1 Introduction

Malnutrition in Ethiopia remains a serious development issue, with high rates of stunting, wasting, micronutrient deficiencies, maternal malnutrition, and disparities between rural and urban areas [1]. Rural dwellers, who are food producers, experience the highest prevalence of malnutrition than the urban population in all its forms [2]. The efforts to address the burden of malnutrition have, until recent years, focused mainly on interventions that are designed directly to address the immediate causes of undernutrition [3]. The persisting challenge of malnutrition requires not only treating the symptoms but also tackling the underlying causes, which are referred to as nutrition-specific interventions. Nutrition-specific and nutrition-sensitive interventions [4, 5] encompass the broad spectrum of efforts towards ending food and nutrition insecurity.

As a strategy to tackle the underlying determinants of food and nutrition security challenges, nutrition-sensitive interventions integrate complementary sectors. Those sectors include, but are not limited to, agriculture, health, social protection, education, water, and sanitation [3, 4, 6, 7]. Nutrition-sensitive agriculture (NSA) is an approach that seeks to ensure the production of a variety of affordable, nutritious, culturally appropriate, and safe foods in adequate quantity and quality to meet the dietary requirements of populations in a sustainable manner [4, 8, 9]. NSA aims to maximize the positive impacts of the food system on nutrition outcomes [10, 11]. NAS to impact better nutritional outcomes requires following certain pathways [12]. The pathways include agricultural food production (production pathway), agricultural income (income pathway), and women’s empowerment [13,14,15]. There are also additional pathways suggested, including nutrition-related knowledge and strengthening local institutions [16]. Depending on the context, the pathway to nutrition-sensitive agriculture could also consider the food environment in the market, natural resource management, water, irrigation, health, and sanitation [12].

The production pathway of agriculture for better nutritional outcomes focuses on making food available, diversified, and accessible through nutrient-dense crop production and animal products such as dairy, poultry, meat, and fish. Availability and access to diversified nutrient-dense foods through on-farm production at the smallholder farm household level are mainly for household consumption [17,18,19]. The problems of food and nutrition security in Ethiopia in subsistence smallholder farm households are mainly characterized by a lack of diversification, low inputs, and being dependent on rainfall [20, 21]. Diversified production by smallholders, on the other hand, is an adaptive and cost-effective risk-reduction strategy [22,23,24]. The production pathway of agriculture, on the other hand, is linked with the available assets of households and agricultural-based livelihoods [12, 25]. The study attempts to shed light on the contribution of smallholder agricultural production to diversified food availability for own consumption. The production pathway, other than income, women's empowerment, and value chain approaches, has been presented as an alternative strategy to tackle the challenges of household food and nutrition insecurity. Measuring agricultural diversity as an indicator of nutrition-sensitive agriculture requires assessing crops and livestock production, including several of crops grown and animal husbandry [26]. The objective of this study is to investigate the production pathway of smallholders to nutrition-sensitive agriculture in northwest Ethiopia. The study primarily seeks to address the research question of how smallholders' plant-sourced and animal-sourced food production is diversified, what types of food groups are produced by farm households, and what factors influence the production of various food groups.

2 Methods

The northwest of Ethiopia represented by four districts including Alefa, North Achefer, Takusa, and Quara, was the study area. The study area has an estimated population of 814,021, which include both rural and urban dwellers. There are about 130,472 registered smallholder farm households, who are expected to feed their own households and support the food demands of the growing rural and urban population on fragmented land. The agricultural-based livelihood of the farm households in the area is based on mixed farming, with crop cultivation as primary and animal domestication as secondary. Meeting the food demand of households predominantly relies on cereals. Household income generation depends on cash crops such as sesame, soybean, cumin, garlic, and onion and the sale of livestock.

A cross-sectional study consisting of quantitative and qualitative techniques was employed. Registered smallholder farm households in selected districts were used as a sampling frame of the study population. The sample size was determined by using the formula suggested by [27] for homogenous population. Data were generated from 478 randomly sampled smallholder households through a structured questionnaire-based household survey. The questionnaire was designed to capture demographic, socioeconomic, and institutional attributes, as well as the number of groups of agricultural commodities produced, including livestock. To substantiate the data collected using household surveys, additional qualitative data were also collected by interviewing experts from the agriculture sector and focus group discussions with farmers.

