1. Training Workshop Welcome to the

2. 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

3. 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

4. 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

5. Overview of the Assessment State of the fishery resource Recommended levels of control ParFish software Stock Assessment Interviews Fishing Experiments Catch-effort Fisher preferences

6. 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.

7. 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

8. Next: Paul

9. Stock Assessment A brief introduction to principles and methods

10. 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

11. 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

12. Data Variables must be:  measurable  relevant Convert from data to indicator  possible to collect Fit in with data collection system Low costs

13. 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

14. 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

15. 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

16. 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

17. Uncertainty  Sources of uncertainty Observation error (sampling) Process error (time series) Structural error (models)  Presentation of uncertainty  Making decisions under uncertainty

18. 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

19. ParFish I  Design incorporates all other methods  Robust  Explicitly deals with uncertainty  Involves fishers: promotes management action  Simple advice?

20. 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

21. Next: PM/SW concepts

22. Bayesian Approach A brief introduction

23. Summary  Introduction to probability  Likelihood  Bayes rule  Decision theory and utility  A practical application: ParFish

24. 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

25. 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.

26. Discrete → Continuous

27. Example Probability Density

28. 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

29. Binomial Likelihood  nCr is the number of ways (combinations) r heads could occur in n trials.

30. Likelihood: 8 Heads 2 Tails

31. Fishing Experiment  Population size on day 0 = n We catch C 0 fish on day 0  Population size on day 1 = n - C 0 We catch C 1 fish on day 1  Population size on day 2 = n - C 0 – C 1  Population size on day t = n -  t C i

32. Fishing Experiment

33. Lake Fishing Likelihood

34. Bayes Rule  Posterior  Prior * Likelihood  Pr(p, n  Data)  Pr(p, n) * L(Data  p,n)

35. 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)

36. Lake Fishing Experiment

38. Lake Fishing Likelihood

39. Lake Fishing Posterior

40. 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

41. Example Utility Curves

42. Decision Theory Combines probability and utility  Bayes action:  Choose the action which will maximise the expected (average) utility

43. Next: PM/SW

44. Software Structure Simulation Model Posterior Parameter PDF PDF 1 PDF 2 PDF N Control Projected catch - effort time series Source Model 1 Source Model 2 Source Model N Probability Modelling Preference

45. Generating Parameter Probabilities ParFish software takes frequency observations, and estimates the underlying probability distribution from which they were drawn Observations Estimate PDF

47. Probability density functions from various data sources can be combined into a single ‘posterior’ PDF Combine Posterior

48. Conventional and New Information Sources Current version uses logistic (Schaefer) as simulation model: r, B cur, B inf and q j Various data types and sources can be combined e.g. Long term catch-effort data models Interviews Fishing experiments Biological parameters Others?

49. Software – probability models

50. 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

51. 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

52. Utility & Decision Theory Utility refers to how good something is for someone Modelling provides a variety of possible outcomes from different decisions 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

53. Preference Interviews Scenario cards - different levels of catch and effort Pair-wise ranking then scoring Score indicates ‘utility’

54. 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.

56. 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

57. 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

58. Understanding the context Fishery and management context Stakeholder Analysis Communications Plan Gather background information

59. Engaging stakeholders Set objectives for the assessment Introduce concepts: uncertainty, fish stock dynamics, probability, overfishing Participatory techniques

60. Communicating to Fishers: Concepts Stock size, growth and fish catch Year 1Year 2Year 3Year 4 Growth Survival Catch Survival Growth Stock Catch Catches will start to fall as the fish stock can no longer support the same size catches

61. Communicating to Fishers: Concepts Over-fishing

62. Communicating to Fishers: Concepts Estimating the number of oranges Illustrates uncertainty, estimating and probability curves 1110121314191815161720 101213141918151617 1214191617 1112131418151617 121314151617 1314151617 13141516 1415

63. 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

64. Next: Narriman

65. Case Study – Kizimkazi, Zanzibar Period of data collection and development of techniques Feed back results and initiate planning process ADD Background to Kizi

66. Framesurvey Information – Kizimkazi No. of FishersNo. of boats K.Mkunguni 16758 K. Dimbani15273 Total319131

67. 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

68. 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) Outer patch reefsInner fringing reefs

69. Dimbani – offshore reefs Mtende – inshore reefs State of the stock under different control levels

70. Management recommendations Effort control Decrease in effortPredicted outcome Inner fringing reefsOuter patch reefs 20% More preferred conditions in the fishery 63%*10-15%Reduce chance of overfishing to 10% Closed area control Area closed to fishingPredicted outcome Inner fringing reefsOuter 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

71. Management recommendations Combination controls Area closed to fishing Decrease in effort Timescale Inner fringing reefs 5%10-20%2-3 years Outer patch reefsRotational 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.

72. 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

73. 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

74. 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.

75. Fisher Knowledge Validity