Dating Saharan dust deposits on Lanzarote (Canary Islands) by luminescence dating techniques and their implication for palaeoclimate reconstruction of NW Africa.

This article is published in: Geochemistry, Geophysics, and Geosystems (2008) Volume 9, Number 2, 14 February 2008, Q02Q07 doi:10.1029/2007GC001658 ISSN: 1525-2027 Dating Saharan dust deposits on Lanzarote (Canary Islands) by luminescence dating techniques and their implication for palaeoclimate reconstruction of NW Africa H. von Suchodoletz*, M. Fuchs, and L. Zöller Department of Geomorphology, University of Bayreuth, Universitätsstrasse 30, D-95440 Bayreuth, Germany (*[email protected]) Lava flow dammed valleys (Vegas) on Lanzarote (Canary Islands) represent unique sediment traps, filled with autochthonous volcanic material and allochthonous Saharan dust. These sediments and the intercalated palaeosoil sediments document past environmental change of the last glacial-interglacial cycles, both on Lanzarote and in NW Africa. A reliable chronology must be established to use these sediment archives for palaeoclimate reconstructions. Owing to the lack of organic material and the limiting time range of the 14Cdating method, luminescence dating is the most promising method for these sediments. However, the fluvio-eolian character of these sediments is a major problem for luminescence dating, because these sediments are prone to insufficient resetting of the parent luminescence signal (bleaching) prior to sedimentation. To check for the best age estimates, we compare the bleaching behavior of (1) different grain sizes (coarseversus fine-grain quartz OSL) and (2) different minerals (fine-grain feldspar IRSL versus fine-grain quartz OSL). The results show that owing to its bleaching characteristics, quartz is the preferable mineral for luminescence dating. On the basis of the fineand coarse-grain quartz OSL age estimates, a chronostratigraphy up to 100 ka could be established. Beyond this age limit for OSL quartz, the chronostratigraphy could be extended up to 180 ka by correlating the vega sediments with dated marine sediment archives. Keywords: luminescence dating; Saharan dust; Canary Islands; insufficient bleaching; land/sea correlation; human occupation. 1. Introduction Lava flow dammed valleys (Vegas) on Lanzarote (Canary Islands) represent sediment traps, filled with autochthonous volcanic material and allochthonous Saharan dust. These sediments are regarded as unique terrestrial archives for the Quaternary paleoclimate reconstruction of northwest Africa (H. von Suchodoletz et al., Geomorphological investigations of sediment traps on Lanzarote (Canary Islands) as a key for the interpretation of a palaeoclimate archive off NWAfrica, submitted to Quaternary International, 2008) (hereinafter referred to as von Suchodoletz et al., submitted manuscript, 2008a). There have been many attempts to reconstruct the paleoclimate for this region. Marine records that may extend back to the Tertiary [e.g., Dupont, 1993; Brunner and Maniscalco, 1998; Moreno et al., 2001] have the advantage that they provide continuous records with generally well-preserved proxies. In contrast, terrestrial archives are commonly discontinuous and represent only short periods, which makes their paleoclimate interpretation difficult [e.g., Gasse et al., 1987; Cheddadi et al., 1998; Lancaster et al., 2002]. However, despite the limitation of interpretation and hiatuses, terrestrial archives are essential for investigating paleoclimate on land, especially the regional patterns and the land-sea interaction. On the Eastern Canary Islands, various terrestrial archives have been used to reconstruct paleoclimate, for example, dune sequences and their intercalated paleosols [e.g., Rognon et al., 1989; Criado et al., 2004; Ortiz et al., 2006], calcretes [e.g., Alonso-Zarza and Silva, 2002], cave sediments [e.g., Coello et al., 1999] and marine terraces [e.g., Meco et al., 2002]. However, these archives show hiatuses and represent limited periods of the Quaternary. Our studies from Lanzarote, where thick Saharan dust had accumulated in volcanic valleys, reveal a generally continuous record, reaching from the Holocene to the Middle Pleistocene (Zo¨ller et al. [2003] and present study), which makes these sediments an outstanding terrestrial archive for a paleoenvironmental reconstruction of Lanzarote and NW Africa. For paleoclimatic interpretation of these archives a sound chronostratigraphy is of crucial importance. Previous studies on the Eastern Canary Islands used the 14C-dating method [e.g., Petit-Maire et al., 1986] as well as U-Th ages [e.g., Hillaire-Marcel et al., 1995], but the results of both methods suffer from the fact that the ages were determined on land snail shells which may represent open systems. Therefore these dates must be regarded as provisional [Edwards and Meco, 2000]. The second shortfall of both methods is that they do not directly date the sedimentation process, a drawback that does not apply to luminescence dating. Sporadic studies from the Eastern Canary Islands using luminescence dating were presented by Pomel et al. [1985] and Bouab and Lamothe [1997]. Since our initial dating results using infrared stimulated luminescence (IRSL) were promising [Zöller et al., 2003], and in most parts of the investigated profiles there are neither organic matter nor land snail shells, we continued to built