Summary 1 The long-term impact of changes in land use, fire and climate on species diversity in Australia are only just beginning to be understood. We combined fine-resolution palaeoecological proxies for terrestrial and aquatic ecosystems to investigate the responses of a tropical rain forest catchment over the last 700 years. 2 Sediment cores were sampled at 1-cm intervals to a depth of 100 cm from Lake Euramoo on the Atherton Tableland of north-east Queensland, dated and analysed for pollen, charcoal, diatom, chironomid and inorganic content. 3 The pollen and charcoal record shows a rapid loss of forest diversity (particularly the long-lived taxa Agathis and Podocarpus) and increased burning coinciding with the arrival of European settlers. The aquatic environment is also subject to rapid changes at this time, with a possible increase in pH and subsequent shifts in local algal and insect communities. This event was outside the historic range of variability in both rain forest and aquatic communities. 4 The present mosaic of vegetation types is a complex function of environmental changes operating across a range of spatial and temporal scales: millennial climate change, short-term climatic variations associated with El Nino events and, most significantly, a shift from indigenous to ‘European’ land-use practices, including clearance and burning activities associated with the timber and farming industry between about ad 1880 and 1920. 5 The establishment of a World Heritage reserve around the lake catchment and the suppression of fire over the last 50 years have not yet restored the terrestrial or aquatic ecosystem to its pre-European state and are unlikely to, given the current predictions of future climate change. This supports the notion that ecological and climate thresholds are not necessarily the same, and that the effects of crossing them are not necessarily reversible. Retrospective studies of the historic range of variability within small catchments can provide an understanding of the limits of natural and human-induced variability that can inform management decisions and resource planning.