Layout 1 INTRODUCTION Concerns regarding the presence of cyanobacterial toxins (cyanotoxins) in drinking water and associated health effects have raised research and public health in- terest worldwide. Microcystins (MCs) are probably the most frequently found cyanotoxins which can be pro- duced by various cyanobacterial genera including water bloom- and scum-forming planktonic cyanobacteria such as Dolichospermum (formerly Anabaena), Microcystis or Planktothrix (Manganelli et al., 2012). Cyanobacteria rep- resenting these genera have been previosly identified in Ghanaian water reservoirs along with other cyanobacterial species potentially producing MCs (Addico et al., 2006, 2009, 2017). MCs are highly toxic for mammals with acute LD50 as low as 50-60 µg kg–1, mouse, i.p. (Bláha et al., 2009; Van Apeldoorn et al., 2007). Their acute effetcs are primarily manifested in liver but MCs have been shown to induce gastrointestinal and renal damage or neu- rological symptoms as well (Manganelli et al., 2012). Chronic exposures to MCs have been linked to tumor pro- moting and carcinogenic effects which is based on labo- ratory animal and in vitro experiments (Svircev et al., 2010) and supported also by results of epidemiologic studies of human population consuming drinking water contaminated by these cyanotoxins (Fleming et al., 2002; Svircev et al., 2009, 2013, 2014; Ueno et al., 1996; Yu et al., 1995; Zhou et al., 2002). In fact, MCs have been clas- sified as possible human carcinogen (class 2B) by the In- ternational Agency for Research on Cancer (Grosse et al., Advances in Oceanography and Limnology, 2017; 8(1): 92-106 ARTICLE DOI: 10.4081/aiol.2017.6323 This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License (CC BY-NC 4.0). Cyanobacteria and microcystin contamination in untreated and treated drinking water in Ghana Gloria Naa Dzama Addico,1* Jörg D. Hardege,2 Jiří Kohoutek,3 K.A.A. deGraft-Johnson,1 Pavel Babica3,4 1CSIR Water Research Institute, Achimota, Accra, Ghana; 2Biological Science Department, University of Hull, United Kingdom; 3RECETOX - Research Centre for Toxic Compounds in the Environment, Faculty of Science, Masaryk University, Brno, Czech Republic; 4Department of Experimental Phycology and Ecotoxicology, Institute of Botany, Czech Academy of Sciences, Brno, Czech Republic *Corresponding author: naadzama443@hotmail.com ABSTRACT Although cyanobacterial blooms and cyanotoxins represent a worldwide-occurring phenomenon, there are large differences among different countries in cyanotoxin-related human health risk assessment, management practices and policies. While national standards, guideline values and detailed regulatory frameworks for effective management of cyanotoxin risks have been implemented in many in- dustrialized countries, the extent of cyanobacteria occurrence and cyanotoxin contamination in certain geographical regions is under- reported and not very well understood. Such regions include major parts of tropical West and Central Africa, a region constisting of more than 25 countries occupying an area of 12 million km2, with a total population of 500 milion people. Only few studies focusing on cyanotoxin occurrence in this region have been published so far, and reports dealing specifically with cyanotoxin contamination in drinking water are extremely scarce. In this study, we report seasonal data on cyanobacteria and microcystin (MC) contamination in drinking water reservoirs and adjacent treatment plants located in Ghana, West Africa. During January-June 2005, concentrations of MCs were monitored in four treatment plants supplying drinking water to major metropolitan areas in Ghana: the treatment plants Barekese and Owabi, which serve Kumasi Metropolitan Area, and the plants Kpong and Weija, providing water for Accra-Tema Met- ropolitan Area. HPLC analyses showed that 65% samples of raw water at the intake of the treatment plants contained intracellular MCs (maximal detected concentration was 8.73 µg L–1), whereas dissolved toxins were detected in 33% of the samples. Significant reduction of cyanobacterial cell counts and MC concentrations was achieved during the entire monitoring period by the applied conventional water treatment methods (alum flocculation, sedimentation, rapid sand filtration