Landrace and GM maize cultivars’ selection choices among rural farming households in the Eastern Cape Province, South Africa

ABSTRACT

Maize plays a significant role among rural farming households in South Africa. The study therefore estimated the drivers of maize cultivars’ selection choices among rural farming households focusing on the commonly grown cultivars from the study area (Landrace and genetically modifies [GM] maize). A cross-sectional survey of 650 respondents randomly selected from Port St Johns and King Sabata Dalindyebo Local Municipalities of the Eastern Cape Province of South Africa were used. Descriptive results revealed that a significant number of the respondents from the study area (65%) were growing Landrace maize cultivars, followed by GM maize (31%), improved OPVs (3%) and conventional hybrids (1%). Multivariate Probit regression estimates show that GM maize cultivars’ selection is positively influenced by the amount of rainfall (1% level), household size (5% level), education (1% level), size of arable land (10% level) and access to cell phones (5% level), whilst negatively influenced by the employment status (5%). Conversely, the selection of Landrace maize cultivars is negatively influenced by the of amount rainfall (1% level), education (1% level), income (10% level), access to cell phone (10% level) and radio (10% level) whilst positively influenced by the number of livestock (5% level). The study therefore argues that GM maize cultivars may be meaningfully promoted in high rainfall areas, focusing on arable land sizes and targeted awareness campaigns. The promotion of Landrace maize cultivars may be targeted in low rainfall areas in a mixed farming setting to enhance the complementarity of maize and livestock.

KEYWORDS:

• Eastern Cape Province
• GM maize cultivars
• Landrace maize cultivars
• smallholder farming
• South Africa
• yields

Introduction

Maize seed cultivars’ selection by smallholder farmers is influenced by many factors, which are considered under different conditions of environmental uncertainty.^1,2 With the vast array of maize seed cultivars’ attributes that the smallholder farmer must evaluate and understand, choosing what seed to plant is very difficult and unique to individual farmers.³ The unique farm conditions and experiences that smallholder farmers encounter with maize seed cultivars as well as various choices dictated by local seed availability, all constrain farmer seed choice in any given agricultural season.³ Generally, the weak agricultural extension services in Sub-Saharan Africa (SSA) as well as the influx of different maize seed cultivars from the non-governmental agencies and the private seed companies further compound this seed selection decision-making process for smallholder farmers.

Despite all these challenges, there is a consensus that maize productivity from this subsector is very low.⁴ Efforts to improve the low maize productivity seem to suggest the use of improved maize cultivars (genetically modified [GM] maize, conventional hybrids and improved Open Pollinated Varieties) as one of the options among others.⁵ Unfortunately, improved maize cultivars do not perform very well under the agro-ecological conditions of most smallholder farmers and this is worsened by their high seed costs.⁶ It is interesting to note that, as improved maize cultivars are introduced to the smallholder farming subsector hoping to address the low productivity, the farmers are choosing Landrace maize cultivars.⁷

Against this background, there is a need to understand drivers of maize cultivars’ selection choices at the household level using a multivariate approach given the possibility of substitutability of these maize cultivars from this subsector. In as much as improved maize cultivars are needed to improve productivity, the preference for Landrace maize cultivars from this subsector is worth understanding to enhance the development of improved maize cultivars that address farmers’ expectations and needs.^8–11

Problem Statement

Numerous studies from the (SSA have highlighted the importance of smallholder farmers adopting improved maize cultivars to increase their productivity.^5,9,12 Though this may be, adoption of such kind of technological innovations that increase productivity by smallholder farmers in Africa is still very low.^13,14 Such kind of productivity enhancing technologies are very costly; hence, the majority of smallholder farmers opt for cheaper Landrace maize cultivars. Moreover, the success of these innovations (improved maize cultivars) is not guaranteed because they do well under good agro-climatic conditions and managerial regimes that in most cases are beyond the reach of many smallholder farmers. Improved maize varieties thrive better under conditions of adequate moisture, good pest management practices and fertile soils.^6,14,15 Conversely, in South Africa, as is the case in most African countries, smallholder farmers reside in areas that are inherently dry, hot and characterized by infertile soils, where these conditions are seldom met.

