Report of the Study Group on Practical Implementation of Discard Sampling Plans (SGPIDS)

The Study Group on Practical Implementation of Discard Sampling Plans (SGPIDS; chaired by Edwin van Helmond, The Netherlands) met 27 June – 1 July 2011 in Co-penhagen, Denmark. Seventeen participants representing 11 countries were present at the meeting, including the outgoing chair, Simon Northridge, of ICES WGBYC (Working Group on Bycatch of Protected Species). SGPIDS was proposed by ICES PGCCDBS (2010) in response to a request from the Regional Coordination Meeting for the North Sea and Eastern Arctic (RCM NS&EA; 2010) to foster an exchange of experience and expertise between experts on discard sampling, planning and imple-mentation of PGCCDBS recommendations and ultimately synchronize coordination and data collection procedures of discard sampling between countries.

To handle the exhaustive list of terms of reference the group split up into subgroups. These dealt with one term of reference each. Wherever necessary, the subgroups col-lected information about the existing discard sampling programmes by represented member state. This information was used to create an extensive overview of tech-niques and protocols used to sample discards onboard commercial vessels. Through-out the meeting all subgroups updated each other during plenary sessions.

The study group identified 21 different discard sampling programmes among the countries present, which were divided into two main types of discard sampling tech-niques: observer and self sampling (including self sampling with a reference fleet). Among observer programmes, differences in the procedures of selecting vessels and allocating sampling effort were identified. For example, nine out of 15 observer pro-grammes use a quasi-random vessel selection method, based on a combination of opportunistic and co-operative criteria. The remaining six programmes use a fully random or otherwise systematic approach to select the vessels for monitoring. It was noted that only 25% of the programmes routinely record refusal rates. Six countries at SGPIDS conduct dedicated self-sampling schemes. Of these, 66% are validated (e.g. comparing biological data with matched or unmated observed trips and/or other in-dependent sources). Vessel selection was a key source of potential bias for both sam-pling techniques. Sampling effort allocation was another major source of bias. Further, it was noted that legal conditions under which discard sampling is taking place, potentially harm the cooperation between industry and scientist in discard sampling programmes and, eventually jeopardize the quality of sampling pro-grammes.

SGPIDS recognised the potential for more standardisation in sampling designs and this should start with a complete description (in English) of sampling designs of all current sampling programmes. SGPIDS created a detailed description, at all levels (i.e. sampling protocols, data processing, data storage procedures, co-operation with industry, observer training and safety procedures) for the 21 programmes. With the aim to standardize discard sampling across countries, it is important that bias and variability associated with their respective sampling programmes are investigated.

The Data Collection Framework (DCF) set out precision levels but did not include any requirements about bias. Bias is introduced to sampling schemes when samples are not representative of the population. In accordance with previous working and study groups (e.g. ICES WKEID, WKACCU), SGPIDS identified a number of poten-tial sources of bias in discard data. There was a general agreement that improving the data quality by reducing bias should be prioritised over increasing precision levels.