Women's Dietary Diversity Changes Seasonally in Malawi and Zambia

Woman selling vegetables in the Luwingu Market (Zambia). Photo Credit: M.Ahern/Bioversity Women’s Dietary Diversity Changes Seasonally in Malawi and Zambia Molly B. Ahern1*, Gina Kennedy2, Gianluigi Nico3, Ousmane Diabre3, Frezar Chimaliro4, Grace Khonje4, and Emmanuel Chanda5 1 Bioversity International, Lusaka, Zambia 2 USAID Advancing Nutrition, Washington, DC 3 United Nations Food and Agriculture Organization, Rome, Italy 4 Small Producers Development and Transporters Association (SPRODETA), Mzuzu, Malawi 5 Self Help Africa (SHA), Luwingu, Zambia *Corresponding Author- Molly B. Ahern Email: [email protected]

Authorship: MA and GK designed the study. MA, FC, GKh and EC undertook data collection and analysis. GN and OD conducted statistical analysis. MA, GK, GN and OD wrote the manuscript. All authors revised and approved of the final manuscript. Conflicts of Interest: The authors declare that this are no conflicts of interest. Acknowledgements: The authors would like to thank Dr. Andrew Thorne-Lyman (John Hopkins University) for his review of this manuscript prior to submission. Financial Support: This research was funded by the International Fund for Agricultural Development (IFAD) Grant Number 2000000974 with project partners McGill University (Montreal, Canada), WorldFish, Bioversity International, Self Help Africa (Zambia), University of Zambia, SPRODETA (Malawi) and the Lilongwe University of Agriculture and Natural Resources (Malawi). Additional support was provided by the CGIAR research program on Agriculture for Nutrition and Health. Ethical Standards Disclosure: This study was conducted according to the guidelines laid down in the Declaration of Helsinki and all procedures involving research study participants were approved by the Malawi National Commission for Science and Technology and the Directorate of Research and Graduate Studies at the University of Zambia. Written informed consent was obtained from all respondents. Shortened version of the title: Seasonal Dietary Diversity of Women in Malawi and Zambia

Abstract: Objective: There is growing recognition of the role that seasonality plays in agricultural production, expenditure, food security, diet quality and nutritional status, however, annual or bi-annual surveys may not capture seasonal or intra-seasonal shifts in dietary intake which can inform agriculture and nutrition policies, programming, and monitoring and evaluation.

Design, Setting and Participants: Seasonal variation in diets of women of reproductive age (WRA) living in rural Malawi and Zambia were measured bimonthly for eleven rounds, from September 2017 to May 2019. Trained enumerators collected data on a sample of 200 women using a qualitative 24-hour list- based recall of food items consumed, based on the ten food groups for Minimum Dietary Diversity of Women (MDD-W).

Results and Conclusion: There were significant seasonal fluctuations in the percentage of women achieving MDD-W, ranging from a low of 18% to a high of 82%. MDD-W followed expected fluctuations, peaking during harvest season and lowering during lean season, however, there were unexpected highs and lows at other times, demonstrating the importance of regular monitoring. The study demonstrated significant seasonal fluctuations in the proportion of WRA achieving MDD-W, having implications for project monitoring and evaluation. The research provides evidence of periods of abundance and scarcity for nutritionally important food groups.

Keywords: Dietary Diversity, Seasons, Seasonal Variation, Humans, Diet Surveys/methods, Female, Zambia, Malawi, Vegetables, Vitamin A, Fruit, Nuts, Seeds, Fish, Food Supply, Food Systems

Acronym List DDS Dietary Diversity Score

DGLV Dark Green Leafy Vegetable

DHS Demographic and Health Surveys

EPA Extension Planning Area (Malawi)

FISP Farm Input Subsidy Programme (FISP) in Malawi

FGD Focus Group Discussion

MDD-W Minimum Dietary Diversity for Women

NGO Non-Governmental Organization

SCLANSA-FVC “Strengthening Capacity of Local Actors on Nutrition-Sensitive Agri- Food Value Chains in Zambia and Malawi” (Project)

SHA Self Help Africa (based in Lusaka, Zambia),

SPRODETA Small Producers Development and Transporters Association (based in Mzuzu, Malawi)

WRA Women of Reproductive Age (15-49y)

1. Introduction Globally, Sub-Saharan Africa has the second-highest percentage of WRA and pregnant women with iron deficiency anemia (1), which increases the risk of early delivery, low birthweight, and is associated with iron deficiency in children postpartum (2). Micronutrient deficiencies, which are exacerbated by diets low in micronutrients, are common in both Malawi and Zambia, where persistent child stunting, high rates of iron deficiency anemia and vitamin A deficiency are also common. More than one-third of children are chronically malnourished, or stunted, in both countries (37% in Malawi and 35% in Zambia) as well as more than one-in-ten children being underweight (12% in both Malawi and Zambia) (3, 4). There are high vitamin-A supplementation rates in both countries, often administered biannually (99% in Zambia and 91% in Malawi) (1), and iron deficiency anemia is a public health concern for children under five years of age and WRA (63% and 33% in Malawi; 54% and 33% in Zambia respectively)(1, 3). The rate of anemia is even greater for pregnant and lactating women due to increased needs for iron during this time (for example, anemia prevalence in Malawi is 45% of pregnant women, 33% of lactating women and 30% of non-pregnant nor lactating women) (3).

Dietary diversification is an important strategy for improving micronutrient adequacy and has been linked to improved probability of micronutrient adequacy in the diet of WRA (5). Inadequate intake of foods rich in iron is reported to be the leading cause of anemia amongst WRA in Sub-Saharan Africa (6, 7). Similarly, it is important to note that inadequate consumption of vitamin-A in the diet can lead to marginal vitamin- A status between bi-annual supplementation (8). A recent study in Malawi found that approximately three-quarters of food energy comes from maize, the primary starchy staple crop, prepared as a thick porridge called nsima, while intake of micronutrient-rich foods – particularly those rich in iron, vitamin A (such as animal protein, dairy and eggs) is low (9, 10).

As in many countries in Sub-Saharan Africa, the rural poor in Malawi and Zambia are heavily reliant on smallholder subsistence agriculture for food security and nutrition, and as such, are particularly vulnerable to seasonal hunger and nutrient deficiencies (11, 12). As in many low-income countries, farming communities in Malawi and Zambia experience seasonal changes in food availability and access, corresponding to agricultural cycles that also impact fluctuations in income and expenditure (9). Although many rural communities may rely on subsistence agriculture, access to markets for buying and selling food are also important for diet quality and food security (13, 14, 15). Data from four Sub-Saharan African countries (including Malawi and Zambia) show that about 45% of foods consumed by rural households were purchased (16).

Seasonal changes in agricultural production not only directly affect the household’s food security and nutrition but also may reduce food and nutrition security and dietary diversity through reduced income

5 and food shortages resulting in higher food prices during the lean season (17, 18). In areas with variable climatic conditions, smallholder producers may be more adapted to coping with prolonged rainy or dry season and may practice ex-ante or ex-post food coping strategies which could be promoted through transfer of knowledge and skills (17, 19). Households vulnerable to seasonal hunger may use ex-ante strategies to smooth consumption, such as on-farm diversification, livestock ownership or crop storage; or ex-post strategies such as eating less preferred foods, lower quality foods or collecting foods from the wild (19). Higher on-farm crop diversity is associated with reduced levels of seasonal hunger and malnutrition (20-23), and a recent study from Tanzania positively correlated increased crop diversity with higher dietary diversity for WRA (15). Annual changes in food availability affect the diet of everyone in the household, with particularly severe consequences for WRA and pregnant women, as they experience increased nutritional needs (24).

Recognizing the role that seasonality can play in agricultural production, market prices, expenditure, food security, diet quality and nutritional status, annual surveys conducted in the same-season, same-site offer a feasible framework for assessing trends in agriculture, food security and nutrition to identify policy and program interventions (18). However, annual surveys may not fully capture these trends throughout the year (18), and changing climatic conditions result in shifts in usual seasonal food intake, requiring research on seasonal dietary trends. Still much of the current evidence focuses on seasonal consumption of specific foods rather than the whole diet (25) or is binary in that data collection takes place in a pre-harvest lean season and a post-harvest period of food abundance (9, 12, 26). These results are timely in highlighting the interpretation of MDD-W data based on seasonality, as there has been widespread uptake of the MDD-W indicator in nationally-representative Demographic and Health Surveys, as well as strong support from the European Union, United Nations agencies and international NGOs (27).

In this study, data was collected at regular intervals (bi-monthly) for a period of 21 months (September 2017- May 2019) to monitor changes in food consumption throughout the year and to assess the seasonal changes in the diets of WRA. We hypothesized that seasonal heavy rains and extremely dry periods of the year affect agricultural production and food availability at different times of year, thus causing fluctuations in dietary diversity for WRA in rural communities. In this paper, we refer to harvest season as the season in which starchy staple crops are harvested – in the survey area in Malawi, the main starchy staple crop (maize) is harvested beginning in March, while in the survey area in Zambia, the main starchy staple crop (cassava) are harvested in April-May.

