Controls on timescales of soil organic carbon persistence across sub-Saharan Africa.

Given the importance of soil for the global carbon cycle, it is essential to understand not only how much carbon soil stores but also how long this carbon persists. Previous studies have shown that the amount and age of soil carbon are strongly affected by the interaction of climate, vegetation, and mineralogy. However, these findings are primarily based on studies from temperate regions and from fine-scale studies, leaving large knowledge gaps for soils from understudied regions such as sub-Saharan Africa. In addition, there is a lack of data to validate modeled soil C dynamics at broad scales. Here, we present insights into organic carbon cycling, based on a new broad-scale radiocarbon and mineral dataset for sub-Saharan Africa. We found that in moderately weathered soils in seasonal climate zones with poorly crystalline and reactive clay minerals, organic carbon persists longer on average (topsoil: 201 ± 130 years; subsoil: 645 ± 385 years) than in highly weathered soils in humid regions (topsoil: 140 ± 46 years; subsoil: 454 ± 247 years) with less reactive minerals. Soils in arid climate zones (topsoil: 396 ± 339 years; subsoil: 963 ± 669 years) store organic carbon for periods more similar to those in seasonal climate zones, likely reflecting climatic constraints on weathering, carbon inputs and microbial decomposition. These insights into the timescales of organic carbon persistence in soils of sub-Saharan Africa suggest that a process-oriented grouping of soils based on pedo-climatic conditions may be useful to improve predictions of soil responses to climate change at broader scales.

MASSAI: Multi-agent system for simulating sustainable agricultural intensification of smallholder farms in Africa.

The research and development needed to achieve sustainability of African smallholder agricultural and natural systems has led to a wide array of theoretical frameworks for conceptualising socioecological processes and functions. However, there are few analytical tools for spatio-temporal empirical approaches to implement use cases, which is a prerequisite to understand the performance of smallholder farms in the real world. This study builds a multi-agent system (MAS) to operationalise the Sustainable Agricultural Intensification (SAI) theoretical framework (MASSAI). This is an essential tool for spatio-temporal simulation of farm productivity to evaluate sustainability trends into the future at fine scale of a managed plot. MASSAI evaluates dynamic nutrient transfer using smallholder nutrient monitoring functions which have been calibrated with parameters from Malawi and the region. It integrates two modules: the Environmental (EM) and Behavioural (BM) ones.•The EM assess dynamic natural nutrient inputs (sedimentation and atmospheric deposition) and outputs (leaching, erosion and gaseous loses) as a product of bioclimatic factors and land use activities.•An integrated BM assess the impact of farmer decisions which influence farm-level inputs (fertilizer, manure, biological N fixation) and outputs (crop yields and associated grain).•A use case of input subsidies, common in Africa, markedly influence fertilizer access and the impact of different policy scenarios on decision-making, crop productivity, and nutrient balance are simulated. This is of use for empirical analysis smallholder's sustainability trajectories given the pro-poor development policy support.

Overcoming constraints of scaling: Critical and empirical perspectives on agricultural innovation scaling.

Scaling is a ubiquitous concept in agricultural research in the global south as donors require their research grantees to prove that their results can be scaled to impact upon the livelihoods of a large number of beneficiaries. Recent studies on scaling have brought critical perspectives to the rather technocratic tendencies in the agricultural innovations scaling literature. Drawing on theoretical debates on spatial strategies and practical experience of agricultural innovation scaling in Ethiopia, this paper adds to the current debate on what constitutes scaling and how to overcome critical scaling constraints. The data for the paper came from a qualitative assessment using focus group discussions, key informant interviews, and document analysis on scaling work done in Ethiopia by a USAID-funded research for development project. The paper concludes with four broad lessons for the current understating of agricultural innovation scaling. First, scaling of agricultural innovations requires a balanced focus on technical requirements and associated social dynamics surrounding scaling targets, actors involved and their social relations. Second, appreciating the social dynamics of scaling emphasizes the fact that scaling is more complex than a linear rolling out of innovations towards diffusion. Third, scaling may not be strictly planned; instead, it might be an extension of the innovation generation process that relies heavily on both new and long-term relationships with key partners, trust, and continuous reflection and learning. Fourth, the overall implication of the above three conclusions is that scaling strategies need to be flexible, stepwise, and reflective. Despite the promises of flourishing scaling frameworks, scaling strategies it would appear from the Africa RISING experience that, if real impact is to be achieved, approaches will be required to be flexible enough to manage the social, processual and emergent nature of the practice of scaling.

Estimating spatially distributed SOC sequestration potentials of sustainable land management practices in Ethiopia.

