On the recurrent Ulva prolifera blooms in the Yellow Sea and East China Sea

Authors


Abstract

[1] A massive bloom of the green macroalgae Ulva prolifera (previously known as Enteromorpha prolifera) occurred in June 2008 in the Yellow Sea (YS), resulting in perhaps the largest “green tide” event in history. Using a novel index (Floating Algae Index) and multiresolution remote sensing data from MODIS and Landsat, we show that U. prolifera patches appeared nearly every year between April and July 2000–2009 in the YS and/or East China Sea (ECS), which all originated from the nearshore Subei Bank. A finite volume numerical circulation model, driven by realistic forcing and boundary conditions, confirmed this finding. Analysis of meteorological/environmental data and information related to local aquaculture activities strongly supports the hypothesis that the recurrent U. prolifera in the YS and ECS resulted from aquaculture of the seaweed Porphyra yezoensis (or nori) conducted along the 200 km shoreline of the Subei Bank north of the Changjiang (Yangtze) River mouth. Given the continuous growth in aquaculture efforts in the region, similar macroalgae bloom events, such as the summer 2008 event, are likely to occur in the future, particularly between May and July. This was confirmed by the 2009 bloom event in the same regions and the same period. The profit of the local P. yezoensis aquaculture industry (∼16,000 Ha in 2007) is estimated as U.S. $53 million, yet the cost to manage the impact of the summer 2008 U. prolifera bloom exceeded U.S. $100 million. Therefore, better strategies are required to balance the economic benefit of seaweed aquaculture and the costs of environmental impacts.

1. Introduction

[2] Coastal eutrophication is a serious environmental problem caused by excessive nutrients and other pollutants derived from agriculture, urbanization, and industries. In the Gulf of California, recurrent phytoplankton blooms are linked to agriculture irrigation and runoff [Beman et al., 2005]. On the west Florida shelf, the frequency of toxic Karenia brevis blooms appears to have increased over 10 fold between the 1950s and the 2000s, in part due to increased nutrient inputs from coastal runoff [Brand and Compton, 2007]. In Chinese coastal waters of the Yellow Sea (YS), East China Sea (ECS), and the Bohai Sea, the number and size of toxic algae blooms also have increased since 1998 [Zhou and Zhu, 2006].

[3] Green macroalgae blooms have also been reported in the world's oceans [e.g., Fletcher, 1996; Blomster et al., 2002; Nelson et al., 2003; Merceron et al., 2007]. These are typically small blooms restricted to coastal areas. However, between May and July 2008, an extensive bloom of the green macroalgae Ulva prolifera (previously known as Enteromorpha prolifera [see Hayden et al., 2003]) occurred in coastal and offshore waters of the YS (Figure 1a). By late June 2008 this bloom had covered 2400 km2 of water as estimated from satellite imagery collected with the NASA Moderate Resolution Imaging Spectroradiometer (MODIS), and thought to be one of the largest in recorded history (see section 3 and Table 1 [Hu and He, 2008]). Various methods were employed to maintain an algae-free water area near Qingdao for the Olympic sailing games, at a cost exceeding U.S. $100 million (Text S1) [Wang et al., 2009].

Figure 1.

(a) Approximate location and distribution of U. prolifera identified from MODIS FAI imagery between April 2000 and May 2009. The background MODIS RGB image on 5 April 2003 shows the extensive sediment plume from the Subei Shallow Bank to the ECS. Nearly all U. prolifera algae slicks in the ECS were found in the downstream portion of this plume, which occurs every year between fall and spring following cross-shelf currents between the Subei Bank and the ECS [Yuan et al., 2008]. (b and c) MODIS FAI images tracing U. prolifera blooms in 1 × 1° areas in the YS and ECS on 31 May 2008 and 17 July 2008, respectively. More examples of the full-resolution MODIS FAI imagery showing the algae slicks, including those after May 2009, are provided in the auxiliary material. Algae coverage statistics are provided in Table 1.