In the interest of promoting smallholder farmer’s maize productivity in rural areas for various reasons (food security, poverty reduction, livestock feed), improved maize cultivars have been promoted by various stakeholders in South Africa under various support programs.^16–19 Post withdrawal of these support programs literature seem to suggest a low uptake of improved maize cultivars^20,21 and a continued preference of Landrace maize cultivars from this subsector.^7,8 Need therefore arises to understand the drivers of maize cultivars’ selection choices among rural farming households using a multivariate approach to allow the simultaneous influences of independent variables on the multiple competing maize cultivars. The study thus far highlights the following research gaps worth understanding. Firstly, literature acknowledge low uptake of improved maize cultivars under rural farming households. Secondly, literature also acknowledge huge efforts to promote the usage and adoption of improved maize cultivars under farming households. Largely missing is a clear understanding of the preference of Landrace maize cultivars by this subsector in a multivariate approach that accommodates the existence of other improved maize cultivars capable of substituting or complementing Landraces. Such information helps with the understanding of specific conditions under which different maize cultivars may be promoted to enhance efficiency and evidence-based programing. This is against the current status quo where improved maize cultivars are promoted in almost every rural area, and post withdrawal of the promotional program, rural farming household resort back to their Landraces.

Objective

This study investigated the commonly grown maize cultivars from the study area and factors that influence their selection by rural farming households. This was motivated by the recent influx of improved maize cultivars on the market, their low uptake and the continued preference for Landrace maize cultivars by the rural farming subsector.

Literature Review

This section presents literature on maize cultivars and factors that influence their selection in rural areas. Initially, a summary of maize cultivars is given followed by factors that influence the selection of maize cultivars.

Genetically Modified (GM) Maize Cultivars

GM maize cultivars can be broadly classified into three categories: Bacillus thuringiensis (Bt) maize, Herbicide Tolerant (HT) maize and Stacked (Bt and HT) maize.

Bacillus thuringiensis (Bt) Maize

Hilbeck et al.,²² noted that Bt maize is a maize cultivar genetically modified to express one or more proteins from the Bacillus thuringiensis bacterium capable of generating some proteins that are lethal to some insects. Thus far, Bt maize as it is popularly known, is of special interest to maize producers because of the insect defense mechanism. This has multiple health and economic benefits to maize producers, and the environment for producers are more likely to use less insecticides during the production cycle of Bt maize.¹⁶ This therefore implies less contact (health benefits) and usage (economic and environmental benefits) during the production cycle. Literature also acknowledges high yield potential and enhanced grain quality for Bt maize.²³ The rapid uptake of Bt maize cultivars is therefore partly attributed to high yields and reduction in insecticide usage.

Herbicide Tolerant (HT) Maize

Brookes and Barfoot,²⁴ noted that GM HT maize is tolerant to a number of herbicides, including glyphosate and glufosinate. The technology therefore enables “over the top” spraying of GM HT maize crops with broad-spectrum herbicides, which target both grass and broad-leaved weeds while causing no harm to the maize.²⁵ The main benefit of HT technology according to Brookes et al.²⁶ is to make weed control more cost effective (less expensive) and simpler for maize producers. Beyond the claimed cost effectiveness benefits, improved weed control will also result in higher yields through reduced weed biomass in maize fields.²⁷

Stacked (Bt and HT) Maize

Stacked Bt and HT maize cultivars have stacked traits of both Bt insect resistance and herbicide tolerance.²³ These maize cultivars offer insect protection, labor savings (reduced manual weeding), health and environmental benefits (reduced usage and contact with chemicals during the production cycle). They are however relatively more expensive compared to Bt and HT maize cultivars. The usage of stacked Bt/HT maize cultivars, especially among smallholder producers, is motivated by higher yields, taste and quality regardless of the higher seed cost.²⁸

