Large-scale Altitudinal and Latitudinal Variability of Vegetation Phenology in a Tropical Montane Landscape: A Remote Sensing Perspective

Abstract:

Mountain ecosystems are particularly vulnerable to climatic change, given the reduced potential of species to disperse uphill. Studies have been concentrated at higher latitudes, and relatively little is known about montane vegetation and ecosystem dynamics in tropical mountains. We assessed climatic, latitudinal and altitudinal effects on leafing phenology for the Espinhaço Range, a South American tropical mountain landscape comprised of a mosaic of savannas, grasslands, rock outcrops, cloud forests, and semi-deciduous to deciduous forests. We assessed a time series of 884 MODIS/NDVI images acquired between 2002 and 2015, at 7-day intervals. We classified broad vegetation types based on elevation (SRTM) and the mean and variance of each pixel in the entire series. We then extracted the phenological indicators of start, end and length of the growing season, green-up and brown-down rates, NDVI peak, and integral of the growing curve, using the TIMESAT algorithm. We also obtained precipitation data from the TRMM dataset, and calculated the Topographic Wetness Index and clear-sky radiation budgets based on the SRTM dataset. Our results show that the start of the growing season was noticeably more variable than the end date, suggesting that season length is an important factor for tropical montane vegetation. The start of the growing season decreased linearly with altitude, with vegetation at higher elevations having a later start, while no clear relation was found for the end of season. For montane vegetation above 800m, we observed shorter season lengths. Green-up rates were higher in woody seasonal formations, and became progressively slower with increasing altitudes. Higher green-up rates were also associated with rainfall patterns, where dry seasons are longer. Brown-down rates had the opposite trend, with rapid brown-down at higher altitudes. Our results quantify previously unreported seasonal, latitudinal and altitudinal variations in vegetation phenology for a large tropical mountain range, and emphasize the high variability of phenological responses in the tropics. Understanding this variability is key to assess the impacts imposed by present and future climate changes.