Temperature variability over Africa during the last 2000 years

Abstract

A growing number of proxy, historical and instrumental data sets are now available from continental Africa through which past variations in temperature can be assessed. This paper, co-authored by members of the PAGES Africa2k Working Group, synthesises published material to produce a record of temperature variability for Africa as a whole spanning the last 2000 years. The paper focuses on temperature variability during the ‘Medieval Climate Anomaly’ (MCA), ‘Little Ice Age’ (LIA) and late 19th–early 21st centuries. Warmer conditions during the MCA are evident in records from Lake Tanganyika in central Africa, the Ethiopian Highlands in northeastern Africa, and Cango Cave, the Kuiseb River and Wonderkrater in southern Africa. Other records covering the MCA give ambiguous signals. Warming appears to have been greater during the early MCA (c. ad 1000) in parts of southern Africa and during the later MCA (from ad 1100) in Namibia, Ethiopia and at Lake Tanganyika. LIA cooling is evident in Ethiopian and southern African pollen records and in organic biomarker data from Lake Malawi in southeastern tropical Africa, while at Lake Tanganyika the temperature depression appears to have been less consistent. A warming trend in mean annual temperatures is clearly evident from historical and instrumental data covering the late 19th to early 21st centuries. General warming has occurred over Africa since the 1880s punctuated only by a period of cooling in the mid 20th century. The rate of temperature increase appears to have accelerated towards the end of the 20th century. The few long high-resolution proxy records that extend into the late 20th century indicate that average annual temperatures were 1–2°C higher in the last few decades than during the MCA.

Keywords

Africa ‘Little Ice Age’‘Medieval Climate Anomaly’temperature variability 20th century

Introduction

With the exception of the equatorial zone, arid and semi-arid conditions prevail over much of the African continent. Consequently, continuous or intermittent drought plagues many African countries and the majority of studies on past climatic variability and change have focused on long- and short-term fluctuations in rainfall. At the same time, temperature is assessed at far fewer contemporary meteorological stations than rainfall and the observational record of temperature is much shorter than that for rainfall. Palaeoclimatic records from Africa are similarly biased, with the vast majority reflecting primarily variations in available moisture (cf. Chase and Meadows, 2007; Gasse, 2000; Gasse et al., 2008). For these reasons, there have been relatively few investigations of past temperature change over Africa beyond analyses of instrumental records from the last few decades. Those studies that are available generally do not provide information for the continent as a whole.

This paper, co-authored by members of the PAGES Africa2k Working Group, synthesises published data on past temperature change in Africa to produce a record of temperature variability for the continent as a whole, spanning the last 2000 years. We focus, in particular, on evidence for temperature variability during the ‘Medieval Climate Anomaly’ (MCA), the ‘Little Ice Age’ (LIA) and the late 19th to early 21st centuries. For general reference, the MCA is defined here as spanning the period c. ad 950 to c. ad 1250 and the LIA from c. ad 1300 to c. ad 1850 (cf. Jones et al., 2001; Matthews and Briffa, 2005). We also aim to highlight some of the shortcomings within both the proxy and instrumental temperature records, and point to possible ways forward. We first consider palaeoenvironmental reconstructions covering all or part of the past two millennia with adequate (century-scale or better) resolution and time control. This is followed by a review of annually resolved historical and instrumental records for the last two centuries.

On multidecadal to centennial timescales, several climate proxy records from East African lakes (most notably Lake Tanganyika and Lake Malawi) provide relatively high-resolution palaeotemperature information. Additionally, a limited number of pollen records have been interpreted as containing temperature signals, and oxygen isotopic records (δ¹⁸O) from cave speleothems in southern Africa and ice-cores in East Africa have also been proposed as potential palaeothermometers. Of coarser resolution, borehole records from southern and East Africa yield equivalent temperature information at the centennial scale. Lake-sediment cores and historical indicators provide some information for the last two centuries, but the instrumental record of temperature generally covers only a fraction of the last century.

As this paper shows, the reconstruction of Africa’s temperature history is difficult. First, the interpretation of proxy indicators is complex and factors other than temperature have exerted considerable influence upon some records (cf. Barker et al., 2001; Gasse, 2002; Verschuren, 2004). Second, the majority of palaeotemperature records are from East and southern Africa, with the remainder of the continent poorly represented. Third, the distribution of modern instrumental records is very uneven, making it difficult to adequately assess spatial variations in temporal patterns across the whole continent. With these limitations in mind, we describe African temperature trends over the last 2000 years based on a consideration of the best data currently available.

Proxy records of temperature variability for the last two millennia

Much of what we know about long-term, pre-instrumental, temperature variability in Africa has been derived from geological proxy records such as lake sediments, speleothems and ice-cores (the locations of key palaeoenvironmental records are indicated on Figure 1). However, the majority of temperature proxies extracted from these archives are also influenced by changes in precipitation. Since the expected magnitude of temperature variations over the last 2000 years is close to the sensitivity limits of the proxy indicators employed, obtaining unambiguous reconstructions is challenging. As a result, our understanding of the temperature history of Africa over this time period is limited to a select group of reconstructions. While these lack the spatial and temporal resolution required to fully characterise continental-scale temperature variability, their similarities provide some indications that temperature changes have been coherent on fairly large spatial scales.

Figure 1.

Locations of key palaeoenvironmental sites noted in the text. 1: Ethiopian Highlands; 2: Mt. Kilimanjaro; 3: Lake Tanganyika; 4: Lake Malawi; 5: Kuiseb River, Namibia; 6: Wonderkrater, South Africa; 7: Cold Air Cave, South Africa; 8: Cango Cave, South Africa. Grey triangles indicate borehole sites used for temperature reconstruction by Huang et al. (2000).

Continuous records of temperature variability for the last two millennia in East Africa have been obtained from Lake Tanganyika (Tierney et al., 2008, 2010a) and from Mt. Kilimanjaro (Thompson et al., 2002). Other studies in the region, including a record from Lake Malawi (Powers et al., 2005a, 2011), pollen evidence from the Ethiopian Highlands (Bonnefille and Umer, 1994) and chironomid analyses from the Ruwenzori Mountains of western Uganda (Eggermont et al., 2010), provide shorter time series, at most covering the period since the MCA. Oxygen isotope data from cave speleothems (Holmgren et al., 2003; Lee-Thorp et al., 2001; Stevenson et al., 1999; Talma and Vogel, 1992), borehole data (Huang et al., 2000) and selected lower-resolution pollen data (Scott, 1996, 1999; Scott and Thackeray, 1987; Scott et al., 2003) offer some indication of long-term temperature trends for southern Africa.

