Zinc is one of the essential mineral elements for human health that are supplied through vegetables in diets. Across all sites and vegetable types, vegetable zinc contents ranged from 12.57 to 134.54 mg/kg (Table 5). On average, Zn content was lowest in vegetables grown in Hombolo (29.10 mg Zn/kg) and highest in vegetables grown in Ihumwa (98.44 mg Zn/kg) (Table 5). The zinc contents in leafy vegetables in this study are higher than the Zn contents of leafy vegetables obtained from Poland markets, which were reported to have an average range of 23.8 in cabbage and 65 mg Zn/kg in lettuce [46] . Another study involving non-leafy vegetables reported Zn concentrations of 0.074 to 4.75 mg/kg in Bandladesh, which are lower than the Zn levels in the present study. Another study reported average Zn concentrations in vegetables in industrial areas of Bangladesh to range from 19.54 to 42.06 mg/kg [17] . The Zn contents in leafy vegetables from Kilombero and Ihumwa are high, which could be due to use of manure and possibly application of foliar fertilizers containing Zn.

The vegetables from Kilombero and Ihumwa are capable of supplying the recommended maximum allowable Zn intake of 20 mg/day for adults [47] based on the Zn concentration and assumption that the daily intake of vegetables in many Tanzanian communities do not exceed 400 g (dry weight) per person per day. Although some vegetables in Ihumwa had higher Zn concentration (>50 mg/kg) that can supply more than maximum allowable Zn intake of 20 mg/day but none of them supply Zn above daily intake of 60 mg Zn/day. The adverse health effects due to excessive Zn intake can occur at Zn intakes of 450 to 660 mg Zn/day, resulting in low Cu absorption and ceruloplasmin level and anemia,

while Zn ingestion of 4000 to 8000 mg Zn is toxic to humans [2] . However, such adverse risks in human health due to excessive Zn intake is unlikely to occur for Zn supply through food, because Zn is bound in food components [47] . On the other hand, all vegetables from Hombolo and vegetable from Mgudeni 2 in Kilombero cannot supply sufficient Zn intake for human health, and may pose risks of Zn deficiencies in humans. [2] reported that an average daily Zn intake of 5.2 to 7.9 mg/day is low and can result in risks of Zn deficiencies in children to the extent that Zn fortification would be necessary. The present results showed variable Zn concentrations due to differences in management practices and not necessarily due to soil levels per se. However, most Zn levels in the studied vegetables would not exceed the maximum allowable daily intake of Zn, and 14% of vegetables studied are likely to pose risks of Zn deficiencies, especially if low absorption of about 15% [5] is considered.

The Cu content in leafy vegetables studied ranged from 0.07 to 52.37 mg/kg (Table 5). The Cu contents in all vegetables grown in Kilombero were very low (<0.10 mg/kg), which is lower than the sufficiency level range of 5 to 10 mg/kg for many leafy vegetables growth and yield [48] . The Cu concentrations in vegetables from Hombolo ranged from 5.77 to 30.55 mg/kg while those from Ihumwa ranged from 12.58 to 52.37 (Table 5). The cupper concentrations in vegetables from Hombolo and Ihumwa are above the critical level for plant growth and yield [48] . Most of the vegetables in Ihumwa and Hombolo are below the critical levels for phyto-toxicity of 19.40 mg/kg for Chinese cabbage (medium sensitive to Cu phyto-toxicity) to 30.9 mg/kg for Celery (tolerant to Cu phyto-toxicity) that would cause 10% reduction in dry matter yield [39] .

