SELECTION EXPERIMENTS ON ATLANTIC SALMON RANCHING .2. VARIATION AMONGRELEASE SITES AND STRAINS FOR RETURN RATE, BODY-WEIGHT AND RATIO OF GRILSE TO TOTAL RETURN

A Nordic project 'Salmon Ranching-Possibilities of Selective Breeding' started in Iceland in 1987. The aim of the project was to study the p ossibilities for breeding and selection in sea ranching of Atlantic sa lmon. Smelts from six salmon strains were released from four different release sites in three year-classes. Performance was estimated for re turn rate, body weight, ratio of grilse to total return and total biom ass per 1000 smelts released, which gives a measure of total yield aft er 2 years at sea. In general, total return rate was low during the ex periment, ranging from 3.04% to 0.96% between year-classes. The mean b ody weights of grilse and two-sea-winter salmon were 2.4 kg and 6.2 kg , respectively. Biomass per 1000 smelts released ranged from 35.6 kg t o 87.5 kg between year-classes. Significant differences were observed among the salmon strains for all traits tested. In general, strains us ed for salmon ranching in Iceland gave the highest yield in total biom ass compared with other strains tested; up to three-fold differences w ere observed. Return rates differed among release sites, with the high est return rates observed at the Kollafjordur release site, where smel ts were released from a riverine trap, compared with releases directly from concrete tanks at Vogavik or from net pens at Silfurlax ranching stations. For the 1991 year-class, 94.0 kg total biomass returning wa s observed at the Kollafjordur release site compared with 52.7 kg at V ogavik and 75.3 kg at Silfurlax. Similar differences in return rates w ere observed between Kollafjordur and the Laros release site, where th e smelts were released from netpens into a freshwater lake. Significan t interactions between salmon strain and release site were detected fo r total return rate and the ratio of grilse to total return, but not f or body weight of grilse. This resulted in change in the order of rank ing of the strains at different release sites, especially in year-clas s 1989. This interaction needs further examination.