up our chronostratigraphy using luminescence dating. Luminescence dating is based on the fact that quartz and feldspar grains accumulate energy in their crystal lattice after burial, that increases in response to ionizing radiation to which the grains are exposed. This signal is discharged when the grains are exposed to sunlight. Measurement of this signal results in the equivalent dose (De). To determine the time that elapsed since the last bleaching, the De is divided by the ionizing radiation (dose rate, . D) of the dated sediment [Singhvi and Krbetschek, 1996; Aitken, 1998; Wintle, 1998]. The advantage of luminescence dating is that it directly dates the last exposure to sunlight, but good optical bleaching prior to burial is of major importance. Pure eolian sediments such as loess show very good bleaching prior to sedimentation, and have been dated successfully for many years [e.g., Berger et al., 1992; Rousseau et al., 1998; Zöller et al., 2004]. It is only recently that the development of coarse-grain dating techniques using optically stimulated luminescence (OSL) facilitated the dating of poorly bleached materials, for example, fluvial, colluvial or playa sediments [e.g., Olley et al., 1998; Fuchs and Wagner, 2003; Bubenzer and Hilgers, 2003]. Being aware of the largely colluvial character of our sequences (von Suchodoletz et al., submitted manuscript, 2008), we have compared three different luminescence dating techniques (IRSL of fine-grained feldspars and OSL of fine and coarsegrained quartz) with different bleaching properties to build up a reliable chronostratigraphy for our archives. This methodological comparison should help to evaluate the reliability of the ‘‘problematic’’ IRSL dates in the older parts of the profiles, where they are the only numerical dating method available. 2. Study Area Lanzarote (28°50’N-29°13’N, 13°25’W-13°52’W) belongs to the Canary Islands/Spain and is situated ~130 km west of the southern Moroccan coast (Figure 1). It is a volcanic island, composed of basic to ultrabasic volcanic rocks. Volcanism started around 15.5 Ma ago and has persisted to recent historical times [e.g., Carracedo et al., 1998]. Owing to its limited altitude (maximum 670 m amsl), Lanzarote does not receive precipitation from rising trade winds. It receives 100 mm (at lower elevation) to 250 mm (at higher elevation) precipitation from boreal winter cyclones. Mean annual temperature is 19.9°C [Jahn, 1988]. Lanzarote is characterized by a very sparse, shrubby and disperse vegetation dominated by xerophytic and halophytic species manifesting a semi desert character. The island is situated at the northern fringe of the Saharan dust plume over the North Atlantic Ocean. Dust is brought to the island during two different synoptic situations: During the first situation, dust is entrained by so called ‘‘Calima’’ winds: low-level continental African Trade winds (Harmattan) deflected toward the west by Atlantic cyclones especially during boreal winter [Criado and Dorta, 2003]. The second situation occurs exclusively during boreal summer, when dust is advected by the northern branch of the high-altitude Saharan Air Layer. Subsequently, the material sinks into the lower atmosphere north of the Canary Islands and is finally transported toward the islands via the Northeast Trade wind [Koopmann, 1981; Bozzano et al., 2002]. Dust can be deposited here during both dry and wet conditions [Criado and Dorta, 2003; Menéndez et al., 2007]. 3. Studied Sites On Lanzarote, paleo-valleys of Miocence to Pliocene age were dammed during the Early and Middle Pleistocene by lava flows and pyroclastic material. Locally these valleys are called vegas. In our studies we used the outcrops situated in loam pits excavated during the past decades in the vegas of Feme´s, Guatiza and Teguise (Valle de San José) (Figure 1). All studied profiles were chosen in a distal position from geomorphologically active slopes. Hence we chose profile Guatiza III instead of Guatiza I and II [Zöller et al., 2003] for further investigations. Two vegas (Guatiza and Teguise) are situated on the fringes of the Famara Massif in the north of the island at a distance of ~6 km from each other. Damming of these valleys was never total, so hiatae can be expected. The valley of Femeés is located in the Los Ajaches Massif in the south of Lanzarote about 30 km from the northern sites, and damming was complete until today [Zöller et al., 2006]. The stratigraphies can be seen in Figure 2. The sediments are composed of interbedded in situ and reworked colluvial material originating from allochthonous Saharan dust and autochthonous volcanic products. Parts of these sediments were subject to pedogenesis and are consequently characterized by reddish color and a vertisol texture, whereas other layers are only slightly weathered and show a yellowish color and a silty granulometry. Some beds were overprinted by carbonatic waters and thus contain carbonate nodules, calcified root channels, or are fully developed as calcrete horizons. Many layers were overprinted by water containing Fe and Mn. Consequently, Fe and Mn stains and concretions occur, the largest in the vega of Teguise. In some horizons, calcified nests of anthophora bees are common. Whereas no coarse colluvial material is found in the lower parts of the sequences, the upper parts consist of anthropogenic colluvium containing pebbles and partly ovicaprid bones. In Femés, coarse volcanic fallout is observed between ~220 and 400 cm. The stratigraphy of this profile is slightly revised compared to that published by Zöller et al. [2003]. Consequently, in the case of differing depths both are indicated in Table 4 in section 4.3 and Table 5 in section 5.2. The properties of the sequences are listed in Table 1. These sedimentary units record the landscape and paleoclimatic history from the Lower Pleistocene to the Holocene. The geomorphic system and its sedimentation processes are described in detail by von Suchodoletz et al. (submitted manuscript, 2008a). 