and chlorination), and MC concentration in the final treated water never exceeded 1 µg L–1 (WHO guideline limit for MC-LR in drinking water). However, cyanobacterial cells (93-3,055 cell mL–1) were frequently found in the final treated water and intracellular MCs were detected in 17% of the samples (maximal con- centration 0.61 µg L–1), while dissolved MCs were present in 14% of the final treated water samples (maximal concentration 0.81 µg L–1). It indicates a borderline efficiency of the water treatment, thus MC concentrations in drinking water might exceed the WHO guide- line limit if the treatment efficiency gets compromised. In addition, MC concentrations found in the raw water might represent significant human health risks for people living in areas with only a limited access to the treated or underground drinking water. Key words: Cyanobacteria; cyanotoxins; drinking water treatment; microcystins; water blooms. Received: October 2016. Accepted: May 2017.No n- co mm er cia l u se on ly Microcystins in drinking water in Ghana 93 2006). In addition, other epidemiological studies associ- ated exposures to toxic cyanobacterial blooms and MCs with chronic liver damage (Chen et al., 2009; Li et al., 2011; Zhang et al., 2015), and MCs were also implicated in neurotoxicity and neurodegenerative diseases (Feurstein et al., 2009, 2010). MCs are therefore regarded as human health hazard. Exposure of human beings to MCs can occur via different routes, such as recreational and sport activities in contaminated water, consumption of contaminated fish products or food supplements, and consumption of contaminated drinking water (Manganelli et al., 2012). The World Health Organization (WHO) set a provisional guideline limit of 1 µg L–1 of MC-LR in drinking water (WHO, 1998). Negative health outcomes resulting from drinking of water contaminated with cyanobacteria or cyanotoxins have been reported world- wide (Bláha et al., 2009; Van Apeldoorn et al., 2007; Wood, 2016). The only documented case of cyanobacte- ria-associated poisoning in Africa has been reported from Harare, Zimbabwe, where annual outbreaks of gastroen- teritis among infants occurred after development of cyanobacteria blooms of Microcystis aeruginosa (KüTZ- ING) KüTZING and Dolichospermum flos-aquae (BRESSON EX BORNET & FLAUHAULT) in Lake Chievero (Zilberg, 1966). However, there is also a documented case of three- time rise of gastroenteritis and 4.3-time increase of liver cancer incidence rate in Harare during the period 1990- 2001 (Ndebele and Magadza, 2006). Although the extent to which this situation is linked to algal toxins is unclear, Johansson and Olsson (1998) reported that MC concen- tration in Lake Chievero was around 13.9 µg L–1 and MC was also detected in municipal tap water. In the last decades, MC occurrence has been investi- gated in some African countries, with MCs reported from Northern Africa (Algeria, Egypt, Morocco, Tunisia), East- ern Africa (Ethiopia, Kenya, Mozambique, Tanzania, Uganda), Southern Africa (Botswana, Lesotho, South Africa) (Mowe et al., 2014; Harke et al., 2016; Ndlela et al., 2016). However, available data about cyanotoxin con- tamination are still very scarce for most African countries and being nearly absent for regions such as West and Cen- tral Africa (Mowe et al., 2014; Harke et al., 2016; Ndlela et al., 2016). These are two large geographical areas with 26 countries and nearly half a billion inhabitants, where Central Africa is represented by nine countries populated by approximately 155 mil people, and West Tropical Africa by 17 countries with a combined population of about 344 mil people (UNSD, 2014). Toxic or potentially toxic cyanobacterial blooms seem to occur frequently in this geographical area (Addico et al., 2006, 2009, 2017; Akin-Oriola et al., 2006; Berger et al., 2006; Haande et al., 2007; Mhlanga et al., 2006; Odokuma and Isirima, 2007). Nevertheless, MCs in this region have been so far reported from Nigeria (Chia et al., 2009a, 2009b; Chia and Kwaghe, 2015) and detected in water reservoirs Brimsu, Kwanyarko, Kpong and Weija in Ghana (Addico et al., 2006; Addico et al., 2017). Studies investigating cyanotoxins in drinking water and their removal during the water treatment have been even more scarce on the