Analysis of wheat traits that determine female farmers’ preferences for wheat varieties: Gender insight from southern tigray, ethiopia

Abstract

Background

Wheat productivity has been constrained by changing agroecological and socio-economic conditions, coupled with a lower uptake of new farm technologies. Gender difference is one major social category that needs systematic estimation to distinguish the adoption of technology and the preference between male and female farmers. Hence, this study analyzes wheat traits that determine female farmers’ preferences for wheat varieties in southern Tigray, Ethiopia.

Methods

The primary data was collected by using semi-structured interviews, focus group discussions, and key informant interviews from 169 female farmers who were selected by using a two-stage sampling procedure. This study used descriptive statistics and a multinomial logit model to estimate the wheat traits that determine the wheat variety preference of female farmers.

Result

The result obtained from descriptive statistics shows the existence of heterogeneity in trait preference of female farmers among bread wheat, durum wheat, and local wheat variety types. The result indicates that risk-averting traits were the most frequently selected traits for both wheat types. Furthermore, the result of multinomial logit model indicates that wheat variety traits such as yield difference, marketability, resistance to drought, and resistance to frost and disease significantly influenced female farmers’ choice of wheat variety to grow.

Conclusion

This study aims to fill the current knowledge gaps and tackle the significant issues faced by wheat-growing female farmers by examining the wheat traits that influence the wheat variety choice of these farmers. The finding scrutinized that even though the female farmers’ choices of wheat variety traits were heterogeneous, the majority of their decisions on the choice of wheat variety were primarily guided by risk-averting and yield traits. This evidence provides significant insight for developing gender-sensitive variety traits in crop breeding programs. Moreover, the findings significantly help policymakers, input suppliers, breeding programs, and extension workers to become more gender-responsive, to increase the productivity of wheat-growing female farmers.

Introduction

Wheat is one of the leading cereal crops grown in the world and the most commonly grown crop from the lowland to highland part of Ethiopia (CSA 2016; Gaya et al. 2017). Ethiopia is one of the largest wheat producers in Sub-Saharan African countries. The increment in wheat production and productivity plays a significant role in achieving agricultural and economic growth in general and household food security in particular (Dibaba 2019; Asfaw et al. 2019; Anteneh and Asrat 2020; Muleta and Mebratu 2024).

Women provide more than half of the agricultural labor force (Harun 2014; Afridi et al. 2024; Abdisa et al. 2024), and contribute up to 70% of household food production in developing countries like Ethiopia (Doss 2014). In this case, men are the top decision-makers in agricultural production, while women have the chance to make domestic and child-care-related decisions. Women contribute huge effort in various farming activities such as land preparation, soil fertility management, and planting, weeding, and harvesting seasons. Nevertheless, most women in developing countries are well-known for lower agricultural productivity because of low adoption capacity for agricultural technologies and insufficient access to production resources such as land, cash income, and animals (Dula 2022; Fahad and Wang 2018).

In Ethiopia, there is a wide gender disparity in agricultural technology adoption. The disparities have emanated from multiple intersectional identities of male and female farmers, such as access and ownership of resources, decision-making power, educational status, access to credit and health services, access to information, and participation in social, economic, and political activities. These factors were identified as a significant determinant of male and female farmers’ adoption of various agricultural technologies in different parts of the country (Abebe 2017; Neway and Zegeye 2022; Gebre et al. 2021).

Several scholars analyze the gender difference in agricultural technology adoption, for example, the studies by Mekonnen (2022), Teklewold et al. (2022), and Aryal et al. (2022) assessed the gender disparity in climate change adaptation; the other studies by Vandercasteelen et al. (2018), Daniso (2022), and Lungu et al. (2024) evaluated the gender-based analysis in Teff variety adoption and also Gebre et al. (2021), conduct a study that assesses the gender difference in improved maize adoption. The results of these and other studies (Beshir and Wegary 2014; Bela et al. 2018) point out that the adoption status of male and female farmers was diverse because of the impact of socio-economic factors and their distinct preferences for agricultural technologies.

