Morphological Characterization and DNA Barcoding of Duckweed Species in Saudi Arabia

Duckweeds, or Lemnaceae, are widespread aquatic plants. Morphology-based identification of duckweed species is difficult because of their structural complexity. Hence, molecular tools provide significant advantages for characterizing and selecting species or clones for sustainable commercial use. In this study, we collected and characterized ten duckweed isolates from nine different regions in Saudi Arabia (SA). Based on the morphological characterization and phylogenetic analysis of intergenic spacer sequences of chloroplast DNA using six barcoding markers, the clones were classified into three genera, represented by seven species: Lemna gibba L., Lemna minor L., Lemna japonica Landolt, Lemna aequinoctialis Welw., Lemna perpusilla Torr., Spirodela polyryiza (L.) Schleid., and Landoltia punctate G. Mey. Lemna gibba was revealed to be a distinct dominant duckweed species in many regions of SA. Five barcoding markers showed that L. gibba, L. minor, and L. punctata were the most widely distributed species in the country. However, L. punctata, L. perpusilla, and S. polyryiza were the dominant species in the Al-Qassim, Madinah-1, and Madinah-2 regions, respectively. Moreover, the morphological traits revealed variations for these clones, relative to other studied duckweed clones. According to the results obtained in this study, three out of six plastid markers (trnH-psbA, matK, and atpF-atpH) helped to identify the dominant duckweed species in Saudi Arabia. Further evaluation based on adaptability, molecular genetic studies, and functional genomics is needed for these species to be used at the commercial level in Saudi Arabia.


Introduction
Under the arid conditions of Saudi Arabia, the agricultural sector plays a significant role in food security for humans and animals, as well as economic, social, and sustainable development. Saudi Arabia, with a Vision 2030 strategic plan, aims to enact sustainable development in food production by saving freshwater resources, the environment, and energy. The government supports research on new feed resources for livestock and poultry that maintain high protein value and high water and fertilizer use efficiency. Duckweed is an aquatic plant that contains crude protein up to 45% of its dry weight under optimum conditions and may be used, without further processing, as a complete feed for animals and fish [1]. Duckweed starch content has been reported to be from 5% to 70% of its dry weight. The high level of starch and low lignin and cellulose levels make duckweed a promising feedstock for biofuel production [2]. Duckweed plants float on the surface or are slightly submerged in water. Leaf-like fronds are modified leaves and stem that mainly function in photosynthesis and vegetative reproduction. As a result, the rate of biomass production is high, providing the basis for various practical applications of duckweed in food and feed sustainability [3]. For commercial utilization of duckweed, knowledge of its taxonomy is required. At the morphological level, delineating duckweed species has been shown to be difficult. Many studies have tried to decrease the anatomical complexity based on the chemical composition of flavonoids [4] and to delineate duckweed species using allozyme, which has helped advance morphology techniques [5,6]. The genome size of duckweed varies 13-fold, ranging from 150 Mb in Spirodela polyrhiza (L.), Schleid. to 1881 Mb in Wolffia arrhiza Wimm. With the development of sequencing technology and bioinformatics, m genomes from Spirodela and Lemna have been sequenced and assembled [7]. Thirty-eight duckweed species have been identified and characterized [8]. With molecular biology tools and techniques (polymerase chain reaction and Q-PCR, DNA barcoding, and highthroughput DNA sequencing tools), a taxonomic analysis at the molecular level offers a more profound understanding than chemotaxonomy. Therefore, among the various DNA analysis methods for taxonomic definition, molecular taxonomy is considered the most suitable method for identifying and evaluating the genetic potential of duckweed species [9]. There are only a few studies in the literature on identifying duckweed genera and species in Saudi Arabia. Masrahi et al. [10] reported the first record of Wolffiella hyalina (Delile) Monod in Saudi Arabia and regarded it as an addition to the flora of Saudi Arabia. The present study aimed to assess the genetic variability based on morphological traits and molecular (DNA barcoding) markers among duckweed clones collected from m