The sustainable land management program (SLMP) of Ethiopia aims to improve livelihoods and create resilient communities and landscape to climate change. Soil organic carbon (SOC) sequestration is one of the key co-benefits of the SLMP. The objective of this study was to estimate the spatial dynamics of SOC in 2010 and 2018 (before and after SLMP) and identify the SOC sequestration hotspots at landscape scale in four selected SLMP watersheds in the Ethiopian highlands. The specific objectives were to: 1) comparatively evaluate SOC sequestration estimation model building strategies using either a single watershed, a combined dataset from all watersheds, and leave-one-watershed-out using Random Forest (RF) model; 2) map SOC stock of 2010 and 2018 to estimate amount of SOC sequestration and potential; 3) evaluate the impacts of SLM practices on SOC in four SLMP watersheds. A total of 397 auger composite samples from the topsoil (0-20 cm depth) were collected in 2010, and the same number of samples were collected from the same locations in 2018. We used simple statistics to assess the SOC change between the two periods, and machine learning models to predict SOC stock spatially. The study showed that statistically significant variation (P < 0.05) of SOC was observed between the two years in two watersheds (Gafera and Adi Tsegora) whereas the differences were not significant in the other two watersheds (Yesir and Azugashuba). Comparative analysis of model-setups shows that a combined dataset from all the four watersheds to train and test RF outperform the other two strategies (a single watershed alone and a leave-one-watershed-out to train and test RF) during the testing dataset. Thus, this approach was used to predict SOC stock before (2010) and after (2018) land management interventions and to derive the SOC sequestration maps. We estimated the sequestrated, achievable and target level of SOC stock spatially in the four watersheds. We assessed the impact of SLM practices, specifically bunds, terraces, biological and various forms of tillage practices on SOC using partial dependency algorithms of prediction models. No tillage (NT) increased SOC in all watersheds. The combination of physical and biological interventions ("bunds + vegetations" or "terraces + vegetations") resulted in the highest SOC stock, followed by the biological intervention. The achievable SOC stock analysis showed that further SOC stock sequestration of up to 13.7 Mg C ha--1 may be possible in the Adi Tsegora, 15.8 Mg C ha-1 in Gafera, 33.2 Mg C ha-1 in Azuga suba and 34.7 Mg C ha-1 in Yesir watersheds.

Modelling soil erosion response to sustainable landscape management scenarios in the Mo River Basin (Togo, West Africa).

The rural landscapes in Central Togo are experiencing severe land degradation, including soil erosion. However, spatially distributed information has scarcely been produced to identify the effects of landscape pattern dynamics on ecosystem services, especially the soil erosion control. In addition, relevant information for sustainable land and soil conservation is still lacking at watershed level. On this basis, using the LAndscape Management and Planning Tool for the Mo River basin (LAMPT_Mo), we (1) modelled soil erosion patterns in relation with land use/cover change (LUCC), land protection regime, and landforms, and (2) examined the efficiency of landscape redesign options on soil erosion amounts at basin scale. We found that Simulated historical net soil loss (NSL) for the Mo basin were approximately 26, 23, 27, and 44t/ha/yr, for 1972, 1987, 2000, and 2014, respectively. These simulated NSLs were higher than the tolerable soil loss limits for the Tropics. Steep slopes (≥15°), poorly covered lands (croplands and savannas), and riversides (distances ≤100m) are critical areas of sediment sources. The local appraisal of soil loss was in line with the simulated outputs even though quantification was not accounted for when dealing with rural illiterate people. Furthermore, results showed that the examined management measures, such as controlling the identified erosion hotspots through land protective measures, could help reduce the NSL up to 70%, to values closer to the tolerable limits for the Tropics. The model implementation in the basin showed insights for identifying erosion hotspots and targeting soil conservation planning and landscape restoration measures.

Mapping Soil Properties of Africa at 250 m Resolution: Random Forests Significantly Improve Current Predictions.

80% of arable land in Africa has low soil fertility and suffers from physical soil problems. Additionally, significant amounts of nutrients are lost every year due to unsustainable soil management practices. This is partially the result of insufficient use of soil management knowledge. To help bridge the soil information gap in Africa, the Africa Soil Information Service (AfSIS) project was established in 2008. Over the period 2008-2014, the AfSIS project compiled two point data sets: the Africa Soil Profiles (legacy) database and the AfSIS Sentinel Site database. These data sets contain over 28 thousand sampling locations and represent the most comprehensive soil sample data sets of the African continent to date. Utilizing these point data sets in combination with a large number of covariates, we have generated a series of spatial predictions of soil properties relevant to the agricultural management--organic carbon, pH, sand, silt and clay fractions, bulk density, cation-exchange capacity, total nitrogen, exchangeable acidity, Al content and exchangeable bases (Ca, K, Mg, Na). We specifically investigate differences between two predictive approaches: random forests and linear regression. Results of 5-fold cross-validation demonstrate that the random forests algorithm consistently outperforms the linear regression algorithm, with average decreases of 15-75% in Root Mean Squared Error (RMSE) across soil properties and depths. Fitting and running random forests models takes an order of magnitude more time and the modelling success is sensitive to artifacts in the input data, but as long as quality-controlled point data are provided, an increase in soil mapping accuracy can be expected. Results also indicate that globally predicted soil classes (USDA Soil Taxonomy, especially Alfisols and Mollisols) help improve continental scale soil property mapping, and are among the most important predictors. This indicates a promising potential for transferring pedological knowledge from data rich countries to countries with limited soil data.