Agroforestry diversity, indigenous food consumption and nutritional outcomes in Sauria Paharia tribal women of Jharkhand, India

Abstract Like several indigenous populations, Sauria Paharias, a vulnerable indigenous tribal group residing in a biodiverse environment of Jharkhand, India, have high levels of undernutrition. We assessed agroforestry and dietary diversity, food consumption especially indigenous food (IF) intake and nutritional status of Sauria Paharia women through a cross‐sectional study conducted in 18 villages of Godda district, Jharkhand. Household level information was elicited through household surveys including a dietary survey and a food frequency questionnaire. Twenty‐four‐hour dietary recalls (24 HDR) and anthropometric assessments were taken on one randomly selected woman per household. An index, Food Accessed Diversity Index (FADI) created to measure agroforestry diversity, showed a low mean score of 0.21 ± 0.15 and range: 0, 0.85. Fifty‐nine percent of women consumed any IF during 24 HDR. Median minimum dietary diversity score for women (MDD‐W) was 3 (acceptable score ≥5). More than 96% of women had intakes below estimated average requirements for all nutrients studied (energy; vitamins A, C, thiamine, riboflavin, niacin, pyridoxine; folate; iron; calcium and zinc) except protein; 41% women were underweight. IF consumption was independently associated with calcium and vitamin A intake. Decision trees developed for micronutrient consumption at different levels of MDD‐W score and IF consumption scenarios revealed 1.3 to 2.9 times higher consumption of micronutrients among women with MDD‐W ≥ 3 or 4. Strategies like agricultural extension programmes promoting indigenous varieties and nutrition education for increasing dietary diversity with IFs have potential to address undernutrition in Sauria Paharia women.


| INTRODUCTION
Globally, more than 820 million people are undernourished, and more than 2 billion people suffer from micronutrient deficiencies (FAO, 2017). More than one fifth of women in India are undernourished with chronic energy deficiency (CED), and more than 50% of them have anaemia (National Family Health Survey [NFHS]-4, 2015). The country is ranked 102nd among 117 nations on the global hunger index, and its status is rated as "serious" on the spectrum of hunger levels (Global Hunger Index [GHI], 2019). Moreover, India is ranked 70th among 187 countries on dietary quality, based on overall dietary patterns (Imamura et al., 2015), with diets that are less diverse, deficient in dairy, fruits, vegetables and protein rich foods (Keats & Wiggins, 2014). Both insufficient quantity and quality of food intake are causing food and nutrition insecurity leading to malnutrition within the country.
Indian agriculture, boosted by the Green Revolution in the mid-60s, mainly focuses on wheat and rice production, with a goal of fulfilling average caloric requirements (Dwivedi et al., 2017;Welthungerlife, 2016). Additionally, the targeted public distribution system, a major government programme under the national food security act (NFSA, 2013), mainly includes food items like wheat, rice, sugar and millets (in specific states). This strategy has promoted poor quality, monotonous and low-diversity diets, especially among vulnerable sections of the population (Welthungerlife, 2016) (Agrawal, 2013;Ghosh-Jerath et al., 2013, 2016Kshatriya & Acharya, 2016;National Nutrition Monitoring Bureau [NNMB], 2009;National Family Health Survey [NFHS]-4, 2015). The diets of Indian tribal communities are considerably deficient in milk and milk products, fruits and vegetables, as well as pulses and flesh foods (meat, fish and poultry) (Ghosh-Jerath et al., 2016;Mittal & Srivastava, 2006;Rao, Kumar, Krishna, Bhaskar, & Laxmaiah, 2015).
This predisposes them to deficiencies of micronutrients like iron, vitamin A (Rao et al., 2015) and also zinc, which could be due to the high phytate content in staples like rice . This nutritional deprivation is further compounded by socio-ecological factors such as geographical isolation, extreme poverty, migration, local agricultural and land use policies and climate change (Bhattacharjee, Kothari, Priya, & Nandi, 2009;Jain et al., 2015;Laxmaiah et al., 2007;Rao et al., 2015).
The Sauria Paharias are one of the particularly vulnerable tribal groups (PVTGs) in Jharkhand, a state that is home to 8.6 million indigenous tribal people constituting 26.2% of the total state population (Census, 2011;Lakra & Kumar, 2017). The Sauria Paharias reside among the hilly ranges and wild dense forests, which provides a rich biodiverse food environment (UNDP, 2008). An ancient tribal group with predominantly agriculture-based livelihoods, they practise cultivation on small areas of land, home gardens (Baaris) and burnt patches of forest land (Kurwa farming). They also source foods from forests, rivers and nearby areas that are indigenous to the community, utilizing these either for household consumption or income generation (Tribal Cultural Heritage in India Foundation, 2018). However, despite a rich biodiverse agroforestry environment, the Sauria Paharias are impoverished and lag behind on all nutrition indicators (Kumar & Sinha, 2016).
Food-based strategies such as incorporating traditional and indigenous foods (IFs) obtained locally from the natural environment through farming and wild harvesting can potentially improve dietary diversity and micronutrient deficiencies (Dwivedi et al., 2017;Sethi et al., 2017). Studies have shown a positive impact of production diversity on child anthropometric outcomes (Jones, 2017); better access to diverse and improved diets through home gardening and consumption of indigenous varieties of fruits and vegetables from forests and other nearby areas could likely improve micronutrient status (Berti, Krasevec, & FitzGerald, 2004;Girard, Self, McAuliffe, & Olude, 2012;Masset, Haddad, Cornelius, & Isaza-Castro, 2012;Tontisirin, Nantel, & Bhattacharjee, 2002). Coincidently, low-income regions with high levels of malnutrition are often rich biodiversity hotspots with nutrient resources that are often underutilized (Herrero

