In vitro bioavailability‐based assessment of the contribution of wild fruits and vegetables to household dietary iron requirements among rural households in a developing country setting: The case of Acholi Subregion of Uganda

Abstract Wild fruits and vegetables (WFV) are believed to contain substantial quantities of micronutrients and are commonly consumed in rural areas of developing countries endowed with natural vegetation. Previous studies that provided evidence on the contribution of WFV to household micronutrient intake in a developing country setting did not consider the effect of antinutritional factors. Therefore, applying the in vitro bioavailability assessment technique and using the Acholi subregion of Uganda a case area, this study examined the contribution of commonly consumed WFV to the pooled annual household dietary requirement for iron. Laboratory analysis showed that the concentration of antinutrients varied with plant species but the pool was dominated by phytate (10.5–150 mg/100 g) and phenolic substances (38.6–41.7 mg GAE/g). In vitro iron bioavailability varied with plant species was quantitatively higher from vegetables than fruits by 27% although total concentration of the micronutrient was higher in fruits than vegetables by 142%. Nutritional computation, taking into account, household composition, and physiological status revealed that consumption of WFV resulted in a median contribution of 1.8% (a minimum of 0.02 and a maximum of 34.7%) to the pooled annual household dietary iron requirements on the basis of bioavailable iron fraction. These results demonstrate that WFV contributes meagerly to household iron needs but may serve other dietary and non‐nutrient health purposes.

and a number of positive outcomes have been achieved over the last decades. A typical example is the reduction in the prevalence of food insecurity from 32.8% to 26.8% observed in SSA between 1990(FAO et al., 2012Fraval et al., 2019). Despite this achievement, SSA remains the region with the highest proportion of hunger affected population globally (1 in every 5 people) (FAO, 2019).
In SSA and other developing regions of the world, food insecurity is highly associated with both macro-and micronutrient undernutrition (Engelbert et al., 2013;Xie et al., 2018). Among micronutrients, iron deficiency remains the most common nutritional challenge of public health importance with 43% of children under 5 years and 29% of women of reproductive age being iron deficient (Stevens et al., 2012;Baker & Greer, 2010;Dupont, 2017;Sharma & Dhandoria, 2019). Besides causing developmental consequences (e.g., stunting, impaired cognitive development, increased childhood, and maternal morbidity), iron deficiency in the context of nutrition transition may present additional deleterious health consequences when combined with obesity and other related noncommunicable chronic diseases (Eckhardt, 2006;Nazia et al., 2019).
Despite the well-known negative health consequences associated with inadequate iron availability in the human body, intake of iron is one of the lowest among essential micronutrients in developing countries. This situation predisposes vulnerable human groups (pregnant women, children, adolescents, and the elderly) in such countries to iron deficiency health complications (Loh & Khor, 2010;Means, 2020). For instance, in Uganda, 53% of children under five are iron deficient, and 32% of women are anemic, while iron deficiency anemia accounts for 75% of all anemia during pregnancy (UBOS, 2017;Horowitz et al., 2013). Low intake of micronutrients such as iron has been found to be critical most especially among the poor segments of the population that live in rural areas in developing countries (Angeles-Agdeppa et al., 2019;Biesalski & Black, 2016). This is notwithstanding the fact that the majority (62%) of the population in such countries live in rural areas (World Bank, 2018) and lack financial resources to access highly bioavailable iron-rich animal source foods (Aguilar & Sumner, 2020).

Information available in literature indicates that WFV, in areas
where they are consumed, contribute significantly to household intake of essential micronutrients such as iron (Borelli et al., 2020;Okidi et al., 2018). This is because wild plant species have been found to be richer in micronutrients compared to the cultivated variants (Mavengahama et al., 2013). Nonetheless, the occurrence of unacceptably high prevalence of anemia among children under five (71%) and women of childbearing age (47%) in certain areas such as the Acholi subregion (UBOS, 2017) raises question on the effective contribution of WFV to household nutritional iron needs in areas where they are consumed. Previous studies that assessed the contribution of WFV to household nutrition in Acholi subregion focused on total iron without due attention to bioavailability of the nutrient (Okidi et al., 2018) and yet plant species in general contain antinutritional factors such as phytates, oxalate, tannins, and saponin that potentially reduce iron bioavailability from plant-based foods (Mihrete, 2019;Natesh et al., 2017) thereby exposing resource constrained rural communities that rely on plant-based foods to high risk of iron deficiency and related health complications. Using the Acholi subregion of Uganda as a microcosm for rural areas in developing countries where WFV are consumed and applying the in vitro bioavailability-based technique and household nutritional needs computation, this study examined the contribution of WFV to household requirement for bioavailable iron. This information is necessary for planning nutritional interventions to improve micronutrient intake in areas where WFV are consumed.

