Variation in the mineral element concentration of Moringa oleifera Lam. and M. stenopetala (Bak. f.) Cuf.: Role in human nutrition

Background Moringa oleifera (MO) and M. stenopetala (MS) (family Moringaceae; order Brassicales) are multipurpose tree/shrub species. They thrive under marginal environmental conditions and produce nutritious edible parts. The aim of this study was to determine the mineral composition of different parts of MO and MS growing in their natural environments and their potential role in alleviating human mineral micronutrient deficiencies (MND) in sub-Saharan Africa. Methods Edible parts of MO (n = 146) and MS (n = 50), co-occurring cereals/vegetables and soils (n = 95) underneath their canopy were sampled from localities in southern Ethiopia and Kenya. The concentrations of seven mineral elements, namely, calcium (Ca), copper (Cu), iodine (I), iron (Fe), magnesium (Mg), selenium (Se), and zinc (Zn) in edible parts and soils were determined using inductively coupled plasma-mass spectrometry. Results In Ethiopian crops, MS leaves contained the highest median concentrations of all elements except Cu and Zn, which were greater in Enset (a.k.a., false banana). In Kenya, Mo flowers and MS leaves had the highest median Se concentration of 1.56 mg kg-1 and 3.96 mg kg-1, respectively. The median concentration of Se in MS leaves was 7-fold, 10-fold, 23-fold, 117-fold and 147-fold more than that in brassica leaves, amaranth leaves, baobab fruits, sorghum grain and maize grain, respectively. The median Se concentration was 78-fold and 98-fold greater in MO seeds than in sorghum and maize grain, respectively. There was a strong relationship between soil total Se and potassium dihydrogen phosphate (KH2PO4)-extractable Se, and Se concentration in the leaves of MO and MS. Conclusion This study confirms previous studies that Moringa is a good source of several of the measured mineral nutrients, and it includes the first wide assessment of Se and I concentrations in edible parts of MO and MS grown in various localities. Increasing the consumption of MO and MS, especially the leaves as a fresh vegetable or in powdered form, could reduce the prevalence of MNDs, most notably Se deficiency.


S-Ethiopia
Moringa rivae subsp. rivae in various countries where Moringa grow show that these species are multipurpose. They are used for food, medicine, fodder, fencing, firewood, gum and as a coagulant to treat dirty water [21,23,[34][35][36][37][38]. The foliage, immature pods, seeds, and roots are used both as food and medicine. Young shoots are also cooked and eaten [23,25]. Leaves are either cooked or consumed raw as vegetables. Moringa leaves are used in a similar way as a cabbage and spinach thereby nicknamed 'cabbage tree' [39]. As a food or forage source, Moringa spp. can supply a wide range of essential macro and micro nutrients [4,25,40,41]. The mean concentration of Ca, Cu, Fe, Mg, and Zn in MO leaves collected from a garden in Jalisco State of Mexico were 16100, 9.6, 97.9, 2830 and 29.1 mg kg -1 dry weight (dw), respectively. Similarly, the concentrations of these elements in MS leaves were 12700, 9.1, 69.9, 3690 and 33.7 mg kg -1 dw, respectively [4]. A mean Se concentration of 0.877 mg kg -1 dw was reported in MO leaves grown at six locations ranging from 0.455 mg kg -1 dw in Rwanda to 2.00 mg kg -1 dw in the Solomon Islands [42]. However, systematic analysis of Se has not been conducted at multiple sites within a country and concentrations of other elements such as iodine have not been reported.