Results
The identification of duckweed clones, collected from ten locations in Saudi Arabia, was made through morphological and molecular characterization ( Figure 1). Morphological variation was recorded based on the different measurements (Tables 1 and 2). All clones showed obovate fronds shape and rounded frond apex, and the turions were present. The frond shape of the clones from the Riyadh, Taif, Tohama, Tanomah, Qassim, and Madinah-1 regions was asymmetric, while those from the remaining regions were symmetric. The frond color of clones from five regions (Riyadh, Tohama, Al-Baha, Qassim, and Madinah-1) were dark green, and, in five regions (Taif, Jazan, Tanomah, Dhahran, and Madinah-2), the frond color was recorded as light green. All the studied duckweed clones showed frond apexes and turions. The analysis of variance for frond and root traits showed highly significant differences among the clones ( Table 2). The number of roots ranged from 1.67 to 9.22; moreover, the Madinah-2 and Al-Qassim clones had the highest number of roots and differed significantly from other clones. The maximum number of roots, i.e., 9.22, was recorded for the Madinah-2 clones, while the Al-Baha clones had the lowest number of roots, i.e., 1.67. The Tanomah clones had the longest roots (3.76 mm), and the Madinah-2 clones had the shortest roots (0.74 mm). Regarding the characteristics of fronds, clones collected from the Al-Qassim region had the highest number of fronds (6.0), while clones from the Riyadh region had the longest fronds (7 mm). The widest and thickest clones were 8.5 mm from the Madinah-1 region and 3.2 mm from the Al-Taif region, respectively.  Table 1. Qualitative morphological traits for studied duckweed clones of the ten locations of Saudi Arabia (n = 10 samples from each clone averaged and replicated three times, total samples = 30 reads). Frond shape-FSh; frond symmetry-FS; frond color-FC; frond apex-FA; turions-T. Table 2. Analysis of variance (ANOVA) for the quantitative morphological traits of the studied duckweed clones. Mean and standard deviation of the evaluated traits (n = 10 samples and three replications, total = 30). 2.00 ± 0.01b 3.0 ± 0.01e 0.4 ± 0.0f 5.0 ± 0.07d 3.0 ± 0.01c 1.54 ± 0.28b 3.22 ± 0.44bc Al-Qassim 8.11 ± 0.78a 4.0 ± 0.01c 0.7 ± 0.0e 6.0 ± 0.06c 5.0 ± 0.01a 2.14 ± 0.09b 6.00 ± 0.01a  To facilitate interpretation of the correlation matrix, produced from the quantitative and qualitative morphological traits for ten duckweed clones, a principal component analysis (PCA) was applied. The PCA plots provided a visual overview of how different traits influenced the distribution of the duckweed clones. The first three principal components explained 88.27% of the total variance (i.e., PC1 46.4%, PC2 24.3%, and PC3 17.51%). The plot for the PC1 and PC2 showed that some traits described the same variation among the clones ( Figure 2). The number of roots formed the underlying dimension for PC1, with a value of 0.95 positive loadings; whereas, frond width (−0.2) was negatively loaded and placed on the left-hand side of the plot. PC2 explained 24.3% of the variance and showed a positive correlation with frond traits, such as frond length (0.79), number of fronds (0.26), and frond thickness (0.22), besides root length (0.26). PC3 was loaded with frond width with positive loading (0.76). The Al-Qassim and Madina-2 clones showed the most positive loading in PC1, with 4.9 and 5.8, respectively, while Madina-1 clones had the most negative loading on PC1 and PC2. Those clones were considered to be the most divergent clones.  Madinah-1 2.00 ± 0.01b 8.5 ± 0.01a 0.9 ± 0.0d 1.0 ± 0.06g 4.0 ± 0.01b 0.84 ± 0.09c 2.89 ± 0.33bc Madinah-2 9.22 ± 2.39a 4.0 ± 0.06c 0.4 ± 0.0f 4.0 ± 0.05f 3.0 ± 0.01c 0.74 ± 0.09c 2.44 ± 0.53bc Tukey HSD 1.60 0.38 0.18 0.45 0.42 0.60 1.10

