Penkalski P.,Geology Museum
Acta Palaeontologica Polonica | Year: 2014
Oohkotokia horneri gen. et sp. nov. is described based on a specimen in the collections of the Museum of the Rockies, Montana, USA. Oohkotokia exhibits a unique combination of characters not seen in other late Campanian North American ankylosaurids: prominent, horn-like, trihedral squamosal bosses, a small, undistinguished median nasal plate on the dorsal surface of the rostrum, a relatively small occipital condyle, a smooth, finely pitted osteoderm external texture, and triangular lateral osteoderms. Other specimens from the Two Medicine Formation are referable to Oohkotokia. O. horneri, Euoplocephalus tutus, Dyoplosaurus acutosquameus, and Scolosaurus cutleri separate stratigraphically. © 2014 P. Penkalski.
In 2007 and 2008 we undertook three (T1–T3) 1 m × 2 m test excavations at Talepu Hill, where large numbers of stone artefacts were found scattered on the surface with loose gravel. The summit of Talepu Hill (4° 22′ 06.5′′ S; 119° 59′ 01.7′′ E) lies 36 m above sea level and 18 m above the floodplain of the Walanae River, which flows 600 m to the east (Extended Data Fig. 1). Geological outcrop conditions are very poor, and thick tropical soils cover the underlying geological formations. The three test excavations near the summit of Talepu Hill proved the occurrence of in situ stone artefacts down to a depth of at least 1.8 m, in heavily weathered conglomerate lenses and sandy silt layers. The same gravel unit occurs on other hilltops to the west and southwest. At Bulu Palece, 850 m west of Talepu Hill, which is the highest hilltop in the vicinity with an elevation of 51 m (see Extended Data Fig. 2), the gravel is at least 13 m thick, but at Talepu Hill only a basal interval of 4.3 m thickness remains. In October 2009, T2 was taken down to 7 m below surface (Extended Data Fig. 2b), at which depth the excavation area was reduced to a 1 m × 1 m square and taken down further to a maximum depth of 10 m. To ensure that this deep-trench operation was undertaken safely, we installed timber shoring as the work progressed (Extended Data Fig. 2c). A new east–west oriented, 1 m × 9 m trench (T4) was excavated at the base of the Talepu Hill, 40 m east of T2. This trench reached a maximum depth of 2 m, revealing the lateral development of the stratigraphy near the base of the hill (Fig. 2 and Extended Data Fig. 2d). Deposits were removed in 10 cm spits within stratigraphic units. Stone artefacts and fossils found by the excavators were bagged and labelled immediately; all other deposits were dry sieved with 5 mm mesh to separate out clasts, including stone artefacts. Pebbles from each spit were weighed; and composition analysis was undertaken on clasts from a representative sample from six spits: average maximum clast diameter was recorded by measuring the longest diameter of the ten largest clasts per spit (Extended Data Fig. 3). Bulk samples of stratigraphic units were taken for sediment and pollen analyses. In October 2010, the excavations at Talepu were continued. A 1 m × 2 m area at the east end of T4 was excavated to a depth of 6.20 m below the surface, thus providing an additional 6 m stratigraphically below the section covered by excavation 2 in 2009. The T4 deposits were removed in 20 cm spits within stratigraphic units. After the excavation of an in situ stone artefact (specimen S-TLP10-1, a flake from sub-unit E at a depth of 2.38 m below the surface) and fossils of Celebochoerus, it was decided to wet-sieve all the excavated sediments with 3 mm mesh to separate out stones and other clasts, including stone artefacts. Wet-sieving of the silty clay deposits from the interval between 2 and 2.4 m depth yielded one more stone artefact (S-TLP10-2; Fig. 3m) and two possible stone artefacts (S-TLP10-3 and S-TLP10-4). Magnetic susceptibility measurements were taken from the excavation profile at 1 cm intervals with a Bartington MS-2 device, to examine the presence of cryptic tephra layers suitable for dating. A sample for 40Ar/39Ar dating was taken at 2.5 m below the ground surface from an interval with elevated magnetic susceptibility values. In October 2012, backfill of T2 and T4 was removed. T4 was enlarged with a 1 m × 2 m extension (T4-B), and both T2 and T4/4-B were taken further down with an additional 2 m and 2.1 m, respectively, to allow for sampling for palaeomagnetic and optical dating methods. In T4-B two more stone artefacts (Fig. 3j–k), originating from spit 31 (depth 3.0–3.1 m depth below ground level), were recovered on the sieves. Stone artefacts were analysed following the definitions and methods in ref. 31. The analysis focused on stone-flaking techniques, sequences of reduction, and sizes of stone-flaking products and by-products (Supplementary Tables 1 and 2). The stone artefacts are stored at the Geology Museum in Bandung. The details for laser ablation uranium-series analysis of skeletal materials were recently summarized14. Uranium-series analyses provide insights into when uranium migrates into a bone or tooth. This may happen a short time after the burial of the skeletal element or some significant time span later. There may also be later uranium-overprints that are difficult to recognize. As such, apparent uranium-series results from faunal remains have generally to be regarded as minimum age estimates. It is very difficult or impossible to evaluate by how much the uranium-series results underestimate