Pb age data (and their limitations)

Pb age data (and their limitations)

Earth and Planetary Science Letters 187 (2001) 131^145 www.elsevier.com/locate/epsl Timing of the Permian^Triassic biotic crisis: implications from n...

348KB Sizes 1 Downloads 17 Views

Earth and Planetary Science Letters 187 (2001) 131^145 www.elsevier.com/locate/epsl

Timing of the Permian^Triassic biotic crisis: implications from new zircon U/Pb age data (and their limitations) Roland Mundil a; *, Ian Metcalfe b , Kenneth R. Ludwig a , Paul R. Renne a;e , Felix Oberli c , Robert S. Nicoll d a

Berkeley Geochronology Center, 2455 Ridge Rd., Berkeley, CA 94709, USA Asia Centre, University of New England, Armidale, NSW 2351, Australia Institute of Isotope Geology and Mineral Resources, ETH-Zentrum NO F, CH-8092 Zu«rich, Switzerland d Department of Geology, Australian National University, Canberra, ACT 0200, Australia e Department of Earth and Planetary Science, University of California, Berkeley, CA 94720, USA b

c

Received 17 July 2000; received in revised form 2 February 2001; accepted 12 February 2001

Abstract The most profound biotic crisis in the Earth's history, causing the near extinction of both terrestrial and marine life, occurred at the end of the Permian period about 253 Myr ago and marks the Paleozoic^Mesozoic era boundary. The cause of this event is still a matter of vigorous debate, with both brief and catastrophic as well as gradual mechanisms having been proposed. Similar to a recent landmark study, this study uses the U^Pb method on zircons from the uppermost Permian/lowermost Triassic ash fall deposits at Meishan (Zhejiang Province, SE China) in order to examine time and rate constraints for these events. The results of both this study and previous work show that for these ash layers, the effects of Pb loss are combined with varying amounts and sources of inheritance, resulting in an age scatter which prohibits the extraction of a statistically robust age in many cases. Though the effects of Pb loss on the zircons analyzed in this study were reduced by leaching the grains in hydrofluoric acid (as opposed to commonly applied air abrasion) prior to analysis, the presence within a single ash layer of multiple generations of older xenocrysts (in many cases only slightly older than the depositional age) has made quantitative interpretation even more difficult. When these combined phenomena bias individual zircon ages by less than a percent, they are extremely difficult to deconvolute, and, if multi-grain analyses are used, can become impossible to recognize (because of the resulting age averaging). Monte Carlo simulations using actual measurements of individual zircon crystals show that age excursions due to Pb loss and xenocrystic contamination for the Meishan bentonites are easily homogenized to the point of undetectability when replicate analyses of multi-grain zircon samples are compared. Thus this study uses only high-precision analyses of single crystals, whether from our work or that of previous studies. Three main conclusions have emerged. First, our data require a significant increase in the age of the Permian^Triassic boundary by more than 2 myr compared to the previous study, which shifts the age to a value older than 253 Ma. Second, neither our data nor those from previous work can confirm or negate the possibility of a very abrupt biotic crisis. Third, even large suites of very high-quality, single-zircon U^Pb analyses for these tuffs cannot, in most cases, yield objective, reliable, and robust dates with

* Corresponding author. Fax: +1-510-644-9201 E-mail addresses: [email protected] (R. Mundil); [email protected] (I. Metcalfe); [email protected] (K.R. Ludwig); [email protected] (P.R. Renne); [email protected] (F. Oberli); [email protected] (R.S. Nicoll). 0012-821X / 01 / $ ^ see front matter ß 2001 Elsevier Science B.V. All rights reserved. PII: S 0 0 1 2 - 8 2 1 X ( 0 1 ) 0 0 2 7 4 - 6

EPSL 5790 6-4-01

132

R. Mundil et al. / Earth and Planetary Science Letters 187 (2001) 131^145

accuracies at the sub-myr level ^ though the temptation to perform arbitrary selection of subsets of the analyses for that purpose is almost irresistible. The last conclusion is not an indictment of zircon U/Pb dating in general (other rocks and other zircon populations can ^ and do ^ behave very differently), and further technical advances will likely improve our ability to prepare grains or sub-grains of adequately enhanced quality for analysis. Consequently, the results of the present study strongly suggest that for problems requiring time-scale accuracy, inferences from zircon U^Pb dating must be based on sufficiently large suites of single-crystal or crystal domain, high-precision analyses ( 6 1% error) that are realistically interpreted. ß 2001 Elsevier Science B.V. All rights reserved. Keywords: Permian^Triassic boundary; mass extinctions; U/Pb; zircon

1. Introduction In order to address questions related to brief events in Earth's history, such as the Permian^ Triassic (P^T) biotic crisis, high-resolution isotopic age data with both precision and accuracy at the permil level are desirable [1,2]. The most promising objects for the age calibration of brief events are volcano-derived airborne deposits interbedded within bio- and/or magnetostratigraphically calibrated sedimentary successions. During the past several years, re¢nement of analytical techniques has permitted analytical precision of individual U^Pb analyses at well below the one percent level1 . It is not always evident, however, that this level of precision can be transferred to the age assignment of a single ash layer. In this study, high-precision IDTIMS (isotope dilution thermal ionization mass spectrometry) U^Pb analyses on single zircon crystals were applied to several volcanic ash fall horizons across the P^T transition exposed in the Meishan sections (SE China). Zircon, which occurs as a trace constituent in many intermediate to silicic volcaniclastics and is considered chemically and mineralogically stable in the sedimentary environment, is highly suitable for the purpose of dating sedimentary sequences given that its crystallization age closely approximates the times of volcanic eruption and deposition. Complications arising from the presence of older cores on which juvenile zircon has nucleated can usually be minimized if single-grain or crystal fragments are an-

1

Uncertainties here and elsewhere herein are given at the 2c or 95% con¢dence level.

alyzed with prior careful selection by microscopic inspection and cathodo-luminescence imaging techniques. Analysis of single zircon crystals containing cryptic, undetected older cores typically produces Pb/U isotopic ratios which are resolved (generally because of signi¢cant 206 Pb/238 U, and frequently 207 Pb/235 U age discordance) from those of samples devoid of inheritance, and are thus readily rejected. If the e¡ects of secondary loss of radiogenic Pb are minimized by the application of suitable techniques (both mechanical and chemical) prior to dissolution, single-crystal analyses from a volcaniclastic layer ideally yield a tight cluster of Pb/U ages, thus de¢ning a precise depositional age. If laboratory contamination is kept su¤ciently low, 206 Pb/238 U age precisions of better than þ 1 Ma (95% con¢dence level) can be attained for individual analyses on single, Permian age zircon crystals containing as little as 1.2 pg (1 pg = 10312 g) of radiogenic Pb. Because for such zircon crystals the precision of the radiogenic 207 Pb/235 U and 207 Pb/206 Pb ratios are relatively poor (207 Pb being about 20 times less abundant than 206 Pb at this age), the mean age has to be calculated from the individual 206 Pb/238 U ages alone. Only a few marine sections which span the P^T transition are both bio- and/or magnetostratigraphically calibrated and contain interstrati¢ed volcaniclastic deposits which allow a chronostratigraphic calibration. Most of these sections are located in southeast Asia on the margins of the former Tethys ocean. Marine P^T sections are rare because most ocean £oor of that age has been subducted and there is a major eustatic sea level low at the P^T transition. The P^T sections of Meishan have been studied in detail and a

EPSL 5790 6-4-01

R. Mundil et al. / Earth and Planetary Science Letters 187 (2001) 131^145

widely accepted biostratigraphic time-scale has been established there, based mainly on conodonts [3]. Depending on the possible causes, mass extinctions may occur over substantially di¡erent extents of time. It is thus essential to examine both the synchronicity with potential causes and the tempo of the extinction [4,5]. For example, geologically brief events such as massive £ood-basalt volcanism [6] or a bolide impact [7,8] would result in geologically rapid environmental extrema, causing a temporally abrupt extinction pattern. In contrast, more gradual events such as the overturn of a strati¢ed ocean resulting in the poisoning of shelf areas with CO2 -rich waters [9^11], transgression-associated anoxia [12,13,14] and environmental extrema associated with a major marine regression should lead to a more gradual extinction [4,15]. However, only very few reliable, high-resolution age determinations exist which constrain the timing of these processes [6]. 2. Geological and biostratigraphical overview During the Permian^Triassic transition period, the Meishan locality (Fig. 1) was located in an o¡shore area at the northern margin of the South China Block, in a shallow marine environment at low latitudes. In the latest Permian (Changhsingian stage), carbonate sedimentation prevailed at

Fig. 1. Location of the sampled quarries near Meishan, South China (Changhsingian stage in gray, modi¢ed from [3]).