According to Jaleta et al. (2023), Massresha et al. (2021) and Diiro et al. (2018) the decision of male and female farmers to adopt or not adopt the latest improved crop varieties was influenced by various economic, social, environmental and demographic factors. Specifically, the studies by (Rai and Bajgai 2023; Sylla et al. 2023; Bacud et al. 2024) mainly elucidate the socio-economic factors that determine the joint decision of male and female farmers to adopt new farm technologies. In addition, some other studies conducted by (Jinbaani et al. 2023; Asante et al. 2023; Namirimu et al. 2024) evaluated the gender-specific differences between male and female farmers on choosing improved crop variety traits. However, the previous studies were limited to investigate female farmers situation separately from male farmers to recognize their unique varietal trait preferences as a country in general and the study area in particular.

Recently a piece of literature reported that to be effective in terms of improving adoption, yield performance, income, or nutritional benefits, breeding programs need to understand farmers’ preferences for varietal traits, especially for female farmers’ (Ashby and Polar 2019; Zimba et al. 2023; Badstue et al. 2022 and Gartaula et al. 2024). Hence, considering female farmers’ situation is necessary to develop client-oriented breeding programs and a general approach to improving overall production (Efisue et al.2008; Bishaw et al 2010; Tadesse et al. 2019; Chipeta et al. 2024).

Similarly in the Tigray region, the majority of rural population is highly dependent on agricultural crop production. Particularly, the study area of the Emba-Alaje district is the potential area for growing bread wheat varieties, durum wheat varieties, and local wheat varieties (Teklay and Tekley 2015 and Tekle and Hagos 2018). The research centers released several improved wheat varieties based on the agroecological conditions of the district. Nevertheless, the adoption status of women farmers was stagnant and only a few wheat varieties have been adopted and the other varieties were abandoned by female farmers (Abebe et al. 2012). However, still now no attention was given to analyze female farmers’ preference for improved wheat variety traits.

This study analyzed wheat traits that determine female farmers’ choice of wheat types to grow. Hence, by investigating female farmers’ choice of wheat varietal traits, this research addresses the existing knowledge gaps and critical challenges faced by wheat-growing female farmers. The outcome of this study has practical implications to increase the relevance and effectiveness of crop breeding programs, extension services, policymakers, and input suppliers by providing insight into what traits are needed by female farmers. Moreover, the study used a mixed-research approach combining quantitative data analysis with semi-qualitative data analysis, that enabled to gain a comprehensive understanding of the perception of female farmers toward wheat trait preferences.

Theoretical framework

The theoretical background of this study emerged from Lancaster’s model of consumer choice (Lancaster 1966) and the random utility models (McFadden 2013). According to the Lancaster assumption raising consumer satisfaction or utility maximization not merely from the good itself but also the characteristics of the goods (Okello et al. 2015). This indicates that the satisfaction or utility of a good is derived from traits of a good rather than the good itself. However, male and female consumers preferred these attributes of goods differently (Mwanga et al., 2021). Due to this Lancaster’s approach has its justification for dividing a product into several attributes. Hence, the preferences of male and female consumers for a product can be examined in terms of their preferences for each attribute, which in this case are the wheat variety traits. Moreover, this study adopted random utility theory (models) that are specified based on maximum likelihood estimation by considering the possibility of making errors in measuring subjective views of people toward the values of the wheat variety attributes (Bhat 2003). In this analysis, the dependent variable was the preference of female farmers among wheat varieties, and that is categorical. Hence, the appropriate econometric model that fits the nature of this variable would be the multinomial logit model (Hassan and Nhemachena 2008).

Conceptual framework

Figure 1 presents the conceptual framework for this study that is constructed based on the above theories. The framework indicates that the most preferred wheat variety trait within three sub-categories such as agronomic traits (high yielding, early maturity, and plant height), risk averting traits (resistance to drought, resistance to disease, and resistance to frost), and quality or appearance traits (grain size, resistance to shattering and marketability). The decision maker (i.e. female farmers) is expected to allocate a utility value for each trait that is important in her choice to maximize their satisfaction and the listed traits were expected to influence the decision of female farmers.