Table 1. Ten Year Record of the Green Macroalgae U. prolifera in the Three Adjacent Regions, as Revealed by MODIS and Landsat Imagerya
 2000200120022003200420052006200720082009
  • a

    The 10 year record is from April 2000 to August 2009. Here “1” means U. prolifera appears in at least one satellite image, while “0” means no U. prolifera is found in any available satellite images. The three regions are defined as follows: Region 1 is the Subei Bank (sediment-rich waters north of the Changjiang River mouth; Figure 1a), Region 2 is the Yellow Seawaters north of the Subei Bank, and Region 3 is the East China Sea. Most information for Region 1 was derived from Landsat, and for Region 2 and Region 3 most information was derived from MODIS. Note that of the 10 years examined, only 2006 showed no U. prolifera for any of the three regions. The maximum algae area for the entire study region estimated using a linear unmixing model with MODIS FAI data is also shown. These crude estimates without ground-truth validation may contain ±20% uncertainties. Assuming 1 kg algae m−2 (wet weight), 1.9 and 1.6 million metric tons of algae biomass were estimated for 2008 and 2009, respectively.

Region 11100100111
Region 20000000111
Region 31111110011
Algae area (km2)<101557<10<10<1002119401600

[4] Using MODIS observations and other means, several studies have examined the summer 2008 bloom and possible causes, including the speculation that there was a high rate of nutrient delivery to coastal waters, and the linkage with seaweed aquaculture [Hu and He, 2008; Li et al., 2008; Liang et al., 2008; Lü and Qiao, 2008; Qiao et al., 2008; Sun et al., 2008; Liu et al., 2009]. However, the 250 m resolution MODIS data could not be used to trace the bloom initiation when the U. prolifera patches are much smaller, and the method to detect U. prolifera suffered from the interference of changing environmental and observing conditions (aerosol type and thickness, solar/viewing geometry). Also, there is great interest in understanding whether this event was restricted to the YS only, whether it was the result of local eutrophication similar to the recurrent phytoplankton blooms in coastal waters worldwide [Beman et al., 2005; Zhou and Zhu, 2006; Brand and Compton, 2007], and whether it was a historically recurring problem that might continue in the future.

[5] Here, by combining multiresolution remote sensing data from Landsat (30 m) and MODIS (250 m), numerical circulation modeling with realistic forcing, meteorological/environmental data, and information related to local aquaculture activities, we examined all macroalgae events in the YS and ECS between 2000 and 2009 to address the questions posed above on of algae bloom origin, occurrence frequency, and potential causes.

2. Data and Methods

[6] MODIS and Landsat data from 2000 to date were obtained from the U.S. NASA and USGS, respectively, and processed to generate RGB and Floating Algae Index (FAI) imagery using the methods described by Hu [2009]. The index was defined as the reflectance difference between a near-infrared wavelength and a baseline. Specifically, Rayleigh-corrected reflectance (Rrc) was first obtained

equation image

where λ is the wavelength, Lt* is the calibrated sensor radiance after correction for gaseous absorption, F0 is the extraterrestrial solar irradiance, θ0 is the solar zenith angle, θ is the sensor (viewing) zenith angle, Δϕ is the relative azimuth, and Rr is the reflectance due to Rayleigh (molecular) scattering. For simplicity, the dependence of Rrc on the solar-viewing geometry (θ0, θ, Δϕ) will be omitted. FAI was then derived as

equation image

where Rrc,NIR′ is the baseline reflectance in the NIR band derived from a linear interpolation between the red and short-wave IR (SWIR) bands. For MODIS, λRED = 645 nm, λNIR = 859 nm, λSWIR = 1240 nm. For Landsat, λRED = 660 nm, λNIR = 825 nm, λSWIR = 1650 nm.

[7] Model simulations and image comparisons showed that the index was more effective in detecting floating macroalgae than other methods primarily because most of the aerosol effects are removed by the baseline subtraction, and FAI is less sensitive to changes in solar/viewing geometry [Hu, 2009]. Floating macroalgae appears as outstanding slicks or patches over near-homogenous water background on the FAI imagery, which can be easily delineated. Further, because of the linear design (equation (2)), FAI is more suitable than other nonlinear indexes (e.g., Normalized Difference Vegetation Index, NDVI) when used for algae area estimates, where pixels with partial algae coverage can be estimated with linear unmixing [e.g., Hu et al., 2010]. For any cloud-free ocean pixel, FAI ≥ 0.02 was assumed to correspond to 100% algae coverage while FAI ≈ 0.0 was assumed to represent 0% algae coverage [Hu et al., 2010]. The lower bound threshold actually varied between −0.001 and 0.001 after calibration through trial and error with careful visual inspection.