Number of Fronds
Means followed with the same letter in the column were not significantly different based on Tukey HSD, p < 0.05.
To facilitate interpretation of the correlation matrix, produced from the quantitative and qualitative morphological traits for ten duckweed clones, a principal component analysis (PCA) was applied. The PCA plots provided a visual overview of how different traits influenced the distribution of the duckweed clones. The first three principal components explained 88.27% of the total variance (i.e., PC1 46.4%, PC2 24.3%, and PC3 17.51%). The plot for the PC1 and PC2 showed that some traits described the same variation among the clones ( Figure 2). The number of roots formed the underlying dimension for PC1, with a value of 0.95 positive loadings; whereas, frond width (−0.2) was negatively loaded and placed on the left-hand side of the plot. PC2 explained 24.3% of the variance and showed a positive correlation with frond traits, such as frond length (0.79), number of fronds (0.26), and frond thickness (0.22), besides root length (0.26). PC3 was loaded with frond width with positive loading (0.76). The Al-Qassim and Madina-2 clones showed the most positive loading in PC1, with 4.9 and 5.8, respectively, while Madina-1 clones had the most negative loading on PC1 and PC2. Those clones were considered to be the most divergent clones. The results generated from the six plastid markers (psbK-psbI, trnH-psbA, atpF-atpH, matK, rpoC1, and rbcL) for species identification, and their accessions numbers, are presented in Table 3, and the sequences generated from each sample are shown in Supplementary Table S1. The clones collected from the Riyadh, Dhahran, Tohama, Al-Baha, Al-Taif, and Tanomah regions represented the same species (Table 3). However, the duckweed clones collected from the Jazan region represented L. aequinoctialis, and all the The results generated from the six plastid markers (psbK-psbI, trnH-psbA, atpF-atpH, matK, rpoC1, and rbcL) for species identification, and their accessions numbers, are presented in Table 3, and the sequences generated from each sample are shown in Supplementary Table S1. The clones collected from the Riyadh, Dhahran, Tohama, Al-Baha, Al-Taif, and Tanomah regions represented the same species (Table 3). However, the duckweed clones collected from the Jazan region represented L. aequinoctialis, and all the sequence data markers confirmed its exact identification, except that rbcL identified L. minor. Similarly, clones collected from the Al-Qassim and Madinah (Madinah-1 and Madinah-2) regions were identified as L. punctata, L. perpusilla, and S. polyryiza, respectively. Table 3. DNA barcoding-based identifications and accessions number among 10 duckweeds clones.

Region
psbK-psbI trnH-psbA matK atpF-atpH rpoC1 rbcL After sequencing, the data were cleaned and aligned using the Bioedit software, and the species-level phylogenetic tree was constructed. According to the six markers, the duckweed clones, isolated from the Riyadh region, were identified as L. gibba by four markers (trnH-psbA, mat K, atpF-atpH, and rbcL), while L. japonica was recognized by trnH-psbA, and L. minor was recognized by rpoC1. Similar results were recorded for the Dhahran, Tohama, Al-Baha, Al-Taif, and Tanomah regions. Most markers identified the Jazan isolate as L. aequinoctialis, but it was identified as L. minor only by the rbcL marker. However, the duckweed clones isolated from Al-Qassim and Madinah-2 were identified as L. punctata and S. polyryiza by all markers, respectively. The clones from Madinah-1 were identified as L. perpusilla with three markers (psbK-psbI, trnH-psbA, and matK). In comparison, two markers (atpF-atpH and rpoC1) identified the isolate as L. aequinoctialis, and rbcL identified it as L. minor. Based on multiple nucleotide sequence alignment ( Figure S1), divergence score matrix ( Figure S2), and the phylogenetic tree (Figure 3), four species were mostly distributed in Saudi Arabia, which were L. gibba, L. aequinoctialis, L. punctata, and S. polyryiza (Figure 3).