The Cu concentrations of vegetables from Kilombero are insufficient for vegetable growth and yield and may contribute to low vegetable yields in these areas. Similarly, balanced Cu nutrition in humans requires a Cu intake of at least 2.4 mg/day for a net Cu gain, while intake of < 0.8 mg/day results in net losses of Cu [49] . Thus, people consuming the vegetables from Kilombero will have low Cu intake, below 0.8 mg/day (for 400g vegetables per day), and may require supplementation of Cu from other types of diets especially meat-based diet. A review by [50] reported the maximum allowable concentration (MAC) of Cu from vegetables to be 40 mg/kg (fresh weight, fw). In the present study all vegetable samples did not exceed the MAC (Table 2). Reference [51] reported Cu concentrations in non-leafy vegetables of Bangladesh to range from 0.946 to 11.78 mg/kg (fw), which is slightly higher than the Cu concentrations in leafy vegetables from Kilombero and Hombolo, but lower than the Cu concentrations in vegetables from Ihumwa.

A study conducted in Morogoro region a decade ago reported Cu contents in vegetables ranging from 8.85 to 13.5 mg/kg [52] , which are similar to Cu concentrations of vegetables from Hombolo, but lower than those from Ihumwa. All Cu concentrations of all vegetables in this study are lower than the Potential Dietary Toxicity (PDT) due to Cu of 338.98 mg/kg [39] . A recent review on the Cu requirement for human health reported limited evidence on health and impairment of various physiological/metabolic functions such as cardiovascular diseases, reduced cognitive ability, arthritis, and cancer for Cu intake range of 0.6 to 3.0 mg/day and low immunity due to Cu intake range of <0.38 mg/day, suggesting further studies on Cu requirements [49] . Therefore, vegetables from Hombolo and Ihumwa can provide sufficient Cu for human health. However, vegetables from Kilombero are deficient in Cu for plant growth and may not be able to supply sufficient Cu for good human health.

Iron contents of vegetables grown in Dodoma and Kilombero ranged from 152.95 to 1780 mg/kg, and averaged 911.13 mg/kg (Table 5). Vegetables grown in soils from Ihumwa had higher average Fe contents ranging from 581.86 to 1655.39, with mean of 1051 mg/kg, than those from Kilombero and those from Hombolo. Vegetables from Hombolo contained Fe contents ranging from 152.95 to 569.86 mg/kg, with a mean of 274.45 mg Fe/kg (Table 5). Among vegetables, Chinese cabbage, kale and Amaranthus had higher Fe contents than other leafy vegetables across the soils (Table 5) and management practices.

All vegetables contain sufficient Fe for human diets, with recommended nutrient intake (RNI) for Fe for adults being 27.4 mg/day for males to 58.8 mg/day for females [2] , obtained with maximum vegetable intake of 400 g/day. All vegetables from the garden soils studied have sufficient Fe contents for growth and yield. In many vegetables Fe contents of 50 to 300 mg/kg are considered sufficient for leafy vegetable growth and adequate yield [48] . The present adequate levels of Fe in vegetables is supported by the soil Fe levels and management practices. The relatively lower Fe contents in Kilombero vegetables than in Ihumwa vegetables despite adequate pH for micronutrient availability may be due to low use of organic manure in Kilombero as reported by [36] . This shows the important role of manure in enhancing Fe availability over the soil Fe content per se. Ihumwa farmers apply more manure in their vegetable gardens than do Kilombero farmers [36] .

Vegetables are a dependable source of Fe especially in poor families, due to higher Fe contents and availability of Fe from vegetables. Many vegetables contain ascorbic acid which enhances Fe bioavailability in the gastrointestinal tract [11] , unlike cereals diets, which contain high phytate that inhibits Fe bioavailability. However, health benefits of phytate have also been reported, such as reduced risks of non-communicable diseases like various cancers, heart-related disease, diabetes and renal stones [53] . Thus, combinations of vegetables and cereals in the diet will ensure both health benefits of protection against non-communicable diseases and increase Fe bioavailability to humans.

