1 INTRODUCTION

Harmful algal blooms (HABs) have been a recurring problem in the recent decades in coastal waters around the world (Hallegraeff, 1993; Anderson et al., 2008). Generally, HABs-causative species can be mainly divided into two groups, the toxin producers, which contaminate seafood or kill fish, and the high-biomass producers, causing hypoxia or anoxia after reaching dense concentrations leading to indiscriminate mortalities in marine life. Meanwhile, some HABs species could be both (Furuya et al., 2010). Thus, HABs generally pose severe impacts on fisheries, marine resources, mariculture, public health and the diversity of the coastal ecosystems.

The South China Sea (SCS) is the largest semienclosed marginal sea in Southeast Asia. It is situated along the tropical and subtropical regions, including the waters around Guangdong, Guangxi and Hainan. Based on the annual reports from the State Oceanic Administration, SCS Branch of China, the coastal areas of Guangdong Province is the area most seriously affected by HABs (SOA, 2016). This has significant implications since Guangdong is also one of the main aquaculture provinces in China, with its mariculture production ranking third in the country (Fisheries Bureau of Ministry of Agriculture, 2017). Therefore, understanding the characteristics of historical HAB events along the coast of Guangdong Province is of particular significance.

Characterization of historical HAB events and the factors causing their formation have already been previously carried out in Guangdong coastal areas (Qian et al., 2000; Li and Lv, 2009). For example, Qi et al. (2004) described the causative organisms reported to cause blooms in the coastal waters of Guangdong and Hongkong in 1998, and speculated on the possible causes of such blooms. Other studies have focused on descriptions of the algal species (e.g. Qi et al., 1991; Chen et al., 1999), bloom dynamics and their correlation with certain environmental factors (e.g. Qi et al., 1993; Xu et al., 2003; Ou et al., 2010; Shen et al., 2012). In this study, we analyzed the spatial and temporal dynamics of HABs occurrences from 1980 to 2016 using geostatistical tools within GIS, and illustrated the distribution of toxic species and their negative effects, thus, providing a framework for urgent measures and baseline data for policy making for HABs management.

2 MATERIAL AND METHOD 2.1 Study area

Guangdong Province (GD), adjacent to the South China Sea, has the longest coastline of 3.37×10⁶ m in China. Our study area was divided into 5 water zones, including the Daya Bay, Zhujiang River estuary, Mirs Bay, and the Western and Eastern Coasts of Guangdong (Fig. 1). Daya Bay is one of the biggest inlets and an important cultural area in Guangdong Province, while the western and northeastern parts of the bay are surrounded by highly urbanized Shenzhen and Huizhou, respectively. The Zhujiang River estuary on the other hand is located midway along the northern boundary of the South China Sea. The coast has a NE-SW orientation and the adjacent shelf is 150-250 km wide. Its watershed covers some of the most highly urbanized and industrialized regions in China. Mirs Bay is a semi-enclosed bay, embedded into the terrene about 18 km north-northwestwards and neighbored by the Zhujiang River estuary westward and the Daya Bay eastward. Lastly, the eastern Guangdong coastal regions include coastal waters of Shantou, Shanwei, and Jieyang. The western Guangdong coastal regions include coastal waters of Zhanjiang, Maoming and Yangjiang.

2.2 Data resources and management

In this study, HABs occurrence records were retrieved from published scientific papers (Zhang et al., 2002; Leng and Jiang, 2004; Deng et al., 2006; Tong, 2006), Bulletin of marine disaster of China (SOA, 2007-2008), Bulletin of Marine Environmental Quality of China (SOA, 1991-1993, 1998, 2001-2003), Bulletin of Marine Environmental Quality of Guang-dong Province (SOA, 2002-2003, 2006-2008, 2010-2017), as well as Bulletin of Marine Environmental Quality of Shenzhen (SOA, 2001-2002, 2005, 2007-2009), Maoming (SOA, 2009), Zhuhai (SOA, 2006) and Shantou city (SOA, 2008-2009) during the period of 1980-2016. It should be noted that some HAB events might have been missed and not reported in the early years, while some HABs were recorded but without any specified causative species. Some other HABs did not have records for the study area.