Discussion
Out of seven different haplotypes identified, four were dominant in the ten regions, identified using DNA barcoding and phylogenetic analysis. The frond shape, frond apex, and turions registered the same obovate, rounded, and present, respectively, for all duckweeds clones. In this study, the number of roots and frond length significantly contributed to the genetic variability among collected clones, confirming that they were the most influential variability components. A similar morphological identification was made earlier by [11] using morphological characterization for species identification among duckweed. This grouping included S. polyrhiza, L. aequinoctialis, L. punctata, and W. globose. It was also suggested that the low degree of differentiation among members of the Lemnaceae family most likely owes to adaptation to the highly specialized way of aquatic life [12].
Similar results were reported by Azer [13], who classified the Lemna genus into three different clusters, based on morphological characterization. The first cluster included L. gibba and L. minor, while the second and third clusters included L. trisulca and L. aequinoctialis. Bog et al. [14] classified the Lemna genus into four sections. It was also suggested that the similarities in specific morphological characteristics were closely connected to L. gibba and L. minor. A molecular study based on DNA barcoding results also recorded L. japonica, L. gibba, and L. minor aggregated in the same group. However, the phylogenetic analysis confirmed this relationship and reported that L. japonica, L. gibba, and L. minor were present in the same group.

Discussion
Out of seven different haplotypes identified, four were dominant in the ten regions, identified using DNA barcoding and phylogenetic analysis. The frond shape, frond apex, and turions registered the same obovate, rounded, and present, respectively, for all duckweeds clones. In this study, the number of roots and frond length significantly contributed to the genetic variability among collected clones, confirming that they were the most influential variability components. A similar morphological identification was made earlier by [11] using morphological characterization for species identification among duckweed. This grouping included S. polyrhiza, L. aequinoctialis, L. punctata, and W. globose. It was also suggested that the low degree of differentiation among members of the Lemnaceae family most likely owes to adaptation to the highly specialized way of aquatic life [12]. Similar results were reported by Azer [13], who classified the Lemna genus into three different clusters, based on morphological characterization. The first cluster included L. gibba and L. minor, while the second and third clusters included L. trisulca and L. aequinoctialis. Bog et al. [14] classified the Lemna genus into four sections. It was also suggested that the similarities in specific morphological characteristics were closely connected to L. gibba and L. minor. A molecular study based on DNA barcoding results also recorded L. japonica, L. gibba, and L. minor aggregated in the same group. However, the phylogenetic analysis confirmed this relationship and reported that L. japonica, L. gibba, and L. minor were present in the same group.
We can infer that the samples collected from the Riyadh, Dhahran, Tohama, Al-Baha, Al-Taif, and Tanomah regions had similar traits (morphological characterization). The traits include long roots, having many fronds with veins that inflate, and few seeds ribbed with L. minor and fewer with L. gibba, but not in L. aequinoctialis and L. trisulca [14]. Our study recorded similar results-i.e., samples collected from the Qassim, Madinah-1, and Madinah-2 regions belonged to three different genera and recorded unique characteristics. relative to other duckweed strains. In contrast, a detailed key is available for identifying species based on morphological traits [11]. A lack of experience and expertise in morphological taxonomy limits using this key to identify duckweed species worldwide. We also used six different plastid markers to identify duckweed species in Saudi Arabia because of limited morphological markers.
Similarly, Klaus et al. [15] suggested that plastid spacer barcoding markers have shown to be reliable tools for identifying duckweed species, especially when used in conjunction with several markers. However, all clones of the same species cannot be differentiated by these markers alone. In this context, several other plastid and mitochondrial markers are good options for obtaining excellent results [8,16], or another global marker (i.e., AFLP) could serve the purpose [14]. Zhang and Azizullah [17] stated that, although duckweeds can occasionally be found in single or more complex communities, they often exist in two or three species communities. Using ISSR markers to assess the genetic variability in populations of seven species in genus Lemna and Spirodela from China and Vietnam, Xue et al. [18] reported genetic differentiation of duckweeds, suggesting that geographic differentiation had a significant influence on the genetic diversity of duckweeds. In a previous report, duckweed clones from tropical Africa showed higher genetic diversity than those from other regions, probably because of higher temperatures and radiation in the equatorial region [19].
