Journal of Hunger & Environmental Nutrition ISSN: 1932-0248 (Print) 1932-0256 (Online) Journal homepage: www.tandfonline.com/journals/when20 Food security variation among Indigenous communities in South-western Uganda Kaitlin Patterson, Lea Berrang-Ford, Shuaib Lwasa, Didacus B. Namanya, James Ford, IHACC Research Team & Sherilee L. Harper To cite this article: Kaitlin Patterson, Lea Berrang-Ford, Shuaib Lwasa, Didacus B. Namanya, James Ford, IHACC Research Team & Sherilee L. Harper (2023) Food security variation among Indigenous communities in South-western Uganda, Journal of Hunger & Environmental Nutrition, 18:4, 620-648, DOI: 10.1080/19320248.2020.1852146 To link to this article: https://doi.org/10.1080/19320248.2020.1852146 Published online: 24 Feb 2021. Submit your article to this journal Article views: 239 View related articles View Crossmark data Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journalInformation?journalCode=when20 https://www.tandfonline.com/journals/when20?src=pdf https://www.tandfonline.com/action/showCitFormats?doi=10.1080/19320248.2020.1852146 https://doi.org/10.1080/19320248.2020.1852146 https://www.tandfonline.com/action/authorSubmission?journalCode=when20&show=instructions&src=pdf https://www.tandfonline.com/action/authorSubmission?journalCode=when20&show=instructions&src=pdf https://www.tandfonline.com/doi/mlt/10.1080/19320248.2020.1852146?src=pdf https://www.tandfonline.com/doi/mlt/10.1080/19320248.2020.1852146?src=pdf http://crossmark.crossref.org/dialog/?doi=10.1080/19320248.2020.1852146&domain=pdf&date_stamp=24%20Feb%202021 http://crossmark.crossref.org/dialog/?doi=10.1080/19320248.2020.1852146&domain=pdf&date_stamp=24%20Feb%202021 https://www.tandfonline.com/action/journalInformation?journalCode=when20 ARTICLE Food security variation among Indigenous communities in South-western Uganda Kaitlin Patterson a,b, Lea Berrang-Ford c,d, Shuaib Lwasa d,e, Didacus B. Namanyad,f, James Ford c,d, IHACC Research Teamd, and Sherilee L. Harper d,g aDept. Of Population Medicine, University of Guelph, Guelph, Ontario, Canada; bSchool of Public Health, University of Alberta, Edmonton, Alberta; cPriestley International Centre for Climate, University of Leeds, Leeds, UK; dIndigenous Health Adaptation to Climate Change Research Group, School of Public Health, University of Alberta, Edmonton, Alberta, Canada; eDept. Of Geography Geoinformatics and Climatic Sciences, Makerere University, Kampala, Uganda; fUgandan Ministry of Health, Kampala, Uganda; gSchool of Public Health, University of Alberta, Edmonton, Alberta, Canada ABSTRACT We assess whether the household is an apt scale of analysis to examine food insecurity of Indigenous Batwa of Kanungu District, Uganda. Our objectives were: 1) estimate the preva lence and range of food insecurity, 2) estimate the variation in food insecurity associated with household and community fac tors. The survey was administered 6 times at 3-month intervals (Jan 2013-April 2014). Multilevel modeling was used to deter mine household and community associations with food inse curity. The Batwa were highly food insecure (97%). Variation in food insecurity that is explained by household and community factors was low. Food insecurity analyses should be considered scale-dependent. KEYWORDS Food Security; Indigenous Peoples; social determinants of health; multi-level modeling; Uganda Background Rising food prices, constraints on agricultural land, the occurrence of severe droughts, land degradation, and population growth have contributed to increasing attention to global hunger and food security. Indeed, the eradica tion of extreme poverty and hunger are the first two Sustainable Development Goals (SDGs).1–5 Concerns regarding climate change impacts on food produc tion and security (SDG 13) have further highlighted the importance of under standing and untangling the complex combination of regional, community, and household-scale stressors that interact to cause food insecurity and under- nutrition burden.6–9 In sub-Saharan Africa, high dependence on agriculture makes the region especially vulnerable to biophysical, climatic, economic, and geopolitical perturbations. For marginalized populations already struggling CONTACT Kaitlin Patterson kpatte08@uoguelph.ca Department of Population Medicine, Ontario Veterinary College, University of Guelph, Guelph, Ontario, Canada, N1G2W1 *Indigenous Health Adaptation to Climate Change (IHACC) Research Team: Cesar Carcamo, Alejandro Llanos, and Victoria Edge. JOURNAL OF HUNGER & ENVIRONMENTAL NUTRITION 2023, VOL. 18, NO. 4, 620–648 https://doi.org/10.1080/19320248.2020.1852146 © 2020 Taylor & Francis Group, LLC http://orcid.org/0000-0002-3444-0472 http://orcid.org/0000-0001-9216-8035 http://orcid.org/0000-0003-4312-2836 http://orcid.org/0000-0002-2066-3456 http://orcid.org/0000-0001-7298-8765 http://www.tandfonline.com https://crossmark.crossref.org/dialog/?doi=10.1080/19320248.2020.1852146&domain=pdf&date_stamp=2023-06-26 with poverty and hunger, stressors on food security may exceed their ability to adapt.10,11 Indigenous populations are believed to be particularly vulnerable to food insecurity due to patterns of neglect by governments, colonialism, lack of access to resources, ethnic discrimination, low employment, extreme poverty, and dependence on subsistence livelihoods.8,10–12,13 Approximately 250 million Indigenous Peoples reside in Africa. The African Commission on Human and Peoples’ Rights (ACHPR) cautions not to use Aboriginality as a synonym for Indigeneity, rather that focus should be placed on 1) self-definition as Indigenous; 2) special attachment to and use of their traditional lands, 3) experi ences of subjugation, marginalization, dispossession, exclusion or discrimination. 14 The Batwa a rural Indigenous peoples (both recognized and self-identified as Indigenous), living in Southwestern Uganda, are among the poorest populations within Africa. They experience a high burden of illness relative to other Ugandans ― including their non-Indigenous neighbors ― and face social barriers accessing health care and education.10,15–18 Economically marginalized populations, subsistence-based food systems, and Indigenous peoples more generally are recognized as highly vulnerable to perturbations in the stressors of food security.19,20 Despite this, there is limited place-based research on Indigenous health and food systems in general, and Indigenous peoples’ food security in sub-Saharan Africa in particular.10,15,21–23 Food insecurity can be conceptualized at a range of scales, from the indi vidual to household, or as aggregate measures for a community or nation. The household-level survey is the current gold standard used most frequently to identify populations at risk,24–26 though individual-level food security esti mates are also common.27,28 Household-level surveys characterize the varia tion in food security across populations to identify the most vulnerable households and frequently also to estimate associations with food security or insecurity. This approach, however, is predicated on the assumption that households within a sampled population are heterogeneous, and that the household level reflects an appropriate scale at which to capture variation in food security potential-associated factors. As highlighted by Pearce29 and Rose,30 the most important determinants of health may vary little within a particular population, and sampling should necessarily seek to capture a scale at which health outcomes and exposures vary. While the household is undoubtedly a highly relevant unit of analysis at which many food production, access, and consumption decisions are made, a focus on household-level variation may neglect empirical interrogation of larger-scale population- based determinants ― at which important policy-relevant variation may occur ― and intervention entry points. The use of household-level surveys for food security has herein remained relatively unchallenged as a unit of analysis, and there has been limited application of multilevel modeling within food security research.31–34 What scale then is appropriate for measurement of JOURNAL OF HUNGER & ENVIRONMENTAL NUTRITION 621 food security, in what contexts, and how can this be determined for a particular population? To what extent does the selection of sampling unit affect the range of potential intervention options inferred from epidemiologic analysis? We herein seek to identify and characterize variation in food security across multiple scales for a Ugandan Batwa population. Objectives include: 1) esti mate the prevalence and range of food insecurity for Indigenous Batwa in south-western Uganda, and 2) estimate the variation in food insecurity that is associated with household and community factors. Methods Study Population The Batwa are an Indigenous People dispersed throughout Central Africa, and within Uganda live within three districts with a combined population of approximately 6,700.35 Peer-reviewed and gray literature consistently cite poor health status and heightened risk among Batwa compared to their non- Indigenous counterparts.22,36–38 Previous research has reported that poor nutri tion and low food security status are high among the Batwa.15,21,22,28 Previously nomadic forest hunter-gatherers, the Batwa of Uganda were evicted from their ancestral homes, with no or negligible compensation, due to the establishment of Bwindi Impenetrable National Park (BINP) in 1991.23 This displacement forced the Batwa to adopt an agricultural and sedentary lifestyle.15 Main food sources for the Batwa since leaving the forest include subsistence agriculture (crop cultivation & small livestock rearing) and food in exchange for manual labor.10,21 In Kanungu District, the Batwa live in settle ments, and share agricultural land purchased and owned by non-governmental organizations (NGOs) and private donors. Socioeconomically, the Batwa are highly impoverished compared to the Ugandan average and their neighbors. Uganda’s national per capital income is significantly higher (550$) compared to the Batwa (97$),10,39 and adult literacy (<10%) is significantly lower than the mean for Uganda’s south western province (neighboring non-Indigenous population) (>73%).10,40 Maternal health, child mortality, infant mortality, and life expectancy are all poorer among the Batwa than among neighboring peoples or the Ugandan average, and the Batwa have been highlighted as one of the world’s most impoverished and marginalized populations.15,41 Despite a wealth of gray literature high lighting concerns related to Batwa food security, there is negligible peer- reviewed research on Batwa health and food systems. Indeed, most articles in the peer-reviewed literature citing Batwa focus on the impact of the Batwa on the health of primates in Bwindi Impenetrable National Park, with limited focus on the Batwa themselves. 