Landrace Maize Cultivars

In South Africa, typical of most African countries, rural farming households produce Landrace maize cultivars.²⁹ A Landrace maize cultivar can be loosely defined as a local variant of a domesticated maize plant that has evolved over time adapting to natural and cultural conditions.³⁰ Returned maize seeds from the previous harvest also fits in this general definition of a Landrace cultivar. Landrace maize cultivars are claimed to be resilience to abiotic stressors and yield stability.³¹ Landraces are also claimed to be very nutritious with significant genes for adaptability to adverse situations (water stress, salinity and high temperatures) common in most rural Africa.^32,33 In certain parts of the world, for example, Mexico and South America, farmers have developed and selected local varieties and landraces of maize to satisfy local consumers’ cultural and culinary needs.^34–36 More recently, even in industrialized countries, there have been attempts to adopt and manage maize landraces, which are shaped not only by different productive priorities, but also by larger projects wherein new practices and values are created simultaneously.³⁷

Conventional Hybrids

Conventional hybrids are a first generation (F1) type of a controlled crossing between two parents who are diametrically opposed.³⁸ Normally, the offspring outperforms the average of the parents with several agronomic, phenotypic, and nutritional improvements.³⁸ These cultivars are popular among commercial farmers (with resources) mainly because of their potential higher yields,³⁹ especially in high rainfall areas with fertile soils. Despite their advantages, farmers are supposed to buy new seed every year, a reason why they are not so popular with smallholder farmers in rural South Africa with limited resources normally located in low rainfall areas with low soil fertility.⁴⁰

Improved Open Pollinated Varieties

Improved open pollinated varieties are developed using random cross-pollination to increase genetic diversity.⁴⁰ Normally, the crops are not uniform but can tolerate low rainfall area and low soil, making them more profitable and long lasting than conventional hybrids.^40,41

Literature suggests that, despite high yields, reduced cost of spraying, health and environmental benefits associated with hybrid maize cultivars, the production is still very low in rural South Africa.⁴⁰ Although their yields are claimed to be low, Landrace maize cultivars are widely produced in rural South Africa for a variety of reasons to include low seed cost and tolerance to harsh agro-ecology.⁴

Factors That Influence Production of GM and Landrace Maize Cultivars

Large amount of literature exists which explains factors that influence adoption decisions of different maize cultivars by smallholder farmers. Both previous and recent studies have highlighted factors such as socio-economic, institutional and technical factors as having a significant influence on adoption of specific maize cultivars by farmers.^{1,2,34,42–44} For instance, according to Martey et al.,⁴³ in Northern Region of Ghana, adoption of drought tolerant maize cultivars was mainly driven by access to seed, labor availability, extension service and location of farm households. These findings are consistent with those reported in most empirical studies that variables such as gender, asset ownership, land holdings and social networks amongst others affect smallholder farmers’ adoption of improved maize cultivars.^35,45,46

Notwithstanding the forgoing, Waldman et al.³ cautions that factors that influence smallholder farmers’ decision to select a specific maize seed cultivar for production are exceptionally complex. These decisions are often affected by such factors as the availability of information on seed performance, local seed availability, the transfer of information to farmers and perceptions of farmers on climate variability amongst others. Moreover, maize seed selection decisions are further complicated by the heterogeneity in growing conditions and their resultant performance under these varying conditions. Again, hybrid maize cultivars are deemed to be high yielding compared with local Landrace cultivars. However, despite these advantages in most parts of SSA, Landrace maize cultivars are still popular amongst farmers over hybrids because of taste, poundability, storability and lower requirements for inorganic fertilizer.^34,47–49 Hence, regardless of the production advantages of hybrids over Landrace cultivars, in most cases, they lack the consumption attributes preferred by farmers that are found in Landrace maize cultivars. Failure to consider consumer preferences and needs of the farmers are some of the criticisms that have been pointed at plant breeders, especially during cultivar development stages targeting marginal farming areas.^34,50

While in most African countries, smallholder farmers mainly rely on informal seed markets to access inputs, South Africa has one of the most advanced, competitive and developed seed industry in Africa.^51,52 Though this may be, access of improved maize seed cultivars in South Africa by smallholder farmers is a major challenge because the seed industry in the country seem to favor the large-scale commercial farmers.^52,53 This makes production of improved maize cultivars such as GM maize a daunting task amongst smallholder farmers. As such, in South Africa, previous studies tend to suggest that limitations in production conditions, cost and access of improved maize cultivars force farmers to opt for local Landrace cultivars.^8,54

Of interest to note is the reliance on Landrace maize cultivars among rural farming households and an interest to promote improved maize cultivars in the same subsector.^7,8 This therefore calls for a clear understanding of the potential drivers of improved and Landrace maize cultivar selection choices from this subsector given several possibilities of complementarity and substitutability between these maize cultivars. Such information is largely under reported in literature, yet these maize cultivars are significant for the subsector. Building on available knowledge skewed toward univariate modeling, the study took a multivariate approach for purposes of enhancing simultaneous interpretation of influences of the independent variables on each of the dependent variable.

