Assessing dietary adequacy and temporal variability in the context of Covid-19 among Indigenous and rural communities in Kanungu District, Uganda: a mixed-methods study

Dietary adequacy is hard to achieve for many people living in low-income countries, who suffer from nutritional deficiencies. Climate change, which alters weather conditions, has combined with other cascading and compound events to disrupt Indigenous communities’ food systems, limiting the consumption of adequate diets. The aim of this work was to conduct a proof-of-concept study exploring dietary adequacy, and to investigate evidence for temporal variation in the dietary intake of Indigenous and non-Indigenous communities in Kanungu District, Uganda in the context of the Covid-19 outbreak. We randomly selected 60 participants (20 mothers, 20 fathers and 20 children aged between 6 and 23 months) from two Indigenous Batwa and two Bakiga settlements. A mixed-methods study with concurrent qualitative and quantitative data collection was conducted. Monthly dietary intake data were collected from each participant from February to July 2021 through 24 h recall surveys using a specially developed Ugandan food composition database included in the online tool myfood24. At the same time, we also collected: (i) demographic and contextual data related to Covid-19; (ii) data on weather and seasonality; (iii) data on the perception of dietary intake over the year, and during the Covid-19 period; (iv) baseline anthropometric measurements. The majority of the participants did not achieve nutrient adequacy over the 6 months period, and household dietary diversity scores were generally low. Pregnant and lactating women consumed a diet which was severely inadequate in terms of nutrient consumption. Caloric and nutrient intake varied over the 6 months period, with the highest food consumption in June and lowest in April. Temporal variation was more evident among Batwa participants. Vitamin A intake varied more over months than other nutrients in adults’ and children’s diets, and none met iodine requirements. Participants characterised the diverse mechanisms by which season and weather variability determined the type and amount of food consumed each month. Dietary intake showed indications of temporal variation that differed between nutrients. Also, they reported that the Covid-19 pandemic influenced their diet. During lockdown, 58% of adults reported changing dietary habits by consuming less—and less nutritious—foods. The findings of this work highlight that the majority of the Batwa and Bakiga participants did not meet the dietary requirements for their age and gender. Also, our research indicates that weather patterns and seasonality may cause variations in smallholder food production with consequences on households’ dietary intake. Emerging evidence suggests that nutrients and caloric intake vary monthly and under different weather conditions. Accurate and time-varying nutrition evaluations would help in identifying seasonal and monthly dietary needs, supporting preventive interventions protecting children and their parents from any form of malnutrition. Consideration of time-varying nutritional intake will become increasingly important as climate change affects the seasonality and availability of smallholder subsistence crops.

. Illustration indicating the four settlements where the study was conducted. The grey area represents Kanungu District where both Batwa and Bakiga communities live. the data collection tools, research methods, anthropometric evaluation and ethical dilemmas concerning research and research participants was conducted online and on-site prior to the start of the study (January 2021). We discussed the ethical implications of studying food insecurity in impoverished communities and followed international ethics guidelines throughout the study [36]. The exploration of food and nutrition security is considered to be vital for the Batwa and Bakiga communities, and their wellbeing, and for this reason a proof-of-concept study was conducted. The sample size was set by feasibility, and we used standard sampling techniques. The study was designed by imposing the absolute minimum of additional risk to the participants, and following the highest standards for obtaining informed consent, and ensuring confidentiality. Theoretical and practical sessions were delivered over one week of training, which were coordinated on the ground by Sabastian Twesigomwe (ST), and supported remotely by Giulia Scarpa.