4. Methods For luminescence dating sampling took place during the night, after removing the outer 30 cm of the profile to avoid any contamination with lightexposed material. Sampling and sample preparation at the Bayreuth luminescence laboratory (University of Bayreuth/Germany) was done under subdued red light (wavelength 640 ± 20 nm). 4.1. Sample Preparation 4.1.1. Coarse-Grain Quartz Samples After sieving, the fraction 63-200 mm was treated with HCl and H2O2 to destroy carbonate and organic matter, and was held in an ultrasonic bath for 30 min to destroy aggregates. Heavy minerals (density > 2.75 g/cm3) and feldspars (density < 2.62 g/cm3) were separated in a lithium heteropolytungstate solution (LST). The quartz material was subsequently etched in 40% HF for 45 min in order to remove any remaining feldspar as well as the alpha irradiated outer layer of the quartz grains. The resultant material was fixed on aluminum cups (diameter 12 mm) using silicone oil. The number of grains per cup was between 200 and 600, representing small aliquots according to Fuchs and Wagner [2003]. 4.1.2. Fine-Grain Quartz Samples To remove carbonates and organic carbon, the fraction <63 mm was treated with HCl and H2O2. Subsequent addition of 0.05 M sodium pyrophosphate and treatment in an ultrasonic bath for 30 min dispersed clays and destroyed grain aggregates. Extraction of the 4–11 mm fraction was done in Atterberg-settling tubes up to 50 times. The separated polymineral fraction 4–11 mm was etched for 4-5 days in pretreated 34% hexafluorosilicic acid to remove any remaining feldspars (for methodology, see Fuchs et al. [2005]). After a short IRSL test on two aliquots that included radiation (50 Gy) and IRSL measurement to identify any possible remaining feldspar contamination, the material was pipetted on to 9.6-mm aluminum discs. 4.1.3. Fine-Grain Polymineral Samples The polymineral fine silt fraction was extracted as described above for fine-grain quartz samples and was directly pipetted on to aluminum discs (diameter 9.6 mm). Each disc had about 1.6 mg of silt. 4.2. Measurements For measurements we used two Risø-Readers TL/OSL-DA-15 combined with a Thorn-EMI 9235QA photomultiplier [Bøtter-Jensen et al., 1999]. Sources used for radiation are listed in Table 2. The software Analyst 3.07b was used for analysis of the data. 4.2.1. Fineand Coarse-Grain Quartz OSL Measurement parameters of fineand coarsegrain quartz samples are listed in Table 3. Equivalent doses (De) were determined using the single aliquot regenerative dose protocol (SAR) [see Murray and Wintle, 2000]. A saturating exponential growth curve was constructed, using six regeneration cycles after one measurement of the natural OSL: four measurements of regenerated doses, one repeated measurement of the first regenerated dose to determine the recycling ratio and one 0-dose regeneration cycle. The De was determined by subtracting the background from the used OSL integral. A check for possible feldspar contamination was done by stimulating the artificially irradiated samples with infrared and detecting in the blue range (390-450 nm). Depending on the availability of the prepared mineral fraction, we tried to measure a minimum of 30 aliquots from the coarsegrained fraction, and 6–7 aliquots from the finegrained samples. In individual cases, the number of measured aliquots was below 30 for coarse-grained samples. To determine a-values of the fine-grain samples, a-irradiation was used. 4.2.2. Polymineral IRSL Measurement parameters for polymineral IRSL samples are listed in Table 3. For De determination, the multiple aliquot additive dose protocol (MAAD) following Lang et al. [1996] was routinely used. After measurement of three natural and three artificially irradiated discs (test dose of 50 Gy) to obtain a rough approximation of the equivalent dose, six dose groups with five discs each were irradiated and stored for 1 month at room temperature. Subsequently, they were measured together with nine natural discs to construct the additive growth curve. The a-value of some representative samples was determined using an a-growth curve with three dose groups of three aliquots each. Owing to limited laboratory capacity, we did not conduct tests for anomalous fading after Auclair et al. [2003], but used instead a simpler protocol following Lang et al. [1996]. We irradiated five discs with the highest irradiation dose, stored them for 3 months at room temperature before measurement and compared them to the measurements done after 1 month of storage. 