River are considered to induce red tides around Kakuijima Island. The longshore current flowing from the Ibo River is estimated to be dispersed offshore by the Chikusa River, which has no dam and flows into the Harima-nada Sea between Kakuijima Island and the Ibo River.

Asahi River is close to the Yoshii River, but the mouths of both rivers flow into Kojima Bay and the outflow of the two rivers is considered to become one flow in Kojima Bay.

4. Discussion

Miyagi Prefecture reported the occurrences of red tide from 1978 to 2011 in Sendai Bay and along the Sanriku Coast, which includes Oppa Bay, Onagawa Bay, Okatu Bay, Sizukawa Bay and Kesennuma Bay [17]. In Kesennuma Bay, the number and areas of red tide occurrences were well documented along with water quality parameters by Ito et al. [7]; from 1972 to 2003 red tide occurrences continued for periods lasting from several days to over seven months.

Ito et al. [8] reported that the mud that had been estimated to be deposited at the inner part of the seafloor of Kesennuma Bay through drainage from factories and that contributed to eutrophication and phytoplankton growth, including red tide, was removed in work conducted from 1976 to 1987. Red tide occurrences in Kesennuma Bay decreased after the removal of the mud (Figure 3), demonstrating a correlation between mud accumulation and red tide occurrences.

In a previous paper, the link between mud delivered from the rivers on which dams had been constructed and red tides was clearly established. However, only one small river without a dam flows into Kesennuma Bay. The situation in Kesennuma Bay is different than that observed in other big bays along the Japan coast [7].

It is considered that mud is not necessarily introduced to bays only by a dam constructed on a river that flows into the bay but also can be carried into the bay by a nearby river. Thus, it is estimated that the Kitakami River had an influence on Kesennuma Bay due to its proximity.

In Figure 3, the construction of Aratozawa Dam was shown to be linked with only low red tide occurrences. The lack of influence due to the Aratozawa Dam is considered to be explained by the fact that the overflow from this dam flows into the Old Kitakami River and flows into Sendai Bay (Figure 2); therefore, this dam does not influence the Sanriku Coast and Kesennuma Bay.

Based on results obtained in this study, the mud that accumulates in Kesennuma Bay is estimated to be deposited by the Kitakami River. In 1995, red tide was also recorded in Shizukawa Bay located south of Kesennuma Bay and north of the Kitakami River [17]. This mud was found to be transported away from the dam and from the mouth of the Kitakami River to the Pacific Ocean where the mud is considered to not be deposited around the mouth of the river but is carried northward by longshore currents (Figure 2) as far as Shizukawa Bay and Kesennuma Bay. The longshore currents are well known to be caused by waves along the coast. After reaching Shizukawa Bay, part of the mud is estimated to be deposited in the bay due to slowing current speed, and red tide is estimated to be induced in this area. Similarly, the longshore current speed is estimated to slow after reaching Kesennuma Bay, and fine particles of mud are deposited there and are estimated to induce red tide.

Red tides were observed from 1981 to 2008 in Oppa Bay, Okatu Bay and Onagawa Bay [17]. Oppa Bay, Okatu Bay and Onagawa Bay have no inflow from rivers with dams. The mud is estimated to be carried southward from the mouth of the Kitakami River by longshore currents as far as Oppa Bay, Okatu Bay and Onagawa Bay to be deposited in the bay, and red tide is estimated to be induced there.

Similar to Kesennuma Bay, Dokai Bay is estimated to be affected mostly by the dams on the Koya River, which is to the east of Dokai Bay, and by the dam on the Onga River, which is located east of Dokai Bay.

During the ebb tidal current, the flow from the Koya River that flows into the Kanmon Channel averages about 5 knots for 5 to 6 h [18]. This flow rate is sufficient for mud to clear the Kanmon Channel of about 21 km length. Thus, mud from the Koya River reaches the mouth of Dokai Bay from the estuary of the Koya River. After reaching the estuary of Dokai Bay, the tidal current slows and mud in the stream is estimated to be deposited around this area. During the period of the full tidal current, part of the mud is considered to flow into Dokai Bay and is deposited there (Figure 4).

