Phylogenetic diversity of Actinobacteria from Momela soda lakes, Arusha National Park, Tanzania

The Momela soda lakes consist of seven small, hypersaline, alkaline lakes, situated in the East African rift valley at Arusha National Park, Tanzania. The lakes are fed by separate underground water sources with slightly varying mineral contents resulting in colour variation and supporting different kinds of prokaryotic and eukaryotic species. In this study, the diversity of Actinobacteria in surface water and sediments of five Lakes were investigated using culture-dependent and culture-independent molecular techniques. A total of 34 out of 112, and 13 out of 85, representatives of Actinobacteria isolates and clones, respectively, were selected for gene sequencing using the CD-HIT program. Analysis of their 16S rRNA gene sequences displayed the presence of species affiliated to 15 different genera, namely Mycobacterium, Rhodococcus, Microbacterium, Isoptericola, Dietzia, Leucobacter, Jonesia, Nesterenkonia, Micrococcus, Streptomyces, Hoyosella, Norcadiopsis, Cellulomonas, Bogoriella, and Agromyces. The results showed 5 and 12 putative new Actinobacteria isolates and clones, respectively. This is the first report of isolation of bacteria from the genus Mycobacterium from a soda lake globally, as well as the genera Hoyosella, Isoptericola, Jonesia, Micrococcus, Leucobacter and Agromyces from a soda lake in East Africa. Because Actinobacteria are known as a source of biotechnologically important compounds, the species revealed set a platform to search for novel bioactive compounds.

Actinobacteria are gram-positive bacteria with high guanine and cytosine (G + C) content in their genomes, and are ubiquitously distributed in different ecosystems. Some Actinobacteria are known to have similarities with fungi as they exhibit some fungal characteristics, but have some features that appear bacterial. However, as they have more features that distinguish them from fungi, they are classified into the Bacteria domain (Das et al. 2008;Chaudhary et al. 2013). Ecologically, Actinobacteria are widely distributed in nature (Barka et al. 2016) and have survived and thrived in extreme environments (Sultanpuram et al. 2015). These include groups of acidophilic, alkaliphilic, psychrophilic, thermophilic, xerophilic, halophilic and haloalkaliphilic Actinobacteria (Al-Tai and Ruan 1994). These kinds of Actinobacteria have the potential to produce various commercially important bioactive compounds (Subramani and Aalbersberg 2012).
In recent years, the quest for novel compounds with industrial and medicinal potential, including, but not limited to, antimicrobial compounds and enzymes, has led to an extensive tapping of Actinobacteria from unexplored extreme micro-habitats. Accordingly, large groups of culturable and non-culturable novel haloalkaliphilic Actinobacteria of various genera have been discovered from soda lakes around the world, and their evolutionary relationships have been established (Learn-Han et al. 2012;Luo et al. 2013). In East Africa, most studies have been done in Kenya's soda lakes, including a report by Mwirichia et al. (2009) who reported three Actinobacteria species, namely Streptomyces, Microbacterium, and Norcadia from Lake Elmenteita, Kenya. Four Actinobacteria species (Rhodococcus, Dietzia, Microbacterium and Nocardia) were reported from Lake Magadi, Kenya (Ronoh et al. 2013). Alkaliphilic Actinobacteria Bogoriella caseilytica (Groth et al. 1997) and Cellulomonas bogoriensis (Jones et al. 2005) have been isolated from Lake Bogoria, Kenya. A novel Streptomyces species has been isolated from Lake Nakuru (Solingen et al. 2001) and a new member of the genus Dietzia named Dietzia natronolimnaios was isolated from Lake Oloidien, Kenya (Duckworth et al. 1998). In Tanzania, the only report on Actinobacteria is of a novel species named Nesterenkonia natronophila, which was isolated from Lake Magadi (Machin et al. 2019). Therefore, comparatively little is known about the occurrence and diversity of Actinobacteria in Tanzanian soda lakes.