| Study design and study area
A cross-sectional study design that made use of primary data from the current study and secondary data from Okidi et al. (2018) was applied. Samples of WFV used for laboratory analyses as well as households that participated in nutritional contribution evaluation were drawn from Amuru and Gulu districts ( Figure 1).
Primary data were collected on the level of antinutritional factors and in vitro bioavailability of iron, whereas secondary data on household consumption of WFV were obtained from Okidi et al. (2018). The two districts were purposively selected because previous studies that documented the consumption of WFV were conducted in them (Okidi et al., 2018;Oryema et al., 2013). Second, these two districts are located in Northern region of Uganda where the highest iron deficiency anemia prevalence prevails (UBOS, 2017;Obai et al., 2016;FANTA & USAID, 2010). The two districts cover a total land area of 11,732 km 2 , comprised of open water and swamps (180 km 2 ), arable land (10,301 km 2 ), national park and game reserves (982 km 2 ), and forest covers (371 km 2 ) (ACF, 2007).
The main agro ecological zone in the two districts is savannah grassland and characteristically experience both wet and dry seasons (Langdale-Brown et al., 1964). The average annual rainfall received is approximately 1,500 mm. The wet season extends from April to October with peaks in May, August, and October. The dry season starts in November and lasts up to March (ACF, 2007).
The main economic activity in the study area is subsistence agriculture (57.8%). Amuru and Gulu districts have a total population

| Sample collection and preparation
Samples of 16 wild fruits and vegetables reported to be mostly consumed in Amuru and Gulu districts (Okidi et al., 2018)  collected. Samples were transported and stored in the refrigerator at 4°C till used. Fruit samples were washed with clean tap water and depulped. The fresh pulps were then oven-dried at 45°C for 72 hr as described by Okullo et al. (2010) and ground to fine powder using an electric grinder (Brooks Crompton Series 2000, Bradford, UK).
For vegetable samples, leaves were picked from stems, washed with tap water, rinsed, oven-dried, and ground using the same electric grinder already stipulated with cleaning in between samples. Ground samples of each fruit and vegetable species were packed in airtight high-density poly ethylene (HDPE) for laboratory analyses.

| Determination of antinutritional factors
The antinutritional factors considered in this study were phytate, oxalate, tannins, saponin, and total phenols. These antinutritional factors were chosen because of their potential to bind and interfere with bioavailability of iron from plant-based foods (Natesh et al., 2017). Phytate was determined using high-performance liquid chromatography (HPLC) procedure previously developed by Lehrfeld (1987 (Schanderi, 1970). The spectrophotometer used for oxalate detection was also applied except that absorbance measurements were performed at 700 nm. The content of tannin was expressed as a percentage. Saponin was determined using the weight difference method previously used by Obadoni and Ochuko (2002) and as slightly modified by Rout and Basak (2015). All the reagents used (ethanol, n-butanol, and sodium chloride) were of analytical grade and obtained from BDH (BDH, Kampala, Uganda). Total phenols were determined using the Folin-Ciocalteu spectrophotometric assay method as previously described by Marinova et al. (2005). Measurements of absorbance were performed at 750 nm using the same spectrophotometer model used for oxalate and tannin quantification. All reagents used (Na 2 CO 3 , Folin-Ciocalteu's phenols) were also of analytical grade and obtained from BDH. The total phenolic content was expressed as mg gallic acid equivalents (GAE)/g dry weight. All analyses were performed in duplicates.