Materials and Methods

The study was conducted in the OR Tambo District of the Eastern Cape Province of South Africa purposively selected. The district is largely rural with a diverse climatic environment.⁵⁵ Some areas receive high rainfall (Port St Johns Local Municipality), whereas others receive low rainfall (King Sabata Dalindyebo Local Municplaity), a scenario the study targeted to understand improved and Landrace maize cultivar selection choices. Household level maize cultivar selection data were solicited using a questionnaire.

The household level data were randomly collected from two purposively selected study areas in the OR Tambo District (Port St Johns Local Municipality and King Sabata Dalindyebo Local Municipality). Agricultural household population estimates of 37,168 and 15,962 for King Sabata Dalindyebo and Port St Johns local municipalities, respectively, were used to calculate the sample size (Eastern Cape Socio Economic Consultative Council).⁵⁶ Following Yamane (1967), the sample size was calculated as illustrated in Equation 1:

(1) $\mathrm{n}=\frac{N}{1+N{\left(e\right)}^2}$(1)

Where n is the sample size, N is the population size, and e is the level of precision.

The minimum sample size was therefore estimated to be 397 agricultural households. For the purposes of increasing estimation accuracy, a sample size of 650 agricultural households was randomly selected from the two study areas (King Sabata Dalindyebo Local Municipality = 325, Port St Johns Local Municipality = 325).

Conceptual Framework

The study conceptualized possible linkages of various factors that influence the decision of rural farming households to select maize cultivars as summarized in Fig. 1 below. These factors include the genetic traits of the maize seed, agro-ecology, institutional factors and socio-economic attributes of the farmer.^1,2,34,43 Rural farming households therefore weigh the perceived benefits they are likely to get from a given maize cultivar with reference to its genetic traits like yield and resilience to abiotic stressors.³¹

Figure 1. Conceptual framework.

Agro-ecological factors of importance for this subsector include, soil types and quality, rainfall, and temperature. Thus far, selection is influenced by location-specific agro-ecological factors, which varies depending on the location of the farmer.⁴³ Maize grows under specific agro-ecological conditions, which may force farmers to select cultivars compatible with their agro-ecology.

In addition, access to extension services, farmer organizations, markets (for inputs and produce) and credit are also important institutional factors considered by rural farming households as they choose maize cultivars to grow. These factors provide information, social capital, input and produce market to drive production. Last, but not least, several socio-economic factors like age, gender, education and income further influences how rural farming household choose different maize cultivars.³⁵ These factors deals with issues related to cultures and belief systems capable of shaping the behavior of human beings as they make decisions. The ultimate selection of a maize cultivar is therefore a combination of all these factors,³ which is best explained by the utility maximization theory as detailed in the next section.

Theoretical Framework

The study assumes that rural farming households grow different crops for various socio-economic reasons (food access and sell surplus to generate household income). Maize is one of the crops grown by rural farming households as a strategic commodity that addresses food security.²⁹ A wide range of maize cultivars is used in South Africa to include improved Open Pollinated Varieties, Landraces, GM hybrids and conventional hybrids.⁵³ Against this background, the study assumes that maize cultivar selection choices are explained by individual farmers’ expected utility subject to various technical, socio-economic and institutional factors. Thus far, what is observable and unordered are the maize cultivars selected by the farmers, whereas the associated utility for each maize cultivar selected is directly unobservable. The random utility maximization theory therefore explains farmers’ maize cultivar selection choices as follows: a rational rural maize farmer is expected to select maize cultivar “j” over cultivar “k” if, and only if, the perceived utility from maize cultivar “j” is greater than that of “k” as shown in Equation 2.