133

Meishan, interrupted by episodic volcanic ash fall deposits derived from explosive, acidic volcanism whose source is unknown. Shortly after the P^T transition, carbonate sedimentation ceased and the sedimentary environment became dominated by ¢ne-grained clastic deposition (mudstones). In marine sequences which span the P^T boundary, a transitional zone has been identi¢ed wherein the earliest typical Mesozoic fossils coexist with Paleozoic relict faunas [4,5,9,15]. The global Permian^Triassic boundary stratotype (reference section) has been selected by the Permian^Triassic Boundary Working Group [16] at section D, Meishan, South China [3] with the boundary between the Permian and Triassic, de¢ned by the ¢rst appearance of the conodont microfossil species Hindeodus parvus [17]. The P^T extinction of marine faunas in South China has been shown to have occurred in three pulses. The most signi¢cant pulse occurred in the uppermost Permian (at the base of bed 24e), with the other two placed in the uppermost Permian and earliest Triassic (at the bases of Meishan beds 25 and 28 at section D; Fig. 4). 3. Results from previous geochronologic studies at Meishan Most attempts to isotopically date the Permian^Triassic boundary have been applied to minerals from a bentonite (bed 25 of [3]) immediately below the biostratigraphic boundary. Zircons from this horizon were analyzed by SHRIMP (sensitive high-resolution ion microprobe) and yielded a 206 Pb/238 U age of 251.2 þ 3.4 Ma [18]. A 40 Ar/39 Ar age of 249.9 þ 0.3 Ma was reported for sanidine from this same unit [6]; however, as noted by [19], there is evidence for bias between the U^Pb and 40 Ar/39 Ar systems. In order to properly compare the two systems, consideration of potential systematic errors would expand the 40 Ar/39 Ar age error to þ 4.6 Ma as discussed for this speci¢c case by [20]. In a very extensive and detailed study, IDTIMS U/Pb zircon ages were recently published [2] for six bentonites in P^T boundary sections at Meishan. In that study, the ages were based on both

EPSL 5790 6-4-01

134

R. Mundil et al. / Earth and Planetary Science Letters 187 (2001) 131^145

multi- and single-grain analyses of abraded crystals. In order to compare these ages with the contrasting results obtained in this study and to illustrate the complexity of the samples, it is necessary to discuss and compare the results obtained by [2] in detail, in particular those which have been extracted from the same horizons or similar stratigraphic levels sampled for this study. For this purpose we con¢ne data selection to single-crystal analyses with precisions better than þ 1% on the 206 Pb/238 U age from the previous study, as multigrain data in contrast to single-grain data tend to disguise rather than expose complications potentially hidden in zircon populations. In most cases, however, these single-grain data scatter beyond their analytical error which either indicates that analytical uncertainties have been underestimated (unlikely, we think) or that the data sets are biased by inheritance and/or Pb loss. The singlecrystal data by [2] are plotted together with the results of this study in concordia space (Fig. 2). Stratigraphic level numbering follows [3], and meter positions are given with respect to the biostratigraphic boundary. 3.1. Bed 7, 336.8 m

Fig. 2. Concordia plots (using [30]) and 206 Pb/238 U ages (sorted by age) showing the results of single-zircon analyses obtained in this study and [2]. Error ellipses (gray, this study; white, [2]) and error bars (black, this study; gray, [2]) are plotted at the 95% con¢dence level.

[2] assigned an age of 253.4 þ 0.2 Ma to a layer (MD96-7) close to the base of the Changhsingian stage (i.e. close to D16 from this study, see Fig. 4), using both multi-grain and single-grain analyses. Arbitrary rejection of both older and younger ages (as the result of xenocrystic inheritance and Pb loss, respectively) was required to obtain coherence within analytical errors. If only the singlegrain analyses are considered, however, the data strongly suggest that the zircon crystals were affected by signi¢cant amounts of Pb loss, in spite of treatment by air abrasion. If this interpretation is correct and if the single-grain analyses are free of xenocrystic components, the oldest 206 Pb/238 U age (254.7 þ 0.2 Ma) may be considered as a minimum age for this horizon (Fig. 2). 3.2. Bed 22, 34.3 m The age assigned by [2] to a bentonite 4.3 m below the P^T boundary is 252.3 þ 0.3 Ma. The

EPSL 5790 6-4-01

R. Mundil et al. / Earth and Planetary Science Letters 187 (2001) 131^145

distribution of the corresponding single-grain analyses from that study show no scatter in excess of analytical error, and therefore the likely absence of both xenocrystic contamination and effects of Pb loss. If the stringent criterion of minimal non-analytical scatter for replicate singlegrain analyses is applied, this age is the only one within the crucial time interval (at or close to the main extinction pulse) from the entire suite published by [2] which is statistically robust (our study did no better, with only a single statistically robust age). The stratigraphic level of this bed was not sampled for this study (it has only been detected in section Z, where the uppermost part of the Changhsingian stage is exposed). 3.3. Bed 25, 30.18 m For bed 25 (MAW-b25, same stratigraphic level as bed D3t), [2] reported an age of 251.4 þ 0.3 Ma, derived from ¢ve coherent analyses with partially overlapping error ellipses out of a total of 20 multi- and single-crystal analyses, rejecting analyses with both younger and older ages. This choice was made in order to maintain the relationship between stratigraphic order and age. An equally plausible age assignment of s 252 Ma can also be inferred from these data, as the authors noticed a second, older data cluster at 252.7 þ 0.4 Ma, which was, however, attributed to inheritance (cf. Fig. 3C of [2]). The distribution of the single-crystal data shown in Fig. 2 suggests that they are severely biased by Pb loss. 3.4. Bed 28, + 0.08 m For Meishan bed 28 (MZ96-(+0.17) = AW3 in this study), a few centimeters above the Permian^ Triassic boundary, an age of 250.7 þ 0.3 Ma was reported by [2]. However, the single-crystal analyses scatter beyond their analytical uncertainty suggesting that the oldest single-grain age can be considered a minimum age for this horizon, as there is no obvious suggestion of xenocrystic contamination from the single-crystal data set displayed in Fig. 2. This example demonstrates the biasing towards a younger age if multi-grain anal-

135

yses are applied and Pb loss is the main biasing factor. 3.5. Bed 33, + 1.75 m An age of 250.4 þ 0.2 Ma was assigned by [2] to bed 33, which is based on seven coherent analyses (six single-crystal and one multi-crystal analyses). If single-grain data alone are considered, the set appears to be slightly biased by Pb loss. 3.6. Ash layer between beds 34 and 35, + 7 m [2] report an age of 250.2 þ 0.2 Ma for this horizon (MD96-293w = D1 in this study). However, this age is entirely based on multi-grain analyses, whose averaging e¡ects can easily result in a biased age from Pb loss and/or xenocrystic contamination, yet display good reproducibility because of the inherent averaging of the age distribution of a zircon population. Complications at the scale documented for the Meishan zircons by the single-grain analyses of both ourselves and [2] can only be recognized if single-grain IDTIMS analyses are applied. Certainly for this particular horizon, as documented later in this paper, the zircon population is characterized by non-analytical scatter, possibly mainly from xenocrystic contamination. 4. Sampling sites and U/Pb zircon ages (this study) Samples were collected from P^T boundary sections in Meishan (Fig. 1) as well as Dong Chuan Ling (Anhui Province) with guidance of researchers from the Nanjing Institute of Geology and Palaeontology and the assistance of local geologists. Also, the sections have been logged independently and then related to bed numbers of previous workers (compiled in [3]). The P^T boundary section at Meishan is exposed in ¢ve closely spaced limestone quarries near the town of Meishan (Changhsing County, Zhejiang Province). Quarry D serves as the type locality for the Changhsingian stage and the Changhsing formation (Figs. 1 and 4). The samples analyzed in this