The most continuous, high-resolution and unambiguous temperature reconstructions currently available from Africa have been generated from measurements of molecular fossil palaeotemperature proxies (specifically the TEX₈₆ index) in the large East African lakes Tanganyika and Malawi. The TEX₈₆ index is based on differences in the relative degree of cyclisation in glycerol-diakyl-glycerol tetraether (GDGT) molecules produced in the membranes of the aquatic Thaumarchaeota, which show a strong correlation with temperature in modern surface-sediment samples (Kim et al., 2008, 2010; Powers et al., 2004, 2005a,b; Schouten et al., 2002; Tierney et al., 2010a) and in mesocosm experiments (Schouten et al., 2007; Wuchter et al., 2004). Although it was first developed for marine palaeotemperature reconstructions (Schouten et al., 2002), the TEX₈₆ index was subsequently found to be applicable in lake systems, albeit with additional complexities associated with the occurrence of methanogenesis or large contributions of terrigenous organic matter (Powers et al., 2004, 2005a,b; Sinninghe Damsté et al., 2012). Because of these complexities, it has been suggested that TEX₈₆ is most appropriate for use in large, deep stratifying lakes, such as Malawi and Tanganyika (Blaga et al., 2009; Powers et al., 2011; Tierney et al., 2010b). Records from these lakes are considered in detail here. A TEX₈₆ record has also been published recently for the relatively shallow (c. 35 m average depth), well-mixed and hydrologically closed Lake Turkana in Kenya (Berke et al., 2012). Chronological uncertainty on the floating, ¹⁴C-dated piston-core section of the data series prevents us from considering this part of the Turkana TEX₈₆ record; only data from the freeze-cored and ²¹⁰Pb-dated uppermost sediment section are discussed.

The TEX₈₆ record from Lake Tanganyika (Tierney et al., 2008, 2010a) shows a progressive decline of lake-surface temperature of nearly 1.5°C (from c. 24°C to 22.5°C) between ad 500 and c. ad 950. A brief warming around ad 1000 is followed by brief cooling, and then an abrupt temperature increase starts an extended warm period from c. ad 1100– 1400, which partly overlaps with, but appears to slightly lag, the MCA (Figure 2). Temperature falls abruptly by c. 1°C around ad 1450, with more variable temperatures during the LIA extending to the late 19th and early 20th centuries. Cooler mean LIA conditions relative to the MCA are not clearly evident from the Tanganyika TEX₈₆ record, but a marked warming since c. ad 1880 is apparent, suggesting that modern temperatures are the warmest of at least the last 1500 years. Lake-surface temperatures have risen from c. 23°C to c. 25.8°C in the last two centuries, a change that is distinctly anomalous in comparison with any period over the last 1500 years. The main long-term trends in reconstructed lake-surface temperature broadly parallel Northern Hemisphere (NH) temperature variation (Jansen et al., 2007; Moberg et al., 2005) (Figure 2), suggesting that temperatures in East Africa were forced by the same factors which control global temperature variation.

Figure 2.

Comparison of Northern Hemisphere mean annual temperature estimates (Moberg et al., 2005) with proxy data sets from a range of African sites (see Figure 1) representing past temperature variation. From top to bottom: a bioclimatic index from Dela Sala swamp in the Ethiopian Highlands derived from the pollen ratio between forest trees and Ericaceae (Bonnefille and Umer, 1994); the composite δ¹⁸O record from five ice-cores (KSIF1 and 2, KNIF 2 and 3 and FURT) from Mt. Kilimanjaro (Thompson et al., 2002); TEX₈₆ organic biomarker records from Lake Tanganyika (Tierney et al., 2010a) and Lake Malawi (Powers et al., 2011; revised using the TEX₈₆ calibration of Tierney et al., 2010a); the percentage of frost-intolerant Salvadora pollen in rock hyrax middens along the Kuiseb River in the Namib Desert (Scott, 1996); principal components analysis (PCA) factor 1 scores calculated from pollen assemblages from Wonderkrater, South Africa (Scott and Thackeray, 1987); speleothem δ¹⁸O from Cold Air Cave (Holmgren et al., 2003; Lee-Thorp et al., 2001) and Cango Cave (Talma and Vogel, 1992), both South Africa; and temperature change relative to present estimated from African borehole records (grey symbols in Figure 1; Huang et al., 2000).

The strength of evidence for a ‘lagged’ MCA at Lake Tanganyika is uncertain. The compound uncertainty of the ¹⁴C-based portion of the age model, which involves a variable old-carbon age offset correction (cf. Blaauw et al., 2011), will almost certainly constitute a sizable part of the apparent time lag in MCA-period warming. However, whereas the principal long-term temperature trend of the Moberg et al. (2005) NH reconstruction broadly conforms to the traditional timing of the MCA (ad 950–1250), other NH-wide reconstructions (e.g. Mann et al., 2008) suggest that medieval warmth (relative to mean annual temperature from 1961 to 1990) was common throughout the period between c. ad 600 and 1450. A real possibility exists, therefore, that the warming inferred to have occurred at Lake Tanganyika between c. ad 1100 and 1400 may represent a distinct regional African, or even a more widespread tropical, dynamic. For example, a synthesis of South American proxy temperature data (Neukom et al., 2011) reveals a sustained anomaly of medieval warming similarly dated to between c. ad 1150 and 1400.

A second TEX₈₆-based record of lake surface temperatures, spanning the last c. 700 years, was generated from Lake Malawi, another large, deep rift lake in East Africa (Powers et al., 2011). To facilitate comparison with the Lake Tanganyika record, we recalculated the Lake Malawi temperatures using the TEX₈₆-lake surface temperature calibration of Tierney et al. (2010a). This calibration has a much smaller error (0.4°C) and is more conservative at the high temperature end of the calibration than the Powers et al. (2011) lake calibration. Using the same calibration for both data sets eliminates any uncertainties in the comparison of absolute temperature estimates, as any errors in the calibration would have an equivalent effect on both records. The recalculated Lake Malawi data series is shown in Figure 2.

The TEX₈₆ palaeotemperature reconstruction from Lake Malawi indicates that lake-surface temperatures fluctuated between c. 24°C and 25.8°C over the past 700 years. The warmest temperatures occurred at around ad 1320, 1575, 1730, 1800 and since c. 1900 (Powers et al., 2011) (Figure 2). Although there are broad similarities between temperature fluctuations in Lake Malawi and variations in reconstructed solar irradiance, the direction of temperature changes associated with these variations is inconsistent. Colder conditions are inferred coincident with the Spörer Minimum (ad 1450–1550) and warmer conditions coincident with the Maunder (ad 1645–1750), Wolf (ad 1280–1350) and Dalton (ad 1790–1820) minima. The degree to which the Lake Malawi and solar records can be correlated is debatable. It may be either that the time resolution and age uncertainty of the lake record masks a stronger relationship, or that the temperature record is also reflecting other factors, such as wind-induced upwelling and/or precipitation (Powers et al., 2011). Unfortunately, the Malawi record does not extend back to the MCA, and thus any relative warming associated with this period is unrecorded.