A comprehensive phylogenetic analysis has also been carried out using the morphological, anatomical characters, flavonoids, and molecular analysis (chloroplast and introns) sequences were used to classify the Lemnaceae species and phylogenetic analysis confirms the lemna, landoltia, spirodela, wolffian, and wolffiella in the grouping [20]. The current study also identifies similar results, i.e., L. gibba, L. minor, L. japonica, and L. aequinoctialis using the morphological and molecular combined data. Wang et al. [8] reported delineating the genus Lemna, based on several plastid sequences. Tippery et al. [21] also investigated the use of various plastid and nuclear markers (ETS and ITS), and Borisjuk et al. [22] used psbK-psbI and atpF-atpH. Similar results were recorded during our study, L. gibba, L. minor, and L. japonica (Riyadh, Dhahran, Tohama, and Al-Baha), while other clones were recorded in different genera (L. perpusilla/L. aequinoctialis, L. punctata, S. polyryiza) using the plastid markers. Landolt [23] assigned the highest morphological primitiveness to L. gibba, while the lowest was assigned to L. minuta and L. valdiviana, which possibly signifies the genus Lemna as the basal species, while the latter represents the most derived species. In several projects, most Lemna species could be easily identified using DNA barcoding.
Xu et al. [24] suggested that duckweeds were the only occasionally observed aquatic plants present in large reservoirs and natural rivers. Our observations show that L. aequinoctialis is the best candidate for future regeneration applications in the Jazan region, where samples are collected from rain pools (a small area of still water, typically one formed naturally), usually receive heavy rain during the year. These pools are distributed on top of mountains and are, perhaps, the best among the pool. The genus Lemna has been studied for several years. However, the taxonomic identification boundaries are not yet adequately resolved, and there are still several botanists who disagree with the identification of Lemna species. Most of its features are changed and significantly overlap. Because of the variability of this genus, taxonomists have had controversy for over a century and treated the Lemna genus into several species [14,25]. It is known that S. punctata, S. intermedia, and S. polyrhiza represent the same genus, called Spirodela [23]. According to morphological traits, S. intermedia is the most primitive Lemnaceae species, and the highest primitiveness is found in this species [9]. By applying molecular tools for taxonomy to the Lemnaceae family, these three species were separated into two different genera, i.e., Spirodela and Landoltia (changing the nomenclature of the species S. punctata to L. punctata), after the analysis of the plastid rbcL sequence [26]. The next generation sequencing technique may be considered, since it is less expensive; additionally, the entire chloroplast genome opens a new avenue to resolve species identification issues. Recent advances in organelles [8] and sequencing of nuclear genomes [27] could mine potential markers further.
According to the rps16 data of [16], the W. globosa grouping is composed of the six genera of the W. globosa species. While the plastid markers rp16 and atpF-atpH showed genetic differences in L. aequinoctialis and W. globosa recorded relatively low diversity in a cosmopolitan environment [28], it is questionable if a longer sequence of clones would significantly increase this genetic diversity. While the genetic background of these clones was also widely distributed, S. polyrhiza was dispersed. The geographical influence on genetic diversity appears to be higher, as variation was found among all duckweed clones in the Jazan south region compared with clones in the northern regions, especially in the L. aequinoctialis species [24]. Our results recorded that the exact geographical positions (Qassim (L. punctata), Madinah-1 (L. perpusilla), Madinah-2 (S. polyryiza)), met the geographic environment. The lower latitude and higher temperatures, and ultraviolet radiation to the south of the Hainan Island could be responsible for this similarity [29]. The presence of L. punctata, L. perpusilla, and S. polyryiza could also contribute to their widespread distribution and adaptability to different environmental conditions in the region.

