Agroecology and gender effects on rural households’ food insecurity and coping mechanisms in Borena District, Amhara Region, Ethiopia

Abstract

Background

Although national and international efforts had been made to reduce hunger, food insecurity continued to affect households in the Borena district. This study examined the food security status of rural households, its key determinants, and coping mechanisms.

Methods

Using a multi-stage sampling technique, 189 households were selected across different agroecological zones. Primary and secondary data were then collected through household surveys, focus group discussions, and key informant interviews. The data were analyzed using descriptive statistics, chi-square tests, and a binary logistic regression model.

Results

The study found that 62.6% of households were food insecure, with higher prevalence in tropical and alpine zones. Female-headed, older, illiterate, and large-sized households were more vulnerable, while farmland size, livestock ownership, off-farm income, irrigation use, food aid, and access to improved inputs reduced the risk of food insecurity. Households employed both short-term (erosive) coping strategies—such as reducing meals, borrowing, and consuming cheaper foods—and long-term (adaptive) strategies, including petty trade, cultivating short-season crops, receiving remittances, and seasonal migration. Food security challenges varied considerably across agroecological zones.

Conclusion

The study underlined the need for targeted policies that empower female-headed households, promote irrigation and off-farm income diversification, enhance access to agricultural inputs, and strengthen social protection and livelihood support in drought-prone areas.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40795-026-01275-7.

Introduction

Globally, approximately one billion people experience hunger annually, despite overall food production being sufficient to nourish the entire population [1]. The eradication of hunger is a significant challenge, and despite national and international efforts, limited progress has been made. As a result, the future standpoint for global food security is uncertain [2].

Food security is a multidimensional concept encompassing four interrelated pillars: availability, access, utilization, and stability [3]. Availability refers to the physical presence of sufficient quantities of food through domestic production or imports; access concerns households’ economic and physical ability to obtain food; utilization relates to the effective use of food for adequate nutrition, considering dietary diversity, health, and sanitation; and stability reflects the consistency of these three dimensions over time. Understanding household food security, therefore, requires examining these dimensions collectively, as shortfalls in any one pillar can lead to food insecurity.

According to the State of Food Security and Nutrition in the World (SOFI) 2025 report by FAO and partners, an estimated 673 million people (about 8.2% of the global population) experienced hunger in 2024 [4].This data highlights the worsening global food insecurity in recent years. Several key factors contribute to this critical situation, including a sharp rise in food prices due to economic constraints, conflicts leading to political instability, and unusual climatic conditions [5–7].

While hunger in Asia and Latin America decreased between 2021 and 2022, it continues to rise in Western Asia, the Caribbean, and every sub region of Africa [8]. Sub-Saharan Africa is one of the world’s most food-insecure regions, with millions suffering from chronic hunger and malnutrition [9]. In 2021, a total of 20.2% of the African population faced hunger [4].

Ethiopia is one of the most food-insecure countries in Sub-Saharan Africa, with a history of frequent droughts and famines. Both temporary and chronic food insecurity affect a significant portion of the population in the nation [10, 11]. Natural and man-made disasters are the primary causes of food insecurity in Ethiopian households [12].The two most commonly cited issues affecting food production, distribution, and access are drought and conflict [13]. High rates of population increase and poverty are also another key factors [14]. Ethiopia is one of the world’s poorest and most food-insecure nations, where a significant portion of the population lives in poverty, and drought claims more lives than any other issue [15]. The poorest rural households in Ethiopia frequently experience food insecurity and struggle to meet their basic needs [12].

The agricultural sector dominates Ethiopia’s economy, contributing more than 36% of GDP, 75% of export revenue, and 73% of job possibilities [16]. This sector mainly consists of small-scale farmers who mainly engage in agriculture reliant on rainfall [17].The irrigation sector represents merely 3% of agricultural produce [18]. Small and medium-sized farms provide a variety of food groups that greatly improve human nutrition [6, 19–21]. However, a substantial percentage of the world’s population experiencing ongoing food insecurity resides in households engaged in smallholder farming [22–25]. Erratic rain fall and low soil fertility are the main challenges faced by smallholder farmers [26, 27]. In addition to these challenges, land fragmentation, land degradation, crop failure, and high food prices have contributed to a significant prevalence of food insecurity in the country [28].

Borena District, located in the South Wollo zone in the northeast part of Ethiopia, is a highly drought-prone area where almost all districts face frequent food shortages. Compared to other parts of highland Ethiopia, this zone has slightly smaller average land holdings (15%t smaller), earns lower incomes, and is less food secure, primarily due to its greater dependence on Belg rainfall (a short rainy season) than other areas [29].

Several studies conducted in Ethiopia have examined food insecurity among rural households. Research by Bahiru et al. [30], Yohannes et al. [31], Masha et al. [32], Tigistu and Hegena [33], Melese et al. [34], Abdulla [35], and Tewdros and Fikadu [36] has provided an overall picture of food insecurity and its determinants at the district level. However, this overall assessment obscures the disparities in food insecurity across various agro-ecological zones within a single district.

This study wants to emphasize evaluating food insecurity status in different agro-ecological zones. A specific type of agro-ecology has significant implications, as it can either limit or facilitate the challenges faced and the potential cropping mechanisms available [37]. According to certain research, disparities in agro-ecological zones within specific areas of various low- and middle-income countries lead to significant differences in how households meet their nutritional needs and ensure food security [38]. Gender differences in society and how they affect the state of food insecurity constitute another weakness in the aforementioned studies. Thus, this study considers the perspective of women, who account for almost half of the global agricultural workforce, and play a vital role in household food security [39].

Although numerous studies have examined the determinants of household food insecurity in Ethiopia Abegaz [40], Eshetu [41], Yitayew & Seyoum [42], Nigussie & Bogale [43], Agidew & Singh [44], most focus on broad regional or national analyses, paying limited attention to how gender dynamics, agroecological variation, and asset distribution interact to shape food security outcomes. Food security often varies across agroecological zones due to differences in rainfall, soil fertility, and livelihood systems, while the gender of the household head influences access to resources, decision-making power, and coping strategies during food shortages.

Analyzing food insecurity through an agroecological lens is therefore crucial, as it helps identify zone-specific risks and design targeted interventions. The Borena District, a highly drought-prone area in the Amhara Region, was selected because recurrent climatic shocks, land degradation, and limited livelihood options have intensified household food insecurity. Despite these conditions, few studies have examined the combined effects of agroecological variation and gender on food security in this district. This study provides practical guidance for interventions such as irrigation development, livelihood diversification, and support for female-headed households, fills an academic gap by integrating agroecological factors into food security analysis, and informs policy for area-specific strategies in drought-prone regions.