24 h recall survey
A 24 h recall survey investigating the type and quantity of food consumed the previous day was collected once per month over 6 months (from February to July 2021) for each participant (n = 60 for 6 months-total of 360 surveys) (Supplementary material). The mothers answered dietary questions for their young children. The leader of the local research team conducted a sampling procedure with the local chairperson, and mobilised households each month. The first time the survey was undertaken, the local researcher explained the type of information needed for the study to the adult participants. Plates, cups, spoons and ladles, which were commonly used by the two communities, were provided before the 24 h recall for each household to facilitate and standardise the measurement of portion sizes across the participants. To measure the portion sizes, first we collected information on the most common tools used to quantify edible food. For example, participants reported that 'plastic cups' (500 ml) were used to measure the quantity of milk or water. Afterwards, two local researchers weighted the quantity of each food contained in the correspondent household tool, for example, a ladle contained 100 g of cooked rice. These measurements were used as a reference to calculate the dietary intake throughout the 24 h recall method.
The 24 h recall survey was recorded with participant consent, and notes were taken by the research team during the interview. The information was later entered into an online nutritional assessment programme, myfood24. myfood24 included a community-specific food composition database, which was previously developed to include the most commonly consumed foods and dishes eaten by the Batwa and Bakiga [37]. It was not possible to record the data on the online database during the fieldwork given the remoteness of the area and related Wi-Fi constraints. More than 20% of the dietary data were assessed for quality-control by a local nutritionist to make sure they were appropriately collected and reported.

Anthropometric measurements
Weight, height, presence of bilateral pitting oedema and middle upper arm circumference (MUAC) were performed using the procedures recommended by World Health Organisation (WHO) [38]. The measurements (n = 60) were collected at the beginning of the study to have baseline anthropometric information. The local nutritionist was in charge of the MUAC, weight and height data collection together with a member of the local research team. We decided to collect both MUAC and weight for height (WHZ) measurements to classify wasting, including severe wasting, in order to avoid missing wasting cases among the Batwa and Bakiga communities. In fact, the majority of children with severe wasting have either severely low WHZ or severely low MUAC [39,40]. However, the degree of overlap between these two measurements is approximately of 40%, and differs by country [41]. The data were reported on paper copies. Two trained researchers with background in nutrition collected the anthropometric measurements separately. After collecting the data, an expert nutritionist on the field compared their measurements to ensure that the differences between their measurements was within the maximum allowed differences (7 mm for length and 50 g for weight). A third measurement was taken if any measurements exceeded the maximum allowed differences. Standardisation sessions were part of the training. During training, the researchers were familiarised with the measuring equipment and techniques. We assessed accuracy and precision by calculating any difference between the measurements of the nutritional expert (considered 'gold standard') and the other researchers while measuring anthropometry of a group of children. During the training, the measurements taken by the expert and the other researchers did not show any systematic bias or overestimation/underestimation of the measures taken.
For this study, we calculated the body mass index (BMI) for adults and classified individuals into categories of underweight (BMI < 18.5), normal weight (BMI 18.5-24.9), overweight (25-29.9) and obese (BMI > 30) [42]. For children under 2 years, we calculated WHZ, weight for age (WAZ) and height for age (HAZ) Z score to check for wasting, underweight and stunting respectively, using the WHO 2006 growth standard as the reference population [43]. Participants who presented bilateral oedema or severe malnutrition were referred to the Bwindi Hospital in Kanungu District.

Socio-demographic & Covid-19 information
During the first set of community visits, we collected current socio-demographic information for the participants. For the adults, we included sex, age, education level, occupation, food access and production (including crops and animals), access to water and sanitation, mosquito net usage, health status, smoking and alcohol use, pregnancy status, access to health facilities, household level information about family size, number of children, and other proxy data for wealth status. For the children, we included data on child sex, place of birth, pre-and post-natal visits, breastfeeding status, general breastfeeding habits and issues, and introduction of complementary feeding.
Additionally, we collected data on household food security during the Covid-19 period using structured and semi-structured questions. We investigated individuals' perception of change in diet and in food accessibility, availability, and consumption from March 2020. Other questions explored the role of humanitarian organisations providing foods to the Indigenous communities during the pandemic. The data were recorded on paper copies by the research team.

Weather and season information
The leader of the local research team ST collected information from internet broadcasting radio on daily weather conditions, extreme climatic events, and temperature (on the day of the fieldwork) over 6 months, and reported on paper copies using monthly calendars for the four selected settlements. Additionally, information on traditional seasons and changes in seasonality in the year 2021, monthly agricultural practices and availability of the main crops in the gardens, markets and shops were collected during the interviews.

