The available freshwater in the African archipelago of Cape Verde (DMS coordinates 15°07′12.51″ N, 23°36′18.62″ W) does not cover its needs. In addition, overexploitation and saline intrusion can impair the quality of groundwater [1
]. Several water storage structures are being planned around the country in order to collect storm water needed to increase irrigated areas and modernize agriculture. However, these waters can carry nutrients of natural and anthropogenic origin, creating conditions for eutrophication and exponential growth of microalgae. These algal blooms have deleterious impacts on public health, water quality, and environmental issues, as well as economic costs due to bottom anoxia, release of noxious products, and toxic metabolites [2
]. Actually, these events are occurring more frequently worldwide, and it is thought that global climatic changes are a major contributor to this problem [4
To evaluate the risk of occurrence of algal blooms in freshwater bodies, it is important to characterize their microbial composition. In non-eutrophic freshwater systems, the most commonly abundant bacterial groups are, in order of decreasing relative abundance, Actinobacteria
(Alpha, Beta and Gamma clades), Verrucomicrobia,
]. In contrast, eutrophic waters contain microbial communities that include large numbers of cyanobacteria, which are able to produce toxins and foul odors and discolor the water [13
]. The cyanobacterial phylum has many genera that produce toxins, also called cyanotoxins, and in freshwater bodies, toxic and non-toxic strains can co-exist and dominate at different times [15
]. Poisoning with cyanotoxins occurs through consumption of contaminated food or water, or during aquatic recreational activities, causing many adverse symptoms like skin irritation, acute diarrhea, and liver and nervous tissue damage, leading to severe health problems, or death in humans, cattle, domestic animals, and wildlife [6
]. Hence, the risk of toxin production occurring increases the need for monitoring plans to prevent toxin-related impairments and costs.
Molecular-based methods combined with sequencing offer the ability to not only identify possible toxin producers but also target species-specific toxins, validating the presence or absence of toxin-related pathways. Nevertheless, DNA-based molecular methods cannot predict if toxins are being produced and released to the environment.
The leveraging of molecular methods provided by Next Generation Sequencing (NGS) allows researchers to gain new insights into microbial community structure in environment samples, identify new community members, and discover alternative bioindicators of water quality.
Environmental DNA samples are often difficult to purify to axenic conditions, so in vitro assembly of genomes recovered from the sequencing reads of metagenomes, MAGs, have been accomplished on several microbial community studies and environmental surveys [20
]. These genomics and metagenomic approaches enable, for instance, the reconstruction of genomes from unculturable samples or fast monitoring of cyanobacterial bloom formation [20
To determine the current potential of risk of toxin production in the Cape Verdean freshwater reservoirs, we performed NGS analysis of 16S rRNA gene amplicon sequences to identify microalgae and bacteria in the reservoirs. Our results show the presence of several cyanobacterial genera well known to produce toxins in all reservoirs. Furthermore, upon the detected dominance of one cyanobacterial strain in Faveta reservoir we decided to perform a metagenomic approach for identification of toxin related genes. We were able to reconstruct, for the first time, the full genome of a potentially toxic cyanobacterium from Cape Verde, based on the full metagenome sequencing data of Faveta reservoir. Analysis of this genome revealed the presence of genetic machinery used to synthesize cyanotoxins. The results of our biodiversity survey, phylogenetic analysis, and genome reconstruction lead us to conclude that toxin risk is a reality and a potential future threat in these reservoirs.
The studied reservoirs show three distinct microbial/microalgae community profiles, despite being located on the same island and in a radius of 15 kilometers from each other: in two sites, bacteria were dominant (Proteobacteria and Cyanobacteria), and in the other reservoir microalgae belonging to the cryptophytes were the most abundant taxa. The dominant species from one of the reservoirs was identified as a M. aeruginosa strain through phylogenetic studies, placing it closer to other strains collected in continental Africa. Microcystis spp. were detected in all three reservoirs, as well as other cyanobacteria known to bloom and produce cyanotoxins. Our analysis of the assembled M. cf. aeruginosa CV01 genome revealed that it can produce toxins, and therefore a potential risk of toxin production can exist in Cape Verdean reservoirs.