2. Materials and Methods 2.1. Study Area Bioversity International led dietary monitoring in the selected project areas of the IFAD-McGill project “Strengthening Capacity of Local Actors on Nutrition-Sensitive Agri-Food Value Chains in Zambia and Malawi” (SCLANSA-FVC), with field support from local partners Self Help Africa (SHA, based in Lusaka, Zambia), and Small Producers Development and Transporters Association (SPRODETA, based in Mzuzu, Malawi). Project locations were Chitipa District in Malawi and Luwingu District in Zambia.

Chitipa district is located in the northwestern corner of Malawi, approximately 100km from Karonga, a larger town situated on the northern shore of Lake Malawi. Chitipa district shares borders with Zambia and Tanzania, and many informal traders from these countries cross through Chitipa, while others use the official border post to the north of Karonga. Chitipa district is divided into Extension Planning Areas (EPA’s), of which Lufita EPA and Kameme EPA were selected for the project interventions and this study. The warm season lasts for 2-3 months, between September and early December, with an average daily temperature of 82 degrees Fahrenheit. The cool season falls between late April and August, with daily temperatures ranging from 48 - 71 degrees Fahrenheit. Rainfall varies by season, with the wet season lasting between late November and April, with peaks for rain in January1.

Luwingu District in Zambia is situated to the west of Kasama (the largest town in Northern Province) and to the northeast of Mansa, the largest town in Luapula Province. Luwingu is north of Lake Banguwelu and the Banguwelu wetlands, and to the southwest of Lake Tanganyika, putting it at a crossroads between two of the major water bodies in Northern Zambia. Luwingu is divided into wards, of which two were chosen for the project and this study: Ipusukilo and Ibale. Luwingu district has a short hot season of 2 months, starting in September and ending in November, with an average high temperature above 88 degrees Fahrenheit. The cool season lasts 3-4 months, between December and April, with temperatures ranging between 49 - 77-degrees Fahrenheit. The rainy season lasts from November to April, with the peak of rainy season in December2.

2.2. Sampling methodology Baseline studies were conducted in each country in December 2016 to inform project activities, with results reflecting that only 18%, and 37% of WRA achieved MDD-W in Zambia and Malawi

1 Data from online source: Weather Spark. Accessed on 8 July 2019. https://weatherspark.com/y/97554/Average-Weather-in-Chitipa-

Malawi-Year-Round

2 Data from online source: Weather Spark. Accessed on 8 July 2019. https://weatherspark.com/y/95868/Average-Weather-in- Luwingu-Zambia-Year-Round

7 respectively. These figures were used to inform the sample size for dietary monitoring, based on a target population of 1,200 household per country with a 95% confidence interval. A sample size of 100 households in each country was chosen (also due to budgetary constraints), resulting in a margin of error of 9.06% for Malawi and 7.21% for Zambia.

Households with at least one WRA (15-49 y) were randomly selected from a list of households developed for the project baseline survey. WRA living in Chitipa District, Malawi (n=100) and Luwingu District, Zambia (n=100) were randomly selected from the two extension planning areas (EPAs), Kameme and Lufita, in Malawi and two wards, Ipusukillo and Ibale, in Zambia. After providing informed consent, the women were enrolled in the panel study.

2.3. Questionnaire adaptation and data collection To develop the list-based dietary diversity questionnaire, focus group discussions (FGDs) were held in the target communities in Malawi in March 2017 and in Zambia in April 2017. Focus groups provided information on available food items in each community, which were categorized into the ten food groups for MDD-W. A questionnaire tool was designed that included a list of food items under each of the ten MDD-W food groups, based on available foods in each country (refer to Appendix A and B to view the dietary monitoring tool/food lists for each country).

A woman from the community leads a session during FGDs (Lufita Men in FGD (Ibale Ward, Zambia, April 2017). Photo Credit: EPA, Malawi, March 2017). Photo Credit: M.Ahern/Bioversity M.Ahern/Bioversity

An enumerator-assisted, list-based dietary diversity questionnaire was administered every-other month for eleven rounds, from September 2017 to May 2019. Enumerators were identified and trained in May 2017 and June 2017, and a refresher training was done in September 2018. The same two enumerators conducted surveys across all completed rounds in Malawi, and one enumerator remained consistent across all rounds of data collection in Zambia. The enumerator in Zambia enlisted the support of two Extension

Officers in local communities to assist with data collection. Enumerators were trained to tick off or circle the food item on the questionnaire based on respondent recall. The research team excluded foods which were consumed in less than the minimum quantity (15 grams, or approximately 1 tablespoon), according to guidelines (28). Enumerators were advised to bring a tablespoon with them when collecting data, to represent the minimum portion of food which the woman must have consumed for it to be considered. If foods were consumed in quantities less than 15g, these were noted separately.

An enumerator completes a test-round of surveying with a woman in Ipusukillo Ward, Zambia. Photo Credit: M.Ahern/Bioversity

The study protocol accounted for potential sources of systematic error such as interpersonal differences in food consumption and day of the week (29). The same women were enrolled in all rounds of data collection, and there was no replacement if the woman was unavailable on the survey day. In Malawi, all 100 women participated in all 11 rounds of data collection (n=1100), as enumerators revisited the household if the woman was not available at first visit. In Zambia, few absences were recorded in some rounds of data collection (total n=1079 for 11 rounds) as enumerators did not revisit households if the

9 woman was not available at first visit. The day of the week on which surveys were conducted remained constant for each community- for example, Community A was surveyed on the Monday of the last week of the month for each round of data collection. This is an important consideration, as it reduces changes in the diet due to weekly fluctuations in food availability due to market days or festive days (eg. weekends).

2.4. Calculation of Indicators from Survey Data Using data collected through the surveys, we construct three main statistical indicators that are expected to capture women’s dietary intake:  Dietary Diversity Score (DDS) was calculated as the sum of the number of food groups (range 0- 10) out of the foods consumed from the ten MDD-W food groups in the 24 hours before the survey and is an ordinal indicator which we used to calculate mean DDS for our sample.  Minimum Dietary Diversity - Women (MDD-W) (28), was calculated as the percentage of women 15-49 years of age who consumed food items from five out of ten food groups during the previous 24 hours (meet MDD-W or achieve MDD-W).  Proportion of women consuming each of the ten food groups in each of the 11 rounds was calculated based on the number of women consuming the food group divided by the total number of responses per round (total responses across 11 rounds is n=1100 for Malawi, n=1079 for Zambia).

2.5. Statistical Methods To test for significance between seasons and quantify changes in the probability to meet MDD-W from one month to the other, we performed a logit regression, as reported in the below formula, separately for each country, where 𝑖 and 푚 are the indices for women and months. Given the nature of the outcome variable associated with MDD-W, which can only take value 0 or 1 (did not achieve MDD-W = 0; achieved MDD- W=1), the logit regression is used to test for significant differences in the probability to achieve MDD-W between the harvest and lean months. Estimated probabilities are based on the cumulative standard logistic distribution, using maximum likelihood estimation.

1 Pr(MDD − W = 1 / X X . . . X ) = 푖,푚 i,m,1, i,m,2, i,m,K 1 1 + ( ) 푒(훽0+훽1푋푖,푚,1+훽2푋푖,푚,2+⋯+훽푘푋푖,푚,퐾)

The probability for the 𝑖′푡ℎ woman in month m to meet dietary diversity 푃푟 (푀퐷퐷 − 푊푖,푚) is estimated as a function of a vector of independent variables (푥푖,푚). The independent variables included in the regression

10 analysis have been constructed as categorical dummy variables. Each categorical dummy is meant to capture the month of data collection, using September as the reference month (as it was the first month of data collection). As data was collected for nearly two years, rounds with corresponding months (September 2017 and September 2018; November 2017 and November 2018; January 2018 and January 2019; March 2018 and March 2019; May 2018 and May 2019; July 2018) were combined, to test if these seasonal variations in WRA achieving MDD-W were consistent across years of data collection, using September as the reference month. The probabilities to meet MDD-W estimated using the above formula with table 1 in the results section showing the marginal effects related to change in probability to meet MDD-W for each month with respect to September as the reference month. It is noteworthy that changes in probabilities to meet MDD-W --with respect to the reference month-- can be affected by other confounding factors (e.g. weather and economic shocks) which are not included in our regression analysis due to the lack of data. Such confounding factors are expected to be captured by the error term in the regression analysis. Estimated coefficients from the logit regression are reported in Appendix C.

3. Results 3.1. Diet Diversity Score (DDS) DDS was calculated for each woman across eleven rounds of data collection (Figure 1 and 2). In Malawi, mean DDS was 4.3, with median of 4 and a range of 2-9, 50% of women had scores between 4 and 6, 25% of women had scores less than 4, and the other 25% had scores higher than 6. Across the 11 rounds, an average of 40% of WRA achieved MDD-W in Malawi, while in Zambia, this figure was slightly higher at 50%. However, in Zambia, DDS varied more than in Malawi, with a mean of 4.7, median of 5 and a range of 1-10. The lowest DDS were in January, July and September in Malawi and in March, July and September in Zambia. DDS were greater in the second year of data collection (September 2018-May 2019) than in the first year (September 2017-July 2018). However, in Malawi, DDS were lower in the second year during the harvest season (March-May 2019 in comparison to March-May 2018).

