Environmental change over the last millennium recorded in two contrasting crater lakes in western Uganda, eastern Africa (Lakes Kasenda and Wandakara)

The last millennium is a key period for understanding environmental change in eastern Africa, as there is clear evidence of marked fluctuations in climate (effective moisture) that place modern concern with future climate change in a proper context, both in terms of environmental and societal impacts and responses. Here, we compare sediment records from two small, nearby, closed crater lakes in western Uganda (Lake Kasenda and Lake Wandakara), spanning the last 700 (Wandakara) and 1200 years (Kasenda) respectively. Multiproxy analyses of chemical sedimentary parameters (including C/N ratios, δ13C of bulk organic matter and δ13C and δ18O of authigenic carbonates) and biotic remains (diatoms, aquatic macrofossils, chironomids) suggest that Kasenda has been sensitive to climate over much of this period, and has shown substantial fluctuations in conductivity, while Wandakara has a more muted response, likely due to the increasing dominance of human activity as a driver of change within the lake and catchment over the length of our record. Evidence from both records, however, supports the idea that lake levels were low from ∼AD 700–1000 AD, with increasing aridity from AD 1100–1600, and brief wet phases around AD 1000 and 1400. Wetter conditions are recorded in the 1700s, but drought returned by the end of the century and into the early 1800s, becoming wetter again from the mid-1800s. Comparison with other records across eastern Africa suggests that while some events are widespread (e.g. aridity beginning ∼ AD 1100), at other times there is a more complex spatial signature (e.g. in the 1200s to 1300s, and from the 1400s to 1600s). This study highlights the important role of catchment-specific factors (e.g. lakemorphometry, catchment size, and human impact) in modulating the sensitivity of proxies, and lake records, as indicators of environmental change, and potential hazards when regional inference is based on a single site or proxy.