How Does Climate Change Influence the Regional Ecological–Social Risks of Harmful Dinoflagellates? A Predictive Study of China's Coastal Waters

ABSTRACT

Harmful dinoflagellates are widely distributed in coastal waters worldwide, posing multiple ecological and socioeconomic threats. Climate change may alter the biogeography of these species; however, few studies have linked shifts in harmful dinoflagellates' ecological distribution to their socioeconomic impacts. This study developed a framework to assess the spatiotemporal ecological–social risks posed by harmful dinoflagellates, identifying these algae as risk sources and considering mariculture and coastal populations as the primary risk receptors. China is the world's largest mariculture producer, with approximately 600 million residents living in coastal areas. Focusing on 14 key harmful dinoflagellate species in Chinese coastal waters, we evaluated ecological–social risks under present conditions and two projected climate scenarios for 2100. Our findings indicate that climate change may lead to reductions in suitable habitats for harmful dinoflagellates in tropical and subtropical regions, while habitats in higher‐latitude areas are likely to remain stable or expand. Risk area expansion is projected for four species and increased average risk intensity for three, with two species experiencing both. Nationally, total risk area is projected to remain stable, while cumulative risk intensity may decline by 16.64%. Regionally, risk intensity is expected to rise in northern provinces (up to 30.46%) and decline across most southern provinces. Importantly, we reveal a potential spatial “decoupling” of risk sources and receptors along the coast of China in the future. This decoupling demonstrates a reduced overlap between harmful dinoflagellate distributions and areas with dense mariculture or populations. Our findings suggest that, contrary to the common assumption that climate change universally exacerbates harmful algal impacts, these effects may vary across regions and species, highlighting the importance of localized adaptation strategies in risk assessment. This study provides a robust tool for understanding harmful dinoflagellate risks under climate change, thereby supporting the sustainable management of coastal ecosystems.

Climate change is expected to shift the risks posed by harmful dinoflagellates in China's coastal waters. In this study, the ecological–social risk is assessed by considering both the potential presence of these algae and their overlap with coastal populations and mariculture areas. The study finds that while some regions and species may face higher risks, others may see reductions, and future risk zones may not always overlap with areas of dense mariculture or population. These findings highlight that the impacts of harmful dinoflagellates under climate change are complex and region‐specific, emphasizing the need for tailored local responses.

1. Introduction

Harmful dinoflagellates are algae that proliferate rapidly under favorable environmental conditions, releasing toxins that pose significant ecological and public health risks (Berdalet et al. 2016; Glibert, Berdalet, et al. 2018; Glibert, Beusen, et al. 2018; P. M. Glibert 2024a). Biotoxins released by harmful dinoflagellates during metabolic processes can travel through the food chain, accumulate in filter‐feeding organisms, and be ingested by higher trophic‐level predators and humans. This can lead to fatalities in both animals and humans (Morquecho 2019). The rapid proliferation of harmful dinoflagellates often leads to harmful algal blooms (HABs), resulting in the closure of aquaculture farms and substantial economic losses (Blasco et al. 2003; Díaz et al. 2019; Telesh et al. 2024). The impacts of harmful dinoflagellates are particularly pronounced in densely populated coastal areas such as bays, estuaries, and semi‐enclosed seas, where the risks to human health and the environment are magnified (Anderson et al. 2002; Skarlato et al. 2018; Glibert et al. 2014). Therefore, identifying the distribution of harmful dinoflagellates and evaluating their potential ecological and social risks are essential. This information provides critical scientific evidence for national and local governmental agencies, the aquaculture industry, and the general public, enabling them to implement effective mitigation strategies and minimize the adverse effects of harmful dinoflagellates (Ding et al. 2022; Stoner et al. 2023).

In recent decades, offshore marine environments have undergone significant changes, including increasing sea temperatures (Zalewska et al. 2024), ocean acidification (Van Der Zwaan and Gerlagh 2016), and nutrient inputs (P. M. Glibert 2020). These changes may directly or indirectly affect the growth of dinoflagellates (Khan et al. 2018), alter the geographic distribution of harmful algal species (Rubio‐Salcedo et al. 2017; Lin et al. 2025), and increase the likelihood of HABs occurring (P. M. Glibert 2020; Moore et al. 2009; Berdalet et al. 2016; Kudela et al. 2015; Wells et al. 2015). Climate warming has been shown to affect the distribution of algal communities (Des et al. 2020), as temperature plays a critical role in regulating the physiology of harmful algae in coastal systems by affecting their spatial and temporal distribution and abundance (Jacobs et al. 2015). For example, the range of the tropical marine dinoflagellate Gambierdiscus toxicus is projected to extend to higher latitudes (Moore et al. 2008). In the North Atlantic, the biogeographic distribution of many algae, including 87 diatom and dinoflagellate taxa, is expected to migrate northward (Barton et al. 2016). In addition, the abundance of harmful diatom and dinoflagellate genera in the North Sea has increased significantly since the 1970s (Nohe et al. 2020). Some HAB events have been observed at higher latitudes. One notable example is the occurrence of Trichodesmium erythraeum blooms, which were first recorded above 30° N in 2010 (Spatharis et al. 2012). Conversely, some algal species may be less suited to their original environment. For instance, the abundance of dinoflagellates such as Prorocentrum spp. in the North Sea has decreased since 2000, whereas paralytic shellfish poisoning toxins have diminished in Scottish waters (Townhill et al. 2018). In general, it is widely recognized that the impacts of HABs are expected to expand as climate change intensifies (Wells et al. 2015). In addition, the species responsible for HABs within the same region may change (Feng et al. 2024), affecting aquaculturists, local beach managers, and other stakeholders (P. M. Glibert 2020; Vieira et al. 2024; Horemans et al. 2023).