[8] The FAI method was implemented to derive MODIS and Landsat FAI imagery between 2000 and 2009. Due to frequent cloud cover, of the 4580 MODIS scenes (or 5 min image granules) covering the YS and ECS (28°N – 37°N and 119°E – 127°E, Figure 1a) between April and July 2000–2008 and between April and May 2009, only ∼400 images containing cloud cover were processed and analyzed. Similarly, about 140 Landsat scenes covering coastal waters of the Subei Bank and near the Changjiang (Yangtze) River mouth (Figure 2) for the same periods were processed and analyzed.

Figure 2.

Approximate location and distribution of U. prolifera identified from Landsat FAI imagery between April 2000 and August 2008. The coverage of the 140 Landsat scenes is outlined by white boxes. The background MODIS RGB image on 23 May 2004 shows the recurrent sediment plume from the Subei Bank to the ECS. The approximate locations of coastal P. yezoensis aquaculture are annotated in purple. The inset shows a Landsat FAI image (17 June 2007) showing U. prolifera algae slicks within 2 km of the Sheyang River mouth. More examples of the full-resolution Landsat FAI imagery showing the algae slicks are provided in the auxiliary material.

[9] A particle tracking experiment was designed and implemented using a high-resolution, unstructured grid ECS Finite-Volume Coastal Ocean Model [Chen et al., 2008]. The model was driven by realistic wind stress and heat flux at the surface, freshwater discharges from the Changjiang River, and eight major astronomical tidal constituents (M2, S2, N2, K2, K1, O1, P1, Q1) on open boundaries. The Kuroshio and Taiwan Warm Currents were simulated by specifying the water transport on the inflow and outflow boundaries under the April climatological fields of temperature and salinity. The model was run for a period of 1 April to 31 May 2008. Two groups of particles were released on the sea surface on 17 April 2008 near the Changjiang River mouth (where and when initial U. prolifera were identified from Landsat FAI imagery) and in nearshore waters of the Subei Bank, respectively, to trace their movements through 10 May 2008 when algae slicks in the YS first appeared in Landsat FAI imagery (see section 3).

[10] The data of annual culture area and production of Porphyra yezoensis were obtained from multiple sources, including statistics from the Jiangsu Statistics Bureau and local news media (Text S2). For years when area data were not available, they were estimated using annual production data because they were approximately proportional to each other.

[11] Air temperature data were obtained from the Jiangsu Meteorological Bureau. Nutrient and pollutant discharge data of coastal rivers were from the State Oceanic Administration (SOA) of China. Pollution index data using a combination of Chinese national water quality standards (GB3097-1997) were obtained from SOA and the Ministry of Environmental Protection (MEP) of China. Because of lack of long-term nutrient concentration data in the YS, annual statistics of pollution level index was used as a surrogate for nutrient availability. Other environmental data were also obtained from various sources and analyzed. These include satellite-estimated sea surface temperature (SST), surface wind, and surface radiation.

3. Results

[12] Figure 1 shows the multiyear sequence of the U. prolifera locations observed from MODIS imagery between April 2000 and May 2009 (more recent imagery are shown in the auxiliary material). Because of the frequent cloud cover, it was difficult to construct a complete time series for every year to show bloom evolution over time. However, it is clear that the summer 2008 U. prolifera bloom event in the YS (pink color in Figure 1a, also see Figure 1b) is only an extreme case of the recurrent blooms between April 2000 and May 2009. Indeed, excluding the year of 2006, U. prolifera slicks were identified in MODIS imagery every year, mostly in the ECS rather than in the YS.

[13] Due to the space limit, the full-resolution MODIS FAI imagery and the temporal evolution of U. prolifera cannot be shown here, but some examples are provided in the Auxiliary materials (Figures S1–S8). The algae slicks identified from the MODIS imagery are typically from April to July, with the earliest on 3 April (2002) at about 370 km east of the Hangzhou Bay. In other years, the first appearance of the algae slicks ranged from mid-April to mid-June, with most slicks found in the ECS, some of which extended to at least 400 km offshore from the Changjiang River mouth. Although the algae slicks in the ECS are much thinner than those in the YS (Figure 1c versus Figure 1b), the surface water area containing the multiyear algae slicks is much higher in the ECS (about 130,000 km2) than in the YS (about 40,000 km2).