EPSL 5790 6-4-01

136

R. Mundil et al. / Earth and Planetary Science Letters 187 (2001) 131^145

study were collected from section D (GSSP for the de¢nition of the P^T boundary) and section A west, respectively (Figs. 1 and 4). As discussed above, IDTIMS single-grain zircon U/Pb ages obtained from most of the Meishan P^T bentonites are di¤cult to interpret, and display complex age patterns within individual layers. In most cases, the e¡ects of subtle Pb loss are combined with varying amounts of inheritance. If inheritance is present in the form of cores, such grains escaping detection during crystal selection may typically be recognized because of the resulting pronounced 206 Pb/238 U^207 Pb/235 U age discordance. However, the presence within a single horizon of multiple generations of older xenocrysts (in many cases only slightly older than the depositional age) complicates age distribution to a degree which renders objective assignment of an age di¤cult or even impossible. When these phenomena bias individual zircon ages at the sub-percent level they are extremely di¤cult to deconvolute, and are impossible to recognize if multi-grain analyses or individual analyses with low precision were used. The zircon crystals were carefully selected by microscopic inspection in transmitted light in order to avoid recognizably inherited components. Zircons from samples D1 and D10/D15 were extracted from polished mounts after cathodo-luminescence imaging. However, the CL images did not make it possible to distinguish between zircons of the original population and xenocrysts. Most grains were subjected to leaching in hot (V80³C), ultrasonically agitated HF prior to dissolution, which for these samples is demonstrably superior to air abrasion in minimizing Pb loss e¡ects (see Section 4.2 on D10/15 results, and also [21^23]), without disturbing the pristine parts of the crystal. The small sample size, in combination with low concentrations of U and the relatively young age, result in minimal amounts of radiogenic Pb for analysis, which require careful attention to analytical blanks. Replicate analyses of laboratory common Pb contamination yielded 1.3 þ 0.8 pg per analysis (for blank composition see Appendix), U blanks being indistinguishable from zero.

4.1. D16: ash layer between beds 3 and 4, 341 m Sample D16 is a 2 cm thick brownish clay which is highly weathered. It occurs near the base of the Baoqing Member, the lower unit of the Changhsing formation. Biostratigraphically, this unit is in the basal part of the Clarkina subcarinata conodont zone and about 7 m above the C. orientalis zone of the topmost Wuchiapingian stage (Fig. 4). Evidently, the crystals in this population with 206 Pb/238 U ages of less than 255 Ma are strongly a¡ected by Pb loss. Moreover, the excess scatter (beyond analytical error) of the remaining seven points (having excluded one crystal with a 207 Pb/ 206 Pb age of 493 Ma clearly documenting inheritance) demands the presence of either residual Pb loss, subtle xenocrystic contamination, or both. Because there is no obvious break in the distribution, the simplest interpretation is that of Pb loss only, in which case the age of the oldest crystal (257.2 þ 0.7 Ma) is the best estimate of the minimum age of the unit (Fig. 2). 4.2. D10/15: within bed 15, 317.3 m Sample D10/15 (8^9 cm thick) is a gray clay which is highly weathered and of yellowish color due to the oxidation of pyrite. Zircons from this horizon demonstrate the e¡ects induced by Pb loss and display further complications resulting from the likely presence of slightly older xenocrysts. Analyses on abraded grains (no HF leaching) are dominated by Pb loss, with four analyses clustering around 249 Ma but still being dispersed beyond analytical uncertainties (MSWD = 4.2). Although an age of 252.0 þ 0.4 Ma can be calculated from a coherent (MSWD = 1.3) cluster of ten analyses (eight on abraded and HF-leached zircons and two on abraded-only grains), this age is extremely di¤cult to reconcile with the results from the other analyzed horizons. From the bracketing layers an age of ca. 255 Ma would be expected for this horizon which, however, is represented by only one analysis (see also Section 5). Only analyses from further bracketing horizons and di¡erent locations, respectively, can help to solve this problem.

EPSL 5790 6-4-01

D16 Z02 Z08 Z05 Z07 Z03 Z04 Z10 Z06 Z09 Z12 Z11 Z13 D10/15 Z51 Z55 Z49 Z05 Z03 Z42 Z52 Z56 Z02 Z41 Z04b Z53 Z54 Z45 Z04a Z21 Z48 Z05 Z46 Z44 Z32 Z20 Z25

Sample

Table 1

1.4 3.0 2.6 2.3 2.5 2.2 2.6 2.0 1.2 2.5 2.5 1.7

0.4 0.9 1.2 1.1 1.7 1.6 0.9 0.7 1.7 2.9 0.9 1.3 0.9 0.7 1.0 1.4 1.4 0.9 4.5 3.4 0.7 5.9 1.1

A/HF A/HF A/C A/HF A/HF A A/HF A/HF A/HF A/C A/HF A/HF A/HF A A/HF A A/C A A/C A/C A A/C A

Wg zirc.

HF HF HF HF HF HF HF HF HF HF HF HF

a

81 189 244 164 84 203 242 220 174 154 288 108 180 309 308 243 198 249 302 240 170 245 280

457 122 179 226 207 321 302 261 757 213 203 370

ppm U

0.47 0.56 0.56 0.53 0.55 0.47 0.41 0.45 0.52 0.47 0.48 0.63 0.57 0.62 0.43 0.56 0.54 0.52 0.51 0.40 0.61 0.50 0.50

0.66 0.59 0.66 0.63 0.58 0.51 0.57 0.58 0.57 0.56 0.59 0.54

Th/Ub Pbr:

4.1 6.1 10.2 6.2 4.9 11.1 7.5 5.3 10.1 15.3 8.8 4.8 5.5 7.4 10.4 11.5 9.4 7.6 45.7 27.1 3.9 46.5 9.8

35.5 12.9 16.0 18.0 18.3 24.2 27.2 18.6 31.1 17.9 16.7 19.9

(pg)

206

1.1 1.2 2.1 1.4 1.0 1.4 1.0 1.1 1.3 2.1 1.4 1.1 1.3 1.5 2.9 1.1 2.6 0.9 3.0 3.9 1.0 4.0 1.2

2.0 2.8 3.0 1.5 3.0 7.6 1.6 2.3 1.6 1.3 1.5 1.5

(pg)

291 405 380 345 370 610 547 360 579 561 500 341 340 372 286 765 287 638 1125 535 297 873 621

1356 362 409 888 466 254 1267 610 1450 1002 825 974

Pbd 204 Pb

206

0.06974 0.06034 0.05173 0.05195 0.05062 0.05154 0.05097 0.05146 0.05132 0.05143 0.05173 0.05167 0.05115 0.05210 0.05154 0.05144 0.05142 0.05144 0.05154 0.05139 0.05214 0.05178 0.05121

0.05703 0.05171 0.05132 0.05163 0.05126 0.05170 0.05146 0.05159 0.05111 0.05130 0.05161 0.05154

Pbe 206 Pb

207

Pbcc: Isotopic ratios

3.0 2.8 3.2 4.3 3.4 1.9 2.3 3.4 2.5 2.1 2.3 4.0 3.6 3.2 5.0 1.5 4.5 2.1 1.0 2.2 4.2 1.3 2.0

0.6 1.6 1.6 0.7 1.3 2.5 0.5 1.2 1.0 0.6 0.7 0.6

þ %

Pbe

1.3992 0.3499 0.2882 0.2874 0.2787 0.2837 0.2804 0.2828 0.2820 0.2825 0.2838 0.2834 0.2805 0.2850 0.2805 0.2794 0.2788 0.2787 0.2783 0.2738 0.2773 0.2675 0.2599

0.4980 0.2902 0.2876 0.2890 0.2863 0.2880 0.2866 0.2871 0.2797 0.2771 0.2765 0.2644

235 U

207

3.24 3.01 3.37 4.55 3.57 2.05 2.43 3.58 2.66 2.26 2.47 4.34 3.77 3.41 5.34 1.63 4.73 2.22 1.08 2.29 4.48 1.39 2.13