the correct age of the sample. Details of the instrumentation, analytical procedures and data evaluation have been modified from those described in detail elsewhere14, 32. All isotope ratios refer to activity ratios. Sequential laser spot analyses were undertaken on cross sections of eight Celebochoerus fossils from the T4 excavation at Talepu. They comprised fragments of six teeth and two bones from sub-unit E found 10–50 cm below the lowest stone artefacts in the same silt layer. Of one fossil (TLP10-1, a Celebochoerus lower canine), two subsamples were analysed (a and b). Each fossil specimen was cut transversely using a dentist drill with a diamond saw blade (Extended Data Fig. 4). Four or five samples were then mounted together into aluminium cups, aligning the cross-sections with the outer rim of the sample holder, which later positioned the samples on the focal plane of the laser. Uranium-series isotopes were measured using the laser ablation multicollector (MC)-ICP-MS system at The Australian National University’s (ANU) Research School of Earth Sciences. It consists of a Finnigan MAT Neptune MC-ICP-MS equipped with multiple Faraday cups. At the time of measurement, the mass spectrometer had only one ion counter. This necessitated two sequential sets of measurements along parallel tracks, one for 230Th and a second for 234U. The ion counter was set either to masses 230.1 or 234.1 while the Faraday cups measured the masses 232, 235 and 238. Samples were ablated with a Lambda Physik LPFPro ArF excimer (λ = 193 nm) laser coupled to the Neptune through an ANU-designed Helex ablation cell. The samples were initially cleaned for 10 s with the laser spot size set to 265 μm followed by a 50 s analysis run with a 205 μm spot size using a 5 Hz pulse rate. Analyses were performed at regular intervals along traverses, all starting from the exterior surface (Extended Data Fig. 4a–i). The data sets of each transect were bracketed between reference standard analyses to correct for instrument drift. Semi-quantitative analysis of uranium and thorium concentrations were derived from repeated measurements of the SRM NIST-610 glass (uranium = 461.5 μg g−1; thorium = 457.2 μg g−1), and uranium-isotope ratios from repeated measurements of rhinoceros tooth dentine from Hexian (sample 1118)33. Age estimates combining all measurements on a specimen were calculated using the iDAD program15, assuming diffusion from both surfaces for the bones (TLP10-6 and 7) and roots of the teeth (Extended Data Fig. 4a–f, h, i) and directional diffusion from the central pulp cavity into the dentine and covering enamel for TLP10-9 (Extended Data Fig. 4g). The enamel data of the enamel samples were omitted as enamel has a different diffusion rate. Generally, results with elemental U/Th <300 are rejected, as these are associated with detrital contamination. However, this applied only to a single measurement. The finite ages are given with 2σ error bands; the infinite results only refer to the lower bound of the 2σ confidence interval (Supplementary Table 3). None of the samples showed any indication for uranium leaching, which is either expressed by sections with 230Th/234U >> 234U/238U or increasing 230Th/234U ratios towards the surface in conjunction with decreasing uranium-concentrations. Five samples had infinite positive error bounds and it was thus only possible to calculate minimum ages. It can be seen that the uranium-series results may change over small distances within a sample. The first data set of TLP10-1 yielded a finite result of 161 ± 15 kyr while the second set yielded a minimum age of >255 kyr. As mentioned above, all uranium-series results, whether they are finite of infinite, have to be regarded as minimum age estimates. If the faunal elements present a single population, the uranium-series results indicate that the Talepu samples are most probably older than ~350 kyr, but certainly older than ~200 kyr (Supplementary Table 3). The large errors do not allow us to further constrain the age. Optical dating provides an estimate of the time since grains of quartz or potassium-rich feldspar were last exposed to sunlight34, 35, 36, 37. The burial age is estimated by dividing the equivalent dose (D , a measure of the radiation energy absorbed by grains during their period of burial) by the environmental dose rate (the rate of supply of ionizing radiation to the grains over the same period). D is determined from the laboratory measurements of the optically stimulated luminescence (OSL) from quartz or the infrared stimulated luminescence (IRSL) from potassium (K)-feldspar, and the dose rate is estimated from laboratory and field measurements of the environmental radioactivity. K-feldspar has two advantages over quartz for optical dating: (1) the IRSL signal (per unit absorbed dose) is usually much brighter than the OSL signal from quartz; and (2) the IRSL traps saturate at a much higher dose than do the OSL traps, which makes it possible to date older samples using feldspars than is feasible using the OSL signal from quartz. However, the routine dating of K-feldspars using the IRSL signal has been hampered by the malign phenomenon of ‘anomalous fading’ (that is, the leakage of electrons from IRSL traps at a faster rate than expected from kinetic considerations38), which gives rise to substantial underestimates of age unless an appropriate correction is made39. Recently, IRSL traps that are less prone to