Tshikana Phillip Rasehlomi ^1* Marjolaine Krug ¹ mari myksvoll ²

• ¹ Oceans and Coasts Research Branch, Department of Environment, Forestry and Fisheries (DEFF), Cape Town, South Africa
• ² Norwegian Institute of Marine Research (IMR), Bergen, Hordaland, Norway

Lagrangian dispersal models are valuable tools for understanding the transport mechanisms and distribution of larvae in space and time. These models primarily use high-quality physical oceanographic data from hydrodynamic ocean models to simulate the advection and connecting pathways of larvae. The paper seeks to achieve the following objectives 1) to establish the dispersal patterns of larval anchovies and sardines spawned in South Africa's coastal marine protected areas (i.e., inshore of the 200 m isobath), and 2) to assess the impact of model resolution on the dispersal patterns by comparing outputs of a 3-km-grid coastal and regional ocean community model, and the 8-km-grid global ocean physics reanalysis product. We demonstrate that the model yields more structured patterns than sporadic patterns obtained from the global reanalysis. The model shows relatively shorter residence times (i.e., < 5 days) and transport confined to the inshore area of the Agulhas Current. The high variability of dispersals observed off the coast is attributable to the significant differences in mean kinetic energy between the model and global reanalysis. This finding directly impacts transport and potential connectivity along coastlines: global reanalysis data overestimated local retention and subsequent recruitment variability. Our findings indicate that a model grid size of 3 km or smaller may be more appropriate for studying the dispersal patterns of anchovies and sardines in the South African coastal zone. These findings add to the growing knowledge of the importance of including sub-mesoscale processes in Lagrangian analysis for dispersal studies of coastal species.