To address these gaps, the study conducts a context-specific analysis that integrates gender and agroecological perspectives to examine household food security, its determinants, and coping mechanisms. By exploring the interactions between these factors, it provides localized, evidence-based insights that are often overlooked in broader studies and contributes to designing targeted interventions that enhance both gender equity and agroecological resilience in rural Ethiopia. Accordingly, the main objective is to analyze the food security status of rural households in the Borena District, along with key determinants and coping strategies, while explicitly considering gender differences and agroecological variations.

Study area and research methods

The study area

Borena District is situated at a distance of 578 kilometers northeast of the capital city of the country, Addis Ababa, and 180 kilometers southwest of Dessie. It is under the administration of the Amhara National Regional State as a part of the South Wollo Zone administration. The absolute location of the district lies between 10°34’00’’N to 10°53’00’’N latitudes and 38°29’00’’E to 39°57’00’’E longitudes (Fig. 1).

Fig. 1.

Study area map

This section provides a brief description of the district’s topography, climate, and soil characteristics, highlighting their relationship to the area’s physiographic aspects. The altitude of the district ranges from 500 m above sea level at the bottom of the canyon of the Abay River on the west side to 3700 m in the north-east corner. The Borena district is divided into four agro-ecological zones, with temperatures ranging from hot to cold. These are wurch (Alpine), which describes highlands that are between 3200 and 3700 m in elevation. Dega (temperate) refers to highlands found between 2300 and 3200 m. Woinadega (sub-tropical) describes highlands that range in elevation from 1500 to 2300 m. Kola (tropical) describes lowlands found between 500 and 1500 m. It has two rainy seasons: summer, the main rainy season, i.e., from June to September, and spring, a shorter and smaller rainy season. According to the Borena District Agriculture and Rural Development Office report for 2020, the annual rainfall is 1500 mm and the mean annual temperature is 23˚C. The study area has various soil types, the major soils being leptosols, cambisols, vertisols, andosols, and luvisols. Soils that cover much of the rural Borena District are leptosols, eutric cambisols, and eutric regosols.

According to the Borena District Agriculture and Rural Development Office report for 2020, there are seven main rivers in the district that flow to the Abay River. These are Endrias, Legekora, Kulbit, Boreda, Lege-Amara, Tula and Yeshum. The River Abay is a natural border separating the district from the East Gojjam Administrative Zone. Most of the tropical kebeles are found within this river valley. Due to the existence of different agro-ecological zones, the district has various flora reserves. Most of the plant species are found in Borena-Sayint Arehimeno National Park, which is a remnant of a forest reserve in the south Wollo. Wanza (Cordialafricana), Digita (Calpurnia aurea), Embis (Allophylus abyssinica), Shekori (Acanthus sennii), Tatesa (Rhus abyssinica) are the commonest plant types and yekok sar (Arthraxon prionodes), wajima (Trifolum spp), serdo Cynadon dactyl), senbelet, are the commonest grass species of the district.

Concerning the population composition, according to the Borena District Finance and Economic Development Office (BDFECO), it was estimated in 2020 that there are 155,594 people living in the district, with 78,461 males and 77,133 females. 90.2% reside in rural areas, whereas 9.8% live in urban areas. The population density of this district is 153.96 persons per km², which is greater than the zonal average of 143.58 people per km². According to the district report from 2021, the main economic activity of the population is rain-fed mixed farming. Maize, sorghum, and teff (Eragrostis tef) in Kolla; teff and wheat in Woinadega; wheat and barley, nug (Guizotia abyssinica), and sunflower are among the food crops in Dega parts, while vitch and chickpeas are common cereal crops in both Dega and Woinadega climatic zones. Onion, potato, carrot, cabbage, tomato, coke, coffee, banana, and sugarcane are produced in very small amounts where there is access to water for irrigation. The average land holdings of the farmers are too small (less than a quarter hectare), and they are known to be low in productivity. Raising livestock, which includes cattle, donkeys, sheep, goats, horses, and mules, is another essential chore that farmers are accustomed to.

Research methods

A mixed research approach was employed to complement quantitative and qualitative data collected through various methods. This approach allowed for a comprehensive understanding of household food security and coping mechanisms. A descriptive cross-sectional research design was used to assess the status and determinants of household food security at a specific point in time. The researcher focused on a defined population, systematically observing, describing, analyzing, and interpreting the findings to provide an in-depth understanding of the issue.

Sampling size and techniques

In the South Wollo Zone, there are 21 rural and 2 urban districts participating in the Productive Safety Net Program (PSNP). The study focuses on Borena District, one of these rural areas and a designated safety net corridor in Ethiopia, where all 34 kebeles are identified as food insecure and PSNP beneficiaries. Sample kebeles and households were chosen using both non-probability and probability sampling techniques. A multi-stage sampling method was employed for data collection. First, Borena District was purposefully selected due to its identification as a food insecure district practicing the PSNP. Next, kebeles were randomly selected from each agro-ecological zone within the district, and respondents were then randomly chosen from the gots (the smallest administrative units of a kebele) within those selected kebeles.

All the kebeles in the district are categorized into four agro-ecological zones: Kola, Woinadega, Dega, and Wurch. From each of these zones, one kebele was randomly selected, based on the assumption that households in similar agro-ecological areas share comparable experiences and face similar opportunities and constraints. The selected kebeles are Dokise from Kola, Debre-senbo from Woinadega, Dega Dibi from Dega, and Chiro-cadis from Wurch agro-ecological zones. The researchers employed Yamane’s method to determine the sample size [45] formula (Eq. 1).

1

Where,

n=Desired sample size

N=Target population of the study area (2648)

e=Margin of error (0.07)

Where n = desired sample size, N = target population of the study (2648) and e = margin of error (0.07). Given the homogeneity of the study population in terms of livelihood and socioeconomic characteristics the sample areas, a margin of error of 0.07 was utilized. n = 189.

Using systematic random sampling techniques, 189 household heads from the four rural kebeles were selected for the questionnaire survey. The sample size for each kebele (stratum) was determined using the proportional stratified sampling formula [46] (Eq. 2).

2

Where: nh=sample size of the strata (Kebele)

n*=Total sample to be taken,

N=total number of households

Nh=household size of the strata (Kebele)

Table 1 summarizes the sample sizes of households in each kebele.

Table 1.

Sample households by selected Kebeles & agro-ecologies

Selected Kebeles	Agro-ecological zones	Number of HHs in sample Kebeles
		Male HHH	Female HHH	Total HHH	Percent	Sample Household Heads
Chiro-cadis	Wurch	359	208	567	21.4	41
Dega-dibi	Dega	487	301	788	29.8	56
Debre-senbo	Woinadega	372	319	691	26.1	49
Dokise	Kolla	345	257	602	22.7	43
Total		1563	1085	2648		189

Table 1 indicates the sample households selected by kebeles and agro-ecological zones. A total of 189 household heads from the four rural kebeles were chosen for the questionnaire survey using systematic random sampling techniques.

