Training Workshop Welcome to the
Jan 16, 2016
Training Workshop
Welcome to the
What is ParFish?• An approach to
stock assessment• Involve fishers and
other stakeholders• Suitable for small-
scale fisheries• Rapid assessment• Appropriate for
data-poor situations
ParFish Process
1. Understand the context
2. Agree objectives with stakeholders
3. Undertake ParFish
stock assessment
4. Interpret results and give feedback
6. Evaluate ParFish process
5. Initiate management
planning
ParFish Toolkit• Guidelines: guidance
for carrying out the process, data collection, assessment and management planning
• Software for carrying out the stock assessment and Software Manual
Overview of the Assessment
State of the
fishery resource
Recommended levels of control
ParFish
software
Stock Assessment Interviews
Fishing Experiments
Catch-effort
Fisher preferences
Learning Objectives
• By the end of today, you will:• Have been introduced to the 6 stages
of ParFish and how to implement them;• Be more familiar with the ParFish
Software and analysis;• Be introduced to various participatory
techniques.
Characteristics of a suitable fishery
• Sedentary local species (not highly migratory e.g. tuna)
• Fishers responsible for the majority of fishing mortality can be identified
• One or more fishing villages involved (depending on resources)
• Co-management situation or wishing to develop co-management
Next: Paul
Stock Assessment
A brief introduction to principles and methods
Principles
Identify measurable indicators related to policy
Identify state of exploited populations (reference points)
Identify controls on fishing Identify and deal with uncertainty Provide relevant advice accounting
for the above
Types of Indicators
Stock size, SSB Catch / Landings Effort / vessels / days-at-sea Fishing mortality Employment Profit / economic rent Non-target catch Interactions / illegal activity etc
Data Variables must be:
measurable relevant
Convert from data to indicator possible to collect
Fit in with data collection system Low costs
Example
Policy statement: “…sustainable utilisation maximising economic benefits.”
Interpret: “…maintain stock size above MSY point, balancing employment and economic rent.”
Indicators and reference points: Stock size and MSY point Total employment and current employment Vessel profits (catch rates) and break-even
point
Data and Analysis
Stock size Total Catch Stock size index
Employment Number of people employed by sector
Vessel profits Economic inputs
Vessel, gear, fuel costs Economic outputs
Landings, prices
Reducing costs
Using Proxies Catch rates:
Population size index Profitability
Number of registered / licensed vessels Employment
Sampling Allow for error: sufficient sampling Sampling design
Co-management
Reference Points
Impact of fishing on populations Link fishing activity to depletion Link stock size to productivity
Use models to interpret data Simple indicators Complex reference points
Uncertainty
Sources of uncertainty Observation error (sampling) Process error (time series) Structural error (models)
Presentation of uncertainty Making decisions under uncertainty
Summary
Interpret independent information in relation to policy aims E.g. Indicators and reference points
reduction Address uncertainty Provide simple understandable
advice Promote management action
ParFish I
Design incorporates all other methods
Robust Explicitly deals with uncertainty Involves fishers: promotes
management action Simple advice?
ParFish II
Target simulation model Build probability density functions of
parameters (encapsulates uncertainty) Apply stochastic projections to
simulation model under possible actions
Identifies management actions best for fishers
Generates standard indicators / reference points
Next: PM/SW concepts
Bayesian Approach
A brief introduction
Summary
Introduction to probability Likelihood Bayes rule Decision theory and utility A practical application: ParFish
Mathematical Probability
Probabilities are between 0 and 1.0 0 = impossible 1.0 = certainty Probabilities often defined as sets of
possible events or outcomes A set of exclusive events, one of
which must occur, sum to one
Subjective Probability
People assess a risk even without direct observations
Some events we may wish to estimate we do not wish to observe, such as nuclear war or overfishing.
Discrete → Continuous
0
0.05
0.1
0.15
0.2
0.25
0.3
0 1 2 3 4 5 6 7 8 9 10
Heads
Pro
bab
ility
0
0.1
0.2
0.3
0.4
0.5
0.6
0 1
Heads
Pro
bab
ility
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
Heads
Pro
bab
ility
Example Probability Density
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.60
0.06
0.12
0.18
0.24 0.
3
0.36
0.42
0.48
0.54 0.
6
0.66
0.72
0.78
0.84 0.
9
0.96
Random Variable
Pro
bab
ility
Den
sity
0
5
10
15
20
25
30
35
40
45
Pro
bab
ility
Den
sity
Likelihood
Probability when p is known: Pr(H) = p Pr(T) = 1-p
Likelihood when H/T is known Pr(p ¦ H) = p Pr(p ¦ T) = 1-p
Binomial Likelihood
nCr is the number of ways (combinations) r heads could occur in n trials.
!!
!