(2) $U_{ij}\left({\stackrel{`}{\beta }}_j{X}_i+{\epsilon }_j\right)U_{ik}\stackrel{`}{\beta }k{X}_i+{\epsilon }_k,\forall j\ne k$(2)

Where:

• $U_{jk}$= denotes perceived utilities of maize cultivar “j” and “k”

• $X_i$ = vector of explanatory variables

• $\stackrel{`}{{\stackrel{ˋ}{\beta }}_{jk}}$= parameters to be estimated

• ${\epsilon }_{jk}$ = error terms (assumed to be independently and identically distributed)

Normally, rural farming households grow a diversity of maize cultivars for substitutability, risk hedging, piloting and sometimes complementary reasons. In that case, a multivariate approach which captures the correlation of the dependent variables can be used to estimate drivers of maize cultivar selection choices among farming households.^57–59

Analysis

For purposes of managing the correlated dependent variables, a multivariate probit model was used⁶⁰ to estimate drivers of farmer level maize cultivar selection choices. Only two maize cultivars commonly grown from the study area (GM and Landrace) were used as the dependent variables. The other two cultivars (OPV and conventional hybrids) were dropped because of the low number of respondents growing them. Several previous studies with correlated dependent variables also used multivariate probit regression.^57,60–62 For this study, two dependent variables were created based on the common maize cultivars grown from the study areas (GM and Landrace). The multivariate probit model was specified as shown in Equation 3.⁶⁰

$y_i=1if\grave{x}{\beta }_i+{\epsilon }_i>0,$

(3) $=0if\stackrel{`}{x}{\beta }_i+{\epsilon }_i\le 0,i=1,2,\dots ,n,$(3)

Where:

• $x=$vector of explanatory variables

• ${\beta }_1,{\beta }_2,\dots ,{\beta }_n=$ conformable parameter vectors

• ${\epsilon }_1,{\epsilon }_2,\dots ,{\epsilon }_n=$ random error terms distributed as multivariate normal distribution with zero means, unitary variance and an n x n contemporaneous correlation matrix $R=\left[{\rho }_{ij}\right]$, with density $\varphi \left({\epsilon }_1{\epsilon }_2,\dots ,{\epsilon }_n;R\right)$.

Results and Discussion

This section presents the study results, initially focusing on basic sample statistics and the commonly grown maize cultivars from the study area. This is followed by the factors that influence GM and Landrace maize cultivars production.

Descriptive Statistics

Table 1 present a summary of the sample statistics from the study area. A sample of 650 respondents was randomly selected from the study area. The sample was dominated with females (67%) and single headed households (70%). A majority (76%) were unemployed although educated (79%).

Table 1. Basic sample statistics of respondents.

Categorical Variables	Frequency (N = 650)		Percentage (%)
Gender
Male	212		33
Female	438		67
Marital status
Single	458		70
Married	192		30
Employment status
Employed	157		24
Unemployed	493		76
Education
Educated	513		79
Uneducated	137		21
Household income
<500	35		5
500–1000	95		15
1000–2000	315		49
2001–5000	158		24
5001–10000	33		5
10001–20000	13		2
>20000	1		0
Access to TV
Yes	453		70
No	197		30
Access to cell-phone
Yes	540		83
No	110		17
Access to radio
Yes	336		52
No	314		48
Livestock ownership
Yes	563		87
No	87		13
Continuous variables	Mean	Std. Dev.	Min	Max
Age	57	14	19	91
Size of arable land	1	1	0.1	8
Household size	7	4	1	25
Number of livestock	26	24	0	161

The majority (49%) of the respondents had an average monthly income between R 1000 – R2000. A majority (70%) had access to television (TV), cell phone (83%) and radio (52%). In addition, a majority (87%) owned livestock with an average arable land size of 1 hectare. The family size, on average, was seven family members with a mean household head age of 57 years. Lastly, households owned on average 26 livestock.

Maize Cultivars Grown from the Study Area

Figure 2 shows the production status of maize cultivars from the study area. The results show that Landrace maize cultivars were produced by 65% of the respondents while GM maize cultivars were produced by 31% of the respondents. The results also show that improved OPV were produced by 3% of the respondents while conventional hybrids were produced by 1% of the respondents. These findings suggest that Landrace maize cultivars are still preferred over improved maize cultivars from the study area possibly because of their adaptability to harsh agro-climatic conditions (water stress, salinity, high temperatures and poor soils) and their ability to be recycled. Previous studies also noted high usage of Landrace maize cultivars in rural areas because of their adaptability to local conditions and yield stability.^4,31,32

Figure 2. Production status of maize cultivars from the study area.