622 K. PATTERSON ET AL. This research took place in Kanungu District, South-western Uganda. As of 2013, there were approximately 750 Batwa living in Kanungu District, scattered throughout 10 communities in 130 households (Fig. 1). These communities are partners with the Indigenous Health and Adaptation to Climate Change Research Team. In 2011, the Batwa com munities identified food insecurity as a key climate-sensitive health prior ity; a series of six longitudinal open-cohort census surveys was identified as the best way to assess food insecurity within these communities.10 Surveys were administered in 3-month intervals; January 2013, April 2013, July 2013, November 2013, January 2014 and April 2014. Due to the relatively small Batwa population, each round of survey implementation aimed to capture a full census of Batwa households. 767 household and food security surveys were collected with an average of 127 households participating in each survey administration. Analytical Framework Defining food security is difficult and as of yet there is no universally accepted definition. Food systems broadly refer to the production, processing, Figure 1. Map of Batwa communities in Kanungu District, Uganda. JOURNAL OF HUNGER & ENVIRONMENTAL NUTRITION 623 distribution, preparation, and consumption of food.20 Food systems can become stressed by a number of factors – including climatic (drought, flood ing), economic (price or demand increases, food shortages), social (food shar ing disruption), and conflict (supply routes destroyed, decreased safety) – which can lead to increased difficulty in securing food at various scales (e.g. regional, community, household, and individual).42 These factors lead to increased difficulty in securing food for the household or self, leading to food insecurity. The FAO defines food security as “all people, at all times, [having] physical and economic access to sufficient, safe and nutritious food to meet their dietary needs and food preferences for an active and healthy life.”43 In other words, when a group is food insecure, their access to, quality of, or quantity of food resources may be lacking.20 Food insecurity in itself is a negative outcome, but it can also indirectly lead to other negative health outcomes: under-nutrition or over nutrition, stunting, wasting, mental illness, adverse pregnancy outcomes and chronic illness (e.g. type II diabetes).44–48 Numerous methods and indices exist to measure food security at a range of scales: global, continental, regional, community, household, and individual levels. Data collected for each scale range from global food supply to self- reported experiences, perceptions, and behaviors of food insecurity.50 The U.S. Department of Agriculture’s (USDA) Household Food Security Survey Module (HFSSM) is one of the most frequently used instruments for assessing and measuring food security and has been validated and used internationally in less developed countries, emergency situations, and among Indigenous populations.50–52 In selecting the food security measures for this study, we considered a variety of options from the economic literature, health literature, and international development literature, among other options. Following our community-led approach to this research53–55 our decisions on food security measures were made in collaboration with Ugandan and community partners. The HFSSM was selected for this research because of its adaptive nature, its focus on the experiences and behaviors that characterize ranges of food insecurity and hunger severity, and a recognition in the literature that it is appropriate for vulnerable or high-risk populations.25,49 Data from the HFSSM were not available for regional or national estimates; however, Kanungu District (and all of Southwestern Uganda) was considered to experi ence none or minimal food insecurity (at least 80% of households are able to meet their food and non-food needs without relying on unsustainable mea sures) by the Integrated Food Insecurity Phase Classification project.56 Data Collection Data were collected using three survey instruments: individual, household head, and food security (FS). The first was administered to all members of the Batwa community (n = 750). The second instrument was administered 624 K. PATTERSON ET AL. to self-identified household heads (n = 130), or their spouse, and the third instrument (food security) was administered to the head of household food preparation (n = 130). Paper questionnaires were used to record responses with each question administered conducted orally in Rukiga (the local language). The individual and household head survey aimed to identify an individual’s characteristics (health, education, employment), and a household’s characteristics (size, composition, assets wealth and agricul tural participation). The food security survey was based upon the HFSSM and aimed to establish the extent and occurrence of food insecurity within the Batwa community from October 2012 to April 2014. The version of the HFSSM used examined the conditions, experiences and behaviors that characterize ranges of food insecurity and hunger severity experienced over the previous 3 months.57 We used a one week recall period to reflect the high level of food insecurity reported by communities, in response to the difficulty with long recall periods during pilot work in 2011, to respect Batwa worldview and conceptualizations of time, and to follow the recom mendations and instructions from our partners. Questions ranged from concerns about food insecurity, experiences of being food insecure and the frequency at which they occurred (Table 1). Additional questions on number of meals and food diversity were collected. Because Batwa com munities share access to agricultural land owned in trust by NGOs, key informants and partners were consulted to identify and collect data for community-level associations with food security, which included crop- raiding, land quality, landscape type, and market access. Food Insecurity Score and the Adapted Vulnerable Populations Score (AVPS) To reflect the high proportion of Batwa with high food insecurity, we adapted the HFSSM (hereafter referred to as the Adapted Vulnerable Populations Score or AVPS) to capture variation in food security along a continuous gradient, reflecting variation at more severe levels of food insecurity (See Table 1). The HFSSM module categorizes score from (0–18) into the following four categories; “high food security,” “marginal food security,” “low food security,” or “very low food security.”57 Based on pilot work we found that 85% of Batwa households were categorized into the “very low food security” category. We used a 26-point score to reflect a wider continuum of food security. As the score increases so does the severity of food insecurity. Our HFSSM-based AVPS incorporates an additional 6 points for households with out children and an additional 8 points for households with children for total maximum scores of 16 and 26, respectively (See Table 2 for comparison between AVPS and HFSSM score/categorization). The additional points allowed us to capture wider variation among the most severe households. Missing data were imputed following standard HFSSM guidelines.57 JOURNAL OF HUNGER & ENVIRONMENTAL NUTRITION 625 Ta bl e 1. A da pt ed v ul ne ra bl e po pu la tio ns s co re (A VP S) 2 6- po in t sc al e de si gn . H FS SM # Q ue st io n N eg at iv e re sp on se s (C od e = 0 ) Affi rm at iv e re sp on se s (C od e = 1 ) Affi rm at iv e se ve re r es po ns e (C od e = 2 ) AV PS w ith ki ds AV PS w ith ou t ki ds 1 In t he p as t w ee k, d id y ou ev er w or ry w he th er t he fo od fo r yo u an d yo ur fa m ily w ou ld ru n ou t be fo re y ou c ou ld g et m or e? ne ve r so m et im es , o ft en 1 1 2 In t he p as t w ee k, d id it ha pp en t ha t th e fo od y ou Bo ug ht /o bt ai ne d di dn ’t la st en ou gh t im e an d yo u co ul dn ’t bu y/ ob ta in m or e? ne ve r so m et im es , o ft en 2 2 3 In t he p as t w ee k, d id y ou ha ve e no ug h va rie d, h ea lth y or ba la nc ed m ea ls t o ea t? ne ve r so m et im es , o ft en 3 3 4 In t he p as t w ee k, w er e th er e tim es w he n yo u co ul d on ly fe ed yo ur c hi ld re n le ss e xp en si ve / lo w er q ua lit y fo od s be ca us e fo od ra n ou t at h om e an d it w as di ffi cu lt to b uy /o bt ai n hi gh er qu al ity fo od ? ne ve r so m et im es , o ft en 4 5 In t he p as t w ee k, w er e th er e tim es w he n yo u di d no t ha ve en ou gh t o fe ed y ou r ch ild re n w ith v ar ie d, b al an ce d an d he al th y m ea ls ? ne ve r so m et im es , o ft en 5 6 In t he p as t w ee k, w he re t he re tim es w he n yo ur c hi ld re n w er e no t ea tin g en ou gh b ec au se y ou ju st c ou ld n’ t bu y/ ob ta in e no ug h fo od ? ne ve r so m et im es , o ft en 6 Ad de d H ow m an y m ea ls d oe s yo ur h ou se ho ld t yp ic al ly e at in a da y? 3 or 4 2 7 4 1 8 5 (C on tin ue d) 626 K. PATTERSON ET AL. Ta bl e 1. (C on tin ue d) . H FS SM # Q ue st io n N eg at iv e re sp on se s (C od e = 0 ) Affi rm at iv e re sp on se s (C od e = 1 ) Affi rm at iv e se ve re r es po ns e (C od e = 2 ) AV PS w ith ki ds AV PS w ith ou t ki ds 7 In t he p as t w ee k, d id y ou o r ot he r ad ul ts in y ou r ho us eh ol d ev er c ut t he s iz e of y ou r m ea ls o r sk ip m ea ls b ec au se t he re w as n’ t en ou gh fo od a t ho m e? no ye s 9 6 8 H ow m an y da ys in t he p as t w ee k di d ad ul ts s ki p or c ut m ea ls ? 1 1+ 10 7 Ad de d Ad ul ts s ki pp ed /c ut m ea ls w ei gh te d 0– 2 3– 4 11 8 5+ 12 9 9 In t he p as t w ee k, d id y ou o r ot he r ad ul ts in y ou r ho us eh ol d ev er e at le ss t ha n yo u fe lt yo u sh ou ld b ec au se t he re w as n’ t en ou gh fo od a t ho m e? no ye s 13 10 10 In t he p as t w ee k, w er e yo u or ot he r ad ul ts in y ou r ho us eh ol d ev er h un gr y bu t di dn ’t ea t be ca us e th er e w as n’ t en ou gh fo od a t ho m e? no ye s 14 11 11 In t he p as t w ee k, d id y ou o r ot he r ad ul ts in y ou r ho us eh ol d lo se w ei gh t be ca us e yo u co ul dn ’t bu y/ ob ta in e no ug h fo od ? no ye s 15 12 12 In t he p as t w ee k, d id y ou o r ot he r ad ul ts in y ou r ho us eh ol d ev er n ot e at fo r a w ho le d ay be ca us e th er e w as n’ t en ou gh fo od in t he h om e? no ye s 16 13 13 H ow m an y da ys in t he p as t w ee k di d ad ul ts n ot e at fo r w ho le d ay ? 