4.3. Dose Rate Determination The dose rate ( . D) is the energy that accumulates in a mineral and thus creates the luminescence signal. The dose rate is composed of the natural radioactivity (α-, β-, γradiation) and the cosmic radiation affecting the sample. Since the outer rim of coarse grains are etched away using HF, α-radiation influences only the fine-grain material. Thus, for fine grains, the a-value giving the effectivity of αcompared to β-radiation must be determined. We determined natural radioactivity by measuring the concentration of the radioactive elements U, Th and K using dry, ground material. U and Th contents were calculated using thick source α-counting (42 mm) at the University of Bayreuth. Potassium concentrations were measured at the Bayreuth Center for Ecology and Environmental Research (BayCEER), Bayreuth/Germany using inductively coupled plasma source mass spectrometry (ICPMS), and at the University of Marburg/Germany using an atomic adsorption spectrometer (AAS). Dose rates were calculated using the conversion factors given by Adamiec and Aitken [1998]. Cosmic dose rates were evaluated according to Prescott and Hutton [1994]. All dose rates are listed in Table 4. 4.4. Water Content For every sample the water content was measured gravimetrically. Since the outcrops have been open for many years and the sediments have been desiccated, measured values are obviously underestimations compared to the paleovalues as already stated by Zöller et al. [2003]. Thus, using the grain size distribution of a sample and the resulting middle pore volume, we estimated potential minimal and maximal water contents of a sample as described by Fuchs [2001]. For age calculation, an average value was taken. The error is assumed to be generally 0.1. Measured and corrected water contents are given in Table 4. 4.5. Age Calculation von Suchodoletz et al. (submitted manuscript, 2008a) demonstrate that most of the sediments filling the vegas are reworked and deposited by colluvial and fluvial processes. These processes are problematic for luminescence dating because they are occurring rather rapidly and often the finer material is transported in the form of aggregates. Hence bleaching of these sediments during colluvial transport was presumably not always complete, thus resulting in overestimated De values as detected in other studies [e.g., Porat et al., 2001; Fuchs and Wagner, 2003]. For fine grains, insufficient bleaching is not easily detectable since every disc contains thousands of grains giving a luminescence signal so that De differences between the discs are averaged. Instead, for coarse grains single aliquot and single grain methods offer the possibility to detect and correct insufficient bleaching by looking at their De scatter and the type of De distribution (Gaussian, left-skewed, right-skewed) [e.g., Olley et al., 1998; Lepper et al., 2000; Bailey and Arnold, 2006; Fuchs and Wagner, 2003]. Since most of our De distributions are positively skewed which may indicate insufficient bleaching, we had to find a way to obtain the De corresponding to the last reworking of the sediments. Different techniques were developed on the basis of the assumption that the last bleaching event is equivalent to the left maximum of a right skewed De distribution. However, some of these techniques are not applicable to our De distributions for different reasons: the approach of Olley et al. [1998] taking the lowest 5% of the De distribution is not applicable owing to a relatively broad rising limb of the De distribution resulting in greatly underestimated De. Accordingly, this method is generally judged to be useful only in case of young, poorly bleached fluvial samples [Bailey and Arnold, 2006]. The leading edge method developed by Lepper et al. [2000] was inapplicable since the quantity of measured aliquots was not sufficient to get a good Gaussian fit through the rising limb of our distributions [cf. Fuchs et al., 2007]. The method of Fuchs and Lang [2001], using an empirically derived threshold to identify the well bleached proportion of an insufficiently bleached sample is also not applicable in our case. This is due to the fact that for many samples the threshold was already exceeded using the first two De from the sorted low to high De value data set. Thus we decided to use the approach of Juyal et al. [2006], modified by Fuchs et al. [2007]. Basically, this is a simplified minimum age model of Galbraith et al. [1999]. De = Demin + 2 * σmax + 4% where Demin is the minimal equivalent dose after subtraction of the lower 5% quantil (see below), σmax is the maximal error occurring in the data from the lower end to the first maximum of the De histogram and 4% are added as general instrumentation error of the used Risø reader. A common problem is the mixing of younger material dropped through pedogenic cracks or root channels as described by Bateman et al. [2003]. We have observed this process in the vega of Guatiza. This phenomenon would cause an underestimation of the De and is visible as individual outliers not linked to the form of the De distribution toward lower values in many histograms (Figure 3). To neutralize this frequently occurring effect, we decided to subtract the lower 5%quantil of the sorted De distribution from all samples prior to statistical ccalculations. Depending on the number of measured aliquots, one to two De had to be rejected. For fine grains, insufficient bleaching is hardly detectable. Thus the arithmetic mean of obtained De from the measurements was taken for age calculation. IRSL ages by Zöller et al. [2003] were recalculated using corrected water contents (see above) and new cosmic dose rates. 