Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
An agent-based model of fisher behavior and fish
population dynamics in the Gulf of Mexico How does fisher behavior affect stock assessment?
Steve Saul, Ph.D.
Assistant Professor
ASU
2
Objectives
• Uncertainty in fisheries and human behavior. • West Florida Shelf agent-based simulation
model configuration and parameterization • Model results. • Stock assessment of simulation model results
and comparison with “known” simulation system dynamics.
• Results, discussion, and implications.
• Future research.
3
Fisher Behavior: Key Source of
Uncertainty in Fisheries
• Resource users sometimes respond to policies in an unintended way.
• Though uncertainty is widely accepted as a problem in fisheries management, attention has only focused on scientific uncertainty around the status of exploited resources.
• Uncertainty due to human responses to management has received much less attention.
• Human behavior dictates the spatial and temporal locations of fishery-dependent observations, often used to infer abundance trends in assessment models.
(See Fulton, et al. 2011 for review)
4
Catch per Unit Effort
Catch per unit effort (CPUE) is used as fishery dependent observations to determine a relative measure of population abundance if the following relation holds:
i
i
ii Nq
f
CU
Where U is the catch per unit of fishing effort, f and q is the catchability coefficient, the proportion of the population captured by a unit of fishing effort
5
Sources of Variation in CPUE
Characteristics of animal population
• Density (=abundance) • Spatial & temporal differences
• Behavior of animals • Aggregation behavior • Competition for gear space or bait • Physiological condition
Fishing power of fishing unit
• Gear characteristics • Vessel characteristics • Fishermen’s skills
CPUE is a stochastic process
• Environmental conditions
• Affecting gear performance • Affecting animal behavior
Economic/management drivers
• Regulated catches • Reporting accuracy • Market value of target species
6
Ability to use CPUE as index of abundance depends on being able to adjust for (i.e. remove) the impact on catch rates of changes over time from factors other than stock abundance
CPUE Standardization
Relative
abundance
• Spatially • Temporally
CPUE Observations
Other sources
of variation
7
Hypothesis Testing Through Simulation
• Reproduce fish population dynamics and fishing fleet behavioral dynamics – allow them to interact across space and time under realistic state conditions.
• Compare metrics derived from simulated fishing fleet observations, with the biological dynamics simulated in the system.
• Agent-based modeling well suited to simulate complex social and biological dynamics and their interactions.
• Fish and fishing vessels each modeled as individual agents.
8
Process of Study Fishing Fleets and Fish Modeled
9
-88 -87 -86 -85 -84 -83 -82 -81
2425
2627
2829
3031
Longitude
Latit
ude
Simulation Model Spatial Partitions20m Depth Strata
Stratum 1Stratum 2
Stratum 3
Stratum 4
Stratum 5
Stratum 6
Stratum 7Stratum 8
Stratum 9
Stratum 10
1 minute by 1 minute (1.8 km2) Grid
Simulation Model Spatial Partitions
20 Meter Depth Strata Shown 20 Meter Depth Strata Shown
Simulation Spatial Definitions
• Grouped Strata (): 3 groupings of Strata
• Strata (): areas of equal latitude where Florida’s coast runs north-south, areas of equal longitude where the coast runs east-west.