| Determination of in vitro bioavailability of iron
Before bioavailability could be assessed, total iron concentration in the samples had to be determined. This was achieved using the flame atomic absorption spectrophotometer (FAAS) (Analytik Jena, content, bioavailability of the micronutrient was determined using in vitro dialysability method. This method determines the fraction of dialysable iron following sequential digestion of the sample in simulated gastric and pancreatic medium (Luten et al., 1996). Both the pancreatic and gastric processes were performed according to Chiocchetti et al. (2018). The only modification was that for pancreatic digestion, a segment of dialysis tubing (Ø = 20.4 mm; MMCO of 10k Da; Sigma-Aldrich, Malaysia) containing NaHCO 3 (an amount equivalent to the moles of NaOH needed for the pancreatic digestion) was used.

| Determination of the contribution of bioavailable iron to household iron requirements
To determine the contribution of bioavailable iron from WFV to the annual household iron requirements, a three-stage process was followed. First, RDAs for the micronutrient for healthy members in the age groups of 7-12 months, 1-3 years, 4-8 years, 9-13 years, 14-18 years, 19-30 years, 31-50 years, 51-70 years, and >70 years segregated by physiological status in terms of pregnancy or lactation (age groups of 14-18 years, 19-30 years, 31-50 years) and sex (male or female) in a given household were aggregated for a day and This RDA reference has been applied before in other studies conducted in Uganda (Isabirye et al., 2020;Okidi et al., 2018). Second, secondary data on the aggregated annual consumption levels of various WFV species (quantities in grams) by each household that participated in a previous study (Okidi et al., 2018) and primary data on iron content and bioavailability levels determined in the current study were used to derive the quantity of bioavailable iron consumed by each household over a one-year period. Third, the contribution of bioavailable iron from WFV to the pooled annual household dietary requirement for iron was computed as a fraction (proportion) of the expected pooled annual household RDA for the nutrient.

| Data analysis
Data were analyzed using IBM Statistical Package for the Social Sciences (SPSS) software version 22.0. Kolmogorov-Smirnov and Shapiro-Wilk tests were performed to check for normality of data for various aspects. Normality tests revealed that data on levels of antinutritional factors, levels of total iron, and levels of bioavailability of iron were normally distributed. However, data on percentage contribution of WFV to the pooled household annual RDA for the micronutrient were not normally distributed. On this basis, differences in the level of each antinutritional factor, total iron, and bioavailable iron among fruits or vegetables species studied were determined using one-way analysis of variance (ANOVA). Means were separated using Tukey's honestly significant difference (HSD) test. Pooled antinutritional factor content, total iron, and bioavailable iron between fruits and vegetables were compared using independent sample t test. Finally, the contribution of WFV to the pooled annual RDA for iron among households in the study area was estimated by calculating the median of the percentage contribution attained for the sample size of 192 households that were involved in the study. For all statistical analysis, the level of significance was fixed at 5%. Table 2 presents the levels of antinutritional factors in various wild fruits studied. Generally, the contents of antinutritional factors were dependent on fruit species and the specific antinutritional factor in question. All fruits had tannin contents less than 1% and in some cases not detectable, except in Capsicum frutescens which was above that level by approximately 0.6%. The contents of phytate in the fruits ranged from 2.36 to 17.26 mg/100 g, with most of them having at least 10 mg/100 g except for Borassus aethiopum, Aframomum angustifolium, and Physalis macrantha.

| Levels of antinutritional factors in wild Fruits and Vegetables
There was some clustering of fruit species in terms of the levels of the antinutritional factors. The levels of phytates were the same in Comparison of pooled concentration of specific antinutritional factors between wild fruits and vegetables is presented in Table 4.
Generally, the levels of all antinutritional factors were not different between fruits and vegetables except phytate which was higher in vegetables than fruits by a factor of 14.

| Bioavailability of Iron from wild Fruits and Vegetables
With regard to fruits, generally, the concentration of total iron varied with the plant species and ranged from 0.81 to 5.97 mg/100 g ( In the case of vegetables, iron was detected in all the vegetable species investigated ( Estimated 78% of the vegetables studied contained biologically available iron above 10%. As observed in the case of fruits, the degree of bioavailability of iron was not concomitant with the level of total iron. Pooled concentration of total iron and iron bioavailability in fruits or vegetables is presented in Table 6. Generally, iron was more abundant in fruits than in vegetables by a factor of about 2 (p < .05). However, iron was more bioavailable in vegetables than in fruits by approximately 3.2% (p < .05).

| Contribution of bioavailable iron by wild fruits and vegetables to household iron requirements
The distribution of the level of intake of total iron, bioavailable iron from wild fruits against the expected annual pooled household RDA  for iron as stipulated under section 2.3 among households (192) that participated in the study is presented in Figure 2.