5. Results 5.1. Dose Rates Results of dose rate ( . D) determination are given in Table 4 with their 1s errors. Uranium contents calculated from thicksource α-counting range from 1.46 to 3.1 ppm and thorium values from 5.0 to 10.7 ppm. Potassium contents determined either by AAS or ICPMS range from 1.7 to 4.1%. Rather low dose rate values generally occur in calcareous horizons (e.g., BT 318, BT 319, BT 308). The occurrence of radioactive disequilibria in calcareous horizons cannot be excluded. However, the U content of calcite presents only a very small proportion compared to the total Ucontent, so that desequilibria in the U-decay chain are not expected to have a significant influence on the total dose rate. Furthermore, Schäfer and Zöller [1996] demonstrated from a Middle Palaeolithic site in Thuringia/Germany showing significant disequilibria detected by lowlevel γ-spectrometry, that using dose rates from U and Th based on α-counting is a good approach if radioactive disequilibrium has regenerated more or less permanently since deposition. Carbonate precipitation posterior to deposition may dilute concentrations of radioelements and thus lower the dose rate. However, carbonate-free layers deposited over carbonate-rich beds indicate that secondary calcification was more or less contemporaneous to sedimentation. The consequences of this effect should not have played a major role although a slight underestimation of dose rates due to this effect cannot be excluded. 5.2. Equivalent Doses Equivalent doses of quartz OSL and feldspar IRSL measurements are listed in Table 5 with their 1σ errors. Owing to their own characteristics and accuracies, the applied luminescence methods (coarseand fine-grain OSL, fine-grain IRSL) are discussed separately. 5.2.1. Quartz Coarse-Grain OSL measurements The potential for total bleaching of the investigated sediments was tested by exposure of the prepared coarse-grain quartz extracts to 10 hours of natural daylight (Bayreuth/Germany, ~50°N, November 2005) with a subsequent OSL measurement. This test was carried out for samples BT 190, BT 195 and BT 196 which showed no OSL signal after daylight exposure. Furthermore, a dose recovery test was carried out using the same samples as for the bleaching test. After the bleaching of the samples they were given b-radiations of 4.28 Gy (sample BT 195), 17.13 Gy (sample BT 196) and 38.55 Gy (sample BT 190). Subsequently, the OSL measurements for De determination were performed with preheat temperatures of 220°, 240°, 260° and 280°C, with four aliquots for each temperature. In Figure 4 the results of these measurements can be seen, with a De plateau between 220° and 280°C within error bars. Thus, in spite of a slight overestimation of given doses, for all further De measurements a preheat temperature of 240°C was applied. All quartz samples show typical exponential saturating growth. From the growth curves it can be seen that saturation, on average, is reached around 350 Gy but can also be somewhat higher for several samples. Thus highest De estimations above 300 Gy have to be treated with caution. Measured aliquots were excluded from further analysis when test dose error, paleodose error, or recycling ratio error were >10%. The same applies to aliquots giving a signal less than three times the standard deviation of the background. This was true for about 7% of the measured aliquots. 5.2.2. Quartz fine-grain OSL measurements Fine-grain OSL De range from 5.6 to 211 Gy. The frequency distribution of the measured aliquots per sample yielded a standard deviation between 2 to 12%. About 3% of the measured aliquots had to be removed owing to the exclusion criteria defined above. All growth curves show an exponential behavior, with saturation attained around 350 Gy (highest calculated De = 211 Gy). Typical measurements of a-values range between 0.03 and 0.05. Consequently, a mean value of 0.04 was used for calculations. 5.2.3. IRSL fine-grain measurements As can be seen from the IRSL growth curves, saturation of the investigated feldspars is attained around 1800-2000 Gy. Hence all investigated samples (highest De ~700 Gy) are within the datable dose range. IRSL growth curves from Lower to Middle Holocene samples show an almost linear behavior, whereas the older ones are clearly saturating exponential. Fitting error of the growth curves was between 0.3 and 3.5%, giving lower values for higher doses. As already stated by Pomel et al. [1985], measured a-values are very low with values between 0.029 and 0.063, being in general higher in Teguise (average 0.06) than in Femés (average 0.04) and Guatiza (average 0.05). Owing to long a-irradiation times only some representative a-values were determined, whereas for the missing values the mean for every profile was taken. Almost half of the samples show significant anomalous fading up to 40%. 