1Pr
rnr
nC
ppCHeadsrp
rn
rnrr
n
where
Likelihood: 8 Heads 2 Tails
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
p
Lik
elih
oo
d
Fishing Experiment
Population size on day 0 = n We catch C0 fish on day 0
Population size on day 1 = n - C0
We catch C1 fish on day 1
Population size on day 2 = n - C0 – C1
Population size on day t = n - t Ci
Fishing Experiment
y = -0.1949x + 8.3521
R2 = 0.9033
0
1
2
3
4
5
6
7
8
9
0 5 10 15 20 25 30
Cumulative Catch
Ind
ex
Lake Fishing Likelihood
0.05
0.1
0.15
0.2
0.25
27
43
59
75
910
0.01
0.02
0.03
0.04
0.05
0.06
0.07
pn
rnrr
n ppCFishrnp 1,Pr
Bayes Rule
Posterior Prior * Likelihood
Pr(p, n Data) Pr(p, n) * L(Data p,n)
Updating Using Bayes
Pr(p, n Data) Pr(p, n) * L(Data1 p,n)* L(Data2
p,n)
Which gives
Pr(p, n Data) Pr(p, n Data1) * L(Data2 p,n)
Lake Fishing ExperimentPrior for n
0
0.5
1
1.5
2
2.5
3
3.5
20 24 28 32 36 40 44 48 52 56 60 64 68 72 76 80 84 88
n
Pro
bab
ility
Prior on p
0
0.2
0.4
0.6
0.8
1
1.2
p
Pro
bab
ility
0 1
Lake Fishing Experiment
0 0.15
0.3 0.45 0.
6
0.75 0.
9
27
43
59
75
91
0
0.5
1
1.5
2
2.5
3
p
n
Prior for n and p
Lake Fishing Likelihood
0.05
0.1
0.15
0.2
0.25
27
43
59
75
910
0.01
0.02
0.03
0.04
0.05
0.06
0.07
pn
Lake Fishing Posterior
0.05 0.
08 0.11 0.
14 0.17 0.
2
0.23
27
43
59
75
91
0
0.02
0.04
0.06
0.08
0.1
p
n
Utility
Score cost / benefits of outcomes in one dimension
Not monetary Used in economics to manage risk Explains why people enter games
where they expect to lose money
Example Utility Curves
00.10.20.30.40.50.60.70.80.9
1
020
0040
0060
0080
00
1000
0
1200
0
1400
0
1600
0
1800
0
2000
0
Variable (e.g. Dollars)
Uti
lity
Risk Averse
Risk Seeking
Decision Theory
Combines probability and utility
Bayes action: Choose the
action which will maximise the expected (average) utility
ImplementNot Implement
Overfished
Not Overfished
0
0.1
0.2
0.3
0.4
0.5
0.6
Utility
Recovery Plan
State of Nature
Next: PM/SW
Software Structure
Simulation Model
Posterior Parameter
PDF 1
PDF 2
PDF N
ControlProjected catch - effort
time series
Source Model 1
Source Model 2
Source Model N
ProbabilityModelling
Preference
Generating Parameter Probabilities
• ParFish software takes frequency observations, and estimates the underlying probability distribution from which they were drawn
Observations
Estimate PDF
1
Bnow PDFBnow frequency
Bnow10.80.60.40.20
Pro
babi
lity
0.064
0.0620.06
0.058
0.0560.054
0.0520.05
0.0480.046
0.0440.0420.04
0.0380.036
0.0340.032
0.030.0280.026
0.0240.022
0.020.018
0.0160.0140.012
0.010.008
0.0060.004
0.0020
• Probability density functions from various data sources can be combined into a single ‘posterior’ PDF
Combine
Posterior
Conventional and New Information Sources
• Current version uses logistic (Schaefer) as simulation model: r, Bcur, Binf and qj
• Various data types and sources can be combined e.g.• Long term catch-effort data models• Interviews • Fishing experiments• Biological parameters• Others?
Software – probability models
Fishing Experiments
• Estimate population size and catchability
• Fishers concentrate their fishing effort in a specific area, catches and effort are recorded
• Complemented by underwater visual surveys of fish population
R2 = 0.70
0
0.5
1
1.5
2
2.5
0 200 400 600 800 1000 1200 1400
Cumulative catch (kg)
CP
UE
(pe
rson
h-1
)
Interviews• Stock assessment interviews
gather fishers’ knowledge about the resource and provide a starting point for the stock assessment
• Preference interview indicates how much fishers would like or dislike different outcomes of catch and effort
Utility & Decision Theory• Utility refers to how
good something is for someone
• Modelling provides a variety of possible outcomes from different decisions
00.10.20.30.40.50.60.70.80.9
1
Variable (e.g. Dollars)
Util
ity
Risk Averse
Risk Seeking
• Decision Theory helps us decide which of a set of actions to take, based on their expected utility (probability of happening times cost)
• Bayes action:Choose the action which will maximise the expected (average) utility
Preference Interviews
• Scenario cards - different levels of catch and effort
• Pair-wise ranking then scoring• Score indicates ‘utility’
Example pairwise comparison
• Keep current work level in the fishery, but get 25% more income/fish, OR
• Keep fishery income the same, but for 25% less time which could be used for other work.