We used ArcGIS technology to process the records, with each HAB event represented by a central point rather than a polygon, which was typically the shape of a single HAB event. This point pattern analysis can be found elsewhere (Wang and Wu, 2009; Wu et al., 2013; Song et al., 2016; Yuan et al., 2017).

From the collected HABs cases, we have made comparisons for the time of occurrence, location and causative species in four periods, namely 1980-1989, 1990-1999, 2000-2009 and 2010-2016.

2.3 Frequency analysis

To identify the most frequent HAB occurrence areas in the Guangdong coasts, a kernel density (KDE) estimation was used. KDE is a method to generate a map that shows the density of the events modeled as continuous field by taking the value of a specific point and spreading it across a predefined area (Gatrell et al., 1996; Wang and Wu, 2009). In KDE, a kernel, namely a circle with a predefined constant radius, is moved across the study area. The weight of each event point is up to its distance from the center of the circle. A point near the center is highly weighted and vice versa (Spencer and Angeles, 2007). Accordingly, the density of points is estimated. To get a smoothened map, different search radius values were tested and a value of 0.5° was finally selected.

3 RESULT AND DISCUSSION 3.1 Spatial distribution

There were a total of 337 HAB events recorded in Guangdong coasts from 1980 to 2016, with 33 occurring in western coast of Guangdong, 46 in the eastern coast, 59 in Daya Bay, 112 in Mirs Bay, and 87 in Zhujiang River estuary. HABs in the study area were usually localized in shallow waters and rarely occurred in open waters (Fig. 2a, b). Furthermore, a very few large-scale HAB events with area greater than 1 000 km² were recorded, nearly half of the HAB events had an affected area of less than 50 km² (Fig. 2a). It should be noted that 37.5% of the HAB events were recorded with no specified occurrence area.

The KDE result showed that the highest density of HABs was observed in Mirs Bay throughout the study period, suggesting that HABs occurred most frequently in this area. However, the waters in Mirs Bay has not reached up eutrophication level (Huan et al., 2016), even not high enough to support high biomass of phytoplankton blooms, indicating that there were possible mechanisms concentrating the phytoplankton during HABs formation (Yin, 2003). Wind direction and speed, which either allow vertical migration or horizontal aggregation of phytoplankton, play important roles in the formation of HABs (Yin, 2003).

A high frequency of HABs was also observed along the west coast of Daya Bay (Fig. 3). From the 1980s to 2000s, a moderate high frequency of HABs was observed in Shenzhen Bay (east coast of the Zhujiang River estuary), while in 1990s and 2010-2016, HABs was only in a moderate low frequency in the same area. For the west and east coasts of Guangdong, the point density was low and/or moderate low, indicating that these two areas were not frequently infested with HABs (Fig. 3).

3.2 Temporal distribution 3.2.1 Inter-annual variations

Figure 4 shows the yearly HABs occurrences from 1980 to 2016. For the entire study area, HABs occurrences were generally less than 10 per year in the 1980s. It however significantly increased during 1990s, especially becoming more frequent in 1990 (18), 1991 (22) and 1998 (16) (Fig. 4). After 2000, HABs occurrences were generally maintained between 6 and 16 except in 2003, when it reached to as high as 25 reports.