[14] The MODIS time series imagery suggests that the algae slicks in the YS followed the northward coastal current and those in the ECS followed the cross-shelf current during the spring. The effect of this latter current can be clearly visualized from the extensive sediment plume from the Subei Bank to the offshore areas in the southeast (Figure 1a [also see Yuan et al., 2008]). Algae slicks in both the YS and ECS can be traced to the very turbid and shallow Subei Bank. However, the 250 m MODIS imagery is limited in detecting small algae patches that might be present in the nearshore bank. The exact detection limit is difficult to determine without field data. However, the linear unmixing of FAI to determine partial pixel coverage may help estimate this limit. The linear unmixing model resulted in ∼2%/pixel coverage for several thin slicks that were visible in the MODIS FAI imagery, corresponding to 2% × 250 m = 5 m for the slick width. For small algae patches that do not form continuous slicks, 2%/pixel coverage corresponds to about 35 × 35 m in algae size. Thus, an algae slick <5 m in width cannot be detected by the 250 m MODIS FAI imagery.

[15] The Landsat image series at 30 m resolution extended the MODIS findings to the nearshore waters, even if their coverage was limited due to cloud cover and due to the 16 day satellite orbit cycle. Figure 2 shows the multiyear sequence of the U. prolifera locations identified from the Landsat FAI imagery. The high-resolution FAI images revealed much smaller U. prolifera slicks or patches near the coast of the Subei Bank, at one time only 2 km from the Sheyang River mouth in June 2007 (Figure 2 inset). In 2008, Landsat FAI imagery showed small algae slicks near the Changjiang River mouth on 17 and 26 April 2008, nearly a month earlier than reported previously using MODIS for the YS [Hu and He, 2008]. Because of the small size, most of the U. prolifera slicks were not observable in MODIS imagery. Although there is no ground-truth data, the thin filaments of 1–2 Landsat pixels (30–60 m) showed a spectral shape featuring a distinctive reflectance peak at 825 nm, which is characteristic of floating macroalgae. On 10 May 2008, extensive slicks of U. prolifera were found in the north of the Subei Bank. Landsat FAI images also showed recurrent slicks in the same locations in May and/or June of 2000, 2001, 2004, 2007, and 2008. Most of the slicks in the shallow bank were oriented in the N-S or NE-SW direction, following the direction of the Yellow Sea Coastal Current (YSCC, see Yuan et al. [2008]).

[16] Although a complete statistics of U. prolifera occurrence is difficult to achieve due to the frequent cloud cover, Table 1 lists whether or not the algae slicks/patches were found from the available near-cloud-free satellite images in each of the three adjacent regions: Region 1 for the Subei Bank, Region 2 for the YS excluding the Subei Bank, and Region 3 for the ECS. U. prolifera was found in Region 3 in every year except 2006 and 2007, and found in Region 1 in 6 years in the 10 year record through August 2009. For Region 2, only in 2007, 2008, and 2009 were U. prolifera found during the summer.

[17] The total algae coverage for each year was estimated as the maximum coverage during that year using one or more FAI images and the linear unmixing scheme. Results are listed in the last row of Table 1. Although the algae areas between 2000 and 2007 (except 2006) are negligible as compared with those found in 2008 and 2009, the surface water areas containing the algae slicks are extensive (Figures 1 and 2). This suggests that the algae slicks were thinner and more diluted between 2000 and 2007. Assuming a conservative 1 kg algae m−2 (wet weight), approximately 1.9 and 1.6 million metric tons of algae biomass were estimated for 2008 and 2009, respectively. These figures are significantly lower than those reported earlier [Hu and He, 2008] because of the linear unmixing method used with FAI data.

[18] In summary, U. prolifera was found from satellite imagery every year between 2000 and 2009 except 2006. The sequence of patterns observed in the MODIS and Landsat imagery (Figures 1, 2, and S1–S8 in the Auxiliary materials) suggest that the 2008 U. prolifera blooms in the YS and ECS as well as the recurrent U. prolifera slicks/patches in other years all originated in nearshore waters of the Subei Bank.