Numerous studies have focused on short‐term forecasting and early warnings of HABs. For example, methods based on satellite remote‐sensing data and improved algorithms have been widely used to monitor the spatial and temporal distribution of HABs (Lekki et al. 2019; Gokul et al. 2019; Fernandes‐Salvador et al. 2021). In addition, isothermal nucleic acid amplification strategies have been used to detect HABs (Toldrà, Alcaraz, et al. 2019; Toldrà, O'Sullivan, and Campàs 2019). Ding et al. demonstrated that the risk of HABs in the East China Sea could be assessed 1–2 months in advance using a sea surface temperature (SST)‐based method (Ding et al. 2022). However, long‐term predictions of climate change‐induced alterations in the biogeographic distribution of harmful algae remain insufficient. Incorporating biogeography into climate change impact studies and utilizing models to translate past observations into future projections could improve our understanding of how environmental changes affect dinoflagellate distribution (Hannah et al. 2002; Anderson et al. 2015). Some researchers have applied coupled oceanographic–biogeochemical models to project future changes in HAB distribution and associated risks (Glibert, Berdalet, et al. 2018; Glibert, Beusen, et al. 2018). For example, a biogeochemical model incorporating genus‐specific physiological rules—based on temperature, salinity, and nutrient stoichiometry—has been used to project potential shifts in HAB distribution across Europe and Asia under climate change scenarios (Glibert et al. 2014). Similarly, Wang, Bouwman, Wang, et al. (2023) used a coupled watershed nutrient export model and coastal hydrodynamic–biogeochemical model to assess future HAB risks in the eastern Chinese coastal seas. Another applicable approach is species distribution modeling (SDM) (Jones 2013; Weinert et al. 2016; Brun et al. 2016). This model assesses how the distribution of HAB species responds to climate change by considering their environmental preferences and potential shifts (Wilson et al. 2019). In recent years, SDM based on machine learning algorithms, such as MaxEnt and random forests, has been used to predict spatial changes in HAB species (Hu et al. 2024). Townhill et al. projected that because of climate change, conditions suitable for HAB species in the northwestern European continental shelf seas would migrate further north (Townhill et al. 2018). Similarly, Borges et al. (2022) predicted that by 2100, the global distribution of three dinoflagellate species predominantly producing paralytic shellfish toxins would shift toward the polar regions, based on SDM (Borges et al. 2022). Wang, Xu, Gu, et al. (2023) predicted the future habitat suitability of Gymnodinium catenatum (Wang, Xu, Gu, et al. 2023). However, these studies seldom link the changes in the ecological distribution of harmful dinoflagellates to the resulting comprehensive socioeconomic risks.

As the world's largest aquaculture producer, China contributes over two‐thirds of the global seafood production, with a mariculture output value of 488.548 billion CNY by 2023 (Meng and Feagin 2019; FAMA, 2024). Furthermore, China's coastal population of approximately 600 million represents 43.5% of the national population, making it one of the most densely populated regions in the world (Tian et al. 2016). Since 2002, China has experienced 1407 red tides, posing significant risks to both socioeconomic conditions and public health (Gao et al. 2024). As the distribution of harmful algae changes, the associated risks to mariculture and public health vary. However, a gap exists in the quantitative assessment and prediction of these risk changes. This study aims to develop a novel methodology for predicting the ecological and social risks associated with harmful dinoflagellates under climate change. Here, risk is defined as the potential socioeconomic impact of harmful dinoflagellate distribution on mariculture and coastal populations. It consists of two key components: risk sources and risk receptors. The risk source is the presence of harmful dinoflagellates, represented by habitat suitability, where higher suitability indicates a greater likelihood of occurrence. The risk receptors include mariculture and coastal populations, with socioeconomic impact measured by the affected mariculture value and coastal population size. We focus on the Chinese offshore region, taking 14 harmful dinoflagellates as risk sources. The objectives of this study were to (1) predict changes in the distribution of harmful dinoflagellates along the Chinese coast under two climate change scenarios for the year 2100 using machine learning‐based SDMs; (2) evaluate the potential impact of these harmful dinoflagellates on mariculture operations and coastal populations; and (3) assess spatial and temporal trends in ecological–social risk, taking into account both the extent and intensity of the risk. This study provides essential data for ecosystem risk management in this region and serves as a scientific foundation for coastal communities to address the challenges posed by climate change.

2. Materials and Methods

2.1. Study Area

China's coastal regions lie along the edge of the East Asian continent, spanning 22° latitude, and include tropical, subtropical, and warm temperate climate zones (Figure 1). The study area encompasses the mainland coast of China (105–127° E, 17–42° N), covering the Bohai Sea, Yellow Sea, East China Sea, and South China Sea coasts (Kang et al. 2021; Liu 2008; Hu et al. 2022). This region benefits from a relatively warm climate, with major rivers, such as the Yangtze and Yellow Rivers, delivering significant volumes of freshwater and nutrients to coastal waters, creating favorable conditions for dinoflagellates (Wang et al. 2021; Xiao et al. 2021). More than 300 dinoflagellate species thrive in these waters. Among these, 21 are the main species causing HABs (Chen, Zhang, et al. 2023; Gu et al. 2022). Excessive algal growth has triggered significant offshore ecological disasters in China (Yu et al. 2023; MEE, 2023). From 2011 to 2022, dinoflagellates were responsible for the most frequent and widespread HAB events. Remote‐sensing data from 2003 to 2022 show that algal blooms affect an average annual area of approximately 180,000 km² along the coastline (Zeng et al. 2024). Given the densely populated coastal areas and extensive aquaculture industries, socioeconomic losses due to HABs can be significant in some years. A single large‐scale bloom may cause direct economic losses of two billion CNY and trigger public health crises, including hundreds of poisoning cases (Zhou and Zhu 2006; Yu and Liu 2016; Gu et al. 2022). Consequently, assessing the risks posed by harmful dinoflagellates in these coastal regions is crucial for mitigating their impact on ecosystems and socioeconomic systems.

FIGURE 1.

Study area extent and distribution of the mariculture industry. Map lines delineate study areas and do not necessarily depict accepted national boundaries.

2.2. Methodology Framework

Of the 21 major harmful dinoflagellate species along the Chinese coast (Chen, Zhang, et al. 2023; Gu et al. 2022), five lacked sufficient distribution records for modeling, and two were excluded owing to sampling and identification controversies. Thus, we selected 14 species with ample data for this study. To assess and predict their potential impacts, we developed an ecological–social risk assessment framework consisting of two core components: a risk source assessment module and a risk receptor assessment module (Figure 2). The risk source module evaluates the spatial likelihood of harmful dinoflagellate presence based on habitat suitability under current and projected climate scenarios for 2100. The risk receptor module quantifies the socioeconomic exposure by mapping mariculture production value and coastal population size in affected areas. Risk area and intensity were then derived from the spatial overlap between sources and receptors. Considering the differences in toxicity among dinoflagellate species (Dorantes‐Aranda 2023; Kudela et al. 2005; Yuan et al. 2024), we incorporated a toxicity coefficient (TC) into the risk intensity calculation. This coefficient is determined by the LD₅₀ values of toxins produced by specific dinoflagellate species, with higher toxicity corresponding to a larger TC value. The detailed calculation process for TC is provided in Appendix I. The risk intensity for each 5‐arcminute grid cell was calculated using the following formula:

$${\text{Risk}}_i={\mathrm{RS}}_i\times {\mathrm{RR}}_i\times \mathrm{TC}$$

where Risk _i represents the risk intensity for a particular harmful dinoflagellate species in grid cell i; RS _i is the risk source index characterized by the habitat suitability of harmful dinoflagellates in grid cell i; RR _i is the risk receptor index in grid cell i, calculated using the formula given in Section 2.4.3; and TC is the toxicity coefficient for the species.

FIGURE 2.

Methodology framework.