Impact of environment on mineral element concentration in Moringa edible parts
The mineral element concentrations in different edible parts of Moringa spp. are affected by the environment in which they grow. For example, the effect of elevation and season on mineral micronutrient concentration of leaves and immature pods of MO and MS was studied in Ethiopia [40]. Concentrations of Ca, Fe, and Zn in Moringa leaves grown in mid-altitude areas during the rainy season were 24800, 578 and 24.3 mg kg -1 dw for MO and 14900, 700 and 24.7 mg kg -1 dw for MS, respectively. In low altitude areas, the Ca, Fe, and Zn concentration in MO leaves during rainy season were 25700, 564 and 26 mg kg -1 dw, while in MS leaves, the concentrations were 24000, 581 and 28.1 mg kg -1 dw, respectively [40]. Other studies have compared MO samples collected from various sites without specifying environmental variables. For example, in their study on the mineral concentration of MO edible parts in two regions of Nigeria, it was reported that Ca, Mg, Fe and Cu concentration in the leaves, pods and seeds were higher in tissues collected from Sheda region than Kuje, Abuja [43]. Similarly, a study conducted in the Punjab province of Pakistan indicated that the Ca, Mg, and Zn concentration in the leaves and pods of MO varied significantly by region [44]. For example, the Ca concentration in MO leaves in Bahawalnagar and Sadiqabad were 22900 and 19000 mg kg -1 dw respectively. A study conducted by Olson et al. [4] indicated variation in leaf elemental concentration between 12 Moringa species grown in a common garden experiment.

Study aims
To our knowledge, no studies have explored the association between plant tissue element concentration of Moringa spp., and the site-specific physico-chemical properties of the soil. Previous studies assessing the variation in the elemental concentration in edible parts of Moringa spp. in various agro-ecological zones have typically been based on generic classifications, e.g., elevation [40]. Furthermore, there is some evidence that Moringa accumulates Se [42] but this has not been widely confirmed in leaves or for other plant parts. Iodine concentrations have not previously been reported in Moringa leaves. The objectives of this study were to: • determine the multi-elemental concentration in the flowers, immature pods, leaves, and seed kernels of MO and MS grown in different agro-ecological zones in Ethiopia and Kenya; • explore the association between MO and MS edible parts mineral element concentration and soil physico-chemical properties; • assess the potential of consumption of MO and MS leaves in alleviating dietary micronutrient deficiencies in sub-Saharan Africa; and • compare the mineral element concentrations in MS and MO edible parts with locally grown cereal and vegetable crops.

Materials and methods
This study was conducted in southern Ethiopia and Kenya. Sample collections from localities in southern Ethiopia were carried out in December 2014 and April 2015, and in July 2015 from localities in Kenya (Fig 2). Edible parts of MO and MS were sampled from plants that were cultivated by Moringa growing households after receiving their consent. The study was carried out on private/communal land with the owners' permission, and it did not involve endangered or protected species. The study received ethical approval from the University of Nottingham, School of Biosciences Research Ethics Committee (SB REC), approval number: SBREC140117A.

Study sites
Edible parts of MO, MS, other food crops, and soil samples, were collected from localities in southern Ethiopia and Kenya (Fig 2 and S1

Plant multi-elemental analyses
Sample collection and preparation. A total of 196 Moringa plant edible parts with ! 3 samples per site for each tissue (i.e., flowers, leaves, immature pods, seeds and roots) were collected from southern Ethiopia and Kenya (). The edible parts collected from MS were limited to leaves due to unavailability of other tissues during the sampling campaign. Cereal grains and vegetable crops were also collected from some of the farmers' fields that grew those crops in combination with Moringa trees in Kenya. Similarly, various cereal and pulse grains were acquired from households that took part in the survey from Ethiopia. Fresh Moringa leaves were washed in the field by using either tap or bottled water. Fresh edible plant samples collected from Ethiopia were air dried and those from Kenya were oven-dried at 40-50˚C at Kenyan Forestry Research Institute (KEFRI) headquarters in Nairobi and transferred to the University of Nottingham, UK, for further processing and chemical analyses. The dried edible parts, and grains were milled using an ultra-centrifugal mill to pass through a 1 mm screen (ZM 200, Retsch GmbH, Haan, Germany).
Nitric acid digestion of plant samples. Subsamples (c. 0.2 g) of the milled plant samples were weighed in triplicate for nitric acid (HNO 3 ) digestion and subsequent multi-elemental analysis. Samples were mixed with 6 mL of HNO 3 (PrimarPlus-Trace Analysis Grade (TAG), Fisher Scientific, Loughborough, UK) in microwave digestion tubes and digested at 140˚C for 20 min (Multiwave PRO, Anton Paar, St. Albans, UK). After cooling, the samples were diluted with 14 mL of Milli-Q water (MQW) (18.2 MO cm; Merck Millipore Milli-Q, Darmstadt, Germany) prior to multi-elemental analysis by inductively coupled plasma-mass spectrometry (ICP-MS; iCAP-Q, Thermo-Scientific, Loughborough, UK) following a further 1-in-10 dilution with MQW.

