This article provides a description of intra-household survey data that were collected in Uganda and Tanzania in 2014 and 2015, respectively. The surveys were implemented using a structured questionnaire administered among 585 households in Uganda and 608 in Tanzania. Information on decision making processes in agricultural production was collected from the principal adult male and female decision-makers in each household. The survey consisted of two parts. Firstly, the decision-makers, both male and female of each household were jointly interviewed. Secondly, individual interviews were carried out, questioning the decision-makers separately. The datasets include both household and individual level data containing numeric, categorical and string variables. The datasets have been shared publicly on the Harvard dataverse.

Currently, there is a high demand for amaranth due to its ability to withstand harsh climatic conditions, making it an ideal crop in the changing climate. There is also increased awareness and education on its nutritional and overall health benefits, and the availability of improved recipes. However, the presence of hazards can hinder the commercialisation of amaranth, which is in most cases traded informally. Food safety issues along the amaranth value chain should, therefore, be addressed to cope with both production and safety demands. The objective of this study, therefore, was to develop a Hazard Analysis and Critical Control Point (HACCP) plan for hazards in the amaranth value chain in Uganda. The seven principles outlined by Codex Alimentarius were followed to develop the HACCP plan. A tree diagram was further used to identify each potential hazard at each processing stage and Critical Control Points (CCPs) along the chain. For the CCPs identified, reliable control mechanisms and corrective actions were established to fulfil the requirements set by the critical limits to guarantee the safety of the products. Verification and records systems were proposed to determine the effectiveness and traceability of the HACCP plan. For each of the identified CCPs, samples were collected purposively and analysed for chemical and microbial contaminants. From the analysis, fifteen processing stages, starting from the land section to cooking and serving, were identified. Out of these, eight stages were defined as CCPs. These were site selection, land and seedbed preparation, irrigation, market display/humidity control, washing before preparation, chopping, cooking, and holding time and serving. At CCP 1, soils were contaminated with lead and cadmium, mercury and aflatoxins but at considerably low levels. At CCP 2, organic fertilisers were only contaminated with E. coli. At CCP3, E. coli was present in irrigation water. Heavy metals were also present in the irrigation water but were below the critical limits. At CCP4, E. coli was absent in water and display surfaces. E. coli was, however, present on raw amaranth. S. aureus was detected on vendors' hands. At CCP5, water was not contaminated with E. coli. At CCP6, only personnel hands were infected with S. aureus and Enterobacteriaceae. No contamination was detected in CCP7 and CCP8. Strict control of E. coli in manure and water and S. aureus and Enterobacteriaceae on personnel hands is required to ensure the amaranth value chain attains good food safety output.

Common beans (Phaseolus vulgaris L) may be contaminated with heavy metals and aflatoxins. Cooked beans may also be contaminated with micro-organisms due to poor hygiene and sanitation practices. Hazard Analysis and Critical Control Point (HACCP), which is a globally recognised food safety program, was proposed as a suitable program to minimise/eliminate the risk of contamination. Therefore, the objective of this study was to develop a HACCP plan for dry common beans in Uganda and an accompanying food safety toolkit. The seven principles of HACCP as outlined by Codex Alimentarius were followed to develop a HACCP plan for the dry common beans value chain in Uganda. A decision tree diagram was further used to identify each potential hazard at each processing stage and Critical Control Points (CCPs) along the chain. The identification of the CCPs was further supported by an evaluation of the actual risk and severity of the hazard. For the CCP identified, reliable control mechanism and corrective actions were established to fulfill the requirements set by the critical limits to guarantee the safety of the products. Verification and records systems were proposed to determine the effectiveness and traceability of the HACCP plan. For identified CCPs, a co-creation methodology was used to develop the food safety toolkit. This was carried out in four sessions that included a background of the chain actors' ambitions to determine the suitability of the toolkit, assessment of CCPs, expert advice on the CCP and an exercise to develop concepts for each CCP. From the analysis, fourteen processing stages starting from land selection to cooking and serving were identified. Out of these, four stages were CCPs. These were land selection and preparation, storage, post-harvest drying, and cooking and serving. Hazards at the CCPs included heavy metals, mycotoxins, and micro-organisms such as S. aureus, E. coli, and Salmonella spp. A combination of good hygiene and sanitation practices and good agricultural practices were recommended as control measures against the hazards. To further equip the value chain actors with mitigation strategies, a food safety toolkit whose usefulness is to give the actors a systematic means to control identified CCPs was developed. In this regard, the toolkit and HACCP plan will complement each other. From the study results, implementation of the toolkit, followed by an assessment of its uptake and impact on livelihoods and food safety risks is recommended.