Qualitative interviews
Qualitative interviews (n = 40 adults) were conducted at the beginning of the fieldwork with the same participants who were part of the 24 h recall. Open questions explored: (i) food availability and access over months and in different seasons, (ii) perception of diet and food nutrition quality change over months, and (iii) perception of changes in nutritional status among the Batwa and Bakiga communities over the year.
Despite a small sample size, we reached data saturation during qualitative interviews. The interviews lasted on average 24 min. The team leader of the local research team ST reviewed all qualitative interviews for quality-control. The interviews were audio-recorded and translated from Rukiga to English language by a proficient speaker ST on the research team.

Data analysis
The socio-demographic data and the information related to Covid-19 responses were statistically analysed using Python 3.8.
For the nutrient and caloric intake assessment, we selected specific macro and micronutrients which are essential for children's growth and for lactating and pregnant women [44], including protein, iron, zinc, iodine, folate, vitamins A and C. Due to missing data in the food composition database, we could not include information on other B vitamins. We drew time-series graphs and used boxplots to explore these nutrient and caloric changes over months and seasons. Mean values for the month with the highest and lower nutrient intake were reported for each macro/micronutrient and calories. We then calculated percentages of nutrient variation over months, with confidence intervals to assess the difference of nutrients and calories in the best and worst months for Batwa and Bakiga women, men, and children. We also calculated household dietary diversity scores (HDDS) by summing the number of food groups eaten by each household every month. The total possible number of food groups is 12, and includes: (i) cereals; (ii) roots and tubers; (iii) vegetables; (iv) fruits; (v) meat, poultry and offal; (vi) eggs; (vii) fish and seafood; (viii) pulses, legumes and nuts; (ix) milk and milk products; (x) oils and fats; (xi) sugar and honey; and (xii) miscellaneous foods, including condiments, coffee, salt and spices [45]. Finally, we evaluated individual nutrient and caloric adequacy by comparing the values to the EAR recommended nutrient intake references (intake as % EAR) for protein, vitamin A, B12 and C, iodine, zinc, iron, calcium and folate [46]. Mean status was defined as adequate (⩾100% of EAR), inadequate (⩾50% to <100% of EAR) or severely inadequate (<50% of EAR).
Finally, we conducted a contextualised thematic analysis [47] using NVivo 12 using the qualitative data collected through interviews. We used the framework of Ford et al [48] to structure the questions and the analysis. We explored Batwa and Bakiga's lived experiences and perceptions on the impact of weather and climatic extremes, and Covid-19 on food security and nutrition. To do that, we investigated the interaction between 'slow' stressors, such as environmental change, socio-economic factors, and knowledge, and 'fast' stressors, such as extreme climatic events, changes to livelihood, food prices, and market access. The analysis involved multiple steps, from data familiarisation to the development of codes, codebook and themes, and revision of the themes [49]. It was performed by a single author, however 10% of the analysis was checked by a second researcher. The qualitative data were integrated with the quantitative results, and were used to identify key narratives about dietary patterns over the year and during Covid-19, and also to gather information about weather, seasonality, and monthly agricultural practices in Kanungu District.

Sample characteristics
The sample for the study included 60 participants; however, seven 24 h recalls out of 360 were missing as participants moved from the community due to work or other family circumstances.
Nearly 80% of the adults, including women and men, had a BMI classified as normal weight, between 18.5 and 24.9 kg m −2 . Five (13%) adults were underweighted (BMI 16-18.5), all of whom were Batwa, and four (10%) were overweight (BMI 25-28), all of whom were Batwa as well. More than half of the children (n = 12, 60%) presented a form of malnutrition (underweight, wasting or stunting), particularly among the Batwa, who were also more affected by stunting (n = 6, 60%) compared to the Bakiga (n = 2, 20%). Two children were classified as wasted, one (5%) of them identified through MUAC and bilateral pitting oedema checks, and the other one (5%) with the WHZ calculation. The majority of the children (n = 16, 80%) were regularly breastfed (more times during the day), and 65% (n = 13) of mothers reported that they started complementary feeding at 6 months of age.
Most of the adults were employed, working as subsistence farmers (n = 26, 65%). Most of the food consumed by the participants was produced in their gardens (n = 28, 70%), and half reported owning at least one animal (e.g. chicken, pig, cow, or guinea pig) (n = 20, 50%).
Most families lived in impoverished conditions with limited access to basic public services: the source of drinking water was generally unprotected (n = 6, 30%), toilet facilities were uncovered pit latrines (n = 18, 90%) and shared (n = 11, 55%), electricity was not available (n = 17, 85%). Only seven families out of 20 (35%) treated the water before using it. Fifty percent (n = 10) of the households had more than three children and five or more family members (table 1).