Soil multi-elemental analysis
Sample collection and preparation. Thirty-three and 62 soil samples were collected from southern Ethiopia and Kenya, respectively (S2 Table). Each sample comprised soil pooled from five locations underneath the canopy of a Moringa tree spp. Bulked samples were air dried and sieved to pass through < 2 mm screen. A subsample of 30 g was taken to the University of Nottingham. From each sample, a 10 g subsample was Agate ball-milled (PM 400, Retsch, Haan, Germany) for multi-elemental analyses.
Multi-acid digestion of soils. Triplicate finely ground soil samples (c. 0.2 g) were digested for two days with 2.5 mL hydrofluoric acid (HF) (40% AR), 2 mL HNO 3 (70% TAG), 1 mL perchloric acid (HClO 4 ) (70% AR) and 2.5 mL MQW in PFA tubes on a Teflon-coated graphite block digester (Model A3, Analysco Ltd, Chipping Norton, UK). On the third day, the hot plate heating was turned off and 2.5 mL concentrated HNO 3 (70% TAG) and MQW were added and heated for 1 h at 50˚C. After cooling, the digestates were made up to 50 mL in plastic volumetric flasks. Multi-elemental analyses were undertaken by ICP-MS following a further 1-in-10 dilution.
Phosphate-extractable soil Se (Se-P). Duplicate soil samples (< 2 mm; c. 2 g) were shaken in SUCT for 1 h on a rotary shaker with 20 mL of 0.016 M potassium dihydrogen phosphate (KH 2 PO 4 ) [45]. The soil suspensions were centrifuged at 2200 rpm for 20 min and 10 mL of supernatant solution was filtered through a SF prior to Se-P analyses by ICP-MS.
TMAH-extractable soil iodine. Finely milled duplicate 2 g soil samples were mixed with 10 mL of 10% TMAH in a SUCT. The soil suspensions were heated in an oven at 70˚C (Memmert GmbH + Co, D 06061, Model 500, Schwabach, Germany) for 3 h and then centrifuged at c. 3000 rpm for 20 min. The supernatant solution was diluted 1-in-10 with MQW prior to analysis for iodine by ICP-MS.
Soil pH. The < 2 mm sieved soil was mixed with deionized water at a ratio of 5 g:12.5 mL in SUCT and shaken for 30 min on a rotary shaker. The pH of the mixture was measured using combined pH meter and electrode (HI-209 pH/mV pH Meter, Hanna Instruments Ltd., Leighton Buzzard, UK). Prior to taking the pH readings, the electrode was calibrated using buffers at pH of 4.01 and 7.00. After each reading, the glass electrode was rinsed by deionized water before measuring the pH of the next sample.

Analytical quality control
For analytical quality control, blanks, duplicates, internal standards and certified reference materials were analysed in all instances of plant and soil analyses. The certified reference materials were tomato leaves (1573A), wheat flour (1567B), and Montana soil II (2711A) from the National Institute of Standards and Technology, Gaithersburg, MD, USA (S3 and S4 Tables). Raw data of the plant and soil sample analytical results is presented as supplementary tables (S34-S39 Tables).