6. Discussion 6.1. Reliability of different luminescence methods 6.1.1. Coarse-Grain OSL Ages Most of our coarse-grain single aliquot histograms are right skewed (mean skewness 1.24), indicating insufficient bleaching during colluvial transport. Several outliers toward lower doses were also observed (Figure 3). Owing to the strong scatter of the equivalent doses, age errors are between 9 and 31%. The occurrence of colluvial reworking during most periods was also proven by micromorphological studies [Sauer and Zöller, 2006]. This occurs in all vegas, however the histogram asymmetry is generally lower in Teguise compared to Femés and Guatiza III. This is explained by the lower catchment area-valley bottom ratio, which means a lower proportion of (insufficiently bleached) colluvial material originating from the slopes relative to in situ eolian dust is found in the vega bottom sediments (von Suchodoletz et al., submitted manuscript, 2008). Correspondingly, lowest errors were observed in the vega of Teguise. Using the approach of Juyal et al. [2006], most De values are now situated at the upper rising left limb of our distribution (Figure 3). Here we believe that the equivalent doses corresponding to the last bleaching event should be located. The general omission of the lower 5% of measured aliquots from all samples strongly improved our De calculation in samples exhibiting outliers toward the lower end of the distribution, and thus De of most of these samples fall at the upper rising limb of our histograms as well. In cases without outliers, owing to the broad rising limb, the resulting shift toward higher De is very small. Thus this method appears to be a justified approach for De determination. For two samples, however, this approach seems to fail: Sample BT 200 shows an exceptional De histogram with a left skewed distribution (Figure 5a). Since the De of this sample is close to saturation level it is very likely that the aliquots normally forming the right (older) tail of the distribution were already in saturation and thus are not present in the histogram. Additionally, the number of aliquots is strongly limited so that a rising limb is not well developed. The arithmetic mean De of this sample gives a value at the rising limb close to the first maximum of the distribution (Figure 5a). However, owing to the non-Gaussian form of the distribution this arithmetic mean can only be a rough estimation. Caused by the limited amount of aliquots, the calculated De of sample BT 197 may be underestimated and should be regarded as a minimum age (Figure 5b). Luminescence tests investigating insufficient bleaching (due to iron stains around the grains) during aeolian transport were carried out with recent Saharan dust from the Canary Islands (H. von Suchodoletz et al., Luminescence bleaching characteristics of Saharan dust – A case study from Lanzarote, Canary Islands (Spain), submitted to Quaternary Geochronology, 2008) (hereinafter referred to as von Suchodoletz et al., submitted manuscript, 2008b). They show that for coarse grains inherited doses with respect to dose rates from Lanzarote are in the range of only 0.1-0.2 ka and are thus negligible. A sample from colluvial material overlying a historic lava flow dating from 1736 AD (sample BT 204) yields an apparent coarse-grain OSL age of about 1.2 ka, thus indicating an overestimation <1 ka. These tests show that coarsegrain OSL basically offers the required conditions for dating Lower Holocene or older fluvioeolian sediments. The problem of a slight overestimation of given doses during the preheat tests (see Figure 4) must be seen against the background of generally relatively large errors of coarse-grain OSL ages:owing to the uncertainty caused by insufficient bleaching, dating errors are rather large, so that the overestimation effect is smaller than calculated errors. This shows, however, that our luminescence ages are points of reference rather than precise dates. 6.1.2. Fine-Grain Quartz OSL Ages Comparison of coarse and fine-grain quartz OSL ages reveals that the latter do not show significant age overestimations as could be expected owing to insufficient bleaching. Young Holocene samples show, however, a regular overestimation of about 1 to 3 ka. (samples BT 189, BT 198, BT 310). Overestimation of the same magnitude is found for recent sample BT 204. Thus bleaching of fine grains is not as complete as for coarse-grain OSL. This could be due to their colluvial transport as aggregates. Since quartz OSL does not exhibit anomalous - Menéndez, I., J. L. Diaz-Hernandez, J. Mangas, I. Alsonso, and P.-J. Sanchez-Soto (2007), Airborne dust accumulation and soil development in the North-East sector of Gran Canaria (Canary Islands, Spain), J. Arid Environ., 71, 57–81. - Moreno, A., J. Taragona, J. Henderiks,M. Canals, T. Freudenthal, and H. Meggers (2001), Orbital forcing of dust supply to the North Canary Basin over the last 250 kyr, Quat. Sci. Rev., 20, 1327–1339. - Murray, A., and A. Wintle (2000), Luminescence dating of quartz using an improved single-aliquot regenerative-dose protocol, Radiat. Meas., 32, 57–73. - Olley, J., G. Caitcheon, and A. Murray (1998), The distribution of apparent dose as determined by optically stimulated luminescence in small aliquots of fluvial quartz: Implications for dating young sediments, Quat. Geochronol., 17, 1033–1040. - Ortiz, J. E., T. Torres, Y. Yanes, C. Castillo, J. de la Nuez, M. Ibanez, and M. R. Alonso (2006), Climatic cycles inferred from the aminostratigraphy and aminochronology of Quaternary dunes and paleosols from the eastern islands of the Canary Archipelago, J. Quat. Sci., 21(3), 287–306. - Petit-Maire, N., G. Delibrias, J. Meco, S. Pomel, and J.-C. Rosso (1986), Paléoclimatologie des Canaries orientales (Fuerteventura), C. R. Acad. Sci. Ser. II, 303(13), 1241–1246. - Pomel, R.