Abstract
Multiple proxies using variation in δ18O, δ13C, mineralogy, and petrography in a newly generated high-resolution record of Stalagmite DP1 from Dante Cave indicate a linkage between changes in hydroclimate in northeastern Namibia and changes in solar activity and changes in global temperatures. The record suggests that during solar minima and globally cooler conditions (ca. 1660–1710 and ca. 1790–1830), wetter periods (reflecting longer summer seasons) in northeastern Namibia were linked to advances of the Inter-Tropical Convergence Zone (ITCZ) and the Inter-Ocean Convergence Zone (IOCZ) southwestward. A slight southward push of the Angola–Benguela Front (ABF) during such intervals could also be expected, bringing more rainfall inland. On the other hand, drier and warmer periods in northeastern Namibia, inferred from the increasing δ18O trend in Stalagmite DP1 after AD 1715, coincide with globally warmer conditions, and thus a northeastward migration of the ITCZ, specifically with more warming of the Northern Hemisphere (NH). This finding agrees with reducing precipitation observed in the summer rainfall zone of southern Africa since ca. 1900. Therefore, predictions of warming in high-latitude regions of the NH in the next century should suggest that the presently semi-arid climate of northern Namibia may become even drier.
Keywords
Introduction
The Inter-Tropical Convergence Zone (ITCZ) responds to changes in interhemispheric temperature contrast (e.g. Broccoli et al., 2006). For example, cooling in the Northern Hemisphere (NH) pushed the ITCZ southward, and this resulted in wetter conditions in the Southern Hemisphere (SH) (e.g. Sachs et al., 2009; Schefuß et al., 2011). This hypothesis was supported by paleoclimatological analyses (Brown and Johnson, 2005; Koutavas and Lynch-Stieglitz, 2004; Leech et al., 2013; Russell and Johnson, 2005; Sinha et al., 2011) and by modeling studies (Broccoli et al., 2006; Chiang and Bitz, 2005; Chiang et al., 2003; Donohoe et al., 2013; Frierson and Hwang, 2012; Kang et al., 2008; McGee et al., 2014). It was also summarized in literature reviews (e.g. Chiang and Friedman, 2012; Schneider et al., 2014). As a result, several locations in the SH became wetter, for example, during the Heinrich Stadial 1 and the Younger Dryas in southeast Africa (e.g. Schefuß et al., 2011) and during the ‘Little Ice Age’ (LIA) in the tropical Pacific (e.g. Sachs et al., 2009). Multiple proxies from Stalagmite DP1 from northeastern Namibia suggested an abrupt transition from drier to wetter conditions at 1.8 ka, when climate in the NH cooled, and another transition from drier to wetter conditions at the ‘Medieval Warm Period’ and LIA transition, when cooling also prevailed in the NH (Sletten et al., 2013). The aim of this paper is to investigate in much detail whether such cooling has persistently affected the hydroclimate in northeastern Namibia at shorter time interval, between ca. AD 1400 and 1950. In this regard, we study the upper portion of Stalagmite DP1 at very high resolution (ca. 1–2 years) to provide a more comprehensive paleoclimate data set from northeastern Namibia and to link information from such data to regional and global conditions.
Namibia is a critical location to study paleoclimate for the following reasons. High-resolution paleorecords from the past millennium prior to the advent of instrumental records are presently scarce in the region (e.g. Neukom and Gergis, 2012), despite the increasing number of paleo-archives in continental Africa (e.g. Nicholson et al., 2013; Neukom et al., 2014b). Thus, records from Namibia should complete gaps in paleoclimate reconstruction in the SH (e.g. Neukom et al., 2014b). Northeastern Namibia is also located in a climatic sensitive region, where rainfall is linked to the ITCZ, the Inter-Ocean Convergence Zone (IOCZ), and the Angola–Benguela Front (ABF) (Figure 1). Namibia is one of the driest countries in Sub-Saharan Africa (Reid et al., 2007), and about one-third of all years can be described as drought- or flood-prone (Jury and Engert, 1999). Recent studies also suggest that the dune systems in the Kalahari basin that are used in pastoral and agricultural activities will be reactivated by 2099 (Thomas et al., 2005). In dry years, Namibia becomes a desert and failures of small farming operations are very likely (Jury and Engert, 1999). Thus, Namibia is among regions in Africa where sensitivity to climate changes is substantial (Devereux and Næraa, 1996; Dirkx et al., 2008; Niang et al., 2014; Thomas et al., 2005).
Map of Africa with thick red lines showing position of the Inter-Tropical Convergence Zone (ITCZ) in austral winter and austral summer and with a thinner red line showing the position of the Congo Air Boundary (CAB), Zaire Air Boundary (ZAB), or Inter-Oceanic Convergence Zone (IOCZ) in austral summer. Contours and colored fields show geographic variation of mean annual atmospheric precipitation for 1979–2011 from the Global Precipitation Climatology Project (GPCP) of the US National Aeronautics and Space Administration (NASA). Currents in the southeastern Atlantic are Southern Equatorial Counter Current (SECC), Angola Current (AC), and Benguela Coastal Current (BCC). ABF is the Angola–Benguela Front in austral winter (W) and austral summer (S). The currents and position of ABF are from Benthien et al. (2002), Veitch et al. (2006), and Dupont et al. (2008). The lines for the ITCZ are compromises between the lines shown by Dupont et al. (2008), Nicholson (2009), and Wang (2009). The line for the IOCZ is a compromise between the lines shown by Van Heerden and Taljaard (1998), Dupont et al. (2008), and Schefuß et al. (2011). Note that, east of about 15°E, land is distributed symmetrically with respect to the Equator, and the winter and summer positions of the ITCZ are likewise positioned symmetrically with respect to the Equator. By contrast, west of about 15°E, African land is dominantly in the Northern Hemisphere (NH), and the ITCZ is restricted to the NH. The location of Dante Cave is shown with a large asterisk.
