The study sites were Chiredzi, Mudzi, Rushinga, Tsholotsho and Uzumba Maramba Pfungwe (UMP), Horticulture Research Station and M-Chiota, as indicated in Table 1. The study sites are in the semi-arid and high-rainfall environments. Ten spider plant landraces were obtained from various sources, including the national gene bank, Marondera University of Agricultural Sciences and Technology (MUAST), University of Zimbabwe and Horticulture Research Institute (HRI) (Table 2). These landraces were evaluated across study sites during the 2021–2022 summer season. The evaluation of the landraces included the assessment of descriptors such as stem colour, flower colour, petiole colour, leaflet lobbing and stem hairiness, as shown in Table 3.

The FFS utilises the learning cycle as the fundamental approach for knowledge acquisition. Participatory research methods are employed in the FFS for cropping systems research and variety selection, with support from extension officers, research institutes and development organisations.

The participants of the FFS have been actively involved in cultivating and consuming spider plant, providing valuable knowledge about spider plant production. To understand the important traits and preferences of spider plant, focus group discussions were conducted with the farmers in each FFS. Initially, these discussions aimed to gather information about key characteristics of spider plant, with a specific goal of compiling a list of their important traits. The discussions then progressed to determine the relative importance of each trait using a scale of 0% – 100% to quantify the extent to which each trait contributes to overall preferences.

Participatory variety selection activities were organised across the study districts to evaluate different spider plant landraces and discuss their production. During the season, individual farmers ranked the varieties according to a few traits and the final assessment rank. The ranking scale used ranged from 1 to 5 (1 = very bad, 2 = bad, 3 = average, 4 = good and 5 = very good). The ranking was translated into the local languages of Shona and Ndebele, which are spoken in the study areas. The percentage of farmers who ranked for individual traits was determined for each trait to assess the strength of the landraces in the farmer environments. The farmers then met to perform a final ranking of the varieties on the basis of their seasonal assessments, which they had conducted during the growth of the crop. At Chiredzi and HRI-R farmers had an agread ranking for each landraces, whereas, at other FFS individual farmers ranked each landrace.

The varietal score for each farmer at Tsholotsho, Rushinga, M-Chiota and Mudzi and the agreed and/or combined ranking of the FFS group at Chiredzi and HRI. The varietal scores from the individual farmers in FFS were subjected to chi-square tests to assess the significance, and Kruskal‒Wallis one-way analysis of variance was used for the combined analysis and individual FFS site using Genstat 17th edition.

Ethical clearance to conduct this study was obtained from the Ministry of Lands, Agriculture, Fisheries, Water and Rural Settlement, Department of Research and Specialist Services (DR&SS), Horticulture Research Institute.

The study sites are in agroecological zones IIA, IV and V, which are the high-rainfall and semi-arid tropics of Zimbabwe and are known for their rich diversity of indigenous vegetables.

Farmer field schools were established with the aim of facilitating learning and experimentation among farmers. More than 80% of the farmers who participated in the FFSs had at least 3 years of farming experience, and they formed collaborative groups. Additionally, a community seed bank was set up at each FFS in semi-arid regions to enable farmers to share seeds, while those farmers in high-potential areas rely on the HRI as their source of vegetable seeds. During the focus group discussions, farmers shared their indigenous knowledge systems to identify the significant traits of spider plant cultivation, as listed in Table 4. Germination of spider plant was highlighted as an important trait in four FFSs, namely, Rushinga, Mudzi, HRI-R and Tsholotsho. Early plant growth rate is an important trait in Rushinga, HRI-R and M-Chiota. Profuse branching was universally regarded as essential across all FFS locations. Farmers expressed the desire for minimal leaf damage to maintain good leaf appearance in terms of pest and disease tolerance. Early leaf harvesting was also considered an important trait by farmers in most of the FFSs. Plant height is an important trait in Rushinga, Mudzi, HRI-R and M-Chiota FFSs. Leaf size was considered important at almost all FFSs, as larger leaves are preferred by farmers. The number of pods, which is linked to seed yield, was particularly important in Rushinga, Mudzi, UMP and Tsholotsho. HRI-R and M-Chiota had similar important traits among the FFSs than among the other FFSs in dry regions. Unlike those close to the HRI research station, the FFSs in drier regions presented additional traits related to seed availability, such as pod number and seed yield (Table 4).