2.3. Risk Source Assessment Method

2.3.1. Data Sources and Pre‐Processing

SDM requires data regarding species occurrence and environmental variables. To minimize bias from uneven sampling in different regions, we used global distribution records for the 14 dinoflagellate species to build the SDMs. Species occurrence data were sourced from the Ocean Biogeographic Information System (OBIS 2022) and the Global Biodiversity Information Facility (GBIF.org 2022), yielding 12,535 occurrence records. The spocc R package (version 1.2.0) was used to filter out duplicates, erroneous entries, and spatial redundancies (Chamberlain 2021). After data cleaning, 8490 occurrence points were extracted for modeling (see Appendix II). Then a method that combines spatial constraints with environmental subsampling was adopted to generate pseudo‐absence points, ensuring environmental heterogeneity (Barbet‐Massin et al. 2012; Shipley et al. 2022). For each species, three times as many pseudo‐absence points were generated as the occurrence points. Details of the generation of the pseudo‐absence points and the corresponding dataset are provided in Appendix III.

To simulate the impacts of climate change, we selected the shared socioeconomic pathways (SSP) 1–2.6 and SSP5‐8.5 scenarios to model the spatial and temporal distribution of harmful dinoflagellates for the present and for 2100 at a resolution of 5 arcmin (approximately 10 km). These scenarios were proposed by the IPCC, providing a standardized framework for projections in climate change research (O'Neill et al. 2017; Riahi et al. 2017). SSP1‐2.6 assumes a sustainable development pathway with low emissions and strong climate mitigation, while SSP5‐8.5 represents a fossil‐fuel‐driven economy with high emissions and minimal mitigation efforts, leading to the most extreme climate impacts. In this study, the environmental variables used in the SDMs include three categories: physical, chemical, and biological parameters, with a total of 44 variables. Appendix IV provides a full list of environmental variables and data resources. An optimized selection process that coupled collinearity control with contribution analysis was adopted to select appropriate variables for each species (Phillips 2017). This process resulted in a set of 11 environmental variables as predictive factors for each species. All variables were resampled to a resolution of 10 km for further analysis. Details of the selection process are provided in Appendix IV.

2.3.2. Habitat Suitability Modeling

An integrated SDM approach, combining maximum entropy model (MaxEnt) and artificial neural networks (ANN), was used to simulate habitat suitability for 14 harmful dinoflagellate species under current and future climate scenarios. The MaxEnt and ANN algorithms were selected as the top performers after testing seven SDM algorithms. The outputs of the two algorithms were assembled by averaging the predicted outputs to obtain stable and robust results (Araujo and New 2007; Marmion et al. 2009; Tanaka et al. 2020). Modeling was performed using the BIOMOD2 R package (Thuiller et al. 2019) in R (version 3.5.1; R Core Team 2017). To ensure reliable results, we independently modeled the habitat suitability for each species following the ODMAP protocol provided in Appendix V (Zurell et al. 2020).

For each species, we ran both algorithms 10 times and averaged the results before combining them into an ensemble model. This process resulted in 280 model runs. For parameter tuning, we used the caret R package (version 6.0) for the ANN model (Kuhn 2008; Breiner et al. 2018) and the ENMeval R package (version 2.0) for MaxEnt (Valavi et al. 2019; Kass et al. 2021). Detailed processes were described in Appendix VI. To reduce the spatial dependence between the training and validation sets and ensure lower spatial autocorrelation in the test set, we used the blockCV R package (version 3.0.0, Valavi et al. 2019) and applied a five‐fold spatial cross‐validation method to evaluate the model's predictive performance (Huang et al. 2024). Model performance was evaluated using the AUC, Cohen's kappa coefficient, and the true skill statistic (TSS), with a minimum TSS value of 0.8 for ensemble models (Swets 1988; Allouche et al. 2006; Phillips et al. 2006; Tanaka et al. 2020). The habitat suitability output of the ensemble models was binarized using thresholds derived from ROC curves, with areas of species presence representing the distribution of harmful dinoflagellates. Notably, to ensure robust extrapolation and avoid underfitting in localized regions, we constructed global species distribution models for each dinoflagellate species using worldwide occurrence data. Modeling at the global scale allowed us to account for the full ecological niche of each species and avoid biases that could arise from restricting the modeling extent. Finally, the global model outputs were clipped to China's coastal waters for regional analysis.

2.4. Risk Receptor Assessment Method

2.4.1. Estimation of the Economic Value of Affected Mariculture

China's coastal regions host some of the largest mariculture areas in the world (FAO, 2020), covering over 20,000 km² (Figure 1). To estimate the spatial distribution of mariculture production value, we divided China's offshore waters into 5‐arcminute (~10 km) grid cells. The mariculture production value for each grid cell was calculated using the following formula:

$$M_i=S_i\times {\mathrm{OV}}_i$$

where M _i represents the mariculture production value in grid cell i, S _i is the mariculture area within grid cell i, and OV _i denotes the annual output value of mariculture per km², which varies by province. The S _i was calculated using the mariculture map derived from high‐frequency remote‐sensing imagery (Liu et al. 2022). Considering that distinct regions engage in the cultivation of disparate species, which consequently yield disparate economic values, the OV _i for each province (Table 1) was estimated based on provincial aquaculture statistics from 2022 (FAMA, 2023). Given the predictive nature of this study, we examined the temporal dynamics of China's mariculture industry. Based on national fisheries statistics in the past 5 years (FAMA, 2019; FAMA, 2024), we analyzed the variation in production value per km². Under discount rates of 5%–6%, the annual average change in discounted output value per km² ranged from +0.36% to −0.58%, suggesting minimal variation over time. Therefore, we assumed that the OV _i would remain relatively stable. We also assumed that the mariculture area remains constant over time. Although mariculture is subject to various sources of spatial disturbance, including long‐term climate change, extreme weather events, and short‐term policy adjustments, incorporating all these uncertainties may introduce additional noise and reduce the robustness of the projections. Moreover, since the risk receptor scoring is based on normalized values, the absolute mariculture area has limited influence on the resulting risk intensity; rather, it is the relative spatial distribution that plays a more critical role. According to national fisheries statistics, the spatial pattern of mariculture—particularly the proportional allocation across provinces—has remained relatively stable in recent years. Therefore, for both current and future risk assessments, we used the present‐day distribution of mariculture area as a fixed input, rather than projecting spatial changes over time. After obtaining the spatial distribution map of the mariculture production values (Figure S1), the distribution areas of harmful dinoflagellates (simulated in Section 2.3) were overlaid onto the aquaculture production areas to calculate the total production value affected by these dinoflagellates.

TABLE 1.

Statistics on mariculture in coastal provinces of China.

Province	Output value of mariculture production (104 CNY)	Area (km2)	Output value per unit area (104 CNY/km2)
Liaoning (LN)	4,556,230	6772.01	672.80
Hebei (HB)	1,960,568	1055.87	1856.83
Tianjin (TJ)*	34,880	10.05	3470.65
Shandong (SD)	11,397,630	6174.64	1845.88
Jiangsu (JS)	3,271,058	1721.88	1899.70
Shanghai (SH)*	—	—	—
Zhejiang (ZJ)	2,657,852	834.39	3185.38
Fujian (FJ)	10,696,136	1679.53	6368.53
Guangdong (GD)	8,427,380	1665.96	5058.57
Guangxi (GX)	2,553,930	673.93	3789.61
Hainan (HN)	832,785	155.94	5340.42

^*In Tianjin, the jurisdictional marine area spans 2146 km², yet only 10 km² is used for mariculture. Within the 5‐arc‐minute evaluation grid, mariculture occupies on average less than 0.5% of each cell—below the minimum precision required for our calculations. Shanghai has no marine aquaculture. Therefore, they were excluded from subsequent province‐level statistical analyses.