Figure 1. DDS by round, Malawi

Figure 2. DDS by round, Zambia

3.2. MDD-W There were significant seasonal fluctuations in the percentage of women achieving MDD-W, ranging from a low of 18% to a high of 82% (see Figure 3).

Figure 3. Percentage of WRA who achieved MDD-W by month and year for Zambia and Malawi.

Percentage of Women who achieved MDD-W by Month and Year- Zambia and Malawi 100% Achieved MDD-W 90% 80% 70% 60% 50% 40% 79% 81% 82% 71% 70% 62% 63% 62% 30% 59% 54% 50% 43% 20% 34% 37% 37% 39% 38% 40% 26% 28% 10% 18% 20%

18-Jul 18-Jul

18-Jan 19-Jan 18-Jan 19-Jan

18-Sep 17-Sep 17-Sep 18-Sep

18-Nov 17-Nov 18-Nov 17-Nov

18-Mar 19-Mar 18-Mar 19-Mar

18-May 19-May 18-May 19-May Malawi Zambia

MDD-W followed expected fluctuations, peaking in the harvest season (March-May) and lowering during the “lean” season (January). Highs and lows in percentage of WRA achieving MDD-W were also seen at other times of the year. Our data suggests that there is possibly a second season of low dietary diversity during the “cold” months – between June and September (seen in our data in September 2017 and July 2018).

The probabilities to meet MDD-W were estimated using the formula in section 2.5 and corroborate many of the findings from the unconditional descriptive statistics that are presented below, with some interesting nuances due to the multivariate nature of the regression models estimated. Table 1 shows the marginal effects related to change in probability to meet MDD-W for a given month with respect to the reference month (September).

Table 1. Estimated probabilities to meet MDD-W

MDD-W Malawi MDD-W Zambia

VARIABLES (Reference month: Sept) (Reference month: Sept)

November 18% *** 20% ***

January 10% ** 13% **

March-(Harvest in Malawi) 43% *** -1%

May-(Harvest in Zambia) 15% *** 29% ***

July -18% *** 2%

Observations 1,100 1,079

*** p<0.01, ** p<0.05, * p<0.1

Our results confirm that agricultural patterns play a key role for women in achieving MDD-W: during the harvest season the probability for women to meet MDD-W increases remarkably, both compared to the base month (September) and compared to the other months corresponding to the off-season. For example, in Malawi the average probability to meet MDD-W is 80% during the month of March (harvest), nearly 43% higher than the corresponding probability in September. In Zambia, women have almost 68% probability to meet MDD-W during the month of May (harvest), a probability which is, on average, 29% higher than September.

3.3. Consumption by food group Figures 4 and 5 display the proportion of women consuming each food group across the 11 rounds with the sum of the proportions of all 10 food groups displaying the mean DDS for each round. In both countries, all women surveyed consumed a starchy staple food in every round of data collection, seen in Figures 4 and 5 as the starchy staple food group accounting for 1 full food group across all 11 rounds. Additionally, all Zambian women and nearly all Malawian women consumed a DGLV in each round. The third most common food group - other vegetables - was consumed by approximately three-quarters of Malawian women and half of Zambian women across most rounds of dietary monitoring, although there were times of lower consumption.

Figure 4. Food groups contributing to the mean DDS in Malawi

Figure 5. Food groups contributing to the mean DDS in Zambia

Beans are consumed nearly year-round in Malawi, with peak proportion of WRA consuming beans during the harvest season (March) and lowest proportion of WRA consuming beans in January. In contrast, nuts

15 and seeds, primarily groundnuts, were consumed on average by half of Zambian women throughout the year, with clear peaks during the harvest season (May) and troughs during the rainy season (January). Flesh foods – primarily small fish species – were consumed on average by half of respondents in Zambia, and approximately one-third of respondents in Malawi, with clear highs in the hot season (September) and lows in the rainy season (January). Vitamin A rich fruits and vegetables and other fruits were consumed on average by approximately one-third of women in Zambia, whereas less than one-fifth of women consumed foods from these food groups in Malawi, with highs between November-March and lows between May-September. Dairy and eggs were not highly consumed in either Zambia or Malawi, with less than one-fifth of women consuming these food groups on average.

Additional qualitative analysis of the food groups most and least commonly consumed as well as the most common species consumed by month can be found in Appendix D and E. These results show that within food groups, there are different species consumed in each country and some variance in species consumed by season. For example, within the DGLV food group, mustard, bean leaves, and pumpkin leaves were most commonly reported in Malawi, and cassava leaves, sweet potato leaves and pumpkin leaves were the most common in Zambia, although consumption of these varies by season. Similarly, the most common fish species consumed in Malawi was Kabwili (Bathyclarius longibarbis), followed by Usipa (Engraulicypris sardella), and in Zambia, Amatuku (Tilapia sparmanii) was most common in dry season, with Imilonge (Clarias theodorae) consumed in rainy season. Appendix G displays the number of food items consumed per food group by all women surveyed in Malawi and Zambia by round of data collection as well as total number of food items reported in initial focus group discussions for developing the survey tools.

During the harvest season (March-May), foods from the other fruits, dairy and eggs food groups are more commonly consumed. Oranges, banana and guava (Other Fruits) were more frequently consumed in March and May. Dairy and eggs are not commonly consumed in the targeted communities, however, these food groups were consumed by more WRA surveyed in Malawi during the harvest season (March) and after the harvest season (May). In addition, some vitamin A rich fruits and vegetables are consumed during harvest season, including orange-fleshed sweet potatoes and pumpkins, although vitamin A rich fruits and vegetables were more widely consumed in both countries during the lean season, such as mangoes (in both countries) and wild fruits such as amasuku, a wild loquat with bright orange flesh (in Zambia).

4. Discussion 4.1. Monthly variation in DDS and MDD-W One critique of DDS and MDD-W is that they capture a ‘snapshot’ of the situation, without considering seasonal variations in the diet (30). Increasingly, studies have tried to address this limitation, however many are cross-sectional at 2 points in time, or do not account for the full cycle of seasons (31-34). In similar studies, regular dietary monitoring was conducted in Malawi on a quarterly basis (35) and in Bangladesh on a monthly basis (30), making it possible to identify seasons of plenty, lean season and a ‘lesser’ season, similar to our study. Our study demonstrated significant variations in dietary diversity for WRA by month, a finding which has been observed in similar studies around the world (30, 32-34). A recent study conducted in Malawi found that seasonality affects both dietary diversity and nutrient adequacy, and that foods consumed are highly sensitive to seasonality in low-resource settings (9). Two studies in Malawi found that meal frequency and energy intake decreased during the rainy season, with clear reductions in consumption of certain food groups (24, 35). In a study similar to ours in Zambia, DDS varied significantly based on 24-hour recall data collected on a monthly basis for 6 months between August and March (31). Our study supports these previous findings of seasonal variation and adds to these findings through repeated bi-monthly panel data collection over 21 months.

The probability to achieve MDD-W was at its highest in the harvest season in both countries, which may be explained by greater production diversity on-farm, or could be due to increased agricultural revenue, allowing more access to diverse foods from the market (36, 37). Although the probability to meet MDD-W in July was low, there was greater variance in DDS in Zambia at this time, as few women reported 6 or more food groups consumed. This can potentially be due to some households practicing different planting and harvesting techniques, preserving and storing food after the harvest season for later consumption, or could be explained by greater revenue available in July from sale of surplus foods after the harvest season. Although access to markets has been demonstrated to increase dietary diversity (13, 14), a study in Malawi found that nutrient-dense crops are often more expensive than starchy staple crops in areas where production diversity is low – suggesting that greater production diversity plays an important role in market access to nutrient-dense foods (38). The probability to achieve MDD-W was low in the period preceding the harvest, which is typical in many countries where prevalence of malnutrition increases in the period preceding harvest and often coinciding with the rainy season (31).

A second period of higher dietary diversity is observed in November, especially in Zambia and in the second year of data collection (2018) in Malawi. This corresponds to a period of calm, warm weather, with intermittent rains as the rainy season approaches. In both November 2017 and November 2018, the number of women who reported consumption of vitamin-A rich fruits and vegetables and flesh foods

(small fish) increased, contributing to greater DDS and percentage of women achieving MDD-W. Fishing activity during this time of year may be greater in preparation for the fish ban (December 1-March 1 in Zambia and 1 November-31 March for Lake Malawi) or due to the calmer weather at this time of year, preceding the heavy rains. The higher dietary diversity in November following greater fishing activity in the dry and hot season (September-October) may be due to greater fish consumption or income for households involved in fishing activities or could be a time of greater market access to fish. In Malawi, the greater availability of vitamin-A rich fruits and vegetables in November each year is primarily due to mango availability, whereas in Zambia, women reported consuming mangoes and imfungo, an orange- fleshed wild fruit.