entire African continent, known to be conducted for ex- ample in Egypt (Mohamed and Carmichael, 2000), Alge- ria (Nasri et al., 2004), South Africa (Harding et al., 2009), and recently in two drinking reservoirs in Central Ghana (Addico et al., 2017). In the present study, we investigated seasonal occur- rence and removal of cyanobacteria and MCs in four treat- ment plants supplying the two major metropolitan areas in Ghana with drinking water. The study provides very rare but important information regarding the efficiency of drinking water treatment, concentrations and health risks of MCs in drinking water in the understudied geographi- cal region of Central and West Tropical Africa. METHODS Study area The Barekese Reservoir is a mesotrophic reservoir (Addico et al., 2009), which lies on latitude 6° 49′ 50.2″ N and longitude 1° 43′ 21.8″ W (Fig. 1). This reservoir was formed in 1970, it has a surface area 6.4 km2 and maximal depth 15 m (Amuzu, 1975). It is located on the River Ofin, which flows through many farming areas be- fore reaching the dam site (Kumasi et al., 2011). The Owabi Reservoir, also mesotrophic (Addico et al., 2009), is located on latitude 6° 44′ 35.7″ N and longitude 1° 42′ 13.4″ W (Fig. 1). It was constructed in 1928 and up- graded in 1954. The reservoir has a surface area about 3.5 km2 and a mean depth 7 m (Akoto et al., 2014). The Owabi Reservoir is fed by seven rivers/streams, all of which flow through the densely populated Kumasi Metro- politan Area and the central business and industrial areas. The Owabi and Barekese Reservoirs are both situated in the Ashanti Region of Ghana and serving as major water supplies to Kumasi Metropolitan Area with a population over 2 mil people. The Owabi reservoir is designed to pro- duce up to 20% of the total potable water requirement (Akoto et al., 2014), while the Barekese treatment plant is providing about 80% of the total piped drinking water to the Kumasi metropolis (Kumasi et al., 2011). The Kpong Reservoir, described as mesotrophic (Ad- dico et al., 2009), is located in the Eastern Region of Ghana on 6° 07′ 1.3″ N 0° 07′ 31.6″ E (Fig. 1). It was con- structed in 1981 on the Volta River system mainly to pro- vide hydroelectricity to supplement power generated from the Volta River dam. It has a total surface area of 38 km2, maximal depth of 15 m with a mean depth of 5 m, and a mean annual flow 1183 m3 s–1 (Ansa-Asare and Ansong- No n- co mm er cia l u se on ly G.N.D. Addico et al.94 Asante, 1998; Quarcoopome et al., 2011). The Kpong Reservoir apart from power generation is also used for drinking water production, irrigation, recreation and also well known for its fisheries, especially the tilapias. The Weija Reservoir, a eutrophic reservoir (Addico et al., 2009), is situated in the Greater Accra Region of Ghana and lies on latitude 5° 34′ 7.1″ N and 0° 20′ 44.8″ W (Fig. 1). It has a surface area of about 38 km2, a mean depth of 5 m and a mean annual flow of 54.2 m3 s–1 (Ansa- Asare and Ansong-Asante, 1998, Asante et al., 2008). The Weija Reservoir was built in 1977 on the Densu River sys- tem. This river system is under intensive threat from heavy pollution mainly from domestic and agricultural wastes. Major crops include maize, cassava, pineapples, pawpaw, banana, sugar cane and vegetables. Fishing is also very intensive in the reservoir sometimes with the use of chemicals. The Kpong and Weija Reservoirs are the two main drinking water supplies serving the Accra- Tema Metropolitan Area with a population about 2.3 mil people, with the Kpong treatment plant providing approx- imately 47% and the Weija plant about 53% of piped drinking water for the metropolis (Stoler et al., 2012). Drinking water treatment procedure The water treatment procedure in the studied treatment plants starts from the water intake. In most reservoirs, raw water was collected from depths between 5 to 7 m (Ad- dico et al., 2006), with the exception of the Barekese plant, where the intake point is placed at the level 1.5-3 m from the surface (Amuzu, 1975). Raw water is sieved using a mesh to remove big objects like plant parts, twigs etc. This step is followed by flocculation using aluminium sulphate (alum) at a 100 mg L–1 dose, mixing and passing Fig. 1. Map of the reservoirs and