Anom a lous Up per De vo nian mer cury en rich ments: com par i son of In duc tively Cou pled Plasma – Mass Spec trom e try (ICP-MS) and Atomic Ab sorp tion
Spec trom e try (AAS) an a lyt i cal data
Grzegorz RACKI1, *, Leszek MARYNOWSKI1 and Micha³ RAKOCIÑSKI1
1 Silesian Uni ver sity, De part ment of Earth Sci ences, Bêdziñska 60, PL-41-200 Sosnowiec, Po land
Racki, G., Marynowski, L., Rakociñski, M., 2018. Anom a lous Up per De vo nian mer cury en rich ments: com par i son of In duc - tively Cou pled Plasma – Mass Spec trom e try (ICP-MS) and Atomic Ab sorp tion Spec trom e try (AAS) an a lyt i cal data. Geo log i - cal Quar terly, 62 (3): 487–495, doi: 10.7306/gq.1419
Mer cury geo chem is try is emerg ing re cently as a hot topic in chemostratigraphical and fa cies re search, ow ing to the di ag nos tic char ac ter of Hg en rich ments as a proxy of vol ca nic ac tiv ity (cru cial in the con text of as sumed causal links be tween vol ca nic cat a clysms and mass ex tinc tions). Thus, as a pre req ui site to such far-reach ing in ter pre ta tions, re li able an a lyt i cal de ter mi na tions of Hg con cen tra tions are nec es sary. In con ven tion ally per formed anal y ses in sed i men tary geo chem is try, In - duc tively Cou pled Plasma – Mass Spec trom e try (ICP-MS) is usu ally ap plied, as an an a lyt i cal stan dard for trace el e ments, in - clud ing Hg. How ever, with a de tec tion limit (DL) of 10 ppb, such mea sured val ues have been ques tioned as a con clu sive geo chem i cal in di ca tor of Hg anom a lies, and, in stead, far more ac cu rate tech niques, such as Atomic Ab sorp tion Spec trom e - try (AAS; DL = 0.2 ppb), are re quested. As a pre lim i nary test of this view, we pres ent com par a tive anal y sis of 91 sam ples from three sec tions en com pass ing the key Frasnian-Famennian and Famennian-Tournaisian bound ary in ter vals in Mo rocco (Lahmida), Ger many (Kahlleite) and Uzbekistan (Novchomok), for which Hg con cen tra tions were de ter mined by both meth - ods in the same sam ples. De spite some dif fer ences, es pe cially at low Hg con cen tra tions, both an a lyt i cal meth ods re veal the same 12 ex traor di narily en riched sam ples in ex cess of 1 ppm (with one ex cep tion, the de ter mi na tion er ror is <20%), as well as sim i lar over all chemostratigraphic pat terns char ac ter ized by a few prom i nent Hg spikes, with a top value of 5.8 ppm. The Hg con cen tra tions de ter mined by ICP-MS and AAS are sig nif i cantly cor re lated, as high as r = 0.98 (Novchomok), even if the first method re veals a gen eral ten dency to ward slightly height ened val ues (by ~15 to 30% for me di ans). There fore, ICP-MS re sults can con clu sively be used in mer cury chemostratigraphy in or der to rec og nize ex traor di nary vol ca nic (or other) sig - nals, at least in the De vo nian geo log i cal re cord. False Hg anom a lies were not gen er ated by these con ven tional ICP-MS de - ter mi na tions.
Key words: Mer cury anom a lies, an a lyt i cal meth ods, geo chem is try, Frasnian-Famennian bound ary, De vo nian -Car bon if er ous bound ary, vol ca nism.
INTRODUCTION
De spite great ad vances in sed i men tary el e men tal geo - chem is try in the last few de cades, fo cused on the use of trace el e ments to re fine palaeoenvironmental re con struc tion (e.g., Sagemann and Ly ons, 2003; Tribovillard et al., 2006; Calvert and Pedersen, 2007; Ramkumar, 2015), mer cury, in con trast to irid ium, mo lyb de num, ura nium, va na dium, bar ium and zir co - nium, used to be largely ig nored in event chemostratigraphy and fa cies anal y sis. How ever, al ready Hildebrand and Boynton (1989) dis cov ered “sub-ppm” Hg anom a lies at three non-ma - rine North Amer i can lo cal i ties of the Cre ta ceous-Paleogene (K-Pg) boun d ary, and con sid ered them as ev i dence for acid rain caused by a large me te or ite im pact (see sim i lar data from Slovenia in Palinkaš et al., 1996).
The over looked, seem ingly triv ial geo chem i cal theme of ex - traor di narily con cen trated lev els of this chalcophile trace metal un ex pect edly re turned sev eral years ago. The es sen tial con tri - bu tion of vol ca nic gas eous ex pul sions to the mod ern global mer cury cy cle is well-es tab lished (e.g., Pyle and Mather, 2003).
Volcanogenic Hg is con tained mainly in gas eous emis sions, not in pyroclastic ash. Prior to de po si tion of ox y gen ated re ac tive Hg via rain, it can be dis trib uted world wide in the at mo sphere be - cause of the Hg res i dence time of ~0.5–2 years. Hg is scav - enged dur ing depositional pro cesses by sed i men tary or ganic mat ter, and also by ab sorp tion on clay min er als, sulphides (in ox y gen-de fi cient con di tions), and hy drous iron ox ides (see Sanei et al., 2012; Percival et al., 2015; Sial et al., 2016;
Bergquist, 2017; Sabatino et al., 2018). In the strati graphi cal re - cord, Nascimento-Silva et al. (2011) and Sanei et al. (2012) con cur rently high lighted Hg anom a lies as a re cord of cat a - strophic vol ca nic erup tions dur ing the end-Cre ta ceous and end-Perm ian mass ex tinc tions, re spec tively, while Hg was used as a marker of Neoproterozoic vol ca nism in Cryogenian de pos - its by Sial et al. (2010).