Agricultural commodities produced by smallholders were categorized into 12 food groups based on the recommendations of [28] and [29] in the FAO and FANTA guidelines for dietary assessment, respectively. For the smallholders’ production, indicators of nutrition-sensitive agriculture recommended by FAO [30] for household availability of various food groups were used. Taking the production scores of each food group, which have binary outcomes, as response variables and demographic, socioeconomic, and institutional factors as explanatory variables, binary logistic regression was used. Simple linear regression analysis was also used to determine the association between livestock ownership and meat production. It was used to look at whether the on-farm availability of livestock in the household has directly translated to meat or access to animal-sourced protein. The results from inferential statistics were supported by descriptions and qualitative explanations which were recorded in verbatim translation.

2.1 Variables

The dependent variables of the study are the production scores of different food groups emanating from plant and animal sources. The dependent variables include the production score of cereals, pulses, oilseeds, vegetables, fruits, tubers, spices, eggs, milk and its products, fish, and honey/sweet, which were used as a measure of diversified food production. The outcomes of response variables are binary (1 = produced, 0 = not produced) for each food group. The amount of meat produced as an outcome variable was measured using standardized meat estimation recommended by [31], and it is a continuous variable. Common explanatory variables (Table 1.) hypothesized to have an effect on both plant- sourced and animal-sourced food productions are presented in the following table.

Table 1 Explanatory variables
Full size table

3 Results and discussion

3.1 Production share of agricultural commodities

Crop-livestock mixed agriculture is a dominant way of food production with a great variation among smallholder farm households in northwest Ethiopia. The variations are not only with the amount and number of household participated but also in terms of diversity of the food groups.

The average production of total crops (all types of crops produced in the production season) from all groups per household per annum is estimated to be 43.53 quintals. There are substantial variations in proportion of smallholder households’ participation and the production share of various food groups. All crops produced by smallholder households were categorized into seven food groups. Household production is dominated by cereals not only in terms of number of the households participating but also the volume produced. With a production participation of 97.91% farm households, the total production share of cereals from all crops produced by smallholder households is 60.85% (Fig. 1). Although, significant numbers of smallholders participate in production of vegetables, pulses, and oilseed, the volume share is not directly proportional.

Fig.1
figure 1

Proportion of household involved and share crop produced

Full size image

Fruit crops constitute fewer shares of the total crops produced, and households participated in the study area. Unbalanced associations between the proportions of farm households that participated and the quantity share of crops produced in each category may be attributed to the time of diversified food availability. Observations during data collection revealed that cereal food groups, followed by pulses, are available throughout the year from households' own production. Despite the nutrient density, especially micronutrients and vitamins, the on-farm availability of vegetables and fruits is seasonal.

Livestock domestication and the production of animal-sourced foods are integral parts of smallholder mixed farming. Depending on the purpose and economic capacity of households, there are variations in livestock ownership. Some types of animals serve as critical household assets. For instance, oxen are crucial because of their multiple roles. They are important sources of traction power to cultivate and thresh crops, sources of income, and the last resort to get out during social and economic crises. Considering multiple uses of oxen, about 86.40% of households have at least an ox or young bull (Table 2).

Table 2 Smallholder participation and average livestock per household (n = 478)
Full size table

Like oxen, cows are also important liquid assets for smallholders. They are important sources of milk and its products, income generation, and major livestock investment capital to increase or replace the stock. Accordingly, 80.54% of smallholder households own a cow or heifer. On average, smallholder farm households own 2.12 cows and/or heifers. Sheep and goats are also considered useful livestock assets of households, mainly for income generation, saving or investment, and consumption. Poultry production is also an integral part of smallholder livestock production for income, chickens, and eggs.

Given their indirect contribution to food, donkeys, mules, and rarely camels have been used as means of transportation and sometimes as traction power. Accordingly, 67.36% and 10.25% of households have a donkey and a mule, respectively. However, only 0.21% of households reported having a camel.

Aggregated livestock ownership in households was calculated using standard conversion factors for tropical livestock. Consequently, the average livestock ownership of smallholder farm households in the study areas was found to be 6.24 tropical livestock units, which is equivalent to 1560 kg of live weight.

The diversity of farm production in smallholder households was determined by converting agricultural commodities produced into twelve food groups derived from both crops and livestock. It was used as an indicator of the nutrition sensitivity of on-farm production by smallholders. As a result, smallholder households in northwest Ethiopia produce an average of 6.08 food groups annually, regardless of the quantity produced, with a standard deviation of 2.13.

