Spatial Change of Dominant Baltic Sea Demersal Fish Across Two Decades

Area of Science:

• Marine Ecology
• Fisheries Science
• Biogeography

Background:

• Species distribution models (SDMs) traditionally use occurrence data, underutilizing quantitative biomass information to define realized niches.
• Local fish densities, rather than just presence, significantly influence ecological processes and ecosystem function.
• Understanding macro-environmental controls on species growth rates can enable prediction of geographical abundance and densities.

Purpose of the Study:

• To model spatiotemporal biomass patterns of four dominant demersal fish species in the Baltic Sea over 20 years.
• To assess the impact of oceanographic features and geographic information on fish biomass.
• To evaluate the performance of seasonally variable models compared to annually averaged models for predicting fish distribution.

Main Methods:

• Collated 20 years (2001-2020) of standardized scientific bottom trawl data.
• Applied hierarchical generalized additive models using biomass (kg km⁻²) for Common dab, European flounder, European plaice, and Atlantic cod.
• Incorporated geographic (position, depth) and high-resolution oceanographic covariates, with seasonally variable smoothing terms.

Main Results:

• Common dab biomass showed relative stasis, influenced by salinity and season-dependent temperature responses.
• European flounder and plaice biomass increased in the western Baltic Sea, with distinct seasonal temperature relationships and flounder exhibiting seasonal salinity shifts.
• Atlantic cod biomass declined significantly in specific areas (Bornholm Deeps, Gdańsk Deeps), with less reliable predictions compared to other species, suggesting the need for dynamic fishing covariates.

Conclusions:

• Seasonally variable models demonstrate superior predictive performance for marine fish biomass, highlighting dynamic species-environment interactions.
• Fish biomass distribution is strongly influenced by seasonal oceanographic changes, necessitating dynamic approaches in species distribution modeling.
• The findings provide valuable insights for biogeographic forecasting, improving the discrimination of high and low biomass areas in space and time.