The interval studied here, AD 1400–1950, is particularly important because it is characterized by regional- to global-scale variations in temperature and insolation. It is characterized by the LIA (Bradley and Jones, 1993; Chambers, 2015; Mann, 2002; Mann et al., 2009) and by the warming of the last century (e.g. Bradley, 2000; Crowley, 2000; Folland et al., 1990) of more than 0.5°C over most parts of Africa (Niang et al., 2014 and references therein). The LIA is recognized in the NH as a time of exceptionally cold temperature, whereas in the SH it seems to be a time distinguished by change in precipitation and winds (e.g. Chambers et al., 2014; Moy et al., 2008). This interval is also characterized by three events of low solar activity (Spörer, Maunder, and Dalton minima), which may have contributed to the LIA’s coldest phases (cf. Hathaway, 2010; Mauquoy et al., 2002; Muscheler et al., 2007; Usoskin et al., 2003, 2007; Figure 5a–c). These conditions in turn could have amplified the temperature gradient between the two hemispheres (cf. Neukom et al., 2014a) and could have affected the dynamics of the ITCZ (e.g. Chiang and Bitz, 2005; Yan et al., 2015).
Stalagmites were investigated because they can be accurately dated (e.g. Dorale et al., 2004; Edwards et al., 1987). They have a potential in storing valuable climatic information in several proxies (e.g. Fairchild and Baker, 2012); stable isotopes, petrography, and mineralogy are the main proxies used. Stalagmite multiple proxy analyses have been shown to provide comprehensive paleoclimate information from regions (e.g. Brook et al., 2009, 2015; Railsback et al., 2014; Sletten et al., 2013) where cave monitoring and/or sample replication is often impossible because of the high cost and/or other environmental regulations (e.g. Sletten et al., 2013).
Setting
Physical setting of Dante Cave
Stalagmite DP1 was collected from the Purgatory Chamber of Dante Cave (19° 24′S; 17° 53′E). It was accidentally toppled, during surveying of the chamber in 1988 (Figure 2). The sample was collected in 2007, and the absence of sediment or staining indicated that the stalagmite had not been submerged after it was toppled. Dante Cave is one of the longest (828 m) and deepest (65 m) caves in Namibia’s Karstveld (Sletten et al., 2013 and references therein). Purgatory Chamber is the cave’s lowest room, with hot and humid conditions that give it its name. The chamber is located at about 100 m from the cave’s entrance shaft and about 60 m below the entrance. It is one of the most difficult chambers of Dante Cave to access. The cave itself is accessible through a single vertical entrance in the roof of the southernmost chamber, which is 9 m above a cone of debris of collapsed rocks.
Photograph of Stalagmite DP1 lying in the Purgatory Room of Dante Cave, with Ben Hardt for scale (photograph by Eugene Marais).
Dante Cave is located in the Kalahari basin (Figure 1 of Thomas et al., 2005), on the northeastern margin of the Otavi Mountainland, also referred to as Namibia’s Karstveld (Irish et al., 1992), in the Otjozondjupa region of northeastern Namibia. The region is characterized by poorly developed underground karst in the carbonate succession of the Otavi Group. It developed in dolomites of the upper Tsumeb Subgroup, one of the two carbonate sequences deposited between 730 and 700 Ma in the Damara Supergroup Complex (Miller, 1983).
Dante Cave is in a semi-arid region, with a mean rainfall of ca. 500 mm/yr (Figure 1 of Sletten et al., 2013; Figure 2 of Rohde and Hoffman, 2012). Vegetation in the Otavi Mountainland is dominated by deciduous woodland savanna with a relatively high diversity of scattered tall trees (Mendelsohn et al., 2002; Rohde and Hoffman, 2012; Sletten et al., 2013). Rohde and Hoffman (2012) showed that, within the region, vegetation cover in areas with rainfall around 500 mm/yr is strongly dependent on changes in rainfall, a relationship further supported by the positive correlation between δ13C and δ18O in Stalagmite DP1 (Sletten et al., 2013).
Climate of northern Namibia
Northern Namibia, including Dante Cave, is in the austral summer rainfall zone of southern Africa and receives most of its monsoonal rainfall from October to May (Burrough et al., 2007; see also Figure 1 of Brook et al., 2015). Rainfall is linked to the annual migration of the ITCZ and the associated IOCZ, or Congo or Zaire Air Boundary (Van Heerden and Taljaard, 1998), a branch of the ITCZ where the Atlantic and Indian Ocean air-streams converge (Figure 1). Rainfall in northern Namibia is also affected by the location of the ABF, the boundary between the south-flowing warm Angola ocean current and the north-flowing cold Benguela current in the tropical SE Atlantic. For example, Nicholson and Entekhabi (1987) found that an anomalously warm SE tropical Atlantic current (with an SST > 28°C off northern Angola) leads to an increase in rainfall amount (above average) along the Angolan (6–17.5°S) and along the Namibian (17.5–29°S) coasts. This is also observed inland (Nicholson and Entekhabi, 1987), and it happens in late austral summer, from February to April (Rouault et al., 2003). A more southerly position of the ABF thus increases evaporation and the level of moisture associated with the IOCZ, bringing more rainfall to Angola and Namibia (Florenchie et al., 2003; Jury and Engert, 1999; Rouault, 2012; Rouault et al., 2003). This latitudinal migration of the ABF is expected to follow the latitudinal migration of the ITCZ, but this aspect has not been fully investigated.
Methods
This paper reports on the upper 261 mm of Stalagmite DP1, providing a record of much higher resolution than that of Sletten et al. (2013). Within that interval, Sletten et al. (2013) reported 3 U–Th ages and 66 stable isotope analysis. In contrast, this paper reports 18 U–Th ages and 264 stable isotope analyses.
