Temperature-driven recycling and euxinia as a marine mass extinction mechanism

Dominik Hülse (  [email protected] ) University of California, Riverside https://orcid.org/0000-0001-5386-6746 Kimberly Lau The Pennsylvania State University Sebastiaan van de Velde University of California, Riverside Sandra Arndt University of Bristol Katja Meyer Willamette Andy Ridgwell University of California, Riverside https://orcid.org/0000-0003-2333-0128

Article

Keywords: nutrient recycling, euxinia, temperature, microbial metabolism, ocean state, carbon cycling, end-Permian mass extinction

Posted Date: November 30th, 2020

DOI: https://doi.org/10.21203/rs.3.rs-80350/v1

License:   This work is licensed under a Creative Commons Attribution 4.0 International License. Read Full License 1 Temperature-driven nutrient recycling and euxinia as a

2 marine mass extinction mechanism

1 2,3 1 3 Dominik Hulse,¨ Kimberly V. Lau, Sebastiaan J. van de Velde,

4 5 1 4 Sandra Arndt, Katja Meyer, Andy Ridgwell

1 5 Department of Earth and Planetary Sciences, University of California, Riverside, CA, USA

2 6 Geosciences and Earth and Environmental Systems Institute, The Pennsylvania State University,

7 University Park, PA ,USA

3 8 Geology and Geophysics, University of Wyoming, WY, USA

4 9 Bgeosys, Geoscience, Environment & Society, Universite´ Libre de Bruxelles, Brussels, Belgium

5 10 Earth and Environmental Sciences, Willamette University, Salem, OR, USA

11 Extreme warming at the end-Permian induced profound changes in marine biogeochemical

12 cycling and animal habitability, leading to the largest extinction in Earth’s history. However,

13 a causal mechanism for the extinction that explains the different proxy evidence has yet to

14 be found. By combining recent modeling developments with global and local redox obser-

15 vations, we show that a temperature-driven increase in microbial respiration can reconcile

16 reconstructions of the spatial distribution of euxinia and seafloor anoxia spanning the Per-

17 mian/Triassic transition. We illustrate how enhanced metabolic rates would have strength-

18 ened upper ocean nutrient recycling, and thus shoaled and intensified the minimum

19 zones eventually causing euxinic waters to expand onto continental shelves, poisoning ben-

20 thic habitats. Finally, we find that the temperature effect on microbial activity can account

1 21 for some of the decline in carbon isotopes at the end-Permian with the implication that car-

22 bon release as inferred from those changes is likely overestimated. Our findings present

23 a novel view of the sensitive interconnections between temperature, microbial metabolism,

24 ocean redox state and carbon cycling during the end-Permian mass extinction with potential

25 far-ranging implications for the interpretation of carbon cycle perturbations during Earth

26 history.

27 Climate warming driven by volcanic greenhouse gas release is widely regarded as the under-

28 lying driver for the largest metazoan extinction event in Earth’s history at the end of the Permian

1, 2 29 Period when ∼90% of marine species were eliminated . Although the extinction event itself has

30 been intensely studied and is relatively well characterized, the specific physical and/or biogeo-

3 31 chemical environmental changes that drove biodiversity loss in the ocean are uncertain . Proxy

◦ 4–6 32 evidence reveals a 7-10 C increase in sea surface temperature occurring in as little as ∼39 kyr

7 33 (Fig. 1a+d), the development of (photic-zone) euxinia (waters containing sulphide ), an expan-

8, 9 34 sion in the extent of seafloor anoxia , and a decrease in the carbon isotopic signature recorded in

10 13 11 35 carbonates (δ Ccarb, Fig. 1e), all approximately reaching their nadir at the extinction horizon

36 (EH, ∼251.94 Ma).

37 Proposed explanations linking these observations with the extinction all require reduced oxy-

38 genation of the ocean, but fundamentally diverge at this point. In particular, previous 3D Earth

39 system model (ESM) studies have required either a sustained collapse of global ocean circulation

12 40 in conjunction with a much weaker , or a well ventilated end-Permian ocean

2 41 in conjunction with a much stronger biological pump driven by enhanced nutrient (phosphate)

13, 14 15 42 availability . However, other modelling work has demonstrated that increasing the ocean

43 phosphate inventory sufficiently to create widespread subsurface euxinia, not only requires an ex-

16 44 cessive increase in phosphate availability, but also results in near-global anoxia in the deep sea

9,17 45 which is in conflict with paleoredox estimates from uranium isotope records (Fig. 1b+c). Other

46 explanations focus on reducing oxygen availability throughout the ocean as a whole, either through

13 47 the oxidation of methane released from hydrates or of a massive reservoir of dissolved organic

18 19, 20 48 matter , or via warming driven by the CO2 release associated with volcanism . These car-

13 49 bon release mechanisms can also drive a pronounced decline in δ Ccarb. However, a gradient in

13 21 50 δ Ccarb from surface to subsurface records suggests a vertical partitioning in ocean geochem-

51 istry occurred and that there is much more to the event than carbon release and climate warming.