[19] The numerical circulation experiment confirmed this finding. Simulated surface particles released on 17 April 2008 in nearshore waters of the Subei Bank (red traces in Figure 3) were advected alongshore to the north, following the pattern of U. prolifera slicks seen in the Landsat FAI imagery of 10 May 2008 (red outlines in Figure 2). In the ECS, particles released near the river mouth moved offshore to the northeast and then to the east (blue traces in Figure 3). The experimental results showed that although algae patches in both locations (north of the Subei Bank in the YS and around the Changjiang River mouth) originated from the Subei Bank, the latter could not be transported to the YS, but were rather advected offshore in the ECS following the water movement. This tendency is consistent with the clockwise low-frequency circulation of the low-salinity plume originated from the Changjiang River [Chen et al., 2008]. In this region, U. prolifera was found farther offshore (as seen in the MODIS FAI image of July 2008; Figures 1a and 1c), a result of advection of the nearshore U. prolifera following the cross-shelf current. These circulation patterns also agree with the recurrent sediment plume patterns observed in typical MODIS RGB imagery collected in this region, for example on 5 April 2003 (Figure 1a) and 23 May 2004 (Figure 2). The circulation patterns are driven by the cross-shelf current from the Subei Bank to the ECS, as revealed by the MODIS image sequence of normalized water-leaving radiance [Yuan et al., 2008]. Clearly, the combination of multisensor remote sensing and high-resolution numerical circulation modeling revealed the nearshore origin for U. prolifera in nearly all years between 2000 and 2009.

Figure 3.

Trajectory of surface particles from 17 April to 10 May 2008 from a numerical model [Chen et al., 2008]. The approximate locations of coastal P. yezoensis aquaculture are annotated in purple. The first appearance of U. prolifera slicks in the north of Subei Bank on the 10 May 2008 Landsat FAI image is annotated with a star.

[20] What would cause the initial accumulation of U. prolifera in nearshore waters? Local news media reported that U. prolifera was often found to accompany the cultured seaweed in coastal waters. Therefore, we speculated that there might be a connection with seaweed aquaculture [Hu, 2009]. Liu et al. [2009] also hypothesized that aquaculture of the seaweed P. yezoensis in coastal waters of Jiangsu Province could account for the 2008 U. prolifera bloom in the YS. The independent study here strongly supports this hypothesis.

[21] Coastal aquaculture in Jiangsu Province provides 95% of the nation's P. yezoensis product. This industry has grown to occupy at least 44,000 Ha in 2008 compared to about 5700 Ha in 1999 (Figure 4 and Text S2). P. yezoensis are cultured using bamboo rafts and rope nets, which are known to be accompanied by U. prolifera as a nuisance species (Text S3 [Wang et al., 2007]). U. prolifera can grow over P. yezoensis on the aquaculture rafts, causing poor harvests. Common techniques to remove U. prolifera from P. yezoensis include lifting the rafts out of the water and suspending them in air or storing them in cold compartments, as U. prolifera quickly dies in air or cold temperature (Text S3). During or before harvesting in May, the cleaning of the culturing rafts or severe storms can release considerable amounts of U. prolifera to the water. Although U. prolifera is known to grow well in sewage waters (D. Bronk, Virginia Institute of Marine Science, personal communication, 2008; B. Lapointe, Harbor Branch Ocenaographic Institution, personal communication, 2009), the dramatic increase in cultured U. prolifera area in 2008 strongly suggests the bloom's connection with an increase in aquaculture biomass of P. yezoensis, since nutrients and other pollutants discharged from local rivers to the YS did not show any apparent increase in 2008 relative to, for example, 2005, when the U. prolifera bloom was only found in the ECS (Figure 5).

Figure 4.

Aquaculture of the seaweed P. yezoensis in coastal areas of Jiangsu Province (in Ha). Also shown are the average air temperatures for the month of May from three coastal cities (Lianyungang, Yancheng, and Nantong) with P. yezoensis aquaculture.

Figure 5.

(top) Annual discharge of the three major pollutants from Sheyang River (location shown in Figure 1a inset) and (bottom) annual percentage area of polluted water, defined by a pollution level index (GB3097-1997), in the YS. The index was used as a surrogate for nutrient availability.