In Dokai Bay, mud that had accumulated on the seafloor and that was estimated to mainly be derived from the drainage of factories located around Dokai Bay was dredged from 1973 to 1976 [19]. The Koya River Dam (C) in Figure 4 with a surface area of 161 ha was built on the Koya River in 1955 and the Rikimaru Dam (F) in Figure 4 with a surface area of 79 ha was built on the Onga River in 1965. It is thought that these dams delivered a considerable amount of mud to the Dokai Bay until 1973.

The number of red tide occurrences in the Suo-nada Sea fluctuated markedly from 1975 to 2005 (Figure 6). To date, red tide occurrences are estimated to be induced by eutrophication brought about by domestic waste water, agricultural effluent and so on. Then these fluctuations are considered to be caused by eutrophication. It is considered that when the number of red tide occurrences increases, the quantity and/or concentration of eutrophic substances such as T-N and T-P becomes bigger and higher. But there is no paper that reports the increase and decrease of eutrophic substances that correspond with the fluctuation of red tide occurrences. Moreover, the main sources of eutrophic substances causing the each peak of the number of red tide occurrences in Figure 6 were not analyzed and identified in Suo-nada Sea.

The obtained results show a correlation between the number of red tide occurrences per year in the Suo-nada Sea and the surface area of dams on rivers flowing into the Suo-nada Sea (Figure 6). Red tides have a tendency to be induced a few years after the dam is constructed. Thus, red tide occurrences can be expected in the bays and enclosed seas several years after starting to receive water from rivers with dams.

The earliest maps showing areas of red tide occurrences in the Suo-nada Sea and Harima-nada Sea are from 1960. Using the map in 1960, it is possible to identify causes of red tide events. However, due to the boom in dam construction from 1960 to 2000, the maps showing areas of red tide occurrences after 1970 are difficult to interpret.

Similar to the mechanism in Kesennuma Bay, occurrences of red tides in Tokuyama Bay are considered to be induced by mud carried by longshore currents from the Saba River on which the Saba River Dam had been constructed in 1955. One of the reasons is that the Saba River is the closest river on which a dam had been constructed, and the other is that the dam was completed five years before the first red tide event was recorded in 1960. Mud originating from factories around Tokuyama Bay and transported from the Saba River by the longshore current is considered to be deposited into the bay as the sea current slows and stagnates there.

Fishermen brought a suit against the companies that were active around Tokuyama Bay for the red tide occurrences and the mediated settlement included the requirement for those companies to remove the accumulated mud from Tokuyama Bay [20]. This illustrates the relationship between red tide occurrence and mud accumulation.

Red tide was observed along the coast of Fukuoka Prefecture, which faces the Suo-nada Sea (also called the Buzen Sea) in 1960 (Figure 6). Until 1960, no big dams had been constructed on the rivers flowing into the Buzen Sea, including the Tono, Ima, Yamakuni and Yakkan rivers. This phenomenon requires further consideration now.

Similar to Suo-nada Sea, in Harima-nada Sea the number of red tide occurrences fluctuates greatly from 60 in 1976 to 20 in 1978 (Figure 8). After 1978, the fluctuation in the number of red tide occurrences continued. There are no reports that identify factors, explain this fluctuation and specify the sources of eutrophication. In this paper, fluctuation in the number of red tide occurrences is explained for the first time.

There is a large industrial area in the area facing Tokuyama Bay and Dokai Bay. Further, there were many food processing factories, including those related to the fishery industry, on Kesennuma Bay before the Great Tohoku Earthquake and tsunami in 2011, but in the area facing Kakuijima Island in the Harima-nada Sea, there are no big industrial areas that compare to those in Tokuyama Bay and Dokai Bay. In the coastal area near Kakuijima Island, there is no fishery industry. However, red tide occurrence was reported in 1960 (Figure 7). From these conditions, the red tide occurrence near Kakuijima Island can not be explained by mud carried from factories. It should be considered that the red tide occurrence was induced by a dam constructed on the Saba River and mud carried from the dam to near Kakuijima Island.

