Coastal phytoplankton support marine food webs and biogeochemical cycling, yet environmental controls of chlorophyll-a (Chl-a) vary among coastal regions. We integrated satellite, ocean reanalysis, atmospheric, and terrestrial data to assess Chl-a variability across 55 Large Marine Ecosystems (LMEs) during 2000–2020. Mean Chl-a varied 17-fold, from 0.18 mg m⁻³ in the East Brazil Shelf to 3.11 mg m⁻³ in the Yellow Sea, with higher values in mid-latitude, enclosed, upwelling, and river-influenced systems. Chl-a increased in 32 LMEs and declined in 23, including 11 significant increases and five significant decreases. Among 11 algorithms, Extra Trees achieved the strongest pooled prediction (test R² = 0.964). SHAP attribution assigned 37.1% of predictive importance to ocean physics, 24.5% to terrestrial influence, and 22.6% to ocean nutrients, which dominated 36, 15, and four LMEs, respectively. Path models explained 52.5–63.3% of Chl-a variability and distinguished three mechanisms. In ocean-physics-dominated LMEs, deeper mixed layers directly reduced surface Chl-a (β = −0.416) through dilution and reduced light exposure, although enhanced macronutrient supply produced a weaker positive indirect effect (β = 0.301 × 0.330 ≈ 0.10). In land-dominated LMEs, basin wetness increased Chl-a through hydrological delivery and plume processes (β = 0.232), while iron promoted Chl-a (β = 0.365) and stronger circulation reduced it (β = −0.363). In nutrient-dominated LMEs, iron had a strong positive association (β = 0.504), whereas deeper mixing (β = −1.103) and thermal forcing (β = −1.274) constrained Chl-a. These findings provide a mechanistic basis for anticipating coastal phytoplankton responses to climatic and land-derived pressures and for developing regionally targeted ecosystem monitoring and management strategies.

暂无