1 1+ 17 14 Ad de d Ad ul ts n ot e at fo r w ho le d ay w ei gh te d 0– 2 3– 4 18 15 5+ 19 16 (C on tin ue d) JOURNAL OF HUNGER & ENVIRONMENTAL NUTRITION 627 Ta bl e 1. (C on tin ue d) . H FS SM # Q ue st io n N eg at iv e re sp on se s (C od e = 0 ) Affi rm at iv e re sp on se s (C od e = 1 ) Affi rm at iv e se ve re r es po ns e (C od e = 2 ) AV PS w ith ki ds AV PS w ith ou t ki ds 14 In t he p as t w ee k, d id y ou ha ve t o re du ce y ou r ch ild re n’ s fo od p or tio ns b ec au se t he re w as n’ t en ou gh fo od a t ho m e? no ye s 20 15 In t he p as t w ee k, d id a ny o f yo ur c hi ld re n ha ve t o sk ip m ea ls be ca us e th er e w as n ot e no ug h fo od a t ho m e? no ye s 21 16 H ow m an y da ys in t he p as t w ee k di d ch ild re n sk ip m ea ls ? 1 1+ 22 Ad de d Ch ild re n sk ip m ea ls n um be r w ei gh te d 0– 2 3– 4 23 4+ 24 17 In t he p as t w ee k, d id a ny o f yo ur c hi ld re n ev er g o hu ng ry be ca us e th er e w as n o fo od a t ho m e? no ye s 25 18 In t he p as t w ee k, d id a ny o f yo ur c hi ld re n no t ea t fo r a w ho le da y be ca us e th er e w as n o fo od a t ho m e? no ye s 26 628 K. PATTERSON ET AL. Independent Variables We selected variables that have been identified as important for food security at the individual, household and community level based on the literature. Table 3 presents the variables that were selected for univariate testing as associations with food insecurity, the type of variable (i.e. continuous, binary, or categorical) expected direction of influence, and the justification for inclu sion in preliminary testing. Wealth variable selection was guided by case studies in Africa among poorer populations and local observations58–60 Principal component analysis was employed to create an asset-based wealth variable; the results were then transformed into quartiles to include in the model “least poor,” “poor,” “very poor” and “poorest” (Table 4). A large proportion of Batwa houses were constructed by the Batwa Development Programme (BDP), a local NGO, thus type, size, and materials of the house hold were not a reflection of wealth. We controlled for seasonal variation;21 using a mixed-methods approach, found that seasonality was a significant component of food security for this population. The Batwa had higher food insecurity in the dry season compared to the rainy season. Analysis Food security scores for each household were calculated as described above; the distribution of the food security variable approximated normal. To exam ine internal consistency of the AVPS we assessed Cronbach’s alpha for the 26 items.61 Univariate multilevel linear regression was used to identify candidate independent variables collected at the individual and household levels using a threshold of p<0.20.62–64 Collinearity among candidate variables was assessed through the Spearman rank correlation analysis with a cutoff point of 0.70.65 When variables were collinear, we proceeded with the variable that had the highest R2, the smallest confidence intervals, and/or the variable with higher causal plausibility based on similar studies in Africa. The multilevel model structure accounts for repeated measure at the household and commu nity levels (n = 170 households, n = 10 communities). Three best-fit models were estimated: 1) household-level variables only (accounting for repeated Table 2. Adapted vulnerable populations score (AVPS) 26-point scale compared to Household Food Security Survey Module (HFSSM) standard categorizations. USDA Categories HFSSM AVPS High food security 0 0–2.32 Marginal food security 1 3–6 Low food security 2 6.96–10.44 Very low food security 3 11–26 JOURNAL OF HUNGER & ENVIRONMENTAL NUTRITION 629 Ta bl e 3. V ar ia bl es in cl ud ed in d at a an al ys is a s po te nt ia l a ss oc ia tio ns w ith fo od s ec ur ity a m on g th e Ba tw a of K an un gu D is tr ic t, U ga nd a fr om N ov em be r 2 01 2- Ap ril 20 14 . D es cr ip tio n Ty pe Ex pe ct ed D ire ct io n Ju st ifi ca tio n D ep en de nt V ar ia bl e Fo od s ec ur ity Ca te go ric al (1 –4 ) ( H FS SM ) Co nt in uo us (1 –2 6) (A VP S) . A s th e sc or e in cr ea se so d oe s th e se ve rit y of fo od in se cu rit y. N /A W e pr es en t b ot h th e U SD A H ou se ho ld h ol d fo od s ec ur ity s ur ve y m od ul e (H FS SM ) a nd th e Ad ap te d Vu ln er ab le P op ul at io ns S co re (A VP S) Fo od s ec ur ity e xi st s on a c on tin uu m r at he r th an a s pe ci fic s ca le . A la rg er r an ge en ha nc es t he a bi lit y to d iff er en tia te h ou se ho ld s w ith m or e se ve re fo od in se cu rit y. In de pe nd en t va ria bl es H ou se ho ld v ar ia bl es Se x of h ea d of h ou se ho ld Bi na ry o ut co m e m al e = 1 , f em al e = 0 In cr ea se sc or e H ou se ho ld s w ith n o m al e he ad h av e be en fo un d to h av e lo w er fo od s ec ur ity 30 . H ig he st le ve l o f a du lt fe m al e ed uc at io n Ca te go ric al v ar ia bl e (N o fo rm al s ch oo lin g = 3 , Pr im ar y sc ho ol in co m pl et e = 2 , P rim ar y co m pl et e or a bo ve = 1 D ec re as e sc or e H ig he r l ev el s of e du ca tio n ha ve b ee n as so ci at ed w ith fo od s ec ur ity , p ar tic ul ar ly e du ca tio n of m ot he rs .66 ,6 7 N um be r of a du lts e m pl oy ed fu ll or p ar t tim e w ith in a ho us eh ol d Co nt in uo us V ar ia bl e D ec re as e sc or e Em pl oy m en t in cr ea se s w ea lth a nd a cc es s to fo od in k in d fo r la bo r, de cr ea si ng fo od in se cu rit y. 66 ,6 8 An y ev en t of c hr on ic il ln es s in ch ild re n or a du lts Bi na ry o ut co m e ye s = 1 n o = 0 In cr ea se sc or e H ea lth a nd C hr on ic d is ea se a re m aj or s tr es so rs o n ho us eh ol d w ea lth , r es ou rc es a nd a bi lit y to w or k. 32 ,4 8, 68 Fo od in se cu rit y ca n al so in cr ea se t he li ke lih oo d of c hr on ic il ln es s. N um be r o f d ep en de nt s un de r t he ag e 18 w ith in a h ou se ho ld Co nt in uo us V ar ia bl e In cr ea se sc or e H ou se ho ld s w ith c hi ld re n ex pe rie nc e a hi gh er fo od in se cu rit y bu rd en 32 ,6 7 N um be r of p eo pl e w ith in a ho us eh ol d di vi de d by t he nu m be r of r oo m s. Co nt in uo us V ar ia bl e In cr ea se sc or e Va ria bl e is u se d as a p ro xy fo r cr ow di ng w ith in t he c om m un iti es .69 W ea lth Ca te go ric al v ar ia bl e, 4 q ua rt ile s D ec re as e sc or e Pr in ci pa l c om po ne nt a na ly si s (P CA ) w as u se d to c al cu la te 4 c at eg or ie s of w ea lth b as ed o n an o w ne rs hi p sc or e of t he fo llo w in g as se ts : s oa p, fu el , l an d ow ne rs hi p, a ni m al ow ne rs hi p, t oi le t ty pe , p ho ne , r ad io , r em itt an ce s. 32 ,5 9, 68 ,7 0– 72 Co m m un ity le ve l v ar ia bl e W ha t ty pe o f l an ds ca pe is m os t pr ev al en t in t he c om m un ity Ca te go ric al V ar ia bl e Fl at = 1 , M ix ed = 2 , H ill y = 3 In cr ea se sc or e Fl at la nd sc ap e is m or e id ea l f or g ro w in g cr op s. H ill y or s te ep in cl in es a re v ul ne ra bl e to er os io n. 73 Ra nk t he c om m un ity ’s ac ce ss t o sm al l m ar ke ts o r sh op s Ca te go ric al v ar ia bl e G oo d = 1 , F ai r = 2 , P oo r = 3 In cr ea se sc or e Ac ce ss t o sm al l m ar ke ts t o bu y or t ra de g oo ds is im po rt an t fo r da y to d ay a cc es s. 66 ,7 4 (C on tin ue d) 630 K. PATTERSON ET AL. Ta bl e 3. (C on tin ue d) . D es cr ip tio n Ty pe Ex pe ct ed D ire ct io n Ju st ifi ca tio n D ep en de nt V ar ia bl e Ra nk t he c om m un ity ’s ro ad ac ce ss in t er m s of p ro xi m ity an d qu al ity Ca te go ric al v ar ia bl e G oo d = 1 , F ai r = 2 , P oo r = 3 In cr ea se sc or e Ro ad a cc es s is k ey d et er m in an t to t ra ns po rt g oo ds t o tr ad e or it em s bo ug ht a t m ar ke ts an d an in di ca to r of in fr as tr uc tu re in t he a re a. 75 Ra nk t he c om m un ity ’s ac ce ss t o la rg e tr ad in g ce nt er s/ m ar ke ts Ca te go ric al v ar ia bl e G oo d = 1 , F ai r = 2 , P oo r = 3 In cr ea se sc or e La rg er c en te rs a nd m ar ke ts e na bl e pe op le t o ne go tia te o r ba rt er in a c om pe tit iv e en vi ro nm en t an d of te n ha ve lo w er p ric es t ha n sm al le r ce nt er s. 66 ,7 4 . H ow o ft en d oe s th e co m m un ity ha ve e ve nt s of c ro p ra id in g Ca te go ric al v ar ia bl e, o ft en = 1 , s om et im es = 2 , ne ve r = 3 D ec re as e sc or e Cr op r ai di ng in cr ea se s th e bu rd en o f f oo d se cu rit y as it r ed uc es y ie ld s an d le ss en s ha rv es t.76 Ra nk t he o ve ra ll qu al ity o f l an d w ith in t he c om m un ity Ca te go ric al v ar ia bl e G oo d = 1 , F ai r = 2 , P oo r = 3 In cr ea se sc or e Ar ab le la nd is a k ey d et er m in an t of y ie ld s an d ha rv es t po te nt ia l.19 ,7 3 Co nt ro l v ar ia bl es Se x of p er so n in c ha rg e of ho us eh ol d fo od p re pa ra tio n bi na ry o ut co m e m al e = 1 , f em al e = 0 N /A Re sp on se s m ay d iff er b y ge nd er , w om en m ay r ep or t hi gh er fo od in se cu rit y th an m en .77 H ow ev er , w om en a re t yp ic al ly r es po ns ib le fo r fo od p re pa ra tio n an d m ay b e m or e aw ar e of t he h ou se ho ld fo od s itu at io n .67 JOURNAL OF HUNGER & ENVIRONMENTAL NUTRITION 631 household measures), 2) community-level variables only (accounting for repeated household measures and clustering within communities), and 3) household and community-level variables combined in a multilevel model (accounting for repeated household measures and community-level cluster ing). A manual iterative stepwise approach was used in model building. Akaike Information Criterion (AIC) was utilized to assess the best-fit model.78 We conducted sensitivity analyses for all excluded candidate variables by testing them in the final best-fit model. We conducted post-estimation assessments of the models, examining normality and homogeneity of variance for the best linear unbiased predictors and a graphical assessment of Pearson’s residuals. Variation accounted for by the model and random affects was recorded to calculate the intraclass correlation (ICC) or variance partitioning coefficient (VPC). These were used to analyze the amount of variation accounted by each level included in the model. Finally, we assessed how well the models were able to predict true food insecurity levels of households in the sample. All analyses were conducted in Stata version 13 (StataCorp13). Results Batwa Demographics A full census of the Batwa residing in Kanungu district was attempted; the response rates ranged between 93% and 99% for individuals, and each admin istration had 99% participation from households. The demographic structure of Batwa households was relatively homogenous within and across communities, with an average of 4.7 people (2.6 dependents). A typical Batwa household lives in a 2-room mud house with iron sheet roofs. Sixty-five percent of households ranked in the two lowest categories of wealth. The majority of households (77%) reported owning