treatment plants under the study. The Barekese and Owabi Reservoirs are located in the Ashanti Region of Ghana and supplying Kumasi Metropolitan Area with drinking water. The Kpong Reservoir is located in Eastern Region of Ghana, and the Weija Reservoir in Greater Accra Region, both reservoirs are providing drinking water for Accra-Tema Metropolitan Area. No n- co mm er cia l u se on ly Microcystins in drinking water in Ghana 95 through baffles to maximise contact time. The flocs are allowed to settle out of the water in sedimentation tanks. The exception is represented by the Kpong Reservoir water treatment plant, where alum flocculation is not reg- ularly applied during water treatment and water is pre- chlorinated before the filtration step. Filtration is then done by the rapid sand filtration method and the pH is ad- justed to between 6.6 and 8.5 using lime. The final stage involves chlorination employing chlorine gas or calcium hypochlorite with a concentration of 0.5 to 1 mg L–1 of residual chlorine after a contact time of about 30 min (Fig. 2). All samples from the water intake to chlorination step were collected consecutively and within the same day. It is important to mention that during one time of sampling at the Owabi treatment plant, algaecide treat- ment with copper sulphate was simultaneously being ap- plied in the reservoir. Sampling and cyanobacteria determination Samples for MC and cyanobacteria analysis were col- lected monthly from January-June 2005 from drinking water treatment plants at the Barekese and Owabi Reser- voirs, and biweekly at the Weija and Kpong Reservoirs. Water samples (1 L) were collected into clean plastic (PET) bottles from each treatment stage, namely: i) raw water at the intake; ii) flocculation (except the Kpong treatment plant); iii) sedimentation tanks or clarifiers; iv) filtered water; and v) final chlorinated water. A total num- ber of 127 samples were analysed for intracellular MCs in the four reservoirs, whilst 59 samples were analysed for dissolved MCs. The lower number of samples analysed for dissolved MCs was due to financial con- straints during the field work in Ghana and losses during the sample transport from Ghana to the United Kingdom. Samples for microscopic determination of cyanobac- terial species composition were collected from raw water and final treated water using plankton net (25 µm) or by simply filling a bucket. Net samples were preserved for taxonomic work with formalin at the final concentration of 2% (v/v), whilst water samples were preserved in Lugol’s solution for quantitative microscopical analysis as described in Addico et al. (2006, 2009) using Olympus BX51 and BX60 microscopes equipped with objectives 10, 20, 40, 60 and 100X (Olympus). Briefly, the aliquots of the samples were transferred into counting chambers for analysis, where all colonies and filaments were counted as individuals. The average number of cells was determined for 20 individuals and cell concentration was calculated. Sub-samples for counting picocyanobacteria were filtered through a 0.2 μm Nucleopore filter prestained with Irgalan black. Cells were stained with Dapi (4-diamidino-2-phenylindole dihydrochloride) and counted under the fluorescence (Excitation 330-385 nm, Emission 510-560 nm). About 300-400 picocyanobacter- ial cells were counted for each sample. All data on abun- dance were expressed as number of cells per mL, including the cells inside colonies. Identification of cyanobacteria species was carried out at the Institute of Botany of the Czech Academy of Sciences, Trebon, Czech Republic under the supervision Prof. Jiří Komárek and using the recent taxonomical literature (Komárek and Anagnostidis, 1999, 2005). Extraction of cell-bound (intracellular) MCs Water samples (1 L) were filtered through preweighed GF/C filter (1.2 µm mesh, Whatman). The cells collected on the filters were frozen overnight and freeze-dried. Freeze-dried cells on filters were stored at -20°C until ex- tracted for HPLC analysis. Extraction of cell-bound (in- tracellular) toxins from freeze-dried cells was done as described by Harada et al. (1999). Cells were extracted with 20 mL of 75% aqueous methanol (Fastner et al., 1998) for 1 hour. This extraction step was repeated three times, the extracts