As with the Lake Tanganyika record, the Lake Malawi record shows a trend towards warmer temperatures since ad 1880. Lake-surface temperatures were at c. 24.2°C by ad 1800, fell by approximately 1°C in the late 19th century, but have risen progressively since, only punctuated by a brief cooling event in the 1950s. The highest reconstructed temperatures in the entire record are from recently deposited core-top (surface) sediments (i.e. ad 1996; 25.5°C). The inferred magnitude of the 20th century temperature rise (1.3°C) is slightly less than that recorded in Lake Tanganyika (2°C). However, the temperature change in the Malawi record is closer to that observed in instrumental records, which indicate a warming of Lake Tanganyika surface water of 1.3°C over the last century. It is not clear whether these differences are real, or if they merely reflect the analytical limitations of the proxy calibration. Tierney et al. (2010a) point out that core-top sediments from Lake Tanganyika yield a TEX₈₆-inferred temperature (25.7°C) that is indistinguishable from lake-surface water temperatures (25.5–26.3°C) today, and that reconstructed temperature changes since 1960 (1°C) are similar to changes in local air temperature (0.8°C), suggesting that the TEX₈₆ proxy is capable of reconstructing temperature changes of this magnitude and resolution with reasonable reliability. Similarly, at Lake Malawi, the reconstructed temperature change of 0.7°C over the last 60 years is close to instrumental air temperature change over the same time period (0.6°C), i.e. fully within the uncertainty of the proxy calibration. Regardless, the data suggest significant warming of the large East African lakes over the last century. Warming is also evident in the uppermost portion of the TEX₈₆ record from Lake Turkana (Berke et al., 2012). Using the Kim et al. (2010) marine TEX₈₆ calibration on account of the lake’s brackish modern-day conditions, Burke et al. reconstructed a rise in lake surface temperature from c. 25°C to c. 26.7°C during the 19th and early to mid 20th century. However, a return to cooler conditions is inferred after ad 1950, with temperatures similar to those in the early 19th century.

Another East African record that has been used to determine past temperature variability is δ¹⁸O in Mt. Kilimanjaro ice-cores (Thompson et al., 2002), although the interpretation of this proxy has been somewhat controversial (e.g. Gasse, 2002; Kaser et al., 2004). A recent comparison with δ¹⁸O variability in diatom silica from nearby Lake Challa shows an inverse relationship throughout the Holocene, suggesting that, at least on the millennial timescale, changes in rainfall are not the dominant control on ice-core δ¹⁸O (Barker et al., 2011). Unlike the records from lakes Malawi and Tanganyika, the Kilimanjaro δ¹⁸O record shows no sustained multicentury trends over the past 1500 years, but it is marked by several 50–100 year long isotopic depletions of 2–4‰, including events at c. ad 800, 1400 and 1800. A comparison with the Tanganyika TEX₈₆ record (Tierney et al., 2010a), indicates, if anything, a broadly inverse relationship between lake-surface temperature and the ice-core δ¹⁸O signal, i.e. inferred cooling on Kilimanjaro correlates with a warmer Lake Tanganyika. This apparent contradiction suggests that the influence of temperature on ice-core δ¹⁸O may be restricted to longer-term trends, and that use of this proxy record for inferring regional temperature variability at shorter timescales must be considered carefully.

There is abundant documentary evidence that the alpine glaciers in East Africa (Ruwenzori, Mt. Kenya, Mt. Kilimanjaro) have been retreating almost without interruption since the early 20th century. Geomorphological proxy evidence that glacier recession started even earlier is consistent with regional warming from the end of the LIA (Hastenrath, 2001; Mölg et al., 2003). However, the dominant causes of this early retreat and of accelerated glacier melting in recent decades are strongly debated (e.g. Hastenrath, 2010; Hastenrath and Kruss, 1992; Mölg et al., 2006; Taylor et al., 2006; Thompson et al., 2011). Comparison of historical sedimentation patterns in Rwenzori lakes that are currently situated in glaciated and unglaciated catchments (Russell et al., 2009) indicates that post-LIA glacier retreat was underway by ad 1870, notably during a regionally wet phase which can be expected to have enhanced, rather than limited, high-mountain snowfall. This study also confirmed that the Rwenzori glaciers have stood at extended LIA positions for several centuries prior to their late 19th-century recession, notwithstanding a regionally dry climate regime (cf. Russell and Johnson, 2007). This finding supports the combined evidence for (at least modest) LIA-period cooling in the African great lakes region.

Other direct and indirect palaeotemperature reconstructions broadly support the trends observed in the East African lake records. For example, Bonnefille and Umer (1994) interpreted pollen-inferred altitudinal shifts of vegetation belts in the Ethiopian Highlands, dated to ad 1500–1800, as indicating that mean annual temperature was c. 2°C colder than today. Consistent with the Tanganyika TEX₈₆ record, warmer temperatures are inferred for the period prior to ad 1500, including the latter part of the MCA (Figure 2). Dramatic warming is evident over the last two centuries, although the low sampling resolution and limited age control of this proxy record limit our ability to constrain its rate and exact timing.

Recent warming is also evident in the different species composition of larval chironomid assemblages extracted from surface and basal short-core sediments in 16 Rwenzori lakes (Eggermont et al., 2010). Although the episodic nature of these data does not constrain the exact timing of warming, it shows that modern-day Rwenzori lake-water temperatures are on average 0.38 ± 0.11°C higher than they were between c. 85 and c. 645 years ago, depending on the time period covered by each short core.

In southern Africa, δ¹⁸O speleothem records from the Cold Air (northeastern South Africa) ( Holmgren et al., 2003; Lee-Thorp et al., 2001) and Cango caves (south coast of South Africa) (Talma and Vogel, 1992) have been used to either calculate or infer palaeotemperatures. Regional reviews based largely on these data have concluded that relatively warmer conditions prevailed during the MCA, and that the LIA was notably cooler than the present day (Tyson and Lindesay, 1992; Tyson et al., 2000). Applying the most up-to-date age models, however, and considering the variability between the records – sometimes from within the same cave – it becomes less clear whether the MCA and LIA can be so simply characterised and/or whether the records can be considered as reliable temperature proxies. At Cold Air Cave, δ¹⁸O depletions thought to be indicative of cooling occur during both the MCA and LIA. Furthermore, variations within the Cango Cave record show no coherency with the Cold Air Cave record, the trends observed in East Africa (Figure 2), or the NH (Moberg et al., 2005). These inconsistencies may be a result of the complex factors that influence δ¹⁸O in speleothems (Asrat et al., 2007; Baker et al., 2007). In southernmost Africa, these include not only temperature and amount effects, but also the significant variability in moisture source and moisture-bearing systems that may bring rainfall to the region (cf. Holmgren et al., 2003).

Greater consistency is observed in the lower-resolution pollen records from the Kuiseb River, Namibia (Scott, 1996), and Wonderkrater, near Cold Air Cave, South Africa (Scott, 1999; Scott and Thackeray, 1987; Scott et al., 2003). At the Kuiseb sites especially, the prevalence of frost-prone Salvadora during the MCA, and their sharp decline in the LIA, would seem to indicate a pattern of temperature change similar to the East African sites (Figure 2). This pattern is also expressed, albeit to a lesser extent, at Wonderkrater, and it may be that inconsistencies in the age model (which we have recalculated here using the most up-to-date calibration data), and/or the low sampling resolution, are masking a greater degree of regional coherence. It is also possible that the PCA-based method does not adequately capture what are subtle variations in pollen data. Inferences of significant post-LIA warming in these records are supported (accepting shortcomings in the chronological control of the pollen records) by palaeotemperature reconstructions derived from 85 borehole records from across Africa (Huang et al., 2000), which indicate relatively cooler temperatures during the LIA in southern Africa, and substantial warming during the 20th century(Figure 2).

