In the central and northern highlands of Ethiopia, native forest and forest biodiversity is almost confined to sacred groves associated with churches. Local communities rely on these 'church forests' for essential ecosystem services including shade and fresh water but little is known about their region-wide distribution and conservation value. We (1) performed the first large-scale spatially-explicit assessment of church forests, combining remote-sensing and field data, to assess the number of forests, their size, shape, isolation and woody plant species composition, (2) determined their plant communities and related these to environmental variables and potential natural vegetation, (3) identified the main challenges to biodiversity conservation in view of plant population dynamics and anthropogenic disturbances, and (4) present guidelines for management and policy. The 394 forests identified in satellite images were on average ~2ha in size and generally separated by ~2km from the nearest neighboring forest. Shape complexity, not size, decreased from the northern to the central highlands. Overall, 148 indigenous tree, shrub and liana species were recorded across the 78 surveyed forests. Patch α-diversity increased with mean annual precipitation, but typically only 25 woody species occurred per patch. The combined results showed that >50% of tree species present in tropical northeast Africa were still present in the 78 studied church forests, even though individual forests were small and relatively species-poor. Tree species composition of church forests varied with elevation and precipitation, and resembled the potential natural vegetation. With a wide distribution over the landscape, these church forests have high conservation value. However, long-term conservation of biodiversity of individual patches and evolutionary potential of species may be threatened by isolation, small sizes of tree species populations and disturbance, especially when considering climate change. Forest management interventions are essential and should be supported by environmental education and other forms of public engagement.

Understanding biodiversity of plants depends on detailed knowledge of floristics and environmental parameters. According to the Flora of Ethiopia and Eritrea (FEE; 1989–2009), the latest academic flora of these two topographically complex countries, the number of taxa (including non-indigenous ones) is 6027 of which 647 are endemic. Since 2009, 56 indigenous taxa have been described from the FEE area. Based on FEE data and later information, the number of indigenous species is here counted at 5219. In the RAINBIO database Sosef et al. (BMC Biol 15:1–23, 2017) counted the number of indigenous species in Ethiopia at 4481 (Eritrea not counted). Studies of local diversity and endemism produce different results with varying methods and sources: FEE data show highest diversity in the southern part of Ethiopia near the borders with Kenya, while RAINBIO data show highest diversity in the south-west and around the town of Harar. Wang et al. (J Syst Evol 58(1):33–42, 2020) found the highest ‘weighted endemism’ in the central Ethiopian mountains, while our FEE data indicates highest single-region endemism near the borders with Kenya and Somalia; Hawthorne and Marshall (Gard Bull Singap 71(Suppl. 2):315–333, 2019), weighting species according to rarity, found the highest ‘bioquality score’ in the same areas as our highest single-region endemism. Studies of altitudinal diversity show a ‘mid-elevation diversity bulge’ at 1400–2100 m a.s.l. More data are needed before we understand the complex diversity of the two countries, including altitudinal diversity. We review needs for additional data gathering and modelling that may help answering outstanding questions.Graphic abstract