At ten different sites in the Kilimanjaro region we installed two identical suction cups (SK 20, UMS, Munich, Germany) at 10, 40, and 80 cm soil depths (6 in all) and set them to a vacuum of -400 hPa to allow continuous soil water sampling into brown Duran glass bottles (SF 1,000 ml, UMS, Munich, Germany) stored 50 cm belowground in a covered bucket. Soil water sampling and in situ GHG exchange measurements were conducted simultaneously. The water was decanted into 1L polyethylene bottles (VWR International GmbH, Darmstadt, Germany) and transported to the Nkweseko research station. There, two subsamples were filtered with 0.2 lm syringe filters (Minisart, Sartorius, Goettingen, Germany) and subsequently frozen and stored for a maximum of 6 months. After transportation to KIT IMK-IFU, Germany, total ammonium (NH4+) and nitrate (NO3-) concentrations were measured by colorimetry using an Epoch Microplate Spectrophotometer (BioTek, VT, USA) according to (Dannenmann et al., 2009). Subsamples were analyzed for total extractable nitrogen (DN) and total extractable organic carbon (DOC) using a multi-N/C 3100 combined chemo-luminescence infrared analyzer (Jena Analytik AG, Germany), as described by Bim€uller et al. (2014). Extractable organic nitrogen (DON) was calculated by subtracting mineral N (NH4+ and NO3-) from DN. Annual leaching rates were calculated from mean annual N concentrations multiplied by seepage water formation as calculated from mean annual precipitation - mean annual ETa (estimated from MODIS).

Soil water is sampled to quantify nutrient losses through leaching.