-S., D. Miallier, J. Fain, and S. Sanzelle (1985), Datation d’un sol brunrouge calcife`re par une coulée volcanique d’âge Würm ancien (51.000 ans) à Fuerteventura (Iles Canaries), Mediterranée, 4, 59–68. - Porat, N., E. Zilberman, R. Amit, and Y. Enzel (2001), Residual ages of modern sediments in an hyperarid region, Israel, Quat. Sci. Rev., 20, 795–798. - Prescott, J. R., and J. T. Hutton (1994), Cosmic ray contributions to dose rates for luminescence and ESR dating: Large depths and long-term time variations, Radiat. Meas., 23, 497–500. - Rognon, P., and G. Coudeé-Gaussen (1987), Changements dans les circulations atmosphérique et océanique à la latitude des Canaries et du Maroc entre les stades isotopiques 2 et 1, Quaternaire, 7(4), 197–206. - Rognon, P., G. Coude´-Gaussen, M.-N. Le Coustumer, J. C. Balouet, and S. Occhietti (1989), Le massif dunaire de Jandía (Fuerteventura, Canaries): Évolution des paléoenvironnements de 20.000 BP a` l’actuel, Int. J. Fr. Quat. Assoc., 37, 31–37. - Rousseau, D.-D., L. Zöller, and J.-P. Valet (1998), Late Pleistocene climatic variations at Achenheim, France, based on magnetic susceptibility and TL chronology of loess, Quat. Res., 49, 255–263. - Sauer, D., and L. Zöller (2006), Mikromorphologie der Paläoböden der Profile Femés und Guatiza, Lanzarote, Bayreuther Geowiss. Arb., 27, 105–130. - Schäfer, D., and L. Zöller (1996), Zur Charakterisierung des weichselzeitlichen Freilandfundplatzes vom Gamsenberg bei Oppurg/Thüringen, Tübinger Monogr. Urgeschichte, 11, 235–246. - Singhvi, A. K., and M. R. Krbetschek (1996), Luminescence dating: A review and a perspective for arid zone sediments, Ann. Arid Zone, 35(3), 249–279. - Trautmann, T., M. R. Krbetschek, A. Dietrich, and W. Stolz (2000), The basic principle of radioluminescence dating and a localized transition model, Radiat. Meas., 32, 487–492. - Wintle, A. G. (1973), Anomalous fading of thermoluminescence in mineral samples, Nature, 245, 143–144. - Wintle, A. G. (1998), Luminescence dating: Laboratory procedures and protocols, Radiat. Meas., 27, 769–817. - Xie, J., and M. J. Aitken (1991), The hypothesis of mid-term fading and its trial on Chinese loess, Ancient TL, 9, 21–25. - Zöller, L., H. von Suchodoletz, and N. Küster (2003), Geoarchaeological and chronometrical evidence of early human occupation on Lanzarote (Canary Islands), Quat. Sci. Rev., 22, 1299–1307. - Zöller, L., D.-D. Rousseau, K.-D. Jäger, and G. Kukla (2004), Last interglacial, Lower and Middle Weichselian—A comparative study from the Upper Rhine and Thuringian loess areas, Z. Geomorphol., 48, 1–24. - Zöller, L., D. Faust, and H. von Suchodoletz (2006), Pedostratigraphie des Quartärs auf Lanzarote, Bayreuther Geowiss. Arb., 27, 1–21. Tables Table 1: Properties of Studied Sites on Lanzarote Femés Guatiza Teguise Latitude N 28°55´32´´ 29°04´08´´ 29°04´52´´ longitude W 13°45´19´´ 13°29´22´´ 13°30´55´´ Altitude (m a.s.l.) 300 100 300 catchment area (km2) 5.07 10.1 3.8 valley bottom (% of catchment area) 18 16.1 35 relative elevation difference in the catchment area (m) 100 (SE slopes) – 200 (NW slopes) > 550 100 time of volcanic damming 1.4 Ma1 170 ka2 1.2 Ma3 volcanic damming complete? yes no no thickness of anthropogenic colluvial deposits (cm) 55 420 80 1 Zöller et al. (2006) 2consistent thermoluminescence and ESR data (H. v. Suchodoletz et al., unpublished results, 2006) 3 Instituto Tecnológico y Geominero de España [2008]. Table 2 : Kind and strengths of used irradiation-sources. Strengths are given in Gy/min and referred to 21 August 2006, except for the external 90Sr/90Y-source which is referred to 15 June 2006. type of irradiation type of source coarse grain fine grain 90Sr/90Y (internal Risø-reader 1) 8.81 - 90Sr/90Y (internal Risø-reader 2) 2.57 2.52 Β 90Sr/90Y (external) - 7.85 Α Littlemore 241Am (external) - 1.26 Table 3: Used measurement-parameters for OSL and IRSL OSL IRSL StimulationLEDs 470 ± 20 nm 845 ± 20 nm Detection-filter U-340 Hoya (transmission 290 - 380 nm) BG 39, GG 400, 2xBG 3 (transmission 390-450 nm) Preheat (rate 10°/s) 240°C for 10 s 220°C for 120 s (age < 10 ka) 220°C for 300 s (age > 10 ka) Cut-heat (rate 5°C/s) 160°C Test dose Reader 1: 2.2 Gy Reader 2: 0.64 Gy Measurement 20 s at 125°C 60 s at room temperature Used integral 0.0 - 0.4 s 0 - 15 s Background interval 16 - 20 s 55 - 60 s Table 4: Luminescence dating: Analytical results Dose rates ( . D) are given in Gy/ka. Uand Th-contents were calculated from the α –countrate. Measured water contents present the ratio moist/dry weight, the correction was done using the clay contents of the samples. The error of water contents is assumed to be 0.1. Sample depth (cm) Sample α - countrate (cpm) U-content (ppm) Th-content (ppm) K-content (%) 1 from ICPMS 2 from AAS . D-β* . D-γ* . Dcosm Measured water content Corrected water content Femés 60 BT 310, (BN-D 207) 0.58* - - - 2.58 ± 0.13 1.12 ± 0.06 0.21 ± 0.011 1.1* 1.19 80 BT 311 0.48 ± 0.012 2.34 ± 0.058 8.00 ± 0.20 2.45 ± 0.072 - - 0.21 ± 0.010 1.13 1.24 100 BT 189 0.46 ± 0.009 2.25 ± 0.044 7.70 ± 0.15 3.19 ± 0.101 - - 0.20 ± 0.010 1.13 1.24 120 BT 312 0.36 ± 0.008 1.75 ± 0.039 6.01 ± 0.13 3.32 ± 0.102 - - 0.20 ± 0.010 1.13 1.24 140 BT 313 0.43 ±0.006 2.09 ± 0.029 7.16 ± 0.10 2.12 ± 0.062 - - 0.20 ± 0.010 1.14 1.22 170 BT 190, (BN-D 209) 0.47* - - - 3.25 ± 0.05 1.08 ± 0.05 0.19 ± 0.010 1.19* 1.25 190 BT 314 0.40 ± 0.008 1.96 ± 0.040 6.71 ± 0.14 1.91 ± 0.062 - - 0.19 ± 0.010 1.17 1.26 210 BT 315 0.44 ± 0.001 2.12 ± 0.004 7.25 ± 0.02 2.61 ± 0.082 - - 0.19 ± 0.009 1.20 1.27 225 BT 316 0.39 ± 0.010 1.87 ± 0.047 6.39 ± 0.16 2.66 ± 0.082 - - 0.19 ± 0.009 1.20 1.28 245 BT 317 0.41 ± 2.01 ± 0.047 6.87 ± 0.16 2.99 ± 0.092 - - 0.18 ± 0.009 1.19 1.30 0.010 270 BT 224 0.44 ± 0.008 2.13 ± 0.039 7.28 ± 0.13 2.89 ± 0.091 - - 0.18 ± 0.009 1.18 1.24 290 BT 318 0.30 ± 0.008 1.46 ± 0.039 5.01 ±0.13 1.78 ± 0.052 - - 0.18 ± 0.009 1.11 1.24 310 BT 319 0.30 ± 0.006 1.46 ± 0.027 5.01 ± 0.09 2.07 ± 0.062 - - 0.18 ± 0.009 1.11 1.23 330 BT 301 0.44 ± 0.009 2.12 ± 0.044 7.27 ± 0.15 2.24 ± 0.072 - - 0.17 ± 0.009 1.14 1.25 350 BT 302 0.43 ± 0.010 2.09 ± 0.048 7.16 ± 0.17 1.70 ± 0.052 - - 0.17 ± 0.009 1.18 1.28 380 (400*) BN-D 211 0.45* - - - 3.27 ± 0.05 1.09 ± 0.06 0.17 ± 0.008 1.15* 1.27 460 BT 191 0.47 ± 0.010 2.28 ± 0.050 7.16 ± 0.17 3.39 ± 0.101 - - 0.16 ± 0.008 1.12 1.26 480 BT 303 0.44 ± 0.007 2.11 ± 0.034 7.23 ± 0.12 2.83 ± 0.081 - - 0.15 ± 0.008 1.17 1.28 630 (670*) BN-D 213 0.45* - - - 3.12 ± 0.16 0.99 ± 0.05 0.14 ± 0.007 1.30* 1.3 Teguise 80 BT 306 0.52 ± 0.018 2.5 ± 0.087 8.57 ± 0.30 2.53 ± 0.082 - - 0.21 ± 0.010 1.05 1.16 90 BT 198 0.65 ± 0.013 3.12 ± 0.061 10.70 ± 0.21 2.49 ± 0.071 - - 0.21 ± 0.010 1.04 1.16 110 BT 235 0.52 ± 0.010 2.51 ± 0.048 8.60 ± 0.17 2.86 ± 0.092 - - 0.20 ± 0.010 1.09 1.27 135 BT 307 0.44 ± 0.016 2.11 ± 0.078 7.22 ± 0.27 3.19 ± 0.102 - - 0.20 ± 0.010 1.19 1.28 160 BT 199 0.46 ± 0.007 2.21 ± 0.033 7.57 ± 0.11 3.69 ± 0.111 - - 0.20 ± 0.010 1.13 1.31 185 BT 308 0.32 ± 0.006 1.56 ± 0.030 5.35 ± 0.10 2.24 ± 0.072 - - 0.19 ± 0.010 1.14 1.26 210 BT 309 0.39 ± 0.008 1.86 ± 0.039 6.38 ± 0.13 2.32 ± 0.072 - - 0.19 ± 0.009 1.17 1.24 250 BT 200 0.56 ± 0.010 2.69 ± 0.048 9.21 ± 0.17 2.90 ± 0.091 - - 0.18 ± 0.009 1.1 1.26 330 BT 236 0.40 ± 0.007 1.92 ± 0.034 6.57 ± 0.12 4.08 ± 0.121 - - 0.17 ± 0.009 1.14 1.26 450 BT 201 0.42 ± 0.009 2.05 ± 0.044 7.03 ± 0.15 3.73 ± 0.111 - - 0.16 ± 0.008 1.18 1.26 610 BT 202 0.43 ± 0.007 2.10 ± 0.034 7.19 ± 0.12 2.93 ± 0.091 - - 0.14 ± 0.007 1.09 1.27 Guatiza III 110 BT 193 0.39 ± 0.008 1.88 ± 0.039 6.44 ± 0.13 2.89 ± 0.091 - - 0.2 ± 0.010 1.09 1.24 250 BT 194 0.40 ± 0.009 1.96 ± 0.044 6.71 ± 0.15 2.53 ± 0.081 - - 0.18 ± 0.009 1.10 1.26 435 BT 195 0.51 ± 0.009 2.49 ± 0.044 8.53 ± 0.15 3.96 ± 0.121 - - 0.16 ± 0.008 1.07 1.21 470 BT 233 0.43 ± 0.006 2.06 ± 0.029 7.05 ± 0.10 2.51 ± 0.081 - - 0.15 ± 0.008 1.11 1.25 495 BT 196 0.48 ± 2.35 ± 0.044 8.03 ± 0.15 2.73 ± 0.081 - - 0.15 ± 0.008 1.08 1.25 0.009 550 BT 234 0.44 ± 0.006 2.11 ± 0.029 7.22 ± 0.10 3.22 ± 0.101 - - 0.15 ± 0.008 1.14 1.26 610 BT 304 0.39 ± 0.011 1.90 ± 0.053 6.51 ± 0.18 4.02 ± 0.122 - - 0.14 ± 0.007 1.13 1.28 660 BT 197 0.53 ± 0.009 2.58 ± 0.044 8.84 ± 0.15 2.99 ± 0.091 - - 0.14 ± 0.007 1.09 1.22 700 BT 305 0.50 ± 0.013 2.43 ± 0.063 8.32 ± 0.22 3.86 ± 0.122 - - 0.13 ± 0.007 1.11 1.23 Guatiza I 210 BN-D 219 0.46* - - - 2.58 ± 0.04 0.85 ± 0.04 0.18 ± 0.009 1.20 - Tahiche recent BT 204 0.42 ± 0.009 2.04 ± 0.043 6.98 ± 0.15 2.00 ± 0.061 - - 0.21 ± 0.011 1.10 - * data taken from Zöller et al. [2003] Table 5: Luminescence dating: Equivalent doses and ages De´s are given in Gy. Coarse grain OSL-De´s were calculated after Juyal et al. [2006]. All ages are given in ka. Measured a-values are given in standard, assumed a-values in italic letters. Sample depth (cm) Code IRSL-De a-value IRSL IRSL age Anomalous Fading Fine-grain quartz OSL-De a-value fine grain quartz OSL Fine-grain quartz OSL age Coarse-grain quartz OSL-De Coarse-grain quartz OSL age Femés 60 BT 310 (BN-D 207) 18.3 ± 4.7* 0.04 ± 0.001* 4.9 ± 1.3 yes* - - - 20.5 ± 1.5 6.4 ± 0.7 80 BT 311 - - - - - - 38.0 ± 3.4 12.4 ± 1.6 100 BT 189 65.5 ± 0.8 0.04 ± 0.002 16.7 ± 1.5 yes 65.3 ± 4.9 0.03 ± 0.01 16.8 ± 1.9 47.9 ± 3.7 13.6 ± 1.6 120 BT 312 - - - - - - - 122 ±11.9 35.4 ± 4.7 140 BT 313 - - - - - - - 150 ± 10.8 56.0 ± 6.5 170 BT 190 (BN-D 209) 166 ± 12* 0.04 ± 0.002* 41.0 ± 3.8 no* - - - 165 ± 31.4 45.9 ± 9.7 190 BT 314 - - - - - - - 194 ± 22.6 55.9 ± 8.2 210 BT 315 - - - - - - - 164 ± 22.8 47.7 ± 7.8 225 BT 316 - - - - - - - 204 ± 25.1 54.8 ± 7.2 245 BT 317 - - - - - - - 193 ± 17.4 57.6 ± 7.2 270 BT 224 328 ± 2 0.04 ± 0.002 91.2 ± 8.1 no - - - 208 ± 16.6 64.8 ± 7.8 290 BT 318 - - - - - - - 213 ± 49.2 101.0 ± 24.9 310 BT 319 - - - - - - - 219 ± 65.7 93.3 ± 29.2 330 BT 301 - - - - - - - 237 ± 21.0 88.3 ± 11.1 350 BT 302 - - - - - - - 275 ± 46.8 125.0 ± 23.8 Figure 4. Results of the dose recovery test of coarsegrain samples measured with OSL, shown with error bars (2σ). Every data point is the average of four single measurements. Dashed lines below a graph indicate the given paleodose. Mean ratios given/measured dose were: 220°C, 1.08; 440°C, 1.09; 260°C, 1.1; 280°C, 1.07. Figure 5. De histograms of coarse-grain OSL samples (a) BT 200 and (b) BT 197 with different De. Figure 6. Vega stratigraphies with relative kaolinite contents adjusted to numerical chronostratigraphy and to marine Al and Fe contents. Luminescence dates giving correct sedimentation ages are shown with their error bars (in kiloannums), whereas the minimum age of sample BT 197 not presented. Shadowed areas in the profiles indicate generally unweathered/less weathered layers visible in the field and used for correlation between the vegas. Horizon depths in centimeters (not in scale) are indicated on the left side of the profiles. Figure 7. Overview of measured luminescence ages with hiatae. Reliable and unreliable ages are discriminated.