Since this dis cov ery, a mile stone in the de bate on vol ca nism ver sus im pact – mass ex tinc tion causal links (see a recentmost
* Corresponding author, e-mail: racki@us.edu.pl
Received: March 5, 2018; accepted: April 21, 2018; first published online: June 13, 2018
al., 2017; Sabatino et al., 2018). Only the Late De vo nian global events have re mained un re ported in this re spect. There fore, the oc cur rence and con cen tra tion of Hg in sed i men tary rocks now ap pears to be a hot topic in event/chemostratigraphical re - search. In ad di tion, mer cury iso topes re corded as a marker of mas sive vol ca nism of Large Ig ne ous Prov inces (LIPs) re mains an as so ci ated theme (Thibodeau and Bergquist, 2017). On the other hand, the di rect role of toxic Hg con tam i na tion in bio - sphere dev as ta tion, orig i nally stressed by Sanei et al. (2012), has shifted to a rather sub or di nate level (“has yet to be fully eval u ated” – Bond and Grasby, 2017: p. 18).
Con se quently, re li able an a lyt i cal de ter mi na tions of Hg abun dances are a pre con di tion for sub stan tive geo log i cal in ter - pre ta tions. In sed i men tary geo chem is try, In duc tively Cou pled Plasma – Mass Spec trom e try (ICP–MS) is ap plied as a stan - dard an a lyt i cal tool for a broad set of trace el e ments, in clud ing mer cury. How ever, such con ven tion ally mea sured Hg val ues, with a de tec tion limit (DL) of 10 ppb, are ques tioned as a trust - wor thy proxy of vol ca nic par ox ysms. There fore, other re fined tech niques, such as Atomic Ab sorp tion Spec trom e try (AAS; DL
= 0.2 ppb), are com monly pre ferred. Fur ther more, the “sus pect qual ity” of Hg an a lyt i cal data pro vided by ICP-MS might be a rea son for re ject ing manu scripts in clud ing this kind of data.
To test this view, we pro vide herein com par a tive ICP-MS and AAS anal y ses of 91 sam ples from three sec tions en com pass ing the key Frasnian-Famennian (F-F) and De vo nian-Car bon if er ous (D-C) bound ary in ter val in Mo rocco, Ger many and Uzbekistan (Fig. 1) for which Hg con cen tra tions were de ter mined by both
biodiversity col lapse (Racki et al., 2018).
SAMPLES AND METHODS
We have ana lysed sam ples from three deep-wa ter marly lime stone suc ces sions (Figs. 1 and 2): Lahmida (Mo rocco; 43 sam ples), Kahlleite (Ger many; 16 sam ples from the F-F bound - ary beds and 7 from the De vo nian-Car bon if er ous tran si tion), and Novchomok (Uzbekistan; 25 sam ples).
In palaeogeographic terms, the nat u ral Lahmida sec tion rep re sents the deep-wa ter Rheris shelf ba sin, on the south ern pe riph ery of the Rheic Ocean. The ex po sure is lo cated ~12 km to the north-west from Erfoud in the east ern part of the Anti-At - las (Wendt and Belka, 1991; Dopieralska, 2003, 2009). The Mo roc can suc ces sion con sists mainly of mo not o nous shales with nu mer ous marly interbeds and con cre tion ho ri zons as well as dark grey lime stones, which to gether with dark grey shales cor re spond to the ex panded Kellwasser fa cies of the Rheris Ba - sin (see Dopieralska, 2003). The thick ness of the in ter val sam - pled is ~36 m. This suc ces sion was dated by cono donts but apart from zonal as sign ments (Fig. 2), fau nal de tails have not yet been pub lished. The suc ces sion shown in cludes the in ter val from the Mid dle Frasnian (MN Zone 5) to the top of the lower Famennian (rhomboidea Zone; Dopieralska, 2003).
The in ac tive (from 2013 on) Kahlleite Quarry is lo cated 1 km south-west of Rüdersdorf near Gera, Thuringia, cen tral Ger -
Fig. 1. Geo graphic and palaeogeographic lo ca tions of the Late De vo nian sites stud ied for Hg abun dances (Fig. 2), com pared to in ferred prox im ity to the co eval Si be rian large ig ne ous prov ince (af ter Kravchinsky, 2012;
Late De vo nian palaeo ge ogra phy af ter Golonka et al., 1994)
many, at the north-west flank of the Berga Anticline (Gereke, 2004, 2007; Weyer, 2016; see also fig. 18 in Hartenfels, 2011) in the Saxothuringian Zone (Eckelmann et al., 2014, Franke et al., 2017). The suc ces sion stretched from the Mid-Up per Frasnian to the Mid-Up per Tournaisian (Weyer, 2016). Two in - ter vals were sam pled, that en com pass marker Up per De vo nian black shale lev els: the Up per Kellwasser (3.5 cm thick) and Hangenberg (30 cm thick). The beds con sist largely of grey nod u lar-micritic lime stones, de pos ited on a deep sub ma rine rise be long ing to the outer south ern shelf of the clos ing Rheic Ocean (Gereke and Schindler, 2012; Eckelkmann et al., 2014;
Saupe et al., 2016; Franke et al., 2017).
