Sensitivity of a tropical montane cloud forest to climate change, present, past and future: Mt. Marsabit, N. Kenya

Abstract During the Last Glacial Maximum (LGM) lowland forests contracted throughout the tropics but, by contrast, many montane forest taxa moved to lower elevations. These taxa are often found in cloud forests, which are globally important ecosystems that depend on the capture of atmospheric water from fog drifting through the canopy, here referred to as occult precipitation. Understanding the response of tropical montane taxa to climate variations is limited by a lack of modern data on fog capture; whereas palaeoecological data only provide indirect evidence for its importance. Hence, the response of vegetation to fog capture is not considered in palaeo-estimates of precipitation. We develop a method that uses satellite Normalized Difference Vegetation Index (NDVI) data to estimate the annual amount of occult precipitation and investigate the sensitivity of a cloud forest to past and future changes in both rainfall and occult precipitation. We apply this method using satellite and meteorological data from 1982 to 2015 collected at Mt Marsabit, which is located in northern Kenya (2.34 ∘ N, 37.97 ∘ E, summit 1707 m a.s.l.). Mt Marsabit has a sub-humid tropical montane cloud forest at its summit that is excessively green for the amount of rain it receives. We estimate the annual amount of occult precipitation for current conditions at about 900 mm y −1 which is more than the average annual rainfall of 700 mm y −1 . This is consistent with the observation that, for the wider Marsabit area, interannual variations in NDVI are more closely linked to changes in cloud-base height ( r 2 = 0.87 ) than to changes in rainfall ( r 2 = 0.67 ). We investigate the sensitivity of forest extent to past and future changes; for the LGM we estimate that cloud-base height decreased by 500 m in response to a 4 ∘ C cooling and that this caused a 20%–100% increase in forest area despite a 30% decrease in rainfall, a 22% decrease in atmospheric humidity and a substantial reduction of atmospheric CO 2 levels (values representative for mountains in Kenya during the LGM). An expected increase of 250 m in the cloud-base height associated with a future 2 ∘ C global warming is likely to reduce forest extent by 50%–100%. Our results indicate that the satellite vegetation record is useful to estimate modern hydrological inputs into drier cloud forests (up to 2000 mm y −1 ) and that this information can be used to estimate the contribution of occult precipitation to altitudinal displacements of tropical montane cloud-forest species during the Quaternary.