• Depth Sub-strata (): 20m isobaths
• Grids (): cells one minute latitude by one minute longitude
10
-84.0 -83.8 -83.6 -83.4 -83.2 -83.0
27.0
27.2
27.4
27.6
27.8
28.0
Red Grouper
Distributed With Spatial Autocorrelation
Spatial Distribution of Abundance
012345678
20 40 60 80 100
Sta
nd
ard
ize
d C
PU
E
Depth Strata (Meters)
c) Red Grouper
Grids 1 and 2 Grids 3 - 7 Grids 8 -10Strata Strata Strata
0.0 1.0 2.0 3.0
0.0
0.2
0.4
Distance (km)
Sem
ivar
ianc
e
a) Gag Grouper
0.0 1.0 2.0 3.0
0.0
0.2
0.4
Distance (km)
Sem
ivar
ianc
e
b) Mutton Snapper
0.0 1.0 2.0 3.0
0.0
0.2
0.4
Distance (km)
Sem
ivar
ianc
e
c) Red Grouper
(Saul et al. 2013. Modeling the spatial distribution of
commercially important reef fishes on the West Florida
Shelf. Fisheries Research 143: 12-20)
11
Validation
0 1 2 3 4 5
2000
4000
6000
8000
Range (km)
Sem
ivar
ianc
e Area 1Area 2Area 3Area 4Area 5Area 6Area 7Area 8Area 9Area 10Area 11
Variogram Fits to 11 Areas
12
Biased Random Walk Migration
Algorithm
Mortality
• Develop pathway to simulate movement of an individual fish to offshore habitat.
• Important part of reef fish life history.
• Fishing effort removes migrating individuals in transit to offshore habitats.
• Establish an age structured population that is properly distributed across space and time.
(Saul et al. 2013. An Individual-Based Model of
Ontogenetic Migration in Reef Fish Using a Biased
Random Walk. Trans. Am. Fish. Soc. 141: 1439-1452)
13
Biased Random Walk Migration
Algorithm
14
Tagging Simulation Results
Empirical Linear Speed
Grid Cells/Day
Freq
uenc
y
0.0 1.0 2.0 3.0
010
030
0
α = 0.7β = 0.2
Simulated Linear Speed
Grid Cells/Day
Freq
uenc
y
0.0 1.0 2.0 3.00
500
1500
α = 0.9β = 0.1
(Wilcoxon rank sum test: P=0.3764)
Red Grouper Example Results
15
Tagging Simulation Results
-85.0 -84.0 -83.0 -82.0
26.0
27.0
28.0
29.0
Longitude
Latit
ude
Empirical Migrating Fish
Red Grouper
-85.0 -84.0 -83.0 -82.0
26.0
27.0
28.0
29.0
LongitudeLa
titud
e
Simulated Migrating Fish
Red Grouper
Comparison of 350 randomly selected empirical and simulated red grouper recaptures from off the coast of
Mote Marine Lab (location of half the releases)
16
Tagging Validation
1 2 3 4 5 6
0.0
0.2
0.4
0.6
0.8
1.0
Age (years)
Pro
babi
lity
Mat
ure
a) Maturity at Age b) Simulated Total Migration Time
Migration Time (years)
Freq
uenc
y0 1 2 3 4 5
050
0015
000
1 3 5 7 9 11 14
Sed. AdultsMigratingSed. Juv.
c) Age and Stage of Tagging
Age (years)
Freq
uenc
y0
500
1500
2500
1 3 5 7 9 11 14
Sed. AdultsMigrating
d) Simulated Recapture Age
Age (years)
Freq
uenc
y0
200
600
1000
1 3 5 7 9 11 14
DiscardsCatch
e) Commercial Catch at Age
Age (years)
Per
cent
05
1015
1 3 5 7 9 12 14
f) Simulated Age at Arrival
Age (years)
Per
cent
010
2030
4050
17
Fish Population Parameterization
• Life history, starting age structure, and recruitment parameters taken from most recent stock assessment reports.
• Spatial distribution of fish abundance validated against the spatial distribution of fish habitat characteristics.
• Migration algorithm validated using tagging data.
18
10 20 30 40 50
0.00
0.04
0.08
Wind Speed (knots)
Prob
abilit
y
Regulated Fisheries OpenRegulated Fisheries Closed
Trip Choice HL: Panhandle
Wind SpeedFisher Decision-Making
• Discrete choice models developed for three decisions: participation, site choice, and trip termination.
• Questionnaire completed by vessel captains determined factors to test in model.
• Panel dataset developed to fit models using commercial logbook data, weather, fish price, vessel characteristics, expected revenue, fuel price, and regulations.