Fig. 1

Conceptual framework of the study

Research methodology

Descriptions of study area

This study was conducted in the Emba-Alaje district located in the southern Zone of Tigray regional state. It is about 85 km far away from the capital city of Tigray Regional State, Mekelle. It has 20 rural and one urban Kebeles. The administrative center of this district is Adi-Shehu. It is a part of the Southern Zone and bordered on the south by Enda-Mehoni, on the Southwest by the Amhara Region, on the north by South eastern Zone, and on the Southeast by Raya Azebo (Gebrewahd et al. 2017). The district contains a total population of 107,972 of whom 52,844 (48.9%) are men and 55,128 (51.1%) are women and 7568 are urban population. The district contains a total of 24,784 households and a total cultivable land of 22,457 hectares. For the land under cultivation in this district, 65.39% was planted in cereals, 24.94% in pulses, and 51 hectares in oilseeds; the area planted in vegetables is missing. The area planted in fruit trees was 57 hectares, while 32 were planted in Gesho. 65.36% of the farmers were use mixed farming systems (both crops and livestock), while 33.63% only grew crops and 1.0% only raised livestock (Teklay and Tekley 2015). Figure 2, shows the map of the study area Emba-Alaje district.

Fig. 2

Source: DLAO, (2019)

Map of study area (Amba Alaje district).

Sampling techniques and sample size determination

In this study, a two-stage sampling procedure was used to select the study area and the representative samples for this study. In the first stage, the Emba-Alaje district and three study villages (locally referred to as ‘Kebeles’, namely Atsella, Sesat, and Ayba) in the district were purposefully selected by considering its great potential for production of bread wheat, durum wheat, and local wheat verities. In the second stage, from the total of 987 female farmers in three Kebeles, l69 respondents were selected by simple random sampling techniques. The number of respondents was determined by considering several factors, such as the desired degree of precision, and the availability of sufficient time and funds. Hence, by considering these factors, the sample size was determined using Yamane’s (1967) formula.

$${\varvec{n}}=\frac{{\varvec{N}}}{1+{\varvec{N}}({{\varvec{e}})}^{2}}$$

(1)

Where, n, total sample size of this study; N,  total female farmers of the three kebeles or Population size (N = 987); e, Confidence level (0.07); The total sample size of this study was = 169.

As shown in Table 1, making the sample size proportional among Atsella, Sesat, and Ayba Kebeles was necessary to reduce the research bias. Therefore, the following formula was used to make the sample size proportional to each kebele respondent:

$${\varvec{n}}{\varvec{i}}=\frac{{\varvec{N}}{\varvec{i}}({\varvec{n}})}{\boldsymbol{\Sigma }{\varvec{N}}{\varvec{i}}}$$

(2)

Where, ni, sample to be selected from ith kebele; Ni,  the total population living in selected ith kebele; ΣNi, the sum of total population in the selected three kebeles; n, total sample size.

Table 1 Proportional sample size distribution for each Kebeles

Data type, sources, and method of data collection

To achieve the purpose of this research, both quantitative and qualitative cross-sectional data were collected from primary data sources. Thus, the primary data was collected through a semi-structured interview schedule, focus group discussions, and key informant interviews from 169 female farmers about wheat production, the source of wheat varieties and their preferences among wheat varieties concerning their traits.

Primary data collection tools

Semi-structured interview schedule: Before collecting the original survey data, the interview questionaries were translated from English to the local language (Tigrigna). Then after, the interview questionaries were pre-tested on non-sample female farmers for further modification of the interview schedule based on the feedback obtained from the pre-test.

Focus Group Discussion (FGD): In this study, focus group discussions were used to collect the data that complemented the interview schedule by providing brief explanations behind quantitative data. Accordingly, to obtain more precise and detailed information a FGD checklist was prepared for wheat-producer female respondents. Hence, one focus group discussion with six members was held in each Kebeles.

