Training Workshop Welcome to the. What is ParFish? An approach to stock assessment Involve fishers and other stakeholders Suitable for small- scale fisheries.

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

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