Data sources and instruments

Both primary and secondary sources were utilized to gather data. Primary data were collected through household surveys, key informant interviews, and focus group discussions. Secondary data were obtained from relevant government offices and institutional reports, including agricultural production, demographic, and rainfall records from the district agriculture and rural development offices. Additional information on national and regional food security trends was sourced from reports by the Central Statistical Agency (CSA) and the Food and Agriculture Organization (FAO). These data were used to contextualize and support the primary household survey findings.

The study primarily utilized a survey questionnaire, supplemented by key informant interviews, focus group discussions, and field observations. The questionnaire included questions about the determinants of rural household food insecurity and coping mechanisms, with translations into Amharic for better understanding. A pilot study was conducted with 10% of the sample in a neighboring kebele prior to distributing the questionnaires to the selected households, allowing researchers to verify the appropriateness of the questions. Additionally, interviews were conducted with officials, including district agriculture and rural development experts, development agents, and kebele administrators.

Furthermore, focus groups were conducted with carefully selected heads of households. The discussions included elderly households of both sexes that shared specific characteristics, aimed at complementing the data collected through other methods. A total of 12 households participated in four focus group discussions, one in each kebele. Relevant brainstorming questions were posed to facilitate the exchange of ideas among participants. Additionally, field observations were incorporated throughout the research to ensure the validity of the information obtained.

Methods of data analysis

Both qualitative and quantitative techniques were employed to analyze the collected data. Surveys are coded, altered, entered, and analyzed using SPSS version 25 for quantitative techniques. The results are then tabulated, summarized, and interpreted to characterize the populations from the samples. The responses of the many stakeholders from interviews, focus groups, and field observations were described using the qualitative technique.

Descriptive statistics, including mean, standard deviation, frequencies, and percentages, were utilized to assess the quantitative data. Inferential statistics, such as the independent sample t-test and chi-square test, were employed to examine relationships between variables. The independent sample t-test was used to determine whether the mean scores of two independent samples differed significantly. For comparing categorical or dummy variables, the chi-square test was applied. Additionally, a binary logistic regression model was used to identify the variables influencing household food security.

Analysis of food security status and determinants food insecurity

A modified version of the Household Food Balance Model (HFBM), originally adapted by [47] from the FAO Regional Food Balance Model and subsequently used by various researchers in this field [44, 48] was employed to calculate per capita food availability (Eq. 3). The Household Food Balance Model (HFBM) was employed to assess household food security status because it provides an objective, consumption-based measure of food availability at the household level. Unlike perception-based indicators such as the Household Food Insecurity Access Scale (HFIAS) or the Food Consumption Score (FCS), which rely on recall and subjective responses, the HFBM quantifies the actual food available for consumption by considering own production, purchases, and food transfers. This approach is particularly suitable for rural agrarian communities like Borena, where subsistence farming is dominant and market participation is limited. The HFBM thus allows for a more accurate estimation of calorie availability per adult equivalent and helps identify households falling below the minimum food energy requirement. Its use overcomes the limitations of other methods that often overlook seasonal variation and self-produced food sources, making it a reliable tool for this context.

In line with the FAO’s [3] conceptual framework, food security is understood as a multidimensional construct encompassing four interrelated pillars: availability, access, utilization, and stability. The modified HFBM primarily captures the availability pillar by quantifying the actual food accessible for household consumption through production, purchase, aid, or transfer. However, indicators associated with the other three pillars—such as access (income, credit, off-farm activities), utilization (diet adequacy through calorie estimation), and stability (seasonal variation and production reliability)—are incorporated within the explanatory variables used in the econometric model. This integration ensures a comprehensive assessment of household food security consistent with the FAO framework.

This model assesses the balance between food grains and waste within a household. It was chosen because it effectively estimates diet adequacy or inadequacy in terms of caloric availability. Given that data sources are household heads, methods like dietary diversity are not suitable. Dietary diversity typically requires asking one or more household members about different food items consumed over a specified period, without recording quantities. Consequently, if it is not feasible to inquire about the frequency of consumption of specific quantities, estimating the extent of dietary inadequacy in terms of caloric availability becomes impossible [49–52].

3

Where,

NGA = Net grain available/year/household

GP = Total grain produced/year/household

GB = Total grain bought/year/household

FA = Quantity of food aid obtained/year/household

GG = Total grain obtained through gift or remittance/year/household

HL = Post harvest losses/year

GU =Quantity of grain reserved for seed/year/household

GS =Amount of grain sold/year/household

GV =Grain given to others within a year

Econometric model

The binary logistic regression model was chosen because the logistic distribution has somewhat fatter tails, meaning that the conditional probability of Pi approaches 0 or 1 more slowly in the logit model compared to the probit model. Additionally, the logit model is mathematically simpler than the probit model [53]. (Eq. 4). The model can be written as.

4

Where Y is the observed food security status of a household,

Xi is the household set of explanatory variables,

℮ is an exponential term; Pi is the probability of household i being food secure. It is 1 if a household is food secure and 0 if it is insecure.

Zi is a function of n-explanatory variables (Xi), which can be expressed in linear form as:

Based on equation three, the probability of a household being food insecure (1 – Pi) (Eq. 5) can be expressed as follows:

5

The odds of can be stated as (Eq. 6):

6

Then is the odds ratio in favour of food security.

It is the ratio of the likelihood that a household would be food secure (Pi) to the likelihood that a household would be food insecure (1-Pi). Finally, taking the natural logarithm of Eq. 3 and assuming linearity produces (Eq. 7):

7

where Li is the logarithm of the odd ratio, which is assumed to be linear for both variables and parameters. If the disturbance term is introduced, the logit model in Eq. 6 is represented by Eq. 8:

8

In Eq. 8, the terms βi are parameters to be estimated, and X1 to Xn are explanatory variables such as: sex of the household head, age of the household head, educational level of the household head, marital status of the household head, family size, farm land size, livestock ownership, fertilizer use, irrigation use, off-farm income, and access to credit, respectively.

Multicollinearity test

Prior to running the logit model, a multicollinearity test was made to check whether there exists a strong association between explanatory variables because multicollinearity will seriously affect the parameter estimates. The simultaneous existence of the two variables will either strengthen or lessen their separate impacts if multicollinearity is found to be significant. On the other hand, a specification bias will result from inaccurately excluding significant interaction factors [54].

