input data
UNCERTAINTIES
We must recognize the uncertainties to
minimize or take them into account
Use different models
Use different parameters
SCENARIOS
Use different
data sources
Ensemble
Sofia Saraiva
Overview
 Methodology to go from the individual model to population
to ecosystem modelling
 The model (as tool) is able to be use in other configurations, other
species, other sites, prepared to switch on and off processes for
processes studies
 The integration of the different levels and processes enable to
quantify the influence of bivalves in the Balgzand area
 Dealing with many processes, many data increases uncertainty
 How do we deal with it? Performing scenarios, produce ranges
Sofia Saraiva
Swedish Meteorological and Hydrological Institute
Thank you!!!
Sofia SARAIVA
[email protected]
[email protected]
Funded by
<<<<<<<<<<<<<<
(SFRH/BD/44448/2008)
. Temperature, food availability
(phytoplankton) and physical
conditions (emersion time and
water depth) control mussel bed
dynamics.
. After 1 year cohort 1 length
ranges from 2 cm to 3.5 cm (0.85
was the initial value)
. Model vs. observations is not very
clear (high variability and sparse
data)
. High variability within a mussel
bed (timing and intensity of the
spawning event)
. In most mussel beds, densities
and biomass are in the same order
of magnitude, although the
biomass results seem to slightly
deviate
April 2009
Mussel Beds
Results