In lean season and pre-harvest months (January-March), diets are less diverse, composed of mainly starchy staples, DGLV and other vegetables. During this time of year, there are marked differences in flesh food consumption and bean consumption (particularly in Malawi), and greater consumption of vitamin A rich fruits and vegetables (mostly due to mango being highly available in rainy season). Flesh food consumption is low during this time, as small fish from small-scale capture fisheries are the main animal source food that people consume in these areas and rainy season presents bad fishing weather as well as fish bans in both countries. However, strategies for smoothing consumption or maintaining dietary diversity during lean season were observed (see section 4.3).

4.2. Monthly variation in food group diversity Our study demonstrates three food groups – starchy staples, DGLV and other vegetables form the basis of the habitual diet of women in our study area and do not demonstrate much month-to-month fluctuation in percent of women who consume these food groups. Similar findings are documented in other studies (31, 40). Starchy staples, DGLV and other vegetables form the core of the diet and are important for food security, however while these foods provide a “base” for the diet, it is vital to promote consumption of foods from other food groups to ensure micronutrient adequacy in the diet. The main starchy staple dish consumed in both countries is called nsima or nshima - a hard porridge often made from maize or cassava flour, which is served in large portions and consumed by forming the nshima into a ball and dipping the ball in “relish” of small portions of nutrient-rich foods such as vegetables or fish. Dietary diversity surveys display that maize was the most common starchy staple food crop in both EPA’s in Malawi in all months of data collection, while in Zambia cassava was the most common starchy staple crop, with differing harvest seasons (March in Malawi, May in Zambia). Our study also documented a wide variety of DGLV species consumed, with some month-to-month variation in species. Fifteen varieties of DGLV were reported in Zambia, with cassava leaves consumed by nearly half of respondents throughout the year (most often in fresh form but at times in dried form), followed by pumpkin leaves, sweet potato leaves

18 and Chinese cabbage. Similarly, 16 varieties of DGLV were reported in Malawi, bean leaves were the most common throughout the year (in fresh form during harvest season and dried form at other times), followed by pumpkin leaves and mustard. Tomato and onion accounted for the majority of the food consumed in the Other Vegetables food group throughout the year, however mushrooms were consumed in the rainy season.

Our study found that beans were consumed by many women across most seasons (with clear lows during rainy season) in Malawi, while beans and peas were not common in Zambia. In Chitipa District of Malawi, many varieties of the Common Bean (Phaseolus Vulgaris) are cultivated and sold in markets as cash crops. Greater bean and maize consumption in target areas of Malawi may be driven by the Farm Input Subsidy Programme (FISP) in Malawi, which aimed at distributing maize, legumes and soya in 2016-2017 (41). On the contrary, nuts and seeds (namely groundnuts) were consumed by many women across the majority of the year in Zambia and consumed by less women in Malawi. The Zambian Fifth National Development Plan (2006-2010) began an era of national development plans in Zambia focused on increased production of starchy staples such as maize and cassava, as well as groundnuts (42), however cassava production varies regionally - there are recognized “cassava belts” in the country – one of which is the Northern Province (where this study took place) (43). The fresh cassava market is highly seasonal, and cassava sales peak in the rainy season when maize prices are high (44). Similar dietary patterns were found in a study in two districts in Zambia, where starchy staple crops (primarily maize and cassava), vegetables (particularly leafy vegetables) and groundnuts were the major foods consumed (40). Caswell et al. (31) found that the most common foods consumed were nshima (maize porridge) with side dishes of leafy vegetables (DGLV) prepared with tomatoes and onion (other vegetables). Chisanga and Zulu- Mbata (45) analyzed changes in income expenditure on food in Zambia, showing the lowest income quartiles consuming starchy staples and vegetables (DGLV and other vegetables), as expenditure on vegetables increased in comparison to more wealthy income groups, who increased expenditure on animal-source foods. Consumption of flesh foods in our study was generally limited to small fish, although there were clear highs and lows in fish consumption related to weather changes and fishing bans. Similarly, two other studies (31, 40) found that fish was the most common flesh food consumed – although in small amounts - with little or no consumption of dairy, eggs, meat, poultry or insects.

4.3. Coping strategies to smooth consumption and maintain dietary diversity Utilizing diverse food sources is a key factor of food and nutrition security, dietary diversity and food systems resilience (28, 46, 47). Our study documented the consumption of diverse foods by few people and foods which were consumed by many respondents across all seasons. Initial focus group data displays knowledge of a wider variety of foods available in local food systems than that which is consumed

19 regularly (Appendix G). The study was able to identify strategies that local people practice to smooth consumption and maintain dietary diversity throughout the year. For example, we observed diversified leafy vegetable production and processing to preserve DGLV and make them available year-round. In Zambia, we observed consumption of less-preferred fish species such as Imilonge, a species which is more common in aquaculture in the area, however, is not consumed by many people due to religious beliefs. Additionally, consumption of alternative animal protein sources (mice and insects) during times of lower fish availability, as well as consumption of wild fruits, however these strategies were not documented in Malawi. In Zambia, wild fruits in the “Other Fruits” category were consumed during the pre-harvest rainy season, suggesting that collection of wild fruits could be a coping strategy for lower food availability during this time. In Malawi, wild or indigenous fruits were reported less than ten times across all months of data collection suggesting that wild fruits are either less available, or women do not collect foods from the wild. Studies in other countries have found higher food group diversity in periods of lower food availability, as a result of increased consumption of wild foods during these times (30, 34), however there is other studies documenting that this dietary adaptation did not occur in Malawi (9).

In our study, this may be partially explained by project partners in Zambia including promotion of seasonally available wild foods in nutrition education programming, a strategy which could be replicated in Malawi. Consistent consumption of animal proteins can be promoted through various methods, including improvements to traditional small fish drying and smoking methods as well as improved storage to preserve fish during times of high availability to consume during times of lower availability, as well as promoting coping strategies such as consuming mice and other flesh foods or insects which are highly available during periods when fish are less available. In addition, the important role that wild or indigenous fruits have in diversifying diets during the lean season should be promoted through awareness campaigns and encouraging cultivation and processing for preservation of wild or indigenous fruits, such as amasuku (wild loquat), to expand consumption of these fruits and preserve them during times of high availability. Year-round access to vitamin-A rich fruits and vegetables can be encouraged through extension services promoting production and consumption of diverse vitamin-A rich crops such as orange-fleshed sweet potato and pumpkin, as well as innovative preservation methods to preserve mango during times of high availability (rainy season) for times of low availability.

4.4. Implications of these findings to programs and policy The significant statistical findings confirm our hypothesis of monthly variation in women achieving MDD-W and diversity of food groups consumed. The most probable explanation follows weather patterns, with seasonal heavy rains and extremely dry periods of the year affecting agricultural production and food availability from own production and in markets at different times of year, thus causing

20 fluctuations in dietary diversity for women in rural communities. Therefore, an effective nutrition program for WRA should be more sensitive to specific months of lower dietary diversity, such as January, July, and September, where the DDS is particularly low in both countries.

Probability to achieve MDD-W varied greatly across seasons, with a maximum difference in probability of 61% greater probability to achieve MDD-W in March in comparison to July (seen in 43% difference between March and the reference month and -18% between July and the reference month, see Table 1 for Malawi), a finding which can inform interpretation of MDD-W data in annual DHS surveys. The MDD- W indicator is best suited for population-level assessment but is also advocated for monitoring change in dietary diversity across time or with interventions (48). The findings from our study reflect significant changes in DDS and probability of women achieving MDD-W across months. This finding has policy implications for use in program monitoring, as the use of MDD-W as a monitoring and evaluation indicator may not lead to accurate assessment if measurements taken at different periods of the year are used to evaluate the success of a program without controlling for the effect of seasonality. A cross- country evaluation of nutrition outcomes using MDD-W as an indicator for evaluation of nutrition interventions in eight countries in Africa found that diet quality can easily change depending on seasonality and annual changes in weather, highlighting the importance of continuous monitoring to adjust program activities (49). 4.5. Study limitations It is important to note that there were limitations of this study - for example, data did not include quantitative information concerning the actual amount of each food consumed, beyond ensuring that the minimum quantity of food was consumed for it to ‘count’ (15g). All women in both settings consumed starchy staple crops every month, mainly maize and cassava. Anecdotally, we observed that these starchy staple foods are consumed in large portions and supplemented by small amounts of nutrient-rich food items. Since portion sizes were not collected in 24-hour recalls, we cannot further analyze nutrient adequacy of the diet beyond dietary diversity. In addition, the study did not collect information on where foods were acquired – from own production or markets. Furthermore, these results may not be generalizable to other areas of Zambia or Malawi or nearby countries, as agroecological conditions vary (15). A final limitation is confounding in our results due to improvements in dietary diversity over time as a result of population level interventions. In our study area, the population received nutrition education as well as training on food processing. We are not able to control for the impact of these interventions on the changes in DDS and MDD-W over time. Similarly, the women participating in the study may have more biased responses over time as they become more familiar with the food groups and they learned more about the importance of dietary diversification.