0.91 1.71 1.90 0.82 1.41 2.70 0.58 1.25 1.11 0.72 0.76 0.64

þ %

Pbe

0.14551 0.04206 0.04041 0.04013 0.03993 0.03992 0.03990 0.03986 0.03985 0.03984 0.03979 0.03978 0.03978 0.03967 0.03947 0.03939 0.03932 0.03930 0.03916 0.03683 0.03857 0.03748 0.03680

0.06334 0.04070 0.04065 0.04060 0.04051 0.04040 0.04039 0.04036 0.03969 0.03918 0.03884 0.03721

235 U

206

0.46 0.31 0.43 0.54 0.37 0.35 0.25 0.45 0.27 0.33 0.29 0.70 0.32 0.40 0.52 0.30 0.51 0.27 0.18 0.22 0.42 0.16 0.42

0.66 0.27 0.81 0.40 0.25 0.51 0.29 0.21 0.29 0.37 0.23 9.17

þ %

0.61 0.68 0.53 0.57 0.57 0.46 0.60 0.54 0.62 0.49 0.55 0.48 0.68 0.56 0.62 0.46 0.59 055 0.47 0.63 0.65 0.56 0.45

0.75 0.48 0.52 0.57 0.46 0.42 0.57 0.48 0.41 0.59 0.46 0.45

bf Pbe

875.7 þ 4.1 265.6 þ 0.8 255.4 þ 1.1 253.6 þ 1.4 252.4 þ 0.9 252.4 þ 0.9 252.2 þ 0.6 252.0 þ 1.1 251.9 þ 0.7 251.8 þ 0.8 251.5 þ 0.7 251.5 þ 1.8 251.4 þ 0.8 250.8 þ 1.0 249.5 þ 1.3 249.1 þ 0.7 248.6 þ 1.3 248.5 þ 0.7 247.6 þ 0.4 244.3 þ 0.5 244.0 þ 1.0 237.2 þ 0.4 233.0 þ 1.0

395.9 þ 2.6 257.1 þ 0.7 256.8 þ 2.1 256.6 þ 1.0 256.0 þ 0.6 255.3 þ 1.3 255.3 þ 0.7 255.1 þ 0.5 250.9 þ 0.7 247.7 þ 0.9 245.6 þ 0.6 235.5 þ 0.4

238 U

206

Pbe

888.6 þ 28.8 304.6 þ 9.2 257.2 þ 8.7 256.5 þ 11.7 249.6 þ 8.9 253.6 þ 5.2 251.0 þ 6.1 252.9 þ 9.0 252.2 þ 6.7 252.6 þ 5.7 253.7 þ 6.3 253.4 þ 11.0 251.1 þ 9.5 254.6 þ 8.7 251.0 þ 13.4 250.2 þ 4.1 249.7 þ 11.8 249.6 þ 5.6 249.3 þ 2.7 245.7 þ 5.6 248.5 þ 11.1 240.7 þ 3.3 234.6 þ 5.0

410.4 þ 3.7 258.7 þ 4.4 256.7 þ .9 257.8 þ 2.1 255.7 þ 3.6 257.0 þ 6.9 255.9 þ 1.5 256.3 þ 3.2 250.4 þ 2.8 248.4 þ 1.8 247.8 þ 1.9 238.2 þ 1.5

235 U

207

Isotopic ages (Ma)

921 þ 61 616 þ 61 274 þ 73 283 þ 98 224 þ 78 265 þ 44 239 þ 53 261 þ 77 255 þ 58 260 þ 49 274 þ 53 271 þ 93 248 þ 82 290 þ 73 265 þ 116 261 þ 35 260 þ 102 261 þ 48 265 þ 23 259 þ 50 291 þ 96 276 þ 30 250 þ 45

493 þ 13 273 þ 37 255 þ 37 269 þ 15 253 þ 30 272 þ 58 262 þ 11 267 þ 27 246 þ 24 254 þ 13 268 þ 16 265 þ 13

Pbe 206 Pb

207

R. Mundil et al. / Earth and Planetary Science Letters 187 (2001) 131^145

EPSL 5790 6-4-01

137

EPSL 5790 6-4-01

Z17 Z06 Z.18 Z01 AW3 Z01 Z02 Z06 Z05 Z04 Z07 Z03

Z16 Z34

Z41 Z45 Z07 Z35

Z37

Z32

Z31

Z44 Z47 Z42 Z36

D3T Z13 Z46 Z33

Sample

HF HF HF HF HF HF HF

A HF NaOH/ HF HF HF HF NaOH/ HF NaOH/ HF NaOH/ HF NaOH/ HF HF HF A NaOH/ HF A NaOH/ HF A A A

a

Table 1 (continued)

0.8 1.4 1.0 4.3 1.1 5.2 2.1

1.1 0.7 0.7 2.9

1.7 4.0

4.4 3.9 0.5 2.3

519 251 348 262 386 108 263

42 67 193 281

213 228

177 129 58 397

304

139

2.6

2.6

159

122 125 101 119

122 77 108

ppm U

3.6

3.4 3.3 3.1 3.6

1.2 2.7 3.2

Wg zirc.

0.48 0.55 0.59 0.63 0.54 0.66 0.61

0.93 0.62 0.69 0.66

0.70 0.73

0.79 0.62 0.65 0.65

0.66

1.19

0.62

0.69 0.91 0.87 0.63

0.72 0.64 0.57

Th/Ub Pbr:

14.4 12.1 12.0 38.7 14.6 19.3 18.9

1.5 1.5 4.0 24.1

12.2 30.4

26.9 17.0 1.0 31.0

27.2

12.5

19.7

14.6 14.1 10.8 14.8

6.6 7.3 12.1

(pg)

206

2.2 1.8 1.9 1.9 2.0 2.8 1.8

2.1 2.0 2.5 6.3

3.6 0.8

0.8 1.3 1.5 1.1

3.3

1.2

1.3

2.4 1.9 2.3 1.3

5.3 1.3 1.2

(pg)

504 509 486 1494 546 524 808

72 74 136 297

265 2714

2509 979 66 2091

625

818

1102

465 558 365 883

108 431 742

Pbd 204 Pb

206

þ %

1.0

0.8

0.7

1.3 1.2 1.5 0.7

5.6 0.3

0.05168 0.05148 0.05165 0.05150 0.05187 0.05195 0.05147

2.3 2.3 2.4 0.9 2.1 2.4 1.5

0.05174 21.7 0.05520 19.5 0.05219 9.9 0.05228 4.0

0.05184 0.05142

0.05120 0.6 0.05123 1.0 0.05385 23.2 0.05133 0.4

0.05151

0.05150

0.05133

0.05193 0.05148 0.05185 0.05168

0.05449 12.5 0.05203 1.5 0.05146 1.1

Pbe 206 Pb

207

Pbcc: Isotopic ratios Pbe

0.2872 0.2841 0.2848 0.2837 0.2855 0.2858 0.2831

0.2709 0.2874 0.2509 0.2477

0.2814 0.2758

0.2823 0.2812 0.2942 0.2798

0.2846

0.2651

0.2843

0.2891 0.2865 0.2882 0.2971

0.3952 0.2917 0.2879

235 U

207

2.45 2.45 2.57 1.05 2.26 2.58 1.56

22.40 20.12 10.22 4.17

5.73 0.40

0.92 1.08 23.96 0.48

1.14

0.92

0.77

1.38 1.31 1.66 0.85

12.94 1.67 1.26

þ %

Pbe

0.04031 0.04002 0.03999 0.03995 0.03991 0.03991 0.03989

0.03797 0.03775 0.03487 0.03437

0.03936 0.03889

0.03999 0.03981 0.03962 0.03953

0.04007

0.04016

0.04017

0.04037 0.04036 0.04031 0.04029

0.05261 0.04065 0.04058

235 U

206

0.23 0.27 0.29 0.50 0.23 0.52 0.25

0.94 0.83 0.48 0.24

0.23 0.24

0.65 0.40 0.95 0.28

0.43

0.20

0.26

0.31 0.46 0.26 0.35

0.71 0.34 0.37

þ %

0.64 0.57 0.56 0.54 0.60 0.45 0.47

0.75 0.78 0.70 0.55

0.56 0.67

0.73 0.48 0.79 0.65

0.49

0.44

0.47

0.44 0.48 0.47 0.51

0.60 0.44 0.46

bf Pbe

254.8 þ 0.6 253.0 þ 0.7 252.8 þ 0.7 252.5 þ 1.3 252.3 þ 0.6 252.2 þ 1.3 252.1 þ 0.6