fading have been identified40, using either a post-infrared IRSL (pIRIR) approach41, 42 or a MET-pIRIR procedure16, 43. The progress, potential and remaining problems in using these pIRIR signals for dating have been reviewed recently17. Dating the samples from Talepu using quartz OSL is impractical because of the paucity of quartz. Furthermore, the quartz OSL traps are expected to be in saturation, owing to the ages of the samples (>100 kyr) and the high environmental dose rates of the deposits (4–5 Gy/kyr). In this study, we applied the MET-pIRIR procedure to K-feldspar extracts from Talepu to isolate the light-sensitive IRSL signal that is least prone to anomalous fading. We also allowed for any residual dose at the time of sediment deposition, to account for the fact that pIRIR traps are less easily bleached than the ‘fast’ component OSL traps in quartz. The resulting MET-pIRIR ages should, therefore, be reliable estimates of the time of sediment deposition at Talepu. The total environmental dose rate for K-feldspar grains consists of four components: the external gamma, beta and cosmic-ray dose rates, and the internal beta dose rate. The dosimetry data for all samples are summarized in Supplementary Table 4. The external gamma dose rates were measured using an Exploranium GR-320 portable gamma-ray spectrometer, equipped with a 3-inch diameter NaI(Tl) crystal calibrated for uranium, thorium and potassium concentrations using the CSIRO facility at North Ryde44. At each sample location, three or four measurements of 300 s duration were made of the gamma dose rate at field water content. The external beta dose rate was measured by low-level beta counting using a Risø GM-25-5 multicounter system45 and referenced to the Nussloch Loess (Nussi) standard46. The external beta dose rate was corrected for the effect of grain size and hydrofluoric acid etching on beta-dose attenuation. These external components of the total dose rate were adjusted for assumed long-term water contents of 20% for the Talepu Upper Trench (TUT = T2) samples and 30% for the Talepu Lower Trench (TLT = T4) sample (TUT and TLT sample numbers refer to the Centre for Archaeological Science laboratory numbers). These values are based on the measured field water contents (Supplementary Table 4), together with an assigned 1σ uncertainty of ±5% to capture the likely range of time-averaged mean values over the entire period of sample burial. To check the equilibrium status of the 238U and 232Th decay chains, each sample was dried, ground to a fine powder and then analysed by high-resolution gamma-ray spectrometry (HRGS). The measured activities of 238U, 226Ra and 210Pb in the 238U series, 228Ra and 228Th in the 232Th series, and 40K are listed in Supplementary Table 5. The activities of 228Ra and 228Th were close to equilibrium for all of the samples, as is commonly the case with the 232Th series. By contrast, the 238U chain of each sample, except TUT-OSL9, was in disequilibrium at the present day. Sample TUT-OSL2 had a 39–45% deficit of 226Ra and 210Pb relative to the parental 238U activity, whereas sample TUT-OSL3 had a 224–345% excess of the daughter nuclides. Samples TUT-OSL1 and TLT-OSL6 had 226Ra deficits of 50% and 26%, respectively, relative to their 238U activities, but the 210Pb activities of both samples were similar to their parental 238U activities. Sample TUT-OSL3 was the only sample with a present-day excess of 226Ra. This sample was from a sandy layer (unit B) through which ground water could percolate, so we attributed the observed 226Ra excess to the deposition of radium transported by ground water. Given the similar 238U activities of TUT-OSL3 and nearby TUT-OSL2, it is reasonable to assume that the parental uranium activity had not changed substantially during the period of burial of either sample, and that the 226Ra excess in TUT-OSL3 most probably occurred recently. The latter can be deduced from the fact that 226Ra has a half-life of ~1,600 years, which is short relative to the ages of our samples (>100 kyr), so any unsupported excess of 226Ra would have decayed back into equilibrium with 238U within ~8 kyr of deposition (that is, five half-lives of 226Ra). The alternative option—that groundwater has continuously supplied excess 226Ra to unit B—is not supported by the disequilibrium between 226Ra and 210Pb: the latter nuclide has a half-life of ~22 years, so it should remain in equilibrium with 226Ra if the latter is supplied continuously and no radon gas is lost to atmosphere. Moreover, as the return of 210Pb to equilibrium with 226Ra is governed by the half-life of the shorter-lived nuclide, it could be argued that the excess 226Ra was deposited within the past ~110 years (five half-lives of 210Pb). Fortunately, the calculated age of TUT-OSL3 is not especially sensitive to different assumptions about the timing or extent of disequilibria in the 238U series. The latter accounts for only 28% of the total dose rate estimated from the HRGS data in Supplementary Table 5; this assumes that the present-day nuclide activities have prevailed throughout the period of sample burial. If, instead, as we consider more likely, the observed excess in 226Ra was deposited recently and the 238U decay chain had been in equilibrium for almost all of the period of sample burial, then the 238U series accounted for only 12% of the total dose rate (that is, using activities of 37 ± 4 Bq kg−1 for 238U, 226Ra and 210Pb). The