Legume integration in smallholder farming systems for food security and resilience to climate change

PLOS One | https://doi.org/10.1371/journal.pone.0327727 August 13, 2025 5 / 20 collectors failed to use the tool or had no smartphone, we employed paper-based data collection and later incorporated. In this case, an extra 10% contingency was added to the estimated household number, adding the total number of house- holds to 859. To ensure data quality, incomplete and wrongly filled forms were discarded and the remaining was combined with ODK data. Interviews included questions about all legume crop species produced by the household, followed by more in-depth questions about the target legume, the production constraints, the seasonal effect on the production and intensification method, and their perception of rhizobia inoculant technology. Data analysis The collected data were downloaded from the Kobo Toolbox repository, while data collected on paper base were filtered and manually curated before combining with the data downloaded from the tool. In total, 847 household responses were subjected to descriptive and inferential statistical analysis using R software version 4.3.2. Farm sizes were characterized as small (<1 ha) and large (≥1 ha) by slightly modifying the FAO category [27], while livestock owned by the farmers was expressed in tropical livestock unit (TLU) and categorized as small and large TLU according to Tache et al. [28]. Correla- tions between demographic variables and farmers’ productivity were analyzed using the agricolae package in R software.

Similarly, multiple linear regressions were used to evaluate the effect of different farm factors on the farmers’ productiv- ity. We employed the ggplot2 package to visualize the results of the analysis. Results Demography of the households To study the factors influencing farm productivity and livelihoods of the farmers, 847 HHs were interviewed, and their responses were analyzed. Accordingly, the ages of the respondents (HHs) ranged from 21 to 83 years of age. The aver- age family size of the HHs was 6.81, representing on average a large family size and a preponderance of rapid population growth in the country. HHs from pastoralist areas had the maximum number of the family size (20), with large farm size holding capacity. The HH size in adult equivalent scale (AES) was estimated to be 4.74. The educational qualifications of the participants were delineated as follows: individuals categorized as illiterate constitute 44.46%, those who have completed primary education account for 39.1%, respondents with secondary education represented 13.22%, and those who have attained post-secondary or university degrees comprised 3.31% (Table 1). Given this, the educational level of the head of the HHs had no effect on agricultural output in the area, since their productivity (in tons) did not differ probably due to lower proportion (3.31%) of the HHs heads with Post-secondary education and above. Relating the demography of the HHs to total Table 1. Household characteristics. Variables Category No. Percent Productivity (in tons) Sex of HH head Male 794 93.74 3.13 ± 0.70

Female 53 6.26 1.86 ± 0.41 Age of the HH Male 21-83 Na 1.86 ± 0.41 Female 30-55 Na 3.13 ± 0.70 HH head Educational Illiterate 377 44.51 2.32 ± 0.41 Primary 331 39.79 2.50 ± 0.43 Secondary 112 13.22 2.82 ± 0.51 Post-secondary 21 2.48 2.58 ± 0.87 University 7 0.83 2.23 ± 1.38 HH Size (AES) Mean of AES 4.74 ± 0.61 Na Na Where: HH = households; AES = Household size in adult equivalent scale, Productivity = Mean farm productivity per year, Na = not estimated. https://doi.org/10.1371/journal.pone.0327727.t001