52 Here we reconcile the varying proxy observations and provide new insights into the controls on

53 subsurface euxinia and hence marine extinction mechanisms, by recognising the important role of

54 temperature in stimulating microbial respiration and thus dictating the redox profile in the water

55 column.

56 Oxygen availability in the water column generally decreases from well-mixed surface waters

57 (few tens of meters), to the oxygen minimum zone (OMZ, typically at a depth of a few hundreds

58 of meters). Along with vertical ocean mixing, this gradient is controlled by the remineralization

59 of particulate organic matter (POM) which consumes oxygen and releases inorganic (and

22 60 carbon) that can be returned to the surface to fuel new primary productivity . Critical here is the

61 interplay between reactivity and sinking rate of POM, as it controls the shape of the remineraliza-

3 62 tion rate depth profile (i.e. the scale depth of POM remineralization) and, thus, the intensity and

16, 23 63 depth of OMZs . We posit that a further factor, and the key to understanding how the marine

24 64 environment changed across the Permian/Triassic boundary (P/Tr), is ocean temperature .

65 A mechanistic representation of the biological pump

66 To demonstrate the importance of a warming ocean in driving subsurface euxinia and potentially

67 widespread extinction across shallow marine environments, we simulate redox distributions for a

25, 26 68 range of P/Tr conditions using the cGENIE ESM . We modify the widely used ’static’ represen-

27 69 tation of the biological pump (i.e. an invariant POM remineralization depth profile ), to allow us

70 to explicitly account for the impact of ocean warming at the end-Permian on POM remineralization

71 (SI). In addition, to reflect the shallower remineralization profile inferred prior to the rise of pelagic

23 72 calcifiers in the early Mesozoic , we decrease the sinking rate of POM in the model by ∼22% –

28 73 scaled to the smaller mean animal biovolume at the end-Permian ( , SI). Finally, we account for

74 changes in POM remineralization as reactive POM compounds react with sulphide (H2S) and be-

29 75 come less susceptible to bacterial remineralization (SI), in a process known as “sulfurization” .

76 Because the time-scale of warming leading up to the P/Tr boundary is slow relative to the adjust-

77 ment time-scale of large-scale ocean circulation (i.e. warming likely occurred over ∼39 kyr or

6, 11 78 more, starting in the C. meishanensis biostratigraphic zone ), a series of (10 kyr) steady-state

79 simulations is appropriate for simulating the biogeochemical response to warming. In these, we

80 prescribe a range of atmospheric pCO2 (1 – 30 × pre-industrial pCO2, i.e. 280–8400 ppmv) chosen

81 to span the increase in tropical Tethys ocean temperatures reconstructed from proxy records (∼22

◦ 82 – 35 C, Fig. 1a+d, SI). Simultaneously, we explore the importance of 1 – 2.5× modern ocean

4 83 phosphate inventories to represent the potential net impact of increased and

30–32 84 regeneration rates as the climate warms and ocean anoxia increases, respectively – thus, creat-

85 ing model ensembles of varying climate vs. ocean nutrient state. Atmospheric oxygen is fixed at

86 modern.

87 Constraining model results with global and local redox proxies

88 Uranium isotopes can provide powerful constraints on ocean models via the reconstructed extent

89 of seafloor anoxia (fanox) . To quantitatively compare the ESM results with a compilation of car-

238 90 bonate δ U data over the P/Tr transition (SI), we use a forward box model that encapsulates the

9 91 uncertainties in the U isotope budget (adapted from Lau et al. ). According to our U-model results,

238 92 the δ U data can be best explained by an abrupt increase in fanox that either coincided with the

93 EH, or preceded it by much less than the onset of the warming event (Fig. 1 b+c). An increase

94 of fanox from a modern value of 0.6% to about 30% (i.e. a factor of 50) represents our preferred

95 scenario as other perturbations fail to simulate the rate of change and magnitude of the shift in the

238 96 δ U data (Extended Data Fig. 4).