BACKGROUND Due to increasing demand for livestock products in sub‐Saharan Africa, increasing livestock productivity is a priority. The core constraint is limited availability of feed of good quality. We assessed optimal harvesting time of three improved grasses, two Urochloa lines (Basilisk a selection from wild population, Cayman – a hybrid, a product of breeding) plus Mombasa, a Megathyrsus selection. All are released in Latin America and Kenya or in the registration in other regional countries. We assessed dry matter (DM) yields and quality at 4, 6, 8 and 12 weeks of age in two sites. RESULTS DM yields (in t ha−1) were of the order Cayman (9.6–14.3) > Mombasa (8.0–11.3) > Basilisk (5.5–10.2) in one site, and Cayman (6.4–9.7) > Basilisk (4.9–7.6) > Mombasa (3.3–5.9) at site two. The harvesting regimes produced DM largely similar for weeks 4 and 6, 6 and 8, 8 and 12. Across the sites quality was of the order Cayman > Mombasa > Basilisk for neutral detergent fiber (NDF), metabolizable energy (ME) and crude protein (CP). With increasing harvesting interval, MJ ME ha−1 and kg CP ha−1 were inconsistent across both sites, but significant differences returned for MJ ME ha−1 unlike kg CP ha−1. CONCLUSIONS Harvesting at either 8 or 12 weeks is not recommendable as quality drops without an increase in DM yield that can compensate despite doubling and tripling time respectively, compared to 4 weeks. We recommend harvesting at 4 through 6 weeks for any of the three grasses based on yield against time, and demand at the intensified cut‐and‐carry smallholder systems. © 2021 The Authors. Journal of The Science of Food and Agriculture published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.

Globally, there has been an explosion of data generation in agriculture. With such a deluge of data available, it has become essential to create solutions that organize, analyze, and visualize it to gain actionable insights, which can guide farmers, scientists, or policy makers to take better decisions that lead to transformative actions for agriculture. There is a plethora of digital innovations in agriculture that implement big data techniques to harness solutions from large amounts of data, however, there is also a significant gap in access to these innovations among stakeholders of the value chains, with smallholder's farmers facing higher risks. Open data platforms have emerged as an important source of information for this group of producers but are still far from reaching their full potential. While the growing number of such initiatives has improved the availability and reach of data, it has also made the collection and processing of this information more difficult, widening the gap between those who can process and interpret this information and those who cannot. The Crop Observatories are presented in this article as an initiative that aims to harmonize large amounts of crop-specific data from various open access sources to build relevant indicators for decision making. Observatories are being developed for rice, cassava, beans, plantain and banana, and tropical forages, containing information on production, prices, policies, breeding, agronomy, and socioeconomic variables of interest. The Observatories are expected to become a lighthouse that attracts multi-stakeholders to avoid “not see the forest for the trees” and to advance research and strengthen crop economic systems. The process of developing the Observatories, as well as the methods for data collection, analysis, and display, is described. The main results obtained by the recently launched Rice Observatory (www.riceobservatory.org), and the about to be launched Cassava Observatory are presented, contextualizing their potential use and importance for multi-stakeholders of both crops. The article concludes with a list of lessons learned and next steps for the Observatories, which are also expected to guide the development of similar initiatives. Observatories, beyond presenting themselves as an alternative for improving data-driven decision making, can become platforms for collaboration on data issues and digital innovations within each sector.