A final palaeoenvironmental proxy, derived from the methylation index and cyclisation ratio of branched tetraethers (MBT and CBT, respectively) based on branched GDGTs, offers potential to reconstruct past temperature. To date, the MBT/CBT proxy has been applied to basin-scale, 20,000-year temperature reconstructions from marine sediments off the Congo (Weijers et al., 2007) and Zambezi (Schefuß et al., 2011) river mouths. First-order coherence is observed between the Congo Basin record and global (Dansgaard et al., 1993; EPICA Members, 2004) and other African palaeotemperature trends (Tierney et al., 2008). The Zambezi record, however, is remarkable in that it is reported to indicate continental temperatures during the Younger Dryas cold reversal as being c. 5°C higher than Holocene averages, and a temperature decline (rather than increase) of c. 9°C during the glacial–interglacial transition. Beyond calling the reliability of the Zambezi record into question, this suggests that the MBT/CBT proxy may require further evaluation and development before it can be applied reliably to the finer-scale temperature reconstruction required for studies of the last 2000 years.

Historical and instrumental temperature data from the 19th and 20th centuries

Historical temperature records

Bridging the proxy/instrumental gap in temperature reconstructions/data is an annually resolved chronology of 19th-century cold season variability for Lesotho in southern Africa (Grab and Nash, 2010) (Figure 3). This chronology, constructed using a variety of historical documentary sources, is not calibrated against instrumental data but rather gives an indication of winter severity that can be compared against other records. The overall trend identified in this record indicates more severe and snow-rich cold seasons during the early to mid 19th century (1833–1854) compared with the latter half of the 19th century. However, a series of severe cold seasons are identified during the mid 1880s, including a run of four winters with unusually high numbers of snowfall events immediately following the eruption of Krakatoa in 1883. A reduction in the duration of the frost season by over 20 days during the 19th century is also identified. The latter half of the Lesotho record agrees well with the Southern Hemisphere composite land-based time series for cold season temperature since ad 1858 (Jones et al., 1986) and may be used to extend this record back to the 1830s (Grab and Nash, 2010). However, the degree to which the cold season proxies can be considered as indicators of annual temperature variation is unclear.

Figure 3.

(Top) Chronology of cold season severity in Lesotho and surrounding areas of South Africa during the 19th century, reconstructed through the analysis of missionary and colonial documents. The timings of major tropical and Southern Hemisphere volcanic eruptions are also indicated. (Bottom) Variations in the seasonality of snowfall events identified from the same sources (data from Grab and Nash, 2010).

Overview of instrumental records

A large number of investigations of temperature changes exist for the instrumental period, so the remainder of this section provides an overview of the most important studies. The regions of the continent for which temperature trends have been examined most intensively are East Africa, including the highland areas, the Sahel, and South Africa. Several studies have evaluated temperature changes in detail for individual countries, such as Cameroon (Molua, 2006), Egypt (Domroes and El-Tantawi, 2005), Libya (Kenaway et al., 2009; Mantimim et al., 2011), Republic of the Congo (Samba et al., 2008), Senegal (Fall et al., 2006), Uganda (Nsubuga et al., 2011) and Zimbabwe (Unganai, 1997). Only a handful of studies have examined larger regions of the continent. Aguilar et al. (2009), for example, described temperature trends across several countries, primarily in Central Africa, including Cameroon, Central African Republic, Democratic Republic of Congo (DRC), Gabon, Guinea Conakry, Republic of the Congo and Zimbabwe. Funk et al. (2012) additionally examined the Sahel and the Horn of Africa. The majority of investigations cover 40 years or less; the analysis by Aguilar et al. (2009) of temperature extremes, for example, considers only the period 1971 to 1995.

The longer time series, such as that of Jones and Moberg (2003) from 1861 onward, are strongly biased by a few individual countries, such as South Africa, where temperature records begin in the 1840s (Hulme, 1996). Africa is included in the global temperature analysis of Alexander et al. (2006), but that analysis is likewise dominated by a few regions, most notably southern and East Africa. Many of the long-term stations lie relatively near the coast. These may be influenced strongly by sea-surface temperature variability and hence may be unrepresentative of land temperatures over the continent. This is particularly relevant for records from the countries of the western Indian Ocean (Madagascar, Seychelles, Mauritius), where land temperatures are strongly correlated with sea-surface temperatures (Vincent et al., 2011).

Continent-wide analyses of instrumental data

Of the continent-wide analyses of instrumental data, the study by Jones and Moberg (2003) is perhaps the most comprehensive (Figure 4). The analysis, which extends back to the 1860s for Africa, is based on a gridded data set and examines various time periods. During the period 1920 to 1945, a warming trend of up to 0.5°C/decade occurred. However, fewer stations were available during these years. A weak cooling trend was also apparent during the period 1945 to 1976. Between 1977 and 2001, temperature records were relatively plentiful and well distributed, although gaps remained in Angola, Zaire, and the east-central Sahara. During this period, a warming trend was observed in all available African grid boxes. The trend was evident on an annual basis and for all seasons except December–January–February (DJF), during which months many grid boxes indicated cooling.

Figure 4.

African temperature variability during the 19th and 20th centuries. From top to bottom: annual mean near-surface air temperature anomalies (1979–2010) for Africa derived using Microwave Sounding Unit total lower-tropospheric temperature data (values expressed in °C; after Collins, 2011); instrumental annual mean surface temperature (ad 1861–2000) for an area bounded by 35°N–40°S and 20°W–45°E (values expressed as anomalies (°C) from the 1961–1990 mean; after Jones and Moberg, 2003). The 19th–21st century sections of the TEX₈₆ records from Lake Tanganyika (Tierney et al., 2010a) and Lake Malawi (Powers et al., 2011) are shown for comparison.

Hulme et al. (2001) also present temperature curves averaged for the entire continent, in a record spanning the period 1901 to 1998. Time series of seasonal and annual temperatures are presented. Although the annual temperature curve is in substantial agreement with that of Jones and Moberg (2003), Hulme et al. demonstrate a temperature increase in all seasons during the period 1977 to 2001.

The few studies of instrumental data that extend back to the 19th century generally show that the lowest temperatures occurred in the 1880s, supporting the document-derived cold season chronology for Lesotho described previously (Grab and Nash, 2010). According to the Jones and Moberg curve (Figure 4), decadal average temperatures were in the order of 0.6°C below the mean for 1961–1990, then rose to roughly 0.8°C above the mean by 2001. Hulme (1996) presents temperature time series commencing in 1881 for the Southern African Development Community (SADC) region (most of Southern Hemisphere Africa) and the Nile Basin. These similarly show that lowest temperatures occurred in the late 19th century. In the SADC region, temperatures generally continued to rise, with abrupt warming occurring as of the mid 1970s. Some of the warming could be attributed to drought (Unganai, 1997). The late 20th century rise was not apparent in the Nile Basin series, but the evaluated time series ended in 1986.

The only study to examine very recent temperature variability over a large part of Africa is that of Collins (2011) (Figure 4). This assessment is based on the Microwave Sounding Unit that provides estimates of lower-tropospheric temperature during the period 1979 to 2010. The following regions were examined: all of Africa, Northern Hemisphere Africa, Southern Hemisphere Africa, tropical Africa, and subtropical Africa. Significant rising trends were apparent in all of these regions during June–August, but during December–February the warming was concentrated in Northern Hemisphere sectors only. Warming accelerated over the continent during the period 2001 to 2010.

