Farmers can effectively use downscaled seasonal forecasts accompanied by local historical climate data when supported by a structured participatory process (Dinku et al., 2018a). However, West African countries do not have the robust climate information (CI) necessary to guide climate adaptation actions (Hansen et al., 2011). Few weather stations exist and the limited data they produce is not edited, well-stored, or widely used (Traore et al., 2014). This step aimed to bolster national capacity to produce good quality, locally relevant CI with broad coverage.

National Meteorological Service (NMS) staff were trained to analyze historical climate records (using Instat/R-Instat) and predict crop yield using SARAA-O/H (Traore et al., 2011; Vintrou et al., 2014; Sultan et al., 2019). Where data gaps constrained access to high-quality local climate records, NMS staff were trained to use the Enhancing National Climate Services (ENACTS) approach. This overcomes data gaps by blending NMS station data with satellite and other proxy data to produce moderately high-resolution historical gridded data (Traore et al., 2014; Dinku et al., 2018a,b). Additionally, trainees learned how to conduct an inventory and to prioritize potential climate-smart options based on three key criteria: the three pillars of CSA (production, adaptation, and GHG mitigation); a cost-benefit analysis; and degree of scalability (Table 2; Bayala et al., 2018a; M'Bo et al., 2019). The prioritization exercise eliminated options that did not fit farmers' needs and highlighted farmer-tested options (Bonilla-Findji et al., 2018; Table 1). This testing process involved CI-based planning using seasonal forecasts long time before the onset of the cropping season and making tactical adjustments during cropping seasons through short-term forecasts (Dayamba et al., 2018; Clarkson et al., 2019). Finally, they graphically formatted the data in probabilistic terms to be easily understood by people across the literacy spectrum (Dinku et al., 2018a).

The analysis of local climate seasonality, trends, and variability provided insights into farmer and value chain actors' risk, as well as whether perceived changes in agricultural performance are driven by climate or environmental changes such as soil degradation (Dinku et al., 2018a). The resulting graphs were critical elements of the Participatory Integrated Climate Services for Agriculture (PICSA) trainings. Participatory Integrated Climate Services for Agriculture is an approach that uses historical climate records, participatory decision-making tools and seasonal climate forecasts to help farmers identify and better plan livelihood options suited to their own circumstances and climate conditions (Dorward et al., 2015).

Participatory Action Research uses past, present, and future CI to choose, test, and validate climate smart options for crops and livestock research and social innovations. In our case, four types of CI were used to inform the CSA selection and testing for the CSVs:

The participatory planning process was based on the TOP-SECAC (Trousse à Outils de Planification et Suivi-Evaluation des Capacités d'Adaptation au Changement Climatique) toolkit developed by the International Union for Conservation of Nature (IUCN) (Somda et al., 2014). The process in this step involves establishing a PAR platform for Climate-Smart Agriculture (PAR-CSA) actors, identifying technological climate adaptation options for on-farm testing, and participatory evaluation and validation, and helps translate the program's common objectives into contextualized actions in each of the benchmark sites in five countries: Burkina Faso, Ghana, Mali, Niger, and Senegal (Bayala et al., 2016; Sanogo et al., 2016). In preparation for the PAR, National Agricultural Research Systems (NARSs) scientists were trained in PAR to test and validate CSA options (Bayala et al., 2016, 2020). Farmers, extension staff, and NGO staff were also trained to use CI to plan activities (Dayamba et al., 2018; Clarkson et al., 2019).

Next, rural communities collaborated to create visions of the desired CSVs and how they would deal with the effects of climate change and related hazards. Participating farmers visited villages with a climate comparable (climate analog sites) to that of their predicted climate, enabling direct observation of production systems and exchanges with locals to offer ideas of potential adaptation approaches. Thus, the climate analog sites were also used to strengthen local actors' capacity to further analyze possible climate changes in their environments and plan for the desired future. The initial outputs were shared with regional rural development stakeholders to fine-tune the model and identify strategic actors. This also helped establish partnerships between researchers, extension services, non-governmental organizations (NGOs), private-sector actors, policy makers, and communities (Somda et al., 2016) founded on developing agricultural systems that promote vegetative cover and land restoration through sustainable intensification.