TA B L E 4 Comparison of pooled concentration of antinutritional factors between wild fruits and vegetables
Generally, for most of the households, intake of total and bioavailable iron from wild fruits was all below the pooled annual household RDA. In terms of total iron, the median contribution of wild fruits to the pooled annual household RDA was 7.6% (a mini-

| D ISCUSS I ON
In the current study, the levels of antinutritional factors in WFV commonly consumed in Acholi subregion of Uganda were examined (
Comparing the pooled content of each antinutritional factor between fruits and vegetables indicates identical levels except phytate which was significantly higher in vegetables than in fruits (Table 4). This suggests that wild vegetables would provide less bioavailable iron due to the high content of the phytate, a principal antinutritional factor that binds iron (Natesh et al., 2017). The results of this study have revealed that bioavailability of iron was dependent on plant species (Table 5). This variation can be attributed to differences in the composition of antinutritional factors and their respective levels in various plant species investigated (Table 2 and 3). Average iron absorption from heme food sources such as meat range from 15% to 35%. However, it varies from 40% during iron deficiency to 10% during iron repletion (Hurrell & Egli, 2010). On the other hand, for plant-based diets iron bioavailability can be as low as 5%-12% (Hurrell & Egli, 2010;Blanco-Rojo & Vanquero, 2018). This low level of iron bioavailability from plantbased diets expose resource constrained households especially those in rural areas of developing countries to persistent iron deficiency and related complications.
There is sufficient evidence to suggest that antinutritional factors in plant-based foods contribute to low iron bioavailability (Biesalski & Black, 2016;Natesh et al., 2017;Welch & Graham, 2009). Among antinutritional factors that lower iron bioavailability, the most cited include tannins, phytate, oxalate, saponin, and total phenolics that were all considered in the present study ( Table 2). The present study recorded iron bioavailability of 11.7% from fruits and 14.8% from vegetables (Table 6). These values are supported by the work of Scheers et al. (2015), but lower than the values reported by Chiocchetti et al. (2018) for pumpkin peels (20% bioavailability).
The lower bioavailability observed in the present study can be attributed to higher antinutritional factors content especially phytate and total phenolics which were greater in plant species used in the present study (Table 2 and 3) than reported in previous studies (Banjari et al., 2013;Ajala, 2009;Ndlovu & Afolayan, 2008;Adeboye & Babajide, 2007;Castro-Alba et al., 2019). This further justifies the rationale for the current study thus signifying limited applica- On the other hand, it has been shown that bioavailability of nonheme iron (in terms of absorption) such as those found in plant foods increases in the presence of heme iron (Kumar et al., 2020;Young et al., 2018). This implies that consumption of food groups that contain heme iron such as meat together with WFV could potentially increase bioavailability of iron from them. However, an important question that has largely remained unanswered is the quantity of heme iron food group that is required to substantially improve It was noted that whereas wild fruits had more iron than the vegetable species, iron bioavailability was higher in vegetables than in fruits (Table 6). This observation is not peculiar to the current study.