It is noteworthy that the regression techniques implemented in this paper are good at testing difference in the probability to achieve MDD-W, but are less successful at finding the root causes. This is because a satisfactory explanation of why some women achieve MDD-W does not only depend on the harvest and lean months but also on a number of geographic, household, cultural and individual characteristics (e.g. the poverty and economic status of the household where women live, the level of education attained, cultural and nutritional factors, etc.), which have not been collected through this survey. Therefore, estimates from the econometric models aim at showing how the probability to achieve MDD-W changes across months of the year, but cannot so easily explain why some women meet MDD-W, while some others do not. Moreover, we expect the probability to achieve MDD-W and to consume more food groups to be highly specific to each target area in this study, which motivated our choice to control for country- specific effects. This is because, while a general relationship between the probability to achieve MDD-W in the lean/harvest months can be established a priori, a complex mix of demographic, economic, political, cultural, household and nutritional factors --that are not captured in the regressions-- are expected to influence both the probability to achieve MDD-W, as well as the likelihood to consume a given number of food groups, at the country level.

5. Conclusion The research provides evidence of periods of abundance and scarcity for nutritionally important food groups. Over 21 months of panel surveys conducted in Zambia and Malawi, an average of 50-60% of WRA do not consume a diverse enough diet to achieve MDD-W, ranging from 18%-82% in different seasons. Dietary monitoring conducted at regular intervals helped to validate pre-harvest lean seasons and post-harvest seasons of food abundance, as well as other seasons of high dietary diversity and low dietary diversity. These periods of high dietary diversity can be explained by months of greater diversity of foods available in local production, while markets are also recognized for their role in improving dietary diversity. Our results reflect up to a 61% difference in probability to achieve MDD-W across different seasons, however seasonal food availability and consumption can be influenced by geographic, demographic, economic, political, cultural, household and individual nutritional factors, highlighting the need for seasonal dietary diversity surveys in other countries and regions to help inform interpretation of national surveys and monitoring of programs. This has important implications in terms of monitoring and evaluation of nutrition projects, as dietary surveys should consider various seasonal changes in food availability which affect consumption, dietary diversity and nutrition.

These results may inform agriculture and nutrition policies and programming, including initiatives to improve nutrition-sensitive processing, storage, access to and awareness of diverse food items from food groups which vary seasonally in local food systems. Information on seasonal food availability should

22 always be considered in assessments, particularly in rural areas, in order to improve planning of targeted and responsive food and nutrition security interventions for improving access to nutrient dense foods (50). Awareness raising of seasonal food availability through nutrition education or promoting ex-ante and ex- post coping strategies to maintain dietary diversity such as diversified production, collection of wild foods and capacity building and technology for food preservation and storage, as well as improved access to markets have potential to positively affect dietary diversity for WRA.

6. References 1. World Bank, (2019). World Bank Open Data. Accessed on 17 November 2019. https://data.worldbank.org/ 2. Allen LH. (2000). Anemia and iron deficiency: effects on pregnancy outcome. Am J Clin Nutr. 2000;71(5 Suppl):1280S-4S. doi:10.1093/ajcn/71.5.1280s 3. DHS, N. S. (2016). Malawi Demographic and Health Survey. Zomba: Government of Malawi National Statistical Office. 4. Zambia Statistics Agency, Ministry of Health (MOH) Zambia, and ICF. (2019). Zambia Demographic and Health Survey 2018. Lusaka, Zambia, and Rockville, Maryland, USA: Zambia Statistics Agency, Ministry of Health, and ICF. 5. Arimond, M., Wiesmann, D., Becquey, E., Carriquiry, A., Daniels, M.C., Deitchler, M., FanouFogny, N., Joseph, M.L., Kennedy, G., Martin-Prével, Y. & Torheim, L.E. (2010). Simple food group diversity indicators predict micronutrient adequacy of women’s diets in 5 diverse, resource-poor settings. J Nutr. 140(11): 2059S–69S. 6. McClure E. M., Meshnick S. R., Mungai P., et al. (2014).The association of parasitic infections in pregnancy and maternal and fetal anemia: a cohort study in coastal Kenya. 2014;8(2) doi: 10.1371/journal.pntd.0002724.e2724 7. Ononge S., Campbell O., Mirembe F. (2014). Haemoglobin status and predictors of anaemia among pregnant women in Mpigi, Uganda. 2014;7(1, article no. 712) doi: 10.1186/1756-0500-7- 712. 8. Tanumihardjo SA, Permaesih D, Muhilal Vitamin A status and hemoglobin concentrations are improved in Indonesian children with vitamin A and deworming interventions. Eur J Clin Nutr. 2004;58:1223–1230. 9. Hjertholm KG, Holmboe-Ottesen G, Iversen, PO, Mdala I, Munthali A, Maleta K, Shi Z, Ferguson E, Kamudoni P. (2019). Seasonality in Associations between Dietary Diversity Scores and Nutrient Adequacy Ratios Among Pregnant Women in Rural Malawi – A Cross-Sectional Study. Food & Nutrition Research. doi.org/10.29219/fnr.v63.2712

10. Schmaelzle S, Kaliwile C, Arscott SA, Gannon B, Masi C, Tanumihardjo SA. (2014). Nutrient and Nontraditional food intakes by Zambian Children in a Controlled Feeding Trial. Food Nutrition Bulletin 35 (1): 60-67. doi: 10.1177/156482651403500108 11. Abdalla, S., Leonhäuser, I. U., Bauer, S., & Elamin, E. (2013). Households coping strategies and food consumption gap in Sudan (Working Paper No. 13/06). Pavia: Natural Resources, Agricultural Development and Food Security-International Research Network (NAF-IRN). 12. Becquey, E., Delpeuch, F., Konaté, A. M., Delsol, H., Lange, M., Zoungrana, M., & Martin- Prevel, Y. (2012). Seasonality of the dietary dimension of household food security in urban Burkina Faso. British Journal of Nutrition, 107(12), 1860–1870. doi:10.1017/S0007114511005071 13. Meerman, J; Aberman, NA; Harris, J; and Pauw, K. (2018). Chapter 2: Indicators for Examining Links Between Agriculture, Food Security and Nutrition. In Agriculture, Food Security and Nutrition in Malawi: Leveraging the Links. International Food Policy Research Institute, 2018. DOI: https://doi.org/10.2499/9780896292864 14. Koppmair, S; Kassie, M; Qaim, M. (2016). Farm Production, Market Access and Dietary Diversity in Malawi. Public Health Nutrition 20(2), pp 325-335. doi:10.1017/S1368980016002135 15. Bellows, A.L., Canavan, C.R., Blakstad, M.M., Mosha, D., Noor, R.A., Webb, P., Kinabo, J., Masanja, H., Fawzi, W.W. (2020). The Relationship Between Dietary Diversity Among Women of Reproductive Age and Agricultural Diversity in Rural Tanzania. Sage Food and Nutrition Bulletin 2020. Vol. 41 (1) 50-60. https://doi.org/10.1177/0379572119892405 16. Reardon, T., Tschirley, D., Minten, B., Haggblade, S., Liverpool-Tasie, S., Dolislager, M., Snyder, J. & Ijumbaa, C. (2015). Transformation of African agrifood systems in the new era of rapid urbanization and the emergence of a middle class. In O. Badiane & T. Makombe, eds. Beyond a middle income Africa. Transforming African economies for sustained growth with rising employment and incomes, pp. 62–74. ReSAKSS Trends and Outlook Report. Washington, DC, IFPRI. 17. Yengoh, G.T., Tchuinte, A., Armah, F.A. et al. (2010). Impact of prolonged rainy seasons on food crop production in Cameroon. Mitig Adapt Strateg Glob Change 15, 825–841 (2010). https://doi.org/10.1007/s11027-010-9241-2 18. Kanungwe Kalaba , Quinn CH, Dougill AJ. (2013). The role of forest provisioning ecosystem services in coping with household stresses and shocks in Miombo woodlands, Zambia. Ecosystem Services, Volume 5, 2013, Pages 143-148. https://doi.org/10.1016/j.ecoser.2013.07.008.

