Information on the length distribution of krill in the catch, and the spatial and temporal patterns of krill length in the population is essential to assess krill population status and dynamics, and the impacts of a fishery on the stocks. Here, we use generalised additive models (GAM) and generalised additive mixed models (GAMM) to evaluate the spatial and temporal patterns of length in Antarctic krill (Euphausia superba) in CCAMLR Subarea 48.1 derived from data collected by scientific observers. Median krill length showed a complex pattern and varied significantly with fishing location, fishing depth, season, month and vessel. The current sampling strategy to observe krill length in SSMUs in Subarea 48.1 needs to be modified in order to gain a comprehensive understanding of the temporal and spatial variability in krill length distribution and to determine the scale of observer coverage in the longer term. Krill length measurements should also be conducted on all vessels in every fishing season to reduce the likelihood of potential biases in the overall krill length estimates.
We also propose a simulation approach to evaluate alternative observer sampling strategies for measurements of particular properties of a krill population that are of interest to the management of the fishery.