Stabilizing smallholder crop yields under changing climatic conditions in Africa will require adequate adaptation strategies focused on soil and water management. In some regions, rainwater harvesting (RWH) is used already to decrease the susceptibility of crops to frequent dry spells. Findings from this thesis show that Africa is likely to see significant changes in rainfall patterns during crop growing seasons, including higher intensity rainfall and more frequent very long dry spells. It is shown that RWH is a valuable adaptation strategy to climate change in Africa for maize, millet, and sorghum for a number of reasons. RWH could bridge ~30% of the yield gaps attributable to water deficits in the 2050s, thereby reducing future irrigation requirements. However, yield increases from improved water availability remain marginal (e.g. ~5-6% for millet and sorghum), unless combined with improved fertility measures (doubling of yields possible). Key benefits, potentially of greater importance than increased water availability from RWH, include protecting seeds, concentrating nutrients, and reducing long-term soil degradation. While RWH strategies show great biophysical potential as adaptation strategies, there remain a number of locally specific barriers to their adoption which need to be addressed to ensure their successful implementation at larger scales. As humans normally respond to perceived risks brought on by certain situations, it was hypothesized that climate change perceptions may be key in pro-moting the adoption of adaptation strategies such as RWH at the field level. In Burkina Faso, farmers had skewed perceptions of climate change (e.g. perceived decrease in precipitation when there are observed and projected increases), and thought of RWH as a central adaptation strategy despite not addressing projected impacts directly. Widespread RWH adoption across three field sites (Burkina Faso, Ethiopia, and Tunisia) rather depended heavily on government and NGO intervention. Overall, RWH could be an integral part of “adaptation packages” aimed at smallholder farmers, but should not be promoted as an independent solution to climate change in rainfed Africa.

The relative influence of precessionally-driven changes in direct insolation and changes in boundary conditions associated with glacial-interglacial cycles on climatic conditions at low latitudes remains uncertain. This thesis presents records of late Quaternary climatic change from Tswaing crater lake in South Africa and from the Naivasha and Nakuru-Elmenteita basins in the Central Kenya Rift, with the aim of increasing our understanding of the nature and causes of climatic change at low latitudes in Africa. The sedimentary sequence from Tswaing crater lake provides some of the longest terrestrial records of palaeoclimatic change in southern Africa, but the confidence associated with these records is limited by chronological uncertainty. This thesis presents five new 230Th/234U dates from the lower, previously undated section of the sequence, which are used to construct a new age-depth model for the sediments. When viewed in light of this chronology, new sedimentological, geochemical, and diatom assemblage records from the sequence indicate that boundary conditions associated with glacial-interglacial cycles determined climatic conditions at the site over the last -150 kyr, and that the obliquity of the earth's axis may have affected conditions between -150 and -350 kyr B.P. Diatomite beds deposited in the Naivasha and Nakuru-Elmenteita basins at the time of the last interglacial document a period in which deep, dilute lakes existed in the Central Kenya Rift. 518Odiatom records from these 40Ar/39Ar-dated beds are used to reconstruct palaeohydrological conditions during this lake-level highstand. The records indicate that lake levels in both basins responded to increases in precipitation driven by peaks in March and September insolation on the equator, and by increased tropical sea-surface temperatures. It is therefore concluded that precipitation in the Central Kenya Rift was influenced by both precessionally-driven changes in insolation and global boundary conditions during this period.