The Novchomok (22) site is lo cated in the east ern part of the Kitab Geo log i cal Re serve Area, ly ing in the Zeravshan - -Gissar Moun tains area (west ern part of Tian-Shan, south Uzbekistan; see Narkiewicz et al., 2017). The De vo nian-Car - bon if er ous bound ary in ter val sam pled, ~17 m thick, com prises the top part of the Yatavluk For ma tion and the basal slice of the Novchomok For ma tion. The lower part (Yatavluk Fm.) con sists mostly of dark grey micritic lime stones with cri noid de tri tus and cor als, and in cludes also cherry-col oured and grey marly shales and marls. The up per por tion (Novchomok Fm.) is com - posed of dark grey, mostly micritic lime stones and brown marly shales, lo cally with cri noids. The D-C bound ary, based on cono - donts, was ten ta tively traced just be low the top of the Yatavluk For ma tion by Narkiewicz et al. (2017). The De vo nian and Car - bon if er ous strata were de pos ited on the pas sive mar gin of the
Tarim plate, as ter ranes of Perigondwana then ap proached Kazakh ter ranes with clos ing of the South Tianshan Ocean (Golonka, 2012; Han and Zhao 2018; see also Stampfli and Borel, 2002).
MERCURY DETERMINATION
The Hg con cen tra tion of bulk rock sam ples from the three sec tions stud ied were ana lysed us ing two in de pend ent geo - chem i cal meth ods (Fig. 2; Ta bles 1–3). Con ven tional ICP-MS anal y ses were pro vided com mer cially by Bu reau Veri tas AcmeLabs, Van cou ver, Can ada. STD DS10 stan dard ma te rial was the prin ci pal ref er ence, with a cer ti fied Hg abun dance of 289 ppb. The mer cury con cen tra tion of the ref er ence ana lysed in the lab o ra tory was within 13% of its ex pected Hg value. How - ever, when an other STD, OREAS45EA, was con sid ered, with a low cer ti fied Hg value of 30 ppb (reg is tered in our 2011 anal y - ses), i.e., near the method’s de tec tion limit (10 ppb), the ac cu - racy er ror is far greater, reach ing >60%. Se lected sam ples were mea sured in du pli cates, show ing reproducibility better than ±5%.
For atomic ab sorp tion spec trom e try (AAS), a two-cell, pyrolyzer -type Mile stone DMA-80 Di rect Mer cury An a lyzer was used, with a de tec tion limit of 0.2 ppb. Anal y ses were per formed at the Fac ulty of Earth Sci ences, Uni ver sity of Silesia (Po land).
Fig. 2. Com par i son of an a lyt i cal Hg abun dances pro vided by ICP-MS and ASS meth ods for Late De vo nian sec tions in Mo rocco (A;
af ter Dopieralska, 2003), Ger many (B; af ter Gereke, 2004) and Uzbekistan (C), show ing con sis tently large Hg anom a lies in both the Hg data sets (Fig. 3), as so ci ated with the F-F and D-C bound aries (Ta bles 1–3), in clud ing the marker black shale Up per Kellwasser and Hangenberg lev els at Kahlleite
FAMENNIAN
LA 44 80 54.5
LA 43/44 40 31.2
LA 42 270 250.0
LA 41/42 270 276.6
LA 40T 50 44.8
LA 38 100 99.8
LA 35/36B 250 283.2
LA 34 250 260.0
LA 32/33 150 207.0
LA 32M 140 153.2
LA 30 400 333.8
LH 29 170 137.5
LH 28 110 114.0
LH 27A2 420 569.1
LA 27/28 260 338.4
LH 27 120 126.4
LA 26/27 360 481.2
LH 26 60 57.3
LA 25/26 190 232.5
LH 25T 180 180.1
LH 25B 210 189.3
FRASNIAN
LA 24/25S 110 36.9
LA 24/25N 1530 1136.4
LH 24T 110 113.2
LH 24B 190 153.7
LH 23T 250 219.0
LH 23B 90 90.0
LH 22 490 464.2
LH 21 120 90.3
LH 20 90 45.7
LH 19 230 123.0
LH 18 250 187.7
LA 17/18T 260 183.2
LA 17/18B 170 147.8
LA 17M 70 99.9
LA 14 1050 1144.9
LA 13T 200 233.1
LA 12 50 68.8
LA 10 <10* 6.8
LA 8 <10* 9.9
LA 6 30 25.9
LA 5 320 312.3
LA 4A 30 36.9
Me dian/mean value 175/226 153.2/217.4
Lin ear cor re la tion co ef fi cient 0.96
Spearman’s rs cor re la tion co ef fi cient 0.95
*Taken as 8 ppb in cal cu la tion of cor re la tion co ef fi cients
The DMA an a lyt i cal curves were pre pared with the di lu tion of a 1 mg L–1 stan dard so lu tion (Merck Darmstadt, Ger many). Mea - sure ments of each sam ple were du pli cated, and anal y ses were re peated when the co ef fi cient of vari abil ity of sam ples ex - ceeded 5%. The in stru ment was cal i brated us ing cer ti fied ref er - ence ma te rial INCT-OBTL-5 (to bacco leaves) prior to the mea - sure ment, with Hg con tent = 20.9 ppb. The ac cu racy did not ex - ceed 2%.
In an other Hg study, with the use of the same analyser type, the ac cu racy and pre ci sion of the de ter mi na tions were es ti - mated as ~8 and 6.5%, re spec tively (Sabatino et al., 2018; see also Sanei et al., 2012). In other pa pers, these qual ity in di ca tors of Hg anal y ses, fre quently partly only re ported, are <10% (e.g., 4 and 9.5%, re spec tively – Nascimento-Silva et al., 2011; Sial et al., 2016; see also Grasby et al., 2015; Jones et al., 2017).