97 We also ground-truth the ESM results with a new compilation of local redox proxies (Ex-

98 tended Data Fig. 1 and Extended Data Table 2). The data-set consists of geochemical, lithologic,

99 sedimentologic, and biomarker evidence for water-column euxinia and bottom water anoxia and

100 distinguishes three phases of the P/Tr transition (Late Permian background, Start Warming and

101 Main Extinction, Extended Data Table 2). The Late Permian data only indicate photic-zone eu-

102 xinia occurring at the Meishan section with no clear evidence for seafloor anoxia (Fig. 3a+g,

5 103 Extended Data Table 2). Immediately before the EH, water-column euxinia expanded to sections

104 in British Columbia (BC), Shangsi, and potentially to the equatorial Panthalassic (Japan), while

105 unequivocal evidence for bottom water anoxia remains limited and only exists for BC (Fig. 3b+h).

106 In contrast, during the Main Extinction, anoxia spread across much of the shallow seafloor, and

107 multiple lines of evidence suggest widespread euxinia impinged on shallow marine habitats (Fig.

108 3c+i).

109 Impact of temperature-driven respiration on the marine redox-landscape

110 Our model experiments show that for a ’static’ biological pump, global POM cycling (i.e. POM

111 export production, rain and burial rates within the sediment) is almost entirely controlled by PO4

112 availability, with a minor increase at higher temperatures attributable to temperature-dependent

113 productivity (Fig. 2a–c). Global mean photic-zone [H2S] shows little temperature sensitivity and

−1 114 remains below 4µmol kg (Fig. 2d) and the depth of maximum [H2S] is generally below 400m

115 (Extended Data Fig. 7). At the same time, the extent of seafloor anoxia increases to values above

116 30% even at moderately high warming and nutrient increases (Fig. 2e). The relative lack of

117 sensitivity of shallow euxinia vs. deep anoxia cannot easily be reconciled with observations for

118 photic-zone euxinia (Extended Data Fig. 7) and expanded seafloor anoxia (Fig. 1b+c), respec-

119 tively.

120 In contrast, a temperature-dependent biological pump results in more efficient POM reminer-

121 alization and nutrient recycling in the upper ocean, which significantly enhances export production

122 (2f, and Extended Data Fig. 5a–d). This positive feedback is further intensified by the assumed

6 23 123 slower-than-modern sinking rate in the Paleozoic (Extended Data Fig. 5c), as it increases the

124 residence time of POM in the upper ocean, leading to more complete remineralization and nutri-

125 ent recycling. Although POM export production increases by up to a factor of three compared to

126 the static model, the POM rain to the does not change significantly (Fig. 2g) due to

127 the temperature driven, efficient POM remineralization in the water column. Mean photic-zone

−1 128 [H2S] is substantially higher (reaching values > 60µmol kg ) and highly sensitive to tempera-

129 ture variations (Fig. 2i). Notably, higher temperatures cause an increase in upper ocean [H2S]

130 and a concomitant shallowing of the – even without the need to invoke any change in

131 nutrient inventory (Extended Data Fig. 8). In addition, increased ocean euxinia leads to higher

132 sulfurization rates causing POM burial rates to increase (Fig. 2h) and the extent of seafloor anoxia

133 to decrease (compare Fig. 2j+e; S6), a negative feedback that may be important in the transition

33 134 back to a more oxygenated ocean .

135 Our model predictions for the spatial distribution of euxinia and seafloor anoxia appear in

136 general agreement with the available evidence for local redox-conditions (Fig. 3) and our U-model

137 results (Fig. 1b+c). In our “Late Permian background” scenario (i.e. lower temperatures and

138 modern ocean [PO4]), only a few very local environments developed [H2S] around 20–30µmol/kg

139 and are limited to depths below 284m (Fig. 3a+d), similar to profiles in the modern Cariaco

34 140 Basin . Instances of upper ocean euxinia were very limited in spatial extent and only occured in

141 the Eastern Equatorial Panthalassic and the Eastern Tethys (South China). The simulated extent

35 142 of bottom water anoxia, 0.7% of the seafloor, is comparable to the modern value of 0.6% ( , Fig.

143 3g). Consistent with proxy reconstructions, climate warming together with a moderate release

7 144 of nutrients (i.e. the “Start Warming” phase), resulted in increasing [H2S], with sulphidic waters

145 impinging onto continental shelves and slopes, especially in warm equatorial waters (Fig. 3b+e).

146 While most of the global seafloor remains oxic (fanox =1.7%), approximately 20% of the seafloor

147 shallower than 1000m exhibits [O2] <60µmol/kg (Fig. 3h, Extended Data Table 4), a typical

148 threshold for hypoxic conditions and considered critical for the survival of many modern marine

22 149 animals . A second increase in temperature together with another slight relaxation of nutrient-

◦ 150 limitation (i.e. a ∼7 C warming of SSTs and a doubling of ocean [PO4] compared to modern) is

151 sufficient to simulate the “Main Extinction”. This scenario satisfies not only local evidence for an

152 expansion of upper ocean euxinia (Fig. 3c) but also predicts more widespread seafloor anoxia that

238 153 has spread to the abyssal plain (Fig. 3i), in broad agreement with inferences from δ U data (Fig.