ages calculated under these two alternative scenarios, using only the HRGS data for estimating external beta and gamma dose rates, range from ~118 kyr to ~143 kyr (Supplementary Table 5). Sample TUT-OSL2 was from the more silty overlying layer (sub-unit A ) and had deficits of 226Ra and 210Pb relative to 238U, but these disequilibria were much smaller in magnitude than those of TUT-OSL3. If it were not continuously leached from the sample, 226Ra will return to secular equilibrium with 238U within ~8 kyr, so the existence of disequilibrium in TUT-OSL2 adds further weight to the argument for recent transport of 226Ra in ground water at Talepu. The alternative is that 226Ra has been leached continuously from this sample, so we performed the same sensitivity test on the dose rates and ages as that performed on TUT-OSL3. For TUT-OSL2, the ages determined using the present-day HRGS data or activities of 41 ± 3 Bq kg−1 for 238U, 226Ra and 210Pb are statistically indistinguishable (130 ± 12 and 125 ± 11 kyr, respectively; Supplementary Table 5), because the disequilibria are much less marked than in TUT-OSL3 and the 238U series makes only a small contribution (10–14%) to the total dose rate of TUT-OSL2. Samples TUT-OSL1 and TLT-OSL6 had deficits of 226Ra relative to 238U, but similar activities of 238U and 210Pb. The latter additionally strengthens our proposition that 226Ra was leached from these sediments recently, because 210Pb should return to a state of equilibrium with 226Ra within ~110 years (five half-lives of 210Pb). For both samples, the ages calculated using the present-day HRGS data were statistically concordant with those estimated by assuming that the 238U chain had been in secular equilibrium for almost the entire period of sample burial (Supplementary Table 5). The same applies to sample TUT-OSL9, since the measured activities of 238U, 226Ra and 210Pb were consistent at 1σ. To calculate the ages of the Talepu samples, we used the beta dose rates deduced from direct beta counting and the in situ gamma dose rates measured at each sample location. The external beta dose rates determined from beta counting and from the HRGS data (Supplementary Table 5) were statistically consistent (at 2σ) for all five samples; such agreement is expected, as both measure the present-day activities. The field gamma dose rates are also based on the nuclide activities prevailing at the time of measurement (214Bi, a short-lived nuclide between 226Ra and 210Pb, being used for the 238U series) and—importantly—take into account any spatial heterogeneity in dose rate from the ~30 cm of deposit surrounding each sample. The in situ gamma dose rates for samples TUT-OSL1 and TLT-OSL6 were consistent at 1σ with those estimated from the HRGS activities, whereas the field gamma dose rates for TUT-OSL2, -OSL3 and -OSL9 were either higher or lower than those calculated from the HRGS data. The lower in situ gamma dose rate of TUT-OSL3 can be explained by the location of this sample close to the boundary with the TUT-OSL2 sediments, which have a smaller beta dose rate (Supplementary Table 4), and vice versa for the elevated field gamma dose rate of the latter sample. This result also indicates that the 226Ra and 210Pb deficits (TUT-OSL2) and excesses (TUT-OSL3) were spatially localized and not pervasive in the 30 cm of deposit surrounding these samples. Under dim red laboratory illumination, the collected samples (see Methods) were treated with hydrochloric acid and hydrogen peroxide solutions to remove carbonates and organic matter, then dried. Grains of 90–180 or 180–212 μm in diameter were obtained by dry sieving. The K-feldspar grains were separated from quartz and heavy minerals using a sodium polytungstate solution of density 2.58 g cm−3, and etched in 10% hydrofluoric acid for 40 min to clean the surfaces of the grains and remove (or greatly reduce in volume) the external alpha-irradiated layer of each grain. For each sample, 8–14 aliquots were prepared by mounting grains as a 5-mm-diameter monolayer in the centre of a 9.8-mm-diameter stainless steel disc, using ‘Silkospray’ silicone oil as the adhesive. This resulted in each aliquot consisting of several hundred K-feldspar grains. The single-aliquot regenerative-dose (SAR) MET-pIRIR procedure introduced in ref. 16 was adapted for the Talepu samples in this study. We modified the original procedure by using a preheat at 320 °C (rather than 300 °C) for 60 s, to avoid significant influence from residual phosphorescence while recording the MET-pIRIR signal at 250 °C (Supplementary Table 6). In addition, following ref. 47, we used a 2 h solar simulator bleach before each regenerative dose cycle, instead of the high-temperature infrared bleaching step used originally, as this proved essential for recovering a given laboratory dose (see below). Example IRSL (50 °C) and MET-pIRIR (100–250 °C) decay curves are shown in Extended Data Fig. 6a for an aliquot of sample TUT-OSL2. The decay curves observed at the different stimulation temperatures are similar in shape, with initial MET-pIRIR signal intensities of the order of a few thousand counts per second. Extended Data Fig. 6b shows the corresponding dose–response curves for the same aliquot. Each sensitivity-corrected (L /T ) dose–response curve was fitted using a single saturating-exponential function of the form I = I (1 − exp−D/D0), where I is the L /T value at regenerative dose D, I is the saturation value of the