2.3. Experimental design and data collection. A complete randomized design with a (4 2 2) factorial treatment structure where factors included four levels of types of trees present on farm; two levels of coffee type and two soil depths was applied in the current study. Lists of coffee producers within each coffee region were obtained from Uganda Coffee Development
Authority (UCDA) from which farmers where the study was conducted were selected. Based on farmers with the similar tree coffee combinations, 40 coffee producers with coffee farms
(0.5 ha) were randomly selected from each of two regions that grow Arabic and Robusta coffee respectively, for soil sampling. For any randomly selected farm, location of the farm was documented using a hand GPS, and information on age and spacing of the coffee plants and owner of the farm were recorded and existing shades trees identified using a para-taxonomists and later confirmed at Makerere University Herbarium. The selected farms that had coffee
agroforestry systems consisted of either Robusta or Arabica coffee, intercropped with either fruit trees, such as Artocarpus heterophyllus Lam. Persea americana Mill. and M. indica L., or non-fruit trees, such as Maesopsis eminii Engl., F. natalensis Hochst. Markhamia lutea K. Schum., Grevillea robusta A. Cunn., Erythrina abyssinica Lam., Senna spectabilis (DC.) H.S. Irwin and Barneby, Albizia chinensis Merr., Albizia coriaria Welw. and Ficus ovata Vahl.
Farms without trees consisted of coffee as a monocrop.