The study was conducted in the Gert Sibande District Municipality in Mpumalanga, South Africa as shown in Figure 1. The district covers an area of 31,841 km², which makes it the largest district in the province. The district is divided into seven local municipalities, namely: Govan Mbeki, Chief Albert Luthuli, Msukaligwa, Dipaleseng, Mkhondo, Lekwa, and Dr. Pixley ka Isaka Seme. To the north, it is bordered by the Ehlanzeni and Nkangala District Municipalities, to the south by KwaZulu-Natal and the Free State, to the east by Swaziland and to the west by Gauteng. The major economic sectors are mining, agriculture, energy and manufacturing. The municipality is chosen because of its high concentration of subsistence farmers, and similarly, SLMP has been mapped out and adopted into this province long before now.

Cross-sectional data were used for this study. Data was collected between December 2019 and August 2020, using a semi-structured survey questionnaire validated by two agricultural economist experts (independent experts). A reliability test was performed on the questionnaire to ascertain its use. The questionnaire contained logic flow questions aimed at farmers' demography, information acquisition of SLMP, social groups and the adoption of SLMP. The survey was conducted through face-to-face interviews; each session with the farmers lasted 40 min. A representative sample size was determined, using Slovin's formula given in Equation (1) after which a total number of 250 questionnaires were administered to the maize farmers in the district by four trained enumerators who translated the questionnaire into local language for farmers to understand. A proportionate random sampling technique was used to select the sample size where each local municipality represent a stratum from which sample were randomly obtained. This was achieved by adopting a quantitative model, as presented below:

Where n is the sample size,

Table 1 shows the distribution of sample size collected according to each municipality or stratum.

Data were analyzed, using both descriptive and inferential statistics. Descriptive statistics such as mean values, standard deviation and percentages were used to describe the farmers' socioeconomics, information acquisition and adoption of SLMP. Subsequently, a seemingly unrelated bivariate regression (SUBP) and recursive bivariate probit (RBP) model were used as inferential statistics to investigate if the decisions to acquire information on SLMP are jointly determined and simultaneously lead to the smallholder maize farmers adopting SLMP.

In reality, the decision of a farmer to adopt sustainable land management practices preceded by information received or acquired on SLMP. This information is mostly shared and acquired by the social capital group. Thus, every member of a social group (mostly farmer groups) decides whether or not to make use of the information. Thus, there occur unobserved characteristics leading into endogeneity problem, in which failure to take account of it will lead into biased and spurious result (Owusu et al., 2020; Oduniyi and Chagwiza, 2021). However, if the net benefit associated with the adoption is positive or greater than otherwise, then the farmer decides to adopt the innovation. In this instance two procedures occur that are dichotomous and mutually decided or simultaneous to each other. This requires joint estimation of the two sets of procedures or equations relating to decision models which do not necessarily involve the same independent variables or regressors but contain the same error terms just like in the instance of seemingly unrelated regression equations (SURE) model. However, the dependent variables involved are the binary options; information acquisition on SLMP can be represented as (y₁ = 1) or otherwise (y₁ = 0). Correspondingly, for adopting or non-adopting SLMP can be represented by (y₂ = 1) or (y₂ = 0), respectively.

With Table 3 presents descriptive statistics of the sample of the explanatory variables used. The average age of the smallholder maize farmers in the study area was found to be 48 years, with an average of 10 years spent in school. The mean visit by an extension officer was found to be at least twice a month. The majority of farmers had an average of 11 years of farming experience and the mean farm size was found to be 123 hectares. Similarly, about 93% adopted at least one practice of sustainable land management while 71% claimed that they acquired information on SLMP.

The study examined the linkage of how information acquired by farmers on SLMP relates to adoption. The two models used explore variables considered to be exogenous to the two dependent variables. The exception is information acquisition in the RBP models, a matter that is subjected to econometric testing. From the SUBP result, the first procedure was to use Wald test (LR test) to evaluate the null hypothesis that rho is zero (0). The value of rho (5.90e-10) was significant at the 1% level (Chi-square = 35.1773, df = 1, p-value = 0.0000). The result indicated the probability that a farmer acquired information was indeed related to the probability of adopting SLMP through unobserved effects captured in the error terms of the models. The positive sign for rho in the SUBP model indicates that the two variables are complementary to each other. A way to think about these results is that information acquisition and the adoption of SLMP worked together as a strategy for improved productivity. Thus, there is a relationship between the information acquisition and adoption of SLMP. This finding is consistent with the submission of Huth and Allee (2002) and Moreno and Sunding (2003) who acknowledged that positive value for rho suggests a complementary decision variable.

In the RBP model, as shown in Table 4, the non-statistically significant (0.1160) results of the Wald test (LR test) for rho = 0 indicated that information acquisition is exogenous which suggested that the decision to acquire information and adopt of SLMP was not jointly decided. This is not surprising as some farmers who adopted SLMP did not acquire information, vice versa. The reason is not farfetched from the fact that most smallholder farmers practice sustainable agriculture unaware. They still practice primitive form of agriculture such as bush fallow, mulching, planting cover crops, crop rotation etc.; which are typical examples of SLMP, thus, they adopted SLMP unknowingly without acquiring information on SLMP. Hence, information acquisition and adoption of SLMP was not jointly decided.