Data analyses
Research data compilation and management were carried out using Microsoft Excel and Access 2016 (Microsoft, Redmond, USA). Statistical analyses of the elemental concentration in edible plant parts and the soils were conducted using IBM 1 SPSS 1 Statistics version 22 (IBM Corp., New York, USA). The Shapiro-Wilk test for normality of the distribution of the data and Levene's test for homogeneity of variance were run to select between parametric and nonparametric analyses of variance (ANOVA) (S5-S16 Tables). In addition, visual assessments of the data distributions were made. Due to the small sample size at each locality, most of the plant and soil elemental concentration data did not meet the assumptions of parametric ANOVA; logarithmic transformation did not improve the non-normal distribution and heteroscedasticity of the elemental concentration data. Hence, Welch's robust test for equality of means was applied to test the variation in elemental concentration by locality. Spearman's rank correlation analysis was conducted using GenStat 1 version 17 (VSN International, Hemel Hempstead, UK) to assess the association between soil physico-chemical properties and Moringa leaves elemental concentration, and relationships between elemental concentrations in edible parts of various vegetables. Box plots of plant and soil elemental concentration and pH were drawn using Tableau 1 Desktop Professional Edition version 10.0.0 (Tableau Software Inc., Seattle, Washington, USA). Outliers were not included in the box plots. Plant edible parts with sample size < 3 per locality were excluded from statistical analyses. For instance, there was only one MO and MS sample at Baringo and Ramogi, respectively. These were not included in the data analyses.

Results
Plant edible parts and soil elemental concentration analytical results for calcium (Ca), copper (Cu), iodine (I), iron (Fe), magnesium (Mg), selenium (Se), and zinc (Zn); and soil pH are reported. The association between plant edible parts elemental concentration and soil properties, and variation in elemental concentration by location are also reported. Furthermore, comparisons are made among Moringa spp. edible parts, maize and sorghum grains, beans, amaranth leaves, baobab fruit, brassica leaves, and enset (Ensete ventricosum a.k.a., false banana), mineral element concentrations.

Moringa elemental concentration
The concentrations of mineral elements in Moringa leaves, immature pods, seeds and flowers and variations by localities are presented below, and summarised in Figs 3-7, S17-S21 Tables and Table 2.
Moringa oleifera leaf elemental concentration. The overall mean concentrations of Ca, Cu, I, Fe, Mg, Se and Zn in MO leaves were 18300, 6.92, 0.218, 202, 5390, 4.25 and 35.6 mg kg -1 dw, respectively (S17 Table). Mineral element concentration of the MO leaves varied significantly (p < 0.05) between localities, except for Ca (Fig 3 and S27 Table). There was no systematic variation in the relative concentration at a given location for these elements, although Kibwezi had the highest values of the trace elements (Cu, Se, Zn).
Moringa stenopetala leaf elemental concentration. The overall mean concentrations of Ca, Cu, I, Fe, Mg, Se and Zn in MS leaves were 21100, 4.53, 0.07, 162, 6440, 1.66 and 22.2 mg kg -1 dw, respectively (S18 Table). Mean Cu, I, Mg and Zn differed significantly between localities (p < 0.05), while Ca, Fe, and Se concentrations of MS leaves did not differ significantly between localities (S28 Table). Moringa stenopetala leaves collected from Kenya (n = 5) had higher median concentrations of mineral elements than those from Ethiopia (n = 36) except Cu and Zn (Fig 4 and S18 Table). MS leaves from Hawassa, southern Ethiopia had significantly (p < 0.05) higher concentration of Zn and lower Mg than those from Baringo Island, Kenya. On the contrary, MS leaves collected from Baringo island contained significantly (p < 0.05) higher concentrations of Se and I than all samples from localities in Ethiopia. The concentration of Cu in MS leaves collected from Baringo island were significantly (p < 0.05) lower than samples from Derashe, southern Ethiopia.
Moringa oleifera immature pods elemental concentration. The overall mean concentrations of Ca, Cu, Fe, Mg, Se and Zn in MO immature pods were 3600, 5.42, 65.4, 2860, 2.36 and 27.6 mg kg -1 dw, respectively (S19 Table). The distribution of MO immature pods elemental concentration in comparison with the overall median value varied between the elements and locations (Fig 5). For example, the median elemental concentration in the immature pods collected from Kibwezi were generally higher than the overall median concentration in Kenya. The median Ca and Mg concentration in immature pods collected from Ramogi and Ukunda were below the overall median concentration in Kenya. There were significant differences (p < 0.05) in the Cu, Fe and Mg but not (p ! 0.05) Ca, Se and Zn mean concentrations of MO immature pods collected from different localities (S30 Table).
Moringa oleifera seeds elemental concentration. The overall mean concentrations of Ca, Cu, Fe, Mg, Se and Zn in MO seeds were 1310, 4.18, 49.2, 3080, 3.59 and 44.8 mg kg -1 dw, respectively (S20 Table). Overall median elemental concentration in MO seeds varied between the elements and locations (Fig 6). There was no significant difference (p ! 0.05) in the mean elemental concentration of MO seeds collected from different localities (S31 Table).
Moringa oleifera flowers elemental concentration. The overall mean concentrations of Ca, Cu, Fe, Mg, Se and Zn in MO flowers were 3650, 6.40, 253, 2830, 2.81 and 32.7 mg kg -1 dw, respectively (S21 Table). The distribution of MO flowers elemental concentration in comparison with the overall median value varied between elements and locations (Fig 7). There were significant differences (p < 0.05) in the mean Ca, Cu, Fe, Mg and Se, but not Zn concentrations of MO flowers collected from different localities (S32 Table).