154 1b+c). The habitable area in the upper ocean is further restricted as now ∼30% of the seafloor

155 above 1000m exhibits hypoxic conditions (Extended Data Table 4).

156 Implications for ocean-redox and carbon cycle dynamics

6 157 Given the relatively prolonged interval of warming in the lead up to the P/Tr boundary (∼39 kyr ),

158 our analysis favors a conceptual model in which higher metabolic rates drive a pronounced verti-

159 cal partitioning in ocean redox. This biogeochemically driven partitioning occurs within an ocean

13, 36 160 that remains not only well ventilated at depth, but consistent with previous modelling work ,

161 increases in overturning strength for warmer states (Figure 3j-l). The temperature impact on

162 metabolic rates alone is hence sufficient to provide an effective kill mechanism for the upper

163 ocean ecosystem. This is in contrast with a physical partitioning such as simulated by Penn et

12 164 al. using the Community Earth System Model (CESM), whereby instantaneous warming drives

8 165 persistent (multi kyr) stratification of the ocean. These two conceptual (and numerical) models

166 also fundamentally differ in the consequent predictions of biological export – vigorous circulation,

167 in conjunction with temperature-driven nutrient recycling substantially enhances export, whereas

168 stratification severely restricts and hence export – with important implications for the

169 stressors associated with the marine extinction. Given that the initial ocean circulation response of

170 the cGENIE ESM to instantaneous warming is very similar to that of CESM immediately after the

171 perturbation (Extended Data Fig. 10), it is clear how critical the time-scale of warming is to fully

172 understand the mechanistic link between anoxia and the extinction, highlighting the importance of

173 further work in refining age models and temperature reconstructions for the latest Permian.

174 Our inference of a metabolically-driven – rather than transient circulation-driven redox par-

4 175 titioning, allows for persistently high temperatures during the Early Triassic to prolong the oc-

9 37 176 currence of extensive seafloor anoxia . Episodic volcanism in the Early Triassic and consequent

177 oscillations between sulfidic and oxic conditions potentially played a role in the protracted biogeo-

178 chemical and biological recovery prior to the Middle Triassic. Our finding that the temperature

179 response of nutrient recycling and ocean redox is somewhat amplified in the model under a slower

180 POM sinking rate also implies that the severity of marine extinction should decrease through the

181 Phanerozoic. Specifically, the advent of pelagic biomineralization in the early Mesozoic Marine

38 23 182 Revolution , leading to a step-increase in POM sinking rates , may have decreased the sensitivity

183 of ocean anoxia to fluctuations in global climate and hence volcanism-related carbon release in the

39 184 latter half of the Phanerozoic .

9 13 185 Finally, our numerical model analysis sheds new light on the δ C decline itself, and specif-

186 ically the cause for substantive variability in what is nominally a globally imprinted signal. Dif-

187 ferences in onset timing, temporal evolution, and maximum isotopic change of the end-Permian

13 10 188 δ C decline have been observed among stratigraphic sections (e.g. the amplitude varies be-

189 tween 4 and 7h), complicating estimates of the size and source of the carbon cycle perturbation.

13 190 We simulate a maximum decrease in the Tethys Ocean subsurface δ C during the P/Tr of 3.9h

13 191 – approximately 60% of the observed δ C decrease in Armenia and Iran, with a smaller shift

192 occurring in colder waters (Fig. 1e). This decline is driven only by temperature-induced changes

193 to the biological pump in the model without invoking changes in the ocean+atmosphere carbon

194 inventory (see SI). The implication is that end-Permian carbon release as inferred from observed

13 3 195 δ C changes (e.g. Erwin ) is likely overestimated, with the total release being smaller and/or the

196 isotopic composition of the source is less negative, than previously assumed. This implies that

197 carbon from volcanism rather than a reduced source (e.g. organic matter) was more dominant. In

198 sum, these results reassess the relationship between temperature and the ocean redox state, and

13 199 carbon cycle perturbations recorded by δ C.

10 200201 1. Payne, J. L. & Clapham, M. E. End-Permian Mass Extinction in the Oceans: An Ancient

202 Analog for the Twenty-First Century? Annual Review of Earth and Planetary Sciences 40,

203 89–111 (2012).

204 2. Burgess, S. D., Muirhead, J. D. & Bowring, S. A. Initial pulse of Siberian Traps sills as the

205 trigger of the end-Permian mass extinction. Nature Communications 8, 1–6 (2017).