Manganese contents in the vegetables studied ranged from 22.18 to 321.82 mg/kg (Table 5). The vegetables from Hombolo and Ihumwa have generally higher Mn concentrations, averaging 75.59 and 174 mg/kg, respectively, than the vegetables from Kilombero with average Mn contents of 47.65 mg/kg (Table 5). The Mn content in all vegetables is within the sufficiency range of 50 to 200 mg/kg for adequate growth and yield of leafy vegetables as reported by [48] . One vegetable from Bwawani1 had a lower Mn content of 22.18 mg/kg, while two vegetables from Shuleni 1 and Shuleni 2 had higher Mn contents of 321.82 and 289.45 mg/kg, than the sufficiency levels (Table 5). However, none of the Mn contents in vegetables reached the Mn toxicity levels of 1940 mg/kg reported by [54] .

Manganese is an essential nutrient for both plants and humans. The Mn contents in vegetables in this study are adequate for plant growth and yield. While no established recommended dietary intake of Mn has been reported due to insufficient information, the MAC for Mn is 11 mg/kg for adults and 6 mg/day for children below 13 years old [55] . The Mn levels in vegetables studied are lower than the 200 mg/kg Mn content in lettuce leaves and 290 mg/kg Mn concentration in beans leaves grown in contaminated soils of Germany [56] , except for sweet potato leaves from two sites in Ihumwa. The Mn levels in the vegetables presently studied supply higher than the MAC of 11 mg Mn/day for adults, assuming 400 g of vegetables are consumed per day. Although the human body has a mechanism of excreting excess Mn from the body, the possibility of side effects due to excessive Mn intake is reported in populations with Fe-deficient anemia [55] . A study conducted in Bagladesh reported a median concentration of Mn in vegetables to be 65 mg/kg [57] . Manganese toxicity in human due to food intake per se is rare despite the fact that some diets supply more than 20 mg/day of manganese [55] . Rare Mn toxicity due to diet is due to homeostatic mechanism to regulate absorption and excretion [55] [58] . Therefore, the vegetables studied have enough Mn for human health, and if sufficient vegetables are consumed in everyday meal, Mn supplementation may not be needed in these areas.

Soil fertility management practices influence soil properties and hence nutrient uptake by plants, which eventually determine the quality of vegetables in terms of their nutrient contents. In this study, the common soil fertility management practices in vegetable production in Kilombero include use of farm yard manure alone (FYM), FYM and inorganic fertilizers (usually Urea and CAN) and use of FYM and compost from available organic materials such as rice husks. The results show that different fertility management practices used in the study areas resulted in significant differences in soil properties, especially soil TN, available P, Zn and Cu in Kilombero and Hombolo soils (Table 6). In Ihumwa, soil fertility management practices resulted in significant differences in soil K, Cu, Fe and Mn (Table 6). Compost and rice husks used in Kilombero for soil fertility management in vegetable production resulted in significantly the highest soil N (0.3%), P (133.8 mg/kg), and Zn (29.3 mg/kg) (Figure 4). Soil available P and Cu in soils treated with FYM alone did not statistically differ from that of soils treated with compost and rice husks. Soils in Kilombero treated with FYM + urea had the lowest soil total N, P, Zn and Cu (Figure 4).

df = degrees of freedom; * = significant at alpha = 0.05; ns = not significant at alpha = 0.05.

In Hombolo, the fertility management practices included application of FYM + Urea, FYM alone, or FYM + foliar fertilizers. The results showed that among these practices FYM + urea resulted in higher soil total N (2.8%), available P (23.3 mg/kg) and Zn (4.5 mg/kg) than other management practices. However, FYM + urea had significantly the lowest available Cu (1.2 mg/kg) in soil (Figure 4). Farm yard manure + foliar fertilizers resulted in statistically similar soil nutrient concentrations, but resulted in significantly greater Cu (1.8 mg/kg) concentration in soil (Figure 4).