Key messages
• Access to agroforestry diversity was poor among Sauria Paharias despite extensive collective information on availability of diverse traditional foods in their environment.
• Women had poor nutrient intake and dietary diversity.
• IF consumers had significantly higher intake of calcium and vitamin A.
• Stratification by MDD-W scores and IF consumption revealed 1.3 to 2.9 times higher consumption of different micronutrients among women with MDD-W equal to or above 3 or 4 and IF consumers.
• Access to agroforestry diversity can be improved through agricultural extension programmes to promote cultivation of indigenous varieties and nutrition education for improving dietary diversity utilizing IFs from the environment. Lachat et al., 2018;Myers, Mittermeier, Mittermeier, da Fonseca, & Kent, 2000). Therefore, leveraging this biodiversity ('Agroforestry and its Benefits|Biodiversity'; Palacios Bucheli & Bokelmann, 2017) to improve access and utilization of traditional foods can facilitate consumption of diverse, affordable diets among the poorer and indigenous communities (Jones, 2017;Lachat et al., 2018). Because the relationship between agroforestry diversity, actual dietary intake and nutritional outcomes is affected by multiple factors (Gómez et al., 2013;Pinstrup-Andersen, 2013), a detailed enquiry can help identify the pivotal determinants affecting this relationship under specific contexts. An integrated approach built on these findings can strengthen and make modern agriculture nutrition sensitive by mainstreaming the traditional food systems into it (Jones, 2017;Welthungerlife, 2016).
In this paper, we explore the association between production and access to diverse foods (agroforestry diversity), IF consumption and dietary diversity with nutrient intake in the Sauria Paharia community. We also analyse the impact of external factors such as socio-economic and demographic profile on this relationship. Finally, we explored quantitative differences in mean micronutrient intakes stratified both by dietary diversity and IF consumption.

| METHODS
This study was part of a larger project that examined IF consumption by tribal groups of Jharkhand and its contribution to dietary diversity and food security among women and children. A detailed description of the larger project and its methodological approaches are reported elsewhere (Ghosh-Jerath et al., 2019). In the present paper, we report the socio-economic, demographic, dietary profile, agroforestry diversity, IF consumption, nutrient intake and nutritional status of Sauria Paharia women. The exploratory cross-sectional survey was conducted in July-August 2018, during the rainy season, in Sauria Paharia villages.

| Study area
The study was conducted in selected Sauria Paharia villages in two blocks of Godda district of Jharkhand (Sauria Paharia population in Godda district = 13,688; Census, 2011).

| Sampling framework and study population
A two-stage cluster sampling design was followed (Figure 1).
From the purposively chosen two blocks (a block is defined as an administrative subdivision of a district), namely, Sunderpahari and Boarijor, and by using a tribal village list, in the first stage, nine villages each from the blocks were randomly selected using probability proportional to size (PPS) sampling. In the second stage, all 18 selected villages were visited, and a house-listing exercise for all Sauria Paharia households (HHs) was carried out to construct the sampling frame of all eligible HHs. The eligibility was based on the overall objective and presence of at least one non-pregnant woman in the reproductive age group (15-49 years) and one child (6 months to 54 months) in the HH. In case of more than one eligible woman in a HH, one woman was randomly selected using Kish

| Sample size calculation
Requisite sample size was calculated based on the difference in mean dietary intake of iron of 4 mg/day (35% increase) with a standard deviation (SD) of 7 mg/day between consumers and non-consumers of IFs, reported in a previous study (Ghosh-Jerath et al., 2016) among women of Santhal tribes. A minimum of 97 women in each group were required to be sampled to detect the suggested minimum difference in iron intake across the two groups with 80% power and a 5% level of significance. A design effect of 2.0 was considered to account for the loss in precision due to cluster sampling and deviation from normality. The above calculation

