Tracking what matters: Monitoring Forest Landscape Restoration in Malawi

Abstract

The thesis examined how Forest Landscape Restoration (FLR) is monitored, governed and implemented in Malawi's mosaic socio-ecological landscapes, and what this reveals about the design of credible, contextually grounded monitoring systems. It responds to the growing reliance on forest restoration targets expressed in hectares and trees, whilst far less is known about the accuracy of the underlying data, the spatial correspondence between mapped priorities and implemented interventions, and how diverse FLR implementing actors define and measure success.
First, three global ten-metre land cover products were evaluated against Malawi's National Forest Inventory. These were Esri Land Cover, ESA WorldCover and Dynamic World. The products achieve only moderate accuracy in miombo-dominated mosaics, and misclassification is directional. Trees, Crops and Built Area are systematically absorbed into Rangeland with little flow in the opposite direction. The same bias has been reported in smallholder mosaics elsewhere in sub-Saharan Africa and beyond, indicating that the limit is structural rather than specific to Malawi. Global products are insufficient for national reporting on their own, and hybrid approaches that integrate them with national inventories and ground-level observations are required.
Second, the thesis analysed where restoration occurs in Mzimba District. Using spatial data on 191 interventions delivered by 83 organisations, it quantified alignment with national priority surfaces, modelled spatial clustering, and profiled actor portfolios. Only five of 191 interventions lay within a 30 m buffer of a mapped priority pixel. FLR interventions clustered along roads and settlement corridors rather than in ecologically strategic areas. NGOs delivered most interventions through short, small- to medium-sized projects, while government agencies and international NGOs implemented fewer but larger and longer projects. Restoration geography is shaped more by accessibility and institutional presence than by ecological targeting.
Third, semi-structured interviews with governmental, non-governmental, civil society, small-enterprise and international actors investigated how restoration success is conceptualised and monitored. Actors operate with distinct theories of change and indicator sets. National frameworks privilege biophysical metrics such as area and tree counts, whereas local organisations routinely track livelihood, governance and gender outcomes that rarely feature in national dashboards. Monitoring choices are strongly mediated by institutional position, mandates and accountability relationships. What each actor implements and what each actor monitors are not independent decisions, and they flow from the same institutional configuration.
The findings demonstrate that FLR monitoring in Malawi is constrained by the technical limitations of global datasets, institutional complexity across governance scales and actor types, and epistemic divergence over what counts as evidence. These constraints reinforce one another rather than operating in isolation. The thesis proposes a hybrid monitoring architecture that integrates global remote sensing, national and district datasets, actor-generated reporting, and Indigenous and Local Knowledge through a multi-actor platform and an independent validation function, with a plural indicator framework spanning ecological, social and governance metrics. The architecture maps onto governance, spatial and temporal scales, supports cadences from quarterly community observation to the 2030 UNCCD Land Degradation Neutrality target, and tracks both restored area and ongoing degradation so that net change, rather than gross restoration, is the reported metric. The institutional costs of integrated reporting across AFR100, the Global Biodiversity Framework, UNFCCC and UNCCD LDN are real and need to be explicitly resourced. The architecture is grounded in northern Malawi, but its design principles apply to other mosaic landscapes in sub-Saharan Africa with comparable smallholder dynamics and multi-actor governance.

Date of Award	20 Jul 2026
Original language	English
Awarding Institution	Bangor University
Supervisor	Eefke Mollee (Supervisor) & James Gibbons (Supervisor)