19. C. Leigh Anderson, Travis Reynolds, Joshua D. Merfeld & Pierre Biscaye. (2018). Relating Seasonal Hunger and Prevention and Coping Strategies: A Panel Analysis of Malawian Farm Households, The Journal of Development Studies, 54:10, 1737- 1755, DOI: 10.1080/00220388.2017.1371296 20. Abdalla, S., Leonhäuser, I. U., Bauer, S., & Elamin, E. (2013). Households coping strategies and food consumption gap in Sudan (Working Paper No. 13/06). Pavia: Natural Resources, Agricultural Development and Food Security-International Research Network (NAF- IRN). [Google Scholar] 21. Afifi, T., Milan, A., Etzold, B., Schraven, B., Rademacher-Schulz, C., Sakdapolrak, P., … Warner, K. (2015). Human mobility in response to rainfall variability. Opportunities for migration as a successful adaptation strategy in eight case studies. Migration and Development, 1(21), 254–274. [Google Scholar] 22. Bacon, C. M., Sundstrom, W. A., Gómez, M. E. F., Méndez, V. E., Santos, R., Goldoftas, B., & Dougherty, I. (2014). Explaining the ‘hungry farmer paradox’: Smallholders and fair trade cooperatives navigate seasonality and change in Nicaragua’s corn and coffee markets. Global Environmental Change, 25, 133–149. doi:10.1016/j.gloenvcha.2014.02.005 23. Devereux, S. (2009). Seasonality and social protection in Africa (Working Paper No. SP07). Brighton, UK: Future Agricultures Consortium. 24. Ndekha M, Kulmala T, Vaahtera M, Cullinan T, Salin M, Ashorn P. (1999). Seasonal variation in the dietary sources of energy for pregnant women in Lungwena, rural Malawi. Ecology Food Nutr 1999; 38: 605–622. doi: 10.1080/03670244.2000.9991599 25. Moreau, M., Garaway, C.J. (2018). “Fish Rescue us from Hunger”: the Contribution of Aquatic Resources to Household Food Security on the Rufiji River Floodplain, Tanzania, East Africa. Hum Ecol 46, 831–848 (2018). https://doi.org/10.1007/s10745-018-0030-y 26. Anderson CL, Reynolds T, Merfeld JD, Biscaye P. (2016). Relating Seasonal Hunger and Prevention and Coping Strategies: A Panel Analysis of Malawian Farm Households. Taylor & Francis Journal of Development Studies 54 (10): 1737-1755. https://doi.org/10.1080/00220388.2017.1371296 27. Hanley-Cook, G. T., Tung, J., Sattamini, I. F., Marinda, P. A., Thong, K., Zerfu, D., Kolsteren, P. W., Tuazon, M., & Lachat, C. K. (2020). Minimum Dietary Diversity for Women of Reproductive Age (MDD-W) Data Collection: Validity of the List-Based and Open Recall Methods as Compared to Weighed Food Record. Nutrients, 12(7), 2039. https://doi.org/10.3390/nu12072039

28. FAO and FHI 360. (2016). Minimum Dietary Diversity for Women: A Guide for Measurement. Rome: FAO. 29. Rosalind S Gibson, U Ruth Charrondiere, Winnie Bell, Measurement Errors in Dietary Assessment Using Self-Reported 24-Hour Recalls in Low-Income Countries and Strategies for Their Prevention, Advances in Nutrition, Volume 8, Issue 6, November 2017, Pages 980– 991, https://doi.org/10.3945/an.117.016980 30. Stevens, B., Watt, K., Brimbecombe, J., Clough, A., Judd, J., & Lindsay, D. (2017). The role of seasonality on the diet and household food security of pregnant women living in rural Bangladesh: A cross-sectional study. Public Health Nutrition, 20(1), 121-129. doi:10.1017/S136898001600183X 31. Caswell, B., Talegawkar, S., Siamusantu, W., West, K., & Palmer, A. (2020). Within-person, between-person and seasonal variance in nutrient intakes among 4- to 8-year-old rural Zambian children. British Journal of Nutrition, 1-22. doi:10.1017/S0007114520000732 32. M’Kaibi, F.K., Steyn, N.P., Ochola, S. et al. (2015). Effects of agricultural biodiversity and seasonal rain on dietary adequacy and household food security in rural areas of Kenya. BMC Public Health 15, 422 (2015). https://doi.org/10.1186/s12889-015-1755-9 33. Hilbruner, C. and Egan, R. (2008). Seasonality, Household Food Security, and Nutritional Status in Dinajpur, Bangladesh. SAGE Food and Nutrition Bullletin. https://doi.org/10.1177/156482650802900308 34. Savy, M et al. (2006). Dietary Diversity Scores and Nutritional Status of Women Change during the Seasonal Food Shortage in Rural Burkina Faso, The Journal of Nutrition, Volume 136, Issue 10, October 2006, Pages 2625–2632, https://doi.org/10.1093/jn/136.10.2625 35. Zimba, TR et al. (2019). Seasonality of Food Availability Influences Dietary Patterns in Two Farming Districts of Malawi. https://www.biorxiv.org/content/10.1101/607606v1 36. Ntakyo, P.R., van den Berg, M. (2019). Effect of market production on rural household food consumption: evidence from Uganda. Food Sec. 11, 1051–1070 (2019). https://doi.org/10.1007/s12571-019-00959-2 37. Ayenew, H. Y., Biadgilign, S., Schickramm, L., Abate-Kassa, G., & Sauer, J. (2018). Production diversification, dietary diversity and consumption seasonality: panel data evidence from Nigeria. BMC public health, 18(1), 988. https://doi.org/10.1186/s12889-018-5887-6 38. Harris, J. and Aberman, N.L. (2018). ‘Chapter 7: Conclusions: The Key Role of Agriculture in Achieving Better Diets in Malawi’. In Aberman, N.L, Meerman, J. and Benson, T. Agriculture, Food Security and Nutrition in Malawi: Leveraging the Links. Washington DC. International Food Policy Research Institute. Pp 67-70.

39. FAO, IFAD, UNICEF, WFP and WHO. (2018). The State of Food Security and Nutrition in the World: Building Climate Resilience for Food Security and Nutriton. Rome. FAO. 40. Alamu, E.O., Gondwe, T., Eyinla, T.E., and Maziya-Dixon, B. (2019). Assessment of Dietary Diversity of Mothers and Children of 6-24 Months from Eastern and Southern Provinces of Zambia. Journal of Nutrition and Metabolism Volume 2019. Article ID 1049820. https://doi.org/10.1155/2019/1049820 41. Logistics Unit (2017). Final Report on the Implementation of the Farm Inputs Subsidy Programme (201617). Lilongwe 42. Republic of Zambia. (2006). Fifth National Development Plan (2006-2010). Accessed on 10 May 2020. Accessed from: https://evaw-global-database.unwomen.org/- /media/files/un%20women/vaw/full%20text/africa/zambia%20- %20fifth%20national%20development%20plan%202006-2010.pdf?vs=1508 43. Haggblade, S. and Nyembe, M. (2007). Cassava Transformation in Southern Africa (CATISA) Startup Task 3. Report on Zambia’s Cassava Value Chain. Accessed on 10 May 2020. Accessed from: https://pdfs.semanticscholar.org/6ef0/8eab62784e2ab02549dd1ea33d6183092168.pdf 44. Muimba-Kankolongo, A. (2018). Food Crop Production by Smallholder Farmers in Southern Africa: Challenges and Opportunities for Improvement. Chapter 9: Roots and Tuber Crops. Elsevier Academic Press. 45. Chisanga, B., Zulu-Mbata, O. (2018). The changing food expenditure patterns and trends in Zambia: implications for agricultural policies. Food Sec. 10, 721–740 (2018). https://doi.org/10.1007/s12571-018-0810-7 46. Herrera de Leon, HJ and Kopainsky, B. (2019). Do you Bend or Break? System Dynamics in Resilience Planning for Food Security. Syst. Dyn. Rev. 35, 287–309 (2019) DOI: 10.1002/sdr.1643 47. Tendall, D.M ; Joerin, J ; Kopainsky, B ; Edwards, P ; Shreck, A ; Le, Q.B ; Kruetli, P ; Grant, M ; Six, J. (2015). Food system resilience: Defining the concept. Global Food Security 6 (2015) 17- 23. https://doi.org/10.1016/j.gfs.2015.08.001 48. Women's Dietary Diversity Project (WDDP) Study Group (2017). Development of a Dichotomous Indicator for Population-Level Assessment of Dietary Diversity in Women of Reproductive Age. Current developments in nutrition, 1(12), cdn.117.001701. https://doi.org/10.3945/cdn.117.001701 49. Trentmann, C. and Weingärtner, L. (2019). “Appropriate Diet for All: Cross-Country Evaluation of Nutrition Outcomes”. Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ) GmbH. October 2019.

50. Wijesinha-Bettoni R, Kennedy G, Dirorimwe C, Muehlhoff E. (2013). Considering Seasonal Variations in Food Availability and Caring Capacity when Planning Complementary Feeding Interventions in Developing Countries. International Journal of Child Health and Nutrition, 2013, 2, 335-352.

Appendix Appendix A. Malawi Dietary Monitoring Tool

Malawi Dietary Monitoring Tool

Date: ______Participant Name: ______Sublocation: ______ID Number______

To be read to the respondent: Now I’d like to ask you about foods and drinks that you ate or drank yesterday during the day or night, whether you ate it at home or anywhere else. I am interested in whether you had the food items I will mention even if they were combined with other foods. For example, if you had a soup made with carrots, potatoes and meat, you should reply “yes” for each of these ingredients when I read you the list. However, if you consumed only the broth of a soup, but not the meat or vegetable, do not say “yes” for the meat or vegetable. As I ask you about foods and drinks, please think of foods and drinks you had as snacks or small meals as well as during any main meals. Please also remember foods you may have eaten while preparing meals or preparing food for others. Please do not include any food used in a small amount for seasoning or condiments (like chilies, spices, herbs or fish powder). I will ask you about those foods separately.