Select regional analyses of instrumental data

The regions of Africa that have been examined in greatest detail for recent decadal-scale temperature trends include South Africa, Zimbabwe, East Africa, Cameroon and Senegal, and the Sahel overall. Temperatures over South Africa have risen considerably between 1960 and 2003, with significant positive trends at 18 of the 24 stations evaluated (Kruger and Shongwe, 2004). Positive trends were greatest in autumn and at interior stations, and smallest in spring and at coastal stations. The monthly trends show little uniformity, either between stations or between months. In contrast, Muhlenbruchtegen (1992) showed little evidence of trends in the preceding half century, again suggesting that the greatest warming occurred in recent decades.

Stern et al. (2011) examined temperature trends in East Africa, using different versions of the University of East Anglia’s Climatic Research Unit data base. They emphasised that the presence of trends depended on which version of the data base was utilised, underscoring the importance of evaluating uniform time periods for comparative studies. Positive temperature trends were generally evident in East Africa in the latter decades of the 20th century (Stern et al., 2011), a conclusion extended by recent studies (e.g. Nsubuga et al., 2011). The analysis by Funk et al. (2012) of six stations in Kenya and Ethiopia, extending from 1960 to 2009 (Figures 5 and 6), shows a similar increase during the March–June and June–September seasons.

Figure 5.

Station observations of temperature change between the 1960–1989 average and the 2000–2009 average for MAMJ (left) and JJAS (right) (adapted from Funk et al., 2012).

Figure 6.

Time series of temperature for two areas (Sudan–Niger–Mali and Kenya–Ethiopia) and two seasons (MAM and JJAS) from 1960 to 2009 (from Funk et al., 2012). Stations utilised are shown in Figure 5.

Other investigations, such as those of Pascual et al. (2006) and Alonso et al. (2011), have also confirmed warming across the East African highlands. Such findings are supported by analyses of surface-water temperatures in Lake Victoria, which have increased by over 1°C between 1927 and 2009 (Sitoki et al., 2010). Water temperatures in Lake Tanganyika have also increased by c. 1°C between 1913 and 2000; this magnitude of warming was evident from the surface to some 200 m in depth (Verburg et al., 2003). Warming of Lake Malawi by c. 0.7–1.4°C over the past 50 years is indicated by water temperature measurements (Vollmer et al., 2005) and, as already noted, by the TEX₈₆ analysis of a sediment core (Powers et al., 2010). However, the reasons for lake warming are complex, so that the lake temperature cannot be assumed to be fully representative of air temperatures over the surrounding land.

In West Africa, 12 stations in Senegal recorded a positive and significant warming trend throughout the country (except in the southeast) for the period 1971 to 1998 (Fall et al., 2006). The trend was strongly correlated with sea-surface temperatures, again raising the question of the representativeness of these stations. In addition, a strong positive trend for the months of July, August and September has been identified for select stations in Cameroon over the period 1960–2000 (Molua, 2006).

Funk et al. (2012) examined temperatures during the March–June and June–September seasons across the Sahel from Senegal to Sudan (with the exception of Chad). During the period 1960 to 2009 there was a consistent increase at nearly every station evaluated (Figure 5). The average increase for Sudan, Mali and Niger was over 1°C during both seasons (Figure 6).

Maximum/minimum temperatures and temperature extremes

The most comprehensive study of African daily temperature extremes is that of New et al. (2006), who analysed daily temperatures during the period 1961–2000 in 13 southern and West African countries (Botswana, Gambia, Lesotho, Malawi, Mauritius, Mozambique, Namibia, Nigeria, Seychelles, South Africa, Tanzania, Uganda and Zambia). A large proportion of stations indicate statistically significant trends for the majority of temperature indices evaluated. For example, the occurrence of hot days and nights (95th percentile) increased by over 8 days/decade. The occurrence of extreme cold decreased at the same time. The greatest trend was in the positive tail of the frequency distribution of daily maximum temperatures.

A similar analysis was performed using data from 38 stations across six equatorial African countries (Cameroon, Central African Republic, the DRC, Guinea Conakry, Republic of the Congo and Zimbabwe), spanning the period 1971 to 1995 (Aguilar et al., 2009). Nine temperature indices were used to identify extremes: warmest day and night, coldest day and night, diurnal range, cold day and night frequency, warm day and night frequency. Statistically significant warming of c. 0.1–0.2°C/decade was evident in both the cold and warm extremes.

Regional analyses of diurnal temperature ranges for Ethiopia and Zimbabwe were presented by Hulme et al. (2001) and compared against similar analyses for South Africa and the Sudan (Jones and Lindesay, 1993). In all cases except Zimbabwe, the diurnal range decreased from the mid to the late 20th century. For the period 1901 to 1995, temperature trends were positive over most of the continent (Hulme et al., 2001). Temperature increases over this period were generally small (< 0.5°C), but much more intense warming was apparent in parts of southern and northwestern Africa (≥2°C). A few areas exhibited cooling trends: coastal sectors of extreme southern Africa and far West Africa; eastern equatorial sectors of the Sudan, Ethiopia, Egypt, and Somalia; and over the highlands of Cameroon and Nigeria.

Surface temperature variations up to 1993 for the entire country of Zimbabwe, and for the stations Harare and Bulawayo, were examined by Unganai (1997). The record commenced in 1897 for the two city stations, but only began in 1933 for most other stations. The national average recorded a warming trend in maximum temperatures, but minimum temperatures indicated either slight cooling or no trend. Harare and Bulawayo both record a warming trend in maximum and minimum temperatures since records began.

In South Africa, the study of temperature variability over the period 1960 to 2003 by Kruger and Shongwe (2004), noted previously, included analyses of minimum and maximum temperatures. The majority of stations (23 out of 26) showed positive trends in annual mean maximum temperature, whilst positive trends in annual mean minimum temperature were apparent at 21 stations. In both cases, although most trends were significant, there was no clear pattern in the trends in diurnal temperature range. However, days and nights with relatively high temperatures generally increased over the analysis period, whilst days and nights with relatively low temperatures decreased. Muhlenbruchtegen (1992) also studied South African temperatures and concluded that minimum temperatures had increased over a 50 year period, while maximum temperatures decreased at the same time. The results are not directly comparable with those of Kruger and Shongwe (2004), since the latter study covered a later time period.

Variations in maximum and minimum temperatures in East Africa have been analysed by Christy et al. (2009), using 60 stations in Kenya and 58 stations in Tanzania. Data were gridded and the analysis period and data density differed among the grid points. At the most data-rich five latitude-longitude degree cells, which included areas around Nairobi, the record commenced in 1905. In this cell the trend in maximum temperatures was not significant, but a significant rise in minimum temperatures was measured.

The combined temperature record

The preceding sections illustrate the limitations of both proxy indicators and the instrumental record for analysing temperature variability across Africa during the last 2000 years. Proxy temperature records covering at least a few centuries are limited to East and southern Africa and there are few high-resolution reconstructions. Furthermore, few proxies appear to give a completely unambiguous temperature signal. Instead, most are influenced to some extent by other climatic and non-climatic factors, making calibration problematic. The most notable shortcomings of the instrumental record are the unevenness of the station distribution in both space and time, and the fact that most studies are based on only 25 to 50 years of data. Few stations have records approaching even a century. The lack of a consistent analytical approach between studies further hinders the assessment of trends. However, by considering the proxy, documentary and instrumental temperature records as a whole, some general conclusions about temperature variability across Africa during the last 2000 years can be drawn. We now focus on conditions during three key time periods – the MCA, LIA and the late 19th to early 21st centuries.

