LAUSR

Abstract This study presents a first attempt at using phytoliths from a high-altitude crater lake as a palaeoecological tool at a tropical montane site. Mt. Muhavura (4127 m asl) at the border of Uganda and Rwanda is part of the Virunga volcanoes located in the Albertine Rift, one of the most biodiverse areas in Africa. To investigate the potential of phytoliths to record grassland history in an Afroalpine environment, a unique and little studied ecosystem, we analysed the phytolith content of 37 sediment samples from a 159 cm sediment core from the crater lake at the summit of Mt. Muhavura representing the last ca. 2400 years BP. Ten plant species from the Alpine belt were also analysed for their main phytolith morphotypes. Phytoliths from the sediments were grouped into 12 main categories and the most frequently occurring individual morphotypes were bilobate (33%), oblong (16%), elongate (17%) and globular psilate (13%). The phytolith assemblage consists predominantly of phytoliths characteristic of grasses (70%). A literal interpretation of the results would be that Panicoideae grasses (mainly C 4 ) that are characteristic of warm and wet climate and C 3 Pooideae subfamily grasses characteristic of cold climate/high altitude have co-existed throughout the late Holocene in this belt. As only C 3 cold climate grass subfamilies are known to occur at this altitude today, the abundance of short cell phytoliths (bilobates) typical of Panicoideae grasses in the lake sediments at this altitude presents a new perspective to the phytolith discussion. The C 3 Pooideae grasses analysed did not produce bilobates. In this study, the phytolith index, Ic that is used as a proxy of past grass subfamily dominance is not representative of the taxonomic significance for which it is known. This study suggests that caution needs to be taken when using bilobate morphotypes and the Ic as indicators of grass subfamily dominance at high altitude on East African mountains as the bilobates may not be taxonomically representative of the grasses for which they are typically known. This highlights the inherent problem of phytolith redundancy even at an altitude where C 3 /C 4 grass distribution is better delineated.