Thus, the Hg con tents from the Pol ish lab o ra tory, con sid ered herein, show sim i lar lev els of an a lyt i cal relatability.
RESULTS
Com par a tive anal y sis of the Hg val ues se ries from the three lo cal i ties, shown in Ta bles 1–3 and Fig ure 2, re veal a close re - sem blance of ICP-MS and AAS de ter mi na tions. Unsurprisingly
in sta tis ti cal terms, there fore, covariation of both data sets is very sig nif i cant, and the Spearman’s cor re la tion co ef fi cient is not
<0.95. Even if back ground (= me dian) val ues vary widely be - tween the suc ces sions, from 36.4 ppb at Kahlleite to 263.5 ppb at Novchomok, they dis play sim i larly low er ror lev els. Com par ing the me dian abun dances, the ICP-MS val ues are no ta bly con - stantly height ened by 12.5% at Lahmida, 17.2% at Kahlleite, and 29.0% at the most Hg-en riched Uz bek suc ces sion. Mean Hg val - ues char ac ter ize a de creased dis per sion, but this es ti mate is ran - domly bi ased by un usu ally high con cen tra tions.
It is es pe cially sig nif i cant that the same ex traor di narily en - riched val ues of >1000 ppb, are re vealed in both data sets, with top value of 5825.3 ppb at Novchomok (if we grant that AAS val - ues ap prox i mately re flect real Hg abun dances in the sam ples ana lysed). In more de tail, for 17 large-scale en rich ments (>500 ppb), the rel a tive de ter mi na tion er rors by ICP-MS method are ran domly dis trib uted and, with one ex cep tion, <20%. In fact, the in creased Hg val ues lo cate in the AAS anal y sis er ror field in the case of >70% en riched sam ples (Fig. 3).
When com par ing par tic u lar sam ples, some di ver gences are very dis tinc tive. As ex em pli fied by sam ple K 1, the very low ICP-MS value be low the DL (10 ppb) is not con firmed by AAS de ter mi na tion (22.5 ppb). By con trast, a fab ri cated “en rich ment”
is noted for sam ple K 4 (20 in stead of 8.5 ppb), and sim i larly Stage
Sam ple Hg Hg
Method ICP-MS AAS
Method de tec tion limit (MDL) 10 ppb 0.2 ppb
FAMENNIAN
K-HANG 7 110 92.9
K-HANG 6 60 58.4
K-HANG 5 1300 1504.9
K-HANG 4 1350 1529.0
K-HANG 3 790 753.3
K-HANG 2 1280 1456.4
K-HANG 1 1380 1438.3
K 8 20 16.2
K 7 <10* 6.9
K 6 20 11.0
K 5 <10* 5.2
K 4 20 8.5
K 3 <10* 9.3
K 2 30 18.9
K 1 <10* 22.5
FRASNIAN
K 0 CH 2380 2517.3
K 01 110 93.0
K 02 30 22.7
K 03 50 23.6
K 05 10 10.6
K 06 40 36.4
K 07 Sx 120 63.3
K 08 60 42.1
Me dian/mean value 50/400 36.4/423.5
Lin ear cor re la tion co ef fi cient 1.00
Spearman’s rs cor re la tion co ef fi cient 0.96
*Taken as 8 ppb in cal cu la tion of cor re la tion co ef fi cients
T a b l e 2 Hg abun dances in sam ples from the Kahlleite sec tion us ing ICP-MS and AAS
meth ods (Hg en rich ments >0.5 ppm in AAS val ues are high lighted)
large dif fer ences are found for the sub-back ground abun - dances of the sam ples NOV 22 /04 and 07 at Novchomok. Like - wise, the Hg con tent of the near F-F bound ary sam ple from Lahmida is shown as three times higher in the ICP-MS mea - sure ment (110 ppb ver sus 36.9 ppb). Among truly en riched sam ples, the only note wor thy de vi a tion is in an Uz bek sam ple NOV 22 /09A, for which Hg val ues are dou bled in the ICP-MS mea sure ments (2430 ver sus 1186.6 ppb). On the other hand, a max i mal “im pov er ish ment” by 17.4% is noted for the ex treme value in sam ple NOV 22 /010 in the same sec tion (4810 ver sus 5825.3 ppb).
DISCUSSION
Since 2011 we have es tab lished a da ta base for Hg abun - dances, de ter mined largely by ICP-MS, for 17 F-F sec tions and 11 D-C lo cal i ties in dif fer ent re gions of the world, re veal ing sev - eral var i ously re corded anom a lous Hg spikes, up to 8 ppm in the F-F bound ary beds (Psie Górki, Holy Cross Mts., Po land) and 18.5 ppm in the D-C bound ary beds (Kronhofgraben, Carnic Alps, Aus tria). How ever, only the re fined AAS val ues are
com monly ac cepted and widely re garded as re li able in the dis - cus sion of po ten tial vol ca nic sig nals. There fore, we have tested the Hg chemo stratigraphic pat tern aris ing from less re li able ICP-MS re sults. Both an a lyt i cal meth ods clearly re veal a sim i lar Hg chemo stratigraphic pat tern in the suc ces sions stud ied (see Fig. 2), as well as show ing high cor re la tion val ues of the mea - sured val ues, even for the low est Hg abun dances (<40 ppb;
Fig. 3A). With one ex cep tion (Fig. 3B), the larg est an a lyt i cal er - rors oc cur no ta bly in val ues be low or at most near the Hg base - lines in the sec tions we stud ied. So, pro por tion ally large dis - crep an cies may be in flu enced not only by the an a lyt i cal pro ce - dure weak ness of the ICP-MS stan dard, but also by in suf fi cient ho mog e ni za tion of the rock pow ders of the sam ples ana lysed, among other fac tors.