2.4.2. Estimation of the Population Affected by Harmful Dinoflagellates

Harmful dinoflagellates produce a variety of toxins that cause different types of poisoning, including paralytic shellfish poisoning, diarrhetic shellfish poisoning (DSP), neurotoxic shellfish poisoning, and others. These toxins accumulate in marine organisms, such as shellfish and fish, enter the food chain, and potentially harm humans who consume contaminated seafood. Based on the records of harmful dinoflagellate outbreaks in China over the past 20 years (Chen, Liang, et al. 2023; Hou et al. 2022; Gu et al. 2022), nearly all poisoning incidents have been linked to local seafood consumption. Therefore, we estimated the size of the potentially affected population based on the distribution range of seafood in specific aquaculture areas and ports. Seafood distribution ranges depend on factors such as seafood type, market demand, and transportation methods. Owing to its short shelf life, fresh seafood typically circulates within local or regional markets tens to hundreds of kilometers from aquaculture sites or ports (Song et al. 2024). Using the locations of mariculture farms and 1154 coastal fishing ports (Figure 1) as central points (Wang, Hu, Ding, et al. 2023), we defined the local market coverage using the size of the coastal municipal administrative districts (Table S1) and created buffer zones to represent the seafood distribution range. We overlaid the harmful dinoflagellate distribution areas with these central points and their buffer zones to identify the at‐risk populations. Population data were obtained from National and Provincial Population and Economy Projection Databases under SSP1‐5, with a spatial resolution of 5 arc‐minutes (Jiang et al. 2022). We extracted the population grid data for 2020 and 2100 to match the risk source predictions.

2.4.3. Calculation of the Risk Receptor Index

The risk receptor index for each grid cell was calculated by normalizing the scores of the mariculture production value and population size using the following formula:

$$\mathrm{RR}=\text{AVERAGE}\left(M_{\text{normalized}}+P_{\text{normalized}}\right)$$

where RR is the risk receptor index, M _normalized is the normalized score of the affected mariculture production value in a specific grid cell, and P _normalized is the normalized score of the affected population size in the grid cell.

3. Results

3.1. Distribution Changes of Risk Sources

Under current conditions, a total of 1.17 million km² of China's coastal waters are considered suitable habitats for harmful dinoflagellates. Projections for future scenarios indicate that the total suitable habitat area will range from 0.69 million km² (SSP5‐8.5) to 1.14 million km² (SSP1‐2.6) by 2100. The main environmental factors that determine the distribution of most species are distance from the shore, bathymetry, and annual mean SST (Table S2). Most dinoflagellate species are found at depths of up to 100 m, within 50 km from the coastline, and in areas where the annual mean SST ranges between 15°C and 20°C (Figure 3a). The environmental niche ranges of dinoflagellates across all variables are provided in Appendix IV. Currently, Karlodinium veneficum has the largest potential distribution, covering 0.41 million km², while Karenia mikimotoi has the smallest, spanning only 0.03 million km² (Table 2). Due to climate change, the spatial distribution of dinoflagellate species along the Chinese coast is projected to shift significantly by 2100 (Figures S2–S15 show the spatiotemporal distribution of each species). Under the SSP1‐2.6 scenario, the distribution areas of five species, including Akashiwo sanguinea, Alexandrium minutum , and Gonyaulax verior , expanded by 12.65%–191.44%. In contrast, the remaining nine species were predicted to experience reductions in their distribution areas, ranging from 3.01% to 63.91% (Figure 3b). Under SSP5‐8.5, A. sanguinea and Prorocentrum lima are projected to increase their ranges by 4.12%–31.43%, whereas the distribution areas of the other 12 species will decrease by 5.85%–84.89% (Figure 3b).

FIGURE 3.

Ranges of key environmental variables (a) and potential habitat loss and gain for the 14 species (b). Species codes are referenced in Table 2.

TABLE 2.

Suitable habitat for each harmful dinoflagellate species under different scenarios.

Species	CODE	Habitat area at present (km2)	Habitat area under SSP1‐2.6 (km2)	Habitat area under SSP5‐8.5 (km2)
Akashiwo sanguinea	SP.1	91,658.80	223,331.63	120,468.90
Alexandrium minutum	SP.2	142,343.23	213,834.96	134,020.32
Alexandrium ostenfeldii	SP.3	154,614.20	55,806.23	87,604.04
Coolia monotis	SP.4	322,566.42	301,972.53	296,637.33
Dinophysis acuminata	SP.5	77,467.16	50,364.32	37,346.43
Gonyaulax spinifera	SP.6	163,043.83	158,135.44	125,697.40
Gonyaulax verior	SP.7	215,115.41	242,324.95	192,280.74
Gymnodinium catenatum	SP.8	312,109.42	161,549.97	47,163.20
Karenia mikimotoi	SP.9	34,892.23	101,688.98	32,544.74
Karlodinium veneficum	SP.10	415,932.48	263,772.47	263,879.18
Margalefidinium polykrikoides	SP.11	78,747.61	63,275.52	53,672.15
Noctiluca scintillans	SP.12	255,662.96	206,792.50	156,748.28
Prorocentrum lima	SP.13	354,470.93	824,822.49	369,089.39
Protoceratium reticulatum	SP.14	384,454.78	298,984.82	298,237.89

The spatial distribution of suitable habitats and the species richness of harmful dinoflagellates along a latitudinal gradient in the coastal waters of China were estimated (Figure 4a). Currently, the East China Sea hosts the majority of suitable habitats, with 42% of the total habitat located between 23° N and 31° N. Under the SSP1‐2.6 scenario, habitat extent remains relatively stable. In contrast, under SSP5‐8.5, suitable habitats in the South China Sea and East China Sea are projected to decline by 47%, with the greatest reduction occurring around 20° N. Meanwhile, habitat areas in the Yellow Sea and Bohai Sea are expected to expand by approximately 8%, with the largest increase occurring near 39° N (Figure 4b). These findings suggest that suitable habitats for harmful dinoflagellates may contract at lower latitudes while remaining stable or expanding at higher latitudes. Currently, the number of species follows a latitudinal pattern, with lower diversity in the middle and higher diversity at both ends. Fewer than eight species were present between 27° N and 29° N, while species richness peaked between 12 and 14 species north of 34° N in the Yellow and Bohai Seas (Figure 4c). In the future, the number of harmful dinoflagellate species in the Yellow Sea and Bohai Sea is expected to remain high, with habitats for species such as A. sanguinea and K. mikimotoi likely to expand significantly in these regions. In contrast, species richness south of 33° N is projected to decline markedly, particularly under SSP5‐8.5. Species such as A. ostenfeldii and G. catenatum may be nearly absent from the South China Sea and southern East China Sea.