The Momela soda lakes situated in Arusha National Park, northern Tanzania, are among the East African rift valley's saline-alkaline lakes with high carbonate salt concentrations and pH ranging from pH 9 to 12 (Hamis et al. 2017). These characteristics make them extreme and
The conventional dilution plate technique was employed: 1 g of the soil sample or 1 ml of water samples were suspended in 9 ml of starch casein broth in a sterile test tube and serially diluted to 10 −6 , then 0.1 ml from each dilution was spread on SCA plates and incubated at 28 °C. Observation of the cultures' growth was done on the 5th, 7th, 14th and 19th days. The developed colonies were picked randomly, based on colony morphology from selected dilution plates and subcultured to get pure isolates. Morphological characterisation of colonies was macroscopically based: shape, aerial mass colour, substrate colour and texture of the colonies were noted and compared with literature. Standard Gram-staining coupled with microscopy was also done.
In this study, sequencing was the confirmatory technique for identification through the use of specific PCR primers. Pure cultures of Actinobacteria were maintained on starchcasein agar plates at 4 °C.

Extraction of genomic DNA from pure culture
For DNA extraction, approximately 0.8 g of pure isolate was scraped from the SCA plate and inoculated into 30 ml of sterile nutrient broth and incubated at 28 °C overnight on a shaker to obtain pure fresh cultures of Actinobacteria. The freshly cultured Actinobacteria isolates were centrifuged at 10 000 × g in a 1.5 ml microcentrifuge tube until pellets of 200 mg were obtained. Thereafter, DNA was extracted and purified, using Quick-DNA™ Fungal/Bacterial Miniprepkit as per manufacturer's protocol (Cat No. D6005, Zymoresearch Corp. USA). The purified DNA sample was stored at −20 °C.

Extraction of genomic DNA from environmental samples
DNA was directly extracted and purified from a total of 10 environmental samples (5 samples of sediment and 5 samples of water, i.e. one sample from each lake) using ZymoBIOMICS™ DNA Miniprep kit according to manufacturer's protocol (Cat No. D4304, Zymoresearch Corp. USA). Soil samples were directly subjected to DNA extraction. Before DNA extraction was performed on the water samples, the water samples were first filtered using a 47 mm membrane filter with a pore size of 0.45 µm (HAWG047S1, Millipore Corporation, Billeria, MA 01821) placed within a sterile stainless steel vacuum filter holder (Sartorius, Germany) attached to a water aspirator (FBL, China). After filtration, the membrane filter containing the retained microbes was cut into small pieces, and subsequently processed for DNA extraction. The purities and concentrations (in ng µl −1 ) of the extracted DNA were determined spectrophotometrically using the NanoDrop™ One Microvolume UV-Vis Spectrophotometer (Thermo Fisher Scientific, USA); the purity and quantity of DNA were within the recommended ranges for PCR (El-Ashram et al. 2016).
PCR cycling parameters were as follows: initial denaturation at 94 °C for 4 min, followed by 35 cycles of denaturation at 94 °C for 30 s, annealing at 60 °C for 50 s, and extension at 68 °C for 1 min. A final extension was performed at 68 °C for 5 min. The reaction mixture was then held at 4 °C. The integrity of amplification products was determined through electrophoresis on 1.2% agarose gel (Electran, VWR International Ltd), stained with EZ vision ® Bluelight DNA Dye and viewed under an ultraviolet transilluminator. The PCR products of the extracted DNA from the water and soil samples were sequenced at a commercial genomics facility in South Africa (Inqaba Biotec Ltd).

Cloning and nucleotide sequencing of the PCR products
The PCR products of the environmental samples were first cloned using the NEB ® PCR cloning kit, as per the manufacturer's protocol (Cat No. 1202, New England BioLabs ® Inc.). The cloned sequences were then re-amplified from the vector and purified (together with PCR amplicons of pure cultures), as per ExoSAP protocol. The purified PCR amplicons with sizes of approximately 640 bp in length were sequenced using BrilliantDye™ Terminator Cycle Sequencing Kit V3.1 BRD3-100/1000 (Nimagen) according to the manufacturer's instructions and afterwards injected on the Applied Biosystems™ ABI 3500xL Genetic Analyser (Cat No. 4406016,ThermoFisher Scientific).

