1 of 3

1

Deepak George Pazhayamadoma, Emer Rogana, Ciaran Kellyb and Edward Codlingc

aSchool of Biological, Earth and Environmental Sciences (BEES), University College Cork, Ireland; bFisheries Science Services, Marine Institute, Ireland; cDepartment of Mathematical Sciences, University of Essex, United Kingdom

Can we manage a fishery if no previous data are available?

Application of quality control charts in management of data limited fisheries

Historical data

Yes

Qualitative risk assessments

Quantitative stock assessments

No

Self Starting Cumulative Sum�SS-CUSUM

YES

No historical data at 0th year

2 of 3

2

SS-CUSUM

Self starting CUSUM (Hawkins, 1998)

Running mean�(Calibrated using real time data)

Three parameters

1. Allowance (k) �2. Control limit (h) �3. Winsorizing constant (w)

  • SS-CUSUM is an indicator monitoring tool.
  • SS-CUSUM do not need a reference point.
  • SS-CUSUM calculate the cumulative deviations of indicator from running mean

Parameters

  • Allowance (k) accommodate the inherent variability in observations
  • Control limit (h) produce signal if the indicator is in an out-of-control (OC) situation
  • Winsorizing constant (w) make self starting CUSUM robust to outliers

Evaluation of SS-CUSUM using a stochastic simulation test

  • A stable fish stock was overfished and indicators were monitored using SS-CUSUM
  • Signals obtained from SS-CUSUM were used to calculate sensitivity and specificity
  • Sensitivity is the probability of getting a true signal when overfishing was applied
  • Specificity is the probability of getting a true signal when there was no overfishing

Indicator observations corresponding to out-of-control situations are omitted while calibrating the running mean

Performance measures used

  • Receiver Operator Characteristic (ROC) curves

3 of 3

3

  • SS-CUSUM was successful in detecting the fishing impact.
  • An indicator is best when the apex of ROC curve is closer to upper left corner.
  • The method performed best with Large Fish Indicators (LF catch numbers, LF catch weight and LF CPUE).

Results (ROC curves)

Conclusion

All stock indicators in the study were useful in detecting fishing impact and hence SS-CUSUM can be potentially used for monitoring data poor fisheries

Reference:

Hawkins, D.,Olwell, D., 1998. Cumulative sum charts and charting for quality improvement: Springer Verlag, pp:162-168.

Best

Good

Worst