Six sam ples un der study dis play con cen tra tions be low the DL of the ICP-MS method (10 ppb). For the pur pose of cal cu la - tion, half of the DL value is com monly adopted (e.g., Riboulleau et al., 2018), i.e., 5 ppb in the case of Hg (Fig. 3A). How ever, ex - clud ing the ab nor mally in con sis tent sam ple K 1, dis per sion of the re main ing val ues (from 5.2 to 9.9 ppb) sug gests that Hg val - ues be tween 7 and 8 ppb are far more rep re sen ta tive.
Both meth ods char ac ter ize the suc ces sions as in clud ing prom i nent Hg ex cur sions, for 12 sam ples >1000 ppb (1 ppm), TOURNAISIAN
NOV 22 /9 120 102.6
NOV 22 /6 340 294.7
NOV 22 /5 490 373.3
NOV 22 /3 130 80.4
NOV 22 /2 140 94.6
NOV 22 /1 160 119.9
NOV 22 /0 130 83.7
NOV 22 /01 230 173.4
FAMENNIAN
NOV 22 /02 630 756.2
NOV 22 /03 190 100.9
NOV 22 /04 250 127.9
NOV 22 /05 1320 1512.5
NOV 22 /06 280 179.9
NOV 22_07 310 151.9
NOV 22 /08 2430 2068.4
NOV 22 /09A 2430 1186.6
NOV 22_09 B 3970 4331.5
NOV 22 /010 4810 5825.3
NOV 22_011 360 263.5
NOV 22 /012 550 352.9
NOV 22_013 680 595.0
NOV 22 /014 260 200.1
NOV 22_015 610 457.7
NOV 22 /016 900 730.3
NOV 22 /020 300 238.7
Me dian/mean value 340/860 263.5/794.8
Lin ear cor re la tion co ef fi cient 0.97
Spearman’s rs cor re la tion co ef fi cient 0.98
in par tic u lar di rectly be low the F-F and D-C bound aries (Fig.
2). No ta bly, Wedepohl’s (1991) av er aged Phanerozoic Hg con cen tra tions range from 30 ppb (lime stone) to 450 ppb (ar - gil la ceous shale). Also, Ketris and Yudovich (2009) cal cu lated the di verse shale-av er aged abun dances as be tween 180
±30 ppb (cherty shales) and 290 ±30 ppb (cal car e ous shale), and an av er aged Hg abun dance of 270 ±30 ppb for black shales. There fore, the Hg anom a lies rec og nized for the first time in Up per De vo nian strata are ap prox i mately one (Lahmida, Novchomok) to two (Kahlleite) or ders of mag ni tude higher com pared to the base lines for those suc ces sions, as well as to the ex pected world av er ages. This sta tus is main - tained even for the most ab nor mally bi ased ICP-MS value of the highly en riched sam ple NOV 22/09 (Figs. 2 and 3).
The as so ci a tion of anom a lous Hg ex cur sions with a cat a - clys mic LIP trig ger at the F-F and D-C global events is an at trac - tive hy poth e sis (Racki et al., 2018), es pe cially in the light of some re cent views on a causal role of the Si be rian Viluy LIP (e.g., Kravchinsky, 2012; Ricci et al., 2013; see also Win ter, 2015; Fig. 1), but other al ter na tives need to be con sid ered. En - rich ments of a sim i lar or der (1570 ppb), de ter mined from ICP - -MS val ues, have al ready been re ported from the sup posed Up - per Kellwasser level in coastal fa cies of Catalonia, Spain, by Moreno et al. (2018), but in ter preted ex clu sively as a sig na ture of lo cal hy dro ther mal ac tiv ity, no ta bly in tense in the clos ing, north-subducting Rheic Ocean (Win ter, 2015; Raumer et al., 2017; see Fig. 1). In fact, an oc cur rence of re gional and lo cal Hg spikes is prob a ble, es pe cially in the palaeogeographic do main strongly af fected by Eovariscan vol ca nism, wide spread also in the Saxothuringian zone (Timmerman, 2008) and other re gions stud ied (Simancas et al., 2005; Michard et al., 2010; Han and Zhao, 2018).
Hg anom a lies may in deed have mul ti ple or i gins, and also re cord “in creased pres er va tion of Hg within sed i ments ei ther through in creases in de po si tion via scav eng ing/ab sorp tion onto par ti cles or post-depositional mi gra tion of Hg within the sed i - ments” (Bergquist, 2017: p. 8675). In case of the K-Pg en rich - ments of Hg at Bidart (France), a hy poth e sis of “post depo - sitional geo chem i cal leach ing and re duc tion pro cesses” has been pro posed by Smit et al. (2016) in place of a Deccan LIP – de rived Hg de liv ery. In ad di tion, as shown by Zheng et al.
(2018) in coal bas ins, Hg con cen tra tion pat terns are sig nif i - cantly af fected by sec ond ary mag matic phe nom ena, such as flows of mag matic-hy dro ther mal and low-tem per a ture hy dro - ther mal flu ids, and by el e vated heat and pres sure due to in trud - ing mag mas. There fore, the dis tinc tion be tween global (=
widely dis trib uted and syn chro nously re corded) sig nals and di - ver sity of re gional noise is a novel chal lenge in Hg chemo - stratigraphy (Racki et al., 2018).