Dietary intake and adequacy
Caloric intake was adequate for Batwa and Bakiga children, but inadequate for all Batwa and Bakiga adults, and especially low for lactating mothers (table 2). Children's calories were derived primarily from carbohydrates (77%), with 10.5% coming from proteins, and 13.5% coming from fats, with no differences among the two communities. Seventy-six percent of adults' calories came from carbohydrates, 10% from proteins and 13% from fats, and this was consistent among both communities.
Most of the nutrient requirements of lactating women were classified as 'inadequate' (nutrient value ⩾50% to <100% of EAR) or 'severely inadequate' (nutrient value <50% of EAR), especially for Batwa women. All children, women and men had vitamin A, iodine and calcium intakes that were inadequate or severely inadequate, and Batwa lactating women had also a very low iron intake.
Generally, more Bakiga women met the caloric and nutrient requirements compared to Batwa women (appendix A), Caloric intake, vitamin A, zinc and calcium were not met by the majority of breastfeeding and pregnant women across months. None met the iodine requirement, and only one woman met the folate requirement. Overall, Bakiga men were more likely to meet the nutrient and caloric requirements than the Batwa men over the 6 months-period. Although iron intake was met each month by less than 50% of men, generally they achieved iron requirements more frequently than women and children. A similar pattern can be seen for zinc intake, which was adequate for more male participants than for women and children. Only a few men achieved vitamin A and iodine requirements. Protein intake was most frequently met by children: every month at least 50% of children met the requirements. Vitamin A was overall inadequate among most of the children (adequacy percentage per month from 5% to 15%). No one met the iodine requirement in the 6 months period.

Caloric and nutrient intake trends over seasons and under different weather patterns
Caloric and nutrient intake showed temporal variation over the 6 months, which spanned wet and dry seasons corresponding to the planting and harvesting periods. Also, crop availability varied through the same period (table 3).
Temporal variation can be seen in caloric intake in both Batwa and Bakiga diets, however in adult Batwa participants the fluctuation was modestly more accentuated. For the adult Batwa community, a drop in calories was seen in the month of April (rainy season, end of planting season), and the highest caloric intake in June (harvesting season). The diet of adult Bakiga varied less seasonally than for the Batwa, with slightly higher caloric consumption in May (beginning of harvesting season). Children's caloric intake had a different pattern, showing less variation in the first 4 months, but with a peak in the month of June in both Batwa and Bakiga children.
Nutrient intake varied over the months of the study period in the diets of Batwa and Bakiga adults and children (appendix B). Each nutrient had its own pattern with peaks and drops in different months. For example, the consumption of calcium in adults was higher in June and lower in May, while folate intake had a peak in May and dropped in June. Nutritional intake of calcium, for example, varied more than 50% in the 6 months period for women, men and children in both communities, with overall higher difference among women and children. Vitamin A was the nutrient which varied more across months in adults' and children's diets, although for children the variation was slightly higher, up to 97% (C.I. 96-99) among Batwa children, and 94% (C.I. 92-96) among Bakiga children. These confidence intervals did not overlap despite the small sample size, pointing to a significant difference in variation of vitamin A over seasons between Batwa and Bakiga children, with variation greater among Batwa children. The differences in carbohydrate intake for all participants was lower compared to the other nutrients over the 6 months period: around 60%-65% for Batwa and Bakiga adults, except for Bakiga children for whom the variation was higher (82%, 95% CI 73-90). Iron intake (appendix C) varied a minimum of 62% among Batwa men to maximum 84% among Bakiga women across the 6 months.