In Ihumwa site, the common management practices consisted of the use of urea alone, FYM + urea, FYM alone or Urea + CAN. The results showed that soil K was significant (p < 0.05) lowest (0.16 cmol_c/kg) in soils managed by FYM + urea compared to soil K in other management practices (ranged from 0.50 cmolc/kg in FYM alone to 0.69cmol_c/kg in Urea+CAN) (Figure 4). Soil Cu was lowest (0.75 mg/kg) in soils managed by use of urea + CAN (Figure 4). Urea alone, FYM + urea, and FYM alone had statistically similar soil available Mn (range 7.32 to 9.59 mg/kg), and Fe (5.49 to 6.73 mg/kg) (Figure 4). Urea + CAN resulted in significantly the lowest Mn of 2.40 mg/kg and the highest soil available Fe of 10.85 mg/kg (Figure 4).

Soil fertility management practices slightly differed among the three sites studied, but the use of FYM and FYM in combination with inorganic N fertilizer were common practices. The soil fertility management effects on soil nutrient status differed among sites due to differences in inherent soil properties among these sites. Reference [22] described the soils of Ihumwa and Hombolo as Haplic Cambisol and Haplic cutanic Acrisols, respectively, with very low P, Ca and Zn, while [17] described Kilombero soils as alluvial clay soil with low to medium Zn, P and K in some locations. Thus, fertility management with addition of P, Zn, Fe, and Mn increased concentrations of these nutrients in the Ihumwa and Hombolo soils. Likewise, high total N, available P, and Zn due to compost and rice husks use is due to high contents of N, P, Zn in rice husks and compost, and slow release of these nutrients from these organic materials used in Kilombero vegetable growing soils. These results further suggest that Zn and Cu in these soils following use of are slightly lower than from the use of compost and rice husks.

In Hombolo, the high soil total N, P, and Zn suggest that the FYM is the major source of N, P and Zn in the soil, because these soils are naturally very low in N, P and Zn [22] . However, the P concentrations in soils under all fertility management practices in Hombolo were still lower than the critical level of 15 mg/kg [32] . Urea supplemented available N, which enhanced plant growth and absorption of other nutrients. It is therefore beneficial to integrate the two sources of nutrients. Higher total N in Hombolo compared to Kilombero is due to differences in leaching and runoff potential, where lower rains in Hombolo may have resulted in less leaching hence higher total N than in soils from Kilombero where rains are high.

Ihumwa soil is deficient in soil K (Table 4; [22] ), hence low K in the soil under FYM + Urea, which is lower than the critical K level of 0.20 cmolc/kg [59] [32] . Low K in FYM + Urea treated soil may be due to low quantity of manure used when combined with urea. In the other management practices soil K is high and it appears that K is supplied by organic manure except for Urea + CAN. Low Cu due to use of Urea + CAN may be explained by non-addition of Cu under this practice and possible antagonistic effect of divalent cation Ca on Cu. This practice’s lowest Cu in soils corresponds to numerically lowest Cu concentration in vegetables from Ihumwa, which was 13.33 mg/kg as compared to 43.33 mg/kg from FYM + Urea and 35.11 mg/kg from Urea alone (data not shown).

Higher Cu concentration in soil but lower Cu concentration in vegetables due to FYM alone may be due to supply of Cu from FYM, and organic matter interference with Cu uptake. It was established that Cu uptake is reduced by SOM because Cu is tightly bound by OM [60] . High soil Fe in the Urea + CAN practice, which corresponds to the highest Fe concentration in vegetables (1429.9 mg/kg), cannot be presently explained. Also the lowest soil Fe in soils supplied with Urea alone cannot be directly explained, suggesting presence of complex factors affecting soil Fe and/or other sources of Fe in soil, which were not captured in this study. The differences in soil nutrient contents due to different management practices can also be explained by differences in inherent soil properties due to parent materials of soils in Kilombero, Hombolo and Ihumwa.