206 3. Erwin, D. H. Extinction: How Life on Earth Nearly Ended 250 Million Years Ago-Updated

207 Edition, vol. 37 (Princeton University Press, 2015).

208 4. Sun, Y. et al. Lethally Hot Temperatures During the Early Triassic Greenhouse. Science 338,

209 366–370 (2012).

210 5. Joachimski, M. M. et al. Climate warming in the latest Permian and the PermianTriassic mass

211 extinction. Geology 40, 195–198 (2012).

212 6. Joachimski, M. M., Alekseev, A. S., Grigoryan, A. & Gatovsky, Y. A. Siberian Trap volcan-

213 ism, global warming and the Permian-Triassic mass extinction: New insights from Armenian

214 Permian-Triassic sections. GSA Bulletin 132, 427–443 (2020).

215 7. Grice, K. et al. Euxinia During the Permian-Triassic Superanoxic Event. Science

216 307, 706–709 (2005).

217 8. Brennecka, G. A., Herrmann, A. D., Algeo, T. J. & Anbar, A. D. Rapid expansion of oceanic

218 anoxia immediately before the end-Permian mass extinction. Proc. Natl. Acad. Sci. U.S.A.

219 108, 17631–17634 (2011).

11 220 9. Lau, K. V. et al. Marine anoxia and delayed Earth system recovery after the end-Permian

221 extinction. Proc. Natl. Acad. Sci. U.S.A. 113, 2360–2365 (2016).

222 10. Korte, C. & Kozur, H. W. Carbon-isotope stratigraphy across the PermianTriassic boundary:

223 A review. Journal of Asian Earth Sciences 39, 215–235 (2010).

224 11. Burgess, S. D., Bowring, S. & Shen, S.-z. High-precision timeline for Earths most severe

225 extinction. Proc. Natl. Acad. Sci. U.S.A. 111, 3316–3321 (2014).

226 12. Penn, J. L., Deutsch, C., Payne, J. L. & Sperling, E. A. Temperature-dependent hypoxia

227 explains biogeography and severity of end-Permian marine mass extinction. Science 362,

228 eaat1327 (2018).

229 13. Winguth, A. M. E. & Maier-Reimer, E. Causes of the marine productivity and oxygen changes

230 associated with the PermianTriassic boundary: A reevaluation with ocean general circulation

231 models. Marine Geology 217, 283–304 (2005).

232 14. Winguth, C. & Winguth, A. M. E. Simulating PermianTriassic oceanic anoxia distribution:

233 Implications for species extinction and recovery. Geology 40, 127–130 (2012).

234 15. Meyer, K. M., Kump, L. R. & Ridgwell, A. Biogeochemical controls on photic-zone euxinia

235 during the end-Permian mass extinction. Geology 36, 747–750 (2008).

236 16. Meyer, K. M., Ridgwell, A. & Payne, J. L. The influence of the biological pump on ocean

237 chemistry: implications for long-term trends in marine redox chemistry, the global carbon

238 cycle, and marine animal ecosystems. Geobiology 14, 207–219 (2016).

12 239 17. Zhang, F. et al. Multiple episodes of extensive marine anoxia linked to global warming and

240 continental weathering following the latest Permian mass extinction. Science Advances 4,

241 e1602921 (2018).

242 18. Rothman, D. H. et al. Methanogenic burst in the end-Permian carbon cycle. Proc. Natl. Acad.

243 Sci. U.S.A. 111, 5462–5467 (2014).

244 19. Cui, Y., Kump, L. R. & Ridgwell, A. Initial assessment of the carbon emission rate and

245 climatic consequences during the end-Permian mass extinction. Palaeogeography, Palaeocli-

246 matology, Palaeoecology 389, 128–136 (2013).

247 20. Clarkson, M. O. et al. Ocean acidification and the Permo-Triassic mass extinction. Science

248 348, 229–232 (2015).

249 21. Meyer, K. M., Yu, M., Jost, A. B., Kelley, B. M. & Payne, J. L. δ13C evidence that high pri-

250 mary productivity delayed recovery from end-Permian mass extinction. Earth and Planetary

251 Science Letters 302, 378–384 (2011).

252 22. Keeling, R. F., Kortzinger,¨ A. & Gruber, N. in a Warming World.

253 Annual Review of Marine Science 2, 199–229 (2010).

254 23. Lu, W. et al. Late inception of a resiliently oxygenated upper ocean. Science 361, 174–177

255 (2018).

256 24. Lopez-Urrutia,´ A., Martin, E. S., Harris, R. P. & Irigoien, X. Scaling the metabolic balance of

257 the oceans. Proc. Natl. Acad. Sci. U.S.A. 103, 8739–8744 (2006).