The Agulhas Current transport of heat and salt from the Indian Ocean into the South Atlantic around South Africa (Agulhas leakage), can affect the Atlantic Meridional Overturning Circulation (AMOC) and, thus, influence global climate. Upper water column reconstructions in the southwest Indian Ocean over the past 100 kyr based on marine sediments from the core region of the Agulhas Current suggest that surface ocean temperature, salinity and planktonic foraminiferal assemblage records from the Agulhas Current exhibit high variability on orbital to millennial timescales. A high degree of similarity in this variability could also be identified in the Agulhas leakage records in the South Atlantic which suggests that changes in the Agulhas leakage can be partly explained by upstream variability in the Current itself. The results of a benthic stable isotope record from the southwest Indian Ocean over the past 270 kyr gives evidence that during glacial periods as well as during Northern Hemisphere Cold Stadials Southern Component Waters substituted for North Atlantic Deep Waters. The recorded hydrographic variability in the deep southwest Indian Ocean is explained in terms of a less vigorous AMOC exporting a reduced amount of NADW into the Southern Hemisphere and/or at shallower depth causing the observed changes in the deep water inventory. A multiproxy data and model integration approach reveals that phases of more humid southeast Africa climate were driven by southward oscillations of the Intertropical Convergence Zone and its associated rain belt over the past two glacial-interglacial cycles. Low-latitude summer insolation changes paced by orbital precession explain the long-term climate variability whereas abrupt climate oscillations in the northern high latitudes are the main driver for the observed millennial-scale wet phases. Southeast African climate variability seems to have been coupled, and anti-phased, with the East Asian Summer Monsoon during the late Pleistocene. Agulhas Current sea surface temperatures changes did not exert a primary control on southeast African hydrology.

A 2°C increase in global mean temperature (ΔTg) has been widely adopted as a benchmark for dangerous climate change. However, there has been a lack of research into the implications of 2°C, or any other degree of warming, for Africa. In this thesis changes in African temperature and precipitation associated with 1°C, 2°C, 3°C, 4°C, and beyond are investigated for the first time, using output from 350 climate model experiments: a collection of simulations from international modelling centres (CMIP3), two Perturbed Physics Ensembles (PPEs), and a group of five regional models. The models project temperature and precipitation anomalies which increase in magnitude and spatial extent as global temperature rises, including a wet signal in East Africa, and drier conditions for African rainforests. The models consistently show that the evolution of change with global warming is gradual, even at 4°C and beyond; but the amplitude and direction of precipitation change at each ΔTg increment vary between models and between datasets. The PPEs project precipitation signals which are not represented by CMIP3, in particular a large drying (>0.5 mm day-1 °C-1) of western Africa. There are also important differences between global and regional models, especially in southern and West Africa (>1 mm day-1). Analysis of atmospheric circulation responses suggests that the higher resolution projections are no more credible in this case. Some of the variation between models can be understood as the result of untrustworthy simulations, leading to constraints on the PPEs, and casting doubt on the strong drying of west Sahel; but model evaluation is found to be limited by observations in the case of the Congo Basin. The implications of global warming are different depending on which models are consulted. The findings emphasise that caution should be exercised in the application of climate model data to inform mitigation debates.

Climate north of the Atlas Mountain belt in NW Africa is dominated by extratropical disturbances. However, climatic controls to the south, where climate transitions from extratropical to tropical regimes, are poorly understood due to a paucity of both instrumental and palaeoclimate data. In this thesis past climate change between the High Atlas Mountains and Sahara Desert is reconstructed using the stable isotopic composition and radiometric dating of speleothems. A high-resolution record from the mid-Holocene and a discontinuous record covering the past 400,000 years are developed. Supplemented by U-Th dating of a further four samples, these records indicate increased humidity in this area concomitant with the wider African Humid Period, and indicate a link between the West African Monsoon and humidity north of 30°N. Reconstructed glacial-interglacial scale increases in humidity overlap with "green Sahara" conditions and evidence a recurrent humid corridor connecting NW Africa and the central Sahara that is highly relevant to discussions of prehistoric human migrations. Evidence for a strong influence of high-latitude and solar forcing on decadal to millennial time- scales in this area is also presented. Further to this work, the potential of cadmium-to-calcite ratios as a novel proxy for palaeo-hydrology is confirmed using an annually-resolved trace element, stable isotope and calcite fabric dataset from a North Moroccan stalagmite. The first measurements of cadmium-to-calcite ratios in natural speleothem are here presented, and the palaeoclimatic significance and potential of this proxy for aiding the quantitative reconstruction of changes in calcite precipitation behaviour are demonstrated.