-88 -87 -86 -85 -84 -83 -82 -81
2425
2627
2829
3031
Longitude
Latit
ude
-88 -87 -86 -85 -84 -83 -82 -81
2425
2627
2829
3031
Fishing PortsFishing Sites
Fishing Ports and Sites
Gulf of Mexico
(Saul et al. Accepted. Modeling the decision making
behavior of fishers in the reef fish fishery on the West
Coast of Florida. Human Dimensions of Wildlife.)
19
Fisher Decision and CPUE Analysis
20 Year
Projection From 2005/2006
State of Fishery
(pre-IFQ)
When to fish
Where to fish
Return to port
Simulated Logbook Data
20
Model Validation
Source: Strelcheck and Farmer (2012),
Personal Communication,
NOAA Southeast Regional Office
21
CPUE Standardization
22
Catch at Size
Length Distribution of Catch
Length (mm)
Den
sity
0 200 400 600 800
0.00
00.
002
0.00
4Red Grouper Year 20
Length Distribution of Population
Length (mm)
Den
sity
0 200 400 600 800
0.00
00.
001
0.00
20.
003
Red Grouper Year 20
0
0.2
0.4
0.6
0.8
1
0 200 400 600 800
Sele
ctiv
ity
Length (mm)
Red Grouper HL SelectivityUsed in Saul Simulation Model as Derived from SEDAR 9
Selectivity
Size Limit
23
0
10,000
20,000
30,000
40,000
50,000
60,000
70,000
80,000
0 4 8 12 16 20
Met
ric
Ton
s
Simulation Year
Biomass Over Time: Red Grouper
Simulation
0
5
10
15
20
25
30
35
40
45
50
0 4 8 12 16 20
Mill
ion
s o
f Fi
sh
Simulation Year
Numbers Over Time: Red Grouper
Simulation
Findings with Stock Synthesis
24
0
10,000
20,000
30,000
40,000
50,000
60,000
70,000
80,000
0 4 8 12 16 20
Met
ric
Ton
s
Simulation Year
Biomass Over Time: Red Grouper
Simulation Base Model
0
5
10
15
20
25
30
35
40
45
50
0 4 8 12 16 20
Mill
ion
s o
f Fi
sh
Simulation Year
Numbers Over Time: Red Grouper
Simulation Base Model
Findings with Stock Synthesis
25
0
10,000
20,000
30,000
40,000
50,000
60,000
70,000
80,000
0 4 8 12 16 20
Met
ric
Ton
s
Simulation Year
Biomass Over Time: Red Grouper
Simulation Base Model Perfect CPUE, init F 1/3 SEDAR
0
5
10
15
20
25
30
35
40
45
50
0 4 8 12 16 20
Mill
ion
s o
f Fi
sh
Simulation Year
Numbers Over Time: Red Grouper
Simulation Base Model Perfect CPUE, init F 1/3 SEDAR
Findings with Stock Synthesis
26
0
5
10
15
20
25
30
35
40
45
50
0 4 8 12 16 20
Mill
ion
s o
f Fi
sh
Simulation Year
Numbers Over Time: Red Grouper
Simulation Base ModelPerfect CPUE Perfect CPUE, est. init. FPerfect CPUE, init F half SEDAR Perfect CPUE, SEDAR init FPerfect CPUE, init F 1/3 SEDAR
0
10,000
20,000
30,000
40,000
50,000
60,000
70,000
80,000
0 4 8 12 16 20
Met
ric
Ton
s
Simulation Year
Biomass Over Time: Red Grouper
Simulation Base ModelPerfect CPUE Perfect CPUE, est. init. FPerfect CPUE, init F half SEDAR Perfect CPUE, SEDAR init FPerfect CPUE, init F 1/3 SEDAR
Findings with Stock Synthesis
27
Findings with Stock Synthesis
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0
0.0 1.0 2.0 3.0 4.0 5.0
Dep
leti
on
(SS
Bcu
rre
nt/
SSB
msy
-pro
xy)
Harvest Rate (Fcurrent/FMSY-Proxy)
Control Rule Plot: Simulated Red GrouperMSYproxy = SPR30%
Base Model, MSYReferenceBase Model, SPRReferencePerfect Index, MSYReferencePerfect Index, SPRReferenceSEDAR 12
SEDAR 12 Update
SEDAR 42
OFL
MSST
28
Findings with Stock Synthesis
0
5
10
15
20
25
30
0 5 10 15 20 25 30
Mill
ion
s o
f P
ou
nd
s
Simuilation Year
Red Grouper MSY Based Projections
