Gain and loss: Human and environmental wellbeing – drivers of Kilimanjaro’s decreasing biodiversity

PLOS One | https://doi.org/10.1371/journal.pone.0334184 October 29, 2025 5 / 23 Climate data For the temperature analysis of the past decades, we used data from the NOAA National Climatic Data Centre (NCDC) (https://www.ncdc.noaa.gov/). The temperature data are much more limited than the precipitation data, as only a few weather stations have collected long-term temperature data. A station with useful data was Kilimanjaro Airport in the southwestern corner of the study area at 850 m a.s.l. The average daily temperature was determined from 4 to 24 mea- surements/day. For precipitation data, we used the published data from Otte et al. [36]. We used a generalised least squares (GLS) regression model to test the statistical significance of the long-term trends. Data of population and economic growth The scale and time frame of our study were determined by the available census and remote sensing data: The smallest spatial unit for which census data were available was the ward level, an administrative structure for several villages, one Fig 2. The nine land cover/land use (LULC) classes used for the study area. The first two lines represent anthropogenic LULC classes. GRA: sub- montane grassland (class 4); AFO: agroforestry (class 2), a: commercial coffee plantation, b: traditional agroforestry system of a Chagga homegarden; FPL: forest plantation (class 7); BTU: built-up area (class 8); AGR: agriculture (class 1), a: rice field, b: wheat field, c: maize field. The last two lines rep-

resent natural LULC classes. FOR: forests (class 3), a: groundwater forest, b: Ficus riverine gallery forest, c: submontane Newtonia ravine forest; REG: forest regeneration (class 5); WAT: waterbody of Lake Chala (class 9); SAV a and b: savanna grasslands and woodlands, c: palm shrubland on alkaline soil with high groundwater table (class 6). https://doi.org/10.1371/journal.pone.0334184.g002