Accordingly, a Variance Inflation Factors (VIF) (Xi) technique will be employed to detect the problem of multicollinearity for continuous variables [47]. Variance Inflating Factor (VIF) was used to check for multicollinearity problems among and between explanatory variables. Each selected continuous explanatory variable (Xi) is regressed on all the other continuous explanatory variables; the coefficient of determination (Ri²) was constructed in each case. If an approximate linear relationship exists among the explanatory variables, then this will show up as a large value for Ri² in at least one of the test regressions. A popular measure of multicollinearity associated with the VIF (Xi) (Eq. 9) is defined as:

9

Where Ri is the squared multiple correlation coefficient between and other explanatory Xi variables. The larger the value of VIF, the more troublesome it is. As a rule of thumb, if the VIF of a variable exceeds 10, the variable is said to be highly collinear. Similarly, for categorical variables, contingency coefficient tests are employed using the following formula (Eq. 10):

10

Where CC is the contingency coefficient, X² is the Chi-square value, and n is the total sample size. For categorical variables, if the value of the contingency coefficients is greater than 0.75, the variable is said to be collinear [55].

Dependent variable

the outcome variable is the households’ food security status.

Independent variables

these are the predictor variables that are believed to be determinants of food security. Thus, it is estimated that these factors and the dependent variable, which is detailed in Table 2 below, will correlate.

Table 2.

Description of independent variables in the study

Variables	Description	Measurement	Values
SEXHHH	Sex of household head	Male = 1, female = 2	+/-
AGEHHH	Age of the household head	Years	-
EDUCHH	Educational attainment of the household head	Number of grades completed	+
HHSIZE	Household size	Number	-
LIVSTOC	Livestock ownership	Tropical Livestock Unit	+
FARMSIZ	Farmland size	Hectare	+
IRRIGUSE	Use of irrigation	User = 1, non-user = 2	+
UFERTZR	Use of chemical fertilizer	User = 1, non-user = 2	+
OFFINCO	Participation in off/non-farm activities	Participant = 1, non-participant = 2	+
FOODAID	Food aid	Beneficiaries = 1, Non-beneficiaries = 2	+
IMPUSAGE	Improved seed usage	Users = 1, non-users = 2	+
CREDIACC	Access to credit service	Yes = 1, No = 2	+/-
REMITTANCE	Remittance receives	Yes = 1, No = 2	-

Analysis of agro-ecology and gender effects on household food insecurity

The study applies cross-tabulation and chi-square tests to evaluate the effect of gender and agro-ecology on household food insecurity status in the community. These tests help to determine the relationship of food insecurity status with agro-ecology and gender in a statistically meaningful way. At first, expected counts are computed assuming that food insecurity has no associations with gender or agro-ecology. The chi-square test is then used to assess the level of significance of the difference that was established between the expected counts and the observed ones. Furthermore, we quantify the strength of the gender and agro-ecology effect on household food insecurity status using metrics like Phi and/or Cramer’s V. Akoglu [56], which offers a framework for analyzing these metrics, classifying the associations’ strengths as very strong (> 0.25), strong (0.15–0.25), moderate (0.10–0.15), weak (0.05–0.10), or no/very weak (0-0.05).

In addition to quantitative household survey data, qualitative information was collected through focus group discussions (FGDs) and key informant interviews (KIIs) with community leaders, extension agents, and local administrators. The qualitative data were analyzed thematically to complement and contextualize the quantitative findings, particularly regarding coping mechanisms and perceived causes of food insecurity.

Result and discussion

The results of this study is presented in three subsections: the first section addresses household food security status, the second focuses on the determinants of food insecurity, and the third explains the coping mechanisms employed by households facing food insecurity. Out of 189 questionnaires, two were excluded during the data coding process due to issues with the responses, resulting in a response rate of approximately 99%.

Characteristics of the study community with their food insecurity status

Accordingly, descriptive statistics are performed for both continuous and categorical variables, including the sex of the household, the age of the household, the educational status, household size, livestock ownership, farmland size, use of irrigation, use of chemical fertilizer, participation in off-farm or non-farm activities, food aid, and improved seed usage. Finally, to determine whether there was a significant difference between the mean scores of two independent samples, an independent-sample t-test was employed. The Chi-square test was used to compare categorical or dummy variables for both food-secure and insecure households, respectively. Regarding the age of the household head, the minimum age of the household heads is 21 years but the maximum age of the respondents is 67. The average age of food-secured households was (M = 40.56, SD = 8.662) years, which is lower than that of insecure households (M = 45.06, SD = 9.482). This implies that food insecurity in households increases with the age of the household head (Table 3). Moreover, the t-test result (t-test = 2.51, P < 0.05) shows that there exists a statistically significant association between the age of the household heads and the food security status of the families. Concerning the education level of the household head, Table 3 shows that food secured household heads have a better educational background (M = 5.02), SD = 3.852) than those foods in secured household heads (M = 3.21), SD = 3.248). Above all, the t-test result (t-test=-5.82, p < 0.01) confirmed that there existed a statistical significance between the educational level of the household heads and the food security status of the whole family. This may be because education empowers men to acquire basic knowledge not only on how to use resources in general but also on the use of agricultural technologies. The family size of the household head is also shown in Table 3. The minimum and maximum family sizes of the respondents are three and nine, respectively. The average family size of the food insecure households (M = 6.54, SD = 5.674) is much greater than the food insecure households (M = 4.45, SD = 2.213). Apart from this, the t-test result (t-test = (-3.96), p < 0.01) also shows a significant statistical relationship between family size and the food security status of the household. From this, we can conclude that as the family size increases, the food security status of the household deteriorates. The focus group discussions and key informant interviewees also mentioned family size as a cause for food insecurity in the study area. As far as the size of cultivated land is concerned, the minimum and maximum land holding sizes of the respondents are 0.12 and 2.13 hectares, respectively. Food-secured households have a greater land size (M = 1.146, SD = 0.6789) than food-insecure households (MD = 0.26, SD = 0.481) (Table 3) The t-test result (t-test = 5.12, p < 0.05) indicates that there is a significant statistical correlation between the land size owned and the food security status of the households. Most of the respondents replied that the slope of their plot of land is steep and susceptible to soil erosion. Apart from this, the fertility of their cropping land is also very poor due to the high topographic configuration of the area and frequent tilling practices year after year without fallowing.

Table 3.

Food security status of households and continuous variables

Variables	Food Security Status
	Food Secure		Food Insecure		t-value	p-value
	Mean	SD	Mean	SD
Age of household head	40.56	8.662	45.06	9.482	2.51	0.010**
Level of education	5.02	3.852	3.21	3.248	5.82	0.000***
Family size	4.45	2.213	6.54	5.674	3.96	0.000***
Number of Livestock	4.322	2.458	2.723	2.234	5.86	0.000***
Land Size	1.146	0.6789	0.26	0.481	5.12	0.020 **

** and *** indicates that the coefficients are significant at 5% and 1%, respectively

The average livestock holding size of the food insecure households (M = 2.7228, SD = 2.2341) is much lower than the average number of livestock owned by the food secured households, i.e., (M = 4.32154, SD = 2.4583). Besides, the t-test result (t-test = 5.86, P < 0.01) also indicates that there is a statistically significant mean difference between livestock holding and food security status of the households (Table 3). From this, we can infer that as the livestock holding size increases, the food security status of households gets better and better. This information is also supported by the responses of the key informant interviews and focus group discussions, which pointed out that livestock are the basic vehicles of the community in that they serve as food, income, and inputs of production and productivity by their own power, and they provide animal dung to increase the fertility of their agricultural land.