Bioinformatic analysis of generated 16S rRNA gene sequences
The integrity of the cloned nucleotide sequences was checked, whereby the partial sequences were manually edited using Chromas Version 2.6.6 (Technelysium Pty Ltd., Australia) . Superfluous vector sequences were identified using sequence analysis tools on National Centre for Biotechnology Information (NCBI) website, namely the Basic Alignment Search Tool (BLAST) and VecScreen, and removed using BioEdit Sequence Alignment Editor version 7.0.5.3 (Tom Hall © ). To obtain representative culture and clone 16S rRNA gene sequences, the cleaned 16S rRNA gene sequences were aligned and their similarity compared using the CD-HIT (Cluster Database at High Identity with Tolerance) program (Huang et al. 2010b).
The resulting representative of culture and clone 16S rRNA gene sequences were compared for similarities with other known sequences in the sequence databases,  Tanzania, showing sampling stations at the Momela soda lakes (Lake Small Momela, Lake Big Momela, Lake Rishateni, Lake Lekandiro and Lake Tulusia) using BLAST at NCBI (http://www.ncbi.nlm.nih.gov). In all cases, one to two closest match sequences were considered for additional alignment analysis. Accordingly, the alignment of the sequences was performed by using AliView software version 1.26 (Uppsalla Universitet, Sweden) through MUSCLE (Larsson 2014). Phylogeny tree construction was done by using Evolutionary Genetics Analysis version X (MEGAX) Version 10.0.5 using a Tamura-Nei as a substitution model, maximum likelihood as a statistical method, the nearest-neighbour interchange as a maximum likelihood Heuristic Method tree inference option (Kumar et al. 2018). The resultant rooted tree topologies were re-evaluated by Bootstrap analysis with 1 000 resamplings of data and, in all trees, Bacillus lentus was used as an outgroup to position the root of the trees. All the sequences used in the study were deposited in the GenBank database and their accession numbers were retrieved.

Physico-chemical parameters
The physico-chemical parameters (water temperature, pH, salinity and conductivity) values recorded during sampling had minor variations from one lake to another. The pH and water temperature had a small variation across all lakes and ranged from pH of 9.20 ± 0.02 at Lekandiro to 10.30 ± 0.00 at Big Momela; a temperature of 21.00 ± 0.01 °C at Lekandiro to 24.40 ± 0.02 °C at Small Momela. The high pH values in the Momela Lakes were attributed to the carbonates within the soda lakes, as previously suggested in various studies (e.g. Hamisi et al. 2017). The Momela lakes' water temperatures were also within the range reported from previous studies (Hamisi et al. 2017;Nonga et al. 2017). Nonga et al. (2017) outlined that the lakes' water temperature might be low because the lakes are relatively cavernous, located at elevated terrain near Mount Meru (with a height of 4 562.13 m), and encircled with very thick forests. The recorded water temperature levels were favourable for mesophilic microorganisms' growth and maintenance (Nyakeri et al. 2018).
The values of salinity and conductivity showed a high variation across the lakes. As such, salinity ranged from 34.00 ± 0.00 practical salinity units (psu) at Small Momela to 46.00 ± 0.00 psu at Big Momela, and conductivity ranged from 7.51 ± 0.05 mS cm −1 at Lekandiro to 21.30 ± 0.02 mS cm −1 at Big Momela Lake. The salinity and conductivity were significantly and positively correlated (r = 0.979, p = 0.0037), implying that the amount of salts dissolved in Momela lakes' water moves in tandem with the positively and negatively charged ions formed. However, the salinity and pH values of Big Momela and Tulusia recorded in this study are relatively higher than those previously reported (Hamisi et al. 2017;Nonga et al. 2017). Such differences can be attributed to sampling period variations. In this study, sampling was done during the dry season, during which there is increased water evaporation resulting in carbonate precipitation, which then contributes to the observed higher values of pH and salinity (Nonga et al. 2017). Salinity and pH are thought to be predominant stress factors that limit the microbial diversity in extreme environments, such as soda lakes (Lanzén et al. 2013). Consistently, the recorded high values of salinity and pH suggest that microorganisms, including Actinobacteria in Momela lakes are haloalkaliphiles (Zvyagintsev et al. 2008;Chavan et al. 2013).
