Unexpected High Diversity of Terrestrial Cyanobacteria from the Campus of the University of the Ryukyus, Okinawa, Japan

Terrestrial cyanobacterial strains were isolated from the Nishihara campus of the University of the Ryukyus, Okinawa, Japan. The 13 sampling sites were distributed in a 200 m radius and appeared as dry, blackened stains. From these small areas, 143 cyanobacterial strains were established. The strains were divided into five morphotypes, including unicells, unicells with baeocytes, non-branching filaments, false-branching filaments, and heterocystous strains. From the strains, 105 partial 16S rRNA gene sequences were obtained and could be classified into 30 generic types. Among them, 22 unique strains and over 1100 bps of data were selected for further phylogenetic analyses. These sequences were positioned into six main clades corresponding to cyanobacterial orders: Nostocales, Chroococidiopsidales, Chroococcales, Oscillatoriales, Pleurocapsales, and Synechococcales. Almost all sequences had no identical matching data in GenBank and many of them had no closely related data. These data suggest that the terrestrial cyanobacteria are very divese even within close sampling areas, such as within the campus of the University of the Ryukyus. The established strains are not only important for classification of terrestrial cyanobacteria but also for possible application studies in the future.


Introduction
Many traditional and even modern buildings are made with white limestone, mortar, and concrete, and such buildings often have black stains, particularly in tropical and subtropical countries. These black stains are sometimes positively considered as atmospheric of past eras for historical monuments, but these stains are usually dealt with as nuisances. These black stains are caused by terrestrial cyanobacterial growth, and many floral studies of such terrestrial cyanobacteria have been reported from many ecosystems in European temperate and Mediterranean regions [1][2][3], India [4][5][6][7], the Middle East [8], Australia [9], China [10], North America [11], Middle and South America including Mexico and Brazil [12,13], the Hawaiian Islands [14], and even from locations with extreme conditions such as Antarctica [15], Arctic Greenland [16], the Alps [17], and Africa [18].
Scientific reports on cyanobacteria in urban areas have appeared in the past few decades. Such black stains have shown to lower values of real estate in Taiwan [19]. Repeated strong radiation and heavy showers give strong light, high temperature, high humidity, and dryness, specific primers CYA106F [38]-CYA1371R(1+2+3) [40] for the remaining strains. PCR cycles were set to 94 • C for 5 min, followed by 30 cycles of 94 • C denaturing for 1 min, 60 • C annealing for 1 min, and 72 • C extension for 1 min, with an additional 72 • C extension for 7 min, and an ending hold at 4 • C. The amplified DNA fragments were checked by electrophoresis on a 1% agarose gel with ethidium bromide staining. Amplified PCR products were sent to Macrogen Japan Co. for sequencing.
For phylogenetic analyses, additional data sequences were obtained from NCBI GenBank and used to construct the phylogenetic tree. All sequences were aligned and analysed using MEGA 7.0 [41]. A maximum likelihood phylogenetic tree based on partial 16S rRNA gene sequences was constructed. The General Time Reversible-parameter model (GTR + G + I) was used and the outgroups were selected as Gloeobacter kilaueensis (NR_121745) and Gloeobacter violaceus PCC 7421 (NR_074282). Bootstrap values were tested 1000 times using the rapid bootstrap option (>50%). amplified DNA fragments were checked by electrophoresis on a 1% agarose gel with ethidium bromide staining. Amplified PCR products were sent to Macrogen Japan Co. for sequencing. For phylogenetic analyses, additional data sequences were obtained from NCBI GenBank and used to construct the phylogenetic tree. All sequences were aligned and analysed using MEGA 7.0 [41]. A maximum likelihood phylogenetic tree based on partial 16S rRNA gene sequences was constructed. The General Time Reversible-parameter model (GTR + G + I) was used and the outgroups were selected as Gloeobacter kilaueensis (NR_121745) and Gloeobacter violaceus PCC 7421 (NR_074282). Bootstrap values were tested 1000 times using the rapid bootstrap option (>50%).