The study examines frequency distributions of variables in addition to employing descriptive statistics for the analysis of respondents’ attributes. This enables a thorough analysis of the data, providing insights into the distribution patterns and occurrences of different variables.

As far as the sex structure of the household head is concerned, as depicted in Table 4, 28.9%of female-headed households are in the food insecure category, while only 13.9% of them are in the food secure category. However, the proportion of male-headed households in the food secure category is 25.1%, which is smaller than the proportion in the food insecure category, which is about 32.1%. In both cases, the food insecure households are much higher than the food secure households; however, the deviations of the female group are relatively far apart from those of the male group. In addition to this, the chi-square analysis (x² = 5.412, p < 0.05) indicates that the food security status and sex of the household head.

Table 4.

Summary of descriptive statistics results for categorical variables

Variables		Food Secure		Food Insecure		Total		X2	P-value
		Frequency	%	Frequency	%	Frequency	%
Sex	Male	47	25.1	60	32.1	107	57.2	5.412 **	0.020
	Female	26	13.9	54	28.9	80	42.8
Irrigation	User	32	45.7	41	35.0	73	39.0	6.647***	0.010
	Non-user	38	54.3	76	65.0	114	61.0
Fertilizer	User	53	75.7	53	45.3	106	56.7	16.501***	0.000
	Non-user	17	24.3	64	54.7	81	43.3
Off Farm	Participant	46	65.7	40	34.3	86	46.0	12.468**	0.030
	Non-participant	24	34.1	77	65.8	101	54.0
Improved seed	User	42	60.0	35	29.9	77	41.2	16.366***	0.000
	Non-user	28	40.0	82	70.1	110	58.8
Food Aid	Beneficiaries	48	68.6	38	32.5	86	46.0	8.32**	0.020
	Non-Beneficiaries	22	31.4	79	67.5	101	54.0

Regarding irrigation use by household heads and food security status, as indicated in Table 4, out of food secure respondents, 45.7% of them had utilized an irrigation scheme, while the majority of food insecure households (65%) did not use irrigation. Additionally, the Chi-square analysis (x² = 6.647, p < 0.01) revealed a statistically significant association between food security status and irrigation utilization among farming household heads. This suggests that an increase in the use of irrigation schemes improves the probability of achieving food security for farming households.

Fertilizer application by households is also another important aspect of the study. As shown in Table 4, the proportion of households using artificial fertilizers in the food-secure category (75.7%) is greater than in the food-insecure category (45.3%). Conversely, the proportion of households that did not use artificial fertilizers is greater in the food insecure category (54.7%) compared to their proportion in the food secure category (24.3%). Additionally, the Chi-square analysis (x²= 16.54, p < 0.01) indicated a statistically significant association between food security status and fertilizer application among farming household heads. This suggests that the use of fertilizer increases the likelihood of farming households achieving food security.

Regarding off-farm activities and the food security status of households, the study results indicates (Table 4) among the sampled respondents, a higher percentage of farming households participating in non-farm income activities are food secure (65.7%) compared to food-insecure respondents than food-insecure respondents (34.3%). Conversely the proportion of households in the food insecure category who did not participate in off-farm activities (65.8%) is greater than their proportion in the food secure category (34. 2%). Additionally, the Chi-square analysis (x² = 12.468, p < 0.05) shows a statistically significant association between food security status and the non-farm income-generating activities of household heads. This implies that farming households engaged in non-farm income-generating activities have a higher probability of achieving food security.

Another important variable is the use of improved seed and the food security status of households. As shown in Table 4, the proportion of households in the food secure category that utilized improved seeds (60%) is significantly higher than that in the food-insecure category (29.9%). Similarly, the proportion of households that did not use improved seeds in the food insecure category (70.1%) exceeds that in the food-secure category (40%). Additionally, the Chi-square analysis (x² = 16.366, p < 0.01) indicates a statistically significant association between food security status and improved seed usage.

The effect of food aid on the food security status of households is also considered in this analysis. As summarized in Table 4, 68.6% of food-secure respondents are beneficiaries of food aid, while 31.4% are non-beneficiaries. In contrast, the majority of food-insecure households (67.5%) did not have access to food aid. Additionally, the Chi-square analysis (x2 = 8.32, p < 0.05) indicates a statistically significant association between food aid and the food security status of farming household heads. This suggests that food aid has a more positive impact on the likelihood of farming households being food secure compared to those that do not receive food aid.

Determinants of the household food insecurity

Before analyzing the data using a binary logistic regression model, a multicollinearity test was made to check whether there existed a strong association between explanatory variables because multicollinearity will seriously affect the parameter estimates. Thus, the variance inflation factor (VIF) and contingency coefficient (CC) are used to test the degree of multicollinearity among the continuous and categorical variables, respectively. As a rule of thumb, if the VIF of a variable exceeds 10, there is multicollinearity. The value of contingency coefficients ranges between zero and one, indicating no association if it is zero and strong association if it is one. To avoid serious problems of multicollinearity, it is quite essential to omit the variable with a value 10 or a higher from the logit analysis [53].

The VIF for five continuous variables is found to be less than 10, implying that there exists no serious problem of multicollinearity. In the same way, the contingency coefficient for six categorical variables indicates that there is no problem of multicollinearity. In this model, household food security is regarded as a dependent variable with a value of 1 if households are food secure and 0 if households are food insecure.

To identify the determinants of food security status among households in the study area, a binary logistic regression model was employed. In this model, the food security status of households served as the dependent variable. The independent variables included in the analysis are: sex of the household head, age of the household head, educational attainment of the household head, household size, livestock ownership, farmland size, use of irrigation, use of chemical fertilizer, participation in off-farm or non-farm activities, food aid, improved seed usage, access to credit services, use of remittances.

Before running the regression model, the appropriateness of the model was tested. The Log-likelihood ratio test (42.65), Pearson chi-square (186.49), Cox and Snell R2 (0.830), Nagelkerke R2 (0.796), H-L model significant level test (11.64) are tested and checked. The model was fitted to this data. The result of the model is shown in Table 5.

Table 5.

