Coastal phytoplankton blooms expand and intensify in the 21st century

Phytoplankton blooms—surface accumulations of microscopic algae—can be natural and beneficial, supporting carbon fixation and fisheries, but harmful algal blooms (HABs) can produce toxins, cause hypoxia, and negatively impact ecosystems and human health. Future climate change is projected to influence bloom frequency and distribution, but heterogeneity in space and time and inconsistent sampling complicate global assessments. Although in situ HAB event databases (e.g., UNESCO IOC HAEDAT) exist, uneven effort and diverse impacts limit trend inference. Satellite ocean color observations since 1997 offer continuous surface monitoring but have faced challenges in optically complex coastal waters. This study develops a global method to map coastal blooms from satellites (2003–2020) to answer: (1) where and how frequently coastal oceans are affected; (2) whether blooms have expanded or intensified globally and regionally over the past two decades; and (3) what the potential drivers are.

The paper situates its work within research on eutrophication and HABs, noting impacts on oxygen depletion and fisheries, and the role of anthropogenic nutrient loading. Prior compilations of HAB events (e.g., HAEDAT) highlight monitoring but are limited for trend analysis due to uneven sampling. Remote sensing methods for bloom detection have advanced, including use of ocean color and fluorescence, with regional applications to HAB species. Links between upwelling systems and HAB development, climate-driven changes in phytoplankton communities, and the influence of mesoscale eddies and ocean circulation on chlorophyll and bloom phenology are established. Studies have also connected aquaculture expansion and nutrient loading with HAB incidence, and emphasized the need for improved models to attribute natural versus anthropogenic drivers.

The authors generated a global, satellite-based dataset of coastal phytoplankton bloom occurrence using 0.76 million daily MODIS-Aqua images (1-km resolution) from 2003 to 2020. An automated detection method identified blooms via satellite-detectable algal fluorescence signals, delineating bloom-affected areas and counting bloom occurrences at the 1-km pixel level. Bloom frequency (dimensionless) was computed by integrating bloom counts and affected area within 1° × 1° grid cells to examine temporal dynamics. Validation against independent satellite samples selected via visual inspection indicated >95% overall accuracy; comparisons with discrete HAEDAT events showed successful identification for 79.3% of historical HAB events. The study domain covered exclusive economic zones (EEZs) of 153 coastal countries and 54 large marine ecosystems (LMEs), encompassing continental shelves and outer coastal current margins. The authors also analyzed trends by continent and LME, and assessed associations between bloom frequency and climate/ocean variables: annual sea surface temperature (SST) averaged over local bloom growth windows, and the spatial SST gradient (proxy for mesoscale current strength/eddy kinetic energy). Relationships with global climate variability (MEI), national fertilizer usage (N or P), and aquaculture production were examined via correlations.