The analysis of variance results showed that none of the micronutrients (Cu, Fe, Zn and Mn) concentrations in vegetables were significantly affected by soil fertility management practices. Soil fertility management practices resulted in significant differences in P, K and Mg concentrations in vegetables from Kilombero, N concentration in vegetables from Hombolo, and K and Ca in vegetables from Ihumwa (Table 7). All other macronutrients did not differ significantly among soil fertility management practices in all sites. In Kilombero, vegetables grown in soils that received compost + rice husks had significantly the highest P (0.73%), K (5.51%) and Mg (0.25%) levels (Figure 5). Kilombero vegetables grown in soils that received FYM alone and FYM + Urea had P levels ranging from 0.54% to 0.61%, K from 3.76% to 5.19% and Mg from 0.09% to 0.13% (Figure 5). Vegetables from Hombolo had the highest concentration of N (3.58%) when FYM + urea were applied as compared to when FYM alone was used (2.48% N) and FYM + foliar fertilizer were applied (2.46% N) (Figure 5). Vegetables from Ihumwa had the highest K (3.08%) when Urea alone was applied and the highest Ca concentration (1.63%) when Urea + CAN were applied (Figure 5).

df = degrees of freedom; * = significant at alpha = 0.05; ns = not significant at alpha = 0.05.

The results showed that management practices directly influenced the absorption of macronutrients and hence nutrient concentrations. The SFMP effect differed among sites due to differences in soil physical and chemical properties. The higher concentrations of P, K, and Mg in vegetables that received compost + rice husks in Kilombero are due to higher concentration of these nutrients in rice husks than under the other fertility management practices. Higher N in Hombolo vegetables due to FYM + Urea suggests better supply of N when organic and inorganic N sources are combined in these sandy loam soils with low SOM [22] . While Urea supplies a high quantity of readily available N, FYM releases N slowly and builds up SOM which reduces leaching losses, hence better N uptake by vegetables. The management practices used in Hombolo and Ihumwa did not lead to differences in P concentration in vegetables despite the fact that there were soil P differences (Figure 4), suggesting that the SFMP used had low ability to supply P. In addition, soils of Ihumwa and Hombolo are also inherently low in P. Thus, lack of significant effects of fertility management practices on vegetable P concentration in Hombolo and Ihumwa may be due the inherently low P in these soils.

In Kilombero soils with soil Zn concentration range of 3.9 to 40.6 mg/kg, soil Zn was positively correlated with K and Mg concentrations in vegetables (Table 8). Concentration of soil Cu ranged from 2.0 to 4.10 mg/kg and was significant and positively correlated with Mg (Table 8). On the other hand soil P ranged from 4.8 to 145.6 mg/kg and was significantly and positively correlated with vegetable K (r = 0.53, p < 0.05) and Cu (r = 0.54, p = 0.05) (Table 8). The soil pH ranged from 5.72 to 6.98, with average of 6.41 in the Kilombero soils studied, and within this range, soil pH was significantly and negatively correlated with Cu concentration in vegetables (r = −0.645; p < 0.05) (Table 8). Similar negative correlation between soil pH and micronutrient availability, including that of Cu, was reported by [61] . Reference [60] reported increased fixation of Cu in long term period with increase in pH in semi-arid alkaline soils of Texas, USA. Although the Cu concentration in soil was sufficient, the availability of Cu to vegetable plants was low as shown by low Cu concentrations in vegetables from Kilombero. Low availability of Cu in these soils was probably because of complex Cu fixation pattern in soils and influence of other soil constituents on availability of Cu for plant absorption. Reference [60] reported best fit of power model kinetics of Cu fixation in alkaline soils in the long term period but second order model in the short run, suggesting that Cu fixation changes over time and in the long term Cu fixation is nonlinear. On the other hand, in the short term, the Cu fixation is controlled by more than one soil property in addition to Cu concentration in soil [60] . The positive correlation between soil P and vegetable K and Cu may be

n = number of observation; SD-standard deviation; Min = minimum; Max = maximum; *Significant at alpha = 0.05.

due to positive effect of P on root growth, which increases root interception and hence absorption of K and Cu by the vegetables [62] . Another positive correlation of P and micronutrients was reported to occur in the presence of vesicular arbuscular mycorrhizal fungi [63] . The negative correlation between soil pH and Cu in vegetables shows that Cu was more available in soils with low pH (<7.0), which is consistent with the availability of cationic micronutrients in low pH as reported by [45] .