Key informant interview (KII): Three key informant interviews were held with each selected three Kebeles. The KII comprised one Kebele extension officer and two female farmers in each kebeles. The interviews were held to gain overall information about female farmers' wheat production and their choice of wheat varieties.

Secondary data

Secondary data was collected from district office reports, books, journal articles, unpublished documents and other related papers.

Method of data analysis

Descriptive statistics

To analyze the overall wheat production status, wheat variety preference concerning the desired traits, and the main source of wheat seeds, this study used descriptive statistics such as mean, frequency, percentages, and graphs. In addition, the correlation coefficient was computed to check the multicollinearity problem among dummy/discrete variables, and the result indicates that there is no multicollinearity problem across the variables.

Econometric model specification

Multinomial logit model

The econometric model specification principally depends on the objectives of the study and the nature of the data available. In this analysis, the dependent variable was the preference of female farmers among wheat varieties, and that is categorical in nature (choice among bread wheat, durum wheat, and local wheat varieties). The analysis considers wheat traits as an independent variable that determine the multiple wheat variety choice of female farmers. Hence, the appropriate econometric model that fits the nature of this variables would be either multinomial logit or multinomial probit regression model. The multinomial logit (MNL) was preferred for this study because it is simple to compute than its counterpart, the multinomial probit model (Hassan and Nhemachena 2008).

As shown in Table 4, during the estimation the multinomial logit model (MNL) by default selects the referent group (base outcome). The model chooses the most frequently occurring wheat variety as the referent group. Therefore, based on the estimation result the MNL model selects bread wheat variety as the referent group (base outcome/category).

In addition, the coefficient estimation of multinomial logit model (MNL) provides the direction of the effect of the independent variables on the response variable. However, the coefficient estimation of MNL cannot measure the actual magnitude of the change. Therefore, it is necessary to estimate the marginal effects from the multinomial logit model, which measures magnitude of the change in choice of wheat varieties by a unit change in an explanatory variable.

The mathematical equation is displayed as follow

$$pij = \frac{{{\text{exp}}\left( {XiB\,j} \right)}}{{\mathop \smallint \nolimits^{{\mathop \sum \nolimits_{j = 0}^{J} {\text{exp}}\left( {xiB\,j} \right)}} }}$$

(3)

With the requirement that $\sum_{j=0}^{J}=$ 0 Pij = 1 for any i, where Pij = probability representing the ith respondent’s chance of falling into category j; X = predictors of response probabilities βj = covariate effects specific to jth response category with the first category as the reference. For appropriate normalization that removes an indeterminacy in the model is to assume that β1 = 0 (this arise because probabilities sum to 1, so only J parameter vectors are needed to determine the J + 1 probabilities) so that exp (Xiβ1) = 1.

Therefore, the above equation can be equivalent to the following equation:

$pr\left(yi=j/xi\right)=pij=\frac{\text{exp}(XiB\,j)}{1+\sum_{i=1}^{J}\text{exp}(XiBi)} for\,j\dots \dots .. \text{0,2}\dots .J$ and

$$pr\left(yi=1/xi\right)=p1=\frac{1}{1+\sum_{i=1}^{J}\text{exp}(XiBi)} for\,j\dots \dots ..\text{ 0,2}\dots .J$$

(4)

Where y = A polytomous outcome variable with categories coded from 0… J. Note the probability of Pi1 is derived from the constraint that the J probabilities sum to 1. That is, pi1=1-∑ = p ij. When the multinomial logit is used to model choices, it relies on the assumption of independence of irrelevant alternatives (IIA) which is not always desirable (Tizale 2007).

Results and discussion

Results of the study

Descriptive statistics results

Wheat varieties grown by female farmers in the study area

As shown in Table 2, female farmers were growing three major wheat types as a source of food consumption and to earn income by selling at local markets. Among the three wheat types, the bread wheat variety was the dominant (commonly grown) wheat type which was grown by 43.2% of respondents in the study area. The other 33.1% of female farmers grow local wheat varieties for their production, marketing, and consumption. The remaining 23.6% of female farmers growing the durum wheat variety. Relative to other wheat types, durum variety was less likely for home consumption and mainly grown for the market purpose to gain more income.