Radiometric dating
A total of 18 samples (including 2 replicates) of 100–150 mg each were drilled for U–Th radiometric analyses to determine the depo-sitional age of the stalagmite (Table 1). Analyses were performed using an inductively coupled plasma mass spectrometry (ICP-MS) on a Finnigan-MAT Element at the University of Minnesota. Uranium and thorium were separated following the chemical procedures described in Shen et al. (2002). Decay constants used for 230Th, 234U, and 238U are similar to those reported in Cheng et al. (2000), which are 9.1577 × 10−6, 2.8263 × 10−6, and 1.55125 × 10−10 yr—1, respectively. Corrected 230Th assumes the initial 230Th/232Th ratio of 4.4 ± 2.2 × 10−6 as explained in Table 1. Dates are reported as ‘Yr BP’ where BP means ‘Before Present’ and ‘Present’ is AD 1950.
230Th dating results. The error is 2σ.
Sample not used in the age model because it has more 232Th relative to DP1-T, a sample collected from the same layer of aragonite of the top of the stalagmite.
Replicate samples, which have lower uncertainty in the value of 230Th (corrected age) relative to the former analyzed sample, are used to reconstruct the age model (see Figure 3).
δ234U = ([234U/238U]activity − 1) × 1000.
Corrected 230Th ages assume the initial 230Th/232Th atomic ratio of 4.4 ± 2.2 × 10−6. Those are the values or a material at secular equilibrium, with the bulk earth 232Th/238U value of 3.8. The errors are arbitrarily assumed to be 50%.
δ234Uinitial was calculated based on 230Th age (T), that is, δ234Uinitial = δ234Umeasured × eλ234×T.
BP stands for ‘Before Present’ where the ‘Present’ is defined as the year AD 1950.
Stable isotopes
A total of 264 samples of 50–100 µg were drilled along the growth axis of the top 261 mm of DP1 using a Micro-Cut dental drill with SSW HP1/4 (ref14820) drill bits in the Sedimentary Geochemistry Laboratory, Department of Geology, at the University of Georgia (UGA). The samples were transferred to round-bottomed 4.5 mL borosilicate vials (Labco Limited, Lampeter, UK) and were analyzed using the GasBench II technique in continuous flow IRMS. Analytical methods were similar to those described in Paul and Skrzypek (2007): (1) the vials were flushed with high-purity He to replace air contained in the vials, (2) about 0.06 mL of 100% ortho-phosphoric acid was injected into the vials to produce sample CO2 gases into the exetainer headspace, (3) the samples were allowed to react with the acid for a minimum of 2 h before the headspace was measured, and (4) the headspace CO2 was then analyzed for δ13C and δ18O (also see Skrzypek and Paul, 2006 for the procedures). Measurements were done on a Delta V Plus at 50°C, with NBS-19 interspersed among the samples (every six samples) to allow calibration relative to the Vienna Pee Dee Belemnite (VPDB) standard. All the analyses were performed at the Alabama Stable Isotope Laboratory of the University of Alabama. Isotope ratios were reported in per mil (‰) relative to VPDB, and isotope error was about ±0.1‰ for both 13C or 18O. As per convention, the 13C/12C and 18O/16O ratios (R) are expressed in the delta notation as follows: δx = [(Rsample − Rstandard)/Rstandard] * 103 (where x represents 13C or 18O and R is the ratio). For comparison with calcite values, δ18O and δ13C values from aragonite layers were adjusted by subtracting 1.7‰ (Romanek et al., 1992) and 0.8‰ (Kim et al., 2007), respectively, to account for the increased fractionation of heavier isotopes in aragonite.
Mineralogy and petrography
Microscopic examination of thin sections using the Leitz Laborlux 12 Pol of the Sedimentary Geochemistry Lab at UGA was done to characterize boundaries between adjacent spelean layers. These boundaries are classified into three categories (Figure 4b; also see Railsback et al., 2013): (1) conformal, when minerals grow without truncation; (2) erosional or Type E surfaces (Railsback et al., 2013), when truncation of the underlying and thus earlier mineral or layer is observed; and (3) Type L surfaces or boundary of lesser deposition (Railsback et al., 2013), which are easily identifiable as a thinning toward the margin of the stalagmites. Type L surfaces can also be identified in hand sample.
Each identified layer-bounding surface was marked on a scanned copy of the thin section (our log image), imported to Adobe Illustrator as a new layer overlain on top of the scanned image of Stalagmite DP1, digitized, and saved electronically (Figure 3b). The distance from the top of the stalagmite to where these surfaces were identified was then converted to age with respect to the age model.
(a) Age model of Stalagmite DP1 based on stalAge (Scholz and Hoffmann, 2011) and petrography (Railsback et al., 2013). ‘H’ indicates the hiatal Type E surface discussed in the text. (b) Scanned image of the uppermost 285 mm of Stalagmite DP1 showing the location of the age trenches (gray lines) and the identified layer-bounding surfaces (see Railsback et al., 2013 for details on these Type E and type L surfaces). A and C denote layers composed of aragonite and calcite. A* denotes the layer of coarse aragonite that has been analyzed under x-ray diffraction and scanning electron microscopy (SEM; see Figure S3b and S3d, available online). P09 and P11 represent locations of series of microphotographs (see Figure S3c, available online) used to illustrate the two main types of layer-bounding surfaces (E and L) identified in Stalagmite DP1.