Comparison of elemental concentration between crops
For comparison, elemental concentration of Moringa spp. edible parts and other vegetables, fruits, and staple cereal crops are presented in Table 2. On a weight-for-weight basis, in Ethiopia, MS leaves contained the highest median concentrations of all elements except Cu and Zn. Median concentrations of Cu and Zn were highest in enset and beans, respectively ( Table 2).
In Kenya, on weight-for-weight basis, Moringa edible parts had the highest median Se concentration ranging from 1.56 mg kg -1 in MO flowers to 3.96 mg kg -1 in MS leaves. The median concentration of Se in MS leaves was 7-fold, 10-fold, 23-fold, 117-fold and 147-fold more than that in brassica leaves, amaranth leaves, baobab fruits, sorghum grain and maize grain, respectively. The median Se concentration in MO seeds was 78-fold and 98-fold greater than sorghum and maize grain, respectively. Seeds of MO had the highest median Zn concentration while amaranth leaves contained comparable quantities of Zn with MO flowers and leaves. The median Zn concentration in MO seeds was 2-fold greater than in maize and sorghum grain. (Table 2).

Soil pH and elemental concentrations
Ninety percent of the soil samples from Ethiopia and 97% of those from Kenya had pH >7 (S37 Table). The soil pH at the three localities in Ethiopia ranged from 6.12 in Hawassa to  Table 3 and S22 Table). Welch's robust test of equality of means showed that the soil pH varied significantly between localities (p < 0.05) (S33 Table). Similarly, Welch's robust tests of equality of mean soil elemental concentrations showed that there was significant difference (p < 0.05) between soils collected from various localities (S33 Table). Descriptive statistics of the soil physico-chemical properties across all localities in Ethiopia and Kenya are summarized in Table 3. Soil samples from Baringo, Kibwezi and Ramogi localities were the three with highest median phosphate-extractable Se concentration. Total Se concentration was highest in soils from Baringo, Hawassa and Ramogi. With respect to total soil iodine, Ramogi, Kibwezi and Mbololo soil samples had the highest concentrations. Total Zn concentration in soil samples from Hawassa were 2-fold, 4-fold and 3-fold more than the median Zn concentration from soils in Ethiopia, Kenya and overall median Zn concentrations.