3.2 Determinants of plant-sourced food groups

Given the agroecology, fertility of the soil, biological, technological, institutional, and managerial aspects, there are many factors affecting smallholder food-based production. The production scores of food groups produced were used to estimate on-farm food availability and diversity. An estimation of the average marginal effects using a binary logistic regression model shows that the age, educational status of the household head, farming experience, the number of oxen owned, and the number of contacts with agricultural extension workers are found to be statistically significant variables affecting the production of cereals (Table 3). As the age of the household head increases by 1 year, smallholder households are 0.2% less likely to produce cereal. Similarly, a farm household with the head having an additional year of schooling is 0.6% less likely to produce cereal. With an additional one-year farming experience of the head of the household and one more additional ox, smallholder farm households are 0.2% and 4.5% more likely to produce cereals, respectively.

Table 3 Factors affecting plant origin food production (n = 478)
Full size table

The result of the study concerning plant-sourced food production suggests that the farming experience of household heads has a positive impact on plant-sourced food production in smallholder households. This is partly because people with ages below the productive group have less contribution to plant-based food production. The negative association between years of schooling and cereal production, on the other hand, is evidence that as household head got more educated, they tend to produce nutrient-dense and diversified plant-sourced-foods such as fruits and vegetables.

Arable land for households and consideration of management difficulties are statistically significant predictors of the pulse and legume production of smallholders. With an additional hectare of land, smallholder farm households are 2.2% less likely to produce pulses and legume-group foods. This is partly attributed to the fact that pulses and legumes are important soil fertility improvement crops, and households with small farm land tend to cultivate these crops frequently. However, households that consider the management difficulties of growing crops are 7.2% more likely to cultivate pulses and legumes.

Oilseed or fat-source crop production by smallholder households has been affected by various factors. The binary logistic regression model result demonstrates that the age and sex of the household head, farming experience, arable land of the household, and livestock ownership are statistically significant variables affecting smallholder oilseed production. Compared to female-headed households, male-headed households are 19.1% more likely to produce oilseeds. Likewise, for each additional year of farming experience, smallholder farm households are 2.4% more likely to produce oilseeds.

Smallholder vegetable production is affected by distance to the nearby marker center, use of irrigation, and consideration of management difficulties. Practicing irrigation for crop production is an advantage in exploring off-rain seasons. Consequently, compared to non-irrigation user farm smallholder households, user households are 18.0% more likely to produce vegetables. Similarly, households that considered the management difficulties of different crops to grow are 14.4% more likely to cultivate vegetables compared to those that did not consider them.

As far as fruit production is concerned, the sex of the household head, educational status of the household head, and consideration of management difficulties are found to be statistically significant variables affecting smallholder farm households. Male-headed households are 14.8% more likely to plant fruit trees compared to female-headed households. Conversely, households considering the management difficulties of crop cultivation are 7.6% less likely to grow fruit trees. The educational level of the household head is also an important factor in growing fruit trees. As it is clearly shown in the average marginal effect, with each additional year of the schooling of the head of the household, smallholder farmers are 14.8% more likely to grow fruit trees.

Smallholders grow tuber and root crops as important sources of food and income. The production of these crops is also determined by socioeconomic and institutional factors. The result of the binary logistic regression models shows that year of schooling, farming experience, arable land, number of oxen, total livestock ownership, number of contacts with agricultural extension workers, and use of irrigation are found to be factors affecting smallholder tuber and root crop production.

3.3 Determinant of animal-sourced food production

Animal-sourced foods significantly contribute to household availability of diversified and nutrient-dense foods. However, their production at smallholder farm households is determined by various factors. Considering the production scores of each animal-sourced food group as response variables, binary logistic regression analysis was used to estimate the effects of demographic, socioeconomic, and institutional factors on households’ economic access to a variety of foods (Table 4).

Table 4 Factors affecting production of animal-sourced foods (n = 478)
Full size table

Smallholder farm households commonly produce eggs using traditional methods. Various factors affect the production of eggs in smallholder households. Total cops produced and contacts with agricultural extension workers were found to be factors determining the production of eggs. For each additional quintal of crop produced and contact with an agricultural extension worker, smallholder farm households are 0.7% and 1.6% more likely to produce eggs, respectively. The age of the household head, farming experience, total livestock ownership, total crops produced, and number of contacts with agricultural extension workers are statistically significant variables affecting the production of milk and its derivatives. Households with an additional year-old head are 0.9% less likely to produce milk and its products.