Scanning Electron Microscopy (SEM) and X-ray diffraction (Figure 4) were performed to analyze a very unusual layer located between 237 and 250 mm from the top of Stalagmite DP1. For the SEM, a chip of the sample was carbon coated and then copper-taped and imaged at high vacuum using the Zeiss 1450EP (Carl Zeiss, Inc., Thornwood, NY) at Georgia Electron Microscopy, a facility at UGA. The Zeiss 1450EP is equipped with an Oxford SDD-EDX system (Oxford Inc., Scotts Valley, CA). For X-ray diffraction, powdered samples of 3 g were extracted and scanned from 20° to 65° 2θ with CoKα radiation using a Bruker D8 X-ray Diffractometer in the Department of Geology of the UGA.
Plot of δ18O of rainfall at Windhoek, Namibia (horizontal axis) plotted against amount of rainfall (left vertical axis) and atmospheric temperature (right vertical axis). Note that winter months have commonly no rainfall (Namibia Weather, 2016; International Atomic Energy Agency (IAEA)/WMO, 2014); thus, the δ18O values presented here are values for summer precipitation. Each filled circle represents 1 month between 1961 and 1987, with additional monthly data from 1999 to 2000. Additional symbols show means and midpoints for two arrays of data (precipitation-δ18O and temperature-δ18O). In blue, an amount effect (smaller δ18O with greater rainfall) seems evident from the value of r2 and from the position of the mean relative to the midpoint (see text for discussion). In red, a less pronounced temperature effect (greater δ18O with higher temperature) is suggested by the position of the mean of data in the direction of higher temperature and greater δ18O relative to the midpoint of the ranges of those parameters. Data are from IAEA/WMO (2014).
Results
Radiometric results and age model
Results from the 18 radiometric analyses for U–Th dating show that Stalagmite DP1 has high 238U concentrations that range from 962.6 ± 1.0 to 5774.0 ± 11.4 ppb and generally low 232Th concentrations that range from 34 ± 1 to 954 ± 24 ppb (Table 1). Three results were rejected. The first sample, DP1-U1, was rejected because of its large 232Th concentration (685 ppt), and it was not in stratigraphic order. It was then replaced by Sample DP1-T from a slightly higher horizon with a smaller 232Th concentration (209 ppt). The two other samples, DP1-U3 and DP1-U10, were rejected because of the large uncertainties. These samples were replicated and replaced by samples with more precise ages from the same horizons (Table 1). The remaining 15 ages were all in stratigraphic order, that is, becoming younger steadily toward the top of the stalagmite (Table 1).
The age model for Stalagmite DP1 was constructed using both the StalAge1.0 algorithm of Scholz and Hoffmann (2011) and recognition of likely hiatal surfaces as proposed by Railsback et al. (2013; see their Figure 9). The former generates a best estimate of a sample’s age at that sample’s specific depth in an uninterrupted sequence, but it is not well-suited to identification or location of hiatuses, whereas the latter specifically identifies hiatal surfaces and thus localizes offsets in age models. Thus, the kink in the StalAge model for Stalagmite DP1 was replaced with a hiatal offset, at a Type E surface (Figure 3).
The U–Th ages indicate that the interval of Stalagmite DP1 studied for this project was deposited from ca. AD 1450 to 1950 (Figure 3a). The modeling of U–Th ages using StalAge and petrographic evidence combines to suggest a hiatus of about 30 years at about 235 mm from the top of the stalagmite. The hiatal Type E surface is supported by an abrupt shift of ca. 4‰ in δ18O values, from −7.8‰ to −11.7‰ relative to VPDB, and by a major change in growth rate from 0.1 to 1.67 mm/yr (Figure 3a). The age of this hiatus is best estimated at ca. AD 1660–1690.
Stable isotopes
After the transformation of the isotopic values from aragonite discussed in the ‘Methods’ section, values of δ18O in Stalagmite DP1 range from −7.3‰ to −11.9‰ relative to the VPDB and values of δ13C range from −6.8‰ to −10.8‰, relative to the VPDB. The Pearson correlation test described in the supplementary document shows that δ18O and δ13C are correlated with great statistical significance (t = 14.02, n = 263, r2 = 0.43, p < 2.2e−16) and a positive linear relationship (Figure S1, available online).
Values of both δ18O and δ13C are significantly greater (between ca. −9‰ and −7‰, vs VPDB) in the time interval from AD 1450 to 1660 (Figure 5). An abrupt decrease in δ18O from −7.8‰ to −11.7‰ versus VPDB and in δ13C from −7.1‰ to −10.1‰ versus VPDB is observed around ca. 1680–1700. Following this change, δ18O shows a general trend toward greater values from ca. AD 1715 to ca. AD 1950 (Figure 5), when growth of Stalagmite DP1 ceased and did not resume prior to the stalagmite’s fall in 1988. Values of δ18O and δ13C are lowest in layers of calcite and highest in layers of aragonite. In addition, low values coincide with Type E surfaces and high values with Type L surfaces (Figure 5c).
(a) Time series of solar activity (Hathaway, 2010; Muscheler, 2007). (b) Paleotemperature reconstructions and simulations from several records of the two hemispheres (Neukom et al., 2014a). (c) The reconstruction of Northern Hemisphere (NH) temperature from Büntgen et al. (2011). The black line is the best estimate, and uncertainty is shown in red. (d) Mineralogical, stable isotope, and petrographic data from Stalagmite DP1. Vertical blue rectangles highlight significant wetting during solar minima (here the Maunder and the Dalton minimum) and intervals of colder NH temperatures relative to the Southern Hemisphere. These intervals coincide with wetter conditions in Dante Cave. The layer-bounding surfaces identified in Figure 3b are plotted as vertical lines.