The absence of halogenase genes in aeruginosin and cyanopeptolin gene clusters is known to have no influence on the metabolite’s next biosynthesis steps [37
]; thus, it can be effectively synthesized by M.
CV01. It has been suggested that halogenase genes were present in an ancient form of cyanobacteria and were repeatedly lost in several lineages [38
]. Regarding the relatively low sequence identity found in the sequences similar to mcyA
detected in the genome, it suggests that M.
CV01 gene degeneration implies loss of function of these genes. The discovery of large insertions and deletions affecting this cluster suggest this scenario [39
]. The detection of selected microcystin and anatoxin genes in a metagenome survey does not imply by itself synthesis in Faveta reservoir, since it can belong to a pool of diverse organisms. Hence, the real toxic potential of these metabolites is difficult to determine, as the actual synthesis pathways are not fully known, and the diversity and plasticity of gene clusters in cyanobacteria allows for the production of a huge variety of analogous molecular structures and chemical activities. Therefore, further toxicological screenings need to be performed in order to fully understand these pathways.
Our microbial biodiversity survey revealed distinct community structures in the three reservoirs (Figure 2
). In fact, Proteobacteria
dominated Poilão reservoir, mainly bacteria from the Acinetobacter
genus, while in Saquinho reservoir microalgae from the Cryptomonas
genus were dominant, and in Faveta reservoir the cyanobacterial strain M.
CV01 was the most abundant. Despite being located on the same island and having common microbial groups, we found that each phyla’s quantitative distribution varied substantially between reservoirs (Figure 1
A). Nevertheless, the profiles from the reservoirs identify groups that are common in other water bodies studied worldwide despite the differences in relative quantities as is the case of Cyanobacteria
, and the three clades of Proteobacteria
]. Actually, the microbial profile of Poilão reservoir resembles those of lakes close to urban areas, where Acinetobacter
is the dominant genus [41
In each reservoir, one genus prevailed with a relative abundance above 50%: in Poilão it was Acinetobacter, in Saquinho it was Cryptomonas, and in Faveta the Microcystis genus.
The analysis of the local diversity indices of the replicates from each reservoir revealed consistency and reinforced the sites’ distinct microbial communities (Figure 2
A–C). The indices also showed that microbial communities presented different dominant genera in each of the reservoirs, as well as abundance of different taxa in all sites as typically found in freshwater bodies around the world [7
Besides the operational starting date differences between the reservoirs and no physical communication between lakes, abiotic factors specific for each site might be involved in the dominance variations within the microbe communities.
The dominance of one cyanobacterial strain in Faveta allowed us to assemble and fully study its genome, and to identify genes, allowing the reconstruction of toxin pathways and assessing the toxin risk inherent in this specific strain. Exploration of the assembled genome also revealed genomic features in common with other M. aeruginosa
genomes (Table 1
). Some phage genes were found integrated in the genome of M.
. Myoviridae “photosynthetic” freshwater cyanophages (Ma-LMM01 and MaMV-DC) were also found. These genes are thought to play an important role during phage infection by supplementing the host with the production of photosynthesis proteins, a process that can be also beneficial to the host during the infection process, as suggested by some authors [31
]. These horizontal gene transfer events are shaping the genome architecture of the Microcystis
genus, providing a supplementary advantage that can be important during cyanobacteria blooms. A region containing chlorophyll a
apoproteins A1 and A2 synthesis genes was also identified, but since these are single copy genes located near transposase sequences in this new genome, they were probably misidentified as having phage origin.