13 258 25. Ridgwell, A. et al. Marine geochemical data assimilation in an efficient Earth System Model

259 of global biogeochemical cycling. Biogeosciences 4, 87–104 (2007).

260 26. Hulse,¨ D., Arndt, S., Daines, S., Regnier, P. & Ridgwell, A. OMEN-SED 1.0: a novel, nu-

261 merically efficient organic matter sediment diagenesis module for coupling to Earth system

262 models. Geoscientific Model Development 11, 2649–2689 (2018).

263 27. Hulse,¨ D., Arndt, S., Wilson, J. D., Munhoven, G. & Ridgwell, A. Understanding the causes

264 and consequences of past marine carbon cycling variability through models. Earth-Science

265 Reviews 171, 349–382 (2017).

266 28. Heim, N. A., Knope, M. L., Schaal, E. K., Wang, S. C. & Payne, J. L. Copes rule in the

267 evolution of marine animals. Science 347, 867–870 (2015).

268 29. Sinninghe Damste,´ J. S., Kok, M. D., Koster,¨ J. & Schouten, S. Sulfurized carbohydrates:

269 an important sedimentary sink for organic carbon? Earth and Planetary Science Letters 164,

270 7–13 (1998).

271 30. Walker, J. C. G., Hays, P. B. & Kasting, J. F. A negative feedback mechanism for the long-

272 term stabilization of Earth’s surface temperature. Journal of Geophysical Research: Oceans

273 86, 9776–9782 (1981).

274 31. Van Cappellen, P. & Ingall, E. D. Benthic phosphorus regeneration, net ,

275 and ocean anoxia: A model of the coupled marine biogeochemical cycles of carbon and phos-

276 phorus. Paleoceanography 9, 677–692 (1994).

14 277 32. Schobben, M. et al. A nutrient control on marine anoxia during the end-Permian mass extinc-

278 tion. Nature Geoscience 1–7 (2020).

279 33. Hulse,¨ D., Arndt, S. & Ridgwell, A. Mitigation of Extreme Ocean Conditions by

280 Organic Matter Sulfurization. Paleoceanography and Paleoclimatology 34, 476–489 (2019).

281 34. Li, X. et al. Particulate sulfur species in the water column of the Cariaco Basin. Geochimica

282 et Cosmochimica Acta 75, 148–163 (2011).

283 35. Helly, J. J. & Levin, L. A. Global distribution of naturally occurring marine hypoxia on

284 continental margins. Deep Sea Research Part I: Oceanographic Research Papers 51, 1159–

285 1168 (2004).

286 36. Montenegro, A. et al. Climate simulations of the Permian-Triassic boundary: Ocean acidifi-

287 cation and the extinction event. Paleoceanography 26 (2011).

288 37. Payne, J. L. & Kump, L. R. Evidence for recurrent Early Triassic massive volcanism from

289 quantitative interpretation of carbon isotope fluctuations. Earth and Planetary Science Letters

290 256, 264–277 (2007).

291 38. Ridgwell, A. A Mid Mesozoic Revolution in the regulation of ocean chemistry. Marine

292 Geology 217, 339–357 (2005).

293 39. Bachan, A. et al. A model for the decrease in amplitude of carbon isotope excursions across

294 the Phanerozoic. American Journal of Science 317, 641–676 (2017).

15 295 40. Elrick, M. et al. Global-ocean redox variation during the middle-late Permian through Early

296 Triassic based on uranium isotope and Th/U trends of marine carbonates. Geology 45, 163–

297 166 (2017).

298 41. Zhang, F. et al. Congruent Permian-Triassic δ238u records at Panthalassic and Tethyan sites:

299 Confirmation of global-oceanic anoxia and validation of the U-isotope paleoredox proxy. Ge-

300 ology 46, 327–330 (2018).

301 42. Zhang, F. et al. Global-ocean redox variations across the Smithian-Spathian boundary linked

302 to concurrent climatic and biotic changes. Earth-Science Reviews 195, 147–168 (2019).

303 43. Schobben, M., Joachimski, M. M., Korn, D., Leda, L. & Korte, C. Palaeotethys seawater tem-

304 perature rise and an intensified hydrological cycle following the end-Permian mass extinction.