Household dietary diversity scores
The mean HDDS ranged between 4.4-5.5 (out of 12) over the 6 months, with no statistical difference between Batwa and Bakiga's households. The HDDS was, in fact, stable through the period of the study with no significant variation over months and seasons. The participants consumed in average 4-5 food groups per day, which is considered a moderate diverse diet. Fruits and vegetables (constituting between 30% and 43% of their diet), and cereals and tubers (constituting between 28% and 38% of their diet) were the most consumed food groups, and dairy and eggs (constituting between 0% and 1% of their diet) the least eaten by the participants over the 6 months. Additionally, all animal proteins coming from meat, poultry and fish (constituting between 0.5% and 3% of their diet) were rarely consumed by the communities (appendix D).

Perceivedand experienced effects of Covid-19 and climatic changes on diets
During the interviews, participants discussed the interacting factors affecting their dietary intake in the first 6 months of 2021.

Environmental change and seasonal shifts
Participants characterised weather, food availability, and consumption for the 6 months of fieldwork and compared this to a 'typical' seasonal calendar. They reported that two of the 6 months during the study period differed from typical expectations in terms of weather and food security: March was expected to be rainier and May sunnier, and this delayed the planting season, decreasing in some cases crop production. In fact, February was typically the time to prepare the field for the agricultural season, and March and April the time to plant. These 2 months corresponded also to a 'critical time' in terms of food availability and accessibility (table 3). Women and men also commented on the quality of food over the months. They reported that in the dry season the type of food consumed was less nutritious, more expensive and less affordable, while in the rainy season the quality was higher: In dry months, when there is a lot of sunshine, there are no greens (which are a mix of vegetables, such as amaranth, also called 'dodo'), which are very nutritious. During the rainy season, greens are available-there are greens in the bush. We do not plant them, but we find and collect them in the bush directly.
Cassava and dipuri (poor quality matoke, plantain species) were the only two 'fresh' foods available during food insecure and dry periods according to the Batwa and Bakiga participants. Indeed, cassava 'could easily grow under any weather condition' and without the need to purchase seeds as 'planting the stem was sufficient' to grow this crop. However, participants agreed that both foods were not 'nutritious': Health workers say that dipuri and cassava are not nutritious, and give suggestions for a balanced diet but they ( Participants added that during the planting season food was usually dried, for example dried beans and maize, and kept in stores. Conversely, in the harvesting season corresponding to May and June, 'the best' period for food quantity and quality according to the participants, people consumed fresh foods (e.g. fresh beans, fresh maize). People felt generally healthy in this season, full of energy to work, and satisfied with the food that is on average consumed over two meals. Some participants explained, however, that food was not enough for their families: During the rainy season there are more foods, but this does not mean that is enough for all the Batwa, but it is certainly more available.

Extreme climatic events
No extreme weather events occurred in the four settlements over the 6 months period. However, participants did report previous experiences with extreme events: when floods or droughts occurred, the food security level was drastically reduced, and the price of foods increased for both Batwa and Bakiga communities:

Food is expensive when droughts or floods occur. […] I usually eat different types of food and I think that the food I consume is nutritious, but not when there is an extreme climatic event as everyone is in crisis of food.
However, some participants who reported no variation in their diet over months explained that there were circumstances in which a change occurred, for example in case of an extreme weather event: I rarely change my diet, but this happens with long drought or floods. Sometimes I need to go to markets as crops do not grow under some climatic changes. Lots of wind destroys banana plantations […]. If coffee trees fall down, we do not have any income.

Socio-economic factors
Others, however, explained that their diet did not change because of the 'permanent poverty condition' , explaining that usually they cannot afford diverse food as they do not have enough money, or thanks to economic stability and availability of lands, which ensured enough food for the family:

No change in my diet-most of the time I have food at home as I have coffee plantations that give me enough money to eat well.
Additionally, 75% of the adult participants reported changing their diet since March 2020 (appendix E). Most of them reported having consumed food in lower quantities and less frequently due to lack of money, but also due to access restrictions for markets and shops, and lack of availability of work. However, two participants reported to have eaten more than usual to protect themselves from Covid-19, and given that their social life was reduced drastically: I eat more since Covid started because we have more time to stay at home due to the lockdown. (Bakiga man, Interview) During the adult interviews, 17 of 20 Batwa and 12 of 20 Bakiga participants reported that their diet usually changed over the year, due to 'climate change' and 'famine' , but also depending on the economic situation of the family, market access and availability of work:

Culture and knowledge
Adults perceived that some foods had higher nutritional properties, while others were consumed only to'not die' from hunger. Generally, beans, groundnuts, meat and matoke were thought to be 'good' foods. However, participants had different opinions about the 'most nutritious' type of food. Some of them called 'good food' food that was fried and rich in fat:

When I eat beans fried with cooking oil or meat with matoke, I get satisfied. In fact, when I eat any food that is fried with oil, I know that I am eating very well. (Batwa man, Interview)
Others explained that greens and protein-rich foods made people healthy. Also, they commonly felt that 'good taste' meant 'having many nutrients': Foods less nutritious are cassava and dipuri because they do not taste very good. I do not like them, but I eat them because I have no options.

Food and market availability and accessibility
Also, more than half of the participants (58%) reported that some foods previously consumed were not available during the Covid-19 period (appendix E). The consumption of certain types of foods such as fish, meat, matoke, rice, posho, and Irish potatoes decreased due to high costs at the market. Nearly 68% of the participants had limited or no access to markets and nearly 88% no access to shops due to the pandemic restrictions. In 2020, however, all Batwa participants received help at least once from humanitarian organisations, which provided food, but this did not happen among the Bakiga community.
A participant also mentioned that during the lockdown the cases of malnutrition increased due to lack of food: In February [2021] people became thinner because there was no food due to Covid-19, which made our situation worse.

Malnutrition as consequence of the interaction between climatic changes, Covid-19 and other interacting factors
Participants reported that the number of malnutrition cases in the communities varied over the year with a peak during the planting season when food was scarce, and individuals were more likely to get sick: People are thinner whenever entering in a new season […], and especially during the planting season. Also, because malaria comes with the rainy season, people get sick and end up malnourished. Flu also makes people thinner. When food is not there, people get malnourished.
Malnutrition was perceived to be exacerbated by extreme climatic events, which affected Batwa more than the Bakiga, and in particular hit breastfeeding mothers as they reported that due to inadequate food, they did not have the strength to breastfeed. The occurrence of flooding and droughts posed additional vulnerability to already vulnerable individuals: With extreme events, Batwa suffer more. There is no food also for Bakiga, but Batwa suffer more because there is no labour for Batwa when there is no work for Bakiga. However, Bakiga can sell something like animals, or they could have saved money before the droughts or floods destroy everything.
However, participants also reported that other compound events contributed to food and nutrition insecurity and increased the number of malnutrition cases, including alcohol abuse, family conflict, and scarcity of lands that according to them were consequences of poverty.