In the next section, factors that influence the selection of maize cultivars are presented focusing on the main cultivars grown from the study area (Landrace − 65%, GM − 31%).

Factors That Influence the Selection of Maize Cultivars

This section presents results of the multivariate probit regression as summarized in Table 2. The correlation matrix from the multivariate probit model of the GM and Landrace maize cultivars reveals that the two outcomes (selection of GM and Landrace maize cultivars) are negatively correlated. Interdependence of the two maize cultivars is therefore confirmed, which suggests substitutability of GM and Landrace maize cultivars by rural farming households subject to their socio-economic, institutional and technical constraints.^60–63

Table 2. Multivariate probit regression results for factors that influence the selection of maize varieties.

Independent Variables	Dependent Variables
	GM Maize Cultivars			Landrace Maize Cultivars
	β	dy/dx	P>|z|	β	dy/dx	P>|z|
(1) Rainfall	.6837	.2679157	0.000***	−.5343	−.1877363	0.000***
(2) Gender	.1107	.031288	0.488	−.0444	- .0077287	0.860
(3) Age	.0066	.0023425	0.154	−.0080	−.0026102	0.102
(4) Marital Status	.0411	.0120139	0.649	.0116	.0026008	0.920
(5) Household size	.2936	.0119641	0.036**	.0002	.0002821	0.960
(6) Education	.2258	.0870306	0.001***	−.2500	−.0852076	0.001***
(7) Monthly income	−.0016	−.0038889	0.857	−.1004	-. 0399088	0.059*
(8) Employment Status	−.2314	−.0956058	0.041**	.1020	.0545556	0.242
(9) Size of arable land accessed	.07961	.0359638	0.077*	−.0450	−.024282	0.219
(10) Number of Livestock	.0006	.0005886	0.553	.0048	.0022804	0.025**
(11) Livestock ownership	.1272	.0374572	0.557	.1748	.068508	0.298
(12) Access to TV	−.0072	−.0135197	0.765	.0629	.0412169	0.356
(13) Access to cell-phone	.3911	.1311579	0.012**	−.3323	−.0929117	0.075*
(14) Access to radio	.1336	.550207	0.183	−.1787	−.0693784	0.085*
Constant	−2.0785	-	0.000	1.6880	-	0.000
Number of observations = 650 Wald Chi^2 (28) = 101.26 Log likelihood = −642.05354 Prob > Chi^2 = 0.0000

Note: *, ** and *** are respectively at the 10, 5 and 1% levels significant p-values.

Rainfall

Results reveal that rainfall positively influences the selection of GM maize cultivars, while also negatively influencing the selection of Landrace maize cultivars among rural maize farming households. A unit increase in precipitation is associated with 27% chance of increasing the selection of GM maize cultivars and 19% probability chance of decreasing the selection of Landrace maize cultivars ceteris paribus. These findings reveal a substitution effect of Landrace maize cultivars by GM maize cultivars as rainfall increases in rural farming areas. GM maize cultivars that are not breed for drought tolerance require high amounts of water because of their high yield traits. Thus far, as rainfall increases GM maize production becomes more productive compared to Landrace cultivars assuming other agronomic variables are in place (high inputs and good management). Rational rural maize farming households are therefore expected to consider GM maize cultivars over Landrace maize cultivars as rainfall increases for the purposes of maximizing yields. These results are consistent with comparable literature that improved maize varieties (which broadly represents cultivars like GM, conventional hybrids and OPVs) thrive better under conditions of adequate moisture, good pest management practices and fertile soils.^6,14,15,64