land; the BDP and Kellerman foundation hold these land titles in trust. No households owned bicycles or motorbikes, and none had electricity. Eleven percent owned phones and 40% owned a radio. Few Table 4. Distribution of assets by principle component analysis category among Indigenous Batwa in Kanungu District, Uganda, October 2012-April 2014. Principle component analysis – Wealth Status Assets Poorest Very Poor Poor Least Poor Soap 0% 100% 5.29% 41.80% Land ownership 70.68% 76.83% 74.07% 83.07% Pigs 0% 0% 10.58% 25.40% Chickens 0% 0% 3.17% 47.62% Cell phone ownership 0% 0% 8.99% 37.04% Radio ownership 0% 0% 77.25% 75.13% Receiving remittances 10.10% 13.41% 10.05% 6.88% 632 K. PATTERSON ET AL. households (10%) had external monetary support or remittances. Sixty-one percent of households had no adult females with formal schooling, 32% had some primary school and 7% of households had an adult female who had completed primary school. More than ninety-five percent of households parti cipated in tending crops and manual labor. Only 10% reported tending animals and less than 5% reported participating in hunting activities. Due to this low variation among and between households and communities, further analysis of these agricultural variables was precluded. Most households did not own any animals (73%); those that did owned chickens (13%), pigs (9%) and goats (13%). Communities ranged in size from 3 to 21 households. Landscape in commu nities ranged from flat to hilly, with most being mixed. No communities had good quality road access, 50% had fair and 50% had poor access. Access to large trading centers was fair or poor; small markets were more accessible. Land quality was generally good (40%) or poor (40%). Crop raiding occurred in communities located near the national park, 50% did not experience any crop raiding. Prevalence of Food Security among Batwa Households The Batwa of Kanungu are severely food insecure (Table 5). According to the HFSSM standard categorization, 97% of households were food insecure, with 13% of households reporting low food security and 84% experiencing very low food security. The AVPS was better able to distinguish variation between households, particularly in the most severe category. Cronbach’s alpha for the AVPS was 0.8931, indicating high internal reliability. Among the Batwa the mean food insecurity score was 14.37 (out of 26) and the median was 15. Comparing these results to the HFSSM categorization, both the median and mean of the adapted score fall within the most severe HFSSM category of food Table 5. Food security results (HFSSM and AVPS) for Batwa households in Kanungu District, Uganda, October 2012-April 2014. Food Security level Households without children (%) Households with children (%) Total (%) HFSSM 86 681 767 Food secure High food security 7 (8) 9 (1) 16 (2) Marginal food security 0 (0) 10 (1) 10 (1) Food insecure Low food security 15 (17) 81 (12) 96 (13) Very low food security 64 (74) 581 (85.31) 645 (84) AVPS 26-point scale Minimum, maximum (Standard deviation) 0–18.57 (4.70) 0–26 0–26 (5.72) Mean 9.92 14.92 14.37 Median 10.44 16 15 JOURNAL OF HUNGER & ENVIRONMENTAL NUTRITION 633 insecurity. Experiences of food insecurity did not vary significantly by community. Ninety percent of households responded that they were worried they would run out of food, that food would not last and that they could not afford balanced meals (Fig. 2). Households with children fed children lesser quality foods, unbalanced meals, and felt they were unable to feed their children enough food. Sixty-eight percent of households reported eating two meals or fewer per day. Seventy one percent of adults reported skipping meals with 31% skipping more than 3 meals in the past week. The majority of adults also reported eating less than they felt they should (81%), feeling hungry (66%), losing weight (68%), and not eating for whole days in the past week (50%). Households reported cutting the size of their children’s meals (89%), skipping meals (57%), children feeling hungry (59%) and children being unable to eat for a whole day (41%) in the past week. Variables Associated with Batwa Food Insecurity Based on unconditional linear regression at the household level, we retained the following household-level variables in the mixed-effect multivariable mod els: adult female education, presence of a household member with a chronic disease, the total number of household members and dependents, number of people sleeping per room, and household wealth. At the community level, we tested the role of landscape type, road access, access to trading centers, and Figure 2. Word cloud of food consumption in batwa households, Kanungu District, Uganda (2013–14). The more frequently a word is mentioned by households, the larger and darker its presentation in the image. 634 K. PATTERSON ET AL. crop raiding. Based on unconditional univariate analysis of these variables, we retained market access and experiences with crop raiding as community-level variables in the mixed-effects multivariable models. All multivariable models accounted for repeated measures of households over multiple surveys (Table 6). In the null model, intra-class correlations at the household level (0.18) exceeded the community level (0.01), both indicat ing low levels of between-group variation. A higher household-level ICC reflects the repeated nature of the data, with households similar across multi ple measurements. In the household-only mixed-effects model, households in the poorest quartiles had significantly higher food insecurity than households in the least poor quartile (4th), 1.62 (p < .01) (1st quartile) and 1.70 (p < .01) (3rd quartile). For each dependent a household had, food insecurity increased by Table 6. Final multivariable mixed effects linear regression model using random intercepts to control for repeated household measurements and community-level clustering for associations with food insecurity among the Batwa of Kanungu District, Uganda, October 2012-April 2014. Model name (description) Null model Model 1 (Household associations only) Model 2 (Community associations only) Model 3 (Household and community associations) Intercept/Constant 14.36 14.23 12.03 Total variation (%) 32.63 (100) 25.75 (100) 32.59 (100) 25.90 (100) Explained by household (%) - 4.27 (25.58) 5.39 (16.53) 3.14 (12.12) Explained by community (%) - - 0.48 (1.47) 0.08 (0.00) Random effects/Unexplained (%) 32.63 (100) 21.43 (83.22) 26.72 (81.98) 22.68 (87.56) Intra-class correlation/ Variance partitioning coefficient Household level 0.18 0.14 0.16 0.11 Community level 0.01 - 0.00 0.002 Independent variables Highest Adult female education - - Primary complete or Above ref. ref. Primary incomplete 0.21 0.17 No formal Schooling 0.99 0.90 Presence of chronic disease - - No ref. ref. Yes 0.71 0.66 Number of dependents - 0.37*** - 0.41*** Wealth quartiles - - Least poor ref. ref. Poor 1.70** 1.67** Very poor −0.03 −0.19 Poorest 1.62** 1.59** Crop raiding - - Never ref. ref. Sometimes 0.34 .80 Often 0.93* 1.28* Access to markets - - Good ref. ref. Fair −1.59* −1.57* Poor −0.17 −0.41 AIC of model 4816.425 4803.353 4815.815 4777.72 JOURNAL OF HUNGER & ENVIRONMENTAL NUTRITION 635 0.37 points (p < .01). Female education and the presence of chronic disease in the household were not significant in this model but contributed to model fit. The data indicated that the presence of educated females may be protective, and that the presence of chronic disease may increase food insecurity. The ICC for the household level (0.14) decreased modestly compared to the null model. In the mixed-effects multivariable community-level model, communities subject to regular crop raiding showed higher levels of food insecurity (0.93- point increase compared to those with no crop raiding). Access to markets was also associated with higher food insecurity. Households in communities with fair access to markets had a food insecurity score 1.59 points (p < .05), lower (i.e. more food secure) compared to households in communities with good access. The explanatory power of the community-level model was poorer (higher AIC) compared to both the null and household-level models. Further, community-level clustering accounted for less than 2% of the varia tion, with variation from household and random effects >98%. These results indicate that food insecurity was highly homogenous among communities, and that community differences did not contribute to explaining variation among households in food insecurity. The final best-fit multilevel model incorporating both household and com munity-level variables and accounting for repeated household measures and community-level clustering retained four household level and two commu nity-level associations (identical to those retained by the household-only and community-only models). Coefficients of the variables varied only marginally between the models; both direction and strength of associations for all vari ables were stable across the three models. Wealth – measured via PCA – had the largest protective impact on the food security score, though even in the case of wealth, the magnitude of impact was relatively low: a difference of only 1.59–1.67 points on a scale of 26 for food insecurity between the poorest and least poor quartiles. Community-level variation, similar to the Model 2 was <1%, while variation explained by the household level decreased to 12.12%. Post-estimation diagnostics were normal, and the model was a good fit for the data. Despite the retention of significant independent variables at both the house hold and community levels in the best-fit models, all models had the poor capacity to explain variation in food security among households and commu nities. Notably, the magnitude of all the regression coefficients was low, even for categorical independent variables with high theoretical predictive capacity (e.g. education, wealth). Discussion This study found that the prevalence of food insecurity among the Batwa was among one of the highest in the published literature, with more than 97% of 636 K. PATTERSON ET AL. households being classified as food insecure over six surveys between January 2013 and April 2014. A new scale, AVPS, was developed to capture variation of food security occurring within the HFSSM’s worst category. The model found that wealth, experiences with crop raiding and proximity to markets were significant associated with food security for the Batwa. However, due to the homogenous nature of severe poverty among Batwa households, the significantly associations with food insecurity found in our models explained little of the variation in food security. The scale at which food security should be measured is debated in