Gownaris et al. compiled 17 published Ecopath models across 13 African lakes plus Lake Kariba reservoir, extracting 15 ecosystem attributes and regressing each on seven physical predictors (including relative lake level fluctuations, RLLF) via LASSO . Interannual fluctuation (RLLFa) was the most frequent predictor, retained in 9 of 15 models: higher fluctuation meant higher primary/total production but lower fish diversity, transfer efficiency, and maximum trophic level. Seasonal fluctuation (RLLFs) was retained in 6 models, positively associated with standing biomass β unexpected, since biomass typically increases with maturity.
High-RLLF systems showed steeper declines of log biomass (slope β0.59 vs β0.31) and log production (β1.4 vs β1.0) with trophic level, meaning more biomass concentrated at lower trophic levels .
Strengths: LASSO is appropriate for n=13β16 with collinear predictors; Theil-Sen trends are robust to outliers; the authors computed pedigree indices themselves (conservatively lower than model authors' own), and found no correlation between model quality/complexity and RLLF, addressing a key confounder. Temporal mismatch between model periods (mostly 1970sβ80s) and 1990sβ2000s hydrology was justified by rΒ² = 0.94β0.96 correlation between periods.
Weaknesses: cross-sectional correlations cannot establish causation; only linear forms tested despite expected dome-shaped productionβfluctuation relationships; Lake Hayq's missing seasonal data reduces seasonal regressions; fish diversity was partly sourced from FishBase lists rather than field data; and Lake Nakuru (RLLFa = 38.7, one fish species) is a potential leverage point in several regressions. Catch showed no relationship with RLLF, possibly because Ecopath catch snapshots and underestimated inland fishery statistics mask real yield-fluctuation links β the authors honestly acknowledge this. The Odum maturity framing is heuristic; the paper does not overclaim, and its 'total biomass alone is insufficient as a maturity indicator' conclusion is well supported by their trophic-distribution analysis.
Confidence: moderate. The directional patterns are internally consistent and align with prior tropical-lake work (e.g., hydrology dominating Amazonian floodplain fish abundance over fishing effort ), but generality beyond African systems is untested. Findings would be weakened if updated, higher-pedigree lake models or within-lake time series failed to reproduce the RLLF-attribute correlations.
Author-specific analyses: N.J. Gownaris Β· K.J. Rountos Β· L. Kaufman Β· J. Kolding Β· K.M.M. Lwiza Β· E.K. Pikitch
Know what changed, what holds up, and what remains uncertain. Every Friday. No ads.