In this discussion of red tide occurrences, the chronology of red tide occurrences and dam construction is an important analysis factor. Magana et al. [13] reported the chronology of red tide occurrences in the western part of the Gulf of Mexico, and I examined the relationship between red tide occurrence and dam construction. Figure 10 summarizes the locations of rivers, constructed dams and red tide occurrences in the areas facing the Gulf of Mexico.

In 1935, there were reports of red tide occurrences along the Mexican coast. Lund reported a red tide off Padre Island, Texas on 30 June 1935, and a fish kill was observed over an area extending from Port Aransas, Texas, southwards for about 84 miles [21]. This report shows that red tide was observed in the Gulf of Mexico and not in Corpus Christi Bay. If the influence of dams constructed on the Nueces River induced the red tide, the red tide would have occurred in Corpus Christi Bay based on the occurrence of red tide in Tokyo Bay, Ise Bay, Osaka Bay and the Ariake Sea after dams were constructed [7]. Thus, this red tide should be estimated to have been induced by mud in the longshore current that flowed from the Rio Grande (Figure 10). Gunter reported that kills of marine fish and other animals were observed in the Gulf of Mexico in the summer of 1935 and that the area of fish kills extended northward over 250 miles from the Rio Grande [22].

Lund’s and Gunter’s papers are considered to report the same event. It is considered that these mortalities were caused by the red tide. It should be considered that the starting point of the bloom was at the mouth of the Rio Grande based on these two reports.

These phenomena are estimated to be due to dams constructed on the Rio Grande. The three dams built in 1934 on the Rio Grande were the San Acacia Diversion (J), Isleta Diversion (I) and Angostura Diversion (H) dams (Figure 10). After these dams were completed, it is estimated that muddy soil was carried into the Gulf of Mexico and then flowed northward and/or southward by the longshore current. This phenomenon is similar to theone identified in Kesennuma Bay and Dokai Bay of Japan.

Gunter reported a mass fish mortality that occurred in the fall of 1948 along the southern end of the Texas coast, and this phenomenon is estimated to be attributable to red tide. In 1935, the El Vado Dam (C) and in 1938, the American Diversion (N) and Caballo (L) dams were built on the Rio Grande (Figure 10). It is considered that this event is attributable to dam constructions and mud carried from the dams [22]. The red tide occurrence was induced about ten years after the completion of these dams. The reason for the long lag between the construction of the dams and the occurrence of red tides needs further consideration.

There are two reports of red tide occurrences in 1955 and they are estimated to be correlated with dam constructions. Wilson and Ray reported an extensive red tide bloom during September 1955 near Port Isabel [23] following the completion of the Platoro Dam (390 ha, 1951) (A), Jemez Canyon Dam (570 ha, 1953) (G) and Falcon Dam (35,400 ha, 1954) (P) on the Rio Grande. It is estimated that mud was carried northward from the mouth of the Rio Grande along the Texas coast, and a portion of it entered into the Lower Laguna Madre at high tide, probably through the channel between South Padre Island and Brazos Island. Since the channel is near the mouth of the Rio Grande, the concentration of mud is estimated to be high. This phenomenon is similar to the red tide occurrence in Kesennuma Bay and Dokai Bay mentioned above.

On the other hand, mud was also carried southward and red tides were induced during October to November 1955 from Tamaulipas to Veracruz as reported by Ramirez-Granados [24]. The Rio Panuco lies between the Rio Grande and the Rio Papaloapan in Mexico. Therefore, the flow of mud from the Rio Grande or the Rio Papaloapan is estimated to be dispersed offshore by the Rio Panuco (Figure 10).

However, there have not been any reports of red tide occurrences in the Gulf of Mexico after 1969 when the Amistad Diversion Dam (N) was built on the Rio Grande. This big dam has a vast surface area of 26,300 ha, and it is estimated that construction of the dam induced big red tide occurrences in the Gulf of Mexico. The mechanism for this phenomenon will require further consideration.