Discussion
Through this proof-of concept study, we assessed the caloric and nutrient intake of 20 Batwa and Bakiga households in south-western Uganda. In spite of the small sample size, we found that calories and nutrient intake varied over the 6 months period with a peak in the month of June and a drop in the month of April. The novelty of this work lies in the longitudinal nature of the study, and in the simultaneous collection and interpretation of qualitative and quantitative information on diet, environment and climatic factors in the context of Covid-19. The qualitative component of the study contributed insight on communities' perceptions of diet change over the year, and in the context of climate change and pandemics. Indeed, food-insecure households can be heavily hit by shocks and food deficits as a consequence of climate change [50]. This study contributes to providing useful data to identify nutritional needs of individuals vulnerable to malnutrition and infectious diseases in south-western Uganda.
The findings indicated that weather patterns and seasonality-and associated variation in smallholder food production-influence households' dietary intake. For example, the peak of caloric intake occurred in June given the higher food production during the harvesting season. This has been reported elsewhere in Kenya [51], and linked with low dietary diversity and seasonal weather extremes events in other Sub-Saharan countries [52]. During the dry season, participants reported that food prices in the market fluctuated, and that they experienced difficulties affording commonly consumed foods, a challenge also reported in other low-income and food-insecure countries [53].
For the majority of the participants, nutritional intake was inadequate overall compared to reference intakes. This trend has been found in many studies in low-income countries, which assessed dietary adequacy of mothers and children [54][55][56][57][58][59]. According to UNICEF data, children aged 6-11 months did not have a diverse diet in Sub-Saharan Africa [60]. Younger children (up to 12 months), who belonged mostly to the Bakiga community, had a nutrient intake lower than the Batwa. The diet of both Batwa and Bakiga participants was mainly composed of cereals, and was poor in animal proteins, which is common in Africa [61]. As reported in the interviews, families in non-harvesting seasons consumed a very limited number and type of foods, and this was reflected in the household dietary diversity low scoring. Although some participants stated that health workers promoted the consumption of diverse food, this was challenging for families to achieve, especially during the dry season.
The inadequacy of dietary intake is a major factor leading to stunting, wasting, undernutrition and nutrient deficiencies, which are especially high among Batwa young children according to our research and previous studies [33]. Deficiencies linked to dietary intake are preventable by providing individuals with sufficient and diverse foods. There are some essential nutrients which are particularly scarce in the Batwa and Bakiga diet. First, consumption of iron, which has many vital functions, such as transporting of oxygen to body tissues [62], was inadequate. Second, none of the study participants consumed sufficient iodine, which is involved in the thyroid hormone synthesis [62]. The consequences of extreme iodine deficiency are irreversible, causing brain damage especially in infants [3]. The literature highlights that Ugandan Districts bordering the Democratic Republic of Congo grapple with iodine deficiency [63]. Finally, the intake of vitamin A, which is an essential nutrient for growth and development, but also for the immune function, vision and reproduction [64], was also low. Vitamin A deficiency is common in populations consuming vitamin A from provitamin carotenoid sources (e.g. leafy green vegetables), such as in Uganda [64], and with a diet very poor in fat [65].
The participants described the cascading effects of climate change on weather and seasonal variability, and food availability, explaining that Covid-19 exacerbated food insecurity and malnutrition among both communities. Given economic instability in 2021, participants reported shifting their diet towards cheaper sources of calories, especially cereals and other foods poor in proteins and fats. This occurred in many countries during the lockdown as food prices increased [66]. There is evidence that during the pandemic a household's ability to meet dietary requirements decreased, while food insecurity increased together with the consumption of poor-quality diets, leading to increased risk of malnutrition [67]. Studies conducted in Uganda, Kenya and Nigeria, in line with our work, showed a deterioration of individual food security status in 2020-2021 [68][69][70]. As the Batwa and Bakiga participants reported, people consumed less food and more alcohol during the pandemic, hitting already food insecure communities hardest. This is likely to increase the rate of malnutrition in the post-Covid-19 era [71]. However, the effects of Covid-19 on the communities in south-western-Uganda are difficult to estimate due to a lack of longitudinal studies and pre-pandemic dietary information, which make it difficult to distinguish the consequences on diet due to seasonal variation and weather-driven variability vs. the pandemic and associated restrictions [72].