Relationships between MO edible parts and other vegetables
The association between elemental concentrations of the edible parts of MO are presented in S40 Table. Calcium concentration in MO flowers showed highly significant (p < 0.01) positive correlation with the Ca, Cu, Mg, and Se concentration in MO immature pods. Contrary to this, the Cu concentration in MO flowers had a significant (p < 0.01) negative correlation with the Ca in the seeds and leaves. The Se in the MO flower was highly significantly (p < 0.01) correlated with the Se in leaves and immature pods. The association between the elemental concentrations in MO edible parts and other vegetables cultivated on the same location are presented in S41-S43 Tables. The Fe concentration in MO leaves showed significant (p < 0.05) positive correlation with the Fe, and negative correlation with the Se concentration in amaranth leaves (S41 Table). Calcium concentration in MO leaves showed significant  Table). Similarly, Ca concentration in amaranth leaves showed a significant (p < 0.01) negative correlation with the Ca, Cu, and Mg concentrations in brassica leaves (S43 Table).

Relationships between Moringa leaves elemental concentration and soil properties
The concentration of Cu in MO leaves was significantly (p < 0.05) positively associated with total soil Cu and Fe concentrations. Similarly, MO leaf Fe concentration also showed a statistically significant (p < 0.05) positive correlation with soil Fe, Se and Zn. Selenium concentration in MO leaves showed a stronger significant (p < 0.05) positive correlation with phosphate extractable (Se-P) soil Se than the total soil Se (S23 and S24 Tables). The Ca and Zn concentration in MO leaves showed no statistically significant (p ! 0.05) correlation with any of the reported soil properties.
Moringa stenopetala leaf Fe concentration was significantly (p < 0.05) positively correlated with soil Mg, Se and Se-P. Similarly, MS leaves Mg concentration showed a statistically significant (p <0.05) negative correlation with soil Zn concentration. The Se content of MS leaves indicated a strong and significant positive correlation with the soil Se-P (S25 and S26 Tables). However, the Ca, Cu and I concentration of MS leaves did not show significant (p ! 0.05) correlation with any of the soil properties.