Results and Discussion
Thirteen samples of mats or scraped Melamine sponges of terrestrial cyanobacteria were collected from within the sampling area (Figure 1), directly obserbed by light microscopy, and precultured for establishing culture strains of terrestrial cyanobacteria. Sampling sites were selected as blackened parts of walls, monument stones, and concrete buildings ( Figure 2). Sampling sites were usually dry but the sites had tendencies to be routes for water or became small water puddles or pools during and after rainfall. Sampling sites were also slightly shaded and not under full sunlight. Direct light microscopic observations of the natural samples indicated that the dominant constituents of the samples were Gloeocapsa and related chroococcalean types from scraped Melamine sponges, whereas Scytonema and related heterocystous filamentous types were dominant in the cyanobacterial mats. After pre-culture, the growing cyanobacteria were isolated by pipette washing method or agar plating to establish culture strains. We established 143 culture strains that belonged to almost all

Results and Discussion
Thirteen samples of mats or scraped Melamine sponges of terrestrial cyanobacteria were collected from within the sampling area (Figure 1), directly obserbed by light microscopy, and pre-cultured for establishing culture strains of terrestrial cyanobacteria. Sampling sites were selected as blackened parts of walls, monument stones, and concrete buildings ( Figure 2). Sampling sites were usually dry but the sites had tendencies to be routes for water or became small water puddles or pools during and after rainfall. Sampling sites were also slightly shaded and not under full sunlight. Direct light microscopic observations of the natural samples indicated that the dominant constituents of the samples were Gloeocapsa and related chroococcalean types from scraped Melamine sponges, whereas Scytonema and related heterocystous filamentous types were dominant in the cyanobacterial mats. After pre-culture, the growing cyanobacteria were isolated by pipette washing method or agar plating to establish culture strains. We established 143 culture strains that belonged to almost all cyanobacterial types, such as unicells, unicells with baeocytes, non-branching filaments, false-branching filaments, and heterocystous strains ( Figure 3). From the strains, 16S rRNA gene sequences from morphologically different strains were analysed to obtain 105 sequences belonging to 30 different genetic types. In this study, we eliminated closely related and relatively short (less than 1100 bp) sequences, and finally, 22 strains were selected for diversity evaluation. In this paper, we used 16S rRNA gene data from Komárek et al. [42] and BLAST [43] as a reference for grouping sequences to identify closely related strains (e.g., to at least genus level). Finally, morphological comparisons were done to confirm the identity of strains using Cyanoprokaryota monographs [33][34][35]. cyanobacterial types, such as unicells, unicells with baeocytes, non-branching filaments, falsebranching filaments, and heterocystous strains ( Figure 3). From the strains, 16S rRNA gene sequences from morphologically different strains were analysed to obtain 105 sequences belonging to 30 different genetic types. In this study, we eliminated closely related and relatively short (less than 1100 bp) sequences, and finally, 22 strains were selected for diversity evaluation. In this paper, we used 16S rRNA gene data from Komárek et al. [42] and BLAST [43] as a reference for grouping sequences to identify closely related strains (e.g., to at least genus level). Finally, morphological comparisons were done to confirm the identity of strains using Cyanoprokaryota monographs [33][34][35].

Phylogenetic Analyses
Twenty-two strains (=sequences) approximately 1100 bp in length were selected and utilized in phylogenetic analyses for diversity evaluation of terrestrial cyanobacterial strains. Seventy-one 16S rRNA sequences were obtained from NCBI GenBank from the data of Komárek et al. [42] and from BLAST similarity searches, and were used to construct the phylogenetic tree. We constructed a maximum likelihood (ML) phylogenetic tree with a total of 105 sequences. Bootstrap values were tested 1000 times using the rapid bootstrap option (Figure 4). The resulting phylogenetic tree was divided into six main clades corresponding to Nostocales, Chroococcidiopsidales, Chroococcales, Oscillatoriales, Pleurocapsales, and Synechococcales. Some strains were closely related with GenBank data; e.g., Ryu2-7DN(D3) was closely related with Hapalosiphon welwitschii AY034793, Ryu4-7 was closely related with Chroococcidiopsis thermalis PCC7203 (CP003597), and Ryu5-15d was closely related with Gloeocapsopsis sp. CENA327 (KT731148). Also, many strains were identifiable to the generic levels, some within the same genus; e.g., Ru1-6 and Ru1-3d belonged to the genus Nostoc, Ryu1A1(C3) belonged to Brasilonema, Ryu1-11 belonged to Scytonema, and Ru4-3d belonged to Nodosilinea. On the other hand, other strains had no close relatives in GenBank, such as strains Ru3-14, Ryu8-6, and Ryu1-3.