Limitations of the study
This study has some limitations. First, the sample size was adequate for a proof-of-concept to assess the feasibility of undertaking longitudinal dietary assessment in a remote and vulnerable population, but not to generalise the findings among the entire Batwa and Bakiga communities. Also, the study was conducted over 6 months only. This limited the assessment of diet over time across a larger population, and generalisation of the findings. However, the results are representative of the diet during a specific time, the Covid-19 outbreak, and provide critical information on how local populations responded to protect their food security, and nutrition in this specific period.
Second, the 24 h recall surveys were conducted only once a month, limiting the possibility to model possible scenarios of dietary intake in the future or to explore day-to-day variation. A year-long longitudinal study with multiple 24 h recall surveys per month, and a higher number of households involved would allow estimation of more accurate usual dietary intakes among the Batwa and Bakiga (through the probability of adequacy calculation applying BLUPs of usual nutrient intakes where appropriate). The 24 h recall approach has some limitations itself as a method given that it is subject to recall bias [73], and under and over-reporting [74]. Also, although we developed a food composition database for the population previously [37] and the information was collected by trained and knowledgeable local researchers, the estimation of portion sizes for the participants was difficult given that usually people eat from a common plate. New methods for the data collection may be proposed in the future to overcome this issue [75].
Third, anthropometric measurements were performed only at the beginning of the study due to budget constraints, and the technical error of measurement was not applied, therefore we were not able to assess the effect of seasonality and weather patterns on participants' nutritional status. This would have enriched our work to assess the link between diet and nutritional status, especially by assessing the children's growth curve. Also, this could have started the discussion about the relation between nutrient/caloric intake and the short stature typical of the Batwa population.
Fourth, there was no information about dietary intake during extreme climatic events. Although Kanungu District was recently hit by a devastating flood (2019), at the time when the data were collected no extreme events occurred. These data would have helped in assessing nutritional requirements in times of very low nutrients intake, when people are most in need.
Fifth, we could not evaluate any difference in dietary intake before and after the pandemic, given that no data on caloric and nutrient intake was available in the literature for these two communities. However, our study is the first one assessing dietary intake among these two food-insecure populations. (1). To adapt to climatic changes and weather variability in Kanungu District, Batwa women highlighted the need for more land to cultivate different crops which could help when food is very expensive and largely unavailable [29,76]. In line with this, adequate facilities to store crops from the garden or sell food in excess could reduce the economic gap [77]. Also, the promotion of Indigenous and wild foods still available in Bwindi National Park could aid in reducing food-related costs at this time of year. (2). Evidence has demonstrated that wild foods may be more resilient to climate change than other crops, and be a protective factor for highly food insecure families [78]. Additionally, researchers have reported the importance of consuming wild fruits and vegetables to improve maternal and child nutritional outcomes [79]. However, eviction and continued exclusion from the forest has almost completely eliminated access to wild foods according to the Batwa [76]. Further research is needed to investigate if wild foods follow a seasonal pattern to identify key food items over the year. Additionally, avoiding the exploitation of endangered animals species for food would be critical when implementing nutrition interventions. (3). To improve the dietary diversity and adequacy, together with nutrition and health messages, adaptive agricultural strategies which consider local climate trends and seasonal patterns, such as the use of plants resistant to droughts, could mitigate the risks of agriculture production losses, and ensure crops' growth to provide enough food to the communities every month [80].

Recommendations and further research
(4). To reduce micronutrient deficiencies, some recommendations need to be taken into account: (i) communication campaigns targeting young children and women of childbearing age have been successful in other low income countries to prevent anemia [81], and may be helpful for the Batwa and Bakiga communities. Indeed, during the complementary feeding period, the need for iron is particularly high to support infant growth, and difficult to meet in food insecure populations [3]; (ii) raising awareness on the importance of consuming food rich in iodine (for example fish), and promoting iodised salt campaigns would help in reducing iodine deficiency's cases among the Ugandan communities [82]. Also, the use of iodine-containing supplements may decrease the risk of iodine deficiencies for pregnant and breastfeeding women [83]; (iii) vitamin A supplementation is critical for the growth of Ugandan children, and improving coverage would help in reducing vitamin A deficiency [64]. (5). Knowledge of the nutrient composition of Indigenous and wild foods, but also common locally-consumed dishes, could help in guiding nutritional interventions to reduce deficiencies and improve nutritional status [37]. For example, poultry interventions, which are well documented in the literature, and are associated with an increase of chicken and egg production and consumption, could prevent vitamin A deficiency [84], but need to be implemented with broader poverty eradication projects, education and interventions which address other compounding social risks, such as excess of alcohol consumption. Indeed, culturally-sensitive and feasible nutritional programs are needed to maintain the integrity of Indigenous and local food systems, but also to contribute to economic development [50].
Despite some limitations, the findings of this study highlight emerging evidence that nutrients and caloric intake may vary each month among rural and Indigenous communities, although there is need for longer longitudinal studies calculating daily variation in dietary intake, and involving a higher number of participants. Ensuring monthly adequate dietary intake, and addressing individuals' nutritional needs in case of food crisis would prevent children and adults from malnutrition and ensure healthy growth [85].