the literature, with no clear agreement.79,80 Policymakers and governments favor the household-level measure as it is cost-effective yet focused enough to deal with context-specific issues. Some argue that the current household-level scale of food security analysis is overly broad and suggest that the indivi dual-level would be more appropriate.79,80 However, there has been negli gible empirical research attempting to integrate broader scale associations with food insecurity into quantitative food security research, despite advances in multilevel modeling.66 Review of the drivers of food insecurity in southern Africa re-iterates the role of external factors on household associations with food security. Misselhorn found that food insecurity resulted from “the interaction between environmental stressors, and socio- economic conditions over various time scales”.66 Developments in public health research highlight growing recognition of “neighbourhood” (or higher level) effects, and multilevel modeling techniques are increasing applied to understand the relative contribution of determinants of health outcomes across scales.81–88 Figure 3. Global comparison of household food security survey module (HFSSM) status. High and marginal food security have been collapsed into one category. JOURNAL OF HUNGER & ENVIRONMENTAL NUTRITION 637 Our findings represent among the highest levels of food insecurity recorded in the peer-reviewed literature (Fig. 3). Ninety seven percent of Batwa house holds in Kanungu were found to be food insecure. Levels of food insecurity among Batwa exceed reported prevalence among the surrounding neighbouring Bakiga population and other Indigenous or low-income popula tions globally, confirming findings of high impoverishment and anecdotal and gray literature reports of severe food insecurity.10,17,18,21,23,38 Other studies have found similar disparities in the experiences of infant death, acute gastro intestinal illness, malaria, malnutrition between the Batwa and their neighbor ing populations.38,41,89–91 The Batwa have been systemically marginalized and discriminated against based on their physical appearance, their identities as hunter-gatherers, and lack of literacy (read, write and speak in English).92,93 Finally, in Uganda the recognition of Indigenous peoples includes all ethnic groups; therefore, policy instruments meant to protect the rights of Indigenous peoples are of little use to the Batwa.23,92 We considered both community and household-level food security vari ables to explain variation in food insecurity among the Batwa. Community variables, though retained in the best-fit model, provided negligible power to explain variations in food security. Several household-level variables provided weak but significant explanatory power. The number of dependents increased household experiences of food insecurity. This finding is similar to other case studies of highly food insecure populations.94–97 Larger households have more numbers to feed, and the Batwa lacked sufficient wealth and assets to ensure adequate food for each member. Reduction in food intake occurred at a higher prevalence for adults than children. This finding was consistent with similar findings elsewhere and supports the theory that households may be protecting children from hunger: adults take on a higher burden and buffer children’s exposure to food insecurity.98,99 Adults within communities’ work for food in- kind or for money, which is then shared with household members unable to work (including children or elderly). Additionally, education and health costs of dependents increased the burden on household wealth.95,96 Adult female education was predictive of food security. Among the Batwa, adult literacy was very low compared to the Ugandan average and women tended to be less educated. Formal educational opportunities – within the Ugandan national schooling system – were largely absent for Batwa until their eviction from the forest in 1991, and many were unable to afford the school fees until the Batwa Development Programme established a scholarship program for all Batwa children to attend school for free in 2006. Older Batwa (>45 yrs.) typically have no formal education, while education of younger adult Batwa remained limited. Young Batwa (<20 yrs.) had the highest levels of education (increas ingly frequent attendance at secondary school and a small number of post- secondary students) and reflect an emerging generational shift in education levels of future adults and household heads. 638 K. PATTERSON ET AL. In lower income countries, the wealth acquisition of rural populations is typically from agriculture or manual labor and/or remittances. Wealth assets usually center around owning animals, radios, cell phones, and modes of transportation.100–102 National associations with food insecurity in Uganda include poverty, living in rural areas compared to urban, having a female- headed household, subsistence farming, owning land and distance to access the land.103 Eighty percent of Uganda’s population depends on agriculture. Among the Batwa, however, owning land was not found to be significant in affecting food security, likely since most land has been purchased through an NGO and is held in trust; indeed, Batwa reported poor land fertility and low yields.104,105 This is further exacerbated by experiences with crop raiding by wildlife; communities within close proximity of the BNIP borders reported higher frequency of raiding events.76,106,107 Owning animals, and having employed members of the household theorized as indicators of wealth within the Batwa population and as potentially associated with increased food secur ity. Owning animals was not significantly associated with increased food security, likely due to low ownership within these communities. Household employment was analyzed through number of members working, number working full-time jobs, number of women employed, and presence of a salaried job. None of these were significantly associated with food security. Lack of variance in both food security outcomes and measured independent variables constrained analysis and identification of the associations with food insecurity among the Batwa. For example, the importance of agriculture often varies greatly from one household to one another, and from one community to one other. However, in our study, we found little variation between and among households and communities, which precluded further analysis of the agriculture variables captured by the questionnaire. Further research should include more sensitive and specific questions in attempts to capture more heterogeneity in responses among participants. For instance, at the household level, time to access to cultivated plots, land size, agriculture yields, type of crops grown, crop diversity, access market information through cell phones, among other factors are often important. At the community level, existence of agricultural extension services, time to access the nearest market, among other factors are often important. Social networks and sharing networks may explain some of the absence of variation in food security outcomes, and the low impact of wealth.108 Data on social networks were not available for this research, but informal discussions and observations indicated regular movement of households and members between communities. Social and reciprocal relationships are often found in Indigenous Peoples globally and may help explain how households are able to acquire food if not through typical sources of wealth, i.e., owning animals or having a wage employment.109–111 However, exposure to adverse realities did not vary among the Batwa: all Batwa reported difficulty accessing sufficient JOURNAL OF HUNGER & ENVIRONMENTAL NUTRITION 639 food. Homogenously extreme impoverishment, as in this case, may preclude development of effective social and reciprocal networks. Rose 30 highlights the shortcomings of seeking within a homogenous popu lation to identify outliers and defining normal as the average within a local population: as a result, “real” determinants – those outside the scale of study – may remain undiscovered. Even exposure variables that are associated with 100% of poor outcomes will explain 0% of variation in that outcome if that variable is invariant in a population.29 In the case of food insecurity among the Batwa, our results suggest that the most significant associations with food insecurity may explain little of the variation in food insecurity – not because they are not important causes of food insecurity, but because they vary little when households are used as the scale of analysis. Interpretation of significant independent variables based on these results could indicate interventions targeted at educating women, eliminating chronic disease presence, reducing family size and increasing wealth may increase food security. However, while interventions targeting education and wealth are likely to be appropriate within this population, the Batwa likely face larger social mediating effects that are invariant at the household scale or even community scale, including exposure to inequality, and lack of social capital and access. In this case-study, even multi-level modeling to include the community level was insufficient. Rose30 provides further insights and suggests that policy-relevant variance may be observable only between populations rather than within them, parti cularly if an exposure is unvarying at the scale of analysis. Herein, multilevel models that include regional comparisons (Batwa versus neighboring popula tions) would be required to fully – and appropriately – identify the distinctive causes of Batwa food insecurity. Limitations There are a few limitations that should be addressed. First, measuring food security is difficult and the measure used here was based on self-reporting and perceptions. this can be subjective and potentially impacted by recall bias.112 However, we shortened the recall period to better respond to community preferences (1 week, rather than 3 months), and we worked with community members to ensure the translations and meaning of the questions were clear through pilot work (four pilot surveys were conducted in 2011–2012). Second, alternative tools to measure food insecurity are available and widely used, and may have resulted in more variance for analysis. We brought several options to the community for feedback and the choice to use the USDA HFSSM and the AVPS to categorize food security was finalized by our community partners. Third, comparative analysis would likely illustrate and underscore the assumptions made here, that food insecurity is worse than neighboring popu lations and may have illuminated