Elemental concentration in edible parts of Moringa spp.
This study is the first comprehensive analysis of Se concentrations in different edible parts of MO and MS grown in various localities. Four previous studies reported Se concentrations in MO leaves, from Niger (27.1 mg kg -1 dw) [46], Solomon Islands (2 mg kg -1 dw) [42], South Africa (363 mg kg -1 dw) [41], and Mexico, Lombardia (0.096 mg kg -1 dw) and San Pedro (1.07 mg kg -1 dw) [47]. Our MO leaf Se concentration (mean = 4.25 and median = 2.73 mg kg -1 dw) is consistent with results from Solomon Islands and Mexico, but differ markedly from the South African data. The results of the analyses from Niger were based on two samples. Taking this into account, and the fact that a MO leaf sample (L-MO-29-MBO) from Mbololo, Kenya, had a mean Se concentration of 21.2 mg kg -1 dw from triplicate analyses (S34 Table), the findings from Niger were reasonably consistent with ours. Our attempt to verify the very high reported concentrations from the South Africa study by contacting the authors was not successful. Iodine concentrations in MO and MS leaves or other parts have not been reported previously to our knowledge.  Table 4 summarises 11 previous studies of MO leaf elemental concentrations alongside data from the present study. Allowing for differences in analytical method and likely inter-study variation in leaf maturity there are many broad similarities. For example, the mean Ca concentration in MO leaves in the present study is the fourth lowest following MO leaf samples collected from Kuje, Nigeria [43], Hawassa, Ethiopia [48] and Jalisco state, Mexico [4], while the mean Zn concentration is the second highest following MO leaves collected from Thailand [49]. Similarly, mean elemental concentrations in MS leaves collected from Ethiopia and Kenya in the current study indicated inconsistent variation when compared with previous studies. For instance, the mean concentration of Ca in MS leaves of 21 g kg -1 dw in the present study is comparable with that reported from Ethiopia [40] (19.8 g kg -1 dw), and greater than the 12.7 g kg -1 dw reported from Mexico [4]. However, the Fe concentration (162 mg kg -1 dw) in the present study is far lower than that reported from Ethiopia (666 mg kg -1 dw) [40]. Mean elemental concentrations in MS leaves collected from Ethiopia and Kenya in the current study indicated inconsistent variation when compared with previous studies. For instance, the mean concentration of Ca in MS leaves of 21 g kg -1 dw in the present study is comparable with that reported from Ethiopia [40] (19.8 g kg -1 dw), and greater than the 12.7 g kg -1 dw reported from Mexico [4]. However, the Fe concentration (162 mg kg -1 dw) in the present study is far lower than that reported from Ethiopia (666 mg kg -1 dw) [40].
The mean concentration of Ca (3,600 mg kg -1 dw) in the immature pods of MO in the current study was higher than that reported from Ethiopia (2,740 mg kg -1 dw) [40] and Pakistan (2,740 mg kg -1 dw) [44]. However, the Fe concentration (65.4 mg kg -1 dw) in the MO immature pods in this study was much lower than that reported from Ethiopia (510 mg kg -1 dw) [40] and Pakistan (510 mg kg -1 dw) [44]. The Cu, Mg and Zn concentrations in the immature pods of MO reported from Ethiopia [40] are comparable to our findings. However, the concentration of Cu in the immature pods of MO reported from Pakistan (26.6 mg kg -1 dw) [44] was higher than the results in this study (5.42 mg kg -1 dw). Elemental concentration in MO seed kernel in the present study showed inconsistent variation as compared to previous studies in two regions of Nigeria [43]. For example, the mean concentration of Cu (4.2 mg kg -1 dw) and Fe (49.2 mg kg -1 dw) in MO seed kernel in the present study is lower than the MO seed kernel samples collected from Sheda region of Nigeria with Cu and Fe concentration of 34.2 mg kg -1 dw and 118.5 mg kg -1 dw, respectively. Calcium concentration of 1310 mg kg -1 dw in the present study is higher than the concentration of Ca (1029 mg kg - 1 ) in MO seed kernel collected from Kuje region of Nigeria [43]. The variation in elemental concentration in immature pods and seeds of MO this study and others can be attributed to the variation in environment in which MO grew, intra-specific variation in the MO, the variation in maturity levels of immature pods, and the difference in analytical method pursued.
There are no studies we are aware of that report the elemental concentrations in the flowers of Moringa species.

Variation in Moringa spp. element concentration
Variation in the elemental concentrations of the different edible parts of MO and MS can be due to the impact of the environment/management, the effect of intra-and inter-specific genetic variation [4], and the interactions between the genetics and environment [40]. The seeds and/or planting materials sources of the Moringa trees from which the edible parts were sampled were not traceable and so this discussion is limited to environment/management factors. Samples were collected from trees that were grown and managed by households in various localities of Ethiopia and Kenya. Different households pursue various tree management regimes, such as, lone trees, hedgerow, woodlot, pollarding, lopping, watering, fertilizing, intercropping, etc. For example, some households place household wastes and manure that can supply nutrients to the trees and some may water their plants during dry season. Stages of growth, for instance, of the leaves at the time of surveying varies due to variation in management regime, and climate and soil type may all contribute further to variation of the elemental concentration in the edible parts. The positive correlation between some of the Moringa edible parts elemental concentration and soil chemical properties indicate the significance of the soil environment in which the plants grow besides the inherent genetic ability of these species to absorb and translocate mineral elements to edible parts. In addition, it is not only the quantity of the mineral element available in the soil which impacts on the Moringa spp. edible parts elemental concentration but also the chemical form in which the element exists in the soil [51]. The stronger positive correlation of Moringa leaves Se concentration with KH 2 PO 4 extractable soil Se than the total soil Se was an indication of the association between Moringa edible parts and phyto-available soil elemental concentration.