The binary logistic regression determinants of food security

Variables	B	S.E.	Wald	Sig.	Exp(B)
Sex of HHH	1.256	1.034	16.278	0.004***	2.581
Age of HHH	-1.400	0.067	26.356	0.036**	0.246
Educational HHH	1.753	0.818	31.738	0.019**	2.194
Household size	-2.340	0.667	43.291	0.000***	0.096
Livestock ownership	4.218	1.205	56.253	0.000***	8.457
Farmland size	0.269	0.042	44.940	0.0430**	2.919
Irrigation	2.340	0.667	1.291	0.004***	10.948
Fertilizer	1.218	1.205	12.253	0.030**	6.348
Food aid	0.97	0.087	22.67	0.0035***	5.92
Improved seed	3.011	0.010	16.195	0.001***	4.236
Off-farm Income	0.934	0.858	1.185	0.000***	9.545
Constant	3.343	1.979	2.853	0.9		12.865

***and ** shows significance at the level of 1% and 5% respectively

Table 5 presents the logistic regression coefficient, Wald test results, and odds ratio for each of the predictors. Thirteen variables are hypothesized to explain factors affecting the food security level of households. Out of these variables, eleven are identified as significant. According to the model, sex of households, household age, educational level, household size, livestock holding, farm land size, use of irrigation, use of chemical fertilizer, participation in off-farm or non-farm activities, food aid, and improved seed usage are identified as the variables that significantly affect households’ food security.

Sex of the household head

The sex of the household head is positively associated with the food security situation in the study area at a 1% significance level. The odds ratio of 2.581 indicates that the likelihood of a household being food-secure is nearly 2.6 times higher for males compared to females. This disparity may be due to women’s limited access to education, productive resources, and off-farm opportunities, as well as cultural norms restricting their economic participation [57, 58].

This study’s findings contrast with those of Mohamed and Ibrahim [59], who reported that males were more likely to be food insecure, with 56.1% of the insecure sample being male compared to 43.9% female. Similarly, Francisco et al. [60] noted that female-headed households had lower probabilities of experiencing food shortages following an extreme event than male-headed households.

Age of the household head

The level of food insecurity increases with the age of the household head, indicating that families led by older individuals are more food insecure than those led by younger heads. Thus, the age of the household head has an inverse relationship with the food security status at a significant level of 5%. The coefficient for the age of the household head was negative, indicating that an increase in age reduces the likelihood of being food insecure, holding other factors constant. This implies that older household heads are more likely to be food secure than their younger counterparts. A possible explanation is that older heads often possess more farming experience, stronger social networks, and better resource management skills, which enhance their ability to secure the household food supply.

This finding is consistent with studies by Gebre [61] and Bogale & Shimelis [62], in Ethiopia, which reported a negative relationship between age and food insecurity due to accumulated experience and knowledge. Similar results were also found by Arene and Anyaeji [63] in Nigeria. However, Haile et al. [64] identified an opposite relationship, noting that older age may reduce productivity and adaptive capacity, thereby increasing the risk of food insecurity in certain contexts.

Level of education

The level of education has a positive relationship with household food security at a 5% significance level, aligning optimistic prior expectations. The calculated odds ratio of 2.194 indicates that, the likelihood of a household becoming food secure increases by approximately 2.2 times for each level of educational attainment of the household head. This may be because education equips individuals with essential knowledge on resource management and provides farmers with information about agricultural technologies, modern inputs, optimal times for ploughing and harvesting, and the use of irrigation schemes to diversify their livelihood options. This study aligns with the findings of Magaña-Lemus et al. [65], who also reported that all levels of education are positively and significantly associated with improved food security and reduced food insecurity.

Household size

As hypothesized, this variable has an inverse relationship with the food security status of households. The coefficient for household size was negative and statistically significant (B = -2.340, p < 0.01, Exp(B) = 0.096), indicating that larger households are more likely to be food insecure. This may be due to the increasing consumption needs associated with a larger family size while available resources remain limited.

This finding aligns with previous studies by Bogale and Shimelis [62], and Ayalneh [66], which reported that household size negatively affects food security. Conversely, studies by Kidane et al. [67] and Amaza et al. [68] found that larger households may be more food secure if additional members contribute productive labor or income. This suggests that the impact of household size depends on household composition and resource availability.

Livestock ownership

Livestock ownership is positively correlated with the food security status of the household at a significant level of 1%. The result of this variable is congruent with what was hypothesized. The odds ratio of 8.457 implies that for every one unit of increase in the TLU of a household head, the likelihood that a household becomes food secure increases by approximately 8.5 times. This may be due to the current market value of livestock, which is extremely high, which helps livestock owners to sell their cattle and buy food crops. This study is in conformity with Aynalem and Mada [69] concluded that homes with a large number of livestock are probable to become food secure significantly and negatively at a less than 1% probability level.

Farm land size

It has a positive relationship with the food security status of the household at a significant level of 5%. This variable agrees with the hypothesis that households with a larger land size are more food secure than those with a smaller land size. Other things remaining constant, the odds ratio of 2.919 indicates that for every one hectare of increase in the land size of a household, the likelihood that a household becomes food-secure increases by approximately 3 times. This study is in congruent with Dube et al. [70], who found that increased in cultivated land by one hectare, decreased the probability that the household was food insecure by 33.1%.

Access to irrigation

The study reveals that irrigation has a positive relationship with household’s food security at a significance level of 1% and also in line with what was hypothesized. The odds ratio of 10.948 implies that the odds that a household becomes food-secure is approximately 11 times greater for irrigation user households compared to households that did not practice irrigation. This implies that irrigation provides not only supplementary food choices but also a cash crop that increases the income and purchasing power of the household. This study is in line with Sani and Kemaw [71], who found that irrigation affects household’s extent of food energy intake. Observations during field visits also revealed limited use of irrigation and soil conservation practices, which substantiates the statistical result showing that households without irrigation access were more food insecure.

Fertilizer application

This variable is positively related with the food security status of households at a significance level of 5% implying that households that use artificial fertilizer are more food-secure than those who did not. The odds ratio of 6.348 indicates that the odds that a household becomes food-secure is approximately 6 times greater for households that apply fertilizer in their farmland compared to those households that did not apply fertilizer in their farmland.

Nevertheless, the cost and accessibility of fertilizer is the major hindrance factor that hampers farmers not to use artificial fertilizers as it was explained in the focus group discussions. This study is contradicts with Teklay and Solomon [72] who found that the probability of households’ food insecurity increases by about 68% if the households use fertilizer as a farm input, holding all other things constant, but it is in line with Tsegamariam and Wakjira [73], who found that fertilizer use and food insecurity have negative relationship, i.e., the mean of fertilizer use is 1.55 quintal with a maximum of 4 quintals and a minimum of zero.

Food aid

Food aid is positively associated with the food security status of the study area, with a significance level of 1%. The odds ratio of 5.92 indicates that the likelihood of a household achieving food security is nearly six times higher for food aid beneficiaries compared to non-beneficiaries. This result is in line with the findings of Agidew and Singh [44].