In Mirs Bay, HABs were observed almost every year, but most frequently in 1990 (17) and 1991 (15) (Fig. 4). HABs also occurred almost every year in the Zhujiang River estuary, showing an increasing trend during 2000-2003 (Fig. 4). In the Daya Bay, the number of HAB events was also observed have increased after 1998. The same was observed in the west and east coasts of Guangdong, where trends increased in the past four decades (Fig. 4). Before 2000, only three HAB events were observed in the west coast of Guangdong, but after 2000, HAB events significantly increased from once per year during 2000-2009 to twice per year after 2010 (Fig. 4). Since October 1997 when Phaeocystis globosa bloomed in Zhelin Bay of Shantou, a sharp increase in HABs occurrences had been observed in the east coast of Guangdong. The increasing trend in HABs occurrences in late 1990s in the eastern and western coasts of Guangdong could be associated with the simultaneous increase in aquaculture production. The excessive wastes generated from the feeds used for breeding led to increased eutrophication in the surrounding waters. The intensive mariculture activities in the area caused self-pollution, causing eutrophication and providing suitable environmental conditions for phytoplankton growth and aggregation. As recorded in Guangdong Statistical Yearbook on Agriculture, aquaculture production generated 50 621 t harvest in the eastern coast and 62 564 t in the western coast in 1992, increasing to 299 071 t and 703 185 t in 1998, and to 596 644 t and 1 585 996 t in 2015, respectively (Fig. 4).

3.2.2 Seasonal distribution

As illustrated in Fig. 5, HABs occurred year around in Guangdong coasts areas, with most observed in spring between March and May, accounting for 44.8% of the total annual HABs events. During spring, the monsoon shifts from northeast to southwest, and a rise in temperature is generally observed in Guangdong. In addition, wind velocity weakens characterized by lower air pressure at the beginning of the rainy season. These are ideal conditions for phytoplankton growth leading to more frequent HABs formation (Qi et al., 2004; Wang et al., 2008). The lowest number of occurrences was recorded in December (11), and the numbers recorded in other months were relative more even between 16 and 28 (Fig. 5).

For the different areas, in the western Guangdong coast area, March to May had the highest number of HABs records. In Mirs Bay, the frequency of HABs peaked during the warmer months of March (26) and April (22). In Daya Bay, majority (87.9%) of HABs occurred between March and September, while in Zhujiang River estuary, the highest number was recorded in April (25) with no HABs in the colder months of September and December. In contrast, November had the highest number of HABs records in eastern Guangdong coastal area, while in other months the number was relatively more even (Fig. 5).

3.3 Causative HAB species

From 1980 to 2016, HAB causative species documented in the Guangdong coastal waters were diverse. Prior to 1990s, HABs were primarily caused by dinoflagellates and diatoms. From 1990, some previously unobserved HAB species such as a raphidophyte (Chattonella marina), prymnesiophyte (Phaeocystis globosa), ciliate (Mesodinium rubrum) were observed. The dominant HAB species were dinoflagellates throughout the study period. However, the percentage of the dinoflagellate blooms was the lowest during 2000s when the Raphidophyceae and Haptophyceae started to appear in higher numbers.

3.3.1 Dominant HAB species

The most frequently occurring species in descending order were Noctiluca scintUlans, Phaeocystis globosa, Skeletonema costatum and Scrippsiella trochoidea. N. scintillans was responsible for blooms on 75 occasions, accounting for 20% of the total HABs events. Nearly all N. scintillans blooms occurred during spring and winter with the highest frequency during April and were almost absent during June to September, with nearly 70% recorded in Mirs Bay (Figs. 6, 7). N. scintillans is a cosmopolitan species and found in almost all of the world's oceans. A progressively stable and muggy weather without heavy rain favor formation of N. scintillans blooms, which could partially explain this species was not observed forming dense blooms in June to September when the surface water temperature was relatively high accompanied by heavy rain (Huang et al., 1997; Mohanty et al., 2007). Its frequent occurrence in Mirs Bay could be associated with the mechanical convergence due to its topographical and tidal features (Huang et al., 1997).

