Intensification of agricultural systems is a major threat to the associated biodiversity and could also affect the dynamics of pests and pathogens. One such system that is currently under an intensification trajectory is the production of Arabica coffee. In this thesis, I studied the relationships between fungal diseases and their natural enemies, the genetic variation in coffee, coffee yield and associated biodiversity along a coffee management gradient in southwestern Ethiopia.

The specific goals of this thesis were to investigate variation in fungal diseases on coffee and their natural enemies along a gradient of management (I, II), how genetic variation in coffee among sites relate to variation in incidence of the fungal diseases (III), and to investigate the trade-offs in biodiversity-yield relationships along the gradient of coffee management (IV). To answer these questions, I selected 60 sites along a gradient of management that ranged from coffee naturally growing in only little disturbed forests to intensively managed plantations. I used both observational studies and molecular approaches.

In paper I, I examined if the severity of the four major fungal diseases on coffee varied along the gradient and assessed the main drivers of variation in disease severity. I found that two of the fungal diseases were more severe in the intensively managed coffee sites, while the other two were more severe in the less intensively managed sites. Altitude was the main driver for the diseases, but related in a different way to the different diseases. In paper II, I examined the temporal dynamics in coffee leaf rust-hyperparasite interaction, the biocontrol potential of the hyperparasite and environmental drivers for the two species for three consecutive years during the dry and wet seasons. I found that the rust was more common during the dry season and in managed sites while the hyperparasite was common during the wet season and in sites that were less managed. My results also revealed that higher hyperparasite incidence during the wet season resulted in a lower growth rate of the rust during the subsequent dry season. In paper III, I investigated if genetic composition and diversity of coffee sites relate to the incidence of the fungal diseases assessed. I found that genetic composition of the coffee stands was linked to the incidence of the four fungal diseases, but genetic diversity among the coffee sites did not relate to the incidence of the diseases. In paper IV, I examined biodiversity-yield trade-offs and shape of the relationships between biodiversity and yield along the gradient of management. I found a steep, concave shape initial decline in biodiversity values as coffee yield increased to a certain level, after which a further increase in yield did not have much effect on biodiversity values.

In conclusion, I found different drivers for the different diseases and for the parasite-hyperparasite interaction. It is difficult to achieve a single management approach that can suit the different pathogen species investigated. High genetic diversity among coffee sites did not reduce disease pressure. While the more complex, less managed sites provide high biodiversity values, and could potentially serve as habitats for natural pest control and in situ conservation for coffee genetic diversity, the yield gap compared to more intensively managed sites was very high. To optimize coffee management and conservation of biodiversity in these landscapes, there is a need to develop strategies whereby the smallholder farmers who depend on coffee and the forest as the main source of livelihood can benefit through for example coffee certification schemes that can pay premium prices for biodiversity-friendly coffee management.