Use of improved seeds

This variable is positively associated with the food security status of households at a 1% significance level, implying that households using improved seeds are more food secure than those that do not. The odds ratio of 4.236 indicates that the likelihood of a household achieving food security is about 4.2 times higher for users of improved seeds compared to non-users. This suggests that the adoption of improved seed varieties enhances crop productivity and income diversification, thereby reducing vulnerability to food shortages. This finding is consistent with previous studies in Ethiopia and other parts of sub-Saharan Africa, which reported that households adopting improved seeds were significantly more likely to be food secure Tura et al., [74]; Kassie et al., [75]; Abebe et al., [76]; Amare & Simane [77]. Evidence from focus group discussions and key informant interviews in the study area also supported this result, highlighting that improved seed use contributes to better yields and resilience against climate variability.

Participation in off-farm activities

Off-farm activities in the study area are both in-cash and in-kind that can help households to generate income and supplement their food deficiency. This variable is found to be positively related to the food security status of the household at a significant level of 1%. The odds ratio of 9.545indicates that the odds that a household becomes food-secure is 9.5 times greater for households that participated on off-farm activities compared to those that did not participate on off-farm activities. This study is inconsistent with Abdulla [35] who found off-farm activity is positively related to household food security.

The qualitative results provide valuable insights that complement the quantitative findings on the determinants, causes, and coping mechanisms of household food insecurity in the study area. Participants consistently cited large family size, limited livestock holdings, and small land sizes as major contributors to food insecurity. Livestock were described as “the backbone of rural livelihoods,” providing food, income, and manure for soil fertility; their decline was seen as directly weakening household resilience.

Households food insecurity coping mechanisms

This section presents the major food shortage coping mechanisms employed by households to cope up with food shortfalls. In this study, a variety of coping mechanisms implemented by the food secured and insecure households both the initial and later stages of food shortfalls are identified. The study employed simple descriptive statistics frequencies to examine the coping strategies employed by households. A total of 11 coping mechanisms are employed by the sampled households. As the survey result (Table 6) shows that the main coping mechanisms used by households are petty trade, reducing number and size of meal per day, harvesting crops that have short growing seasons, eating less preferred and cheap foods, borrowing cash or grain, receiving remittance and food aid, working as a daily laborer, and others.

Table 6.

Coping strategies practiced by households during food crisis in the study area

Coping Strategies	Food Secured (n = 70)		Food Insecure (n = 117)		Total (n = 187)
	Frequency	Percent	Frequency	Percent	Frequency	Percent
Borrowing cash or grain	5	7.1	85	72.6	90	48.1
Working as a daily laborer	3	4.3	76	65.0	79	42.2
Send children to work	2	2.9	48	41	50	26.7
Harvesting crops that have short growing seasons.	20	28.6	82	70.1	102	54.5
Seasonal migration	4	5.7	34	29.1	38	20.3
Reducing number and size of meal per day	9	12.9	93	79.5	102	54.5
Receiving remittance and food aid	12	17.1	68	58.2	80	42.8
Sale of livestock	2	2.9	72	61.5	74	39.6
Participate in petty trade	54	77.1	67	57.3	121	64.7
Eating less preferred and cheap foods	12	17.1	80	68.4	92	49.2
Withdrew children from school	6	8,9	70	59.8	76	40.6

According to respondents’ response reducing number and size of meal per day (79.5%) and borrowing cash or grain (72.6%) are the major coping mechanisms of food insecure households in the study area. According to the focus group and key informant interviewees responses, borrowing money from local wealthier households had a severe impact on the food security status of the household because if one borrows 1000 birr during harsh times of food shortage, he will return 2000 birr for the next year; the double payment rate continues until the farmer finishes paying the total money. This further aggravates the food insecurity of the rural community not only in the present but also in the future. Borrowing food from locally wealthier farmers had an impoverishing impact on the poor because there existed high return rates, i.e., if one borrows a kilogram of food crop, he will return two kilograms of the same food crop in the next production season.

Households in the Borena district employ a variety of coping mechanisms to deal with food insecurity. These strategies can be broadly categorized into short-term erosive and long-term adaptive responses.

Short-term (Erosive) Coping Mechanisms

These include actions that help households survive immediate food shortages but undermine their long-term livelihood sustainability. Such strategies were more common among food-insecure households. Examples include reducing the number and size of meals (79.5%), eating less preferred foods (68.4%), borrowing cash or grain (72.6%), selling livestock (61.5%), sending children to work (41%), and withdrawing children from school (59.8%). These responses provide temporary relief but deplete household assets, reduce human capital, and perpetuate vulnerability—findings consistent with Maxwell et al. [78] and Devereux [79], who note that erosive coping mechanisms are indicators of severe food insecurity and livelihood stress.

Long-term (Adaptive) Coping Mechanisms

Adaptive strategies involve efforts to adjust livelihood systems to improve resilience against future shocks. These were more commonly practiced by food-secure households, including engaging in petty trade (77.1%), harvesting short-season crops (28.6%), receiving remittances and food aid (17.1%), and seasonal migration (5.7%). Such strategies diversify income sources and enhance adaptive capacity. This aligns with the findings of Dercon [80] and Little et al. [81], who emphasize diversification and migration as key long-term adaptation strategies in rural Ethiopia.

Overall, the evidence indicates that food-insecure households rely heavily on short-term erosive mechanisms that aggravate poverty and vulnerability, whereas food-secure households adopt adaptive strategies that enhance resilience and livelihood sustainability. Strengthening access to sustainable livelihood opportunities is therefore critical to reducing dependence on erosive coping behaviors in the Borena district.

According to focus group discussants, borrowing money or grain from wealthier households and reducing meal frequency were common coping strategies during food shortages—verifying the quantitative findings that these were the most frequently used responses. Key informants also emphasized that limited access to irrigation water and declining soil fertility were major factors reducing agricultural productivity, consistent with regression results identifying farmland size and input use as significant determinants of food security.

The coping strategies analyzed in this study were limited to those most frequently reported by respondents during household surveys, focus group discussions, and key informant interviews. While other coping mechanisms—such as consuming lower-quality or cheaper food, using food held for the next season, feeding children at the expense of adults, or skipping meals—are recognized in the broader literature (e.g., [relevant citations]), these were rarely mentioned by participants in the Borena context. This suggests that households primarily relied on the identified coping options, reflecting local patterns of food insecurity response and cultural preferences. Accordingly, the analysis was confined to the strategies most representative of the study area.

The effects of agro-ecology and gender on the household food insecurity

The gender of the household and the agro-ecological zones they live in frequently have a significant impact on the level of food insecurity in the household (specially in Ethiopia). In this particular instance, 47% of study participants are female, and 57% of study participants are male household heads. Figure 2 illustrates that while the number of food-insecure households is higher in Kolla and Wurch, the number of food-secure households is significantly higher in the Woinadega and Dega areas.

Fig. 2.