240.2 þ 2.2 238.9 þ 2.0 220.9 þ 1.1 217.8 þ 0.5

248.9 þ 0.6 246.0 þ 0.6

252.8 þ 1.6 251.6 þ 1.0 250.5 þ 2.4 249.9 þ 0.7

253.3 þ 1.1

253.8 þ 0.5

253.9 þ 0.7

255.1 þ 0.8 2551.1 þ 1.2 254.8 þ 0.7 254.6 þ 0.9

330.5 þ 2.3 256.9 þ 0.9 256.4 þ 0.9

238 U

206

Pbe

256.4 þ 6.3 253.9 þ 6.2 254.5 þ 6.5 253.6 þ 2.7 255.0 þ 5.8 255.3 þ 6.6 253.1 þ 3.9

243.4 þ 54.5 256.5 þ 51.6 227.3 þ 23.2 224.7 þ 9.4

251.7 þ 14.4 247.3 þ 1.0

252.5 þ 2.3 251.6 þ 2.7 261.9 þ 62.7 250.5 þ 1.2

254.3 þ 2.9

254.7 þ 2.3

254.0 þ 1.9

257.8 þ 3.6 255.8 þ 3.4 257.1 þ 4.3 256.3 þ 2.2

338.2 þ 43.8 259.9 þ 4.3 256.9 þ 3.2

235 U

207

Isotopic ages (Ma)

271 þ 53 263 þ 53 270 þ 55 263 þ 20 280 þ 49 283 þ 55 262 þ 33

274 þ 497 420 þ 435 294 þ 226 298 þ 92

278 þ 128 260 þ 7

250 þ 14 251 þ 22 265 þ 523 266 þ 8

264 þ 23

263 þ 19

256 þ 16

282 þ 29 263 þ 27 279 þ 35 271 þ 17

391 þ 281 287 þ 35 262 þ 26

Pbe 206 Pb

207

138 R. Mundil et al. / Earth and Planetary Science Letters 187 (2001) 131^145

EPSL 5790 6-4-01

HF HF HF HF HF A/HF HF HF HF HF HF HF A/HF A/HF HF

a

1.4 1.6 3.5 3.1 1.5 2.2 2.0 3.8 2.2 1.9 1.4 2.5 1.2 1.8 3.3

Wg zirc.

382 27 172 219 296 216 509 155 241 268 383 292 308 244 389

ppm U

0.66 0.66 0.60 0.60 0.65 0.69 0.65 0.60 0.67 0.62 0.63 0.62 0.63 0.70 0.70

Th/Ub Pbr:

19.5 1.6 20.8 23.4 15.3 16.4 36.1 20.3 18.2 17.5 18.4 25.3 12.6 15.0 43.7

(pg)

206

1.5 1.4 2.7 1.8 1.4 1.5 1.2 1.4 1.5 1.4 1.6 3.3 1.3 3.8 1.7

(pg) 950 104 594 972 829 826 2204 1058 918 928 844 589 745 305 1937

Pbd 204 Pb

206

þ %

0.05137 1.2 0.05334 12.5 0.05153 2.0 0.05166 1.2 0.05143 1.4 0.05066 1.9 0.05141 0.6 0.05155 1.1 0.05046 2.0 0.05139 1.3 0.05146 1.4 0.05172 2.1 0.05156 1.5 0.05157 3.9 0.05126 0.7

Pbe 206 Pb

207

Pbcc: Isotopic ratios Pbe

0.3008 0.3097 0.2861 0.2860 0.2846 0.2803 0.2843 0.2850 0.2779 0.2831 0.2827 0.2839 0.2817 0.2818 0.2799

235 U

207

1.66 13.24 2.16 1.29 1.51 2.06 1.20 1.21 2.19 1.43 1.47 2.25 1.65 4.13 0.74

þ %

Pbe

0.04248 0.04211 0.04027 0.04016 0.04013 0.04013 0.04011 0.04010 0.03995 0.03995 0.03984 0.03981 0.03963 0.03963 0.03960

235 U

206

1.03 1.01 0.62 0.27 0.19 0.28 1.01 0.14 0.45 0.38 0.18 0.34 0.29 0.35 0.25

þ % 0.67 0.77 0.45 0.46 0.52 0.52 0.86 0.54 0.45 0.43 0.55 0.46 0.46 0.68 0.48

bf Pbe

268.2 þ 2.8 265.9 þ 2.7 254.5 þ 1.6 253.8 þ 0.7 253.7 þ 0.5 253.6 þ 0.7 253.5 þ 2.6 253.4 þ 0.4 252.5 þ 1.1 252.5 þ 1.0 251.8 þ 0.4 251.6 þ 0.8 250.5 þ 0.7 250.5 þ 0.9 250.4 þ 0.6

238 U

206

Pbe

267.0 þ 4.4 273.9 þ 36.3 255.5 þ 5.5 255.4 þ 3.3 254.3 þ 3.8 250.9 þ 5.2 254.1 þ 3.1 254.6 þ 3.1 249.0 þ 5.4 253.1 þ 3.6 252.8 þ 3.7 253.7 þ 5.7 252.0 þ 4.2 252.1 þ 10.4 250.6 þ 1.8

235 U

207

Isotopic ages (Ma)

257 þ 28 343 þ 282 265 þ 45 270 þ 27 260 þ 33 226 þ 45 259 þ 14 266 þ 26 216 þ 47 258 þ 30 261 þ 32 273 þ 49 266 þ 35 267 þ 89 253 þ 15

Pbe 206 Pb

207

Uncertainties of individual ratios are given at the 2c level and do not include decay constant errors. Ratios involving 206 Pb are corrected for initial disequilibrium in 230 Th/238 U adopting Th/U = 4 for the crystallization environment. a NaOH/HF, ultrasonically leached in NaOH and/or HF (80³C) prior to dissolution; A, abraded; C, ion exchange chemistry applied. b Present day Th/U ratio calculated from radiogenic 208 Pb/206 Pb and age. c Total common Pb including analytical blank (analytical Pb blank is 1.3 þ 0.8 pg per analysis). d Measured value corrected for tracer contribution and mass discrimination (0.20 þ 0.06%/AMU). e Ratios and ages corrected for mass discrimination and fractionation, tracer contribution and common Pb contribution. f Correlation coe¤cient of radiogenic 207 Pb/235 U versus 206 Pb/238 U.

D01 Z64 Z65 Z69 Z71 Z70 Z59 Z68 Z72 Z66 Z67 Z63 Z63b Z60 Z43 Z62a

Sample

Table 1 (continued) R. Mundil et al. / Earth and Planetary Science Letters 187 (2001) 131^145 139