Occurrence and phylogenetic analysis of Actinobacteria isolates DNA was extracted from a total of 117 out of 120 isolated strains of Actinobacteria. All isolated strains were phenotypically considered Actinobacteria based on colony characteristics and were all Gram-positive bacteria. Upon sequencing of the PCR products from the 117 isolates, 112 isolates from either sediment (71) or water (41) from different lakes (eight from Small Momela, 37 from Big Momela, 15 from Rishateni, 27 from Lekandiro, and 25 Tulusia) resulted in readable and reliable sequences for additional analysis. Although this study was not quantitative, it was clear that the number of Actinobacteria recovered from sediment samples was much higher compared with those from water samples. The fact that the majority of the detected Actinobacteria species were from the lakes' sediments is in agreement with other studies (Akhwale et al. 2015;de Groot et al. 2019). George et al. (2012) suggested that Actinobacteria live in a variety of habitats in nature but are primarily soil inhabitants. Their presence in water continues to corroborates evidence that several Actinobacteria are frequently distributed in water habitats as a result of mixing (Goodfellow and Williams 1983). The observed differences between numbers of Actinobacteria in sediment and in water can also be explained by the sediment-water interface acting as either a potential source of nutrients or as a sink (Ghanem et al. 2000). The insignificant difference of Actinobacteria from one Momela Lake to another could be attributed to close proximity of these lakes and flamingo movements that may transfer organisms, like bacteria, from one lake to another.
The comparison of the sequences using the CD-HIT program revealed that some of the sequences were similar to each other forming 34 clusters. From each cluster, one representative was selected and the closest match analysis of the resulted sequences after BLASTn analysis showed that the representative isolate sequences possessed a higher sequence similarity of 97-100% to their conspecifics (Table 1). The 34 representative isolates were affiliated to 15 genera of Actinobacteria species, namely (1) and Agromyces (1).
The phylogenetic reconstruction of partial 16S rRNA gene sequences of the isolated Actinobacteria with their closely related match strains from the database is shown in Figure 2. The isolated strains clustered with Actinobacteria species are from other soda lakes and other environments, such as marine environments, sludge, different plant sources, soils polluted with crude oil, grassland soil and clinical isolates.
The genus Streptomyces consisted of the highest number of isolated strains followed by Dietzia species (Figure 2). The majority of the isolates (30)   gene sequence similarities above 98.65% identical to known species of Actinobacteria. The remaining four isolates 34SI (MT322826), 42BI (MT322860), 14TI (MT322915), and 17BI (MT322916) had less than 98.65% sequence similarity match with known species of Actinobacteria, which shows that they could be putative new species, because 98.65 % 16S rRNA gene sequence similarity is the threshold for differentiating two species (Kim et al. 2014). The majority of the isolated Streptomyces strains were closely (99-100%) related to existing isolates in GenBank. Accordingly, our isolated strains 104RI (MT322917) and 55BI (MT327747) were closely related (100%) to strain Streptomyces sp. (MN446723) and Streptomyces sp. (JX051287), respectively, isolated from a marine sponge in China and coastal sediment in Turkey (unpublished). Five isolates were closely related to soda lake environments, 108BI (MT322921) and 101TI (MT322920) were closely related (99.8%) to Streptomyces sp. (KC779046) and Streptomyces sp. (KC779045), respectively, both isolated from soda lake sediment in the Ethiopian rift valley (Plessis 2011); 71TI (MT322919) and 2BI (MT192564) were closely related (99-100%) with Streptomyces sp. (MH645743) isolated from grassland soil (unpublished), Streptomyces chumphonensis (LT800131) isolated from a saline and alkaline soda lake called Barex in Hungary (unpublished) and type strain Streptomyces chumphonensis (AB738400) isolated from marine sediment in Bangkok, Thailand (Phongsopitanun et al. 2014); 23TI (MT322918) was closely associated (100%) with Streptomyces sp. (MH430525) and the type strain Streptomyces alkaliterrae (MH430523) isolated from a meteoric alkaline soda lake in India (Świecimska et al. 2020). The predominance of the Streptomyces in nature has been reported widely (Chater 2016), including East African soda lakes such as Lake Nakuru (Solingen et al. 2001) and Lake Elmenteita (Mwirichia et al. 2010;Akhwale et al. 2015) in Kenya. As Streptomyces is the largest genus of Actinobacteria (Kämpfer 2006), their presence in Momela Lakes may be important because these species are among the well-known and economically important species owing to their significant role in medical science, ecology and the biotechnology industry (Alam et al. 2010). A large number of antiparasitic agents, herbicides, immune suppressants and several enzymes pertinent in the food and other industries have been sourced from Streptomyces spp. (Alam et al. 2010 the isolates from this study were closely related to known Dietzia species from East Africa soda lakes (Duckworth et al. 1998;Ronoh et al. 2013). The third most represented species were Nocardiopsis spp., Nesterenkonia spp. and Hoyosella spp. Nocardiopsis spp. were represented by three strains, namely 13TI (MT322456), 69TI (MT322775) and 21TI (MT322785), and were related (99-100%) to Nocardiopsis exhalans (EU430537), Nocardiopsis sp. (FJ898297) and Nocardiopsis alba (MH843127), respectively (Suihko et al. 2009;Yang and Lou 2011). Nocardiopsis spp. has already been isolated from Lake Elmenteita, another East African soda lake (Mwirichia et al. 2010), as well as other hypersaline environments (e.g. Quesada et al. 1982;Jose and Jebakumar 2012). Bennur et al. (2016) reported that the halotolerant Nocardiopsis spp. produce extremozymes, compatible solutes, surfactants and bioactive compounds, such as antimicrobials, to survive in a hypersaline environments.
