Inter-annual variability in abundance of key fish and pelagic crustacean species in the Lake Victoria 'natural laboratory' : from ecosystem understanding to fisheries management

Lake Victoria (East Africa), the world's second-largest freshwater lake, is home to capture fisheries that are vital to regional food security and economic development. However, knowledge of the accurate biomass variability of pelagic stocks, specifically Caridina nilotica (Caridina), Rastrineobola argentea (Dagaa), haplochromines, and Lates niloticus (Nile perch), has been limited due to present method limitations hampering progress towards ecosystem-based fisheries management. Annual fisheries-acoustic surveys have been conducted for ca. 20 years, including species-by-species biomass estimation employing analysis methods prone to systematic bias. These include attempts to estimate Caridina biomass by scaling acoustic backscatter (assumed to originate from Caridina) using Target Strength (TS) derived from Antarctic krill; the use of single-target detection for Nile perch, despite its inapplicability to densely aggregated fish; and water column partitioning based on fixed species depth niches, which fails to account for vertical overlap and dynamic behavioral variability. We present direct approaches for estimating species-specific biomass using acoustic backscatter serial subtraction. Fish schools are first detected using the SHAPES algorithm (Coetzee 2000). Dagaa schools are then classified using a Random Forest model trained on manually labelled data. The remaining schools are grouped via K-means clustering, revealing four distinct clusters—three corresponding to haplochromines and one to juvenile Nile perch, as validated by trawl data. Caridina biomass is estimated using a dB-differencing method and a species-specific TS derived from actual geometric measurements and a Distorted Wave Born Approximation (DWBA) model. After quantifying biomass for Caridina, Dagaa, and haplochromines, the remaining unclassified backscatter is attributed to adult Nile perch. The resulting pelagic biomass time-series are lower than previous estimates and exhibit pronounced inter-annual variability, but within the confidence intervals established under the previous analyses. These new insights provide a foundation for exploring how environmental factors influence fish production, enabling forecasting production dynamics and supporting sustainable, ecosystem-based fisheries management.