The most important records from continental Africa covering the MCA give ambiguous signals. Warmer conditions during the MCA are clearly evident in Lake Tanganyika, the Ethiopian Highlands and in the Cango Cave, Kuiseb River and Wonderkrater records from southern Africa. Evidence from the Mt. Kilimanjaro ice-core δ¹⁸O record (Thompson et al., 2002) is less clear. The warming, within the limits of dating, appears to have been greater during the early MCA in some parts of southern Africa (e.g. Cango Cave (Talma and Vogel, 1992); Wonderkrater (Scott, 1999; Scott and Thackeray, 1987)) and during the later MCA in Namibia (Scott, 1996), Ethiopia (Bonnefille and Umer, 1994) and at Lake Tanganyika (Tierney et al., 2008, 2010a) (Figure 2).

LIA cooling is evident in Ethiopian and southern African pollen records (Bonnefille and Umer, 1994; Scott, 1996; Scott et al., 2003) (Figure 2). According to the Lake Malawi record, coolest conditions occurred during the middle of the LIA (Powers et al., 2005a, 2011), but LIA-period temperature depression at Lake Tanganyika appears to have been less persistent through time (Tierney et al., 2008, 2010a). The LIA record from Lake Malawi also indicates several periods of intermittent warming, but without data from the MCA, these fluctuations remain without good context. The Mt. Kilimanjaro ice-core record (Thompson et al., 2002) shows decadal–centennial isotopic depletions at the start and end of the LIA, but δ¹⁸O values similar to those from the MCA during the remainder of the LIA.

Instrumental temperature data for the late 19th to early 21st centuries show a clear warming trend (Figure 4). The few available instrumental records that extend back beyond 1900 suggest that this warming started at least as early as the 1880s (the last cold decade of the 19th century across the continent; Jones and Moberg, 2003). This is consistent with documentary reconstructions from southern Africa (Grab and Nash, 2010) (Figure 3). The proxy record of African alpine glacier recession suggests that post-LIA warming was already underway by the 1870s (Russell et al., 2009). The 20th century warming pattern is also evident in TEX₈₆ records from lakes Tanganyika and Malawi (Figure 4) and in borehole records from southern Africa (Huang et al., 2000). A brief period of cooling occurred in the mid 20th century, but warming has been nearly continuous from the 1970s to the present day.

The rate of temperature increase appears to have accelerated during the past two decades (Figures 5 and 6), and is evident in all seasons. Analyses of extreme temperature and diurnal range do not show uniform results. However, radiosonde data from South Africa suggest that a cooling trend is evident in the upper troposphere (Bencherif et al., 2006). This is consistent with surface warming resulting from the trapping of radiation by greenhouse gases.

It is important to point out that the average temperature trends identified in compilations of instrumental and satellite data are not necessarily representative for all regions of Africa. Global warming is expected to alter the patterns of atmospheric circulation, including the meridional (i.e. north–south) airflow. Thus, some areas, particularly in the extra-tropical extremes of Africa, could conceivably experience relatively lower temperatures within a warmer world.

In conclusion, our analysis has revealed the limited but growing number of data sets available to assess temperature variability across Africa over the last 2000 years. The most promising proxy for reconstructing longer-term temperature fluctuations appears to be the TEX₈₆ organic biomarker. Provided that lake-specific confounding factors can be properly controlled, studies in other deep East African lakes may permit powerful comparisons with the Lake Malawi and Tanganyika records. Geographically, much more focus is needed on production of new high-resolution proxy temperature records from North and West Africa, and on the development of better temporally resolved records from new and existing sites in southern Africa, to allow continent-wide patterns of temperature variability to be revealed. Efforts are also required at national and international level to ensure that the network of meteorological stations recording temperature data is maintained and, if possible, expanded. Only in this way will the scale of temperature change already identified across Africa during the late 20th and early 21st centuries be adequately monitored, and data made available for future modelling investigations.

Footnotes

Acknowledgements

This paper is dedicated to the memory of Dr Mohammed Umer, co-convenor of the PAGES Africa2k Working Group, who sadly passed away on 27 November 2011.

Funding

The authors would like to thank PAGES for financial support of an Africa2k workshop held in Ghent in 2010, when the idea for this study was first discussed. BM Chase was further supported by the European Research Council (ERC) under the European Union’s Seventh Framework Programme (FP7/2007-2013)/ERC Starting Grant ‘HYRAX’, grant agreement no. 258657. SE Nicholson was supported by NSF Grant 1158984 and NOAA Project NAO80AR4310731.

References

Aguilar

Barry

Brunet

. (2009) Changes in temperature and precipitation extremes in western central Africa, Guinea Conakry, and Zimbabwe, 1955–2006. Journal of Geophysical Research 114: D02115, DOI: 10.1029/2008JD011010.

Alexander

Zhang

Peterson

. (2006) Global observed changes in daily climate extremes of temperature and precipitation. Journal of Geophysical Research 111: D05109, DOI: 10.1029/2005JD006290.

Alonso

Bouma

Pascual

(2011) Epidemic malaria and warmer temperatures in recent decades in an East African highland. Proceedings of the Royal Society B – Biological Sciences 278: 1661–1669.

Asrat

Baker

Umer

. (2007) A high-resolution multi-proxy stalagmite record from Mechara, Southeastern Ethiopia: Paleohydrological implications for speleothem paleoclimate reconstruction. Journal of Quaternary Science 22: 53–63.

Baker

Asrat

Fairchild

. (2007) Analysis of the climate signal contained within δ¹⁸O and growth rate parameters in two Ethiopian stalagmites. Geochimica et Cosmochimica Acta 71: 2975–2988.

Barker

Hurrell

Leng

. (2011) Seasonality in equatorial climate over the past 25 k.y. revealed by oxygen isotope records from Mount Kilimanjaro. Geology 39: 1111–1114.

Barker

Street-Perrott

Leng

. (2001) A 14,000 year oxygen isotope record from diatom silica in two Alpine lakes on Mt Kenya. Science 292: 2307–2310.

Bencherif

Diab

Portafaix

. (2006) Temperature climatology and trend estimates in the UTLS region as observed over a southern subtropical site, Durban, South Africa. Atmospheric Chemistry and Physics 6: 5121–5128.

Berke

Johnson

Werne

. (2012) A mid-Holocene thermal maximum at the end of the African Humid Period. Earth and Planetary Science Letters 351: 95–104.

10.

Blaauw

van Geel

Kristen

. (2011) High-resolution ¹⁴C dating of a 25,000 year lake-sediment record from equatorial East Africa. Quaternary Science Reviews 30: 3043–3059.

11.

Blaga

Reichart

Heiri

. (2009) Tetraether membrane lipid distributions in water-column particulate matter and sediments: A study of 47 European lakes along a north–south transect. Journal of Paleolimnology 41: 523–540.

12.

Bonnefille

Umer

(1994) Pollen inferred climatic fluctuations in Ethiopia during the last 3000 years. Palaeogeography, Palaeoecology, Palaeoclimatology 109: 331–343.

13.

Chase

Meadows

(2007) Late Quaternary dynamics of southern Africa’s winter rainfall zone. Earth-Science Reviews 84: 103–138.

