The results obtained through microscopic analysis and Illumina MiSeq analysis had in common that species belonging to specific taxa exist abundantly from the class level to the family level. However, in terms of the total number of taxa detected, the results of Illumina MiSeq analysis were superior to those of the microscopic analysis. Exceptionally, it was possible to detect relatively more diverse species via microscopy by limiting it to the species level, although these results tended to be biased toward specific taxa in the upper taxonomic level. Based on the above findings, diatoms in microalgal communities can be effectively detected via Illumina MiSeq [13]. However, this method seems to have obvious limitations in the delicate detection of taxa at the species level [37], possibly due to the lack of registered information regarding the marker gene of each diatom species [12, 13, 37, 38]. Therefore, more accurate analyses of diatom communities can be guaranteed only when abundant information is available [12, 38, 39]. Although microscopic analysis yielded relatively biased results for specific taxa, it was able to identify more taxa at the species level [38–40]. Thus, this method is considered to be relatively effective for the detection and identification of taxa at the species level [40]. This suggests that it is relatively ineffective for the analysis of communities, but advantageous for the collection of detailed information about diatom species [40–42]. There are commonalities between the results obtained via microscopy and Illumina MiSeq (Figs, 2–4). In both analyses, the abundances of species belonging to different phyla at different taxonomic levels, i.e., Bacillariophyceae (at the class level; 83.87% via microscopy and 62.92% Illumina MiSeq), Bacillariales (at the order level; microscopic: 31.18%, Illumina MiSeq: 13.48%), Naviculales (at the order level; microscopic: 27.96%, Illumina MiSeq: 34.83%), Bacillariaceae (at the family level; microscopic: 31.18%, Illumina MiSeq: 13.48%), and Naviculaceae (at the family level; microscopic: 21.51%, Illumina MiSeq: 13.48%), were high. Although both methods tended to detect similar species compositions, clear differences were noted. Illumina MiSeq analysis (which identified 5 classes, 21 orders, 32 families, and 89 species including unclassified results) was able to detect higher diversity within each taxonomic category, from the class level to the species level, compared with microscopic analysis (which revealed 3 classes, 12 orders, 17 families, and 93 species, including unclassified results). Diatom species belonging to the class Mediophyceae and the orders Rhopalodiales, Chaetocerotales, Rhizosoleniales, Rhaponeidales, Biddulphiales, Cymatosirales, Eupodiscales, and Hemiaulales could not be detected via microscopy. Although only this method detected order Licmophorales, generally speaking, from the class level to the order level, the results obtained with Illumina MiSeq were superior in terms of the number of detected taxa. At the family level, the tendency of Illumina MiSeq to detect relatively higher taxonomic diversity was more pronounced. Seventeen families (Achnanthidiaceae, Diploneidaceae, Sellaphoraceae, Stauroneidaceae, Rhopalodiaceae, Aulacoseiraceae, Chaetocerotaceae, Stephanopyxidaceae, Rhizosoleniaceae, Skeletonemataceae, Thalassiosiraceae, Staurosiraceae, Rhaphoneidaceae, Biddulphiaceae, Cymatosiraceae, Eupodiscaceae, and Hemiaulaceae) were detected only via Illumina MiSeq, whereas 4 families (Cocconeidaceae, Cymbellaceae, Thalassiosiraceae, and Ulnariaceae) were detected by microscopic analysis. Although 13 families (Entomoneidaceae, Bacillariaceae, Gomphonemataceae, Rhoicospheniaceae, Achnanthaceae, Amphipleuraceae, Diploneidaceae, Naviculaceae, Pleurosigmataceae, Surirellaceae, Catenulaceae, Melosiraceae and Fragilariaceae) were commonly detected at the family level by both methods (microscopic and Illumina MiSeq analysis), only 9 species (Nitzschia sp., Rhoicosphenia abbreviata, Navicula cryptocephala, Navicula gregaria, Navicula perminuta, Navicula sp., Amphora sp., Melosira nummuloides, and Melosira varians) were commonly detected at the species level. The above findings suggest that the lower the taxonomic level, the greater the difference between the results obtained from the two methods. Ironically, more taxa were detected via Illumina MiSeq for each taxonomic category, from the class level to the family level, but a higher number of species were identified via microscopy. Based on these results, we support Illumina MiSeq analysis as a more effective method for the analysis of diatom communities. However, considering that extensive genetic information concerning each diatom species must be available to conduct accurate analyses via Illumina MiSeq.