140

R. Mundil et al. / Earth and Planetary Science Letters 187 (2001) 131^145

Nonetheless, the systematics emerging from the D10/15 zircon data are useful for the detailed evaluation of the e¡ects of the HF leaching pretreatment : ¢rst, the e¡ects of Pb loss were signi¢cantly reduced (by 2% on average) compared to the abraded-only zircons; and second, there is no evidence for laboratory-induced age bias (e.g. by preferential leaching of U over Pb) in the HFleached zircons ^ they simply occupy the older part of the distribution of the unleached crystals (Fig. 2). 4.3. D3t: bed 25, 30.18 m Sample D3t, traditionally named the `boundary clay', occurs about 18 cm below the P^T boundary and has been the target of most isotopic dating studies related to the P^T boundary (Figs. 2 and 4). Lithologically, the horizon consists of a gray clay (yellow if weathered) with abundant pyrite which is covered by a black layer which probably represents the background sediment predominating after carbonate production was temporarily shut down by the ash fall deposit (bed 26). Initial analyses in this study were performed on abraded individual crystals from this unit. They are associated with large uncertainties, as the sample size is substantially reduced and the common Pb correction has a signi¢cant impact. It can be demonstrated though, that this treatment can yield a coherent (due to larger individual errors) but biased mean age, as inferred by the mean 206 Pb/238 U age of 249.0 þ 0.8 Ma (four analyses) which clearly is too young. The D3t data are an excellent example of an assemblage of a large number of high-quality single-zircon analyses (19, including the two obviously inherited, strongly discordant crystals listed in Table 1) which nonetheless defy any precise interpretation. There is no real plateau of statistically equivalent ages, and there is a slight jump in the ages of the two oldest (concordant) grains which look suspiciously, but not convincingly, like slightly older xenocrysts. All that one can conclude is that either the ash is older than about 257 Ma (the apparent age of the two oldest concordant grains) and thus being in con£ict with the

age of layer D16, or if these two grains are xenocrysts, that the ash is slightly older than 254 Ma old (the mean age of the ¢ve next oldest, statistically similar grains). A weak conclusion indeed, given the e¡ort involved (Fig. 2). 4.4. AW3: bed 28, + 0.08 m Bed AW3 is the ash layer closest to the biostratigraphic boundary and occurs between the Permian Changhsingian Fm. (limestone) and the Triassic Yinkeng Fm. (¢ne-grained clastics). Layer AW3 is about 4 cm thick and contains abundant phenocrysts of feldspar. Zircons from this horizon yield a statistically coherent age, de¢ned by a tight cluster of six concordant analyses with a weighted mean 206 Pb/238 U age of 252.5 þ 0.3 Ma (MSWD = 0.9; Fig. 2). The reproducibility of the individual ages suggests that the e¡ects of Pb loss were eliminated (or at least reduced to the low permil level). One analysis yielded a slightly older age but is resolved from the remaining cluster and can thus be rejected (as a likely xenocryst) with reasonable objectivity from the calculation of the mean. 4.5. D1: between beds 34 and 35, + 7.0 m Bed D1, a 5^12 cm thick grayish clay with abundant secondary pyrite, occurs within the lower Triassic Yinkeng Formation. Twelve partially overlapping error ellipses of zircons from this bed display 206 Pb/238 U ages ranging from 250.4 to 254.5 Ma, scattering well beyond their analytical error (Fig. 2). Although the oldest seven analyses scatter only slightly beyond their analytical error (253.5 þ 0.4 Ma, MSWD = 1.7), this age is discordant with its stratigraphic position, and might be biased by xenocrysts that are only slightly older than the true age of this layer (Fig. 2). 5. Discussion and conclusions The high precision of IDTIMS enables recognition of subtle biases due to both inheritance or xenocrystic contamination on the one hand, and Pb loss on the other. IDTIMS data on single

EPSL 5790 6-4-01

R. Mundil et al. / Earth and Planetary Science Letters 187 (2001) 131^145

grains from the zircon populations of this study (whether ours or other workers') typically reveal an age distribution whose scatter signi¢cantly exceeds analytical errors. Because no physical or chemical characteristic (save 206 Pb/238 U, and occasionally, 207 Pb/235 U age discordance) of these zircons permits a priori rejection of unsuitable analyses, the only remaining criteria for `bene¢ciation' of the data sets are statistical. However, even when `straightforward'-type methods are applied (such as ¢nding age `plateaus', rejecting analyses whose apparent ages suggest a subtle discontinuity in the age distribution as xenocrysts), the results are both unconvincing (when examined individually) and do not yield a consistent picture when they are compared to those from either different laboratories or di¡erent samples from a single laboratory. It is tempting to use stratigraphic order as a basis for deconvolution of the problem: that is, when one has reliable ages for beds that stratigraphically bracket an ash layer, to apply these ages as a constraint in interpreting the zircon apparent ages from the bracketed ash layer. Such an approach is certainly valid when each of the two bracketing beds is dated accurately ^ that is, without bias and with well-established uncertainties. Either Bayesian or Monte Carlo methods can then assist in improving the error estimate for the bracketed bed. However, to use the bracketing age constraints as an aid in choosing which singlezircon analyses to pool for calculating an accurate age for the bracketed bed is much more problematic. First, the criterion that the bracketing ages and their uncertainties must be reliable can be met neither for our data, nor, we argue, for those of previous studies. Second, the process cannot be used to reliably assess the uncertainty calculated for a bed's data set which shows clear evidence for the various complications (a signi¢cant fraction of zircons with more than a few permil of Pb loss; subtle xenocrystic contamination) that are so pervasive in the zircon populations analyzed for this study. In other words, one may use reliable bracketing ages (and uncertainties) to constrain the age of a bracketed layer. One may not use those ages to validate the statistical uncertainty

141

of a sub-population chosen to meet those criteria, unless the whole population has a near-uniform apparent age distribution and a size large enough to recognize such a distribution. It is evident from all previous data sets (both SHRIMP and IDTIMS) as well as the results from this study, that subtle inter- and intra-grain heterogeneities exist in most zircon populations from the Meishan bentonites, and that these heterogeneities are resolved only through (1) analysis of small samples (individual crystals or fragments thereof in combination with extremely low analytical blanks) in order to avoid averaging of discrete components and e¡ects, and (2) use of high-precision analytical methods to resolve the e¡ects of geological scatter at the percent level. The importance of the application of single-grain or fractional-grain as opposed to multi-grain analysis is illustrated in Fig. 3, which shows a synthetic multi-grain data set derived by combination of groups of 11 analyses randomly drawn from the actually measured single-grain data. This example, which is relevant to any zircon population with minor

Fig. 3. Arti¢cial multi-grain data set (gray) generated by Monte Carlo simulations from real zircon analyses on single abraded crystals (white) pooled to 11 grains. The inferred precise age of 249.0 Ma is in sharp contrast to the true minimum age of 252.0 Ma (arrow) from replicate real single analyses of HF-leached grains (see Fig. 2). Replicate simulations yielded coherent results (errors are assigned by randomly selecting from a realistically estimated error range; synthetic analyses were forced to fall on Concordia within their assigned errors).

EPSL 5790 6-4-01

142

R. Mundil et al. / Earth and Planetary Science Letters 187 (2001) 131^145

age dispersion arising from Pb loss or xenocrystic contamination, demonstrates that an apparently coherent and precise, but inaccurate age will commonly be calculated from a multi-grain sample simply due to homogenization of real grain-tograin dispersion. In many cases (especially in the later Phanerozoic as in the present case), the resulting biased 206 Pb/238 U and 207 Pb/235 U ages will be concordant, as shown in the example of Fig. 3, contributing to a false sense of security. One of the main ¢ndings of this study is that of a signi¢cant bias between our inferred ages and those of the previous study [2], which leads us to question their conclusions. Our results do not indicate that the entire Changhsingian stage is characterized by high sedimentation rates, and therefore contradict the concept of extreme rapidity for the late Changhsingian end Permian extinction. Also, the hypothesis of a very short (i.e. 165 kyr in [2]) duration of the negative N13 C excursion near the P^T boundary is severely undermined, as the time span between beds D3t and AW3 appears to be much longer than postulated by these authors. However, as many of the data obtained in the present study are no less complex (with regard to their lack of coherence) than the equivalent (i.e. single-grain) data of the previous study, we must emphasize that it is extremely dif¢cult to derive an unambiguous conclusion from the Meishan bentonites with respect to the tempo of mass extinction (Fig. 4). Nonetheless, the data of this study indicate that either the sedimentation rate (and thus the extinction rate) for the uppermost Permian is lower than for the rest of the Changhsingian stage, or that there is a depositional hiatus beneath layer AW3 (indicated by the lithological change from limestone to mudstone), if our age result for this layer is accurate. All of the available U^Pb data, however, indicate that the Permian^Triassic boundary, and thus the main stage of the mass extinction, must be slightly older than our IDTIMS age for bed AW3 (8 cm above the boundary) of 252.5 þ 0.3 Ma. The age for this sample is also older than the internally coherent IDTIMS U^Pb age reported by [24] for the Norils'k intrusion, which is structurally and genetically related to the Siberian Trap volcanism (the Siberian Trap £ood ba-

salts have been suggested to be an important cause of the P^T extinction, e.g. by [6]). However, because the interval between the emplacement of the Norils'k intrusion and the main stage of Siberian Trap volcanism is unknown, and the latter may have initiated signi¢cantly before the Norils'k intrusion [25,26], the available U^Pb data can neither convincingly support nor rule out a cause^e¡ect relation between trap volcanism and mass extinction. High-resolution 40 Ar/39 Ar data [6] from the traps and bed 25 in Meishan are indistinguishable and support the hypothesis of causal relationship. Further U/Pb ages directly measured on primary, i.e. non-redeposited, genuine members of the Siberian Trap volcanics are highly desirable, although di¤cult to obtain, as U bearing phases in basaltic rocks are rare. Furthermore, additional 40 Ar/39 Ar data for the ash fall horizons in Meishan and other localities would be of great use in further constraining the rates of both sedimentary processes and extinc-

Fig. 4. Detailed stratigraphy for the Permian^Triassic boundary interval at Meishan (South China) showing the ash layers and assigned ages, the three extinction levels (horizontal arrows) and the N13 C curve [3]. The column to the left shows the stratigraphic sequence exposed at the proposed Permian^ Triassic boundary stratotype, section D and section AW, with volcanic ash layers (black bands) and samples collected for isotopic age dating (D16, D10/15, D3t, AW3 and D1).