exponential curve and D is the characteristic saturation dose. The D values are shown next to each dose–response curve in Extended Data Fig. 6b. For a total of 38 aliquots drawn from all 5 samples, we calculated the D values for the 250 °C MET-pIRIR signal; these are plotted in Extended Data Fig. 6c. On a ‘radial plot’ such as this, the most precise estimates fall to the right and the least precise to the left. If these independent estimates are statistically consistent with a common value at 2σ, then 95% of the points should scatter within a band of width ±2 units projecting from the left-hand (‘standardized estimate’) axis to the common value on the right-hand, radial axis. The radial plot thus provides simultaneous information about the spread, precision and statistical consistency of experimental data48, 49, 50. The measured D values range from ~220 to ~600 Gy, with the vast majority consistent at 2σ with a common value of ~360 Gy. The average D value (calculated using the central age model49) is 358 ± 14 Gy, with the standard error taking the extent of overdispersion (16 ± 4%) into account. If we adopt the D values corresponding to 90% (2.3D ) and 95% (3D ) of the saturation level of the typical dose–response curve as the upper limits for reliable estimation of D 43, 47, 51, 37, then the maximum reliable D values that we can determine using the 250 °C MET-pIRIR signal are ~820 Gy and ~1070 Gy, respectively, for these samples. To validate whether the MET-pIRIR procedure is applicable to the Talepu samples, we conducted dose recovery, anomalous fading and residual dose tests. For the latter, four aliquots of each sample were bleached for 4–5 h using a Dr Hönle solar simulator (model UVACUBE 400). The residual doses were then estimated by measuring these bleached aliquots using the modified MET-pIRIR procedure (Supplementary Table 6). The residual doses obtained for each of the TUT samples are plotted against stimulation temperature in Extended Data Fig. 7a. The IRSL signal measured at 50 °C has a few grays of residual dose, which increases as the stimulation temperature is raised, attaining values of 16–20 Gy at 250 °C. The size of the residual dose is only about 2–3% of the corresponding D values for the 250 °C signal, which were subtracted from the D values for the respective samples before calculating their ages. It was noted in ref. 52 that a simple subtraction of the residual dose from the apparent D value could result in underestimation of the true D value if the residual signal is large relative to the bleachable signal. Accordingly, it advocated the use of an ‘intensity-subtraction’ procedure instead of the simple ‘dose-subtraction’ approach for samples with large residual doses. The dose-subtraction approach should be satisfactory for the Talepu samples, however, given the small size of the residual doses compared with the D values obtained from the MET-pIRIR 250 °C signal. A dose recovery test49 was conducted on sample TUT-OSL1. Eight aliquots were bleached by the solar simulator for 5 h, then given a ‘surrogate natural’ dose of 550 Gy. Four of these aliquots were measured using the original MET-pIRIR procedure16, with a ‘hot’ infrared bleach of 320 °C for 100 s applied at the end of each SAR cycle (step 15 in Supplementary Table 6). The other four aliquots were measured using the modified MET-pIRIR procedure (Supplementary Table 6), with a solar simulator bleach of 2 h used at step 15. The measured doses at each stimulation temperature were then corrected for the corresponding residual doses (Extended Data Fig. 7a), and the ratios of measured dose to given dose were calculated for the IRSL and MET-pIRIR signals. The dose recovery ratios are plotted in Extended Data Fig. 7b, which shows that a hot bleach at the end of each SAR cycle results in significant overestimation of the known (given) dose; for the MET-pIRIR 250 °C signal, an overestimation of 48% was observed. For these same four aliquots, we obtained a ‘recycling ratio’ (the ratio of the L /T signals for two duplicate regenerative doses) consistent with unity (1.00 ± 0.03), which indicates that the test-dose sensitivity correction worked successfully between regenerative-dose cycles. The overestimation in recovered dose, therefore, implies failure of the sensitivity correction for the surrogate natural dose: that is, the extent of sensitivity change between measurement of the surrogate natural and its corresponding test dose differs from the changes occurring in the subsequent regenerative-dose cycles. The surrogate natural and regenerative-dose cycles differ only in respect to the preceding bleaching treatment (that is, a solar simulator bleach was used for the former and a hot bleach for the latter), so we compared these results with those obtained for the four aliquots that were bleached at the end of each regenerative-dose cycle using the solar simulator. The dose recovery results improved significantly using this modified procedure (Extended Data Fig. 7): all of the measured/given dose ratios were consistent with unity (at 2σ) for the signals measured at different temperatures, with a ratio of 1.02 ± 0.03 obtained for the MET-pIRIR 250 °C signal. The results of the dose recovery test on sample TUT-OSL1 suggest that the MET-pIRIR procedure could successfully recover a known dose given to K-feldspars from Talepu, but only when a solar simulator bleach was applied at the end of each SAR cycle. We therefore