from the individual steps were com- bined and then dried using a rotary evaporator. The con- centrated extract was dissolved in 400 µL methanol prior to HPLC analysis, filtered through 0.45 µm nylon syringe filter (Millipore). Fig. 2. Summary of drinking water treatment process employed in Ghanaian plants Barekese, Owabi, Kpong and Weija. No n- co mm er cia l u se on ly G.N.D. Addico et al.96 Extraction of dissolved (extracellular) MCs Filtrates of water samples (1 L) filtered through GF/C filter (see above) were processed according to Harada et al. (1999). Briefly, filtrates were treated with sodium thio- sulphate (2 mg L–1), acidified with trifluoroacetic acid (TFA, 0.1%, v/v) and concentrated using solid phase ex- traction by ODS cartridges (Supelclean LC-18, 3 mL Tube, Supelco). Cartridges were activated with 5 mL of methanol and rinsed with 5 mL of distilled water prior to the application of the sample. MCs were then eluted with 15 mL of 0.1% TFA in methanol, the eluate was evapo- rated to dryness by rotary vacuum evaporation (45 C) and then redissolved in 400 µL methanol in an ultrasonic bath. Identification and quantification of MCs MCs were identified and quantified using high per- formance liquid chromatography (HPLC Agilent 1100 se- ries) system, coupled with a diode array detector (DAD). MCs were separated on a C-18 column Luna 150×4.60 mm, 5 µm (Phenomenex) at 30°C using a flow rate of 1 mL min–1. The binary gradient of the mobile phase con- sisted of (A) H2O+0.05% TFA and (B) acetonitrile +0.05% TFA, with a linear increase from 30 to 70% B be- tween 0-30 min. The injection volume was 20 µL. Chro- matograms were recorded at 238 nm. UV spectra (200 to 300 nm) of all chromatographic peaks were carefully checked and compared to the spectra of MC standards: MC-LF, -LR, -LW, -RR and -LY (Alexis Biochemicals). Peaks possessing the UV spectrum characteristic for MCs were quantified using a calibration curve (n=5, r2=0.999) of the corresponding standard with the matching retention time. Unidentified peaks possessing the UV spectrum characteristic for MCs but not matching the retention time of the standards were quantified as MC-LR equivalents using the calibration curve of MC-LR (McElhiney and Lawton, 2005). The detection limit of the method (LOD) was 0.01 µg L–1 for the individual MC variant. RESULTS Cyanobacteria removal In this study, we complemented previous data on cyanobacterial concentrations in the raw water (Addico et al., 2009) with a new data set on cyanobacterial cell counts in the final treated water, and also with MC analy- ses, in order to discuss relationships between MC occur- rence, cyanobacterial diversity, and their removal during the drinking water treatment. As reported, all four reser- voirs were dominated by cyanobacteria, which accounted for 70-90% of phytoplankton biomass (Addico et al., 2009). Detailed results of microscopical analyses of cyanobacterial species composition are summarized in Tabs. 1-4, the complete list of the identified species is pro- vided in the Supplementary Tab. 1. Representatives of pic- ocyanobacterial genera Cyanogranis, Aphanocapsa and Geitlerinema were among the most aboundant species Tab. 1. Concentration of cyanobacterial cells (cell mL–1) in the water intake and in the final treated water at the Barekese drinking water treatment plant during the Jan-May 2005. Barekese January February March April May Average Intake* Final Intake* Final Intake* Final Intake* Final Intake* Final Intake Final Removal (%) Anabaena austro-africana 0 0 35 0 140 0 89 0 926 0 238 0 100.0 Anabaena nygaardii 7768 0 9010 54 11,509 10 4923 0 1922 0 7026 13 99.8 Chroococcus cronbergae 756 0 467 0 899 0 1281 0 874 0 855 0 100.0 Cyanogranis ferruginea 201,870 98 229,018 1143 191,002 475 48,594 45 125,000 87 159,097 370 99.8 Cylindrospermopsis raciborskii 1007 11 4005 28 2086 12 4272 0 2760 0 2826 10 99.6 Merismopedia punctata 2987 0 1998 0 1254 0 995 0 1075 0 1662 0 100.0 Merismopedia tenuissima 7098 0 5998 0 5990 0 3709 0 2136 0 4986 0 100.0 Microcystis aeruginosa 501 0 429 0 557 0 400 0 566 0 491 0 100.0 Oscillatoria princeps 5783 0 6602 0 4998 0 5340 0 7251 0 5995 0 100.0 Planktolyngbya minor 3056 0 2955 0 1565 0 1427 0 5073 0 2815 0 100.0 Planktothrix lacustris var. solitaria 2008 3 3090 