O K
Outputs of Analysis
• Output reference points, fishery states etc. as probabilities
• Limit and target control levels:• Recommended (target) control levels • Limit control levels with acceptable chance
of overfishing
Participatory Framework
• Involve fishers at an early stage
• Helps their acceptance of assessment results
• Participatory framework draws on Adaptive Learning, Participatory Action
Plan Development, Consensus Building Methodology, participation literature
• Supports co-management
Understanding the context
• Fishery and management context • Stakeholder Analysis• Communications Plan
• Gather background information
Engaging stakeholders
• Set objectives for the assessment
• Introduce concepts: uncertainty, fish stock dynamics, probability, overfishing
• Participatory techniques
Communicating to Fishers: Concepts
Stock size, growth and fish catch
Year 1 Year 2 Year 3 Year 4
Growth Growth
Survival Survival
Catch Catch Catch
Survival
Growth
Stock
CatchCatches will start to fall as the fish stock can no longer
support the same size catches
Communicating to Fishers: Concepts
• Over-fishing
Communicating to Fishers: Concepts
• Estimating the number of oranges
• Illustrates uncertainty, estimating and probability curves
1110 12 13 14 191815 16 17 20
10 12 13 14 191815 16 17
12 14 1916 17
11 12 13 14 1815 16 17
12 13 14 15 16 17
13 14 15 16 17
13 14 15 16
14 15
Feedback and Planning
• Communicating the results of the assessment to fishers and fisheries management institutions;
• Building consensus on problems and possible solutions for the fishery;
• Developing a management plan or action plan;
• Evaluating the process
Next: Narriman
Case Study – Kizimkazi, Zanzibar
• Period of data collection and development of techniques
• Feed back results and initiate planning process
• ADD Background to Kizi
Framesurvey Information – Kizimkazi
No. of Fishers No. of boats
K.Mkunguni 167 58
K. Dimbani 152 73
Total 319 131
Data Collection in Kizimkazi 2003
• Techniques developed and tested and used to provide an assessment for the fishery
• Experiments• Carried out for inner fringing reef (Mtende) and patch reefs (Dimbani)
• Interviews • 43 fishers in Dimbani• 39 fishers in Mtende & Mkunguni
Results – Kizimkazi handline fishery
• Uncertainty about the current state of the stock
• 50% chance that it is overfished (less than half of the unexploited biomass remaining)
Resource State0.90.80.70.60.50.40.30.20.10
Pro
babi
lity
1.6
1.5
1.4
1.3
1.2
1.1
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.506
Resource State0.90.80.70.60.50.40.30.20.10
Pro
babi
lity
1.4
1.35
1.3
1.25
1.2
1.15
1.1
1.05
1
0.95
0.9
0.85
0.8
0.75
0.7
0.65
0.6
0.55
0.5
0.45
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0.425
Outer patch reefs Inner fringing reefs
Dimbani – offshore reefs Mtende – inshore reefs
State of the stock under different control levels
Management recommendationsEffort control
Decrease in effort Predicted outcome
Inner fringing reefs Outer patch reefs
20% 20% More preferred conditions in the fishery
63%* 10-15% Reduce chance of overfishing to 10%
Closed area controlArea closed to fishing Predicted outcome
Inner fringing reefs Outer patch reefs
6% 0% Most preferred conditions in the fishery
35% 5% Reduce chance of overfishing to 10%
All assessments suggested a decrease in fishing effort would be advisable for fringing and patch reefs
The patch reef assessment suggests closed areas would not be acceptable to the fishers, although a small closed area may be acceptable on the fringing reef
Management recommendations
Combination controls
Area closed to fishing
Decrease in effort
Timescale
Inner fringing reefs
5% 10-20% 2-3 years
Outer patch reefs Rotational closure of patch reefs
10% Each reef closed for 1-12 months
• A combination of a closed area and a reduction in effort decreases the chance of overfishing and gives a lower recommended reduction in effort.
• Monitored closed areas would provide additional information on recovery rates and unexploited biomass to update the assessment.
Management recommendations
• Further data collection should be continued to reduce the uncertainty of the assessment, e.g. monitor catch and effort, monitor closed area recovery
• Fishing experiments should be repeated at the beginning of the good fishing season
• Results of any management actions should be monitored
Follow-up for Kizimkazi • Initial results were fed back to fishers who
agreed that a reduction in effort may be required
• A workshop was held to discuss and agree management recommendations
• Types of issues raised: •Controls: effort/closed area•Enforcement •Visiting fishers •Monitoring
• Needs further follow-up to turn workshop recommendations into concrete actions
Case Study: Turks and Caicos Is.
• Interviews carried out with conch fishers;
• Catch and effort data used from 1976 – 2002;
• Stock had declined in 1980s.
Fisher Knowledge Validity
0
100
200
300
400
500
600
700
800
1974
1976
1978
1980
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
Years
CP
UE
Original Projected On 1.68 quota