Sources of bias in the RFID tag-recapture data used in the stock assessment of North East Atlantic Mackerel

Abstract

Northeast Atlantic Mackerel (Scomber scombrus) is an immensely valuable fish stock for Norway and several other nations. The dynamic stock is spread throughout many nations, making the stock assessment an effort of large international cooperations. Many nations cooperating means that there will be more room for error in the estimate. IMR has made an admirable effort in assessing the mackerel stock using steel tags for over 40 years (1969 2010), with manual processes of recapturing fish from conveyor belts with metal detectors during processing at fish factories. Technological changes with the introduction of RFID tag recapture technology in 2011, resulted in more automatic processes, more effective data handling and a much larger proportion of the catch scanned. With mackerel being the important resource that it is, errors and biases in the stock assessment could have large implications. It is therefore essential to detect the possible sources of errors. This thesis uses data gained from RFID-tags from various nations and factories, as well as data from the old steel tag series to assess the possible sources of biases in the tag-recapture time series as input to the current assessment model. One source of bias is related to the use of age length key (ALK) at release to estimate age at recapture in the RFID data, compared with actual age reading from otoliths of recaptured fish in the old steel tag time series. The results showed that the estimated age from ALK is slightly higher than the actual age reading of the mackerel per length group, suggesting a bias. Tagging mortality due to crowding, waves and presence of birds was evaluated so see if it affected the stock estimate. In general there was no effect of tagging mortality. Spatial factors such as region, country and factory were also compared to each other to see if stock estimates of mackerel varied between them. In general region provided similar estimates, but estimates varied from each other when going into country and factory level. Norwegian factory Pelagia Austevoll seemed to have the largest variation in estimates when compared to other factories. Variations might be due to different levels of mixing of year classes in spawning and fishing areas. RFID is a new approach to estimating mackerel stock and is still lacking a certain level of accuracy. The initial assessment indicates that the RFID-tags will provide a consistent estimate and that data needs to be evaluated over a longer period of time to pinpoint the sources of bias more accurately.