Farming experiences, crops produced, and total livestock owned by smallholder farm households affect the production of milk and its products. Households with one additional year of farming experience are 0.8% more likely to produce milk and its products. Similarly, for each additional tropical livestock unit ownership and contact with agricultural extension workers, smallholders are 1.9% and 2.4% more likely to produce milk and its products, respectively.

Fish and honey production are rarely reported as animal-sourced foods produced by smallholders. Environmental suitability and access to water resources determine the production of fish and honey. The results from the analysis of binary logistic regression indicate that sex, year of schooling, and farming experience of household heads, arable land, total crops produced in a given year, distance to the nearest market center, and contacts with agricultural extension workers are statistically significant factors affecting the production of fish. Production of honey, on the other hand, is affected by the year of schooling of the household head, arable land, number of oxen, livestock ownership, and number of children between the ages of five and fourteen.

Meat production is an integral part of smallholder food production, with high variations among households. Though the quantity produced and frequencies of accessing meat vary, no household was found to have a zero production score for meat. The annual meat production of smallholders ranges between 10 and 2451 kg per household. There is a statistically significant association between livestock ownership and meat production. Simple linear regression model analysis results show that for a unit change in tropical livestock, the amount of meat produced increases by 11.21 kg in smallholder farm households at the 99% level of the confidence interval.

3.4 Blockades of smallholder food-based agricultural production

Food-based agricultural production by smallholders is constrained by various external and internal factors, including climatological, technical, technological, institutional, and managerial pitfalls. Food-based production is heavily dependent on rain-fed crops and livestock. Only 13.6% of households use small-scale irrigation to cultivate different crops, mainly fruits and vegetables, during off-rain seasons, whereas the majorities (constituting 86.4%) are contingent on rain-fed agriculture. Consequently, as pointed out by [31], seasonality has become one of the challenges of agriculture and nutrition linkage through production. Ever-increasing climate variability, which includes erratic rainfall along with extreme weather events, puts smallholder agricultural production at risk. Evidence from qualitative data indicates that a poor agricultural input supply system for seeds, chemicals, and fertilizer, uneven agricultural extension support, and weak institutional support have negatively affected smallholder production. The result of the study is also similar with the findings of [32, 33], and [34].

Managerial, environmental, and institutional bottlenecks, on the other hand, have significantly constrained smallholder livestock production and productivity. The absence of grazing land and affordable animal feeds also affects the livestock production of smallholders. 78.66% of households reported that they do not have access to affordable animal feeds in their locality. Moreover, the traditional open-grazing herding system and the expansion of agricultural land for crops, coupled with poor, inaccessible, and unaffordable veterinary services, are some of the bottlenecks of livestock production. Observations and reports of focus group discussions revealed that unregulated and unsafe agricultural chemical applications like pesticides and herbicide residues are found to be the hidden threat to open-grazing-based livestock production. This, in turn, can affect not only the quantity of food being produced but also the safety of the food and nutrients supplied by agriculture.

4 Conclusion

Smallholder production pathways can be fully integrated into nutrition-sensitive agriculture to ensure the availability of diversified food for farm households in northwest Ethiopia. With all its limited resources, including land and technologies, smallholder agricultural production is nutrition-sensitive and can provide economic access to diversified foods. Regardless of the quantity, the average on-farm production diversity score of smallholder farm households has the potential to supply a minimum of six food groups. However, cereal crops dominate plant-sourced food production, with a larger share of volume and the number of households participating.

As important components of agricultural production, crops and livestock production complement each other and constitute food-based production and income generation. However, demographic, socioeconomic, institutional, and managerial factors affect the production of various plant-sourced and animal-sourced food groups by smallholder farm households. Unless measures are taken, the ambitious need to meet the food demands of an ever-increasing urban and rural population and ensure nutrition security will remain a challenge. Moreover, when it is added to the existing crises of food insecurity and malnutrition, observations on food safety issues are evidence that the worst situation with catastrophic effects is looming. However, improving agricultural extension services, promoting small-scale irrigation schemes, and introducing nutrient-dense animal feeding with good crop and livestock management techniques could contribute to the availability of diversified foods.