Mineralogy and petrography
Mineralogy is distinct below and above the hiatus at 235 mm. Below the hiatus, the interval of Stalagmite DP1 studied consists solely of aragonite with different fabrics: (1) the common white, soft, opaque, and fibrous crystals of aragonite (Railsback, 2000) and (2) the uncommon pale-yellow, compact, translucent, and prismatic crystals of aragonite (Figure 3b). The uncommon crystals of aragonite, at 237–250 mm from the top of Stalagmite DP1, are much wider than the aragonite in the rest of the stalagmite (Figures S3a and S3b, available online). This aragonite is very similar to the ‘ray aragonite’ described by Frisia et al. (2002) and to the elongated columnar aragonite described by Duan et al. (2012). It has consistently high values of δ18O and δ13C (Figure 5). Above that hiatal surface, the interval of Stalagmite DP1 studied consists of alternating thick layers of calcite (8) and aragonite (8) with a mean thickness of ca. 15 mm (Figures 3b and 5). We did not find any evidence of post-depositional alteration of aragonite to calcite in Stalagmite DP1 (cf. Green et al., 2015).
For petrography, 26 Type E surfaces and 8 Type L surfaces were identified. Type E surfaces are surfaces below which layers are truncated, particularly toward the crest of the stalagmite, in wetter conditions. Type L surfaces are surfaces below which layers thin toward the flank and thus indicate reduced deposition rate in response to lesser dripwater in drier conditions (Railsback et al., 2013, 2014).
Discussion
Interpretation of Stalagmite DP1’s multiple proxies
Oxygen isotopes
The δ18O of stalagmite CaCO3 reflects the δ18O of the dripwater and the temperature in the cave (Hoefs, 1997). Temperature inside a cave usually represents the mean annual atmospheric temperature (McDermott, 2004). Thus, its effect is minor on speleothem δ18O compared with the effect of water. However, long-term changes in atmospheric temperature could be expected to leave signals in speleothem δ18O (see, for example, the trend in Figure 5).
In contrast, dripwater δ18O is determined by δ18O of atmospheric precipitation and its subsequent evaporation, both during precipitation events (Lee et al., 2012) and thereafter (Cuthbert et al., 2014). The δ18O of atmospheric precipitation is in turn controlled by the δ18O of the vapor source, the distance of transport from the source (the ‘continent effect’ resulting from Rayleigh distillation), the amount of precipitation (the ‘amount effect’), and the atmospheric temperature during precipitation (the ‘temperature effect’) (Dansgaard, 1964; Kurita et al., 2009; Lachniet, 2009; McDermott, 2004; Rozanski et al., 1993).
In Namibia, precipitation occurs almost exclusively in summer (Namibia Weather, 2016), and δ18O is influenced by changes in amount of summer rainfall, with more negative values in wetter months and more positive values in drier months (Figure 4). This relationship between δ18O and monthly rainfall in Namibia appears very similar to the amount effect defined by Dansgaard (1964: 445). However, Dansgaard (1964: 445) attributed the amount effect to differing degrees of cooling in the atmosphere, to differing amounts of falling water in the atmosphere, and to evaporation from falling raindrops (as was documented by Lee et al., 2012). Namibia’s semi-arid climate suggests that evaporation from falling raindrops may indeed be important in determining δ18Ow of rainfall. However, the rainfall-δ18Ow relationship in Namibia may be additionally and perhaps better ascribed to varying latitudinal migration of the ITCZ during austral summers. That is because monsoonal air is moister, lessening evaporation from falling raindrops during each rain event while allowing more rain events and thus more rainfall in months when the ITCZ is present.
In addition to evaporation from falling raindrops (Lee et al., 2012), evaporation outside or within the cave additionally leads to higher δ18Ow values of soil water and dripwater and thus ultimately to higher δ18Oc values of spelean CaCO3 (e.g. Cuthbert et al., 2014; Deininger et al., 2012; Fleitmann et al., 2007).
Carbon isotopes
The δ13C of stalagmite carbonate is determined by the δ13C of the total dissolved inorganic carbon of the dripwater. This variable is in turn controlled by a number of factors (e.g. Brook et al., 2006; Quade, 2004; Wong and Breecker, 2015). First is the δ13C of atmospheric CO2. It is known to have changed because of the Suess effect (Suess, 1955; Verburg, 2007) wherein burning of fossil fuels has added 13C-poor CO2 to the atmosphere. For some comparisons, we have used the equation of Verburg (2007) to correct our δ13C data from Stalagmite DP1 for the Suess effect, but the change is small compared with the variability of δ13C in the stalagmite (Figure S2, available online). Second is the biological input through roots and/or decay of plant matter in soil. High input of soil CO2 could decrease the speleothem δ13C values during wetter and warmer seasons (e.g. Baldini et al., 2005; Hesterberg and Siegenthaler, 1991; Lauritzen and Lundberg, 1999). Third is the flow conditions above the cave that could decrease (open system) or increase (closed system) the δ13C values (e.g. Clark and Fritz, 1997; Fairchild and Baker, 2012). An open system reflects a longer residence time of the percolating water, allowing it to exchange CO2 with the soil. A closed system reflects a shorter residence time of the percolating water before reaching the cave, and this is often the case with less vegetation cover and/or with less soil bioproductivity (e.g. Brook et al., 2006, 2015). Fourth is degassing. After passage of water through the soil and bedrock, degassing to the cave atmosphere occurs because of the difference in PCO2 between the cave dripwater and the PCO2 of the cave atmosphere, and the result is an increase in δ13C (Dreybrodt and Scholz, 2011). Degassing depends on the soil biological activity (e.g. Dreybrodt and Scholz, 2011), but it can also depend on cave ventilation (e.g. James et al., 2015). The fifth and last factor is ‘prior calcite precipitation’ or more broadly ‘prior carbonate precipitation’ (PCP). It can result in an increase in speleothem δ13C in response to Rayleigh-type enrichment in 13C with progressive degassing (e.g. Mickler et al., 2004; Oster et al., 2010) prior to calcite deposition on the speleothem. The effect of PCP is also more likely during drier times when saturation states are higher and cave PCO2 may be lower (e.g. Cross et al., 2015; Johnson et al., 2006).