other inequities in social determinants of 640 K. PATTERSON ET AL. health. Again, during the study design Batwa communities preferred focusing on Batwa-only sampling. In the future, in partnership with Batwa commu nities, a comparative study may be supported to illustrate and analyze the inequities between communities. Fourth, the positionality of the researchers may have impacted answers.113–116 Some studies have found that the level of food insecurity may be over-emphasized to receive potential aid, or that participants respond with answers they believe the researchers want to hear.112 Extensive pilot work was conducted in an effort to integrate commu nities into the research process and prevent this. The research team acknowl edged their positionality and attempted to reduce the asymmetry in power between the researchers and participants. However, given the inequity this community experiences, the power imbalance may still have impacted responses by community members. Conclusion Our results suggest that variation in food insecurity may be poorly reflected in household-level, or even community-level analyses and that variation and key predictors of food security – and therein appropriate intervention entry points – are likely occurring at the regional scale. Food insecurity analyses should be considered scale-dependent, with increased use of multilevel mod eling approaches to identify methodologically appropriate and policy-relevant scales of analysis. The AVPS may be appropriate for analyses where a large proportion of the population are food insecure. The scale dependence of food insecurity among the Batwa, may indicate the scale dependence of other burdens of disease. A comparative study between the Batwa and the Bakiga may provide valuable insight into the determinants of health outcomes in the District. While the Batwa do have extremely high burdens of food insecurity, they have an extensive history of adaptation and resilience. Moving forward, the Batwa will need external support to negate historical injustices and part nerships strengthening their capacity. Funding This work was supported by the Canadian Institutes of Health Research [298312]; CIHR/ NSERC/SSHRC and IDRC Tri-Council Initiative on Adaptation to Climate Change [106372- 003,106372-004,106372-005]. ORCID Kaitlin Patterson http://orcid.org/0000-0002-3444-0472 Lea Berrang-Ford http://orcid.org/0000-0001-9216-8035 Shuaib Lwasa http://orcid.org/0000-0003-4312-2836 JOURNAL OF HUNGER & ENVIRONMENTAL NUTRITION 641 Didacus B. Namanya http://orcid.org/0000-0001-6906-4617 James Ford http://orcid.org/0000-0002-2066-3456 Sherilee L. Harper http://orcid.org/0000-0001-7298-8765 References 1. Hinrichs CC. Regionalizing food security? imperatives, intersections and contestations in a post-9/11 world. J Rural Stud. 2013;29(January):7–18. doi:10.1016/J. JRURSTUD.2012.09.003. 2. McMichael P, Schneider M. Food security politics and the millennium development goals. Third World Quarter. 2011;32(1):119–139. doi:10.1080/01436597.2011.543818. 3. Schmidhuber J, Tubiello FN. Global food security under climate change. Proc Natl Acad Sci. 2007;104(50):19703–19708. doi:10.1073/pnas.0701976104. 4. Wheeler T, Braun, JV. Climate change impacts on global food security. Science. 2013;341 (6145):508–513. doi:10.1126/SCIENCE.1239402. 5. Godfray H, Charles J, Beddington JR, et al. Food security: the challenge of feeding 9 billion people. science. 2010;327(5967):812–818. doi:10.1126/science.1185383. 6. Challinor A, Wheeler T, Garforth C, Craufurd P, Kassam A. Assessing the vulnerability of food crop systems in Africa to climate change. Climatic Change. 2007;83(3):381–399. doi:10.1007/s10584-007-9249-0. 7. Confalonieri U, Menne B, Akhtar R, Ebi KL, Hauengue M, Kovats RS 2007. “Human health.” In climate change 2007: impacts, adaptation and vulnerability: contribution of working group ii to the fourth assessment report of the intergovernmental panel on climate change. 8. Fussell E. The cumulative causation of international migration in latin America. Ann Am Acad Political Soc Sci. 2010;630(1):162–177. doi:10.2307/20743992. 9. Tschakert P. Views from the vulnerable: understanding climatic and other stressors in the sahel. Global Environment Change. 2007;17(3–4):381–396. doi:10.1016/J. GLOENVCHA.2006.11.008. 10. Berrang-Ford L, Kathryn Dingle JD, Ford CL, et al. Vulnerability of indigenous health to climate change: a case study of Uganda’s batwa pygmies. Soc Sci Med. 2012;75 (6):1067– 1077. doi:10.1016/J.SOCSCIMED.2012.04.016. 11. Lynn K, Daigle J, Hoffman J, et al., The Impacts of Climate Change on Tribal Traditional Foods. Climatic Change 2013;120:545–556. doi:10.1007/978-3-319-05266-3_4 12. IWIGA. The Indigenous World 2019 [Internet]. Copenhagen; 2019 [cited 2019 Nov 5]. Available from: https://www.iwgia.org/images/documents/indigenousworld/ IndigenousWorld2019_UK.pdf 13. Maldonado JK, Colombi B, Pandya R, eds. Climate Change and Indigenous Peoples in the United States, 2014. cham:springer international publishing. 10.1007/978-3-319- 05266-3 14. ACHPR. 2005. “Report of the African commission’s working group of experts on indigenous populations.” Banjul.15. Balenger S, Fried S, Kanchev K, Between Forest and Farm: Identifying Appropriate Development Options for the Batwa of Southwestern Uganda. Washington, DC: First Peoples Worldwide; 2005. 15. Balenger S, Fried S, Kanchev K, Between Forest and Farm:Identifying Appropriate Development Options for the Batwa of Southwestern Uganda. Washington, DC: First Peoples Worldwide; 2005. 16. Jackson, D., & Payne, K. 2003. “Twa women, Twa rights in the Great Lakes region of Africa.” London: Minority Rights Group International. Retrieved from. https://minorit 642 K. PATTERSON ET AL. https://doi.org/10.1016/J.JRURSTUD.2012.09.003 https://doi.org/10.1016/J.JRURSTUD.2012.09.003 https://doi.org/10.1080/01436597.2011.543818 https://doi.org/10.1073/pnas.0701976104 https://doi.org/10.1126/SCIENCE.1239402 https://doi.org/10.1126/science.1185383 https://doi.org/10.1007/s10584-007-9249-0 https://doi.org/10.2307/20743992 https://doi.org/10.1016/J.GLOENVCHA.2006.11.008 https://doi.org/10.1016/J.GLOENVCHA.2006.11.008 https://doi.org/10.1016/J.SOCSCIMED.2012.04.016 https://doi.org/10.1007/978-3-319-05266-3%5F4 https://www.iwgia.org/images/documents/indigenousworld/IndigenousWorld2019_UK.pdf https://www.iwgia.org/images/documents/indigenousworld/IndigenousWorld2019_UK.pdf https://doi.org/10.1007/978-3-319-05266-3 https://doi.org/10.1007/978-3-319-05266-3 https://minoritrights.org/publications/twa-women-twa-rights-in-the-great-lakes-region-of-africanovember-2003/ rights.org/publications/twa-women-twa-rights-in-the-great-lakes-region-of-africano vember-2003/ 17. Tumushabe G, Musiime E, Living on the margins of life: the plight of the Batwa communities of South Western Uganda. ACODE Policy Research Series. Kampala, Uganda: Advocates Coalition for Development and Environment; 2006. 17. 18. Warrilow F, Wood A, Levin E, Hinton J, The right to learn: batwa education in the great lakes region of Africa. Kampala, Uganda: Minority Rights Group International; 2008; 19. Foley JA, Ramankutty N, Brauman KA, et al. Solutions for a cultivated planet. nature. 2011;478(7369):337–342. doi:10.1038/nature10452 20. Gregory PJ, Ingram JSI, Brklacich M. Climate Change and Food Security. Philos Trans R Soc Lond B Biol Sci. 2005;360(1463):2139–2148. doi:10.1098/rstb.2005.1745. 21. Patterson K, Berrang-Ford L, Lwasa S, et al. Seasonal variation of food security among the batwa of Kanungu, Uganda. Public Health Nutr. 2017;20(1):1. doi:10.1017/ S1368980016002494. 22. Lewnard JA, Berrang-Ford L, Lwasa S, et al. Relative undernourishment and food insecurity associations with plasmodium falciparum among batwa pygmies in uganda: evidence from a cross-sectional survey. The American Journal of Tropical Medicine and Hygiene. 2014;91(1):1. doi:10.4269/ajtmh.13-0422 23. UOBDU, COPACSO, and FPP. 2015. “Indigenous peoples in uganda: a review of the human rights situation of the batwa people, the benet people and pastoralist commu nities.” Kanungu. 24. Cafiero C, Melgar-Quiñonez HR, Ballard TJ, Kepple AW. Validity and reliability of food security measures. Ann N Y Acad Sci. 2014;1331(1):230–248. doi:10.1111/nyas.12594. 25. Coates J. Build it back better: deconstructing food security for improved measurement and action. Global Food Security. 2013;2(3):188–194. doi:10.1016/J.GFS.2013.05.002. 26. Jones AD, Ngure FM, Pelto G, Young SL. What are we assessing when we measure food security? a compendium and review of current metrics. Adv Nutr. 2013;4(5):481–505. doi:10.3945/an.113.004119. 27. Lohman BJ, Stewart S, Gundersen C, Garasky S, Eisenmann JC. Adolescent overweight and obesity: links to food insecurity and individual, maternal, and family stressors. J Adolesc Health. 2009;45(3):230–237. doi:10.1016/j.jadohealth.2009.01.003. 28. Weiser SD, Fernandes KA, Brandson EK, et al. the association between food insecurity and mortality among HIV-infected individuals on HAART JAIDS. 2009;52(3):342–349. doi:10.1097/QAI.0b013e3181b627c2. 29. Pearce N. Epidemiology in a changing world: variation, causation and ubiquitous risk factors. Int J Epidemiol. 2011;40(2):503–512. doi:10.1093/ije/dyq257. 30. Rose G. Sick individuals and sick populations. Int J Epidemiol. 2001;30(3):427–432. doi:10.1093/ije/30.3.427. 31. Fisher M, Lewin PA. Household, community, and policy determinants of food insecurity in rural Malawi. Develop Southern Afr. 2013;30(4–05):451–467. doi:10.1080/ 0376835X.2013.830966. 32. Hadley C, Linzer DA, Belachew T, Mariam AG, Tessema F, Lindstrom D. Household capacities, vulnerabilities and food insecurity: shifts in food insecurity in urban and rural ethiopia during the 2008 food crisis. Soc Sci Med. 2011;73(10):1534–1542. doi:10.1016/j. socscimed.2011.09.004. 33. Nalty CC, Sharkey JR, Dean WR. Children’s reporting of food insecurity in predomi nately food insecure households in texas border colonias. Nutr J. 2013;12(1):15. doi:10.1186/1475-2891-12-15. JOURNAL OF HUNGER & ENVIRONMENTAL NUTRITION 643 https://minoritrights.org/publications/twa-women-twa-rights-in-the-great-lakes-region-of-africanovember-2003/ https://minoritrights.org/publications/twa-women-twa-rights-in-the-great-lakes-region-of-africanovember-2003/ https://doi.org/10.1038/nature10452 https://doi.org/10.1098/rstb.2005.1745 https://doi.org/10.1017/S1368980016002494 https://doi.org/10.1017/S1368980016002494 https://doi.org/10.4269/ajtmh.13-0422 https://doi.org/10.1111/nyas.12594 https://doi.org/10.1016/J.GFS.2013.05.002 https://doi.org/10.3945/an.113.004119 https://doi.org/10.1016/j.jadohealth.2009.01.003 https://doi.org/10.1097/QAI.0b013e3181b627c2 https://doi.org/10.1093/ije/dyq257 https://doi.org/10.1093/ije/30.3.427 https://doi.org/10.1080/0376835X.2013.830966 https://doi.org/10.1080/0376835X.2013.830966 https://doi.org/10.1016/j.socscimed.2011.09.004 https://doi.org/10.1016/j.socscimed.2011.09.004 https://doi.org/10.1186/1475-2891-12-15 34. Nawrotzki RJ, Kristin Robson MJ, Gutilla LM, Hunter WT, Norlund P. Exploring the impact of the 2008 global food crisis on food security among vulnerable households in rural South Africa. Food Security. 