The results from the SUBP model revealed that the number of years spent in school (education); access to agricultural extension service; number of agricultural extension visits and the number of years of farming jointly influenced information acquisition and adoption of SLMP. The years of farming or farming experience of a farmer was found statistically significant. The result shows that the lower the years of experience, the lower or less likely a farmer acquire information and/or adoption of SLMP, vice versa. This was in support to a report by Alam (2015) who reported that farmers with more experience in agriculture are more likely to adopt agricultural innovation.

The number of years spent in school signifies the education of the head of the household. The result in Table 5 shows that education was found to be statistically significant and it positively influenced information acquisition and adoption of SLMP. This advocates that an educated or literate household head farmer has a propensity to acquire more information and adopt SLMP. This is confirmed by the findings of LaFerrara (2002) and Haddad and Maluccio (2003) who reported that higher education encourages farmers to seek and acquire more information. This was contrary to Thuo et al. (2014) who affirmed that farmers with more years of education are less likely to adopt new agricultural practices.

Access to agricultural extension services was found to be statistically significant and it negatively affected information acquisition and adoption of SLMP. The result suggests that farmers with access to agricultural services are less likely to acquire information and adopt SLMP. The possible reason could be that, despite having access to the agricultural extension service, little or few information related to SLMP were shared. However, the farmers claimed to have acquired more related information on SLMP from a social capital group, such as farmer's group and cooperative group. This result goes against the study carried out by Katungi et al. (2008) in Uganda who reported that extension activity in the village is an important determinant of information exchange related to agricultural technologies among rural people.

The number of agricultural extension visits were statistically significant and positively influenced the information acquisition and adoption of SLMP. The result suggests that the more visits a farmer received the more likely he/she acquired more information about SLMP. Most farmers explained that more visits provide an avenue to ask questions on the information which they have acquired from the social capital group on SLMP, which becomes easy to adopt. Normally, a farmer with higher extension activity is more likely to engage in a two-way information exchange compared to those with less frequent extension activity. This result is buttressed by Ntshangase et al. (2018) and Oduniyi (2018), who affirmed that access to extension service is not enough; the intensity of the extension services is critical in determining the level of adoption.

Other factors that influence information acquisition and adoption of SLMP independently are:

The gender of the farmers was found to be positive and statistically significant (p < 0.05) in influencing information acquisition. The result explained that male farmers are more likely to acquire information on SLMP. The reason for this could be that male farmers dominate the farming industry and they have better access to agricultural resources such as land, agricultural inputs, including information acquisition. Female heads of households compared to their male counterparts are likely to be disadvantaged in their access to a social capital group that facilitates information flow. This result is confirmed by Katungi et al. (2008) who reported that female heads of households are expected to acquire less information on agricultural technologies and new practices compared to their male counterparts.

The age of the household heads was found to be significant (p < 0.01), and it contributes to information acquisition. This suggests that the older the household head, the better the chances of acquiring information on SLMP. This is not surprising as most old farmers are used to soil management practices and cultivation practices, such as mulching and crop rotation, which are an example of SLMP. Thus, it is easier for them to relate and acquire information on SLMP. This is interesting as it negates many findings that reported that old age discourages an individual from acquiring information (Alesina and La Ferrara, 2002; Godquin and Quisumbing, 2006). It should be noted here, that age and farming experience are synonymous. Thus, this conform to the earlier explanation that a farmer with few farming experiences is less likely to acquire information, while a farmer with more years of experience is more likely to acquire information.

The Marital status of the household head was statistically significant (p < 0.05) and negatively influenced information acquisition. This explains that a married household head farmer is less likely to acquire information on SLMP. Of course, a married household head will discuss information with his/her partner before acquiring information; however, in most cases, the influence of the other partner rejects the acquisition.

The cost attached to the adoption of SLMP was statistically significant (p < 0.01) and found to be positive; thus, influencing the adoption of SLMP. The study suggests that cost involvement increases the probability of SLMP adoption in the study area. The result is surprising as smallholder farmers are not rich and they are always skeptical to cost attached to farm innovation. It is always believed that smallholder farmers' willingness to pay for technology is far low. Although, some farmers claimed that if costs are involved with the opportunity of improving crop yield, then they would not mind paying for it. This result is conformed to Alemu et al. (2021) who reported that smallholder farmers' willingness to pay for SLMP in the Upper Blue Nile basin in Ethiopia was found to be 76%.

Similarly, a study conducted by Takele and Umer (2020) in Homosha, Ethopia found revealed that the total willingness to pay for SLMP among small-scale farmers was found positive and significant. The study concluded that household's willingness to pay more than 66 percent cost for any SLMP to improve agricultural production by examining their total willingness to pay for SLM. It is expected that if net benefits exceed the cost of adoption, then the farmer decides to adopt the SLMP. However, farmers have always looked at new technologies as a way to reduce costs. All the same, this result was supported by The Organisation for Economic Co-operation Development (OECD) (2001) which explained that to survive, farm production must be cost/price-driven. New technology is therefore needed to increase productivity. Farmers must keep up with improvements in technology to stay in business.