The dependence of soil biological activity, extent of degassing, and PCP on wetness and thus rainfall combine to suggest that greater δ13C of stalagmite carbonate is typically an indicator of drier conditions (e.g. Cross et al., 2015; Johnson et al., 2006). The linkage of δ13C of Stalagmite DP1 to rainfall is further supported by the historical relationship between vegetation distribution and spatial distribution of rainfall in inland Namibia (Rohde and Hoffman, 2012).
Mineralogy and petrography
The deposition of spelean aragonite rather than calcite is largely affected by temperature (Moore, 1956; Railsback et al., 1994) and trace or minor element composition of the dripwater (Caddeo et al., 2011; Fischbeck, 1976; González and Lohmann, 1988; Railsback et al., 1994; Riechelmann et al., 2014), which can in turn be a function of bedrock lithology and PCP (Riechelmann et al., 2014). However, because drier conditions lead to higher saturation states and (concomitantly) more extensive PCP, deposition of aragonite is most commonly linked to drier conditions (Cabrol and Coudray, 1982; Frisia et al., 2002; Murray, 1954; Pobeguin, 1965; Railsback et al., 1994; Siegel, 1965; Thrailkill, 1971). In these regards, the overall higher δ18O and δ13C values of aragonite layers of Stalagmite DP1 suggest dryness, and the overall lower δ18O and δ13C values of calcite layers suggest wetness.
For petrography, Type E and Type L surfaces in Stalagmite DP1 are characterized by low and high δ18O, respectively (Figure 5). This agrees with the findings of Railsback et al. (2013), suggesting that Type E surfaces are indicative of exceptionally wet conditions and Type L surfaces are indicative of exceptionally dry conditions.
Summary
The four paleoenvironmental proxies reported from Stalagmite DP1 studied here agree with the generalization of Fairchild and McMillan (2007) that speleothems are indicators of wet and dry periods. Lower δ18O and δ13C values, calcite, and Type E surfaces combine to suggest wetter conditions. Higher δ18O and δ13C values, aragonite, and Type L surfaces combine to suggest drier conditions (Figure 5).
Paleoclimate significance inferred from Stalagmite DP1
The hiatus of ca. 1660–1690
One of the most striking features of the DP1 record is the hiatus of about 30 years in the late 1600s (Figure 3). That hiatus is required by an abrupt change in U–Th ages of 46 years in the interval from 238 to 234 mm from the top of the stalagmite, which is by far the most abrupt change of age in the series of U–Th ages from DP1. It coincides with a Type E surface that resulted from dissolutional corrosion, suggestive of exceptional wet conditions (Railsback et al., 2013), and it coincides with an abrupt shift of both δ18O and δ13C to more negative values, likewise suggestive of a transition to wetter conditions, as discussed above. We therefore conclude that the most parsimonious interpretation of the hiatus ca. 1660–1690 is one of dissolution and non-deposition during an exceptionally wet period.
Wetter conditions in northeastern Namibia during low solar activity and cold intervals in NH
The records from Stalagmite DP1 suggest that wet intervals in northeastern Namibia are linked to low solar activity and cold intervals in NH (Figure 5). At ca. AD 1660–1710, the decrease in both δ18O and δ13C, the presence of Type E surface at ca. 237 mm from the top of Stalagmite DP1, and the abrupt change in growth rate (from 0.1 to 1.67 mm/yr) combine to suggest wetter conditions. This interval coincides with the Maunder minimum (Figure 5a) and a decreased temperature (specifically, colder NH relative to SH) in the global reconstructions (Neukom et al., 2014a; Figure 5b and c). At ca. AD 1790–1830, the decrease in δ18O and δ13C and the presence of a calcite layer bounded by Type E surfaces suggest another wetter period in northeastern Namibia. This younger interval coincides with the Dalton solar minimum and with the coldest period in the NH temperature reconstructions by Büntgen et al. (2011) and Neukom et al. (2014a) (Figure 5).
Long-term trends in δ18O in DP1 from AD 1750 to 1950
Values of δ18O in Stalagmite DP1 show an increasing trend from ca. AD 1750 to 1950 (r2 = 0.82). The increase starts from ca. −11.8‰ to −8.5‰ (Figure 5). In contrast, there is no significant trend in δ13C (r2 = 0.02), even if those δ13C data were transformed to account for the Suess effect using the equation of Verburg (2007) (Figure S2, available online).
The δ18O trend could be ascribed to a more northward migration of the ITCZ, when the NH is warmer (Figure 5; Neukom et al., 2014a), so that Namibia went from being on the fringes of the Tropics in the 18th century to being more arid and subtropical. This new climatic condition in turn leads to reduced precipitation and drier austral summers, combining to increase the raindrop re-evaporation effect (Lee et al., 2012) and the extent of evaporation from the land surface. The coeval increase in global temperature and increase in Stalagmite DP1 δ18O after ca. 1715 (Figure 5) suggest that although instrumental data show a weak relationship between δ18O and temperature (Figure 4) (IAEA/WMO, 2014), northeastern Namibia not only became drier but also could have become warmer.
Regional comparison
The wet interval observed between ca. AD 1660 and 1710 in northeastern Namibia is not a local event. Similar climatic response is observed in Cold Air Cave, South Africa (Holmgren et al., 1999; Sundqvist et al., 2013) (Figure 6). Although these authors mainly use δ18O as a proxy for temperature reconstruction, their stable isotope records from two stalagmites, T5 and T7, from Cold Air Cave around AD 1650–1750 (Figure 4 of Holmgren et al., 1999, also see Sundqvist et al., 2013) could also suggest wetter conditions. These wet intervals commonly coincide with the Maunder solar minimum and with periods of globally cold climate, with colder NH relative to the SH (Figures 5 and 6). They collectively fall in the period of coldest temperatures in several locations around the world from AD 1570 to 1730 (Bradley and Jones, 1993). Wetter conditions in northeastern Namibia and in South Africa also coincide with the most extreme minimum in the time-averaged record of English instrumental records of temperature (AD 1690–1699) (Manley, 1974), with AD 1684, the coldest winter in English instrumental records of temperature (Manley, 2011).