4. Discussion

[22] In summary, four lines of evidence support the hypothesis that U. prolifera in the YS and ECS was connected to coastal aquaculture of P. yezoensis along the 200 km shoreline of the Subei Bank, specifically, (1) proximity (several kilometers) to shore of the first appearance of U. prolifera in Landsat imagery; (2) recurrent U. prolifera slicks between 2000 and 2009 all traced to the Subei Bank (the slicks may have occurred in other years before 2000 in nearshore waters, but this is unknown due to the lack of MODIS and Landsat data before 2000); (3) connectivity revealed by the particle tracing numerical experiment; and (4) U. prolifera growth with P. yezoensis aquaculture nets. For the same reasons, the summer 2008 bloom event in the YS resulted from the dramatic expansion of the P. yezoensis aquaculture (Figure 4).

[23] However, some questions still remain unanswered, such as why U. prolifera patches were not found in satellite imagery in 2006. Similarly, it is unclear why extensive blooms did not occur in the YS in 2000, 2001, and 2004, when small U. prolifera patches appeared in the north of the Subei Bank.

[24] Laboratory studies showed that U. prolifera preferred low-light environments, and optimal growth required temperatures between 15 and 25°C [Taylor et al., 2001; Wang et al., 2007]. We have examined all environment variables that can potentially affect U. prolifera growth, including air temperature recorded in several local cities, sea surface temperature (SST) from MODIS (2003 to present) and AVHRR (Advanced Very High Resolution Radiometer, 1985 to present), photosynthetically available radiation (PAR) estimated from SeaWiFS measurements (Sea-viewing Wide Field-of-view Sensor, 1997 to present), and surface wind patterns from QuikScat measurements (2000 to present). However, there was no conclusive interannual pattern that could explain why U. prolifera was not found in 2006. For example, air temperature in May 2006 was significantly higher than in May 2002 (Figure 4), yet extensive U. prolifera slicks were found in the ECS in 2002 while no U. prolifera was found in 2006. We speculate that the occurrence of U. prolifera in 2006 may have been missed in the satellite time series due to clouds. For 2005, only one MODIS image on 23 June 2005 showed some U. prolifera slicks about 200 km east of the Changjiang River mouth (Figure 1a). If clouds had covered the entire ECS on this day, our statistics would show no U. prolifera for 2005 either. The same argument also applies to the much less frequent Landsat imagery, which could have easily missed some U. prolifera slicks in the Subei Bank.

[25] Hence, we conclude that small patches of U. prolifera existed in the Subei Bank every year at some point between April and June, a result from the continuous P. yezoensis aquaculture. Indeed, MODIS SST data showed that the low bound threshold temperature (15°C) for optimal growth of U. prolifera occurred in the month of April when P. yezoensis was harvested. This also agrees with the MODIS/Landsat observations in Figures 1 and 2. Depending on the water movement directions, the U. prolifera patches can be transported to ECS and/or the YS north of the Subei Bank. Once in the sediment-poor YS, the initial amount of U. prolifera and nutrient availability collectively determine whether a massive bloom may occurs during subsequent months. For example, May 2000 and 2001, although some U. prolifera patches were found in the relatively clear waters north of the Subei Bank, low nutrient levels in this region (Figure 5 (bottom)) probably did not provide conditions for massive blooms during the following months. In contrast, during May 2008 when excessive amount of U. prolifera was transported from the Subei Bank to the sediment-poor YS, a massive bloom occurred even if available nutrients (inferred from the water pollution index for the region) were not significantly higher than previous years (2003–2007). Thus, U. prolifera blooms were not a result of increased nutrients, but very likely a result of expanded aquaculture. Other environmental factors may play a role, yet their interannual temporal patterns were not conclusive.

[26] Coastal aquaculture of P. yezoensis in nearshore waters of the Subei Bank started in the 1960s, and has expanded rapidly in the last decade (Figure 4). Although complete statistics for 2009 are not available, a recent news report suggests that in Yancheng City alone, about U.S. $17 million was invested to increase the aquaculture area by another 5400 Ha in 2009 (Text S2). It is therefore natural to ask whether a similar bloom event as in summer 2008 will occur in 2009.