The microscopic and Illumina MiSeq analyses yielded conflicting results in terms of diatom classification at low taxonomic levels, such as the genus and species levels. Microscopy (Fig. 5, Table 1) confirmed that diatoms of the genera Nitzschia and Navicula dominated at the genus level, and some species belonging to them were predominant. In contrast, Illumina MiSeq confirmed the low abundance of Nitzschia and the high abundance of Melosira, Entomoneis, and Amphiprora at the genus level. In addition, in many Illumina MiSeq results, the species name was classified as “sp.”. Based on the above, it was confirmed that microscopy can accurately identify taxonomic categories down to the species level, whereas Illumina MiSeq is limited in this task. This is possibly due to the strengths and weaknesses of the two methods [8, 12, 23, 38, 40]. Although identification through microscopic observation can vary greatly depending on the skill and experience of the observer, it can provide accurate identification down to the species level [41, 42]. On the other hand, Illumina MiSeq, which depends on the target sequence for identification, may interpret data differently depending on the database used as the reference for identification [13, 43]. However, although detailed identification of each diatom cell is important, sample size must also be considered to obtain reliable results when analyzing diatom communities [10, 38]. One of the reasons why microscopy is limited in the analysis of communities (even though species-level identification is possible) is that substantial human resources and time are required to process samples [8, 12]. Moreover, because microscopy is applied to relatively small samples, the results obtained can be fatally biased [12, 41]. Illumina MiSeq is not affected by these issues and can efficiently analyze relatively large samples in a relatively short time; consequently, this method is superior to microscopy in the analysis of microbial communities [12, 13]. Although microscopy allows accurate identification, it has obvious limitations in community analysis. Therefore, it is necessary to use Illumina MiSeq to efficiently analyze large-scale samples. However, to improve the accuracy of the Illumina MiSeq results, precise and extensive information is required, and microscopy can significantly contribute to the creation of such databases.
Table 2 summarizes the biological parameters for the sampled diatom communities calculated based on results of microscopic and Illumina MiSeq analyses, it was confirmed that the two methods yielded results that were inconsistent for all parameters; in any case, they did not converge to 100% even though the same samples were analyzed. High dominance values were accompanied by low diversity and evenness values, and results with a high number of species tended to be accompanied by a high richness value [44–46]. Moreover, no correlation was observed between the microscopy and Illumina MiSeq results of the biological parameters (dominance, diversity, and evenness; number of species, and richness). On the other hand, a particular regularity could be easily found between TDI and DAIpo values [47, 48]. Although both methods detected regularity between the calculated biological parameters, the clear differences observed between the results obtained from the same samples indicated that the effectiveness of these methods needs to be evaluated [45, 47]. In the analysis of diatom communities, either method may have limitations or both methodologies (microscopy, Illumin MiSeq) may have significant limitations [12]. To accurately analyze diatom communities using biological parameters, such as dominance, diversity, richness, evenness, TDI, and DAIpo, accurate diatom community analysis results must be the basis [44–48]. Therefore, it is important to evaluate the effectiveness of each method and suggest solutions to its limitations.
For effectiveness of microscopy and illumina MiSeq in the analysis of diatom communities, microscopy and Illumina MiSeq were used to analyze diatom communities, and the results obtained were compared (Figs. 2–5; Tables 1 and 2). The results obtained at the species level (the lowest taxonomic level) generally showed a clearer difference compared with those at the class level (the upper taxonomic level) [8, 10, 12]. The strength of microscopic observation is that it allows accurate identification at the species level [42]. This is an important factor for the analysis and understanding of microbial communities [10, 40]. In addition, in this study, the results obtained from microscopy showed high discrimination of the taxonomy of each diatom cell. This strength provided the basis for research on the diatom community to be carried out using microscopy [9, 40, 42]. However, microscopy-based analysis also has obvious limitations as the accuracy of identification is based on the observer’s experience and knowledge [43, 49]. The number of experts is limited and currently declining, and their lack directly decreases the reliability of the results obtained through microscopy [12, 49]. On the other hand, a strength of microscopy is that it requires no special materials other than a microscope and sample preparation [42]. Therefore, this method can be used to analyze diatom communities at a relatively low cost [42, 50]. However, extensive human resources and time are required to obtain an adequate amount of data, which makes it difficult for microscopy to have effetely valid merits [12, 50]. Furthermore, a vast amount of data would be desirable in the study of diatom communities, but microscopy has serious limitations in this regard [49, 50]. In contrast, Illumina MiSeq is less affected by these limitations, which is one of the reasons why it is used for the analysis of diatom communities [39, 49]. Analysis via Illumina MiSeq does not require skilled observers; therefore, there are no concerns about the reliability of data, which would vary depending on their level of expertise [13]. In addition, this method allows researchers who are inexperienced in taxonomic identification to easily analyze microbial communities [11, 39]. Furthermore, compared with microscopy, Illumina MiSeq can effectively process large amounts of data in a relatively short time [13, 37]. Although Illumina MiSeq is undoubtedly one of the most powerful tools for the analysis of diatom communities, it also has obvious limitations [40, 43]. The procedures required, such as DNA extraction and PCR, may introduce significant bias into the results [13, 23, 43]. Moreover, in addition to the kits and procedures required to perform sequencing, the database upon which the analysis of the results is based is also a significant factor [23]. As the analyzed sequences are taxonomically assigned based on this database, their quality is revealed in the results [37, 38, 43]. Therefore, it can be risky to analyze communities exclusively using Illumina MiSeq [12, 39, 40]. However, microscopy and Illumina MiSeq can complement each other because each method has strengths that can compensate for the limitations of the other [38, 43]. This can be accomplished by constructing a high-quality database based on accurate taxonomy information obtained through microscopy and analyzing diatom communities more effectively via Illumina MiSeq using the improved database.