Phylogenetic Analyses
Twenty-two strains (=sequences) approximately 1100 bp in length were selected and utilized in phylogenetic analyses for diversity evaluation of terrestrial cyanobacterial strains. Seventy-one 16S rRNA sequences were obtained from NCBI GenBank from the data of Komárek et al. [42] and from BLAST similarity searches, and were used to construct the phylogenetic tree. We constructed a maximum likelihood (ML) phylogenetic tree with a total of 105 sequences. Bootstrap values were tested 1000 times using the rapid bootstrap option (Figure 4). The resulting phylogenetic tree was divided into six main clades corresponding to Nostocales, Chroococcidiopsidales, Chroococcales, Oscillatoriales, Pleurocapsales, and Synechococcales. Some strains were closely related with GenBank data; e.g., Ryu2-7DN(D3) was closely related with Hapalosiphon welwitschii AY034793, Ryu4-7 was closely related with Chroococcidiopsis thermalis PCC7203 (CP003597), and Ryu5-15d was closely related with Gloeocapsopsis sp. CENA327 (KT731148). Also, many strains were identifiable to the generic levels, some within the same genus; e.g., Ru1-6 and Ru1-3d belonged to the genus Nostoc, Ryu1A1(C3) belonged to Brasilonema, Ryu1-11 belonged to Scytonema, and Ru4-3d belonged to Nodosilinea. On the other hand, other strains had no close relatives in GenBank, such as strains Ru3-14, Ryu8-6, and Ryu1-3.

Morphological Comparisons between Molecular Phylogeneticaly Related Strains
All strains require further detailed studies for formal characterization.

Nostocales
Ru1-3d: Nostoc sp. (Figure 3A, collection point 1) This strain had heterocysts, with the heterocysts presented at the end of the filament having a rounded, cylindrical, and somewhat conical shape, and the intervening heterocysts being spherical. The diameter of cells was relatively equal, about 4 µm wide, and only one filament was covered with a mucilaginous sheath. We observed that there were two or more splitting faces, with a part of the cells splitting at the face perpendicular to the filamentous body, and the filamentous body had a meandering shape. This Nostoc strain was morphologically related with isolates from Cycas revoluta and particularly close to N. punctiforme or Nostoc sp. [35]. Ru1-6: Nostoc sp. (Figure 3B, collection point 1) This strain also consisted of heterocysts and was intervening. The diameters of cells were sometimes slightly irregular, 4.1-4.7 µm wide, in young trichomes or, in hormogonia, 6-9.1 µm. Several filamentous bodies were wrapped in a spherical colony covered with a mucilaginous sheath. There were two splitting faces, and some of the cells were in two rows. The cells were olive in color and contained granules inside. Akinetes were not observed. This strain was closely related with Nostoc edaphicum Kondrateva, 1962 [35].
Ryu2-7DN(D3): Hapalosiphon sp. (Figure 3C, collection point 6) The strain Ryu2-7DN(D3) ( Figure 3C) was genetically closely related with Hapalosiphonwelwitshchii AY034793 (Figure 4). This strain had heterocysts and two cell division faces, and the filament were true branches. The branches were almost at right angles to the stem. The stem cells were barrel-shaped, 4.8-5.9 µm width × 5.9-8.5 µm length, and the branch cells were an elongated rod shape. The cells contained granules. Akinetes were continuously produced. In some cases, most of the cells of a filament were akinetes. These akinetes were 5.5-8.0 µm width × 5.3-6.4 µm length. These morphological characters were close to those for Hapalosiphon welwitschii W. et G. S. West, 1897, but the cell sizes of the strain were slightly larger than those of Hapalosiphon welwitschii. Also, Hapalosiphon welwitschii inhabits freshwater streams, different from the strain observed here. Thus, we identified strain Ryu2-7DN(D3) as Hapaloshiphon sp.
Ryu5-18(F2): Tolypothrix sp. ( Figure 3D, collection point 9) This strain was related with Calothrix deseritica PCC 7102 (KM019960) (Figure 4) but the character of having no hairy ends of filaments did not fit with the genus Calothrix. In this strain, a heterocyst appeared at the tip of the filament. Filaments had slight narrowing to the tip and no hairy ends, but had false branching, with one filament in one sheath included. Each filament did not form colonies and was isolated. Trichomes were 5-11 µm wide, and dark blue-green or olive-green in coloration. Due to morphological similarity, we identified strain as Tolypothrix sp.
Ryu1-11: Scytonema sp. 1 ( Figure 3E, collection point 5) Heterocysts were confirmed at both the head and stem parts. Most of the vegetative cells were blue-green, 6.3-9.5 × 10-20 µm wide, but several heterocysts were pale yellow. The filament had a thin sheath. The strain certainly belonged to the genus Scytonema by molecular phylogenetic and morphological results. Ryu1A1(C3): Brasilonema sp. 1 ( Figure 3F, collection point 5) The filament had heterocysts and was mediate. Many filaments were singularly false-branching. Most of the cells were yellowish brown, but several cells were pale white, and 6-9 × 10-20 µm wide. The genus Brasilonema has been established by separating from the genus Scytonema, based mainly on molecular phylogenetic analyses [43]. Hence, the current strain should be carefully compared with known Scytonema species that as of yet have no 16S rRNA gene sequences available publically. and there were some distances between other cells. The cell divisions were probably on three faces. After division, the cells grew to their original size and shape and then proceeded to the next division. Morphologically the strains were closely related with Aphanocapsa parietina Nageli 1849 [33], but there is no authentic data for this genus in GenBank.