(a) Plot of solar activity (Hathaway, 2010). (b) Stable isotopes of oxygen profile of Stalagmite DP1 from Dante Cave (this study). (c) Stable isotopes of oxygen profile of Stalagmites T5 and T7 from Cold Air Cave (Holmgren et al., 1999; Sundqvist et al., 2013). The two dashed vertical lines represent the lower and upper limits of the Maunder minimum. The shaded area represents the estimated wetter interval in Dante and Cold Air Cave, with darker shade highlighting very wet interval within the Maunder sunspot minimum according to Stalagmite DP1, T5 and T7 records. Note that δ18O in Dante Cave is more depleted than δ18O in Cold Air Cave.
Global implications
Latitudinal migration of the ITCZ linked to solar activity and global conditions
Hydroclimate response in southern Africa to the latitudinal migration of the ITCZ is revealed in the paleorecords between AD 1400 and 1950. The wet intervals inferred from Stalagmite DP1 ca. AD 1660–1690 and ca. AD 1790–1830 in northeastern Namibia and the wet interval inferred from Stalagmite T5 and T7 from South Africa around AD 1700 coincides with solar minima and globally colder conditions, specifically a much cooler NH (Figure 5). This ‘solar activity–global temperature–rainfall’ relationship is very similar to the findings of Sachs et al. (2009) in the Pacific Ocean between AD 1400 and 1850 during the LIA. Sachs et al. (2009) concluded that lesser solar irradiance, which could have caused the colder conditions in the NH (e.g. Mauquoy et al., 2002; Shindell et al., 2003), may have driven the ITCZ south. Thus, the wet interval in Namibia could be the result of a southward migration of the ITCZ (Arbuszewski et al., 2013; Broccoli et al., 2006; Chiang and Bitz, 2005; Chiang and Friedman, 2012; Chiang et al., 2003; Donohoe et al., 2013; Haug et al., 2001; Koutavas and Lynch-Stieglitz, 2004; Peterson et al., 2000; Sachs et al., 2009; Schefuß et al., 2011; Sinha et al., 2011).
In contrast, the drying trend since ca. AD 1750 inferred from Stalagmite DP1 in northeastern Namibia coincides with an overall higher solar activity and globally warmer conditions, when NH is warmer (Figure 5). This condition favors a northward migration of the ITCZ toward Earth’s warmer hemisphere (Frierson and Hwang, 2012; Kang et al., 2008; McGee et al., 2014; Sachs et al., 2009).
The DP1 record suggests that specific sunspot minima (i.e. Maunder and Dalton) associated with a significant temperature gradient between the two hemispheres (i.e. cooler NH than SH) could have intensified rainy seasons in northeastern Namibia, parallel with a southward migration of the ITCZ. This southward migration of the ITCZ could have also affected the IOCZ. As a result, summer seasons in northeastern Namibia must have been wetter and could have lasted longer than they are today. In contrast, warming of the NH since the end of the LIA (Figure 5b) has driven the ITCZ northward, leading to shorter wet summer seasons and longer dry winter seasons in northeastern Namibia.
Estimates of the latitudinal shift of the ITCZ with NH cooling vary considerably, but they could be used to estimate the change in rainfall in northeastern Namibia during those extended wet periods. McGee et al. (2014) suggested a shift of at most 1° of latitude during the Last Glacial Maximum, whereas Sachs et al. (2009) inferred a shift during the LIA of as much as 500 km or about 4.4° of latitude. In northern Namibia and southern Angola, rainfall increases northward at a rate of about 100 mm/yr per degree of latitude. In reference to this latitudinal variability of rainfall in Namibia, the estimates of McGee et al. (2014) and Sachs et al. (2009) would suggest increases of 100–440 mm/yr of rainfall respectively, which would be significant compared with the present rainfall of about 500 mm/yr at Dante Cave.
Possible movement of the ABF accounting for wetter conditions in northern Namibia
The widely accepted notion of southward shifts of the ITCZ with cooling of the NH (Arbuszewski et al., 2013; Broccoli et al., 2006; Chiang and Bitz, 2005; Chiang and Friedman, 2012; Chiang et al., 2003; Donohoe et al., 2013; Haug et al., 2001; Koutavas and Lynch-Stieglitz, 2004; Peterson et al., 2000; Sachs et al., 2009; Schefuß et al., 2011; Sinha et al., 2011) is understood as an atmospheric phenomenon: a southward shift of the convergence of the easterly trade winds. However, because of the significance of those easterly winds to low-latitude ocean circulation (e.g. Knauss, 1963), it seems likely that a southward shift of the ITCZ could lead to some southward shift of low-latitude currents. For example, with regard to the eastern tropical Atlantic, Li and Philander (1997) concluded that sea-surface temperature there varies seasonally as a ‘passive response of the ocean to the changes in the winds’ and thus to seasonal change of the ITCZ. If that thought extends to the Angola Current, the coastal down-current expression of the Atlantic’s Southern Equatorial Countercurrent (Figure 1), it suggests that the southern limit of the Angola Current at the ABF may have shifted southward during southward shifts of the ITCZ (as it does seasonally). The southern limit of the Angola Current is relevant to rainfall in northern Namibia because southward advances of that warm current promote offshore evaporation that leads to more rainfall inland (Rouault et al., 2003), and the position of the ABF has been shown to vary interannually as well as seasonally (Colberg and Reason, 2007). Thus, the linkage between wet intervals in northeastern Namibia and the southward movement of the ITCZ inferred above, combined with the possible linkage of the southern limit of the Angola Current to the position of the ITCZ, suggests that rainfall in northeastern Namibia may have been linked to changes in ocean currents resulting in an SST anomaly that favored rainfall on the continent.
