Fisheries Research 75 (2005) 29–39 Choosing the best sampling unit and auxiliary variable for discards estimations L. Borges a,∗ , A.F. Zuur b , E. Rogan a , R. Officer c a Aquaculture and Fisheries Development Centre, Department of Zoology, Ecology and Plant Science, University College Cork, Lee Maltings, Prospect Row, Cork, Ireland b Highland Statistics Ltd., Newburgh, UK c Marine Institute, Galway Technology Park, Parkmore, Galway, Ireland Received 3 September 2004; received in revised form 19 April 2005; accepted 17 May 2005 Abstract The effects of two sampling units (haul and trip) and three auxiliary variables (total landings in weight, effort in hours fished and in numbers of fishing trips) commonly used in discard estimations were compared. Considering the sampling unit at haul level will increase variability in the final estimation. In addition, there was a significant linear relationship between discards per trip with both landings and effort, but not at haul level. The different auxiliary variables used to raise discard samples to population levels give significantly different results, with effort in hours fished giving the highest discard estimates. Fishing trips was the recommended variable to estimate discards in six out of ten fleets studied. These results highlight the importance of preliminary studies to determine the methodology to be used in discard estimations, in order to avoid considerably imprecise estimations. © 2005 Elsevier B.V. All rights reserved. Keywords: Auxiliary variables; Discards; Fishing effort; Landings; Sampling units 1. Introduction Discarding is recognised as a global problem in fish- eries (Alverson et al., 1994). Over the last 30 years, con- siderable effort has been made to increase the knowl- edge about discarding in worldwide fisheries. Sampling programmes have been established with the aim of esti- ∗ Corresponding author. Tel.: +353 1 8228200; fax: +353 1 8205078. E-mail address: lisa.borges@marine.ie (L. Borges). mating the amounts discarded. The level and scale of these programmes vary greatly and are influenced by the programme objectives, geographical area, fishery characteristics, cost and logistics. Most programmes do not reach 100% coverage, resulting in only a part of the fishery being observed. Thus, the discard data collected by these programmes must be raised to represent the population i.e. the whole fishery. This process requires the use of an auxiliary variable, serving as a proxy for fishing activity, that has a linear relationship with the amounts discarded. 0165-7836/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.fishres.2005.05.002