14.

Christy

Norris

McNider

(2009) Surface temperature variations in East Africa and possible causes. Journal of Climate 22: 3342–3356.

15.

Collins

(2011) Temperature variability over Africa. Journal of Climate 24: 3649–3666.

16.

Dansgaard

Johnsen

Clausen

. (1993) Evidence for general instability of past climate from a 250 kyr ice core record. Nature 364: 218–220.

17.

Domroes

El-Tantawi

(2005) Recent temporal and spatial temperature changes in Egypt. International Journal of Climatology 25: 51–63.

18.

Eggermont

Verschuren

Audenaert

. (2010) Limnological and ecological sensitivity of Rwenzori mountain lakes to climate warming. Hydrobiologia 648: 123–142.

19.

EPICA Members (2004) Eight glacial cycles from an Antarctic ice core. Nature 429: 623–628.

20.

Fall

Niyogi

Semazzi

FHM

(2006) Analysis of mean climate conditions in Senegal (1971–98). Earth Interactions 10, Article number 5.

21.

Funk

Michaelsen

Marshall

(2012) Mapping recent decadal climate variations in precipitation and temperature across eastern Africa and the Sahel. In: Wardlow

Anderson

Verdin

(eds) Remote Sensing of Drought: Innovative Monitoring Approaches. Cambridge University Press, Chapter 14, 484 pp.

22.

Gasse

(2000) Hydrological changes in the African tropics since the Last Glacial Maximum. Quaternary Science Reviews 19: 189–211.

23.

Gasse

(2002) Kilimanjaro’s secrets revealed. Science 298: 548–549.

24.

Gasse

Chalié

Vincens

. (2008) Climatic patterns in equatorial and southern Africa from 30,000 to 10,000 years ago reconstructed from terrestrial and near-shore proxy data. Quaternary Science Reviews 27: 2316–2340.

25.

Grab

Nash

(2010) Documentary evidence of climate variability during cold seasons in Lesotho, southern Africa, 1833–1900. Climate Dynamics 34: 473–499.

26.

Hastenrath

(2001) Variations of East African climate during the past two centuries. Climatic Change 50: 209–217.

27.

Hastenrath

(2010) Climatic forcing of glacier thinning on the mountains of equatorial East Africa. International Journal of Climatology 30: 146–152.

28.

Hastenrath

Kruss

(1992). The dramatic retreat of Mount Kenya’s glaciers between 1963 and 1987: Greenhouse forcing. Annals of Glaciology 16: 127–133.

29.

Holmgren

Lee-Thorp

Cooper

GRJ

. (2003) Persistent millennial-scale climatic variability over the past 25,000 years in Southern Africa. Quaternary Science Reviews 22: 2311–2326.

30.

Huang

Pollack

Shen

P-Y

(2000) Temperature trends over the past five centuries reconstructed from borehole temperatures. Nature 403: 756–758.

31.

Hulme

(1996) Climate change within the period of meteorological records. In: Adams

Goudie

Orme

(eds) The Physical Geography of Africa. Oxford: Oxford University Press, pp. 88–102.

32.

Hulme

Doherty

Ngara

. (2001) African climate change: 1900–2100. Climate Research 17: 145–168.

33.

Jansen

Overpeck

Briffa

. (2007) Palaeoclimate. In: Solomon

Qin

Manning

. (eds) Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, pp. 433–497.

34.

Jones

Lindesay

(1993) Maximum and minimum temperature trends over South Africa and the Sudan. Proceedings of the Fourth International Conference on Southern Hemisphere Meteorology and Oceanography. Hobart, Boston: American Meteorological Society, pp. 359–360.

35.

Jones

Moberg

(2003) Hemispheric and large-scale surface air temperature variations: An extensive revision and an update to 2001. Journal of Climate 16: 206–223.

36.

Jones

Osborn

Briffa

(2001) The evolution of climate over the last millennium. Science 292: 662–667.

37.

Jones

Raper

SCB

Wigley

TML

(1986) Southern Hemisphere surface air temperature variations: 1851–1984. Journal of Applied Meteorology and Climatology 25: 1213–1230.

38.

Kaser

Hardy

Mölg

. (2004) Modern glacier retreat on Kilimanjaro as evidence of climate change: Observations and facts. International Journal of Climatology 24: 329–339.

39.

Kenaway

AMEL

Lopez-Moreno

Vincente-Serrano

. (2009) Temperature trends in Libya over the second half of the 20th century. Theoretical and Applied Climatology 98: 1–8.

40.

Kim

J-H

Schouten

Hopmans

. (2008) Global sediment core-top calibration of the TEX86 paleothermometer in the ocean. Geochimica et Cosmochimica Acta 72: 1154–1173.

41.

Kim

J-H

van der Meer

Schouten

. (2010) New indices and calibrations derived from the distribution of crenarchaeal isoprenoid tetraether lipids: Implications for past sea surface temperature reconstructions. Geochimica et Cosmochimica Acta 74: 4639–4654.

42.

Kruger

Shongwe

(2004) Temperature trends in South Africa: 1960–2003. International Journal of Climatology 24: 1929–1945.

43.

Lee-Thorp

Holmgren

Lauritzen

S-E

. (2001) Rapid climate shifts in the southern African interior throughout the mid- to late Holocene. Geophysical Research Letters 28: 4507–4510.

44.

Mann

Zhang

Hughes

. (2008) Proxy-based reconstructions of hemispheric and global surface temperature variations over the past two millennia. Proceedings of the National Academy of Sciences of the United States of America 105: 13,252–13,257.

45.

Mantimim

El-Tantawi

AMM

Schaefer

. (2011) Recent trends of temperature change under hot and cold desert climates: Comparing the Sahara (Libya) and Central Asia (Xinjiang, China). Journal of Arid Environments 75: 1105–1113.

46.

Matthews

Briffa

(2005) The ‘Little Ice Age’: Re-evaluation of an evolving concept. Geografiska Annaler 87A: 17–36.

47.

Moberg

Sonechkin

Holmgren

. (2005) Highly variable Northern Hemisphere temperatures reconstructed from low- and high-resolution proxy data. Nature 433: 613–617.

48.

Mölg

Georges

Kaser

(2003) The contribution of increased incoming shortwave radiation to the retreat of the Rwenzori Glaciers, East Africa, during the 20th century. International Journal of Climatology 23: 291–303.

49.

Mölg

Rott

Kaser

. (2006) Comment on ‘Recent glacial recession in the Rwenzori Mountains of East Africa due to rising air temperature’. Geophysical Research Letters 33: doi:10.1029/2006GL027254.

50.

Molua

(2006) Climatic trends in Cameroon: Implications for agricultural management. Climate Research 30: 225–262.

51.

Muhlenbruchtegen

(1992) Long-term surface temperature variations in South Africa. South African Journal of Science 88: 197–205.

52.

Neukom

Luterbacher

Villalba

. (2011) Multiproxy summer and winter surface air temperature field reconstructions for southern South America covering the past centuries. Climate Dynamics 37: 35–51.

53.

New

Hewitson

Stephenson

. (2006) Evidence of trends in daily climate extremes over southern and west Africa. Journal of Geophysical Research – Atmospheres 111(D14): D14102, DOI: 10.1029/2005JD006289.

54.