Enumerator: Begin with the first row and ask “Yesterday during the day or at night, did you eat or drink any foods made from …Read the name of the food group and provide some common examples. Go to the next food group and ask the same question continuing to the final column. Mark an “X” next to each food item, there can be more than one food item per food group or no foods from that food group.

Nuts & Vitamin A rich Starchy staples Beans & Peas Dairy Flesh foods Eggs DGLV Other vegs Other fruits seeds fruits & vegs Yes/No Yes/No Yes/No Yes/No Yes/No Yes/No Yes/No Yes/No Yes/No Yes/No

If yes, please check or circle the food items eaten from the food group below Iziba Bonongwe Amungu Amakoma Amakondo Busambia Amaji ya Anyezi Amankolo (Cow Izembwa Imyungu Makoma (Coco Yam) (Sesame) Boto amahanga (Onions) milk) Woto (Wild blite) (Pumpkin) (Palm Fruit)

(Guinea fowl eggs) Amangagu Izima ya Chigwada Ivilombe Ground mbuzu Amaji ya baka Karoti Bogha Amampangwa Boma Katapa Vingoma nut (goat Busipa (duck eggs) (Carrots) (Mushroom) Mamphangwa (Cassava Leaf) (maize) milk) Usipa Amai ya Ubaka nkhuku Chinese Amayawo (small Mangoe Kabichi Buyole Masumbi (Chinese Amamphesya (Cassava) ground Bye-Bye (Mango) (Cabbage) (Chicken cabbage) nut) Bai Bai eggs) Mbatata Mbatata Amaji ya Chinkhundya Olenji Amapeyala Mayengala (Yellow Sweet Icholonko inkwale (Quail Isalwa (Orange Mapeyala (Calabash) Potato) Chambo eggs) (Bean leaf) Sweet (Avocado) Potato) Mbatata Chisoso Inzelu Pawpaw Amatobo Imbalala Kapunika Tomato (White Sweet Daga (Papaya) (Snot apple) (Blackjack) Potato) Kapenta Insyababa Chiungwa Ndozi Indofani Ikhusi Uponda Tubala Amelwasama (Irish Potato) (Pumpkin (Okra?) (Garden Imphuni Leaf) Peas) Inthonga Isokolo Inkhombwe Mphange Zumba Chimwamwaji Swaswa (Plantains) Nandolo(Pige Inkhumba (Okra?) (Watermelon) (Cowpea leaf) on Peas) Mulambo Inzubu Izubu Isubu Malezi Mzama (Groundnut Amasele Chizovu (Millet) Inkolo-kolo (Bambara Nkholokolo leaf) nuts) Iweyu Isoholo Galegadeya Mapemba Kalembula Imphwa Isokolo (Granadilla, (Sorghum) (Sweet potato (Eggplant) Cowpeas Insipa Passionfruit) leaf) Kabaya Impwapwa Mpunga Kwayiti Kale (Small white Isongwe (Rice) (White Ivolo eggplant) beans)

Luni Indelema Vittuwo Kabulangeti Mziwa (Big Imbosya (Yams) Mabulangeti Kabaji (Cat’s mushroom) Ubaji whiskers)

Mpilu Kokwa Kalomo Imbula Kabwili (Mustard) (Camelina) Bonya

Katyetye Mugagala Imfungo Mandolo

Lusaka Muzaza Insungwa Maso Luhunju

Mtela Intonongwa Makhembi Mutela Pupwe (Wild Mutera blueberry?)

Rape Inthumbusya Maluwa Peza Urepu Nthumbusya Upeza

Mandondo Ukwipa Isongwi Sango Mangulungul Aboyela Jam-Jam Unchesya u (Cicada) (Mulberry) Lemoni Mankhamba Baka (Duck) Mandimu (Lemons) Masipengile Ma epozi Mwasipengile Imbuzi (Goat) (Dorset apple) (Sugar bean) Imphombo Masusu Mafwisa (Deer) Ing’ombe Magulwi Njunje (Cow) (Peaches) Inkhunda Soya Mapwama (Pigeon) Inthafu Masuku

(Grasshopper) (Loquat) Kalulu Matecha Uharulu

(Rabbit) Kayonga Matubu (Wild animal) Lupenga Mguava

(Wild animal) (Guava) Makanga Mulewe Ikhanga Umaleba (Guinea fowl) Mphalata Imbewe Ntochi (Black flying ants) (Bananas) not included in MDD-W score Nguluwe Olenji

(Pig) (Oranges) Nkhuku Tangerine (Chicken) Imbewa Tukomekomne (Mice) Imbelele Umpwele

(Sheep) Namatandela Inkwale

(Quail) Uhingwa

(small deer) Usindi

(Wild animal) Utuyuni

(Birds) Amabungu (Caterpillar) not included in MDD-W score

Appendix B. Zambia Dietary Monitoring Tool

Zambia Dietary Monitoring Tool

Date: ______Participant Name: ______Sublocation: ______

ID Number______

Vitamin A Nuts & Starchy staples Beans & Peas Dairy Flesh foods Eggs DGLV rich fruits & Other vegs Other fruits seeds Yes/No Yes/No Yes/No Yes/No Yes/No Yes/No vegs Yes/No Yes/No Yes/No Yes/No

If yes, please check or circle the food items eaten from the food group below Impupu Umukaka Amani ya Chinese Chilungunt Amale Ilanda Amayembe (Pumpkin (Cow ifibata (Chinese anda/ Ocra Amabungo (Millet) (Cowpeas) Amatuku (mango) Seeds) milk) (Duck eggs) cabbage) (Okra) Amani ya Ingolyolyo Ipupu Cibwabwa Amasaka inkoko Eggplant Amacungwa (Pigeon (Sunflow Babomba/ (Pumpkin Carrots (Sorghum) (Chicken (purple) (Orange) peas) er seeds) Bomba leaves) eggs)

Amani ya Cimkamba/ Impwa Intoyo Imbalala Amakolobwe Amataba makanga Cimpapila Ifipushi (Small (Garden (Groundn Bapolwe/ (prickly (Maize) (Guinea fowl (Bean (Pumpkin) white peas) uts) Polwe cucumber) eggs) Leaves) eggplant) Ifyumbu Kabulangeti Ifilashi (Irish Ibondwe (Orange Onyoni Amapela (Common Potatoes) Calukuwa (Amaranth) sweet (Onions) (Guava) bean) potatoes) Ifyumbu Lusaka Kacesha Amasuku (Whtie Sweet (Common (Cowpea Imfungo Tomato Cise (Wild Loquat) potatoes) bean) leaves) Ubowa Tute Popo Soya beans Kafulu (Mushroom Amateke (Cassava) (Papaya) Imbilya s) Kalembula Yellow White beans (Sweet pepa Mdrifei/ Umungu (common potato (Yellow Grapes bean) Imene leaves) pepper) Yellow bean Katapa Cikanda Umupunga Ichibimbi (Common (Cassava (African (rice) (Cucumber) bean) Imiita leaves) polony) Icikabi/ Lubanga Ifimbita Imiliba (Watermelon)

Ifinanashi Imilonge/ Lungulungu Umulonge (Pineapple)

Ifisali Pupwe Imimbulwe (Sugarcane)

Ifisongole Ulumanda Imintesa (Wild fruit) Umulembwe Ifitungulu Imishipa Katali Cibangankon Impundu Impata de

Indimu Rape Impende (lemons) Inkonde

Impifu (Bananas0

Insafwa Inkomo

Insongwa Inkundu

Kapenta Kotapela

Kolongwe (Avocado) Malberry/ Malubeni Nkonkonko (Mulberry) Manda/ Mandalena Ubushimba (Mandarine) Amakile

Ulukomo (Snot apple) Ifishimu/ Mumpa (caterpillar)- not included in MDD-W score Inshinge (Black flying ants) not included in MDD-W score Amakanga /Ikanga (Guineafowl) Duc/Imbata

(Duck) Fulwe

(Turtle) Imbushi

(Goat)

Imfuko

(Mole) Impanga

(Sheep) Impombo

(Antelope) Ingombe

(Cow) Inkoko (Village and

broiler chickens) Inkumba

(Pig) Insengele

(Guinea pig) Insenshi Inshimba Kapanga

(Rat) Kalulu

(Rabbit) Kolwe

(Monkey) Kote

Appendix C. Estimated coefficients from the logit regression

(1) (2) MDD-W Malawi MDD-WZambia VARIABLES

November 0.719*** 0.814*** (0.205) (0.206) January 0.398* 0.510** (0.205) (0.204) March-(Harvest) 2.005*** -0.048 (0.231) (0.207) May-(Harvest) 0.599*** 1.222*** (0.205) (0.211) July -0.767*** 0.076 (0.291) (0.253) Constant -0.619*** -0.490*** (0.148) (0.146)

Observations 1,100 1,079 Standard errors in parentheses

Appendix D. Most commonly consumed food items for Zambia (color coded by food group)