Phaeocystis globosa, reported to cause 38 HAB events occurring every month and peaking in November, with more than half occurring in Eastern Guangdong coastal waters (Figs. 6, 7). The species was originally identified to be Phaeocystis pouchetii. However, DNA sequencing confirmed that it was actually P. globosa, and that it was endemic, being different from those observed in the European and American waters (Medlin et al., 1994; Qi et al., 2004). In the coastal waters of Guangdong, it spread from the west to the east and has been reported to prefer higher temperature and higher salinity (Xu et al., 2003, 2017). The annual average temperature in the coastal waters of Guangdong Province is around 21℃, which is suitable for the growth of P. globosa. In addition, the SCS strain has a unique adaption to extremely low- or high-light environments (Xu et al., 2017), thus, allowing it to frequently form blooms during different seasons in the study area.

Blooms of S. costatum occurred from March to October, January and peaked in May (Fig. 6). This species dominated the HABs in western Guangdong coast and Zhujiang River estuary (Fig. 7). S. trochoidea bloomed predominantly during July to September (Fig. 6), mainly distributed in Daya Bay (Fig. 7). The S. trochoidea blooms were initiated by the germination of cysts from the sediment surface of Daya Bay (Xiao et al., 2001, 2003).

3.3.2 Toxic HAB species

Some HAB species were reported to exert harm by the accumulation of their toxins through the food chain, culminating in the illness or death of humans and other marine life. Among them, dinoflagellates Gonyaulax spinifera, Procentrum lima, Alexandrium sp. and Gymnodinium sp. and diatom Pseudonitzschia pungens have been reported to cause red tide along the Guangdong coasts. However, it has rarely been reported to have any direct negative impacts and is not known to have rendered shellfish or other feeders to be toxic.

Ichthyotoxic blooms are more common in the Guangdong coasts and have been reported to result in catastrophic economic losses to the fisheries and mariculture industry. Taxa known to produce Ichthyotoxins mainly include species of the raphidophycean flagellates Heterosigma akashiwo and Chattonella marina, dinoflagellates Karenia mifomotoi, and prymnesiophyte Phaeocystis globosa.

As discussed previously, Phaeocystis globosa was one of the most frequent HAB forming species along coast of Guangdong. However, blooms of this species associated with mortality of fish as well as severe economic losses were only recorded in Eastern coast of Guangdong. In 1997, major fish kills of Scombermorus guttatus, Pampus chinensis, Lutjianus chrysotaenia and Pagrosomus major occurred in Zhelin Bay in Raoping, resulting in severe economic losses estimated at 65 million Chinese Yuan (Qi et al., 2004). In December 2002, a bloom resulted in apparent reduction in production of elvers, some losses on Porphyra tenera and oyster. Another major event occurred in November 2003, causing a significant decline in productions of cultured Porphyra tenera and shellfish. Aquaculture losses were also reported in July 2007 in Shantou, where the dead fish and crab were very visible. In September 2007 in Shanwei, the bloom resulted in an economic loss of 1 million Chinese Yuan. The Chinese strains of Phaeocystis globosa have been confirmed to have hemolytic properties (Shen et al., 2005; Peng et al., 2005; Wang et al., 2010), and the components of the hemolytic compounds extracted from filtered P. globosa cells have been identified as glycolipids, with 1'-Oheptadecadienoyl-3'-O-(6-O-α-D-galactopyranosyl-β-D-galactopyranosyl)-glycerol as the major constituent (He et al., 1999).

Blooms of Karenia mikimotoi, initially described as Gymnodinium mikimotoi, have also been reported in many regions of the world, including the eastern North Atlantic, Japan, Europe, Australia, South America, North Africa and China (Landsberg, 2002; Long and Du, 2005; Gentien et al., 2007; Yao et al., 2007; Zhang et al., 2009). The occurrence of K. mikimotoi in China was first recorded in the coastal waters of Xiamen (Zhang et al., 1988). Its large-scale bloom was then reported along the Guangdong coast from Huidong (Daya Bay) to Yangjiang (western coast of Guangdong) in March to April 1998. This bloom was responsible for mortalities of several economically important species such as Seriola sp., Pagrosomus major and Epinephelus epistictus (Qi et al., 2004) and resulted in a 0.35 billion Chinese Yuan economic loss to fisheries. After 2010, K. mikimotoi blooms sporadically appeared in Mirsbay, Shanwei and Zhanjiang, but with no reports of fish mortalities.

