Ten sites covering six different ecosystems were selected for detailed situ N2O and CH4 exchange measurements. The six different ecosystems represent main (semi) natural forest types along an elevational and thus climatic gradient (savanna, lower montane, Ocotea, and Podocarpus forest) as well as the main land-use gradient between 1,300–1,800 m a.s.l. (lower montane forest, homegarden, and coffee plantation). The initial sampling started in February 2011 on one transect (Machame) and included savanna, coffee plantation, homegarden, lower montane, Ocotea, and Podocarpus forest and lasted until July 2013. From September 2013 until end of 2015, we focused on additional forest sites and conducted more measurements at a lower montane, Ocotea and Podocarpus forest located at another transect (Mweka), Each site was equipped with at least 5 spatially replicated chambers (25.2 x 15.2 x 14.7 cm. Gas sampling was performed with 60 ml gas tight syringes equipped with a one-way LuerLock stop-cock (VWR International, Darmstadt, Germany). Each chamber was sampled five times during chamber closure(t0 = 0, t1 = 15, t2 = 30, t3 = 45, and t4 = 60 min). A total of 45 ml of headspace sample was taken and used to flush and finally over-pressurize(5 ml) 10 ml glass vials (SRI Instruments, Bad Honnef, Germany). The samples were shipped within 4 weeks to KIT, IMK-IFU (Garmisch- Partenkirchen, Germany) for further analysis using an auto sampler (HT200H; HTA s.r.l, Brescia, Italy) coupled to a gas chromatograph (8610 C; SRI Instruments, Torrence, USA) equipped with an electron capture detector (ECD N2O) and a flame ionization detector/methanizer (FID: CH4 and CO2). Samples were continuously calibrated with standard gas samples (N2O: 406 ppb; CH4: 2,110 ppb; CO2: 407.9 ppm, Air Liquide, Düsseldorf, Germany). Flux rates were calculated with R version 3.2.0 including HMR package 0.3.1 for calculating the GHG flux rates by linear increase or decrease in gas concentration over time (N = 5).