- Spatial distribution and occurrence: - Blooms detected in 126 of 153 coastal countries. - Total bloom-affected area: 31.47 million km² (≈24.2% of global land area; 8.6% of global ocean area). - Global median bloom count: 4.3 per pixel per year over 2003–2020. - Continental contributions to affected area: Europe 9.52 million km² (30.3%), North America 6.78 million km² (21.5%), Australia 2.84 million km² (9.0%); Africa and South America showed the highest median bloom frequencies (>6.3 per year); Australia had the lowest frequency (2.4 per year). - Hotspots include eastern boundary current systems (California, Benguela, Humboldt, Canary), northeastern USA, Latin America, Baltic Sea, Northern Black Sea, and Arabian Sea. - Top LMEs by frequency (annual median >15): Patagonian Shelf, Northeast US Continental Shelf, Baltic Sea, Gulf of California, Benguela Current. - Estuaries exhibited more bloom events than coasts without major river discharge (P<0.05). - Long-term trends (2003–2020): - Global bloom-affected area expanded by 3.97 million km² (+13.2%), at ≈0.14 million km² yr⁻¹ (P<0.05). - Global median bloom frequency increased by 59.2% (+2.19% yr⁻¹, P<0.05). - Areas with significant increasing frequency trends were 77.6% larger than those with decreases. - Spatial patterns: Southern Hemisphere largely showed increased frequency; notable weakening at low-latitude Northern Hemisphere coasts (<30° N), especially California Current System and Arabian Sea; strengthening in northern Gulf of Mexico and East/South China Seas (smaller magnitudes); weakening at higher latitudes in the northeastern North Atlantic and Okhotsk Sea. - Largest frequency increases (yr⁻¹): Oyashio +6.31%, Alaska +5.22%, Canary +4.28%, Malvinas +3.02%, Gulf Stream +2.42%, Benguela +2.30%. - Drivers and correlations: - SST vs bloom frequency: significant positive correlations at high latitudes (>40° N): Alaska Current r=0.44, Oyashio r=0.48, Baltic Sea r=0.41 (P<0.05), consistent with extended bloom-favorable seasons under warming. - SST gradient (proxy for mesoscale currents/EKE) vs bloom frequency: strong positive correlations in several current systems—Canary r=0.84, Malvinas r=0.83, California r=0.81, Benguela r=0.73, Gulf Stream r=0.61, Oyashio r=0.58 (P<0.05). Subtropical upwelling regions showed bloom changes aligned with upwelling strength changes inferred from SST gradient. - ENSO: MEI minima (2010 La Niña) preceded low bloom frequency (2011); MEI maxima (2015 El Niño) preceded highest frequency (2016). - Anthropogenic inputs: Positive correlations between fertilizer use trends and bloom frequency in China, Iran, Vietnam, and the Philippines; countries with significant positive correlations (r>0.5, P<0.05) between aquaculture production and bloom frequency include Finland, China, Algeria, Guinea, Vietnam, Argentina, Russia, and Uruguay. Some regions showed decreased fertilizer use yet increased bloom frequency, implying countervailing climate or other effects.

The study demonstrates a clear global increase in coastal phytoplankton bloom extent and frequency from 2003 to 2020, addressing the long-standing question of whether blooms are changing at a global scale. Regional contrasts are pronounced, reflecting interacting natural and anthropogenic drivers. At high latitudes, warming (SST increases) is associated with higher bloom frequency, consistent with extended growth seasons. In many boundary current systems, mesoscale circulation changes captured by SST gradients tightly track bloom trends, indicating roles for upwelling strength and current jets in nutrient delivery and bloom stimulation. ENSO variability imparts interannual signals to bloom frequency. Anthropogenic nutrient enrichment and aquaculture expansion also correlate with increased bloom incidence in several countries, though attributions vary and can be confounded by concurrent climate effects. The satellite-based approach cannot distinguish harmful from benign blooms, but its consistent methodology across space and time provides robust trend detection and spatial baselines for risk assessment and management. The lack of simple relationships between extent and frequency in some countries and LMEs underscores bloom complexity and the need for continuous monitoring and integrative models.

The authors provide the first spatially and temporally consistent global daily mapping of coastal phytoplankton blooms at 1-km resolution over 2003–2020. They find significant global increases in bloom area (+13.2%) and frequency (+59.2%), with strong regional heterogeneity linked to climate warming, mesoscale ocean circulation changes, anthropogenic nutrient inputs, and aquaculture development. The dataset establishes a global baseline to inform mechanistic understanding, forecasting, and policy decisions aimed at mitigating harmful bloom impacts while sustaining beneficial blooms. Future research should develop ecosystem models integrating terrestrial and oceanic nutrient transport and species-specific nutrient–plankton relationships, and conduct region-focused studies to disentangle drivers of mesoscale circulation changes and quantify the relative contributions of natural versus anthropogenic forcing.

Satellite detection is constrained by sensor sensitivity and algorithm thresholds, leading to underestimation of bloom incidence relative to in situ HAB records, especially for low-cell-concentration events. The method cannot distinguish harmful from non-harmful blooms. Bloom counts are averaged across all 1-km pixels within EEZs, differing from discrete sampling regions in HAEDAT. Despite these limitations, uniform methods across space and time support robust trend inference. Some regions show weak or no relationships between bloom extent and frequency, and attribution of mesoscale changes to specific drivers (wind, stratification, current shear) requires regional investigation. Potential confounding between climate effects and anthropogenic inputs (e.g., fertilizer reductions concurrent with warming) complicates causal attribution.