Household Size

The results indicate a positive relationship between household size and selection of GM maize cultivars at 5% significance level. Per every positive unit increase in household size, holding other independent variables constant, results reveal a 1.2% chance of selecting GM maize cultivars among rural maize farming households. The results suggest that bigger household sizes are more likely to consider the use of GM maize cultivars in comparison with households with few members. Given the labor saving properties of GM maize cultivars, no manual weeding (HT maize) and reduced spraying (Bt maize),^23,26 the observed correlation may be explained by the food security and income potential of GM maize. Larger households are therefore logically expected to select GM maize cultivars over Landrace maize cultivars to take advantage of their high yields, which may address their food access and generate the much needed household income after selling surplus maize (more mouths to feed). Comparable previous studies suggest that large households are likely to adopt high-yielding crop varieties because of the food security incentive as the main driver.^42,65–68 Labor benefits may not largely apply for the stacked (Bt and HT) GM maize cultivars for the labor intensive exercises (weeding and insect sprays) are significantly reduced.

Education

There is a negative relationship between educational level and the selection of Landrace maize cultivars and a positive relationship with the selection of GM maize cultivars at 1% significance levels. Per every unit increase in education level ceteris paribus, there is an 8.5% chance of decreasing the selection of Landrace maize cultivars and an 8.7% chance of increasing the selection of GM maize cultivars among rural maize farming households. The results indicate a substitution effect of Landrace maize cultivars by GM maize cultivars as education levels increase among the rural farming households. The observed association suggests that education enhances rural farming households to understand benefits associated with new technologies like GM maize (high yields, reduced insect spraying, reduced manual weeding), which may condition their decision to select them over Landrace maize cultivars. In addition, educated households are more likely to have access to information on new technologies (GM maize) and gainfully employed than their uneducated counterparts, enhancing them to purchase expensive GM maize and resources to support their production. Several comparable studies on improved maize cultivars also highlight the tendency of more educated farmers to shun production of Landrace maize cultivars in favor of improved maize cultivars.^45,68–71 This is mainly because of access to information,^28,72 capital⁷³ expected higher yields⁷⁴ and positive perceptions of GM technology.^75–77

Monthly Income

Results reveal a negative association between monthly income and the selection of Landrace maize cultivars at 10% significance level. A unit increase in monthly income is associated with a 4% chance of decreasing the selection of Landrace maize cultivars holding all other factors constant. As income increases, the social classes of human beings also change, normally migrating from old technologies (commonly viewed as inferior – Landrace maize cultivars) to new technologies (commonly viewed as modern). In this case, rural farming households most likely substitute Landrace maize farming with other modern livelihood activities leveraging their income (spaza shops, transport business, etc.). Previous similar studies also reached the same conclusions that new techniques improve farmers’ income.^78,79

Employment Status

The results show a negative relationship between employment status and the selection of GM maize cultivars at 5% significance level. Per every positive unit change in the employment status of the household head ceteris paribus, results reveal a 10% chance of decreasing the selection of GM maize cultivars among rural maize farming households. Formal employment opportunities tend to discourage farmers from engaging in productive agricultural activities like GM maize farming because of time constrains. These results are consistent with those of Wang et al.,⁸⁰ who noted that in Kenya, households who had other off-farm employment opportunities were less likely to adopt new maize varieties compared with their counterparts who were mainly engaged in agriculture as a main livelihood source. More recently, Zamisa and Taruvinga⁸¹ also noted an inverse relationship between employment status and participation in GM maize production in South Africa because of time constraints.

Size of Arable Land

A positive association between land size and the selection of GM maize cultivars is revealed at 10% significance level. Results indicate that a positive unit increase in land size is associated with 3.6% chance of increasing the selection of GM maize cultivars ceteris paribus. Given arable land size limitations in rural areas, available arable land competes for old and new maize cultivars, where farming households assess the relative risk level associated with new maize cultivars. As access to arable land increases, rural farming households tend to allow the introduction of more new risk maize cultivars without substituting old trusted cultivars. Also, GM maize cultivars require mandatory isolation in line with available regulations, which may be more suitable to those farmers with large land sizes. Similar comparable findings were also noted by previous studies.^42,74,81

Number of Livestock

The number of livestock is positively correlated with the selection of Landrace maize cultivars at 5% significance level. Results indicate that a positive unit increase in livestock numbers is associated with a 0.2% chance of increasing the selection of Landrace maize cultivars ceteris paribus. There is a huge complementary relationship between the production of Landrace maize cultivars and livestock in rural areas (drought power, manure, stock supplementary feed). These results are not surprising given the fact that the majority of smallholder farmers in the study area resides in areas with poor soil fertility and are constrained financially to buy fertilizers and pay for tractors to assist with tillage operations. Thus, manure from livestock become very handy as a basal dressing in the production of Landrace maize cultivars. Similarly, livestock such as cattle more often benefit from maize by-products as a source stock feed. Similar findings were noted by Salasya et al.,⁸² arguing that increased livestock units improve availability of manure for the production of maize in Kenya, reducing on the need to buy fertilizer by smallholder farmers.