The Hoyosella isolates from this study, 42BI (MT322860), 14TI (MT322915) and 17BI (MT322916), had 96-98% sequence similarity with species of the genus Hoyosella, suggesting that the isolates from Momela Lakes may be novel species (Kim et al. 2014). At the time of publication, the isolated Hoyosella species are reported for the very second time from a soda lake globally and for the first time from an East African soda lake. To date, three species of the Hoyosella genus exists: Hoyosella rhizosphaerae, a halotolerant Actinobacterium isolated from rhizosphere soil of the dried saline lake, Suaeda Salsa in the Hebei Province, China (Li et al. 2016); Hoyosella altamirensis isolated from a complex cave biofilm (Altamira cave) in Spain (Jurado et al. 2009); and, Hoyosella subflava, which has been reclassified from Amycolicicoccus subflavus (Hamada et al. 2016), which was isolated from saline soil polluted by crude oil in China (Wang et al. 2010). Furthermore, the Mycobacterium species reported in this study are the first to be isolated from a soda lake to date. Strain 33R1 (MT322827) isolated from Lake Rishateni was closely related (99%) to Mycobacterium sp. (X93026) from clinical isolates (Springer et al. 1996).
Other species that have been found in this study and previously reported from East African soda lakes include Cellulomonas, Rhodococcus, Microbacterium and Bogoriella. Strain 27LI (MT328253) was very closely related (100%) to Cellulomonas sp. (HQ413084) isolated from Alkali Lake in the USA (unpublished). Cellulomonas bogoriensis has previously been isolated from sediment of the littoral zone of Bogoria soda lake in Kenya (Jones et al. 2005 Figure 2: Phylogenetic affiliations of partial (≈640 bp) 16S rRNA gene sequences of Actinobacteria isolates retrieved from the Momela Lakes and the type strains of the most closely related genera. The inference tree was constructed through MEGAX using the Tamura-Nei as a substitution method, maximum likelihood as a statistical method, and nearest-neighbour interchange as a maximum likelihood heuristic method. Bootstrap values are expressed as percentages, based on 1 000 resamplings of data. Bootstrap values, >50% are shown at branch points. Bacilus lentus T (MN122294) was used as an outgroup to position the root of the tree genus is their ability to degrade cellulose using enzymes, such as endoglucanase and exoglucanase, that produce propionic or acetic acid as byproducts of sugar metabolism (Glazer and Nikaido 2007). Isolate 35SI (MT328423) from Small Momela was closely related (100%) to Rhodococcus ruber (MN252046) isolated from sewage sludge in China, which also possessed the ability to biodegrade (Tsuruo 1975). The Rhodococcus sp. has also previously been isolated from Lake Magadi (Kenya) by Ronoh et al. (2013). Microbacterium spp. were represented by the strain 36SI (MT328425) from Small Momela and 5LI (MT328426) from Lekandiro and were closely related (98-100%) to Microbacterium sp. (MK578295) and Microbacterium sp. (MH671848) isolated from rice leaf and tomato roots (unpublished), respectively. Species of Microbacterium have also been previously found from Lake Magadi in Kenya (Ronoh et al. 2013).