RDA Total Bioavailable
Previous studies have shown that bioavailability of iron decreased with increase in the concentration of iron in plant foods (Hurrel & Egli, 2010) ostensibly due to higher contents of antinutritional factors (Natesh et al., 2017;Acipa et al., 2013). From a nutritional point of view, nutrients in any food source are useful to the body if they can be utilized by the cells and tissues to support body functions (Moughan, 2018). Whereas nutrient bioavailability is a proximate indicator of the nutritional value of a food source to the human body (Baree et al., 2018), knowledge of bioavailability alone may not be adequate to construe nutrient adequacy at household level, unless meaningfully translated to reflect RDA of household members.
Indeed, in this study, the fraction of total iron in WFV available for metabolism over a one-year period was assessed. Contrary to the general belief that wild food plants are essential to households' nutrition among rural households in developing country setting such as the Acholi subregion of Uganda (Okidi et al., 2018), the contribution of bioavailable iron to household annual RDA was very marginal ( Figure 4). for which bioavailability is not constrained by antinutritional factors. Therefore, the limited nutritional value of WFV in terms of iron should not limit households from consuming them.
From a nutritional point of view, it is well known that no food type contains all nutrients in adequate quantities but nutrient complementarity can be achieved through diet composites derived from various food sources. This is indeed the fundamental basis for emphasis on dietary diversity (Alowo et al., 2018). Therefore, households should be encouraged to include other high iron-bioavailable food sources such as meat in the diet. Relatedly, whereas dietary diversity is believed to be a good indicator of diet adequacy (Alowo et al., 2018), information on optimal combination of food types to ensure diet adequacy for micronutrients of public health importance such as iron, vitamin A, and zinc for use by economically disadvantaged households in developing countries such as those in Acholi subregion of Uganda is largely lacking. This is a potential subject for future research.
In looking at these results, it should be appreciated that calculation of the contribution of bioavailable iron from WFV to household RDA did not take into account processing methods (practiced in Acholi subregion such as sun drying, boiling, and sun drying, and salting and sun drying) that can improve bioavailability of iron from plant foods (Bighaghire, 2019). Thus, the abysmal contribution observed in the current study may be an underestimation. Future studies should evaluate the effect of those processing methods on iron bioavailability from WFV. When segregated botanically, the contribution of bioavailable iron to household RDA was below 2% for both fruits and vegetables although contribution in terms of total iron intake was more for fruits than vegetables by a difference of 13.3%.
This observation is not surprising because iron from fruits was less bioavailable than from vegetables (Table 6).
Generally, majority of rural households such as those living in Acholi subregion of Uganda depend mostly on plant-based foods for which bioavailability of iron and other essential micronutrients such as zinc is largely constrained by the presence of antinutritional factors (Biesalski & Black, 2016). Okidi et al. (2018) reported that WFV contributed adequately to household iron requirement in Acholi subregion on the basis of total iron content of the wild food plants.
An important argument that formed the basis of the current study was that contribution based on total iron content of wild food plants This study encountered two limitations. First, nutrient interactions are one of the factors that interfere with iron bioavailability.
However, this study did not assess the effect of nutrient interaction on iron bioavailability. Future studies should design methods that allow examination of the contribution of antinutritional factors and other nutrients to iron bioavailability separately. Second, this study used secondary data built on the assumption that food distribution in households is in accordance with individual household member food needs which may not be the case. Therefore, nutrient adequacy determined based on pooled household estimate may not reflect the intrahousehold nutrient adequacy.

| CON CLUS IONS
This study has demonstrated that the pool of antinutritional factors in wild food plants in Acholi subregion is dominated by phytate and total phenolics, while bioavailability of iron is higher from wild vegetables than from wild fruits despite the later having higher content of iron than the former. Taking into consideration the limitations of this study, the contribution of bioavailable iron to household annual requirement for the micronutrient was very marginal. Thus, wild fruits and vegetables alone cannot be relied upon to guarantee adequate intake of iron among rural households in Acholi subregion of Uganda. Future studies should consider establishing optimal food combinations that can enhance uptake of iron from WFV, the effect of bioenhancers such as vitamin C and nutrient interaction on bioavailability of iron from wild food plants as well as the effect of traditional processing methods on levels of antinutritional factors, bioavailability, and annual intake of bioavailable iron at individual household member level.

ACK N OWLED G M ENTS
The authors are grateful to local leaders of Gulu and Amuru districts, research assistants, and rural communities in the two districts for their active participation in the study. We extend our appreciation to the Regional Universities Forum for Capacity Building in Agriculture (RUFORUM) for funding the research work (Grant number: RU/2017/NG-MCF-01). RUFORUM had no role in the design, analysis, or writing of this article.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

Ethical clearance was obtained from Gulu University Research Ethics
Committee (Approval number: GUREC-079-18). Permission from Chief Administrative Officers was also sought before data collection in the respective districts. Last but not least, informed consent of residents who helped in the identification and collection of study samples was obtained prior to sample collection.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data that support the findings of this study are available from the corresponding author upon reasonable request.