The selected CSA technologies and practices were then tested. Tested options were comprised of a combination of:

In most cases, successfully deployed activities were based on existing techniques or innovations in the community that were adjusted to fit the new context. The exception was Ghana, where land reclamation techniques were new to the communities involved. In this country, there was a tendency to screen a large number of techniques across integrated soil fertility management techniques (crop rotation, manure and inorganic fertilizer application), new soil and water conservation practices (ridges and tied ridges, stone lines, etc.), and new crop varieties (maize, soybean, and cowpea). Emphasis was placed on interventions that were likely to be more beneficial to women and resource-poor individuals, including nutritional education e.g., soybean preparation, village savings and loans groups, and income generating activities for short-term wins. In other countries, activities that promote women's involvement include gardening, tree propagation (grafting), fruit tree planting, tree products processing, value added post-harvest handling, dry season supplementary activities, and nutrition and health information.

Progress was participatorily evaluated at the middle and end of each cropping season via field visits and community meetings. The results were discussed at the end of each rainy season in annual review and planning meetings at both community and regional levels to make adjustments when and where needed. As a result, the list of options narrowed significantly (Tables 1, 2).

All participatory evaluation processes were supported by rigorous monitoring and evaluation (M&E) (Bonilla-Findji et al., 2018; Partey et al., 2018). Monitoring and evaluation (M&E) was designed and implemented in a participatory manner, so adaptation and mitigation activities are sustained by “adaptive behavior” leading to increased food security. Behavior change from program participation was monitored at farmer and community levels using the Most Significant Changes technique (Davies and Dart, 2005) integrated into a step-by-step toolkit for planning and M&E of climate change adaptation (Somda et al., 2011). Through storytelling, significant changes enacted by individual farmers and by gender-differentiated groups were gathered and analyzed. Changes were substantiated and used to learn about newly initiated behaviors and any remaining constraints farmers face in maintaining these new behaviors.

Through this work, climate information services (CIS) have increasingly become a key entry point to guide farmers' decisions and selections of crops, varieties, agro-sylvo-pastoral systems, technologies, production area, degree of intensification, production timelines, and investment levels (Sanogo et al., 2016). Climate information services are also more available and accessible by farmers thanks to national meteorology offices, mobile phone text messaging, rural radio broadcasts, and the various PICSA trainings.

Capacity building to bolster stakeholders' understanding of local climate data and evidence-based decision making (Dorward et al., 2015; Dayamba et al., 2018; Clarkson et al., 2019; ICRAF, 2019) was an ongoing component of this work. A total of 18 PICSA trainings, 4 CSA trainings, 8 data collection trainings, and 2 M&E trainings directly reached 630 extension agents, researchers, NGO staff, NMS staff, and NARS staff. Stakeholders were trained to generate and use quality historical CI at the relevant scale; assess the climate-smart potential of projects and programs; PAR; crop yield prediction modeling; CIS delivery; and leverage participatory M&E methods (Bayala et al., 2016; Somda et al., 2020). These trainings effectively developed local expertise to help sustain regional climate-smart action, thus building a foundation for CSA mainstreaming in future projects or programs (Ouédraogo et al., 2019; Bayala et al., 2020). As a result, there is evidence of CI use (Dayamba et al., 2018; Clarkson et al., 2019) and even willingness to pay for access (Ouédraogo et al., 2018). In addition, farmers received training through demonstration plots, farms-of-the-future, farmer field schools, field days for farmer-to-farmer learning, traditional annual farming festivals, local radio programming, mobile phones, etc. (Bayala et al., 2016).

Peer-to-peer conversation and collaborative learning methods within and between communities are powerful participatory ways to strengthen existing social learning approaches for scaling CSA (Somda et al., 2020). Through our M&E processes, the IUCN social learning approach team captured gender-differentiated social learning methods, institutions, and sociocultural events (Somda et al., 2020). The prediction of scalable production needs by administrative unit (sub-national, national, and regional) through the use SARRA-O/H models was meant to enabling management to plan for risks associated with the prospective volume of production (Traore et al., 2014). Finally, the co-developed CSA options have contributed to the compendium of CSA technologies developed by the CCAFS program for scalable CSA (CCAFS/ICRAF, 2020; Figure 2). These proven climate-smart technologies serve to design evidence-based effective CSA profiles and investment plans to catalyze their broad adoption at both national and regional scales (Figure 3).

There is a clear need for institutional support on CI delivery. Issues exist surrounding organizational landscape, organizational capacities, and human resources (Figure 3). Expanded weather station networks and human capital development to provide advisory services on time and at the relevant scale for planning or data-based decision making would represent significant improvements. However, skepticism around the accuracy of CI is a further obstacle to its use in decision making at all levels. In order to address this, CI must be available, and stakeholders must know how to apply the information. Effectively designed policies and investment plans to assess impacts can improve user confidence. Finally, national education and research systems must embrace climate issues with adapted curricula and production systems (Figure 3).