Our isolate 99LI (MT328036) from Lekandiro Lake was closely (99%) related to Bogoriella caseilytica (NR_029328) isolated from Lake Bogoria in Kenya (Groth et al. 1997). The isolate is of interest because it has not been reported since 1997.

Occurrence and phylogenetic analysis of Actinobacteria clones
The clone library consisted of 85 clones (15 from Small Momela, 17 from Big Momela, 14 from Rishateni, 19 from Lekandiro and 20 from Tulusia) that contained readable and reliable Actinobacteria clone sequences. The analysis of the low number of clones was attributed to financial limitations within this study, whereby only 20 clones could be analysed for each sample. The results showed that a number of the cloned sequences were similar to each other forming 13 clusters. From each cluster representative, clone sequences of Actinobacteria that had a percentage similarity of 90-99% with their closest match in GenBank after BLASTn were chosen (Table 2).
Almost all Actinobacteria clones could not be classified or linked to isolates of either genus or type species with a similarity of 90% or higher, except clone 15SC (MT313136), 213BC (MT322435) and 121SC (MT322442), which had 92-93% sequence similarity with type strain Actinomarinicola tropica (MN638853) isolated from sea sediment from southern China (He et al. 2020), a new marine Actinobacterium belonging to the Iamiaceae family. This demonstrates that majority of recovered Actinobacteria clones in these lakes may be unculturable under the protocol used for the isolation of strains in this study. These results are similar to those reported from Lake Elmenteita, a soda lake from Kenya (Mwirichia et al. 2009) where out of all 655 sequenced clones, 80.15% (525) were related to uncultured clones. One clone sequence (213BC (MT322435)) showed more than 98% sequence similarity to their closest type match clone. The remaining 12 clones had a lower sequence similarity suggesting that they are novel, and additional studies to identify these species are required.
The results of phylogenetic analyses of the Actinobacteria clones are shown in Figure 3. The constructed clones' phylogenetic tree depicts six major clusters, which evidently indicates the presence of diverse groups of non-cultivable Actinobacteria across Momela Lakes. Eight of the clones may be considered putative new species because they were less than 98% similar to other clones in GenBank as they formed their own phyletic branches. Unfortunately, the majority of matched clones from the database were unpublished; hence, speciation could not be confirmed.
Almost half of the clones, namely 352RC (MT313100), 15SC (MT313136), 316RC (MT322237), 128SC (MT322253), 110SC (MT322304) and 213BC (MT322435) were similar (90-99%) with environmental sequences (uncultured Actinobacterium clones) from soda lake environments. These were Actinobacterium clone (EU532544) from saline brines in China (Unpublished), Actinobacterium clone (HM106311) from soda Lake Chitu in Ethiopia (Unpublished), Bacterium clone (HM050983) from Lake Xingyunhu in China (Unpublished), Actinobacterium clone (AF454307) from the Mono soda lake in California, Bacterium clone (HM050983) from Lake Xingyunhu in China, Actinobacterium clone (EU703275) from Lake Qinghai in China, respectively. This closeness and formation of the clusters between the sequences of Actinobacterium clones from the Momela soda lakes with other Actinobacterium clones from other hypersaline soda lakes may indicate that these Actinobacteria are not easy to culture. Literature suggests that only a small percent (<1%) of microbes in nature can be cultured using already known standard techniques (Malkawi and Al-Omari 2010), however culturing techniques for the majority of microbes have not yet been ascertained (Su et al. 2012).
The sequences of the remaining seven clones did not cluster with sequences from soda lake environments, but did cluster with sequences that have been retrieved from clones isolated from extreme and marine environments. These were: clone 222BC (MT322445) closely related (98%) with Actinobacterium clone (HQ265288) from a Tibetan hot spring, China (unpublished); clone 215BC (MT322448), 23BC (MT322444) and 220BC (MT322447) were closely related (97-98%) to Actinobacterium clone (JN874667) and Actinobacterium clone (JN874665) both cultured from Poomarichan coral reef sediment, India (unpublished); clone 214BC (MT322446) and 36RC (MT322229) were closely related (94-96%) to Actinobacterium clone (KT714868) and Actinobacterium clone (KT714673), respectively, both isolated from reef coral Porites lutea (Kuang et al. 2015); and, clone 121SC (MT322442) was closely related (94%) to Actinomycete clone (JX242791) isolated from tidal beach sand in China (unpublished). This is the first time that these clones have been reported to occur in soda lakes. Further studies need to be conducted to enumerate the Actinobacteria from these lakes in order to corroborate the current findings.