EPSL 5790 6-4-01

R. Mundil et al. / Earth and Planetary Science Letters 187 (2001) 131^145

tions at the P^T boundary, although they might be less useful for high-accuracy calibration in terms of absolute age because of uncertainties in the half-life and in standardization. The ¢ndings of this and other studies on volcanic ash fall deposits call for stringent criteria on zircon preparation and treatment of U^Pb data. Precise and statistically meaningful ages can only be obtained if the e¡ects of both Pb loss and contributions by older components can su¤ciently be minimized or recognized at an adequate level of con¢dence. The requirement of sub-percent analytical precision (to resolve the complications of minor degree of Pb loss and of contamination by slightly older zircon xenocrysts) argues against the application of the ion microprobe U^Pb technique to zircon populations similar to those from Meishan. However, we note that, though the problem of subtle xenocrystic contamination is inherently intractable by ion microprobe, at least to the extent that such contamination is unavoidable by the IDTIMS analyst, the magnitude of the Pb loss problem faced by the latter is not necessarily as great for the ion microprobe. The former consumes hundreds or thousands of nanograms of zircon for each analysis, and has inspected at high resolution only a surface of the analyzed crystal (if extracted from a polished grain mount), whereas the latter consumes just about a nanogram of a grain, whose inspected surface characterizes an immensely larger fraction of the analyzed zircon. Therefore, it is possible that the material selected for ion microprobe analysis could be su¤ciently superior (i.e. absence of cracks, inclusions, material with inferior polishing hardness) that, on average, the biasing e¡ects of Pb loss on the ¢nal age (calculated from the combination of many tens of spots) are smaller than for IDTIMS. The di¤culty, unfortunately, lies in testing this assertion, which has never been convincingly attempted (though protocols to do so exist ^ e.g. reliable IDTIMS and ion microprobe dating of a bed in two locations, one with signi¢cant Pb loss complications, and one without). In the light of the complexities of the U^Pb isotopic data from the Meishan bentonites, additional con¢rmation of the few less ambiguous re-

143

sults is needed in order to convincingly constrain the duration and tempo of the mass extinction. Additional age data from biostratigraphically as well as magnetostratigraphically calibrated P^T sections are essential. The P^T section of Shangsi (Guangyuang county, northern Sichuan Province) may be useful in this regard, as it comprises a longer time interval than the Meishan section, and contains numerous bentonites (likely to be altered volcanic ash) throughout the uppermost Permian stages as well as the lowermost Triassic. Acknowledgements This research was supported by the Ann and Gordon Getty Foundation (R.M., K.R.L. and P.R.R.) and by an Australian Research Council Large Grant to I.M. We wish to thank Wang Cheng-yuan from the Nanjing Institute of Geology and Palaeontology (Nanjing, China) for guidance in the ¢eld. The help of Hannelore Derksen, Abed Jaouni, James Lin and Martin Meier with mineral separation and in the chemistry lab is gratefully acknowledged. Reviews by Jim Mattinson and Sam Bowring helped to improve the manuscript.[SK] Appendix. Analytical procedures U^Pb age determinations for samples D16, D10/15, D3t, AW3 and D1 were performed at the Berkeley Geochronology Center. Initial analyses for sample D3t were performed at the Institute of Isotope Geology and Mineral Resources (IGMR) at ETH Zu«rich. Rock samples weighing from 4.5 to 24 kg were collected with great care in order to avoid contamination from adjoining layers. The bentonites were disintegrated in dilute acetic acid using an ultrasonic disintegrator. Mineral concentrates were then puri¢ed by standard heavy-liquid and magnetic separation techniques. Individual zircon grains were pre-selected using a binocular microscope. These grains were then examined by transmitted light microscopy and in some cases by cathodo-luminescence imaging. Only euhedral,

EPSL 5790 6-4-01

144

R. Mundil et al. / Earth and Planetary Science Letters 187 (2001) 131^145

clear grains devoid of optically recognizable cores were chosen for analysis in order to avoid inherited components. Grains with cracks or other damaged domains susceptible to Pb loss were also excluded from analysis. For further suppression of surface-correlated secondary lead loss, some of the crystals were subjected to stepwise air abrasion [27], which, however, did not su¤ciently control the problem. Most of the zircons were treated with hot (80³C), concentrated acids (ultrasonically agitated in the case of HF). However, for most samples Pb loss could not be entirely suppressed, even though experiments in our laboratory have shown that HF leaching is superior to air abrasion for these samples. Selected grains were long prismatic and showed characteristic features of volcanic zircons (skeletal growth). Th/U ratios were not a useful criterion to identify xenocrysts in the studied populations, probably because the xenocrysts crystallized from the same magmatic environment. The average U concentration for all concordant analyses was 240 ppm. Prior to dissolution the grains were: (1) cleaned in ultrasonically agitated ethanol; (2) immersed for 15 min in warm conc. HNO3 , (3) leached for 2 h in hot (80³C), ultrasonically agitated 50% HF+HNO3 where indicated (see Table 1), and (4) washed in clean HNO3 . Zircons were then transferred to Krogh-type PTFE capsules, and spiked with 205 Pb^233 U^235 U tracer solution. After addition of 200 Wl 50% HF+15 Wl conc. HNO3 to the capsules, the zircons were heated to 215³C for 5 days. After dissolution, the dried sample^tracer mixtures were taken up in 40 Wl of 8 M HCl+5Wl 0.25 M H3 PO4 , dried down, and loaded together with silica gel+H3 PO4 on outgassed Re ¢laments. Isotope ratios were determined on a Micromass Sector 54 mass spectrometer using a Daly-type ion counter positioned behind a WARP ¢lter. Pb (as Pb‡ ) and U (as UO‡ 2 ) were run sequentially on the same ¢lament. We cannot rule out the possibility of a systematic bias between results from our analyses and those conducted in other labs due to tracer calibrations, but any o¡set larger than several permil seems unlikely. The 205 Pb^233 U^235 U tracer solution used in this study was calibrated repeatedly

against solutions derived from certi¢ed standards of isotopically pure 206 Pb and natural U (NIST SRM-991 and CRM-145, respectively). To provide the highest level of con¢dence in the tracer calibration, the standard solutions were themselves checked against solutions derived from NIST-certi¢ed standards. The natural U standard was calibrated against solutions derived from NBL-CRM111-A (enriched in 233 U) and NBLCRM135 (enriched in 235 U), while the 206 Pb standard was calibrated against solutions derived from NIST Pb-wire standards SRM-981 (common Pb) and SRM-982 (equal-atom Pb). All of these calibrations (e¡ectively three independent calibrations each for U and for Pb) agreed within the precision of the measurements, so that we are con¢dent of tracer Pb/U to within 1 permil. In addition, the same calibration procedure has been applied to the mixed tracer solution used at IGMR, ETH Zurich, con¢rming the Pb/U ratio obtained in previous calibrations at the ETH laboratory. Furthermore, the ETH solution has been used for an interlaboratory comparison of zircon measurements with the participation of the ROM (Toronto), CGA (Ottawa) and Max-Planck (Mainz) laboratories [28], wherein the interlaboratory bias was shown to be insigni¢cant. Some age values given in earlier abstracts and deviating from the ages presented here were preliminary or are slightly biased (30.4% o¡set) because of a miscalibrated tracer solution (or a combination of both) in the early course of the study (for details please make inquiries to the senior author). Repeat measurements of the total procedural blank averaged 1.3 þ 0.8 pg Pb (U blanks were indistinguishable from zero), with 206 Pb/ 204 207 Pb = 18.58 þ 0.80, Pb/204 Pb = 15.47 þ 0.60, 208 204 Pb/ Pb = 37.89 þ 0.9 (all 2c of population), and a 206 Pb/204 Pb^207 Pb/204 Pb correlation of +0.82. These ratios and uncertainties were propagated into the age and age error calculations [29]. Common Pb in excess of the analytical blank was assumed to have the same composition as given above since the impact on the ¢nal result using model Pb isotopic compositions is insigni¢cant. Mass fractionation of U during analysis was controlled by U double-spike techniques, whereas Pb mass fractionation was corrected by 2.0 þ 0.6 per-