Fourteen spider plant preference traits were further prioritised on the basis of the scores and prioritised among the FFSs. In the Mudzi district, the FFS prioritised the first four traits in descending order: leaf size > taste > number of branches > late maturing. Conversely, in Tsholotsho, the order of breeding traits was germination > number of branches > size of leaves > late maturing. In UMP, the traits were ranked as follows: taste > number of pods > number of branches > size of leaves. In Rushinga, the order of traits was taste > number of branches > early leaf growth > tolerance to drought. At HRI-R, the most prioritised traits were in the following order: number of branches > size of leaves > germination > taste/late maturity. In the M-Chiota FFS, the traits were ranked in the following order: early leaf growth > taste/size of leaves > late maturity. The most frequently mentioned breeding traits among the FFSs were number of branches (mentioned in 6 FFSs), size of leaves (6 FFSs), taste (5 FFSs), late maturity (5 FFSs), pest and disease tolerances (4 FFSs), number of pods (4 FFSs) and dark green leaves (3 FFSs). On the contrary, the least common traits were larger grain size (1 FFS), stem hairiness (1 FFS), green stems (2 FFSs), tolerance to drought (2 FFSs) and taller plants (2 FFSs). When the total scores across the FFSs were considered, the order of importance for the breeding traits was as follows: the size of the leaves was the most highly ranked trait, followed by the number of branches, taste, late maturity, germination and early leaf growth, which was ranked sixth (Table 5).

Farmers preferred landraces such as CGNPGRC353, CGMRGP-Marondera and CGKENYA, which exhibited minimal leaf and shoot damage (Table S1). In terms of plant height, the landraces CGNPGRC355, CGGUR and CGKENYA were highly preferred, suggesting a preference for taller plants. Similarly, the landraces CGNPGRC356, CGMRGP-Marondera and CGKENYA were favoured for their larger leaf sizes (Table S2). The landraces CGGUR, CGMRGP-Marondera and CGUZG1 were selected because of their relatively high branching habits. The individual FFS-selected landraces were in the following order: Tsholotsho significantly (p < 0.001) preferred CGGUR (4.09), Rushinga significantly (p < 0.001) preferred CGKENYA (4.22) and Chiredzi significantly (p < 0.001) preferred CGMRGP-Marondera (4.35). The high-potential areas FFSs, HRI-R equally preferred CGMRG1 and CGKENYA (5.0), whereas M-Chiota showed significant (p < 0.001) equal preferences for CGZIM and CGUZG1 (5.0). The across-site analysis revealed significantly (p < 0.001) greater preferences for CGMRG-Marondera (3.9) > CGMRG1 (3.39) > CGNGRC355/CGKENYA (3.34) (Table 6).

This study presents findings from FFSs concerning indigenous knowledge and cultivation of spider plant, which is well adapted to the farming systems of the participants. Spider plants have untapped potential in terms of food and nutrition security, revenue creation and livelihood improvement for rural, peri-urban and urban farmers. They fetch high prices when sold in local and international markets, including restaurants, hotels and supermarkets (Mwadzingeni et al. 2021). Participatory rural appraisal methods were employed to prioritise breeding goals for spider plant improvement, a recognised approach for transitioning agriculture from subsistence to market-oriented enterprises in sub-Saharan Africa (Campbell 2001).