[27] During preparation of this manuscript between late 2008 and early 2009, the most recent MODIS imagery in May 2009 showed extensive U. prolifera slicks in the ECS (Figure 1a). Landsat imagery on 29 May 2009 showed that the algae slicks in the Subei Bank were only 20 km south of the location where U. prolifera was initially found in May 2008 [Hu and He, 2008]. We predicted that once the algae slicks reached the sediment-poor waters in the north, a similar bloom to the summer 2008 case would occur. Indeed, at the time of this writing in October 2009, we have observed a bloom event in the YS that was nearly identical in dimensions to the 2008 bloom event, except that due to different wind directions the massive bloom did not end up in Qingdao. On 2 July 2009 when the bloom had reached its maximum extent, of the 39,000 km2 waters containing the algae in the YS, about 1450 km2 were covered by the algae. In comparison, during the 2008 bloom event, on 25 June 2008 (maximal bloom extent), of the 24,000 km2 waters containing the algae in the YS, about 1840 km2 were covered by the algae. The algae area coverage in the two years is similar, although some of the algae slicks in July 2009 drifted northeast and reached South Korea. These results suggest that with the continuous expansion of the coastal aquaculture of P. yezoensis, similar bloom events are likely to occur in the future, posing significant environmental concerns to local government and management agencies.

[28] Approximately 25% of the global human population resides in coastal zones [Small and Nicholls, 2003]. Correspondingly, coastal aquaculture of seaweed in many countries, especially in Asia but also including North America, supports the local economy and provides a significant portion of the global seafood. For example, a net profit of about U.S. $53 million was obtained from the P. yezoensis aquaculture in Jiangsu Province in 2007 (Text S2). However, the profit is completely compromised by the environmental management cost (>U.S. $100 million to manage the algae bloom in 2008; Text S1 [Wang et al., 2009]). Proper management, with more efficient methods to suppress U. prolifera growth during both culturing and harvesting, is key to avoid future massive U. prolifera blooms in the YS. On the other hand, similar to the role of the brown macroalgae Sargassum spp. in the Gulf of Mexico and the South Atlantic Bight, recurrent U. prolifera in the ECS as a biproduct of the coastal aquaculture may serve as an important habitat (shade and food) for fish, shrimp, turtles, crabs, and other organisms [South Atlantic Fishery Management Council, 2002]. Whether or not this is true, however, deserves further investigations.

[29] In the global perspective, green macroalgae blooms have been reported in many coastal oceans. The effectiveness of the FAI and the free availability of MODIS and Landsat data have demonstrated great potentials in establishing a long-term record of the algae blooms for any coastal ocean region. Such a record can help understand local environmental changes in response to anthropogenic activities and global climate change.

5. Conclusion

[30] The summer 2008 U. prolifera bloom event in the YS stimulated several studies for the bloom's origin, cause, and ecological consequence. This multidisciplinary work resulted in several new findings. First, application of a novel index (FAI) to multiresolution remote sensing data showed that U. prolifera occurred not only in 2008, but also in almost every year between 2000 and 2009, although the area coverage of the macroalgae before 2008 was much smaller than those in 2008 and 2009. Combined with high-resolution numerical circulation modeling, the results revealed that the recurrent U. prolifera was all traced to the nearshore waters of the Subei Bank. Finally, analysis of local aquaculture data and other environmental data strongly suggests that the recurrent U. prolifera slicks/patches (including the massive blooms in summer 2008 and summer 2009) in both the YS and ECS were unlikely a result of coastal eutrophication but rather connected to coastal aquaculture of P. yezoensis in this region. Because of the continuous expansion in aquaculture, similar bloom events such as those in summer 2008 and summer 2009 may also occur in the future.

Acknowledgments

[31] This work was supported by the U.S. NASA (NNX09AE17G) and NOAA (NA06NES4400004), by the National Science Foundation of China (60638020), and by a special project of the Qingdao government (09-2-5-4-HY). We thank the U.S. NASA/GSFC and USGS for providing MODIS and Landsat data, and two anonymous reviewers for their constructive comments. We also thank Deborah Bronk of the Virginia Institute of Marine Science and Brian Lapointe of the Harbor Branch Oceanographic Institution for their input on nutrient requirements of coastal U. prolifera.

Ancillary