Evaluation of sampling bias in growth and skin health assessment methods of New Zealand-farmed king salmon (Oncorhynchus tshawytscha)

Sampling bias was investigated in New Zealand farmed king salmon ( Oncorhynchus tshawytscha ) at the end of a feed trial. Two sampling techniques - a non-lethal manual assessment (MA) and a lethal hook and line (HL), were evaluated for the representativity of their samples, with regard to two performance indicators, weight and skin health. Both sampling techniques assessed (MA n = 60, HL n = 4) were biased when compared to the analysis of all the fish in each pen. However, MA samples were biased towards fish with better skin health and higher weight (+ 0.20 kg), while HL selected fish with poorer skin health than the pen population average and lower Fulton's Condition Factor (− 0.25) than MA samples. Our results emphasised the significance of understanding sampling bias so research interpretation is correct. • Manual assessment was biased for weight and skin health when compared to the analysis of all the fish in each pen. • This study is novel as while most of fish sampling is biased it is rarely admitted and even more rarely investigated. • This work is highly relevant to any aquaculture research which relies on fish sampling