Moringa spp. role in human Se nutrition
Selenium deficiencies are widespread in sub-Saharan Africa [3,33]. For instance, based on 2009 food supply data from the Food and Agriculture Organization, national level Se deficiency risks in Ethiopia and Kenya were estimated to be 35.5% and 58.3%, respectively [33]. Based on seven day dietary recall survey conducted in the year 2010-2011, Joy, Kumssa et al., [3] estimated that 81% of Malawian households had insufficient Se to meet dietary requirements. Similarly, in northwest Ethiopia, Gonder town, a cross-sectional study on school children (n = 100) using blood serum concentration of mineral nutrients reported 62% of the children were deficient in Se [52]. Gashu et al. [53] reported Se deficiency risk in school children in the Amhara region of Ethiopia to be 58% (n = 349).
Moringa spp. edible parts contain high concentrations of Se and the leaves have similar levels of the 6 other reported mineral elements to other leafy vegetables grown in the same localities. Table 5 summarizes the Recommended Daily Allowances (RDA) for an adult male of Ca, Cu, I, Fe, Mg, Se and Zn, concentrations of these mineral elements in MO and MS leaves collected from various localities of Ethiopia and Kenya, and percentage of RDA fulfilled by consuming 100 g of fresh Moringa leaves per day. The RDA is a daily nutrient intake level that fulfils the nutrient requirements of~98% of the healthy individuals in an age-and sex-specific population [54]. Moringa oleifera grown without Se fertilizer can provide 100% of the RDA of a healthy adult man which is comparable with Se obtained from a similar quantity of carrots biofortified with 1 kg ha -1 of Se fertilizer [55], and maize biofortified with 5 g of Se ha -1 at the level of the Malawian population maize consumption [56]. A daily consumption of 100 g fresh leaves of MS grown in Ethiopia can fulfil 41% of the Se RDA, while MS grown in Kenya can provide 144% of the Se RDA for a healthy adult man. Consumption of fresh leaves or leaf powders of MO and MS, for example, can help at least to reduce the many MNDs and alleviate Se deficiency if interventions target vulnerable populations living in localities where these Moringa species grow vigorously. Besides, Moringa leaf powders can be stored for use during the dry season and transported and traded with areas where Moringa is not cultivated to fight against MNDs. In areas where rain fed agriculture is practiced, other vegetables, for example, Brassica can be used to diversify sources of dietary mineral elements and Moringa leaf powders can be stored and used when they are needed most during the dry season.

Conclusion
In addition to the high selenium concentration, Moringa spp. leaves are rich in proteins and β-carotene [4,50], possess anti-oxidant properties [17], contain low concentrations of antinutrients [57][58][59][60], may be used in treating ailments [18,61], the seeds are used as water coagulant [62], and they grow under marginal environmental conditions providing much needed ecological services (for example, shade, wind break, etc.). Moringa oleifera is naturalized while MS is indigenous to Kenya and Ethiopia. Where these species grow, the population have indigenous knowledge of their multiple uses including the high nutritive values [63]. Nonetheless, the utilization of these species as food is limited to specific localities and communities [8], they are neglected in terms of research and development, and the trees can be classed as underutilized crops [64,65]. Agricultural and health extension work to popularise the production and consumption of MO and MS may be a useful strategy to complement efforts to alleviate dietary MNDs through dietary diversification, and use of Moringa leaf powders to fortify meals in the dry season when other leafy green vegetables are not available. In addition, variations in mineral micronutrient concentrations suggest that breeding efforts to increase the nutritional value of Moringa foliage may be successful. However, research is also required to determine the bioavailability of nutrients from Moringa edible parts. The Moringaceae belongs to the same order (Brassicales) as Brassicaceae [5] which are known to be Se accumulators [66,67]. Hence further studies on the Se concentrations in edible portions of the other Moringa species are important to understand and exploit the potential of the family in the fight against human Se undernutrition. Supporting information S1