Access to Cell Phone

Access to cell phone has a positive influence on the selection of GM maize cultivars and an inverse relationship with Landrace maize cultivars at 5% and 10% significance level, respectively. Per every positive unit change in access to cell phone ceteris paribus, results indicate a 13% chance of increasing the selection of GM maize cultivars. Conversely, per every positive unit change in access to cell phone, the results suggest a 9.3% decrease in the selection of Landrace maize cultivars holding other independent variables constant. These results show a substitution effect of Landrace maize cultivars by GM maize cultivars as access to cell phones increases among the rural farming households. Cell phones are the emerging ICT devices used in rural areas for internet access and accessing crop production and market information. Rural farmers with access to cell phones thus far, receive a lot of crop production information and new technologies. Previous empirical studies have also indicated the same, that household ownership of mobile phones increase the adoption of new maize varieties, as they are vital sources of information for new technologies.^42,83 As such, mobile phone services are key strategic platforms that are normally used for public education on adoption of new technologies such as GM maize cultivars. The negative association between cell phones and Landrace maize cultivars may suggest non/low internet coverage of traditional maize cultivars deemed to be old fashioned and inferior.

Access to Radio

Lastly, results also indicate that access to radio has a negative relationship with the selection of Landrace maize cultivars at the 10% significance level. A positive unit change in access to radio is associated with a 6.9% decrease in the selection of Landrace maize cultivars ceteris paribus. This may be because nowadays mass media such as radio are mainly used to convey information about adoption of new technologies over traditional technologies deemed inferior and old fashioned like Landrace maize cultivars. Previous studies argued that farmers who had access to assets such as television and radio were more likely to get information on new farming technologies like improved maize cultivars and thus increase their level of adoption.^9,42,83 These studies suggest non/low coverage of traditional technologies by mass media in favor of new technologies. Access to these mass media platforms (marketing new technologies) by rural farming households creates a narrative that discourage the selection of traditional maize cultivars.

Conclusions and Recommendations

The study estimated factors that influence the selection of maize cultivars among rural farming households. The study noted that a majority of rural farming households from the study area are still producing Landrace maize cultivars compared to improved maize cultivars. The study also revealed that the selection of GM maize cultivars among rural farming households is positively influenced by the amount of rainfall, household size, education, size of arable land and access to cell phones. Employment status negatively influences the selection of GM maize cultivars. The selection of Landrace maize cultivars is negatively influenced by the amount of rainfall, education, income, access to cell phone and radio. The number of livestock on the other hand positively influences the selection of Landrace maize cultivars among rural farming households. The study therefore makes the following conclusions: firstly, although GM maize cultivars are not grown by a majority of rural households, their uptake may be promoted in areas with high rainfall using mass media communication platforms like cell phones. Secondly, although Landrace maize cultivars are very popular among rural farming households, their production may decline in high rainfall areas as they are substituted by GM maize cultivars highly promoted through mass media communication platforms like cell phones and radio. They are also easily substituted by GM maize cultivars as the education of rural farming households increases.

The study makes the following recommendations: the promotion of GM maize cultivars in rural farming areas should be in high rainfall regions where their genetic potential is maximized. Arable land size in such areas should be addressed to accommodate the isolation requirements and the adoption processes without substituting other crops. The promotion of Landrace maize cultivars in rural farming areas should be in low rainfall regions where they are highly adaptable compared to GM maize cultivars. Mixed farming (livestock and maize) in such areas should be promoted to enhance the complementarity of the two enterprises (drought power, manure and stock feed).

Acknowledgments

We acknowledge the support and cooperation of our respondents and the various communities visited for data collection and our research assistants who aided our data collection.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.