305 Gondwana Research 26, 675–683 (2014).

16 18 δ Oapatite (‰ VSMOV)● ●

22 21 20 19● 18 ●

● ● Armenia ● Iran ● (a) ● (b) (c) SLM cubic fit ● ● LOESS fit ● ● ● ●●● U/Pb dates ●● ● ● Bed 28 ● ● ●● ● ● ● PTB ● ● Dajiang ● ● ● Zuodeng ● ● ● ● Guandao Bed 25 ● ● ● Daxiakou ● Zal ● ● Taşkent Age (Ma) ● ● Jesmond ● ● ● Induan ● Kamura ●● ● ●● ● ●●● ● ●● ● ●● ●●●●● ●● ●●● 252 251●● ● 250 ● ●●●● ●● ● ●● ● ● ●● (d) (e) Chang. Olenekian Permian Triassic Bed 22

20 25 30 35 40 −0.6 −0.2 0 10 20 30 40 252.120 252.0 25 30 251.9 35 251.8 40 -2 0 2 4 238U 13 Temperature (°C) δ fanox Temperature (°C) δ Ccarb

Figure 1: Temporal relationship of changes in isotope records and U mass balance modeling

for the P/Tr extinction. (a) Temperature reconstructions from the δ18O of biogenic apatite from

South China (black circles)4. (b) Carbonate δ238U data (colored symbols)9,17, 40–42. U isotope for- ward model results are shown, with gray lines representing individual iterations from the Monte

Carlo routine with variable isotopic fractionation into anoxic sediments, riverine δ238U, and car- bonate diagenetic offset. The black line represents the moving average of all model runs (SI). (c)

Varying perturbations in seafloor anoxic fraction (fanox) that drive the U-model in panel b (indi- vidual iterations, grey lines; average, black line). (d + e) Conodont apatite δ18O and δ13C6,43 (SI).

Grey shading in (e) represents the range of cGENIE δ13C values simulated for the Tethys Ocean for the experiments shown in Fig. 3.

17 Export production POM rain POM burial Mean photic-zone [H2S] Bottom water anoxia (PgC yr-1) (PgC yr-1) (PgC yr-1) (μmol kg-1) (%) 2.5 2.5 2.5 2.5 2.5 20 1.6 1.8 0.5 0.7 70 2 3 80 60 (a) (b) (c) 0.4 0.6 (d) (e) 1.6 1 70 50 3 1.4 3 3 3 2 60 3 2 15 2 2 0.3 0.5 2 2

modern 1 40 80 15 1.4 0.4 50 1.2 0.2 70 4 1 30 40 0.3 60 2 2 1.2 2 2 20 30 2 1.5 1.5 1 1.5 0.1 1.5 1.5 50 1 0.2 10 20 10 40 10 30

Static ModelStatic 10 0.1 Ocean PO 0.8 1 1 0.8 1 1 1 1 1 1 1 1 24 26 28 30 32 34 24 26 28 30 32 34 24 26 28 30 32 34 24 26 28 30 32 34 24 26 28 30 32 34

2.5 2.5 2.5 0.9 2.5 2.5 1.8 1 30 70 50 1.6 0.8 (f) 25 35 (g) (h) 0.9 (i) (j)30 20 60 1.6 0.7 60 20 1.4 0.8 30 50 45 0.6 0.7 40 30 3 1.43 3 3 10 3 2 2 2 0.5 0.6 2 50 2 1.2 10 20 30 modern 40 1 10 20 1.2 0.4 0.5 40 10

4 20 1 0.3 0.4 25 1 35 2 2 2 0.3 2 2 1.5 15 1.5 0.8 1.5 0.2 1.5 1.5 30 0.8 30 0.1 0.6 0.2 10 20 Ocean PO 20 1 25 1 0.6 1 0.1 1 1 Dynamic Model Dynamic 1 1 1 1 1 24 26 28 30 32 34 24 26 28 30 32 34 24 26 28 30 32 34 24 26 28 30 32 34 24 26 28 30 32 34 Tethys SST (°C) Tethys SST (°C) Tethys SST (°C) Tethys SST (°C) Tethys SST (°C)

Figure 2: Global sensitivity of particulate organic matter (POM) cycling and ocean redox to temperature and nutrient changes for the static and dynamic representation of the biological pump: POM flux values (in PgC yr−1) of export production (a+f), POM rain to the seafloor (b+g)

−1 and POM burial in the sediments (c+h). Mean photic-zone [H2S] (in µmol kg , d+i) and bottom

−1 water anoxia (e+j) as fraction of seafloor area with [O2] ≤ 5µmol kg (in %). The red diamonds indicate results for the experiments presented in Fig. 3.