Magana et al. [13] wrote that a red tide bloom was observed in 1974 along the Mexican coastline from south of the Rio Grande to Tampico, Mexico. From 1970 to 1973, three dams were constructed on the Rio Grande. Those are the Galisteo (1970, F), Heron (1971, B) and Cochiti (1973, E) dams. The completion year and the letter designation in Figure 10 are shown in parenthesis. It is considered that this red tide was induced by mud flowing southward from the Rio Grande in the longshore current.

Cortes-Altamirano et al. reported toxic red tides along the coast of Mexico near Veracruz in 1994-1995 [25]. It is estimated that these red tides were attributable to the completion of the Cerro de Oro Dam (Q) built on the Rio Papaloapan in 1989.

The Rio Grihalva is a big river in the south of Mexico on which four dams (T-W) were constructed (Figure 10). These dams were constructed from 1966 to 1986, but there were no reports of red tide occurrences and/or a mass fish mortality after 1966 around the mouth of the Rio Grihalva. This phenomenon will require further consideration and research.

It is known that red tides have occurred off the western coast of Florida for more than 160 years. Those events often include harmful algae blooms, which are toxic to fish and marine animals, according to the Florida Fish and Wildlife Conservation Commission [26]. Red tides have also been observed along the northwestern coast of Cuba [27,28].

On the other hand, there are no big rivers on the Florida Peninsula, making it is impossible for red tides to be induced by rivers in and/or near the Florida Peninsula. However, in the Gulf of Mexico, there is a Loop Current and the sea currents flow clockwise offshore from the Florida Peninsula (Figure 10) [29]. It is estimated that the outflow of the Mississippi River extends southeast into the Gulf of Mexico and that part of the outflow reaches the west coast of the Florida Peninsula and the northwestern coast of Cuba.

All red tide events and/or a fish mortality events cited in the chronology edited by Magana et al. [13] are not explained by the construction of dams. In particular, events of the massive fish kill observed between 1648 and 1875 in Veracruz, Mexico could not be explained by the construction of dams. Most events occurring after 1986 in the Gulf of Mexico also need further consideration.

However, it is estimated that there are common mechanisms that cause red tides and/or a mass fish mortality among small Japanese bays such as Kesennuma Bay, Dokai Bay, nearby Kakuishima Island in Harima-nada Sea, Tokuyama Bay in Suo-nada Sea, and the Gulf of Mexico. The mud flowing from rivers on which dams have been built is estimated to be carried from the mouth of the river along the coast by longshore currents. When the longshore current enters a small bay, the current slows and fine particles in the mud are readily precipitated, causing red tide in the bays.

5. Conclusions

The relationship between red tide occurrences and dam construction was researched and is discussed in this paper using chronological data from Kesennuma Bay, Dokai Bay, Tokuyama Bay in Suo-nada Sea and nearby Kakuishima Island in Harima-nada Sea. As a result, the same tendency was observed in these four sea areas.

Red tides occurring in these four areas were estimated to be induced by mud overflowing from dams constructed on rivers that flow into the sea near the four areas. The mud flow is estimated to be carried north and/or south along coast by a longshore current, and the flow rate is estimated to become slow or stop in the four sea areas, depositing mud deposits there and inducing red tides.

The red tides observed in the Gulf of Mexico as compiled by Magana et al. [13] were estimated to be induced by mud carried from rivers such as the Rio Grande and the Rio Papaloapan. However, some of the red tide records cited by Magana et al. need further research and consideration.

The red tides observed off the western coast of the Florida Peninsula are considered to be induced by mud carried from the mouth of Mississippi River. It is well known that there is a loop current in Gulf of Mexico and mud carried from the Mississippi River is estimated to be carried off the western coast of Florida Peninsula and northwest coast of Cuba and deposited there.

6. Acknowledgements

I would like to thank the following facilities for providing data on their websites: The Japan Dam Foundation and the Florida Fish and Wildlife Conservation Commission.

REFERENCES