| Data collection
A pretested questionnaire developed on an electronic data capture (EDC) platform using Samsung (Model SM-T385) and the software CS pro (Version 7.1) was administered at the HH level to elicit information on socio-economic and demographic profile.
Information on availability of and access to different food sources was collected using HH survey tool and agriculture diversity tool.
The software had extensive in-built checks (context, range and logic checks). Frequency of intake of different foods at HH level with a reference period of 1 month was assessed using a food frequency questionnaire (FFQ). A 24-h dietary recall (24 HDR) and anthropometric assessment on one selected woman was also done.
Paper forms were used for FFQ and 24 HDR and administered by nutritionists and nutrition interns after due training. The field investigators administered the HH questionnaire on EDC platform and did anthropometric assessments after due trainings and retraining on need basis. Standard procedures were followed for anthropometric assessments, and routine back-end data checks were conducted to maintain data quality.

| Study variables
The outcome variables for this study included intake of macronutrient, that is, energy (Kcal), protein (g), fats (g) and micronutrients including iron (mg); calcium (mg); zinc (mg); vitamin A (μg); vitamins B1, B2, B3 and B6 in (mg); folic acid (μg) and vitamin C (mg) and anthropometric status of women. The exposure variables included IF consumption at the individual level, minimum diet diversity for women of reproductive age (MDD-W; a validated measure of dietary diversity; Minimum Dietary Diversity for Women-A Guide to Measurement, 2016) and an index developed for agroforestrydiversity at HH level. Other independent variables included socio-demographic profile of the HHs, HH wealth index and access to different sources of food like agricultural land, forests and water sources. A brief description of each these variables is given as follows: Measurement of nutrient intake: A nonconsecutive, 2-day multiple pass 24 HDR method was used for nutrient intake estimation of one selected woman per HH. The selected women were asked to recall all food items consumed over the past 24 h and show (wherever possible) or describe the foods eaten in each meal (i.e., each food item consumed along with a detailed recall of ingredients used in preparation and method of preparation). Food recall kit (standard utensils and food picture flip book) was used for portion size estimation. At the end, a final probing was done to elicit information of foods consumed but inadvertently forgotten. Self-reported amounts of raw foods were used to determine the daily intake of individual food items belonging to the different food groups such as cereals, pulses, flesh foods, green leafy vegetables (GLVs), other vegetables, roots and tubers, fruits, sugars and fats. We developed and pretested a 300-item FFQ with the community.
A list of IFs reported by the community along with their taxonomic classification (wherever available) is provided as Table S1.

| IF consumption
This variable was calculated from the list of food items reported in the 24 HDR. If a woman consumed any IF (as listed in table S1) over the recall period of 2 days, she was considered as a consumer of IF.

| Nutritional status
Anthropometric measurements (heights and weights) were taken to assess the nutritional status of the women from whom DRs were taken. These measurements were carried out using standard protocols and equipments that included 'Seca' digital flat scale Model 813 for weight measurement and 'Seca' stadiometer Model 213 for height measurement.

| Socio-demographic characteristics
A pretested structured questionnaire was developed to elicit information on HH members' characteristics, which included family type, age profile of family members and educational and occupational status of family members (selected women participant and head of the household [HOH]). HH information like type of roofing material, number of rooms, presence of a separate kitchen, source of electricity and drinking water, possession of assets (cot, chair, kerosene stove and mobile phone) and monthly expense on food were elicited and used to generate the HH wealth index variable.

| Access to food sources
To elicit information on food access during the past 1 year, self-reported information on HHs ownership of agricultural land and kitchen garden (baari), access to forest land for "Kurwa farming" and for food gathering and access to water bodies were collected.

| Measurement of agroforestry diversity
Information of total number of foods accessed from different sources (as mentioned above) was obtained through HH questionnaire and an agricultural diversity tool. The latter was administered on conveniently selected 60 HHs.
In previous studies, agricultural diversity has been simply calculated by the total number of food groups grown and/or types of animals raised for food (Hirvonen & Hoddinott, 2016;K. T. Sibhatu & Qaim, 2016). In our study, we modified the agroforestry diversity component by constructing an index called Food Accessed Diversity Index (FADI), which has been adapted from crop diversity index (CDI) (Michler & Josephson, 2017). We divided the total number of foods grown, gathered or accessed as well as animals raised in a particular HH by the corresponding maximum possible number of foods grown, gathered, accessed and raised in a particular village (N).
Instead of standard number of food groups, we preferred to do counts of individual food items. This is because nutritive value of several food items under a specific food group may vary a lot. The Thus, FADI is a relative expression. Foods accessed from the market were not included in this index. Lower values of FADI indicate lower diversity in production and access to foods and vice versa. This index helped in capturing the inequality in production and access between HHs of the same village, while adjusting for agro-climatic conditions, specific to a village (Michler & Josephson, 2017).