The variation of food secure insecure households with their agro-ecology

Within the research topic, the purpose of this study is to look into how gender and agro-ecology affect household food security. As shown in Table 7, cross-tabulation and Chi-square testing were used to accomplish this. Regarding the case of gender, it was noted that the counts in both food-secure and food-insecure households varied from the expected values. The expected counts show what one would expect in the absence of any relationships. In other words, the expected and observed counts would agree closely if gender had no bearing on the level of food insecurity.

Table 7.

Food security status of the household-by-household sex and agro-ecology

Variables	Measures	food secured		food insecure		Total
		Count	Expected Count	Count	Expected Count	Count	Expected Count
sex	male	47	41.8	60	65.2	107	107
	female	26	31.2	54	48.8	80	80
Total		73	73	114	114	187	187
Agro-Ecology	Kolla	11	16.4	31	25.6	42	42
	WainaDega	25	18.7	23	29.3	48	48
	Dega	29	21.9	27	34.1	56	56
	Wurch	8	16	33	25	41	41
Total		73	73	114	114	187	187

In this case, Table 7 shows that the number of homes experiencing food insecurity is significantly higher for both males and females when compared to households that are secure. Nonetheless, compared to males, the percentage of secure and insecure families among females is far higher. Nonetheless, the comparison of the percentage of secure and insecure families among the males and females has shown a much bigger difference. According to Table 7, roughly 49 females are from food insecure households and 31 of them are from food secure households. Accordingly, 65 males are food insecure households and 42 of them are food secure households. As shown in Table 7, the observed and expected counts differ from one another. The important question here is whether the variation between expected and observed counts is significant or not.

Accordingly, since no cell has an expected count lower than 5%, the chi-square test indicates that the test’s assumption is not violated. 31.23 is the bare minimum predicted count. With one degree of freedom and a p-value of 0.113 and the Pearson chi-square value of 2.511 suggests that the outcome is not statistically significant. This implies that although there is an association between gender and the level of household food insecurity, the correlation is not high enough to support the premise. The idea that gender influences the level of households’ food insecurity is therefore unsupported by the data.

A comparable assessment of agro-ecology was carried out. According to Table 7, food-secure households in the Kolla and Wurch agro-ecological zones have higher expected counts than the observed ones. Compared to the expected count for the Woinadega and Dega agro-ecological zones, the observed count in the group of food-secure households is significantly higher. However, the opposite pattern is found in the group of households that experience food insecurity. The cross tabulation also shows the size of the discrepancy between the expected and observed counts of households in the food secure and insecure groups. Showing how significant this difference is is crucial in this case, since it suggests that agro-ecology may or may not affect the degree of food insecurity in the households.

The chi-square assumption is not violated because it shows 0 cells (0.0%) have an expected count less than 5, and the minimum expected count is 16.01. Accordingly, the Pearson chi-square value is 16.738 with df 3 and p-value 0.001 indicate that the result is statistically significant, which says that there is a significant association between agro-ecology and household food insecurity status. That means, household food insecurity status is not independent of the agro-ecology zone in which the household resides. Based on the crucial value provided in the technique section, this evaluation is further assessed to determine the degree of the association or the impact of agro-ecology on the food security status of families.

A Phi or Cramer’s V value greater than 0.25 indicates a strong association. In this study, the computed value of 0.29 demonstrates that agroecology has a very strong effect on household food insecurity status, and the corresponding p-value of 0.001 confirms that this relationship is statistically significant. This finding reveals that food security challenges vary considerably across agroecological zones. Households in lowland areas are more vulnerable to food insecurity due to recurrent droughts and limited irrigation facilities, whereas those in highland and midland zones tend to be more food secure because of favorable rainfall patterns and diversified crop production. These results highlight the importance of agroecology-specific policy interventions, such as promoting drought-tolerant crops in lowlands, expanding small-scale irrigation schemes, and strengthening soil and water conservation measures in highland areas.

The analysis revealed that female-headed households were more prone to food insecurity compared to male-headed ones. This could be attributed to differences in access to productive resources, household labor, and market participation. Similar patterns were observed in studies by Fekadu and Mequanent [82] and FAO [83], which emphasized that gender-based disparities in land ownership and income opportunities exacerbate vulnerability to food shortages. This finding supports the entitlement theory of Sen [84], which posits that access to assets and resources is a key determinant of food security.

In the Borena district, gender-based food security disparities are influenced by agroecological conditions, with female-headed households in lowland areas facing fewer livelihood opportunities than those in highland zones. This aligns with studies by Amare and Simane [77], Tesso et al. [85], and Belay et al. [86], which found that lowland areas are more vulnerable due to drought, limited rainfall, and restricted access to productive resources. However, it contrasts with Tadesse et al. [87], who reported greater food insecurity among highland female-headed households, underscoring the role of local context in shaping gendered food security outcomes.

Conclusion and recommendations

The study revealed that 62.6% of households in the district are food insecure, mainly due to limited productive assets such as small landholdings and low livestock ownership. Key determinants of food security included sex, age, education, farmland size, family size, livestock ownership, agricultural input use, food aid, and off-farm income. Households with larger families and fewer resources were most vulnerable. While short-term coping strategies provide temporary relief, adaptive mechanisms such as petty trade, cultivation of short-maturing crops, and seasonal migration enhance resilience.

To improve food security, efforts should focus on empowering women, expanding access to productive assets, promoting climate-resilient agriculture, and diversifying livelihoods. Agroecology-sensitive policies are also needed—supporting drought-tolerant crops and irrigation in lowlands, mixed farming and input access in midlands, and soil conservation and diversification in highlands. Cross-cutting actions, including credit access, market linkage, early warning systems, and gender-responsive programs, are vital for strengthening household resilience and achieving sustainable food and nutrition security.

Strengths and limitations of the study

This community-based cross-sectional study, which combined quantitative and qualitative methods, provided a comprehensive understanding of the determinants of household food insecurity. The use of standardized tools (HFBM) and the inclusion of multiple agroecological zones enhanced the validity and representativeness of the findings. However, the cross-sectional design limits causal inference, and self-reported data may have been affected by recall and social desirability biases. As the study was confined to a single district, the results may not be generalizable to other contexts. Future longitudinal and multi-site studies are recommended to strengthen the evidence base.

Authors’ contributions

Muluemebet Worku Gurmu **:** Methodology, validation, formal analysis, investigation, data curation, supervision, and writing - review & editing. Berhanu Mitiku Zeleke: Conceptualization, methodology, software, data curation, formal analysis, and writing - original draft.

Funding

No funding was received from any organization.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

The study was approved by the Ethics Committee of the Department of Geography and Environmental Studies (Departmental Review Board approval Ref. No. GeES 023/22). The researchers and data collectors obtained informed consent from respondents, interviewees, participants in focus group discussions, and secondary data sources to ensure anonymity and confidentiality of the data.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.