EPSL 5790 6-4-01

R. Mundil et al. / Earth and Planetary Science Letters 187 (2001) 131^145

mil/AMU, based on multiple analyses of NBS 981 (the rather high value arising from an V0.6x/ AMU mass discrimination contribution by the WARP ¢lter/Daly assembly).

[14]

References

[16]

[1] R. Mundil, P. Brack, M. Meier, H. Rieber, F. Oberli, High resolution U^Pb dating of Middle Triassic volcaniclastics; time-scale calibration and veri¢cation of tuning parameters for carbonate sedimentation, Earth Planet. Sci. Lett. 141 (1996) 137^151. [2] S.A. Bowring, D.H. Erwin, Y.G. Jin, M.W. Martin, K. Davidek, W. Wang, U/Pb Zircon Geochronology and tempo of the End-Permian Mass Extinction, Science 280 (1998) 1039^1045. [3] H. Yin, W.C. Sweet, B.F. Glenister, G. Kotlyar, H. Kozur, N.D. Newell, J. Sheng, Z. Yang, Y.D. Zakharov, Recommendation of the Meishan section as global stratotype section and point for basal boundary of Triassic System, Newslett. Stratigr. 34 (1996) 81^108. [4] D.H. Erwin, The Great Paleozoic Crisis: Life and Death in the Permian, Columbia University Press, New York, 1993, 327 pp. [5] A. Hallam, P.B. Wignall, Mass Extinctions and Their Aftermath, Oxford University Press, Oxford, 1997, 320 pp. [6] P.R. Renne, Z. Zichao, M.A. Richards, M.T. Black, A.R. Basu, Synchrony and causal relations between PermianTriassic boundary crises and Siberian £ood volcanism, Science 269 (1995) 1413^1413. [7] L.W. Alvarez, W. Alvarez, F. Asaro, H.V. Michel, Extraterrestrial cause for the Cretaceous Tertiary extinction, Science 208 (1980) 1095^1108. [8] D.-Y. Xu, S. Ma, Z.-F. Chia, X.-Y. Mao, Y.-Y. Sun, Q. Zhang, Z.-Z. Yang, Abundance variation of iridium and trace elements at the Permian/Triassic boundary at Shangsi in China, Nature 314 (1985) 154^156. [9] A.H. Knoll, R.K. Bambach, D.E. Can¢eld, J.P. Grotzinger, Comparative Earth history and Late Permian mass extinction, Science 273 (1996) 452^457. [10] Y. Isozaki, Permo-Triassic boundary superanoxia and strati¢ed superocean; records from lost deep sea, Science 276 (1997) 235^238. [11] Y. Kajiwara, S. Yamakita, K. Ishida, H. Ishiga, A. Imai, Development of a largely anoxic strati¢ed ocean and its temporary massive mixing at the Permian/Triassic boundary supported by the sulfur isotopic record, Palaeogeogr. Palaeoclimatol. Palaeoecol. 111 (1994) 367^379. [12] P.B. Wignall, A. Hallam, Anoxia as a cause of the Permian/Triassic mass extinction; facies evidence from northern Italy and the Western United States, Palaeogeogr. Palaeoclimatol. Palaeoecol. 93 (1992) 21^46. [13] P.B. Wignall, A. Hallam, Griesbachian (Earliest Triassic) palaeoenvironmental changes in the Salt Range, Pakistan and Southeast China and their bearing on the Permo-Tri-

[17]

[15]

[18] [19]

[20] [21]

[22] [23]

[24] [25]

[26]

[27]

[28]

[29] [30]

145

assic mass extinction, Palaeogeogr. Palaeoclimatol. Palaeoecol. 102 (1993) 215^237. P.B. Wignall, R.J. Twitchett, Oceanic anoxia and the end Permian mass extinction, Science 272 (1996) 1155^1158. D.H. Erwin, The Permo-Triassic extinction, Nature 367 (1994) 231^236. C. Spinosa, Vote on Meishan as GSSP for the Permian^ Triassic Boundary, Permophiles 35 (1999) 4^5. H. Kozur, M. Pjatakova, The conodont Anchignathodus parvus n. sp., an important guide fossil for the lower Triassic, Proc. K. Ned. Akad. Wet. Ser. B Palaeontol. Geol. Phys. Chem. 79 (1976) 123^128. J.C. Claoue-Long, Z. Zhang, G. Ma, S. Du, The age of the Permian-Triassic boundary, Earth Planet. Sci. Lett. 105 (1991) 182^190. K. Min, R. Mundil, P.R. Renne, K.R. Ludwig, A test for systematic errors in 40 Ar/39 Ar geochronology through comparison with U-Pb analysis of a 1.1 Ga rhyolite, Geochim. Cosmochim. Acta 64 (2000) 73^98. P.R. Renne, D.B. Karner, K.R. Ludwig, Absolute ages aren't exactly, Science 282 (1998) 1840^1841. R. Mundil, K.R. Ludwig, Improving the accuracy of high resolution single zircon U/Pb data: An example from the Permo-Triassic boundary, Geol. Soc. Am. Abstr. w/Progs 30 (1998) 312. J.M. Mattinson, A study of complex discordance in zircons using step-wise dissolution techniques, Contrib. Mineral. Petrol. 116 (1994) 117^129. T.E. Krogh, Identi¢cation of concordant zircons using etch techniques. Abstracts of the Eighth international conference on Geochronology, cosmochronology, and isotope geology, Geol. Surv. Circ. C 1107 (1994) 180. S.L. Kamo, G.K. Czamanske, T.E. Krogh, A minimum U-Pb age for Siberian £ood-basalt volcanism, Geochim. Cosmochim. Acta 60 (1996) 3505^3511. P.R. Renne, Excess 40 Ar in biotite and hornblende from the Norils'k 1 intrusion: Implications for the age of the Siberian Traps, Earth Planet. Sci. Lett. 131 (1995) 165^ 176. A.R. Basu, R.J. Poreda, P.R. Renne, F. Teichmann, Y. Vasiliev, N.V. Sobolev, B.D. Turrin, High-3 He plume origin and temporal-spatial evolution of the Siberian £ood basalts, Science 269 (1995) 822^825. T.E. Krogh, Improved accuracy of U-Pb zircon ages by the creation of more concordant systems using an air abrasion technique, Geochim. Cosmochim. Acta 46 (1982) 637^649. M. Wiedenbeck, P. Alle, F. Corfu, W.L. Gri¤n, M. Meier, F. Oberli, A.F. von Quadt, J.C. Roddick, W. Spiegel, Three natural zircon standards for U-Th-Pb, Lu-Hf, trace element and REE analyses, Geostand. Newslett. 19 (1995) 1^23. K.R. Ludwig, Calculation of uncertainties of U-Pb isotope data, Earth Planet. Sci. Lett. 46 (1980) 212^220. K.R. Ludwig, Isoplot/Ex 2.45, A Geochronological Toolkit for Microsoft Excel, Berkeley Geochronology Center, 2000.

EPSL 5790 6-4-01