CONCLUSIONS
A pre lim i nary test of two an a lyt i cal meth ods of Hg de ter mi na - tion ne gates the view that ICP-MS val ues are un re li able, and gen er ate of false sig nals. De spite some dif fer ences be tween the Hg data sets, es pe cially at low Hg lev els (spo rad i cally even two to three times over stated by ICP-MS), both an a lyt i cal meth ods re veal the same ex traor di narily en riched sam ples in the F-F and D-C bound ary beds un der study. The over all chemo stratigraphic pat tern is also sim i lar, char ac ter ized by a few prom i nent Hg ex - cur sions in ex cess of 1 ppm. The val ues of Hg de ter mined by ICP-MS and AAS are, there fore, sig nif i cantly cor re lated, even if the first method re veals a gen eral ten dency to wards slightly height ened val ues (by ~15 to 30% for me di ans).
In sum mary, con ven tional ICP–MS re sults can con clu sively be used in mer cury chemostratigraphy in or der to rec og nize ex - traor di nary vol ca nic (or other) sig nals, un doubt edly re lated to the geo log i cal re cord of De vo nian global events. False anom a - lies are not gen er ated by the Hg de ter mi na tions.
Ac knowl edge ments. The study was sup ported by the MAESTRO grant 2013/08/A/ST10/00717 from the Na tional Sci ence Cen tre – Po land (to Grzegorz Racki). We warmly thank Z. Be³ka, M. Gereke, U. Rakhmonov, N. Abdiev, F. Salimova, and other par tic i pants in the grant field works. Con struc tive re - marks by the jour nal re view ers Z. Saw³owicz, J. Kalvoda and T. Becker are greatly ap pre ci ated. We also thanks I. Miliñska and R. Fikier for lab o ra tory works.
Fig. 3. Cor re la tion of low, less than 40 ppb (A), and high,
>500 ppb (B), mer cury abun dances pro vided by ICP-MS and ASS for three Up per De vo nian sec tions (Fig. 2) Note the gen er ally highly co her ent covariation of both Hg data sets, with the most sig nif i cant ex cep tion of the sam ple NOV 22 /09A from the D-C tran si tion in Uzbekistan. The dark grey field shows a max i - mum er ror range in the AAS mea sure ments (taken as 10% af ter Sabatino et al., 2018; in cludes 70.6 % ICP-MS val ues), and the light grey field in di cates a 20% de ter mi na tion er ror range, that en com - passes 94% ICP-MS de ter mi na tions. Val ues be low the de tec tion level were ap prox i mated af ter Riboulleau et al. (2018)
Calvert, S.E., Pedersen, T.F., 2007. El e men tal prox ies for palaeo - climatic and palaeoceanographic vari abil ity in ma rine sed i - ments: in ter pre ta tion and ap pli ca tion. De vel op ments in Ma rine Ge ol ogy, 1: 567–644.
Dopieralska, J., 2003. Neo dym ium iso to pic com po si tion of cono - donts as a palaeoceanographic proxy in the Variscan oce anic sys tem. Ph.D. the sis, Justus-Liebig-Uni ver sity, Giessen.
Dopieralska, J., 2009. Re con struct ing sea wa ter cir cu la tion on the Mo roc can shelf of Gond wana dur ing the Late De vo nian: ev i - dence from Nd iso tope com po si tion of cono donts. Geo chem is - try, Geo phys ics, Geosystems, 10: 1–10.
Eckelmann, K., Nesbor, H.-D., Königshof, P., Linnemann, U., Hofmann, M., Lange, J.-M., Sagawe, A., 2014. Plate in ter ac - tions of Laurussia and Gond wana dur ing the for ma tion of Pangaea — con straints from U–Pb LA–SF–ICP–MS de tri tal zir - con ages of De vo nian and Early Car bon if er ous siliciclastics of the Rhenohercynian zone, Cen tral Eu ro pean Variscides. Gond - wana Ressearch, 25: 1484–1500.
Franke, W., Cocks, L.R.M., Torsvik, T.H., 2017. The Palaeozoic Variscan oceans re vis ited. Gond wana Re search, 48: 257–284.
Gereke, M., 2004. Das Profil Kahlleite Ost – die stratigraphische Entwicklung einer Tiefschwelle im Oberdevon des Bergaer Sattels (Thüringen). Geologica et Palaeontologica, 38: 1–31.
Gereke, M., 2007. Die oberdevonische Kellwasser-Krise in der Beckenfazies von Rhenoherzynikum und Saxothuringikum (spätes Frasnium/frühestes Famennium, Deutsch land). Kölner Fo rum für Geologie und Paläontologie, 17: 1–228.
Gereke, M., Schindler, E., 2012. “Time-Spe cific Fa cies” and bi o log - i cal cri ses – the Kellwasser Event in ter val near the Frasnian/Fame nnian bound ary (Late De vo nian). Palaeo ge - ogra phy, Palaeoclimatology, Palaeo ec ol ogy, 367–368: 19–29.
Golonka, J., 2012. Pa leo zoic paleoenvironment and Paleolitho - facies Maps of Gond wana. AGH Uni ver sity of Sci ence and Tech nol ogy Press.
Golonka, J., Ross, M.I., Scotese, C.R., 1994. Phanerozoic paleogeographic and paleoclimatic mod el ing maps. Ca na dian So ci ety of Pe tro leum Ge ol o gists Mem oir, 17: 1–47.