We identified four NRPS/PKS gene clusters that could synthesize potentially toxic metabolites: three well-known metabolites (aeruginosin, cyanopeptolin, and microviridin, represented in Figure 3
A) and another metabolite from one yet unknown gene cluster.
Phylogenetic markers placed the Cape Verdean strain among others from freshwater bodies from Africa, albeit Cape Verde being a distant archipelago from the continental Africa. The identification in all reservoirs of other cyanobacterial genera known to be toxin producers like Phormidium, Planktothrix, and Cylindrospermopsis increases the potential risk of toxin production. Other studies in African water bodies have identified these and other potentially cyanotoxin producers, raising the possibility of future occurrence of other cyanobacterial genera in Cape Verdean freshwater reservoirs.
Cyanobacterial blooms occur in freshwater reservoirs distributed worldwide where M. aeruginosa
is one of the most frequently detected species. Actually, many long-term studies have reported toxic blooms in lakes and rivers from Kenya, Uganda, Senegal, Morocco, and South Africa [44
], often dominated by M. aeruginosa
, as we also detected on the island of Santiago. Moreover, a recent review on the occurrence of cyanobacterial blooms in Africa [51
] shows that there is limited information from western African countries, including Cape Verde, exposing the need for further studies in countries were water quality is threatened and scarce. Therefore, our study increases the available information on cyanobacterial communities described for the western African region. The scarcity of renewable freshwater resources of archipelagic states like Cape Verde is aggravated by the terrain that favors torrential water flows and strong anthropogenic pressures on the environment leading to eutrophication of its freshwater bodies and increased risk of toxic algal blooms.
The main threat and concern from our analysis was the identification of a bacterial community dominated by M.
CV01, which signals the possibility of toxic blooms in Cape Verdean reservoirs, since exponential growth is typical of this species. Alongside cyanobacteria, Enterobacteriaceae
family members were also detected, although they are considered to be transient of the bacterial community in freshwaters, as they have anthropogenic and zoonotic origin [8
The occurrence of blooms and toxin production are potential life-threatening risks to public health, so monitoring plans are very important. The costs involved in these control and containment strategies can be prohibitive, especially for low and middle-income countries. Hence, approaches like the one proposed in this work, which enabled the identification of potentially toxic cyanobacterial genera through 16S rRNA gene markers, could be an interesting alternative, without the time-consuming and expertise-dependent microscope identification of toxin-producing organisms or mass spectrometry-base identification of toxins. NGS is still not widely available, but DNA sampling kits are easy to use and can be sent to sequencing facilities at cost-effective prices. Other more sophisticated technics are possible such as lab on a chip, mass spectrometry, or even portable NGS devices, which can be adapted to use our workflow in the field, but if a simple molecular lab is available, a PCR assay could also be efficient at detecting the presence of specific putative cyanotoxin genes. These strategies can alert authorities and populations before bloom formation and toxin production.
Cyanobacteria are currently being developed and used for bio-production of metabolites and biomass in algal farms, for example, in the production of human dietary supplements, fertilizers, animal feed or biodiesel, to name a few uses [52
]. Cape Verde has little land suitable for agriculture, but it has temperature and light conditions that are speculated to be suitable for simple bioreactors for biomass production using photosynthetic microorganisms like green algae or cyanobacteria. The observation of spontaneous blooms of cyanobacteria in freshwater reservoirs lends credibility to this hypothesis, and opens the way for a new productive industry in the archipelago.
The present work identifies the existence of real risk for cyanotoxin production in Cape Verdean freshwater reservoirs. Similar structures are planned, which will also need to be studied and monitored. Future work should include studies on the dynamics of the local microbial communities, as well as characterize how environmental factors are affecting their organization, in order to predict and control the impact of water impairment and toxin production on public health and on the economy. In this study, we made use of many freely available open source tools, which represent, to our knowledge, innovative research strategies in Cape Verde. This study will open the way for further research on microbial biodiversity and other genomic studies in the archipelago, and raise questions relevant for different areas of research and application.