Table 2 Summary of commonly growing wheat varieties in the study area

Source of wheat varieties in the study area

Figure 3 presents the major source of wheat seed varieties for planting in the stud area. Based on the survey result, female farmers had four major sources of seed to grow both bread, durum, and local wheat varieties. Among the major seed sources, farmers’ cooperatives covered 44.5% of the total seed supply to female farmers. The district agricultural office provides 37.3% of the seed supply for female wheat producers. The research centers and also farmers' self-seed stores (own seed) proved 13.2% and 4.9% of wheat seed supply in the study area, respectively.

Fig. 3

Source: Own survey (2019)

Source of Wheat seed varieties in the study area.

Female farmers’ preference for wheat varieties

Based on the results in Table 3, female farmers were asked about their perception toward various wheat variety attributes and their preference among these traits (attributes) of wheat varieties. The traits were ranked according to the mean score computed from survey data. According to the mean score of durum wheat traits, disease resistance (3.7), resistance to frost (3.4), and resistance to shattering (3.1) were ranked as the first, second, and third attributes preferred by female farmers. While, high-yielding (3.6), disease resistance (3.4), and resistance to shattering traits (3.3) were ranked first, second, and third favored bread wheat traits, respectively. From local wheat variety traits, resistance to frost (3.7), resistance to shattering (3.5), and high-yielding (3.4) were ranked as the first, second, and third preferred attributes. The result of this study indicates that female farmers identified both agronomic, risk-averting and quality traits, but risk-averting traits were the most frequently preferred traits by female farmers in the study area.

Table 3 The rank of wheat variety traits based on the mean score as perceived by female farmers

Econometric model results

In the econometric analysis various wheat traits such as high-yielding, early maturity, grain size, disease resistance, marketability, resistance to drought, resistance to frost, and plant height were incorporated as determinants of female farmers variety preference or choice in the study area. The result of multinomial logit model is displayed in Table 4.

Table 4 Estimation of multinomial logistic regression for female farmers’ varietal preference

Discussion

Female farmers wheat seed varieties source in the study area

During the survey year (2019), the majority of female farmers used seeds from the formal sector for planting wheat crops. About 44.5% of female farmers used farmers’ cooperatives as a foremost source of seed for growing wheat. Moreover, agricultural offices provide 37.3% of seed supply, whereas the research centers supply 13.2% of seed for female farmers in the study area. According to Semahegn et al. (2021), farmers in the Oromia region Sire and Dodota districts accessed improved seed from agricultural offices, while farmers in Hitosa district predominantly used farmers’ cooperatives as a source of wheat seeds in different growing season. The remaining 4.9% of female farmers used their own saved seed store as a supply of seed for growing wheat. This result is consistent with the findings that report men and women farmers mutually used their own saved seed from the previous year (Bishaw et al. 2010).

Female farmers’ preference for wheat varieties for production, marketing, and consumption

The farmers’ wheat variety preferences were heterogenous and shaped by their perceptions toward traits of improved varieties. As displayed in Table 3, diseases resistance, resistance to frost, resistance to shattering, and high-yielding traits of durum wheat varieties were taking the first, second, third, and fourth ranks, respectively. Female farmers show greater concern for traits associated with disease resistance across the study Kebeles. Focus group discussion participants (FGD) also reported that disease resistance traits of wheat were vital because the majority of female farmers were economically poor and did not afford chemical pesticides, and had limited information on the suitable chemicals to control the diseases. This result is in agreement with the previous study by Tadesse et al. (2019) that report female farmers were more likely to adopt disease resistant varieties than male farmers. However, the other studies by Seifu et al. (2018), indicate that both male and female farmers prefer to adopt disease tolerant wheat varieties. This implies that both male and female farmers prefer disease and insect tolerant crop varieties. Hence, a significant improvement in these attributes of wheat would enhance farmers adoption of improved crop varieties.