adopted this procedure to measure the D values for all five Talepu samples. Previous studies of pIRIR signals have shown that the anomalous fading rate (g value) depends on the stimulation temperature, with negligible fading of MET-pIRIR signals stimulated at temperatures of 200 °C and above16, 17. Accordingly, no fading correction is required for these high-temperature MET-pIRIR signals. To check that this finding also applied to the Talepu samples, fading tests were conducted on six aliquots of sample TUT-OSL3 that had already been used for D measurements. We adopted a single-aliquot procedure similar to that described in ref. 53, but based on the MET-pIRIR signals. Doses of 110 Gy were administered using the laboratory beta source, and the irradiated aliquots were then preheated and stored for periods of up to 1 week at room temperature (~20 °C). For practical reasons, we used a hot bleach (320 °C for 100 s) instead of a solar simulator bleach at the end of each SAR cycle, but this choice should not have affected the outcome of the fading test, given the aforementioned recycling ratio of unity obtained using the hot bleach. Extended Data Fig. 7c shows the decay in the sensitivity-corrected MET-pIRIR signal as a function of storage time for these six aliquots, normalized to the time of prompt measurement (which ranged from 720 s for the 50 °C IRSL to 1480 s for the 250 °C MET-pIRIR signal). The corresponding fading rates (g values) were calculated for the IRSL and MET-pIRIR signals (Extended Data Fig. 7d). The highest fading rate was observed for the 50 °C IRSL signal (5.5 ± 0.4% per decade), and decreases as the stimulation temperature is increased, falling to 0.94 ± 0.92 and 0.17 ± 1.13% per decade for the 200 and 250 °C signals, respectively. The latter g value is consistent with zero at 1σ, so we used the D value obtained from the 250 °C signal to date each of the samples. We note, however, that the g values for the 200 and 250 °C signals have large uncertainties, owing to the difficulty in obtaining precise estimates at low fading rates, so our data do not exclude the possibility that the high-temperature signals may fade slightly. On the basis of the results of the performance tests described above, the MET-pIRIR procedure in Supplementary Table 6 was used to estimate the D values for all four TUT samples, as well as one sample (TLT-OSL6) collected from near the base of the stratigraphically underlying deposits in the TLT. The D estimates obtained for the TUT samples using the MET-pIRIR 250 °C signal are shown in Extended Data Fig. 8. Most of the estimates are distributed around a central value, although the spread is larger than can be explained by the measurement uncertainties alone. The overdispersion among these D values is ~20% for three of the TUT samples and almost twice this amount for TUT-OSL9, the latter arising from a pair of low D values measured with relatively high precision. To estimate the age for each of these samples, we determined the weighted mean D of the individual single-aliquot values using the central age model49, which takes account of the measured overdispersion in the associated standard error. As a further test of the reliability of our D estimates for the TUT samples, we have plotted the estimates of the central age model as a function of stimulation temperature in Extended Data Fig. 9a. These plots show that the D values increase with stimulation temperature until a ‘plateau’ is reached at higher temperatures for each of the TUT samples; the plateau region (marked by the dashed line) indicates that a non-fading component is present at these elevated temperatures. The existence of a plateau can be used, therefore, as an internal, diagnostic tool to confirm that a stable, non-fading component has been isolated for age determination. For all four TUT samples, a plateau is reached at temperatures of 200 °C and above, from which we infer negligible fading of the MET-pIRIR 250 °C signal. We calculated the sample ages, therefore, using the D values obtained from the 250 °C signal. The corresponding weighted mean D values, dose rate data and final ages are listed in Supplementary Table 4. For sample TLT-OSL6 from the TLT, four of the eight aliquots measured emitted natural MET-pIRIR 250 °C signals consistent with the saturation levels of the corresponding dose–response curves (for example Extended Data Fig. 9b). This implies that the IRSL traps were saturated in the natural sample, which further supports our conclusion that the MET-pIRIR 250 °C signal had a negligible fading rate. It would be hazardous to estimate the age of sample TLT-OSL6 from the D values of the four non-saturated aliquots, as these may represent only the low D values in the ‘tail’ of a truncated distribution. If we adopt the average 2.3D value for the MET-pIRIR 250 °C signal of all five Talepu samples (~820 Gy) as an upper limit for reliable D estimation, then this corresponds to a minimum age of ~195 kyr for sample TLT-OSL6 (Supplementary Table 4). The MET-pIRIR 250 °C ages for the four samples dated from the TUT (=T2) are in correct stratigraphic order, increasing from 103 ± 9 kyr (at ~3 m depth) to 156 ± 19 kyr (at ~10 m depth). They thus span the period from marine isotope stage 6—the penultimate glacial—to marine isotope stage 5, the last interglacial. This coherent sequence of ages also supports our contention that the Talepu samples were sufficiently bleached before deposition. The