The isotopic shift at ca. 1680: Change in cave hydrology
The significant change in growth rate (from 0.1 to 1.67 mm/yr), the difference in isotopic values before and after ca. 1680, and the significant decrease in stable isotope values (of ca. 4‰ and 3‰ for δ18O and δ13C, respectively) around ca. 1680 could represent a significant change in the cave hydrology linked to ITCZ and the ABF in northeastern Namibia. Climate in northeastern Namibia was drier prior to ca. AD 1680 as suggested by the stable isotope values and mineralogy (Figure 5), and it became wetter during the Maunder minimum. After the wet Maunder minimum, rainfall has receded. In this scenario, the alternating long-term and pronounced wet and dry periods agree well with the north–south migration of the ITCZ.
A different point of view: Possible changes in moisture flux before and after ca. 1680
The difference in isotopic signals in Stalagmite DP1 before and after ca. 1680 could also be linked to changes in moisture flux from the primary western Indian Ocean source and the secondary tropical SE Atlantic source that converge over northern Angola and southern Congo (Rouault et al., 2003). Moisture fluxes examination by Rouault et al. (2003) suggests that the size of rainfall anomalies in western Angola/Namibia is influenced by the local circulation anomaly imposed on the mean easterly from the western Indian Ocean. This local circulation anomaly seems to originate in tropical SE Atlantic (Rouault et al., 2003). Although the δ18O of the Indian Ocean and the Atlantic water only differs by ca. 2–3‰, the greater speleothem δ18O values prior to ca. 1680 could suggest a dominance of moisture flux from a relatively near marine source, presumably the nearby Atlantic Ocean, which is today the source of 18O-rich rainfall (δ18Op = ~−2‰ to −3‰; Bowen, 2013) to the Namib Desert along Namibia’s Atlantic Coast, west of the IOCZ. This condition could be similar to the 1995 event, when SST anomaly in tropical SE Atlantic was largest but the inflow into Angola/Namibia from the Indian Ocean was weaker (Rouault et al., 2003). According to Jury and Engert (1999), this climatic condition is similar to an Atlantic El Niño, when summer rainfall over northern Namibia was 50% below normal. In contrast to a southward push of the ABF, climatic conditions in northeastern Namibia prior to ca. 1680 could have been linked to a northward push of the ABF by the cold Benguela current, resulting in much drier conditions over the region.
This condition immediately shifted at ca. 1660 with a southward push of the ITCZ and the ABF, as discussed above. At the same time, easterly flux from the Indian Ocean could have been enhanced. This new condition is similar to the 1984 and 1986 events discussed in Rouault et al. (2003). Thus, the much lower δ18Oc values after ca. 1680 may have resulted from transport from a more distant source, presumably the Indian Ocean, which is today the dominant source for more 18O-depleted (δ18Op = ~−4‰ to −5‰; Bowen, 2013) rainfall to eastern Namibia, east of the IOCZ (Figure 1). This dominance of moisture flux from the Indian Ocean is supported by the more enriched values of speleothem δ18O in Cold Air Cave, South Africa (Holmgren et al., 1999; Sundqvist et al., 2013), and the more depleted speleothem δ18O in Dante Cave, Namibia (Figure 6).
If this point of view on moisture flux is correct, it implies that hydrology over northeastern Namibia around the Maunder minimum was controlled by not only a southward shift in ITCZ and ABF but also a strengthening of the easterly flux from the Indian Ocean.
Conclusion
Examination of the uppermost 261 mm of Stalagmite DP1 suggests that climate in northeastern Namibia responded to changes in solar activity and changes in global temperature. The record shows that not only the LIA in general, but specific sunspot minima, that is, Maunder and Dalton, associated with a significant temperature gradient between the two hemispheres (i.e. cooler NH than SH), have driven the ITCZ and the IOCZ southwestward and also have driven the ABF southward. As a result, wet summer seasons in northeastern Namibia could have lasted longer than they are today. In contrast, warming of the NH since the end of the LIA, estimated around AD 1715 in northeastern Namibia, has driven the ITCZ northeastward, leading to shorter wet summer seasons in favor of dry winter seasons in northeastern Namibia.
The practical significance of these conclusions lies in their implications for future changes in rainfall in northeastern Namibia and more generally in the summer rainfall zone of southern Africa (e.g. Niang et al., 2014; Thomas et al., 2005). Warming of climate in the coming century is predicted for Earth in general and especially for high-latitude regions of the NH (IPCC, 2013, Figure SPM.8). The correlation of past dry periods in northeastern Namibia with warmer conditions in the NH (both reported in this paper and in Sletten et al., 2013), and the inferred northward shift of the ITCZ toward a warmed NH, suggests that warming will lead to less rainfall in the future in a region with a presently semi-arid climate and will cause dune reactivation by 2099 (Thomas et al., 2005).
Footnotes
Acknowledgements
Collection of Stalagmite DP1 was carried out with the permission of the Namibian Ministries of Environment and Tourism and Mining and Energy as well as that of Jurg-Reiner and Renate Otto. The X-ray diffraction analysis was conducted at the XRD lab of the Department of Geology under the supervision of Paul A Schroeder.
Funding
Funding was provided by the National Sciences Foundation Grant NSF 0002193 to George A Brook and by Natural Sciences Foundation of China Grant NSFC 41230524 to Hai Cheng.