FIGURE 4.

Spatial distribution maps of harmful dinoflagellates under different scenarios and their distributional changes along the latitudinal gradient. (a) Spatial and temporal distribution map of dinoflagellate species richness. (b) Changes in average habitat area of 14 species along latitude. (c) Changes in species number along latitude. Map lines delineate study areas and do not necessarily depict accepted national boundaries.

3.2. Changes in Risk Receptors

In the marine aquaculture sector, Protoceratium reticulatum and C. monotis have the greatest potential impact, with annual production values estimated between 184.1 and 185.2 billion CNY. Under both climate change scenarios, the scale of mariculture impacted by five species, including K. mikimotoi, A. sanguinea , and P. lima , is expected to increase by 3.82%–181.13% (Figure 5a). Meanwhile, mariculture potentially affected by nine other species, such as G. verior , A. ostenfeldii , and C. monotis , may decrease by 1.80%–75.72% (Figure 5c,d). This decrease suggests a potential decoupling between the risk sources and marine aquaculture as the risk receptor (Table S3). Although the impacts on individual species vary significantly with climate change, the total production value of marine aquaculture affected by these 14 species remains relatively stable. Across all scenarios by 2100, the total production value potentially affected by harmful dinoflagellates is projected to remain between 212.0 and 212.2 billion CNY.

FIGURE 5.

Changes in risk receptors under different climate change scenarios. (a) Mariculture production value affected by each dinoflagellate species, with species codes referenced in Table 2. (b) Population numbers affected by each dinoflagellate species. (c, d) Spatial distribution changes of the affected mariculture industry and its product distribution range, illustrated using K. mikimotoi and A. ostenfeldii as examples. Province abbreviations are referenced in Table 1. Map lines delineate study areas and do not necessarily depict accepted national boundaries.

Regarding the affected population, C. monotis , A. minutum , and P. lima are projected to impact the largest populations, estimated to range from 115.74 to 213.41 million people. In contrast, G. catenatum was expected to impact the smallest population size, with estimates ranging from 4.63 to 8.91 million (Figure 5b). By 2100, it is projected that the number of people affected by the 13 dinoflagellate species will decrease by 20%–45%. However, K. mikimotoi stands out as the only species for which the impacted population is expected to increase (Figure 5c). Currently, the total number of people potentially affected by these 14 harmful dinoflagellates is estimated at 228 million. Under the SSP1‐2.6 and SSP5‐8.5 scenarios, this number is projected to decline to 109 and 116 million by 2100, respectively. This suggests that the population within the impact range of harmful dinoflagellates will decrease significantly, indicating a potential widespread decoupling between the affected population, as the risk receptor, and the risk sources (Table S4).

3.3. Ecological–Social Risk Under Different Climate Scenarios

Across China's coastal regions, the largest risk areas are concentrated along the coast of Shandong Province in the Yellow Sea and near Liaoning Province in the Bohai Sea (Figure 6a). With climate change, the extent of risk zones in different regions is expected to shift. In southern provinces such as Guangdong and Guangxi, risk areas are expected to decrease, with the proportion of risk zones south of 31° N projected to decrease from 39.8% to 31.9% and 35.3%, respectively. In contrast, the risk areas in northern provinces, such as Liaoning, are expected to expand further (Figure S16). However, due to the compensatory effects of increases and decreases across regions, the total risk area at the national scale is expected to remain relatively stable across different scenarios, ranging from approximately 53,273 to 53,441 km². Among the 14 species, P. reticulatum and C. monotis consistently exhibited the largest risk areas across all scenarios, covering 36,045–36,297 km² (Figure 6b). Climate change is expected to expand the risk areas for A. sanguinea , K. mikimotoi, P. lima , and Noctiluca scintillans by 13.43%–268.97%, whereas the risk areas for other species are projected to remain stable or decrease (Figure 6b).

FIGURE 6.

Spatiotemporal changes in risk area extent. (a) Total risk area extents along the Chinese coast and their changes under different scenarios, displayed using 6000 km² hexagonal grids and provincial boundaries. Province abbreviations are referenced in Table 1. (b) Projected changes in risk area extent by 2100 for each species under different scenarios, with species codes referenced in Table 2. Map lines delineate study areas and do not necessarily depict accepted national boundaries.

The spatial patterns and trends of risk intensity differ from those of risk area (Figure 7a). Across all scenarios, Shandong and Guangdong provinces exhibit the highest cumulative risk intensity, ranging from 23.46 to 40.65, while Hainan shows the lowest, with values between 0.36 and 0.44. Under the influence of climate change, cumulative risk intensity is projected to increase in northern provinces along the Yellow and Bohai Seas—particularly Shandong, Hebei, and Liaoning. The most significant increase is expected in Liaoning under the SSP1‐2.6 scenario, with a rise of 30.46%. In contrast, provinces along the East and South China Seas are likely to experience a decline in cumulative risk intensity, with the largest decrease occurring in Guangxi under SSP5‐8.5, where a 47.53% reduction is projected (Figure S17). At the national scale, cumulative risk intensity under SSP1‐2.6 shows no significant change compared to the current scenario, while under SSP5‐8.5, it is projected to decline by 16.64%. Among the 14 species, A. ostenfeldii , G. catenatum , and K. mikimotoi are projected to exhibit relatively high mean risk intensity (i.e., the average risk level calculated across spatial grid units) across all scenarios, ranging from 0.045 to 0.056. In contrast, C. monotis , Gonyaulax spinifera , and Dinophysis acuminata are predicted to have lower mean values, ranging from 0.020 to 0.029. Under SSP1‐2.6 or SSP5‐8.5 scenarios, the average risk intensity of K. mikimotoi, M. polykrikoides, and P. lima is expected to increase, while that of other species is likely to decline. The largest projected increase in risk intensity occurs for K. mikimotoi under SSP1‐2.6, with a rise of 21.01%.

FIGURE 7.

Spatiotemporal changes in cumulative risk intensity values. (a) Projected changes in total cumulative risk intensity values along the Chinese coast under different scenarios, displayed using 6000 km² hexagonal grids and provincial boundaries. Province abbreviations are referenced in Table 1. (b) Projected changes in average risk intensity for each species under different scenarios, with species codes referenced in Table 2. Map lines delineate study areas and do not necessarily depict accepted national boundaries.

Overall, under the influence of climate change, both the risk area and risk intensity of harmful dinoflagellates are projected to increase in Liaoning, which may warrant prioritization in future risk management. At the species level, the risk areas of A. sanguinea , K. mikimotoi, P. lima, and N. scintillans are expected to expand, while the average risk intensity of K. mikimotoi, M. polykrikoides, and P. lima is projected to increase. These findings suggest that both the spatial extent and intensity of risk associated with the toxin‐producing species K. mikimotoi and P. lima are likely to be exacerbated under climate change, highlighting the importance of prioritizing these species in future risk management efforts.