14 3163 8 5146 14 2895 25 3260 13 99.6 Planktothrix sp. 98 33 801 99 2675 10 1226 26 3925 34 1745 40 97.7 Pseudanabaena recta 6780 0 6675 10 3727 4 3888 7 925 2 4399 5 99.9 Radiocystis fernandoi 0 0 96 0 230 0 0 0 431 0 151 0 100.0 Romeria elegans 56 0 0 0 18 0 53 1 36 5 33 1 96.3 Total (cell mL–1) 239,768 145 271,179 1,348 229,813 519 81,343 93 155,795 153 229,813 153 Removal (%) 99.9 99.5 99.8 99.9 99.9 99.9 *Data adapted from Addico et al. (2009). No n- co mm er cia l u se on ly Microcystins in drinking water in Ghana 97 within the cyanobacterial communities in the studied reservoirs (Tabs. 1-4). In the Barekese and Owabi Reser- voirs, which are located in the same ecological zone in the Ashanti Region (Fig. 1), Cyanogranis ferruginea (F. WAWRIK) HINDAK ex Hindak accounted for the majority of cyanobacterial cells. C. ferruginea population in the raw water at the Barekese treatment plant ranged between 60-85% of total cyanobacteria cell counts (Tab. 1). This species was the most abundant in the treated water as well, accompanied also with Planktothrix agardhii (Gomont) K. ANAGNOSTIDIS & J. KOMáREK, Planktothrix lacustris (KLEBAHN) I. UMEZAKI & M. WATANABE, and eventually by Cylindrospermopsis raciborskii (WOLOSZYNSKA) SEENAYYA & SUBBA RAJU, Pseudanabaena recta KOMáREK & CRONBERG, Anabaena nygaardii CRONBERG & KOMáREK (Tab. 1). Concentrations of cyanobacterial cell in the final water from the Barekese Reservoir ranged between 93-1,348 cell mL–1. In the Owabi treatment plant, C. ferruginea represented 95-97% of the total cyanobacteria cell counts (Tab. 2). In addition to C. ferruginea, cyanobacteria Aphanocapsa hol- statica (LEMMERMANN) G. CRONBERG & KOMáREK, P. recta, and Leptolyngbya sp. were detected most frequently in the treated water from Owabi, with total cyanobacterial counts between 95-1099 cells mL–1 (Tab. 2). Geitlerinema unigranulatum (C. AGARDH EX GOMONT) ANAGNOSTIDIS was the most abundant cyanobacterium in the Kpong Reservoir, representing 65-78% of total cyanobacterial cell counts in the raw water samples (Tab. 3). However, P. agardhii was in average the most abundant species found in the final water from the Kpong reservoir, followed by G. unigranulatum and C. raci- borskii, while other species were detected in the treated water only occassionally. Total cyanobacterial cell counts in the final water were between 173-845 cell mL–1 during the sampling period (Tab. 3). Cyanobacterial community in the Weija Reservoir was the most diverse one (Tab. 4), when the most abundant cyanobacterial species Aphanocapsa nubilum KOMáREK & H.J. KLING accounted only for 18-26% of total cyanobacterial cell counts throughout six months of sam- pling, while being accompanied with Merismopedia tenuissima LEMMERMANN (14-19%), Planktolyngbya minor (GEITLER & RUTTNER) KOMáREK & CRONBERG (8- 15%), P. recta (6-18%) and others. The most abundant species in the treated water was Chroococcus cronbergae J. KOMáREK & E. NOVELO, which penetrated into the final stage throughout the study, along with A. nubilum, M. aeruginosa, A. nygardii, P. agardhii and C. raciborskii. The cyanobacterial cell counts in the final water from the Weija Reservoir were found to be between 369-3,055 cell mL–1 (Tab. 4). Overall, the drinking water treatment process eliminated >97-99.9% of cyanobacterial cells, however, cyanobacteria were detected in 100% samples of treated water collected from all four treatment plants during the entire sampling period. MC removal Commonly occurring MC variants identified in the ex- amined reservoirs were MC-LR, -LF, -RR and -YR. The highest diversity of MC variants was observed in the Weija Reservoir, where also two additional peaks possessing MC- like UV absorption spectrum were identified. Out of the 26 samples of raw water, 17 samples (65%) contained intra- cellular MCs (Fig. 3, Tab. 5). During the water treatment process, concentrations of both intracellular as well as ex- tracellular toxins generally decreased with the treatment Tab. 2. Concentration of cyanobacterial cells (cell mL–1) in the water intake and in the final treated water at the Owabi drinking water treatment plant during the Jan-May 2005. Owabi January February March April May Average Intake* Final Intake* Final Intake* Final Intake* Final Intake* Final