Nutrient concentrations in vegetables grown in Hombolo had significant and positive correlation with Soil Fe, CEC and pH. Concentration of K in vegetables was significantly and positively correlated with soil Fe concentration (r = 0.447; p < 0.05) (within the soil Fe range of 3.29 to 42.50 mg/kg) but negatively correlated with soil pH (r = −0.686; p < 0.05) (within the soil pH range of 6.41 to 8.73) (Table 9). Zinc concentration in vegetables was negatively correlated with soil pH and CEC (within the CEC range of 10.8 to 24.20), with r = −0.571 and −0.501, respectively (Table 9).

These results shows that in soils of Hombolo the soil Fe influenced availability of K in vegetables, while soil pH influenced availability of K, P, and Zn in vegetables. The negative influence of pH on P and Zn concentrations in the vegeta-

n = number of observation; SD-standard deviation; Min = minimum; Max = maximum; *Significant at alpha = 0.05.

bles suggests that the high soil pH in Hombolo soils, with a mean pH of 8.08, rendered P and Zn unavailable due to P fixation and precipitation of Zn with hydroxyl ions at those high pH levels [45] [64] . The negative correlation between soil CEC and vegetable P and Zn concentration shows that relatively more P and Zn are available at low CEC within the range 15.73 to 24.20 cmolc/kg). It appears that the increase in total negatively charged colloids increased cationic Zn adsorption, which reduced its availability especially in high pH soils. On the contrary, in Kilombero soils with high CEC (range of 12.4 to 33.60 cmolc/kg) and low soil pH (5.72 to 6.98), CEC did not affect P and Zn concentrations in the vegetables. Therefore, in Hombolo, soil Zn, soil Fe, CEC and soil pH individually or jointly influenced concentrations of K, P, and Zn in the vegetables.

The concentration of Zn in soils was significantly and negatively correlated with vegetable Zn (r = −0.704) and Fe (r = −0.587) concentrations but positively correlated with Mn (r = 0.543) concentration in vegetables grown in Ihumwa soils (Table 10). Soil Fe (range 3.29 to 12.48 mg/kg) and vegetable Fe concentration

n = number of observation; SD-standard deviation; Min = minimum; Max = maximum; *Significant at alpha = 0.05.

were significantly negatively correlated (r = −0.587). Soil CEC (9.6 to 20 cmolc/kg) was significantly and negatively correlated with vegetable Zn (r = −0.411) and vegetable Fe (r = −0.600) (Table 10). No significant correlation was observed between soil pH and nutrient concentrations in vegetables in Ihumwa soils.

The results show that for Ihumwa, micronutrient concentrations in vegetables were influenced by soil Zn, soil Fe and CEC. The negative correlation of vegetable Zn and soil Zn concentration was unexpected, but further alludes to complex phenomena and possibly nonlinear relationship between soil constituents and micronutrient availability in soils [60] , especially in high pH semi-arid soils like those of Dodoma. The Zn concentration in soils of Ihumwa was variable, ranging from deficient (0.35 mg/kg) to sufficient (3.45 mg/kg), while Zn levels in vegetables were also in the sufficiency range of 29.11 to 144.13 mg/kg (Table 5). The negative correlation between vegetable Fe concentration and soil Zn can be attributed to antagonism and competition between Fe and Zn cations for plant absorption, where in this case it appears that high levels of soil Zn in these soils tend to reduce absorption of Fe. Like in the vegetable Zn and soil Zn correlation, the negative correlation between vegetable Fe and soil Fe was unexpected, which suggests some complex association since the soil Fe concentration was not in the toxic range. However, the soil and vegetable Fe concentrations are in the sufficiency range and do not raise any management concerns. Therefore, in Ihumwa, soil Zn, soil Fe and CEC in combination or individually determined the concentration of Zn, Fe and Mn concentrations in the vegetables.