4. Discussion

4.1. Distribution and Changes in Harmful Dinoflagellates

Analysis of environmental variables indicated that distance from the shore, bathymetry, and SST were the primary drivers of dinoflagellate habitat distribution. Although temperature and nutrients are often considered crucial for determining algal distribution (Boivin‐Rioux et al. 2021; Heisler et al. 2008), our study highlights the significant influence of topographical factors, suggesting that essential physical oceanographic processes may play an underestimated role in shaping dinoflagellate distribution. A large‐scale survey in the East China Sea found that major dinoflagellate blooms commonly occurred between 27° N and 31° N in areas where the seabed topography changed sharply at depths of 30–50 m. Researchers have suggested that key ocean processes driven by topographical variations directly or indirectly affect water mass structure as well as temperature and salinity conditions, which in turn either constrain or enhance dinoflagellate growth (Zhou and Zhu 2006). Additionally, the distance from the shore likely correlates with nutrient input, particularly in nearshore areas of the East China Sea and South China Sea, where inputs from the Yangtze River estuary and Pearl River delta contribute to the concentration of suitable dinoflagellate habitats (Li et al. 2009; Wang et al. 2018; Gu et al. 2022). Nearshore‐offshore gradients may also be closely linked to nutrient stoichiometry. A study conducted in the coastal waters of eastern China reported that, over a 30‐year period, DIN concentrations increased by more than 80%, while DIP increased by 16%–39% in areas shallower than 20 m. These disproportionate increases led to a rapid rise in DIN:DIP and TN:TP ratios in shallow, nearshore waters, potentially promoting the growth of dinoflagellates (Wang, Bouwman, Van Gils, et al. 2023). Collectively, these factors—and their complex interactions—shape the distribution of suitable habitats for dinoflagellates in the coastal waters of China.

The findings of this study indicate that under current conditions, the East China Sea (23–31° N) contains the largest area of suitable habitat for harmful dinoflagellates in China's coastal waters. Over the past 30 years, 59% of the HAB events and 64% of the affected areas in China's coastal regions have been recorded in this zone (Zeng et al. 2019; He et al. 2021; Gu et al. 2022). Satellite‐based remote‐sensing studies have identified algal bloom distribution patterns across China, showing that the East China Sea has the highest bloom intensity (Zeng et al. 2024; Chen, Liang, et al. 2023). Xiao et al. (2019) analyzed the frequency and duration of HABs over 45 years (1970–2015), highlighting the particularly high frequency and duration of blooms between 28° N and 33° N. These observations are consistent with our estimates of harmful dinoflagellate distribution and reveal a strong correlation between suitable habitat availability and HAB occurrence. As climate change progresses, we found that both suitable habitats and species richness of harmful dinoflagellates are expected to decline at low latitudes and stabilize or increase at higher latitudes. In China's coastal waters, this is reflected by a decline in harmful dinoflagellates in the South and East China Seas and an increase in the Yellow and Bohai Seas.

This latitudinally divergent pattern under climate change may result from the fact that many major HAB species thrive in temperatures between 15°C and 25°C, and as southern seas warm, conditions may become less suitable for these species. Predictive studies have indicated a northward shift in the distributional centroids of dinoflagellates (Borges et al. 2022). A 60‐year time‐series study of North Atlantic phytoplankton showed that dinoflagellate distributions closely tracked shifts in isotherms (Chivers et al. 2017), aligning with projections that marine plankton migrate toward the poles as temperatures rise (Thomas et al. 2012; Goldsmit et al. 2020). The redistribution of dinoflagellates under climate change conditions may lead to community decline in some areas. For instance, a decrease in dinoflagellate HAB groups has been observed in the northeastern Atlantic (Hinder et al. 2012), whereas other species are expanding their ranges. In North America, A. catenella blooms are expected to reach the Gulf of St. Lawrence and the Scotian Shelf along Canada's east coast (Boivin‐Rioux et al. 2021). Long‐term observational data from China corroborate these projections. The emergence and expansion of numerous new HAB species along the coastline of China have been observed over the past four decades (Gu et al. 2022). Moreover, a comparison of historical (2003–2013) and recent (2013–2022) remote‐sensing data revealed a decline in the occurrence of algal blooms in the Yangtze River Estuary in the East China Sea (Liu et al. 2022). Conversely, a marked increase was observed in the northern Yellow Sea near Jiangsu and in the Bohai Sea near Tianjin and Liaoning (Peng et al. 2022; Wang et al. 2020). These trends may be driven not only by long‐term climate change, such as rising SSTs, but also by shifts in land‐based nutrient inputs in certain regions.

4.2. Spatiotemporal Pattern of Ecological–Social Risk

The results indicate divergent trends in the future risk distributions of different species. For instance, the risks associated with P. lima and K. mikimotoi are projected to increase in most areas, whereas G. catenatum is expected to nearly disappear in the South China Sea. These results are partially consistent with the projections of Glibert et al. (2014), who predicted an expansion trend for Prorocentrum and Karenia species in Northeast Asia (including China), primarily driven by increasing temperature and shifts in nutrient stoichiometry. The contrasting trends of different species likely stem from differences in the adaptability of specific dinoflagellates to local environmental changes, particularly their physiological thresholds (Aquino‐Cruz et al. 2018; Bravo et al. 2010; Rial et al. 2023; Zhang et al. 2024). For example, P. lima and K. mikimotoi exhibit significant thermal tolerance, with the ability to grow at temperatures up to 30°C (Aquino‐Cruz et al. 2018; Zhang et al. 2024). In contrast, G. catenatum , which is typically found in cooler waters, has an optimal growth range of 18°C–23°C (Bravo et al. 2010). Climate projections indicate that by the end of the 21st century, SSTs in China's coastal waters will rise by 2°C–4°C (Legg 2021). This warming trend is expected to create more favorable conditions for P. lima and K. mikimotoi while potentially exceeding the physiological tolerance of G. catenatum in some regions, leading to population declines or even local extinction. Similar ecological thresholds also apply to other environmental factors, such as salinity and dissolved oxygen, further influencing the distribution and survival of these species (Bravo et al. 2010; Trainer et al. 2020).