| Data analysis
Quantitative variables were reported as Mean ± SD, and qualitative variables were summarized using counts and percentages. Outliers, if any, were retained unless there was a data entry mistake. The information on number of food sources accessed were converted to scores. HH wealth index score was obtained using the principal component analysis (PCA), an adaptive statistical technique for reducing the dimensionality of large data sets. The HHs were then classified into five groups based on the quintiles of the HH wealth index score. MDD-W was computed by taking median of total food groups consumed over 2 days. MDD-W was later classified into two levels less than median and more than or equal to median. BMI was calculated as weight (in kg) divided by height (in m 2 ) based on standard practices. The women were then classified as being underweight or, otherwise, using the BMI cut-offs. The women were further classified under different levels of CED using the standard BMI cut-offs (WHO Expert Consultation, 2004).
The food intake data from 2 days 24 HDR were converted to nutrient intakes using a validated software 'DietCal' (Version 8.0;

Profound Tech Solution), which is based on values from Indian
Food Composition Table, (Gibson, 2008). Details are provided in Table 2. The usual intake between IF consumers and nonconsumers and MDD-W below or above median were compared using independent samples t-test.
To account for the nested structure of the data, LMER technique with villages (n = 18) as the random effect was adopted, to investigate linear relationship of usual intake of each nutrient with IF consumption and MDD-W adjusting for other factors (presented in Table 4). The usual nutrient intakes obtained were also observed to be skewed and hence were again transformed using box-cox procedure as illustrated above.

| Ethical considerations
The study protocol was approved by the Institutional Ethics  3.4 | Nutrient intake of women in the reproductive age group (15-49 years) To explore the contribution of IFs to nutrient intake, we compared the 24 HDR data in women who consumed IFs in the past 2 days with the women who did not ( In this, major food commodities including staple food grains, such as wheat and/or rice, sugar and kerosene oil (a fuel used for cooking) are distributed through a network of public distribution shops (also known as ration shops) at subsidized prices. Possession of a PDS ration card (an official document entitling the holder to a ration of food) under various categories of poverty, that is, APL , BPL and AAY, a category based on degrees of poverty, entitles the holder to access the food product at highly subsidized rates.

| Factors associated with nutrient intake of women with reference to dietary diversity and IF intake
The LMER technique was adopted with villages (n = 18) as the random effect. FADI was associated with educational status of women who had education till secondary level and above (P = 0.006). Similarly, MDD-W was also associated with education status of HOH who had education till secondary level and above (P = 0.005  In the present study, education status, FADI, MDD-W and HH wealth index were considered to confound the relationship between IF consumption status and each nutrient intake. Taking into account the confounding effect of these factors, the adjusted findings obtained using LMER are presented in Table 4. Estimates in Table 4 are presented in terms of percentage change in response variable per unit change in the independent variable.

T A B L E 3 Comparisons of mean usual intakes between the IF consumers and categories of MDD-W (N = 204)
IF consumption was observed to be significantly associated with higher vitamin A and calcium intake after controlling for MDD-W and other factors in the model (Table 4). An IF consumer was expected to consume 42.6% more of Vitamin A and 10.1% more of calcium than non-IF consumer. MDD-W was observed to be significantly associated with all the nutrients after adjusting for IF consumption and other factors in the model.
Decision trees were developed to perform an objective assessment of various micronutrient consumption at different scenarios of MDD-W score and IF consumption status among the study population ( Figure 2).
These assessments revealed 1.3 to 2.9 times higher consumption of different nutrients among the small proportion of women with MDD-W above or equal to 3 or 4 compared to those with MDD-W less than 3. For example, though the mean vitamin A intake of the study population was 30 μg/day (considerably lower than EAR), those who had a MDD-W of more than or equal to 4 had mean intake of 88 μg/day (only 9% of women), whereas those with MDD-W less than 4 had mean intake of 24 μg. IF consumption among those with MDD-W lower than 4 improved the vitamin A intake to 35 μg/day (52% of women). Similarly, the mean intake of vitamin C was also low in the study population (14 mg/day). However, the intake improved to 21 mg/day among those who had MDD-W score of more than or equal to 4 (only 9% of the population). The mean calcium intake of the study women was 95 mg/day. However, women with MDD-W of more than or equal to 4 had a calcium intake of 123 mg/day. Among women with low MDD-W (equal to 3), the calcium intake among IF consumers (101 mg/day) was observed to be higher as compared to non-IF consumers (91 mg/day). IF consumption also improved calcium intake of women with MDD-W score of less than 3 with an intake of 90 mg compared to non-consumers with an intake of 79 mg.

