Hasşerbetçi C., Tunçelli G., Can Tunçelli İ.
SU URUNLERI DERGISI, cilt.43, sa.4, ss.372-382, 2026 (TRDizin)
Özet
The Mediterranean mussel Mytilus galloprovincialis is the principal cultured bivalve in Turkish coastal waters, and Dynamic Energy Budget (DEB) models forced with satellite ocean colour offer a low-cost route to growth prediction. In practice, published parameter sets underestimate shell growth in coastal settings, and the reason is unresolved. We tested a published covariation-method DEB parameterisation against ten grow-out experiments in Türkiye (43 cohort groups in the Aegean, the Dardanelles, the Marmara and the Black Sea), forced with reanalysis sea-surface temperature at culture depth and satellite Level-4 chlorophyll-a. Applied as published (XK=2,1 µg L-1) the model under-predicted at every experiment (RMSE 19.8 mm, bias mm). Recalibrating the half-saturation constant alone, with no biological parameter changed, reduced error to 12.7 mm (95% CI 8.1–17.9) under leave-one-experiment-out cross-validation with every data-derived choice refitted inside each fold, and out-performed a satellite-to-in-situ ratio correction by 4.5 mm. We then tested the measurement-scale explanation for that improvement and did not find support for it: at the one site where in-situ chlorophyll is resolved in time and the product observes the site itself the ratio varied three-fold between seasons, rescaling XK is algebraically identical to rescaling the forcing, and at two sites the product returns no valid retrieval at the deployment at all. The corrected model is therefore an empirical forcing-context calibration, whose residual absorbs unmodelled food quality and culture-system effects, rather than a measurement-scale conversion. Its limits matter as much: a near-zero overall bias masks a mean absolute per-experiment bias of 9.6 mm, and no site-ranking skill was demonstrable against baselines using no chlorophyll or no model. Satellite-forced DEB can reproduce the central tendency of Turkish mussel growth after calibration, but does not yet resolve the between-site differences on which farm siting depends.