Abstract: Appropriate use of agricultural technologies and diversifying the farming activities is critical to addressing food security problems in Africa, including Ethiopia. The country is experimenting with the new Agricultural Innovation System (AIS) approach alongside the well-established Transfer of Technology (ToT) approach. This paper analyzes the gaps between policy discourses (as reflected in policy documents and strategic orientation documents) and extension practices (as reflected in the daily exchanges between farmers and the frontline staff of the Ethiopian extension system). It provides insights into the challenges faced and emphasizes the need for better coordination between policy formulation and implementation to enhance extension services. Policymakers, practitioners, and researchers can benefit from the valuable perspectives the findings offer. The study contributes to understanding the relationship between policy discourses and extension practices, and its implications can inform policy design and implementation in similar contexts. A qualitative research approach was deployed to analyze policy discourse and practice. Data were collected in Fogera, a district in Northwest Ethiopia, between August 2018 and February 2019. The data for the paper were obtained from 23 Focus Group Discussions conducted with men and women. 13 Informant Interviews (KIIs) were also carried out with personnel at different levels of government agricultural services and departments. Transcripts of recordings of the Focus Group Discussions (FGDs) and Key Informant Interviews (KIIs) were analyzed using a deductive approach. The study focuses on rice crops in the Fogera district, which are crucial for food security and reducing poverty. Although the geographic area is limited, the results can be used to improve the extension system in other areas facing similar challenges. Specifically, the study suggests switching from the traditional transfer of technology approach to the agricultural innovation system approach. Furthermore, the study's techniques, such as qualitative interviews, may have limitations and not fully capture the intricacies of policy and extension practices. The findings demonstrate that, although the policy documents strongly adhere to agricultural innovation system principles, top-down transfer of technology approaches continues to dominate in practice. Moreover, we have found potential discrepancies between the training content delivered and the specific needs of smallholder farmers. Practically, prescriptive systems are still used because agricultural innovation system approaches are not well understood by the Extension Agents. To realize a genuine agricultural innovation system, Ethiopia's extension apparatus should move forward with building committed and robust relationships between farmers, extension agents, researchers, private sectors, and non-governmental organizations. To this end, more research, enhanced training, and improved institutions are needed on what genuine agricultural innovation system could look like at the grass-roots level. This also includes understanding the roles that different actors within Ethiopia's development army should assume how a multi-actor policy dialogue can be organized.

Abstract: Climate change is causing alterations in precipitation patterns, leading to adverse ecological consequences in many ecosystems. Recently, an increasingly persistent weather pattern has emerged, characterized by lengthening the duration of alternating dry and wet periods, which is more complex than exclusively drought or increasing precipitation. It is currently unclear how soil microbial communities respond to these new regimes in relation to their interactions with plants, especially in precipitation-sensitive ecosystems, such as tropical rainforests and grasslands. In this thesis, we explored responses of soil bacterial and fungal communities to increasing weather persistence in rainforests and grasslands, using high throughput sequencing technology. We firstly investigated the resistance and resilience of microbial communities to prolonged drought in a mature seasonal tropical rainforest which experiences unusually intensive dry seasons in the current century. Through excluding rainfall during and after the dry season, a simulated prolongation of the dry season by five months was compared to the control. Our results indicate that as rain exclusion progressed, the microbial communities increasingly diverged from the control, indicating a moderate resistance to prolonged drought. However, when the drought ceased, the composition and co-occurrence patterns of soil microbial communities immediately recovered to that in the control, implying a high resilience. To further investigate the ecological roles of soil microbial communities in response to increasing weather persistence, we set up grassland mesocosm experiments. In these experiments, precipitation frequency was adjusted along a series, ranging from 1 to 60 consecutive days alternating of dry and wet periods, while keeping the total precipitation constant. Our results show that microbial community assembly tended to be more stochastic processes at intermediate persistence of dry and wet alternations while more deterministic processes dominated at low and high persistence within 120 days regime exposure. Moreover, more persistent precipitation reduced the fungal diversity and network connectivity but barely impacted that of bacterial communities. The prior experiences of persistent weather events for one year caused legacy effects. The soil microbial legacy induced by soil microbial communities subjected to prior persistent weather events was more enduring in subsequent fungal communities than bacterial communities, likely due to slower growth of fungi compared to bacteria. However, a minor effect of soil microbial legacy  was observed on plant performance. In addition, we kept the grassland mesocosm experiment for two growing seasons. The effects of precipitation persistence on soil microbial communities increased in the second year. The dissimilarities of microbial communities between the first and second year were less with more persistent precipitation, potentially resulting in more vulnerable microbial communities, due to some taxa disappearing and a reduction in functional redundancy under more persistent weather. To conclude, our findings provide a comprehensive theoretical understanding of soil microbial communities in response to the current and future climate change, drawing from both natural and experimental systems. It helps in predicting and managing the impacts of future climate change on ecosystems mediated by microbial communities. Additionally, the findings of microbe-mediated legacy effects on grassland ecosystems can provide practical guidance for their application in agriculture, specifically for using an inoculum to mitigate the impacts of climate change.