sample analysed from ~8 m depth in the TLT (=T4; sample TLT-OSL6) yielded a minimum age of ~195 kyr, corresponding to marine isotope stage 7 (the penultimate interglacial) or earlier. We have not yet dated the other sediments exposed in the TLT, but expect that the 6 m of deposit immediately overlying TLT-OSL6 will be older than 156 ± 19 kyr, as they stratigraphically underlie sample TUT-OSL9 in the TUT. We interpret the ages for the TUT samples as true (finite) depositional ages, based on the existence of D plateaux (Extended Data Fig. 9a) and the increase in D with depth (that is, ordered stratigraphically). This is the most parsimonious reading of our data. The measured fading rate of 0.17 ± 1.13% per decade for sample TUT-3 allows for the possibility, however, that the MET-pIRIR 250 °C signal may still fade slightly and that our samples had reached an equilibrium state of trap filling and emptying (so-called field saturation54). If so, then the increase in D with depth could, instead, be due to a systematic decline in fading rate with increasing depth. Any such a trend cannot be verified or rejected from laboratory measurements of the g value, owing to the size of the associated uncertainties at low fading rates (Extended Data Fig. 7c, d). The ages for the TUT samples could, therefore, be viewed conservatively as minimum ages (as for sample TLT-OSL6), given the uncertainties in the measured fading rate of the 250 °C signal and the exact level at which the signal saturates. The measured age of the uppermost sample in the sequence, TUT-OSL1, would increase by about 15% and 40% after correcting39, 55, 56 for assumed fading rates of 0.5 and 1% per decade, respectively. Similarly, the measured ages of TUT-OSL2, -3 and -4 would increase by about 17, 23 and 28%, respectively, after correcting for an assumed fading rate of 0.5% per decade. Thus, whether viewed as true ages or as minimum ages, the TUT sediments were deposited more than ~100 ka. Samples for palaeomagnetic polarity assessment were taken from the baulks of excavations Talepu 2 (T2) and Talepu 4 (T4) (Fig. 2). Samples were taken at 20–30 cm intervals using non-magnetic tools. Preferably samples in non-bioturbated silty deposits were taken. The upper conglomeratic interval of T2 was omitted because of its coarser grain size and because it appeared heavily affected by soil formation and plant root bioturbation. From each sample level, five oriented sample specimens were retrieved by carving the sediment using non-magnetic tools and fitting them into 8 cm3 plastic cubes. The samples were labelled according to excavation, baulk and depth. In the laboratory all specimens were treated by an alternating field demagnetizer. The mean magnetic directions for each sample are presented in Supplementary Table 7. Demagnetization was performed with intervals of 2.5–5 mT to a peak of up to 80–1,000 mT. The magnetization vectors obtained from most samples showed no more than two separated components of natural remanent magnetization (NRM) on the orthogonal planes, which means that the specimens had been affected by secondary magnetization. However, secondary magnetization was easily removed with a demagnetization of up to 5–20 mT, while the characteristic remanent magnetizations (ChRMs) could be isolated through stepwise demagnetization of up 20–40 mT, in some cases up to 50 mT. Above 40 mT most samples were completely demagnetized (Extended data Fig. 4j and Supplementary Table 8). The mean magnetization intensities and palaeomagnetic directions are plotted against stratigraphic depth in Extended Data Fig. 5. The 90–98% intensity saturation was achieved from 1.30 × 10−4 to 3.81 × 10−3 A m−1 before demagnetization, and between 8.52 × 10−6 and 1.49 × 10−4 A m−1 after demagnetization at 20–40 mT. The direction of ChRMs is determined from the orthogonal plots in at least four or five successive measurement steps between 20 and 50 mT using principal component analysis57 (PuffinPlot58 and IAPD 2000 software59) with the maximum angular deviations setting at <5°. Although there are no well-defined criteria for the acceptability of palaeomagnetic data available, the k > 30 and α95 < 15° criteria of ref. 60 were used to accept the average remanence direction for sampled levels. On the basis of these tests, all the samples (n = 24) throughout the Talepu sequences yielded acceptable ChRMs directions and showed a normal polarity. The ChRM directions were relatively constant throughout the sequences, except the direction of samples taken in T2 at 6.5 and 7.5 m depth, which showed steep inclinations of 56–68°. Such steep inclinations are unusual for near-equatorial regions. One possible interpretation is that post-depositional mass-movement disturbances, such as creep or a landslide, resulted in rotational movements of this interval. The equal-area projections show that the dispersion of within-site means of the remanence directions re-group more closely together after demagnetization, and no significant change in the major remanence direction occurs with depth. The major remanent direction corresponds closely with the present magnetization direction (Extended Data Fig. 5b). Sample TAL-10-01 was taken from T4, sub-unit E , at a depth of 2.5 m below the surface. Euhedral sanidine crystals up to 250 μm in length were hand-picked following standard heavy liquid and magnetic separation techniques. Crystals were loaded into wells in aluminium sample discs (diameter 18 mm) for neutron irradiation, along with the 1.185 Myr Alder Creek sanidine61 as the neutron fluence monitor. Neutron irradiation was done in the cadmium-shielded CLICIT facility at the Oregon State University TRIGA reactor. Argon isotopic analyses of gas released by CO laser fusion of single sanidine crystals (Supplementary Table 9) were made on a fully automated, high-resolution, Nu Instruments Noblesse multi-collector noble-gas mass spectrometer, using procedures documented previously1, 62. Sample gas clean-up was through an all-metal extraction line, equipped with a −130 °C cold trap (to remove H O) and two water-cooled SAES GP-50 getter pumps (to absorb reactive gases). Argon isotopic analyses of unknowns, blanks and monitor minerals were performed in identical fashion.40Ar and 39Ar were measured on the high-mass ion counter, 38Ar and 37Ar on the axial ion counter and 36Ar on the low-mass ion counter, with baselines measured no less than every third cycle. Measurement of the 40Ar, 38Ar and 36Ar ion beams was performed simultaneously, followed by sequential measurement of 39Ar and 37Ar. Beam switching was achieved by varying the field of the mass spectrometer magnet and with minor adjustment of the quad lenses. Data acquisition and reduction was performed using the program ‘Mass Spec’ (A. Deino, Berkeley Geochronology Center). Detector intercalibration and mass fractionation corrections were made using the weighted mean of a time series of measured atmospheric argon aliquots delivered from a calibrated air pipette. Decay and other constants, including correction factors for interference isotopes produced by nucleogenic reactions, are as reported in ref. 62. The resulting age probability diagram for single sanidine crystals (Extended Data Figure 10) shows a wide range in ages with a dominant population around 9.4 million years ago (Late Miocene). This indicates that the sanidine crystals from the sample do not represent a single volcanic event, but were predominantly derived from erosion of the Miocene volcanic rocks west of the Walanae Depression and/or from Late Miocene marine sediments of the Walanae Formation.
Brumm A.,Griffith University |
Brumm A.,University of Wollongong |
Van Den Bergh G.D.,University of Wollongong |
Storey M.,Copenhagen University |
And 25 more authors.
Nature | Year: 2016
Recent excavations at the early Middle Pleistocene site of Mata Menge in the So'a Basin of central Flores, Indonesia, have yielded hominin fossils attributed to a population ancestral to Late Pleistocene Homo floresiensis. Here we describe the age and context of the Mata Menge hominin specimens and associated archaeological findings. The fluvial sandstone layer from which the in situ fossils were excavated in 2014 was deposited in a small valley stream around 700 thousand years ago, as indicated by 40Ar/39 Ar and fission track dates on stratigraphically bracketing volcanic ash and pyroclastic density current deposits, in combination with coupled uranium-series and electron spin resonance dating of fossil teeth. Palaeoenvironmental data indicate a relatively dry climate in the So'a Basin during the early Middle Pleistocene, while various lines of evidence suggest the hominins inhabited a savannah-like open grassland habitat with a wetland component. The hominin fossils occur alongside the remains of an insular fauna and a simple stone technology that is markedly similar to that associated with Late Pleistocene H. floresiensis. © 2016 Macmillan Publishers Limited. All rights reserved.
Meijer H.J.M.,Institute Catala Of Paleontologia Miqual Crusafont |
Meijer H.J.M.,Smithsonian Institution |
Kurniawan I.,Geology Museum |
Setiabudi E.,Geology Museum |
And 5 more authors.
Palaeogeography, Palaeoclimatology, Palaeoecology | Year: 2015
The So'a Basin in central Flores contains the earliest known evidence for hominins on this remote Indonesian island, with stone tools from this region dating back at least 1.0 Ma. At least 16 late Early to early Middle Pleistocene terrestrial fossil localities record evidence for highly insular endemic faunas and proxy evidence for hominins. The hominin presence in the So'a Basin raises many questions as to how, when, and from where hominins arrived on the island, and key issues regarding their adaptation to this insular environment remain poorly understood. Here, we provide palaeoenvironmental data based on avian remains recovered during recent, systematic excavations of late Early to early Middle Pleistocene deposits at Mata Menge and Bo'a Leza in the So'a Basin. At least six species of birds in five avian orders were recovered, forming the oldest bird remains recovered from Wallacea. Two species, a swan (Cygnus sp.) and an eagle owl (Bubo sp.), no longer occur on Flores. The avian assemblage from Mata Menge and Bo'a Leza described here is indicative of an open environment with a strong open, freshwater component and nearby grasslands, and with forests at a distance. In that, it differs from the Late Pleistocene limestone cave site of Liang Bua, the find locality of Homo floresiensis, and is more similar to the Early Pleistocene Homo erectus sites of the Sangiran Dome on Java. Although little is known regarding the identity of the So'a Basin toolmakers, the presence of an open, savannah-type environment in the late Early to early Middle Pleistocene of Flores may have facilitated their dispersal from the Asian shelf into Wallacea. © 2015 Elsevier B.V.