Relationship between Actinobacteria isolates and clones of Momela Lakes
When aligned, the sequences of Actinobacteria isolates and clones from environmental samples from the Momela soda lakes were closely related to different genera or families of order Actinomycetales (Figures 2 and 3; Supplementary  Figure 1). The detection of different strains of Actinobacteria by using culture dependent and culture independent approaches may show that the cultured strains were in the minority and favoured by the culturing conditions. Similar findings have also been reported from the Ethiopian Rift Valley Lakes (Plessis 2011), in whcih they ascribe such findings to the extreme environment that soda lakes provide for microorganisms, hence they are less amenable to culture dependent studies as a large number of them are in a viable or dormant state, but non-culturable. The dormant state is a vital survival approach, which enables bacteria to survive when growth conditions are not optimal (Roszak and Colwell 1987). One of the methods to capture the full diversity in such a metapopulation could be the use of genus-specific primers to identify bacterial subpopulations. This has been achieved through the use of broad, but highly specific, clone libraries (Jiang et al. 2006;Malkawi and Al-Omari 2010;Plessis 2011). Therefore, there is the possibility that a larger diversity of Actinobacteria in the Momela soda lakes exists and further investigation using suitable laboratory-based culturing techniques and growth conditions for various subpopulations of, as yet, uncultured Actinobacteria is required. Although there is a tremendous improvement on development of various Actinobacteria-specific PCR primers (Stach et al. 2003), it is also possible that the primers in this study were unable to detect the diverse Actinobacteria in the environment. This can be remedied by using the sequences derived from cultured isolates in combination with extremophilic Actinobacterial 16S rRNA gene sequences to design new   Figure 3: Phylogenetic affiliations of partial (≈640 bp) 16S rRNA gene sequences of Actinobacteria clones retrieved from Momela Lakes and the type clones of the most closely related genera. The tree was constructed through MEGAX using the Tamura-Nei as a substitution method, maximum likelihood as a statistical method, and nearest-neighbour-interchange as a maximum likelihood heuristic method tree inference option. Bootstrap values are expressed as percentages, based on 1 000 resamplings of data. Bootstrap values >80% are shown at branch points. Bacilus lentus T (MN122294) was used as an outgroup to position the root of the tree improved primers. This discrepancy between the cultured diversity and culture independent diversity of Momela Actinobacteria emphasises the requirement for multifaceted approaches towards studies involving community ecology (Hill et al. 2000), especially the involvement of the modern next-generation DNA sequencing (NGS) technologies (metagenomics) towards the discovery of untapped Actinobacteria diversity and novel natural products (Gomez-Escribano et al. 2016).
For the first time, this study records the diversity of Actinobacteria communities in the Momela soda lakes at Arusha National Park, Tanzania. It reveals the diverse composition of Actinobacteria isolates and clones that are similar to findings from other soda lake subpopulations, like those in Kenya (Groth et al. 1997;Solingen et al. 2001;Jones et al. 2005;Mwirichia et al. 2009;Ronoh et al. 2013), China Xing et al. 2009) and India (Sultanpuram et al. 2015). However, this study reports on putative new strains in the soda lakes for the first time with yet unknown potential. The presence of Mycobacterium and Hoyosella from a soda lake, as well as Isoptericola, Jonesia, Micrococcus, Leucobacter and Agromyces from a soda lake in East Africa, is also reported for the first time. Conversely, some subpopulations identified in other soda lakes (i.e. Arthrobacter, Terrabacter and Nocardia species) were not detected in this study. Because Actinobacteria are known to be a potential source of biotechnologically important bioactive compounds, the species isolated from Momela Lakes set the precedent for the search for novel biotechnologically significant bioactive compounds of anthropogenic importance. Therefore, the findings call for microbiological bioassays to screen for such bioactive compounds in order to establish or affirm their potential uses as postulated by this molecular-based study.