The results in Table 3 also indicate high grain yield, disease resistance, and resistance to shattering were ranked as the first, second, and third most preferred bread wheat attributes by the female farmers in the study area. This result implies that female farmers were more likely to adopt wheat varieties that have better pre-harvest and post-harvest attributes. This result is in contrast with the finding by Christinck et al. (2017), that indicates the likelihood of women farmers toward preferring post-harvest traits, while men showed their preference to agronomic treaties. The other study reported that grain yield was the most preferred trait of bread wheat varieties followed by disease resistance, adaptation and early maturity for both male and female farmers (Semahegn et al. 2021).

From local wheat varieties, resistance to frost, resistance to shattering, and high-yielding attributes were taking the first, second, and third rank according to the mean score. During a key informant interview, female farmers explained that local wheat varieties have better environmental adaptability than other improved wheat varieties. This finding is in agreement with the previous study reported that environmental adaptability is a positive trait of local seed varieties and this negatively affected farmers adoption of improved varieties because this quality is mainly found in local varieties (Beshir and Wegary 2014).

Wheat traits that determine female farmers' preferences for wheat varieties

High yielding: as shown in Table 4 the yield performance of wheat varieties had positively and significantly influenced female farmers’ preference for wheat varieties at a 5% significant level. The marginal effect result implies that as female farmers need to increase their wheat yield in one unit, their preference for selecting durum and local wheat varieties relative to bread wheat declines by 2.9% and 2.2%, respectively. This implies that a bread wheat variety has better yield performance compared to durum and local wheat varieties in the study area. Previous studies on gendered trait preferences in agricultural crops confirmed that yield and yield attributes were the most important factors influencing a farmer’s decision to grow a variety (Nelson 2013; Semahegn et al. 2021). Likewise, Alebachew (2012) also identified grain yield and spike size as important criteria for choosing a wheat variety. In addition to grain yield, there is a considerable difference in grain size, and color between local and improved bread wheat varieties, consequently female farmers preferred improved wheat varieties over local varieties (Bishawa and Alemu 2017).

Early maturing: this attribute of wheat had a significant influence on female famers’ preference for wheat varieties at a 10% level. It is recognized that farmers need crop varieties that mature within a short period. As shown in Table 4, when female farmers required the shortest maturing wheat variety, the probability of choosing durum and local wheat varieties relative to bread wheat varieties declined by 0.1% and 4.7%, respectively. This result revealed that bread wheat varieties are slightly better at early maturing than durum wheat, while bread wheat varieties are highly better than local wheat varieties at early maturing. During focus group discussion (FGD) in Atsella Kebele the participants stated that: “Some wheat varieties have been abandoned because they don’t mature fast, taking several months to mature before harvest. The late maturity trait has two problems for us, the first one is we cannot wait for a long time because most of us haven’t reserved food for our children if the crop is not matured on time, and the second reason is the rain will stop before the crop matured and this reduces or failed the output of our crop” (Atsella women farmer FGD). Previous studies indicate that early maturity trait was the most frequently preferred trait by women farmers on various crops, such as earliness of maize in Ethiopia (Mulatu and Zelleke 2002), quinoa in Ecuador (McElhinny et al. 2007), early maturing potatoes in Kenya (Mwende Mutiso et al. 2024), sorghum and millet in Malawi (Zimba et al. 2023) and earliness of cowpea in Ghana (Jinbaani et al. 2023).

Resistance to drought: the crop varieties that resist drought stress are highly demanded by female farmers residing in drought-prone agroecology like the study area. This drought-resistance trait of wheat varieties influenced the seed preference of female farmers at a 1% level of significance. The estimated marginal effect also confirmed that with a unit increase in drought resistance, the female farmers’ preference for choosing durum wheat relative to bread wheat declined by 7% (Table 4). The key informant interview participants likewise explained the adverse impact of changes in weather conditions on crop failure, such as shortage of rainfall, rise of temperature and irregular patterns of rainfall were destroying the crops they are growing. Hence, farmers need drought-resistant wheat varieties to respond to this problem of production. The earlier study also confirmed a similar result that both male and female farmers were willing to sacrifice yield for the drought tolerance trait of the maize crop (Marenya et al. 2022). Moreover, the study conducted by Semahegn et al. (2021) on bread wheat variety identified that resistance to drought traits had positive and significant effects on the adoption of new and improved wheat varieties.