Nsubuga

FNW

Olwoch

Rautenbach

CJdeW

(2011) Climatic trends at Namulonge in Uganda: 1947–2009. Journal of Geography and Geology 3: 119–131.

55.

Pascual

Ahumada

Chaves

. (2006) Resurgence in the East African highlands: Temperature trends revisited. Proceedings of the National Academy of Sciences of the United States of America 103: 5829–5834.

56.

Powers

Johnson

Werne

. (2005a) Large temperature variability in the southern African tropics since the Last Glacial Maximum. Geophysical Research Letters 32: L08706.

57.

Powers

Johnson

Werne

. (2011) Organic geochemical records of environmental variability in Lake Malawi during the last 700 years, Part I: The TEX86 temperature record. Palaeogeography, Palaeoclimatology, Palaeoecology 303: 133–139.

58.

Powers

Werne

Johnson

. (2004) Crenarchaeotal membrane lipids in lake sediments: A new paleotemperature proxy for continental paleoclimate reconstruction? Geology 32: 613–616.

59.

Powers

Werne

Johnson

. (2005b) The development of TEX₈₆ for continental paleotemperature reconstruction: Problems and promise. Geochimica et Cosmochimica Acta 69: A78–A78.

60.

Powers

Werne

Vanderwoude

. (2010) Applicability and calibration of the TEX₈₆ paleothermometer in lakes. Organic Geochemistry 41: 404–413.

61.

Russell

Johnson

(2007) Little Ice Age drought in equatorial Africa: Intertropical Convergence Zone migrations and El Niño-Southern Oscillation variability. Geology 35: 21–24.

62.

Russell

Eggermont

Taylor

. (2009) Paleolimnological records of recent glacier recession in the Rwenzori Mountains, Uganda-D. R. Congo. Journal of Paleolimnology 41: 253–271.

63.

Samba

Nganga

Mpounza

(2008) Rainfall and temperature variations over Congo-Brazzaville between 1950 and 1998. Theoretical and Applied Climatology 91: 85–97.

64.

Schefuß

Kuhlmann

Mollenhauer

. (2011) Forcing of wet phases in southeast Africa over the past 17,000 years. Nature 480: 509–512.

65.

Schouten

Forster

Panoto

. (2007) Towards calibration of the TEX86 palaeothermometer for tropical sea surface temperatures in ancient greenhouse worlds. Organic Geochemistry 38: 1537–1546.

66.

Schouten

Hopmans

Schefuss

. (2002) Distributional variations in marine crenarchaeotal membrane lipids: A new tool for reconstructing ancient sea water temperatures? Earth and Planetary Science Letters 204: 265–274.

67.

Scott

(1996) Palynology of hyrax middens: 2000 years of palaeoenvironmental history in Namibia. Quaternary International 33: 73–79.

68.

Scott

(1999) Vegetation history and climate in the Savanna biome South Africa since 190,000 ka: A comparison of pollen data from the Tswaing Crater (the Pretoria Saltpan) and Wonderkrater. Quaternary International 57–58: 215–223.

69.

Scott

Thackeray

(1987) Multivariate analysis of Late Pleistocene and Holocene pollen spectra from Wonderkrater, Transvaal, South Africa. South African Journal of Science 83: 93–98.

70.

Scott

Holmgren

Talma

. (2003) Age interpretation of the Wonderkrater spring sediments and vegetation change in the Savanna Biome, Limpopo province, South Africa. South African Journal of Science 99: 484–488.

71.

Sinninghe Damsté

Ossebaar

Schouten

. (2012) Distribution of tetraether lipids in the 25-kyr sedimentary record of Lake Challa: Extracting reliable TEX86 and MBT/CBT palaeotemperatures from an equatorial African lake. Quaternary Science Reviews 50: 43–53.

72.

Sitoki

Gichuki

Ezekiel

. (2010) The environment of Lake Victoria (East Africa): Current status and historical changes. International Review of Hydrobiology 95: 209–223.

73.

Stern

Gething

Kabaria

. (2011) Temperature and malaria trends in highland East Africa. PloS One 6: doi:e24524.

74.

Stevenson

Lee-Thorp

Holmgren

(1999) A 3000-year isotopic record from a stalagmite in Cold Air Cave, Makapansgat Valley, Northern Province. South African Journal of Science 95: 46–48.

75.

Talma

Vogel

(1992) Late Quaternary paleotemperatures derived from a speleothem from Cango Caves, Cape Province, South Africa. Quaternary Research 37: 203–213.

76.

Taylor

Milcham

Tindimugaya

. (2006) Recent glacial recession in the Rwenzori mountains of East Africa due to rising air temperature. Geophysical Research Letters 33: L10402, DOI:10.1029/2006GL025962.

77.

Thompson

Mosley-Thompson

Davis

. (2002) Kilimanjaro ice core records: Evidence of Holocene climate change in tropical Africa. Science 298: 589–593.

78.

Thompson

Mosley-Thompson

Davis

. (2011) Tropical glaciers, recorders and indicators of climate change, are disappearing globally. Annals of Glaciology 52: 23–34.

79.

Tierney

Mayes

Meyer

. (2010a) Late-twentieth-century warming in Lake Tanganyika unprecedented since AD 500. Nature Geoscience 3: 422–425.

80.

Tierney

Russell

Eggermont

. (2010b) Environmental controls on branched tetraether lipid distributions in tropical East African lake sediments. Geochimica et Cosmochimica Acta 74: 4902–4918.

81.

Tierney

Russell

Huang

. (2008) Northern Hemisphere controls on tropical southeast African climate during the past 60,000 years. Science 322: 252–255.

82.

Tyson

Lindesay

(1992) The climate of the last 2000 years in southern Africa. The Holocene 2: 271–278.

83.

Tyson

Karlen

Holmgren

. (2000) The Little Ice Age and medieval warming in South Africa. South African Journal of Science 96: 121–126.

84.

Unganai

(1997) Surface temperature variation over Zimbabwe between 1897 and 1993. Theoretical and Applied Climatology 56: 89–101.

85.

Verburg

Hecky

Kling

(2003) Ecological consequences of a century of warming in Lake Tanganyika. Science 301: 505–507.

86.

Verschuren

(2004) Decadal to century scale climate variability in tropical Africa during the past 2000 years. In: Battarbee

Gasse

Stickley

. (eds) Past Climate Variability Through Europe and Africa. Dordrecht: Kluwer, pp. 139–158.

87.

Vincent

Aguilar

Saindou

. (2011) Observed trends in indices of daily and extreme temperature and precipitation for the countries of the western Indian Ocean, 1961–2008. Journal of Geophysical Research – Atmospheres 116: D10108, DOI: 10.1029/2010JD015303.

88.

Vollmer

Bootsma

Hecky

. (2005) Deep-water warming trend in Lake Malawi, East Africa. Limnology and Oceanography 50: 727–732.

89.

Weijers

JWH

Schefuß

Schouten

. (2007) Coupled thermal and hydrological evolution of tropical Africa over the last deglaciation. Science 315: 1701–1704.

90.

Wuchter

Schouten

Coolen

MJL

. (2004) Temperature-dependent variation in the distribution of tetraether membrane lipids of marine Crenarchaeota: Implications for TEX₈₆ paleothermometry. Paleoceanography 19: PA4028.