The size of the leaves was the most important breeding trait (Table 5), indicating its economic value in terms of production and consumption. Leaf size is also a significant trait in other African indigenous vegetables (AIVs) (Mncwango et al. 2021; Nakyewa et al. 2021). The second-ranked trait, branching, was recognised as a crucial trait by the FFSs, as it results in a bushy plant with numerous harvesting points, a trait also valued in vegetable amaranth (Dinssa et al. 2018; Mncwango et al. 2021; Ndinya et al. 2020). Taste preferences varied among the FFS participants, with some favouring a bitter taste and others not. Further studies are needed to explore preferences on the basis of gender, age group and regional factors, as cultural influences and food availability may play a role (Jan van Rensburg et al. 2009; Thovongi et al. 2021). Additionally, variability in tannin content, which governs the bitterness, has been identified among Zimbabwean spider plant accessions, suggesting the need to involve a diverse range of genotypes and consumers in future studies (Kutsukutsa et al. 2014).

Early growth rate was identified as an important breeding trait, with farmers highlighting the benefit of combining it with late flowering. These two traits extend the duration of the leaf harvesting period, allowing for multiple harvests during the season. Similar considerations regarding the length of the harvesting period have been reported in studies on amaranth (Dinssa et al. 2022). Germination has emerged as a priority breeding trait, as poor germination due to seed dormancy negatively impacts stand establishment and yield (Shilla et al. 2016). Previous studies on PVS of spider plant also emphasised the importance of good germination, while poor germination led to the rejection of a variety (Ndinya et al. 2020). To address this production challenge, there is a need for high-quality seeds, which can be achieved through seed treatment/coating, improved storage practices and the development of improved varieties (Shilla et al. 2016; Sohidji et al. 2020; Thovhogi et al. 2021; Zharare 2012). Previously, Asian germplasm was shown to have a relatively high germination percentage; hence, Asian germplasms can be incorporated into the development of varieties with relatively high germination rates (Blalogoe et al. 2020). A number of seed pre-treatment options have been identified, which increase germination and even seedling vigour, including the use of gibberellic acid (Sohidji et al. 2020; Tapfumaneyi et al. 2023). The capacity to produce an abundance of pods and achieve good seed yield is crucial, particularly for farmers in drier areas. Similar observations of a lack of spider plant seeds have been reported in Namibia and Malawi (Chataika et al. 2022). Participatory research has the advantage of exposing farmers to diverse germplasm, which can be harvested and stored for future use, and the seeds can be kept in their seed banks and shared in the future. This leads to improved adoption, as farmers would have experimented with the varieties. Traits such as green stems, stem hairiness and larger grain size were ranked as lower breeding priorities. Larger grain size may require seed coating to improve seed handling during planting to minimise seed wastage and improve germination. Stem hairiness could be associated with pest tolerance. The need for green stems requires further assessments and discussions with farmers.

Among the different FFS sites, the most preferred genotype was CGGUR in Tsholotsho FFS, which exhibited green stems and petioles; high branching, tall plants; and minimal leaf damage. In Rushinga, farmers preferred CGKENYA, which displayed higher plant height, larger leaf size and minimal damage. In Mudzi FFS, the preferred landrace was CGMRG-Marondera, which displayed low leaf damage, high branching, green stems and medium hairiness. In Chiredzi, the best cultivar was CGMRG1, which has the following traits: greater leaf size and minimum leaf damage, a mixture of green and purple stems, and medium leaf hairiness. The most preferred landraces in M-Chiota were CGZIM and CGUZG1, whereas HRI-R preferred CGMRG1 and CGKENYA. The selection of landraces was in agreement with the need to prioritise the size of leaves, number of branches, taste, late maturity, early harvesting and greater plant height, which has been reported in previous studies by Ndinya et al. (2020) and Chataika et al. (2022). The selections based on all FFSs were in the following order of preferences: CGMRGP-Marondera > CGMRG1 > CGKENYA/CGNPGRC355. This study provides preliminary results concerning farmer preferences, and further studies might consider more on-farm sites, urban consumers, retailers and hotels.