Highly marketable: as shown in Table 4 the marketability attribute of wheat significantly influenced the wheat variety preference of female farmers at the 1% significant level. As farmers need to increase the marketability of wheat products, their probability of selecting durum wheat relative to bread wheat declines by 8.1%. This result implies that the durum wheat variety has a low demand relative to the bread wheat variety in the local market of the study area. The study conducted on cassava trait preferences in Nigeria depicts that the marketability or market price trait of cassava had a significant influence on female-headed farmers’ cassava production (Bela et al. 2018).

Resistance to disease: the disease resistance attribute significantly influenced the seed variety preference of female farmers at the 1% significance level (Table 4). The estimated marginal effect confirmed that with a unit increase in resistance to disease, the probability of female farmers preferring local wheat varieties relative to bread wheat varieties declined by 12.5%. This result implies that local wheat varieties were easily vulnerable to disease compared to improved bread wheat varieties. The study conducted in selected wheat-producer zones of Ethiopia points out that improved wheat varieties that have better resistance to diseases were highly preferred by small-holder wheat producer farmers (Semahegn et al. 2021 and Euler et al. 2024).

Resistance to frost: wheat varieties’ resistance to frost stress significantly influenced female farmers' wheat variety preference at 5% level of significance. As one unit increases in resistance to frost stress, the probability of female farmers preferring local wheat varieties relative to bread wheat varieties increases by 3.9%. This result implies that the local variety has a better resistance capacity to frost stress than other wheat varieties in the study area (Table 4). During a focus group discussion farmers explained that; “Sometimes the local wheat variety was more desirable because it has better environmental adaptability such as resistance to frost and resistance to waterlogging. In addition, local varieties have better baking quality and are suitable for local food preparations and consumption” (Sesat Kebele women FGD). The study by Nelson, (2013) indicates resistance to rust disease has a significant effect on both male and female farmers adoption of improved wheat varieties.

Conclusion and recommendation

Understanding consumer preferences is a well-known concept in private industry to elevate their profitability, but relatively less common in the public sector of developing countries that produce wheat varieties for farmers. This study elicited wheat variety traits preferred by female farmers in three Kebeles of Emba-Alaje district southern Tigray.

The result of this study concluded that female farmers’ choices of wheat variety traits were heterogeneous. According to the result of descriptive statistics, the most frequently selected traits by female farmers were high grain yield, disease resistance, resistance to frost, and resistance to shattering. This implies risk averting traits were the most frequently selected traits for both bread wheat, durum wheat, and local wheat types. This result is contradicted by earlier research findings that conclude women farmers mostly prioritize food quality (post-harvest) traits while men preferred agronomic traits. Moreover, the result of the multinominal logit model identified resistance to drought, high yielding performance, early maturity, marketability, resistance to disease, and resistance to frost as the major varietal attributes shaping the choice of female farmers for wheat variety to grow. This study infers that high-yield traits are desirable by female farmers, but it is not always the most preferred attribute, besides quality traits such as marketability and risk-averting traits including drought tolerance, resistance to disease, and frost determining female farmers' choice of wheat variety to grow. Hence, this study concludes that female farmers’ preference for wheat varietal traits is heterogeneous according to the situation, and variety traits have significant influence on female farmers' choice of wheat to grow.

The result of this study has the following implications: since farmers’ trait preferences vary across situations, wheat breeders should take a real holistic approach to consider traits prioritized by farmers, especially women and other marginalized groups of farmers. Research centers, seed suppliers, extension officers, farmers' organizations, and other concerned stakeholders must collaborate effectively to improve female farmers’ access to newly improved wheat varieties. This will enhance female farmers' adoption of newly developed improved varieties that will boost sustainable production and food security of smallholder farmers.