Strong partnership is crucial to effective implementation of PAR-CSA at the national level. Our experience suggests that the engagement of 10–12 national partners is indispensable (Bayala et al., 2016). Partnerships at the regional level are equally vital, including coordination with CCAFS programs, and CGIAR centers such as ICRAF, and other international and regional organizations like IUCN, and AGRHYMET, respectively. Such partnerships enable broad changes in knowledge, attitudes, and practices (KAP) (Somda et al., 2011). Finally, connecting PAR-CSA achievements to national policies through a science-policy dialogue is key. Some progress has already been achieved in our pilot areas in this regard, and much remains to be done (Partey et al., 2018; Raile et al., 2019; Zougmoré et al., 2019).

Throughout implementation, a concerted effort to connect activities with national adaptation and mitigation priorities is vital. In our experience, one major challenge has been capturing a breadth of farmer experiences in sampling replicates while controlling costs (Figure 4). In addition, time constraints and a lack of perceived benefits mean farmers tend to execute activities individually on their farms rather participating collectively. This is partly due to the fact collective actions on communal lands are associated with governance issues, including land and tree tenures (Bayala et al., 2016).

Two main categories of barriers to CSA adoption were identified. The first category is about access to inputs/knowledge in broad terms: limited access to information, knowledge and new skills on CSA options, limited availability, and access to inputs and equipment, all together leading to poor technical capacity (Ouédraogo et al., 2018). The second category relates to the institutional, cultural, policy, and regulatory environments as well as governance. The last category has implications about moving from plot or household levels to landscape scale as well as on power relations.

To address these barriers, knowledge sharing and communication are crucial to effective CSA. A strong knowledge bank, such as Evidence for Resilient Agriculture (ERA; available at https://era.ccafs.cgiar.org/) may be crucial to supporting risk-informed adaptation planning while successes and lessons learnt inform efficiency in future work. Outreach, meanwhile, helps leverage investments. The high upfront costs of CSA make funding an integral part of scalability. Establishing clear links from local needs to national or regional strategic directives can help bring funding to the local level. Technical competency through capacity building is also indispensable to ensure that available funding is utilized in evidence-based programming at the appropriate scale. Communication may be improved by: (1) strengthening the communication capacity of national programming teams; (2) making further use of information communication technology (ICT), including local communication channels and context specific program information dissemination (Etwire et al., 2017). This work has generated 32 publications, including peer-reviewed articles, manuals, occasional papers, info notes, a special issue, and more.

In some cases, these publications have ultimately informed scalable CSA projects. For example, evidence from our work at the regional level catalyzed the European Union-funded Projet d'Appui à la Résilience Climatique pour un Développement Agricole Durable (PARC-DAD) and the World Bank-funded Projet d'Appui à l'Agriculture Sensible aux Risques Climatiques (PASEC). Our results also helped bring about the UTZ/Rainforest Alliance-funded Consultancy in view of developing a training curriculum on climate-smart agriculture for small cocoa farmers in Côte d'Ivoire. Further, the Conseil Ouest et Centre Africain pour la Recherche et le Développement Agricole (CORAF), funded under the West Africa Agricultural Productivity Program (WAAPP), a project on Capacitating Stakeholders in Using Climate Information for Enhanced Resilience in the Agricultural Sector in West Africa (CaSCIERA-TA). Based on the CaSCIERA-TA experience, the Benin team has expanded its work to other parts of the country and Burkina Faso's Meteo service has developed its independent project on Strengthening national capacities for Early Warning System (EWS) Service Delivery in Burkina Faso. Similar scaling up initiatives are ongoing at both national and regional levels across the region.

Policy support is needed for such initiatives to spread at lager scale (Raile et al., 2019), and in many cases a clear link between field action achievements and the national policy is lacking. Synthesizing the results of in-field PAR in a format that speaks to policymakers can help ameliorate this issue. We found that multi-stakeholder national science-policy dialogue platforms on CSA at the regional level helped filling this gap. These dialogue platforms used scientific evidence to create awareness of climate change impacts on agriculture and to advocate for CSA development plans (Zougmoré et al., 2016; Partey et al., 2018). Compendia of national agricultural policies as well as a plan, program, or strategy for each country were developed based on desk reviews and meetings with decision-makers. Similarly, catalogs of CSA options were produced that help farmers adapt to climate change and variability. These two actions targeted relevant development planning for climate change and CSA mainstreaming toward agricultural policy or investment initiatives (Zougmoré et al., 2019).