Grasby, S.E., Beauchamp, B., Bond, D.P.G., Wignall, P.B., Tala - vera, C., Gal lo way, J.M., Piepjohn, K., Reinhardt, L., Blomeier, D., 2015. Pro gres sive en vi ron men tal de te ri o ra tion in NW Pangea lead ing to the lat est Perm ian ex tinc tion. GSA Bul le - tin, 127: 1331–1347.
Han, Y.G., Zhao, G.C., 2018 (in press). Fi nal amal gam ation of the Tianshan and Junggar orogenic col lage in the south - west ern Cen tral Asian Orogenic Belt: con straints on the clo - sure of the Paleo-Asian Ocean. Earth-Sci ence Re views, https://doi.org/10.1016/j.earscirev.2017.09.012
Hartenfels, S., 2011. Die globalen Annulata-Events und die Dasberg -Krise (Famennium, Oberdevon) in Europa und Nord - -Afrika – hochauflösende Conodonten-Stratigraphie, Karbonat - -Mikrofazies, Paläoökologie und Paläodiversität. Münstersche Forschungen zur Geologie und Paläontologie, 105: 17–527.
Hildebrand, A.R., Boynton, W.V., 1989. Hg anom a lies at the K/T bound ary: ev i dence for acid rain? Meteoritics, 24: 277–278.
Jones, D.S., Mar tini, A.M., Fike, D.A., Kaiho, K., 2017. A vol ca nic trig ger for the Late Or do vi cian mass ex tinc tion? Mer cury data from south China and Laurentia. Ge ol ogy, 45: 631–634.
Ketris, M.P., Yudovich, Y.E., 2009. Es ti ma tions of Clarkes for car - bo na ceous biolithes: world av er ages for trace el e ment con tents in black shales and coals. In ter na tional Jour nal of Coal Ge ol ogy, 78: 135–148.
Tectonophysics, 492: 1–24.
Moreno, C., Gon za lez, F., Sáez, R., Melgarejo, J.C., Suárez- Ruiz, I., 2018. The Up per De vo nian Kellwasser Event re corded in a re gres sive se quence from in ner shelf to la goonal pond, Cata lan Coastal Ranges, Spain. Sedimentology, 65: doi:
10.1111/sed.12457
Narkiewicz, K., Rakociñski, M., Corradini, C., Racki, G., 2017.
New cono dont data from the De vo nian-Car bon if er ous bound ary in ter val in the Kitab Re serve Area (Uzbekistan). Cuadernos del Museo Geominero, 22: 183–184.
Nascimento-Silva, V.M., Sial, A.N., Ferreira, V.P., Neumann, V.H., Barbosa, J.A., Pimentel, M.M., Lacerda, L.D., 2011. Cre ta - ceous-Paleogene tran si tion at the Paraíba Ba sin, north east ern, Brazil: car bon-Iso tope and mer cury subsurface stratigraphies.
Jour nal of South Amer i can Earth Sci ences, 32: 379–392.
Palinkaš, A.L., Drobne, K., Durn, G., Miko, S., 1996. Mer cury anom aly at the Cre ta ceous-Ter tiary bound ary: Dolenja Vas, Slovenia (ab stract). In: In ter na tional Work shop Postojna ‘96.
The Role of Im pact Pro cesses in the Geo log i cal and Bi o log i cal Evo lu tion of Planet Earth: 31–32.
Percival, L.M.E., Witt, M.L.I., Mather, T.A., Hermoso, M., Jenkyns, H.C., Hesselbo, S.P., Al-Suwaidi, A.H., Storm, M.S., Xu, W., Ruhl, M., 2015. Glob ally en hanced mer cury de po si tion dur ing the end-Pliensbachian ex tinc tion and Toarcian OAE: a link to the Karoo-Ferrar Large Ig ne ous Prov ince. Earth and Plan e tary Sci ence Let ters, 428: 267–280.
Percival, L.M.E., Ruhla, M., Hesselbo, S.P., Jenkyns, H.C., Mather, T.A., White side, J.H., 2017. Mer cury ev i dence for pulsed vol ca nism dur ing the end-Tri as sic mass ex tinc tion. Pro - ceed ings of the Na tional Acad emy of Sci ences of the United States of Amer ica, 114: 7929–7934.
Pyle, D.M., Mather, T.A., 2003. The im por tance of vol ca nic emis - sions for the global at mo spheric mer cury cy cle. At mo spheric En vi ron ment, 37: 5115–5124.
Racki, G., Rakociñski, M., Marynowski, L., Wignall, P.B., 2018.
Mer cury en rich ments and the Frasnian-Famennian bi otic cri sis:
a vol ca nic trig ger proved? Ge ol ogy, 46: 543–546.
Ramkumar, M. ed., 2015. Chemostratigraphy. Con cepts, Tech - niques, and Ap pli ca tions. Elsevier, Am ster dam.
Rampino, M.R., 2017. Cat a clysms. A New Ge ol ogy for the Twenty - -first Cen tury. Co lum bia Uni ver sity Press, New York.
Raumer, J.F., Nesbor, H.D., Stampfli, G.M., 2017. The north - -subducting Rheic Ocean dur ing the De vo nian: con se quences for the Rhenohercynian ore sites. In ter na tional Jour nal of Earth Sci ences, 106: 2279–2296.
Riboulleau, A., Spina, A., Vecoli, M., Riquier, L., Quijada, M., Tribovillard, N., Averbuch, O., 2018. Or ganic mat ter de po si - tion in the Ghadames Ba sin (Libya) dur ing the Late De vo nian: a multidisciplinary ap proach. Palaeo ge ogra phy, Palaeo climato - logy, Palaeo ec ol ogy, 497: 37–51.