In terms of overall risk, this study projects a future decline in the East and South China Seas. Supporting this perspective, a 20‐year global analysis found that while the spatial extent and frequency of algal blooms have increased globally, bloom intensity has weakened in tropical and subtropical regions of the Northern Hemisphere under climate change (Dai et al. 2023). In addition to the impact of climate, risk assessments based on data prior to 2010 indicated that dinoflagellate HABs were closely linked to nutrient concentrations and nutrient stoichiometry in China's coastal waters (Wang, Bouwman, Wang, et al. 2023; Wang, Bouwman, Van Gils, et al. 2023). Between 1980 and 2000, dinoflagellate‐dominated HAB risk zones expanded by 127% along the eastern coast (Wang, Bouwman, Van Gils, et al. 2023). National‐scale observational data also show that the risk of harmful dinoflagellates rose from the 1990s and peaked around 2003; however, both the annual frequency and spatial coverage of HABs have declined since the 2010s (Sakamoto et al. 2021; Guan et al. 2022; Fang et al. 2017; Zeng et al. 2019). Recent evidence indicates that the previously increasing trends in DIN and DIP concentrations in China's coastal waters have reversed. From 2012 to 2020, declining nutrient concentrations and fluxes were recorded in the lower Yangtze River (Wang, Gao, Ming, and Zhao 2023; Zhong et al. 2025). A similar reversal has been observed in the Yellow and Bohai Seas, where the excess of dissolved inorganic nitrogen relative to soluble reactive phosphorus began to decrease after 2011, following a prior long‐term increase (Zheng and Zhai 2021). Model‐based projections suggest that if nutrient concentrations—especially nitrate—are reduced by half by 2050 compared to 2020 levels, dinoflagellate blooms in the East China Sea could be substantially suppressed (Zhou et al. 2022). However, other studies suggest that under SSP scenarios, high N:P ratios may persist along China's coasts through 2050, maintaining favorable conditions for dinoflagellate dominance (Wang et al. 2021). In addition, Wang, Bouwman, Van Gils, et al. (2023) argued that although nutrient‐driven eutrophication historically intensified dinoflagellate HAB risk, the system may now be approaching ecological saturation with respect to nutrients, making temperature an increasingly important driver of HAB dynamics in the future. Taken together, these findings suggest that future HAB risk will likely be shaped by the combined effects of nutrient dynamics and climate change. While climate change is generally expected to elevate the risk of harmful dinoflagellates (Gobler 2020; Griffith and Gobler 2020), our results highlight that such risk trajectories are highly region‐ and scale‐dependent.

On the other hand, the reduction in risk in China's coastal waters may have resulted from a decoupling trend between risk sources and receptors. In the mariculture industry, this decoupling manifests as reduced spatial overlap between the biogeographic distribution of harmful dinoflagellates and aquaculture areas. With climate change, harmful dinoflagellate species are expected to gradually shift away from existing aquaculture areas. China's mariculture hotspots are concentrated around the Liaodong Peninsula, the southern Shandong Peninsula (Liaoning and Shandong provinces), and Guangdong and Fujian provinces (Duan et al. 2021; Meng et al. 2024). However, climate‐driven SST increases in Guangdong and Fujian are likely to exceed the preferred ranges for some dinoflagellate species. In terms of affected human populations, decoupling is reflected in the reduced population density within the potential impact zones for harmful dinoflagellates. Population decline in China's coastal areas under the SSP scenarios is likely driven by sea level rise, natural disasters, and land subsidence (Fang et al. 2017; Cai et al. 2022; Fang et al. 2020), along with factors such as resource limitations, population policies, and restricted living spaces (UN 2024; Chen et al. 2020). However, this trend may differ in other regions such as South Asia and coastal Africa, where rising coastal populations could increase potential risks (Merkens et al. 2016).

Although the risks posed by the most harmful dinoflagellate species along the coast of China are declining, some high‐risk species are becoming increasingly prevalent. For example, the risk areas and intensities of K. mikimotoi and P. lima have been increasing. These species are bloom‐forming dinoflagellates that produce distinct toxins. K. mikimotoi releases gymnodimines, and P. lima generates DSP toxins. They are widely distributed in temperate, subtropical, and tropical coastal waters. Since 2008, they have been the dominant HAB species in China (Lu et al. 2022; Chen et al. 2019; Yan et al. 2022; Gu et al. 2022; Zhang, Lim, et al. 2020). In particular, blooms of K. mikimotoi have grown substantially over the past two decades, with major outbreaks occurring after 2003. Some blooms have spanned tens of thousands of square kilometers, causing mass fish die‐offs in aquaculture areas (Zhou and Zhu 2006; Baohong et al. 2021). A. sanguinea , first recorded in 1998 near the Shandong Peninsula, spread to all four major Chinese seas by 2017, with blooms stretching over 20° latitude and causing mass mortality of fish, invertebrates, and birds (Luo et al. 2017; Chen et al. 2019; White et al. 2014). P. lima , a benthic toxin‐producing species, is associated with widespread human poisoning. Although large‐scale blooms of P. lima have only been documented once in China's coastal waters over the past 15 years (Zhang, Su, et al. 2020), environmental shifts could potentially enable it to dominate future blooms, similar to K. mikimotoi. Studies have indicated that climate change, characterized by rising temperatures, altered coastal benthic habitats, and increasing ocean acidity, may enhance both the growth rates and toxin production in P. lima (Aquino‐Cruz et al. 2018). Moreover, a shift in the dominant HAB species has been observed along the coast of China in recent decades (Gu et al. 2022; Hou et al. 2022; Wu et al. 2013; Wang, Bouwman, Wang, et al. 2023). Thus, future risk management may require a differentiated approach that focuses on monitoring and early‐warning efforts for these high‐risk species. These include the development of early‐warning systems that monitor key environmental indicators such as SST, nutrient concentrations (particularly nitrogen and phosphorus), and water column stratification to detect bloom‐favorable conditions; the implementation of habitat restoration measures such as reducing land‐based nutrient inputs and constructing ecological buffer zones to mitigate eutrophication; and the optimization of aquaculture zoning and management practices to avoid high‐risk areas and minimize the exposure of sensitive production systems to HAB events. Successful examples from Japan and South Korea demonstrate that integrated approaches combining pollution control and spatial planning can effectively reduce HAB risks (Guan et al. 2022; Imai et al. 2021; Kim 2010).

4.3. Limitations and Uncertainties

Various methods are available for studying harmful dinoflagellates, analyzing their spatial distribution, and predicting their associated risks. For example, approaches based on satellite remote‐sensing data and algorithm improvements are commonly used to monitor the spatiotemporal distribution of HABs (Lekki et al. 2019; Gokul et al. 2019; Fernandes‐Salvador et al. 2021). Additionally, both numerical and statistical modeling methods have been employed (Glibert et al. 2014; Pei et al. 2022; Hu et al. 2024; Goncharenko et al. 2021). The methodology proposed in this study not only quantifies the spatiotemporal distribution of harmful dinoflagellates but also integrates regional socioeconomic factors to predict ecological and social risks in terms of both spatial extent and risk intensity, which is more comprehensive than previous methods. However, this study has some limitations. One key limitation is that the risk assessment was based on the distribution of harmful dinoflagellates rather than on the specific locations of HABs. Although the distribution of suitable habitats for harmful dinoflagellates closely aligned with the observed HAB locations, they were not identical. Algal bloom events rely on specific environmental conditions at certain times (Griffith and Gobler 2020), such as wind speed, temperature, and nutrient concentrations, which are influenced by episodic factors (Gu et al. 2022; Kang et al. 2025). Therefore, the distribution of suitable habitats does not fully equate to the distribution of algal blooms; this distinction should be considered when interpreting the results of the present study. Another limitation is that the selection of species in this study was constrained by the availability and quality of occurrence data. While this allows for robust modeling, it may lead to an underestimation of overall risk, particularly in regions where emerging or less‐studied species—currently lacking adequate records or facing taxonomic uncertainties—could become significant risk contributors under future environmental change. Similar challenges may also arise from the potential introduction of harmful nonnative species. Additional species may be introduced from other regions in the future through either natural or human‐mediated pathways. For example, ballast water transport is a likely pathway for the introduction of harmful marine microalgae, with nearly all known HAB species found alive in ballast waters (Hallegraeff 2015). This limitation highlights the need for continued efforts in species monitoring, data sharing, and taxonomic refinement to improve the comprehensiveness of future risk assessments.