Intake Final Removal (%) Anabaena nygaardii 65 0 0 0 0 0 16 0 0 0 16 0 100.0 Aphanocapsa holsatica 2092 76 1980 66 2541 97 1997 45 2672 76 2256 72 96.8 Chroococcus cronbergae 67 0 0 0 10 25 0 0 0 0 15 5 67.5 Cyanogranis ferruginea 225,317 34 220,001 47 165,002 901 157,005 37 278,430 98 209,151 223 99.9 Cylindrospermopsis raciborskii 24 0 15 0 24 0 45 0 23 0 26 0 100.0 Leptolyngbya sp. 946 0 882 27 98 0 107 13 905 9 588 10 98.3 Merismopedia tenuissima 62 0 77 0 102 0 91 0 75 0 81 0 100.0 Planktolyngbya limnetica 772 0 1372 0 1532 0 2109 0 815 0 1320 0 100.0 Planktolyngbya minor 2008 0 1247 0 2349 0 2129 0 3761 0 2299 0 100.0 Pseudanabaena recta 1465 56 2165 35 1645 76 1705 0 868 20 1570 37 97.6 Total (cell mL–1) 232,818 166 227,739 175 173,303 1099 165,204 95 287,549 203 227,739 175 Removal (%) 99.9 99.9 99.4 99.9 99.9 99.9 *Data adapted from Addico et al. (2009). No n- co mm er cia l u se on ly G.N.D. Addico et al.98 step in all four individual treatment plants (Fig. 3). Statis- tically significant (P<0.05) correlation (Spearman’s rank correlation coefficient ρ) between MC concentration and the order of treatment step was found: ρ=0.996 (Barekese), ρ=0.861 (Owabi), ρ=0.987 (Kpong) and ρ=0.899 (Weija) for intracellular toxins, and ρ=1 (Barekese), ρ=0.911 (Owabi), ρ=0.965 (Kpong) and ρ=0.980 (Weija) for dis- solved toxins. However, increases in concentration of both intracellular and dissolved MC were observed in some cases after the flocculation/sedimentation steps of water treatment (Fig. 3 and Supplementary Figs. 1-4). Five sam- ples of the treated water (17%) contained intracellular or Fig. 3. Combined data on MC concentrations at different stages of four drinking water treatment plants in Ghana during Jan-Jun 2005. Boxes plot median values (middle lines), 25th and 75th percentils (boxes), 10th and 90th percentils (error bars) and outliers (circles). Ratio between median intracellular (IC) and dissolved (DIS) MC concentrations was calculated for different treatment steps and ploted as a line graph. Hash indicates significant difference between concentration of IC and DIS MC at the particular step of drinking water treat- ment (P<0.05, Mann-Whitney test). Asterisks indicate significant difference between MC concentration in the water intake and a par- ticular treatment step (P<0.05, Mann-Whitney test). Values below the method LOD (0.01 µg L–1) were susbstituted with LOD/2. Tab. 3. Concentration of cyanobacterial cells (cell mL–1) in the water intake and in the final treated water at the Kpong drinking water treatment plant during the Jan-May 2005. Kpong January February March April May Average Intake* Final Intake* Final Intake* Final Intake* Final Intake* Final Intake Final Removal (%) Chroococcus cronbergae 690 0 382 0 656 0 609 0 541 0 575 0 100.0 Coelomoron tropicale 0 0 13 0 0 0 0 0 40 0 11 0 100.0 Cylindrospermopsis cuspis 1226 0 1232 0 2401 0 1003 11 3980 0 1968 2 99.9 Cylindrospermopsis raciborskii 2625 30 1806 39 2216 87 3873 35 2082 60 2520 50 98.0 Geitlerinema unigranulatum 31,584 127 39,562 89 49,325 179 30,252 389 33,546 16 36854 160 99.6 Merismopedia punctata 1204 0 924 0 840 0 550 0 1082 0 920 0 100.0 Merismopedia tenuissima 2033 0 3693 0 2991 0 1563 0 3865 0 2829 0 100.0 Planktolyngbya minor 55 0 61 0 90 0 61 0 1531 67 359 13 96.3 Planktothrix agardhii 2475 65 4123 45 4070 293 2846 410 4352 41 3573 171 95.2 Pseudanabaena recta 673 0 904 0 719 0 646 0 320 0 652 0 100.0 Total (cell mL–1) 42,562 222 52,698 173 63,306 559 41,400 845 51,336 184 51,336 222 Removal (%) 99.5 99.7 99.1 98.0 99.6 99.6 *Data adapted from Addico et al. (2009). No n- co mm er cia l u se on ly Microcystins in drinking water in Ghana 99 particle-associated MCs (maximal detected concentration was 0.61 µg L–1), and two samples of the treated water (14%) contained dissolved MCs at concentrations 0.57 µg L–1 (Kpong) and 0.81 µg L-1 (Weija) (Tab. 5). Concentra- tions of intracellular toxins significantly correlated with the concentrations of the cyanobacterial cells in the treated water (ρ=0.561, P<0.01). In the Barekese Reservoir, MCs were detected during two out of six sampling months (Tab. 5, Supplementary Fig. 1). In one instance (February 10th), intracellular MCs were found in the sedimentation step and then in the sam- ple of treated