| CED in women
Based on anthropometric assessment, about 41% women had various degrees of CED, with about 13.4% women falling in the category of CED III (<16 kg/m 2 ). CED was, however, not associated with calorie intake (P = 0.422; Table 5).  (Kumar & Sinha, 2016;Satyam, 2017). We evaluated the agroforestry diversity in the context of access to food sources by developing an index called the Food Access Diversity Index (FADI), which is adapted from CDI (Michler & Josephson, 2017). We have added a new dimension to the CDI for a more robust view of foods produced and food sources accessed contextual to our study population. This crucial new dimension included sources of food collected and gathered from forests, water bodies apart from cultivated foods. A low FADI score was observed in the study reflecting poor access diversity. This was despite the   (Powell et al., 2015;Powell, Bezner Kerr, Young, & Johns, 2017). The lack of diversity and overall poor socio-economic profile of the study population was echoed in the poor nutritional status (41% underweight and poor nutrient intake) of the women, a finding similar to previous studies in tribal communities of India from Jharkhand and other states (Das & Bose, 2015;Ghosh-Jerath et al., 2016.

| DISCUSSION
Other studies on women from resource poor settings in India and other Asian countries have also reported inadequate energy and micronutrient intake to varying extent. However, our study reports extraordinarily lower levels of nutrient intakes reflecting severe nutritional deprivation. We must also consider that the methodology used to estimate nutrient intake in these studies is not uniform and some have not reported usual intakes as we have (Martin-Prevel et al., 2017;Pathak, Kapil, Kapoor, Dwivedi, & Singh, 2003;Thankachan, Muthayya, Walczyk, Kurpad, & Hurrell, 2007;Torheim, Ferguson, Penrose, & Arimond, 2010).
Although many indigenous varieties of cereals, GLVs, roots and tubers, other vegetables and fruits were known and potentially available, the access to these foods and frequency of their consumption was low. A typical day's diet for a large majority included rice and some  Higher MDD-W scores in our study were associated with better micronutrient intake. Other studies in low-income settings from developing countries have also shown poor energy intake and less diverse diets contributing to poor micronutrient intake among women (Arsenault et al., 2013;Henjum et al., 2015). This is crucial from the to impact the nutrient intake. (iv) We did not assess B 12 intake in study population, as majority of the women (except 12) did not report consuming any dietary sources of vitamin B 12 in their diets; further, there is no information on B 12 content of foods in the IFCT (Longvah et al., 2017). (v) Since the FFQ list was long, we expect reporting and recall bias, and there could also be recall bias while we used 24-HDR method. (vi) Despite using food recall kit and a portion size estimation flip book, we do expect some level of portion size estimation error in the recalls. (vii) The FADI explored the foods accessed from the natural sources and did not consider the market foods as the purpose of this index was to explore the access to natural IF sources within these communities. (viii) Owing to the cross-sectional nature of the design, our study presents only associations, and no causal inference can be drawn from the findings. It is also possible that there could be unmeasured sources of confounding in this study.

| CONCLUSION
The Sauria Paharia community of Jharkhand had collective information on several IF sources but was not optimally accessing their diverse agroforestry system. This was reflected in their poor dietary intake and nutritional status. However, a small proportion of this population with higher MDD-W scores and higher consumption of IFs showed better nutritional intakes. Strategies to promote better access to IFs like promoting production of IFs, information education and communication on nutritive value of IFs and ways of incorporating these in daily diets, school feeding programmes, supplementary feeding programmes and public food distribution system in this community have the potential to complement ongoing nutrition interventions and improve nutritional status of women. data collection platform for the study, which has enhanced the quality and efficiency of field data collection manifold. Authors would also like to thank all the field staff, interns and nutrition consultants and ASHAs and Anganwadi workers in the study villages for their support during all phases of the study and their warmth and cooperation that has made this study an enriching and mutually satisfying experience. Lastly, we would owe a debt of gratitude to the Sauria Paharia families especially for the time given during the sowing season, sparing those invaluable hours for us amidst a critical phase of their agricultural cycle. We hope that our findings will be able to compensate for their contribution in some small measure.

CONFLICTS OF INTEREST
The authors declare that they have no conflicts of interest.