Oscillatoriales
Ru5-34: Phormidium sp. (Figure 3N, collection point 4) This strain was closely related with Oscillatoria acuminata PCC 6304 (CP003607) (Figure 4). The filamentous body did not have a heterocyst or calyptora. Each filament did not have a sheath. Each cell had a rectangular shape with a short side parallel to the filamentous body, and the tip cells at the end were narrowing and bent, 1.9-3.2 µm long × 5.5-7.5 µm wide. In addition, cells had gliding motility with active rotation. These characters agree well with many morphologically related species in Phormidium [34], and further comparative studies are required.

Pleurocapsales
Ru3-1ND: Chroococcopsis sp. (Figure 3O, collection point 2) Cells were spherical, elliptical, polygonal, etc., and varied in size. Additionally, the cells showed polarity. There were numerous cell division faces; growth by endospores (baeocytes), and their parental sheaths did not remain after divisions. The cells were yellowish brown, and 15-18 µm in diameter. These morphological characters agree with Chroococcopsis [33], but there is no authentic data for this genus in GenBank.

Synechococcales
Ru3-34: Pseudophormidium sp. (Figure 3R, collection point 2) Filaments were solitary or in small groups, composed of thin, colourless sheaths, which were sometimes slightly telescopically widened at the ends. False branchings were not very common. Trichomes were not constricted at cross-walls and were slightly attenuated towards the ends, and rare, screw-like coiled sheaths were observed. Cells were blue-green, shorter in length than width, 1.2-1.6 µm long × 2.3-2.6 µm wide. Pseudophormidium batrachospermi (Starmach) Anagnostidis et Komarek 1988 was morphologically related [34] but there are no data for this species in GenBank. Ryu1-2: Leptolyngbya sp. ( Figure 3S, collection point 5) In this strain, short filaments were densely entangled to form a thin mat that adhered to the substrate. In addition, grown colonies had a bundle shape with filiforms tangled. The sheath was thin, colorless, and adhered to the filamentous body. The filament was slightly curved, resulting in many filaments being tangled. False branching was not abundant. Cells were blue-green and clearly constricted at the cross walls. Cells at the terminal were rounded, the width of the cells was 1.9 ± 0.1 µm, and the length was half to the same as the width, 1-2.5 µm. From these morphological features, the strain was thought to be closely related with Leptolyngbya henningsii (Lemmermann) Angnostidis 2001 [34].
Ru5-44: Leptolyngbya sp. (Figure 3T, collection point 4) The Ru5-44 strain had a bundle shape with long, tangled colonies that did not adhere to the substrate. The filament was straight, slightly wavy or coil-like, and frequently branched. The sheath was colorless, thin, and adhered to the cell. The cells were light bluish green, and the cells were not constricted. Terminal cells were rounded at the ends but never pointed apart. The width of cells was 2.2-2.8 µm, which was slightly shorter than the normal width [34].
Ru4-3d: Nodosilinea sp. (Figure 3V, collection point 3) The filamentous body did not have ant heterocysts, and no calyptra was confirmed. The sequence was sister to that of Nodosilinea epilitheca Perkerson & Casamatta [44] (Figure 4), and morphologically the strain agreed well with this genus [42], but further studies are required to confirm this.