It was demonstrated that K. mikimotoi produces several toxic agents, such as hemolytic toxins (Arzul et al., 1994; Mooney et al., 2007; Zou et al., 2010), cytotoxic polyethers (Satake et al., 2002), and reactive oxygen species (ROS) (Yamasaki et al., 2004). However, the fish and invertebrates mortalities associated with this species were thought to be mainly caused by hemolytic substances. Wang et al. (2001) observed histopathological changes in the gill of intoxicated fish exposed to K. mikimotoi blooms in the coastal waters of South China Sea in 1998. The authors noted severe damage to the gill, including branchial lobule epithelial hyperplasia, accretion of adjacent branchial leaflets, exfoliation of epithelial cells, and angiorrhexis (Wang et al., 2001; Yang et al., 2011).

Furthermore, there were 13 HABs events caused by Chatonella marina in Guangdong coasts, one recorded in Shanwei in 1998, four in Mirs Bay and the others in Daya Bay. The first report of C. marina bloom in China was in March 1991 in Mirs Bay (Qi et al., 1991), which inflicted an economic loss amounting to 1.5 million Chinese Yuan in aquaculture and death of hundreds of thousands of fish seedlings. In Daya Bay, two blooms caused by this species were also associated with economic losses.

The underlying mechanisms causing toxicity in Chattonella remain controversial. Previous studies have shown that the toxin of C. marina contained putative brevetoxins, reactive oxygen species such as hydrogen peroxide (Oda et al., 1994), free fatty acids EPA (Marshall et al., 2003) and light-dependent hemolytic cytotoxin (Kuroda et al., 2005). However, results from Shen et al. (2010) showed that the previously reported ichthyotoxins from C. marina were unlikely the principal toxic molecules, and that traditional organic extraction methods could not extract and preserve majority of C. marina toxins. Alternative approaches for protein or small molecule compounds have been suggested to be adopted or developed (Shen et al., 2010).

Another raphidophycean flagellate, Heterosigma akashiwo, has been reported to cause massive fish kills in America and Japan (Khan et al., 1997; Twiner et al., 2001; Lewitus et al., 2012). However, although it was reported to cause blooms in Daya Bay and Mirs Bay, fish kills have not been observed with blooms of this species.

4 CONCLUSION

In summary, the present study used GIS approach to analyze the spatial and temporal distributions of HABs as well as the causative species in the coastal waters of Guangdong Province 1980-2016. Results revealed that the areas with frequent HABs were Mirs Bay and the west coast of Daya Bay.

Temporally, after 2000, there was an increasing trend of HAB events observed in the study area except for Mirs Bay. The sharp increase of aquaculture production appeared to play an important role in accounting for the increasing trend of HABs. Meanwhile, different study region shows distinct seasonality, with blooms frequently observed in spring for the western Guangdong coast area, March and April for Mirs Bay, March to September for Daya Bay, April for Zhujiang River estuary, and November for the eastern Guangdong coast area.

The dominant HAB species were Noctiluca scintillans, Phaeocystis globosa, Skeletonema costatum and Scrippsiella trochoidea. Most N. scintillans blooms were observed in Mirs Bay, mostly occurring during spring and winter. Meanwhile, P. globosa occurred year round, with more than half in Eastern Guangdong coastal waters. Blooms of S. costatum occurred during March to October and January, appearing to be widely distributed in western Guangdong coast and Zhujiang River estuary. S. trochoidea bloomed predominantly from July to September in Daya Bay.

Ichthyotoxic blooms are still common in the Guangdong coasts and are associated with tremendous economic losses to the fisheries and mariculture industry. Phaeocystis globosa, Chattonella marina, Karenia mikimotoi and Heterosigma akashiwo are the important causative species.

5 DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.