17-Sep 17-Nov 18-Jan 18-Mar 18-May FOOD GROUP KEY FOOD % FOOD % FOOD % FOOD % FOOD % CASSAVA 84 CASSAVA 88 CASSAVA 93 CASSAVA 91 CASSAVA 86 STAPLES CASSAVA 72 CASSAVA LEAVES 68 MANGO 85 MAIZE 49 GROUNDNUTS 82 DGLV LEAVES SWEET POTATO PUMPKIN SWEET POTATO MAIZE 44 40 51 46 ORANGE 48 LEGUMES LEAVES LEAVES LEAVES SWEET POTATO ORANGE SWEET TOMATO 35 GROUNDNUTS 39 39 GROUNDNUTS 37 39 NUTS AND SEEDS LEAVES POTATOES CASSAVA PUMPKIN WHITE SWEET GROUNDNUTS 32 TOMATO 33 35 36 38 ANIMALS LEAVES LEAVES POTATOES RAPE 24 MAIZE 29 MAIZE 31 BEAN LEAVES 32 MAIZE 23 DAIRY CHINESE ONIONS 24 MANGO 29 MUSHROOMS 26 OKRA 26 23 EGGS CABBAGE AMATUKU CASSAVA 23 MUSHROOMS 25 GROUNDNUTS 16 20 TOMATO 22 VIT. A (TILAPIA/FISH) LEAVES UMULEMBWE IMFUNGO (WILD 23 24 TOMATO 14 ORANGE 20 GUAVA 22 OTHER VEG KATALI (WILD) FRUIT) KABULANGETI AMASUKU IMILONGE IMILONGE SWEET POTATO 20 23 13 17 21 OTHER FRUIT (BEAN) (WILD LOQUAT) (CATFISH) (CATFISH) LEAVES 18-Jul 18-Sep 18-Nov 19-Jan 19-Mar 19-May FOOD % FOOD % FOOD % FOOD % FOOD % FOOD % CASSAVA 91 CASSAVA 88 CASSAVA 89 CASSAVA 93 CASSAVA 93 CASSAVA 90 CASSAVA PUMPKIN PUMPKIN GROUNDNUTS 67 GROUNDNUTS 58 55 86 86 GROUNDNUTS 78 LEAVES LEAVES LEAVES CHINESE CASSAVA 41 CASSAVA LEAVES 53 GROUNDNUTS 50 MUSHROOMS 56 GROUNDNUTS 55 49 CABBAGE LEAVES WHITE SWEET CASSAVA 36 RAPE 38 MANGO 49 GROUNDNUTS 53 47 CHICKEN EGGS 46 POTATOES LEAVES AMATUKU CASSAVA TOMATO 30 34 MUSHROOMS 45 47 TOMATO 34 RAPE 41 (TILAPIA/FISH) LEAVES AMATUKU AMATUKU AMATUKU MAIZE 21 TOMATO 32 39 MANGO 43 32 40 (TILAPIA/FISH) (TILAPIA/FISH) (TILAPIA/FISH) SWEET POTATO ORANGE 21 MAIZE 31 37 TOMATO 37 MAIZE 27 TOMATO 34 LEAVES UMULEMBWE AMASUKU AMATUKU SWEET POTATO Amataba BEAN LEAVES 20 27 35 31 24 33 KATALI (WILD) (WILD LOQUAT) (TILAPIA/FISH) LEAVES (Maize) MAIZE, SWEET AMATUKU 17 CHICKEN EGGS 16 MAIZE 32 POTATO 25 MILLET 23 SOYA 31 (TILAPIA/FISH) LEAVES CHINESE SWEET UMULEMBWE AMASUKU 15 CABBAGE and 15 TOMATO 27 24 ONIONS 23 POTATO 24 KATALI (WILD) (WILD LOQUAT) ONIONS LEAVES

Appendix E. Most commonly consumed food items for Malawi (color coded by food group)

17-Sep 17-Nov 18-Jan 18-Mar 18-May FOOD GROUP KEY FOOD % FOOD % FOOD % FOOD % FOOD % STAPLES 1 MAIZE 99 MAIZE 98 MAIZE 99 MAIZE 98 MAIZE 99 DGLV 2 TOMATO 71 TOMATO 74 TOMATO 77 TOMATO 76 TOMATO 80 LEGUMES 3 ONIONS 52 ONIONS 69 ONIONS 60 ONIONS 59 BEAN LEAVES 62 NUTS AND SEEDS BEAN 4 KABWILI (FISH) 41 KABWILI (FISH) 28 BEAN LEAVES 49 50 ONIONS 52 ANIMALS LEAVES YELLOW 5 MUSTARD 32 RAPE 25 MANGO 42 PUMPKIN 48 SWEET 51 DAIRY POTATO YELLOW SWEET UBAKA (SMALL BUYOLE WHITE SWEET 6 25 24 MUSTARD 19 47 38 EGGS POTATO GROUNDNUTS) (BEAN) POTATO MASIPENGILE MASIPENGILE 7 24 BOMA (BEAN) 22 SOYA 16 COW MILK 37 38 VIT. A (BEAN) (BEAN) PUMPKIN UBAKA (SMALL 8 BEAN LEAVES 22 BEAN LEAVES 18 KABWILI (FISH) 16 34 37 OTHER VEG LEAVES GROUNDNUTS) PUMPKIN MASIPENGILE 9 22 16 RAPE 15 GUAVA 33 BUYOLE (BEAN) 24 OTHER FRUIT LEAVES (BEAN) 18-Jul 18-Sep 18-Nov 19-Jan 19-Mar 19-May FOOD % FOOD % FOOD % FOOD % FOOD % FOOD % 1 MAIZE 99 MAIZE 100 MAIZE 100 MAIZE 100 MAIZE 100 MAIZE 100 WHITE SWEET WHITE SWEET 2 MUSTARD 72 TOMATO 88 TOMATO 84 MANGO 79 96 80 POTATO POTATO BEAN MASIPENGILE 3 TOMATO 70 ONIONS 75 MANGO 79 86 BEAN LEAVES 62 LEAVES 64 (BEAN) PUMPKIN UBAKA (SMALL 4 ONIONS 44 KABWILI (FISH) 73 ONIONS 67 TOMATO 84 62 54 LEAVES GROUNDNUTS) MASIPENGILE MASIPENGILE 5 31 35 KABWILI (FISH) 54 ONIONS 67 TOMATO 48 COW MILK 25 (BEAN) (BEAN) YELLOW SWEET PUMPKIN 6 21 SOYA 24 MUSTARD 42 24 GUAVA 48 AVOCADO 20 POTATO LEAVES PUMPKIN MASIPENGILE 7 19 POTATO LEAVES 24 34 CASSAVA 22 PUMPKIN 38 PAPAYA 18 LEAVES (BEAN) UBAKA (SMALL CAT'S UBAKA (SMALL 8 16 MUSTARD 20 SOYA 24 13 31 TOMATO 18 GROUNDNUTS) WHISKERS GROUNDNUTS) PUMPKIN 9 RAPE 16 COW MILK 15 16 AMARANTH 12 ONIONS 31 KABWILI (FISH) 17 LEAVES

Appendix G. Number of foods reported in each food group by round including the number of foods reported during Focus Group Discussions (FGDs), from which the dietary monitoring tool was based.

Flesh Foods Nuts and Other Other Staples Legumes Dairy Meat, Poultry Eggs DGLV Vit. A TOTAL Seeds Fish Veg Fruit and Insects Malawi 11 23 3 2 17 21 4 18 5 12 33 149 FGDs Zambia 8 9 3 1 22 19 3 15 6 8 22 116 Malawi 6 8 1 1 6 4 1 10 2 5 2 46 17-Sep Zambia 5 6 1 1 9 4 1 10 2 4 8 51 Malawi 8 11 1 2 5 3 3 11 3 6 2 55 17-Nov Zambia 3 4 2 1 10 5 1 10 2 6 6 50 Malawi 8 7 1 2 3 2 1 13 2 3 4 46 18-Jan Zambia 7 5 1 0 4 3 1 10 2 5 3 41 Malawi 5 8 2 2 5 2 3 12 5 4 9 57 18-Mar Zambia 6 3 1 1 6 5 2 8 4 7 4 47 Malawi 8 10 1 1 8 2 3 8 2 6 5 54 18-May Zambia 6 8 1 1 3 4 1 11 3 7 4 49 Malawi 7 8 1 2 3 4 1 7 2 6 2 43 18-Jul Zambia 6 5 1 1 5 3 2 10 1 4 5 43 Malawi 7 4 1 2 2 0 1 10 0 2 1 30 18-Sep Zambia 5 8 1 1 6 2 1 11 3 4 7 49 Malawi 7 5 1 0 3 2 1 11 1 3 3 37 18-Nov Zambia 4 7 1 1 8 4 2 11 3 6 7 54 Malawi 4 3 1 1 2 1 1 8 1 7 1 30 19-Jan Zambia 6 7 1 0 6 4 2 12 3 5 7 53 Malawi 5 7 1 1 2 0 1 7 1 5 2 32 19-Mar Zambia 6 6 1 0 6 4 2 11 3 4 5 48 Malawi 5 5 1 1 5 0 1 7 1 4 4 34 19-May Zambia 7 7 1 1 6 4 2 10 2 4 6 50