In conclusion, these results and data suggest that terrestrial cyanobacteria is very diverse in Okinawa, even within a small area such as inside the University of the Ryukyus campus. In this study, we suceeded in establishing 143 strains from 13 samples. A total of 105 partial 16S rRNA gene sequences were obtained, and they were divided into 30 generic types of six main cyanobacterial order clades [40]. This means that the terrestrial cyanobacterial diversity at the University of the Ryukyus campus is more diverse than in locations from other surveys, including in India [4][5][6][7], South Korea [21,22], the main Haiwaiian islands [14], Brazil, and Mexico [12,13].
Many archaeologically important stone temples and mortar monuments with artistic structure, as well as the surfaces of painted buildings, are disfigured by changed colour surfaces and caused aesthetic and structural damage due to the colonization of cyanobacterial biofilms [7,43,45]. The diversity distribution of particular cyanobacteria on monuments in urban regions are thought to be related mostly to environmental conditions of the climate, such as water availability and air circulation, and to the position and orientation of the hard surface [43]. Correlations among other environmental factors have been tested to identify other factors that could substitute for the effects of humidity and light intensity [22]. To examine such environmental factors, it is likely that selecting a few sampling points in the Nishihara campus of the University of the Ryukyus should be enough due to the very high diversity of terrestorial cyanobacteria at this location.
In this study, 22 selected strains were used for phylogenetic analyses. From the results, no identical sequence data were present in GenBank, and surprisingly, many strains' sequences had no closely related data. Among these strains, some could be identified to generic level by morphology, and hence it is required to study in detail both the molecular phylogeny and morphology of each strain. Finally, we successfully established a very diverse, 143 terrestrial cyanobacterial strain collection, and the collection may contain useful strains for applications in the future.

Conclusions
We collected terrestrial cyanobacteria from blackened parts of walls, monument stones, and concrete buildings at a total of 13 sampling sites in the Nishihara campus at the University of the Ryukyus (Figures 1 and 2). The dominant constituents of the samples were Gloeocapsa and related chroococcalean types from scraped Melamine sponges, whereas Scytonema and related heterocystous filamentous types were dominant in the cyanobacterial mats. We used BG11 and BG11-N agar plating to attempt to establish culture strains from all 143 isolated strains belonging to almost all cyanobacterial types (Figure 3). From morphologically different strains, 16S rRNA gene sequences were analysed to obtain 105 sequences belonging to 30 different genetic types. Twenty-two strains (=sequences) of approximately 1100 bp in length were selected and utilized in phylogenetic analyses for a diversity evaluation of terrestrial cyanobacterial strains with 71 16S rRNA sequences obtained from NCBI GenBank/BLAST and the data of Komárek et al. [42]. The resulting phylogenetic tree was divided into six main clades corresponding to Nostocales, Chroococcidiopsidales, Chroococcales, Oscillatoriales, Pleurocapsales, and Synechococcales. Also, many strains were identifiable to the generic level, with some within the same genus; e.g., Ru1-6 and Ru1-3d, Ryu1A1(C3), Ryu1-11, and Ru4-3d. On the other hand, other strains had no close relatives in GenBank, such as strains Ru3-14, Ryu8-6, and Ryu1-3. These results provide useful baseline diversity data for research in the future.