Ricci, J., Quidelleur, X., Pav lov, V., Orlov, S., Shatsillo, A., Courtillot, V., 2013. New 40Ar/39Ar and K–Ar ages of the Viluy traps (East ern Si be ria): fur ther ev i dence for a re la tion ship with the Frasnian–Famennian mass ex tinc tion. Palaeo ge ogra phy, Palaeoclimatology, Palaeo ec ol ogy, 386: 531–540.
Sabatino, N., Ferraro, S., Coccioni, R., Bonsignore, M., Del Core, M., Tancredi, V., Sprovieri, M., 2018. Mer cury anom a lies in up per Aptian-lower Albian sed i ments from the Tethys realm.
Palaeo ge ogra phy, Palaeoclimatology, Palaeo ec ol ogy, 495:
163–170.
Sageman, B.B., Ly ons, T.W., 2003. Geo chem is try of fine-grained sed i ments and sed i men tary rocks. Trea tise on Geo chem is try, 7:
15–158.
Sanei, H., Grassby, S.E., Beauchamp, B., 2012. Lat est Perm ian mer cury anom a lies. Ge ol ogy, 40: 63–66.
Saupe, A., Hartenfels, S., Becker, R.T., 2016. Ag glu ti nat ing foraminifers around the Annulata Events and Dasberg Cri sis (Famennian, Up per De vo nian) – palaeo ec ol ogy and palaeo - diversity. In: 87th An nual Con fer ence of the Paläontologische Gesellschaft, Dresden, 2016: 134.
Sial, A.N., Gaucher, C., Silva Filho, M.A., Ferreira, V.P., Pimentel, M.M., Lacerda, L.D., Silva Filho, E.V., Cezario, W., 2010. C-, Sr-iso tope and Hg stratigraphies of Neoproterozoic cap car bon - ates of the Sergipano Belt, North east ern Brazil. Pre cam brian Re search, 182: 351–372.
Sial, A.N., Chen, J., Lacerda, L.D., Frei, R., Tewari, V.C., Pandit, M.K., Gaucher, C., Ferreira, V.P., Cirillis, S., Peralta, S., Korte, C., Barbosa, J.A., Pereira, N.S., 2016. Mer cury en rich - ments and Hg iso topes in Cre ta ceous-Paleogene bound ary suc ces sions: links to vol ca nism and paleoenvironmental im - pacts. Cre ta ceous Re search, 66: 60–81.
Simancas, J.F., Tahiri, A., Azor, A., González-Lodeiro, F., Mar ti - nez Poyatos, D.J., El Hadi, H., 2005. The tec tonic frame of the Variscan–Alleghanian orogen in South ern Eu rope and North ern Af rica. Tectonophysics, 398: 181–198.
Smit, J., Koeberl, C., Claeys, P., Montanari, A., 2016. Mer cury anom aly, Deccan vol ca nism, and the end-Cre ta ceous mass ex - tinc tion: com ment. Ge ol ogy, 44: e381.
Stampfli, G.M., Borel, G.D., 2002. A plate tec tonic model for the Pa leo zoic and Me so zoic con strained by dy namic plate bound - aries and re stored syn thetic oce anic isoch rones. Earth and Plan e tary Sci ence Let ters, 196: 17–33.
Thibodeau, A.M., Bergquist, A.B., 2017. Do mer cury iso topes re - cord the sig na ture of mas sive vol ca nism in ma rine sed i men tary re cords? Ge ol ogy, 45: 95–96.
Timmerman, M.J., 2008. Palaeozoic magmatism. In: The Ge ol ogy of Cen tral Eu rope: Pre cam brian and Palaeozoic (ed. T.
McCann): 665–748, 1. Geo log i cal So ci ety, Lon don.
Tribovillard, N., Algeo, T.J., Ly ons, T., Riboulleau, A., 2006. Trace met als as paleoredox and paleoproductivity prox ies: an up date.
Chem i cal Ge ol ogy, 232: 12–32.
Wedepohl, K.H., 1991. The com po si tion of the up per earth’s crust and the nat u ral cy cles of se lected met als. Met als in nat u ral raw ma te ri als. Nat u ral Re sources. In: Met als and their Com pounds in the En vi ron ment (ed. E. Merian): 3–17. Verlag Chemie (VCH), Weinheim.
Wendt, J., Belka, Z., 1991. Age and depositional en vi ron ment of Up per De vo nian (Early Frasnian to Early Famennian) black shales and lime stones (Kellwasser Fa cies) in the East ern Anti- At las, Mo rocco. Fa cies, 25: 51–90.
Weyer, D., 2016. Re view of some Frasnian ahermatypic coral lo cal i - ties from Ger many and de scrip tion of a new ge nus Spinaxon (Anthozoa, Rugosa, Up per De vo nian). Geologica Belgica, 19:
147–163.
Win ter, J., 2015. Vulkanismus und Kellwasser-Krise – Zirkon - -Tephrostratigrafie, Identifizierung und Herkunft distaler Fall - out-Aschenlagen (Oberdevon, Synklinorium von Dinant, Rheinisches Schiefergebirge, Harz). Zeitschrift der Deutschen Gesellschaft für Geowissenschaften, 166: 227–251.
Zheng, L., Sun, R., Hintelmann, H., Zhu, J., Wang, R., Sonke, J.E., 2018. Mer cury sta ble iso tope com po si tions in mag matic - -af fected coal de pos its: new in sights to mer cury sources, mi gra - tion and en rich ment. Chem i cal Ge ol ogy, 479: 86–101.