SS Yield Base Model - OFLBase Model - OY Base Model - FcurrentPerfect CPUE - OFL Perfect CPUE - OYPerfect Index - Fcurrent
29
0
5
10
15
20
25
30
0 5 10 15 20 25 30
Mill
ion
s o
f P
ou
nd
s
Simuilation Year
Red Grouper SPR30% (MSY Proxy) Based Projections
SS Yield Base Model - OFLBase Model - OY Base Model - FcurrentPerfect CPUE - OFL Perfect CPUE - OYPerfect CPUE - Fcurrent
Findings with Stock Synthesis
30
• Direction and magnitude of CPUE index matters and drives biomass trend estimation in stock assessment models.
• An index that perfectly fits biomass produces results that estimates nearly identical biomass trends.
• Accurately anchoring fleet fishing mortality at the beginning of the time series in stock assessment models is critical to obtaining properly scaled results.
Results and Implications
31
• Historical data such as landings can be useful in tuning the early phases of a fishery.
• Runs with higher fishing mortality at the start underestimated spawning stock biomass, and therefore estimated less biomass and numbers of fish, and vice versa.
Results and Implications
U.S. Red Grouper Fleet Historical Catch
012345678
1880
1885
1890
1895
1900
1905
1910
1915
1920
1925
1930
1935
1940
1945
1950
1955
1960
Year
Millio
ns
of
Po
un
ds
West Florida Shelf Campeche Bank Other Gulf of Mexico Locations
32
Spatial Distributions of Fishing Effort
and Biomass
33
Results
• Fishing effort spatially and temporally affected by weather: less fishery dependent observations during times and at locations of high wind.
• Trip duration sometimes limited by hold size (filled vessel to capacity) which suggests effort saturation effect on CPUE.
• Spatial and temporal fishing effort distributions driven by closures indicate stock level catchability cannot be constant in space and time.
• Local depletion - spatial distribution of fishing effort and fish populations not uniform. —People tend to have their fishing sites that they know
work and only exploratory fish a fraction of the time if at all (survey responses suggest 25%).
34
Fisher Behavior and Oil Spill
35
Where did the fishing effort go?
36
Full Model
Evaluate socioeconomic and fisheries assessment and
management affects
Source: The Sea Around Us Project, University of British Columbia
37
Acknowledgements
This research was made possible in part by a grant from The Gulf of Mexico Research Initiative, and in part by CIMAS at the University of Miami, NOAA Cooperative agreement NA17RJ1226, the NOAA/Sea Grant Population Dynamics Fellowship Grant Award NA080AR4170763, and an International Light Tackle Tournament Association Scholarship. Data for this project will be made publicly available through the Gulf of Mexico Research Initiative Information & Data Cooperative (GRIIDC) at https://data.gulfresearchinitiative.org.
38
“Managing fish is managing people”
A.P. Bell and Starfish Market Abrahm’s Seafood Ariel Seafood Buddy Gandy Seafood Cox Seafood Fish Busterz (Madeira Beach) Holiday Seafood Jenson Tuna Little Manatee Fish House Madeira Beach Seafood Sammy’s Seafood Save On Seafood Water Street Seafood
Glen Brooks David Krebs Jason de la Cruz Bobby Spathe
-Ray Hilborn, 2007
Gulf Fisherman’s Association Southern Offshore Fishing Assn. Reef Fish Shareholder’s Alliance
39