18 Late Permian background Start Warming Main Extinction 25.6.°C SST Tethys 29.5°C SST Tethys 33.0°C SST Tethys

2.5 x pCO2, 1.0 x [PO4] 7.5 x pCO2, 1.5 x [PO4] 20.0 x CO2, 2.0 x [PO4] 90 90 90 60 60 60 200 (a) (b) (c) 30 30 30

200 100 100 100 200 200 100 0 0 100 200 0 100 mol kg-1) 100 200 100 200 200

S < 1000m 100 μ 2 S (

-30 -30 -30 2 H -60 -60 -60

Max H 0 -90 -90 -90 -180 -90 0 90 180 -180 -90 0 90 180 -180 -90 0 90 180 90 90 90

S 60 60 60 81 2 (d) (e) (f) 30 30 30 175 284

0 0 0 411

558 -30 -30 -30 Depth (m) 729 -60 -60 -60 Depth of Max H Depth -90 -90 -90 928 -180 -90 0 90 180 -180 -90 0 90 180 -180 -90 0 90 180

90 90 200150 90 0 15010060 60 200 50 10060 5050 05

5 5 2 60 60 60 05 200 100 60 200 100150 5060 5 60 5 (g) (h) (i) 0 150 30 30 60 30 5

200 100 5 50 5 0 5 200

100 5

200 100 150100 200 200 150100 6050 5 5 150 0 0 60 50 0 50 100 60

50 100 50 mol kg-1)

5 0 150 5 μ

50

60 60 5 5 60 50 0

50 ( 150 Latitude Latitude Latitude 0 6050 60

5 2 200 60 5 -30 150 -30 -30 6050100 200 200 100 150 50 100 150 50 O 0.7% 1.7% 60 50 29.3% Bottom water O water Bottom -60 -60 150 -60 60 0 -90 -90 150 -90 100 -180 -90 0 90 180 -180 -90 0 90 180 -180 -90 0 90 180 Longitude Longitude Longitude 0 0 0 (j) (k) (l) 15 1 1 1 10 5 2 2 2 0 5 MOC (Sv)

Depth (km) Depth 3 3 3 10

Ocean circulation 15 4 4 4 90 60 30 0 30 60 90 90 60 30 0 30 60 90 90 60 30 0 30 60 90 Latitude Latitude Latitude

Figure 3: Ocean redox conditions during the P/Tr transition using the dynamic cGENIE model: (a-c): Simulated maximum [H2S] between 81 and 928m. (d-f): Depth where the maximum in [H2S] is observed. (g-i): Simulated extent of seafloor anoxia (values indicate fanox). Model results for [H2S] and [O2] are superimposed by proxy observations: Evidence for euxinia/anoxia is represented by circles; evidence against by crosses; ambiguous evidence or dynamic redox- conditions by triangles (see Extended Data Table 2). (j-l): Meriodional overturing circulation

(MOC, in Sv).

19 306 Acknowledgements We gratefully acknowledge data contributions from M. Joachimski and advice on

307 how to align proxy-records. D.H. is supported by a postdoctoral fellowship from the Simons Foundation.

308 S.J.V. is a NASA Postdoctoral Program fellow. D.H. and A.R. acknowledge support from the Heising-

309 Simons Foundation.

310 Author contributions D.H., K.V.L. and A.R. conceived the study; D.H. and A.R. designed and conducted

311 cGENIE experiments; K.V.L. adapted the U-model and conducted the experiments; K.V.L. and D.H. com-

312 piled and analyzed the proxy-data; All authors analyzed model output. D.H., K.V.L. and A.R. wrote the

313 manuscript with input from all authors.

314 Competing Interests The authors declare no competing interests.

315 Supplementary Information is available for this paper.

316 Correspondence and requests for materials should be addressed to D.H. (email: [email protected]).

317 Data availability

318 The locations of all data used in this study are provided in the supplementary information.

319 Code availability

320 The version of the code used in this paper is tagged as release v0.9.15 and has a DOI of 10.5281/zen-

321 odo.4008865. Necessary boundary condition files are included as part of the code release. Con-

322 figuration files for the specific experiments presented in the paper can be found in the installation

323 subdirectory: genie-userconfigs/MS/huelseetal.2020. Details of the experiments, plus the com-

324 mand line needed to run each one, are given in the readme.txt file in that directory. A man-

20 325 ual describing code installation, basic model configuration, and an extensive series of tutorials

326 is provided. The Latex source of the manual and pre-built PDF file can be obtained by cloning

327 (https://github.com/derpycode/muffindoc).

21 Figures

Figure 1

Temporal relationship of changes in isotope records and U mass balance modeling for the P/Tr extinction.

Figure 2

Global sensitivity of particulate organic matter (POM) cycling and ocean redox to temperature and nutrient changes for the static and dynamic representation of the biological pump. Figure 3

Ocean redox conditions during the P/Tr transition using the dynamic cGENIE model.

Supplementary Files

This is a list of supplementary les associated with this preprint. Click to download.

SIHuelseeaEndPermian.pdf