Our projections are also subject to several uncertainties, particularly regarding factors that were not explicitly included in the modeling framework. One such uncertainty is the potential for climate change to alter toxin expression in harmful dinoflagellates. Climate change may influence toxin production (Brandenburg et al. 2019; Glibert 2024b); for example, increased CO₂ concentrations and nutrient limitations have been shown to enhance toxin levels in certain species, such as K. veneficum and Pseudo‐nitzschia fraudulenta (Fu et al. 2010; Tatters et al. 2012; Glibert 2024b). However, the extent and direction of these changes remain uncertain due to species‐specific and region‐dependent responses (Lima et al. 2022; Bui et al. 2024). Furthermore, most studies on toxin variability have been conducted under controlled laboratory conditions, limiting their applicability to real‐world ecosystems (Laabir et al. 2013; Van de Waal et al. 2014). Therefore, our study primarily focuses on species' spatial distribution dynamics and known toxin differences while acknowledging these uncertainties. Another source of uncertainty lies in the assumption that the locations and production of coastal aquaculture remain relatively stable. However, the analysis does not account for potential shifts in industrial patterns driven by global environmental change or local policy interventions, which may influence the distribution of risk receptors. Coastal aquaculture industries in regions such as China and other Asian coastal areas are highly vulnerable (Handisyde et al. 2017; Froehlich et al. 2022), shaped by natural and social factors. The major drivers of China's mariculture, such as spatial sea‐use policies, marine environmental governance, and local development policies (Yu et al. 2020), were not included in this study. In addition to affecting the assessment results of risk receptors, changes in human population density and the aquaculture industry can indirectly influence nutrient levels in coastal waters, altering the conditions that affect dinoflagellate distribution (He et al. 2021; Chen, Liang, et al. 2023; Hou et al. 2022; Xiao et al. 2019). For example, research has indicated a strong link between aquaculture size, river pollutant discharge, and algal bloom outbreaks (Zeng et al. 2024). Given the complexity of the factors influencing harmful dinoflagellates, future research will benefit from incorporating a broader array of natural and anthropogenic variables, such as nutrient stoichiometry (e.g., N:P and DIN:DIP ratios), organic nutrient forms, water column stability, residence time, riverine nutrient inputs, and coastal human activities, including agricultural/aquaculture intensity and spatial layout, all of which have been shown to influence HAB occurrence and associated risks (Glibert et al. 2014; Wang, Bouwman, Van Gils, et al. 2023; Zahir et al. 2024).

5. Conclusions

This study developed a novel framework for assessing and projecting the spatiotemporal ecological and social risks of harmful dinoflagellates under both current and future climate scenarios. A key innovation of our approach lies in the integration of regional socioeconomic receptors—specifically population distribution and aquaculture activity—into ecological risk assessments. By linking projected habitat shifts of 14 harmful dinoflagellate species with spatial data on human and economic assets, we offer a more comprehensive understanding of not only where species are likely to expand or contract, but also where these changes may translate into actual risk. Our study provides reliable distribution maps for 14 harmful species under two climate scenarios, illustrating clear latitudinal shifts in habitat patterns. The total suitable habitat for these species is expected to decline from 1.17 million km² to between 0.69 and 1.14 million km² by 2100. Risks in the southern coastal areas are projected to decrease, while risk areas and intensities in the northern coastal regions are expected to remain stable or increase, particularly in the Yellow and Bohai Seas. Notably, species such as P. lima and K. mikimotoi are expected to expand their risk ranges as well as increase their risk intensities. Liaoning was identified as the priority province for monitoring and risk management. Our study suggests that, although climate change is generally expected to exacerbate the impacts of harmful algae on coastal systems, these impacts may follow different trends in specific regions, particularly in tropical and subtropical areas. This finding highlights the importance of localized adaptation when assessing risks. In conclusion, this study contributes a scalable, integrative framework for understanding and managing the risk posed by harmful dinoflagellates in a changing climate. Importantly, the framework is transferable to other coastal regions globally, provided that species occurrence, environmental, and socioeconomic data are available, offering a valuable tool for global‐scale assessments of harmful dinoflagellates' risks under climate change.

Author Contributions

Shangke Su: data curation, formal analysis, methodology, visualization, writing – original draft, writing – review and editing. Zhaohe Luo: data curation, formal analysis, funding acquisition, investigation, project administration, writing – review and editing. Jianhua Kang: methodology, supervision, writing – review and editing. Xinyuan Guo: data curation, formal analysis, writing – original draft. Changyou Wang: data curation, investigation. Rui Jin: writing – original draft. Jianguo Du: funding acquisition, resources. Xinqing Zheng: funding acquisition, investigation. Kieng Soon Hii: data curation, investigation. Shifeng Fu: data curation, investigation. Wenjia Hu: conceptualization, funding acquisition, methodology, supervision, validation, writing – original draft, writing – review and editing. Bin Chen: conceptualization, project administration, supervision, writing – review and editing.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Data S1.

Su, S. , Luo Z., Kang J., et al. 2025. “How Does Climate Change Influence the Regional Ecological–Social Risks of Harmful Dinoflagellates? A Predictive Study of China's Coastal Waters.” Global Change Biology 31, no. 7: e70323. 10.1111/gcb.70323.

Data Availability Statement

The data that support the findings of this study and all appendices are openly available in Figshare at https://doi.org/10.6084/m9.figshare.27719091. Species occurrence data were obtained from the Ocean Biogeographic Information System (OBIS) at https://obis.org/ and the Global Biodiversity Information Facility (GBIF) at https://doi.org/10.15468/dl.f4mxxc, https://doi.org/10.15468/dl.nv6fjb, https://doi.org/10.15468/dl.ufxr7g, https://doi.org/10.15468/dl.7r67aw, https://doi.org/10.15468/dl.ny6445, https://doi.org/10.15468/dl.vtyj95, https://doi.org/10.15468/dl.sw2zk4, https://doi.org/10.15468/dl.2y3h82, https://doi.org/10.15468/dl.y8wd48, https://doi.org/10.15468/dl.3wua4j, https://doi.org/10.15468/dl.dqs6q2, https://doi.org/10.15468/dl.r9hrr3, https://doi.org/10.15468/dl.n8q57a, and https://doi.org/10.15468/dl.pvz8wa. The environmental data for habitat suitability modelling are sourced from Bio‐ORACLE version 3.0 at https://www.bio‐oracle.org/downloads‐to‐email.php and from Global Fishing Watch Inc. at https://www.globalfishingwatch.org. National and provincial population and economy projection data are available via the Science Data Bank at https://doi.org/10.57760/sciencedb.01683.