Plant Material
The duckweed clones analyzed in this study were collected from ten different regions of Saudi Arabia (Figure 4). The plants were rinsed in clean water, cultured in half-strength Hoagland solution, and acclimated for one week before starting the experiments. A single plant from each clone was cultured at 28 • C, in 14 h of light per day, for two weeks for multiplication ( Figure 1). The study was conducted at the greenhouse of the College of Food and Agriculture Sciences, King Saud University, Riyadh, Saudi Arabia.

Morphological Characteristics
The duckweed clones were transferred to a polypropylene container (15 × 10.5 × 7.5 cm 3 ) filled with 750 mL of half-strength Hoagland solution. Three replicates were represented in each clone. Two weeks after establishment, from each replicate, ten plants for each clone were harvested. The following data were recorded, according to Ceschin et al. [30] as follows: qualitative characteristics (such as frond shape, symmetry, color, apex, and turions), while quantitative characteristics (frond length (mm), width (mm), thickness of aerenchyma layers, number of contiguous fronds (n), length, width, number of roots, and root length) were used in the evaluation.
All data was subjected to the analysis of variance (ANOVA) for the quantitative

Morphological Characteristics
The duckweed clones were transferred to a polypropylene container (15 × 10.5 × 7.5 cm 3 ) filled with 750 mL of half-strength Hoagland solution. Three replicates were represented in each clone. Two weeks after establishment, from each replicate, ten plants for each clone were harvested. The following data were recorded, according to Ceschin et al. [30] as follows: qualitative characteristics (such as frond shape, symmetry, color, apex, and turions), while quantitative characteristics (frond length (mm), width (mm), thickness of aerenchyma layers, number of contiguous fronds (n), length, width, number of roots, and root length) were used in the evaluation.
All data was subjected to the analysis of variance (ANOVA) for the quantitative morphological traits of the studied duckweed clones. Mean and standard deviation of the evaluated traits (n = 10 samples and three replications, total = 30). Mean separation was based on Tukey HSD at the 0.05 probability level.

Molecular Markers Analysis: DNA Barcoding
Genomic DNA was extracted from a single plant from each replicated clones after grinding duckweed fronds in liquid nitrogen using a DNA extraction kit (Qiagen, Hilden, Germany), following the company's instructions. For DNA barcoding, six plastid markers (psbK-psbI, trnH-psbA, atpF-atpH, matK, rpoC1, and rbcL) were used for DNA barcoding and identified different clones of duckweeds. The markers were selected based on duckweed reference sequences, as proposed by the plant working group of the Consortium for the Barcode of Life (CBOL), including three non-coding spacers (psbK-psbI, trnH-psbA, and atpF-atpH) and four plastid coding genes (matK, rpoB, rpoC1, and rbcL) were used to identify and characterize duckweed clones [31][32][33]. The amplicon sizes were estimated according to Wang et al. [8]. The PCR reaction comprised 50 ng genomic DNA, five pmol of each marker, 1X PCR Master Mix ((Promega, Fitchburg, WI, USA), and the volume completed to 25 µL. The PCR reaction was set as follows: initial denaturation at 94 • C for 5 min, denaturation at 94 • C for 1 min, with the annealing temperature set for each marker, and extension at 72 • C for 1 min for 35 cycles. However, the final extension time was 72 • C for 5 min. The PCR products were purified using Wizard SV gel and a PCR clean-up system (Promega, Fitchburg, WI, USA). The purified PCR samples were sent to Macrogen, South Korea (https://www.macrogen.com/en/main/index.php) (accessed on 6 December 2020) for sequencing by ABI3730 automated sequencer using the same markers as in the PCR reactions. The strands were both sequenced and checked for ambiguous nucleotides. 30 bp at the ends of each read were removed, the length of amplicon products was measured, and multiple DNA sequence alignments were generated using ClustalW alignment software [34,35]. The phylogenetic tree was constructed using MEGA X to group the clones with following matrix: DNA weight matrix = IUB, transition weight = 0.50, delay divergent cutoff 30% [36].

Conclusions
This study characterized duckweed clones collected from ten different Saudi Arabia regions at both morphological and molecular levels. The results should help to inform research on the indigenous duckweed clones adapted to the climate of Saudi Arabia. Three out of six plastid markers (trnH-psbA, matK, and atpF-atpH) helped to identify the dominant duckweed species in Saudi Arabia. Evaluation of performing L. punctata, L. perpusilla, and S. polyryiza clones in environments such as in Saudi Arabia is needed for future uses such as animal feed supplies and bioremediation programs.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/plants10112438/s1. Figure S1: Multiple nucleotide sequence alignments of the concatenated sequences of psbK-psbI, trnH-psbA, and atpF-atpH intergenic spacer region, and matK, rpoC1, and rbcL genes generated using ClustalW alignment software. The identical nucleotides are shaded in black; Figure S2: Divergence score matrix of the concatenated sequences of psbK-psbI, trnH-psbA, and atpF-atpH intergenic spacer region, and matK, rpoC1, and rbcL gene, based on a ClustalW alignment with the BLOSUM62 matrix; Table S1: All sequences with their accessions number used for species identification.