2014;6(2):283–297. doi:10.1007/s12571-014-0336-6. 35. UOBDU. 2007. “Batwa Data of Dec 2007 about Kisoro, Kabale, Kanungu, Katovu, Mbarara and Ntungamo.” Kanungu. 36. Donnelly B, Berrang-Ford L, Labbé J, et al. Plasmodium falciparum malaria parasitaemia among indigenous batwa and non-indigenous communities of Kanungu District, Uganda. Malar J. 2016;15(1):254. doi:10.1186/s12936-016-1299-1. 37. Kulkarni MA, Garrod G, Berrang-Ford L, et al. Examination of antibody responses as a measure of exposure to malaria in the indigenous batwa and their non-indigenous neighbors in Southwestern Uganda. Am J Trop Med Hyg. 2017;96(2):330–334. doi:10.4269/ajtmh.16-0559 38. Sauer J, Berrang-Ford L, Patterson K, et al. An analysis of the nutrition status of neighboring indigenous and non-indigenous populations in Kanungu District, Southwestern Uganda: close proximity, distant health realities. Soc Sci Med. 2018;217 (November):55–64. doi:10.1016/j.socscimed.2018.09.027. 39. World Bank. 2015. “Uganda | data.” 2015. https://data.worldbank.org/country/uganda. Accessed May 10, 2017 40. UBOS. 2016. “national population and housing census 2014 – main report.” Kampala. 41. Ohenjo N, Willis R, Jackson D, et al., Indigenous Health 3: health Of indigenous people in Africa. Lancet. 2006;367:1937–1946. doi:10.1016/S0140-6736(06)68849-1 42. Sherman M, Ford JD. Market engagement and food insecurity after a climatic hazard. Global Food Security. 2013;2(3):144–155. doi:10.1016/J.GFS.2013.08.003. 43. FAO. 1996. “Rome declaration on world food security.” food and agriculture organiza tion of the United Nations. 44. Cook JT, Frank DA, Berkowitz C, et al. Food insecurity is associated with adverse health outcomes among human infants and toddlers. J Nutr. 2004;134(6):1432–1438. d oi:10.1093 /jn/134.6.## 45. Kursmark M, Weitzman M. Recent findings concerning childhood food insecurity. Curr Opin Clin Nutr Metab Care. 2009;12(3):310–316. doi:10.1097/MCO.0b013e3283298e37. 46. McIntyre L, Tarasuk V, Li TJ. Improving the nutritional status of food-insecure women: first, let them eat what they like. Public Health Nutr. 2007;10(11):1288–1298. doi:10.1017/S1368980007702902. 47. Vozoris NT, Tarasuk VS. Household food insufficiency is associated with poorer health. J Nutr. 2003;133(1):120–126. doi:10.1093/jn/133.1.120. 48. Seligman HK, Laraia BA, Kushel MB. Food insecurity is associated with chronic disease among low-income NHANES participants. J Nutr. 2010;140(2):304–310. doi:10.3945/ jn.109.112573. 49. Ford JD, Berrang-Ford L. Food security in igloolik, nunavut: an exploratory study. Polar Rec. 2009;45(3):225–236. doi:10.1017/S0032247408008048. 50. Kaiser LL, Melgar-Quiñonez H, Townsend MS, et al., Food insecurity and food supplies in latino households with young children. J Nutr Educ Behav. 2003;35(3):148–153. doi:10.1016/S1499-4046(06)60199-1 51. Melgar-Quinonez HR, Zubieta AC, MkNelly B, Nteziyaremye A, Maria Filipinas DG, Dunford C. Household Food Insecurity And Food Expenditure In Bolivia, Burkina Faso, and the Philippines. J Nutr. 2006;136(5):1431S–1437S. doi:10.1093/jn/136.5.1431S. 52. Studdert LJ, Frongillo EA, Valois P. Household food insecurity was prevalent in java during indonesia’s economic crisis. J Nutr. 2001;131(10):2685–2691. doi:10.1093/jn/ 131.10.2685. 644 K. PATTERSON ET AL. https://doi.org/10.1007/s12571-014-0336-6 https://doi.org/10.1186/s12936-016-1299-1 https://doi.org/10.4269/ajtmh.16-0559 https://doi.org/10.1016/j.socscimed.2018.09.027 https://data.worldbank.org/country/uganda https://doi.org/10.1016/S0140-6736(06)68849-1 https://doi.org/10.1016/J.GFS.2013.08.003 https://doi.org/10.1097/MCO.0b013e3283298e37 https://doi.org/10.1017/S1368980007702902 https://doi.org/10.1093/jn/133.1.120 https://doi.org/10.3945/jn.109.112573 https://doi.org/10.3945/jn.109.112573 https://doi.org/10.1017/S0032247408008048 https://doi.org/10.1016/S1499-4046(06)60199-1 https://doi.org/10.1093/jn/136.5.1431S https://doi.org/10.1093/jn/131.10.2685 https://doi.org/10.1093/jn/131.10.2685 53. Charron DF, Ecohealth research in practice. Ecohealth Research in Practice. 2012.255– 271. New York: Springer New York. doi:10.1007/978-1-4614-0517-7_22 54. Cochran PALL, Marshall CA, Garcia-Downing C, et al. Indigenous ways of knowing: implications for participatory research and community. Health Policy Ethics. 2008;98 (1):22–27. doi:10.2105/AJPH.2006.093641. 55. Koster R, Baccar K, Harvey Lemelin R. Moving from research on, to research with and for indigenous communities: a critical reflection on community-based participatory research. canad geograph/le géographe canadien. 2012;56(2):195–210. doi:10.1111/ j.15410064.2012.00428.x 56. IPC. 2014. “Report of the integrated food security phase classification analysis for Uganda.” 57. Bickel G, Nord M, Price C, Hamilton W, Cook J, Guide to Measuring Household Food Security. Washington, D.C: United States Department of Agriculture; 2000. 58. Barrett CB, Marenya PP, Mcpeak J, et al. Welfare dynamics in rural Kenya and Madagascar. JDev Stud. 2006;42(2):248–277. doi:10.1080/00220380500405394. 59. Filmer D, Scott K. Assessing asset indices. demography. 2012;49(1):359–392. doi:10.1007/s13524-011-0077-5. 60. Hargreaves JR, Morison LA, Gear JSS, et al. Assessing household wealth in health studies in developing countries: a comparison of participatory wealth ranking and survey techniques from rural South Africa. Emerg Themes Epidemiol. 2007;4(1):4. doi:10.1186/1742-7622-4-4. 61. Tavakol M, Dennick R. Making sense of cronbach’s alpha. Int J Med Educ. 2011;2 (June):53–55. doi:10.5116/ijme.4dfb.8dfd. 62. Bursac Z, Gauss CH, Williams DK, Hosmer DW. Purposeful selection of variables in logistic regression. Source Code Biol Med. 2008;3(1):17. doi:10.1186/1751-0473-3-17. 63. Mickey R, Greenland S. The impact of confounder selection criteria on effect estimation. Am J Epidemiol. 1989;129(1):125–137. doi:10.1093/oxfordjournals.aje.a115101. 64. Bendel RB, Afifi AA. Comparison of Stopping Rules in Forward “Stepwise” Regression. J Am Statistic Assoc. 1977;72(357):46. doi:10.2307/2286904. 65. Tabachnick BG, Fidell LS, Using Multivariate Statistics. 7th Edition, ed. Boston: Pearson; 2019. 66. Misselhorn AA. What drives food insecurity in southern africa? A meta-analysis of household economy studies. Global Environment Change. 2005;15(1):33–43. doi:10.1016/J.GLOENVCHA.2004.11.003. 67. Metallinos-Katsaras, E., Must, A., & Gorman, K. (2012). A Longitudinal Study of Food Insecurity on Obesity in Preschool Children. Journal of the Academy of Nutrition and Dietetics, 112(12),1949–1958 https://doi.org/10.1016/j.jand.2012.08.031 68. Osei, A., Pandey, P., Spiro, D., Nielson, J., Shrestha, R., Talukder, Z.,. . . Haselow, N. (2010). Household Food Insecurity and Nutritional Status of Children Aged 6 to 23Months in Kailali District of Nepal. Food and Nutrition Bulletin, 31(4),483–494 https://doi.org/10.1177/156482651003100402 69. Willows, N. D., Veugelers, P., Raine, K., & Kuhle, S. (2009). Prevalence and socio demographic risk factors related to household food security in Aboriginal peoples in Canada. Public Health Nutrition, 12(8),1150–1156 https://doi.org/10.1017/ S1368980008004345 70. Azam, J.-P., & Gubert, F. (2006). Migrants’ Remittances and the Household in Africa: A Review of Evidence. Journal of African Economies, 15(Supplement 2),426–462 https:// doi.org/10.1093/jae/ejl030 71. Booysen, F., van der Berg, S., Burger, R., Maltitz, M. von, & Rand, G. du. (2008). Using an Asset Index to Assess Trends in Poverty in Seven Sub-Saharan African Countries. JOURNAL OF HUNGER & ENVIRONMENTAL NUTRITION 645 https://doi.org/10.1007/978-1-4614-0517-7%5F22 https://doi.org/10.2105/AJPH.2006.093641 https://doi.org/10.1111/j.15410064.2012.00428.x https://doi.org/10.1111/j.15410064.2012.00428.x https://doi.org/10.1080/00220380500405394 https://doi.org/10.1007/s13524-011-0077-5 https://doi.org/10.1186/1742-7622-4-4 https://doi.org/10.5116/ijme.4dfb.8dfd https://doi.org/10.1186/1751-0473-3-17 https://doi.org/10.1093/oxfordjournals.aje.a115101 https://doi.org/10.2307/2286904 https://doi.org/10.1016/J.GLOENVCHA.2004.11.003 https://doi.org/10.1016/j.jand.2012.08.031 https://doi.org/10.1177/156482651003100402 https://doi.org/10.1017/S1368980008004345 https://doi.org/10.1017/S1368980008004345 https://doi.org/10.1093/jae/ejl030 https://doi.org/10.1093/jae/ejl030 World Development, 36(6),1113–1130 https://doi.org/10.1016/J.WORLDDEV.2007.10. 008 72. Wamani, H., Tylleskär, T., Astrøm, A. N., Tumwine, J. K., & Peterson, S. (2004). Mothers’ education but not fathers’ education, household assets or land ownership is the best predictor of child health inequalities in rural Uganda. International Journal for Equity in Health, 3(1),9. https://doi.org/10.1186/1475-9276-3-9 73. Reilly, J. M., & Schimmelpfennig, D. (1999). Agricultural Impact Assessment, Vulnerability and the Scope for Adaptation. Climatic Change, 43, 745–788. https://s3. amazonaws.com/academia.edu.documents/40856086/Reilly_Ag_impact_assess_vul_ Climatic_Change_1999.pdf?response-content-disposition=inline%3Bfilename% 3DAgricultural_Impact_Assessment_Vulnerabi.pdf&X-Amz-Algorithm=AWS4- HMAC-SHA256&X-Amz-Credentia 74. Tusiime, H. A., Renard, R., & Smets, L. (2013). Food aid and household food security in a conflict situation: Empirical evidence from Northern Uganda. Food Policy, 43,14–22 https://doi.org/10.1016/J.FOODPOL.2013.07.005 75. Rosegrant MW, Cline SA. Global Food Security: Challenges and Policies. Science. 2003; 302(5652):1917–9. doi:10.1126/science.1092958 76. Mwakatobe, A., Nyahongo, J., Ntalwila, J., & Røskaft, E. (2014). The impact of crop raiding by wild animals in communities surrounding the Serengeti National Park, Tanzania. International Journal of Biodiversity & Conservation, 6(9),637–646. https:// doi.org/10.5897/IJBC2014.0753 77. Matheson, J., & McIntyre, L. (2014). Women respondents report higher household food insecurity than do men in similar Canadian households. Public Health Nutrition, 17 (1),40–48. https://doi.org/10.1017/S136898001300116X 78. Sclove SL. Application of model-selection criteria to some problems in multivariate analysis. psychometrika. 1987;52(3):333–343. doi:10.1007/BF02294360. 79. Deitchler M, Ballard T, Swindale A, Coates J, Validation of a Measure of Household Hunger for Cross-cultural Use. Washington, DC: Food and Nutrition Technical Assistance (FANTA), USAID; 2010. 