water (0.45 µg L–1), while dissolved MCs were found in the flocculation stage (Supplementary Fig. 1). In April, intracellular MCs were detected in one sam- ple of raw water at the concentration 0.46 µg L–1, which further increased in the flocculation stage, but then de- creased below the detectable levels in the next treatment step (Tab. 5, Supplementary Fig. 1). The Owabi Reservoir was found to be more contami- nated with MCs. All raw water samples from the Owabi treatment plant contained intracellular MCs (Tab. 5, Sup- plementary Fig. 2). The highest detected concentration of intracellular MCs in the intake water from the Owabi Reservoir was 8.73 µg L–1, and intracellular toxins were detected also in the final water in one instance (0.07 µg L–1, March 17th). Dissolved MCs could be found in sam- ples from all treatment stages of the Owabi treatment plant with the exception of the final stage (Tab. 5, Sup- plementary Fig. 2). In the Kpong Reservoir, MCs were detected relatively less frequently. Intracellular toxins were found only in two out of seven samples of intake water (Tab. 5, Supplemen- tary Fig. 3). However, the Kpong drinking water treatment plant, which does not have a flocculation stage, had two out of eight samples contaminated with intracellular MCs at the final chlorination stage (0.13 and 0.46 µg L–1, March Tab. 4. Concentration of cyanobacterial cells (cell mL–1) in the water intake and in the final treated water at the Weija drinking water treatment plant during the Jan-May 2005. WeijaJanuary February March April May Average Intake* Final Intake* Final Intake* Final Intake* Final Intake* Final Intake Final Removal (%) Anabaena austro-africana 398 0 330 0 1,075 0 961 0 2807 0 1,114 0 100.0 Anabaena nygaardii 2104 21 7080 475 5173 67 6485 35 11,297 15 6428 123 98.1 Anabaenopsis ambigua 177 0 763 0 738 0 792 0 45 0 503 0 100.0 Anabaenopsis tanganyikae 34 0 109 69 60 0 879 64 557 0 328 27 91.9 Aphanocapsa holsatica 6302 25 2677 0 821 0 2151 0 6350 0 3660 5 99.9 Aphanocapsa nubilum 24,567 12 20,538 883 20,832 80 26,883 42 19,511 67 22,466 217 99.0 Chroococcus cronbergae 3141 515 5398 515 5284 962 5989 483 4586 224 4880 540 88.9 Coelomoron tropicale 52 0 20 0 0 0 0 0 0 0 14 0 100.0 Cyanogranis ferruginea 0 0 80 0 1208 0 903 0 1773 0 793 0 100.0 Cylindrospermopsis cuspis 10 23 12 14 96 0 76 0 87 0 56 7 86.8 Cylindrospermopsis raciborskii 1284 17 5148 97 4565 41 5051 25 4112 12 4032 38 99.0 Geitlerinema unigranulatum 8 0 6 0 0 0 0 0 9 0 5 0 100.0 Lyngbya sp. 6 0 0 0 5 0 6 0 12 0 6 0 100.0 Merismopedia punctata 1414 0 627 0 263 0 3006 0 1461 0 1354 0 100.0 Merismopedia tenuissima 20,859 48 20,907 34 17,794 0 15,059 5 12,916 0 17,507 17 99.9 Microcystis aeruginosa 692 53 1782 744 2958 66 3051 47 2183 33 2133 189 91.2 Microcystis viridis 0 0 0 0 0 0 0 0 493 0 99 0 100.0 Microcystis wesenbergii 250 0 0 0 562 0 0 0 318 0 226 0 100.0 Planktolyngbya circumcreta 162 0 75 0 416 47 439 0 219 0 262 9 96.4 Planktolyngbya limnetica 3645 0 2774 0 5060 0 1843 0 2220 0 3108 0 100.0 Planktolyngbya minor 13,843 0 16,883 0 12,610 0 11,736 0 7069 0 12,428 0 100.0 Planktothrix agardhii 2395 11 2903 224 1311 37 2181 66 1757 18 2109 71 96.6 Planktothrix lacustris var. solitaria 8723 0 6873 0 5299 0 2443 0 6130 0 5893 0 100.0 Pseudanabaena recta 13,446 0 15,838 0 18,835 0 9624 0 5058 0 12,560 0 100.0 Radiocystis fernandoi 3992 0 3395 0 2516 0 3089 0 1330 0 2864 0 100.0 Romeria elegans 0 0 20 0 8 0 1 0 10 0 8 0 100.0 Total (cell mL) 107,500 725 114,231 3055 107,485 1300 102,643 767 92,304 369 107,485 767 Removal (%) 99.3 97.3 98.8 99.3 99.6 99.3 *Data adapted from Addico et al. (2009). No n- co mm er cia l u se on ly G.N.D. Addico et al.100 Ta b. 5 . S um m ar y of M C a na ly se s fr om d if fe re nt tr ea tm en t s te ps o f t he fo ur G ha na ia n tr ea tm en t p la nt s. In tr ac el lu la r M C s (µ g L –1 ) D is so lv ed M C s (µ g L –1 ) (R es er vo ir T re at m en t N um be r of s am pl es C on ce nt ra ti on M ed ia n N um be r of s am pl es C on ce nt ra ti on M ed ia n (s am pl in g pe ri od ) s te p To ta l >L O D > 1 µg L –1 ra ng e T ot al > L O D > 1 µg L –1 r an ge B ar ek es e In ta ke 4 1 (2 5% ) 0 < L O D -0 .4 6 < L O D 3 0 0 < L O D < L O D (1 0 Ja n- 7 Ju n) F lo cc ul at io n 5 1 (2 0% ) 1 (2 0% ) < L O D -1 5. 50