Diatoms as paleolimnological indicators: a reconstruction of Late Quaternary environments in two East African salt lakes

Lakes Magadi (Kenya) and Manyara (Tanzania) occupy closed basins in the southern Gregory Rift valley. Water in these lakes is presently shallow and saline, testifying to the dominance of evaporation (E) over precipitation (P). Past changes in the P: E ratio, and hence in palaeoclimate, can be reconstructed from evidence of the former extent of these lakes. Lake-level fluctuations engender marked variation in water chemistry, and consequently on the composition of the limnological biota. One approach is to examine the sedimentary record of diatoms (unicellular algae), which are excellent indicators of water chemistry and relative water depth, and whose modem distribution is sufficiently well known to allow the quantitative reconstruction of chemical parameters. Diatom analysis of 116 samples from a series of radiometrically dated (14C and U/Ib) sediment cores has revealed significant changes amongst the diatom assemblages during the Late Quaternary. Conductivity and pH have been estimated from the fossil samples by transfer functions (Gasse unpublished, Gasse 1986b). However, the interpretation of fossil diatom assemblages is often problematical in hypersaline environments. Difficulties arise as a result of the operation of taphonomic and diagenetic processes which can severely alter the composition of the diatom assemblagesfr om the ambient population at the time of deposition. Probably the most important factor responsible for assemblage diagenesis in saline lakes is silica dissolution, and this is explored further by a series of laboratory experiments. Results indicate that silica dissolution acts differentially between species, by removing the smaller, more delicate taxa first, and causing the relative enrichment of large robust forms in the fossil samples. A similar dissolution gradient may be reflected in modem samples studied near hot springs at Magadi. Differential dissolution is potentially an important source of error in palaeoenvironmental reconstructions, but, with the outcome of these experiments, it has been possible to assess the extent to which the dissolution process may have shaped the diatom records from Magadi and Manyara. The bulk of the palaeolimnological evidence is focussed upon two periods, 30,000-20,000 BP and 12,700-9,500 BP. The earlier period is most clearly dated in the core from Manyara, where the diatom record suggests the development of an intermediate level lake between c. 27,500 BP and c. 26,000 BP. This is a more complete representation of the same lake phase found in earlier studies from Manyara by Holdship (1976) based on diatoms, and by Casanova (1986a) on stromatolites 20M above the present lake. This time interval may also be represented by the central portion of the Magadi cores NF1 and NF2 but here dating is more problematical. The period 12,700-9,500 BP was one of major lacustrine transgression across Africa although the fine-structure of this event is less well known. Cores NF1 and NF2 from Magadi provide a detailed register of this phase indicating a major highstand from c. 12,700-11,000 BP when the lake became deep enough to stratify and deposit laminated couplets. At c. 11,000 BP the diatoms show that salinity increased greatly from fresh-oligosaline to meso-hypersaline which was probably a consequence of lake level falling.