80. Quisumbing AR. Generating evidence on individuals’ experience of food insecurity and vulnerability. Global Food Security. 2013;2(1):50–55. doi:10.1016/J.GFS.2012.11.004. 81. Arcaya MC, Tucker-Seeley RD, Kim R, Schnake-Mahl A, Marvin S, Subramanian SV. Research on neighborhood effects on health in the united states: a systematic review of study characteristics. Soc Sci Med. 2016;168(November):16–29. doi:10.1016/j. socscimed.2016.08.047. 82. Roux AVD, Mair C, Neighborhoods and health. Ann N Y Acad Sci. 2010;1186. doi:10.1111/j.1749-6632.2009.05333.x 83. Kumanyika S, Grier S, Targeting interventions for ethnic minority and low-income populations. Future Child. 2006;16(1):187–207. doi:10.1353/foc.2006.0005 84. Pickett KE, Pearl M. Multilevel analyses of neighbourhood socioeconomic context and health outcomes: a critical review. J Epidemiol Comm Health. 2001;55(2):111–122. doi:10.1136/jech.55.2.111. 85. Obayelu, A. E. 2012 Households’ food security status and its determinants in the North- Central Nigeria. Food Economics. 9(4):241–256. 86. Guo, Y., Berrang-Ford, L., Ford, J., Lardeau, M. P., Edge, V., Patterson, K., . . . & Harper, S. L. 2015 Seasonal prevalence and determinants of food insecurity in Iqaluit, Nunavut. International journal of circumpolar health, 74(1), 27284. 87. D’andreamatteo, C., & Slater, J. 2017 Measuring food security in Canadian homeless adult men. Canadian Journal of Dietetic Practice and Research, 79(1),42–45. 646 K. PATTERSON ET AL. https://doi.org/10.1016/J.WORLDDEV.2007.10.008 https://doi.org/10.1016/J.WORLDDEV.2007.10.008 https://doi.org/10.1186/1475-9276-3-9 https://s3.amazonaws.com/academia.edu.documents/40856086/Reilly_Ag_impact_assess_vul_Climatic_Change_1999.pdf?response-content-disposition=inline%3Bfilename%3DAgricultural_Impact_Assessment_Vulnerabi.pdf%26X-Amz-Algorithm=AWS4-HMAC-SHA256%26X-Amz-Credentia https://s3.amazonaws.com/academia.edu.documents/40856086/Reilly_Ag_impact_assess_vul_Climatic_Change_1999.pdf?response-content-disposition=inline%3Bfilename%3DAgricultural_Impact_Assessment_Vulnerabi.pdf%26X-Amz-Algorithm=AWS4-HMAC-SHA256%26X-Amz-Credentia https://s3.amazonaws.com/academia.edu.documents/40856086/Reilly_Ag_impact_assess_vul_Climatic_Change_1999.pdf?response-content-disposition=inline%3Bfilename%3DAgricultural_Impact_Assessment_Vulnerabi.pdf%26X-Amz-Algorithm=AWS4-HMAC-SHA256%26X-Amz-Credentia https://s3.amazonaws.com/academia.edu.documents/40856086/Reilly_Ag_impact_assess_vul_Climatic_Change_1999.pdf?response-content-disposition=inline%3Bfilename%3DAgricultural_Impact_Assessment_Vulnerabi.pdf%26X-Amz-Algorithm=AWS4-HMAC-SHA256%26X-Amz-Credentia https://s3.amazonaws.com/academia.edu.documents/40856086/Reilly_Ag_impact_assess_vul_Climatic_Change_1999.pdf?response-content-disposition=inline%3Bfilename%3DAgricultural_Impact_Assessment_Vulnerabi.pdf%26X-Amz-Algorithm=AWS4-HMAC-SHA256%26X-Amz-Credentia https://doi.org/10.1016/J.FOODPOL.2013.07.005 https://doi.org/10.1126/science.1092958 https://doi.org/10.5897/IJBC2014.0753 https://doi.org/10.5897/IJBC2014.0753 https://doi.org/10.1017/S136898001300116X https://doi.org/10.1007/BF02294360 https://doi.org/10.1016/J.GFS.2012.11.004 https://doi.org/10.1016/j.socscimed.2016.08.047 https://doi.org/10.1016/j.socscimed.2016.08.047 https://doi.org/10.1111/j.1749-6632.2009.05333.x https://doi.org/10.1353/foc.2006.0005 https://doi.org/10.1136/jech.55.2.111 88. Vahabi, M., Damba, C., Rocha, C., & Montoya, E. C. 2011 Food insecurity among Latin American recent immigrants in Toronto. Journal of Immigrant and Minority Health, 13 (5), 929–939. 89. Busch J, Berrang-Ford L, Clark S, et al., Is the effect of precipitation on acute gastro intestinal illness in southwestern uganda different between indigenous and non-indi genous communities? PLoS One. 2019;14(5):e0214116.IHACC team. doi:10.1371/ journal.pone.0214116. 90. MacVicar S, Berrang-Ford L, Harper S, Huang Y, Bambaiha DN, Yang S, Whether weather matters: evidence of association between in utero meteorological exposures and foetal growth among Indigenous and non-indigenous mothers in rural Uganda. PLoS One. 2017;12(6):e0179010. doi:10.1371/journal.pone.0179010 91. MacVicar S, Berrang-Ford L, Harper S, et al. How seasonality and weather affect perinatal health: comparing the experiences of indigenous and non-indigenous mothers in Kanungu district, Uganda. Soc Sci Med. 2017;187(August):39–48. doi:.10.1016/J. SOCSCIMED.2017.06.021 92. Mukasa N, The batwa indigenous people of uganda and their traditional forest land: eviction, non-collaboration and unfulfilled needs. indigen policy j. 2014;24:4. 93. Kidd C. Inventing the ‘pygmy’: representing the ‘other’, presenting the ‘self.’. History Anthropol. 2009;20(4):395–418. doi:10.1080/02757200903216387. 94. Altman M, Hart TGB, Jacobs PT. Household food security status in South Africa. agrekon. 2009;48(4):345–361. doi:10.1080/03031853.2009.9523831. 95. Amaza PS, Adejobi AO, Fregene T, Rastilantie M. Measurement and determinants of food insecurity in northeast nigeria: some empirical policy guidelines. J Food Agric Environ. 2005;6:92–96. 96. Bogale A. Vulnerability of smallholder rural households to food insecurity in Eastern Ethiopia. Food Security. 2012;4(4):581–591. doi:10.1007/s12571-012-0208-x. 97. Maharjan KL, Joshi NP. Determinants of household food security in nepal: a binary logistic regression analysis. J Mount Sci. 2011;8(3):403–413. doi:10.1007/s11629-011- 2001-2. 98. Hadley C, Lindstrom D, Tessema F, Belachew T. Gender bias in the food insecurity experience of ethiopian adolescents. Soc Sci Med. 2008;66(2):427–438. doi:10.1016/j. socscimed.2007.08.025. 99. McIntyre LN, Theresa Glanville KD, Raine JB, Dayle BA, Battaglia N, Do low-income lone mothers compromise their nutrition to feed their children? CMAJ. 2003;168:6. 100. Antwi-Agyei P, Dougill AJ, Fraser EDG, Stringer LC. Characterising the nature of household vulnerability to climate variability: empirical evidence from two regions of ghana. Environ Develop Sustainable. 2013;15(4):903–926. doi:10.1007/s10668-012- 9418-9. 101. Demeke AB, Keil A, Zeller M. Using panel data to estimate the effect of rainfall shocks on smallholders food security and vulnerability in rural Ethiopia. Climatic change. 2011;108 (1–2):185–206. doi:10.1007/s10584-010-9994-3. 102. Échevin D. Measuring vulnerability to asset-poverty in Sub-Saharan Africa. World dev. 2013;46(June):211–222. doi:10.1016/J.WORLDDEV.2013.02.001. 103. UBOS & ICF, and UBOS. 2017. “Uganda Demographic and Health Survey 2016: Key Indicators Report.“ Kampala, Uganda: UBOS, and Rockville, Maryland, USA: UBOS and ICF.” 104. Beyene T. Factors for implementing public-private partnership (ppp) in the develop ment process : stakeholders ’ perspective from Ethiopia. Int j sci res. 2014;3:792–797. JOURNAL OF HUNGER & ENVIRONMENTAL NUTRITION 647 https://doi.org/10.1371/journal.pone.0214116 https://doi.org/10.1371/journal.pone.0214116 https://doi.org/10.1371/journal.pone.0179010 https://doi.org/10.1016/J.SOCSCIMED.2017.06.021 https://doi.org/10.1016/J.SOCSCIMED.2017.06.021 https://doi.org/10.1080/02757200903216387 https://doi.org/10.1080/03031853.2009.9523831 https://doi.org/10.1007/s12571-012-0208-x https://doi.org/10.1007/s11629-011-2001-2 https://doi.org/10.1007/s11629-011-2001-2 https://doi.org/10.1016/j.socscimed.2007.08.025 https://doi.org/10.1016/j.socscimed.2007.08.025 https://doi.org/10.1007/s10668-012-9418-9 https://doi.org/10.1007/s10668-012-9418-9 https://doi.org/10.1007/s10584-010-9994-3 https://doi.org/10.1016/J.WORLDDEV.2013.02.001 105. Holden S, Shiferaw B. Land degradation, drought and food security in a less-favoured area in the ethiopian highlands: a bio-economic model with market imperfections. Agric econ. 2004;30(1):31–49. doi:10.1016/J.AGECON.2002.09.001. 106. Hartter J. Attitudes of rural communities toward wetlands and forest fragments around kibale national park, Uganda. hum dimensions wildlife. 2009;14(6):433–447. doi:10.1080/10871200902911834. 107. Hill CM. Conflict of interest between people and baboons: crop raiding in Uganda. Int j primatol. 2000;21(2):299–315. doi:10.1023/A.1005481605637. 108. Perkins JM, Subramanian SV, Christakis NA. Social networks and health: a systematic review of sociocentric network studies in low- and middle-income countries. Soc sci med. 2015;125(January):60–78. doi:10.1016/j.socscimed.2014.08.019. 109. Egeru A. Role of indigenous knowledge in climate change adaptation: a case study of the teso sub-region, Eastern Uganda. Ind j trad knowl. 2012;11:217–224. 110. Beaumier MC, Ford JD. Food insecurity among inuit women exacerbated by socio economic stresses and climate change. Sour canad j public health/revue canadienne desanté publique canadian public health association canadian journal of public health. 2010;101:196–201. 111. Osbahr H, Dorward P, Stern R, Cooper S. Supporting agricultural innovation in Uganda to respond to climate risk: linking climate change and variability with farmer percep tions. experiment agric. 2011;47(2):293–316. doi:10.1017/S0014479710000785. 112. Tadesse G, Abate GT, Zewdie T. Biases in self-reported food insecurity measurement: a list experiment approach. food policy. 2020;92(April):101862. doi:10.1016/j. foodpol.2020.101862. 113. Valentine G, Theorizing and researching intersectionality: a challenge for feminist geography*. pro geogr. 2006;59:10–21. 114. Kitchin R, Tate N. Conducting research in human geography: theory, methodology and Practice - rob kitchin, nick tate - Google books. New York: Routledge; 2013. 115. Winchester HPM. Ethical issues in interviewing as a research method in human geo graphy. Aust geograph. 1996;27(1):117–131. doi:10.1080/00049189608703161 116. Baxter J, Eyles J. Evaluating qualitative research in social geography: establishing ‘rigour’ in interview analysis. trans inst br geogr. 1997;22(4):505–525. doi:10.1111/j.0020- 2754.1997.00505.x. 648 K. PATTERSON ET AL. https://doi.org/10.1016/J.AGECON.2002.09.001 https://doi.org/10.1080/10871200902911834 https://doi.org/10.1023/A.1005481605637 https://doi.org/10.1016/j.socscimed.2014.08.019 https://doi.org/10.1017/S0014479710000785 https://doi.org/10.1016/j.foodpol.2020.101862 https://doi.org/10.1016/j.foodpol.2020.101862 https://doi.org/10.1080/00049189608703161 https://doi.org/10.1111/j.0020-2754.1997.00505.x https://doi.org/10.1111/j.0020-2754.1997.00505.x Abstract Background Methods Study Population Analytical Framework Data Collection Food Insecurity Score and the Adapted Vulnerable Populations Score (AVPS) Independent Variables Analysis Results Batwa Demographics Prevalence of Food Security among Batwa Households Variables Associated with Batwa Food Insecurity Discussion Limitations Conclusion Funding ORCID References